COMPARATIVE PHILOLOGY

AND THE TEXT OF

THE OLD TESTAMENT
With Additions and Corrections

BY

JAMES BARR

EISENBRAUNS
W inona Lake, Indiana
1987
©1968 by Oxford University Press
Postscript ©1987 by Eisenbrauns
All rights reserved
Reprinted 1987, with additions and corrections,
by permission of Oxford University Press
Printed in the United States of America

Library of Congress Cataloging‫־‬in‫־‬Publication Data

Barr, James.
Comparative philology and the text of the Old Testament.
Bibliography: p.
Includes index.
1. Hebrew language—Grammar, Comparative—Semitic.
2. Semitic languages—Grammar, Comparative—Hebrew.
3. Bible. O.T.—Criticism, Textual. I. Title.
PJ4564.B28 1987 221.415698‫׳‬4 87‫־‬
ISBN 0-9 3 1 4 6 4 1 ‫־‬33‫־‬
 PREFACE

T h e research for this book has been done over a number of years,
but the basic information was gathered in the excellent facilities of
the Speer Library at Princeton Theological Seminary. I owe
thanks to the staff of this library, and to Mr. Terence Fretheim,
who as a graduate student and departmental assistant worked on
the collection of data. My first attempt to present the problem in
public was at the meeting of the Catholic Biblical Association of
the United States in the summer of 1964. During 1965 my research
was greatly assisted by the generosity of the John Simon Guggen­
heim Memorial Fund, which enabled me to travel in the Near East
and study certain relevant linguistic problems. Among scholars
who have assisted me with advice and criticism I am particularly
indebted to the Revd. John A. Emerton, Reader in Semitic
Philology at Oxford and now Regius Professor-elect of Hebrew
at Cambridge. I owe much to the opinions of colleagues at
Manchester, especially Mr. P. R. Weis, Dr. Meir Wallenstein, Dr.
P. Wernberg-Moller, Dr. J. D. Latham, Dr. T. L. Fenton, and
the late Mr. Arie Rubinstein. Mr. W. G. Lambert of Birmingham
favoured me with an opinion on a point of Accadian. I have been
greatly helped by the excellent secretarial assistance provided by
Manchester University, in the persons of Mrs. Rowena Scaife
and her assistants.
J. B.
The University
Manchester
 CONTENTS

I. TEXTUAL TREA TM ENT AND

P H IL O L O G IC A L T R E A T M E N T 1

II. S O M E E X A M P L E S IN G R E A T E R D E T A IL 14
(1) ‫כלם‬ ‘speak* 14
(2) ‫טוב‬ ‘speech* 16
(3) ‫יקל‬ ‘be impudent, shameless* 17
(4) ‫ידע‬ not meaning ‘know* 19
(5) ‫דעה‬ ‘call* 23
(6) ‫להקה‬ ‘body of elders* 25
(7) ‫‘ בצ קלו‬fresh vegetables* 26
(8) ‫נשר‬ ‘herald* 26
(9) ‫אדם‬ ‘pleasant, delightful*and ‫‘ קול‬dig, bore’ 28
(10) ‫זמר‬ ‘protect* and ‫‘ עזי‬warrior* 29
(11) A Grammatical Example 30
(12) ‘Enclitic M em* 31
(13) Some General Statements 34

III. SO M E H IS T O R IC A L A S P E C T S 38
(1) T h e D isuse of H ebrew am ong the Jews 38
(2) L inguistic Elem ents in Jewish Interpretation 44
(3) Early Intra-linguistic Relations 50
(4) Aspects of the H istory of Jewish Gram m atical
Studies 60
(5) M ore Recent T rends 65

IV . A S P E C T S O F C O M P A R A T IV E
P H IL O L O G IC A L M E T H O D 76
(1) H istory 76
(2) Sound 81
(3) M eaning 86
(4) G eneral 92
 CONTENTS
V. P R E L IM IN A R Y Q U E S T IO N S IN
PH ILO LO G ICA L TR EA TM EN TS 95
(1) G eneral 95
(2) M etathesis and Dialect 96
(3) Loan-w ords and W ords of non-Sem itic O rigin 101
(4) Area Preferences w ithin Com parative Philology 111
(5) Problems of the Lexicographical T radition 115
(6) ‘Aramaisms* and Similar T erm s 121

V I. TH E D IS T R IB U T IO N OF H O M ON Y M S 125
(1) G eneral 125
(2) H om onym s and Com m unication 134
(3) T h e C ount of K now n H om onym s 145
(4) H om onym y and Style 151

VII. TH E D IS T R IB U T IO N OF LEXICA L
R E SO U R C E S IN T H E S E M IT IC
LANGUAGES 156
(1) G eneral 156
(2) Biliteral T heories 166
(3) Semantic Fields 170
(4) W ords w ith Opposite M eanings (’Addad) 173
(5) Patterning of Roots and Com patibility o f
Consonants 178
(6) W ords K now n T hrough Personal Names 181
(7) Lexicostatistics or G lottochronology 184

VIII. TH E M ASSORETES, V O C A LIZA TIO N AND

EM ENDATION 188
(1) G eneral 188
(2) Fallibility in the Consonantal T ex t 191
(3) T h e Im portance of the Vocalization 194
(4) Evidence for Pre-M assoretic Vocalization 207
(5) Conclusions 217

IX . LATE HEBREW AND THE LOSS OF

VOCABULARY 223
 CONTENTS IX
X. THE USE OF EVIDENCE FROM THE
VERSIO N S 238
(1) G eneral 238
( 2 ) T h e Q uestion of the H ebrew T ex t 245
(3) T h e Question of the Versional T e x t 247
(4) Im precise M ethods of T ranslation 249
(5) T h e U se of Favourite W ords 251
(6) Etymologizing 253
(7) Free Rew riting 255
(8) A dditional Points in the U se of the Versions 259
(9) U ncertainty about the M eaning of th e Version 262
(10) T h e Versions and the G ram m ar of the Original 265
(11) Conclusions 266

X I. SOME PA R T IC U L A R L IN G U IS T IC , LITER -
ARY, AND CULTURAL PROBLEMS 273
(1) Onomatopoeia 273
(2) Some L inguistic-cultural Relations 276
(3) Parallelism 277
(4) Religious Factors 282
(5) T h e A rgum ent from A ctuality 285

XII. SU M M ING -UP 288

A p p e n d ix : C o m p a r is o n o f S y r ia c a n d H e b r e w V er bs 305

ABBREV IA TIO N S 308

BIBLIOGRAPHY 310

INDEX OF EXAM PLES 320

INDEXES 338

POSTSCRIPT 355

PHILOLOGY AND EXEGESIS 362

UGARITIC AND HEBREW “SBM”? 388

LIMITATIONS OF ETYMOLOGY AS A LEXICOGRAPHICAL

INSTRUMENT IN BIBLICAL HEBREW 412
 I

TEXTUAL TREATMENT AND

PHILOLOGICAL TREATMENT
T he Hebrew manuscript text of the Old Testament shows a high
degree of uniformity. This characteristic constitutes a peculiarity
of Old Testament textual criticism, and provides a considerable
contrast with the situation found in many classical authors or in
the New Testament.
This uniformity, indeed, should not be exaggerated. Large col­
lections of variants were made by Kennicott and de Rossi. As all
investigators have recognized, however, most of these variations
are of a comparatively minor nature. While they undoubtedly
merit attention and may form useful evidence for the history of the
text, they are generally not such as to lead to, or provide the chief
clues for, the resolution of the major difficulties which have been
found in the reading of the Old Testament.
It is true that, with the discovery of the Qumran scrolls, access
has at last been gained to a Hebrew text which shows in some places
a substantial variation from the text previously known to us. This
qualification also, however, is limited in its effect. Only a smallish
portion of the Old Testament text has as yet been found at Qumran
and in the associated discoveries. Apart from the book of Isaiah,
the amount of text where real alternative controls have been made
available is still small. Moreover, the Qumran texts, being un­
vocalized, do not provide exactly the same kind of information as
the Massoretic text does. Again, many of the fragments, like the
later manuscripts, vary very little from the Massoretic text; and^
it is widely agreed that the text-type which we call the Massoretic
in a broad sense was already in existence in the Qumran period, or
the latter part of it.
Therefore we are justified in reiterating that the Hebrew text of
the Old Testament shows a striking uniformity in comparison with
the text of some other types of ancient literature. For a very large
proportion of the serious points of difficulty which the reader
encounters, and where he might pause and wonder whether the
 2 TEXTU A L TREA TM ENT AND PHILOLOGICAL TREA TM ENT
text is in order, the Hebrew manuscripts provide him with no
series of substantial alternatives from which to select.
The reader of the Old Testament, indeed, is not dependent
solely upon the Hebrew text of that literature. There are also the
ancient translations, ‘versions* as they are called in the technical
convention, in Greek, Latin, Aramaic, and other languages. These
translations were made at relatively early dates, in some cases
before the Christian era. Thus the Massoretic manuscripts are
antedated by several centuries by the origin of the versions, and
even by some of the extant manuscripts of them. A version, once
translated and in use, originated a line of transmission different
from that of the Hebrew text. This difference does not always mean
absolute separation; attempts were made to bring the Greek ver­
sion into line with the Hebrew text as it was in the post-Christian
era. But in general the versions enjoyed some considerable inde­
pendence from the historical development of the Hebrew text.
They have therefore always been highly valued in Old Testament
studies, for they provide that choice between substantial alterna­
tives which is so often lacking in the Hebrew text itself.
Nevertheless the scholar cannot use the ancient versions as if
they were actual Hebrew texts. The translators may have mis­
understood the original Hebrew, so that their version is not a good,
but a very bad, guide to what the original text said. Finding a
difficult passage in Hebrew, they may have just guessed at the
sense. They may not have translated literally, but have given a
rough paraphrase of what was said. They may be literal at one
place but paraphrastic at another; and in some versions, like the
LXX, the translating techniques differed from book to book, and
even between sections of books.
For these and similar reasons, the very high importance which
attaches to the ancient versions does not alter the fact that they are
not Hebrew manuscripts. The effect they have upon our thinking
is, in respect of directness and complexity, quite different from the
effect of manuscript evidence in Hebrew. The attention we give
to a reading in a version is usually proportionate to the degree of
difficulty which we find in the Hebrew text itself. In general,
therefore, the existence and the importance of the ancient versions
do not really alter the peculiar aspect to which I have drawn
attention, namely the relatively high uniformity of the Hebrew text
in points of substantial effect on meaning.
 TEXTUAL TREA TM ENT AND PHILOLOGICAL T REA TM ENT 3
This leads to an important distinction. The starting-point of a
textual discussion is different where the texts are divergent and
where the texts are uniform. If the text of a book varies, the normal
starting-point for a textual discussion will be the fact that different
readings exist. Given this fact, it obviously becomes important to
discuss which reading may have been the original, or to classify the
readings or assess them in some other way. This discussion has
to take place because the texts vary; its necessity is not removed if
several, or indeed if all, of the readings appear to lend a satisfactory
sense within the context.
Where there is no substantial variation in the text, however, a
textual discussion usually begins from a different starting-point,
which we can name only vaguely as a ‘difficulty’. The reader finds
‘a difficulty’ in the text which he is reading. He feels that it ‘does not
make sense’. The grammar is ‘wrong’, i.e. does not fit with usual
patterns of usage. The use of words is anomalous. Or perhaps the
text contradicts what is said elsewhere in the same literary work, so
that it seems to ‘spoil the effect’ of the whole; or it may contradict
something well known from altogether other sources. These are
simple examples of what is a ‘difficulty’.
Now so long as the reader is confident that the text is right, he
has to resolve his difficulties through various linguistic and literary
explanations: perhaps the grammar should be recognized as an
anomaly; perhaps we have a case of poetic licence, an unusual
meaning for a word, an ellipse of something usually expressed
explicitly, a metaphor, or an allegory with a hidden meaning. Or,
indeed, the reader may just give up, and decide that he does not
know the meaning and cannot know it with the data he now pos­
sesses.
The more sophisticated reader, however, will know that texts
may be wrongly transmitted; and, after trying the expedients men­
tioned above, or even before trying them, he may begin to suspect
an error in the text. If there is no manuscript reading to support
his conclusions, what he produces will be a conjectural emendation,
which he will support by arguing that it makes much better sense.
Even a conjectural emendation, however, will point out some
kind of relation between the reading conjectured and the text
actually found. There may be some features in common; or it will
be possible to show how the conjectured text, once misunderstood
or miswritten, could naturally have led to the text actually found;
4 TEXTUAL TREATM ENT AND PHILOLOGICAL TREA TM ENT
or some other such relation will be suggested. In other words, even
conjectural emendations are seldom purely conjectural in nature;
they normally take departure from, or have some logical link with,
some aspect of the existing text.
I repeat, then, my generalization. With a non-uniform text we
may find variant readings, and textual discussion begins from these
variant readings, even if all of them ‘make sense’. With a text of
high uniformity, however, textual discussion will more frequently
begin from the feeling that there is a ‘difficulty’; the procedure
will be more independent of the existence of variant readings, and
conjectural emendation will take a larger place in the discussion.
This may be illustrated vividly from the difference between Old
and New Testaments. In the latter it is quite common to find a
variant which differs substantially in meaning and yet each of the
readings ‘makes sense’. It is quite uncommon to find that by the
judgement of competent readers the passage does not ‘make sense’,
to the degree that the extant texts must be despaired of and resort
be made to a conjectural emendation found in none of them. Thus
conjectural emendation has been used with considerable restraint
in the New Testament. Metzger1 tells us that
the apparatus of the twenty-fourth edition of Nestle’s Greek New
Testament includes from various sources about 200 conjectures, 90 of
which are identified by the name of the scholar who first suggested them.
The average student, however, if we may hazard a guess, will
find it hard to recall a single one of these conjectures, unless it be
the addition at I Peter 3. 19 of the name ‘Enoch’ as the person who
preached to the spirits in prison. Not only for the average student,
but for the general current of New Testament scholarship, the
procedure of conjecture is decidedly a marginal one, considerably
more marginal than the recording of 200 examples would suggest.
It is in fact extremely seldom that a scholar judges the text so
desperately hopeless that a conjecture seems better than all the
attested variants.
In the Old Testament our average experience is just the oppo­
site. It is quite seldom that textual problems which are substantial
in point of meaning arise because variant Hebrew readings exist.
On the other hand it is quite normal experience to find that a read­
ing is almost unanimously supported by Hebrew manuscripts but
1 B. M. Metzger, The Text of the New Testament, p. 185.
 TEXTU A L TREATM ENT AND PHILOLOGICAL TREA TM ENT 5
that scholars turn to emendation to find a text which seems to be
viable. The well-known example at Ps. 2. 11-12 reads materially
alike in all Hebrew manuscripts:
. . . ‫ע ב ד ו א ת ״ י ה ו ה ב י ר א ה וגי לו ב ר ע ד ה נ ש קו ־ ב ר‬
This high unanimity of the manuscripts has been equalled, how-
ever, by the high degree to which scholars have preferred emenda-
tion. The text contains difficulties. If we take the AV and read:
‘Serve the Lord with fear, and rejoice with trembling.
Kiss the Son . . .’
the prima-facie difficulties are: (1)‫‘ ב ר‬son’ is not a Hebrew, but an
Aramaic, word; (2) the sequence ‘rejoice with trembling’ seems to
make poor sense. We need not here decide whether these considera-
tions are final. Let us observe that in fact scholars have generally
been inclined to favour such a rearrangement of the text as:
: ‫ י ה ו ה ב י ר א ה ב ר ע ד ה נ שקו ב ר ג ל י ו‬- ‫ע ב ד ו א ת‬
‘Serve the Lord in fear; in trembling kiss his feet.’
This removes both difficulties. It implies that corruption took place
by a fairly simple loss of sequence for a few letters, which explains
the corruption fairly well. It involves no very drastic rewriting of
letters in the text beyond this. Finally, it produces good con-
formity to known types of expression in Hebrew.
Thus the beginning of a textual discussion arises not primarily
from the existence of variant readings but from the perception of
difficulties in the Hebrew text. The solution just described does
not even have direct support from any of the versions, though these
differ from each other in their understanding of the text. The
position is not materially changed, however, when active use is
made of the versions.

Thus far I have tried to establish a characteristic of the situation

where a relatively uniform text exists, and to show that this is so in
the Old Testament. Now in this situation there are in principle two
quite different and indeed opposed methods which can be used for
the achievement of understanding. One I shall call textual and the
other I shall call philological. The distinction is of central impor-
tance for this book.
A textual treatment works on the hypothesis that an error has
6 TEXTUAL TREA TM ENT AND PHILOLOGICAL TREA TM ENT
occurred in the graphic transmission. A scribe has misread some
letters in the manuscript being copied, or has missed out some
words and later added them in the wrong place, or inserted a mar­
ginal note as if it was part of the text. If variant readings exist, they
probably imply such an error in the past; if there is no variation,
but conjectural emendation is used, the same possibilities are
implied as a hypothesis.
A philological treatment does not suggest a differing original
text, corrupted by graphic error; rather, it elucidates the meaning
of the existing text through the application of linguistic evidence
hitherto ignored. It thus justifies the existence of the rare or anom­
alous words which had constituted the original difficulty, and by
removing the difficulty it undercuts the foundations of the textual
treatment. The principal evidence used in a philological treatment is
the linguistic usage of the cognate languages, and with it the usage
of other stages of the same language, Hebrew. For a difficult form
in the existing text the scholar will consider words in cognate
languages which might be related. This consideration, if success­
ful, may suggest for the Hebrew form a meaning other than that
which has normally been acknowledged, and this new-found mean­
ing now removes the original difficulty.
Thus the philological treatment, compared with the textual,
means a shift from problems of writing, of scripts, of scribes, to
problems of meaning; it leads not to a text which has been copied
wrongly, but to a meaning which has been obscured or interpreted
wrongly. It does not rewrite the characters of the text but explains
them in a new way.
The limited scope of the Old Testament provides a reason in
favour of the philological approach. It is a comparatively small
body of literature; there is little direct external evidence for the
Hebrew of biblical times, and the post-biblical language has some
striking differences. Not surprisingly the Old Testament contains
many rare or unique expressions, which are difficult because they
are unexampled elsewhere in Hebrew. Is it not then natural to turn
to the large resources of the cognate languages, such as Arabic?
These may suggest words which in their form could be cognate
with a Hebrew form and which by their known meaning could
suggest a suitable sense for it.
Isa. 44. 8 has the expression If the text is right, the
verb is a hapax legomenon, so unexampled elsewhere that the
 TEXTUAL TREA TM ENT AND PHILOLOGICAL TREA TM ENT 7
dictionaries are uncertain what verb it is in the first place; some
register it as ‫ י ל ה‬, others as ‫ ל ה ה‬. Emendations have been suggested,
such as that to ‫ ת ל א ו‬, from the familiar ‫‘ י ל א‬fear’, and that to
‫ ת ל ה ב ו‬, which would mean ‘be arrogant’. The Qumran reading
‫ ת י ל א ו‬, though almost identical with the former emendation, does
not settle the matter, since it may have been caused by exactly the
same motives which led modern scholars to suggest the emenda-
tion. The question remains: is there any reason to suppose that a
form ‫ ת ל לו‬may have existed P1
If there is suitable evidence in the cognate languages, then our
form, though unique in the Bible, may yet in some sense be ex-
plained or justified. This is done in dictionaries which quote an
Arabic verb wariha, said to mean ‘be stupefied’ or the like.2 The
argument tacitly runs as follows: if this word, known in Arabic,
had a cognate in Hebrew, the latter would, in respect of form, come
close to the form of the M T ; moreover, in respect of meaning, the
sense known for the Arabic word, or a sense relatable to it by a
reasonably probable analogy, fits the contextual setting of the
Hebrew word. This being so, the Hebrew text has been both justi-
fied and explained; an account has been given of the form of MT,
and a meaning has been suggested which appears to fit the passage
well. The resources for this solution are supplied by comparative
philology.
Thus in principle the philological treatment, if right, cancels out
textual treatment; the question is not one of exploring hypothetical
scribal errors, but one of exploring the forms and meanings of
words in cognate languages.

Having established the difference between the two types of

treatment, one must go on to say that these are not distinct in the
sense that one must consistently follow one and ignore the other.
A competent worker must understand both, and, as we shall see,
it is very common for practice to mingle the two.
At any one point, however, the distinctness between the two
may be very marked. They move the scholar in exactly opposite
directions. The textual approach leads towards the detection of
a corruption in transmission and therefore, if the text is uniform,

1 Cf. below, pp. 166, 188, 231 f.

2 Freytag, iv. 459b; but does Freytag’s entry really justify this sense?
8 TEXTU A L TREA TM ENT AND PHILOLOGICAL TREA TM ENT
towards a conjectural emendation, with or without support from
the versions. The philological approach suggests the rightness of
the text (for without it there is no basis for a search in the cognate
languages) and leads on towards an exploration of resources,
hitherto neglected, in the other Semitic languages. Formally speak­
ing, the former should be registered in a critical apparatus to the
text; the latter should mean a modification to the Hebrew lexicon.
Thus, although the two methods are frequently mingled or their
differences obscured in a number of ways, it is important to observe
that at any one point they offer sharply distinct possibilities.
This is a reason for the writing of this book. For the textual
criticism of the Old Testament adequate guidance already exists
in the standard works of Wurthwein and Roberts. But the whole
weight of their work lies, and quite properly, on the textual
kind of treatment, and the philological approach receives only the
barest mention. Wurthwein devotes about a page (pp. 78 f.) to it,
mentioning one example (Hab. 3. 6f.); he does not discuss the
criteria or the ultimate implications of the method. The subject
requires not only to be mentioned, but to be discussed in detail,
with plentiful examples, and with consideration of the criteria by
which it can be guided, of the further implications which it carries,
and of the lines of new research which it appears to demand. In
this book the existing studies of the textual approach will be com­
plemented with a general survey of the philological approach and
of the interaction between the two.
Modern textual criticism, even when it adopts an ‘eclectic’ policy
in its decision about readings, does not work atomistically, but
depends on a general survey of the history of the manuscript
tradition and the modes by which alterations may occur. For
philological treatments, however, such a general discussion has
hitherto been lacking.
This is illustrated by the way in which the results of philological
treatments have been made public. Very often this has been in the
form of small and disparate notes, which are scattered throughout
the technical journals and neglect to discuss the problem system­
atically. Even dictionaries and commentaries, which might be
expected to gather together all that has been published in detailed
notes, have frequently not done so in fact. Sometimes scholars
proposing philological treatments seem to have paid little attention
to the proposals of other scholars on the same passages.
 TEXTUAL TREA TM ENT AND PHILOLOGICAL TREA TM ENT 9
If this breakdown in communication and discussion is a difficulty
for the scholar, it is a very much more serious one for the student,
who is much less able to judge for himself. He may be so much in
awe of the linguistic erudition of suggestions that he hesitates to
use his intellect critically upon them. This book is intended to
provide him with the necessary critical equipment.
Moreover, the discrimination of students is often hindered
rather than fostered by works devoted to the ‘flood of light’ (or
some such cliché) shed upon the Old Testament by modern dis­
covery. Such works often spend more time in admiring the advance
of knowledge about the cognate languages than in examining the
difficulties which attend the application of this knowledge. They
give the impression that, so long as advance in the knowledge of
Semitic languages is being made, one can somehow rest assured
that the elucidation of Old Testament passages formerly obscure
will follow. To suppose this, as we shall see, is to ignore some
awkward problems.
The fact that many attempts at philological treatment are to be
found only in scattered notes in the journals is one reason why this
present work was undertaken in the first place. I found it necessary
for my own information to gather these suggestions together ; this
I did simply by card-indexing. The mere existence of an index,
however, was itself an invitation to survey the principles and
problems involved. The juxtaposition of so many attempts could
not fail to raise questions which might be missed so long as one
looked at each suggestion individually.
For it should be observed that the number of philological treat­
ments has become very large. The question I raise is not a marginal
or an occasional one. In recent years interest in the philological
approach has so expanded that a competence in it may possibly
have become more important than the traditional training in the
textual approach. The passages so treated number many hundreds,
if not thousands, even if one discounts obviously incompetent or
fantastic solutions. It is not uncommon for a scholar in a brief article
to publish notes which claim to add to the Hebrew vocabulary a
score of words not previously recognized. One article of 34 pages
by Eitan has provided my index with over 40 suggested new words,
most of them otherwise quite unrecognized, from the book of
Isaiah alone.1
1 T he article is H U C A xii-xiii (1937-8) 5 5 8 8 ‫ ־‬.
Io TEXTUAL TREATM ENT AND PHILOLOGICAL TREATM ENT
It remains for this introductory chapter to say something about
the limitations which will be observed in this work.
Our subject is the application of philological means to elucidate
Old Testament passages which would otherwise be regarded as
obscure or corrupt. This book is not intended to be an introduction
to comparative philology itself. Certain aspects of this discipline,
however, call for special remark, and for this reason some discus-
sion of Semitic comparative philology will be given. This discussion
is not intended, however, to be a substitute for the actual study of
comparative philology.
Again, this book will not discuss the problems involved in the
wider study which might be called comparative literature. For
example, there are relations between Babylonian and Hebrew
creation stories. It is possible that this study of the larger literary
complexes, their content, themes, and styles is the most fruitful
way in which the Semitic background can be made to illustrate
the Old Testament. Nevertheless this form of research will not be
discussed here; we shall work not on the macrocosmic scale of the
literary forms, but on the microcosmic scale of the lexical forms.
Moreover, within the study of lexical forms, there are a number
of ways in which comparative data may be used. Sometimes com-
parative data may enrich with additional nuances our appreciation
of a word the basic sense of which is already known. Examples of
this, however, will not generally be discussed here. Again, some-
times a Hebrew word is quite familiar but comparative studies
have been invoked in order to identify more exactly the referent to
which the word is applied—for example, animal names like ‫ר א ם‬
or ‫נ ש ל‬. This also will not be discussed in this book. In such cases,
although an important service is rendered if the meaning of the
word can be more closely identified, the word was not in the first
place of such obscurity as to constitute for the reader a source of
doubt about his text; he knows that ‫ ר א ם‬is some sort of strong
animal and ‫ נ שר‬some kind of bird.
In general, then, this book concentrates mainly on the situation
where the use of comparative study is most critical, because with-
out it the reader would be likely to question his text. It does not
attempt to give equal attention to all the various kinds of contribu-
tion that comparative philology can make.
Readers should observe that some of the words and meanings
mentioned in this book have a somewhat hypothetical character.
 TEXTUAL TREA TM ENT AND PHILOLOGICAL TREA TM ENT 11
This follows from the nature of the enterprise; we are discussing
suggestions which have been put forward rather than facts which
are certainly known. If I mention that an Arabic word with a cer­
tain meaning has been cited in the course of a scholarly suggestion,
the reader should not conclude that I guarantee the accuracy of the
information cited. I have not intentionally left unremarked, how­
ever, anything which seems to me to be misleading. Nor have I,
except in special cases, used the asterisk which is conventional for
hypothetical or reconstructed forms.
For my purpose it has not been particularly important to trace
the scholar who first made a particular suggestion. Our question is of
the logical status, rather than the historical priority, of philological
treatments. I have not, therefore, made any great effort to discover
the first author of a suggestion, especially since, as I have found,
later workers often remained ignorant of earlier suggestions in any
case. Conversely, a treatment cited may sometimes later have been
abandoned by its author; but this makes no difference to its logical
status. I have cited treatments, where possible, in modern sources
of suitable form and accessibility.1 I apologize in advance if I seem
to have failed to give credit where it is due.
This book does not contain many philological treatments of my
own suggestion. The positive need of the present time is not the
production of larger numbers of such treatments, but the elucida­
tion of the criteria by which they can be sifted and evaluated. The
examples cited below are, I believe, a fair example of the contribu­
tion which philological study has made; but it may be more impor­
tant that we use facts critically than that we store facts themselves,
and more valuable that we should learn to appraise suggestions
than that we should produce them. The thoughtful restudy of past
scholarship is not criticism for the sake of criticism, but an attempt
to elucidate the principles involved in the discovery of truth. In
doing this, however, it is right that we express our gratitude and
respect to those whose work is being used and restudied, and with­
out whose pioneering zeal and daring the present evaluation could
not have been attempted.
Various chapters of this book discuss different criteria, which

1 Thus the idea that at Num. 23. 10 means ‘dust’ (Index, nos. 294-5)
can be found in a much earlier form in Jacob, Z A W xxii (1902) 111; but the
circumstances of the modem treatments are lacking, even if the suggestion has
the same result.
12 TEXTUAL TREA TM ENT AND PHILOLOGICAL TREATM ENT
may be applicable to the various examples cited. For many exam­
ples I have not been able to give a final decision at the point where
each is considered, because this would involve other criteria than
the one at present under discussion and would thus immensely
complicate the presentation. In order to avoid the impression that
a decision is being evaded, however, I have sometimes offered
a summary decision, even if I have not been able to state the
grounds for it in full. In any case the logical structure of the argu­
ment does not lie in the rightness or wrongness of individual
examples, each to be decided on its own merits, but in the elucida­
tion of the criteria.

Earlier in this chapter attention was drawn to a difference be­

tween scholarship in the field of the Greek classics or of the New
Testament and scholarship in the Old Testament field. To this
another difference may be added. It is the relative prominence of
languages other than the language in which the texts were written.
The Old Testament is a relatively restricted amount of text for
use as a sample for the language of the ancient Israelites. In
classical or Hellenistic Greek the bulk of text available is very
much greater. Interpretative discussion of specific points will
usually appeal to evidence observable elsewhere in extant Greek
literature; a scholar does not very often explain a word as one not
elsewhere extant in Greek, the existence of which may, however,
reasonably be postulated from a similar word in Sanskrit or
Lithuanian or Gothic. Only on the margins of Greek textual
scholarship are such explanations likely to occur, and the propor­
tion of them is surely infinitesimal in relation to the whole. In
Hebrew scholarship, however, such forms of explanation are quite
normal. Probably no body of ancient literature has benefited (or
suffered ?) so much as the Old Testament from explanations based
upon languages other than that in which it is written. The only
exceptions are literatures which have had no form of continuous
literary transmission from the ancient world into the modern, and
which, being discovered only by modern archaeological research,
have had to be deciphered anew wholly on the basis of comparative
data. Ugaritic is the prime example. The decipherment and inter­
pretation of the Ugaritic texts and others of the same kind are
indeed a triumph of scholarship. But the growing emphasis on
this kind of scholarship has affected our relationship to a language
 TEXTUAL TREA TM ENT AND PHILOLOGICAL TREA TM ENT 13
like Hebrew, in which we do have a transmission of meaning by
tradition from the past to the present. The increasing prominence
of other languages than Hebrew and the increasing dominance of
comparative methods have had effects upon the study of the Old
Testament which have not yet been fully investigated.
So far this subject has been considered with some generality.
Only one or two examples have been given. Our next step will be
to look in much greater detail at a number of examples, in order
to consider the methods and the logic by which they operate, and
to bring to light some of the questions which emerge from them.
 II

SOME EXAMPLES IN GREATER DETAIL

T h is chapter will be devoted to the setting out of some examples

of philological treatment of the Old Testament text. The examples
have been chosen to illustrate the various kinds of problems which
may arise and the various kinds of evidence which may be adduced.
In these respects they form a representative cross-section of the
suggestions which philological work has produced. Some of them
are, in my opinion, very convincing, while some others are doubt-
ful or improbable. At this stage, however, I have tried to hold back
my own judgement, and simply to present the arguments implied
in the treatment under review.

(1) ‫‘ כ ל ם‬speak‫׳‬
Judges 18. 7 has long been considered difficult. Five scouts of
the Danites came to Laish. They found the people living quietly,
with a sense of security; and then the text goes on:
‫־ו אי ך ס ? לי ם ד ב ר ב א ל ץ‬
Moore translated ‘there was no one to put them to shame (or,
insult them) in anything’; but he pronounced this to be ‘wholly
irrelevant’.1 Many scholars have emended the text to read:
‫ו אין מלזסור כ ל ־ ד ב ר ב א ר ץ‬
—a phrase the strength of which lies in the very similar locution
a few verses later (18. 10). The meaning would then be: ‘there was
no lack of anything in the land’. Moore himself preferred to emend
to ‫ מ כ ל א‬, giving the sense ‘there is no one to restrain (us) from
anything in the land’.
This discussion has thus far assumed for ‫ כ ל ם‬the sense ‘humil-
iate, insult, reproach’ which is normal in Hebrew. If, however, the
difficulty causes us to look for help in the cognate languages, we

1 Moore, Judges (ICC), p. 392.

 ‫‘ כ ל ם‬SPEAK* 15

at once think of the very common Arabic verb kallama ‘speak‫ ״‬and
the noun kalam ‘speech, word‫״‬.
This Arabic sense, when applied to the passage, gives what ap-
pears to be a good sense, ‘no,one uttering a word in the land’. This
fits well with the quiet security of life at Laish. The removal of
the difficulty thus abolishes the original ground for emendation. I
do not know which scholar first proposed this interpretation; it is
already known to Reider, who in 1954 uses the same evidence for
another difficult passage.
The testimony of the ancient versions may be added. Moore
stated baldly that ‘the versions give no help‫״‬. But in Judges the
LXX has two different versions. The B text has καταισχύνων,
which confirms the M T ‫ מ כ ל י ם‬by translating it with its usual
sense ‘make ashamed‫״‬. The A text, however, has και μη δυναμένους
λαλησαι ρήμα, and the λαλήσαι ‘to speak’ appears at first sight to
confirm the interpretation made from Arabic.1 This is not argued,
however, by Reider himself.
Reider applied this result to Mic. 2. 6. A prohibition, usually
taken to mean ‘do not preach‫״‬, is followed by the words: 1‫י ס‬
‫ני ל מו ת‬.
Reider not only finds here ‫‘ כ ל ם‬speech‫ ;״‬he holds the entire
phrase to be ‘really an Arabism‫״‬, corresponding to nasaja l-kalam
‘he forged speech‫״‬. The sense is ‘they shall not forge speeches‫״‬,
and is parallel to the earlier prohibition of preaching. Reider thus
identifies also a verb 102 ‘forge‫ ״‬in Hebrew. Previous scholarship
had some considerable uncertainty about the meaning, and emen-
dation has been tried.
We shall not decide whether Reider‫״‬s suggestion is right; we
note only some general characteristics of the method:
(a) The existence of a difficulty, with a previous resort to emen-
dation.
(b) A dependence on the text, and, accordingly, a rejection of
extensive emendation, as a starting-point for the philological
treatment.
(c) A use of the ancient versions as a source which may show
that a sense, now disclosed to us only through comparative
philological methods, was already known in antiquity.
1 Further examination, however, shows that the A text should not be inter-
preted in this way; it is not λαλησαι, but δνναμενονς, that comes from ‫ מ כ ל י ם‬. The
form in the Hebrew, whether identical with M T or not, was taken as from ‫ י כ ל‬.
16 SOM E EX A M PLES IN G R E A T ER D E T A IL

(2) ‫* טו ב‬speech’
Anyone who knows any Hebrew knows the familiar word ‫ ב‬1‫ט‬
*good’. There are certain places where this word appears not to
make very good sense. Hos. 14. 3 (EV 14. 2), a passage difficult
also in other respects, seems to suggest that the repentant man
should address God with the phrase ‫* ק ח ~ טו ב‬accept that which
is good’ (so RSV).
There seems to be something unsuitably condescending in the
idea that repentant sinners should ask God to accept that which is
good, and especially so in the teaching of Hosea. Moreover, the
beginning of the same verse has the phrase ‫ ד ב ר י ם‬DD/3V ‫ק חו‬
which would seem to address the repentant with the command
‘take with you words’ in their returning to the Lord.
It is thus no surprise that Gordis holds ‫ ב‬1‫ ט‬here to mean
*speech’. He writes:
The biblical and rabbinic root ‫‘ ל ב ב‬speak’ (Cant. 7. 10), from which
‫‘ ל ב ה‬report, evil report’ (Gen. 37. 2, Num. 14. 37) is derived (cf.
Accadian dababu ‘speak, charge’), apparently has a cognate ‫ ב‬1‫ט‬, ‫ ט ב ב‬.
Thus ‫ ל ב ה‬is rendered as ‫ ט י ב א‬by Onqelos and as tbhzvn by Peshitta in
Gen. 37. 2, and by ‫ ט א ב א‬by the Targum on Prov. 10. 18.
For these reasons, along with other aspects of the context into
which we shall not enter, Gordis concludes that the meaning is ‘ac-
cept our speech’. This interpretation appears to overcome certain
of the difficulties of the context; and it furnishes a close parallelism
between ‫ ב‬1‫ ט‬and ‫ ד ב ר י ם‬, both of which mean *speech’ or ‘words’.
Not only this; for Gordis goes on to cite other places, e.g. Neh.
6. 19:
: ‫גם טו ב תיו דייו א מ רי ם ל פ ני ו ך ^ו־י ז^יר מו צי אי ם ל ו‬
He renders:
‘His utterances they were wont to repeat to me, and my words they
would bring to him.’
Here ‫ טו ב תי ל‬is represented in the LXX (II Esdras 16. 19) by
το ύ ς λό γο υ ς α ύ τ ο ΰ : κ α ι το υ ς λ ό γο υς α ύ το ΰ η σαν λ εγ ο ν τε ς ττρός μ ε
κ α ι λό γο υ ς μ ο υ ησ α ν εκ φ ερο ντες α ύτω .
Previous scholars (Geiger, Low) had proposed an emendation
to ‫ ; ט ב ת י ו‬this is favoured by Rudolph in his commentary,1 and is
1 Esra und Nehemia (1949), p. 137.
 ‫* ט ו ב‬s p e e c h 1 17
incorporated in KB.1 It implies that the Aramaic word ‫ ט ב א‬or
‫‘ ט ב ה‬rumour, report’ existed in the Hebrew of Nehemiah, and
this in itself is of course possible. The proposal of Gordis,
however, seems to suggest that the meaning ‘speech’ is present {a)
without any emendation and (b) without reliance on an Aramaic
loan-word in Hebrew.
A third case quoted by Gordis is Ps. 39. 3 (EV 39. 2), where
‫ נ א ל מ תי דו מי ה ה ח שי תי פ!טוב‬might then mean2 ‘I was dumb
and silent; I refrained from speech’.
To the three points of general interest raised by our previous
example the following further characteristics may now be added:
(d) Multiple exemplification of the same solution, once it has
first been found. There exists not only one case where ‫ט ו ב‬
‘speech’ is identified, but several; and these several, after the first
identification is made, appear to support and confirm one another.
Gordis indeed identified yet other instances at Job 34. 4 and
Hos. 3. 5, but he was less sure about them.
(1e) The identification of a new homonym. In addition to the
familiar ‫ ב‬1‫‘ ט‬good’ there is another ‫ ב‬1‫‘ ט‬speech’ which is horn-
onymous with it. This is not all, for wide recognition has been given
to yet another ‫ ב‬1‫‘ ט‬perfume’, related to the Arabic fib with that
sense and identified at Isa. 39. 2, Jer. 6. 20, Cant. 7. 10 in senses
like ‘the perfumed oil’ ( ‫) ה ש מן ה ט ו ב‬, ‘the perfume stalks’ (‫ק נ ה‬
‫) ה ט ו ב‬, and ‘the perfumed wine’ (‫) יי ] ה טו ב‬. Another ‫ ט ו ב‬is a
place-name. For yet other homonyms which have been suggested,
see Index, nos. 147 8 ‫ ־‬.

(3) *be impudent, shameless’

The story of Korah’s rebellion begins in Num. 16. 1 with the
words ‫ ו י ק ה ק ר ח‬.
If this was the familiar verb ‫‘ ל ק ח‬take’ one would expect an
object. AV, following Ibn Ezra, supplies one, namely ‘men’, and
RSV follows this also. The 1962 Jewish American version trans-
lates ‘betook himself’, but in a note says the word is literally ‘took’
and adds ‘Heb. obscure’. Gray in the ICC (1912, p. 189) did not
try to make anything of it, being sure that part of the text had
1 KB, p. 346a.
2 There is a complication in Gordis ,s own treatment, for he takes ‫ דו מי ה‬also
to mean *speech*, though it is usually taken to mean ‘silence’; I leave this aside
at present for simplicity’s sake.
18 SOM E E X A M PL E S IN G R E A T E R D E T A IL
become entirely lost. B H 3 offers us as a probable emendation the
too facile ‫‘ וי ק ם‬and he rose u p \
The early versions present a varied picture. LXX says ελάλησ€
*he spoke’; the Targum and Syriac say 4he split off, broke away’ or
the like (Tg. ‫ ;) א ת פ ל י ג‬the Vulgate, which begins ecce autem Core
followed by a list of Korah’s assistants, has no verb at all (could
ecce be from a transliteration?).
Eitan argued that this is a Hebrew word corresponding to
Arabic waqiha ‘be impudent, shameless’.
This might be supported by a piece of evidence which Eitan
himself did not use, but which B H 3 cites: the Hexaplaric note
that d Ε β ρ α ίο ς has ύπερηφανβύθη ‘was insolent’.1
Two things should be noted about this interpretation which
have not been exactly exemplified in the previous instances.
(а) If this explanation is correct, it is probable that the similarity
of the (presumably uncommon) verb ‫ י ק ח‬to certain comparable
parts of the familiar verb ‫‘ ל ק ח‬take’ was itself an actual cause
which contributed to the loss of understanding of the meaning,
and even of the existence, of the former. The familiar verb over-
shadowed the less familiar until the philological treatment restored
it to our sight. This in turn depends on the accident that the verb
‫ י ק ח‬is found only in the imperfect; a perfect foim, which in the
1st singular would be *‫ י ק ח תי‬, would be much less liable to con-
fusion with ‫ ל ק ח‬.
(б) In many cases of this type the new identification will probably
require a change of vocalization. Thus for a verb cognate with
Arabic waqiha we would expect an imperfect qal form like ‫יי ק ח‬. or
‫לקח‬., without the daghesh which we find in the M T of Num. 16. 1
and which gives a form identical with that from ‫‘ ל ק ח‬take’. Or,
conversely, the mistaking of the form for one from the verb ‫ל ק ח‬
‘take’ carried with it the wrong pointing of the form.
Neither of these arguments is actually made by Eitan at this
point; but they are very frequently implied in philological treat-
ments.
Eitan finds another case of this verb at Job 15.12, where the M T
reads:
: ‫מ ה ־י קן ס ף ל ן ף ו מ ה ־י ך ץ מון עיגי ף‬

1 Cf. again below, p. 271.

 ‫‘ י ק ח‬BE I M P U D E N T , S H A M E L E S S * 19
Eitan’s translation is:
*What does your heart dare, and why are your eyes lifted up ?*
Concerning this interpretation some brief remarks may be made.
(a) One reason why a scholar may favour such an explanation as
this is a doubt whether ‫* ל ק ח‬take’ can really be used, in the words
of BDB, *figuratively, of passion carrying one away’.1
(b) Though Eitan does not say so at this point, one might wonder
whether his interpretation is not supported by the LXX:
r t €τόλμησ€ν ή καρδία σου‫י‬
η τ ί έπ ή νβ γκα ν ο ΐ οφ θ α λμ οί σου ;

(c) Eitan uses a philological treatment for the first verb in the
verse, but for the second he resorts to textual treatment by conjee-
tural emendation, reading ]‫ י רו מו‬from the verb ‫* לו ם‬be high’.

(4) ‫ י ד ע‬not meaning ‘know’

The standard dictionaries GB, BDB, and KB recognize and
register only one verb ‫ י ד ע‬, the familiar and extremely frequent
one meaning *know\ There are, however, a number of places
where this does not appear to give satisfactory sense, and in some
of these scholars have proposed other meanings supported by the
existence of words in the cognate languages, especially Arabic.
This example is of special interest because of two things: (a) the
familiarity of the word ‫* י ד ע‬know’, and (b) the large number of
passages in which philological treatments have claimed to identify
a different word. Several different such identifications, indeed,
have been proposed.
(1) the most important of these is a sense *make quiet, make
submissive, subject to discipline or humiliation’, which has been
explored particularly by Winton Thomas. The clearest case per-
haps is at Judges 8. 16, where Gideon obtained certain instruments
and
: ‫ר ד ע ב ה ם א ת אנ שי ס כו ת‬
If the verb is related to Arabic wada*a, this might mean:
*With them (i.e. the instruments) he made submissive the men of
Succoth.’
1 BDB, p. 542b; the sense ‘carry away* appears also in translations like AV
RSV, and in the work of scholars like Dhorme (Job, p. 193).
20 S O M E E X A M P L E S IN G R E A T E R D E T A IL
The sense 'teach’, represented by AV and RSV, is drawn of
course from the usual ‫' י ל ע‬know’; but it is doubtful if this is used
without an object stating what is 'known’. The favourite treatment
has been by emendation to giving the sense ‘and he
threshed with them the men of Succoth’. The philological treat-
ment removes both the anomaly and the need for emendation.
If this line of thought is sound, one can extend it to quite a large
number of other passages. At some of these, while an understand-
ing based on the sense ‘know’ for ‫ י ל ע‬may be more plausible than
at Judges 8. 16, nevertheless the success of the other approach at
this text may encourage one to try again the sense which was
helpful there.
At Isa. 53. 3 the well-known ‫ ע ל ל י‬1‫ י ל‬may mean not ‘ac-
quainted with grief’ or ‘knowing sickness’ (these do not, in the
opinion of some scholars, explain why the passive participle is used;
and this is the only case where the passive participle of ‫ י ל ע‬is
used in such a context; Deut. 1. 13, 15 are not parallel). The mean-
ing may be rather 'humbled, afflicted by sickness’, which fits well
with the context as a whole. At 53. 11, similarly, it may be argued
that ‫ ב ל ע ת ו‬means 'by his chastisement’; it is by his chastisement,
rather than by his knowledge, that the servant brings justification
to others.
At Qoh. 10. 20 we have the word ‫ מ ל ע‬. AV and RSV take this
as ‘thought’, implying that the word belongs with the verb ‫י ל ע‬
‘know’.1 The text reads:
: ‫גם ב מ ד ע ך מ ל ך א ל ת ק ל ל ו ב ח ד ר י מ ש כ ב ך א ל ת ק ל ל ע שי ר‬
Here the parallelism with the clear words ‘in your bedchamber’
has suggested that ‫ מ ל ע‬means not ‘thought’ but ‘repose’ or even
‘bedroom’. This sense, or one like it, indeed, can be reached by
another route, which is tried by KB; it attaches it to ‫ י ל ע‬in its
sexual sense. KB itself is uncertain of this explanation, which in-
deed seems unlikely; but it is sure that ‘bedroom’ rather than
‘thought’ is in general the right kind of meaning.2 It prefers, how-
ever, the emendation ‫ ב מ צ ע ף‬, with the word ‫‘ מ צ ע‬couch’ (from
the root ‫) י צ ע‬, found also in Isa. 28. 20. The suggestion of the
1 So, for example, BDB, p. 396a.
2 KB, p. 497b. Semantically, it is one thing to say that *know’ can have a
sexual sense, and quite another to suggest that a word like *knowing-place* would
thereupon be coined for a bedroom!
 ‫ י ד ע‬N O T M E A N IN G ‘K N O W ’ 21
sense ‘repose’ related to Arabic wada'a makes it unnecessary either
to emend or to appeal to the sexual sense of ‫‘ י ד ע‬know’.
Another example: Prov. 14. 33 reads in the M T
: ‫ב ל ב נ בון תנו ח ח כ מ ה ו ב ק ר ב כ ס י ל י ם ת ו ל ע‬
The RSV, to avoid declaring that wisdom is known in the heart
of fools, took the fairly drastic step of supplying the word ‘not’,
which, as its note ingenuously says, is lacking in the Hebrew, though
present in the Greek and Syriac. It is possible that this difficulty
would be overcome if the sense were as in the other examples just
cited, giving a rendering like:
‘Wisdom rests in an understanding mind
but in the heart of fools it is laid low.’
If this interpretation is right, it means that the ‘not’ of the Greek
and Syriac versions is not evidence of a superior text; rather, it
suggests that these ancient authorities were going through exactly
the same condition of puzzlement, resolved by guessing, which the
RSV translators suffered many centuries later.
These are by no means the only cases where senses such as ‘make
submissive’, ‘humiliate’ and ‘chastise’ have been found for ‫ י ל ע‬.
Without setting out the evidence, I shall mention also Prov. 10. 9,
Jer. 31. 19, Job. 20. 20, and the variant ‫ ת ד ע‬at Sir. 7. 20.
Another impressive case is Judges 16. 9, where it is said of
Samson ‫ ו ל א נ ו ד ע כ ח ו‬. *So his strength was not known’, renders
the AV. But this is difficult, for his strength was known very well.
RSV gets over this by saying that ‘the secret of’ his strength was
not known. But one can also follow Winton Thomas and consider
‘and his strength was not brought to submission’, ‘laid low’,
which fits the context well.1
Thus perhaps about ten cases can be quoted where this treat-
ment of passages with the verb ‫ י ל ע‬has been suggested.
(2) At least one prominent example has been treated as meaning
‘take leave of, dismiss’. This is I Sam. 21. 3 (EV 21. 2):
: ‫ו א ת ־ ה נ ע ר י ם י ו ל ע ת י א ל ־ מ ק ו ם פ ל ני א ל מ ו ני‬
If the verb were from ‫‘ י ל ע‬know’, this would have to be
1 In the end, however, my own opinion would be that the sense ‘know* is more
probable, for the repeated asking of Delilah implies that knowing or understanding
the source or nature of Samson’s strength is the real issue at stake; cf. the
repeated question ‫( ב מ ה כ ח ך ג דו ל‬vv. 5, 6, 15, cf. 10, 13).
22 SOME EXAM PLES IN G R EA TER D E T A IL
explained as ‘I have caused to know, i.e. I have directed’; so
BDB, but BDB itself prefers to emend to a form from the verb
‫‘ י ע ל‬appoint’, and so also BH 3. If it is from ‫‘ י ל ע‬know’, this is
the only case found of this theme, the poel.
Eitan says that this exceptional form has in Arabic a parallel
which is quite usual; it is wadaa (III theme, corresponding to
Hebrew poel), and means ‘say farewell, take leave of, abandon,
leave’. Thus the sense of the sentence is ‘I sent (or, dismissed) the
young men to the place of so and so.’ The verb ‫ י ל ע‬in its usual
sense ‘know’ appears in the very same verse.
(3) Several cases have been interpreted as meaning ‘care for,
keep in mind’. Of these the most prominent is Exod. 2. 25:
: ‫וי ר א א ל ה י ם א ת ־ ב נ י י ש ר א ל ו י ד ע א ל ה י ם‬
It is peculiar to find ‫‘ י ל ע‬know’ used absolutely without object
in a context like this, and this has led to emendation. If the sense
could be ‘and God cared’, evidenced from Arabic wadi a in this
sense, the difficulty might be removed. Winton Thomas argued
that this is the meaning of y-d- in ESA theophoric names, giving
a sense such as ‘cared for by II’ and the like.
If this is sound, then there is a case for seeing the same meaning
in an example like Job 9. 21:
:‫ת ם ־ א נ י ל א ־ א ז ־ ע נ פ שי א מ א ס חיי‬
which might be rendered:
‘I am blameless; I care not for myself; I reject my own life.’
This example illustrates how the result achieved by a philo-
logical treatment may be so close to one achieved through the
traditional understanding that only a narrow partition separates
them; for ‘I do not know about myself’ (from traditional Hebrew
‫‘ י ל ע‬know’) is not really distant semantically from ‘I care not for
myself’.
(4) At Hos. 7. 9 it has just been said that Ephraim is a cake, and
then it is twice added ‫ ו ה ו א ל א י ל ע‬. The more traditional under-
standing of this would be ‘and he did not know’; but Hirschberg
argues that the sense is rather ‘and he did not wrap it up’, quoting
wada'a meaning to wrap up an article so as to preserve it. This
interpretation, being connected with a very special metaphor, not
surprisingly does not recur again, so far as I know.
 ‫ י ד ע‬N O T M E A N I N G ‘K N O W ’ 23
(5) It has also been argued that, while the normal Hebrew form
meaning ‘sweat’ is ‫ י ז ע‬, a dialect form ‫ י ל ע‬also exists. This ex-
planation was tried by Noeldeke in some of the passages mentioned
above, most prominently in Isa. 53. 11, where ‫ ב ל ע ת ו‬is taken to
give the sense ‘by his sweat he shall justify’, and also in Prov. 10. 9,
32, 14. 33. Dahood more recently has followed this line further.
Such, then, are some suggestions relating to the verb ‫ י ל ע‬. Still
others could be added. For instance, the word ‫ מ ל ע‬at Qoh. 10. 20,
which we above saw interpreted as ‘bedroom’, has also been inter-
preted by Dahood as ‘messenger’. More than one of these sugges-
tions may be right, but they cannot all be right.
Thus once again we observe how homonyms are multiplied in
many kinds of philological treatment. Even if we regard cases (4)
and (5) as improbable, as I should be inclined to do, we have sug-
gestions for three meanings which are, or seem to be, substantially
distinct from the familiar sense ‘know’.
It will doubtless occur to the reader at this point that these
meanings might, if all the facts were known, be seen to be related
in some way.1The implications of this will be fully discussed later.2
For the present we observe only that the discovery of new meanings
has, for our purpose, practically the same effect whether or not
later research classifies the words as cases of homonymy (‘different
words’ which are formally identical) or polysemy (different senses
of ‘the same word’). The distinction does not make great difference
to the problems now in hand.
The complexity of the situation raised by multiple philological
treatments is not lessened when we consider the next case, the
suggestion of the existence of a verb ‫‘ ל ע ל‬call’.

(5) ,‫‘ ל ע ל‬call’

The standard dictionaries do not recognize a verb ‫ ל ע ל‬, though
they do take it into account for the personal name ‫ א ל ל ע ל‬.
Philological treatments, however, have identified such a verb,
corresponding to the familiar Arabic da a ‘call’. The Hebrew
meaning is usually stated rather as ‘ask, desire’.
If a verb ‫ ל ע ל‬exists in Hebrew, it is not homonymous with
1 e.g. M. D. Goldman, A B R iii (1953) 46, who argues that Hebrew ‫י ל ע‬
‘know* originally had the sense ‘put* or ‘lay down’, and that the Israelite thus
knew what was ‘laid down’ before him—a dubious enough argument, indeed.
2 See below, pp. 142 ff.
 24 SOM E EX A M PL E S IN G R E A T E R D E T A IL
‫‘ י ד ע‬know’ in the entirety of its paradigm; but in part of that
paradigm, and especially so if we consider the vocalization to be
uncertain, there will be a homonymy, or at least a homography
within the unvocalized writing system. If the existence of a ‫ד ע ה‬
‘ask, desire’ has been concealed from us until now, this has been
because its forms were classified as forms of ‫ י ד ע‬and its sense
correspondingly confused with the sense ‘know’.
The most impressive instance is Prov. 24. 14:
: ‫כ ן ד ע ה ח כ מ ה לנ פ ש ך‬
By the philological treatment, this might mean:
‘So ask for (seek) wisdom for thy soul’
—a sense which fits well with the following clause ‘if you find it’.
The sense ‘know’ is not very easy, and one may doubt whether
RSV is successful with its ‘Know that wisdom is such to your soul’.
The difficulty of understanding the word as ‫‘ י ד ע‬know’ is indicated
by the crop of suggested emendations. The vocalization of the M T
is anomalous for the imperative of ‫ ; י ד ע‬hence the suggestion to
vocalize as ‫ זי ע ה‬and to suppose that some words have fallen out.1
Other passages may be mentioned briefly. Hos. 6. 3
‫ ונ ך ע ה נ ך ך פ ה ל ד ע ת א ת ־ י ה ו ה‬:
might mean
‘let us desire, and pursue, the knowledge of God’;
and Prov. 29. 7
‫י ד ע צ די ק דין ד לי ם ר שע ל א ־י ב ץ ד ע ת‬
might mean:
‘the righteous knows the right of the poor
the wicked does not understand (his) suit (claim)’.
These examples would then include in the same verse both the
word ‫ ד ע ה‬and the familiar ‫‘ י ד ע‬know’.
At Prov. 10. 32:
‫ש פ תי צ ד י ק י ך עון ר צון‬
one might understand as:
‘the lips of the righteous desire (call for) what is acceptable’,
perhaps vocalizing the verb as ] ‫י ך ע ו‬..
1 The anomalous form is discussed by GK § 48 1; see Gemser, Sprüche, p. 89.
 ‫* ד ע ה‬c a ll ’ 25
In addition there have been suggestions of a sense ‘pull down,
destroy’ for !157*7, cognate with Arabic da'a III ‘destroy’ (of a wall).
(Index, no. 98.)
In examples of this kind it is suggested that a word or words,
knowable to us from the cognate languages, existed in biblical
Hebrew, but became obscured, so that such traces as were left
were interpreted as forms from the paradigms of other words. The
next example illustrates a very different situation. The form is quite
distinctive from any other known Hebrew word. Its very rarity has
caused it to be taken as a corruption of some kind, until compara­
tive philology succeeded in showing that there was reason to
believe in the existence in Hebrew of the form found, and with
a meaning which fits the text.

(6) ‫‘ ל ה ק ה‬body of elders’

At I Sam. 19. 20 Saul sent messengers to apprehend David, and
they found, as M T has it,
‫א ת ־ ל ס ק ת ה ? בי אי ם נ ^ אי ם‬
with Samuel at their head. What is this word ‫ ? ל ה ק ה‬b d b pro-
nounced it dubious. It has generally been held that the reference is
to a ‘band’ or ‘company’, which fits the fact that many prophets of
this time lived and worked in groups; so for example AV with its
‘company’, and RSV likewise. The question would then be: how
did this sense come to be represented by this word ?
There are two obvious ways. Firstly, it could be taken as an
historical linguistic metathesis of the familiar stem ‫ ק ה ל‬, in which
case the text may be right but a very peculiar dialect feature is
postulated, with vocalization also peculiarly altered. This approach
is, rather doubtfully, favoured by BDB (p. 530a). Secondly, it
could be a textual error. The original text was ‫ ק ה ל ת‬, and by some
strange accident, which had nothing to do with dialects or phon-
ology, this familiar word was changed by the scribes in the course
of written transmission into the totally unfamiliar (and indeed
nonsensical) ‫ ל ה ק ת‬. This process is implied by those who emend
to ‫ ק ה ל ת‬, as B H 3 advises.1
But a stem l-h-q is familiar to anyone with some knowledge of
Ethiopic; it is known especially in the word bhiq ‘old man, elder’
1 Cf. below, pp. 231 f., 26711., 270 f.
26 S O M E E X A M P L E S IN G R E A T E R D E T A IL
and a solution along these lines has been propounded not only by
Driver (1928), Winton Thomas (1941), and Ullendorff (1956), but
even some centuries earlier by Ludolf in his dictionary of Ethiopic,
cited recently by Ullendorff. If this interpretation is right, the
word is probably a collective of feminine form and means some-
thing like ‘group of elders* among the prophets (senatus prophe-
tarum, said Ludolf); and the word, though a hapax legomenon in
Hebrew, seems to be deserving of acceptance.

(7 )‫‘ ב צ ק ל ן‬fresh vegetables’

II Kings 4. 42 relates that a certain man brought Elisha a variety
of presents, including ‫ כ ר מ ל ב צ ק ל נ ו‬. The word ‫ כ ר מ ל‬occurs
elsewhere, and most scholars take it to refer to ripe corn or produce.
The following ‫ ב‬has often, not unnaturally, been taken to be the
preposition ‘in’, and this has left a strange word ‫ )?( צ קן ל ן‬to be
interpreted as ‘garment’ or ‘wallet’ (BDB), ‘sack’ (RSV), or ‘husk’
(AY). The reading ‫ ב ק ל ע ת ו‬, a fairly drastic emendation based on
the Arabic word gala a, said to mean ‘wallet’ or ‘bread-bag’, is
preferred by BDB (p. 862b) and KB (p. 841b); this suggestion is
traced to Lagarde.
A Ugaritic text, however, shows Danel praying for the chance to
see bsql growing in his dry land, and the word is repeated several
times.1 Driver translates ‘green corn’; he attributes the identifica-
tion to Cassuto. If this text may be taken as guidance, and the sug-
gestion seems a very probable one, we may assume that the letter
‫ ב‬of the biblical text is not the preposition ‘in’, but rather part of
a Hebrew word, pointed perhaps ‫ ב צ קן לון‬in the absolute state,
meaning ‘fresh vegetables’ or ‘green plants’ or, perhaps most likely,
in view of the termination /-on/ which is lacking in the Ugaritic
word, ‘garden’ or ‘plot’ where such plants are grown.
Thus the discovery of a text in a hitherto unknown language
serves in a remarkable way to sustain the reading of a Hebrew text
which had long been taken to be obscure or corrupt.

(8) ‫‘ נ^זר‬herald’
The beginning of Hos. 8. 1 reads:
‫ ר ע ל ־ ב י ת י הו ה‬# ‫ ן‬3 ‫א ל ־ ח פ ף ש פ ר‬
1 The text is Aqhat 1 ii 13 ff.; Driver, C M L, pp. 60,164; Gordon, UHt p. 180,
lines 62 if.
 ‫* ג ש ר‬h erald * 27
This extremely difficult passage has begotten numerous emenda-
tions and some artificial explanations. Taken literally as it stands,
the text would seem to mean:
*To your palate the trumpet, like the eagle upon the house of the
Lord . . .’
We may think that *palate’ here must mean gums or lips, so
BDB, p. 335a; we may think that the bird is not an eagle but a
vulture, though it is not clear what difference this will make to the
present problem. AV supplied the words ‘he shall come’ before
*like an eagle’, which makes good sense except that the words ‘he
shall come’ are just not there. RSV says:
*Set the trumpet to your lips,
for a vulture is over the house of the Lord’;
and this is a fairly simple emendation, i.e. to read ‫* כי‬for’ instead
of the preposition ‫* כ‬like’. B H 3 doubtfully considers an emenda-
tion which would eliminate the difficult bird; it reads ‫ נ צ ר‬, which
would mean *watchman over the house of the Lord’ and would be
addressed to the trumpet-blower.
A possible philological treatment has been proposed by Tur-
Sinai. On the basis of an Arabic naSSar meaning ‘herald’, who is
thus the one who blows the trumpet, he reads a Hebrew 3^ ‫ ר‬or
‫( נשל‬the former would be the more normal correspondence, if the
word was of direct descent from proto-Semitic in both languages,
but Tur-Sinai seems to prefer the latter). The passage would then
mean:
‘Set the trumpet to thy mouth, as a herald . . .’
This gives good sense, with no emendation except for the punc-
tuation.
Tur-Sinai tries the same solution on another passage, Job 39. 25.
In the description of a battle-scene, the war-horse hears something
which in M T is:
‫ע ה‬T‫ת רו‬: ‫ש·רי ם ו‬T ‫־ר ־ע ם‬
perhaps *the thunder of officers and the war-cry’.
Finding this phrase puzzling, he reads it as
‫ר ע נ שרי ם ו ה ר ו ע ה‬
‘the noise and shouting of the heralds’.
28 S O M E E X A M P L E S IN G R E A T E R D E T A IL
In this case, however, an emendation of the consonantal text is
required; moreover, it is doubtful whether the original difficulty
was very great, or whether the new solution is very good. This may
be another instance of a solution which, proposed with some
reason for a really desperate text, goes on to generate similar solu­
tions for passages for which a better explanation was already in
existence.

(9) ‫‘ א ד ם‬pleasant, delightful’

‫‘ ק ו ר‬dig, bore’
The reader will by now have realized that, if philological treat-
ments are numerous, there may be a good number of words in the
Old Testament of which he previously did not know. The conse-
quent sense of surprise or dismay may be savoured by studying a
more complex example, involving several words in a brief text.
At Prov. 12. 27 M T reads:
‫ןהון־אידם;זקר ייריץ‬
AV renders: ‘but the substance of a diligent man is precious’.
This might indeed be the right sense; but it fails to indicate one
essential point, namely that this sense cannot be obtained from the
text with the word order as it now is. The usual treatment is to
transpose two of the words, which can give u s:
‫ו הון י ק ר א ד ם ח רו ץ‬
‘A valuable treasure is a man who is diligent’ (Gemser, p. 60).
Or, with a different transposition:
‫ו הון א ד ם ח רו ץ י ק ר‬
‘The wealth of the diligent man is much’ (Driver).
Or, with transposition plus the preposition : ‫ל‬
‫ו הון י ק ר ל א ד ם ח רו ץ‬
‘The diligent man will get precious wealth’ (BH3, and so RSV also).
A treatment, proposed by Eitan, is both textual and philological.
He reads ‫ י ק ר‬as ‫ י קו ר‬. This change of vocalization makes the form
into part of the verb ‫‘ ק ו ר‬dig, bore’, found at Isa. 37. 25 and more
familiar from the noun ‫‘ מ קו ר‬spring, source’. He then takes the
 ‫* א ד ם‬PLEASANT, D E L I G H T F U L * A N D T i p *DIG, BORE* 29
word ‫ א ל ם‬to be not the familiar word for ‘man’, but an adjective
meaning ‘delightful’, following a sense known from Ethiopic. Thus
the phrase ‫א ד ם‬ is equivalent to the ‫ א ו צ ר נ ח מ ד‬of Prov. 21.
20, and the sense of our verse is:
‘The diligent (man) digs out a delightful treasure.’
The effect of this on the reader is considerable. He knows in-
deed that the sentence as it stands has its difficulties. But in his
uncertainty there seems to be some firm ground in the two very
familiar Hebrew words ‫‘ א ד ם‬man’ and ‫‘ ילןל‬precious’. Both of
these certainties are now removed. Of four words in the sentence,
two mean something quite other than what would at first occur to
the Hebrew reader; and in one of these two the meaning now sug-
gested is not evidenced from within Hebrew at all.
It is indeed possible to argue that a connexion between the senses
‘man’, ‘red’, and ‘pleasurable’ did exist in the historical development
of the Semitic languages;1 or one can point to places where that
which is ‫‘ א ד ם‬red’ is in fact delightful or is so regarded. It is
another thing to say that anything of this is the actual meaning of
‫ א ד ם‬in Hebrew.

(10) ‫‘ ! מ ר‬protect’ and ‫‘ ע ז י‬warrior’

Exod. 15. 2, M T ‫ ז מ ך ת י ה‬1 ‫ ע זי‬, is familiar in English as ‘the
Lord is my strength and my song’. Each of the two nouns, however,
has been interpreted otherwise. The sense of ‫ ז מ ר‬, it has been
argued, is not ‘sing’ but ‘protect’ (cognate with Arabic datnara, and
much used in theophoric names in ESA). Arabic words cognate
with ‘make music’ have the first consonant /z/, not /d/. Meanwhile
‫ ע זי‬is interpreted as related to Arabicgdzi ‘warrior’, gaza ‘go forth
to war’. The total sense is then:
‘Warrior and protector is Yah.’
Thus in a short phrase of three words (one of which occurs only
thrice in the Bible elsewhere), two are given novel interpretations.
On the other hand, the words are not such familiar ones as ‫ א ד ם‬and
‫ י ק ר‬, and the archaic style of the poem may lead the reader to
expect unusual words.
For example, Ullendorff in V T vi (1956) 191 f.
 30 S O M E E X A M P L E S IN G R E A T E R D E T A IL
In favour of this interpretation the LXX may be quoted: βοηθάς
καί σκεπαστής *a helper and a shelterer’. Neither the Targum
(‫ ) ת ו ק פ י ו תו ש ב ח תי ד ח י ל א ירי‬nor the Vulgate (fortitudo mea
et laus mea Dominus) support the LXX in this at Exod. 15. 2, while
it itself renders the same Hebrew at Isa. 12. 2 as ‫ )ל‬δόξα μου καί ή
αιν^σίς μου and at Ps. 118 (117). 14 as Ισχνς μου καί ΰμνησίς μου,
following the sense ‘make music, praise’.
Yet other philological treatments have been proposed. KB, also
alleging support in ESA, holds that ‫ ז מ ר ה‬means ‘strength’, cf.
the phrase ‫ ? מ ר ת ה א ר ץ‬of Gen. 43. 11, which might mean ‘the
strength, i.e. the best products, of the earth’. If ‫ ע ז י‬were taken as
from ‫‘ עז‬strength’, this would give a good parallelism. Rabin, on
the other hand, associates ‫ ע ז י‬with Arabic cazc? ‘patience, con-
solation’, and holds that a Hebrew cognate is found in the personal
names ‫ י ע ז י ה ו‬and ‫ י ע ז י א ל‬.

(11) A grammatical example

The examples cited up to now have been mainly lexical in
character. It is by the new identification of a lexical item that
cognate languages have been used to clear up a difficulty in mean-
ing. Problems of grammar, however, can be dealt with in the same
way, and sometimes the two are interlinked.
Semitic verbs may in some forms indicate gender. In normal
Hebrew, however, this is not so in the 3rd plural of the perfect
tense, where the form ‫ ק ט ל ו‬.is used for both masculine and feminine.
In classical Ethiopic, however, there is a distinction here, with the
vowel /u/ used for the masculine and /a/ for the feminine. There is
reason to believe that this is the older Semitic state. Now consider
these three passages:
1.1 Sam. 4 .1 5 : ‫ו עיניו ק מ ה‬
‘his eyes had grown dim’
(B H 3 emends to ‫ ק מ ו‬, with the Oriental Qere).
2. Neh. 1 3 .1 0 : ‫מניו ת ה לוי ם ל א נ תנ ה‬
‘the gifts of the Levites had not been given’
(B H 3 emends analogously to ‫)נתנו‬.

3. Gen. 4 9 .2 2 : ‫בנו ת צ ע ד ה ע לי׳־ עזו ר‬

perhaps ‘daughters (i.e. branches) run up over the wall’
 A G R A M M A T IC A L E X A M P L E 3i
(BH3 emends; Skinner, Genesis, p. 530, says the discord of number is
harsh, in spite of GK § 145k, which tried to argue for a feminine
singular verb with plural subject; Skinner himself thought the pas­
sage defied explanation altogether.)
Clearly, if it can be maintained that this is an isolated survival
of the distinctive feminine termination in /a/ for the 3rd plural
(which appears, incidentally, in certain readings in the Aramaic
sections of the Old Testament), then the need for emendation is
removed. This possibility is discussed in GK § 145k and § 44m but
rejected; Brockelmann in his Grundriss stated that the distinctive
feminine in this part of the verb had completely disappeared in
Hebrew.1 But in his recent work on syntax he takes the contrary
view, citing two of our three passages explicitly, and saying that
though the ending was no longer recognized by the Massoretes it
was preserved by them because they understood it as a singular.2
It is evident that the choice between a textual and a philological
treatment may be present in grammatical matters of this kind just
as it may be present in matters of vocabulary.

(12) ‘Enclitic mem’

This is another grammatical example. The Ugaritic poems con­
tain a number of examples of /m/ added after a word; there is no
certainty what the vocalization might be. This phenomenon has
been called ‘enclitic’, which seems to mean little more than that it
is written at the end of a word and as part of it, not having the
word-divider between· There has been some discussion whether
the enclitic means anything, i.e. whether its presence or absence
makes any difference, and if so what kind of difference. Into this I
shall not enter, except to say that philological treatments which
have identified an enclitic mem in Hebrew appear for the most part
to have treated it as if it had no distinct meaning. It may be added
briefly that there are in a number of Semitic languages particles
which appear at the end of words and include /m/; of these per­
haps the most noticeable is the final ma of Accadian, which is fully
discussed in the grammars of that language.
Now, if such a phenomenon was present in ancient Hebrew, and
if it later ceased to be recognized or understood, it would not be
1 Grundriss, § 262g, p. 575.
2 Hebraische Syntax (1956), § 50a, p. 50.
32 SOM E EX A M PLES IN G R E A T ER D E T A IL
surprising if it had left traces which to later scholars would consti-
tute a difficulty, leading not improbably to emendation to remove
the unwanted mems. Philological treatment, by exposing the mems
for what they are, would remove the difficulty.
An example where this has been tried is Ps. 29. 6:
‫ד ך ק י ל ם גי מו ־ ע ג ל ל בנון‬
: ‫ ךין )י מו ב ן ־ ך א מ י ם‬$ ‫ף‬
This text, by the traditional grammar, should mean:
‘And he made them dance like the calf of Lebanon,
and (made) Sirion (dance) like the young wild ox/
The difficulty is that one would expect Lebanon and Sirion to be
parallel, and the two animals to be parallel, so that the text would
be rather like this:
‘And he made Lebanon to skip like a calf,
and Sirion (to skip) like the young wild ox/
This happy result can be produced in more than one way. One
possibility is emendation; one can follow the suggestion of B H 3and
emend the verb to ‫ ד ל ק ל‬, thus simply removing the mem which
here, with its vowels, forms the suffix /em/ and gives the verb an
object ‘them’; once this is removed, Lebanon becomes the object.
Alternatively, one can explain it as the enclitic mem. Since this
had slight or no effect on the meaning, it implies that the verb had
no object suffix and has the same syntactic function as if it were
‫ ו י ר ק ד‬without/m/. Thus the philological explanation has the same
effect as the emendation but involves no change of text.
The question also depends on the location of the caesura. The
division of the verse as it is cited above follows that printed in BH 3.
But the Massoretic division, marked with the athnach> fell before
the word ‘Lebanon’, so that the sense is:
‘And he made them skip like the calf,
Lebanon and Sirion like the young wild ox/
This is the division followed, for example, by AV. This division
makes sense of the mem, which is here the object suffix of the verb,
and this object ‘them’ balances with the other object ‘Lebanon and
‘Sirion’ in the other half-verse. On the other hand it makes the first
half-verse much shorter than the second and thus appears to spoil
the parallelism in that way. In any case it is of interest to note how
 E N C L IT IC M E M 33
the emergence of a difficulty in the eyes of scholars is related to a
loss of confidence in the Massoretic accentuation.
The example which has just been described was one of the first
where the existence of enclitic mem in the Bible was suggested, in
fact by Ginsberg.
Another good example is a suggestion of Reider1 at Nah. i. 10,
where the phrase: □‘,KIDD □KDDDI has been a source of diffi-
culty. If the first mem can be treated as enclitic, the text may be
understood as ‘and as the drunken are getting drunk’.
No grammatical philological treatment has been pressed harder
than enclitic mem. Hummel’s article in 1957, often taken as an
exemplary discussion, listed thirty-one instances already dis­
covered, and went on to suggest seventy-six others which seemed
probable.2 Though he admitted that some of these might be ex­
plained otherwise, Hummel was confident that this loss would be
balanced with the discovery of many others still unnoticed. The
large numbers cited appeared to put the phenomenon beyond
reasonable doubt.
It therefore seemed a remarkable scepticism that Driver, himself
an enthusiast for the philological approach if ever there was one,
should doubt the existence of enclitic mem in Hebrew.3 Moran
wrote :4
After H. D. Hummel’s completely convincing study on the subject,
a scepticism which prefers to suspect the text rather than accept a
linguistic feature attested in Amorite, Ugaritic and Amarna (Jerusalem!)
should be virtually impossible.
We shall later consider some issues of principle raised by this kind
of argument. Meanwhile one only will be stated: the issue is not
only the presence of enclitic mem, but the scale or frequency of its
occurrence, once this argument is granted. Very many Hebrew
words end with D; it is common as a plural ending and in pronoun
suffixes. If in every such case it is likely to be suspected to be an
enclitic of no meaning, a very large field of variability is laid open.
Hummel’s 107 instances are quite good if we first assume that
enclitic mem was a frequent phenomenon; as a proof of its existence,
if the latter is in doubt, the list is not very strong.
1 m iii (1952) 79 ·
2 JBL lxxvi (1957) 85-107.
3 CML, pp. 129 n., 130 n .; JSS x (1965) 116.
4 In Wright, The Bible and the Ancient Near East, p. 60.
34 S O M E E X A M P L E S IN G R E A T E R D E T A I L

(13) Some general statements

I have now given enough examples to indicate some charac­
teristics of the philological treatment.
It remains to consider some general principles which have been
stated in justification of the philological approach. Though the
publication of strings of short notes has often left rather vague the
general logic of the method, there are certain points of principle
which frequently find utterance.
The first of these is the hostility to emendation as the way out of
difficulties. Driver wrote in 1927 :l
The time has come to lay down the rule that no word> and especially
no verb, in the Hebrew Bible, if only it presents a truly Semitic form,
may be emended. Many, if not most, such words will find an explanation
some day in the cognate languages, while there will generally be no
reason to suppose that those which cannot be so explained have been
incorrectly handed down although their meaning remains hidden to us.
Or let us hear Professor Winton Thomas:2
It must be regarded as the first business of the Old Testament linguist
to explain by comparative philology the forms he finds in Hebrew, and
not, save in the last resort, to emend. Emendation is based upon the
false assumption that all that can be known of Hebrew is known—it
perpetuates the known as the norm by which language is gauged.
Comparative philology, however, adventures into the unknown, and
discovers new criteria by which language can be adjudged possible or
impossible. . .. This revolt against emendation of the Hebrew text has
restored the reputation of the MT.
That emendation was based on a false assumption of knowledge
has also been suggested by Albright, who refers to:
our ignorance of Hebrew poetic vocabulary, which has led in recent
decades to innumerable erroneous emendations.*
Sometimes the hostility to emendation has become almost
hysterical. Guillaume writes:
Inasmuch as the text of Job has been subject to *emendation’, i.e.
deliberate falsification of the evidence, to an appalling degree . . . I
determined to read it as though it were an Arabic work.4
1 JTS xxviii (1927) 287.
2 In Record and Revelation, p. 401.
3 Peake’s Commentary on the Bible, 2nd edn., 1962, p. 62.
4 In the Hooke Festschrift (1964), p. 108.
 SOM E G EN ER A L S T A T E M E N T S 35
It is hard to see how such a phrase as ‘deliberate falsification of
the evidence’ is justified. Emendations do not in any way falsify the
evidence. The emendation is not inserted in the text, but only
suggested in the appendix. Translations such as the RSV, when
their renderings depend on an implied emendation, usually make
this plain by a note in the bottom margin. The evidence remains
intact for anyone who knows better. Nor can one see what is the
basis for the term ‘deliberate’ ; this could be responsibly said only
if the authors of textual emendations actually knew perfectly well
what the existing text meant, which clearly is not the case. Guil­
laume’s statement shows no understanding or sympathy for the
difficulties and uncertainties which led to the use of emendation.
In any case our purpose at this point is only to note that hostility,
sometimes intemperate, to textual emendation is common in the
literature of philological treatments ; yet, as we have shown and will
show further, many of those who have used philological treatments
have also used emendation quite lavishly.
The second important point of principle is that this devotion to
the text, and hostility to any conjectural alteration of it, apply
only to the consonants of the text, and not to the vowel signs or
other aspects of the Massoretic punctuation. Sweeping and far-
reaching rewritings of the punctuation are a frequent, though not
quite a universal, feature of philological treatments. This can be
found stated as a general principle. Thus Driver writes:
The solutions of difficult words and phrases here put forward are
based on the assumption that alteration of the consonantal text must
wherever possible be avoided but that the vowel-points are only of
secondary importance and may be emended with considerable freedom.1
Or again:
On ne tient pas compte des voyelles, qui n’ont de valeur que celle
d’un commentaire, presque médiéval.2
Such judgements seem indeed to be implied by the wide prac­
tice of philological treatments. The question then is whether it is
ultimately consistent to look with such deep veneration on the
Jewish transmission of the consonants of the Bible while holding
such deep scepticism towards the Jewish transmission of the
vowels.
Also we may ask whether an approach to Hebrew which considers
1 VT i (1951) 250. 2 ETL xxvi (195°) 34 &n. 7.
36 S O M E E X A M P L E S IN G R E A T E R D E T A IL
the vocalization with such doubt does not thereby logically deprive
us of a great deal of our existing knowledge of Hebrew. Does not
the organization of Hebrew into a coherent body of knowledge
depend on the vocalization system, which forms the logical basis
for our analysis of the grammar ? May not the philological treat­
ment, starting out by claiming to extend our limited knowledge of
the vocabulary and usage of Hebrew, logically end up by placing
us in a thorough scepticism of even such knowledge as we have ?
Thirdly, it is implied or stated that, where a difficulty is found in
Hebrew, almost anything from anywhere in the Semitic languages
can conceivably be invoked as a guide to the restoration of the right
meaning. It is true that Semitic languages have not all been drawn
upon equally; for the main body of philological treatments has
depended on a central group of sources, represented especially by
Aramaic, classical Syriac, Accadian, Ugaritic, and classical Arabic.
Nevertheless it does seem to be implied that any linguistic
phenomenon in any one, or in any group, of these languages, if it
appears to fit the needs for the healing of a difficulty found in the
Old Testament, and if no obstacle or impediment is at once obvious,
may be likely to have existed also in Hebrew. Existence in a cognate
language is taken to constitute prima facie evidence for existence in
Hebrew. We have just quoted Moran’s argument1 that if a pheno­
menon exists in Amorite, Accadian, and Ugaritic it is unwarranted
scepticism to doubt its existence in Hebrew.
The implication is that there was a very close sharing of lin­
guistic phenomena, and in particular of lexical items, between the
Semitic languages. The picture is one of a group in which very
great community of features existed. Driver wrote:
These languages stand far more closely together than, for instance,
French, Spanish and Italian; but Ethiopic, like Rumanian, is in many
respects farthest removed from the common type.2
And again:
The further back the enquiry is pushed the closer the resemblance
between the various languages becomes. In fact, early inscriptions show
Phoenician and Hebrew and Aramaic and even Arabic in a stage of
development in which they stand in almost the same relation as
1 See above, p. 33.
2 In The People and the Book (1925), p. 75. But Driver goes directly on to
argue that nevertheless great caution must be used in explaining an unknown
word in one language from a cognate word in another.
 SOM E G EN ER A L S T A T E M E N T S 37
Babylonian and Assyrian to each other and must indeed not so very
far behind this stage have been a single language.1
We have now sufficiently illustrated the procedure of philological
treatments and initiated some discussion of the principles involved.
Before we study these more closely, something should be said about
the history of the understanding of Hebrew. This will be done in
the next chapter.
1 Analecta Orientalia xii (1935) 70. Cf. the discussion below, pp. 184-7.
 Ill

SOME HISTORICAL ASPECTS

T h is chapter will provide a brief survey of certain historical

matters which are of importance for our subject. We are not con­
cerned primarily with the history of the text or the history of the
Hebrew language, or, again, with the history of exegesis and theo­
logical interpretation, though all of these are connected with our
main theme. That main theme itself, however, is the history of
basic linguistic understanding. How well did people at various times
know what Hebrew sentences and words meant ? If they were un­
certain, where did they look for guidance ?

(i) The Disuse of Hebrew among the Jews

Philological treatments generally imply that the meaning of a
rare word came to be lost in the Jewish tradition, and for this
reason has to be recovered by research in the cognate languages.
We have therefore to consider the fact that Hebrew in the course of
time ceased to be the normal daily speech of Jewish communities,
and along with this the date and the manner of this disuse of
Hebrew.
It is commonly held that before the time of Christ Hebrew had
ceased to be a ‘popular' language, and had been replaced among the
Palestinian Jews by Aramaic. The language spoken by Jesus, which
has been the centre of Christian discussion of the matter, is believed
to have been Aramaic, though this view does not exclude the possi­
bility that in discussion with Jewish scholars Jesus may have spoken
Hebrew. It is commonly held also that the Hebrew of the Mishnah
was an ‘artificial’ language of the schools, comparable with ecclesi­
astical Latin, and that it is in any case a strongly Aramaized form
of Hebrew. Where such views are held, they may naturally encour­
age an emphasis on the early and cognate materials (such as
Ugaritic), rather than on the literature of post-biblical Judaism,
as the place to look for the true sense of biblical words.
 T H E D IS U S E O F H E B R E W A M O N G T H E JE W S 39
The following considerations appear to be relevant:
{a) The view that the general language of Palestine in the first
century a .d . was Aramaic, and accordingly that this was the lan­
guage of Jesus, has come to be challenged in recent years; Birke-
land has even tried to prove that Jesus spoke Hebrew.1 Perhaps
these newer theories will not in the end find acceptance, and the
Aramaic theory will maintain its ground; even so it would seem
that the older certainty about the dominance of Aramaic no longer
holds good, and that fresh thinking about the matter is needed.
This present book, however, is not concerned with the language of
Jesus, and I mention it only as a striking example of the issue.
(b) It seems rather too obvious and simple to ascribe the disuse
of Hebrew to the Babylonian Exile. It is true that some scriptural
passages in Aramaic appear from the time of the Return, and the
Jews of Elephantine used only Aramaic. It does not follow that this
can be generalized completely and made to mean that Hebrew was
universally in decline. If Aramaic had been generally accepted in
Nehemiah’s time, that statesman could not well have been shocked
by the existence of children (obviously a minority of children) who
could not speak Hebrew. Moreover, the wrath of Nehemiah may
have had some success in reviving Hebrew, or may have been part
of a wider reaction in favour of Hebrew than is indicated in the one
saying of Neh. 13.24 f. As for the passage Neh. 8. 2-8, where inter­
preters make plain the sense of the law, in spite of the ancient tra­
dition which makes this the origin of the Targum and in spite of
modern arguments to the same effect, the passage may be better
construed otherwise: it is a reference to explanation, rather than to
translation into a different language.
Late books of the Old Testament, including the great literary
complex of the P Document, continue to use Hebrew, and certain
linguistic changes (visible, for instance, in Chronicles or Esther)
show relations with the later Mishnaic language. Sirach wrote in
Hebrew, and in Daniel it is the parts more definitely connected
with the time of Antiochus Epiphanes which use Hebrew. Of the
Qumran documents, a larger body are in Hebrew than in Aramaic.

1 Birkeland, The Language of Jesus; for further references and a summary of

the issues see Emerton in J T S ns xii (1961) 189-202; extensive references in
Fitzmyer, The Genesis Apocryphon from Qumran, p. 20 n .; recently Rabinowitz
in Z N W liii (1962) 229-38, holding the New Testament word fycfxida to be
Hebrew.
40 S O M E H IS T O R I C A L A S P E C T S
Thus a widespread loss or disuse of Hebrew cannot be simply
ascribed to the exilic period. It would appear rather that the disuse
of Hebrew proceeded very gradually, or, indeed, that some initial
decline of Hebrew in the exilic period was later compensated for by
a certain revival, so that Hebrew actually increased in influence
before it later began to decrease again. As late as the war of a .d .
i 32-4, when Simon son of Kosba wrote letters to his officers, he wrote
them in Hebrew. It can indeed be argued that the preference for
Hebrew was a product of the nationalistic upsurge of the time; and
this might be true, just as the inscriptions on Jewish coins were in
Hebrew. But even if this argument is right, it does not diminish the
reality of the continued use of Hebrew; it rather furnishes one
reason for that reality, and a reason not unsupported by Nehemiah’s
similar motives centuries earlier.
(c) It has indeed been argued that the Hebrew literature of Qum-
ran (and equally of Sirach or of Daniel) uses an archaizing style.
This could be so without implying the disuse of the Hebrew lan­
guage. Hellenistic poets like Apollonius Rhodius were passionate
archaizers, but this does not cast doubt upon the vitality and
influence of Greek as a means of communication in their time. If
Sirach or Daniel or the Manual of Discipline archaized in style, this
means only that they chose to express stylistically their continuity
with the earlier sacred literature. It does not mean, or even sug­
gest, that the archaizing took place because Hebrew was no longer
in use in any other form than that of a past literature. Moreover, as
is commonly the case, attempts at archaizing were accompanied by
clear evidences that the language was in fact in process of change.
(d) Mishnaic Hebrew is not an Aramaized Hebrew.1 There are
indeed words adopted from Aramaic, as was the case already in the
Bible; but there are also many words, especially nouns, adopted
from Greek. Especially striking is the fact that, in the stock of
verbs, a significant group which was held in common by biblical
Hebrew and by Aramaic is not used in Mishnaic Hebrew, while of
the 300 verbs occurring in Mishnaic which do not appear in the
Bible only a fairly small proportion are adoptions from Aramaic.2
Typical Aramaic devices like the emphatic state were not generally
adopted into Hebrew, and, of the lexical items which charac­
teristically differentiate Aramaic from biblical Hebrew, most are
1 See Segal, A Grammar of Mishnaic Hebrew, esp. pp. 1-20, 46-57, 98 f.
2 Ibid., pp. 46 if.
 T H E D IS U S E O F H E B R E W A M O N G T H E JE W S 4i
not found in the Hebrew of the Mishnah. Many features which
appear similar, such as the particle tP, are nevertheless not deriva­
tions from Aramaic, though they might be conceived as Hebrew
developments which have been encouraged or fostered by the
presence of analogy in Aramaic. One has the impression that,
though there was clearly much bilingualism, the separateness of the
two idioms was in general well maintained; elements from the
one were found in the other, but speakers knew whether it was
Hebrew or Aramaic that was in use.
(e) The late Hebrew of Mishnah and Midrash is not an ‘artificial’
or an ‘unnatural’ language; it was formed by processes of growth
and change such as are general in linguistic history. If it is true that
its growth took place in particular geographic areas or in particular
social or professional strata, this only means that these are the par­
ticular conditions for a linguistic development under normal
processes. One suspects that the ‘artificiality’ often felt by the
Western or Christian scholar lies less in the language system than
in the content and style. The style fits the content, and expresses
the conventions of a peculiarly stylized approach to the problems of
Judaism at that time. This provides no ground for the idea that the
language itself was artificial, as if it were a contrivance remote from
the actual functioning of languages yet illegitimately pretending to
the status of one.
(/) It is often said that Mishnaic Hebrew was not a ‘popular’
language or a ‘vernacular’, but only a specialized language of the
schools. But the spread of its vocabulary is not such as to suggest
use only in learned contexts. Its stock of words for such unlearned
matters as shopping or cooking is much greater than that of biblical
Hebrew. Similarly, in the Qumran period we find contracts for
land transactions sometimes in Hebrew and sometimes in Aramaic.
The terms ‘vernacular’ and ‘colloquial’ have lately come to be used
once again for Mishnaic Hebrew (‘vernacular’ here not necessarily
in the sense of the only vernacular or universal language of an area,
but in that of a language having some general use apart from purely
learned application).1

1 See Rabin, Qumran Studies, p. 67 n., and in general his stimulating article
in ScrH iv (1958) 144-61; also Greenfield in H U CA xxix (1958) 204, who refers
to Mishnaic Hebrew as ‘a vernacular raised to a literary language* and again ‘a
language with roots in the daily preoccupations of its speakers—agriculture,
the handicrafts, animal husbandry*.
42 S O M E H IS T O R I C A L A S P E C T S
(g) The model of the ‘vernacular’, often openly illustrated by the
example of ecclesiastical Latin among the medieval vernaculars,
fails to recognize the distinctive sociology of the Jews. Jews were
not like medieval Germans or Englishmen. On the one hand the
Jewish scholar was socially closer than the medieval cleric to the
occupations which in Christendom were ‘lay’; on the other hand
the relation of the ‘lay’ Jew to Hebrew texts as used in study, in
prayer, and in the synagogue service, was infinitely closer than the
relation of the medieval Christian to Latin texts. The respective
places of study and of ‘lay’ vocation were very different.
Linguistically the model we have to consider is rather that of
bilingualism, a situation quite different from the relation of Latin
to the monolingual vernacular speakers of medieval Europe.1
(h) In any case, the question of the ‘vernacular’ status of Hebrew
is not the decisive one for our purpose. It is relevant to our study to
know how widespread the living use of Hebrew was, but in itself it
is not decisive. Our question is not about numbers but about
quality; it is about the mode and the precision of the transmission
of Hebrew meanings. If Hebrew was not widely spoken, then the
transmission of meanings was not accompanied by a continuing
productive corpus of non-scholarly usage. This, if true, narrows
down in an interesting way the field in which we conceive the mode
of transmission to lie; but it in no way decides the question of the
reliability or the perceptiveness of that transmission.
Therefore in arguing, as I have done, against an excessively early
date for the general disuse of Hebrew, and against too categorical a
denial of ‘vernacular’ status for late Hebrew, I by no means imply
that those conceptions of the meaning of biblical words which were
current in late times were therefore ‘right’. Since Hebrew changed
as all languages do, the survival of Hebrew in popular usage would
not have universally favoured the correct transmission of the mean-
ing of biblical words. Since the words which we shall discuss are
ex hypothesi difficult or obscure ones in any case, it is possible that
late colloquial Hebrew would furnish no guide to understanding,
and may even have positively obscured it through the development
of new forms and new senses. Thus the continuity of Hebrew from
1 Such bilingualism is asserted by Kutscher, ‫ הלשל] ו ה ר ק ע‬, p. 10; while Rabin
in ScrH iv (1958) 152 speaks of a ‘trilingual’ situation in the Persian period, and
Goshen-Gottstein, op. cit., p. 135, speaks of a ‘quasi-trilingual’ situation in the
Qumran period. For modern studies of bilingualism see U. Weinreich, Languages
in Contact; von Weiss, Zweisprachigkeit, and literature there cited.
 T H E D IS U S E O F H E B R E W A M O N G T H E JEW S 43
the biblical down to the Mishnaic period does not in itself provide
a clear basis for the preservation of the meanings of rare words in
biblical texts. It does, however, make a significant difference to the
perspective in which we approach the whole problem.
(j) If it is true that the Jewish transmission of meanings was of a
scholarly rather than a ‘popular’ kind, this is not as unnatural or
artificial as it has often seemed. On the contrary, the fact of histori­
cal linguistic change made natural and necessary a scholarly struc­
ture for the transmission and interpretation of meanings belonging
to the older stages. The social preference for that which seems to
belong to ‘the people’ and ‘real life’ should not be allowed to
dominate our assessment of linguistic evidence. Even if it is true
that the main burden of the transmission of Hebrew meanings was
borne by a scholastic tradition, this does not prove anything about
the accuracy of that transmission. That scholars should retain
valuable knowledge of a language long out of popular use is not so
very surprising a thing. Western Europe learned classical Greek
from scholars whose life was lived in a milieu in which classical
Greek, and even Hellenistic Greek, had not been popular usage for
centuries. It is rather ironic that modern scholars, whose own ex­
perience of Hebrew is often formed exclusively through scholarly
reading, should regard the transmission of meaning through
scholars in ancient times as a ground for scepticism.
On the other hand, late Jewish scholarship was never of a purely
linguistic type; its linguistic memories were maintained within the
context of religious and legal interpretation, and this in turn may
have reacted upon the senses ascribed to words in the biblical texts.
We shall later examine some instances of this. Moreover, finally,
there was a change not only in the linguistic medium of Palestinian
daily life; there was also one in the medium of scholarly conversa­
tion, with the transition from the Tannaitic to the Amoraic period.
It is possible that this scholarly move to Aramaic may have been,
for the transmission of meanings, as important as the popular move
to Aramaic which occurred earlier.
These, then, are some ways in which the situation of late Hebrew
may be relevant for our study. These considerations do not in any
way decide the importance or the place of late Hebrew; at least,
however, they may clarify the perspective in which we regard it,
and open some questions which have too often been taken to be
closed.
44 S O M E H IS T O R IC A L A S P E C T S

(2) Linguistic Elements in Jewish Interpretation

Certain tendencies in Jewish biblical interpretation, though they
did not depend upon ignorance of Hebrew or loss of command of
Hebrew in the religious community, may perhaps have produced
a climate in which it was factually easier for linguistic information
to become confused, diffuse, or apparently immaterial.
Ancient Rabbinic interpretation differs from the modern philo­
logical approach. Firstly, there was a certain striving for the pro­
duction of multiple meanings. The modern philological approach
has on the whole looked for one ‘correct’ meaning, that which is
justifiable on the basis of linguistic evidence. The production of
multiple meanings seems to work in the opposite direction, and
reduces the emphasis on linguistic evidence as the source of precise
discrimination between interpretations. Discrimination is exercised
rather through an evaluation of the results of interpretation (meas­
ured against the prevalent religious structure), and through com­
parison with other passages in the authoritative Scripture. The
accuracy of the linguistic basis is not a supreme criterion.
Secondly, interpretation might not only fasten on to the literal
form of the text but might attach meaning to segments of it which
are, from our point of view, at a sub-meaningful level. Religious
interpretation could be attached not only to words as a whole but
to segments of words; it could be attached to letters of unusual
shape or position; it could build on senses which the words have
elsewhere, or on the senses of similar words or words associated in
other passages; and it could build on senses excluded by the present
context just as well as it could build on those favoured by the con­
text. Thus, while the acceptance of multiple meanings diminished
the centrality of a clear procedure from linguistic evidence, the
finding of meaning in sub-meaningful elements enabled interpre­
tation also to be closely literal. This ‘linguistic-form allegory’,1
both literal and allegorical at the same time, favoured the multipli­
city of meanings.
This style of interpretation, which in the later Jewish way we
may characterize as the derashy by no means in itself directly occa­
sioned a loss of command of Hebrew; nor was it logically dependent
on such a loss of command on the part of the community. The
1 For this phrase see my Old and New in Interpretation (London, 1966), pp.
114 f., 117.
 L IN G U IST IC ELEM ENTS IN JEWISH INTERPRETATION 45
multiplicity of meanings meant that interpreters could produce
extremely artificial senses while at the same time they knew per-
fectly well what we would now call the real meaning. The procedure
did, however, make it less important and decisive that the meanings
of words, as normal linguistic usage indicated them, should be
kept in mind.
Several particularly relevant types of rabbinic interpretation
should be specially mentioned.
Firstly, there is the finding of meaning by arbitrary analysis of
strange words. B. Ber. 54b displays such an analysis of the words
‫‘ א ב נ י א ל ג בי ש‬hailstones’ to mean ‘stones’ (‫ ) א ב נ י ם‬which ‘stood’
or remained suspended ‘on the back of’ or ‘for the sake of’ a man
(‫ ) ע ל ג ב אי ש‬and came down for the sake of a man; the first ‘man’
was Moses, the second Joshua. The unusual word ‫ת ל פ י ו ת‬
similarly figures in B. Ber. 30a in an interpretation where the word
is taken as the ‘hill’ (‫ ) ת ל‬to which all ‘mouths’ (‫ ת‬1‫ ) פ י‬turn—thus
furnishing scriptural evidence for the practice of turning towards
Jerusalem in prayer: for Cant. 4. 4 says that ‘Thy neck is like the
tower of David builded with ‫ ת‬1‫’ ת ל פ י‬.
Secondly, it is hard to separate this from the ‘etymological’ inter-
pretation of words. Why, for instance, Genesis Rabba asks with
reference to Gen. 1.10, did God call the dry land ‫ ? א ל ץ‬The answer
is: because she ‘conformed’ (‫ ) ר צ ת ה‬to his ‘will’ (]‫) ר צ ו‬. Etymolo-
gizing interpretation of this kind, though found particularly in
connexion with personal names, is to be found in all sorts of other
connexions also.
Pseudo-etymological connexions with similar words constitute
one way in which multiplicity of meaning is achieved. In B. Ber.
29b the Mishnah speaks of a ‘time of crisis’ ( ‫)> פ ך ש ת ע ב ו ר‬. What
does this phrase mean ? One speaker says that it applies to the time
when God is filled with ‘wrath’ (‫ ) ע ב ר ה‬against enemies, like a
‘pregnant woman’ ( ‫) א ש ה ע ו ב ר ת‬. Another opinion refers it to the
time when people ‘transgress’ (‫ ) עו ב רי ם‬the words of the Torah.
Thus the discussion gathers together the similar forms ,‫ע ב ו ר‬
‫ ע ב ר ה‬, ‫ ע ו ב ר ת‬, and ‫ ע ו ב ר י ם‬and provides from this connexion a
multiple network of possible meaning.
One interesting variation, when exegesis of a biblical text is being
attempted, is the al-tiqre (‫ ) א ל ~ ת ק ר י‬interpretation.1 At B. Ber.
1 See H. Torczyner in Encyclopaedia Judaica, ii. 74-87. Also Gordis, Biblical
 46 SO M E H IS T O R I C A L A S P E C T S
64a, for example, a biblical passage (Isa. 54. 13), is quoted, which
reads:
:‫וכל־ןיגיןד למודי יהוה‬
R. Eleazar says:
‘Do not read “thy children” but “thy builders” ’ (‫) בוני ך‬.
This enables a connexion to be made with a series of other texts
concerning building and houses. Again, at B. Ber. 57a, we read:
If one dreams he has intercourse with a betrothed maiden, he may
expect to obtain knowledge of Torah, since it says: Moses commanded
us a Torah, an inheritance (‫ ) מו ר ש ה‬of the congregation of Jacob.
Do not read ‘inheritance’ (HtZH^) but ‘one betrothed’ (‫ ך ש ה‬1‫) מ א‬.
It should not be supposed that these adjunctions ‘not to read’
such and such imply that the text is still in a fluid state. On the
contrary, they are if anything an evidence that the text was already
largely fixed, even though the vocalization signs were not yet
written. The text from which departure is made is the Massoretic
text; it is it that is ‘not’ to be read.
Secondly, these interpretations are not confined to difficult
vocabulary elements. Nor do they imply that the real meaning has
been lost. The interpreter may know perfectly well the general
usage and reference of a word at the same time as he is producing
an artificial analysis of it in quite another sense. These are devices
of the derash, and are by no means to be taken at quite their face
value. Similarly, legends can be generated by linguistic peculiarities
in the text. Num. 21. 14 has a phrase which in the AV margin is
‘Vaheb in Suphah’. The Hebrew is ‫ א ת ~ו ה ב‬. There is a
certain linguistic anomaly in the word ‫ ב‬HI, for very few Hebrew
words begin with /w/. A Tannaitic story, retold in B. Ber. 54a,
resolves the terms into a sense ‘Eth and Heb in the rear’ and
explains that Eth and Heb were two lepers who followed in the
‘rear* (HD1D) of the camp of Israel and later saw the discomfiture
of the Amorites.
These interpretative devices, then, do not by any means imply
that meanings have become unknown; they are, very frequently,

Text, pp. 78 f. For a recent consideration of the relation with the New Testament,
P. Borgen, Bread from Heaven (Leiden, 1965), esp. pp. 62-67. See also further
below, pp. 212 and 214.
 L IN G U IS T IC E L E M E N T S IN JE W IS H IN T E R P R E T A T IO N 47
additions to the plain meaning rather than replacements of it.
Nevertheless, in a general atmosphere in which such methods were
not only possible but popular and attractive, it is understandable
that certain useful and valid linguistic information, grounded in
actual usage, would escape notice and come to be lost.
The etymologizing treatment of words has its roots far back in
the history of Semitic literature. This popular etymology had no
historical reference, and all the etymologizing of antiquity was alike
in this.1 It was a literary device used in the development of narra-
tive or poetic effect, an interpretative device by which special
meanings were extracted, and (as with Aquila) a translation tech-
nique by which features in the original text which must, because it
is a sacred text, be pregnant with meaning, are reproduced, so far
as is possible, in the translation.
The Bible itself has a number of examples of popular etymology,
especially in personal names. The name Eve, Hebrew ΓΠΠ, was
etymologized in Gen. 3. 20: Adam called her by this name because
she was ‫ א ם כ ל ־ חי‬, ‘the mother of all living’. Here the LXX rose
remarkably well to the occasion, saying that her name was Zoe,
life: Ζωή, o n αϋτη μήτηρ πάντων των ζώντων.
Such etymological plays are beloved of the early story-tellers,
and form part of the dramatic colour, especially at the birth of a
child, but also when a name is changed or some other significant
turn in the course of events occurs. Etymologizing seemed very
natural because most Hebrew names were phrases which had
meaning as other phrases of the language had. Some were prayers,
like ‫‘( י ח ז ק א ל‬may God strengthen’) ; some were declarations, like
‫‘( ע ו ב ד י ה ו‬servant of the Lord’); some were statements of events,
like ‫‘( א ל נ ת ן‬God gave’). The names etymologized in the old
traditions, however, were for the most part not names of the normal
Hebrew type. The explanations of names like Cain or Noah or
Naphtali had an artificiality which did not attach to the under-
standing of actual Israelite names.
To us, indeed, it is clear that some of the etymologies do not fit.
The name ‘Noah’, in spite of Gen. 5. 29, does not come from 0 Π2
‘comfort’; and in spite of I Sam. 1. 20 the name ‘Samuel’ is not

1 For an article on another ancient literature comparable with the biblical

see J. Gonda, ‘The Etymologies in the Ancient Indian Brâhmanas’, Lingua v
( 1955‫ ־‬6) 61-85 .
48 S O M E H IS T O R I C A L A S P E C T S
connected with ‫‘ ש א ל‬ask, borrow’.1 This latter discrepancy has
sometimes been taken to show that the story is really about the
birth of Saul, for the explanation would really fit that name. But
this is being too logical; the etymologies did not depend on having
the right ‘root’, as we should call it.
The phenomenon of popular etymology cannot be strictly sepa-
rated from a whole series of other stylistic devices, such as asson-
ance or paronomasia. To quote two familiar examples:
Ps. 137.5::‫ימיני‬.‫אם־אקקחך ירושלם תקפח‬
Tf I forget you, o Jerusalem, let my right hand wither’ (RSV).
This is, in respect of the meaning of the second verb, usually
esteemed to be a better translation than the Tf I forget thee, o
Jerusalem, let my right hand forget her cunning’ of AV.
AV had to supply (from a supposed ellipse) an object for the
second verb; but its rendering at least conveyed that there was an
assonance, and thus caught something for which the writer was
striving; for the point is that the verb in each case is ‫ ש כ ח‬.
In Isa. 5. 7, on the other hand, though the consonants are not
the same, a striking effect is produced by the partial similarity of
the words:
:‫י_קו למקפט והנה מקפח לפךזקה!־לונה צעזקה‬:‫_ו‬
‘He looked for order, what he saw was murder;
he looked for right, what he heard was the cry of fright.’
(The writer’s translation.)
Such literary devices attracted attention to the way in which
words are made up and to the effect of their use in combination
with other words containing the same elements.
The use by prophets of a keyword, as the divinatory guidance
conveyed by the seeing of an object, probably also had an effect.
Two famous cases are at Amos 8. 2, where the prophet sees a
basket of summer fruit, ‫קיץ‬., and on this basis God tells him that
1 Sometimes a modern philological treatment can give a much more satisfying
interpretation of these names. Cain is usually understood after Arabic qain
*smith’, and this fits well with certain characteristics of the Kenites. For the
name Noah we have the very attractive suggestion of Driver in E T L xxvi (1950)
350, after Eth. nofyd ‘be long’, that Noah’s name meant something like ‘long (in
life)’. This would seem to fit with Oriental traditions of the relation between the
Deluge and the quest for immortality, and accords with the Sumerian name
ZI.UD.SUD.RA. These meanings, even if correct, had probably disappeared
from consciousness by the time the biblical stories were written.
 L IN G U IS T IC E L E M E N T S IN JE W ISH IN T E R P R E T A T IO N 49
the end, ‫ ק ץ‬, has come upon Israel; and at Jer. 1. 11, where the
prophet sees a rod of almond, ‫ ש ק ל‬, and indeed God is watching,
‫ ש ק ל‬, over his word to perform it.
In Dan. 5. 26-28 the words ‘Mene, Tekel, Peres’ are already
written; the situation is closer to that of interpreting an authori-
tative but mysterious written document. Its elements have a double
or triple meaning, depending on words which have the same root
or at least look similar. Thus, by a common interpretation, mene is
the coin ‘mina’ but also the verb ‘is counted’; tekel is the coin
‘shekel’, but also the verb ‘is weighed’; and peres or parsin is the
two small half-units, but also secondly the verb ‘divided’, and
thirdly the name of the Persians, one of the two nations to whom
Belshazzar’s kingdom will be divided and given.
In late times, then, etymologizing interpretation is on the in-
crease. A certain amount is found in the New Testament, related
to personal names, like Melchizedek in Heb. 7. 2; and here there is
also an etymologizationof the place-name Salem as ‘peace’. Hebrews,
however, does not venture far into the etymological wilderness;
both of these are simple cases, and the writer does not try us with
a similar treatment of (say) Levi, or Aaron, or Phinehas. A more
systematic etymologization of names, both of persons and of
places, can be seen in Philo and in Christian commentators; it was,
after all, a way of finding sense in lists of names, such as the bare
lists of places where the Israelites stopped on their way from Egypt
to Canaan. The symbolical use of place-names was on the increase,
whether with or without etymologization; the symbolical interpre-
tation of ‘Lebanon’, for instance, goes back as far as the history of
post-biblical tradition can be traced at all.1 In such cases, if an
etymological slant was obvious, it was utilized; if not, no matter,
the results could be reached without it. The explanations are usually
based, more or less obviously, on Hebrew; they rest also in part
upon the Stoic etymological tradition, but Palestinian Judaism
fostered the process independently.
We may then sum up this point by saying that the growth of
etymologizing interpretation was favoured by certain genuine
elements in the nature of Hebrew language and literature; but that
on the other hand the major cause of that growth was a loss of
appreciation of the ancient literary styles for what they were, and
1 Vermes, Scripture and Tradition} p. 36, and pp. 26-39 generally.
50 S O M E H IS T O R I C A L A S P E C T S
the passage to a more scholastic method of study. This newer
method of study was connected with the status of the literature as
a holy scripture and the corresponding expectations of the kinds
of meaning to be derived from such a scripture, in whole and in
its smallest parts. With the loss of sense for the original communi­
cative literary forms, meaningful stress comes to be put upon
elements which do indeed occur in the text but do not carry
substantial independent meaning within it. The great etymolo-
gizers were also the great allegorizers. There grows up a self-
conscious, undiscriminating emphasis on formal linguistic features,
while the overall continuity of resultant interpretation is provided
by legal and dogmatic systems.1 It is not hard to see how some loss
of awareness of exact linguistic meanings could have occurred in
such circumstances. Moreover, as we shall see shortly, the charac­
teristics which thus arose within Hebrew interpretation also
affected the understanding of Hebrew words and texts on the part
of translators who tried to put them into Greek and other languages.

(3) Early Intra-Linguistic Relations

The science of comparative philology as we know it is a modern
creation. But the idea that Hebrew is related to other languages is
very ancient.
There is a considerable period during which Hebrew and
Aramaic lay together in the consciousness of many Jews. There is
a lack of direct evidence in the form of actual statements about the
relation between languages.2 But some relations between the
vocabulary of one language and that of the other must have been
evident to those who knew something of both.
Four categories in lexical similarity or difference between the
two languages seem obvious:
Firstly, there exists a long series of words in which a similarity
between the two languages in form and meaning is fairly plain, for
example:
king Aram. Hebr. *j1?»
heaven vcm n 'm
hear v im v im
say *‫־‬m “iax
1 On this generally see my Old and New in Interpretation, pp. 107 ff., etc.
2 We can quote St. Jerome, however: vicina est Chaldaeorum lingua sermoni
Hebraico: PL xxix. 25 f.
 E A R L Y I N T R A - L I N G U I S T I C R E L A T IO N S 51

A perception of these similarities would probably be very natural.

Secondly, there are words in which one consonant differs but the
similarity of the word as a whole and of its meaning may have made
the correspondence fairly evident. The obvious instances are where
Aramaic *7, D, 57, and D correspond to Hebrew T, X, X, and 2?.
These are found in familiar words such as:
gold Aram.‫ד ה ב א‬ Hebr. ‫ז ה ב‬
run ‫רהט‬ ‫רו ץ‬
land ‫ארעא‬ ‫ארץ‬
sit, dweii ‫יתב‬ ‫ישב‬
This series would not always be as evident as it is to the modern
philologist. The relation between the Aramaic ‫ מ ט א‬and the
Hebrew ‫ מ צ א‬might not be apparent, since the meanings differ,
the Aramaic word meaning ‘come’ (Hebrew mostly ‫ א‬1‫ ) ב‬and the
Hebrew meaning ‘find’; and while in Aramaic ‘find’ is ‫ א ש כ ח‬,
Hebrew ‫ ש כ ח‬is normally ‘forget’.
Thus a person who knew both Hebrew and Aramaic would be
in a position to notice a long series of correspondences in form and
meaning and also to be warned against generalizing such cor-
respondences.
A third category is composed of words which exist in the one
language but are very much more frequent in the other, though
meaning more or less the same in both. The Hebrew ‫‘ א ת ה‬come’
is poetical and infrequent, but the Aramaic ‫ א ת א‬is frequent and
standard. In biblical Hebrew the verb ‫‘ י ה ב‬give’ is found only in
the imperative, but in Aramaic it is the usual word for ‘give’, which
in Hebrew of course is 1‫ת ן‬
A fourth category is the number of important and frequent words
which are completely different between the two languages.1 For
example:
go up Aram. ‫ס ל ק‬ H ebr.‫ע ל ה‬
go down ‫נחת‬ ‫ירד‬
go in ( ‫בו א ) נ כ נ ס‬ ‫על‬
1 There are indeed very occasional cases in which the *Aramaic* word is found
in Hebrew, e.g. ‫ א ס ק‬in Ps. 139. 8 and a number of cases of Jim and ‫ ש ב ח‬. I am
not trying to argue that Hebrew and Aramaic are completely exclusive in this
regard. The contrast is not between Hebrew and Aramaic, between which certain
overlaps occurred, but between normal words which were closely similar in the
two languages and normal words which were very different, even if sporadic
overlaps bridged the gap.
 S O M E H IS T O R IC A L A S P E C T S
serve ‫פלח‬ ‫עבד‬
fear ‫דחל‬ ‫ירא‬
witness ‫שהד‬ ‫עו ד‬
praise ‫שבח‬ ‫הלל‬

Though the modern philologist may sometimes see contacts

between the word in one language and some cognate in the other,
in the ancient world the words would usually seem to be totally
different.
Thus there was a basis upon which certain comparative insights
could develop; but they would also be limited by other aspects of
the material. The knowledge of Aramaic might confuse the tradition
of the meaning of Hebrew; it might also assist the preservation of
that tradition.
Though the similarities between Hebrew and Aramaic can
hardly have escaped the Targumists, the differences were suffi‫־‬
ciently frequent and obvious to save them from the temptation to
read Hebrew as if it was Aramaic, even where a Hebrew word had
the same consonants as those of an Aramaic word. The Targumists
were in general too experienced to fall into the obvious mistake of
writing the same root in Aramaic as they found present in Hebrew.
For example, ‫ ר ח ץ‬means ‘wash’ in Hebrew but ‘trust’ in Aramaic,
and no sense could be got out of the frequent references to washing
in the book of Leviticus if they were translated by ‫ ל ח ץ‬in the
Targum.1
In fact a survey of the Targums suggests that the cases where the
Hebrew word is translated by an Aramaic cognate word, and where
the meaning is also different, are not numerous. In other words, the
Targumists did not generally pursue that etymological fancy in
translating which might have led to a preference for a word like
the Hebrew where such could be found.
There was an additional reason why such close adherence to the
very forms of the original Hebrew was not pursued by the Tar-
gumists. Their translation is, more than most translations are, of
the character of a paraphrase. Quite substantial interpretative
additions are made. For those who are not accustomed to the
characteristics of Targums, an example may be given from Isa.
40. 1-3. Here is the Targum text as translated by Stenning; and

1 Cf., however, the Greek example below, p. 54.

 E A R L Y I N T R A - L I N G U I S T I C R E L A T IO N S 53
in it we italicize all those elements which are substantial additions
to the Hebrew:
O ye prophets, prophesy consolations concerning my people, saith
your God. Speak ye comfortably to Jerusalem, and prophesy concerning
her, that she is about to be filled with the people of her exiles, that her
transgressions have been forgiven her, for that she has received the cup
of consolations from before the Lord, as if she had been smitten twice
for all her sins. The voice of one that crieth, Prepare ye a way in the
wilderness before the people of the Lord, tread down paths in the desert
before the congregation of our God.
Where there is no hesitation to insert either renderings which
make precisions far beyond those required by the Hebrew or
phrases which are pure quantitative additions to the Hebrew, it is
unlikely that an attempt to represent the sheer outward form of the
Hebrew text will be a normal procedure. It would have been quite
possible for the Targum to exploit the Hebrew word ‫‘( צ ב א ה‬her
period of service’) and treat it as suggesting the Aramaic ‫צ ב א‬
‘goodwill’. This could have given a good sense: ‘goodwill for her is
fulfilled’. The Targum has rather followed the normal Hebrew
sense, namely ‘army, host’, and expanded this with its comment
‘the people of her exiles’. The tendency to expand and paraphrase
has linguistic effects opposite to those produced by a painful
concern for the formal and quantitative aspect of the Hebrew.
Sometimes, however, the same source may use both approaches.
The LXX does not, like the Targum, generally introduce large
amounts of matter quite unrepresented in the Hebrew; its tendency
is often rather to abbreviate, as is evident especially in Jeremiah
and Job. To take a simple example, χριστός ‘anointed one’ occurs
in the Greek Psalms and Isaiah in exactly the same places (ten and
one in number respectively) where the M T has ‫ ; מ ש י ח‬but the
Targum has several cases of ‫ מ שי ח א‬which are pure additions to
the Hebrew. Similarly, it writes ‫ מ שי ח א‬for the ‫‘ ש ב ט‬sceptre
(comet?)’ of Num. 24. 17.
There are, however, cases where the other procedure seems to
have been followed and the Targum uses the same root as the
Hebrew.1At Amos 2. 13 there is a word ‫ מ עי ק‬which the Targum
renders by ‫(‘ מי תי ע ק א‬I am) bringing distress’. In so doing it
associates the rare Hebrew verb ‫עולן‬, which in fact occurs only
1 For some illustrations of the problems this may cause in the understanding
of the Targum and the Syriac see below, pp. 263 if.
54 S O M E H IS T O R I C A L A S P E C T S
here and has remained a point of uncertainty to the present day,
with the Aramaic ‫ ע ק א‬. This is quite a common word, and is used
quite frequently to translate Hebrew words like ‫‘ צ ר ר‬to distress’.
The influence of Aramaic may be seen not only in the Targums,
but also at times in other translations, and also in general Jewish
comment and interpretation.
It has often been held that the LXX shows Aramaic influence
in certain of its renderings. Aramaic usage seems to have affected
the translators’ understanding of Hebrew. This would not be
surprising, since some Jews in Egypt had certainly spoken
Aramaic, and one may surmise that most contacts with inter-
national Jewry were made in Aramaic. Two examples will be
quoted:
At Isa. 53. 10 the Hebrew ‫ ד כ א ו‬means ‘to crush him’, but the
Greek says καθαρίσαι αυτόν ‘to purify him’. This seems to be well
explained if the LXX translator had in mind the common Aramaic
word ‫‘ ד כ א‬purify’; this differs from the cognate Hebrew verb by
the normal correspondences. We cannot, however, be quite sure
of this explanation. The translator could simply have misread the
‫ ד כ א ו‬of the text, or, looking at it cursorily, have been sufficiently
impressed by the similarity to the Hebrew verb ΓΟΤ, which would
give the sense ‘purify’. The explanation based on Aramaic, there-
fore, is not entirely necessary. It is, however, somewhat
strengthened by the fact that the Targum takes the same line here,
translating by
‫ל מ צ ר ף ו ל ד כ א ה י ת ש א ר א ד ע מ י ה ב ד י ל ל נ ק א ה מ חו בין‬
:‫נ פ ש הון‬
‘To refine and purify the remnant of his people, in order to cleanse
their soul from sin’.

Another striking case, perhaps the most striking in the Greek

Bible, occurs at Ps.60 (59). 10 and again at 108 (107). 10. M T reads:
‫מו א ב סי ר ר ח צ י‬
‘Moab is my wash-bowl’.
LXX has:
Μωαβ λββης της ελπίδος μου
‘Moab the bowl of my hope.’
 E A R L Y I N T R A - L I N G U I S T I C R E L A T IO N S 55
This seems to depend upon the Aramaic ‫ ר ח ץ‬, which has
already been mentioned above. However, the Targum itself does
not follow this interpretation, but reads at 60. 10:
‫ב ע טיי ת מו א ב אי‬
T trod down the Moabites’,
and at 108. io:
‫דו שי שי ת מו א ב אי‬
‘I trampled down the Moabites’,
both of which, of course, are very rough paraphrases. The LXX
renderings, though they seem to depend on a mental reference to
Aramaic, are not the effect of the actual Targum, or not at any rate
of the Targum which we now have.
Quite considerable lists of cases showing dependence by the
LXX on the Aramaic rather than the Hebrew meaning of occas-
ional words can be collected from the works of scholars.1Not all of
the suggestions which have been made can be accepted, however,
for sometimes other explanations are possible.
The Aramaic language continued to affect the understanding of
the Bible long after it had itself fallen into practical disuse.
Examples of renderings which depend upon Aramaic can be found
in the AV. At Judges 7. 3 M T reads ‫ ה ג ל ע ד‬T O T O ‫־‬:‫}־‬.
Now the verb ‫ צ פ ר‬is a hapax legomenon of very uncertain mean-
ing. Aramaic had a familiar word of this root, ‫ צ פ ר א‬, which means
‘morning’. The later Jewish commentators, such as Rashi and
Kimchi, took this into account in dealing with the passage; and
AV, when it translates ‘depart early from mount Gilead’, is follow-
ing this.2 Interestingly enough, the Targum itself did not take this
line; its translation is ‫ י ת ב ח ר‬, which has no apparent connexions
with Aramaic words of the root ‫ צ פ ר‬. Again, the translation of the
obscure ‫( מ ך ה ב ה‬Isa. 14. 4) as ‘the golden city’ in AV follows
Kimchi and depends on the connexion with the Aramaic ‫ל ה ב א‬
‘gold’.
1 For a considerable list, some of them very dubious, see Wutz, pp. 150 if. A
few are quoted by Swete, p. 319 n., from Nestle; see also I. L. Seeligmann, pp.
49 f. Some scholars have gone so far as to suggest that Aramaic rather than
Hebrew was the Semitic language truly known to the translators; see Vollers
in Z A W iii (1883) 224 f.; also Flashar in Z A W xxxii (1912) 251, etc.
2 See G. F. Moore, Judges, p. 203.
56 S O M E H IS T O R I C A L A S P E C T S
The Targum was productive in this way in the Middle Ages, in
areas where Aramaic had not been spoken anyway, because it was
studied by the commentators and its interpretations were often
taken up by them. Many of these interpretations do not depend on
the similarity between Hebrew and Aramaic words, but rather the
opposite: the Hebrew being difficult or controversial, the Targum
rendering may be free from these particular uncertainties and thus
be taken up into the commentary. This continuing general esteem
of the Targum, which had now become a sort of additional sacred
text supplementing the Hebrew text, is a basis from which there
could always spring interpretations depending upon particular
linguistic phenomena of Aramaic.
In sum, then, the existence of the Targum, and the wide use of
Aramaic in Jewry, formed a kind of practical introduction to prob-
lems of relating Hebrew to another cognate language, and thus
occasionally introduced a kind of primitive linguistic comparison.
The influence of Aramaic and other languages on the under-
standing of Hebrew can not only be traced through indirect
evidence from translations; Rabbinic sources sometimes give
direct statements purporting to rest on the knowledge of other
languages. An introductory phrase sometimes found is 11‫ל א ה‬
‫‘ י ד ע י ר ב נ ן מ אי‬our teachers did not know the meaning o f . . .
u n til.. .’, followed by the circumstances in which the sense came to
be known.1
A famous instance is 1‫ ט א ט א ת י ה‬, of which the meaning was
unknown until the maid of Rabbi was heard to say:
‫ש קו לי ט א טי ת א ו ט א טי בי ת א‬
‘take up the broom and sweep the house’.
Other examples cite languages heard when travelling in foreign
parts. Rabbi d. Bar Hana tells that he had not known the sense of
‫ י ה ב‬in the verse Ps. 55. 23 ‫יהבף‬. ‫ ה‬1‫ ה ש ל ך ע ל ~ י ה‬until one day
when he was carrying a heavy burden and an Arab said to him
‫ ש די א ג מ ל אי‬1 ‫ ל י ה ב ך‬1‫ש ק‬
‘lift up your yehab and throw it on my camel’.
Again, we hear that ‘Rabbi says that in coastal cities ( ‫) כ ר כ י הי ם‬
1 See B. Rosh ha-Shanah 26a-b; also B. Meg. 18a. For a discussion of the
relation to Arabic of these passages see S. Krauss in ZDM G lxx (1916) 321-53,
esp. pp. 338‫ ־‬49‫״‬
 E A R L Y I N T R A - L I N G U I S T I C R E L A T IO N S 57
selling was called ‫ ; י כ י ר ה‬and this piece of philological information
was supposed to be useful for the understanding of Gen. 50. 5:
‫ ב ק ב רי א שר כ רי תי לי‬.
Again, a tradition reports: ‘R. Simeon b. Laqish said: When I
was in the province ‫ ק ] נ ש ריי א‬, one called the bride ‫ נ י נ פ י‬and
the hen ‫ ש כוי‬.‫ ני נ פ י ׳‬is applied to Ps. 4 8 .3 : ‫י פ ה נו ף מ שו ש כ ל ־‬
‫ ה א ר ץ‬, and the word ‫ נו ף‬is thus associated with the distantly
similar Greek vviujyq. Similarly, the interpretation of ‫ ש כוי‬is
applied to Job 38. 36, a passage the obscurity of which often
excited the interest of the older scholars.
Appeals of this kind to foreign languages are sometimes made at
points where the Hebrew is intrinsically difficult or obscure; but
sometimes also they provide an additional or midrashic explanation
to a text where the normal sense must have been quite well known.
An instance of the former is found at Gen. 49. 5:
!‫ש ס עון ן לוי א חי ם ? ל י ח מ ס ?; כ ר תי ה ם‬
In Gen. R . the last word is declared to be the Greek /xa^atpa
‘sword’, which of course gives a seemingly fitting sense. The word
has long been considered difficult. BDB, p. 468b, registers it under
‫ כ ו ר‬, but seems doubtful; GB, p. 423a, is also at a loss, and KB,
p. 523a, states only that the word is ‘unexplained’. A very attractive
philological treatment is to take it as cognate with the Ethiopic mkr
‘advise’ (in the theme II. 1, ‫י‬amkard, ‘consilium dare, suadere’,
Dillmann, col. 199; cf. also III. 3); Ullendorff1mentions this con-
nexion, already seen by Ludolf. The sense would then be: ‘Weapons
of violence are their counsels.’ The appeal to a foreign word in the
Midrash is related to an acknowledged obscurity in the Hebrew
itself.
The position is quite different in the following example from
Pesikta de-R. Kahana, xl.: the words spoken in leading up to the
sacrifice of Isaac, ‫‘ שה ל ע ל ה‬a lamb for a burnt-offering’ (Gen.
22. 7 f.), are interpreted with the statement that this is a Greek
word, i.e. 06‫‘ ־‬thee’; thus the sense of ‫ ש ה ל ע ל ה‬is ‫א ת ה הו א‬
‫‘ ה ק ר ב ן‬thou art the sacrifice’.2 There was, of course, no difficulty
1 Index, no. 200; the discussion in Skinner’s Genesis, pp. 516 f., shows the
uncertainty of the older interpreters. None of the versions seems to have known
the sense, cf. my remarks below, p. 270.
2 See A. Briill, Fremdsprachliche Redensarten (Leipzig, 1869), p. 26; this work
gives many relevant examples.
58 S O M E H IS T O R I C A L A S P E C T S
in the word ΠΦ for the interpreter; but, by suggesting that it was
a Greek word here, he was able to give apparently literal evidence
for what was indeed, as every reader knew, the situation in the
incident described. The introduction of a suggestion of a foreign
word may thus be a means to a midrashic interpretation, and not
connected with any intrinsic difficulty in the text.
These scraps of ‘philological· information can of course by no
means be solemnly taken at their face value. They do not neces-
sarily represent the mode by which an interpretation was reached.
Even when a tradition records that T did not know the meaning of
word x until. . .’ it is by no means certain that this is absolutely true
even for the Rabbi concerned, and his interpretation might use the
information from foreign sources only for one exegesis among
others of a relevant text. Needless to say, there was no systematic
approach to philological study and no attempt to distinguish be-
tween languages which are cognate with Hebrew (such as Arabic)
and those which are not (such as Greek).
The examples are nevertheless significant, in that they may show
u s: (a) that some uncertainty about the sense of unfamiliar words
was admitted, (b) that the legitimacy of appeal to other languages
than Hebrew was already accepted, though in a very confused
form, in the Talmudic period; and this held open the possibility of
something which the medieval lexicographers used to much greater
effect.
In general Greek lacked the obvious similarities to Hebrew
which Aramaic had, and it did not provide, as Aramaic did, a
potential entry into a rudimentary comparative philology. Never-
theless, the practice of Greek translators sometimes provides
further illustrations of the tendencies we have been describing.
Occasionally they seem to have chosen Greek words which had
some similarity to the Hebrew words being rendered. The render-
ing ήχου for ‫ א חי ם‬at Isa. 13. 21 is a probable example;1 and at
Isa. 51.8, where the Hebrew has the two nouns ‫ ע ש‬and DO, both
meaning ‘moth’, σης is used for the rendering of the latter, though
it usually renders the former; this may well be attributed to the
verbal similarity. Stock examples from the later Greek translators
are αυλών for ‫ ן‬1‫ א ל‬at Deut. 11. 30 and 0e/)a7r€1a for ‫ ת ר פ י ם‬at
I Sam. 15. 23 (if this text is reliable).2
1 On this passage cf. below, pp. 243, 250.
2 See Swete, Introduction, p. 41, and Fields note 34 on I Sam. 15. 23.
 E A R L Y I N T R A - L I N G U I S T I C R E L A T IO N S 59
Again, Greek translators sometimes put into practice in Greek
the tendencies of arbitrary analysis and etymologization which we
have already traced in Hebrew and Aramaic sources. The use of
Ικανός for the divine name ‫ ש לי‬depends on a resolution into ‫ש‬
‘which’ and ‫‘ ל י‬sufficient’. For the enigmatic ‫ א ש ל ת‬of Deut.
33. 2, the Vulgate’s ignea lex and later the AV’s ‘a fiery law’ imply
a resolution into ‫‘ א ש‬fire’ and ‫‘ ל ת‬law, religion’. These methods
continued throughout the Middle Ages, as in the resolution of
□‫‘ ש מי‬heaven’ into phrases like ‫‘ ש ם מי ם‬water is there’ or + ‫א ש‬
‫‘ מי ם‬fire and water’, mentioned by Rashi on Gen. 1. 1.
Such devices are an attempt to display or exploit in translation
something of the formal characteristics of the original; it is easy for
us to see how such an emphasis could damage the consciousness of
linguistic meaning, though the intention was doubtless quite
different. The literalizing emphasis on the formal and quantitative
was pressed hardest of all by Aquila, and accompanied by extensive
etymologizing.1
Up to the sixth century or so it seems fair to say that the main
mental effort of Judaism took a form which favoured such methods.
The elaboration of the halachic law implied a concentration on
legal distinctions; while the Midrash, which followed the actual
contours of the biblical material more closely, also included wide
freedom for legendary expansion and, because of its fanciful and
often more humorous style, allowed the plain sense of the text to be
submerged under highly imaginative plays upon words and phrases.
These conditions could have been detrimental to the maintenance
of a good and sober awareness of meanings in biblical Hebrew.
An important change of direction is signalled by Qaraism. In
spite of the compromises and uncertainties of the Qaraitic move-
ment, it does seem to restore an emphasis on the commentary form
and an appeal to the scripture. By this time (Qaraism takes its rise
in the eighth century) there are several other changes of position to
mention. The work upon the text of the Bible, which we associate
with the name of the Massoretes, was under way. Codices were
copied by the families of Massoretes and provided with vocaliza-
tion and accent signs under a number of systems of increasing
refinement. Systems of vocalization were being devised also for the
Quran in Islam and for the texts of the Syriac-speaking Christians.
It is not necessary for our purpose to fix the exact dates and
1 For an introduction to Aquila,s methods see Swete, Introduction, pp. 31-42.
 60 S O M E H IS T O R I C A L A S P E C T S
priorities of these operations. In a later chapter we shall return to
the Massoretes and the nature of their work.
In the Islamic period Jews were able to enjoy an expansion of
their philosophical and scientific horizons. Grammatical studies
began and expanded. The first considerable name in the history of
Hebrew linguistic studies, Saadia Gaon (892-942), was also a con­
siderable commentator on books of the Bible, and a philosopher
of note.
Meanwhile Arabic quickly took over from Aramaic the place of
prominence as the language with which Hebrew might usefully
be compared. Many of the great Jewish thinkers normally used
Arabic. Maimonides wrote his Guide for the Perplexed in Arabic,
and it was Ibn Tibbon who translated it into Hebrew. The pious
Jewish scholar still read his Targum in Aramaic and his Talmud
which is mainly in Aramaic; but the linguistic medium for the most
active Jewish thinking had come to be Arabic.
In contrast with the earlier Judaistic development, in which the
tendency to derash had so been fostered, the period of the Masso-
retic development and of the rise of Jewish grammatical studies
can be characterized as the time of the peshat. It is reasonable to
consider that a major stimulus towards this emphasis on the plain
sense came from the confrontation with Qaraism within Judaism,
and with Christianity and Islam without.1 The religious values of
the derash were by no means abandoned or forgotten; but for the
questions which now became foremost in scholarship it seems
no longer to have been supposed that answers would be found in a
mere appeal to the tradition of derash. The new lexical and gram­
matical studies, rudimentary as they were, represented an attempt
to find guidance in linguistic evidence rather than in traditional
religious interpretations.

(4) Aspects of the History of Jewish Grammatical Studies

This section will provide some elementary introduction to a sub­
ject which has been widely neglected but is relevant to the histori­
cal understanding of our general theme. Scholars will perceive the
extent to which I depend on the learned expositions of Bacher.
Amid the complicated and varied work of the medieval Jewish

1 For a simple presentation of this view see E. I. J. Rosenthal, *Medieval

Jewish Exegesis: Its Character and Significance*, J S S ix (1964) 265-81.
 H IS T O R Y O F JE W IS H G R A M M A T IC A L S T U D I E S 6!
grammarians I have picked out elements which concern us here.
They help to illustrate, among other things, (a) the continuing
presence of comparative methods, (b) the methods of analysis used
on biblical words in medieval times, and (c) the nature of the
Massorah, which will concern us again later.
‘The Massorah is the real cradle of Hebrew grammar’, writes
Bacher.1 It provided a careful registration of the written form of
the Bible, and was a kind of school in which the mind was trained
to recognize with painstaking care the detailed points which might
discriminate between right and wrong reading. The Massoretes
thus come to the verge of grammatical study. Their notes occasion­
ally take cognizance of such distinctions as masculine and feminine,
or the difference between final and penultimate stress.
Yet Massorah is not grammar, and the use of grammatical cate­
gories is not general or predominant in it. Its purpose is to assure
the correct writing and reading of the text. Semantic problems are
not generally discussed. This is not to be despised as a fault. The
value of the Massoretic registration is its careful fixation of the
formal characteristics of the text at a time when interpretative
methods had long been too fluid to permit any simple perception
of the relation between form and meaning. Only late in the course
of the Massoretic activity, with Aaron ben Moshe ben Asher in the
early tenth century (a time therefore to which certain important
Hebrew manuscripts still extant go back), does a combination of
grammatical discussion with Massoretic activity appear to be found.
Even then it is not an independent linguistic description, but rather
a setting forth of the rules implied in Massoretic work, under some
influence from Arabic grammatical science.
It was either ben Asher or his greater contemporary Saadia who
took the important step of first noticing and formulating the dis­
tinction between what would now be called root and afformatives.2
The importance of this is to us so great (our Hebrew grammar
having been developed upon the basis of this distinction) that it is
surprising to realize how late in the history of Hebrew it was recog­
nized. Saadia also concerned himself with the rare words of the Bible,
mostly hapax legomena, and wrote a treatise on ninety of them. In
his translations into Arabic, and in the commentary related to them,
he made much use of Arabic material in the explaining of Hebrew
1 ZDM G xlix (1895) 8.
2 See the discussion by Bacher, ZD M G , p. 48.
62 S O M E H IS T O R I C A L A S P E C T S
words. He was deeply indebted to Arabic grammatical ideas, and
followed an Arabic example in conceiving his task as the prescrip-
tive one of teaching correctness to Jews who had not the slightest
idea of what was correct Hebrew.
About the same time as Saadia, we find Yehuda ibn Koreish,
shocked to hear that the Jews of Fez had discontinued the reading
of the Targum, arguing for a kind of interdependence of the know-
ledge of the three great languages, Hebrew, Aramaic, and Arabic.
Did not the Scripture itself make it clear that the Hebrew, Ara-
maean, and Arab peoples were genetically related through Terah,
Abraham, Laban, and Ishmael ? Aramaic words could be found in
the Bible itself; indeed, the Bible contained words not only of
Arabic, but of Romance and of Berber origin. There is, he argued,
even a Greek word in the Old Testament, for the ‫ ל ?ן ל ס‬of Ezek.
16. 31, which a millennium later was still troubling the editors of
2?£f3, is from the Greek word καλώς. Thus Yehuda ibn Koreish
represents a strong if rudimentary comparativist position. He wrote
a long list of word-comparisons to illustrate these principles.
Menahem ben Saruk (about 960), on the other hand, paid little
attention to Arabic and departed from the Arab theoretical basis
which earlier workers had adopted. He cited Aramaic and Mish-
naic Hebrew more frequently. The emphasis of his work was
lexical rather than grammatical; his dictionary is the first full die-
tionary of biblical language in Hebrew. Following a principle of
explaining words from the context in which they are found, he was
led to a heavy stress on the stylistic form of parallelism. O ne half
of a verse teaches about the other.’
A particular emphasis is placed upon the strict separation of the
various roots. Yet the carrying out of this was still under serious
limitations, for the triliterality of roots was not yet an accepted
principle. Menahem’s dictionary had an entry on ‘the root ‫’ צ‬,
which had seven sections. To these belonged (1) words meaning
‘fly’, like 2) ,‫נ צ א‬, ‫ ) צ י ץ‬words meaning ‘bloom’, like | ‫נ צ‬, ‫ צ י ץ‬,
(3) words meaning ‘quarrel’, like 4) ,‫ מ צ ה‬, ‫ ) נ צ ה‬the word ‫צ י‬
‘ship’, (5) the word ‘drought’, 6) ,‫ ) צ י ה‬a kind of animal, ‫ צ י י ם‬,
(7) the word ]‫ צי ל‬, taken to mean ‘monument’, ‘mark’. Schooled
by a long tradition to analyse by looking for a triliteral root, we find
it hard to understand that in the tenth century this was still not an
established procedure.
Menahem’s work was bitterly criticized by Dimash ibn Labrat,
 H IS T O R Y O F JE W IS H G R A M M A T IC A L S T U D I E S 63
who in particular was able to refine and improve the identification
and discrimination of roots, and to emphasize the predominance of
the triliteral against other explanations. He emphasized once
again the value of explanations from Arabic and gave a list of about
170 words as examples of how explanation from the Arabic could
be correct and necessary.
Yehuda ben David Hayyug (early eleventh century) was after­
wards remembered as the founder of a great advance in Hebrew
grammar. His study laid particular emphasis upon the verbs with
‘weak’ radicals, i.e. consonants like y or w which appear only in
some forms and disappear in others. Until this was clear, the tri­
literal principle could not be carried through consistently. By
making it into a consistent one, he used this principle to sort out
the verb into regularity and classify the behaviour of the various
weak consonants. The performing of this work, needless to say,
involved a careful study of the vocalization; for it is by the vocal­
ization that discrimination can generally be reached between
uncertain possibilities of weak consonants. Appeals to the vocal­
ization had already been made by workers like Dunash in criticism
of earlier suggestions; in the handling of the problem of weak
consonants this appeal becomes more intensive and more exact.
Abulwalid Merwan ibn Janah produced, also in the eleventh
century, a systematic approach to the grammar, including the
parts of speech, the possible combinations of vowels, the principles
of noun-formation, and (a particularly interesting departure) a
discussion of the use of tropes, i.e. cases where a word is used
(or, in the case of ellipse, omitted) for stylistic or rhetorical effect
in places where another would have been more normal. The
question how far a difficulty in the biblical text can be explained by
appealing to an anomalous usage for the sake of effect is obviously
one which runs through all of our present discussion. Later, for
instance, the AV got itself out of all sorts of trouble by an implicit
appeal to ellipse, which in the English it acknowledged by supply­
ing the omitted elements in italic type.
Most of these latter advances took place in the Muslim lands,
including Spain; and it was Abraham ibn Ezra (1092-1167) who
brought these achievements most fully into the lands of Christen­
dom. His works were written in Hebrew, and not in the Arabic
which for northern Jewry had been an obstacle to the understanding
of many grammatical researches. It was in southern France that
64 S O M E H IS T O R I C A L A S P E C T S
the Kimchi family lived—first Joseph, and then his sons Moses and
David. David Kimchi (1160-1235) wrote a grammar usually called
the Michlol (‘completeness, compendium’) and a dictionary known
as the ‘Book of Roots’. In this latter he added further etymological
comparisons, taken from late Hebrew and Aramaic, to the many
which had come down from the earlier scholars.
The systematic study of Hebrew vocabulary, which is implied in
the production of dictionaries, poses the question of synonyms,
words which are different in form but identical or closely similar in
meaning, and also that of homonyms, words identical in form but
different in meaning. Whole books were devoted to the collection
and analysis of synonyms; we may mention in particular the Hotem
Tokhnit of Abraham Bedarshi (later thirteenth century). Homonyms
were handled by Isaac Ha-levi ben Eleazar in his work Ha-Riqmah.
Let us summarize a few salient aspects of this rise of grammatical
study.
Firstly, the development of grammatical science began, in spite
of its careful attention to many details, with what would seem to us
now a very poor analysis of Hebrew grammar, so that it was an
achievement to recognize and formulate the triliteral root and the
difference between radicals and afformatives. By the end of the
development, on the other hand, a position had been reached which
shows remarkable similarity to the grammar generally taught in
modern times.
Secondly, comparative elements, with a stress on Arabic and
Aramaic, were prominent; although scholars varied in the degree
of their reliance on comparative study, the possibility of it was
present throughout the discussion. Even those who did not speak
Arabic, which was the living language of the southern grammar-
ians, had a strong basis for comparative linguistic consciousness
in the Targum. Comparative studies were hindered, however, b
the unsatisfactory degree of clarity then attainable about the phono*
logical correspondences. It is not that no correct correspondences
were known. Important correspondences such as that of Hebrew
‫ ש‬to Arabic /§/, or of Hebrew ‫ ש‬to Arabic /t/ (e.g. ‫ = שלג‬talj
‘snow’), were perfectly well known. But alongside these well-
founded correspondences there were many of more dubious char-
acter.1
1 For many examples in a recent publication see Wechter, Ibn Barun, pp.
54-60 and related notes.
 H IS T O R Y O F JE W IS H G R A M M A T IC A L S T U D IE S 65
Thirdly, the production and refinement of the grammatical
analysis of Hebrew was related to the use of the Massorah. The
grammarians found themselves required to refer to it with great
care, for it afforded them means of discrimination within all the
mass of originally undiscriminated material. The ability to distin-
guish, let us say, forms of the verb ‫ ח י ל‬from forms of the verb
‫ י ח ל‬, something that becomes particularly necessary once com-
parative information begins to press upon the possibilities of under-
standing one or the other, depends upon the careful collection and
comparison of the vocalization in all relevant cases. For this reason
it is important to realize that the basic fixation of the Massoretic
pointing was already done before the rise of grammatical and lexi-
cal inquiries. Even the manuscript B 19a, from which the text of
B H 3was taken, was written in 1008; the Cairo codex of the prophets
was made in 895 by Moses ben Asher. The pointing was done from
tradition, and did not depend on the results of the incipient
scholarly inquiries. Conversely, the fact that a classification of the
grammar was possible forms something of a witness to the value
of the vocalization in general.
Fourthly, on the lexical side it appears that for many words,
especially out-of-the-way words, the Jewish scholars of this period
did not attempt to rely on tradition for information about meaning.
Whether they worked from comparative sources such as Arabic, or
used the guidance of parallelisms and other elements of the Hebrew
text, they did not make a simple appeal to tradition or to current
usage. This does not necessarily mean that they were indifferent to
either; perhaps it was rather that tradition would give contradictory
answers. The emphasis on the peshat or plain sense made it im-
possible to accept the fanciful interpretations found in some cur-
rents of tradition.

(5) More Recent Trends

Only one or two remarks should be made about the general
change of scene as we move into the modern period.
Firstly, one notices a certain shift on the part of Christian opinion
away from dependence on the Jews for deciding the meaning of
Hebrew items. St. Jerome, in his reliance on the hebraica ueritas,
had had to turn to Jewish informants for his linguistic guidance.
The same situation is generally true in the Middle Ages, wherever
66 S O M E H IS T O R IC A L A S P E C T S
study of Hebrew among Christians went on. A particularly strong
influence was that of Rashi (Rabbi Solomon ben Isaac, of Troyes,
i 030-1105), who was, and among Jews has remained, by far themost
popular of the Jewish biblical commentators.1 By the mid-twelfth
century Rashi’s works were being used in the school of Hugo of St.
Victor, and Hugo’s notes include admonitions on the necessity of
understanding the Hebrew—since, after all, ‘the Greek texts are
truer than the Latin, and the Hebrew texts are truer than the
Greek’. It was, however, Nicholas of Lyra, who died in 1349, who
most fully brought the effect of Rashi’s exegesis into the currents
of Christian thinking.
After the sixteenth century, however, when the great Christian
Hebraists like Reuchlin made Hebrew an integral part of Christian
culture and education, there seems to be a lessening in the sense of
dependence on Jewish opinion. The central Jewish authority in the
sixteenth century, in respect of influence upon the Christians, was
Elias Levita (1469-1549).2 But no equally central figure followed
him in the respect of Christians, and the progressive current of
discovery which had characterized Jewish biblical study through
the Middle Ages seemed, once its fruits were passed on to the
Christians, to have become somewhat torpid. The study of Hebrew
was cultivated and passed on among the Christians, and the con­
ditions gradually came about in which it was possible for the
Christians to become unaware of their debt to Jewish grammatical
study and to forget that the Jews might be the ones who would
know what Hebrew words meant.
This, however, did not happen immediately. The Buxtorfs in
their handling of the text showed very deep respect for the Rab­
binic tradition; and in 1620 the elder Buxtorf published his Tiberias
sive commentarius masorethicus triplex, showing a close and lively
interest in the Massorah, here following in the steps of Levita. In
more recent times, however, there has come about a tendency to
give little attention to the Massorah, and the way in which it func­
tioned for the clarification and classification of the language has
been generally less well known.
Another relevant historical matter is the dispute over the inspira­
tion of the vocalization signs. Against the tradition that the points
were, like the consonants, given to Moses at Sinai, Ludwig Cap-
1 See particularly Hailperin, Rashi and the Christian Scholars.
a See G. E. Weil, Site L&vita.
 M ORE R E C EN T T R EN D S 67
pellus in 1624 published an argument showing that the points had
been added after the fifth or sixth centuries a .d . The more trad-
itional view, conversely, now extended the conception of biblical
inspiration so as to include explicitly the divine inspiration of the
vowel points. This debate led to renewed interest in textual
criticism and to an increased awareness that existing texts could
err. The remembrance of this debate has, on the other hand,
left another and a less happy inheritance, in the form of a feeling
that the vowel points are freely disposable at the pleasure of
the critic. Actually, as we shall see, the vowel signs are historical
evidence just as the consonants are, and while neither history nor
dogma can affirm that they are certainly right, history does mean
that they must be examined as evidence and not arbitrarily
replaced.
The seventeenth century is also notable for improved knowledge
of the Semitic languages, especially Arabic. Several polyglot lexica
were published, the best known being that of E. Castell, the Lexi-
con Heptaglotton (1669), which gave comparative registration of
Hebrew, Aramaic, Syriac, Samaritan, Ethiopic, and Arabic, along
with separate listing for Persian.1
All this implies a departure from the older idea that Hebrew was
in some way a divine or angelic speech, and leads towards a treat-
ment of it as a human language; to this one had only to add that it
was a derivative human language, and the movement of Hebrew
study into a more historical realm was accomplished. Albert
Schultens2 made this notably plain, for he regarded Arabic as the
purest and clearest of the Semitic languages and attached Hebrew
to the Semitic group as one dialect among many.
His Dissertatio theologica-philologica de utilitate linguae arabicae
in interpretanda sacra lingua3 argued that Arabic was in principle
just as old as any other Semitic language and could therefore be
applied to the understanding of Hebrew passages. That ‫ צ ל ה‬at
I Sam. 1. 6 meant ‘co-wife* ( = Arab, darra), and that ‫ ל ע ש‬at
I Sam. 14. 32 meant *bent, turned’,4 are points which Schultens
affirmed (not without the support of earlier scholars like Jerome or
Kimchi) and which have enjoyed reiteration in the twentieth

1 On this period see Kraus, pp. 70 ff.

2 See in particular J. Flick, Die arabischen Studien in Europa, pp. 105 ff.
3 Found in his Opera Minora (Leiden, 1769), pp. 487-510.
4 For further comment on this example see below, pp. 69, 98, 246 f.
68 S O M E H IS T O R IC A L A S P E C T S
century. Many of his other suggestions are less happy, and the
discerning modern Arabist Flick speaks of his work as a ‘misuse of
Arabic, driven to the farthest point’.1 Johann David Michaelis
(1717-91) may be mentioned as one who, following in Schultens’s
line, did even more to place the understanding of Hebrew within
the framework of the philology of Arabic and other Semitic lan­
guages. But meanwhile a further change was under way, for Arabic
studies were freeing themselves from the position of a servant to
sacra philologia, i.e. the position of being consulted only when
some obscurity in Hebrew required elucidation.2
In the nineteenth century the logic of textual emendation may
be comparable with the logic of the source analysis which was
equally dominant at the time. In source analysis scholars reacted
against the artificial expedients by which the inconcinnities of nar­
ratives had been covered over. Such inconcinnities, they held, did
not belong to the texts originally; rather, they had arisen by the
careless compounding of different sources. The scholar could trust
in his own sense of unity and sequence, and thus could analyse the
material in a way which would remove the need for artificiality.
Similarly, in the area of verbal difficulties, the scholar could rely on
what he knew; if the text made no sense, one need no longer hide
this fact, as the older exegesis had done, and one certainly need not
try artificial explanations for the anomalies of the text. Rather let
the text be emended to what it had been before careless scribes
corrupted it, and the difficulties would be gone. In conjectural
textual criticism the careless scribe plays a role somewhat analogous
to the role of the clumsy redactor in source criticism. One does not
need to press this parallel; but within limits it is true. Both at the
level of the larger literary units (source criticism or ‘literary’ criti­
cism) and at the level of the smaller verbal or graphic groupings
(textual criticism) we have an outburst of activity by the scholar
who does not recognize that the text, in the form in which it now
stands, and in the continuity in which it has been transmitted, is
final.
At any rate, in the highly productive scholarship of the later
nineteenth century, it does seem that textual and literary criticism,
rather than comparative philological operations, succeeded in oc­
cupying much of the time and energy of scholars. The great Well-
hausen, who was later to achieve fame as an Arabist, in his textual
1 Fuck, p. 107. 2 Flick, p. 122.
 M ORE R E C EN T T R EN D S 69
studies on the Old Testament seems to have concentrated almost
entirely on the textual approach, emphasizing the LXX in particu-
lar. At I Sam. 14. 32, for example, he does not even mention the
Arabic explanation suggested by Schultens, though more recent
scholars have reiterated it.1 At Amos 7. 4, where the rendering
‘creation* or ‘creatures* for the obscure ‫ חל?ן‬is a fairly easy sugges-
tion on the basis of familiar Arabic usage, Wellhausen as late as
1898 simply has nothing to suggest.2 S. R. Driver’s Notes on the
Hebrew Text and the Topography of the Books of Samuel, on which
whole generations of English-speaking students were trained, also
laid all its emphasis on textual and palaeographical, rather than on
comparative philological, insights; though occasionally it noted the
latter, as in the case of ‫‘ צ ר ה‬co-wife* at I Sam. 1. 6.
The predominance of textual treatments can be best seen in the
standard critical text which emerged from this period, namely B H 3.
This text gave space to hundreds of emendations, but quoted
rather few cases of philological treatment. In Hosea I find over 300
places where the editor has registered textual changes in the lower
apparatus; of philological treatments, however, I find only one.
This is at 7. 16, where the editor suggests the reading ‫ח ל ל עג ם‬
(itself assuming a conjectural rearrangement of the order of con-
sonants, plus an assumed haplography); the suggested explanation
is that this ‫ ז ל‬would be the Arabic zala ‘ceased*, or rather a
Hebrew cognate thereof, which in BDB is not recognized except
as the root of ‫‘ זו ל ת‬except*.
In Job, on the other hand, B H 3 registers a much larger number
of philological treatments. The editor, Beer, registers in the lower
apparatus some fifteen cases where he explains the reading through
reference to other languages than biblical Hebrew.3 In many of
these, however, the philological treatment does not go with a reten-
tion of the M T but with an emendation. These are, of course, only
a very small proportion of the philological treatments which have

1 J. Wellhausen, Der Text der Bücher Samuelis (Göttingen, 1871), p. 93.

2 J. Wellhausen, Die Kleinen Propheten (Berlin, 1898), pp. 7, 89. For a dis-
cussion of this possibility see below, pp. 260 f.
3 See 3. 14, Arab, haram ‘pyramid’; 15. 12, MH ‫‘ ל מז‬nod’; 15. 30, MH ‫נשר‬
‘cut down’; 18. 3, MH ‫‘ ט מ ם‬make blunt, stupid’; 19. 26, Accad. and Targ. ‫חו ר‬
‘look’; 20. 18, MH ‫* ל ע ס‬chew’; 21. 24, MH ‫* ע ט ם‬thigh’; 21. 27, Syr. ‫ה מ ס‬
*think’; 25. 3, MH ‫‘ בו ר‬boorish’: 33. 25, Accad. tapäsu ‘be sleek’; 36. 18, Aram.
‫* ח מ ה‬beware’; 36. 33, MH ‫* ע ל ע ו ל ה‬storm’; 37. 21, Syr. ‫‘ ב ה ר‬obscure’;
41. 18, Syr. ‫‘ ש לי ה‬spear’.
 70 S O M E H IS T O R I C A L A S P E C T S
been suggested for the book of Job; but it is interesting that they
did find a place in this way in the apparatus of B H 3. It is also inter­
esting to see how many of these suggestions depend on late
Hebrew, and how few are drawn from the Arabic and Accadian
sources which have recently become considerably more prominent.
In any case, in spite of the somewhat greater registration of philo­
logical treatments in Job and some other books, it remains true that
B H 3 in its notes remains overwhelmingly a monument to the
textual approach.
Nevertheless the strength of the tendency towards textual treat­
ment did not pass without challenge. Sometimes this challenge
came from scholars of a conservative tendency, whether Jewish or
Christian, who found it hard to accept the judgements now
fashionable among scholars and felt that these solutions were
arbitrary. Such scholars felt that it was better to explain the text
than to rewrite it by emendation or to analyse it by source-critical
methods; and if it had to be explained, the use of other languages
was one way to explain it. Moreover, some of the scholars of such
a conservative tendency were men of formidable philological
ability. Franz Delitzsch (1813-90) is the most obvious example,
and, in a somewhat different category, we may mention Dillmann
also; from various points of view, indeed, scholars who disagreed
with the reigning hypotheses, or who worked on aspects which were
not then in the centre of interest, contributed to keeping the scene
of scholarship varied.
The discovery of Accadian caused an important shift of emphasis
in Old Testament study. We need not discuss the impact of the
new knowledge about Accadian literature and religion. On the purely
linguistic side the discovery greatly altered the perspective from
which Hebrew could be seen, and in particular provided a very
different angle of view from that which had been mainly informed
by Arabic and Aramaic.
Friedrich Delitzsch in his Prolegomena (1886) is thus found using
some arguments which were to be re-echoed again and again 70 or
80 years later. It is dangerous, he argued, to remove unusual
locutions or hapax legomena by emendation, since ancient Hebrew
literature is very fragmentary. Emendation can thus simply remove
the genuine from the text. Accadian comes as a providential saviour
to the MT, and much reserve must be used in emendation (p. 69).
The whole study of vocabulary reflects upon the reliability of the
 MORE R EC EN T TR EN D S 71
M T and thus upon all other text-critical questions (p. vii).
Delitzsch’s work, not without its one-sided aspects, sets the tone for
much later philological work.
About the turn of the century we have two men who worked on
the text and whose notes provided numerous examples of both
textual and philological treatments, though with the emphasis
rather on the former than on the latter. Felix Perles in his Analekten
(1895, second series 1922) made many lexical suggestions; for
example he identified at Job 40. 17 a verb ‫ ח פ ץ‬meaning ‘make
straight, stretch out’, different from the usual ‫ ח פ ץ‬meaning ‘be
pleased’. Arnold Ehrlich in his Randglossen provided a long series
of textual and philological suggestions. Both of these writers seem
to have been a stimulus to later workers on the philological side.
In spite of these varying influences, however, the use of textual
treatment remained generally dominant in the earlier twentieth
century, and we have registered already its place in the apparatus of
B H 3. In the twenties and thirties, however, we should take note of
two points of view which stimulated discussion.
The first is the work of the voluminous if erratic scholar Wutz.
The main point of Wutz’s work, the transcription theory, i.e. the
opinion that the LXX was translated from a Hebrew text which
had already been written in Greek script, may fortunately be passed
over here; it is fully reviewed in the standard works on textual
criticism. It does concern us, however, that in quite a large number
of cases Wutz, finding a Hebrew word to be translated in a surpris-
ing way by a Greek word, accounted for this, subject to the tortuous
complications of the transcription theory, by explaining that the
Hebrew word had a sense made evident to us today by a cognate
Arabic or Aramaic or other Semitic word.
For example, there are certain places where the familiar Hebrew
‫ ד ר ך‬, usually meaning ‘way’, is translated by words for ‘sin’ like
α μ ά ρ τ η μ α . Wutz believed1 that the Hebrew word was actually an
(otherwise non-existent) 7‫י ך ב‬, explained in his laconic notes as
identical with (ar. darb vitiunC. That is to say, the original text had
this word ‫ ד ר ב‬, which was quite rightly rendered by the LXX but
in the transmission of the Hebrew text was mistakenly corrupted
into the familiar ‫ ד ר ך‬. Such explanations, often thrown out

1 Transkriptionen, p. 462. The word αμάρτημα occurs in MS. B at Hos. 10.13,

while αμαρτία is found at III Kingdoms 22. 53.
72 S O M E H IS T O R I C A L A S P E C T S
without any attempt at full justification, are quite frequent in
Wutz’s work.
Like the rest of Wutz’s work this aspect has received some
strong criticism. As for the ‘Arabic darV meaning ‘vice’, Nyberg1
flatly denied the existence of this word, as well as the necessity for
such an interpretation. At the best the suggestions of Wutz are
often poorly argued or substantiated; at the worst, they seem often
to be simple mistakes.
Nevertheless Wutz found an echo for his argument that LXX
renderings might derive from real Hebrew meanings which are now
recoverable only through appeal to the cognate languages. Driver
in 1934 enunciated the ‘important rule’ that:
The origin of the Greek rendering must be sought in the vocabularies
of the cognate, especially the Aramaic-Syriac and the Arabic, languages
before assuming that it represents a divergence from the Hebrew text.2

In saying this he made reference to Wutz’s arguments in

Transkriptionen i. 150-2. So careful a scholar as Winton Thomas
concedes that the writings of Wutz ‘contain a great deal of material
which will repay study by the Hebrew lexicographer’.3 Thus the
work of Wutz did something to stimulate the rise of philological
treatments.
Another, and a more important, criticism of the common use of
textual treatments came (1934-5) from the Swedish scholar Ny­
berg. Nyberg was highly critical of the proneness to emendation in
contemporary scholarship. The proud ich lese and the dogmatic
legendum seemed to him to be not a demonstration, but a denial, of
good scholarship. It was assumed that the extant text was tho­
roughly corrupt; but, Nyberg argued, if the text was really corrupt,
so that it became somewhat like the verse of Aristophanes supposed
to represent Persian (Acharnae 100), then the honest thing was not
to emend but to abandon altogether the attempt at interpretation.
Nyberg set out what he held to be the tacit assumptions of the
current fashion, and against these he formulated counterpositions
of his own. The following points are those most relevant to our
subject.
Firstly, the prevailing fashion, as Nyberg analysed it, assumed
1 Studien %um Hoseabuche, p. 81 n. Wutz, one may surmise, based himself on
the plural ’adrab ‘vitium, corruptio* of Freytag, ii. 19a.
2 ZAW M i (1934) 308. 3 O T M S y p. 255·
 M ORE R EC EN T TR EN D S 73
that ‘the understanding of the Old Testament books in the Jewish
community came very early to an end\ Since the text was only
half understood, the transcription was a mechanical one. This led
to all sorts of errors: words, meanings, and constructions which
never existed in the living language. If this is the case, then
The punctuation is nothing but a quite late attempt to wring from
this partially meaningless text a sense appropriate to a Rabbinic train of
thought.
Against all this Nyberg thought that the transmission, being live
and oral, though it certainly involved changes and the evolution of
new understandings, had never been mechanical and had never
consisted in mere guesses at the sense of a text mechanically pre­
served in writing.
Secondly, the fashion assumed that the Old Testament is written
in a uniform language, with the exception of the Aramaic sections
and the limited sections of ‘late’ Hebrew, such as Qoheleth and
Esther. In thinking this, Nyberg held, scholars were deceived by
the unitary punctuation system and the unitary grammar involved
in it. There are, on the contrary, regionally and temporally distinct
strata in the Old Testament.
Thirdly, the fashion supposed that the M T was a late and poor
form of the text, and that better and earlier guidance was to be
found in the versions, and especially the LXX. Nyberg refused to
give wholesale judgements on this kind of thing. He emphasized the
need for careful study of the origin of every variant, of the meaning
of the versions and their translation techniques, and of the history
of each version. In places he was ready to decide that the consonan­
tal text behind the LXX was better than that of the M T.1 But in
general the emphasis of Nyberg’s work was upon the reliability of
the MT. The versions are derivative from M T ; that is, they
generally represent poor departures from M T or a text close to it,
rather than good independent bases. He ended with a strong call
to the scholar to get back to the M T as the text which he had to
interpret.
Nyberg also held that the growth of the fashion for emendation
had gone along with, and partly been a result of, an isolation of Old
Testament scholarship from more general oriental philology.2 The
1 See the summary list, Hoseabuch, p. 114.
2 Nyberg, Z A W lii (1934) 243; Hoseabuch, pp. 6 f.
74 S O M E H IS T O R IC A L A S P E C T S
Old Testament could not be isolated from what was known of the
development of tradition in Islam, in Zoroastrianism, or in Man-
daism. The use of these wider perspectives, he believed, would
overcome the isolation in which the emending procedures had
grown and flourished.
Nyberg in his general remarks did not discuss the place of com-
parative philology in solving difficulties through appeal to the
cognate languages. In his detailed work on Hosea, however, he
uses this method quite frequently. Thus at Hos. 8. 9 he finds it
impossible to make sense of the word ‫ פ ל א‬if it has the usual mean-
ing ‘wild ass’; it means a ‘shoot’, following the Accadian^zV’z/.1 At
10. 10 he comments on the phrase 3 ‫ א ס ך ם‬. One would expect this
to be ‘in their binding’, but this is meaningless. He argues there-
fore:
The corresponding Arabic bi-asrihim is a comprehensive expression
of totality and means ‘all without exception’. . . . If the Arabic expres-
sion has been able to undergo this development of meaning, why
should the same be impossible for the Hebrew ‫ א ס ל‬which means
exactly the same thing ?2
Thus the sentence is to be translated: ‘All the tribes without
exception will assemble themselves against them.’
At Hos. 7.16 he says that ‫ ז ע ם‬obviously cannot have the normal
sense, i.e. ‘wrath’. But, he goes on, ‘anyone who knows Arabic
thinks naturally of zaKam\ The LXX and Syriac, with their render-
ings of ‘shamelessness, lack of discipline’, were at this point more or
less correct.3
Though some appeal was thus made by Nyberg to comparative
philology, it would be wrong to exaggerate his dependence on this
source. To a rather greater degree his arguments depend upon
showing the existence of unusual or unrecognized usages within
Hebrew. Himself the author of a Hebrew grammar, he strove to do
justice to the variety of possible expression in the language. Thus
the difficult phrase ] ‫ כ ^ ל י ב י כ ל‬at Hos. 4. 4 is dealt with
primarily through a syntactic treatment, the main appeal being to
other examples in the Hebrew Bible, and only a secondary appeal
being to Arabic and other Semitic languages.4
1 Hoseabuch, p. 64. 2 Ibid., p. 79.
3 Ibid., p. 60. Ar. zaam is approximately *claim’.
4 Ibid., p. 24.
 M ORE R E C EN T TR E N D S 75
In general, then, the importance of Nyberg’s work for us lies (a)
in its strong challenge to the fashion for emendation, and (6) in his
insistence on raising important questions of principle, in particular
about the relation between language and understanding.
In recent decades the original production of emendations has
decreased, and the tendency to resort to philological rather than
textual treatments has been growing. But the degree of movement
towards the philological approach has been very varied. In England
very distinguished philological work came from Professor G. R.
Driver of Oxford and Professor Winton Thomas of Cambridge. In
America some considerable emphasis on philological treatments,
based especially upon Ugaritic, has been seen in the work of some
of Albright’s followers. Jewish scholars like Eitan and Tur-Sinai
have also been extremely prominent.
German Old Testament scholarship, on the other hand, seems
to have been less prolific in philological treatments. Some of the
post-war commentaries continue to show a marked acceptance of
the guidance of B H 3 at points of textual difficulty. A detailed work
like Kraus’s Psalms resists the zeal for emendation found in the
older scholars like Duhm or Gunkel, but is not on the other hand
very productive of new appeals to the cognate languages. While it
discusses a number of those which have been made, the general
tendency is a sober and moderate defence of the M T as Hebrew,
rather than a fresh appeal to cognate material.
 IV

ASPECTS OF COMPARATIVE
PHILOLOGICAL METHOD

As we have seen, there had long existed a certain somewhat rudi­

mentary comparative approach to Hebrew, in which its words were
related to similar words in Aramaic and Arabic. It was only in the
nineteenth century, however, that comparative philology in its
modern form grew up. The modern movement can be distinguished
from the older in respect of {a) the degree to which it sought to state
consistent or systematic relations, as distinct from more isolated
examples; (b) the insistence on an historical approach. We shall dis­
cuss certain aspects of comparative philology as it is related (i) to
history, (2) to sound, (3) to meaning.
(1) History
Comparative philologists would normally claim that their ap­
proach to language is an historical one. The comparative perspectives
of medieval linguistic scholarship, even when accurate observations
of similarities and differences were made, and even when certain
historical data (like the difference between biblical and Mishnaic
Hebrew) were known, did not assume an historical form. The com­
parative philology which developed in the nineteenth century, by
contrast, was historical in method and outlook in the sense that the
comparative relations which it observed were set within an historical
framework of change and development.
The more obvious aspect of this historical approach was the
emphasis on the dating and the strict chronological control of the
available written sources. In biblical Hebrew this was connected
with the literary criticism which departed from the traditional
dating of books and evolved theories of their composition from
sources. New evidence, like inscriptions and the Amarna letters,
was likewise set in relation with previous knowledge through care­
ful chronological ordering. It thus became possible to write a
history of a language, provided that evidence over a sufficiently
 H IS T O R Y 77
long period was available, as was very obviously so, for instance,
with Greek or English.
More important for our purpose, however, the comparative
relations between two languages like Hebrew and Aramaic were
now worked out as the result of historical development from a com­
mon ancestor, and the reconstruction of this common ancestor
language, though not necessarily undertaken in detail, was implied
or assumed in comparative work. The characteristic of comparative
philology was not only historical organization of extant evidence
but also the use of extrapolation to produce a picture of previous
stages, themselves not directly evidenced, which would serve as a
basis for an historical and developmental account of the stages which
are directly known from evidence.
It might indeed be supposed that there is a quite different task
which is more properly ‘comparative’, that is, the study of the
differences and similarities between languages which are not re­
lated in origin—Chinese and Hebrew, let us say, or Turkish and
English. Comparisons of this kind, though not without their
interest, are not what has usually been meant by the term ‘com­
parative philology’. This term has meant the comparative study of
language groups within which signs of a common historical origin
can be detected; ‘comparison’ is not a general discussion of simi­
larities and differences, but the construction of an historical common
scheme within which the material of related languages can be
placed. It is thus possible to say that ‘comparative’ was not quite
the right word; Hoenigswald1 refers suggestively to ‘the process of
triangulation known misleadingly as the comparative method’.
Obviously the historical emphasis fits in easily with this
emphasis on the study of languages related in origin. If we are
interested in the differences between Chinese and Hebrew, we do
not necessarily advance the subject at all by tracing each back to an
earlier historical stage, for at that earlier stage they will be still
equally unrelated. If we are discussing Hebrew and Aramaic, how­
ever, the extension back into an earlier stage may bring us closer
to their common origin. Thus to be comparative, as comparative
philologists have construed the term, means to be historical.
Conversely, for certain early stages of development, it may seem
that there is no way to be historical except by being comparative in
one’s approach. The point of origin of the Romance languages is to
1 Hoenigswald, Language Change, p. 2.
78 A SPE C T S OF CO M PA RA TIV E P H IL O L O G IC A L M E T H O D
a considerable extent directly evidenced in the form of Latin. But
for the languages of the ancient world the previous historical stages
are not directly known. We have no direct evidence for Hebrew
before the earliest written texts in Hebrew. If we wish to know
something of Hebrew as it was before these earliest texts, we can
have recourse to the comparative study of Arabic or Aramaic.
Moreover, we can no longer rightly confine our study to one such
related language: in principle the prehistory has to be worked out
from the evidence of the whole group of related languages. Pre­
historic Hebrew was not related uniquely to Arabic or to Aramaic,
but had its own history, for the reconstruction of which we may
have to consider the evidence of Ethiopic or Mandaic or a modern
South Arabian dialect. Comparative study therefore involves in
principle the requirement that it should be the general comparative
study of a language family.
The reconstruction afforded when the other Semitic languages
are taken into account may enable us to state probable conjectures
of the nature of Hebrew before the time of the existing records. We
may then perhaps conclude that Hebrew at an earlier stage had
case-endings in its nouns, as classical Arabic still has. Again,
observing that Hebrew appears to have three sibilants, D, ft?, and
ft?, where Arabic has the two, /§/ and /s/, and that Hebrew /&/ (ft?)
often corresponds to Arabic /§/ while Hebrew /§/ (ft?) corresponds
to Arabic /s/, and combining this with similar observations apply­
ing to other Semitic languages, we may perhaps conclude that the
ancestor language had a series of three, which we may represent as
/s/, /s/, and /§/; and that these three have been reduced to two in
Arabic in such a way that the Arabic resultant phonemes com­
monly appear to be opposite to the pair with which they are
obviously comparable in Hebrew.
The total reconstruction of an ancestor language like proto-
Semitic is, of course, understood to be extremely hypothetical.
Indeed, for the purpose of the kind of philological treatment dis­
cussed in this book, no complete explicit reconstruction is required
or assumed. Most treatments offer the particular evidence of a word
with its meaning in two, three, or four of the known Semitic
languages. Thus the very hazardous task of stating what the proto-
Semitic form and meaning were is not necessarily required. Never­
theless the potentiality of such a reconstruction, within the limits
set by our information, is a background assumption of the whole
 H IS T O R Y 79
procedure. Most work, no doubt, has been done on the basis of
what Murtonen calls ‘the theory of the descent of all the Semitic
languages from one single mother tongue structurally similar to
them’.1 Such a theory may be right or wrong, and certain points in
our study may be relevant to a judgement on such a theory; but for
our practical purposes here we do not require to determine what
proto-Semitic was like, but only to observe that the existence of
such an entity was and is one of the assumptions in method made
by comparative philology.
The importance of the common ancestor of languages like
Arabic, Aramaic, and Hebrew has not lain in our reconstruction of
this language; rather, the reconstruction serves as a basis from
which the extant languages can be described historically as the
products of fairly consistent changes. How rigorous this consis­
tency must be is one of the questions which will have to be discussed
shortly.
Though comparative philology is thus historical in the sense that
the relations it observes are set within a framework of historical
change, there is another sense in which it is more ambiguously
related to historical method. The construction of a common ancestor
for extant languages is only a partially historical procedure, for no
direct historical evidence of the common ancestor is available; the
common ancestor is a hypothetical construction sufficient to ac­
count for later phenomena actually extant. Any addition to the
evidence known (as, for example, the discovery of a language like
Ugaritic) may mean that the picture of the common ancestor has
to be revised. Conversely the discovery of a previously unknown
language may provide confirmation of reconstructions which philo­
logists had previously made.
This constructive character of the work of comparative philology
is seen not only in the picture of ancestor languages like proto-
Semitic but also in other relations. When it is said that such and
such a form ‘originally’ had a certain sense, or that a Hebrew
meaning had developed from such and such a pre-Hebrew mean­
ing, these are constructions which rest upon evidence in the
historically known languages but are not in themselves directly or
empirically verifiable.
This constructive character of the work of comparative philology
is important for several aspects of the argument which will follow
1 JfSS xi Ci 966) 150 n.
 80 A S P E C T S O F CO M PA RA TIV E P H IL O L O G IC A L M E T H O D
elsewhere in this book. Philologists have at times tended to speak
rather apodeictically, as if their opinions rested on direct evidence.
This is not always justified. To say this is not to question the validity
of comparative philology; we are merely drawing attention to com-
plications inherent in its methods. The construction of linguistic
stages previous to those directly evidenced, though it is an impor-
tant procedure of historical philology, is a complicated and pre-
carious undertaking which may have to be subjected to constant
revision. One reason for this is that, in the circumstances here
obtaining, the law of economy of hypotheses does not apply: the
simplest connexion between two phenomena need not be the best
explanation, and much more complicated relations may be equally
probable. This constructive character to which we have referred
does not mean that comparative philology is unhistorical or without
factual basis; but its operations are indirect rather than direct.
It is true that the comparative method is not the only one avail-
able for the reconstruction of previous historical stages of a lan-
guage. Something can also be achieved by what has been called
‘internal reconstruction*.1 In Hebrew, for instance, the evidence of
the ‘segholate* nouns may suggest that the earlier form of ‫מ ל ך‬
was malk, while the earlier form of ‫ ס פ ר‬was sifr. This result is
achieved primarily not through comparison with other cognate
languages but through a consideration of phenomena within
Hebrew, such as the suffixed forms ‫ מ ל כ ך‬and ‫ ס פ ר ך‬. Much of
our information about the prehistory of Hebrew is attained in this
way.
Nevertheless a larger proportion of our historical reconstructions,
and for our purpose a more important part, rests upon a compara-
tive method using cognate sources. ‘Much better opportunities for
reconstruction exist where the older stage can be triangulated from
two or more independent later stages into which the speech com-
munity has separated/2 The possibilities offered by internal re-
construction are restricted if it cannot be accompanied and
strengthened by the presence of comparative method.
In summary, then, comparative philology has operated in the
construction of a developmental scheme within which the material
1 For a brief review, wrapped in impenetrable technical terminology, see
Hoenigswald, Linguistic Change, pp. 68 ff.; more simply, Lehmann, Historical
Linguistics, pp. 99-106.
1 Hoenigswald, op. cit., p. 69. Cf. Thieme, in Hymes, pp. 585-97.
 H IS T O R Y 8i

of related or ‘cognate’ languages may be set and seen historically.

For persons who, like Old Testament scholars, work at a point
near to the earliest available historical evidence of the language
involved, i.e. Hebrew, comparative methods seem to be a principal
resource for the obtaining of a view of the earlier history.

(2) Sound
In considering the modes in which the evidence of extant texts is
used in comparative philology, we may first give attention to the
study of sound and sound-laws. As is well known, this played an
important part in the rise of comparative philology, and Grimm’s
law in the field of Germanic is particularly famous among the
general public.
Actually much of comparative philology has avoided a real con­
cern with sound in the strict sense, i.e. with phonetics and
acoustics. Its work has, certainly in the Semitic field, been based
for the most part on the written signs. Where work has been done
on the modern languages and dialects, interest in phonetics has been
more immediately necessary. For the older languages, however,
where no direct phonetic information was available in any case,
philology has often devoted its primary attention to correspon­
dences of the written signs, and has shown considerable naivety or
vagueness when statements about sound going beyond this were
attempted. When a philologist tells us that a correspondence exists
between Hebrew jzj and Aramaic /d/, and that these represent
different developments from proto-Semitic /d/, he to a large extent
has in mind the letters T and *7 in the Hebrew or Aramaic script,
along with the sounds generally associated with them in the various
modern universities. He probably would not be able to give a
description of the changes in voice production which are involved
in his own statement, and the fact that his statement did not seem
to be dependent on such a description is one main reason why a
detailed philology, including assertions about ‘sound-laws’ and
‘phonology’, could grow up in the company of a very considerable
ignoring of the description of sound.
Philologists have known, nevertheless, that it was in the sound
rather than in the writing that the changes they trace had occurred,
and a very important place is given to the correspondences between
the different languages of a cognate group. Any comparative
82 A SPE C T S O F CO M PA RA TIV E P H IL O L O G IC A L M E T H O D
grammar will furnish a table of the sounds of the (constructed) parent
language, with the sounds into which each has developed in the
later languages. Thus we may hear that proto-Semitic had ft/ {th)>
Arabic /t/, Hebrew /§/, Aramaic /1/ and Ethiopic /s/. Hebrew /§/
then ‘corresponds’ to Aramaic /t/, as in ‫ שו ר‬or ‫ ר‬1‫‘ ת‬ox’. If Hebrew
‫ ש‬has a corresponding ‫ ש‬in Aramaic also, as in ‫‘ ק ד ש‬holy’, this
goes back to proto-Semitic /s/ or is explained in some other way.
An exhaustive statement of such correspondences is part of the
structure of a comparative grammar.
It has remained, however, a matter of some dispute how strictly
the correspondences which are normal are to be taken also as
certain and invariable. One school of nineteenth-century philo-
logists maintained the position that ‘sound laws admit no excep-
tion\ Others, however, have thought it possible to treat the
normal phonological correspondences more lightly. Meir Fraenkel,
for example, tries to argue that the familiar Hebrew |jn ‘syn-
agogue cantor’ is a ‘Hebrew variant’ of the Arabic haddäm
‘servant’. This at once involves a conflict with the usual correspon-
dences, since a Hebrew /z/ corresponds to a /z/ or a /d/ in Arabic,
but not to a /d/ in that language. Fraenkel goes on, however:
However, sound rules are not sound laws. Lagarde and Nöldeke
point out exceptions in the Semitic sound rules. . . . The ears and the
mouth of men are not machines, which follow iron physical laws and
are without exception. We know how the sound laws of Grimm have
been criticized. . . . The tables of sound correspondences, which we
find in Bauer and Leander or in Brockelmann, are correct in general,
but we cannot admit that they are right without exception for all
linguistic phenomena. These are not laws, but only tendencies of sound
change.1
Obviously the question is of vital importance for our subject.
We may consider separately two questions: (1) how far the sound
changes, by which the correspondences between cognate languages
have evolved, are regular; (2) how this applies to the situation of
difficult passages in the Old Testament.
As for the first of these questions the opinion of modern linguists
seems to be distinctly in favour of the regularity of sound change.
To argue this does not require, and indeed is not supported by, the
use of the concept of ‘iron physical laws’, or indeed of ‘laws’ of any
1 H U CA xxxi (i960) 69.
 SOU N D 83
kind at all. Examination of the modes by which changes occur,
including the statistical element and the fact that phonetic change
is a social phenomenon, leads still to the assertion that ‘phonetic
change is regular’ and that ‘whenever the proper conditions obtain,
phonemic change occurs without exception’.1
The consideration of dialect adds additional complexity to this
picture without altering it in principle. On detailed examination
the total entities called ‘languages’ may be found to break down into
local diversities, which may have differences from the correspon­
dences shown by the standard forms usually quoted. Thus, to take
a well-known example, proto-Germanic /‫־‬k/ should become /-x/
(commonly spelt ch) over the whole High German area. Dialect
geography, however, shows that the line between ik and ich is
different from the line between maken and machen2 Thus ‘the
various isoglosses in any bundle seldom coincide exactly’.3 The
units examined in linguistics are not absolute and homogeneous
entities which form separate ‘languages’; remarkable local diversity
exists.
This does not, however, in itself disprove the assertion that
changes are regular under given conditions. It means that the state­
ment of the conditions is more complicated than can be achieved
through the simple specification of the language concerned as a
whole. It is thus always possible that the detailed situation in some
area was more complicated than is suggested by a normal corre­
spondence such as that between Hebrew jzj and Aramaic /d/.
The existence of dialect (which will be discussed further in the
next chapter)4 and other complications does not, then, constitute
a valid objection to the conception that sound changes are regular.
This regularity is not only an important principle for compara­
tive philological operations in general; it is one of particular
importance for philological treatments in the Hebrew Bible. The
passages under discussion are, ex hypothesis obscure ones; the
relations of meaning between them and known Hebrew words, and
between them and suggested Semitic cognate words, are an open
1 Quotations are‫־‬from Gleason, pp. 3 9 4 7 ‫ ־‬. Cf. already Saussure, who insists
that phonetic changes are *absolutely’ and ‘perfectly’ regular, Course, pp. 143 if.,
while this does not depend on the use of any simple concept of *law’, ibid.,
pp. 91-95. For another recent summary statement see Robins, pp. 311 f.
2 For a simple statement, with diagram, see Lehmann, pp. 124 f.
3 So Gleason, p. 401.
4 See below, pp. 98-101.
84 A SPEC TS OF C O M PA RA TIV E P H IL O L O G IC A L M E T H O D
question. Since this semantic relation is a very open one, it is desir­
able that the phonological relations should be very closely estab­
lished ; otherwise we have two loose probabilities or surmises, with
no more coercive factor than the conviction that the result is a
suitable meaning for the text. The looser the phonological corre­
spondences, the more the weight of proof must fall on the semantic
suitability of suggestions offered; but semantic suitability, under
these circumstances, is perhaps not much more than mere guessing.
It is wrong that such guessing should be justified by the argument
that phonological correspondences are not rigorously mandatory,
and that the sense achieved ‘fits perfectly weir.
An otherwise unexpected variation from the standard corre­
spondences can well be accepted as important evidence where there
is no substantial doubt about the meaning of the terms in question
(no one doubts that ik on one side of a German isogloss means the
same as ich on the other side); but when the philological operation
is being conducted in order to identify words previously unknown
the matter is quite otherwise. The same is true of dialect; it is one
thing to use evidence known to belong to a particular dialect, but
quite another to call a form ‘dialectal’ when there is no evidence
for its belonging to any particular dialect and no series of dialectal
features into which to fit it; this latter procedure, in other words,
is merely using the general idea of dialect irresponsibly in order to
excuse an ignoring of the normal correspondences in a particular
case.
Many examples in the history of philology show how the deter­
mination to take phonological correspondences rigorously has led
to the discovery of new rules and thus to the extension of know­
ledge. It was precisely because rules such as those of Grimm were
taken to be strict and universal that scholars were forced to worry
over them and through this worrying produced new refinements,
such as Verner’s law relating stress to voicing. If the rules had not
been taken as strict and universal in the first place, the phenomena
would have been supposed to be haphazard and inexplicable, and
no new results would have been forthcoming.
Thus attention to strict phonological correspondence is one of
the ways in which suggestions, which seem at first to be seman­
tically satisfying, can be further probed.
The average person would be very likely to affirm the philo­
logical identity of English day and Latin dies. Yet this identity does
 SOU N D 85
not conform to the normal correspondences, and it appears that in
fact the words are not cognate. If phonological correspondences
were treated as a light matter, these scholarly doubts would have
been swept aside by the impatient sense that the semantic agree-
ment made it unnecessary to consider the problem of the phono-
logical difference. Only strictness in the application of sound
correspondences prevents the student from quick and easy conclu-
sions based on semantic identity or similarity.
In an article in 1956 Ullendorff, listing words held in common by
Hebrew and ancient South Arabian, included the equivalence
between Hebrew ‫‘ ש פ ת‬place, set’ and ESA sft ‘give’. This is
semantically an easy equivalence, since a similar spread of meaning
can be easily seen in the common verb ‫‘ נ תן‬give’. Beeston,
however, in a later note, observes that this equivalence involves an
irregular correspondence of sibilants, for the ESA /§/ usually
corresponds to Hebrew ‫ש‬, as in ‫‘ ש ל ה‬field’, ESA sdw. This being
so, Beeston is led to suggest a completely different etymology for
the words in question.1
It is not necessary for us to argue between the suggestions
advanced by these distinguished scholars. What is important is to
observe that the desire to support the normal correspondence is the
motive for further research into the matter. But for such a motive
the semantic obviousness of the suggestion as formulated by Ullen-
dorff would seem entirely satisfying. It is the desire to render
justice to the normal correspondences that leads to a further
examination of the question, and to the offering of solutions which
are more out of the way. Conversely, one can say that where the
rules of normal correspondence have been taken lightly, all sorts
of vague guesses from cognate languages have been offered.2
The statement of the phonological correspondences, then, forms
a kind of basic logic for the work of comparative philology.
1 Index, no. 310.
2 The results of an extremely cavalier treatment of the correspondences can
be seen in the work of John Gray on Ugaritic. See examples quoted in the review
of the 2nd edition of his The Legacy of Canaan by Pope in jf SSx i (1966) 228-41.
Correspondences work in only one direction: when we have established that
Hebrew x corresponds to Arabic y , this does not constitute a proof that Hebrew y
corresponds equally well to Arabic x. The correspondences are reversed if
Ugaritic and Hebrew rbs are related to Arabic rbq and Ugaritic 7m to Hebrew
slm; see Pope, op. cit., p. 231. Pope writes: *Gray appears to operate on the
assumption that it is permissible to reverse any process of permutation or sub-
stitute freely on either side of a phonological equation.’
 86 A SPE C T S O F CO M PA RA TIV E P H IL O L O G IC A L M E T H O D

(3) Meaning
Thirdly, we observe that the structure of comparative philology
includes an irreducible semantic element. Even where correspon-
dences in form are emphasized, this should not disguise the fact
that in the setting up of these same correspondences a semantic
element is involved. If we argue that Greek β in certain positions
corresponds to Latin v, and if we exemplify this by comparing
Greek βαίνω and Latin venio, we do this because we think that the
Greek word and the Latin are semantically close enough to be good
evidence. If we use as an example the relation between Hebrew
‫' ש ד ה‬field’ and Accadian sadu 'mountain, country’, we imply also
that we can see some kind of possible semantic relationship between
the meanings of the two words, even though this is a relation of
greater dissimilarity than that between βαίνω and venio. Putting it
conversely, it is doubtful whether we would accept words as illus-
trations of a phonological correspondence unless we could perceive
a semantic relation, or a possibility of such relation, between them.
It is true that a series of phonological correspondences can be set
up without this dependence on semantic similarities. Elements
such as inflexions, rather than words supposed to be of like mean-
ing, may be taken as the basis for comparison. In early Indo-
European philology it was the comparison of elements like verb
endings that allowed scholars to avoid the traps involved in the use
of words of like meaning.1 The same can be done in the Semitic
field. The comparison of the Accadian series iprus, taprus, taprusi
and the Hebrew series ‫ י ק ט ל‬, ‫ ת ק ט ל‬, ‫ ת ק ט ל י‬gives information
about vowel and consonant correspondences, even though it is
known that semantically the function of the Accadian tense is not
the same as that of the Hebrew tense. It is thus to some degree
possible for a series of phonological correspondences to be set up
without major semantic decisions about the similarity or dis-
similarity of the meaning of words.
Nevertheless only a sketchy comparative phonology could be
built up on this basis alone. Comparative philological works
usually contain long strings of actual word comparisons. Thus
Brockelmann2 gives a list of words all meaning 'beard’: Arabic
daqan, Hebrew ]pT, Syriac daqna, Accadian ziqnu. No semantic
1 On this see Jespersen, Language, p. 38, with reference to the work of Rask.
2 Grundriß, i. 335.
 M E A N IN G 87
difficulty is involved. But sometimes the semantic relations are
more complicated. In Hebrew TftX means approximately ‘say’;
but the Arabic ’amora is rather ‘command’, Accadian amd.ru is
‘see’, and Ethiopic ’amara, ’ammard is ‘know’ or ‘show’. This may
be in vague terms ‘the same word’; but the relation between the
meanings is much less simple and obvious than with the words for
‘beard’.
A large number of the decisions involved in setting up a system
of phonological correspondences contain, then, a semantic element.
Where possible, philologists may try to use for their basic corre­
spondences examples where the semantic element is minimal
(such as inflexional affixes and the like) or where it would be
accepted as highly obvious (such as a series of words all meaning
‘beard’, as quoted above). Nevertheless almost all comparative
grammars will use semantic criteria in setting up their lists of
correspondences and providing illustrations.
One reason for this is the statistical variation in the frequency of
the different phonemes of a language. Where a phoneme occurs
frequently and in a variety of contexts, it may be relatively easy to
obtain evidence of series of cognate words in which it occurs. In
other cases the evidence may be much more sparse, and therefore
more ambiguous. In Indo-European, for instance, the incidence
of the phoneme /bh/ was very much higher than that of /b/. It is
correspondingly more difficult to produce a convincing multiple
demonstration of the occurrence of realizations of jbj in the
various languages. This rarity means that scholars will probably
scan with greater caution the semantics of words purporting to
illustrate the correspondence. In Hebrew the incidence of 2? is
substantially lower than that of (say) 57, and this is one of the
reasons why the phonological correspondences involving the
sibilants are a matter of greater uncertainty; and since there is some
uncertainty about the normal correspondences, greater weight
must lie on the semantic convincingness of examples quoted.
Moreover, Semitic languages use in inflexional affixes and the like
only a very limited number of their stock of consonant phonemes,
and 2? is not one of them. Thus all attempts to state correspon­
dences for 127 involve the semantics of individual words.
In the research operations of comparative philology the scholar
uses a list of basic correspondences which have been built up
wherever possible with plentiful examples and with the use of words
88 A SPE C T S OF CO M PA RA TIV E P H IL O L O G IC A L M E T H O D
(or grammatical elements), such as the words for ‘beard’ just cited,
which do not present immediate semantic uncertainties. Only
because this list is fairly stable can the more adventurous research
into the highly doubtful words be undertaken with confidence.
This, however, introduces us only to the elementary aspects
of semantic problems in comparative philology. The meanings of
words were not only used in order to construct a basic series of
correspondences; study had also to be given to the various mean­
ings of cognate words in different languages, and to the historical
change of meanings in temporal stages of the same language; and,
as in our philological treatments of the Hebrew Bible, compara­
tive methods had to be harnessed to the task of discovering
meanings.
In the Semitic language field, one may assert without injustice,
the classical discipline of comparative philology showed a much
greater weakness in questions of semantics than in other aspects of
its work. Its careful and meticulous erudition in the classification of
forms and the tracing of their history was not matched by an equal
sophistication in the semantic area. Here on the contrary a remark­
able degree of naivety and even some considerable guesswork is to
be found, while purely quantitatively the work put into aspects like
phonology and morphology was vastly greater than that put into
semantics. In such respects one may say that semantics formed the
Achilles’ heel of comparative philology.
For this weakness on the semantic side it is not difficult to sug­
gest some reasons. Firstly, the generally empirical emphasis of
comparative philology encouraged an emphasis on forms rather
than on meanings. Forms are empirically attested in a way that
meanings are not. This is true in spite of the constructive character
which, as we have seen, attaches to the work of comparative philo­
logy. Even schemes which are highly creative and constructive can
cite actual forms which are attested, being reducible to visible signs
on paper or other mediums of writing. Such forms are the hard core
of evidence on which philological constructions rely and to which
they can in the last resort be referred.
In contrast with forms, meanings are rather slippery to handle.
What kind of empirical evidence is there for the meaning of a form
at this time or that ? The evidence for what was written at such-and-
such a time, and, behind that, the evidence for what was audibly
heard at such-and-such a time, appears to have a hard and tangible
 M E A N IN G 89
character which does not apply to assertions about what was meant
at the same time.
The historical emphasis of comparative philology reinforced
this failure to develop semantic doctrine. In the history of forms
written evidence may enable us to know just when any particular
form was in use. Even within the history of forms, we may note, it
is a much more precarious matter to decide when a form came into
use or fell out of use; for the empirical evidence, in the nature of the
case, is normally only of use and not of non-use, so that for non-use
one is dependent on one of the less desirable forms of argument
from silence. It is a still more uncertain and complex operation when
we move from forms to meanings and try to state just when and
how and why one meaning changed and was replaced by another.
Such historical semantic judgements are indeed possible where the
development of a language and literature can be followed from a
series of contemporary documents, as is the case in medieval
French, or indeed, to some extent, in the Old Testament itself as we
move from the older documents to the later. But historical philo-
logy has never been satisfied to follow the course of development
documented by extant documents; it has also sought to reconstruct
the history of the time before such documents existed. But for this
period, naturally, no empirical evidence exists. This has damaged
the study of semantic changes much more than it has damaged the
study of form changes.
Phonological development, as traced by comparative philology,
seemed to show a remarkable unity. If the phoneme /t/ passed over
into /t/, it seemed to do this by a more or less universal drift, which
affected the language as a whole and was largely (though not neces-
sarily entirely) independent of questions of meaning in particular
words. Such a change could be historically fixed in relation to
other changes, so that an historical account could be given in relative
sequence, even if no exact chronological data were available.
This kind of precision and logicality was not available for se-
mantic change. Historical approaches did not succeed in making it
fall neatly into sequence. For instance, we may suppose that the
Hebrew ‫ מנ ח ה‬underwent a change of meaning from the more
general ‘offering’ to the more particular ‘cereal offering’ which is
its sense in Leviticus. This change of meaning, however, has
nothing to do with general characteristics of words with the con-
sonants jmj or /n/ or /h/, and the understanding of it cannot be
9o A SPE C T S O F C O M PA RA TIV E P H IL O L O G IC A L M E T H O D
reached by gathering evidence empirically discernible through the
presence of certain consonants. The change is, indeed, something
that can be well studied through considering other words in the
same semantic field (such, for example, as the rise to prominence
‫ ״‬n r ® ; but the study of this involves us at once in a much more
literary and less empirical type of analysis. The movement of word
meanings is not easily statable in the general form which sound
changes have, for it depends on forces operative within, and pecu-
liar to, the particular semantic fields concerned.1Thus, though any
semantic change may gradually have come to exert influence on
the language more generally, semantic changes do not provide an
easy means of setting up a relative linguistic chronology such as
can be furnished on the basis of sound changes.
The comparative emphasis, like the historical, tended to make
an appreciation of semantic realities rather more difficult. We all
know the type of philologist who, when asked the meaning of a
word, answers by telling us the meaning of its cognates in other
languages. This over-etymological approach is the result of exces-
sive reliance on comparative thinking. The meaning of a word is its
meaning in its own language, not its meaning in some other. To
say this is not to deny that it is of considerable interest to know the
meaning of cognate words in cognate languages. But the character-
istic procedure of many scholars has been to start with comparative
data; and the attempt to state the meaning in the actual language
under study (in our case, Hebrew) has often been biased by a
striving to fit this meaning into a possible derivative process start-
ing from the comparative material. Thus the comparative emphasis,
which has done so much to clarify fields like phonology and morph-
ology, has often tended to confuse the field of semantics.
One prominent semantic operation is the statement of
analogies.
There is a familiar Hebrew word ‫ ב ט ח‬which means ‘trust, feel
safe’. This is commonly related to the Arabic bataha; but this latter
means rather ‘prostrate, fall down, lie low’. Scholars have some-
times maintained that the latter sense exists in the Hebrew also. It
has been identified at Jer. 12. 5:
: ‫ו ב א ר ץ ש לו ם א ת ה ב ו ט ח ו אי ך ת ע ש ה בג און ה י ר ד ן‬

1 C t below, pp. 170-3.

 M E A N IN G 91
The more traditional type of rendering is:
‘and though in a land of peace thou art secure,
yet how wilt thou do in the pride of Jordan ?’ (RV)
Driver, however, offers:
‘If thou fallest flat on thy belly in a land of peace,
how then wilt thou fare in the rising surge of the stream?’
(Cf. RSV ‘fall down’.)1
Again, at Job 40. 23, Driver argues that the text
‫ עי ח! ך דן‬-‫ ט ח כי‬3‫י‬
means that he (i.e. the crocodile) is lying flat on the mud.
Such a relation between a sense of ‘trust’ and one of ‘fall’ or ‘lie
down’ has been cited not only for this word but for others also. The
Hebrew ‫‘ ש ק ט‬be quiet, be at rest’ has been compared to Arabic
saqata ‘fall, drop’. Within Hebrew ‫ ^ ש ל כ תי‬literally means ‘I was
thrown’ but in context seems to mean ‘I was made to depend or
rely’. Blau, more remotely, compares Arabic ndma ‘sleep’, along
with ndma ’ila ‘rely upon’.2
Again, it may be argued that there is a relationship between ‫י ח ל‬
‘wait’, ‫ ח י ל‬or ‫ ל‬1‫‘ ח‬writhe’, and the root of ‫‘ חי ל‬strength’. If this
is argued, it may also be supported by adducing the example of
‫‘ קרה‬hope’, which is said to be equivalent to Arabic qawiya ‘be
strong’, while in Hebrew a cord is called 1‫ ק‬because it is twisted.3
The use of chains of semantic analogy in this way is character-
istic of comparative philology. The chains may cover different
meanings in one language, or may adduce examples from several.
At times we find Semitic philologists also offering examples from
outside the Semitic area. Thus meanings in the area ‘wind’ and
‘spirit, soul’ can be found for related or identical words not only
in Semitic languages but also in Latin and Greek. The relation
between Hebrew ‫‘ א מ ר‬say’ and Ethiopic ’ammara ‘show’ can be
paralleled with that between Latin dicere and Greek SeiKvvfju*
1 But contrast Kohler’s example of a ‘verfeinerte Semantik’, in O T S viii
(1950) 144 f., followed also in KB, p. 118b.
2 I n V T v i (1956) 244.
3 Cf. also ‫( מ תן‬Index, no. 208), and ‫‘ ה מ תין‬wait’. The problem with these
chains of analogy is that they do not give direct historical information, and the
process of development in one may be quite different from that in another
which looks logically alike.
4 This example already in my Semantics, p. 118.
92 A SPE C T S O F CO M PA RA TIV E P H IL O L O G IC A L M E T H O D

(4) General
Thus far we have spoken of comparative philology as a study
which sought to provide comparative historical statements organ­
izing the data from the various languages of a group, in our case the
Semitic family. But the period in which this work was at its height
(the nineteenth and early twentieth centuries) was also the period
in which large numbers of new texts from ancient times became
known. It was a time of discovery; inscriptions were found and
published, new texts and indeed whole new languages became
centres of scholarly awareness. Accadian and Ugaritic are the most
prominent Semitic examples; in Indo-European a similar situation
was formed by the discovery of Hittite and Tocharian.
The importance of this is that philological scholarship had to
deal not only with the organization of known linguistic evidence
but also with the processing and interpretation of new material. In
all this exploration comparative insights were of great importance.
Comparative methods, indeed, were not the only ones used.
Egyptian hieroglyphic was deciphered, and the grammar and
lexicon of this language worked out, without any substantial body
of comparative material to serve as a guide; such help as was pro­
vided came from bilingual texts, and from later stages (Coptic) of
the same language. Some languages, Sumerian for example, came
to be known, even though no comparative affiliations have been
certainly discovered.
Nevertheless comparative studies have often formed a very large
part of the scholarly apparatus used in work on a new text or a new
language. If a newly discovered Phoenician inscription contains a
word not previously known, the scholar will at once start to look
more widely around in the Semitic field. In the case of Ugaritic
the meaning of the texts has been worked out through a network
of comparative information which gradually became more refined.
This network provided by comparative identification eventually
fixes the contours of the poems sufficiently well for us to make good
surmises of the sense of words for which no comparative explana­
tion can at present be given.
The decipherment of entirely new texts thus gave a high prac­
tical importance to comparative philology; and the occupation of
so many scholars with the study of new texts contributed to the
great upsurge of philological, as against textual, treatments in
 GENERAL 93
recent decades. Nevertheless this practical application of compara­
tive philology was not without its dangers to that subject itself. It
meant that many scholars, when they used comparative perspec­
tives on a language like Ugaritic, were not comparative philologists
in the strict sense, and did not make themselves responsible for the
total task of synthetic organization of data from the Semitic field.
The primary interest of the scholar was to find an interpretation of
a new text. In the zeal and pressure of discovery, he might use
comparative methods in a way which, if it had been part of a truly
systematic comparative philology, would have been seen to be
inviable. The use of abnormal phonological correspondences is a
prime example. If these occurred so freely and so generally as is
implied by the philological suggestions of some scholars (if taken
cumulatively), then their effect would be no less than to shake the
whole fabric of comparison which philologists themselves have
carefully worked out. Thus a wide use of and appeal to comparative
philology does not necessarily mean that philological methods and
insights have been properly used or observed.
Another aspect of comparative philology which calls for com­
ment is that it has, on the whole, been lacking in introspection into
its own methods and has often failed to provide a satisfactory
justification for each decision as it was reached. To some extent
this is a natural effect of the working out of the method, and can be
seen in classic works such as Brockelmann’s Grundriß. The logic
by which comparative philological decisions are made includes the
use of an extremely complicated series of examples, many of which
present in themselves a series of different problems which demand
different answers. The complication of providing a rigorous logical
demonstration for every element in the structure of a comparative
grammar would therefore be immense.
One of the ways in which this difficulty has been met has been for
the philologist to present his results rather than his argument: a
kind of great hypothesis, which asks the reader to accept it as true
if he finds that it covers the data. Within such a corpus we often
look in vain for an argued justification of any particular section.
The method as a whole is taken to be common ground, and its
practical success in accommodating the evidence is the reason for
accepting it. This may help to explain the apodeictic and assertive
air of many philological treatments: it springs from the general
mode of operation of comparative philology.
 94 A SPE C T S O F C O M PA RA TIV E P H IL O L O G IC A L M E T H O D
In contrast with the historical and developmental emphasis of
the older philology, modern linguistics has laid greater stress on the
synchronic study of a language, as it operates at a particular time
as the means of communication of a speech-community. This does
not remove an interest also in the diachronic study of language.
Indeed, it can be argued that language study can be truly historical
only when it works with a picture of succeeding synchronic states
of the language as a whole, and that in this respect the older philo­
logy with its emphasis on historical development, because it failed
to see the languages as synchronically functioning systems, para­
doxically failed to be historical. By isolating the elements from the
system within which they worked, it sometimes actually tended to
damage historical appreciation.1
To sum up, then: an appreciation of philological treatments of
the Hebrew Bible has to include a sympathetic but also critical under­
standing of the discipline of comparative philology from which
they arose. On the one hand, the earlier forms of this discipline
contained certain weaknesses, the effects of which later appeared
in individual suggestions applicable to Hebrew. Some of these
weaknesses can be mended if account is taken of more modern
developments in the study of language. But my argument by no
means depends on a kind of linguistics entirely different in scope
from the older philology. The questions which we shall have to
uncover and develop were already very plainly present in the older
procedures. And, on the other hand, where philological treatments
of the Bible have been faulty, this has often been not because of
weaknesses in the basic philological discipline, but because the
canons of that philological discipline itself were poorly observed
by those who built upon it.
1 For an earlier criticism of ‘atomistic* study see Goshen-Gottstein in ScrH
iv (1958) 101 ff.
 V

PRELIMINARY QUESTIONS IN
PHILOLOGICAL TREATMENTS

(1) General
I f a word in the Hebrew Bible is to be identified, not as the
Hebrew word which it has normally been taken to be, but as
another word known from cognate sources elsewhere in Semitic,
it is clear that an immense learning would be required for the
proper handling of the matter. One would have to know not only
that such-and-such a word exists in Arabic or Syriac or Ugaritic,
but also where it is used, in what connexions, and with what
frequency. One would have to consider yet other Semitic langu-
ages, since evidence in another such (possibly Accadian or South
Arabian) might conceivably make it impossible to maintain the
simple connexions conceivable if only Arabic or Syriac is taken
into account.
Many philological treatments do indeed rest upon such an
encyclopedic scholarship and upon a refined judgement in the
wide fields of the Semitic literatures. It must nevertheless be
confessed that this required degree of erudition has not always
been present. It is possible, though it is not desirable, to short-
circuit some of this learning. The instrument which permits such
a short-circuiting is the dictionary. It is not a superhuman task to
learn what the dictionaries of the various Semitic languages are,
and to become familiar with their various scripts and their modes of
ordering material. Once this is known, all that is required is to
know the possible correspondences for any given Hebrew word
and look them up one by one.
The number of such correspondences will depend on which
consonants are involved, and which languages. For a Hebrew ‫כ ת ב‬
the student would have to consider only one correspondence in
the Arabic dictionary, namely k-t-b. For a Hebrew ‫ ע ז ב‬, on the
other hand, he would have to consider more. The ‫ ע‬could normally
correspond with either a /'/ or a /g/ in Arabic; and the T could
 96 Q U E S T IO N S IN P H IL O L O G IC A L T R E A T M E N T S
correspond with either a /d/ or a /z/. There would be theoretically
at least four possibilities. In the event there might well be less,
for not all of the groups theoretically possible might be in use in
Arabic. But where ‘weak’ letters such as /w/ are involved, the num-
1_ _ ‫״‬ ‫ ״‬c — __ _ u 1‫״‬ ‫_____________ ״‬ t >_______ ‫ _____________ ״‬a .
O ci U i p U S b lL U I iU C Ö JLilitJ UC I c t i g C . J t u p c 111 <X I C C C l l t <UL 1C 1C U1JL 111J \ u

in Job 9. 31 tells us that more than thirty combinations are

theoretically possible in an attempt to state the root and thereby
achieve a comparative philological contact;1 only a few of these,
however, are found to exist. Even in these more complicated
instances, however, it is still easy to work out what the theoretical
possibilities are.
The procedure, while a little laborious to describe, is thus quite
a simple one, assuming some gift for languages and a fair amount
of ingenuity. Some of the more facile examples of philological
treatments probably rest on some such simple dictionary-searching
procedure. From time to time, indeed, protests have been raised
against philological treatments on just this ground, that they are
based on searches in the dictionary, without any adequate know-
ledge of the literatures themselves.
One need not say more about this unfortunate situation. It is
right, however, that it should be mentioned. Because philological
treatments exude an air of immense linguistic scholarship, it is
only right to point out that the procedure may be a quite simple
one, even if applied with ingenuity. In any case, this leads us to
some further discussion of the phonological correspondences, and,
in due course, of the weaknesses in the available dictionaries.

(2) Metathesis and Dialect

Consonants cannot be treated entirely individually; a consonant
may in a certain environment be altered in such a way as to
produce a correspondence which in other environments is ab-
normal. This is normally described in the classical works as
assimilation and dissimilation. Hebrew D, for example, does not
normally correspond with Arabic /t/. But there is little doubt that
Hebrew ‫‘ ק ט ל‬kill’ corresponds to Arabic qatala with the same
sense. The departure from the correspondence otherwise normal
is related to the environment of the /q/ and the /l/.2 Our certainty
1 J B L lxxxiii (1964) 270; Index, no. 170.
2 Brockelmann, Grundriß, i. 154, gives a note on the development from proto-
Semitic ft/ following /q/ in Hebrew and Aramaic; some curious variations exist.
 M E T A T H E S IS A N D D IA L E C T 97
in accepting the unusual correspondence depends on the clear
semantic closeness of the words in the two languages.
Further, a word may be represented in a cognate language by
a word which has the same consonants but in a different order;
in the standard works this is usually known as metathesis.1 It may
not only be found between one cognate language and another; one
language may itself display alternative forms, with the same
consonants in a different order. The standard Hebrew examples
are ‫לה‬/?‫ ש‬or ‫‘ ש ל ^ ה‬dress’, and ‫ כ ב ש‬or 5‫‘ ש ב‬lamb’.
The probability of a metathesis will vary according to the
language with which Hebrew is being related; Ethiopic, for
example, seems to have a higher proportion of secure cases, in
relation to Hebrew, than Arabic has.
Moreover, many explanations which have been based upon
metathesis have not found general agreement, and must be con-
sidered uncertain. Brockelmann2 relates Arabic jhanal\ to Hebrew
‫נ ח ש‬, both meaning ‘serpent'; but this is doubtful, and is rejected
by BDB.3 The attractive relating of Ethiopic mahdrd ‘have com-
passion' to the Hebrew stem ‫ ר ח ם‬of the same meaning has been
disputed,4 though it is regarded as ‘probably' right by Leslau in a
recent discussion.5 Often one has to consider carefully the possi-
bility of a completely different explanation than metathesis, so that
the words cited have no common origin at all. Scholars who have
made liberal use of explanations depending on metathesis, such as
J. Barth in his Etymologische Studien, have incurred a good deal of
criticism as a result.6
In addition, the traditional comparative grammars have erred in
lumping together two quite different phenomena. Firstly, it may be
meant that a form existed historically and that thereafter a change
in the order of its phonemes occurred. This, one presumes, is in-
tended when it is said that ‫ ש ל מ ה‬is a metathesis of ‫ ש מ ל ה‬. Special
reasons brought about a change of sequence in this word, but did
not act so universally as to eliminate the older form from use.
At any rate an historical change from one form to the other is
supposed to have happened.
1 For a survey see Brockelmann, Grundrifi, i. 267-78.
2 Ibid., p. 275. This involves also a doubtful correspondence with the ‫; ש‬
and other explanations of the Hebrew word have been given.
3 BDB, p. 638a. 4 Praetorius, quoted in BDB, p. 933a.
5 Leslau, Contributions, p. 50.
4 See for example Brockelmann, Grundrifi, i. 268, and references there.
98 Q U E S T IO N S IN P H IL O L O G IC A L T R E A T M E N T S
Secondly, what is meant may be not an historical change, but
only the fact that the sequence is different from that which one
would ‘logically’ expect if everything were without anomaly. We
read that ‫ ^ ת מ ר‬is a ‘metathesis’ from an assumed *‫ ; ה ת ש מ ר‬but
it is not necessarily true that *‫ ה ת ש מ ר‬actually existed at some
previous historical point after which a shift was made. Perhaps,
under the given conditions, the /t/ was always put after the first
radical; there was no historical change, but only a difference from
what happened with other consonants.1 The incidence of meta-
thesis should therefore not be exaggerated by the inclusion of
false cases, and these latter should not make us overestimate the
probability of historical changes of order of phonemes.
Moreover, ‘metathesis’ means a linguistic change of sequence
observable in speech, and not a graphic disturbance, i.e. the
mistaken copying of characters in the wrong order. Occasionally
the word has been used in what appears to be the latter sense.
For ‫ וי ע ש‬at I Sam. 14. 32 Reider maintained that the verb ‘must
have been’ ‫ = ש ע ה‬Arabic sa'a ‘betake oneself’, ‘run’.2 He con-
tinues: ‘the error can be accounted for through metathesis, which
is frequent in Hebrew as in other Semitic languages’. The mention
of an ‘error’ appears to mean a graphic error. If so, this is different
from ‘metathesis’ as it is usually meant in comparative grammars.
Explanations involving metathesis were already produced by
the medieval philologists. Ibn Janah, for example, at Ps. 45. 2
‫ ר ח ש ל ב י ד ב ר ט ו ב‬, explains that the heart is ‘sweating out’ good
words, a metaphor for ‘producing’. This explanation depends on
the use of the Arabic rasaha ‘sweat’.3
In summary, then, we must judge that the existence of the
phenomenon traditionally described as metathesis, in the relations
between one Semitic language and another, is indubitable; but
that the appeal to it, in cases where the identity of the Hebrew
word is itself in considerable doubt, is statistically not very prob-
able. Only occasionally can we expect it to lead us to a true solution.
To the matter of metathesis we may conveniently join also the
matter of dialect. We have already discussed the place of dialect in
1 The two types are still confused in Moscati, Comparative Grammary p. 63.
2 H U C A xxiv (1952-3) 85; cf. above, pp. 67, 69, and below, pp. 246 f.
3 Cf. Eppenstein, p. 14. For a large number of additional examples, of which
many luridly illustrate the dangers of reliance on metathesis, while some others
are of substantial value and interest still, see Wechter, Ibn Barun, p. 55, and
especially n. 310, pp. 183 f.
 M E T A T H E S IS A N D D IA L E C T 99
general, and it remains only to show ways in which the appeal to
dialect has worked in particular cases.
Discussing the use of Arabic in the elucidation of Hebrew words
Guillaume writes:1
Arabic is a language of far greater content than ‘classical· Arabic
which is based on the language of the poets of Najd. Arabic includes
the dialects of all the tribes in all areas of the peninsula; consequently
it is legitimate to cite dialectical forms that are known and forms that
have arisen through metathesis or the interchange of consonants.
In the result, however, the acceptance of several variables,
namely the use of (i) any or all dialects alike, (2) metathesis, (3)
unusual consonantal correspondences, simply means a very vague
relationship between the Hebrew words and the Arabic words to
which they are being referred. The reader may judge for himself
from a random selection such as:
Arab, laja a ‘seek protection’ Hebr. 1?XI ‘redeem’
kamala ‘be complete’ *Ifrl ‘finish*
waraqa ‘strip of leaves’ p 1?*7 ‘locust’
maqqa ‘suck the breast’ p*ll ‘suckle’
Guillaume’s articles2 contain large numbers of examples no
more convincing than the above, often involving multiple meta­
thesis and/or abnormal phonological correspondences, plus great
deviation on the semantic side between close agreement and remote
distance. These examples may indeed have some interest when
considered in relation to the theory of biliteral bases ;3 but as direct
lexical equivalences they reduce to the absurd the very principles
to which they appeal.4
This is no argument against the use of dialect forms in the
search for guidance concerning difficult Hebrew words. There is
every reason to expect that, if cognates of Hebrew words have
been preserved in other languages, some of these have been

1 Hooke Festschrift, p. 109.

2 In Abr-Nahrain from 1959; republished as Hebrew and Arabic Lexicography
( 1 9 6 5 ) .

3 See below, pp. 166-70.

4 Cf. also the principle that ‘if certain consonants are known to be interchange­
able in Arabic . . . such changes can legitimately be used to provide a parallel
from Arabic to a word in Hebrew’, Abr-Nahrain i (1959-60) 4.
1 oo Q U E S T IO N S IN P H I L O L O G I C A L T R E A T M E N T S
preserved in non-standard forms, such as the dialects of non-
classical Arabic. Rabin, for example, has furnished some very
probable explanations in this way. The Yemenite kurkur ‘deep
river gorge’ provides an attractive annotation to the Hebrew
phrase ]1.‫ כ כ ר ה י ל ד‬This seems to make better sense than the
usual explanation which relates the word to the sense ‘circle’ or
‘round plain’.
Thus there is every reason to welcome studies which try to
adduce non-standard Arabic dialect forms for the elucidation of
Hebrew words. But there is also every reason to expect some
consistency between various instances of relation between Hebrew
and any one Arabic dialect, while the variety of dialects does not
mean that any or every dialect form may be used as an obvious
correlate to a Hebrew word which seems semantically to suit it.
Dialects not only exist in the cognate languages such as Arabic
but may also have existed in Hebrew. The rather uniform grammar
of M T may have smoothed out dialectal variations.2 One may
expect these especially in writers of particular known localities,
such as Hosea. But, of all the obscure expressions in the Bible,
not more than a small percentage are likely to receive convincing
explanations on the ground of dialect, on the basis of the informa-
tion at present existing. The attempt to press explanations based
on dialect for a large number of examples is likely to be self-
defeating.
These problems are particularly acute in special cases, such as
Dahood’s argument for a strong Canaanite background in Qoheleth.
The argument includes some rather surprising elements. There is
a familiar Hebrew word ‫ פ ע ל‬, which means ‘make, do’. There are
several places in Qoheleth (and in other books) where forms with
the sequence ‫ ב ע ל‬have usually been taken to belong to the familiar
Hebrew ‫‘ ב ע ל‬master’ or ‘owner’; but where, Dahood argues, the
word is in fact a dialectal variation with a /b/ replacing the /p/ of
the standard form. Passages in question include Qoh. 8. 8b, where
‘iniquity cannot save its doers (‫’) ב ע ל י ו‬, Isa. 54. 5, Job 31. 39
(where ‫ ב ע ל י ה‬means ‘her workers’), and Prov. 3. 27.
In support of this proposal Dahood adduces certain variations

1 Rabin, Ancient West-Arabian, p. 28.

2 Cf. the argument of Nyberg, cited above, p. 73. Some of the variations
most commonly noticed, e.g. nV&tP/n&VtP and tPM/ltZD, can hardly be called
dialectal in any case, since they alternate within the same sources.
 M E T A T H E S IS A N D D IA L E C T IOI
which seem to occur in Canaanite dialects, including Ugaritic:
Ugaritic lbs or lps9 nbk or npk, tbt or tpt\ Ugaritic 'rpty but Hebr.
'rbt\ Ya’udi nbs or tips. So he decides ‘that biblical Hebrew
should possess a number of by-forms is, accordingly, no cause for
surprise.’1
It is, indeed, no cause for surprise that this should be possible.
It is quite another thing to suppose that it has thus been proved
to be so, or even to be probable. Just to quote one counter-argu-
ment, one has to consider the possibility that the variations cited
are produced only in certain environments and are, in the language
of the traditional philology, assimilations and dissimilations; and,
moreover, that these are related to the peculiar phonetic character
of the sounds in Ugaritic and to the fact that this language had a
total phoneme stock substantially different from that of Hebrew.
Thus one of the variations occurs after /1/, a phoneme non-existent
in Hebrew; two others occur before /§/. The word for ‘soul’ in
Hebrew occurs not only elsewhere, but in Qoheleth itself, only in
its normal Hebrew form with /p/; the same is true of ‫‘ ש פ ט‬judge’
(Qoh. 3. 17). It is thus an entirely precarious undertaking to argue
that, when /b/ and /p/ may be free variants in certain Canaanite
dialects in certain environments, this may then be simply extended
and supposed to apply to quite any word in Hebrew.
In any case the series of proposals here under discussion, namely
that words including the sequence ‫ ב ע ל‬should be taken to mean
‘do, make’, naturally depends also on another circumstance,
namely that the sense ‘master, owner’ does not give equally good
or better sense; which, however, is far from clear.
It is important in general, therefore, in the circumstances of Old
Testament research, that we should not allow the phenomenon of
dialect to allow wild and arbitrary conclusions. As we have seen,
a considerable degree of regularity in phonological correspondences
has been fundamental to comparative philology. This is not altered
by the fact that dialect exists; dialect only makes more complicated
the statement of the relations.
(3) Loan-words and Words of non-Semitic origin
The lists of normal correspondences are made up for, and apply
only to, the words which have descended straight into the various
languages from their common ancestor. Many words, however,
1 Biblica xliii (1962) 361 £.
102 Q U E S T IO N S IN P H I L O L O G I C A L T R E A T M E N T S
were absorbed later from one Semitic language into another;
Hebrew adopted many Accadian words, and Arabic adopted many
from Aramaic. Such adoptions commonly diverge from the normal
correspondences, and the philologist in setting up his lists of
such correspondences must avoid using probable loan-words as
evidence. The form which the word takes in the receiving language
will depend on the way in which it was heard and phonemicized
at the time and place of its reception and assimilation. If we have
enough examples, it may be possible to state general rules. In the
older Semitic languages, however, we sometimes have only the
vaguest means of determining when and where a word was taken
over, and we may not have enough examples to provide any
consistent: pattern.
Thus words of Arabic origin in the Ethiopic lexicon, or words of
Aramaic origin in the Arabic lexicon, may not show the correspon-
dences with Hebrew which are normal for Ethiopic and Arabic
respectively; nor can they be used as evidence to establish the
validity of unusual correspondences between these languages and
Hebrew.
Sometimes, indeed, a philological treatment will depend on
just this principle, and argue that a Hebrew passage can be
understood because it includes loan-words from other Semitic
languages which precisely for that reason do not follow the normal
correspor. dences. The verb and noun ‫ כ ר ה‬occur as hapax legomena
at II Kings 6. 23, ‫ ו י כ ר ה ל ה ם כ ך ה ג דו ל ה‬. This text may mean
‘and he made for them a great feast’. The words are known in
this sense in Accadian and are registered in BDB, though some
faintheartedness made that lexicon rather doubt the text. Rabin
argues that these are loan-words from Accadian, and that the real
Semitic root is that found in Arabic qara ‘invite to a feast’, cf. also
Ethiopic ’aqrdyd ‘invite or receive as a guest’ (?), Ugaritic qry dbh
‘offer a sacrifice as food’. If this root descended direct to Hebrew
from proto-Semitic, its first consonant would be /q/ rather than
/k/, unless a quite unusual exception were being made. Therefore
it is more likely that the word has been absorbed from Accadian,
where /q/ has passed into /k/ before /r/.1 Thus there is no reason to
doubt the text at II Kings 6. 23. Moreover, examples of this same
Semitic q-r-y remain in Hebrew, spelled with a p because they
have not come in from Accadian. Rabin finds a pair at Exod. 3. 18,
1 Rabin does not in his note give evidence for this shift in Accadian.
 L O A N -W O R D S A N D W O R D S O F N O N -S E M IT IC O R IG IN 103
5. 3, where the somewhat puzzling phrase ‫א ל ה י ה ע ב ר י ם נ ק ר א‬
‫ ע ל י נ ו‬is taken by him to mean ‘God has asked us to hold a ban-
quet for him’.
What has just been said about words absorbed from other
Semitic languages applies much more forcibly to words which have
been taken over from outside the Semitic languages altogether.
Such words may not normally be used for the establishment of
phonological correspondences between the Semitic languages;
and they will be unlikely to provide solutions for difficult Old
Testament passages except in special circumstances.
The situations which involve words of non-Semitic origin fall
clearly into two types. The first is where a difficulty in a Hebrew
passage may have been caused by the presence of an unrecognized
non-Semitic word. The correct identification of this word, as a
word already adopted into Hebrew and used by the biblical writer,
would then remove the difficulty. The second situation is that
which arises when information related to non-Semitic words
comes to be unwittingly mixed up in the philological reasoning
which seeks to discover the true sense of a Hebrew word.
The first type of problem does not require much discussion here.
The main stock of words of non-Semitic origin in the Old Testa-
ment is well enough known.1 Most of these are words about the
meaning of which there is no real doubt, such as the familiar ‫ ה י כ ל‬,
used of the temple or part of it (from Sumerian through Accadian),
or ‫‘ פ ח ה‬governor* (Accadian).
There are also words the meaning and derivation of which is in
some doubt, but which bear unusual forms and thereby make
probable a non-Semitic origin; the ]‫ א פ ר ל‬of Cant. 3. 9 is an
example. Again, sometimes scholars have felt that a Hebrew word
has no adequate etymological background in the Semitic languages
and have therefore sought an explanation elsewhere. The promi-
nent example is flOD ‘passover*. Since the Semitic etymologies are
not very satisfactory, some scholars have suggested explanations
from Egyptian, and this can be supported by the tradition that the
passover originated in Egypt. Some of these explanations have the
peculiar feature that the D of the Hebrew word is interpreted not
as a root consonant but as the definite article of Egyptian (cf.
Index, no. 262).2
1 For a recent treatment see Ellenbogen, Foreign Words.
2 Cf. the recent survey in J. B. Segal, The Hebrew Passover, pp. 96 ff.
104 Q U E S T IO N S IN P H IL O L O G IC A L T R E A T M E N T S
Non-Semitic words are, it would seem, most likely to be identi-
fied if the words are nouns1 and if the reference seems to be to
foreign officials, institutions of foreign origin, unusual artefacts
and foods likely to be transported from distant lands; and par-
ticularly, perhaps, if foreigners are depicted in the text as speaking
or being addressed. Many such identifications though interesting
do not pose a very critical problem for the understanding of the
text. Either the word is known generally, e.g. as a commodity in a
list of such,2 so that the philological identification may serve to
clarify the exact identity of the referent; or the referent may be
taken to be known, as in the case of ‫‘ ת כיי ם‬peacocks’, and the
philological task is rather to trace the source of this word. Some-
times philological study may do something of both. One of the
advantages which study of non-Semitic origins may confer on us
is freedom from bondage to false etymologies from within Semitic
or from within Hebrew itself; thus, if ‫ חני ת‬is not a Semitic word,
we can be free from the dubious derivation of meaning from ‫ח נ ה‬
‘bend’ (‘as flexible!', BDB, p. 333b),3 and if ‫ ש לי ש‬should not be
Semitic, we may be free from a derivation from the ‘third man in
the chariot’.4
Where a non-Semitic origin for a word in the Old Testament is
considered, attention should be given to the date both of the
passage itself and of the events described in it. The probability of
Hittite words would be higher in the earlier period. Accadian
words may have come into Hebrew from very early times down to
the Babylonian exile. Persian words are conceivable from the
sixth century or so onwards, but acknowledged examples of these
(and also of Greek words) in the Hebrew parts of the Old Testa-
ment are few; the Aramaic sections contain many more.
Suggestions of Egyptian words have been most plausible when
they have been found in stories of the Israelite sojourn in Egypt,
1 Of twenty-two likely Hittite words discussed by Rabin in Orientalia xxxii
(1963) 113-39, two are verbs in Hebrew (‫ ת רג ם‬and 0‫) ק ל‬. T he predominance
of nouns may be explained partly through the modes of cultural contact, and
partly because borrowed verbs are harder to trace on account of the forms which
they must assume; cf. below, p. 132. Ellenbogen’s collection appears to include
only one verb (‫) א ס ף‬.
2 See for instance Rabin, ‘Rice in the Bible*, J S S xi (1966) 2-9.
3 Cf. below, p. 291 n.
4 A Hittite origin was suggested by Cowley, J T S x x i (1920) 326 f .; see discussion
by Rabin, Orientalia xxxii (1963) 133 f., who in the end returns to a Semitic
derivation, but not from the root which means ‘three*.
 L O A N -W O R D S A N D W O R D S O F N O N -S E M IT IC O R IG IN 105
or in materials like prophetic oracles against Egypt. A striking
series of names among the Levitical lists—Moses, Phinehas,
possibly Miriam, Putiel, Hophni—can be given plausible Egyptian
explanations. It was in Egypt that the crowd shouted ‫ א ב ר ך‬as
Joseph went by (Gen. 41. 43), and it is reasonable to look for an
Egyptian explanation, even if scholars have suggested too many
such rather than too few.1 The word ‫‘ י א ר‬river’, mainly used of
the Nile, has usually been taken to be the Egyptian 'itrw. The word
‘passover’ has already been mentioned. For the rare ‫ צ א ל י ם‬, some
plant under which Behemoth shelters in Job 40. 21-2, and taken
by BDB to be a kind of lotus, Humbert2 appealed to the Coptic
cal ‘branch’, related to earlier Egyptian d'rt ‘a fruit’.
Most examples of this kind, however, though significant for the
appreciation of Hebrew meanings, are a little different from what
has been meant in this book by a ‘philological treatment’. They
are mostly attempts to deepen our understanding of a strange
word; but they are not generally a resort in situations where the
understanding is in such uncertainty that only a new philological
identification can save the text from emendation. We shall not go
farther into this subject, because our main interest is in the use of
the cognate languages for the illumination of the Bible. It is just
because the languages are cognate that the appeal to them has
seemed so powerful and so convincing. It will always be likely that
some few difficult passages will find clarification through the
identification of a non-Semitic word, but we must on our present
knowledge expect such cases to be statistically infrequent in
comparison with the examples for which solutions are found
through the use of Semitic languages.
The second problem with non-Semitic words is, as we have
said, the possibility that forms and meanings of non-Semitic
words and information about them may involuntarily be taken up
into discussions which are logically valid only if the material is
Semitic in origin. To make mistakes in this regard is not as
difficult as might be imagined. One source of it lies in dictionaries
which do not differentiate words of foreign origin.
The lexical stock of Ethiopic, for instance, and still more that
of the modern Ethiopian dialects, includes many words from

1 See BDB s.v., pp. 7b-8a; B H 3 apparatus ad loc.

2 Humbert Z A W lxii (1950) 206; cf. Driver in Levi Della Vida, i. 237.
106 Q U E S T IO N S IN P H IL O L O G IC A L T R E A T M E N T S
Cushitic and other African sources. Perhaps the most striking
example of all, however, because the most likely to be a cause of
error, is .Arabic. Classical Arabic in certain areas included many
Persian words. In works of reference these are not always marked
as Persian; and the risk of confusion is all the greater in that some
of the shorter Persian words bear forms not unlike common
Arabic formations.
Thus, to take a prominent example, Arabic has dana (root dyri)
‘borrow’, of Semitic origin, and also dana (root dyn)y with a whole
series of derivatives of standard Arabic form, meaning ‘profess a
religion’, which is of Persian origin; and, moreover, it is possible
that we should distinguish, in respect of origin, between din
‘religion’ ( = Iranian daena) and din ‘judgement’ in the familiar
phrase yam al-din ‘the day of judgement’. The latter probably
involves borrowing from Aramaic or Hebrew, since the sense
‘judge’, basic to this phrase, seems to be characteristic of these
languages rather than of Arabic itself.1A similar situation obtained
within Persian itself, where we have pairs like dar ‘gallows’
(Iranian origin) and dar ‘house’ (Arabic), to say nothing of pairs
which are very similar though not identical, such as mehr ‘love,
friendship, sun’ (root as in the name Mithra) and mahr ‘marriage
portion’ (Arabic, cf. Hebrew 2.(‫ מ ה ר‬The ancient philologists knew
of the existence of this problem, but certainly did not have the
means to identify every case rightly.3
There is a Hebrew personal name ‫ י ר ד‬, andN othinhisP^rso^n-
namen (p, 231) relates this to the Arabic ward ‘rose’. This in itself
is semantically a possible suggestion, since there are other Hebrew
personal names which use names of plants, such as Habakkuk
(a plant name known from Accadian) or Tamar (palm-tree). But,
as Kopf points out,4 the Arabic zoard is of Persian origin. Noth’s
explanation will therefore not work. Kopf goes on to say that there
is an Arabic word of real Semitic origin, ward ‘bold, brave’, which
is actually used in Arab personal names and which fits the Hebrew
name very well. Later Plebrew, of course, displays also the word

1 For an analogous situation in Hebrew itself see below, p. 109.

2 For these words see A. K. S. Lambton, Persian Vocabulary, pp. 65, 162.
3 Use was made of a rudimentary study of the rules of consonantal com­
patibility, which will be discussed below, pp. 178-81; see references in Greenberg,
Word vi (1950) 163.
4 V T viii (1958) 179.
 LO A N -W O R D S A N D W O R D S O F N O N -S E M IT IC O R IG IN 107
‫לרד‬. ‘rose’. This is an example where, as mentioned above, the
Persian word is deceptively similar to a Semitic formation.
In fact quite large numbers of examples can be cited where
scholars have cited words from Arabic without noticing that they
are of Persian (or other non‫־‬Semitic) origin and therefore probably
not relevant for texts like the Old Testament. The Arabic kinnara
‘piece of linen’ is not of Semitic origin and therefore probably will
not do to explain the Ugaritic knrt.1 Wutz at Hos. 12. 2 discovered
a word ]EffiH which he supposed to exist also at Ps. 68. 32 and
which he )explained as identical with ‘ar. dusfan nuntius lend—to
quote^the whole of Wutz’s laconic note. As Nyberg points out,
however, this word came into Arabic from Middle Persian and is
not relevant.2 Again, there is a somewhat obscure phrase ‫אגן‬
‫ ה ס ה ר‬at Cant. 7. 3. Most commonly, perhaps, it has been taken
as ‘a rounded bowl’ (so RSV; English numbering, 7. 2) or the like.
Reider, offering a different interpretation,3 used the Arabic jauna,
which he understood to mean ‘disc’ (of sun or moon). Kopf,
criticizing this attempt, says that it rests on a misunderstanding of
the Arabic dictionaries, which refer to the colour rather than the
shape of the sun, and that in any case the word in question is a
Persian word, the classical gun ‘colour’. This word in turn, in-
terestingly enough, did in the end find its way into Hebrew in the
later ]11), cstr. ]1), ‘colour, manner’; but this does not make it a
valid comparison for Old Testament times.
This may be particularly troublesome if the sense obtained by
reference to a word of Persian origin is not very far removed from
the sense obtained by reference to a word of Semitic origin. For
a moment of light relief, we may illustrate this from the case of
‘checkmate’ in English. One often hears that this means ‘the king
is dead’, shah being the familiar Persian word and mat being taken
as the familiar Arabic word ‘is dead’, cognate with Hebrew ‫ מ ת‬.
I have often heard this example taught to beginners in Hebrew as
a mnemonic to assist them in learning the word ‫ מ ת‬. But in chess
the king is not dead; and if the word for ‘king’ were Persian one
might expect the word for his death to be Persian also. Actually,
both words are Persian, and mean ‘the king is helpless’, which

1 Ullendorff, in J S S vii (1962) 342, against Gray, Legacy, p. 86.

2 Wutz, Transkriptionen, p. 321; Nyberg, Hoseabuch, p. 94 n.
3 Reider in H U C A ii (1925) 94; Kopf in V T v i (1956) 293 f. See Lane, p. 490c.
108 Q U E S T IO N S IN P H I L O L O G I C A L T R E A T M E N T S
actually fits the situation of the game.1 The Oxford English
Dictionary, avoiding bigotry, gives one explanation under ‘check-
mate’ and the other under ‘mate’. The matter shows how easy
it is to fail to notice the difference between Persian and Arabic
words.
Sometimes a contamination of senses between words of different
origin in the same language may have taken place. In Jer. 10. 18:
‫ח?ני קולע את״יוקזבי האךץ‬
the verb is commonly rendered ‘sling out’; the noun ‘sling’ is
‫ ק ל ע‬.· Driver, however, has argued that the meaning is ‘uproot’ or
‘remove’, on the basis of the Arabic q-l- ‘remove from its place’.
He goes on to trace the semantic development, and his first series
of words includes the following:
Hebr. ‫‘ ק ל ע‬uproot’; Arab, q-l- ‘extract’; qalaa ‘a loosened clod’
(as extracted from the ground), maqla ‘stone quarry’, qaVa ‘fort-
ress’ (as originally a rocky fastness), q-l- ‘large stones’ [ = qalaa
of Lane, p. 2992?]; Syr. qulaä ‘clod, mud’.
The obvious weak point in the series quoted is qaVa ‘fortress’,
for the idea that this got its name from being a rocky place, while
not impossible, is not immediately convincing.
In fact there has been some suggestion that this word for
‘fortress’., a quite common word, is not of Semitic origin.2 Belardi
in a recent article derives it from an Iranian kaläta ‘hill fortress’.3
If this is so, he goes on, any connexion made in the Arab con-
sciousness with the verb q-l- ‘sling’ would be only through
popular etymology trying to give an a posteriori justification to the
sense by attaching it to a military function.
Whether Belardi is right in his suggestion I would not presume
to judge; but it does indeed look attractive. It does not in any case
have any immediate effect on Driver’s suggestion in respect of
Jer. 10. 18. But it does show us that words may be of non-Semitic
origin even when they appear to fit in with some kind of Semitic
1 It occurs to me that Arab users of the phrase may have thought it meant ‘is
dead*, which would be a further interesting complication; but I do not know if
this is so.
2 Fraenkel, Fremdwörter, p. 237, already doubted that this word was originally
Arabic. Siddiqi, p. 70, makes a suggestion similar to that of Belardi, but without
the explanation of how the word came to be related to the stem q-l-.
3 Belardi in the Annali of the Istituto Universitario Orientale, Naples, Sezione
Linguistics, i (1959) 147-50» following H. W. Bailey, op. cit. 118-20.
 L O A N -W O R D S A N D W O R D S O F N O N -S E M IT IC O R IG IN 109
etymology, and even when they may be felt so to fit by speakers of
a Semitic language. Though we can trace separately the origin
of Semitic and non-Semitic words in relation to a given language,
we cannot be sure of separating the two in their semantic inter-
relations once both have come to be used together. Thus non-
Semitic words may come to be fitted into existing linkages of
form and meaning in a Semitic language. This, if true, will greatly
complicate the task of tracing a semantic history and applying it to
biblical Hebrew in which the non-Semitic component will not be
relevant.
The same kind of problem may be illustrated for Hebrew with
the familiar word ]‫ ל י‬. Driver, I think rightly, holds that the example
at Esther 1. 13 is ‘religion’, from the Iranian daena\ this is con-
firmed in the phrase as a whole, *pTI ‫ ל ת‬, because the other word,
m , is certainly Persian. The Hebrew dictionaries are therefore in
a sense wrong in listing under ]‫‘ ל י‬judgement’. But it is quite
probable that Hebrew speakers did the same thing themselves, so
that this word, in so far as it continued to be used, became con-
taminated with ]‫‘ ל י‬judgement, justice’. For ‘religion’ the Jews
later said rather 1.‫ל ת‬
Where words from two different origins have come closely to-
gether in a Semitic language, it may be very much a matter of
chance whether we have knowledge to see their distinct origin.
There is a Syriac mlat meaning ‘to study, take care’, and it would
seem theoretically a possible semantic development to relate it to
the sense ‘rubbed’, ‘smooth’, which can be related to Hebrew ‫מ ל ט‬
‘escape’ and in Syriac itself appears as mlat ‘smear’ (with oil or
chalk). Analogies for such a development might be thought of—
for example ‘polite’, or ‘erudite’ (from rudis ‘rough’). In fact,
however, the Syriac mlat is derived from the Greek fieXerr) and is
thus a homonym with the mlat of Semitic origin.2 Examples of
this kind may have been much more frequent than our information
enables us to know, and would be particularly difficult to detect if
the senses of the two words overlapped in some degree or if popular
etymology devised notions which bridged the gaps between.
Another example shows how one may be misled by non-Semitic
material and also by false trains of semantic analogy. The word for
‘sailor’ in Hebrew is 1 .‫ מ ל ח‬have several times heard this taken as
1 Cf. above, p. 106.
2 Schulthess, Homonyme Wurzeln im Syrischen (Berlin, 1900), p. 37.
no Q U E S T IO N S IN P H I L O L O G I C A L T R E A T M E N T S
from ‫‘ מ ל ח‬salt’. Not only does it have exactly the same consonants,
but the noun formation is the familiar one used for many occupa-
tions. A moment’s thought, however, should suggest that the
meaning by analogy would probably be ‘salt-worker’ rather than
‘sailor’.
In fact, however, the word is of Sumerian origin, from ma
‘ship’ and lah ‘propel’; so that there is no connexion with salt at
all. (It is, when one considers the beginnings of navigation in the
ancient Near East, not obvious why it would be connected with
salt in any case.) Clearly, the impulse to interpret ‫ מ ל ח‬as some-
thing to do with ‘salt’ arises from the English usage ‘an old salt’
—itself a modern usage, first recorded for 1840 by the OED.
This analogy is an entirely misleading one.
This bads us nevertheless to the interesting realization that the
semantics of this Hebrew word, since it is the name of an occupa-
tion, have acted to draw it into the familiar qattal pattern used
for so many occupations. This may be the reason for the double
/l/, which is not explained by the Sumerian original, or by the
Accadian version of it. Thus, though the word was not of Semitic
origin, and though it was not derived semantically in any sense
connected with salt, nevertheless it took its place within the series
of patterns available for words coming from the Semitic root
m-l-h and conformed in part to the semantic information conveyed
by the choice of a member from that series. If we did not have
clear information from Sumerian about the origin of the word, we
would have been largely groping in the dark about the semantic
development. Another striking example is the ‫ שלג‬of Job 9. 30,
which is probably not ‘snow’ but ‘soapwort’. This is not a semantic
derivation from the sense ‘snow’, but a word from outside Semitic
altogether. Yet the context, taken along with the natural con-
nexions between water, snow, and washing, makes it very under-
standable that confusion should arise.
Thus non-Semitic words, even when we are in a position to
identify them clearly, once admitted to a Semitic language can
begin to have effects in word formation and semantics. There may
be many hundreds of cases where this has happened but where
we are unable to trace it, the only evidence left being a perplexing
variety in the semantics of words which in form seem to be plainly
Semitic.
To sum up this section, then, all philological treatments have to
 L O A N -W O R D S A N D W O R D S O F N O N -S E M IT IC O R IG IN h i

exercise caution lest words of non-Semitic origin are involved in

the formal or semantic histories which they endeavour to trace.
Some such words are fairly obvious from their form and are
correspondingly less likely to cause confusion. Others may easily
pass unnoticed. All philological treatments appeal to formal and
semantic histories in cognate languages, but many of these histories
may involve the influence of non-Semitic words, which influence
is not necessarily, and not even probably, transferable to other
Semitic languages.

(4) Area Preferences Within Comparative Philology

Are there, among the languages cognate with Hebrew, some
which are likely to be more productive of philological suggestions
than others ? We shall later (ch. 7) discuss the extent to which the
lexical stocks of the various Semitic languages overlap. Meanwhile
it is right to mention some more general factors, which have
influenced scholars in their preference for one group rather than
another, within the Semitic languages, as the source for attempted
elucidations of Hebrew.
One such factor, more social than genuinely linguistic, is the
love of the scholar for his own specialization, e.g. Accadian,
Egyptian, or Ugaritic. This is especially evident when a new area
of study is being opened up, as with Accadian in the later nine­
teenth century and with Ugaritic in the twentieth. There is then a
strong impulse to take the new knowledge and apply it to the
solution of old problems. Driver thus speaks of the ‘pan-Ugaritism’
which has arrived and will pass as other fashions like the ‘pan-
Babylonian’ have done.1
Sometimes these forces are supported by other kinds of inter­
pretative special pleading. Yahuda, who produced a long if
unreliable series of Egyptian terms in the Pentateuch, was animated
among other things by a will to disprove the ‘Wellhausen’ type of
historical criticism.
Scholars are also affected by the simple matter of what they
generally get to know. To many Hebrew scholars Arabic has been
the most familiar cognate language, because it is taught at most
universities and made familiar through visits to the Near East.
Post-biblical Hebrew itself, on the other hand, has often suffered
1 In J S S x (1965) 117.
112 Q U E S T IO N S IN P H IL O L O G IC A L T R E A T M E N T S
neglect in Christian scholarship, apart from special cases like the
Dead Sea Scrolls. The modern Ethiopic languages, and even
classical Geez, are also less widely known. Ugaritic, by contrast,
has had great prominence because of its closeness in time and style
to biblical texts, and the attention it has received has thus in
turn made it easy and natural to use it as a main source for philo­
logical explanations. Not without importance, conversely, is the
size of the vocabulary in the relevant languages. The immensity
of the recorded vocabulary of Arabic is a main reason for its
prominence in philological treatments of Hebrew. Given a con­
sonant sequence in Hebrew, there is perhaps a better chance of
finding a possible formal correspondence in Arabic than in any
other cognate language. The extent of its vocabulary, and the
variety of senses attributed to its words, form the subject of not a
few jokes and proverbs. The size of the Arabic lexical stock,
however, carries with it certain dangers, which we shall discuss
shortly: especially there is the question whether all this vocabulary
is properly and accurately recorded.
Though scholars have in fact often displayed a preference for
one cognate source as against another, it is doubtful if any decisive
reason in principle can be given for any such preference. In
principle all Semitic material is likely to be relevant. It is doubtful,
therefore, if it is a valid criticism to say (for example) that a
scholar depends on Arabic rather than Accadian or Ugaritic.
Such criticism in any case, apart from its logical precariousness,
has often been very unfair. Dahood, for instance, has sometimes
in his enthusiasm for Ugaritic let it appear that Albright, using
Ugaritic and Phoenician, forms a contrast with Driver who
appeals to Arabic.1Even if it were true that Ugaritic and Phoenician
had a higher value, it would be false to ignore the dependence on
Arabic (and Hebrew!) for the original decipherment of Ugaritic;
while it is mere folly to suggest that Driver’s work has neglected
Ugaritic or limited itself to Arabic. There are, indeed, scholars
whose work has been vitiated by an over-exclusive attention to one
branch of the Semitic languages, but Driver can hardly be included
in this number; Dahood himself would be a more justified target
for criticism on this score.
If any principle for preference between rival sources were to be
sought, three different scales might be considered: (i) Date:
1 Biblica xxxviii (1957) 306; more recently, xlvii (1966) 403 f.
 AREA PREFERENCES W IT H IN COMPARATIVE PHILOLOGY 113
sources nearer to the Old Testament in time might be considered
more likely to produce solutions. (2) Geographic proximity:
sources closer to Palestine might be considered more reliable.
Even if late in date, they might go back to similar earlier cultural
conditions and might retain traces of common vocabulary.
(3) Closer affinity within the classification of the Semitic family.
This would favour North-west Semitic languages against East
Semitic languages like Accadian or South Semitic languages like
Ethiopic. Experience, however, suggests that lexical isoglosses do
not coincide with the broader groupings of the languages and dia­
lects; certainly many sound comparisons between Hebrew and
Accadian have been made, and the same is true of Ethiopic.
In fact, while these scales deserve some consideration, it is clear
that nothing can be decided by them for the individual case,
though certain statistical trends may be noticed. Dialects of recent
date, of remote locality, or of different group within the classifica­
tion of Semitic, may yet provide good parallels. Ullendorff1 cites
the word nfr ‘to fly’, ‘bird’, as an instance of a word found
only at the opposite local and temporal extremes of the Semitic
spectrum, in Ugaritic and in Tigrinya, and doubtless other such
instances could be found.
Sometimes preferences are made on the grounds of social,
historical, and anthropological judgements. The most important
example is Arabic; it has often been supposed that it is a very
primitive tongue, barely differing from proto-Semitic. This
linguistic view is associated with the social belief that the mode of
life of the Arabian bedouin is identical with the cultural situation in
which Semitic languages were first spoken.
Moscati, for example, in a recent work,2 maintains that circum­
stances in Arabia, particularly poor communications, ‘make for a
greater degree of archaism, whether linguistic or ethnic’, and
concludes both that Arabic provides the best basis for the re­
construction of proto-Semitic and that the bedouin of Arabia come
close to the ethnic type of the ancient Semites. Even in anthro­
pology and social history, however, this view might be questioned ;3
while, on the side of linguistic history, a quite different view can
be quoted from Bergstrasser:
1 Orientalia xx (1951) 273 f.; V T vi (1956) 197.
2 The Semites in Ancient History, p. 33.
3 Cf. the works of Caskel cited in the bibliography.
114 Q U E S T IO N S IN P H I L O L O G I C A L T R E A T M E N T S
In spite of these features of high antiquity [the reference is to the
retention of the vocalic and consonantal stock of original Semitic],
Arabic is on the whole the most complete representative of a later
shaping of the Semitic linguistic character: old freedoms, individual
anomalies- and unevennesses are removed, in the morphology through
the strict carrying out of unitary analogies, and in the syntax through
firm regulation of the possibilities of usage and precise delimitation of
the area of meaning of all syntactic means of expression.1
It would thus be wrong to suppose that the experts have given
any clear or simple approval to the view that Arabic is peculiarly
close to the ancestor of all Semitic languages.
Idiosyncratic elements in Arabic, that is points at which it has
developed away from the original Semitic heritage and con-
structed patterns and elements peculiar to itself, will be less likely
to provide evidence for the elucidation of Hebrew texts. For
instance, Ps. 68. 9 and Jud. 5.5, both ancient poems, have a phrase
‫ סיני‬HI, which has long puzzled commentators. Some have taken it
as a rather stupid gloss, meaning ‘this is (i.e. refers to) Sinai’;
others have considered emendation to ‫סיבי‬ - * T
‘Sinai trembled’.
The modern fashion is to use a philological treatment, and to
say that the sense is ‘He of Sinai’, ‘the Possessor of Sinai’. God is
mentioned immediately before and after, and this therefore gives
a good parallelism. The treatment cites the well-known Arabic
du ‘the possessor of’; and one may add the South Arabian names
for deities, such as dt hmym.
This looks like a solution. Birkeland, nevertheless, has argued
that it will not work.2 For, he maintains, this development of du
‘possessor of’ is peculiar to Arabic, and accordingly will not apply
to this Hebrew text.
We do not have to decide the rights and wrongs of this here;
but it illustrates the effect of lexical idiosyncrasy upon all argu-
ments which seek to transfer phenomena and meanings from one
Semitic language to the elucidation of texts in another.3

1 G. Bergstrasser, Einfuhrungy pp. 134 f.

2 Birkeland in Studia Theologica ii (1948) 210 f., followed by Kraus, Psalmen,
p. 467; but contrast very favourable acceptance by Albright and his followers,
e.g. Moran in Wright, op. cit., p. 61. Some considerable semantic discussion
would be required in order to decide whether the instances adduced by Moran
from Ugarit and Mari are a real analogy.
3 For other examples in which the question of idiosyncrasy is discussed see
below, pp. 119, 172, 193 n., 224; also the discussion of Arabic vocabulary items,
P R O B L E M S O F T H E L E X IC O G R A P H IC A L T R A D I T I O N 115

(5) Problems of the Lexicographical Tradition

Mention has already been made of the danger in any excessive
dependence on the dictionary. Such dependence, as shown by any
simple procedure of searching the dictionaries for possible words,
becomes even more dangerous if the dictionaries themselves are
faulty.
Of modern dictionaries not so much need be said, since students
will be aware of the main problems involved in the use of them.
Nevertheless one or two points will be mentioned as particularly
deserving caution. Many modern dictionaries have been affected
by the etymological emphasis characteristic of much comparative
philology.1 This in itself can hardly be simply called a fault; but
it is a source of greater difficulty for our present purpose than is at
first evident. Because of it a dictionary may strive to state a ‘basic
meaning’ or ‘original meaning’ in addition to the meaning actually
found; or it may strive to give some other sort of comparative basis
as the starting-point of the article. Once this is done, the structure
of the article may be moulded in order to provide a justification for
the claim that the extant meanings flow from this basis or origin.
If this method is followed, it becomes possible not only that
doubtful etymological material will be inserted (because the
article will seem incomplete without it), but also that the statement
of meanings actually found will be distorted by the force of the
attempt to provide an etymological rationale. Even where lexico­
graphers only classify the meanings of words under various headings,
this will very frequently imply theories of semantic history. Thus,
paradoxically, for the investigator pursuing a philological treatment
of an Old Testament text, the etymological interest which past
lexicographers have had may be a very mixed blessing.
In some languages different dictionaries give different degrees
of prominence to etymological information or etymological
hypotheses. The Chicago Assyrian Dictionary, for example, does
not generally give etymological comparisons except at the end of
the article, and does not allow them to alter the information set
out in the article on the basis of the actual Accadian literature.
The corresponding dictionary of von Soden, on the other hand,

below, pp. 116, 162, 165 f., 16911., and that of glottochronology in general,
below, pp. 184-7.
1 Cf. above, p. 90.
116 Q U E S T IO N S IN P H IL O L O G IC A L T R E A T M E N T S
will normally classify a word at the beginning by reference to a
Semitic comparative etymology.
Apart from etymology some modern dictionaries may include
words anc. senses which are too specialized or too remote to form
a good parallel to a Hebrew word.1When the sense ‘deceitfulness,
vain speech’ for Arabic haraba has to be found from Dozy (i. 356),
and when even there it depends on the peculiar form hurbayr, this
is strong evidence that the sense is a special development within
Arabic, and unlikely to be applicable to Hebrew ‫( ח ל ב‬Index,
no. 138). The normal sense in Arabic is ‘destroy’.
Much more serious problems attend us when we consider the
practice of the ancient and medieval lexicographers. Mention has
already been made of the large size of the vocabulary recorded for
Arabic, and anyone who has done even a little work with a large
Arabic dictionary such as Lane’s knows of the bewildering variety
of meanings which can be registered in it. With some further
experience the scholar often comes to have some scepticism of
explanations which rest entirely upon the Arabic dictionaries, for
experience has shown him how easily some kind of suitable sense
for a difficult word can be gained from them. The best analysis
known to me of the problems as they impinge upon the Old
Testament is by L. Kopf; and, having no specialized competence
in this field, I shall refer to his work and state the main points as
I understand them to be.2
Firstly, the great Arabic dictionaries familiar to Western scholars,
such as those of Frey tag and Lane, do not rest directly upon the
literature itself, and are not formed by excerpting and citation
from actual texts, but are themselves compilations from older
Arab dictionaries. Sometimes interpretations within these older
dictionaries have suffered misreading or misunderstanding on the
part of later workers, whether within the development of Arabic
lexicography or in the process of translation into English or Latin.3
So industrious a user of the Arabic lexicon as Driver can thus be

1 Cf. also below, pp. 165 f., 265, etc.

2 See Kopf in V T vi (1956) 286-302; BO xii (1955) 134-6 (a review of Al-
Yasin); also Fuck in ZD M G cvii (1957) 340-7; Wild in ZD M G cxii (1962)
291-8; most recently Wild, Das Kitab al-A in. Wild agrees in the main with
Kopf (see pp. 41-57 of his book) while at one point qualifying the generality of
Kopf’s judgement (p. 57, n. 197). Haywood, Arabic Lexicography, does not con-
tribute to our question.
3 Instances in Kopf, BO xii (1955) 135, V T vi (1956), e.g. 290 if.
P R O B L E M S O F T H E L E X IC O G R A P H IC A L T R A D I T I O N 117
found to deplore the lack of a really scientific lexicon of Arabic,
which would have to be founded upon the actual literature.1
Secondly, the medieval Arabic lexicography did not act very
critically or discriminatingly towards the material which it
assembled. Quite secondary applications of words are found
quoted alongside normal and frequent usages, and metaphorical
applications may receive the same kind of treatment as normal
ones. A striking case is quoted by Kopf from the Qâmüs. The
familiar word kursi (cf. Hebrew ‫ ) כ ס א‬receives in this dictionary,
alongside the normal sense ‘chair’, the very surprising one ‘know-
ledge’. This goes back to the Quranic verse 2. 255 which says:
‘His seat comprehends heaven and earth’. The tradition of exposi-
tion and theological discussion which has gathered around this
verse has found its way into the lexicon as if it was one of the actual
meanings of the word.2
Let us quote Kopf again:
The Arabic poets developed a great number of metaphorical désigna-
tions for the limited group of things and ideas which formed the object
of their descriptions. Hence the supposed richness of Arabic in syn-
onyms, which are alleged to amount to hundreds of designations for
the camel, the sword and so on. In fact in many cases these are adjectives
which express only certain properties of the objects designated but
because of frequent use developed into standard terms.3
Or again:
The Arabic dictionary, in the form in which we now have it, usually
explains words empirically; that is, the explanations are mostly fitted
to the use of the respective words in a particular context, and therefore
are very often too specific. Herein lies one of the chief reasons for the
very numerous senses, widely diverging from one another, which the
Arabic dictionaries accord to various derivatives of the same root.4
1 See V T S iv (1957) 6. Cf. also Flick’s judgements in his Die Arabischen
Studien in Europa \ on Freytag—‘only a somewhat enlarged and improved edition
of Golius’ (p. 166); on Lane—‘in respect of method it does not go beyond Golius
and Freytag, for it fails in principle to adduce the factual usage of the literary
works extant’ (p. 169); in general—‘an Arabic lexicon resting on the texts them-
selves remains even today one of the most urgent tasks of Arabic studies in
Europe’ (p. 84).
2 On this see also Kopf, ‘Religious Influences on Medieval Arabic Philology’,
Studia Islamica v (1956) 33-59; also Wild, op. cit., p. 50, who cites other even
more extravagant cases, such as the glossing of dhk ‘laugh’ as ‘menstruate’ at
Sura 11. 71.
3 Kopf in V T vi (1956) 298.
4 Ibid.
118 Q U E S T IO N S IN P H I L O L O G I C A L T R E A T M E N T S
According to Kopf, the Arabic dictionaries explain expressions
in a purely practical way, without indicating the basic meaning and
the semantic development (p. 300). Or, as we might put it, there
is no ade quate discrimination between the information given by a
word (which we might call its ‘meaning’) and general character­
istics of that to which it refers (its ‘reference’).1 To give a very
simple illustration, we may call a woman a flower, but if we then
make a dictionary and give ‘woman’ as a gloss under the entry
‘flower’ we shall be likely to cause some confusion.
Again, many of the meanings given are not real linguistic
information but are a product of the lexicographical process itself.
This can happen [a) through etymologizing, i.e. the effort to
explain a word through a favoured derivation; (b) through the
telescoping of past exegetical discussions;2 (c) through the collec­
tion, and the representation as different existing senses, of the
suggestions made by different scholars. On this last Kopf says:
A large part of the vocabulary, which the Arabic philologists listed
and explained, was not known to them either from daily use or from
extended reading. Their chief task thus lay not in the establishment of
exact and fitting explanations for words which were current to every
educated man, but in the discovery of the meanings of infrequent and
unknown words, which they probably met for the first time in the course
of their professional activity.3
Similarly Wild points out that the interest of lexicographers was
not in ordinary daily speech but in literary usage. The examples
cited are, where possible, taken from poetic passages, a fact which
increases the possibility that very special applications of words
may be cited misleadingly.4 Genuine examples taken from the
poetry could even be supplemented with instances constructed ad
hoc by the lexicographer.5 Meanwhile common and frequent
words are by-passed without notice of any kind; Wild cites kalb
‘dog’, katlr ‘much’, and kull ‘all’, left without any mention in the
Kitab al-A in .6
Finally, the English of a dictionary like Lane’s may be am­
biguous, or may not be a certain rendering of its Arabic sources,
matters well exemplified by Kopf.
1 This distinction of information and reference is taken up again below,
pp. 291 ff, 2 Kopf, op. cit., pp. 299 f., if I understand him rightly.
3 Kopf, op. cit., p. 296. 4 Wild, op. cit., pp. 41 f.
5 For an example see Wild, op. cit., pp. 44 f.
6 Wild, op. cit., p. 41 and n. 2.
P R O B L E M S O F T H E L E X IC O G R A P H IC A L T R A D I T I O N 119
That the Arabic lexicographical tradition is not wholly reliable
for our purpose (a purpose, needless to say, entirely foreign to
Arab lexicography itself) is no new idea. Jouon in a memorable
phrase wrote:
Si minces que soient les filets d'eau coulant de la source hébraïque,
il est d’une bonne méthode de les exploiter à fond avant d’adresser ses
pas aux eaux étrangères de l’infini et complaisant Qamous.1
Similarly Ullendorff, in discussing the hapax legomenon ‫ ה ב ל‬of
Isa. 47. 13 and proposing the sense ‘worship, do homage’ (which
fits quite well), takes issue with the attempt made by Al-Yasïn to
connect the word with Arabic habr ‘depressed land’, and writes:
It seems to me impermissible to exploit the Arabic dictionary in this
manner (thus treated it will yield almost any meaning desired of it) by
detaching a highly idiosyncratic meaning-variant from a specific poetic
context.
Similar problems are occasionally found with Syriac, where a
meaning has been taken from the indigenous lexicography. They
occur also in the use of Accadian lexical tablets. Sometimes the
equivalences these offer have been construed as evidence for the
existence of words and meanings not clearly found in continuous
texts.
At isa. ii . 15, ‫ ו ה ח ר י ם י ה ו ה א ת ל שון י ם ־ מ צ ר י ם‬, the text has
often been altered to ‫‘ ו ה ח ר י ב‬dry up’ (cf. LXX ip yj^œ a et); the
familiar verb ‫ ה ח ר י ם‬is not elsewhere used with cosmic elements
like the sea as object. Driver wishes to retain the text, appealing to
an Accadian haramu which he says means ‘cut off’. This word
appears to exist, however, only in a lexical list, glossing Sumerian
kud ‘cut’. One of the other words used to gloss kud is bataquy and
of this word the note sa më ‘with reference to water’ is added.2
It is quite precarious to extend this to haramu ; von Soden rather
attaches the latter to the Semitic root hrm, which would seem to
imply the more general sense of ‘separate’.3
Similarly, Driver, trying to show the existence of a ‫‘ ש מ ל‬cast out,
reject’, quotes a syllabary which has the equation sum-mu-ru ==
bu-us-su-ru and argues that the latter is related to the Aramaic
‫‘ ב ס ל‬despise’.4 The former word is thereby identified with Syriac
1 Mélanges de la faculté orientale (Beirut), iv (1910) 10.
2 Von Soden, p. 114; actual use of this verb for interruption of water-courses
can be quoted.
3 Von Soden, p. 323. 4 Von Soden, p. 142; cf. below, p. 141.
120 Q U E S T IO N S IN P H I L O L O G I C A L T R E A T M E N T S
s-m-r ‘send, dismiss, discharge; reject’, and then in turn a Hebrew
cognate is identified. Again, in the argument that there is a Hebrew
‫ י צ א‬meaning ‘shine’, as of gold and silver, it is quoted that Ac-
cadian csu and namaru are used alike as synonyms explaining the
same Sumerian verb.
A famous instance which involves these questions is the word
‫ שי ל ה‬in Gen. 49. 10; ‘the sceptre shall not depart from Judah,
nor the ruler’s staff from between his feet’ (thus far RSV) ‫ע ד‬
‫ כי־יבאעזילה‬....
• T

The meaning of this has long been a subject of dispute.1

It is not easy to know what function the place-name Shiloh
would have, if it were meant here. Traditional Christianity took
‘Shiloh’ to be a Messianic name.2 The too obvious emendation
1‫ מ ש ל ה‬his ruler’ has been tried; and perhaps the commonest ap-
proach has been to construe as something like ‫‘ ש ל ה‬to whom it
belongs’ (so RSV, and cf. Ezek. 21. 32).
The philological treatment, made known independently by
several scholars like Noetscher and Driver, and accepted by so
distinguished a student of Messianic ideas as Mowinckel,3 relates
the Hebrew word to an Accadian selu or silu ‘ruler’. This sugges-
tion has had considerable influence. Yet, as Moran shows in a
recent comprehensive survey,4 there is considerable doubt
whether this word with this sense exists in Accadian at all,
though terms for kings, princes, and potentates are tolerably
numerous therein.
It is from the lexical tablets that the word was first identified,
and it provides a good illustration of the difficulties involved in the
use of them.
The identification of silu was based on the 12th tablet of the
lexical series Id-A-naqu, which gives thirty-five equivalents for
the Sumerian sign U when pronounced u.5 Among these thirty-
five are si-lum and also some words which mean ‘ruler’. These
latter (e.g. sarrum ‘king’ and belum ‘lord’) are separated from
si-lum by i-lum ‘god’, and the list contains such diverse items as
‘totality’, ‘ear’ (uznu)y ‘hand’ and ‘earth’. It was from this list that
1 See for instance Rankin, Jewish Religious Polemic, index, s.v. Shiloh.
2 Cf. now also the Qumran text 4Q Patriarchal Blessings.
3 He That Cometh, p. 13 n.
4 Biblica xxxix (1958) 405-25.
5 The list is published in Cuneiform Textsy pt. X II (London, 1901).
P R O B L E M S O F T H E L E X I C O G R A P H I C A L T R A D I T I O N 121
Zimmern first identified the silu or selu which was then applied
to Gen. 49. 10. In this identification he seems to have been unduly
influenced by the context, and especially by the equivalences
which follow the word rather than those which precede it. In fact
the sense of silu is now well known: it means ‘hole’, ‘rent’, or
‘fissure’.
There is also a vocabulary list in which for the Sumerian nam.
lugal ‘kingship’ the Accadian equivalent is given as sa-lu-tum.
Landsberger, in editing these materials, himself declares this to
be an erroneous writing, and corrects the text to the familiar
sarrutum, which is of course the normal Accadian word for
‘kingship’.1
Such, then, are some of the problems which may be met in any
attempt to use the older lexicographical writings as a means to
appropriate the material of the cognate languages.

(6) ‘Aramaisms’ and Similar Terms

We may end this chapter with an analysis of such terms as
‘Aramaism’ or ‘Arabism’, which we meet fairly often in the litera-
ture of philological treatments.2 These terms are often lacking in
clarity. The following possibilities may be intended:
(1) The term may mean only a statistical displacement towards
what is more frequent in Aramaic but more infrequent in Hebrew.
It is common knowledge that ‫ א ת א‬is the normal Aramaic word for
‘come’. The same word occurs, however, in Hebrew, and there
are about twenty cases in the Old Testament. Most of these, if
not all, are in poetry. Suppose, then, that we find a text where ‫א ת ה‬
becomes the more frequent word for ‘come’, or where it occurs
outside of the poetical context; then the situation in this regard is
more like that which exists in Aramaic, and someone may say that
this is an ‘Aramaism’. But there is no question that the phenomenon
itself is real Hebrew; the only difference is in the distribution and
frequency. All discussions of ‘Aramaisms’ in late books of the
Old Testament like the Song of Songs should try to distinguish

1 See Materials for the Sumerian Lexicon iv (Rome, 1956), p. 32, 1. 55. For
some discussion of silu see von Soden in Orientalia N .S . xvi (1947) 81-3, xviii
(1949) 402; Goetze inyC.S‫ ׳‬i (1947) 256, n. 19. Goetze attaches the word to the
Arabic verb sala*a *split, cleave’.
2 For a helpful survey see Driver, *Hebrew Poetic Diction’, V T S i (1953)
26-39; earlier, Kautzsch, Aramaismen. A recent full survey is by M. Wagner.
122 Q U E S T IO N S IN P H IL O L O G IC A L T R E A T M E N T S
between phenomena which are not normal Hebrew at all and
phenomena which are only statistically unusual.
(2) Se condly, usage may be identified by means of an appeal to
Aramaic, where this usage has not previously been recognized as
existing in Hebrew though it is well known in Aramaic. The
identification is intended as an identification of a normal, if un­
common, usage in Hebrew. Many cases which probably fall into
this category can be found in the literature. Thus at Isa. 57. 13,

Driver says that the word ‫ ק ב ו ץ‬is here an ‘Aramaism’, used in

the same sense as the Syriac qbaa ‘fixing’ and then ‘statue’ (cf.
qba ‘fix, found’; qbi'ta ‘fixation, form, sign’). The sense would
then be:
‘Will your statues save you when you call?’
Driver surely means that the usage is a native Hebrew one,
inherited from earlier Semitic; and when he calls it an Aramaism,
this means that the sense was discovered from Aramaic. If the word
had been actually taken over from Aramaic, or used as a deliberate
imitation of Aramaic usage, it would surely have retained the ‫ע‬
of the /uramaic form.
Sometimes it is not so clear what was meant. At II Kings 3. 27,
where the text reads:
‫ ?ן ? ף ־ ג ד ו ל ע ל ־ י ש ר א ל‬.‫וי הי‬
Driver feels it hard to explain this as the wrath of a Moabite god.
The noun ‫ ק צ ף‬therefore may be supposed to have a ‘weakened
Aramaizing’ sense, comparable to the Syriac qsap which means
*be angry’ but also ‘be sad’; cf. also qsipa ‘sad’, and Mishnaic
‫‘ ק צ ו פ ה‬sorrow’. A Greek variant, he notes, reads Xv7nq here.
‘Aramaizing’ here could mean that the biblical writer, aware of
the shades available in the Aramaic word, applies the same to the
Hebrew word. More probably, however, it means not an imitation
of Aramaic usage by the Hebrew writer, but the use of normal
Aramaic usage as evidence to uncover occasional Hebrew usage.
(3) Thirdly, a term like ‘Aramaism’ may mean that an expres-
sion of Aramaic type was deliberately used, or that, if not deliber-
ately, at least in fact, the existence of an Aramaic phenomenon is
actually affecting the choice and the character of Old Testament
 ‘A R A M A IS M S ’ A N D S IM I L A R T E R M S 123
usage. It is sometimes plausible that contact with Aramaic speakers
may explain unusual locutions by northern Israelite speakers like
Hosea. Contact with Accadian is also a possibility; it is somewhat
less likely than Aramaic influence, because Accadian must have
been to the average Israelite a much more strange and difficult
language than Aramaic.
A possible case for such an Aramaism would be the ‫ מ ל ך י ל ב‬of
Hos. 5. 13. The ‘King Jareb’ of AV has long been discredited,
since no such name appears to exist. A generation of students has
been taught that this text should be read as ‫ מ ל כ י ל ב‬, meaning
‘the great king’. The termination /‫־‬i/ is interpreted as ‘an old
nominal termination’.1
The case for an accidental wrong word-division is not strength-
ened by the fact that exactly the same phrase appears again at
Hos. 10. 6. This leads to the search for another explanation.
Driver argues that the text is quite right, and that the word is an
adjective identical with the Syriacyireb ‘be great’, cf.yaributa ‘size’.
That the M T is right is shown by LXX Iape^ and Pesh. mlk'
dyrb. Thus, in Driver’s opinion, the word is an ‘Aramaizing form’
from the root ‫ ל ב ה‬or ‫ ל ב ב‬, ‘and is nothing but sarru rabu [the
Accadian for “the great king”] in Aramaic guise.’2
(4) A term like ‘Aramaism’ may be used when scholars hold
that a text was originally written in one language and then trans-
lated into another, and that characteristics of the diction of the
former state have been carried over into the latter. This has been
argued particularly for the book of Job. Tur-Sinai in a lengthy
commentary has maintained that Job was written in Aramaic and
later translated (rather badly, it would seem) into Hebrew.3
Similar problems arise with those theories of other books, such
as Qoheleth, which attribute their linguistic peculiarities to
translation from another language.
These problems will receive no further attention in this book.
They depend on special theses which have to be specially proved
for the books concerned and which, whether valid or not for those

1 So, somewhat obscurely, BDB, p. 937a, who further add confusion by citing
the word under the root ‫‘ רי ב‬contend’, although they do not themselves believe
this to be the right meaning.
2 Nyberg, Hoseabuch, p. 38 f., also wants to retain the text; he thinks of an
Assyrian deity, referred to in the text as ‘the Melek Yareb’.
3 Cf. also the quotation from Guillaume, above, p. 34.
124 Q U E S T IO N S IN P H I L O L O G I C A L T R E A T M E N T S
books, cannot be extended therefrom over other books which are
not the product of the same translation process. Some particular
examples which will be cited in this book do, nevertheless, come
from works which contain these special theses.
To sum up, then, the above four categories classify the ways in
which terms like ‘Aramaism’ seem to be used.
 VI

THE DISTRIBUTION OF HOMONYMS

(1) General
P h i l o l o g i c a l treatments tend to increase the number of
homonyms known to exist in Hebrew. The method frequently
implies that a familiar word had alongside it another word of
identical or almost identical form. The rarer homonym then be-
came concealed or forgotten through mistaken identification with
the more familiar, or may otherwise have fallen into disuse. Philo-
logical research is able, from its resources in the cognate languages,
to identify a homonym thus lost; and since treatments do not
generally identify words which previously were formally com-
pletely unrecognized (to do so would usually mean departure
from the consonantal text) a substantial proportion of new
identifications are homonymic with a word already known.
Nevertheless the cumulative implications of this multiplication of
homonyms have seldom been noticed.
A good example is ‫ ע י ר‬, familiar in the sense ‘city’. A homonymic
‫‘ עי ר‬excitement’ (root ‫‘ ע ו ר‬arouse’) has been identified for one or
two places, principally Jer. 15. 8 ‫‘ ע י ר ו ב א ל ו ת‬excitement and
terror’, by BDB and others.1 Driver, however, here finds another
word, ‫ ע י ר‬meaning ‘invasion’, cognate with Arabicgara. At Deut.
34. 3 (customarily ‘city of palm-trees’) and some other places he
identifies an ‫ עי ר‬which is not ‘city’ but ‘small depression’. At Job
3. 8 he finds yet another word, the root of which is ‫ ע י ר‬, meaning
‘revile’, in the phrase ‫‘ ה ע תי ד י ם ע ר ר לוי תן‬those who are ready
to revile Leviathan’. This involves taking the parallel phrase as
‫( א ך ר י ; ם‬m t ‫) א ך ך י יו ם‬, meaning ‘those who curse the Sea (i.e.
sea-monster)’. This verb is cognate with Ethiopic taayydra ‘re-
vile’, Arabic gara II ‘revile’. This sense ‘reviling’ was also applied
by Guillaume to the phrase at Hos. 11. 9: ‫ ו ל א א ב ו א ב ע י ר‬,
which was put by BDB under the word for ‘excitement’ (though
with some doubts about the text).
1 BDB, p. 735b. The word ‫* עי ר‬city’ is registered on p. 746.
126 T H E D IS T R IB U T IO N OF HO M ON Y M S
Nyberg meanwhile had seen at this same place an ancient word
‫ עי ר‬for ‘fire’ or ‘heat’. Gray discovers another ‫‘ עי ר‬inmost recess
(of temple)’ at II Kings 10. 25, and Dahood has an ‫‘ ע ך י ם‬gods’
which at least in the plural is homonymous. To all this we may
add a verb ‫ )?( ע י ר‬meaning ‘bore’, found only at Isa. 50. 4, and
cognate with the Arabic gara ‘sink in the earth’, gar ‘pit’, which
gives to the phrase ] ‫ י עי ר לי אז‬the sense ‘he boreth me an ear’.
The reader, unless he is experienced in the literature of philo-
logical treatments, will probably have found this account some-
what bewildering. Yet the situation I have described is far from
an uncommon one. There are several homonyms if all, or even if
most, of the scholars concerned have been right in their identifica-
tions. Something can, indeed, be done to sort out this list of real
homonyms, near-homonyms, and apparent homonyms, and we
shall shortly offer some distinctions which will help in classifica-
tion. But one could continue indefinitely with the mere listing of
new homonyms in Hebrew identified in the course of philological
treatments.
What I have not found is any perception that the production by
scholars of such large and increasing numbers of homonyms con-
stituted any kind of problem or difficulty. The identification of
roots and the organizing of the vocabulary with adequate distinc-
tion of meanings has long involved the discrimination of homony-
mic roots, and this has been a creative and salutary part of the
progress of lexical studies. One does not therefore see at once how
a process which has solved problems for so long may now begin to
create fresh problems.
Several aspects of homonymy will now be discussed in general,
in order to establish some preparatory approaches to the matter.
The following distinctions provide a preliminary analysis of
various kinds of homonyms relevant to our subject:1
(1) Some homonyms are products of phoneme mergers
traceable through reference to other Semitic languages, and
when this is so it may be possible to separate them clearly.
1 Homonymy was known to Arab and Jewish medieval scholarship; the usual
Arabic term was muitarik and the Hebrew ‫ שם מ שו תף‬. Distinctions of types of
homonymy were drawn up, but are more logical than linguistic in character and
do not correspond with our present series. Homonyms for our purpose mainly
fall within the class which Maimonides called ‘absolute homonyms*, see Efros,
Maimonides’ Treatise on Logic, p. 59. For the Arabic term see Ibn Khaldun,
Muqaddimah (ed. Rosenthal, 1958) iii. 26.
 GENERAL 1 2 7

(2) Some homonyms are complete, in that all items in the

paradigm of a word are identical with those of its homonym;
others are partial, the coincidence extending over only
certain forms within the paradigm.
(3) Accordingly, roots may be identical without the production
of homonyms in actual forms, and the difference between
identity of roots and homonymy of actual forms must
always be kept in mind.
(4) Accordingly, problems with homonymy are likely to be
particularly noticeable in the case of verbs.
We now turn to consideration of these points in greater detail.
(1) It is useful to make a distinction between two classes of
homonyms, according to the relation they have to phonological
changes perceptible between one Semitic language and another.
If proto-Semitic has two phonemes, x and y, and if these merge to
form x in Hebrew, then words of the patterns xab and yab will
probably become homonymous. If there is another language
in which this merger does not take place, then the homonyms in
Hebrew may be distinguishable through comparison with that
language. The two phonemes which in Arabic are /'/ and /g/, and
which correspond to a phonemic distinction in proto-Semitic
also,1 find their representation in Hebrew in the one phoneme /'/,
i.e. ‫ ע‬. The verbs ‫‘ ע נ ה‬sing’ and ‫‘ ע נ ה‬answer‫ ״‬form such a pair of
homonyms. Among the Hebrew consonants, the phonemes which
can normally be so analysed, stated as set against Arabic, are six in
number: T(/d/and /z/); ‫( ח‬/h/ and /h/); initial ‫( י‬/w/ and /y/); ‫( ע‬/‫׳‬/
and /g/); 7) ‫ צ‬s/, /d/ and /$/); and 8f(/t/ and /s/).
Where a word has more than one of its consonants belonging to
groups which have converged in this way, multiple possibilities
of homonymy may theoretically emerge. Thus Hebrew, as BDB
sets it out, has ‫‘ ח צ ר‬enclosure, court‫( ״‬cf. Arab, hasara ‘encom-
pass, surround ‫)״‬, ‫‘ ח צ ר‬village, settlement‫( ״‬cf. hadara ‘be present,
settle, dwell’), and ‫‘ ח צי ר‬herbage‫( ״‬cf. hadira ‘be green’). Seldom,
however, are there actual occurrences of homonymy in the same
number as the merger of phonemes from proto-Semitic would
theoretically allow.
In cases of this kind, other things being equal, it will commonly
1 The view that /£/ was proto-Semitic has been questioned by RuziSka, but
his view has not been generally accepted. In any case it makes no difference to
my argument; I could equally well use the illustration of /ft/ and /&/.
128 T H E D IS T R IB U T IO N OF H O M ONYM S
be possible to distinguish homonyms quite clearly by pointing to
cognates possessing phonological differences which are no longer
operative in Hebrew. Such an identification makes quite clear
that an actual homonymy has occurred and precludes the pos-
sibility that the multiplicity of meaning should be explained in
some other way, e.g. by derivation of the sense of one word from
the sense of the other. For instance, I have heard ‫‘ ע נ ה‬sing*
derived from the phrase ‫ ענ ה ב שי ר‬as an instance of sense-deriva-
tion by ellipsis (from the sense ‘answer in song’) ; but the Arabic
cognates eliminate this possibility. Thus homonyms which can be
identified by reference to different words in another language,
such as Arabic, in which the phoneme mergers found in Hebrew
have not taken place, form peculiarly certain examples of horn-
onymy in comparison with other types.
Not all homonyms which have been detected, however, can be
dealt with in this way. In fact, only a minority of the Hebrew
consonantal phonemes permit an analysis by comparison with
multiple phonemes in a cognate language. In languages like
Arabic, in which we know of relatively few mergers of phonemes
from proto-Semitic,1 correspondingly few of the homonyms
existing can be classified or accounted for in this way.
The historical question of the date when the various mergers
took place may be mentioned briefly. It is conceivable that some
Hebrew texts originated at a time when the relevant merger was
not yet complete, so that (for instance) /'/ and /g/ were still distinct
at the time of the original production, in spite of the identity of
letters in the text now extant. At the difficult place Jer. 12. 9,
m t ‫ה‬T ‫ל י‬V ‫ע‬T ‫ס ·ב י ב‬T ‫עי ט‬ ‫ל ·תי ·ל י ה‬T ‫ ־ח‬: ‫ע ־נ‬- ‫צ ב ו‬T ‫ה עי ט‬
---------

Driver identified the first word ‫ עי ט‬as a word meaning ‘lair, den’,
cognate with Arabic gata, and thus different from the more
familiar ‫‘ עי ט‬bird of prey’, which follows. If this is right, the sense
is roughly:
‘Is my heritage a hyena’s den, with birds of prey around it over-
head?’2
1 The obvious case in Arabic is /s/. One may contrast s-k-r ‘be drunk*
( = Hebr. ‫ )ש‬and s-k-r ‘close* ( = Hebr. 0 in 0‫ ;)ג ר‬likewise in the verbs samara,
samuray one would distinguish the sense ‘converse by night, stay awake* ( =
Hebr. ‫ ש‬in ‫ ש מ ד‬, if BDB, p. 1036a, is right; otherwise GB, p. 847b, KB, p. 993a)
and the sense ‘nail’ ( = Hebr. 0 as in ‫) ס מ ר‬.
2 This solution involves also the identification of ‫ צ ב ו ע‬as ‘hyena*; cf. below
p. 235. For gata (y) intravit et latuit cf. Freytag, iii. 306b.
 GENERAL 129
—rather than the more familiar (but not very clear)
‘Is mine heritage unto me as a speckled bird of prey? Are the birds
of prey around it?’ (RSV).
If this identification at Jer. 12. 9 were right, and if /'/ and /g/ had
merged into one phoneme in Hebrew before Jeremiah’s time, then
the two words were homonyms, and it is possible that play on the
two homonyms may have been part of the writer’s intention.1 If,
on the other hand, the merger had not yet become complete, there
would be no homonymy and the words would be distinct in sound as
well as in meaning; the writer might then still be supposed to have
played on assonance, but not on homonymy. It was long thought
that in the biblical period /'/ and /g/ were still distinct and that
Greek transcriptions like .Ta£a and JTo/xoppa proved this even for
quite late times; but, since the incidence of these Greek spellings
appears not to coincide with the difference between the consonants
in the probable etymologies of the Hebrew words, this opinion is
no longer generally held.2 In my judgement, the mergers listed
above (p. 127) were probably already complete in the later
biblical period, if not earlier. We may only mention also that, if ‫ש‬
were pronounced generally like D (cf. Judges 12. 6), this would
produce certain homonymies, while if ‫ ש‬were pronounced like ‫ש‬
it would produce certain others. In general, however, it does not
appear that study of the date of the mergers will make a very
great difference to our estimate of homonymy in general, though
it may pose intriguing problems in particular cases.
To summarize, then: it is useful to distinguish, among the total
group of homonyms, between those which can be related to
phonological mergers and those which, within the limits of our
present knowledge, cannot be so related.
(2) Secondly, it will be useful to consider a distinction between
complete and partial homonymy. Where homonymy is complete,
all items in the paradigm of a word will coincide with the corre-
sponding parts of its homonym. Where homonymy is partial, it
will be found only in part of the range of the paradigm. There is a
rare construct plural ‫ ש ב ע ת‬which is found from both ‫ש בו ע‬
‘week’ and ‫‘ ש בו ע ה‬oath’ and is thus in a way a homonym; but

1 Cf. below, pp. 151-4.

2 See recently Moscati, Comparative Grammar, p. 40; earlier, Ruzi£ka in Z A
xxi (1908) 293-340 and elsewhere.
130 T H E D IS T R IB U T IO N O F H O M O N Y M S
there is no homonymy in other parts of the words, and the statis-
tical degree of overlap is extremely low. Similarly, the words ‫ב ק ר‬
‘morning’ and ‫‘ ב ק ל‬cattle’ would have partial homonymy in all
probability in the plural; but it is doubtful if overlapping cases
are textually sound; and since both words occur with an over-
whelming predominance in the singular, the degree of overlap, if
any at all, is extremely low.
Partial homonymy, though it is a source of trouble to language
learners,1 is not a very serious obstacle to the use of a language in
communication. For the native speaker in the vast majority of
examples the combination of grammatical and semantic con-
textual indications selects the right sense infallibly. The situation
is not particularly different from that where two different items in
the paradigm of the same word coincide, though in other words
of the same class they generally differ. Thus in Hebrew ‫ ל ע ל ה‬. may
be ‘he goes up’ (qal) or ‘he puts up’ (hiphil), though these are
normally marked as distinct. Paradigms do not always provide
equal marking for every distinction which has a place within the
paradigm.
We may argue then that, while partial or occasional homonymy
or overlap of forms is something with which the language system
may be expected to deal in its stride, complete homonymy, where
two words will overlap in all their forms, may be a more serious
matter.
(3) Thus, thirdly, it is important to see that the problems re-
lating to homonymy depend primarily not on the overlap of roots
but on the overlap of forms.
Under the sequence ‫ ע ל ם‬BDB classifies three different roots.
But, though the radical sequence is common to all three, there are
extremely few cases in which any homonymic forms occur. One
of the three roots, for instance, appears only in the noun ‫; ע ו ל ם‬
but this form does not appear in the other two roots. In fact, of
all the forms registered for the three roots, allowing some leeway
for textual uncertainties and so on, I think there is only one
homonymous form, the qal passive participle which occurs once
from the root meaning ‘conceal’ at Ps. 90. 8 and coincides exactly
1 There are old books of homonyms for learners of Latin, which are designed
to enable :hem to see synoptically the various possibilities of forms like facies
(either 2 s ing. fut. of facio or nom. sing, of the noun facies); many hundreds of
such instances can easily be found.
 GENERAL 131
with the noun meaning ‘youth’ which occurs several times. Thus,
out of quite a large number of forms registered under the three
roots, only very few fully overlapping homonymic forms, and pos-
sibly in fact none at all, actually occur. It is possible, of course,
to register ‫‘ ע ו ל ם‬eternity’ under the root ‫ ל‬1‫ ע‬, but this does not
materially alter the situation.
The problem of understanding how homonyms functioned as
discriminatory communicative signals depends on sound rather
than on writing, and depends on the whole sound of the word
concerned and not upon the abstraction we call the ‘root’.1 There
may be two roots 0 ‫ ל ל‬, but since the forms from one of them (‫ס ל‬
and ‫ ס ל ס ל ה‬, both meaning ‘basket’) do not occur with the other
(a verb meaning ‘lift up, cast up’), there is no factual homonymy
in the extant texts. This is particularly evident when a root is not
used as a verb and is recorded in the dictionaries only in order to
state a root for particular noun formations.
(4) It should now be obvious that the phenomenon of homony-
my will, under the conditions of Semitic languages, be especially
noticeable in verbs. The noun ‫ ד ב י ר‬is not homonymous with the
noun ‫ זי ב ר‬or the noun ‫ די ב ר‬, whether they come from the same
root or not. ‫‘ ע ל ב‬raven’ is never homonymous with ‫‘ ע ל ב‬even-
ing’, though the root sequences are identical. Nouns have particular
patterns of formation which, not in all cases but in many, will
distinguish them from other nouns of a different pattern, whether
the root sequence is the same or not.
In verbs, however, this is not so to anything like the same extent.
Semitic does not have the verb-classes or ‘conjugations’ which are
so prominent in Indo-European languages. In Latin, for example,
we can cite two verbs with stem ven-; but venor ‘hunt’ will not be
confusable with veneo ‘be for sale’. The stem fer- can be found in
two verbs which will nevertheless usually be distinct: ferio ‘strike’
and fero ‘carry’. In Avestan we can list two verb stems kar-. One
means ‘make’ and is cognate with Skt. krnoti ‘build’; its present is
car- (class I). The other means ‘enter’ and is cognate with Skt.
1 Similarly in Greek there is no homonymy between pairs like apa and apa ,
καλώς and κάλως, since the pitch accent is distinctive; the difference had,
however, no written marking until late times. Strictly, a more correct term for
our study would be homophone; two words alike in writing but not in sound
would be homographs. I have ventured, however, to retain the more traditional
term homonym.
132 T H E D IS T R IB U T IO N OF H O M O N Y M S
carati, Greek 7tcAo/jmu, Latin colo\ its present is cara- (class IV).1
Distinctions of this kind are not normally available in Semitic;
identity of radical sequence in verbs will produce homonymy.
In Hebrew it is necessary for the functioning of a verb that it
should fall within a number of formation patterns; if it does not,
it just cannot mean anything. A form like */qitll/ could not function
as a verb. In this respect nouns are different. A noun may be taken
over from another language and offer (as in ]!‫ א פ ר י‬or ‫ ש ע טנז‬and
many post-biblical words of Greek origin) a pattern quite unex-
ampled in Hebrew, but this does not prevent its functioning. A verb,
on the other hand, functions only through its having one of the
patterns such as /qatal/ which are used for verbs. If therefore two
verbs are homonymous in their sequence of root consonants, they
will necessarily, through the nature of the verb system, be honaony-
mous in their entirety, with only the following qualification.
There are in Semitic the themes or binyanim, that is, the types
like qal 2nd piel. Sometimes a distinction in use of theme prevents
verbs of identical root from being homonymous. In Hebrew there
are the two ‘homonymic’ verbs ‫‘ ח ל ה‬be weak, sick’ and ‫‘ ח ל ה‬ap-
pease’. But the latter is found only in the piel, while in the former
the piel is infrequent and other themes are frequent. Though the
radicals are the same, only limited overlaps in usage occur, and
homonymy is only partial. The analysis of ‫ ע ל ב‬in BDB sets out
six roots with this sequence; but of these only three have verbs
extant in biblical Hebrew, and of these three only one is found in
the hithpael and one in the hiphil. All three, however, are registered
for the qal. Again, ‫‘ ז מ ר‬make music’ occurs only as piel, while ‫ז מ ר‬
‘trim’ occurs onlyasqalandniphal. Similarly,‫ ה‬1‫‘ ל‬join, accompany’
occurs in qal and niphal, ‫ ה‬1‫‘ ל‬borrow, lend’ in qal and hiphil.
Here are some other examples: ‫— ב ר א‬I ‘create’, qal, niphal only;
II ‘be fat’, hiphil only; III ‘cut down (trees)’, piel only; ‫— ח ל ל‬I
‘bore’, qal, piel, pual, poel, poal; II ‘play the pipe’(denom.), qal, piel,
polel; III ‘pollute, profane’, niphal, piel, pual, hiphil; IV ‘begin’,
hiphil, hophal only; ‫— ע נ ה‬I ‘answer’, qal, niphal, (hiphil once, but
doubtful); II ‘be occupied’, qal only; III ‘be afflicted’, qal,
niphal, piel, pual, hiphil, hithpael; IV ‘sing’, qal, piel, only.
These statements have to be made with caution when verbs are
rare in any case, so that we cannot say with certainty that particular
1 Quoted from Reichelt, Awestisches Elementarbuch, pp. 434 f.
 GENERAL 133
themes were not used. Nevertheless there is sufficient ground to be
confident that a certain number of verbs, the roots of which were
alike, were in practice partially discriminated because only
limited themes were used and these differed between one verb and
another. In general, however, it remains true that the conflict of
homonyms may have been rather more prominent in the verb.
Certain wider ramifications are implied by points (3) and (4).
Philological treatments have generally emphasized the search
for a cognate root rather than the word formations in which the
forms are found. Many treatments give the impression that, so
long as a root sequence suitable for the Hebrew consonants can be
established, and so long as this sequence can provide an adequate
meaning, the existence of the corresponding formation in Hebrew
can be merely assumed. We shall later discuss the suggestion that
Hebrew had a ‫‘ ל א ם‬ruler’, cognate with Accadian llmu, limmu
‘magistrate, eponym’. Hebrew ‫ ל א ם‬, however, seems to belong to
the formations qutul or qutl, while the Accadian belongs to qitil or
qitl. This fact has to be given due recognition in any argument which
seeks to derive the sense of one word from the sense of the other.
This requirement is avoided, of course, if it is roundly declared
that the vocalization in Hebrew is wrong and should be emended,
e.g. to a form like *‫ ל א מ י ם‬. This drastic step, however, carries
with it other consequences which we shall have to consider. For
the present we observe simply that the low valuation placed upon
the vocalization in many philological treatments agrees with the
comparative neglect of the word formations.
The semantic importance of word formation applies to all uses of
comparative philological information, and not only to cases where
the Hebrew text is itself in difficulty. Hebrew ‫ ע ו ל ם‬and Arabic
calam not only have corresponding roots, but the formation also
is closely similar or identical. If, on the other hand, we compare
the famous word ‫‘ ע ל מ ה‬young woman’ with the Aramaic ‫ ל י מ א‬1‫ע‬
or ‫ ע ו ל י מ ת א‬, though the root is the same, the formation is a differ-
ent one; the Aramaic had the pattern qutail (with or without
feminine ending). Any drawing of semantic conclusions from one
word to the other has to take account of the different formation.1
1 This is one reason why the meaning of Hebrew ‫ ע ל ^ ה‬cannot be quite simply
and directly established through citation of forms like the Palmyrene ‫עו לי מ ת א‬
‘hetaira’; cf. the use of this argument (among others) by Driver in T L S, 19 May
1961, p. 309, with reference to the translation of Isa. 7. 14.
134 T H E D IS T R IB U T IO N O F H O M O N Y M S
These then are some distinctions which may help us in studying
the possible interactions of homonyms in Hebrew.
(2) Homonyms and Communication
I have already mentioned the atomistic nature of philological
treatmen ts. The scholar is seeking a solution to one detailed verse
or phrase. The solution may involve the identification of a new
homonym in Hebrew. In producing this solution the scholar may
not notice that the cumulative effect of his work, especially if the
work of e ther scholars working on similar lines is added, produces
a new question. It has long been common knowledge that
homonyms exist in Hebrew. If we have two words formally
identical, why not four or six? Under the circumstances, the
question whether homonyms can be infinitely increased in number
is not usually asked. In this the scholar is looking at things from a
somewhat self-centred point of view. He is thinking of how he, the
modern scholar, can identify this word in relation to other
languages. He is not thinking about the other question, namely,
how did the Israelites under these circumstances know what they
were talking about? The production of new homonyms raises a
question about the communicative efficiency of Hebrew.
So far as I have found, the producers of philological treatments
have taken notice of this question only in very isolated cases.
Tur-Sinai notices it, for instance, when he proposes the suggestion
that the word ‫ ל ב‬should in certain places be understood not as
‘great, numerous’ but as another word meaning ‘weak, powerless,
afraid’.1
job 4. 3 : ‫י ס ך ת ר ב י ם ד ד י ם ר פ ו ת ה ח ז ק‬.‫ה ב ה‬
can then read:
‘If thou hast supported the powerless
and strengthened the weak hands’
—which, of course, gives a good parallelism. Other instances
suggested by Tur-Sinai are Job 4. 14, 26. 3, 35. 9. He goes on
to remark that:
The pronunciation of this word was apparently different from that
of ‫ ל ב‬in the ordinary sense; otherwise, it would not have been possible
effectively to contrast ‫‘ ל ב‬numerous, great’ with ‫‘ אי] כ ל‬powerless’
(II Chron. 14. 10).
1 Tur-Sinai, Job, p. 76 f.
 H O M O N Y M S A N D C O M M U N IC A T IO N 135
The text at II Chron. 14. 10 reads:
: ‫ף ל?זר בין רב לאין כח‬$‫איךע‬
Whatever we may think of Tur-Sinai’s solution, it is of real
interest that he has noticed the problem of reduction of com-
municative efficiency caused by homonymy, and has adjusted his
solution to it by the virtual addition of a qualification making clear
that in the original situation there cannot, in his judgement, have
been a homonymy.
In very general terms, this question of the communicative
efficiency of Hebrew is the question which we now have to discuss.
In putting it in this way, we do not suggest that homonymy in
itself constitutes any insuperable obstacle to communication.
That homonyms exist in many languages is very well known.
We do have to consider, however, such questions as these: can
we say something more definite about the conditions in which
homonymy exists in Hebrew? Are these conditions such that the
addition of still more homonyms through philological investigation
is favoured or disfavoured? Can we see in homonymy a factor
which worked for linguistic change in Hebrew ? What effect does
the recognition of homonymy have upon our judgements about
Hebrew style ? When a high incidence of homonymy is established
in Hebrew through comparative studies in the cognate languages,
how far does this situation in Hebrew agree with the situation in
each or all of the cognate languages considered for themselves?
What effect does this in turn have upon the making of comparisons
between Semitic languages ?
One or two more distinctions may help us to discriminate be-
tween ways in which homonymy affects communication.
Firstly, we may distinguish between the homonyms of like gram-
matical function and those of unlike grammatical function. This
can be easily illustrated from English, in which there is a high
number of homonyms but the number is greatly reduced if we
remove those which are of unlike function.1 Thus if we take see
and seay one is a verb and the other a noun. Week is a noun, but
weak is an adjective. In these circumstances the context, or in
1 This is already noticed by Menner, Language xii (1936) 231. Arguing that
homonyms are not likely to interfere unless they are the same part of speech, he
cites Old English earm *poor’ and earm ‘arm’. Similarly, while cleave ‘sunder’ is
usually transitive, cleave ‘adhere’ is intransitive and always has a preposition;
these verbs, now homonymous, were not so in Old English.
136 T H E D IS T R IB U T IO N O F H O M O N Y M S
other terms the slot occupied by the unit, does a very great deal to
select foi us between the possible homonyms and exclude as
irrelevant what would be significant if the word occupied a
different slot. In French, something similar can be said for ou> où
and août. There are, of course, many cases where this simple
distinction does not apply: pair and pear are both nouns; so are
piece and peace. Nevertheless the distinction serves to overcome a
great deal of the ambiguity which one would expect to arise from
homonymy. In Hebrew, of course, patterning and inflexion are
such that a verb will relatively seldom be homonymous with a
noun. The relevance of the point is that the high incidence of
homonyms in English is much reduced in its effect upon com-
munication through this difference of grammatical function, and
therefore cannot be used in a simple fashion as an argument for
the viability of large numbers of homonyms where the circum-
stances are different.
Here, however, we can add another discriminating factor,
namely the more general context, which may establish a semantic
field within which a word will fit. We know that in the sentence
They're a happy . . . the next word is likely to be pair rather than
pear. General context will similarly form a good guidance in most
cases for the choice between piece and peace.
Now it would be of general interest to know how far the in-
cidence of homonyms (or alleged homonyms) in Hebrew is
correlated with the existence of general contexts which, by defining
a semantic field, serve to select the correct word from among
several homonyms in the lexical stock. We may illustrate this
first from the group of three homonymie verbs ‫ צ ל ל‬, which occur
in precisely identical forms.
In the first instance, at Exod. 15. 10:
; ‫ א ד·י־ר י ם‬-‫נ •מ י ·ם‬: ‫ר ת‬V ‫פ‬V ‫־ל־ ל ו ־כ ע ו‬: T‫ם צ‬T‫ס מ ו י‬T ‫·כ‬
the first phrase ‘the sea covered them’ has already done much to
define the semantic field; and, after the verb ‫ צ ל ל ו‬has come, the
unambiguous words ‘like lead in the great waters’ form very
substantial guidance to select the meaning ‘sink’ rather than that
of the homonyms meaning ‘tingle’ and ‘be dark’.
For the second verb ‫ צ ל ל‬, in every one of the four cases known the
semantic field is defined by the presence of a subject which is an
organ of speech or hearing (ears or lips). These subjects follow the
 H O M O N Y M S A N D C O M M U N IC A T IO N 137
verb, but in each context the semantic field of hearing has already
been established, before the utterance of the verb, by other phrases
or words.1
The third *]‫ צ ל ל‬occurs at Neh. 13. 19, in a phrase apparently
meaning ‘and it happened, as the gates of Jerusalem became
shadowy at the beginning of Sabbath’. I do not see that the seman-
tic field has been so clearly defined by the context as in the other
two. The reference to the coming of the Sabbath might furnish
some indication. At Ezek. 31.3, where the subject is an enormous
tree, the definition is better. In any case, even if the third ‫ צ ל ל‬was
used without much clear definition of the semantic field, the
presence of clearer definition for the *other two may have been
sufficient to avoid much ambiguity.
Sometimes the presence of homonymy, along with usage of
both homonyms in closely related semantic fields, may lead to the
coalescence of the two senses. Orlinsky suggests that there were
two homonyms ‫ כ ב‬1‫ ע‬. The former has the familiar sense ‘lie’; the
other means ‘pour out’, like Arabic sakaba.2 The Hebrew phrase
‫ ש כ ב ת ־ ז ר ע‬then means ‘outpouring of seed’. The original sense
was forgotten, because (a) the sense ‘lie’ was normal for the verb,
while (b) the semantic field was close to that of ‫ ש כ ב‬as ‘lie with’.
Indeed both words occur together in Num. 5. 13:
‫ ב ר ע‬-‫ו ש כ ב אי ש א ת ה ש קנ ב ת‬
This being established, Orlinsky goes on to expound Job 38. 37:
‫ת ב ל י ש מי ם מי י ש כי ב‬
as:
‘Who can pour out the bottles of heaven?’
Such a treatment, whether in the end it is right or not, shows an
intelligent judgement of the possible effects of homonymy, gives a
relation to a well-established Arabic sense, and perhaps makes
more sense of Hebrew ‫ ש כ ב ה‬than a derivation from ‫‘ ש כ ב‬lie’.
It is a quite different matter if homonyms not only lie within the
same semantic field but mean something quite different, or even
directly opposite, within it.
1 The cases are: I Sam. 3. 11; II Kings 21. 12; Jer. 19. 3; Hab. 3. 16.
2 Cf. already GB, pp. 824 f. The phrase ‫ ש כ ב ת ה ט ל‬at Exod. 16. 13, however,
seems to mean a *layer’ more naturally than an ‘outpouring’ of dew; this some-
what weakens Orlinsky’s suggestion. The LXX has καταπανομένης της §ρόσον>
i.e. ‫ כ‬plus ‫שבת‬.
138 T H E D IS T R IB U T IO N O F H O M O N Y M S
A classic case is English let. Alongside the common let meaning
‘permit’ there is a let meaning ‘prevent’. This is a homonym
produced by convergence through sound changes; in Old English
the words are respectively Idetan and l§ttan. Hamlet says: ‘I ’ll
make a ghost of him that lets me’, and AV renders at Rom. i. 13:
‘I purposed to come unto you, (but was let hitherto)’. In the
seventeenth century this use was already becoming archaic. But
the corresponding noun let is still found in special contexts, like
the legal let and hindrance. (Note that there is no noun form let
meaning ‘permission’; though there is a let referring to house
rental.)1
The existence of two homonyms with directly opposite senses
within roughly the same semantic field may have led to the disuse
of one of the homonyms. ‘After it had become homonymous with
let “permit” , this word must have been singularly ineffective’,
says Bloomfield.2 Thus, he goes on, ‘it is likely that homonymy
plays more than an occasional part in the obsolescence of forms’.
If we apply this to the Old Testament, two possibilities emerge.
If the existence of a homonymy was such as grossly to damage the
communicative efficiency of language, possibly we should regard
a proposed new homonym as on that ground unproved. If, on the
other hand, homonymy can be a factor in the obsolescence and
disappearance of words, this may explain why a rare homonym
became unknown, and thereby confirm the identification of one.
A good illustration is the disappearance of words developed from
Latin gallus ‘cock’ in south-western France, because the word
would have become homonymous with the word for ‘cat’ (/gat/).
This explanation was put forward by Gillieron.3
1 Even in the AV the grammatical context usually selects the sense: let (‘pre­
vent’) is absolute, with no following infinitive, and let in let me go could not mean
‘prevent’.
2 Language, pp. 398 f.
3 An exposition is conveniently available, with map, in Bloomfield, Language,
pp. 396 ff. Other sources I have found helpful are Elise Richter, *Uber Homo­
nymie’, in the Kretschmer Festschrift (Vienna, 1926), pp. 167-201; Menner,
op. cit. Another instance is that of queen and quean, which are different words,
though connected by Ablaut, OE cwen ‘princess, queen’ and cwene ‘woman,
servant, harlot’. Menner, p. 232, says that: ‘a survey of the distribution of quean
in English dialects clearly corroborates the view that confusion with queen is the
cause of its disappearance’. He also suggests that semantic interference may not
develop to the point of excluding one homonym from the language but may
result only in a limitation of meaning or a new division of meanings. This may
have happened also in Hebrew.
 HO M ON Y M S AND C O M M U N IC A T IO N 139
The view of Gillieron is still a matter of controversy; but the
discussion remains suggestive for Hebrew. Here are some examples.
The familiar word ‫ ע ז ר‬means 'help’. At Job 30. 13, however,
Driver detected a homonymic ‫ ע ז ר‬meaning ‘hinder’ in the phrase
‫ ל א ע ז ר למל‬. Dillmann, followed by B H 3, had emended to ‫ ע צ ר‬,
meaning ‘restrain, hold back’. This is the conjecture probably
followed by RSV:
‘They break up my path,
they promote my calamity;
no one restrains them/
Driver, however, argued that the conjecture was unnecessary,
for the Arabic '‫־‬z-r ‘reprove, hinder’ shows that the phrase means
‘not hindering them’ in any case, without emendation of the text.
He added that the Accadian ezeru ‘scold’ attests the early diffusion
of this root.
The difficulty is the conflict between two homonyms, identical
in all forms, of which one means ‘help’ and the other means
‘hinder’, so that they lie within approximately the same semantic
field. We cannot argue that the word meaning ‘hinder’ is normal
in Job and remove the difficulty in this way. Job 26. 2 ‫מ ה ע ז ר ת‬
‫ כ ח‬- ‫ ל ל א‬is a clear case of ‘help’, while 29. 12, only one chapter
before our example, has the phrase
‫וי תו ם ו ל א ־ ע ז ר לו‬
‘an orphan with no one to help him’.
This phrase, with its very close similarity to that at 30. 13, makes
it difficult to suppose that the same writer would make such a
change in so short a space.
These considerations in themselves are far from final; but they
do indicate the kind of problems involved in homonymy. These
cannot easily be simplified, for yet another homonym, ‫‘ ע ז ר‬be
valiant’, has been identified by Driver for three passages in
Chronicles and one in Psalms; it is cognate with JJgar. gzr ‘hero,
heroic might’.1 At I Chron. 12. 1 the sense of
‫ו ה מ ה ב ג בו ךי ם ע ז רי ה מ ל ח מ ה‬
1 C M L t p. 142; Ullendorff in J S S vii (1962) 347 adds the comparison with
Eth. *2rt taazra ‘impetum facere’ (Dillmann, col. 1003) and other comparative
information. Dillmann himself, however, by no means accepts this sense. The
application to Ps. 89. 20 is found also in Albright, The Archaeology of Palestine
(1949), p. 233, but with the sense *youth’. Cf. further Index, no. 236.
140 T H E D IS T R IB U T IO N O F H O M O N Y M S
is
‘and these are among the mighty men, heroic in war’.
Meanwhile for Zech. 1. 15 Eitan appealed to yet another
Arabic word, gazura ‘be copious, abundant’. Arguing that the
usual word ‘help’ cannot really be extended to mean ‘help forward,
increase, further’ (cf. RSV here), he argued that the contrast was
between God who was angry a little and the nations who were
copious or abundant in the affliction.
Again, the verb ‫ ע ז ב‬is familiar in the sense ‘leave, abandon’. At
certain ])laces, however, Driver has identified a sense ‘help’,
which is known for the Ethiopic fazzdba (Dillmann, cols. 1003-4).
At Jer. 49. 25 the M T is difficult:
: ‫ ב ה ע י ר ת ה ל ה‬-‫אי ך ל א־ ע ץ‬
Taken li:erally, this would seem to mean:
Oh how the city of praise has not been abandoned!*
Rudolph in B H 3 and in his commentary (p. 252) preferred to
omit the word ‘not’ (which, indeed, is not represented by the
Vulgate), He argued that the particle is an angry gloss by a reader
who thought that the title ‘city of praise’ could not apply to
Damascus, of which the oracle is speaking.
The emission of the negative by the Vulgate is not so very
powerful an argument when all the other versions have it. If we
follow the Ethiopic word ‘help’, Driver argues, we have the good
sense:
‘How helpless is the city of praise’.
He then goes on to find another case in Jer. 49. i o - n , and an
even more striking one in Exod. 23. 5. If one finds one’s enemy’s
ass in trouble, the text says:
: ‫ ־ע ז ב ע מו‬: ‫ע ז ב ־ת‬T ‫ ־ע ז ב לו‬: ·‫ת ׳מ‬T ‫ל‬: ‫ד‬- ‫ח‬T ‫ו‬:
‘and (if) you are reluctant to help him,
(nevertheless) you must certainly help him.*1
RSV had already seen the advantage of a sense like ‘help’ and
actually translated in this way; but since they note only that the
Hebrew was ‘obscure’, they probably did not have a clear idea
how their rendering could be justified.
1 The more traditional approach is to say the sense is ‘let loose, set free’ and
make this a component in the usual sense ‘leave, forsake*.
 H O M ON Y M S AND C O M M U N IC A T IO N 141
At job 9. 27, ‫ א ע ז ב ה פני ו א ב לי ג ה‬, Driver found yet another
homonym, meaning ‘be cheerful, be agreeable’, with its cognate
in the Arabic '-d-b. The sense therefore is T will look cheerful’.
Once again, then, we have in addition to the familiar word
‘leave, abandon’ a homonym which means ‘be cheerful’ and one
which means ‘help’. Of these two, the former presents less of a
problem; it may be that the context of facial expression would
suffice to select the right homonym. The latter, however, is a
homonym which almost exactly contradicts another homonym
within the same semantic field.
In addition to these the standard dictionaries themselves
recognize a homonym ‫ ע ז ב‬at Neh. 3. 8:
: ‫! י ע ז ב ו י רו ש ל ם ע ד ה חו מ ה ה ר ח ב ה‬
and this is usually taken as ‘restored’ or ‘paved’; it is the only
occurrence. Once again, then, we have a total of perhaps four
verbs ‫ ע ז ב‬, all homonymic.
In the case of the ‫‘ ע ז ב‬help’ suggested by Driver, we have, indeed,
a philological difficulty of a more orthodox kind, in that there is
some uncertainty about the existence of the Ethiopic usage cited
as comparison. Dillmann cites only one case, with the careful
qualifications lut videtur’ and Hncerta est haec significatio’. Leslau
in his Contributions offers nothing on this root.
A similar example can be seen in Driver’s treatments of ‫ ש מ ר‬.
Even for the verb ‘keep’ he identifies a sense ‘tend, cultivate’
(Accadian summuru)> which, applied to Hos. 4. 10-11, with re-
pointing, gives
: ‫כ י ־ א ת ־ י ה ו ה ע ז ב ו ל ש מ ר ןנו ת‬
‘For they have forsaken Yahweh to cultivate whoredom’.

A second verb is ‫‘ ש מ ר‬rage’ (Accadian samaru). A third verb ‫ש מ ר‬

means ‘cast out, reject’. Hitherto recognized only in Syriac, it is
now identified at Ps. 37. 28b :
mt : ‫ר ת‬τ ‫כ‬: ‫ש *עי ם ·נ‬τ ‫ר‬: ‫וז—ר ע‬: ‫מ רו‬τ ‫ש‬: ‫ל ם *נ‬T ‫ל ע ו‬:

Here many scholars, noting that from the acrostic form the
half-verse should begin with ‫ ע‬, have exploited the άνομοι of the
Greek to write a subject ‫ ; ע ו ל י ם‬and, since this makes the wicked
into the subject of the first verb, either one must emend to ‫נ ש מ דו‬
142 TH E D IS T R IB U T IO N OF HOM ONYM S
or follow Driver and say that the ‫ נ ש מ רו‬of M T means ‘are cast
out’.1
In any case, our example shows how the familiar ‫‘ ש מ ר‬keep’ is
taken to have a homonym which is so sharply opposed in sense as
to make confusion likely. A context in which persons are ‘preserved’
by God is a context in which we might also hear that they are
‘cast out, rejected’. A similar treatment by Tur-Sinai at Job 14. 16
gives the sense:
mt ‫ט א •ת י‬T ‫עז מו ר ע ל ־ ־־ח‬ ‫ל א־ ת‬
: ·

‘thou dost not overlook my sin’

(reading as piel, and with the Syriac-Mandean sense of not ‘watch’
but ‘release, absolve’; parallelism with ‘thou numberest my steps’).2
If this is right, conflict of homonyms is again likely.3
Having given these illustrations, we now have to consider an-
other matter of general interest. It is rather difficult, and perhaps
not absolutely necessary, to draw a clear line between homonyms,
i.e. words which are alike in form but mean different things, and
the phenomenon of polysemy, or the existence in the same word of
a number of different senses.
At least some linguists have held that there is no absolute
distinction between the two assessments of a phenomenon.
Bloomfield, for example, writes: ‘In many cases we hesitate
whether to view the form as a single form with several meanings
or as a set of homonyms.’4
In some of the illustrations I have given, as the discriminating
reader will have noticed, it is possible to argue that the number of
homonyms is less than I have represented, because two of the
homonyms quoted are alternate senses of the same word. Thus,
for instance, I cited ‫ טו ב‬meaning ‘good’ and ‫ טו ב‬meaning ‘perfume’
as two homonyms. The philologist whose mind is accustomed to
1 I myself think the M T to be right in its normal sense: ‘they [i.e. the saints,
just mentioned] are preserved for ever, but the seed of the evil is cut off’. The
acrostic sequence ignored the particle ‫ ; ל‬cf. 37. 39, where the ‫ ת‬verse begins
with ‫ך‬. The Greek άνομου is from a variant or doublet.
2 Eitan treats the same passage in the opposite way, keeping ‫ ש מ ר‬as ‘keep,
watch’ and holding that ‫ ח ט א ת‬is ‘step’ (Arab. Jjata, hatwa).
3 The verbs here discussed are of course additional to the II and III ‫ש מ ר‬
registered as the putative roots of ‫‘ ש מ רי ם‬dregs’ and ‫‘ שמ*ר‬adamant’.
4 Language, p. 150; cf. also p. 145. It may be that the decision between homo-
nymy and polysemy can be reached through componential analysis, as outlined,
for instance, by Nida, Science of Translating, pp. 90 ff.
 H O M ON Y M S AND C O M M U N IC A T IO N 143
working on comparative, etymological, and historical lines may
reply that these are actually ‘the same word’, and that the sense
‘good’ is derived from the sense ‘sweet-smelling’ of the same word,
a relation made visible through the South Arabian tyb, a word for
a kind of incense. One sense, accordingly, may be said to ‘derive’
from the other. Similarly, with ‫ ע ז ב‬, where we noted a homonymy
between the word with sense ‘leave, abandon’ and the word with
sense ‘help’, Driver himself, whose interpretation we have been
considering, thought that a continuous semantic development
could be seen from a sense ‘left’ through the passive sense of
‘left; let alone; was indulgent to’ to the active sense of ‘helped’.
It seems to be wrong, however, to suppose that the difficulties
and problems caused by the recognition of homonyms are in any
way disposed of if we are able to see that an historical development
has taken place from one example to another. This is a clarification
only for the historically oriented philologist, not for the normal
user of the language. Menner seems to be right in arguing:
From the point of view of the speaker ignorant of origins, the em-
barrassment and confusion caused by multiplicity of meanings is likely
to be as great when a form represents two or more etymologically
distinct words as when it represents one. Most students of homonyms
and most semanticists pay little attention to this fact, but Jespersen
pertinently remarks that ‘the psychological effect of those cases of
polysemy, where “one and the same word” has many meanings, is
exactly the same as that of cases where two or three words of different
origins have accidentally become homonyms/1
In other words, ‫ טו ב‬meaning ‘perfume’ may, for the etymologi-
cally oriented philologist, be ‘the same word as’ ‫‘ טו ב‬good’; but
from the point of view of the communicative problems related to
homonymy, it makes little difference. The Hebrew speaker did
not know the history of the language; and, since there is a differ-
ence between ‘perfumed wine’ and ‘good wine’, nothing is achieved
1 Menner in Language xii (1936) 243 f. Cf. the procedure of the Chicago
Assyrian Dictionary, which provides a separate entry (marked by A and B, etc.)
where meanings are sufficiently different, thus providing for clarity and distinc-
tiveness even at the price of increasing the number of apparent homonyms.
There is no striving to secure a clear distinction between polysemy and homo-
nymy. See the statement in the preface of the first volume to be published, vol. vi
(£[), p. v. In not striving for any absolute distinction between homonymy and
polysemy, we appear to be in accord with the medieval Hebrew and Arabic
approach. The medieval tendency was to treat the matter as one of varying senses
of the same word. See Maimonides, in Efros, pp. 59 f.
144 TH E D IS T R IB U T IO N OF HOM ONYM S
to overcome the difficulty by saying that ‫ יי] ה טו ב‬historically may
have meant both.
Finally, an instance which shows how conflict of homonyms
may, through the perplexity it caused, have produced a graphic
confusion in the text. I Sam. 14. 25-26 uses the word ‫‘ י ע ל‬honey-
comb’. The MT is difficult, and a clever restoration by Wellhausen1
suggests that the normal word ‫‘ ל ב ש‬honey’ was added as a gloss to
‫ ; י ע ל‬the gloss in turn rendered ‫ י ע ר‬itself unnecessary, so that it
came to be understood in its other and commoner sense, i.e. ‘wood’.
Meanwhile the gloss ‫ ל ב ש‬through its similarity caused a confusion
and loss of a form of ‫‘ ל ב ל י ם‬bees’. Even if hardly quite certain,
this reconstruction illustrates the interdependence of linguistic and
textual transmission. Apart from the LXX, the versions follow the
sense ‘wood’.2
In order to indicate succinctly the kind of problems involved in
the suggestions which have been made, I attach now, without
detailed discussion of each case, a list of some typical examples in
which homonyms have been detected in recent philological
treatments. The second column gives the homonym (or the sense)
which is generally familiar; the third that which has been recently
suggested.
1. ‫אהבה‬ a. ‘love’ b. ‘leather’
2. ‫ברית‬ a. ‘covenant’ b. ‘splendour’
3· ‫מבשר‬ a. ‘messenger of good’ b. ‘refuter, opponent’
4. ‫גוי‬ a. ‘people’ b. ‘field, land’
5 · ‫ז־לק‬ a. ‘burn’ b. ‘pursue’
6. ‫דעך‬ a. ‘be extinguished’ b. ‘attack’
7· ‫דעין‬ a. ‘be fat’ b. ‘be hidden,
shrouded’
8. ‫הזון‬ a. ‘vision’ b. ‘magistrate’
9· ‫הטאת‬ a. ‘sin’ b. ‘step, walk’
c. ‘penury’
10. ‫עברה‬ a. ‘excess’ b. ‘wrath’
11. ‫עמס‬ a. ‘hoist, support’ b. ‘set in cement, lay
bricks’
12. ‫עש‬ a. ‘moth’ b. ‘pus, rottenness’
c. ‘bird’s nest’
1 Set out in detail by S. R. Driver, Notes on Samuel, pp. 113 f.
2 Most do not, however, fail to recognize the feminine ‫( י ע ל ת‬which has no
homonym) at 14. 27, or ‫ י צ ל‬itself at Cant. 5. 1 (but LXX there άρτον).
 H O M ON Y M S AND C O M M U N IC A T IO N 145
13. ‫רוץ‬ a. ‘run’ b. ‘break in’ (a horse)
14. ‫רעזע‬ a. ‘evil’ b . ‘ample, rich’
15· ‫_ עזב‬ a. ‘returning’ b. ‘running about,
(participle) straying’
16. ‫עזיר‬ a. ‘sing’ b. ‘travel’
!‫ל‬. ‫אתנן‬ a. ‘gift, price (of harlot)’ b. ‘effigy’

(3) The Count of.Known Homonyms

In estimating the likelihood of any suggestion which involves the
addition of a new homonym to the Hebrew lexical stock, it is
obviously relevant to know how many homonyms we have already.
Conceivably scholars may suppose that, since Hebrew has a large
number of homonyms in any case, there is no valid reason for
objecting to the addition of a few more.
We may try, therefore, to estimate the number of ‘agreed*
homonyms, that is, of homonyms the existence of which is
generally recognized and which provide a base for measuring any
increase of homonyms. I have not, however, found previous
studies which have produced counts of homonyms on a basis of
modern knowledge and in a form adapted to the questions I am
seeking to answer. The obvious source is the registration in the
standard dictionaries. The organization of these dictionaries is,
however, far from well adapted to our question. One cannot
simply go through them and count the number of places where
words are distinguished as I ‫ ע ב ר‬, II ‫ ע ב ר‬, and so on. These in-
elude many cases where the root is identical but no actual overlaps
in form occur. They use the same distinction for homonyms in
the normal sense and for words of special kinds such as personal
names and place-names. Moreover, the dictionaries vary between
themselves in the way in which they organize the differentiation of
various words. Nevertheless some approximate counts may be
made on the basis of the existing dictionaries.
Using as a criterion the formal one of enumeration under Roman
figures such as I and II,1 BDB is found to list just over 500
homonymic nouns. A very large proportion of these, however, are
proper names, which we may ignore for the purposes of this study.
The proper name ‘Smith* is homonymic with the noun ‘smith*,
but communication allows for this fact and it should not commonly
1 The figures are, very occasionally, inadvertently omitted by BDB.
146 TH E D IS T R IB U T IO N OF HOM ONYM S
stand as an example of interference with communication through
homonymy.
Once personal and place-names are removed, there are about 200
cases of homonymy involving nouns, counted from BDB’s entries.
Among these, however, some are limited in their degree of overlap
or of potential conflict.
A few are instances of homonymy between a noun and a word of
some other class. Thus the homonymy between ‫‘ א ו ל ם‬porch’ and
‫‘ א ו ל ם‬bu t’ is not a likely source of confusion.
Sometimes the semantic fields are so completely distinct that
overlapp ing is not significant. There is little serious overlap between
‫* ב ת‬daughter’ and ‫ ב ת‬meaning the measure ‘bath’. Moreover,
these words are formally distinct in the plural, and may have been
so in the suffixed forms also.
Sometimes one noun belongs to a fairly general semantic field,
while another is much more technical, being the name of an animal,
an instrument, a measure, or the like. This applies to pairs like
‫‘ א נ ק ה‬crying’ and ‫‘ א נ ק ה‬ferret, shrew-mouse’; ‫‘ ד ר ו ר‬release’ and
‫‘ ד ר ו ר‬swallow’; and the two words ‫ ב ת‬, just quoted.
Again, sometimes there may have been no homonymy within
the same text, or the same period. It has often been argued in
favour cf philological treatments that Hebrew is not a uniform
mass but has variations according to place, time, and style. This
same argument, however, may reduce the number of agreed
homonyms and thereby may reduce the probability that should be
accorded to suggestions of new ones. In other words, the homony-
my is a product of the lexicographical process, in which all Hebrew
words are gathered together; but in the original situation, the
homonymous words did not all exist together in the same register,
place, or time, so that homonymic conflict did not take place.
A real homonymy in the same texts is that of 1) ‫‘ ) איי ם‬foreign
coastlands, islands’, (2) ‘wilderness animals’ (traditionally jackals).1
Though this is a full homonym within the same register of speech,
the words are easily distinguishable because ‘foreign coastline’ is
excluded in the contexts which imply an uncanny howling being,
for all cases of the latter are in a list of several wilderness animals.
Similarly, the Song of Songs has not only the common ‫?? ד ב ר‬

1 Torrey, Second Isaiah, pp. 289 fF., and KB take these senses as related in
origin; even if true this does not much affect my argument.
 TH E CO U N T OF KNOW N HOM ONYM S 147
‘wilderness’ but also the hapax legomenon usually taken to
mean ‘speech’ or ‘mouth’. Occurring as it does (4. 3) within a list
of the physical characteristics of the woman, the use runs little risk
of confusion. But at least these examples are found together within
the same texts.
It is different with the ,,X ‘not’ of Job 22. 30 *’j?T‫‘״‬,X. Assuming
that this is not a textual error (as KB takes it to be), it is possible
that it belonged to the vocabulary of Job but not to that of the
classical prophets. Conversely, Job does not use *,X for remote
lands, islands, jackals, or goblins, although these are all themes
which might well come within the purview of his poem. Thus it is
possible that this homonymy, though it appears in the catalogue
structure of the lexicon, did not in fact operate; and the same
argument could be made for the *,X which appears twice in
Qoheleth. Thus although a dictionary is forced to register perhaps
four words of the form *,X, it is also possible that in the usage of a
particular place, time or style only two possibilities, or even only
one, existed.
Once all these various considerations and qualifications have
been taken into account, the gross number of some 500 homonymies
involving nouns may be reduced to a number of practical
homonymies, involving nouns in such a way as to create any real
ambiguity or conflict, of a few dozen in all.
We should consider not only how many forms there are which
have homonyms but also, where a form has homonymy at all, how
many homonyms it is likely to have. The received dictionaries
may from time to time list as many as four, five, or six homonymous
words, but such cases form a small proportion of the total number
of homonymies.
Thus under *?‫!״‬X, according to BDB, there are four homonyms,
meaning ‘ram’, ‘chief’, ‘pillar’, and ‘tree’. One may wish to modify
this: the first and second might be classed as a polysemy; the sec­
ond and the fourth do not actually appear in the absolute singular
form which we have quoted. But there are still probably three or
four homonyms, and we can even add more if we take the construct
form 1?X (the writing as 1?*,X makes no difference for our purpose);
for then there is the familiar divine name 1?X, and also, whether it is
‘a different word’ or not, the 1?X (sometimes taken to mean ‘power’)
of the phrase *,,T ^X*?~t2P. Thus for the form 1?X (1?‫״‬,X) we
could possibly speak of six homonyms.
148 T H E D IS T R IB U T IO N O F H O M O N Y M S
Perhaps a better example is that of ‫ כ פ ר‬. BDB, GB, and KB
register four homonyms, with meanings of ‘ransom’, ‘pitch’, the
name of 21 plant, and ‘village’. But the third is a plant name and
the fourth is conceivably a dialect form.
In any case this situation with four homonyms or more is
relatively rare. There are four also with ‫ ח רו ץ‬and with ‫ צ י ר‬, and
more precariously with ‫ צ ר‬. But of the total number of forms
where homonyms occur, the vast majority among the nouns offer
only two possibilities, once proper names are excluded. Only
quite rarely have large clusters of identical homonyms, with
like grammatical function and excluding proper names, achieved
general recognition.
With verbs, as with nouns, the dictionaries often list what appear
to be homonyms, but on examination these turn out to be from
texts of widely separate provenance, so that no synchronic homo-
nymy occurred. Of the four homonymic verbs ‫ ע נ ה‬, the second
(‘be occupied’) occurs only in Qoheleth, which book, however,
does not use the third (‘be afflicted’)—though affliction, one might
say, is a ;prime theme of this author—or the fourth. Possibly ‫ע נ ה‬
‘sing’ was losing ground before the more frequent ‫( ש י ר‬which alone
had a noun ‘song’ to go with it), and by the time of Qoheleth was
obsolete, thus enabling the newer ‫‘ ע נ ה‬be occupied’ to find a place.
The word ‫‘ ג א ל‬defile’ occurs in Deutero-Isaiah, in which ‫ג א ל‬
‘redeem’ is also prominent. One might expect therefore to find
some considerable conflict. It is interesting therefore that prac-
tically no overlap of actual forms occurs, the only case being the
form ‫ נ ג א ל ה‬at Zeph. 3.1. The forms usually registered as niphal
and hiphil are both unusual forms, which do not occur for ‫ג א ל‬
‘redeem’-‫ נ ג א ל ו ־‬and ‫ א ג א ל תי‬. Thus actual homonymy of forms is
very slight, and possibly the avoidance of homonymy has fostered
the production or retention of the peculiar verbal forms. Caution
must, however, be exercised, because forms which are not attested
may nevertheless have occurred.
Another relevant test is a comparison of the incidence of
homonyms in Hebrew and in the other Semitic languages. It
would be paradoxical if philological treatments produced in
Hebrew an incidence of homonyms greater than that which is
found in the other Semitic languages themselves.
The incidence of homonyms might be expected to vary in
 TH E CO U N T OF KNOW N HOM ONYM S 149
relation to the phonological changes in the prehistory of each
language. Where phoneme mergers have been frequent, the
incidence of homonymy should be expected to be higher. Other
things being equal, there should be more homonyms resulting
from such merger in Hebrew than in Arabic and Ethiopic, but
fewer than in Accadian or Mandean. It is in fact not unlikely that
Hebrew roughly occupies such an intermediate position. This,
however, is far from an exact account of the matter, because
phoneme merger is not the only cause of homonymy. The in­
fluence of loan-words, for instance, has also to be considered, and
is quite high in Accadian, with its many Sumerian words.
Preliminary surveys I have done in two languages fairly close to
the circumstances of Hebrew, namely Ugaritic and Syriac, do not
encourage us to suppose that Hebrew had an incidence of homo­
nymy substantially higher than has already been recognized; on the
contrary, they suggest that the normal degree of homonymy was
somewhat lower. The Ugaritic glossary, when studied cursorily,
appears to display numerous homonymies; but most of these
would disappear when the texts were vocalized. One can, for
instance, expect that hmr ‘ass’ would be as distinct in Ugaritic
from hmr ‘clay, mud’ as is the case in Hebrew. The same would
apply to rh ‘spirit’ and rh ‘smell’. In Ugaritic verbs a number of
homonymies can be found, but not very many. With hwy, for
example, there is no overlap of themes; the verb ‘bow down’
occurs only in a theme unknown for hwy ‘live’. In some cases,
homonymy appears by the reconstruction of one scholar but not
by that of another. There are two homonyms slh according to
Gordon (i. ‘send’, ii. ‘cast’ or ‘beat out’ a metal), and Driver
construes the situation in the same way; but Aistleitner tries to
include both under the same word (his no. 2610). The form t<T
from the verb ‘sweat’ is registered under yd* by Driver and thus
becomes homonymic with y d ' ‘know’; but Gordon has it under
d' and Aistleitner under (w)d\ in which cases it would probably
not be homonymic when vocalized. The number of clear cases of
homonymy in the Ugaritic stock of verbs, once the necessary
qualifications are made, is quite low.1
1 Driver’s glossary appears to yield 16 cases, if we exclude cases like nd (i.e.
nwd) and nd (i.e. ndd); they are: *any, b*l, hwy, hrr, }iss, t*n, yd*, kss, *ms, *ny, qry,
sb*t slh, tr, t*r, t*r. But a considerable number of these would be considered
doubtful and are otherwise construed by other scholars. Some others are not
true homonyms by the criteria we have already established above.
150 T H E D IS T R IB U T IO N O F H O M O N Y M S
Syriac is another language which like Hebrew can present,
when surveyed in the dictionary form, remarkable groups of
‘homonymic’ roots. A good instance is that o f'rb, for which Brock-
elmann registers ten separate entries as against the six in Hebrew
according to BDB; but very few of these have really overlapping
homonymic/oms. In Syriac we have the advantage of a discussion
of homonymy on fairly modern philological principles, by Schul-
thess (1900). His discussion, however, is of homonymic roots, and
does not explore the somewhat different problem of homonymic
forms. The total number of roots discussed by him is forty-eight.1
I have also tested Syriac by taking the number of homonyms
beginning with a particular letter. Brockelmann marks only four
real homonyms with first radical /d/:
dgl (1) ‘aim’ (2) ‘lie, deceive’
dwl (1) ‘move’ (2) no real homonymy (3) ‘serve, be humble’
dwq (1) ‘pound’ (2) ‘inspect’
dH (1) ‘neglect’ (2) ‘cover, close’.
Dll and drr, though marked, are perhaps not real cases.
In Syriac, if the general incidence of homonymy were the same
as in Hebrew, one would expect the proportion including /d/ to be
higher than in Hebrew, because Syriac /d/ results from merger of
/d/ with /d/, while pre-Hebrew /d/ became jzj in Hebrew. In so far
as such merger is a cause of homonymy, one would expect a higher
incidence in Syriac. In fact the incidence is lower than that already
recognized for Hebrew. BDB, even if we eliminate some im-
probable entries (like two words ‫) ל ג ל‬, offers fairly good examples
such as
‘1) ‫‘ )מדבר‬wilderness’ (2) ‘speech’ or ‘mouth’
1) ‫‘ )מלוה‬sickness’ (2) ‘garment’ (p. 551b)
1) ‫‘ )דל‬door’ (2) ‘poor’

1 Though Schulthess discusses the way in which homonyms come into exis­
tence, he has, typically of the older philology, no thoughts about the problem of
communicative efficiency, and he does not really offer anything relevant for our
present discussion of homonymy in general. Nor are the problems which con­
cern us seen by Nöldeke in his review, ZD M G liv (1900) 152-64. For our pur­
pose the study of Schulthess is of interest mainly for its count of homonymic
roots, and for some careful establishments of distinctions between homonyms.
The forty-eight homonymic roots treated by him are: bdl, bl\ b't> glby gppy dglt
d iif xhr, zw f zlly hblf hggy hmyy hsd, tw \ kws, kssf m&r, mit, nglt sbrt shr, 7/, 'mdy
*rb, prgy pTy siv\ y/Z, spp, qbbt qtmf qply qsr, rhbt rht, r \ rpt, sgm, sgr, ihm, ihr,
iwh (Jyh)y ill, Up, iqp, irb, tkk.
 TH E CO U N T OF KNOW N HOM ONYM S 151
1) ‫‘ )דלה‬hair, thrum’ (2) ‘the poor*
1) ‫‘ )דמה‬be like’ (2) ‘cease’
1) ‫‘ )דמם‬be silent’ (2) ‘wail’
1) ‫‘ )דרור‬freedom’ (2) ‘swallow’.
Modern philological suggestions would certainly, if accepted,
add to this number; see Index, nos. 81-4, 87-101. One would
expect Syriac to be a suitable comparison for Hebrew—not far
removed in period, in cultural situation, or in type of literature.
Yet, if the sample can be generalized, the incidence of homonymy
in Syriac is actually lower than that already recognized in Hebrew,
to say nothing of the substantial additions to Hebrew consequent
on philological suggestions. It would then seem possible that
philological treatments, though assuming a close overlap in lexical
resources between the Semitic languages, nevertheless bring
about a large disparity between them in respect of homonymy.
It is conceivable, on the other hand, that there have been periods
of unusually high homonymy in the development of a language,
and that in Hebrew the biblical period was such an epoch; by the
end of the biblical era many such homonyms had been eliminated
and a certain levelling of the vocabulary brought about. Syriac
would then be more comparable with the post-biblical era in
Hebrew. This possibility, though difficult to prove, deserves
further research.

(4) Homonymy and Style

Another aspect of homonymy in Hebrew is its relation to style.
It is by no means intrinsically impossible that Hebrew contained
larger numbers of really interchangeable homonyms than (let us
say) Syriac or modern literary Arabic. If this were so, however, it
would almost certainly have an effect upon style. Certain kinds of
Chinese poetry are said to build much of their stylistic effect upon
the homonymic and ambiguous nature of the elements used.
This is not an argument that Hebrew is like Chinese in this re­
gard, unless we go on to say that Hebrew style, like the correspond­
ing Chinese style, is actually built upon the subtle appreciation of
ambiguities in language. Such an assertion about Hebrew style I
have never seen. But in any case, if homonymy was a major
feature of classical Hebrew, we might expect it to have an effect on
style, whether by the development of devices through which clarity
15* T H E D IS T R IB U T IO N OF H O M O N Y M S
in ambiguous contexts might be found, or alternatively by the
development of a style through which the effects produced by
recurrent homonymy are prized and cultivated. A style of this
latter land is found in some medieval Hebrew poetry.
Now there are quite a large number of places at which scholars
have claimed to identify the use of a pair of homonyms in juxta-
position as a literary device. Not all of these command agreement
by all scholars, but some can be considered very likely. A widely
accepted case is Ps. 137. 5, already quoted.1
Play upon homonymies and double meanings was a feature of
the ancient riddle style. A clear instance comes from the master
of this style, Samson:
Judges 15.16:
‫ב ל ח י ה ס מו ר ס מו ר ס מ ך תי ם‬
: ‫ב ל ח י ה ס מו ר ה כי ת י א ל ף אי ש‬
‘With the jawbone of an ass, mass upon mass,
with the jawbone of an ass I slew a thousand men.’
There is exploitation of the homonymy between ‫‘ ח מ ו ר‬ass’ and
‫‘ ח מ ו ר‬heap’. Even if the phrase ‫ ח מ ו ר ח מ ר ת י ם‬is taken in an-
other sense (e.g. ‘shave, scrape, flay’, supposedly after an Arabic
hamara—accepted by GB, p. 242b),2 the play on homonymy
remains.
A play on homonyms is probable also in the answer to the riddle
about the honey, Judges 14. 18:
: ‫ מ ת ו ק מ ך ב ש ו מ ה עז מ א רי‬- ‫מ ה‬
There is an Arabic ‫י‬ary ‘honey’, and the riddle played upon a
Hebrew cognate, more or less homonymous with ‫‘ א ר י‬lion’.3
Play upon homonyms merges into the more general juxtaposition
of words which merely have some similarity; this can produce all
sorts of effects, from the emphatic to the comic. Isa. 5. 7 has been
1 See above, p. 48.
2 Cf. B H 3. The Greek e£a\ei<j>cov igrjAen/‫׳‬a does not support the interpretation
followed by GB. More probably it was influenced by the other verb ‫ ח מ ר‬, used
of the wiping on of pitch in Exod. 2. 3 (wiping on and wiping off can probably
be covered by the same verb, cf. the history of ‫) כ פ ר‬. It was also influenced
by the range of e£aAe1<£a> in Greek, which could be used for the ‘wiping out’ of
people, e.g. Aes. Ch. 503, !17] ,gaAdiftys crnepua ilcAomSah‫׳‬, and thus fits nicely
with the total sense found in the Greek.
3 See J. R. Porter in J T S N.s. xiii (1962) 106-9.
 HOM ONYM Y AND STYLE 153
quoted already.1 A striking case is Isa. 42. 1-4, with the assonance
of ‫ ר צ ץ‬and ‫ ר ו ץ‬, and the use twice of ‫( כ ה ה‬perhaps two homo-
nyms meaning ‘be dim’ and ‘rebuke’). The exact image of this
song, however, remains obscure.
At some other places, at which the stylistic exploitation of
homonymy has been argued, stylistic considerations seem to tell
against the suggestion. At Isa. 55. 1, for instance, it has been
suggested that ‫ כ ס ף‬is used firstly to mean ‘food’ (cf. the Accadian
kispuy kasapu, kusipatu, etc.; von Soden, pp. 487, 453, 514) and
then with the usual sense of ‘silver, money’. The text would then
mean something like:
Ho, everyone who thirsts, come to the water; and he who has no food
(‫) כ ס ף‬, go, buy and eat; and go, buy without money (‫ ) כ ס ף‬and without
price wine and milk.
The effect of the collocation, if this suggestion were right, would
surely be to produce a poor and stupid comic jingle.2 That the
text repeats twice the familiar sense of ‘money’ seems infinitely
more probable.
Elsewhere, again, a collocation of homonyms may have taken
place more or less accidentally, without any deliberate striving
after effect. At Qoh. 7. 6:
‫כי ס קו ל ה סי רי ם ת ח ת ה ס י ר‬
‫ סי ר‬means firstly ‘thorn’ and secondly ‘cauldron’, but it is possible
that the homonyms are juxtaposed without either the intention of
an effect or the result of one. Or does this passage from one
meaning to another of a rather unimportant word symbolize
futility and foolishness P3
Other places where the collocation of homonyms has been sug-
gested include
Isa. 32. 6 ‫עשה‬ (1) ‘do’ (2) ‘conceal·4
Jer. 12. 9 ‫עיט‬ (1) ‘den’ (2) ‘bird of prey’5
Mic. 1· 7 ‫אתנן‬ (1) ‘effigy’ (2) ‘harlot’s hire’
Ps. 5· 4 ‫בקר‬ (1) ‘morning’ (2) ‘offering for oracle’
Ct. 3 - 8 ‫חרב‬ (1) ‘war’ (2) ‘sword’
1 Above, p. 48; for others, see Driver in V T iv (1954) 242.
2 A good deal worse in this respect is Dahood’s discovery of ‫‘ ע רי ם‬gods*
at Jer. 2. 28, and 19. 15, giving for the former a sense ‘For your gods, O Judah,
were the number of your cities/gods\
3 Gordis, p. 259, takes this to be an example of the writer’s literary skill.
4 Eitan, p. 58. 5 Cf. above, pp. 128 f.
154 T H E D IS T R IB U T IO N O F H O M O N Y M S
Several places suggest play on the similarity of ‫‘ י ל א‬fear’ and ‫ל א ה‬
‘see’, e.g. Ps. 40. 4, 119. 74, Job 6. 21.
Sometimes, where a pair of homonyms is held to exist but
where the text in question uses only one of them, it is possible that
paronomasia or punning is intended; that is, that one form is
used with the intention of suggesting two quite different meanings.
For example, a word ‫‘ א ה ב ל‬leather’, in addition to the familiar
‫‘ א ה ב ל‬love’, has been found at Hos. 11. 4
5 ‫ז כ ם ב ע ב ת ו ת א ה ב ה‬# ? ‫ס ג ל י א ך ם א‬
where the senses in question are ‘bands of leather’ and ‘bands of
love’ (parallel ‘cords of hide’ ?), and Cant. 3. 10, where Solomon’s
palanquin is
‫ס ך כ ב ו א ת מ ן תו כו ז ל צוף א ה ב ה‬
In ttiis latter example, since the parallels are substances like
‘silver’, ‘gold’, and ‘purple’, there seems to be strong reason for
accepting the rendering ‘leather’.
On the other hand one can hardly ignore the fact that love (‫) א ל ב ה‬
is precisely the main theme of the Song of Songs, and, in a some-
what different sense, of Hosea also. It is therefore a little hard to
contemplate that a word ‫ א ה ב ה‬would be used prosaically for
‘leather’ in these two works, without the striking similarity to the
word ‘love’ being noticed.
To sum up, on the one hand it is a reasonable possibility that,
when a rare homonym occurred in the Old Testament, its own
homonymy was a factor in its own desuetude and thereby in the
loss of understanding of its meaning. Where philological treat-
ments claim to uncover homonyms previously not detected, this
circumstance is in their favour. In relatively few cases, however,
so far as my research goes, have such treatments tried to document
with evidence this obsolescence of words because of homonymy.
Treatments which use this argument should in principle attempt
to show in what way semantic confusion may have arisen, in the
circumstances of the relevant contexts, and in relation to the
historical distribution both of homonyms which coincided with
the word in form and of synonyms which, by coinciding roughly
with it in meaning, may have taken over its semantic function
when it came to be lost.
On the other hand we have not found evidence that homonymy
 HOM ONYM Y AND STY LE 155
was a very widespread feature of biblical Hebrew. Its total incidence
of homonymy, though probably higher than that of Greek or
Latin or Arabic, was much lower than that of English or French.
Philological treatments, if too many of them are accepted un­
critically, would seem arbitrarily to raise the incidence of homonyms
above the level justified by general evidence, and thereby to make
Hebrew a language more troubled by homonymy than the very
Semitic languages (apart from Accadian) which are being used as
sources; and this procedure, in turn, must carry peculiar im­
plications about style and communication in Hebrew, which
implications have not generally been realized. These arguments
favour reserve in accepting large additions to the count of
homonyms in Hebrew.1
1 Cf. recently D. F. Payne, ‘Characteristic Word-play in “ Second Isaiah” :
a Reappraisal’ in J S S xii (1967) 207-29.
 VI I

THE DISTRIBUTION OF LEXICAL

RESOURCES IN THE
SEMITIC LANGUAGES

(i) General
P hilological treatments often appear to depend logically on
the assumption of a high degree of community or overlap between
the various Semitic languages in their use of lexical resources. It is
implied that, a cognate word having been found, this will form a
probab le lead to (a) the existence of a related Hebrew word and (b)
the meaning of that Hebrew word. The existence of a word in a
cognate language is taken as a sort of prima-facie case for the
existence of a corresponding word in Hebrew, and this prima-facie
case is then clinched by the fact of a Hebrew text which seems to
fit this word and give good sense.
The question of principle can be put in a quite simple form. How
great is the degree of coincidence (leaving aside words of non-
Semitic origin) between the various Semitic languages in their use
of vocabulary ? If this degree is high, then there will be a high de­
gree of probability (other things being equal) in a claim that, if a
word exists with a known meaning in language A (let us say,
Arabic) it can therefore confidently be expected to occur in a
recognizably related form and with a recognizably related meaning
in language B (in our case, Hebrew). If the general degree of
coincidence is low, then there will be a correspondingly low degree
of probability in this claim. When we find it argued or implied that
the existence of a word or form in a cognate language is prima-facie
evidence for its existence also in Hebrew, such an argument or
implication clearly rests on the assumption that the degree of
coincidence is very high.
It should be noted that the question of related form cannot in
this context be treated usefully apart from the question of related
meaning. Where the same form, or a form corresponding under
 GENERAL 157
known conditions, is found in more than one language, is the mean­
ing also substantially the same ?
If we have ‘the same word’, but if it is found with a different
meaning in language B from that known in language A, this does
not make it difficult to claim that the word exists also in B, but it
means that it will be hard to know what it means in B, and that
therefore we shall have less chance of reaching any solution of the
problem at all. For, we must remember, the whole situation under
discussion is one in which the existence of the word in language B
has not hitherto been recognized; and this means that there is a
lack of evidence for its meaning in language B, apart from the
evidence of the cognate language.
Moreover, since we are using as further evidence the context of
a particular passage, the relevance of that context is itself depen­
dent upon the degree of assurance we have about the meaning of
the word we now claim to recognize for the first time. To put it
simply, it will not help us much if we identify the root x-y-z, well
known in Arabic, if we still do not know what this would mean in
Hebrew. For these reasons it is of real importance to consider, from
our general knowledge of the Semitic languages apart from the
exigencies of particular difficult passages, what is the degree of
community in the use of formally corresponding lexical items, and
what is the degree of community in the meanings with which they
function in the various languages of the group.
Now there is no difficulty, to begin with, in assembling a sub­
stantial list of words which in form (allowing for the normal cor­
respondences) and in meaning (allowing for slight and easily
explicable differences) run fairly uniformly across the whole series
of the ancient Semitic languages or a large number of them. Such
a list is offered by Bergsträsser in the appendix to his comparative
study of the Semitic languages.1 It contains about 170 items, of
which Bergsträsser claims that they comprise the ‘relatively certain
correspondences of the five chief branches of the Semitic languages’
(excluding loanwords from one branch to another). This, he says,
is far from exhausting the lexical stock of proto-Semitic; for all
words which are lacking in one or more of the branches, or which
have been altered beyond recognition, are omitted.
This list contains some very fundamental words: basic words
for human beings and relations, animal names, parts of the human
1 G. Bergsträsser, Einführung, pp. 181-92.
158 LEXICA L R E S O U R C E S IN T H E S E M I T I C L A N G U A G E S
body, some cosmic elements like day and night, and a number of
common verbs, along with the numerals and a few prepositions.
We may quote a few of the words in their Hebrew form: ‫א ב‬
‘father’, ‫‘ י ל ל‬bear’ (a child), ‫‘ ע ק ר ב‬scorpion’, ‫‘ ז ר ע‬seed’, ]‫ע י‬
‘eye’, CH ‘blood’, ‫‘ יו ם‬day’, ‫‘ מי ם‬water’, ‫‘ ב י ת‬house’, ‫‘ נ שא‬lift’,
‫‘ ק ר ב‬approach’, ‫‘ ב כ ה‬weep’, ‫‘ ש ב ר‬break’.
For all these words, about 170 in number, close correspondences
in form and sense can be found running across the whole field of
the basic branches of the Semitic languages.
The closeness of the agreement running through this list should
not, however, impress us too much. This list is itself a very limited
segment of the vocabulary of any one of the languages. Against it
we have to set the vocabulary which is used in one language, or in
two, but is not represented in the others. It is easy to produce
impressive lists of words which are common in two or three of the
relevant Semitic languages, but which are entirely absent from
others or else occur in them only with meanings substantially
different.
Dillmann1 gave a list of about twenty-five important words
which are shared by Hebrew and Ethiopic but do not appear in
Arabic at all or appear in it only with very different meanings.
They include, for instance: ‫‘ א ש‬fire’; ‫‘ ע ץ‬tree’; m ‘stone’;
‫‘ א ש ך‬testicle’; ‫‘ י כ ל‬be able’; ‫‘ י צ א‬go out’.
Following this, Ullendorff2 offers a list of ‘an impressive number
of words shared by Hebrew and South Arabian for which either no
equivalent roots are attested in the other Semitic languages or else
with such sharply differing meanings as to make the semantic and
structural identity doubtful’.
Ullendorff’s list comprises twenty-two examples and includes
such prominent words as ‫‘ אי ש‬man’; ‫‘ ה ר ג‬slay’; ‫‘ י ר ה‬shoot’;
‫‘ י ש ע‬salvation’; ‫‘ ק ה ל‬assembly’.
Delitzsch used similar lists in his arguments against the priority
of Arabic for the elucidation of Hebrew vocabulary. One such list
endeavours to show how Arabic presents, in comparison to Hebrew,
Aramaic, and Accadian, narrowed or clearly derivative meanings.3
This he applies, for example, to :
1 Ethiopic Grammar, pp. 6-7. Some of these words are shared also by Ac-
cadian and Ugaritic. 2 V T vi (1956) 195 ff.
3 Prolegomena, pp. 27 f.; list showing Hebrew-Aramaic community, pp. 32-
35, and Hebrew-Accadian community, pp. 45 ff. Cf. above, p. 70.
 GENERAL 159
‫‘ אמר‬say’—Arabic 'amara ‘command’
‫‘ בוא‬come’—Arabic baa ‘return home’
‫‘ בעל‬possess’ and also ‘marry’—Arabic baala specifically ‘marry’
‫‘ עזב‬leave, forsake’—Arabic '-z-b particularly ‘be unmarried’
These indications, then, encourage us to pursue further the
question of the degree of overlap in lexical stock between the
various Semitic languages.
While Bergstrasser’s list took a particular meaning and showed
the correspondence of words with that meaning throughout the
Semitic group, there seem to be other meanings (such as the words
for going in a direction) which tend to have different words in the
cognate languages:
Hebrew Arabic Aramaic Accadian Ethiopic
‘go down’ ‫ירד‬ n-z-l n-h-t aradu wàrâdà
‘fear’ ‫ירא‬ hâfa d-h-l adâru fârhà
‫פחד‬ palahu
‘know’ ‫ידע‬ '-r-f y-d- edü ’a’màrâ
'-l-m
In some of these one language has a form closely similar to that
of another in the same line, but with a meaning substantially differ-
ent. Arabic w-r-d means not ‘go down’ but mainly ‘arrive’; Hebrew
‫ נ ז ל‬means not ‘go down’ but ‘flow’; Arabic f-r-h is ‘be lively’;
very few Semitic languages show a cognate with Hebrew ‫י ר א‬
‘fear’;1 though Arabic f‫־‬r ‫־‬/ is a very common word ‘know’, the
relevant Hebrew ‫ ע ל ף‬exists only as a noun ‘neck’ and as a deno-
minative verb ‘break the neck’; Hebrew ‫ ע ל ם‬means ‘hide’.
These differences, which refer to the normal words in use, are
not removed by the occasional appearance in Hebrew of a form
normally Aramaic, such as ‫‘ נ ח ת‬go down’, 0 ‫‘ ל ק‬go up’.
Conversely, one may take a given form and, allowing for the
normal correspondences, consider what meanings attach to it in
the other languages. For example: (a) Hebrew ‫‘ א מ ר‬say’: the
senses of cognates are—Accadian ‘see’, Arabic and Aramaic mainly
‘command’, Ethiopic ‘show, know’, (b) Hebrew ‫‘ ל ק ח‬take’: in
Arabic this is ‘conceive’ or ‘impregnate’; Ethiopic laqha means
‘lend’, not ‘take’ ;2 in Syriac the root does not exist except for laqha
1 It occurs in Ugaritic, but not frequently; Leslau, Contributions, p. 42, gives
a rather remote parallel from Tigre for ‫ פ ח ד‬.
2 Leslau, Contributions, p. 29, against KB.
160 L E X IC A L R E S O U R C E S I N T H E S E M I T I C L A N G U A G E S
*planities circa urbem’,1which is semantically remote. For ‘take’ the
Aramaic word is ‫ ב‬0 ‫ נ‬, which is not found in Hebrew; the Aramaic
,ithpeel of ‫ ל ק ח‬is rare and means ‘be taken as a wife’; Ethiopic
*ahazd is cognate with a Hebrew word, but also often appears with
the sense ‘begin’. Cognates of the Hebrew are found in Ugaritic
and Accadian (Iqh, lequ).
Our purpose, we should remember, is not to discover an etym-
ology or identify the same root in another language. The traditional
etymological consciousness, because it is directed towards questions
somewhat different from ours, can indeed cause us some confusion.
In particular, (1) it may be content to find in the cognate languages
a word which has the same root, though that word is actually in a
quite different formation; (2) it may overemphasize the fact that a
cognate has the same root and underemphasize the fact that the
meaning is substantially different, or it may emphasize the histori-
cal connectedness of the two meanings in such a way as to ignore
their difference in function; (3) it may fail to balance the recognition
of cognates with a recognition of the number of languages in which
a cognate is not found. For example:
(1) There is a familiar Hebrew word ‫‘ חו מ ה‬wall’. The etymo-
logist will probably connect this with verbs meaning ‘protect’ in
cognate languages. This, however, does not in itself mean that such
a verb, which would be ‫ ח מ ה‬, is in use in Hebrew with this mean-
ing. The etymology of any particular word does not establish the
existence of other formations from the same root.
It is useful to distinguish between roots which are productive
and those which are not. Hebrew has a root ‫ ל א ך‬, which appears
in the familiar words {a) ‫‘ מ ל א ך‬messenger’, (b) ‫‘ מ ל א כ ה‬work’.
An obvious cognate is the common Ethiopic la aka ‘send’. Unlike
the situation in Ethiopic, however, it is improbable that the verb
‫ ל א ך‬existed in Hebrew as a free form. Though the root ‫ל א ך‬
exists, it cannot therefore be used to predict occurrence in other
formations; nor, for the purpose of calculating the extent of lexical
overlap, can it be counted in the same way as free and productive
roots. Moreover, while we could rightly divine the Hebrew mean-
ing of ‫ מ ל א ך‬from that of the Ethiopic verb, it is doubtful if we
could do this for ‫ מ ל א כ ה‬if the meaning of the latter were in fact
unknown.
1 Brockelmann, Lexicon Syriacum, p. 370.
 GENERAL 161
Other examples can easily be added, e.g. the roots of ‫ל מ י ל‬
‘always’; ‫‘ נ ב י א‬prophet’; ]!"ID ‘priest’; ‫‘ ק ו ל‬voice’.
(2) There is a familiar Arabic jalasa ‘sit’. Hebrew ‫ ג ל ש‬provides
an exact correspondence and is probably cognate; it seems, how-
ever, to mean ‘glide down’, and is used of a herd of goats coming
down a hill (Cant. 4. 1,6. 5). Perhaps etymologists can surmise a
mode of semantic connection between the two. But, if the meaning
in Hebrew were in fact unknown, it is very unlikely that we should
be able to reconstruct it correctly from the Arabic.
The following are some other words, the meanings of which in
Hebrew could probably not be divined from cognate languages, if
the meanings had not been known in the first place: ‫‘ ל ב ר‬speak’;
‫‘ פ ח ל‬dread’; ‫‘ ב ר א‬create’; ‫‘ ג ב ו ל‬boundary’; ‫‘ ב ר י ת‬covenant’;
and ‫‘ ע ל ם‬hide’.
Sometimes, even if the root is known, the derivation of particular
meanings does not follow the same analogy in different languages.
We know that ‫ ל ל ך‬means ‘road’. The verb ‫ ד ר י‬means ‘tread’.
But the words in other Semitic languages which mean ‘road’ are
not cognates of ‫ ; ל ל ל‬and, moreover, they are often not related
to the word which means ‘tread’ in the relevant language. Aramaic
‫ א ו ל ח א‬appears to associate itself with the verb ‘travel’, Arabic
tariq with a verb ‘knock’, sari* with a verb ‘enter, begin’, and
Ethiopic fanot with a verb ‘send’. It is by no means certain that,
even if a ‘root-meaning’ is known, analogy will be available in such
a way as to ensure a right conclusion to the meaning of a given form.
(3) It is important to register not only the presence but also the
absence of cognate words. The latter is frequently neglected. Non-
existence of cognates is not usually made explicit by the diction-
aries, and it is dangerous to treat their silence as evidence that
forms are not found. Moreover, because they are trying to explain
or illustrate a Hebrew word, they will often cite a form in another
Semitic language with a meaning which comes close to the Hebrew
meaning, without making it clear that in the language quoted this
is not by any means the usual meaning. In these three respects,
then, our purpose is something different from the traditional
interest in etymology.1
Another test is to go through the known vocabulary of a given
language and consider how many of the items have a Hebrew word

1 Cf. again below, p. 299.

162 L E X IC A L R E SO U R C E S IN T H E S E M IT IC L A N G U A G E S
which shows adequate correspondence with them in form and in
meaniig.
In itrabic, as has already been implied, a very little thought is
enough to disclose a large number of familiar words which either
have r.o formal correspondence in Hebrew or else have a formal
correspondence only where no close semantic similarity exists. Let
the reader consider, for example, such familiar Arabic words as
the following:
iayh ‘old man, sheikh’ *aql ‘intelligence’ gayr ‘other than*
Him ‘knowledge’ fadl ‘favour’ hasan ‘beautiful’
jama a ‘collect’ fiqh ‘legal science’ jard ‘run’
jihad ‘holy war’ 'ajüz ‘old woman’ jänib ‘side’
jahila ‘be ignorant’ *arsala ‘send’
Some of these Arabic words have, indeed, been ‘found* in
Hebrew by scholars: ‫ ע ל ם‬at Qoh. 3. 11 has been interpreted as like
Arabic Him ‘knowledge’,1the root ofjihadhas been found at II Kings
4. 34 (MT 2,(!.‫ יגהר‬and Tur-Sinai identifies in the ‫ ד ק ה י ל‬of
Job 11. 10 a verb ‫‘ ק ה ל‬forget’ cognate with jahila (Index, no.
277). Such suggestions must, at this stage of the argument, be held
in suspension; we cannot allow our comparison to be dominated at
the start by large numbers of suggested words, some of which have
hardly come to be widely known, much less accepted.
Again, samples taken from various kinds of Arabic literature (e.g.
early poems, modern legal documents) suggest that the percentage
of wor ds used which have a cognate of similar sense in ancient
Hebrew will seldom exceed 30-40 per cent.
In order to follow out this procedure more systematically, one
may se t out an area of the vocabulary of a Semitic language, e.g.
all words beginning with a certain consonant, and set against them
the cognates in Hebrew which have, or do not have, reasonably
close similarity. I have done this, for example, with the Syriac
verbs beginning with /b/, as listed in Brockelmann.3
I have listed the Syriac verbs, and set against them in several
columns a mark indicating the degree of their agreement with
cognate Hebrew verbs. If the cognate Hebrew word is close in
sense to the Syriac, the mark is in column A; if it is remote, the
mark is in column B; if a word from the same root appears in
1 So Hitzig, according to Gordis, Koheleth, p. 221.
2 Montgomery, Kings (ICC), p. 372.
3 See Appendix, pp. 305 ff.
 GENERAL 163
Hebrew, but not as a verb, the mark is in column C ; and if there
is no cognate in Hebrew, the mark is in column D. The total num­
ber of roots under which an entry was made is sixty-nine; the total
number of entries is seventy-eight. The difference is because I
have followed Brockelmann in entering two homonymous verbs in
Syriac, or else have given separate entries to separate meanings of
a Syriac verb, where these differ in their relation to meanings found
in Hebrew. I have also noted some occurrences in Mishnaic
Hebrew, which may also be relevant in several ways. The entries
by columns come out as follows:
A B C D
(sense close (sense remote (Hebrew, but (not found
to Hebrew) from Hebrew) not as a verb) in Hebrew)
26 13 9 3°

These figures cannot be applied very strictly, for a number of

uncertain factors exist: one may be uncertain what to do with
denominative verbs, one may doubt the real existence of some of
the meanings registered in dictionaries, one may be uncertain
where to place a word which deviates from the normal phono­
logical correspondences. For deciding whether a meaning in Syriac
is close to or remote from one in Hebrew, my rough criterion has
been the judgement whether one meaning is close enough to the
other for us to be able to divine the meaning rightly from the
cognate in the situation of normal philological treatments.
In spite of these qualifications the sample seems to be a reason­
ably fair one. Syriac and Hebrew are languages which one would
expect to be fairly comparable, since they belong to roughly the
same culture area and their literature is to a fair extent devoted to
similar subjects.
If the sample is a fair one, it suggests that even in quite compar­
able languages like Syriac and Hebrew the number of verbs in the
one which have closely corresponding cognates in the other with
closely similar meanings may be only about 40 per cent., while the
number of verbs which have no cognate verb at all in the other
language is actually higher than this. Another survey I have done,
of Syriac and Hebrew verbs beginning with /'/, suggested that the
number of verbs showing close correspondence in form and mean­
ing was about 25-30 per cent, of the total number of Hebrew verbs
and about 33 per cent, of the total number of Syriac verbs. Samples
164 L E X IC A L R E SO U R C E S IN T H E S E M IT IC L A N G U A G E S
with other Semitic languages confirm the general direction of this
argument. In Ethiopic verbs beginning with / ' / 1 find the number
having reasonably close agreement with Hebrew cognate verbs to
be about one in six, along with a number of places where we find
less direc t agreements (e.g. Ethiopic has cof ‘birds’ but not a verb
corresponding to Hebrew ‫) ע ו ף‬.
Thus the evidence on the whole, while confirming some con-
siderable community in the use of lexical resources in the Semitic
languages, appears not to favour a degree of overlap or coincidence
so high that the presence of a phenomenon in one language will
easily form a prima-facie case for its presence in another. The
overlap 13 somewhat higher if it refers only to roots, or to purely
formal correspondences; but the philological treatments of which
we are speaking go further than this. They involve a relation be-
tween the form-sense relationship in one language and the form-
sense relationship in another; and this is just what is not, on the
evidence from known examples, predictable in a very high degree.
At this point we may profitably recall experience with Indo-
European, where similar phenomena occur. Words may show
formal correspondence of some closeness, but not mean the same
thing. German sterben is from the same root as English ‘starve’, but
the sense is not ‘starve’ but ‘die’. At first sight one might suppose
this to be an analogy to the relation between Arabic ja a ‘be hungry’
and Hebrew ‫‘ גו ע‬expire, perish’, a familiar term of the P docu-
ment. It would be unwise, however, to accept this impression; for
sterben and starve appear to go back to a sense ‘be numb, stiff’ rather
than ‘be empty, hungry’.1
If we l ake a particular term or concept and collect the words in
that field in the various Indo-European languages (the approach
taken by Bergstrasser in his list already quoted for Semitic), we
usually find not one root in use but a large number.2 For ‘die’, for
instance, Buck lists about seven different roots. If we take sub-
divisions of Indo-European, such as the Germanic languages
alone, we find a greater community of use; but this does not apply
to the Indo-European language group as a whole. It is of course
possible 1:0 suppose that Semitic never became so diversified in this
1 This example illustrates how the logical study of features of the referent,
producing the argument that ‘if you starve you die’, misrepresents the linguistic
facts entirely; on this see below, pp. 290 if.
2 For a compendious display of this procedure see Buck, Dictionary of
Selected Synonyms.
 GENERAL 165
regard as Indo-European did; but at least it is salutary to consider
the position of the latter.
It may be briefly mentioned that the view that Hebrew is a
‘mixed language’ has sometimes been used as part of a plea for the
existence in Hebrew of a large incidence of agreements with several
cognate languages. There is some difficulty, however, in knowing
just what is meant by ‘mixed’, and I think that it is a confusing
term. I shall not, therefore, pursue the matter farther.1
It is also possible that one could find certain semantic fields in
which a higher degree of lexical community is found between
Semitic languages than in other fields. For instance, Low in his
researches on Aramaic fish-names observed that very few fish-
names of common Semitic origin were to be found.2 One might
with further research be able to list fields in which considerable
community existed, and set against this other fields in which the
degree of community was low; if this were done, then the pro-
bability of success for philological treatments would be high in the
former and low in the latter.
Finally, the scholar must bear in mind that a sense quotable for
a cognate language like Arabic may be a rare or specialized develop-
ment of that language and therefore not represent an older Semitic
stage likely to be shared also by Hebrew.3 Even if the sense ‘stay in
a place’ is a legitimate one for ganiya, is not this a doubtful basis
for an identification of Hebrew ‫( ע נ ה‬Index, no. 251), when the
major Arabic sense is ‘be rich’ (Lane, pp. 2301 f.)? Can the sense
‘advise’ be ascribed to ‫( ה שי ל‬no. 306), when the main sense of the
Arabic is ‘sign; make a sign’? Arabic bin (no. 54) is handled by
Lane (p. 288) and Freytag (i. 179~80a) in a way that suggests clear
dependence on the sense ‘between’, and it is unlikely that an
independent sense ‘region, field’ existed in such a way that it could
plausibly be attributed to Hebrew also. Does jarama mean ‘com-
plete’ (no. 79) and so justify a word for a ‘landing’ on a stair, when
the statistically dominant sense is ‘commit a crime’ and the likely
1 The 4mixed’ character of Hebrew was argued particularly by Bauer. For
developments of this kind see Driver in The People and the Book, p. 109; J T S
xxxi (1930) 275; Winton Thomas in Record and Revelationf p. 401. For a critical
note see Harris, Development of the Canaanite Dialects, p. 11 n .; for a judgement
that the term 4mixed’ is confusing, Goshen-Gottstein, ScrH iv (1958) 135.
2 4Aramäische Fischnamen’, p. 550. This circumstance tells against Low’s own
identification of a Hebrew ‫ ע מ ל ץ‬, see below, pp. 236 f.
3 Cf. already above, p. 116.
166 L E X IC A L R E SO U R C E S IN T H E S E M IT IC L A N G U A G E S
primitive sense is ‘cut off’ (wool, fruit) (Lane, pp. 412 f.)? The
identification of 1‫־‬111‫‘ ל‬be astounded’ (no. 86) depends on an Arabic
cognate ‘be stupefied’, but dahiyy and related words surely mean
‘be smart, be cunning’ (Lane, pp. 927 f.). When dana means
mainly ‘be vile, weak’ (adj. dun), is it likely to yield a Hebrew
sense of m ‘provide abundance’ (no. 87) on the limited grounds
of V ‘perfecta opulentia fruitus est vir’ (Freytag, ii. 73-74)? Though
baha is glossed as ‘cognovit, dignoviV by Freytag (i. 181b), does not
the general Arabic usage suggest a basically sexual reference, ‘lie
with (a woman)’ and the like (cf. Lane, p. 278b), which makes
unlikely a Hebrew cognate ‘remember’ (no. 53)? When kadama
really means ‘bite’, and when the sense ‘firmiter vinctus fuit (capti‫־‬
vusy ut effugere non posset)' (Freytag, iv. 18a) is probably derivative
from the sense ‘bite’, is it likely to provide a sense ‘hold fast (in
fetters)’ for a newly identified Hebrew ‫( כ ד ם‬no. 172)? Is there
enough evidence in the solitary V inbaqaa ‘he went away quickly’
(Lane, p. 235a) to justify ‘go away’ for ‫( ב ק ע‬no. 6 2)?
Thus, even when dictionary entries are quite accurate for their
respective languages, one has to consider whether senses registered
are likely to be central enough and old enough to allow of a sharing
with biblical Hebrew.
Sometimes, however, research in other Semitic languages will
confirm the probability of an early date. One might have doubted
the early date of the sense ‘mortgage, pledge’ for Arabic rahana
(no. n o ), but its antiquity is made certain by its frequent use in
Nabataean inscriptions. A Hebrew ‫‘ ח ט א ת‬penury’ has been
identified on the basis of Ethiopic hatiat. Dillmann himself (col.
621) characterizes this Ethiopic sense as ‘extremely rare’; but the
verb and other noun forms in Ethiopic tend to suggest that this is
indeed an ancient element of meaning, such as might be found in
other Semitic sources much earlier (no. 126).

(2) Biliteral Theories

The normal elementary picture of the structure of the Semitic
word is that of a triliteral pattern formed by three consonants
which, with certain known exceptions, are fairly stable in all
changes of the word. Biliteral theories, on the other hand, maintain
that the triliteral structure is itself the product of extension or
supplementation of a base of two consonants, through the addition
 B I L IT E R A L T H E O R IE S 167
of a third. The important class of ‘hollow’ verbs, like fflj? ‘stand’,
is then the residuum of an earlier state. A number of roots have in
common two consonants, but not three, and it is possible that they
represent the supplementation of a basic biliteral form with differ­
ent third consonants. Again, there is often a similarity between
verbs with two main consonants (the ‘hollow’ and ‘geminated’
types) and those which have this base supplemented by a ‘weak’
consonant (most commonly /y/ or /w/ or /n/) either before or after.
The various theories here involved deserve much discussion in
themselves. Their relevance to our theme in this book, however, is
as follows. The classical picture of development from a parent lan­
guage (‘proto-Semitic’) implies the theoretical construction of this
parent language, and the use of correspondences which reflect the
phonological changes in words the consonantal sequences of which
were already formed in the proto-Semitic period. If, however, the
supplementation of biliteral bases by the addition of a third radical
had not already occurred in any standard way in the proto-Semitic
period—if, in other words, the development of some words did not
take place by normal sound changes from pre-existing full forms,
but occurred einzelsprachlich with individual development from
earlier bases—then some questions fall upon the entire system of
comparison which has been understood up to now in this book.1
A comparative philology which operates by close control of the
phonological correspondences found in cognate languages appears
to assume that the root sequences were already firm in the ancestor.
The root being x-y-z in proto-Semitic, it will have in the various
Semitic languages in normal circumstances the normal correspon­
dents of x-y-z. Suppose, however (and this is what biliteral theories
appear to suggest), that in the proto-Semitic stage the only firm
root for this case was x-y, for which a rather vague meaning can be
given. The forms which will appear from this root will then (in
most cases) not be correspondents of x-y. They will take a variety
of patterns; in one language we shall find x-y-z, in another x-y-y,
in another x-z-yy in another z-x-y, in another x-y-p, in another
q-x-y.
In the circumstances envisaged by such hypotheses, only a
rather vague statement can be made either about the formal
1 A survey of the literature concerning biliteralism up to 1952 is given by
G. J. Botterweck, pp. 11-30; more recently, Moscati, Comparative Grammar,
PP· 7 2 5 ‫· ־‬
168 L E X IC A L R E SO U R C E S IN T H E S E M IT IC L A N G U A G E S
characteristics of the root or about its meaning. The biliteral bases
cannot be fitted into strict phonological correspondences in the
way that is pictured in Bergstrasser’s list (which applies to known
complete words). For instance, Driver, offering long strings of
examples, talks of a base gz-ks-ks-qsjs ‘sever’, for which he cites
fourteen words in Hebrew alone.1Another base is zr^sr-sr-sr-sr and
dr-tr-tr ‘surround’ or ‘turn round’. This can be evidenced from a
number of cognate languages in tables like this:
Heb:‫־‬. ‫צור‬ ‘besiege’ ‫צרר‬ ‘tie up’
‫אצר‬ ‘store up’ ‫חצר‬ ‘enclosure’
‫עצר‬ ‘restrain’ ‫עצרת‬ ‘band (of people)’
Arab‫׳‬. *asara ‘confine’ hasara ‘surround’
Accad. qasaru ‘bind’
Such tables, repeated for other forms of the manifold and vari-
able character which such bases are supposed to assume, total up
to between forty and fifty words from assorted Semitic languages,
all of which have some form of the base contained in them and all
of which can in some degree be explained semantically (or so it is
argued) from the meaning of the base.
Out of these very numerous words, however, only very small
groups, usually of two or three here and there, present the kind of
precise phonological correspondences of which we have previously
spoken, while very considerable diversities in meaning also occur.
Consider, for instance, the semantic relation between Hebrew ‫ח ד ר‬
‘room’, glossed by Driver as ‘enclosed chamber’, and Syriac hdar
‘go round’. Even if one accepts that this relation is rightly stated,
one can hardly claim that, if the meaning of one of the words was
unknown, the meaning of the other would supply an adequate
guide to its discovery.
The normal correspondences function in a kind of prediction.
Given, let us say, a normal Arabic form such as simal ‘left (hand)’,
we car! predict with fair accuracy what the normal Hebrew form or
Aramaic form would be. But, given the base zr-sr-sr-sr-sr, and even
given an Arabic word which is an instance of its use, we cannot
predict whether a Hebrew word using the same base will have a
third root consonant or what that consonant will be,2 or whether
it will use the same one (or the corresponding one) of the five forms
1 Driver, Problems of the Hebrew Verbal System, pp. 4 f.
2 ‘A ll consonants may be used as “determinants” *, writes Moscati, p. 74;
italics mine.
 B I L IT E R A L T H E O R IE S 169
of the base; nor can we tell what it would mean, except that it will
be something to do with (or something which in its past semantic
history may have had something to do with) surrounding, turning
round, going round, or being round.
If there is some truth in biliteral theories, they may help to ex-
plain certain points which have already been discussed. For in-
stance, where words in two Semitic languages do not show normal
correspondences but are very similar in meaning, it may perhaps
be possible to say that they are different expansions of the same
base. This would account for their closeness without requiring the
argument that sound correspondences in general should be treated
loosely. Similarly, some of the words which have been cited* as
instances of ‘metathesis’ might rather be regarded as different
expansions of a biliteral base, with the third consonant put in a
different place.1
It is doubtful, however, whether this argument can be used
helpfully in the context of philological treatments, where the mean-
ing of the word in Hebrew is itself not certainly known. For reasons
which have already been stated,2 such treatments are not likely to
carry cogency and conviction if phonological correspondences are
fluid and uncertain. Where the relatedness of cognates depends on
the appeal to a hypothetical common base, the chances of a con-
vincing philological treatment are not good.
Conceivably there were two sectors in Semitic vocabulary. The
first was already in a fairly fixed form in the proto-Semitic period.
The words of this sector will then appear in the historical languages
in forms showing the normal correspondences. In the second
sector, however, the root sequences were still not firm in the proto-
Semitic period, and various branches developed their words
independently later, through expansion of a vague common base.3
If this were so, philological treatments would work mainly in the
first sector of the vocabulary, while in the second they would have
a very much lower degree of probability.
1 Cf. above, pp. 96 if.
2 Cf. above, pp. 83 if.
3 An example might be the words meaning ‘naked’; the common base for this
seems to be the sequence represented in Hebrew as ‫ ; ע ל ה‬but Hebrew and
Arabic develop types like ‫ ע ר ו ם‬, and Syriac develops *rfl and Ethiopic 'rq. One
cannot help wondering whether Arabic, which may be held to have become sepa-
rate at an early stage in Semitic language separation, did not develop many of its
characteristic lexical items thereafter on its own; if this is so, its idiosyncratic
productions are not helpful for philological treatments in Hebrew.
170 L E X IC A L R E SO U R C E S IN T H E S E M IT IC L A N G U A G E S
Thus, in general, while the hypothesis of original biliteral bases
is relevant to the general assessment of philological treatments in
the Old Testament, it is nevertheless unlikely that specific appeals
to bili :eral bases will, in the present state of our knowledge about
the prehistoric stages of Semitic languages, provide an important
or independent basis for the plausibility or probability of particular
suggestions. On the contrary, emphasis on biliteral bases, carrying
with it; some implication that these bases were expanded indepen-
dently in the separate languages, tends on the whole to undermine
rather than to support the use of philological treatments.
(3) Semantic Fields
In each language words function in relation to other words in the
same or contiguous semantic field.1 The meaning of ‫ מ נ ח ה‬, for
instance, can be described only in relation to the other words exist-
ing in the Hebrew of a certain time in the field of sacrifice, gift, and
tribute. Its meaning is then a choice within the series of possibilities
available within Hebrew. Again, the sense of a familiar word like
‫״ ח ט א‬sin’ depends on its relation to other words in similar fields,
such as ‫ ר ש ע‬and ‫ פ ש ע‬, and to other words in opposing fields.
Sometimes quite systematic structures can be stated, such as the
terminology of the sacrificial system in Leviticus, or the two pairs:
‫‘ ק ד ש‬ho1y’ ‫‘ ט ה ו ר‬clean’
‫‘ ח ו ל‬not holy‫‘ ׳ ט מ א‬unclean’
What is ‫ ט ה ו ר‬is not thereby ‫ ; ק ד ש‬and what is ‫ ח ו ל‬, tradition-
ally ‘profane’, though not ‘holy’, ‫ ק ל ש‬, is not thereby in any way
wrong or evil. It is wrong, however, if the distinctions between
members of pairs are obscured, or if that which is ‫ ט מ א‬is mixed
with that which is ‫ ק ל ש‬.
Not all semantic fields are systematically diagrammed in this
way; rather than a system, there is a bundle of meanings. Even
then, however, the meanings are dependent on the meanings of
other words in the same language at the same time. Even though
languages are cognate and have a large number of individual cog-
nate words, the make-up of these bundles may be, and indeed is
likely to be, very substantially different.
This may explain why, when in language A form x-y-z is found
with a certain meaning, no corresponding x-y-z may be found in
1 Cf. above, pp. 89 f., 136 if.
 S E M A N T IC F IE L D S 171
language B. In language A it has a significant function through its
contrast with forms a-b-c and p-q-r. In language B, however, a-b-c
does not exist, while p-q-r has developed a different meaning and
therefore moved into a different semantic field, and the ground oc-
cupied by x-y-z is occupied by d-e-f and l-m-n. The method of
identifying cognates works only for individual words; but the
semantic functioning of words is within word-groupings which are
quite asymmetrical in one language as compared with another,
even when the languages are closely cognate.
We can illustrate this from the identification of the verb !HDD
interpreted as meaning ‘announce’ (hiphil) at Prov. 15. 2. M T
reads: 1‫ ·?שרן ךן>;מים תי טי ב ד ע ת‬Driver, reading the verb as
‫ ת ט ב‬or ‫ ת ט ב‬from the root ‫ ט ב ב‬, gives the sense as ‘the tongue of
the wise announces knowledge’. A sense like ‘understand’ or ‘heal’
might similarly, we may add, be proposed for Prov. 17. 22. This
root ‫ ט ב ב‬is indeed found in several Semitic languages. In Syriac
it seems to have the sense ‘know, discover’; but it occurs chiefly in
nouns like tebba ‘knowledge; story’ and adjectives like tbiba
‘famous, experienced’, and in derived verbal themes with meanings
like ‘make known’ and ‘announce’. In Arabic also it is registered as
‘know’, but the most familiar form is tablb, mainly specialized as
‘doctor’; and the modern sense of the verb is mainly ‘treat medic-
ally’. In Ethiopic it means ‘be wise’, while verb forms exist mean-
ing ‘educate, inform’.
This information, however, should be related to our knowledge
of other words in the semantic field. Ethiopic, for instance, does
not have the common root h-k-m in this field; and while it has the
root y -d -\ from which Hebrew has ‫‘ י ד ע‬know’, this is not used in
Ethiopic in the sense ‘know’, but only in the causative ‘announce,
inform’. Again, while Ethiopic has the root b-y-n> which supplies
a sense ‘know, be intelligent’ in certain contexts in Hebrew, the
sense in Ethiopic is rather different, being rather ‘distinguish,
notice, make clear’. Again, the normal word for ‘know’ in Ethiopic
is *-m-r in the theme II. 1, as has already been observed in another
connection.
Thus, though all of these elements in Ethiopic can be given
parallels and connections in other Semitic languages so long as they
are taken individually, the make-up of the total bundle is quite
different from what we find in other Semitic languages. The fact
that t-b-b is present and important in Ethiopic, therefore, has to be
172 L E X IC A L R E SO U R C E S IN T H E S E M IT IC L A N G U A G E S
related to the other relevant elements, and may be partially ex-
plained by them. The fact that t-b-b exists somewhere in the
Semitic languages by no means implies that it will necessarily exist
in a lani^uage which also has h-k-m, y -d -\ and b-y-n in the senses
in which they are found in Hebrew.
Similar changes of balance in the lexical stock can be traced in
the languages which have h-s-s in this semantic field, such as
Accadian and Ugaritic.
Another example is the organization of words for social systems,
such as government. Sarru in Assyria was the king’s title, as ‫מ ל ך‬
was in Israel. Hebrew had a word ‫ ש ר‬, cognate with the Accadian
sarru, but it not only did not mean the same thing, it does not
appear to have designated one clearly institutionalized political
office at all. Most other Semitic languages appear not to have a
cognate with sarru at all. Another term representing an institution
in Accadian was limu, limmu. Against the idea that words cognate
with this should be identified in Hebrew1 one has to set the possi-
bility that the Accadian semantic development was peculiar to the
institutional framework of Mesopotamian society. In general,
though cognates can usually be found for any of the terms for
government office in any Semitic language, the make-up of the
group of terms in any one is likely to be markedly different from
that of any other.
This point deserves very much more extended research. For our
present purpose it must, however, be left here, and its importance
can be summarized as follows. Firstly, traditional comparative
method has tended to deal with individual words and has failed to
give equal place to their function in relation to other words. This in
itself may be a weakness in the method. Secondly, the consider a-
tion of the groups of words within a semantic field in a Semitic
language may help us to understand how particular words, of
which cognates are known in other languages, may yet not be
present in the vocabulary of the one principally being studied.
Thirdly, the same consideration may show how the semantic
development of words may not follow directly from an original
‘basic meaning’ and may thus be quite other than is foreseeable even
from accurate information about cognates. Given a form in one
Semitic language, we can predict what form there may be in
another, if normal correspondences are followed; but the meaning
1 Cf. pp. 133, 254 *·
 S E M A N T IC F IE L D S 173
cannot so easily be predicted from the sense of the former, because
in both cases it is dependent on interrelations with still other words,
which interrelations are not predictable at all.
(4) Words with Opposite Meanings (’Addäd)
This section will consider briefly the position of those words
which either (a) in the same language have two opposite meanings
or (b) in one language have a meaning opposite to that which is
found in another cognate language.
For this phenomenon Arabic grammar has used the technical
term 5addäd (singular didd). In Arabic studies this is normally used
for the former of the two cases specified, i.e. where a word exists
with two opposite meanings in Arabic itself. Some recent studies,
however, such as those of Guillaume, have extended the term and
applied it where the opposite meanings are found in two different
but cognate languages such as Arabic and Hebrew.
The subject of ’addäd was considerably cultivated in the work of
Arabic philologists, and there grew up a genre of literature with
the title Kitäb al- Addäd. Sometimes opinions have become pre-
valent according to which Arabic is particularly rich in words of
this type; and one hears half-joking references to a language in
which every word can mean a certain thing and also the opposite of
that thing. Modern studies, however, have indicated that this
picture of the Arabic vocabulary is extremely exaggerated. Weil
writes:1
The opinion which has long been maintained, that Arabic, contrary
to all the other Semitic languages, contains a very large number of
such ’addäd is no longer tenable. If all that is false and all that does not
belong here are cut out of the list, there remains also in Arabic only a
small residue.
We may then abandon the conception that words having two
completely opposite meanings are extremely common in Arabic or
any other Semitic language, and that our semantic decisions can be
taken on the basis of such a view.2
1 G. Weil, Encyclopaedia of 1slam, 2nd edition, vol. i, p. 184. The first ed., p.
131, interestingly enough had itself included the false but deceptive example of
Hebrew ‫‘ ע ש ר‬riches’, Arabic ‘usr ‘poverty’; the words are not in fact cognate
(cf. Aramaic ‫) ע ת ר‬.
2 For a careful study of the Arabic words see Nöldeke’s article ‘Wörter mit
Gegensinn’; an earlier survey of the problem in Hebrew is Landau, Gegensinnige
Wörter (Berlin, 1896).
174 L E X IC A L R E SO U R C E S IN T H E S E M IT IC L A N G U A G E S
It is rather more common to find that words have a meaning in
one language opposite to that which they have in another cognate
language. Arabic wataba commonly means ‘spring up’, while the
corresponding Hebrew ‫ י ש ב‬means ‘sit’ or ‘sit down’. There are
Arabic dialects (Himyarite) which are said to have the same mean-
ing as the Hebrew. Hebrew itself has the interesting ‫ !י ש ב‬at
Gen. 15. 11, where the meaning seems to be that Abraham ‘started
up’ or ‘set in flight’, i.e. ‘scared away’, the birds. LXX has
ovv€Kdd‫׳‬,G€vy following the ordinary sense of ‫י ש ב‬. Aquila, though
usually an etymologizer, renders with aTreaofirjcjev, which is the
right sense; the Vulgate similarly has abigebat and the Syriac
makkes. The Targum has ‫‘ א ת י ב‬turned away’, a word which
probably seemed close in form to the Hebrew; but some MSS.
have ‫ א פ ר ח‬, ‘made to fly (away)’, giving the sense but not the
word-similarity. Rashi understood the word as from the verb
‫‘ נ ש ב‬blow’.
The existence of words, where apparently obvious cognates yet
had opposite senses, was already known to the medieval lexico-
graphers. A familiar instance was ‫‘ א ב ה‬be willing’, which means
the opposite of the cognate Arabic *aba (y) ‘refuse’.1
The development of opposite meanings may often have occurred
through semantic change. A word having a certain range of mean-
ing, the middle of the range drops out of use (perhaps through
being taken over by another word altogether), leaving two extremes,
which we may or may not have clues to connect. This is another of
the main causes of homonymy, in addition to phoneme merger and
the adoption of loan-words, discussed already above.
The following examples have arisen in the context of modern
philological treatments:
The verb ‫ י ל ל‬usually means ‘go down’, but it is sometimes said
that it also occasionally means ‘go up’.2 Driver says that it some-
times means ‘go south’, while ‫ ע ל ה‬in the same circumstances
means ‘go north’; in any case there are places where some special
explanation seems to be needed, e.g. Judges 11. 37, where Jeph-
thah’s daughter says ‫ ו י ך ל ה י ע ל ~ ה ה ך י ם‬. Cf. also David’s ‘going

1 See, for instance, Wechter, Ibn Barun, pp. 56, 61. Similarly contrasting
meanings have been pointed out for familiar words like ‫ ל א ה‬and ‫ ; י כ ל‬see
Driver, C M Ly p. 158; Botterweck, Triliterismus, p. 39.
2 See KB, s.v., and literature there cited; Driver, Z A W lxix (1957) 7 4 7 ‫; ־‬
Leslau, Z A W lxxiv (1962) 322, who gives the support of Ethiopian parallels.
 W O R D S W I T H O P P O S I T E M E A N I N G S ( ,A P D A D ) 175
up’ from Maon to Engedi, I Sam. 24. 1. Driver, discussing this
instance, cites some other paradoxical phenomena, such as Arab.
fa ra a , said to mean either ‘go up’ or ‘go down’ a mountain.
Again, Kopf, discussing the meaning of ‫ ב ט א‬in Hebrew, says
that if this means ‘thoughtlessness, over-hastiness’, and if we com-
pare Arab, batua ‘be slow, hesitate’ (}L·), then this is an example
of opposed senses in the two languages.1
Discussing ‫‘ ש ר ב‬heat’, Driver2 takes note of an Accadian sarbu
‘shower’, surubbu ‘(cold) fever’, suribu ‘cold’, and says that the con-
nection between the senses of cold and heat is illustrated by Lat.
urere.
Going back to the Middle Ages we may remark that Saadia,
discussing the difficult ‫ ג י ל ו ב ר ע ל ה‬of Ps. 2. 11,3 argued that
the peculiarity could be explained by use of the analogy of the
Arabic i-r-i, which serves as an expression of both fear and joy.4
Perhaps the most striking example, however, of the way in which
philological treatments have produced an alleged double opposition
of meaning within Hebrew itself, is the familiar word ‫ ב‬, normally
supposed to mean ‘in’. Ugaritic evidence in particular is interpreted
with the sense ‘from’. This sense, says Gordon, is common, and is
‘in accordance with a Hamitic-Semitic feature whereby preposi-
tions meaning “in” or “to” tend to connote also “from” ’.5 There
are indeed places in Hebrew where the sense ‘from’ would at first
sight appear to make good or better sense, e.g. Ps. 68. 19, ‫ל ק ח ת‬
‫□ מ תנו ת ב א ד‬, which would then mean ‘thou hast received gifts
from men’ (so easy a solution that it may be too facile). There has
been no shortage of voices assuring us that Ugaritic evidence
makes it clear that ‘from’ is the actual meaning, all difficulty being
thereby removed. There are, however, certain objections against
this course of argument.
The first is the question of the communicative efficiency of a
language in which the word for ‘to’ and the word for ‘from’ are
apparently identical. This is, as we have seen, one of the obvious
problems of homonymy. Possibly Ugaritic b, when meaning ‘from’,
might have had a different vocalization from b with the sense ‘in’,
1 V T viii (1958) 165.
2 J T S xxiii (1922) 410.
3 Cf. above, p. 5, and below, p. 284.
4 See Eppenstein, op. cit., p. 9. The sense given by Wehr for tariba in modern
literary Arabic is *be moved (with joy or grief)’.
5 UH, p. 217. Cf. also Gordon’s detailed discussion, XJHy pp. 83 f.
176 L E X IC A L R E SO U R C E S IN T H E S E M IT IC L A N G U A G E S
so that no homonymy existed. Again, it is possible that ‘from’ is not
the sense of b in itself but only the sense of b when in collocation
with certain other words, which supply the guidance to select the
sense ‘from’. If this were so, it might follow (a) that the sense ‘from’
could not be defined as such apart from such contexts, and (&) that
the relation of b to such contexts might be confined to Ugaritic and
certainly could not automatically be transferred to Hebrew. For
instance, one of the most convincing cases for the sense ‘from’ ap­
pears i:a tb' bbth 4they departed from his house’. But perhaps the
sense of b here is defined by collocation with tb' ; and since tb' does
not have a cognate in Hebrew we cannot be sure whether this is
relevant for comparison with Hebrew at all. In trd b'l bmrym spn
or td yi$t bbhtn, which Gordon cites with translations as ‘drive Baal
from the heights of Sapan’ and ‘the fire went away from the houses’,
a great deal depends on the exact usage and sense of the verbs,
which may be beyond our power to determine.
Secondly, the whole setting of b in Ugaritic is not comparable
with the setting of D in Hebrew, for the simple reason that Hebrew
has ]Ü ‘from’, while Ugaritic does not. The meanings of words
are fun ctions of choices within the given vocabulary at one time;
and in Hebrew the choice between D and ]D furnished an oppo­
sition entirely lacking in Ugaritic in the literary texts. In any one
language the meaning of the prepositions is a highly subtle, diffi­
cult, and idiosyncratic structure of possibilities and choices, which
cannot be broken into by clumsy assertions that a cognate language
has another meaning for a given item. Some Semitic languages
have bn for ‘from’, but not mn\l some Ethiopian languages have k
with ti e meaning of ‘from’ but not with the familiar Semitic sense
of ‘as, like’. In each case the differences between prepositions de­
pend on the total stock in each individual language, and these
meanings cannot be arbitrarily shifted around from language to
language merely on the grounds that the languages in question are
cognate.
Thir dly, some question may be raised about the kind of ‘mean­
ing’ which attaches to the sense ‘from’ for b in Ugaritic itself. Not
1 South Arabian bn has been regarded as a product from the common mn by
phonetic change; so Brockelmann, Grundriß, § 252b r) (p. 497), § 84c a (p. 226).
But the existence also of In and hn may suggest rather that ‘the whole ESA series
bn, In, hn are morphologically only enlarged forms of b /‫־‬, h to which the
differentiated meaning “from” has been attached’. (So Beeston, Descriptive
Grammar, p. 57.)
 W O R D S W I T H O P P O S I T E M E A N IN G S ( ,A p D A D ) 177
all cases which have been quoted are incapable of interpretation
in another way. T h u s:
stym bkrpnm yn bks hrs dm 'sm
is translated by Gordon as:
‘drink wine from jars, the blood of vines from cups of gold’.
But it is not wholly impossible to consider the sense ‘in’; one may
drink ‘in’ or ‘with’ a cup, as well as ‘from’ one. This can apply also
to bph rgm lys a bspth hwth. While it seems to us natural to speak of
a word going forth ‘from’ the mouth or lips, it is not impossible to
see the sense as ‘in her mouth (or, by her mouth) the word goes
forth’ etc. The insistence of scholars on the sense ‘from’ may
sometimes rest on no more secure foundation than the fact that an
English translation will use the word ‘from’. But a translation can
give a correct general rendering of a passage, without providing
in its equivalences a correct understanding of particular lexical
items.
These paragraphs are not written in order to deny that the
meaning of Ugaritic b may be stated as ‘from’. They do, however,
indicate some of the problems implied if it is really thought that in
Ugaritic the same word could freely and equally mean either ‘to’
or ‘from’; and they suggest that, whatever is true for Ugaritic, the
position is likely to be different in Hebrew because the presence of
]ft in the latter language makes the whole network of prepositional
meanings quite different. These points have commonly been neg­
lected when prepositional meanings discovered in Ugaritic have
then been indiscriminately discovered in the Hebrew Bible; the
same applies to the case of L
Such, then, are some difficulties which may arise when hom­
onymy takes the acute form of completely opposite meanings for
the same form, or when forms which correspond in cognate lan­
guages nevertheless display quite contrary senses. In general, if it is
supposed that cognate words are yet likely to have quite contrary
senses, this consideration reduces the probability of success with
philological treatments on any large scale. Such treatments imply
that, where a cognate form is discovered, its (known) meaning will
suggest that of the (hitherto unknown) Hebrew word. If it is to
suggest this meaning only by a reversal of its own meaning, the
process is likely to degenerate into guesswork.
178 L E X IC A L R E SO U R C E S IN T H E S E M IT IC L A N G U A G E S

(5) Patterning of Roots and Compatibility of Consonants

In the familiar triconsonantal root pattern Semitic languages do
not tolerate any three phonemes in any order, but only certain
combinations. The rules under which consonants are compatible
differ, however, to some extent between the various languages.
(1) Certain combinations do not occur, or are very rare. While it
is common for the same consonant to occur in positions 2 and 3, it
is not generally tolerated for it to occur in positions 1 and 2. In
positions 2 and 3, according to Moscati,1 the same consonant may
be found twice, but not two different consonants with a similar
point of articulation (e.g. Hebrew ‫ ב‬and D). Exceptions to these
restrictions are more frequent in nouns, e.g. ‫ ל י ל ה‬, ‫ שמ ש‬, ‫ ; ש ר ש‬a
number of noun patterns find their way into verbs through the
formation of denominatives, e.g. Hebrew ‫‘ ש ר ש‬uproot; take root’.
(2; The restrictions on compatibility may have somewhat broken
down in the course of time. Greenberg, whose article is the starting-
point for recent research,2 tries to detect distinctions between the
restrictions attributable to the ancestor language and those ob-
served in historical times; he adds a control comparison with
Egyptian. Similarly, words of Aramaic origin may eventually have
disturbed the patterns observed in native Hebrew.3
(3) Rules of compatibility are not identical in all Semitic lan-
guages. According to Moscati,4 ‘in Accadian /g/ and /z/ are never
found in third position, nor can all three radicals be voiced’. If this
is true, Accadian will not contain a word corresponding to Hebrew
‫ ב ג ל‬under the normal correspondences for individual consonants.
Similarly, Moscati says that in Hebrew the sequences /t-q/ and
/q-t/ are compatible, but not /q‫־‬t/; while in Arabic /q-t/ is found,
but iiot /t‫־‬q/.4 This does not quite agree with Greenberg’s tables,
which show one case of /t-q/ in Arabic in positions 1-2, and four
in 2-3. It may be remarked in passing, in any case, that the study
of compatibility rules may give a much better account of the phen-
omena which were traditionally handled under assimilation, dis-
similation, and metathesis; thus the difference between Hebrew
1 Moscati, Comparative Grammar, pp. 74 f.
2 Word vi (1950) 162-81.
3 The calculations of Koskinen, ZD M G cxiv (1964) 16-58, eliminate certain
words from consideration on the grounds that they are *clearly* of Aramaic
origin.
4 Moscati, pp. 74 f.
 R O O T S A N D C O M P A T IB IL IT Y O F C O N S O N A N T S 179
‫ ק ט ל‬and Arabic qatala is to be ascribed to differing compatibilities
rather than to an ‘assimilation’. We should then have an approach
to general pattern systems, instead of lists of individual assimila-
tions, dissimilations, and metatheses; and these lists could not be
used as an argument in favour of exceptions to general linguistic
statements.1
Again, Hebrew has no verb with H in position 2 and a sibilant in
position 3.2 But in Arabic there are several (in Greenberg’s tables
4 before /§/, 1 before /s/, 2 before /z/, none before /s/: a total of 7
out of 127 verbs with /h/ in position 2). Similarly, according to
Koskinen /'/ and /h/ can occur in Arabic and Syriac in the same
verb (e.g. /*-h-d/), but not in Hebrew (nor Ugaritic and Palestinian
types of Aramaic).3
Again, in positions 2 and 3 Hebrew greatly prefers the sequence
/1-p/ over the sequence /l‫־‬b/. No true verb has the latter: ‫ ח ל ב‬and
‫ כ ל ב‬are noun roots only, ‫ ש ל ב‬is probably a denominative verb,
and ‫ ג ל ב‬is commonly taken to be a loan-word. With /1-p/ real
verbs seem to include about seven or eight; this depends on how far
roots are distinguished: ‫ א ל ף‬, ‫( ל ל ף‬two roots ?), ‫( ח ל ף‬two roots ?),
‫ ס ל ף‬, ‫ ע ל ף‬, ‫ ש ל ף‬. The ratio in Hebrew is thus six, seven, or eight
to zero.4 In Arabic verbs, however, Greenberg’s corresponding
figures for /£/ and /b/ are sixteen and fifteen. If these figures are even
approximately right ,we should not expect to find in Hebrew words
cognate with familiar Arabic words like talaba ‘slander’, jalaba ‘get,
gain’, talaba ‘seek’—not, at any rate, with the regular sound cor-
respondences.
Thus in general there appear to be some differences in statistical
preference, either for one consonant as against another, or for one
position of the same consonant as against another, between cognate
Semitic languages.5
An interesting illustration can be provided from Ugaritic.
Ugaritic, like Arabic, has a phoneme which we mark as /t/, and
its normal correspondent in Hebrew is ‫ ש‬. But the statistical
1 Cf. the discussion above, pp. 96 if.
2 Koskinen, p. 57. 3 Koskinen, p. 29.
4 Koskinen, p. 55; his reckoning, including some non-verbal roots, is not
identical with mine, and is less suitable for direct comparison with Greenberg.
5 Some remarkable differences emerge from a direct comparison of Koskinen’s
figures with Greenberg’s; e.g. /t/ is in Hebrew least common in position 3, but
in Arabic it is most frequent in that position. But little can be done with this until
exactly comparable figures are worked out.
180 L E X IC A L R E SO U R C E S IN T H E S E M IT IC L A N G U A G E S
frequency of /t/ in Ugaritic seems to be different from what it is in
Arabic. By Greenberg’s figures, in Arabic /t/ occurs in first position
with exactly the same frequency as does /t/, namely fifty times. In
Ugaritic verbs the frequency is different. Driver’s glossary registers
eighteen Ugaritic verbs with /t/ in first position, and this excludes
special denominative types such as derivatives from numerals, e.g.
tnn ‘do a second time’. For verbs with /t/ in first position, the
corresponding figure is only eight.1
This might be accidental, but a real difference in root patterning
between Ugaritic and Arabic is more probable. A notable instance
is the verb tt' ‘fear’, which is cognate with the Phoenician ‫ ש ת ע‬and
has served to identify ‫ ש ת ע‬in Hebrew also. The sequence /t-t/ in
positions 1 and 2 is not a compatibility allowed in Arabic; Green-
berg registers no instance at all. In Hebrew, on the other hand,
where there is no phoneme /t/, a word like ‫ שת ע‬falls within the
very frequent type of words with ‫ ש‬in position 1, followed by a stop
in second position (cf. especially ‫?צתל‬, ‫ שתם‬, ‫) ש ת ק‬. Clearly roots
may come to be used in one branch of the Semitic languages when
in another they do not fall within the accepted patterns.
One reason for this may be the very different phonetic realization
in each language of the phonemes to which we apply the same
conventional signs (such as /t/) and through which we establish our
comparative correspondences. Though Arabic /f/ and /j/ ‘cor-
respond" to Hebrew D and ‫ג‬, their phonetic realization is very
different, and analogous differences in ancient times may have
affected the degrees of compatibility.
To sum up, it is not likely that these considerations will fre-
quently provide a direct criterion for the evaluation of philological
treatments. No excessive reliance should be placed upon the
statistics until Hebrew lexicography has been tested afresh in this
respect. Nevertheless the rules of patterning and compatibility are
a matter of real importance, and they probably affected the make-up
of the different Semitic vocabularies. In addition to the semantic
considerations which have already been touched on,2 they suggest
some reason why non-coincidences in lexical stock exist on the
formal level also. They show how each of the languages has a

1 Even if we identify the words in a way different from Driver’s, the results
are in this regard not very different; e.g. Gordon’s positive identifications of
verbs in UH seem to be about z with /t/ in position i and io with /t/.
2 See above, pp. !7off.
 R O O T S A N D C O M P A T IB IL IT Y O F C O N S O N A N T S 181
certain individuality, which cannot be broken down on the ground
that other languages are cognate and that many individual words
have close correspondences of form arid meaning. Conversely, they
permit no excessive optimism that a known form and meaning in
one language will guide us directly to the meaning of a mysterious
form in another.

(6) Words Known Through Personal Names

A number of words which are well known in other Semitic
languages appear in biblical Hebrew only in personal names.1
These words seem to constitute a prima-facie argument for the
idea, which we shall see to be probable on other grounds, that the
vocabulary of ancient Hebrew was substantially larger than that
which is found in the Bible. Sometimes we might hardly know the
meaning of these elements in Hebrew names but for the evidence
of cognates, so that they seem to represent a primary success for
a philological approach.
This list is chosen to illustrate some words of fairly general
interest; it does not include words of narrowly specific types such
as names of animals or plants. The words are listed in the order of
the root as it would normally be quoted in Hebrew, assuming the
interpretation cited to be correct.
‫‘ או ש‬give’, found in the name Jehoash^KirP), also in the Elephantine
form ‫ ; י או ש‬Noth, p. 171. Some derive the name Josiah also from
this; for an alternative see the item two below.
‫‘ א ש ל‬rejoice’, found in the name Asarel (‫) א ש ל א ל‬, interpreted by
Noth, p. 183, on the basis of Ar. 9asira, as ‘God has filled with
j°y’·
‫‘ א ש ה‬heal’, which may perhaps be found in the name Josiah (1‫; ) י א ש י ה‬
Noth, p. 212, prefers this to the explanation through ‫ ש‬1‫‘ א‬give’;
it gives a name parallel to .‫י ר פ א ל‬
‫ ב ר ע‬The name Beriah (1‫ ) ב ר י ע ך‬is taken by Noth as related to Ar.
bora a ‘excellence’; p. 224 n. Contrast popular etymology from
‫ ר ע‬, I Chron. 7. 23.
‫‘ ד ר ק‬hard’—the name Darkon (]‫ ; ) ר ר ק ל‬so taken by Noth, p. 225, on
the basis of Ar. darq ‘hard’ (?—so Freytag, ii. 24 b).
1 Some names which occur in the Bible are names not of Israelites but of
Midianites and others (cf. ‫ א ל ד ע ה‬, above, p. 23); linguistic elements contained
in such names, even though genuinely Semitic, are not necessarily evidence for
actual Hebrew.
182 LEXICAL RESOURCES IN THE SEMITIC LANGUAGES
‫הדה‬ ‘lead’, in the name Jahdai (‫ !ד י‬1‫ ;) י‬see Noth, p. 196. ‫ ה ד ה‬at
Isa. 11. 8 may be the same word; but if so, then the sense (‘to
put’ the hand) is rather different from that suggested by the
name (divine guidance, cf. Ar. mahdi ‘guided one’).
‫‘ ז ב ד‬give’, Noth, p. 46 f.; common in names, e.g. Elzabad,
Zebadiah, but very rare before the exile (one under David and
two during the monarchy). The verb, and also the noun ‫ ז ב ד‬,
occur elsewhere only in the explanation of the name Zebulon,
which of course is not from this verb, though the words have a
similarity (Gen. 30. 20).
‫‘ ז מ ר‬protect’, Noth, p. 176. The familiar biblical name is Zimri
(‫) ז מ רי‬, but the Samaritan Ostraca provide names with the fuller
form such as ‫ ב ע ל ז מ ר‬and ‫ ז מ רי ה ו‬. The cognate verb is com-
mon in South Arabian names.
‫‘ ח מ ה‬protect’, found in names like Jahmai (‫ ;) י ח מ י‬Noth, p. 196 f.
Cf. the noun ‫‘ חו מ ה‬wall’, but no verb; see above, p. 160.
‫‘ נ מ ה‬bring tidings’, suggested by Guillaume to fit the Qumran
reading at Isa. 41. 27 (see below, p. 193). The name involved
is Nemuel (‫) נ מ ו א ל‬. Against the idea that this name contains a
verb ‫ נ מ ה‬there are two arguments: (1) this gives no sense
paralleled in Israelite names (2) the forms ‫ א ל‬1‫ י מ‬and ‫ א ל‬1‫ ל מ‬,
which look related, count against a derivation from .‫נ מ ה‬
‫‘ עו ש‬help’, in the name Jeush (‫ ש‬1‫) י ע‬, Ar. gâta; Noth, pp. 176,
196. This verb actually occurs in the MT at Joel 4. 11, but
the text has been doubted (e.g. BDB, p. 736a), perhaps un-
necessarily. The sense ‘help’ seems to have been unknown to the
ancient versions (arwaOpoi&crdey ‫י ת כנ שון‬, erumpitè).
‫עמר‬ ‘live’, in the name Omri (‫ ; ) ע מ ר י‬Noth, p. 63, who mentions that
this name might be of Arab origin. The cognate is the frequent
Ar. 'amara ‘live’.
‫ערש‬ ‘plant’, suggested by Noth, p. 203, for the name Jaareshiah
(‫)י ע ר שי ה‬, on the basis of Ar. garasa and Accad. erësu, the latter
of which is extremely frequent in personal names.
‫עתל‬ ‘tall’ (so Noth; or perhaps ‘carry’, Ar. catala?); found in the
name Athlai (‫) ע ת ל י‬, and cognate with Accad. etellu ‘high’,
(‘prince’?); Noth, p. 191.
‫‘ קלת‬nourish’, found in the name Jekuthiel (‫ ; )י ק ל תי א ל‬Noth, p. 203.
Cognate with Ar. qdta. Cf. the place name ‫ י ק ת א ל‬, b d b ,
p. 430,
 W ORDS KNOW N TH R O U G H PER SO N A L NAM ES 183
‫תלה‬ ‘protect’. Noth’s interpretation of the name Ithream (‫) י ת ר ע ם‬,
p. 197, on the basis of Accad. taru (Bezold). Noth admits the
possibility of a relation rather with the known Hebrew root ‫י ת ר‬
‘be more, be left over’, and this is on the whole more likely.
If these identifications are right, they seem to be indubitable
evidence of Hebrew words beyond the normally recognized
vocabulary. Some of these names were used quite frequently.
There were three Omris besides the great king of that name, and
a whole host of Zebediahs, Zabdiels, and others with names from
the root ‫ ז ב ל‬. Nowhere would philological solutions be so com-
pelling as when Hebrew personal names themselves provide
evidence for the words suggested. Yet, of the list given above, few
examples have figured in philological treatments known to m e; and
some of those which do so figure, such as ‫ נ מ ה‬, seem to be rather
precarious ones. The recognition of new words from the cognate
languages in Israelite personal names themselves does not, as one
would expect, lead to clear identifications of related words in the
running text. Can some other explanation be found ?
The fact of poetical parallelism should also be taken into account.
Of the words listed above a number relate to concepts quite com-
mon in Hebrew literature—‘give’, ‘nourish’, ‘heal’, ‘protect’, ‘help’,
and ‘live’. We should expect that unusual words might appear as
second elements (‘B‫־‬words’) in parallelisms, even if they were not
used independently of such contexts. This is particularly so where
there is no adequate normal parallel for a word. Thus ‘give’ is very
common in poetry, yet there is no adequate common parallel for
‫ נ ת ן‬. It is surprising that ‫ ז ב ל‬was not used with some frequency
to provide such a parallel. The same can be said of ‫‘ ע מ ר‬live’,
which would give an excellent parallel for the heavily-worked ‫ ח י ה‬.
Thus the evidence of personal names hardly encourages us to
suppose that, simply because a word may have had currency in
Israelite onomastics, it may therefore confidently be expected to
be found in the running text of the Bible. The non-usage, or very
limited usage, of plausible terms like ‫ ע מ ר‬and ‫ ז ב ל‬tends to lead
us to the opposite conclusion, surprising as it is: that, even when a
good explanation from cognates is available and even when usage
in personal names is quotable, the lexical stock used in the biblical
text was rather closer to the traditional picture than to that which
would be constructed by the addition of numerous new words
deduced from cognate parallels.
184 L E X IC A L R E SO U R C E S IN T H E S E M IT IC L A N G U A G E S
We may suggest a reason for this. It is possible that certain words
existed in Hebrew but were used only in the formation of proper
names; or, alternatively, that they were used, but only in other
reg: sters of the language than those which found their way into the
biblical text. This is, of course, no more than a suggestion, which
may not be capable of proof. I would mention, however, one
analogy which, though not complete, is at least partially valid. The
‘Amorite’ names are names which notably diverge from the stan-
dard Accadian of the running texts in which they are found; so
that in these materials the names are ‘Amorite’ though the general
language is not. It may be that Israelite personal names used
lexical elements which were not general in Hebrew; and possibly
social and linguistic history might provide some such explanation.1

(7) Lexicostatistics or Glottochronology

Finally, some mention should be made of the recent approach
known as lexicostatistics or glottochronology.2 The basic concep-
tion is this: since the rate of change of vocabulary is fairly constant,
the study of changes between cognate languages in a basic central
word list will indicate the chronology of the separation of the differ-
ent cognate branches from their ancestor. The procedure involves
several assumptions, which have been partly tested in certain
linguistic areas, and in particular:
(a) the assumption that some parts of the vocabulary are less
subject to change than others; this enables the investigator
to set up a basic core vocabulary, the rate of change of which
is used in the calculation;
(b) the assumption that the rate of retention in the core vocab-
ulary is constant through time; thus a certain percentage of
words will be lost every thousand years;
(c) the assumption that the rate of retention and loss of basic
1 Noth lays a heavy emphasis on an Aramaic stratum; e.g. Personennamen,
pp. 4 3 1 7 1 ,7‫ ־‬n., 176 n. One might speculate that such a common Aramean
background in Israel’s prehistory has left behind in personal nomenclature some
words which ceased to be general in other usage. We may compare names like
Jacob (‫)י ע ק ב‬, which surely meant ‘(God) protect’ but which is not accompanied
by any free use of the verb in this sense in the biblical text.
2 See Sarah C. Gudschinsky, ‘The ABC’s of Lexicostatistics (Glottochron-
ology)’, Word xii (1956) 175-210, reprinted in shorter form in Hymes, Language
in Culture and Society, pp. 612-22; and cf. the bibliography of the subject there
also, pp. 622 f.
 L E X IC O S T A T IS T IC S OR G L O T T O C H R O N O L O G Y 185
vocabulary is approximately the same in all languages.
According to Gudschinsky, this has been tested in thirteen
languages, and the results range from 86*4 to 74*4 per cent,
of words retained per thousand years. All but two of these
thirteen, however, were Indo-European languages.
The procedure is then to set up a basic core list of words and
compute from the incidence of cognates in a pair of languages the
date of division of the primitive speech community to which both
belonged. Needless to say, the technique is of particular interest
for groups of languages which are known to be cognate but of which
no historical records over a long period of time exist.
A good deal of uncertainty appears still to surround the approach
of glottochronology, and I do not suggest that it is a certainly valid
one. In any case its object is in principle a different one from that
of the present study. We are not trying to establish the date at
which the proto-Semitic ancestor speech community broke up,
but rather to establish the degree to which cognate elements can be
expected to appear in the various Semitic vocabularies. Thus it is
by no means my purpose to argue for the rightness of the approach
of glottochronology.
Nevertheless this approach may well be suggestive for us. Even
the question of the date of separation of the proto-Semitic language
community is not without some interest for us. Where scholars
speak as if a very high degree of community in the use of lexical
resources existed between Semitic languages, one sometimes has
the impression that they suppose the proto-Semitic stage to be
chronologically not very greatly removed from the appearance of
the first linguistic documents—rather, indeed, as if the story of the
Tower of Babel had rightly portrayed the remoteness of the origin
of linguistic diversity! If the diversity of the Semitic group goes
back not four or five thousand but ten or twenty thousand years—
and it is at least conceivable that the approach of glottochronology
might demonstrate this statistically—our expectancy for the find­
ing of cognates in Hebrew because a form exists in another Semitic
language must be somewhat reduced.1
It may be that the discussion of glottochronology may prove
1 Cf. Driver’s statements on the closeness of the Semitic languages to one
another, quoted above, pp. 36 f. The Romance languages, with which comparison
is there made, can be traced from a known ancestor over a very limited period.
Driver’s argument appears to disagree with the approach of glottochronology.
186 L E X IC A L R E S O U R C E S IN T H E S E M IT IC L A N G U A G E S
stimulating in the following regards. Firstly, the idea that there is a
core vocabulary, in which the rate of change is lower than for other
words, might fit with the observation we have already made in this
chapter, namely that while one sector of the vocabulary can be
found to have a high degree of agreement in many Semitic lan­
guages, outside of this sector other sectors can be found which have
much greater non-agreement (e.g. words for going in a direction—
up, down, out, and so on).
Secondly, however, it might be that the contents of such a core
vocabulary would have to be set up differently for Semitic from the
group which has been used for other languages.
Thirdly, it is conceivable that, for reasons which we cannot at
present decide, the rate of vocabulary change in Semitic languages
has not been the same as that found in the study of other language
groups.1
Thus the approach of glottochronology remains potentially
constructive and actually suggestive for our problem. Our study
has tended to suggest that, while philological treatments of the Old
Testament text have assumed a high degree of community in the
use of lexical items, the degree actually observable in the Semitic
languages may be somewhat lower. Where languages are cognate,
only a limited percentage of their vocabulary can be expected to
occur in cognate and corresponding forms. This has been indicated
from other observations already in this chapter; the difference
made by glottochronology is to relate this to a statistical scheme of
probability over a chronological scale. If, as Gleason indicates,2 the
rate; of loss of items is about 19 per cent, per thousand years for the
core vocabulary, then two cognate languages after a thousand years
will probably have about 66 per cent, of the stock in common, and
44 per cent, after twice that period. Since the rate of loss outside
the basic core vocabulary is even higher, the probability that
unusual and obscure items in one language can have their sense
pre dicted on the basis of another is not statistically very high.
It remains to summarize the results of the discussion in this
1 Some limited application of the approach through glottochronology to
Arabic dialects will be found in Hymes, ‘Rate of morpheme decay in Arabic*,
IJ A L xxv (1959) 267-9.
2 Gleason, Introductionf p. 450. The calculation assumes the most probable
event, i.e. that the two languages will not lose the same item s; language A
will retain 81 per cent, of those retained by language B, and also 81 per cent,
of the 19 per cent, lost by language B.
 L E X IC O S T A T IS T IC S OR G L O T T O C H R O N O L O G Y 187
chapter. The discussion has not been conclusive, and has not
produced any clear and definite estimate of the probability that
lexical elements found in one Semitic language will be shared by
another. It has shown, however, that caution has to be exercised in
this matter, and that it is premature and injudicious to assume that
the presence of material in one Semitic language forms a prima-
facie case for its existence in another. Comparative study, when
directed squarely upon the problem, itself shows that this is not so.
Thus, in arguing against too quick an assumption of lexical overlap
between the Semitic languages, we are not arguing against the
comparative method, but applying it properly, and applying it to
the lexical stocks as wholes, rather than to individual items within
them. To put it negatively, a hyper-comparative approach, which
has been overconfident in a high assessment of the degree of lexical
overlap between languages, has been damaging to a truly compara­
tive understanding of the situation.
 VI I I

THE MASSORETES, VOCALIZATION

AND EMENDATION

(1) General
M any philological treatments hold firmly to the consonantal text,
resisting suggestions that it should be emended; they are extremely
free, however, towards the vocalization, often implying that it is a
late and ill-informed interpretation which may be modified by
scholars at will.1
Exceptions, indeed, can be found. For the enigmatic ‫ו א ל ~ ת ר ה ו‬
at Isa. 44. 8 Driver’s suggestion of an explanation on the basis of
Arabic daha implied an emendation to ‫ ו א ל ~ ת ך ה ו‬affecting the
consonants only, or even to ‫ ו א ל “ ת ד ה ו‬affecting also the vowels
and making the verb a niphal.2 In this approach the philologist
uses his imagination to detect forms close to, but not identical with,
the consonantal text, which promise a successful philological
treatment.
Conversely, a philological treatment will sometimes start from,
and use as evidence, a feature of the existing pointing; and, after
using a cognate language to explain the meaning, it will return to
the existing pointing, which it has thus explained and justified.
At Neh. 5. 7 Kopf notes the unusual niphal in the phrase ‫וי מ ל ך‬
‫ ל בי ע ל י‬and, comparing the Arabic sense, ‘take possession’,
construes, as ‘I was beside myself’. Though the clue to the sense is
found in Arabic, the suggestion starts from and returns to the
Massoretic pointing.
A similar example is found at Hos. 8. 4. Here the verb ‫ מ ל ך‬is
taken by Driver to mean ‘advise’, after the sense in Aramaic. The
verse reads:
: ‫ ל כ ו ו ל א מ&ני ה ע ד רו ו ל א מ ־ ע תי‬$ ‫ה ם ה‬
1 Cf. Driver, cited above, p. 35; Dahood in Biblica xliv (1963) 291, with
reference to Albright in Peake, pp. 62 f.
2 Cf. above, pp. 6 f., 166, and below, p. 231.
 GENERAL 189
and the meaning found by Driver is:
‘they have taken counsel, but not of me;
they have got advice, and I know not of it.’
The first verb means ‘caused advice to be taken’, thus justifying
the hiphil; and the second is related to Arabic ’asara and is in
Hebrew a denominative hiphil meaning ‘obtained advice’. Thus,
though the senses suggested are novel, the interpretation supports
and justifies the hiphil pointing of the MT.
These illustrations show, then, that philological treatments can
and do at times either (a) involve emendation of the consonantal
text or, conversely, (b) confirm the punctuation as well as the
consonantal text. It nevertheless remains generally true of philo-
logical treatments that many of them involve a departure from the
Massoretic punctuation. The exceptions can hardly hope to become
the rule.
Moreover, many philological treatments, though they do not
abandon or emend the major consonants (generally speaking, the
radicals, apart from ‘weak’ letters), nevertheless involve other
changes (commonly supposed to be ‘minor’), such as displacements
of word divisions and confusion between w and y. If these ‘minor’
changes are used with high frequency in a small space their effect
is not greatly different from that of outright emendation. When this
is so, it is legalistic to claim that the consonantal text is being left
intact. Even if basic root consonants are not altered into others,
the text may be in effect rewritten. The following examples will
show different degrees of this process. Moderate changes are
involved in the quite attractive treatment of Ezek. 27. 19 by
Millard. The M T offers the rather impenetrable:
‫; ו ךן ד ו ן ??אוזל‬
Of this AV made
‘Dan also and Javan going to and fro.’
B H 3 emends, so also RSV. Millard takes the text as:
‫ ו תי _יין מ אי ז ל‬:
meamng:
‘and casks of wine from Izalla’.1
1 y S S vii (196a) 201 if. Cf. the identification of Izalla already in GB, p. 15 b.
190 T H E M A SS O R E T E S, V O C A L IZ A T IO N A N D E M E N D A T IO N
The word ]‫‘ ל‬cask’ is identified on the basis of Accadian dannu
(cf. also Ugar. dn). The textual alterations are fairly minor.
Other suggestions are more complex and involve very far-reach-
ing changes in word division and vocalization. Hab. 3. 6-7 has the
very difficult passage:
mt ‫ ת ח ת אגן ר אי תי‬: ‫ה לי כ ו ת ע ו ל ם ל ו‬
These are the words rendered by the ‘his ways were as of old’ and
‘I saw . . . in affliction’ of RSV. Albright, reconstructing this,
produces a text which reads (apart from the last word):
: ‫ה ל כ ו ת ע ו ל ם ל ת ח ת או ן‬
and means:
‘eternal orbits were shattered’.
This suggestion, while not changing any consonant, is an ex-
tremely radical reshuffling of the consonants into a completely
different series of words and clauses. Such a reconstruction is
really no less an emendation than a conjectural alteration of one or
two consonants would have been; the effect is entirely as drastic.
We may add Tur-Sinai’s reconstruction of Job 37. 17:
m t : ‫ ב ג ד י ף ח מי ם ב ה ש קי ט א ל ץ מ ד רו ם‬- ‫א ש ר‬
Tur-sinai: ‫ ר ב ? ב ר כ ח ה מי ם ב דז קז קי ט א ך ץ ז מ ך ך ם‬# ‫א‬
The meaning of the reconstructed text is said to b e:
'So that, when the waters become forceful
and cause their wet clay to fall down.’
The new philological identification is that of a ‫ מ ל ר‬, cognate with
Ethiopic mddr ‘earth’, Arabic madar. Though some of the textual
changes are individually easy ones (e.g. interchange of ‫ ל‬and ‫) ל‬,
the reconstruction is in fact extremely far-reaching, with a liberal
assortment of dittographies and haplographies in the space of some
nine consonants. Though Tur-Sinai says that ‘the punctuators
failed to understand the text’, it is hard to see how they failed to
understand the supposedly original ‫ ב ג ב ר כ ל ה מ י ם‬, which is not
difficult in comparison with most of the text they had to point in
Job.
These instances show that changes of vocalization, or shifts in
consonant order or word division, implied in philological treat-
ments may sometimes be so drastic in effect as to make it legalistic
 GENERAL 191
to claim that the consonantal text is unchanged. We may now go on
to consider the questions more systematically.

(2) Fallibility in the Consonantal Text

Philological scholars, then, though critical of emendation, have
often been inconsistent and have used emendation, or else, as
argued above, have produced virtual emendations through re­
shuffling of the consonants. It may be answered, indeed, that
inconsistency is not a serious charge, since the transmission of the
text was itself subject to inconsistency, so that a responsible scholar
will have to use different methods at different times. In particular
it may be suggested that emendation is legitimate, but only after all
attempts at a philological treatment have resulted in failure.
This, however, hardly goes far enough. To admit the legitimacy
of emendation at all is to admit that the consonantal text is fallible;
and to admit this is to admit that the basis for philological treat­
ments in the consonantal text is not wholly secure. Wholesale state­
ments that emendation is illegitimate, if meant literally, imply that
the consonantal text was infallibly preserved. Given the conditions
of transmission of the Hebrew Bible, if the consonantal text is not
necessarily right, then emendation may at times be necessary.
Once this is admitted, it follows that from the beginning the
alteration of the text has to be considered on equal terms in prin­
ciple with the possibility of a philological explanation. It is not
enough to agree that emendation may be considered only after all
forms of philological explanation, however remote, have been
exhausted. The difference between the two approaches lies from
the beginning in a balance of probabilities. This follows from the
admission that the text may be faulty, and is not invalidated by the
argument that many textual emendations in the past have proved
to be wild and arbitrary; the same is true, after all, of many philo­
logical explanations.
There is substantial, and indeed decisive, evidence that conson­
antal texts could be written and transmitted incorrectly. Such evi­
dence can be found in the Old Testament itself in passages which
appear twice; it can be found in the Qumran documents; and it
can be found in the literature of cognate languages such as Ugaritic.
A comparison of passages which appear twice, such as Ps. 18,
which is also II Sam. 22, suggests that errors in the copying of the
192 T H E M A SS O R E T E S, V O C A L IZ A T IO N A N D E M E N D A T IO N
consonantal text are probable. One cannot, indeed, assume that all
such differences between parallel texts are a result of textual cor-
ruption; some may go back to free variants in oral tradition. The
variation between ‫( כ ף‬II Sam. 22. 1) and ‫( י ד‬Ps. 18. 1) may per-
haps be explained in this way; on the other hand, it could easily
be a scribal change.
It is also possible that some of the differences between parallel
passages are genuine linguistic, rather than textual, differences;
that is to say, the parallel texts have words genuinely different
though approximately parallel, rather than forms which have arisen
by scribal error.
Ps. 58. 11 has a phrase which seems closely parallel to one in Ps.
68. 24. The former is:
‫הרשע‬
τ τ τ
‫בדם‬ ‫פעמיו ירחץ‬
‫ ־‬: I ‫ ־‬:· τ τ :

‘he washes his feet in the blood of the wicked’.

This presents no difficulty. But at 68. 24 the words in M T are:
‫למען תמחץ בעלף מ־־ם‬
Now since ‫ מ ח ץ‬in Hebrew usually means ‘strike’, it has been
common practice to emend the verb at 68. 24 to ‫ ת ר ח ץ‬, thus pro-
ducing agreement with 58.11; so for example B H 3ywith the alleged
support of the versions.1Delitzsch,2 however, argues that there was
an Accadian mahasu meaning ‘pour over’ (synonymous with
balalu)., and gives actual examples in literature. If this is right, the
phenomenon has to be treated as a linguistic rather than a textual
one. Finally, the sense is possibly only the normal Hebrew one
‘strike’: ‘that you may strike your foot in the blood . . Λ
Nevertheless, it is not likely that all differences in parallel pas-
sages could be removed or explained in this way, and many of them
should probably be ascribed to scribal errors, as textual critics have
generally done. Ps. 18. 11 has an uncommon verb in the phrase:
‫ר!ךא על־כןפי־תח‬
which means perhaps:
‘and he swooped upon the wings of the wind’.
1 Recently Kraus, p. 467. It is doubtful whether the versions really support
this emendation. LXX for example has (67. 24) όπως αν βαφτ} 6 πονς σου £ν αΐματι.
This is not very good evidence that the Hebrew translated was ‫ ל ח ץ‬, which is
never rendered with βάπτω in the LXX. Contrast 57. 11 (verb νίφεται). The
Greek just gave a rendering such as the context seemed to demand.
* Proleg., pp. 69 if. More recently cf. von Soden, p. 581a, ‘(5) besprengert mit\
 F A L L I B I L I T Y IN T H E C O N S O N A N T A L T E X T 193
The parallel II Sam. 22. 11 (along with some manuscripts in the
Psalm) has the reading ‫ ל ך א‬, which would mean:
‘and he appeared upon the wings of the wind’.
Probably this latter is simply a wrong text, and shows clearly the
mistaking of ‫ ל‬for ‫ ל‬by copyists and its influence even in the text
of the Psalm.
Thus the parallel passages afford strong evidence for occasional
scribal corruption of the consonantal text. That such corruption
exists is, of course, no new idea, and has in fact been the normal
accepted belief among scholars. There should indeed be no need
to repeat the evidence for it, were it not that arguments in favour
of philological treatments have sometimes come close to regarding
the consonantal text as sacrosanct; and for this reason it is right to
reiterate some evidences of its fallibility.
The Qumran material also shows quite clearly that the con-
sonantal text has been liable to corruption. This is so of course
whether the Qumran texts are right or wrong when they differ from
MT.
There are indeed places where it has been suggested that the
Qumran text will yield to a philological treatment while the M T
will not. An interesting example is the difficult Isa. 41. 27:
: ‫ר א שון ל צ י ו ן הנ ה הג ם ו לי רו ש ל ם מ ב ש ר א תן‬
This, the M T, seems to mean literally:
‘A first for Zion, behold, behold them;
and to Jerusalem I will give a messenger of good.’
The obscurity of this text has led to numerous emendations, for
which see B H 3. The Qumran text iQIsA has ‫ מ ה‬1‫ הנ‬where M T
has ‫ ה נ ם‬. Guillaume argues that ‫ מ ה‬1‫ נ‬is clearly a participle, in
parallelism with the ‫ מ ב ש ל‬of the second part. It is to be explained,
he says, from the Arabic nama 4bring tidings’, for which he quotes
cases in old Arab poems.1

1 On this word see above, p. 182. Driver, The Judaean Scrolls, pp. 435, 444,
gives the same interpretation as Guillaume. He appears also to find this word in
iQIsA 35. 7, where the M T has *and papyrus’; but from the photographs
the word seems clearly to be exactly as in the M T except for the added
waw. The Arabic nama, while certainly a real verb the use of which can easily be
checked in early poetry, is probably an idiosyncrasy of Arabic, and seems to have
no cognates in other Semitic languages. Thus the evidence for its existence in
Hebrew is weak; nor, in Isa. 35. 7, is it clear what sense it would make.
194 T H E M A SS O R E T E S, V O C A L IZ A T IO N A N D E M E N D A T IO N
In any case the Qumran text, whether better or not, adds to the
evidence i:hat texts varied. If readings in the Qumran texts are
superior, it means that in the absence of the Qumran evidence
(which might, after all, never have been discovered) the conjectural
emendations might have been right and would certainly have been
justifiable, There are in fact cases, as is well known, where the
Qumran scrolls have been found to contain readings which had
previously been conjectured by scholars.
A third place where we can see evidence of faulty transmission
of consonantal texts is the Ugaritic literature. In places this is
highly repetitive, quite long sections being reiterated in almost
exactly the same words. In such passages some fairly obvious spel­
ling errors have been detected;1 and this is a literature the line of
transmission of which to us ceased in the fourteenth century B.C.!
There is every reason to expect, therefore, that the transmission
of texts Las included errors in the consonantal as well as in the
vocalized writing; and, in spite of the exaggerated reliance on
textual emendation which has sometimes been shown, it will
always remain an important possibility that difficulties have arisen
by graphic error rather than by loss of linguistic understanding.
Textual criticism must retain its traditional place and not be com­
pletely displaced by a too purely philological approach.
In order to make further progress, we have now to give closer
consideration to the transmission of the vocalization.

(3) The Importance of the Vocalization

The picture implied in philological treatments is one of (a) a
long period during which the consonantal text was carefully
cherished, and transmitted, and (b) a late and arbitrary process by
which a vocalization was more or less imposed on this text by men
who indeed tried their best to understand it but were handicapped
by the limitations of their knowledge of Hebrew (now to them a
long-dead language) and by the narrowness of their understanding
of these particular texts. Is this a credible picture? Does it not
raise in another form the question which Nyberg raised against the
older textual criticism, namely that it assumed that the Jews very
1 For some examples see Gordon, UH 4. 16 (pp. 17 f.). For instance we have
the word bt'rth 4in its scabbard’, once written bVrtp (3 Aqhat i 29). This can be
assuredly explained as the writing of the two strokes of Ugaritic p in place of the
three of h.
 T H E I M P O R T A N C E OF T H E V O C A L I Z A T I O N 195
early lost the understanding of what their own Scriptures meant,
while they transmitted them by a mechanical copying procedure?—
And, it would appear (in this we go beyond what Nyberg argued),
having transmitted this consonantal skeleton of a Scripture by a
mechanical copying procedure, did they then centuries later
attempt to clothe it in the flesh of a newly created vocalization ?
Firstly, the vocalization of the Hebrew Bible was not invented
when the written marks of vocalization were invented. A distinc­
tion has to be made between the actual existence of vocalization
and its written marking.
Biblical manuscripts long ago had no marking of vocalization in
the form now dominant, i.e. the Tiberian point system. This system
grew up from about the sixth century a .d . But long before this
some degree of discrimination of vowels had been provided in the
form of the vowel letters or matres lectionis. While it is possible to
write Hebrew with no vowel letters at all, this was not the practice
in any extant biblical texts. Even the Siloam inscription, from
Hezekiah’s time, has some vowel letters, though not exactly the
distribution to which we have become accustomed from the M T.
The Qumran scroll iQIsA, as is well known, has a more copious
use of vowel letters than the MT, and (somewhat like modern
Hebrew) uses them in indicating ‘short’ vowels which were not so
indicated in MT. Thus, even in the period before the pointing
system was introduced, a written indication of vocalization did
exist; and indeed in certain texts the written indication was greater
than is to be found in the M T if the pointing of the latter is re­
moved.
Thus even when it was without vowel points the text was not
without indications of vocalization. Certainly this vocalization was
not ‘complete’; it did not try to indicate or discriminate all the
phonemes of the text. But this is no absolute peculiarity of Hebrew;
it is true of most writing systems that they do not systematically
mark all the phonemes. For instance, stress and pitch are phonemic
in English, but there is no way of representing this in the normal
writing system.1 In classical Greek vowel length was phonemic,
but it has graphic representation only for certain of the vowels and
not for others; and the pitch accent was phonemic, but was not
marked until long after the classical period.2 Consonant length is
1 On this see Gleason, pp. 40-50.
2 See Gleason, p. 419.
196 T H E M A SS O R E T E S, V O C A L IZ A T IO N A N D E M E N D A T IO N
phonemic, and is extremely important, in Ethiopian languages, but
has r.o written marking (though some modern grammarians have
introduced it). In failing to provide a clear one-to-one marking for
all vc wel phonemes, the Hebrew matres lectionis departed from the
consistent early Phoenician writing system, which had systemati­
cally omitted the entire sub-system of vowels; instead, they formed
a series of optional markings for certain groups within the vowel
sub-system.1 The fact that at this stage there was not a well-pro­
portioned correspondence between the phonemic and the graph-
emic systems is nothing extraordinary. Thus the texts of biblical
Hebrew, even when they were still ‘unvocalized’, were not devoid
of sensitivity to the indication of vowel distinctions in writing.
When the complete vocalization, which included not only the vowel
points but also other diacritic signs like the daghesh and also the
accent system, came to be applied, it was related to a more rudi­
mentary series of vowel indications which was already present.
More important, however, for our purpose is the tradition of
vocalization which was passed on in addition to the written indica­
tions Week by week and year by year, Moses and the prophets
were read. It is a mistake to think of a written consonantal text
which was handed on through generations without vocalization
and then afterwards was ‘vocalized’. This is true of the stages of
the written text when taken in a somewhat artificial isolation; but
it is not true of the history of the tradition. The Massoretes began
with a text lacking vowel points and proceeded to point it and
accent it; but this does not mean that they invented the vocaliza­
tion. What they invented was a series of increasingly subtle systems
for the marking of the vocalization which was already in use.
Thus the term ‘vocalization’ is somewhat ambiguous. The
system of points was applied late (after the completion of the Tal­
mud), and previously texts were unvocalized in the sense that these
signs were absent; even then the vowel letters provided a partial
marking of vowels. But the biblical text was transmitted also in a
spoken form; the text was publicly read and the mode of reading
was passed on. There is no evidence of a stage at which this linguis­
tic transmission of the text had come to be interrupted, so that
Jews had to start again with no material but a written text, from
which the further linguistic elements such as vowels, which were
unmarked in writing, had to be deduced or invented.
1 See Gleason, op. cit., p. 419.
 T H E IM P O R T A N C E O F T H E V O C A L IZ A T IO N 197
It is not probable that Jews learned to read the biblical text by
looking at the written signs and guessing at the vowels which may
have accompanied them. Rather, they learned by reading with a
teacher, or listening to a synagogue reader; and this teacher or
reader knew from his own learning how the text was to be read.1
In this respect there was a difference from the reading of a casual
communication like a letter, which would certainly have to be read
without any tradition of vocalization, but in which the intrinsic
difficulty of the text would be likely to be less, while a wrong
reading would not carry religious consequences.2
It is commonly held to be a characteristic of the Semitic lan­
guages that the consonants are a basis or backbone and provide all
the cardinal ideas, while the vowels indicate only modifications of
these ideas. Only to a limited extent is this true.3 Often one cannot
tell a ‘basic idea’ from the consonants alone. The mere consonants
DX do not make clear whether the idea is ‘father’ or ‘ghost’ or ‘bud’.
The vowels are critical as the consonants are. In the reading of an
extended passage discrimination is made through the context. This
can be done even when the vowels are not marked. It is not that the
ideas come through to the mind independently of vowels; rather,
the context guides the reader to select the vocalization which makes
sense of the passage (or—and particularly if a passage is read
silently—to select senses which imply a particular vocalization, and
thereby to select, from the number of vowel patterns possible
within the consonant sequence, the ones which make sense). The
consonant sequences do not in themselves furnish the basic sense;
taken just as they are written, they do not make clear which con­
sonants are ‘root consonants’ and which are not. While we have
learned to look for triradical consonantal roots and work outwards
from these, we should be aware that (a) this procedure was
developed by a grammar which itself presupposed the vocalization
as its means of analysis, and (b) this procedure is not necessarily
the way in which ancient and medieval Jews read the text (or any
other document). This is relevant especially because the ancient
reader normally read aloud and not silently.
1 Cf. the Talmudic passage quoted below, pp. 213 f.
2 Cf. my article ‘Vocalization and the Analysis of Hebrew*, p. 4, and below,
pp. 208 f., where I suggest that the procedure of translators like the LXX differed
from that of normal reading in Hebrew.
3 Cf. Ullendorff, Orientalia xxvii (1958) 69 if.; Moscati, Comparative Grammar,
p. 72.
198 T H E M A SS O R E T E S, V O C A L IZ A T IO N A N D E M E N D A T IO N
The nainly consonantal nature of Semitic writing systems,
however, might be thought to show the contrary and demonstrate
that the consonants were after all regarded as the main centres of
meaning But is it true that Semitic writing systems are so over­
whelmingly consonantal? It is not true of Accadian writing; nor
is it true of Ethiopic, in which the consonants cannot be written
without an attached vowel. Ugaritic script indicated vowels for one
of the consonants, namely aleph. Arabic script, when not pointed,
does not mark all vowels; but it is equally characteristic of it that
it always marks some vowels: normally, in fact, all long vowels are
marked,1 and only short vowels are unmarked. The commoner
Semitic characteristic is not the marking of consonants exclusively,
but the marking of only some vowels. The use of a script which is
strictly without marking of vowels is indeed rather exceptional.
The undoubted fact that writing systems which can work without
any vowel marking are among those which have emerged in the
Semitic world is not sufficient cause for us to conclude that there
is a mysterious identity between the consonants and the basic
meaning, Thus the relation between vowel and consonant in the
writing of Semitic languages does not in itself give us reason to
treat lightly the tradition of vocalization or to suppose that the
vocalization was marked without very considerable guidance from
previous tradition.
It is possible, indeed, to suggest explanations of the primarily
consonantal nature of many Semitic writing systems without
making any reference to a bearing of the basic meaning by the con­
sonants, and indeed without making reference to meaning at all;
such explanations would depend rather on the actual formal struc­
ture of Semitic languages.
It is possible that the marking of a phoneme in writing was con­
nected with the occurrence of that phoneme in initial position. In
the Semitic languages which are relevant for the early development
of the alphabet, all word patterns of importance had a consonant
in initial position, while none had a vowel in initial position. Con­
versely the alphabet provided a grapheme for each phoneme
which might occur in initial position. The arrangement thus
automatically segmented words into syllables with a consonant in
first place. If this is plausible, the character of the system had
nothing to do with the idea that consonants bore the core of
1 Morag, Vocalization Systems, p. 13.
 T H E IM P O R T A N C E OF T H E V O C A L I Z A T I O N 199
meaning; it derived from the formal structure of languages which
had no relevant patterns except with consonants in initial position.
In Greek adaptations of the script, however, vowels were early
marked, and the necessity to do this may similarly be associated
with the fact that many important Greek patterns have vowels in
initial position.
We can state this in another way. The common and traditional
view is that the basic meaning lay in the root, i.e. the consonants of
the root, and that the vowels indicated modifications of this idea.
But, even within the terms of this view, we observe that ‘modifica-
tions’ may equally include consonants, in which case they are
marked in the script, or not include consonants, in which case they
are not (we assume at this stage an absence of matres lectionis). Thus
if ‫ ק ט ל‬is to be taken as a ‘modification’ of the basic ‫ ק ט ל‬, surely
exactly the same status belongs to ‫ ; ה ק ט ל‬but the latter has a
consonantal addition which is marked in the script. If the conson-
antal nature of the script had had anything to do with the idea that
the consonants should be marked because they formed the basic
meaning, then it would have been natural for a script to develop in
which the ‫ ה‬of the hiphil was not marked any more than the vowels
of the piel. It is more consistent to maintain that what was to be
marked was decided by structure. The traditional view comes
nearer to the truth in this respect, that the pattern of many Semitic
words, and especially verbs (where we can best use the term ‘dis-
continuous morphemes’),1favoured the practicability of some such
writing system. Even so, as we have seen, most systems quite
rapidly introduced the graphic marking of some vowels.
This suggestion is made here only to show that alternative ideas
about the primarily consonantal nature of the scripts can be enter-
tained; but the remainder of my argument does not depend on the
rightness of this suggestion.
It is well known that the pronunciation of Hebrew has changed
historically. Certain spellings in the Qumran texts, for instance,
have suggested to scholars that the 3rd masculine pronoun was
pronounced as /hu’a/ in the Qumran period, while in the Massoretic
period it was pronounced as /hu/. None of the pronunciations
1 The term will be found in Gleason, p. 72 f. Gleason correctly states that
Hebrew is unusual for the high number of such discontinuous morphemes which
it contains; it is inexact when we go farther and make this the basic principle of
Hebrew, as the popular view often does.
200 T H E M A S S O R E T E S , V O C A L IZ A T IO N A N D E M E N D A T IO N
extant in modern times, probably, preserves a series of distinctions
identical with those intended by the Massoretes. That Hebrew
pronunciation should have changed historically is only what we
should expect from our knowledge of languages generally.
Equally we know that the pronunciation of Hebrew has differed
regionally and dialectically. This was already so in ancient times,
and in more modern times there are the whole series of differences
between Ashkenazi and Sephardi Hebrew.1 These differences may
be considerable. The vowel sign qames is an [a] vowel in Sephardi,
but an [o] vowel in Ashkenazi. A D in post-vocalic position is an [s]
in Ashkenazi, a [t] in older Sephardi, and a [t] in modern Israeli.
There have also been different written systems of marking the
texts : the Babylonian punctuation, for example, not only uses
signs of different shape from the Tiberian, but is a different system,
having a different total number of vowels.
All of these, then, are ways in which the pronunciation of
Hebrew has varied along co-ordinates of time and place. The
question now is whether this variety is any reason why the tradition
of vocalization should be considered intrinsically unreliable and
therein separable in principle from the tradition of the written text.
It would seem that many scholars, knowing of the existence of this
variation in pronunciation, have concluded that the tradition of
vocalization is an unchartable chaos, the produce of arbitrary
decisions on the part of men who were out of touch with the real­
ities of the texts under treatment.
This, however, does not necessarily follow. The fact of historical
change and regional variation in the pronunciation of Hebrew,
including the known existence of different written systems of
vocalization, does not in itself prove that the vocalizations existing
in tradition are unreliable or arbitrary in relation to the questions
which are being studied in this book.
Firstly, we are not concerned with achieving a phonetically
accurate reproduction of the words of, let us say, the prophet
Isaiah. Since the pronunciation of Hebrew has become phonetic­
ally different, this would be a very difficult task. Such questions,
however, are quite marginal to our subject. The existence of
phonetic change in time is irrelevant except in so far as it can be
shown to have led in actual evidenced cases to the obscuring of
1 For a statement describing nine different Jewish regional pronunciations
see Bauer and Leander, Historische Grammatik, pp. 170 ff.
 T H E IM P O R T A N C E O F T H E V O C A L IZ A T IO N 201
information given by linguistic elements, for example, through pro­
ducing a situation where one word comes to be confused with
another.
One might have expected that philological treatments where they
involve a change in the traditional vocalization might include an
argument that known historical examples of phonetic change apply
to the words under treatment. Not all elements of Hebrew have
been equally affected by phonetic change. If such change were a
main cause of the unreliability of the vocalization, we should expect
the incidence of philological treatments to be heaviest at points
where phonetic change is known to have occurred and lightest at
points where no substantial change has occurred. No such correla­
tion, however, is observable in a survey of philological treatments,
nor have I seen reason to believe that scholars who produced treat­
ments have considered the possible importance of such a correlation.
Moreover, sound changes do not operate at a constant rate on all
sounds. While certain sounds in a given language may alter very
greatly in a very short time, others may persist with little change
over thousands of years. It is therefore wrong to expect that the
undoubted fact of sound change in Hebrew will necessarily mean
that the late pronunciation traditions are in a level and general
manner unrepresentative of ancient practice. Moreover, a language
may preserve important distinctions even though very great
changes in phonetic character have taken place. A striking example
is that of the stress accent in modern Greek, the position of which
is to a large extent the same as that of the pitch accent in ancient
Greek, even though the general pronunciation of Greek and the
whole character of the accent itself have changed very greatly.1
Where philological treatments have involved a departure from
the traditional vocalization, not only have they not generally given
evidence of relevant phonetic change, but they are themselves
mostly inconsistent with the possibility of explanation by appeal to
phonetic change. One of the important aspects of phonetic change
is its high relative constancy, such as is found in the differences
between Ashkenazi and Sephardi pronunciations. Where philo­
logical treatments have required a change in vocalization, however,
they have seldom or never implied phenomena of high constancy.
Generally they have demanded the revocalization of this one parti­
cular word, irrespective of the effects this would have on other
1 Cf. Thumb, Handbook, p. 28.
202 T H E M A SSO R E T E S, V O C A L IZ A T IO N A N D E M E N D A T IO N
words employing the same patterns. For instance, it is not un-
common for scholars who find a verb pointed as (say) hiphil in the
Massoretic text to rule that it should in fact be a piel. Quite possibly
this kind of decision may be right. But it cannot be justified on the
ground of phonetic changes general in the transmission of Hebrew
pronunciation. All known forms of Hebrew pronunciation provide
distinctions adequate to discriminate between piel and hiphil in
normal circumstances. This is so even if the various pronunciations
are very different from each other. A confusion in transmission
between piel and hiphil could normally be attributed to historical
phonetic shifts only on the hypothesis that these obscured the dis-
tinctions between the two sets of forms. Since, in the texts under
examination, this does not appear to be so, and since adequate
marking of the distinctions continues to be accepted apart from
the isolated cases where the vocalization is emended, the pheno-
menon can hardly be attributed to general phonetic shifts. It is not
easy to see how such shifts could mean that the vowels of a word
somehow got picked up and redistributed, as seems to be implied
when we regard the vocalization as something which the scholar
can redistribute at will.
For instance, at Prov. 26. 28, M T !‫ישנא ד כ י‬. ‫ל שון ״ ש ק ל‬
apparently means ‘a lying tongue hates its victims’. It has been
suggested (Index, no. 91) that the last word be read ‫‘ ל כ ו י‬ac-
quittal’ (as Aram. ‫ = ל כ א‬standard Hebr. POT). If this suggestion
means that an original writing ‫ ל כ ו י‬came to be written as 1‫ל כ י‬
and was followed by a change of vocalization, then it is basically a
graphic corruption and belongs to the textual rather than the
philological type of treatment.1 But even then it leaves unclear
how communities, in which /dikkuy/ had been read, suddenly
changed over to the reading /dakkaw/.
If, on the other hand, a noun pronounced /dikkuy/ came to be
pronounced /dakkaw/, this is no ‘mere’ change of vocalization, but
a complete shift from one word to another, which cannot be ex-
plained by phonetic drift or other change of pronunciation. Any
such change would have affected also other words of the type
‫ ג לוי‬, such as ‫ צ פ ו י‬or ‫ ש ק ו י‬.
In fact philological treatments have generally implied a more
semantically based process of vocalization; they have suggested
1 Cf. again below, p. 218.
 T H E IM P O R T A N C E O F T H E V O C A L IZ A T IO N 203
that the Massoretes were governed, not by the antecedent tradition
of pronunciation, but by opinions about the meaning of the text.
Seeing a consonant group in the unvocalized text, they are thought
to have analysed it for grammar and meaning and marked it with
the points implied by the results of this analysis. If the vocalization
is ‘wrong’, this is because the Massoretes identified the word
wrongly and, failing to discern the meaning, attached an erroneous
vocalization.
There are, however, some serious difficulties in the way of this
conception of the work of the Massoretes. One such difficulty is
the traditionalism which appears to have inspired their work.
There have, indeed, in recent years been some opinions according
to which the Massoretes were rather linguistic innovators. This
will be discussed shortly. But it seems better for the present to
follow Bergstrasser’s view:
What we know of their activity shows an obstinate clinging to the
smallest details of what was transmitted to them.1
Similarly Bacher had earlier written:
These signs [i.e. the vowel points] are the embodiment of the vocalic
pronunciation, which the Massoretes had received in oral tradition
over many generations along with the written text of the biblical books.
The knowledge of the linguistic forms, handed down in the reading of
the text, reaches back into times not too far distant from that in which
Hebrew was still a living language.2
It is true that various schools and stages of punctuation existed.
But the process is an increasingly detailed attempt to mark what
was already there, rather than the making of decisions entailing a
departure from it.
The discussion traceable to the Massoretes is of a primarily non-
semantic nature. The Massoretic treatises known to us, such as the
Ochlah we-Ochlah or the Treatise on the Shewa, are concerned with
classification of phonemic and graphic differences. These are not,
indeed, wholly unrelated to semantic differences. But the mode of
discussion hardly leaves room for the conception that the Mas­
soretes first made up their minds what the text meant and then
pointed it accordingly.

1 O LZ xxvii (1924) 582-6. See Kahle, Cairo Genista (2nd ed.), p. 188.
2 ZD M G xlix (1895) 13.
204 T H E MASSORETES, VOCALIZATION AND EM ENDATION
The same can be said about the marginal notes, commonly
called the Massorah, appended to the Hebrew Bible. These are
not to any great extent of lexicographical or of semantic nature.
Most commonly they are listings of writings. In these listings the
obvious semantic questions—for instance homonymy—are com-
monly ignored. For the sense of a biblical passage the right identi-
fication of a homonym can be of obvious importance, but the
Massorah gives little such guidance.
There are two homonymic nouns ‫ ; מ ק ב ת‬one means ‘hammer’
and the other means ‘hole, excavation’. The latter occurs only at
Isa. 51. 1, but the Massorah of the Aleppo Codex neither marks it
as an unusual form nor gives any indication that it should be under-
stood otherwise than its homonym. In the same sentence, however,
the Massorah marks the phrase ‫ ר‬1‫‘ א ל ~ צ‬unto the rock’ with an
annotation that this group with this writing occurs thrice, though
no semantic problem is touched by this. Again, at Isa. 13. 8 the
form ‫‘ ל ה ב י ס‬flames’ is marked with a note that it occurs thrice
in all. The other two cases are ‫ ל ה ב י ם‬at Gen. 10. 13, I Chron.
1. 11, where the meaning is ‘Lybians’. The Massoretic note, being
concerned with writings and not with meanings, makes no attempt
to separate the two types.1
It is true that certain notes of semantic type may be found, but
these are so occasional as to show that the exception cannot be made
the rule. For instance, at Prov. 25. 14, where the word ‫נ שי אי ם‬
occurs in a sense other than the common ‘prince’, the Massora
Parva of Codex B 19a gives the note ‫] ענן‬1‫‘ ל טי לי ] ל ש‬four
(occurrences of) words in the sense of “cloud” ’.2 There are other
places where the Massorah, using the technical term ‫‘ ל שון‬has
the sense of’, provides some semantic guidance, very often in rela-
tion to unseemly expressions. Such notes, however, form so
infinitesimal a proportion, in relation to the total number of Mas-
soretic notes or in relation to the total number of places where
semantic uncertainty is found, that we may suppose that they were
usually prompted by some special reason; and thus these excep-
tions do not alter the fact that the main body of Massoretic annota-
tions was non-semantic in nature.
Hapax legomena indeed are often marked with the sign ‫( ל‬for
1 See Weil in V T S ix (1963) 276 n.; cf. Textus iii (1963) 119.
2 The other three cases are at Jer. 10. 13, 51. 16; Ps. 135. 7.
 T H E IM P O R T A N C E O F T H E V O C A L IZ A T IO N 205
‫‘ ל י ת‬there is none’, i.e. no other instance).1 But the identification
of rare words in our lexical sense is by no means the purpose of
these notes. The mark ‫ ל‬is used on graphic forms which are unique,
and this has nothing intrinsically to do with the uniqueness of the
word in a lexical sense. A form will be so marked even if the word
is a very common one, provided that the writing is an unusual one
(e.g. in respect of ‫ מ ל א‬or ‫ ר‬0 ‫ ח‬in mode of writing) or one other-
wise liable to confusion.
Generally, then, the Massoretic lists concern the exact writing
and reading. Often the procedure is semantically indifferent,
though it is important to the scribe, whose concern is with exact
copying of text. Even where the lists have semantic implications,
these often remain implications rather than express conclusions,
and the statement expressly made by the lists is one about the
pronunciation. There is a list which, under the consonantal form
‫ י ח ל‬, records the number which have the vowel patah and the
number which have qames.z To us this involves the two different
roots ‫ ח ל ל‬and ‫ י ח ל‬, and thus different meanings; but the form
which the Massoretic annotation takes is a declaration about the
variation between one vowel and another. All this makes it difficult
to sustain a view of the Massoretes as men who pointed the text as
they ‘understood’ it; if this had been so, one would have expected
a primarily semantic sifting and cataloguing of the material.
The picture of the Massoretes as interpreters who worked out,
or guessed, from the consonantal text what its meaning might be
and then vocalized it accordingly, has other difficulties which ap-
pear after very little thought. One of these is the fact that it is
extremely difficult, in cases where a text is intrinsically obscure,
to know how the Massoretes ‘understood’ it at all. If they had in
fact worked out an understanding of it, and then pointed the text
on this basis, one would expect it to be rather more transparent
what they intended than is actually the case. The difficulty of the
existing text tends at many points to suggest that the Massoretes
transmitted a received text with its own difficulties, rather than
iron these out into something which by their then knowledge was
smooth and satisfactory.
1 Cf. Rabin, ‘Millim Bod*dot’, in Encyclopaedia Miqra'it.
2 Weil, V T S ix (1963) 280 n. Weil rightly points out that this failure to make
a distinction between forms from different roots cuts entirely across our lexical
approach and forms a substantial difficulty for the modem student of the
Massorah.
206 T H E M A SS O R E T E S, V O C A L IZ A T IO N A N D E M E N D A T IO N
This is a somewhat different matter from the standardization of
grammar. It is often held that the dialectal and temporal variations
in ancient Hebrew have been overlaid by the fairly unitary system
now represented in the pointing. Such a standardization does in
fact seem to have taken place, though I think it was not the work
of the Massoretes, i.e. of those who introduced the pointing, but
rather was introduced in the liturgical reading tradition much
earlier (which seems much more natural). It is reasonable to sup­
pose that this tradition introduced a more uniform and standard
morphology. It would, in fact, be necessary to do so unless the
readers of the Bible were endowed with phonetic, historical, and
literary discriminations of a quite exceptional order. The synagogue
reader could no more be expected to imitate the different Hebrew
pronunciations of the Song of Deborah, the prophet Amos, and
the wisdom teacher Qoheleth than the modern English reader tries
to reproduce the phonetics of Shakespeare, Pope, and Scott when
he reads these authors. Intelligibility would require that a levelled
system should be employed. It is quite a different thing to suggest
that within such a levelled system the discriminations between
morphemes were arbitrarily altered by the Massoretes in order to
achieve a meaning which seemed to them to be satisfactory. Thus
the fact of the standardization of the grammatical system, if it is a
fact, does not in itself justify the conception of an arbitrary mould­
ing of the text by the Massoretes on the morphemic-semantic level.
One must also ask how the Massoretes, if their work involved
innovations on the level of meaning, succeeded in getting these
innovations accepted by the community, in view of the inertia of
religious traditions and practices. Substantial changes on a seman­
tic level would lead immediately to exegetical and theological
decisions. How could such innovations be carried through, amid
the sensitivity to such questions in the contemporary controversy
between Qaraites and Rabbanites ? How could the opposite party
fail to point out that the reading of the text had been wilfully altered
by the new traditions in the vocalization ?
It should be added that the term ‘the Massoretes’ seems some­
times to be used in two senses, a narrower and a wider. I have used
it in the narrower sense, to denote the scholarly families which
worked upon the copying of the text and the provision of vocaliza­
tion and accents in the period of roughly a .d . 600-1000. But the
term may also be used of the more general transmission of the text,
 T H E IM P O R T A N C E O F T H E V O C A L IZ A T IO N 207
without special and exact reference to the punctuators. When we
read in the context of modern philological treatments that ‘the
Massoretes failed to understand the text' this may be meant to
apply not to the actual punctuators like the Ben Asher or Ben
Naphtali families, but rather to the general tradition which already
lay behind them and was assumed by them. If this is so, then some
of the arguments I have just made do not apply in the same way.
The conception that the text was misunderstood, if no longer
applying to the Massoretes in the stricter sense, is not liable to
the criticisms I have just advanced. The meaning may be that the
misunderstanding, and the ‘correction’ of the text in order to fit a
new understanding, had already taken place before the Massoretes
recorded the tradition.
If this is what is meant, then my arguments do not have the same
effect; but neither is the problem solved thereby. For if ‘the Mas­
soretes’ is a term used to designate the general tradition of reading
and understanding over some centuries, then the work of the actual
punctuators can hardly be stigmatized with terms like ‘artificial’;
for their decisions were not produced in the process of providing a
punctuation but must, right or wrong, have been inherent in the
tradition long before the process of punctuation began. The
decisions now are no longer ‘late’, but may have taken place at any
time between the formation of the original text and the fixation of
the Massoretic punctuation. Only detailed research can tell us
whether they had already been made by the time of the Qumran
texts, or even by the time of Ezra.
The difficulty of attributing to the Massoretes great initiative in
semantic interpretation leads on the whole to the judgement that
the source of confusion often lay early, rather than late, in the
history of transmission.

(4) Evidence for Pre-Massoretic Vocalization

The last section discussed the general relation between the
Hebrew vowels and the language as a whole, along with some
relevant aspects of the work of the Massoretes. We now go on to
discuss some areas in which evidence may be found for the vocaliza­
tion of Hebrew before the Massoretes registered it with points. The
evidence is complicated, and will receive only a partial survey here.
Moreover, it is not new evidence; but it has generally been used in
208 T H E M A SS O R E T E S, V O C A L IZ A T IO N A N D E M E N D A T IO N
the study of different problems from our present one. Our concern
is not with purely phonetic or even purely grammatical differences,
but with differences which have a substantial effect on meaning.
My purpose is to show that certain often-quoted sources and
theories do not, on examination, disprove the continuance of a
tradition of vocalization long before the time of the Massoretes.
Finally, the many anomalies of the Massoretic pointing itself are
an argument against the idea that the vocalization was the product
of an artificial process.
One of the early sources of evidence is the Septuagint. The stan­
dard works mention with some emphasis that the text, from which
this translation was made, was unvocalized. Thus Swete writes:
Lastly, almost every page of the LXX yields evidence that the
Hebrew text was as yet unpointed. Vocalization was in fact only tradi­
tional until the days of the Massorah, and the tradition which is en­
shrined in the Massoretic points differs, often very widely, from that
which was inherited or originated by the Alexandrian translators.1
What is true of the Alexandrian translation, however, does not
necessarily apply to the way in which the Hebrew text was pre­
served within those communities which read in Hebrew and/or
Aramaic.
Firstly, all the evidence indicates that knowledge of Hebrew in the
Egyptian Jewish community was poor.2 Philo, their most learned
man, is usually supposed not to have known any. The Egyptian
synagogue did not (according to the usual view, at any rate) read
the Bible in Hebrew at all, in this respect differing from the other
practice in which the Hebrew text was read along with a Targum
into another language. For this reason LXX renderings which
would seem to imply erratic vocalizations in Hebrew do not
necessarily reflect the transmission of the text where the Hebrew
was read in the synagogue.
Secondly, it is a possible hypothesis that the use of the then
written text alone, i.e. the unpointed text, was a special feature of
the work of some translators, which in this respect differed from
the processes used in the transmission of the Hebrew text. These
translators worked from the written text, deciphering it and pro­
viding a sense in Greek which seemed to correspond. It is not
necessarily correct to say that they ‘vocalized’ it. Rather, given the
1 Introduction, p. 322.
2 For further comment on this see below, pp. 267 if.
 E V ID E N C E F O R P R E -M A S S O R E T IC V O C A L IZ A T IO N 209
sequence of written signs, they may have selected, out of the
various possibilities of meaning which were provided by that
sequence, one which seemed to them, from a number of varying
and sometimes competing criteria, to be good, and this they repre­
sented in Greek. They did not submit the text to a grammatical
analysis, and for this reason we do wrong when we say, as we some­
times do, that they took such and such a form to be a participle, or
that they supposed it to be a piel when it is really a hiphil, and the
like.1 These considerations were quite foreign to their situation.
Given a sequence of characters x-y-z, they may have neither
vocalized this nor analysed it grammatically, but may have had in
their minds a series of Greek words which might be the meanings
of x-y-z, and from these they chose one.
Thus the LXX were actually doing, if this suggestion is right,
what common scholarly theory has depicted the Massoretes as
doing: given an unpointed text, they deduced a sense and fixed
that sense by their translation, just as the Massoretes are pictured
as having deduced a sense and fixed it thereafter by their pointing.
The difference is, however, that the Massoretes were in fact doing
something quite different: they were not fixing a sense, but record­
ing a tradition of reading.
The suggestion here advanced is consistent with the fact that the
word divisions are often placed by the LXX at points which not
only are different from those of the M T but also are intrinsically
extremely improbable. In construing the sequence of written signs,
they permitted themselves greater latitude than that which the lan­
guage system, as we know it, in fact allowed, both to take elements
out of sequence and to vary the boundaries between one element
and another. Until the Middle Ages there was no grammatical
analysis which made clear the conditions under which Hebrew
consonants might seem to disappear, to intrude, or to occur in
peculiar positions; and, until this analysis was done, phenomena
which to us seem quite regular may have seemed to justify other
kinds of disregard of sequence which to us are entirely irregular.
The task of perceiving and fixing a sense, forced upon the trans­
lator by the nature of his task, was greatly eased by this latitude in
the sequence in which elements might be taken.2 In this notional
1 Cf. also below, pp. 265 f. Contrast the picture of St. Jerome, below, pp. 211 ff.:
he, following the work of the later Greek translators (and the Hexapla), knew that
the vocalization could not be neglected. 2 Cf. examples cited below, p. 267.
210 T H E M A SS O R E T E S, V O C A L IZ A T IO N A N D E M E N D A T IO N
altering of the sequence, the translators were not, like the modern
textual critics who have done something similar, trying to establish
the original text. They did not ‘read’ something different; they
worked in such a way that, if sense could be found by what for us
implies a change of sequence, then that sense would do well
enough.
To resume the main point, then: in some ancient translation
techniques it was possible to work from a text which did not
register in writing all the phonemes of the language, without this
implying that no tradition of these phonemes existed.
The later translations into Greek and Latin may also furnish evi-
dence for the vocalization (a) through their renderings, which may,
when due account is taken of the techniques of the same translator
elsewhere, constitute evidence for a mode of reading and therefore
implicitly for a vocalization (b) through their transcriptions of
Hebrew words into Greek or Latin scripts.
The former is illustrated at Ps. 68. 32 (Gk. 67. 32):
mt ‫כרש ת רי ץ י ד י ו ל א ל ה י ם‬
Aq. ΑΙΘίοφ δρομώσει χ^ΐρα αύτοΰ τω 06ώ

For ‘run’ (‫ רו ץ‬qal) Aquila uses τρβχβιν (δραμ,ειι‫)׳‬, e.g. Ps.

19 (18). 6, Prov. 6. 18, Qoh. 12. 6 (though in this last the sense is
different). The formation Βρομούν is his characteristic coinage to
represent the hiphil type. Aquila certainly did not know the form
‫ ת ר ץ‬suggested in B H 3—one of its excursions into the philological
approach—but either (a) he knew the vocalization to be /taris/ as in
M T (or something semantically equivalent), or (b) at least he
deduced the same consequence from the graphic form.
Aquila’s renderings, however, do not necessarily agree with the
implication of the M T vocalization: e.g. his μβτα σου έκουσιασμοί
for M T ‫נ ד ב ת‬ at Ps. n o (109). 3, or his οϊσουσιν for ‫י א תיו‬
at Ps. 68 (67). 32. While this could be ascribed to poor discrimina-
tion of vowels later clearly separated in MT, another explanation
is possible: the renderings may be attributed to Aquila’s etymolo-
gizing technique. This technique, by concentrating on the con-
sonantal shape of words, tended to depreciate the discriminatory
importance of the vocalization. With these qualifications, however,
and with adequate understanding of the translation techniques,
the renderings of Aquila and others often support the vocalization
 E V I D E N C E FOR P R E - M A S S O R E T I C V O C A L I Z A T I O N 211
of M T against any alternative grossly different in its semantic
effects.
Secondly, evidence may be found in the transcriptions of Hebrew
in Greek or Latin scripts. Aquila can cite quite out-of-the-way
words in forms close to MT, e.g. Isa. 6 0 . 13 daSaap kcll daaoovp
(‫ ך ה ר ו ת א שו ר‬1‫)ר‬. The second column of the Hexapla wrote Ps.
46 (45). 1-3 as follows:1
Xapbavaaorj Aaj3vr]K0p aX aXptcod cap eXcoeip, XavoV /xacre ovo£ e£p
ficrapajd vepuaa puoS aX ycv Aco vcpa ^aapLtp aaps ovfiaficoT apip, j8Aej8
tapupL.

A high proportion of vowels are very close to MT.

Similarly, Jerome has many words which suggest a firm tradition
of vocalization.2 The following are taken from the early chapters of
Genesis: bresith (3 ‫) ר א ש י ת‬, eden (]‫) ע ד‬, meccedem (‫) מ ק ד ם‬, hissa
(‫) א ש ה‬, arom (‫) ע ר ו ם‬, thardema (‫) ת ר ד מ ה‬, hatath (‫) ח ט א ת‬. Here
is a complete sentence (Gen. 4. 15):
m t ‫ו י א מ ר ל ו י ה ו ה ל כ ן כ ל ־ ה ר ג קי ן ש ב ע ת י ם י ק ם‬
Jerome: uaiomer 10 adonai lochen chol oreg cain sobathaim ioccamo
There is, indeed, reason to inquire about the phonetic basis of
some of these words (e.g. sobathaim, ioccamo); but they do not sup-
port any claim that vagaries of vocalization were such as to produce
any widespread semantic confusion.
Jerome has a comment on the word ‫ ל ב ר‬at Jer. 9. 21, which has
often been quoted as evidence for the absence of vocalization.3 He
writes:
uerbum Hebraicum, quod tribus litteris scribitur ‘daleth, beth, res’—
uocales enim in medio non habet—pro consequentia et legentis arbitrio
si legetur ‘dabar’ ‘sermonem’ significat, si ‘deber’ ‘mortem’, si ‘dabber’
‘loquere’.
From what has already been said, we are in a position to see this
remark in a wider context. Jerome’s remarks cannot be univers-
alized into the conception that in vocalizing the consonantal text
every man did that which was right in his own eyes. It is by no
1 Quoted after Swete, Introduction, p. 62.
2 On Jerome,s use and analysis of Hebrew see my articles in B JR L xlix
(1966-7) 281-302 and J S S xii (1967) 1-36.
3 e.g. B. J. Roberts, Text and Versions, p. 49.
212 T H E M A SS O R E T E S, V O C A L IZ A T IO N A N D E M E N D A T IO N
means normal for him to offer us a choice between every possible
vocalization of a consonant sequence; his normal procedure is to
give a vocalized transcription without any sign of hesitation. At
Jer. 9. 21 Jerome is facing a particular problem, i.e. that the earlier
translators had been divided. The original LXX had missed out the
word altogether, in accordance with its common tendency to
abbreviate the text in Jeremiah. The Hexapla added θανάτω, which
came from Theodotion. Aquila and Symmachus had λάλησον ( —
MT). Thus translations already existing had provided for two of
the three possibilities which Jerome enumerates.1 For this reason
(and perhaps some other reason which we do not know) he men-
tions the possibility that the written characters can be construed in
these different ways. The observation therefore does not represent
either Jerome’s normal practice or that normal in his own time. We
may add that Jerome may very probably have known, and been
influenced in his remarks here by, the practice of the al-tiqre
interpretation,2 which was certainly current in the Jewry of his
time, but which, as we have seen, by no means implies any un-
certainty in the tradition of vocalization.
That the temptation to use for literary or homiletic effect other
vocalizations which are theoretically possible and which provide
attractive extra meanings or levels of meaning (which, as we have
seen, is the essence of the al-tiqre interpretation) was known to
Jerome, can be seen from examples like his handling of ‫ ר עי ם‬at
Jer. 6. 3.3 If the word is read as reim it will mean ‘lovers’; read as
roim, it will mean ‘shepherds’—satis eleganter, Jerome remarks,
revealing the literary appreciation which he has of the double
meaning. For in the previous verse (6. 2) Zion has been compared
to a beautiful woman who attracts lovers, while now (v. 3) we hear
of the gathering of shepherds for the attack on the city. Thus the
alternative vocalization is a fancy which suggests an additional way
in which two sides of a metaphor are linked. Jerome knew perfectly
well that ‫ ר ע י ם‬meant ‘shepherds’ and was vocalized as roim\ this
is fixed by the following ‫ ע ד ר י ה ם‬, and the versions, including
Jerome’s own, took it so.
The striking difference made by gross variations in implied
1 Cf. also Hab. 3. 5, where a similar discussion by Jerome notes the LXX
λόγος for ‫ ד ב ר‬, but Aq. λοιμός,
2 See above, pp. 45 f., and below, p. 214.
3 Text in CC Series Latina, lxxiv, pp. 63 f.
 E V ID E N C E F O R P R E -M A S S O R E T IC V O C A L IZ A T IO N 213
vocalization can be seen well in the Psalms, where Jerome’s own
translations represent, firstly, a version remote from MT, and,
secondly, one close to it. This is Ps. 102 (101). 24 f:
(i) LXX αττεκρίθη αντω εν όδω ισχνός αυτόν
την όλιγότητα των ημερών μον ανάγγειλόν μοι.
Psalt. Gall.
respondit ei in uia uirtutis suae
paucitatem dierum meorum nuntia mihi.
(ii) Symmachus
εκάκωσεν iv όδω την Ισχνν μον
εκολόβωσε τάς ημέρας μου. ερώ, ό θεός μον . . .
Psalt. iuxta Hebr.
adflixit in uia fortitudinem meam,
adbreuiauit dies meos. dicam: deus meus. . .
The Hebrew implied is:
(i) : ‫ ך ךן ד כ ה ו ק צ ר ; מ י א מ ר א ל י‬3 ‫ע נ ה‬
(ϋ) ‫ א מ ר א ל י‬: ‫ ך ךןי כ ח י ) ק ׳ כ חי( ק צ ר ; מ י‬.‫ענ ה ב‬
The vocalization implied by Symmachus and the iuxta Hebraeos
is semantically equivalent with the M T and departs wholly from
the approach of LXX, though the latter can be attached to exactly
the same consonantal text.1
Similar remarks may be made about the Talmudic evidence.
Here is Baba Bathra 21a-b, which deserves to be quoted in full,
and is one of the passages commonly cited to show that vocalization
signs were not yet in use:2
It is written: for Joab and all Israel remained there until he had cut
off every male in Edom (I Kings 11. 16). When Joab came before David,
the latter said to him: Why have you acted thus [i.e. killed only the
males] ? He replied: Because it is written: Thou shalt blot out the males
(‫ ) ז כ ל‬of Amalek (Deut. 25. 19). Said David: But we read, the remem-
brance 0/(Π!3Τ) Amalek. He replied: I was taught to say HDT. He [Joab]

1 See Gordis, Biblical Text, p. 59, who cites it as a passage where the Gallican
Psalter ‘substantiates beyond the shadow of a doubt a full verse and more of our
Massoretic Text and yet has not one word in common with it!’ The recon­
structed pointing for LXX is that of Gordis.
2 Quoted here after the Soncino version. The passage is mentioned by Roberts,
Text and Versions, p. 49.
214 THE MASSORETES, VOCALIZATION A N D E M E N D A T IO N
then went to his teacher and asked: How didst thou teach me to read ?
He replied : ‫ ז כ ר‬. Thereupon Joab drew his sword and threatened to kill
him. Why do you do this ? asked the teacher. He replied: because it is
written: Cursed be he that doeth the work of the law negligently. He said
to him: Be satisfied that I am cursed. To which Joab rejoined: [It also
says] Cursed be he that keepeth back his sword from blood. According to
one report, he killed him; according to another, he did not kill him.
This passage not only shows that at the time no system of vowel
points existed. It also shows that a reader of the biblical text was
understood to receive the vocalization from a ‘teacher’, who him-
self in turn received it from the tradition of teachers before him.
This is, after all, not entirely unlike the situation in modern
English, another language in which the orthography represents the
linguistic realities with at least as little precision as the representa-
tion of Hebrew in unpointed script. Children therefore learn at
school the spelling and pronunciation, i.e. the socially accepted
linkage between the language system and the writing system. In
the situation of the Talmudic legend the reader stood within a
tradition, even if more than one such tradition existed, and did not
invent the vocalization for himself. This need for teaching and
learning was perpetuated in Jewish usage even after the vocaliza-
tion was written, because the synagogue practice is still to read from
an unpointed text.
Moreover, the passage shows that a pronunciation tradition
agreeing with the later Massoretic marking was already in exis-
tence. The play in the legend on ‫ ז כ ר‬and ‫ ז כ ר‬is a midrashic
device very similar in type to the al-tiqre interpretations; and these,
as we have seen, are not evidence that the received text was not in
existence or was in any way in doubt. David’s reading, character-
istically, was the same as the Massoretic.1
Some discussion should now be given to Kahle’s well-known
theory according to which the Massoretes made substantial innova-
tions or restorations in the grammar of Hebrew.
The three points of Kahle’s argument are the following. Firstly,
for centuries before their time the laryngals or gutturals were not
1 It is slightly misleading when Roberts, p. 49, says that ‘it is significant that
the argument was settled, not by resorting to textual study, but by an authorita-
tive statement of exegesis’. More correctly, there never was any argument at all,
but only a legendary story built up on the similarities and differences of two texts,
with the familiar device of a change of vowels worked into a narrative form.
 E V ID E N C E F O R P R E -M A S S O R E T IC V O C A L IZ A T IO N 215
pronounced; the evidence for this is found in transcriptions, state­
ments by Jerome and so on. The Massoretes made every guttural
carry a consonantal value, followed by its own vowel. Secondly,
and on the basis of similar sources, plus non-Tiberian punctuation
and certain considerations of comparative philology, he held that
the 2nd person pronoun suffix was /-ak/, and the Massoretes altered
this to /‫־‬eka/. Thirdly, in the M T the six letters bgdkpt have a
double pronunciation, but according to Kahle there is no earlier
evidence of this. The double pronunciation fixed by the Massoretes
‘was earlier completely unknown to the most authoritative Jewish
circles’.1
Now if Kahle was factually right in claiming that these changes
were made by the Massoretes, it is conceivable that certain of them
might produce or contribute to confusions and losses of important
distinctions, and thus bring about the mistakes concerning the sense
of a word which only a philological treatment can disentangle.
These possibilities are for the present, however, no more than
simple theory. In my long list of philological treatments I have not
found examples which attributed the loss of understanding of the
text to the specific changes upon which Kahle has laid so much
weight. Conversely, we may say, philological treatment has not so
far confirmed the hypothesis of Kahle by producing solutions the
explanation of which would be assisted by his hypothesis.
The double pronunciation of bgdkpt is not important for our
present problem. The difference is non-phonemic and the two
possibilities are allophonic;2 the phoneme is realized in one form
or the other, determined by position. Taken alone it is not of great
semantic importance. This may, indeed, be one reason why some
early transcriptions ignored the matter; another reason is the
defectiveness of transcriptions in the writing systems of Greek and
Latin, in which the sets of possible phonemes and allophones are
very different from those of Hebrew.3
Thus, though Kahle’s view that the Massoretes were linguistic
innovators has had great influence, the innovations alleged by him,
even if true, are not such as to prove that the vocalization is generally
arbitrary, and especially so in its effect on meaning. Rather, they
are a group of limited alterations which in themselves could not be
1 Cairo Genista, p. 182.
2 Morag, Vocalization Systems, p. 24 and note.
3 See my article in J S S xii (1967) 1 3 6 ‫ ־‬, especially pp. 9 1 3 ‫ ־‬.
216 THE MASSORETES, VOCALIZATION A N D E M EN D A T IO N
responsible for large-scale changes of vocalization, intended to
produce a particular sense. Thus the position I have argued is not
in itself affected if Kahle’s views are correct.
In any case these views may be mistaken; they were disputed
when they were first announced, and important arguments have
recently been directed against them.1 Views of the Massoretic
activity based on them must now be reconsidered.
Moreover, philological considerations themselves raise certain
difficulties against the conception of the Massoretes as men who,
often ignorant of the older state of the Hebrew language, vocalized
it to fit their own conceptions of the meaning of the text. One such
consideration is offered by those philological treatments in which
the word studied is pointed anomalously in relation to the normal
Massoretic procedure. In such cases the existing pointing fits with
and supports the philological treatments, and can hardly be re-
garded as a screen of normalization cast over an ignorance of the
meaning. A good example is ‫ ד ע ה‬at Prov. 24. 14, which has been
interpreted as meaning ‘call’ or 'seek’.2 If this was the familiar verb
‫‘ י ד ע‬know’, then the vocalization is anomalous; GK (§ 48 1) is at
pains to discuss it, and has no good analogy to offer. If the verb is
a ‫‘ ד ע ה‬call’ or the like, the anomaly of the form is much less. But
if we accept this solution, we must also accept one of two alterna-
tives: either that the Massoretes actually knew the right sense here,
or that, though they took the word to be a form from ‫‘ י ד ע‬know’
(or simply did not analyse it semantically at all), they did not
normalize its punctuation into what would be expected under their
own usual procedure.
A similar case is the word ‫‘ ת ל מ ה‬treachery’.3 The anomaly
which made this a difficulty in the first place lay in the vocalization,
which did not conform to usual M T patterns. The Accadian parallel
advanced by Dossin, if right, confirms the Massoretic vocalization.
If the philological explanation is right, then the Massoretes refrained
from interfering with the vocalization of a word which was ab-
normal.
In addition to the existence of anomalies among the forms of
1 See in particular the (in my opinion devastating) criticisms of Kutscher,
J S S x (1965) 21-51; also, inter alia> Goshen-Gottstein in ScrH iv (1958) 117 f.,
and ‘The Tiberian Bible T ext’, esp. pp. 90 if.
2 Cf. above, pp. 23 if.
3 Index, no. 299. On the anomaly see Moore, Judges (ICC), p. 259; ‘an un-
exampled and really inconceivable type of noun’.
 E V ID E N C E F O R P R E -M A S S O R E T IC V O C A L IZ A T IO N 217
M T the apparent linguistic antiquity of many such forms has to
be taken into account. Wherever comparative philology succeeds
in fitting data from the M T into a comparative framework which
also accommodates material from other Semitic languages, this
success is a testimony to the general plausibility of the tradition of
M T.1 Phenomena like the waw consecutive were doubtless often
marked wrongly in detail; but the general nature of the pheno-
menon would hardly be known to us if the Massoretes had not
registered it adequately in many places, though it was something
they could not possibly have known about from the contemporary
linguistic environment. A vast amount of the detail in cur com-
parative grammars presumes and permits an affirmation of the
linguistic credibility of the work of the Massoretes.

(5) Conclusions
The traditional philology gave little place to the antiquity of a
liturgical style of reading a sacred literature such as the Bible. Much
philological study has preferred to find certainty in comparative
evidence, rather than in the analysis of received streams of tradition,
and has been sceptical towards the tradition of reading. But modern
linguistic methods have reopened the question, and striking results
have been achieved by the work of Morag on the reading traditions
of the Yemenite Jews.2 It is by no means true, as scholarly prejudice
has often supposed, that such traditions are ‘artificial’—which
would mean, presumably, the product of factors unconnected with,
and irrelevant to, the original situation of the literature being read.
The agreement between certain Yemenite phonetic features and
elements in the Babylonian system of pointing, the fact that the
Yemenites have different pronunciations for their reading of the
Bible and their reading of the Mishnah, and the existence in their
reading of biblical Hebrew of phones non-existent in their own
Arabic vernacular, all suggest the presence of valuable historical
evidence.
1 This argument is logically the same as that which Bergstrasser is said to
have made when Kahle first announced his theory of an innovating activity by
the Massoretes. The innovators, Bergstrasser argued, must in that case at least
have read Brockelmann’s smaller comparative grammar; how else could the
innovations have been reconcilable with use in a comparative reconstruction ? See
Hempel in ZAW lxi (1945-8) 251; Kutscher in JSS x (1965) 43.
2 S. Morag, ‫( העברית שבפי יהודי תימן‬Jerusalem, 1963)· See review article
by E. Y. Kutscher in JSS xi (1966) 217-25.
218 THE MASSORETES, VOCALIZATION A N D EM E N D A T IO N
Morag has discussed the question whether such traditions were
stabilized at a time when the original language was still a living
one. Such may be the case, he has suggested,1 when three con-
ditions are fulfilled:
(a) that the system used is stable and consistent in itself;
(b) that it does not conform to features known in the vernacular
of the time;
(*:) that its structural relation to other forms of the same language
group is capable of formulation in terms of historical
linguistics.
The growth and preservation of a liturgical reading tradition is
no doubt a special case within linguistic history; but its special
character by no means makes it unrelatable to the general processes
of linguistic change: only the conditions for such change are pecu-
liar. If, then, the Massoretes registered in their pointing the state
of such a tradition in their own time, there is no reason to doubt
that this tradition was connected with earlier stages of the Hebrew
language by lines of development analogous to other known pro-
cesses of linguistic change.
In preference to the sharp distinction between a reliable con-
sonantal text and an unreliable vocalization, one might rather hold
that the two aspects were interdependent. For example, graphic
error in the written text has sometimes been the occasion for con-
fusions in the vocalization. I would take this to be so in the parallel
texts quoted above, where Ps. 18. 11 ‫ ! י ל א‬is likely to be the right
text.2 The easy graphic error 1‫( י ל א‬II Sam. 22. 11) then gene-
rates the vocalization ‫ ל ל א‬which is natural to this writing.3 This
leads towards a textual rather than a philological solution; indeed,
what we are saying is only what was logically implied in the
traditional textual criticism. Moreover, this argument suggests
that it is quite wrong to believe in a reliable consonantal text later
wrongly vocalized through misunderstanding. The two are inter-
dependent, and an erroneous writing may sometimes generate a
vocalization which is suitable for that writing.4
1 In a paper given to the International Conference on Semitic Studies,
Jerusalem, 1965. 2 See above, pp. 192 f.
3 Cf. similar remarks about the case of ‫ ד כ י ו‬, above, p. 202.
4 If we suppose a stage at which the /d/ and /r/ were marked by the same
grapheme, then the case would be one of error in interpretation of a grapheme
which had two phonemic values.
 C O N C L U SIO N S 219

Sometimes we can see how analogy with another biblical passage

could have caused a new but wrong vocalization. One of the most
convincing philological treatments is Ginsberg’s suggestion that
at Prov. 26. 23, where M T has
‫ סף סיגי ם סצסה על־?!ךש‬5
we should read rather CT31DDDD 'like glaze’, as in Ugaritic spsg,
itself from Hittite or paralleled in Hittite. This gives a sense of:
'like glaze set over an earthen vessel . . .’. The close association of
‫' כ ס ף‬silver’ with ‫ סיגי ם‬in several other passages (Isa. 1. 22,
Ezek. 22. 18, and above all Prov. 25. 4, just a little before our
passage) may easily have suggested the reading of MT, once the
original ) 0 D0 was forgotten. Even here a graphic disturbance
(shift of word division) was probably included in the corruption.
Where wrong readings of the text have grown up, the indications
are that we should generally attribute them not to the ignorance of
late Massoretes but to a much earlier stage in the tradition. The
following points should be considered:
(a) The earlier period (let us say, before the second century a . d .)
was the time in which wide variations in the written text still
existed. These are not mere scribal errors, i.e. failures to copy
correctly a text already fixed in its letter. The state was a much
more fluid one, with different orthographies and different ways of
reading. It is comparatively easy to picture how false readings may
have arisen, where, for instance, a text written in one orthography
is construed as if it had been written in another; or, again, where
a text previously known orally came to be written down in the
conditions of varying orthographies. All this was before the
selection of one text-type limited the variety which gave licence
to error. We may contrast the great vagaries in the Hebrew text of
Sirach, a book which presumably did not enjoy the stability
accorded by this kind of official status for the text.
(b) Wrong vocalizations might well arise if a work had circulation
as a written document in certain narrow circles for some time before
it came to be read in the wider community, i.e. as a holy text in the
synagogue. Even if an esoteric group, such as a circle of prophetic
disciples, had known the right reading of a text, this knowledge of
theirs might fail to be passed on to the wider public; and in a more
extreme case, a book might be taken up as a written document,
without any mediation of a tradition of reading (must not this have
220 T H E M A SS O R E T E S, V O C A L IZ A T IO N AND E M E N D A T IO N
happened to Josiah’s Book of the Law?). Once a document had
come to be generally accepted reading in the synagogue, a careful
transmission of readings would begin. The absence of an efflores-
cence of substantial variants in the centuries from the second to the
eighth or ninth confirms this. The early stage provides a far more
plausible setting for confusions of vocalization than the time of the
Massoretes, or even Talmudic times, where there already was a
very firm tradition of reading.
(c) Something would depend on the degree to which the status
of ‘Scripture’ at the times concerned was deemed to imply close
and exact respect for the letter of the text. Such close and exact
respect, when applied at points where the text had in fact already
suffered graphic error, might be the origin of forced vocalizations.
(d) The testimony of the versions very often agrees with this. It
is often, and rightly, observed in philological treatments that the
versions like LXX, even if they understand the text wrongly, testify
to the same text as M T ; and, where this text is taken to be in error,
it implies that the origin of the error was early, possibly going back
to several centuries B.C. To take the example of Prov. 26. 23 studied
just above, the vocalization and division ‫ כ ס ף סיגי ם‬can hardly be
ascribed to ‘the Massoretes’; the various versions, though other-
wise differing, are clear that the word is ‘silver’. The text was
already like M T in this respect, at a time the best part of a
thousand years before the Massoretes.
(1e) This suggestion may also help to deal with a question which
may have arisen from our discussion of the vocalization. Since the
pronunciation of Hebrew changed, is it not likely that the Mas-
soretic vocalization is increasingly unreliable the older the text we
are considering? Will there not be varying degrees of reliability in
the vocalization, and perhaps in the text generally, between the
Song of Deborah, the sayings of Jeremiah, and the Hebrew parts
of Daniel?
The most drastic changes probably took place quite early within
the Old Testament period itself. Phonetic and grammatical change
may have caused obscurity in the Song of Deborah, but such
changes had probably already done their work by the time of
Jeremiah. A simple progression along a time scale therefore does
not at all measure correctly the degree of change. The changes in
question have the character rather of quanta, discrete steps of
change. To take an obvious example, it is possible that the Song of
 CONCLUSIONS 221

Deborah had vocalic inflexional endings; any disorganization

caused by the loss of these took place early, and probably took place
in one major step. On the other hand, since some of the confusion
of vocalization has arisen from the graphic form of the text, a late
text like the Hebrew Daniel is not thereby necessarily particularly
lacking in confusion.
It is now time to sum up the issues involved in the evaluation of
the Massoretic vocalization stated or implied by so many philo­
logical treatments. The following general estimates seem at present
to be possible.
Firstly, no one doubts that the traditional vocalization is subject
to error and may deserve emendation by scholars under the safe­
guard of proper consideration of the factors involved. There is no
evidence, however, that entitles us to carry this so far that we begin
to regard the vocalization as entirely arbitrary or chaotic and there­
fore subject to alteration on no greater basis than the liking of the
modern scholar. The vocalization is historical evidence just as
other aspects of the text are; it has to be explained and not merely
altered. In very few cases examined by me have philological treat­
ments given any full and adequate explanation of the vocalization.
Secondly, not only is the vocalization historical evidence formed
in the transmission of the text, like the consonantal or written text;
but it also is a part of the text which goes back to origins long before
the graphic marking of the vocalization by the Massoretes began.
This does not mean that the vocalization is ‘right’. It does mean
that it has to be investigated in relation to the modes of interpreta­
tion existing in Israel from early post-biblical times on; and only
through reference to the history of these modes can we determine
whether the vocalization is a natural linguistic growth from earlier
stages.
Thirdly, philologists in their constant appeal to the cognate
languages, when it is linked with a low evaluation of the vocaliza­
tion of Hebrew, often fail to realize how much their own system of
knowledge depends upon and draws upon the Massoretic vocaliza­
tion of Hebrew, both directly in that they repeatedly continue to
quote it as if it were reliable, and indirectly in that the modern
organizations of Hebrew grammar are themselves lineal descen­
dants from the earlier Hebrew grammar; and this earlier grammar
succeeded in making order out of the chaos of extant Hebrew
forms only because it carefully followed the Massoretic vocalization
222 T H E M A SSO R E T E S, V O C A L IZ A T IO N A N D E M E N D A T IO N
as a guide. The tendency to ignore the Massoretic vocalization is a
tendency to bring Hebrew into the category of languages such as
Ugaritic where no intermediate tradition of pronunciation exists.
But the possibility of organizing a scientific study of these languages
depends, perhaps more than has been realized, upon the existence
of the languages where an intermediate tradition of pronunciation
does exist.
All in all, of the various aspects in which one has to evaluate
modern philological treatments of the Old Testament text, the
position of the vocalization appears to be the most unsatisfactory.
Many of the necessary preliminary studies have not been done;
adequate proofs are seldom given; and the wide learning of com­
parative philology has often failed to give precision and care to the
establishment with evidence of the position it accords to the Mas-
soretes. In future, philological treatments will have to account for
what the Massoretes did, and not simply push it aside as the product
of ignorance.
 IX

LATE HEBREW AND THE LOSS OF

VOCABULARY

P h i l o l o g i c a l treatments imply that our difficulties have arisen

because words or their meanings came to be forgotten; correspond­
ingly, the vocabulary of Hebrew in biblical times must have been
larger than has normally been recognized. In order to consider
these implications, and to study the method of philological treat­
ments properly, we have to examine the vocabulary of post-biblical
Hebrew.
In such an examination there are a number of different aspects,
and, since these are commonly interlinked in the discussion of any
single example, I shall first mention five, and thereafter go on to
give illustrations which will involve some or all of them. The
aspects are the following:
(i) Most obviously, the disuse of a word in late times may be the
reason why its meaning ceased to be understood. If this is so,
the sense has to be recovered from cognate languages precisely
because within Hebrew itself it was lost from consciousness at a
relatively early date. This argument is particularly persuasive, of
course, in the case of those words which (a) are very rare within
the Old Testament itself, or (b) occur only in very early sources
within it, or (c) occur only in certain very specialized contexts
within it.
The situation is the same if a word, instead of ceasing to be used
in later Hebrew, came to be used with a different sense. This also
might explain why the older meaning came to be forgotten.
In neither of its forms, however, can this argument be used
without some examination of the evidence of post-biblical Hebrew.
One cannot assume that disuse in late Hebrew provides an argu­
ment for the loss of understanding of a word unless one has first
examined the resources of the post-biblical vocabulary, in order to
see how far they included words similar to or related to the biblical
word in question.
224 L A T E HEBR EW A N D T H E L O S S OF V O C A B U L A R Y
Our present reference works do not make this task easy for us.
The dictionaries of biblical Hebrew generally make no effort to tell
us about the history and currency of items after biblical times, with
some exception made only for Sirach and the Qumran texts. The
dictionaries of post-biblical Hebrew, on the other hand, suffer from
some defects in their planning, their comprehensiveness, and their
degree of scientific discrimination. The use of these reference
works is in any case made more uncertain for many biblical
students by their poor training and experience in post-biblical
texts. These difficulties, however, do not remove, but actually
increase, the importance of a proper consultation of post-
biblical Hebrew for any study of the transmission of linguistic
understanding.
(2) Secondly, a number of philological treatments do in fact
make their identification of obscure biblical words through appeal
to post-biblical Hebrew, or else include post-biblical Hebrew
evidence in addition to evidence from cognate languages. It may
be held that a treatment which can quote post-biblical Hebrew
evidence has more to be said for it than one which can quote
evidence only from other languages. At least it has succeeded in
showing that the word or the root in question did at some time exist
in Hebrew. It is thus free, other things being equal, from the suspi­
cion that the word and meaning appealed to are entirely an idio­
syncrasy of the vocabulary of another language; and it does not
require to use the doubtful principle that anything which occurs
in any Semitic language may be supposed to have occurred in
Hebrew. In this respect the use of post-biblical Hebrew evidence
appears to be a source of strength.
Against this, however, we have to weigh another consideration.
If late Hebrew evidence for our word is readily available, and
affords a sense which clarifies the biblical passage, it is no longer
quite so easy to see how a failure in understanding took place and
led to a loss of knowledge of the sense of the biblical word; thus
some additional explanation may be required. Relatedness to a
post-biblical Hebrew phenomenon is a kind of evidence which
may work negatively as well as positively.
(3) Thirdly, as we have seen, philological treatments commonly
tend to enlarge the vocabulary of Hebrew, and it has been argued
that the vocabulary must, for all the purposes of everyday life, have
been very much larger than that which is preserved in the Bible. I
 L A TE H EBREW AND T H E LO SS OF VOCABULARY 225
think that this is certainly true, but that its bearing on our present
discussion has not commonly been rightly seen.
The argument is that there are many objects which the Israelites
had, and actions which they did, in biblical times which are not
mentioned in the Bible and for which we therefore do not know the
name in biblical Hebrew. Thus Driver, for example, has repeatedly
emphasized that Hebrew vocabulary was larger than has previously
been recognized.1 It is, he urges, mistaken ‘to consider that which
is preserved in the M T sufficient even for the limited needs of daily
life in ancient Palestine’.2 Albright has stated that ‘the known
biblical Hebrew vocabulary cannot represent over a fifth of the
total stock of North-west Semitic words used between 1400 and
400 B.C.’3
In a sense, probably no one would doubt this. The Bible is a
very limited segment of all that was said in ancient Hebrew. Words
have been found in our quite small Inscriptional material which
were certainly normal Hebrew but which do not appear in the Bible.
The ‫ א שו ח‬or ‫ א שי ח‬which appears in Moabite in the Mesha*
inscription, and in later Hebrew at Sir. 50. 3, meaning ‘reservoir’,
was surely normal Hebrew. The form ‫ י ם‬D at I Sam. 13.21 was not
recognized, and the text was regarded as corrupt, until quite recent
times, when the evidence of artefacts showed that this was the name
of a weight, ‘pirn’ or the like. Nevertheless modern discoveries of
extra-biblical documents earlier than the Mishnaic period should
make us cautious about claiming that a very large additional
vocabulary is likely quite easily to be found. Though Palestinian
inscriptions, the Hebrew Sirach, and the Qumran documents have
produced some new words, their number should not be exag-
gerated; it is in fact quite limited.
Good examples from Sirach are ‫‘ ש ר ק‬shine’ (43. 9; 50. 7; cf.
Arab. ’asraqa), ‫‘ ל צ ל‬observe’ (14. 22; cf. Arab, rasada) and ‫ל ק ו ת‬
‘punishment’ (9. 4; cf. the place-name Eltekeh, which indeed
implies rather a sense of ‘meet’). The Dead Sea Scrolls also show
only a limited number of words of Semitic origin not found in the
Bible; a good example is the phrase ‫‘ א ו ח ז י א ב ו ת‬intercessors’,
known in Syriac and going back to Accadian.4 Driver himself

1 e.g. J T S xxxi (1930) 275.

2 J B L lxviii (1949) 57 f-
3 In Peake’s Commentary on the Bible, 2nd ed. (1962), p. 62.
4 For discussion see Wernberg-Moller, p. 53.
226 LA TE HEBREW AND T H E LO SS OF VOCABULARY
asserts only that ‘a very few’ words require explanation from other
languages; he identifies among others a fll*! Vehemence’ (with
Accadian cognate), a ]1‫‘ ש א‬purpose’ (cf. Arab, saa), and an ]‫ע י‬
‘company, group’.1 Of other new words found in the Qumran
documents, a number are not of Semitic origin, e.g. ‫נ ח שי ר‬,
usually understood as ‘hunting, pursuit’.2 In some cases the inter-
pretation is itself highly controversial. Anyway, it cannot be said
that the discovery of fresh documentary evidence has produced a
large number of words not already known in biblical vocabulary.
It nevertheless remains true that we should expect some con-
siderable number of words for things and actions not mentioned in
the Bible, and the lack of such in Sirach and at Qumran could con-
ceivably be accounted for on the ground that these documents are
devoted to the same general kind of subject matter which is found
in the Bible. Now post-biblical Hebrew contains a large number of
just such words, i.e. names for things which are not named in
biblical Hebrew. It is generally probable that the words designating
such objects in the Mishnah or the Talmud are also the words
which were used in biblical times, unless there is some particular
reason to the contrary. Doubtless some of these words are new
formations, replacing an older word of which we have no know-
ledge; but it remains probable that many of them are old. I have in
mind such words as $ 3 3 ‘preserves’, ‫‘ ג ח ר‬jetty’, ‫‘ זו ל‬cheapness’,
‫ךת‬.‫■ ןחש‬hump’, ‫ל קן ת‬$ ‘scar’, ‫■ ש ע ה‬hour‫׳‬, 3 ‫‘ ג פ ה‬stone fence’,
and ‫‘ ג ב ב ה‬rakings’.
Thus, in so far as the argument for philological treatments
involves the appeal to objects and actions which are not named in
the Bible, we should expect it fairly frequently to point to the
language of Mishnah and Talmud as a source of evidence—always
providing, of course, the necessary safeguards against mistaken
use of words for processes which were technologically new in the
Hellenistic period, or other innovations of post-biblical times, such
as new religious or political developments.
Here, however, we come to another important distinction.
1 See Driver, The Judaean Scrolls, p. 435. The suggestion of ‫( עין‬or ‫) ע ץ‬
by Yalon is discussed by Wemberg-Moller, p. 59. Even among the suggestions
listed by Driver, not many have found general agreement among scholars.
2 Cf. de Menasce, V T vi (1956) 213 f. Rabin, Orientalia xxxii (1963) 132 f.,
interprets it rather as ‘terror, panic*.
3 On this word see my Biblical Words for Time, pp. 102 if.
 L A TE H EBREW AND T H E LO SS OF VOCABULARY 227
Though the point we have just made justifies us in expecting that
research will make more extensive the Hebrew vocabulary known
to us, this does not apply to quite any area of that vocabulary. It
applies well, as we have seen, to objects for which we do not have
a name in the Bible. It does not, however, apply so well to objects
or actions for which we already have a name, or even two or three
roughly synonymous names, in the Bible. In point of fact, in the
philological treatments which I have collected, not a large propor­
tion refer to objects or actions for which we hitherto have not known
the name. On the contrary, the larger proportion have reference to
objects or actions for which names are already well known.
Thus the argument that the vocabulary of ancient Hebrew must
have been much larger than is known from the biblical text, while
undeniable in itself and in theory, does not serve to justify a parti­
cularly large number of the new identifications of words which
philological treatments have produced. The validity of the argu­
ment depends on the semantic field of the words concerned. For
the semantic fields which were primarily involved in biblical litera­
ture (particularly religion and the human states and institutions
connected with it) it is quite possible on the contrary that biblical
usage exploited almost all of the lexical resources which were
available.
(4) Fourthly, we have the problem of detecting what I shall call
‘restorations’ in the post-biblical literature. When a word used in
biblical Hebrew appears also in post-biblical material, this may
mean that a continuity of general usage existed throughout both
periods. It may, however, also mean that a word found in the
biblical text has been taken up and renewed in later literature, on
the basis not of continuity in general usage but of reference solely
to the limited biblical texts where the word occurs. The word is
thus one quoted, and from this quotation it enters into a new lease
of life. The sense which it now has, however, is not necessarily
identical with that meant by the author in the original biblical
milieu. On the contrary, it may be a sense attached solely to an
occurrence in the biblical text as it was later read and understood;
and this text may have been affected by graphic accidents and
errors of transmission which have nothing to do with the con­
tinuity of linguistic meaning.
It may not be easy to decide in a post-biblical text whether
restoration of this kind has taken place. Two criteria may be
228 LA TE HEBREW AND T H E LO SS OF VOCABULARY
helpful: firstly, literary type, and secondly, frequency of occur-
rence.
Restorations, in the sense here intended, are most common in a
highly allusive style, in which deliberate arcane reference is made
to known biblical locutions. Such locutions may become technical
in an esoteric current of exegesis, or may be cultivated in a learned
poetic style. The Qumran hymns and books of discipline, and the
type of liturgical poem called piyyut, are examples. If the stylistic
setting is more conversational and straightforward, the likelihood of
such restorations is somewhat lower.
Frequency of occurrence is a criterion which more or less follows
from the above. If a usage occurs with some frequency in the large
late Jewish literature, it is on the whole less likely that an arcane
reference to a particular biblical text is being made. This is not a
certain criterion, for a usage might arise by restoration and then
spread so as to become popular; but at least it is a consideration
worth some weight.
(5) A fifth consideration is the influence of Aramaic. Those Jews
who, in post-biblical times, were in a position to discuss the mean-
ing of biblical words and to relate them to current usage would
generally know Aramaic as well as Hebrew. Thus, for the purpose
of studying the transmission of understanding and the loss of
understanding, similarities between biblical words and Jewish
Aramaic words have to be taken into account in the same way as
similarities between biblical and post-biblical Hebrew. A distinc-
tion should also be made between types of Aramaic which were
probably current among Jews and types which were not; lexical
idiosyncrasies of the latter should not usually be brought into this
comparison.
These then are the five general considerations which should be
borne in mind. I shall now proceed to discuss some examples in
which one or more can be seen in practice. I shall first of all
exemplify some restorations, since of the various considerations
this is the one which is most easily isolated from the others.
A good example is the word ‫ ח ל כ ה‬or ‫ ח ל כ א י ם‬which occurs
thrice in the difficult Ps. io.1 BDB doubted the text, but also tried
an explanation of the word as related to an Arabic halaka ‘be
1 The division of the word into two parts, ‫ ח י ל כ אי ם‬, by the Qere at Ps. io. io
is an example of what might begin as a graphic change but achieve an effect on
later linguistic restorations, as suggested above, p. 218.
 L A T E HEBREW AND T H E LO SS OF VO CA BU LARY 229
black*, and passing from there into the sense ‘unfortunate*. This
is itself not a difficult semantic shift, being paralleled by ‫ ק ל ל‬for
example. KB has even greater doubts, and declared the word to be
‘unexplained*. Komlos suggested that the ‫ ל‬was a formative ele-
ment, and derived the word from ‫‘ ח כ ה‬hook*.1 There continues
to be some doubt about the sense and etymology.
Now the Qumran hymns show four examples of this word (1QH
3. 25, 26; 4. 25, 35). For the most part scholars seem to take the
sense to be something near the ‘unfortunate’ of BDB, but make it
perhaps a little stronger. Mansoor, on the other hand, pointing out
that the parallelism at 3. 25 and 4. 35 is with ‘the wicked*, thinks
that the sense is ‘wicked, tyrants*.2
Even if we can certainly work out what the word means in the
scrolls, we do not thereby necessarily show that this was the sense
in the Bible. At several places in the scrolls it seems likely that a
particularly obscure word or phrase in the Bible is picked up and
given a heavily loaded sense; and one of the reasons for this may
be precisely that the word was so enigmatic in the first place. This
is likely to be done by any exegetical technique which holds the
very details of the Bible to contain mysterious, yet quite specific,
references to incidents and persons of a later time.
Meanings may have changed during the long history of use of
the texts. Firstly, if it is true that certain stock parallelisms were
held in common with pre-Israelite poetry, it is possible that items
which are formally identical with Canaanite or Ugaritic materials
may nevertheless have come to be understood and used with another
sense in Hebrew. Secondly, within Hebrew itself, in texts like the
Psalms which were used over a long period, it is possible that another
sense has sometimes taken the place of that originally meant, and
that homonyms which we can distinguish philologically may have
come to contaminate the sense of one another.3 Where literature
was in continual use, like the Psalms, therefore, it is an over-
simplification to suppose that there is one Hebrew meaning which
has to be found. Changes of meaning may have taken place not only
as biblical texts came to be interpreted in post-biblical times, but
1 In J S S ii (1957) 243-6. 2 Mansoor in V T S ix (1963) 316.
3 Thus the words ‫‘ י ל ה‬throw* and ‫‘ י ל ה‬instruct’, though probably philo-
logically distinguishable (so GB, pp. 317 £.; KB, pp. 402 f., but not BDB, p. 434b;
the older tradition is to derive the latter sense from the former in the meaning
of throwing or casting lots), may have come to contaminate each other in the
course of time, especially through certain forms such as ‫ תו ר ה‬.
230 LA TE HEBREW AND T H E LO SS OF VOCABULARY
also within biblical times themselves, if these texts were used and
re-used over a prolonged period.
Another instance is the ΠΤΓΙ or ‫ חזרת‬which occurs only in
Isa. 28. 15, 18. A sense of ‘agreement, contract, covenant’ is sug-
gested by the parallelism with ‫ ב ל י ת‬and also by the evidence of
most versions (LXX συνθήκας; Vulg. pactum; Targum ‫ש ל ם‬
‘peace’; but Syriac hezwa, perhaps meant as ‘vision’, and in any
case a caique on the Hebrew word). A completely different under-
standing is found only in the Hexaplar entry άλλος· καταφυγήν.1
Driver has provided comparative evidence which confirms this
sense (ESA hdyt ‘agreement’, especially in the phrase bhdyt ‘in
conformity with’). This seems greatly to surpass the attempts to
explain the word, on the basis of the verb ΠΤΠ ‘see’, as some kind
of prophetic advice (so BDB, p. 302b; GB, p. 221a; KB, p. 285,
accepts emendation into ‫ ח ס ל‬for both cases). We may hold it very
probable that the versions quoted were right. The word is then
cognate with Arabic h-d- ‘be opposite’ etc., and thus also with the
familiar ΠΤΠ ‘breast (of a sacrificial animal)’.
This example is interesting because the meaning has survived
into medieval and modern Hebrew. Curiously, however, I do not
find a case quotable from the Talmudic period. Perhaps the later
Hebrew use is a restoration, prompted by the parallelism with
‫ ; ב ר י ת‬but if so it is a right restoration. I am on the whole inclined,
however, to consider it more likely that the tradition of meaning
survived; but, if this is so, I am uncertain how to account for the
lack of use in‫ ׳‬the Talmudic period.
A similar case is the obscure word ‫ ת כ ו‬. This hapax legomenon
occurs only at Deut. 33. 3. BDB, GB, and KB have no explanation
of the meaning; B H 3 offers the choice of two colossal emendations.
The sentence is:
mt ‫והם תכו לרגלך‬
Vulg. et qui appropinquant pedibus eius
Now Stummer has pointed out2 that in a piyyut of Yannai (sixth
century a .d . ?) the verse occurs:
‫י ען תו כ ה ל ר ג ל ך ו ב צ ד ק ק ר א‬
‫ע ל כן ל ר ג לו צ ד ק הו ק ר א‬
1 Is this the result of taking the Hebrew word as one related to nODD, which
is translated καταφυγή by LXX at Ps. 45 (46). 2, 103 (104). 18?
2 Notscher Festschrift, pp. 265 if.
 LA TE HEBREW AND T H E LO SS OF VOCABULARY 231
meaning perhaps:
*because he approached your foot and called in righteousness;
therefore righteousness is made to meet his foot.’
This may possibly be evidence that the verb was used in a con-
tinuity of meaning from biblical times down, and may thus serve
towards a recognition of its actual sense in the biblical text; it may,
on the other hand, be deemed to have been a restoration.1 In that
case, Yannai belongs to the same tradition which is represented by
Jerome; but this would not mean that this tradition is *right’.2
Again, medieval Hebrew has a verb ‫‘ ל ה ק‬gather together’, and
the noun ‫ ל ה ק‬. The prayer by Qalir for Rosh-ha-Shanah has the
phrase ‫* שו אגי ם ב ל ה ק‬shouting aloud in unison’; similarly,
Saadia writes ‫ ו ה מ ל ך י ק ר א נ ה ל ע ם ב ל ה ק י ם‬, it would seem
likely that this is a restoration (and new formation) on the basis of
the assumed sense of ‫ ל ה ק ת‬at I Sam. 19. 20.3
Two hapax legomena which are mentioned elsewhere in this book
are the ‫ א ל׳ ״ ת ר ה ו‬of Isa. 44. 8 and the ‫ וינהל‬of I Sam. 7. 2.4
Both of these occur in later poetry. A piyyut published by Dr.
Wallenstein includes the lines:5
‫ו ד ב ר י ד ב רו ו ב ל י ך הו‬
‫מ מ סי כ י ר ע ל ש ד מו ת ם‬
*And they will surely speak and not be afraid
of those who pour confusion (on) their fields.’
A poem ascribed to Saadia6 includes both of our two words to-
gether, and the senses are *fear’ for ‫( ל ה ה‬here hiphil, *frighten’)
and ‘follow obediently’ or the like for :‫נ ה ה‬
‫ר אי ת ה צ ר י ל מ ח ת ה ל כ ל ־ נ ו ה י ה ם‬
‫ ב ט ד ו ד י ה ם מי ר אי ה ם ו מ ך הי ה ם‬$

1 For a recent discussion of the evidential value of late (medieval) poetry in

textual criticism see A. Mirsky in Textus iii. 159-62.
2 Another treatment is to read as one word ‫ ו ה מ ת כו‬and construe as from
‫‘ מ כ ך‬be low’, giving the sense ‘they prostrate themselves’: so, for instance,
Cross and Freedman, JB L lxvii (1948) 200 f.
3 See above, pp. 25 f.
4 See above, pp. 6f., 166, 188, and below, 264 f.
5 Wallenstein, Piyyutim, p. 26.
6 Siddur R. Saadia Gaon, p. 4 1 7 , 1. 21.
z32 L A T E H E B R E W A N D T H E L O S S O F V O C A B U L A R Y
We may translate:
‘You have fated the destruction of adversaries along
with all their followers—(the destruction of) the
scourge of those who dominate, frighten and terrify them.’
Both of these words, then, have to be considered as possible
restorations.
In order to give a full estimate of the nature and course of
probable restoration, one would have to trace the development of
styles in the use of vocabulary right up to the Middle Ages. Rabin
writes:1
When the written use of biblical Hebrew was revived, after a long
pause, in the middle ages, all stages of biblical Hebrew served equally
as model, with special emphasis upon poetic and rare expressions.
Rabin traces the beginning of this revival of biblical Hebrew to
the ninth century.
A possible test for restorations may be the following: if a word
is found in both post-exilic and post-biblical sources, the chances
for linguistic continuity in use are greater; if the word is found in
the Bible only in very early sources, and then not again until late
post-biblical times, the likelihood of a restoration is the greater.
For example, ‫ ל ה ק ה‬falls within the latter type.

Having exemplified the problem of restorations, we may now go

on to look at other possibilities.
As we have said, the disuse of a word in late times may be a
reason why its meaning ceased to be understood. There are a
number of ways in which this may have happened.
A difficult example is the history of understanding of ‫י ש פי קו‬
at Isa. 2. 6:
: ‫_ילדי נסרים יקזפיקו‬5‫ ו‬. . . ‫כי מלאו מ?ןךם‬
The root ‫ ש פ ק‬or ‫‘ ס פ ק‬be enough’ actually occurs in the Bible
(the verb in I Kings 20. io, a noun form ‫ ? צפ קו‬at Job 20. 22),
and several cases are found in Sirach, there spelt 0 ‫ פ ק‬. The sense
‘have enough’ is very familiar in later Hebrew, and in Isa. 2. 6
it provides a good parallelism with the other verb ‫ מ ל א ו‬. Yet
most interpreters have rather turned to the somewhat artificial
1 ScrH iv (1958) 149.
 LA TE HEBREW AND T H E LO SS OF VOCABULARY 233
interpretation as ‘strike hands’ (i.e. in a bargain), which has some
support in Arabic idiom and, with limitations, in the rendering
of Symmachus, quoted by Jerome as et cum filiis alienis applause-
runt. Buchanan Gray,1 rejecting the sense ‘be enough’, says that
‘the use by Isaiah of a verb with such a history is not very probable’.
But it is probable that the LXX understood the word as ‘be enough’
or ‘be plentiful’, in rendering:
/cat τέ κ ν α πολλά α λλό φ υλα €γ€νηθη α ύ το ΐς.

The Vulgate has et pueris alienis adhaeserunt‫ י‬and this also illus-
trates the influence of late Hebrew. Jerome writes:
pro quo scriptum est in Hebraeo, iesphicu, quod Hebraei
interpretantur, έσφ ψ ώ θησαν, et nos uertimus,
adhaeserunt.2
The έσφ ηνώ θ ησ α ν of the ‘Hebrew’ interpretation seems to
depend on the post-biblical verb ‫‘ פ ק פ ק‬drive in a wedge’ (Jas-
trow, p. 1211a).3 It is not clear to me how Jerome got from this to
his adhaeserunt.
The sense ‘strike the hands’, already known in antiquity, came
from the biblical verb ‫ ס פ ק‬or ‫ ש פ ק‬.
It is interesting that the more obvious late Hebrew sense, that
of ‘be enough’ or ‘have enough’, which is also the more probable
sense philologically,4 nevertheless did not succeed in maintaining
itself in ancient times against other interpretations, one of which
also rested on post-biblical Hebrew but on a much less satisfactory
basis in it. It may be something in the exegetical tradition of the
passage that has caused this. In any case, it is clear that the influence
of post-biblical Hebrew did not always work in the way that would
seem to us most obvious.
At Hos. 8. 13, m t ‫ ז ב ח י ה ב ה ב י יז ב חו‬, one suggested inter-
pretation follows a post-biblical word ‫‘ ה ב ה ב‬singe, roast light-
ly’;5 the meaning would be something like ‘burnt-offerings’. At
1 ICC, Isaiah, p. 58.
2 Jerome’s iesphicu here is another vocalization, which, whatever we say about
the ef is clearly in agreement grammatically with the hiphil form of M T.
3 A ‫ ש‬, where it is first radical in verbs, is sometimes taken as if it were a pre-
formative as in shaphel forms. Cf. Isa. 41. 10 ‫ א ל״״ ת ש ת ע‬, LXX μη πλανώ—
surely a construction as from the common ‫‘ ת ע ה‬go astray’.
4 This sense is supported for Isa. 2. 6 by Winton Thomas in Z A W lxxv
(1963) 88 f., and also in KB, p. 928.
5 Index, no. 103; cf. earlier BDB, p. 396b; GB, p. 172b.
234 L A T E H EBR EW A N D T H E L O SS OF V O C A B U L A R Y
Prov. 30. 15 a similar word could then mean ‘burn (with erotic
passion)’. Yet it is doubtful if the versions show evidence of in-
fluence from the post-biblical word. Renderings with α γ α π ά ν
suggest simply an analysis as from ‫ א ה ב‬, while those with φερ€
φ6ρ€ are an analysis as from the exhortation ‫ ה ב‬, ‫ ה ב ה‬. The
Targum’s ‫‘ ד ב ח י ן ד מ ג ב ן מן או ני ם‬sacrifices which they collect
from extortion’, and the Syriac dbJi dgbyt\l also suggest that the
existence of ‫ ה ב ה ב‬did not have much influence on the under-
standing of the passages.
Another example is the Talmudic Hebrew ‫‘ ל ב ו ד‬compact,
solid’, used of the building of walls; Syriac has the corresponding
verb, meaning ‘thicken, solidify’. Tur-Sinai believes that the Bible
had this word. He reads Job 38. 30 as:
: ‫כ א מ ן מי ם י ס ח ב א ו ו פני ת הו ם י ס ל ב ד ו‬
and translates:
‘The waters are hidden as (behind) stone,
and the face of the sea sticketh together.’
He takes this to mean that there is no hole in the heavenly firma-
ment. Since the M T is ‫ י ת ל כ ד ו‬, the explanation assumes a con-
sonantal corruption, as also at 41. 9 (but not at Isa. 5. 8). Since,
however, ‫ ל ב ו ד‬was a technical term in Talmudic law, it is a little
difficult to suppose that, if it was originally present in the text, it
would fall out through misunderstanding. The evidence of late
Hebrew is therefore somewhat unfavourable to the interpretation.
The rare word ‫‘ י ק ה ה‬obedience’ occurs only at Gen. 49. 10,
Prov. 30. 17. The identification of this word is well established
from comparative sources, e.g. both North and South Arabic. Late
Hebrew and Jewish Aramaic on the other hand seem to show no
trace of continued use. Translations such as LXX π ρ ο σ δο κία ,
Vulgate exspectatioy probably derive from the likeness to the
common ‫‘ קו ה‬hope’ (Lam. 2. 16 προσδοκάν). The only faint
association in late Hebrew is with ‫‘ ק ה ה‬faint, long for’ (Jastrow,
p. 1322a). Yet the Targum with its ‫ו ל י ה י ש ת מ עו ן ע מ מ י א‬
seems to indicate that the right sense was known. The complete
disuse of a word, perhaps over many centuries, does not seem to
have damaged the preservation of its form and sense; even the
vocalization is in a credible form, and Rashi pointed out that the
1 Brockelmann, Lexicon Syriacum, p. 100b.
 L A T E H EBR EW A N D T H E L O SS OF V O C A B U L A R Y 235
formation is the same as in ‫ ש מ ח ת‬. The absence of any other
obvious word with which confusion might have arisen may have
assisted its preservation.
We mentioned above1 the identification by Driver at Jer. 12. 9
of a word ‫‘ צ ב ו ע‬hyena’. In favour of the identification of such a
word, and against the more normal construction as the passive
participle of the verb ‘to dye, colour’, one or two facts can be
quoted. There is not only the biblical place-name ‫ צ ב ע י ם‬,
which has been construed as ‘valley of hyenas’ ;2 there is also the
post-biblical noun ‫ צ ב ו ע‬, commonly taken to be a hyena, leopard,
or other beast.3 Once again, however, this very fact, while it con-
firms the likelihood that a word ‫ צ ב ו ע‬for hyena existed in the
biblical period, also makes it harder to understand how the con-
sciousness of this meaning at Jer. 12. 9 was lost if in fact this had
been the right meaning. Not only did the word ‫ צ ב ו ע‬exist, but
some discussion of the attributes of this strange animal is found in
the literature.4 Moreover, the translation of ‫ עי ט צ ב ו ע‬by
<j7TrjAcuov vaLvrjs in the LXX here may itself be not a right diag-
nosis of the biblical sense, but a construction derived from aware-
ness of the post-biblical 5.‫צ ב ו ע‬
The possible influence of Aramaic is illustrated in the suggested
identification of a ‫‘ ש ל ה‬pour out’, with the noun ‫‘ ש לי‬downpour,
outpouring’.
Ezek. 1. 24: MT ‫‘ קו ל ~ ש לי‬the voice of the Almighty’
read ‫‘ קול״״ ש לי‬the sound of a downpour’
II Sam. 1. 21: M T ‫‘ ש לי ה לו מ ת‬fields of offerings (?)’
read ‫‘ ש לי ת לו מ ת‬outpourings of the depths’
But if this Hebrew word existed, it is unlikely that interpreters
would lose touch with it, because of the presence of the Aramaic
‫ ש ל י‬, which is very common. The Aramaic ‫ א ש ל‬also means ‘pour
out’ e.g. Targ. Mic. 1. 4. In B. Ber. 54b the biblical noun ‫א ש ל‬
(Num. 21. 15) is interpreted as ‘poured out’ (‫ ;) ש פ ך‬we can be sure
that this Aramaic verb was present in the minds of interpreters.
These considerations make it more difficult to accept the sugges-
tion of a Hebrew ‫ ש ל ה‬with the same sense.
1 Cf. above, p. 128. 2 See BDB, p. 84
3 See Jastrow, p. 1257b. 4 See references in Jastrow, ibid.
5 Cf. the discussion below, pp. 260 ff.
236 LA TE HEBREW AND T H E LO SS OF VOCABULARY
We may conclude with the curious example of a word which,
reconstructed for the biblical text, even achieved mention in a
dictionary of modern Hebrew, and yet probably had no genuine
history at all: the supposed ‫‘ ע מ ל ץ‬shark’. The M T of Ps. 74. 14
has:
‫ ציי ם‬1? ‫ת ה עו מ א כ ל ל ע ם‬
and a reorganization of this text as
‫ת ת עו מ א כ ל ל ע מ ל צ י ; ם‬
‘you gave it as food to the sharks of the sea’1
has received considerable recognition and acceptance. It came from
I. Low, and Koehler calls it ‘simple and brilliant’ ;2 it is registered
in KB, p. 715b.
The positing of a word ‫‘ ע מ ל ץ‬shark’ seems intrinsically very
attractive; in the end, however, one must doubt its correctness.
Apart from Hebrew itself, it seems quite doubtful whether cognates
of the supposed ‫ ע מ ל ץ‬have ever existed in any Semitic language
as names for any kind of fish. The appeal made by Low was to the
Arabic m-l-s. The change from an assumed ‫ א מ ל ץ‬to ‫ ע מ ל ץ‬is
not in itself very difficult. Greater difficulty lies in the meaning of
the Arabic itself. The verb m-l-s seems to mean ‘be slippery’.3 A
form ‫י‬amlas is actually cited with the sense of a ‘smooth-headed
man’, and also of ‘a thing that slips out of one’s hand by reason of
its smoothness’. But for ‘a fish that slips from the hand by reason
of its smoothness’ the word is malisa. All senses quoted by Lane
and relevant to fish are senses for fish which slip out of the hand.
It is quite doubtful whether 5amlas (or any other word from this
root) is the name for a kind of fish.
Thus the comparative basis for the suggestion is not strong.
When we turn to consider the evidence in extra-biblical Hebrew,
the position is again doubtful. Has the word ‫( ע מ ל ץ‬or ‫) ע כ ל ץ‬
ever been used in Hebrew ? The dictionary of Grossmann, revised
by Segal, indeed enters this word with the gloss ‘shark’, but marks
it as a ‘modern’ word. I have not been able to find any previous
history of usage. There is no entry in Ben-Jehuda. The suggestion
1 Cf. the confusion of ‫‘ ים‬sea’ and the plural ending implied by a familiar
reconstruction at Am. 6. 12.
2 See Koehler, D Lz nf ii (1925) 1055; O T S viii (1950) 151.
3 See Lane, vii. 2736. Cf. also the consideration mentioned above, p. 165.
 L A T E H EBR EW A N D T H E L O SS OF V O C A B U L A R Y 237
for Ps. 74. 14 is indeed mentioned in that Thesaurus, xi. 5459b; but
even there it is treated only as a suggestion, and a doubtful one at
that. This hardly suggests that the word had an authentic history;
the entry in Grossmann-Segal seems itself to be a modern
proposal made on the basis of Low's suggestion for Ps. 74. 14! If
so, it seems hardly to have succeeded, for the word actually used
for ‘shark’ is tZTHS.1
•T
Such then are some illustrations of the use of evidence from
post-biblical Hebrew. Much more requires to be done to investi­
gate the lexical relationships between the biblical and the post-
biblical language. The more obvious similarities, we may expect,
have long been observed; while modern investigation has tended
to look rather towards the other cognate languages than towards
late Hebrew. But even when suggestions have been inspired
primarily by evidence from outside Hebrew, the post-biblical
Hebrew evidence can do much to furnish a cross-check and to help
us evaluate the probability of the suggestions.
1 I now find that our strange word is a technical term for one species of
shark in modern usage. This however does not alter the argument.
 X

THE USE OF EVIDENCE FROM

THE VERSIONS

(i) General
T h is chapter will be devoted to a discussion of the way in which
the early Versions’, i.e. the translations from the Hebrew Bible
into Greek, Latin, Aramaic, and other languages, have been used
in philological treatments.
In textual treatments the versions have commonly been evaluated
as a separate witness to the original text. That is to say, a version
like the LXX, in the absence of substantial variants in the Hebrew
MSS., gives testimony to a text many centuries earlier than the
MT, and thus enables us to get behind the M T and reconstruct an
earlier form of the Hebrew text. This reconstruction fits with the
context, explains how the LXX reading originated, and provides
some reasonable possibility, under the accepted canons for prob­
able scribal mistakes, of explaining also the origin of the present
M T reading. Actually the procedure is considerably more compli­
cated than this; but this statement will suffice for our purpose.
The essential thing is that, in this approach, the M T being con­
sidered difficult or doubtful, the LXX (or other version) is used as
a source from which a Hebrew text different from M T is recon­
structed. This implies retranslation from the version.1 This is
sometimes carelessly phrased by saying that the LXX ‘read’ such
and such a Hebrew word.
The traditional term ‘read’ is ambiguous in this context. When
scholars say that the LXX ‘read’ such and such a Hebrew word,
this may mean that the Greek text, if retranslated into Hebrew,
would provide such and such a Hebrew text; it may also mean that
such and such were the actual signs on the manuscript from which
the translation was made. We have no direct means of knowing
what they ‘read’ in the latter of these senses, and all we can tell
1 Goshen-Gottstein, in a penetrating recent study of methods in the use of
the versions, uses the term ‘retroversion’; see Textus iii (1963) 130-58.
 GENERAL 239
directly is what they wrote in Greek (or, respectively, in another
language). From what they wrote in Greek, taken in relation to
other known characteristics of the version, we can form hypotheses
about the Hebrew text from which the translators worked.
Knowledge of this Hebrew text, however, is indirect rather than
direct. In the use of a version there are two steps to set against one
in the transmission or quotation of a text in the same language.
There are the signs seen in the original text, and there is the pro­
cess of creating a new text in a new language. Since we have direct
access only to the final product, i.e. the text produced in the new
language, it is by a process of indirect decision that we determine
how much of this new text is to be attributed to decisions of the
translation process and how much is to be attributed to the signs
of the original text.
This indirectness of the use of the versions is not mentioned in
order to suggest that the versions are not reliable as a source of
guidance; that the procedure is an indirect one is, in my opinion,
a mere matter of fact, which has to be borne in mind in all our
judgements about the relation of versions to the text from which
they were translated.1
In any case the procedure of reconstruction of the original text
by retroversion from the LXX or other translation is a very
familiar one, perhaps the most familiar of all textual procedures
with Old Testament texts at the present time. Many hundreds of
suggestions in B H 3 follow this pattern.
This procedure is akin to, and often complementary with, the
procedure of emendation. Since the version does not provide direct
evidence of a different Hebrew text, there is a conjectural element
in any textual suggestion, however probable, which depends on
versional evidence, which would not be present if we were weighing
different Hebrew manuscripts. Thus textual reconstruction on the
basis of versions is easily allied with the more general practice of
emendation, which may either have or lack versional support; and
a very large number of emendations in fact make some sort of
appeal to the versions. There is, however, a wide diversity in the
degree to which emendations ‘on the basis of LXX’ (or similar
1 My position is, I think, exactly the same as that taken by Goshen-Gottstein
in the sample edition of Isaiah; e.g. p. 13, *a reading inferred by retroversion,
however plausible, is not the same as a reading in Hebrew, in black and white*;
and p. 24: *almost every assumption of a variant by means of retroversion is a
matter of doubt*. See my review in jfS S xii (1967) 113-22.
240 T H E U S E O F E V ID E N C E F R O M T H E V E R S IO N S
phrases) really rest on versional evidence (at one end of the scale)
or (at the other end) merely quote the versions in connexion with
a reconstruction which is logically almost entirely conjectural.
Philological treatments use the versions in a rather different way.
They do not necessarily deny the approach which has just been
outlined. But their main interest lies in another direction.
In philological treatments the versions have been evaluated
primarily not as witnesses to a different text, but rather as witnesses
which, while they follow from the same text (with room for altera­
tions of vocalization as discussed above), provide confirmatory
evidence for, or clues towards, a different understanding of it.1
These understandings are taken to agree with the philological re­
interpretation of the same text on the basis of cognate languages.
As with textual treatments the value of the clues provided by
a version does not depend on the rightness of its translation. A
rendering in a version is quite frequently taken to be valuable
evidence of linguistic understanding even though at the point
where it is found it is a wrong interpretation. The fact that the
version could make this particular error is evaluated as evidence
that the meaning in question, even if erroneously applied in trans­
lating at this point, was nevertheless well known to the translators.
Thus, it is argued, it furnishes evidence for their linguistic under­
standing, and may be applied as confirmation at points where
philologists have seen clues from cognate languages pointing in
the same direction.
Thus the practice of philological treatments has often been to
use the LXX or other version not as a direct corrective of M T but
as evidence for the identification by comparative methods of
Hebrew words or senses previously unknown. For the general
approach we may quote Professor Winton Thomas:2
The Septuagint. . . frequently presupposes Hebrew words which are
1 It should be added that there is a considerable difference of opinion about
the value of the versions in relation to philological treatments. Driver has on the
whole valued them highly. An opposite position is taken by Albright who,
claiming that scores of passages in the Psalms have been cleared up on the basis
of Ugaritic, writes (Peake’s Commentary, p. 63): *In virtually every case the LXX
translators were just as ignorant of the true meaning as were the Massoretes.’
Cf. recently Dahood, Psalms, p. xxiv. Albright and Dahood, however, actually
use versional evidence, inconsistently, a good deal more than this would suggest.
See my further remarks below, p. 268. It is Driver’s approach that will mainly
concern us here, since it at least makes the versions worth studying.
2 In Record and Revelation, p. 396.
 GENERAL 241
only explicable by reference to one or other of the Semitic languages,
more particularly to Arabic. ‘Arabisms’ in the LXX have, of course,
been suspected before now, and indeed some have already been
recovered. Recently, however, the extent to which the Septuagint
translators could ‘Arabize’ has gradually become more apparent.
Some examples are next given, and the passage continues:
How are these ‘Arabizing’ renderings to be explained ? It is not to be
supposed that the LXX translators had any knowledge of Arabic, any
more than had Ben Sira’s grandson, who frequently ‘Arabizes’ when
translating his grandfather’s work. On the contrary, it can only be
assumed that these words originally belonged to the common stock of
the Semitic languages, and that they formed at one time part of the
Hebrew vocabulary, but that their meaning was generally lost, to be
retained only by the Egyptian Jews, and traceable today only through
the medium of Arabic. In the same way must be explained those render­
ings in the LXX which presuppose other Semitic languages.
A good first example is Ps. 47. 10 η » ‫» »’■ ־‬
literally apparently ‘the shields of the earth’. The LXX (46. 10) has
ol KparaLOL τ η ς γ η ς ‘the powerful of the earth’, cf. also Syriac
5whdnyh d Y y ‘the powers of the earth’.1 The normal Hebrew sense
‘shield’ has been felt to be strange: are there ‘shields of the earth’
which belong to God? Emendations have been suggested which
produce a sense like ‘princes’; in these the versions are used as
clues to construct a consonantal text different from MT. A philo‫״‬
logical treatment is offered by Driver, who says that the LXX here
provides ‘far the earliest evidence’ for the root of Arabic majin
‘bold’.2 Perhaps, then, there was a Hebrew ‘bold, insolent’,
preserved only through the versional evidence. KB, following
Driver, registers this as a Hebrew word.
The textual and the philological treatments both result in
roughly similar senses (‘princes’ or ‘insolent ones’, against the
traditional ‘shields’), but the mode by which this result is reached
is different. In the one case it is reached by altering the text, in
the other by offering a different explanation of the same text.
We do not need to decide now whether this philological explana­
tion is the right one. In addition to these possibilities, one can also
say that the meaning is ‘shields’ and that this is a figurative
1 Brockelmann, Lexicon Syriacum, p. 12a.
2 But what if the Arabic sense ‘bold’ is derived from the senses *to joke, jest,
jeer, make insolent jokes’, etc. ?
24* T H E U S E O F E V ID E N C E F R O M T H E V E R S IO N S
expression for the rulers or the mighty ones of the earth. This
explanation through metaphor, if correct, also removes the original
difficulty. It implies that the LXX were vague about the meaning
and gave a general guess; or that in the translation they abandoned
the metaphor and gave expression to that to which the figure
referred, rather than reproduce the figure itself in Greek.1
Job io. 22 has the phrase:
‫ ו ל א ס ד רי ם‬. . . ‫א ך ץ עי פ ת ה‬
Now the familiar Hebrew ‫ ס ד ר‬means ‘order’. It is not impossible
to take the sentence in this way and understand as ‘a land of dark-
ness and chaos’, but there are certain peculiarities. Firstly, a better
figure might be absence of light rather than absence of order;
secondly, the plural of ‫‘ ס ל ל‬order’ is unusual; thirdly, ‫ס ד ר‬
‘order’, though a very familiar late Hebrew word, is not found
elsewhere in the Bible. Now the LXX rendering is:
els γ η ν σ κό το υς αιω νίου, ου ούκ e a n v φ έ γ γ ο ς
‘into a land of eternal darkness, where there is no light’.
Using this as confirmation, Driver cites Arabic sadira ‘be dazzled
by glare’, and thus identifies a sense of ‘rays of light’ for the Hebrew
*.‫ר ״י ם‬τ ‫ד‬: ‫ס‬
More important, because affecting a more frequent and more
central word, is the case of ‫ שו ב‬. Driver has identified for certain
passages a verb which is quite distinct from the familiar ‫ ב‬1‫ש‬
‘return’. The word ‘return’ is cognate with Arabic taba (w), but the
new word is cognate with Arabic saba (y) ‘run, wander at random’,
used especially of beasts.
At Jer. 8. 6 M T has (K) ‫ ס ר צ ו ת ם‬3 ‫ כ לי ה שב‬. Since the refer-
ence seems to be to the roaming or straying of animals, and ‘return’
does not seem to give good sense, Duhm emended to ‫‘ שט‬go to
and fro’. But Driver, retaining the M T and explaining it as
‘wander at random’, cites the LXX:
hieXnrev 6 τ ρ έχ ω ν άττο το υ δρόμου α ύ το ΰ
1 Cf. further below, pp. 251 if.
2 He omits the words ‫ כ מו א פ ל צ ל מו ת‬as a gloss. To that extent the treat-
ment is textual; but presumably the gloss, if gloss it is, would have been an
explanation of the strange word ‫ ס ד רי ם‬and thus would be a textual gloss with
a philological reason. In a completely different philological treatment, Tur-Sinai,
Job, pp. 184 ff., takes ‫ ס ד רי ם‬as exorcistic ‘arrangements’ of words.
 GENERAL 243

and claims that the Peshitta with its mhlkyn ‘walking about’ pro-
vides further support.1
One of the forms which shows substantial homonymy in Hebrew
is ‫ ע נ ה‬. This is further increased if we follow those scholars who
have found yet another verb, cognate with Arabic ganiya ‘stay in
a place’. This has been identified at Mai. 2. 12 in the difficult
phrase ‫ ע ל‬, which is thus taken to mean ‘gadabout and
stay-at-home’; and also at Isa. 13. 22:
‫תיו‬T ‫מ נו‬: ‫ל‬: ‫א‬- ‫ב‬: ‫איי ם‬
· ‫״‬
‫ענ ה‬ ‫ו‬
T T :

where the sense would be :

‘and jackals will dwell in its palaces’.
At this latter place the LXX renders in just this way:
K a l o v o K e v r a v p o L € K € l K a r o L K r jc r o v a iv ,

So it can be argued that the LXX were familiar with the verb in
the sense already known for the Arabic cognate. At Mai. 2. 12,
however, this is not so; the rendering Ta7T€1vcodfj depends on the
more familiar verb ‫‘ ע נ ה‬be humbled’. Thus the claim for recogni-
tion by LXX is modified by non-recognition on their part at
another place.2
The procedure outlined above is not in any way confined to the
LXX itself. It can be used in relation to the other ancient Greek
versions, as preserved in the fragments of Origen’s Hexapla and
elsewhere, and also to versions in other languages, such as the
Aramaic Targum.
Take the attractive case of the word ‫ ע ש‬, which is normally
understood to mean ‘moth’. At Hos. 5. 12, however, it occurs in
1 One may doubt, however, whether the Syriac word when taken in context
means this. The whole phrase is klhwn mhlkyn hsbynhwn ‘all of them walk in their
own good pleasure’, very close to the Targumic ‫; כו ל הון מ סגן ב ר עו ת נ פ ש הון‬
cf. also the Greek rendering of 6 Σύρος. Probably □‫ ב מ ר צו ת‬was diagnosed as
a form from ‫ ל צ ה‬, and this fact then led to the rendering of the verb as ‘walking’;
this produces the commonplace ‘walking in one’s own will or pleasure’. The
Vulgate with its omnes conuersi sunt ad cursum suum agrees with the more
traditional understanding.
2 Cf. further below, p. 250. Non-recognition in Mai. 2. 12 is itself no insuper-
able objection. It is perfectly possible that the Isaiah translator knew a word
which was unknown to the translator of the Dodecapropheton. In so far, how-
ever, as there is any force in the argument that a word is recognized by several
translators, the converse must also be acknowledged, that it is proper to notice
when recognition by more than one translator does not take place.
244 T H E U S E O F E V ID E N C E F R O M T H E V E R S IO N S
parallelism with ‫‘ ל ק ב‬rottenness, pus’. Symmachus has ζνρώ ς,
while Aquila has β ρ ω σ τή ρ . It is then possible to follow Driver and
hold that there are more words ‫ ע ש‬than one. One of them, meaning
‘moth’, corresponds to Arabic 'utt; another, meaning ‘pus, rotten‫־‬
ness’, corresponds to Arabic gatt. Driver also identifies a third
word ‫ ע ש‬meaning ‘bird’s nest’, but for the sake of clarity we leave
this aside for the present. We note simply that the evidence comes
not from the LXX but from the later versions.1
Symmachus is not necessarily right in his rendering; he writes
βύρώς ‘rot’ at places where the meaning is certainly ‘moth’, such
as Isa. 50. 9, Ps. 39. 12. But the erroneous rendering may still
be good evidence. At Ps. 31. 11 we have similar renderings of the
verb ‫( ע ש ש‬Symmachus βνρω τίασαν). At Job 27. 18, where the
M T has ‫ ע ש‬, the Targum has 2.‫ל ק ב ו ב י ת א‬
Thus, though the LXX will be the most frequently quoted,
other versions may supply evidence equally well.
Erroneous renderings, as we have seen, may be taken as good
evidence for the translators’ knowledge of Hebrew; and under
these circumstances, not surprisingly, arguments of some consider-
able subtlety can be found. Consider the following example.
Tob it 1. 13 reads:
κ α ί eScoKev [/xoc S] 6 ν φ ισ το ς χά ρ ιν κα ί μ ο ρ φ ή ν ένώττιον jEW/zea‫־‬
σαρου.

Now Perles believed that there was a Hebrew word ‫ ש כ ל‬meaning

‘form’, cognate with Arabic sakl ‘form’. He thought that the
original Hebrew of Tobit here was 1‫] ש כ ל‬Π. In Tobit, however,
in his opinion the word ‫ ש כ ל‬actually meant ‘sense, understanding’,
which is the normal meaning of this word in Hebrew. The transla-
tion thus rendered the (putative) Hebrew wrongly. Nevertheless
Perles took the Greek text to be a reliable witness to the existence
of a Hebrew ‫ ש כ ל‬meaning ‘form’. The other place where he found
evidence is in I Sam. 25. 3, where we hear of the attractive Abigail:
‫ ל וי פ ת ת א ר‬5 ‫ו ה א ש ה טו ב ת ~ ש‬
Perles took this to mean:
‘fair of form and beautiful of appearance’.
1 For further discussion of this example see below, pp. 252 n., 279.
2 Since this word is found specifically of the decay caused by moths, however
(Jastrow, p. 1496a), it is by no means clear evidence against an understanding of
the original as ‘moth’.
 GENERAL 245

Here, however, the LXX said α γα θ ή avveaet, ‘good of under-

standing’.
By this argument, then, the evidence for the existence of a ‫ש כ ל‬
‘form’ is a passage where the original Hebrew does not exist and
where, according to the theory itself, the Greek translated wrongly;
while, in the one place where the word (again according to the
theory) actually occurred in the Bible, the Greek translated wrongly
again. The argument is not an impossible one; but readers will
grant that it is an exceedingly delicate one. This, however, is an
extreme case.
With this, then, as a preliminary statement of the way in which
the versions are used in philological treatments, we may now go on
to discuss the problems involved in such use and to consider the
criteria by which we may judge whether such use has been valid.

(2) The Question of the Hebrew Text

Firstly, we have to consider the possibility that the LXX may
have been translated from a Hebrew text differing from that of the
MT. This is, after all, the hypothesis upon which enormous labours
of classical textual criticism have been performed. It is clear that
this hypothesis, if overstressed or used uncritically, can be most
unfortunate in its results. Where the Greek text gives a sense
different from the Hebrew, the hypothesis that it was translated
from a different Hebrew text is only one of a number of possibili­
ties. It may have had the same text, but misread it; or been careless
in handling it, or guessed at the sense, or paraphrased, or assimi­
lated it to another passage; or, indeed, it may have sincerely trans­
lated the same text in a way which we judge to be ‘wrong’ and which
thus gives us the impression that the text was different. There are
many possible relations between what the translators wrote down
in Greek and what was on the page in Hebrew. Only when we
eliminate a number of these possible relations are we entitled to
translate back the Greek into a Hebrew text and say that the
translators ‘read’ this text. It makes for a more sophisticated ap­
proach if we consider the quite different possibility that the Greek
may represent a different linguistic assessment of a Hebrew text
close to M T or identical with it.
Nevertheless it is unlikely that any degree of such sophistication
can eliminate the possibility that there is really a differing Hebrew
246 T H E U S E O F E V ID E N C E F R O M T H E V E R S IO N S
text behind the LXX translation. Scholars who have stressed the
value of the LXX for discovering new meanings for the Hebrew
words of M T have not really denied this. Just as, in spite of their
warnings against emendation, they have frequently emended, so
also, in spite of their use of the LXX as a clue to a different linguis-
tic assessment of the same text, they have like other scholars
frequently continued to use the LXX as evidence of a varying text,
In particular the discovery at Qumran of new Hebrew texts
which agree with the LXX at points where it materially differs
from the MT, seems, if this evidence has been rightly assessed, to
prove conclusively that Hebrew texts differing from M T were
sometimes used for the Greek translation.1
What one has to inquire therefore is whether, at the points where
philological treatments have been offered, the question of a differing
text has been properly considered as a possible explanation for the
LXX rendering. Sometimes one wonders whether the attractive
prospect of a philological treatment does not lead scholars into an
over-confidence towards those texts which seem to support it—an
over-confidence which they do not consistently show towards the
interpretations of the Greek version.
For example, it has at times been argued that there is a verb
‫ ע ש ה‬which is not the familiar ‫‘ ע ש ה‬do’ but another verb ‫ע ש ה‬
cognate with Arabic gasiya and meaning ‘turn oneself, incline,
towards or away’. This can be well exemplified at I Kings 20. 40:
‫ נו‬3‫ולהי ע ס י ף ע ש ה הנ ה ו הנ ה ו הו א אי‬
‘Your servant turned in one direction and in another, but he was gone.*
The traditional sense ‘do’ is hard here, as was remarked by BDB,
p. 244b. Now it has sometimes been said that the argument was
clinched by the LXX of I Sam. 14. 32:
mt (Kethibh) ‫ר ע ש ה ע ם א ל ־ ש ל ל‬
LXX κ α ί €κλίθη 6 λαός els τ α σ κ ΰλ α
which appears to coincide exactly with the interpretation of the
Hebrew as ‘turn oneself, incline,> as the philological treatment
maintained.
Here, however, there is a rather peculiar Qere 1‫ י ע ט‬. This word
1 On such Qumran texts see, for instance, Cross, Ancient Library of Qumran,
e.g. PP· 13345‫·־‬
 T H E Q U E S T IO N O F T H E H E B R E W T E X T 247

occurs elsewhere only in I Sam. 15. 19, where LXX has ώρμησας
‘rushed’ (upon the spoil), and 25. 14, where we have:
m t ‫וי ע ט ב ה ם‬
V τ - ‫—ך‬

LXX /cat ίξόκλινζν απ αυτών.

Thus, whatever the meaning of the strange word ‫ לי ע ט‬, and
whatever the origin of its occurrence as a textual variant at I Sam.
14. 32, the translation at I Sam. 25. 14 makes it extremely probable
that the LXX is a translation not of the present Kethibh at 14. 32
but of the present Qere ‫ לי ע ט‬. Thus the Greek translation, con-
trary to first appearance, does not show any sign of knowing a
Hebrew ‫‘ ע ש ה‬turn oneself, incline’, although there is a reasonable
case for the identification of this word on other grounds.
Driver argued that there is a word ‫ ע ל‬meaning ‘time’. Jer. 11. 12
has ‫‘ ב ע ת ר ע ת ם‬in the time of their trouble’, and, a little later,
in a continuation of the same passage, 11. 14 has ‫ ב ע ל ר ע ת ם‬.
Here, however, some Hebrew manuscripts read ‫ ב ע ת‬as in 11. 12,
and editors commonly follow this easier reading. But, Driver points
out, the LXX writes iv καιρω in both passages; the Vulgate has in
tempore. Evidence can be found also at Isa. 30. 8, 64. 8, Ezek.
22.4 ,30. in this last, m t ‫ ב ע ל ה א ר ץ‬, LXX has iv καιρω της
γης, a rendering which means ‘time’ though there is no doubt that
Ezekiel means ‘on behalf of’.
There is reason to suppose, however, that some of the evidence
is the product of assimilation and other textual processes. The LXX
iv καιρω at Jer. 11. 14 is probably a translation from the assimilated
reading ‫ ב ע ת‬, or else an independent result of the same assimila-
tion. At Isa. 30. 8 it is doubtful whether ‫ ע ל‬is rendered by any
form of καιρός, depending on textual variations in the Greek. Most
of the versional evidence, in fact, can be accounted for through
assimilation to other nearby phrases, along with the influence of
other words like ‫( מ ל ע ל‬frequently translated by καιρός) and Ara-
maic ‫·עוז‬ Since none of the Hebrew contexts intrinsically favour
the sense ‘time’, the suggestion therefore should probably not be
accepted.
(3) The Question of the Versional Text
Secondly, we have to bear in mind the history of transmission
of the Greek (or other) versional text and its possibilities of corrup-
tion, which may be quite independent of the Hebrew. Here again
24S T H E U S E O F E V ID E N C E F R O M T H E V E R S IO N S
there is sometimes a danger that we may overvalue a particular
form of the Greek text and set upon it a confidence which lies less
in its own intrinsic merit than in the attractiveness of the philo-
logical treatment which it appears to confirm.
An important instance is ‫ ג ב ו ל‬, normally taken to mean
‘frontier’, at Ps. 78. 54. Arabic has the familiar jabal ‘mountain’,
and gbl ‘mountain’ has been identified in Ugaritic. Now in Ps.
78. 54 ‫ ג ב ו ל‬is translated by ορος ‘mountain’:
m t : ‫ ז ה ק נ ת ה י מי נ ו‬- ‫ ג ב ו ל ק ד שו ה ר‬- ‫ו י בי א ם א ל‬
LXX (77· 54) Ka'L eia^y ay ev αντο υ ς 61ς ορος α γ ιά σ μ α το ς α υ τό ν,
ορος τούτο, ο έ κ τη σ α το η 8e£1a αύτοϋ.

This, not surprisingly, has been taken as evidence that ‫ג ב ו ל‬

was known to mean ‘mountain’ by the LXX. But the Greek texts
may be in error. The trouble may be caused by two facts: (a) the
great similarity of the two Greek words, ορος ‘mountain’ and ορος
(or opiov) ‘frontier’; (2) the fact that ορος ‘mountain’ occurs
(translating ‫ ) ה ל‬in the second clause, a circumstance which could
easily lead to an error in transcription. Grabe therefore suggested
the emendation opiov, which is accepted by Rahlfs and printed as
his text. A corruption would be even more understandable if the
LXX had written ορος in the first place. Though they did not
usually use ορος for ‘frontier’, they may have done so here precisely
because of the juxtaposition with ορος ‘mountain’ in the same
verse. All other cases of ‫ ג ב ו ל‬in the Psalms are rendered by
opiov.1
Thus the Greek evidence for a Hebrew ‫‘ ג ב ו ל‬mountain’ dis-
solves into an accident of the Greek text, caused by the remarkable
similarity of the Greek words, along with the parallelism with
‘mountain’ in the text translated. We may add the absence of traces
in other early versions or of survivals in later Hebrew.
Even more important than the history of the text, however, is
the technique of translation used by the version, which may vary
substantially from book to book and even between one part of a
book and another. All claims that a version either presupposes a
different text or shows a different linguistic understanding of the
same text have to be tested by relating them to the general methods
of the version in question. The sections which follow will illustrate

1 Cf. also I Sam. 10. 2, and Wutz, Onomastica Sacrat p. 546.

 T H E Q U E S T IO N O F T H E V E R S IO N A L T E X T 249
some of the methods of translation which are important for this
question.

(4) Imprecise Methods of Translation

At Ps. 84. 7:
mt : ‫ ? ל כ ר ת י ע ט ה מו ל ה‬- ‫ג ם‬
LXX /cat yap evXoyiag Scocret o vo/ioOertov
Rabin has suggested the presence of a Hebrew ‫‘ ע ט ה‬give’, cognate
with the familiar Arabic ’a ti. But in this Psalm the translator, who
is normally a fairly sober and literal worker, has by the end of v. 6
(unlike the earlier part of the Psalm) got himself thoroughly lost.
He missed ]‫ מ ע י‬, which might have told him that the context
was something about water; he thus fails to see that ‫ ב ר כ ו ת‬are
‘pools’, and takes them as the commoner homograph ‘blessings’.
With ‫ מ ו ל ה‬the sense ‘rain’ does not occur to him, though it is not
uncommon in contexts clearly watery, as at Joel 2. 23 and (with
‫ ) י ו ר ה‬at Deut. 11.14, Jer. 5. 24. He takes it as the much commoner
word ‘teacher, lawgiver’. Thus it turns out that the sentence is
about a lawgiver who does something to blessings.
The verb is not a common one, and the translator varies his
rendering according to the context. If there is a reference to cloth-
ing, he usually uses 7r6p1/3aAAecr#a6 or a similar word (LXX,
1 0 8 . 1 9 , 2 9 ), and so also, by a familiar metaphor, where the subject
is shame (LXX, 7 0 . 13 ). At 89 (8 8 ). 4 6 , again of shame, he uses the
metaphor not of clothing but of pouring upon. Given this degree
of variability, he may well have written 866061‫ ־‬in our passage as a
general rendering, which would give a passable sense, whatever
the exact original metaphor might be.1 The Gallican Psalter fol-
lowed the LXX with etenim benedictionem dahit legislator, but the
word ‘give’ disappears as soon as Jerome consults the Hebrew and
writes benedictione quoque amicietur doctor. The Syriac, with its
w*p bwrkf yt'tp s'm nmws\ ‘and he also who gives the law will be
wrapped in blessing’, agrees with the LXX in the interpretation of
‫ מ ו ר ה‬but not in the interpretation of the verb. This sense of ‫מ ו ר ה‬
was, of course, prominent at Qumran and elsewhere in late
Hebrew.
1 On this mode of translation in the Psalms see Flashar in Z A W xxxii (1912),
passim.
250 T H E U S E O F E V ID E N C E F R O M T H E V E R S IO N S
It is quite improbable, therefore, that the passage is evidence of
knowledge of a Hebrew word meaning ‘give’.
Similar doubts are raised by the claim1 that κ α τ ο ι κ ε ΐ ν at Isa.
13. 22 is evidence of a word ‫‘ ע נ ה‬stay in a place’. This would be
plausible if the Isaiah translation were a precise one, carefully
rendering each word with a close equivalent. We can test this in
two ways: firstly, by looking at the way in which κ α τ ο ι κ ζ ΐ ν is used
in the Greek Isaiah; secondly, by looking at the verses just before
and after 13. 22 in order to see how closely the translator is follow-
ing his text at that point.
The verb κάτοικον is used in Isaiah from time to time where it
is not a close equivalent. At 9. 1 (8. 23) it is an expansion, and there
is no Hebrew verb ‘dwell’. 14. 23 is much the same. At 16. 7 it is
used to translate the obscure word 2.‫ א שי שי ם‬At 27. 10 it is hard
to see what it is translating. At 62. 5 it is a general (rather than a
close) rendering of ‫‘ ב ע ל‬live with’, where the original reference is
to marriage. Clearly, the use of κ α τ ο ι κ ε ΐ ν in Isaiah does not
necessarily suggest an exact equivalence to any Hebrew word.
Similarly, in the immediate context of Isa. 13. 22 a number of
striking divergences (e.g. ovSe μ η β ΐ σ ί λ θ ω σ ι ν els α υ τ ή ν for M T
1‫ ל א ת ש כן‬, and ο υ δ έ μ η δ ί έ λ θ ω σ ι ν α υ τ ή ν for M T ‫)]* ל א׳ ״ ל ה ל‬
will show how far the translator is from attempting any precise
rendering of the Hebrew text word for word.
As commonly in Isaiah, the translation includes wide divagations
from an exact rendering. This is so even in words which were
undoubtedly entirely familiar to the translator, such as ‫ ש כן‬. This
verb, which ironically does mean κ α τ ο ι , κ ε ΐ ν , is twice given other
and very imprecise renderings in the space of two verses. Words
and phrases are missed out altogether, and there is a tendency to
re-use a word which has just been used before, or one close to it (so
elaeXdeLv, SteXdeLV, and ά ν α π α υ β σ θ α ι ) . It is unlikely that the use
of κ α τ ο ί κ β ΐ ν here for Hebrew ‫ ע נ ה‬constitutes evidence of exact
linguistic knowledge.
Sometimes a Greek text will agree with a possible philological
treatment, but will leave us uncertain whether it does so by acci-
dent or on a basis of real knowledge. At I Sam. 1. 18:
m t ‫ופך?! ל ^ ־ ה י ו ־ ל ה ע ו ד‬
LXX κ α ί τ δ π ρ ό σ ω π ο ν α υ τ ή ς ο υ ovverreaev erri
1 See above, p. 243. 2 On this word cf. also below, p. 254.
 I M P R E C IS E M E T H O D S O F T R A N S L A T IO N 251
we may adduce the sense ‘fall’, known for ‫ הו א‬at Job 37. 6.
Against this, however, we may point out:
(a) Other versions do not support the LXX.
(b) The Greek locution may be an assimilation to a familiar
passage like Gen. 4. 6:
lva τ ί συνβπβσζν το πρόσωπόν σου;
(^) Even if ‘fair is a good gloss for Job 37. 6, it remains uncertain
whether the kind of ‘falling’ used in the phrase ‘his face fell’ could
be expressed with this verb. We should not be over-persuaded by
the fact that we use the same English gloss ‘fall’. Senses known in
Hebrew apply only to snow, and (in the noun ΓΠΠ, ‫ ) ל ו ה‬to
‘disaster, calamity’.
(1d) It is one thing to find a ‫‘ הו א‬fall’ in the peculiar vocabulary
of Job, another to find it in a prose section of Samuel, in which
‫ ה י ה‬in the common sense ‘be’ is very much more frequent than
in the poetic sections of Job.
The probability of a sense ‘fall’ for ‫ היו‬at I Sam. 1. 18 is thus
somewhat below 50 per cent.

(5) The Use of Favourite Words

The second point requiring attention in the use of the versions
is the fact that translators often have certain ‘favourite’ words. It
is a stylistic feature of the Greek Psalter to use one word freely for
a considerable number of different Hebrew words. Such an in-
stance is the group κραταίωμα or κραταίωσή and the adjective
κραταιός.
The word ‫ סו ל‬usually means ‘secret; private council’. The
translation by κραταίωμα at Ps. 25 (24). 14 has been taken to point
to a Hebrew word cognate with the Arabic siid ‘chieftaincy’. Again,
we have seen how the ‫ מגני א ל ץ‬of Ps. 47 (46). 10 is translated by
οι κραταιοϊ τής γής> and this has been taken as evidence related
to Arabic words so as to give a sense ‘bold, insolent’.1
But κραταίωμα (κραταίωσις) is clearly a word freely used in
the Greek Bible. We find it rendering 0‫‘ ל ע‬rock’ and ‫מ ע ו ז‬
‘refuge’2 as well as ‫‘ ע ז‬strength’ and ‫‘ מ שג ב‬refuge’. The adjective
1 See above, pp. 241 f.
2 This rendering is an etymologizing one (the word is taken as from the root
of ‫* עז‬strength’).
252 T H E U S E O F E V ID E N C E F R O M T H E V E R S IO N S
κραταιός is used not only for obvious words like ‫ ע ז‬and ‫ ע צו ם‬but
also for ‫ ע ר י ץ‬and ‫ א ד י ר‬.
An even more marked example in the Psalms is ταράσσω ‘dis-
turb’. This word is employed in numerous contexts, where a
variety of meanings is found in Hebrew. We find it translating
such various words as ,‫ ע ש ש‬, 1 ‫רגז‬, ‫ ס ח ר‬, ‫ ש ח ח‬, ‫ ח מ ר‬, ‫ר ע ש‬
‫ ל ה ט‬, ‫ נ ד ל‬, ‫ ש ל ל‬, ‫ פ ע ם‬, ‫ צ מ ת‬, ‫ ח ל ל‬, etc. Such a Greek translation
can hardly be used as evidence for a philological reinterpretation
of a Hebrew word. It seems rather that the idea of disturbance or
non-disturbance appeals to the translator, and is used by him
without concern for the way in which it obscures the difference
between the Hebrew meanings in the verses translated.
Evidence for a Hebrew ‫ מ י ד‬, cognate with Arabic mada (y ) ‘be
violently agitated’, has been seen in Hab. 3. 6:
m t ‫י מ ך ד א ר ץ‬1 ‫ע מ ד‬
LXX έσττ], καί εσαλβνθη ή γη.
This solution avoids the emendation ‫! י מ ע ד‬.
But σ α λζνεσ θα ι is a favourite LXX word; Hatch and Redpath
register no fewer than twenty-three different Hebrew terms ren-
dered with it. In the Dodecapropheton it is used also for ,‫בלע‬
‫ מוג‬, ‫ מ ס ס‬, and ‫ ר ע ל‬, and it occurs in fact only a few verses
earlier, at Hab. 2. 16 ( σαΧ ςύθητι or Β ίασαλβνθητι, M T ‫ ח ע ר ל‬,
presumably reading as ‫) ה ר ע ל‬. The rendering is only an assimila-
tion to a probable sense of the context and to the thought of passages
like Ps. 18. 8,46. 4, 68. 9, Nah. 1. 5; it is no sign of special linguistic
knowledge.
To put it in another way, the ‘shaking’ of the earth can be treated
as a stereotype of prophetic thought. The same is true of the idea of
‘making firm’ the creation, as at Isa. 51.6:
mt ‫כ י ־ ש מ י ם כ ע שן ג מ ל ה ו‬
τ ν V ‫״‬
L a a οτι ο* ουρανος
‫י‬ \ *
ως καττνος ‫י ו‬ //‫ר‬
6στ€ρ€ωαη.

The Greek verb is not a clue to good linguistic knowledge of the

sense of ‫ ; מ ל ח‬it is a stereotype of creation, found of ‘heaven’ at
Ps. 33 (32). 6 (‫) ע ש ה‬, Isa. 45.12 ( ‫)נ ט ה‬, 48.13 ( ‫) ט פ ח‬, and of ‘earth’
at Isa. 42. 5, 44. 24 (‫) ר ק ע‬. The Targum, which construes Isa.
1 This is of interest for the equivalence ώς ταραχή = ‫ כ ע ש‬at Hos. 5. 12; cf.
above, pp. 144, 243 f., and below, p. 279. Cf. also Index, no. 130.
 T H E U S E O F F A V O U R IT E W O R D S 253
51. 6 of the passing away of the heaven like smoke, is following a
different stereotype.
Dahood, arguing that the verb formation with infixed /-t-/ re-
mained alive through the biblical period, cites the translation of
VIS *Vllpil by άπέστρβψβν τό ττρόσωττον αντοϋ and auertit faciern
suam.1 The form was thus taken by the translators to be from
‫‘ סו ר‬turn away’.
Even a brief study of the contexts shows, however, that this must
have been a stylistic preference in Greek, and has nothing to do
with the original Hebrew verb formation. Ά ποστρίφζιν is a favourite
LXX verb. The translation of ‫ ה ס תי ר‬with it occurs almost
only where the object is ‘face’. (If we exclude cases where there is
no LXX rendering, as sometimes in Job, there are forty-five cases
where the object of 0‫ ת ר‬is not ‘face’, and forty-three of these are
rendered with ‘hide, shelter’, etc., i.e. κρύπτω , αποκρύπτω , or
σκβπάζω; the only partial exceptions are Isa. 57. 17, Ps. 89. 47.)
The Vulgate, cited by Dahood, is not independent evidence, for
this is the Galilean Psalter, translated from the LXX and there-
fore naturally agreeing with it. Where Jerome translated from the
Hebrew, i.e. outside the Psalter, the rendering is almost always
‘hide’, and in the Psalter iuxta Hebraeos Jerome with complete
regularity corrected from auerto with ‘face’ as object to abscondo.
The ‘hiding’ of the face is a quite genuine and an important as-
pect of Hebrew religious idiom. Dahood’s interpretation destroys
this, and his citation of the LXX is not evidence in favour of his
interpretation, but only shows that they too, through some
stylistic or theological preference, had lost, or were unable to
reproduce, this insight.

(6) Etymologizing
Etymologizing is the procedure of interpreting a word by refer­
ence to the meaning of another (usually a better-known) Hebrew
word which had a similarity to it and could, in more modern
terminology, be taken as its root; or a word may be analysed into
separate units from which it is taken to have been made up. More
extreme still, a translator may strive, as Aquila sometimes did, to
represent words of common root in Hebrew with words of common
root in Greek. Though the etymologizing tendency is stronger in
1 Dahood, Psalms, i. 64.
254 T H E U S E O F E V ID E N C E F R O M T H E V E R S IO N S
the later Greek translators, it appears sporadically also in the
LXX.
Isa. 16. 7 has a rare word ‫ ; א שי שי ם‬Driver has argued that it
means ‘luxurious dwellers’, after Arabic ‫י‬atta ‘be luxuriant, luxu-
rious; live comfortably’. Whether this is so or not, it is not sup-
ported by the tols ev^paivopuzvoLs of Symmachus; this is an
etymologizing guess from the similarity to the familiar ‫ שיש‬he-
joice’, so translated by him at Jer. 32 (39). 41, cf. L X X atIsa.61.10.
Etymologizing is also the probable explanation for the rendering
of ‫ ל א מ י ם‬by apxovres at Gen. 27. 29, Isa. 34. 1, 41. 1, 43. 4, 9,
and by /SacrtAefs’ at Isa. 51. 4. At first sight, the frequency makes
this identification very impressive. Driver writes: ‘One cannot
deny this sense, so frequently attested’; but he holds that this sense
is never right at the places where the LXX find it; it is actually
found at Ps. 148. 11, 7. 8. The former reads:
: ‫מ ל כ י ־ א ר ץ ו כ ל ־ ל א מ י ם ש רי ם ו כ ל ־ ש פ ט י א ר ץ‬

and would then mean:

‘The kings of the earth and all rulers;
ministers, and all the judges of the earth.’
The familiar sense ‘peoples’ seems at first sight to fit less well into
the sequence. Ps. 7. 8 is a less clear case; an ‘assembly of nations’ is
more likely to surround one than an ‘assembly of rulers’, and in
7. 9 the Lord judges ‘nations’ (1.(‫ע מ י ם‬
In later Hebrew ‫ ל א ם‬was little used, and ‫ א מ ה‬was becoming
more frequent. It is unlikely that the sense ‘nation’ came to be
unknown, though it is possible that, when even this sense became
infrequent, very occasional other senses (viz. that of ‘ruler’ in this
case) came to be forgotten. Even in the biblical period ‫ ל א ם‬was a
somewhat traditional word. It was mainly used in the plural, was
almost entirely poetical, and in parallelisms is almost always a ‘B-
word’, i.e. a word taking second position in a parallel pair.2
That the LXX rendering is the result of etymologization
can be seen from the ] ‫ ל א שי א ל מי‬which is a Targum of the tribal
name ‫ ל א מ י ם‬at Gen. 25. 3.3 The word was analysed as ‫ ל‬+ ‫ א מ ה‬.

1 Cf. above, pp. 133, 172.

2 See Boling in J S S v (i960) 221-55, and especially p. 233. In the Psalms
Boling found only one exception out of nine occurrences.
3 See Gen. R. ad loc. The same is the reading of the Neofiti text.
 E T Y M O L O G IZ I N G 255
Jerome knew the same interpretation: Laomim uero φυλάρχους,
id est principes multarum tribuum atque populorum.1
The rendering in LXX and Targum is probably affected also by
stock parallelisms. All instances have ‫ ל א מ י ם‬as the B-word; the
A‫־‬word is usually ‘nations’ or the like, once ‘man’ (Isa. 43. 4).
The Targum has ]‫‘ מ ל כ ו‬kingdoms’ at Gen. 27. 29, and similarly
at Isa. 34. 1, 41. 1, 43. 4, 43. 9; the tendency to produce a paral-
lelism of nations/kingdoms can be seen at 43. 4, where ‫ א ד ם‬is
rendered by ‫ ע מ מ י א‬. The rendering of ‫ ל א ם‬as ‘ruler’ never occurs
in an independent sentence where the point would be substantially
different if the sense were ‘ruler’ rather than ‘people’.
Further arguments against this identification of ‘ruler’ are:
(a) This development is surely peculiar to Accadian, and even
there the sense is not ‘ruler’ generally, but refers to the eponym of
the year. The institutions are entirely Mesopotamian. In the
Ugaritic phrase ybmt Vimm, the second word is usually taken to be
‘peoples’, rather than ‘rulers’ as Driver would have it.
(b) Even at Ps. 148. 11 in the Hebrew itself it is questionable
whether the sense ‘rulers’ fits better. It would certainly do so if it
were clear that lists of four items in a context of this type must
always be homogeneous in order to give good sense. But the list
may be one in which the liking for homogeneity is interfered with
by the liking for supplementary parallelism, producing the very
common parallelism of ‘kings’ and ‘peoples’.2 The verse then
returns to continue the list of types of rulers. The pattern can be
described as A-B-A-A.3 This sense is not in any way a poor one.
Since the sense ‘peoples’ for ‫ ל א מ י ם‬was a very familiar one, there
was no strain in passing from ‘kings’ to it and then back to other
words for rulers. These considerations reduce the degree of
difficulty in the accepted sense ‘peoples’, and thus reduce the
original basis for the philological treatment.

(7) Free Rewriting

The fourth procedure on the part of the translators is that of
very general paraphrasing. The rendering is a free rewriting of the
passage, producing a sentiment which is really the translator’s
1 Hebr. Quaest. in Gen., ad 25. 3.
2 Cf. Ps. 72. II (with 135 ;(‫גולם‬. IO (with ‫ גולם‬first).
3 Cf. Exod. 4. 11: dumb/deaf/clear-sighted/blind = a-a- b -a ; Isa. 2. 12 (see
below, pp. 280 f.): high/lofty/uplifted/low = a‫־‬a‫־‬a‫־‬b .
256 T H E U SE OF E V ID E N C E FROM T H E V E R S IO N S
idea, connected here and there with the words of the original.
This applies particularly to books like Proverbs and Job. These
are books which contain many textual-philological difficulties and
which for this reason have produced large numbers of philological
treatments. They are also books in which the translators felt free
to handle the text very loosely.1 Painfully exact renderings may be
succeeded in the next verse by examples of extreme freedom and
complete rewriting.
Driver, arguing the existence of a Hebrew ‫ ז ר‬meaning ‘enemy’,
says that the use of exdpos to render ‫ ז ר‬in one passage where it is
clearly wrong (Prov. 6. 1) proves its currency. The texts are:
mt ‫מ י א ם ־ ע ר ל ת ל ר ע ף תקן^ת ל ז ר כ ? ך ף‬
t ‫־‬w
f ‫יי ׳‬ ‫י‬ / \ / /\
LXA VL€y eav eyyvrjdrj gov (piAov,
7rapaSa)G€Ls orjv xefpa ixdpcp.
Now the correspondence of one word to another is not decisive
here; for the translator, while he has related himself to the text
before him, has made quite a different thought out of it. In the
original the whole verse is a conditional clause, stated in two
parallel forms; the sequel does not come till the next verse. The
sense is:
‘My son, if you have become surety for your neighbour,
or shaken hands in pledge for a stranger . . .’
The Greek rendering is:
‘My son, if you pledge your friend,
you will give over your hand to an enemy.’
Here the thought is already complete within the verse. Since he
has seen the two parts of the sentence as a contrast, the translator
has made it natural to see the ‫ ז ר‬of the second part as part of the
total contrast in terms of friendship and enmity. The interpreta-
tion as ‘enemy’ may be no more than part of this understanding.
It is doubtful whether it suggests any intrinsic linguistic knowledge
that ‫ ז ר‬could mean ‘enemy’, and so it does not provide strong
evidence for the identification of a ‫ ז ר‬meaning ‘enemy’ elsewhere.
To say this is not to deny this identification, for which there is
quite a good case on other grounds. But it does mean that we
1 On these books see G. Gerleman, Studies in the Septuagint (Lund, 1946
and 1956).
 FREE R E W R I T I N G 257
cannot accept the simple equivalences of Hebrew and Greek
words, as registered for instance in Hatch and Redpath, as
evidence for new linguistic identifications without considering the
way in which the whole context is handled by the translator.
Moreover, {a) it was possible for the translator just to take ‘strange’
as equivalent to ‘enemy’ or ‘evil’ on quite non-linguistic grounds,
as he does with the ‘strange woman’ of Prov. 5. 3 (Gk. φ α ύλη
ywanci, πόρνη ); (b) he may have been influenced by ‫‘ צ ר‬enemy’.
The wide separation between the ideas of the Hebrew text and
those of the Greek rendering can be seen at Prov. 17. 14:
m t : ‫ ג ל ע ה רי ב גטו ש‬1‫פ ו ט ר מי ם ר א שי ת מ ד ו ן ו ל פ נ י ךזן‬
‘The start of strife is one who lets out water;
so let go before a dispute breaks out.’
LXX εξουσίαν δίδω σιν λόγους ά ρ χη δικαιοσύνης
π ρ ο η γ ε ίτ α ι δε τ η ς ένδειας σ τά σ ις κ α ι μ ά χ η .

‘The beginning of righteousness gives authority to words;

but quarrelsomeness and fighting lead to poverty.’

The rendering σ τά σ ις κ α ι μ ά χ η does not help to identify a

Hebrew ‫ ש‬1‫‘ נ ט‬clash of battle’. The translator has produced an
admirable Hellenistic-Jewish sentiment of his own, which has
occasional contacts with elements in the Hebrew, e.g. δικαιοσύνη
for ]‫ מ ל ל‬, taken as from ]‫‘ ל י‬justice’, and ‫ ה ת ג ל ע‬, construed in
the sense of ‘litigation’ (Jastrow, p. 250b).
If such a word ‫ נ טו ש‬is to be discovered, a better basis lies in the
‫ ו ת ט ש ה מ ל ח מ ה‬of I Sam. 4. 2; but here no support can be found
in the Greek.
In books where the translation is of free or paraphrastic type,
such as Proverbs, quite a large proportion of the evidence is
affected by this argument.
A few more examples may be of interest. At Prov. 23. 31:
m t : ‫ י י ן כי י ת א ד ם‬. ‫א ל ־ ת ך א‬
LXX μ η μ εθ ύ σ κ εσ θ ε οϊνω

the Greek is not really evidence in favour of the identification of a

verb ‫‘ ר ו ה — ל א ה‬drink’, whatever evidence there may be for this
from other sources; and Driver is over-literal in saying that ‘there
258 T H E U S E O F E V ID E N C E F R O M T H E V E R S IO N S
is no harm in looking on wine, whatever its colour be’.1 The
Greek version removed the subtlety of the Hebrew idiom, cutting
out the indirectness of ‘looking at’ wine and going straight for the
absolute moral issue of drunkenness. It is no evidence of linguistic
knowledge.
A substantially stronger case, in my opinion, can be made for
the identification of a verb ‫‘ ח ק ל‬despise’ at Prov. 28. 11;
m t : ‫ח כ ם ? עיניו אי ש ע שי ר ו ד ל מ בין בסיקךבזי‬
LXX σοφός π α ρ ’ έα υτω άνηρ πλούσ ιος,
πόνης δέ νοημω ν κ α τα γν ώ σ ζ τα ι αυτού

Here Targum and Peshitta have renderings such as bsr Ih, giving
also the sense ‘despise’, though the Vulgate with its pauper autem
prudens scrutabitur eum does not support them; and with Jerome
there stand Aquila and Theodotion whose όξιχνιά σ ει follows the
more normally recognized sense of this verb. It is not easy to
maintain that the LXX reached their rendering at 28. 11 by mere
guesswork from the context, and the philological suggestion (by
Winton Thomas), while hardly absolutely certain, is well worth
consideration. The ‫ ח ק לי ״ ״ ל ב‬of Judges 5.16 might then be thought
of as ‘scornings of the heart’, which in the context would be a
good sense. Another example occurs at Prov. 25. 27, but there the
LXX is very remote from the Hebrew and probably offers no
guidancec
Finally we may mention, for the sake of its intrinsic interest, the
peculiar rendering of ‫‘ ב א ל ץ‬in the land’ as μ ό λ ις ‘scarcely’ at
Prov. 11. 31:
m t : ‫סן צ ד י ק ב א ך ץ י ש ל ם א ף כ י ־ ר ש ע ו חו ט א‬
L X X et ο μβν δίκ α ιος μ ό λ ις σ ώ ζετα ι,
ο άσββης κ α ι α μ α ρ τω λ ό ς 7του φ α ν ^ΐτα ι;

Driver may be right in attaching this to the influence of the

Samaritan word cited by him as ‫‘ א ל ץ‬compel, coerce’. The
Peshitta agrees with the LXX in this rendering (ImaJuen).
1 See also Driver’s recent intricate restatement of his interpretation in J S S ix
(1964) 348 f. Whatever one thinks of the argument from the physics of wine in
a cup, the LXX is not relevant evidence. The translators felt, just as Driver now
argues, that drinking (or, rather, becoming drunk) is what matters, and they
short-circuited the whole suggestive imagery of the verse.
 F R E E R E W R I T IN G 259
These then show the variety which is to be found in the transla­
tion technique of books like Proverbs. As has been said, however,
only very few of the examples sustain the weight of use as linguistic
evidence in a philological treatment of the Hebrew.

(8) Additional Points in the Use of the Versions

Firstly, it is on the whole a support to a philological treatment
if it rests on evidence from more than one version, or from
versions in several languages. Take the instance of ‫ ד מ י‬, which has
been identified as meaning ‘hair (Driver, KB). The text for this
is isa. 3 8 .1 0 : ‫ ב ד מ י ; י מ י אלכה‬. LXX with its iv τω ϋφζί seems to
have taken this as ‫ ב ר ו ם‬or ‫ ב ר מ י‬. But among the other Greek
translations (though not Aquila, Symmachus, or Theodotion, who
took the word to mean ‘weakness’ or ‘silence’) we have the
annotation ol erepor iv ήμίσβι των ήμβρών μου, and this is sup-
ported by Peshitta bplgwt, Vulgate in dimidio. The argument that
‫ ל מ י‬could mean half is supported by the analogy of Accadian
mislu; the half is that which is ‘like’ the other half (is this con-
vincing semantically?). Ezek. 1. 13 has the phrase iv μέσω where
the Hebrew has ΙΥβΎΙ, and Driver argues:
If there had been no such word as ‫ ד מ י‬or ‫ ת‬1‫‘ ד מ‬half, it would have
been impossible for them so to translate 1‫ ו‬1‫ ד מ‬in a passage where the
context did not suggest that meaning.
The claim that it would be ‘impossible’ to suggest a different
explanation is certainly too strong; assimilation is not unlikely.
Nevertheless the agreement of versions in several languages does
appear to strengthen materially the case for this interpretation at
Isa. 38. 10, and Driver may very well be right in his identification.
While, as I have said, it is on the whole a support to a philo-
logical treatment if it rests on evidence from several versions
rather than one version, this is not true without some qualification
and reservation. Firstly, there is the possibility of interdependence
between versions in different languages: the LXX may, for
instance, at places be influenced by an Aramaic Targum, while
Peshitta and Vulgate may have followed the guidance of LXX.
Secondly, even where one version has not been influenced by
another, there remains a possibility that several versions have
alike been guided by the same tradition of Jewish interpretation.
260 T H E U S E O F E V ID E N C E F R O M T H E V E R S IO N S
In spite of this qualification, it still remains generally true that the
agreement of several versions in the interpretation of a Hebrew
word is a sign of some positive value, and that where only one
version can be quoted the chances of some idiosyncratic or ac-
cidental element in interpretation are the higher.
The evidence of the versions can also be related to our knowledge
of late Hebrew and Jewish Aramaic. A particularly interesting case
is that of ‫ ח ל ק‬. Many scholars, struggling with the difficult phrase
of Amos 7.4, m t ‫ ו א כ ל ה א ת ־ ה ח ל ק‬, must have thought of the
common Arabic halaqa ‘create’ and wondered if the sense might
be:
'It (i.e. the great fire) was eating up the creation’.
This was one of the first philological suggestions to occur to my
own mind when I first learned Arabic; later I found it also in the
work of scholars like Montgomery. I certainly did not wish to
abandon so promising a lead.
Such a verb ‫ חלק‬appears to occur in the Hebrew of Sirach,
where the Greek has κτίζω (about six cases). The same sense
‘create’ might also be applied to Job 38. 24:
m t : ‫אי ־ז ה ה ך ר ך י ח ל ק או ר‬

which could then mean something like:

'What is the way to where light is created ?’
The suggestion is at first sight an extremely attractive one; yet
important questions remain. Even if the late Hebrew of Sirach had
‫‘ חלק‬create’, how did this insight not penetrate any of the versions
at Amos 7. 4?1 Does the fact that Sirach had this verb ‘create’
suffice to show that the Old Testament had a noun ‫ ח ל ק‬meaning
‘creatures, creation’ ? No such noun is found in Sirach.
Moreover, ‫ ח ל ק‬in the Hebrew Sirach does not anywhere quite
clearly cover the field of cosmic creation and thus coincide with the
familiar ‫ ב ר א‬. The objects of the verb (or, if niphal, subjects) are in
no case the familiar cosmic terms like ‘heaven’ or ‘earth’. At 7. 15
the object is work, ‫ צבא מלאכה ועבדה‬, LXX γεω ρ γία . At 34.
13 the reference is to the evil eye; LXX (numbered 31. 13) πονη-
ρότερον οφθαλμού τ ί εκ τισ τα ι; At 34· 27 the subject is wine; LXX
1 I do not imply that there is any intrinsic impossibility in the knowing of a
word by Sirach and his translator which was unknown to the translator of the
Minor Prophets. The question is still a valid one.
 A D D IT IO N A L P O IN T S IN T H E U SE O F T H E V E R S IO N S 261
(31. 27) κ α ί α υ τό ς € κ τισ τα ι elg €νφροσννην ανθρώπους. At 38. I it
is the doctor and his work. At 39. 25 it is ‘good things’, Ü11D;
LXX α γα θ ά τ ο ΐς ά γα θ ο ΐς ε κ τ ισ τ α ι άττ α ρ χής. At 40. I it is labour,
‫ עסק גדול‬, Greek α σ χ ο λία μ β γά λ η . At 44· 2 it *s honour, ‫כבוד‬,
Greek 8όξα.
The sense is always derivable from the familiar biblical sense of
‫‘ חלק‬allot, distribute, provide’. The range of meaning is not such
that it can be extended with probability to the Amos passage.
There is some proximity to a reference which could verge on
creation, but the actual function of the word in Sirach is not
‘create’. It is thus unlikely that we can conclude to a noun ‫ח ל ק‬
meaning ‘creation’ as early as Amos, or, indeed, in Sirach himself.
The noun actually found, i.e. ‫ מ ח ל ק ת‬, seems to have a sense close
to ‘allotment’.1 Thus after all the Hebrew Sirach does not give
clear evidence of a word ‫‘ ח ל ק‬create’; and this negative view is
supported by the absence of such a word from other post-biblical
Hebrew sources.
It may be thought that, even in passages where the Hebrew is
lacking, the Greek κ τίζ ω nevertheless suggests a sense which is
strictly that of ‘create’. But this in turn depends on the sense of the
Greek word. Though it is used with a clear reference to cosmic
creation, it is also well documented with a sense like ‘make (some-
body so and so, e.g., free)’, and generally ‘bring about (such and
such a state)’. The Greek Sirach seems to use κ τίζ ω in both ways.
The full sense ‘create’ is clear in passages like 17. 1, κύριος €κτισ€ν
€κ γ η ς άνθρω πον and several others; but, from the evidence of the
Hebrew fragments so far known, we may suspect that the Hebrew
was ‫ ברא‬or ‫יצר‬, or at any rate not ‫ חל ק‬. The word is clearly a
favourite one of the Greek Sirach, which has about twenty-three
cases, as against about forty in all the rest of the Greek Old Testa-
ment. If we take the nouns for ‘creation’, i.e. κ τ ίσ ις and κ τ ί σ μ a,
where these can be set clearly against the Hebrew fragments, the
Hebrew words are ‫ ב ל י ה‬and ‫ מ ע ש ה‬. Thus the Greek evidence
also does not vindicate the sense ‘create’ for ‫ ח ל ק‬. The suggestion
‘creation’ for Amos 7. 4 has to rest on its own merits in the context
of Amos, and not on the Hebrew or Greek evidence in Sirach.
1 Smend, p. 68, gives Verteilung as the sense; at 42. 3 the Greek for this noun
is δοσυ; 41. 21 is not very clear. Our view has the support of Nöldeke in Z A W
xx (1900) 85 f., who argues that the sense in Sirach shows at most a transitional
meaning on the way that led to the Arabic sense ‘create’.
262 T H E U S E O F E V ID E N C E F R O M T H E V E R S IO N S
If we can show that the rendering of a version reflects late
Hebrew or Jewish Aramaic usage, this does not necessarily in-
crease its evidential value. For the discovery of the original sense of
a biblical passage it may be a point either of strength or of weak-
ness. It may be a point of strength in that a late Hebrew usage
may in fact have existed in biblical times. It may be a point of
weakness in that the late usage may have displaced understanding
of the biblical usage.
A verb to which some mystery attaches is ‫ ל ג ל‬. Ps. 20. 6 has
1!!‫ ו נ ת ל‬1‫ ש ם ~ א ל ה י‬. Symmachus here translates: τ ά γ μ α τ α τ ά γ μ α τ α
Β ίαστελοΰμεν. I would take this to mean ‘we shall divide into
separate companies’, and this is surely related to the use of ‫ ל ג ל‬now
known from the War Scroll. The Targumic ‘set soldiers in
order’ agrees in general with this. It is not quite correct to say that
the AV ‘set up banners’ goes back to these translations;1 rather, it
goes back to an interpretation on the basis of the meaning ‘flag’
for ‫ ; ד ג ל‬τ ά γ μ α is a standard and frequent LXX rendering for ‫ ל ג ל‬.
I think it likely that the philological interpretation of ‫ נ ל ג ל‬as ‘wait
upon’ (related to Accadian dagalu ‘wait for, look for, regard’) or
something similar is right; but the versions do not in fact provide
support for it and do not suggest that any ancient translator knew
of this meaning. LXX with its μ €γα λννθ ήσ ομ€θ α tried to render
as if the consonants wereg-d-l; the Vulgate followed with magni-
ficabimur. Jerome in thtiuxta Hebraeos wrote ducemus choros, which
is taking them as if they were d-l-g. All this gives the impression
that no one really knew the meaning.

(9) Uncertainty about the Meaning of the Version

Another point which should not be left out of consideration is
that the meaning of a Greek text is not always clear and unam­
biguous. We should not suppose that, where a Hebrew text is
obscure and uncertain, the consideration of a Greek rendering
necessarily brings us out into the clear light of certainty.
In Lam. 4. 15 we read:
m t : « ?T r n ‫־־‬n x Ti O
LXX ό τι άνήφθησαν κ α ί ye εσαλβυθησαν

1 Driver in H T R xxix (1936) 174 f.

 UN CERTA IN TY ABOUT TH E M EANING OF TH E VERSION 263
This has been taken as evidence for a Hebrew verb cognate with
the Arabic nasa ‘be joined’ (cf. Freytag, iv. 290, V ‘coniunctus
fu it\)
Does ά νάπτβ σθαι, however, in this text mean ‘be joined,
touch, be attached’ ? In the LXX this is the usual sense of the
simple verb ά π τβ σ θ α ι (most commonly translating ‫) נ ג ע‬. In close
proximity to our verse, in 4. 14 and 4. 15, ηψ αντο βνδνμάτω ν α υτώ ν
and ά π ό σ τ η τ ς μ ή απτ€σθ6 are both clear cases of a sense ‘touch’.
But the compound άνάπτ€σθα ί means not ‘touch’ but ‘kindle’ or
‘light’, whether this is the exact sense of the Hebrew (commonly
‫ ) י צ ת‬or only its more general sense (e.g. ‫ י צ א‬with fire as subject).
In twenty-four cases listed by Hatch and Redpath I think there is
no exception, and two are in Lamentations itself (2. 3, 4. 11), the
latter only a few lines earlier. The Greek sense, then, is not ‘be
joined’. The sense is: ‘do not touch them (i.e. the impure); for
they are set on fire, they are shaken . . .’. The translator construed
the word (a) as from ‫‘ י צ ת‬kindle’, or (b) as from ‫‘ י צ א‬burst out’ (of
fire), or (c)—in my opinion most likely—as from the root of ‫ני צו ץ‬
‘spark’ and ‫‘ נ צנ ץ‬be kindled’, found in the Midrash in 1 1 ‫נצה ב‬
‫‘ ר ו ח ה ק ד ש‬the Spirit of holiness was kindled in him’.1
Even within a familiar version like the LXX, then, the determina-
tion of the meanings of Greek words is not a simple m atter; and at
times the Greek support for a philological treatment has rested on
a doubtful or erroneous understanding of the Greek.
This difficulty is even greater when highly special and artificial
techniques of translation are used, as is most obvious with Aquila.
Some of his words are hapax legomena, special coinages not used
elsewhere in Greek; and some, though formally extant elsewhere
in Greek, have in Aquila’s work special senses intelligible only as
renderings of Hebrew and thus different from normal meanings in
Greek.
A word in an ancient translation, then, cannot be taken at its
face value. Similar difficulties may arise in the Aramaic and
Syriac versions, which may at times attempt to imitate the form of
the Hebrew original.
Sometimes this can produce very difficult problems of meaning.
This is particularly so if a word is not found, or is very rare, in
Aramaic apart from the Targum itself. If the root is the same as

1 Cant. R. 1. 12; Jastrow, p. 929b.

264 T H E U S E O F E V ID E N C E F R O M T H E V E R S IO N S
that of the Hebrew word translated, and if the Hebrew word itself
is of uncertain meaning, there are two opposite possibilities:
(a) the Targum may simply have produced a caique of the Hebrew
word; (b) the Targumic rendering may be evidence which pro-
vides a right comparative philological lead to the sense of the
Hebrew. Thus at I Sam. 7. 2,
m t : ‫רעהו כ ל ־ ב י ת י ש ר א ל א ח ר י י הו ה‬
the verb ‫ רנ הו‬has often been suspected and emended. The
Targum has
‫ו א תנ הי או כ ל בי ת י ש ר א ל ב ת ר פ ל ח נ א דיוי‬

The opinion of S. R. Driver1 was that the Targumic word was

itself based on the Hebrew, i.e. my possibility (a) above. This
might fit I Sam. 12. 14, where the same Aramaic renders. .. DjTPill
‫ א ח ל‬. But it is still not clear what the Aramaic word means.
One tradition has it to mean ‘be gathered’,2 (cf. the use to translate
‫ ונ קוו‬at Jer. 3. 17); another has it to mean ‘follow eagerly’,3 and cf.
Stenning’s ‘by our devotion to his words’ for ‫ב ד נ ת נ ה י ל פ ת ג מ ו ה י‬
at Isa. 53. 5.
The view that the Aramaic word is chosen because it translates a
similar Hebrew word is plausible at I Sam. 7. 2 and 12.14, but does
not explain the other passages, such as Jer. 3.17, 30. 21 (MT ‫)!עג ש‬,
31. 22 (remote from Heb.), 33. 13, Isa. 53. 5 (remote from Heb.),
Hos. 2. 17 (MT ‫) ו ע נ ת ה‬, etc. Thus even if the Aramaic word
received some kind of start in use as a rendering for Hebrew ‫נ ה ה‬
or like forms, it seems to have gone beyond this and to have
gained a currency of its own. I do not find examples in Aramaic
outside the Targum.
On the whole, I am inclined to believe that the Targum has a
right rendering of 1.‫ י נ הו‬at I Sam. 7. 2; that its rendering was with a
genuine cognate which gives a good explication of the difficult
Hebrew word; and that this is one of those words which are
found in the Aramaic of the Targums but not in other dialects
such as Syriac. The sense would most likely be ‘follow after, be
devoted’. The sense ‘follow’ is known also in post-biblical Hebrew,
e.g. Saadia writes DITriil ‫‘ ל כ ל‬to all who follow them’.4

1 Notes on Samuel, p. 62. 2 So Levy, ii. 94b.

3 So Jastrow, p. 881b. 4 Cf. above, p. 231.
 UN CERTA IN TY ABOUT TH E M EANING OF TH E VERSION 265
Another interpretation is found in KB,1following Driver,2 giving
a sense ‘hold to’ (KB says ‘hold with’, but means ‘hold to, adhere
to’) on the basis of Arabic naha (y ) glossed as ‘come’.
One of the difficulties of this solution is the semantics of the
Arabic word itself. Though we do have evidence for senses like
‘suivre le conseil de quelqu’un dans une affaire’,3 there is no
doubt that the massive preponderance of meaning in Arabic lies
in senses like ‘prohibit, terminate’. The sense upon which this
solution relies is probably remote from the chief or the earliest
meaning of the Arabic word. This does not make the solution
impossible, but must be noted as a factor in evaluating it. It may
be that this consideration makes the Aramaic evidence just quoted
more important and the Arabic evidence less so. The gloss
‘come’, offered by KB, certainly seriously over-simplifies the
matter.
Similar problems may be found with the Syriac translation.
Thus, though under Vt we find a noun ‘terror’ and a verb ‘terrify,
rush in, invade, oppress’, all cases registered by Brockelmann are
in the Old Testament and almost all are at places where the
Hebrew itself has ‫ ב ע ת‬. Some sense-borrowing from Hebrew may
have taken place. Cf. also the interesting case at I Sam. 20. 13
m t : ‫כי ־יי ט ב א ל ־ א בי א ת ־ ה ר ע ה ע לי ף‬
Pesh. ’n ’tb mn ’by bysf d ‘lyk
meaning ‘if I learn from my father evil which is against you’.4
The Syriac appears to have chosen a word similar in form to the
Hebrew, which also produced a possible sense in the context,
though that sense differed from the Hebrew sense.

(10) The Versions and the Grammar of the Original

One further point may be mentioned briefly. The versions will
not generally give reliable guidance on the grammar of the
original.5 A philological treatment will at times depend on the
assessing of a Hebrew form as originally, let us say, a noun or
an adjective or a passive verb. Sometimes one finds the argument
that a version took the Hebrew as such and such a part of speech,
1 KB, p. 599a. 2 J T S xxxiv (1933) 377·
3 Dozy, ii. 730, for intahd fi l-shay Hla ra'y fulani.
4 Cf. Brockelmann, Lex. Syr., p. 265b. Cf. also above, p. 230 (‫) חז ה‬.
5 See already above, pp. 209, 253.
266 T H E U S E O F E V ID E N C E F R O M T H E V E R S IO N S
because this was the part of speech used in the version. Such an
argument is very precarious unless we know from samples of the
version in this book that its technique included reproduction in
the rendering of the same parts of speech, so far as is possible,
which the original had. It will generally not be found that this is
so, and especially where the text is an obscure one anyway.
Thus at Jer. 2. 31 M T has ‫ א ר ץ מ א פ ל י ה‬but Ehrlich wants to
read ‫ ; מ א פ י ל ה‬and Driver supports this,1 saying that ‘all the
versions support a participial or adjectival form’. This is more than
can be asserted on the basis of the versions, which generally trans-
late the sense and reference of the original without trying to
reproduce its grammar. Similarly at Job 31. 23
M T: ‫כ י פ ח ד א ל י א י ד א ל‬
LXX φόβος γ ά ρ κυρίου σ υνέσ χεν μ ε
it is very precarious to argue2 that the σ υνεσ χεν μ ε ‘indicates a
verb for ‫ א י ל‬with pronominal object at the end of the clause’.
There are not, in fact, many instances in which philological
treatments have tried to obtain from the versions guidance about
the grammar and syntax of the original, and we need not give
further space to the matter, providing that students will be
warned against reliance on this kind of argument except where
specific evidence for grammatical awareness on the part of the
translators can be offered.

(11) Conclusions
The above discussion has taken account of the main factors and
problems which may be met in the use of the versions as evidence
for philological treatments. It remains to try to sum up the
matter with a general evaluation of the use of versional evidence.
The first thing to realize is that many passages which are difficult
for us today were difficult for the ancient translators also. In such a
position of difficulty these translators had to make what they could
out of the context and out of such indications as the text (i.e.,
primarily, the unpointed written text) had to offer. These indica-
tions might include ‘etymological’ similarities to other words,
1 Driver in J T S xli (1940) 165 f.
2 Driver in A J S L lii (1935-6) 165; cf., however, his somewhat different
treatment in Biblica xxxii (1951) 182.
 C O N C L U S IO N S 267
especially to words which were more familiar; they might oc-
casionally include suggestions and influences from the vocabulary
of other languages known to the translators; and they quite
commonly included a practice in which the letters were taken in a
different sequence or otherwise jumbled, or arbitrary word-
divisions were implied.1
In taking guidance from the context the ancient translators
worked from the more familiar words to the more uncertain; and
they were influenced by parallelisms and the general tenor of the
passage, and by associations with words which accidentally had
occurred just before. Their procedure was not entirely different
from that of the modern philological interpreter, and so it is not
surprising if from time to time their results agree with or seem to
support those produced by philological treatments. Though many
other considerations of the modern interpreter were absent from
the mind of his ancient predecessor, the influence of context
worked in a similar way. When the two agree, it is often the
result not of knowledge of a rare word but of analogous divination
from the context.
My own study of past examples, where scholars have offered
versional evidence to support the identification of Hebrew words
not previously recognized, has led on the whole to an unfavourable
judgement. Out of well over one hundred examples closely studied,
using the criteria and considerations discussed in this chapter,
only a few seem to me to be certainly valid examples, while another
proportion of perhaps 15 per cent, may present some reasonable
probability. In a considerable majority of examples the versional
evidence can be explained more easily in some other way than as an
indication that unusual Hebrew words were exactly known to the
translators.
It should not be altogether surprising if this conclusion is
reached. Let us consider particularly the position of the LXX.
Philological treatments, as I have shown, often imply a very op-
timistic picture of the knowledge of Hebrew vocabulary on the
part of the Greek translators. But this optimistic picture is by no
1 See examples like ‫ ד ג ל‬, p. 262; ‫ ש כ ב ת‬, p. 137 n.; ‫ ת ש ת ע‬, p. 233 n . ; ‫ ל ה ק ת‬,
pp. 25 f., 231 f., 270 f. Similarly the rendering of the unusual verb ‫ נ כ ר‬by
7re7rpaK€v at I Sam. 23. 7 is probably not evidence of knowledge of a rare verb
meaning ‘sell’; faced with a puzzling form, the translators just guessed that it
belonged to ‫* מ כ ר‬sell’, which luckily enough gave good sense. Cf. above, p. 209;
Index, no. 214.
268 T H E U S E O F E V ID E N C E F R O M T H E V E R S IO N S
means that which the classical tradition of Septuagintal studies has
maintained. Swete for instance writes:1
The majority of the translators had probably learnt the sacred
language in Egypt from imperfectly instructed teachers, and had few
opportunities of making themselves acquainted with the traditional
interpretation of obscure words and contexts which guided the
Palestinian Jew.
Thus philological opinion itself has traditionally been far from
unanimous in its optimism about the guidance to be found in the
versions. My own assessment of the versions is far from a negative
one. It seems to me in general that the ancient translators did
their task remarkably well, considering the circumstances.2 Their
grasp of Hebrew, however, was very often a grasp of that which is
average and customary in Hebrew. Our concern, on the other
hand, has been for abnormal or rare words or meanings. Rarity in
the sense of quite out-of-the-way terms in subjects like architecture
and tools did not necessarily leave the translators too much at a
loss. Where it is a matter, however, of obscure words in normal
contexts and of strange meanings for common words, there was a
strong tendency towards the levelling of the vocabulary and the
interpretation of that which was rare as if it was that which was
more normal.
Sporadic aberrant renderings may then quite accidentally
provide an apparent agreement with an unusual identification
made by us today.
In the case of the LXX a significant point is its setting within
the Egyptian Jewish community. The complete domination of
Greek within this community rendered it particularly lacking in
access to reliable knowledge of the sense of obscure Hebrew words.
Thus, in so far as a translation was likely to depend on local sources
1 Swete, Introduction, p. 319; and, for the opinion of a master in philology,
see Nöldeke, Die alttestamentliche Literatur (Leipzig, 1868), p. 246. Cf. also
above, p. 240 n.
2 Terms like ‘incompetence’ have, however, been quite freely used; e.g. Katz,
‘Septuagintal Studies’, p. 200, summarizing and confirming the conclusions of
earlier workers: ‘The translator of Isaiah who worked at an early date was com­
pletely unequal to his task. Many Hebrew words were unknown to him . . .’. I
differ, however, from workers like Dahood whose principle is that their philo­
logical identifications from Ugaritic are right and that the LXX, which does not
recognize these, is therefore wrong. Against many of Dahood’s interpretations
I would consider that the absence of support from the LXX shows that the latter
did know Hebrew quite well. Cf. p. 240, n. 1.
 C O N C L U S IO N S 269
of knowledge of the original tongue, the setting in Egypt counts
against a high expectancy of accuracy in the LXX.
This is particularly so, secondly, because the LXX bears the
marks of an origin very early in the total history of Bible trans­
lating. Its translation is extremely uneven. Later the Jewish com­
munity, after growing experience of the trouble which could be
caused by inaccurate translation, especially when it could be
exploited in theological controversy, came to seek greater accuracy
and uniformity; and so also did the Christians at a later time. The
later Greek translations may on the whole be expected to have
shown greater care and diligence in finding out what was really
known of Hebrew idioms; but etymologizing and literalizing
techniques may in return have obscured much of this diligence.
In general, then, the setting in the Egyptian community, and at
so early a date, should tend to make us doubt any great claims for
knowledge of rare lexical items on the part of the LXX. This does
not mean that such unusual knowledge may not sporadically
appear; but, to state the argument conversely, if we find that true
instances of confirmation of philological solutions by LXX
evidence are rare, then this would not conflict with our general
knowledge of the origin of this version.
Another way in which the evidence of the versions may be
tested is to consider cases in which the meaning, as discerned by
philological treatments, has not been seen by the versions. Even
those scholars who have most frequently maintained that the LXX
or other version discerned rightly a sense which philology has now
demonstrated to be correct have even more frequently had to say,
or at least to imply, that the versions missed the sense entirely.
Out of all the philological identifications which have been offered,
the number which can reasonably claim some degree of right
recognition by the ancient translators is not high—well below
25 per cent., I should say, even at a generous estimate, and
perhaps something nearer to 10 per cent. Thus, while there is
every reason to expect that versional evidence will occasionally
show correct recognition of a word now known to us through
comparative philology, the record of philological work itself does
not lead us to expect that such correct recognition will be
statistically very frequent.
Again, if we take not the total number of philological treatments
but the number of those which seem to us to be highly successful,
270 T H E U S E O F E V ID E N C E F R O M T H E V E R S IO N S
once more we find the relation of the versions to the right sense to
be rather weak statistically. I would consider it to be a highly
probable treatment when Ginsberg discovers a ‫‘ ר ב ע‬dust’ in Num.
23. 10:
‫מי מנ ה ע פ ר י ע ק ב ו מ ס פ ר א ת ר ב ע י ש ר א ל‬
But if the word meant ‘dust’, this was unknown to LXX (τις εξάρι-
θμήσεται δήμους Ισραήλ;), to Jerome (nosse numerum stirpis Israel),
to the Targum, which paraphrases on the basis of the sense ‘four’,
as does Aquila (τον τετάρτου Ισραήλ), or to the Peshitta (rwb'h
4the quarter’).1
Again, the attractive explanation of Π1ΤΓΠΡ?? at Gen. 49. 5 as
‘counsels, plans’ makes good sense and has good philological
support.2 The LXX rendering, εξ αίρέσεως αυτών, cannot be given
a quite clear interpretation, but it seems rather probable that the
εξ represents the ‫ מ‬of the Hebrew word, and that no root ‫ מ כ ל‬was
identified. Aquila with άνασκαφαί certainly saw here the verb PHD
‘dig’. Other versions give either a distant paraphrase or the sense
‘weapons’, which, even if it should be right, does not indicate any
exact understanding of the word ‫ מ כ ר ת י ה ם‬, since this sense would
probably be deduced from the context, and especially from the
preceding ‫ כ ל י‬, in any case. There is no evidence that any version
had linguistic knowledge which would support any philological
suggestion.
Again, I would consider that the identification of Πp ‫ ל ה‬at I Sam.
19. 203 is an attractive and probable treatment. The versional
renderings, however, are more naturally interpreted as general
surmises of the sense than as correct identifications of a rare word.
LXX has εκκλησίαν, which could be taken to imply that their
Hebrew manuscripts had ‫( ק ה ל ת‬so the traditional textual treat-
ment) but more probably is rather a construing of a written ‫ל ה ק ת‬
as if it were ‫ ק ה ל ת‬. The Peshitta likewise has knP ‘assembly’.
The Targum’s ‫‘ ס י ע ת‬company’ is a very general rendering, and
the cuneum of the Vulgate probably means simply a ‘number’ or
‘quantity’.4 The later Greek translators all have words simply
1 Index, no. 294, cf. no. 295. Cf. also above, p. 11 n.
2 See above, p. 57. A review of the evidence will be found in Skinner, Genesis
(ICC), p. 516 n. 3 See above, pp. 25 f., 231 f., 267 n.
4 Ducange provides an entry of the sense ‘any number of men or things’,
and from Jerome’s own writing one can cite adv. Iovin. 2. 8: quasi quidam pertur-
bationum cunei ad arcem nostrae mentis intraverint; cf. also ep. 92. 3.
 C O N C L U S IO N S 27 1

meaning ‘group’, ‘gathering’—Aquila όμ ιλό ν , Theodotion σ ύ σ τ η μ α ,

Symmachus σ υ σ τ ρ ο φ ψ . No translator shows any sign of knowing
the sense ‘old man, old age’, which is a component basic to the
comparison with Ethiopic.1
Even when a versional rendering agrees with a philological
identification, one may still have difficulty in deciding whether the
agreement betokens precise and correct knowledge of the sense by
the version, or whether on the other hand it may be a probable
divination from the context and other factors. At Num. 16. i
I consider the identification of a verb Hp*’ ‘be insolent’2 to be highly
attractive. Yet the agreement of the Greek νπ€ρηφαν€νθη may be
the result of a good guess rather than a right knowledge of the
sense of the word. We do not have much exact knowledge of the
version known as d Ε β ρ α ίο ς , but its tendency seems to have been
one of paraphrastic remoteness rather than accurate adherence to
the Hebrew.3 Since many versions at Num. 16. i show wide
diversity of understanding, along with considerable paraphrasing,
the chances are on the whole that o cΕ β ρ α ίο ς should be classified in
the same way. He was right, we may surmise, without knowing
how or why he was right.
We repeat, then, that the considerable number of identifications
which seem strong philologically but which have little support in
the versions forms a further reason why we should not expect that
versional evidence will support the philological treatment in a high
proportion of cases.
To this we add another argument. Philological treatments, where
they have appealed to versional evidence, have appealed much
more often to the LXX than to other versions, such as the Aramaic
Targums. But is this pre-eminence of the LXX not paradoxically
an argument against the procedure? If there were rare words in
biblical Hebrew, the knowledge of which was already dying out in
ancient times, is it not much more likely that this knowledge
survived among Aramaic-speaking Jews than among Greek­
speaking? Yet I have found philological treatments to appeal
relatively seldom to the Targums, in comparison with the number

1 LXX γήρας μητρός for at Prov. 30. 17 is surely a striking

coincidence rather than sound evidence; cf. Greenfield, H U C A xxix (1958)
212 ff. 2 See above, p. 17 f.
3 See Field, pp. lxxv-lxxvii, and especially the paraphrastic instances cited
at the top of p. lxxvi.
272 T H E U S E OF E V ID E N C E FR OM T H E V E R S I O N S
of appeals to the LXX. May it not be the weaknesses rather than
the strengths of the LXX translation techniques which have
provided apparent evidence ?
With this in mind we ought to reconsider the argument from
mistaken renderings. As we noticed above, scholars have appealed
to mistaken renderings by the LXX (or other versions) as evidence
of real linguistic knowledge, even though the rendering was
admittedly a wrong one at the point where it appeared. After
consideration of the examples, one must judge that this argument,
though a possible one, is not a probable one. Where a translator is
wrong, there will often be other explanations more likely than the
view that he has in mind some rare word known to him correctly
(though not correctly for the passage in question) but generally
unknown to scholars until now. The chances that he is confused,
or is guessing, or is paraphrasing, or is rewriting altogether, are
considerable. In the nature of the case most examples of alleged
4correct mistakes’, if we may so call them, are extremely isolated,
and great caution should be used before they are taken as evidence
of real linguistic knowledge. Finally, something should surely be
granted to the argument that a translator who is rendering wrongly
is, just because he is wrong, displaying some degree of incom­
petence in handling language.
 XI
SOME PARTICULAR LINGUISTIC,
LITERARY, AND CULTURAL
PROBLEMS
T chapter will gather together a number of other factors
h is
which may have an effect on our estimate of the probability of
philological suggestions. Some are more directly linguistic in
nature; others are more literary; and yet others belong to the
springs of human motivation, ordinary or religious, which may
underlie the words and passages investigated.

(1) Onomatopoeia
Professor Driver has laid some emphasis on the onomatopoetic
origin of Semitic words, and this has been apparent especially in
attempts to identify biblical birds and animals from their names.
Thus:
Many of the names of birds will be found to be in Hebrew as in
other languages onomatopoeic in origin; but no exact reproduction of a
bird’s cry must be expected. The onomatopoeon may represent a real
attempt to reproduce the original sound or may be a mere echoing
repetition of a single note to give the effect of a monotonous cry; and it
may undergo every form of linguistic assimilation or dissimilation.1
The principle is then extended to all sorts of other words.
Driver argues, for instance, that the element ‫ ג ע‬, as found in ‫ג ע ש‬, is
a noise made in the abdomen when relaxing or in the bottom of the
throat when vomiting; it is then applied to groaning or to disgust.
When used of rivers and waters, it means ‘rise up’, and this
explains ‫ ו תג ע ש‬at Ps. 18. 8.2
The principle is further extended so that it decides what is
primary and what is secondary in meaning. The familiar ‫ ע‬1‫ג‬
4expire, die’ is related, by similar onomatopoetic formation, to
1 PEQ April, 1955, p. 5; cf. ZAW lxv (1953) 255, 258; JSS vii (1962) 15 f.
2 ETL xxvi (1950) 341; cf. Greenfield in HUCA xxix (1958) 205 f.
274 L IN G U IS T IC , L IT E R A R Y , A N D C U L T U R A L PR O B L E M S
many words for throat or bowel sounds, and therefore must
‘originally’ have meant ‘gasp for breath’ and only secondarily
‘die’. By the same token the state of death is ‘properly’ expressed
by ‫ מ ל ת‬, originally ‘be mute’ (like many other words using/m/), and
so ‘be silent in death’.1
These reasonings, then, have an effect on the estimation of
probabilities in linguistic history. In the question whether there
is a verb ‫ ף‬0 ‫‘ כ‬be pale’, a start is made from the assertion that the
biliteral base /g, k, q + z, s, s/ expresses onomatopoetically the
sound of cutting and tearing. Then, it is argued, there is an easy
development from breaking to paleness, and kaspu, like Arabic
Jidda, will originally have described silver as ‘broken stuff’, since
it was purified by breaking two or three times.
Thus the onomatopoetic theory is used to state the sequence of
semantic development. The meanings of Hebrew words cannot be
deduced from cognate languages without some view of general
semantic probabilities. An onomatopoetic theory, because by its
nature it attempts to state something extremely early in language
formation, tends to dominate the assessment of developments in
meaning. It therefore has considerable importance for our present
subject.
Of this theory of onomatopoeia, it must be said that it has not
approved itself to the majority of Semitic scholars, nor does it
appear to have the support of modern general linguistics.2 There
has been a general abandonment of the attempt to find in onoma-
topoetic formation the origin for any substantial element in lang-
uage. Even for the names of birds and the like such an origin is
not widely found—for English, consider such names as sparrow,
thrush, or eagle, alongside names like cuckoo, peewit. Even for
sound-imitating words like ding-dong, it has long been noticed that
these vary from language to language, and in each language fall into
1 JSS vii (1962) 15 f.
2 On this see for example Jespersen, Language, pp. 396-411; Bloomfield,
Language, p. 156 f .; Hockett, Course, p. 298 f. Some of the words which look as if
they were onomatopoetic can demonstrably be shown to have become so only
through sound change; Saussure, Course, p. 69, cites French fouet, fouetter
‘whip’, derived from the Latin fagus ‘beech-tree’, which neither sounded like
nor meant a whip. From a more philosophical point of view, Ziff, Semantic
Analysis, p. 25, though himself using onomatopoeia as an argument against
the complete conventionality of language, admits that ‘onomatopoeia is of no
great importance in language’. A particularly good discussion is in Ullmann,
Semantics, pp. 82-96 and elsewhere.
 O N O M A T O P O E IA *75

the normal phonemic patterns of that language, although the

natural sounds do not so vary at all. Even these words, therefore,
have a strong element of the conventionality which characterizes
language. The elements commonly called onomatopoetic may
perhaps be better expressed as motivated terms, motivated ‘as
stylizations of acoustic impressions'.1 It may be felt that the word
is something like the natural sound; this is a sense or *motivation'
of the language user, but has nothing necessarily to do with origins.
This is not to say that there is no reality at all in the phenomenon
of onomatopoeia. The principal weakness of the idea, as used in
the philology of Semitic languages, is its application to the
origin of words. Applied to their function, as for instance in the
idea that certain words may be felt by their users to symbolize or
represent actual sounds, rather more value can be granted to it.
But the application of it to the origins of words, with the con­
sequent establishment of a ‘basic meaning’ through onomatopoeia
and the associated idea that language, though now conventional,
was originally not so, is extremely precarious. This, we may add, is
especially so with Semitic languages; their somewhat rigid re­
striction on possible formations leaves rather little room for sug­
gestion through word-formation. For example, it has sometimes
been thought that the vowel /i/ ‘is admirably adapted to convey an
idea of smallness’ ;2 but it is hard to see how this could work in
Semitic languages, where the choice of vowels is generally dictated
by the pattern in which words are cast.
Applied to Semitic the onomatopoetic theory is ultimately a
very limited scheme according to which labial stops like /p/ in­
dicate bursting open, dental stops like /d/ indicate cutting, nasals
like /m/ indicate silence or low humming sounds, and so on; but
this explains very little. There are indeed some Semitic words to
which such explanations appear to fit; but such cases are of little
importance, for they cannot be made to support a general theory
except by arbitrary selection of meanings and precarious histories
of semantic development.
The onomatopoetic theory therefore does not appear to have
positive value in the present discussion, and arguments about
semantic development which depend upon it are of no great
weight. Much more importance attaches to the theory of biliteral
1 Thieme, in Hymes, p. 586b.
2 See Ullmann, Semantics, p. 87.
276 L IN G U IS T IC , L IT E R A R Y , A N D C U L T U R A L PR O B L E M S
bases, which has often been associated with the idea of onoma­
topoeia ; but the effects of this have already been discussed above.
The importance of the biliteral base does not depend on association
with onomatopoeia; indeed, on the contrary, the extremely limited
range of meanings attached to the bases as normally identified
could only imply a high degree of conventionality, and cor­
respondingly a marked distance from the directness of
onomatopoeia, in the words developed from these bases.
(2) Some linguistic-cultural relations
There remain one or two conceptions about language in its
relation to thought and culture which may be mentioned briefly
because they have appeared in the context of philological treat­
ments. One of these is the idea that the original sense of words was
always a ‘concrete’ one. Thus Eitan writes:
It is a philological axiom that every word, whether noun or verb,
had (or its root had) originally a concrete significance and that only by
way of metaphor did it receive subsequently derived meanings, as need
for them arose.1
One may doubt, however, whether this axiom has great validity,
and still more may one doubt the attempts to use it as a basis for
argument in particular cases. Eitan’s own example is at
Exod. 17. 13; the sense, he believes, is that Joshua ‘carried off’
or ‘snatched away’ Amalek. So, he argues, the Palestinian Arabic
halasa ‘reap with a sickle’ is merely a specialized shade of the more
general classical halasa ‘carry off, snatch away’.
Thus, he goes on:
The Hebrews, a predominantly agricultural people, would naturally
have derived from this general signification a more special one, ‘to
reap’, i.e. to carry off with a sickle.
Eitan goes on to assert that this sense ‘would have been’ (sic)
preserved by Palestinian peasants up to the Islamic invasions (and
so passed from Hebrew into Palestinian Arabic).
I do not think that this particular type of argument has appeared
very frequently in philological treatments. I mention it here, how­
ever, because it illustrates how general conceptions of the relation
between language and culture may influence a scholar’s picture of
a probable semantic history.
Eitan, p. 35.
 S O M E L I N G U I S T I C - C U L T U R A L R E L A T IO N S 277
Occasionally solutions show some other kind of cultural prefer-
ence. Of the interpretations offered by Hirschberg in a recent
article,1 for instance, a large proportion have a sexual reference.
The notorious crux at Amos 5. 26 is interpreted by him through
taking both thePat^av of LXX (understood as a Hebrew ‫ ל י פ ה‬,
cognate with Arabic ra yyif‘fertile’) and the ]VD as sexual objects
or symbols.2
Philological procedures may at times be guided by particular
lines of cultural assessment, and the student should be aware of
this possibility if he is to evaluate philological arguments properly.

(3) Parallelism
Philological treatments frequently place a very heavy emphasis
upon the phenomenon of parallelism. That parallelism is a
marked feature of Hebrew poetry is a matter of common know-
ledge. But the importance of parallelism becomes even greater
when the establishment of text or of meaning is in doubt. If we
know the meanings, we notice and appreciate the parallelism, and
the parallelism in turn sets the meanings in relief and forms them
into a striking pattern. When we do not know the meanings,
however, the parallelism becomes one of the principal guides by
which we discover the meanings.3 Among philological treatments
of poetical texts, a large proportion make some considerable
appeal to parallelism. Can so heavy a load be placed upon it?
Some sophisticated analyses of parallelism have been produced.
These depend for the most part on the assumption that the mean-
ings of the various Hebrew terms are known. When the study is
rather an attempt to discover the meaning of a word, the situation
is different. The question is not: given the meanings of two parts
of a sentence, what is the nature of the parallelistic relation between
them? It is rather: given a sentence which appears to be paral-
lelistic, and given the meaning of one part, what chances are there
that this will help us to know the meaning of the other part ?
The obvious difficulties are:
(1) We do not know in the beginning whether the verse will be a
synonymous parallelism or some other kind; we may not
even have complete certainty that it is parallelistic at all.
1 In V T x i (1961) 373-85. 2 Ibid., pp. 375 f .; cf. Index, no. 224.
3 Cf. the saying of Menahem ben Saruk, above, p. 62.
278 L IN G U IS T IC , L IT E R A R Y , A N D C U L T U R A L PR O B L E M S
(2) Even where a verse is certainly parallelistic, we do not
certainly know what effect this will have on a particular
word within it. Two elements in a parallelism may be
identical or synonymous within the sense of the customary
definitions of these terms, and yet the words in one element
may have quite considerable differences in meaning from
the words in the other.
The former of these need, perhaps, not be illustrated with
examples; for the second, a simple illustration in the English will
suffice:
Ps. 24. 3: ‘Who shall ascend into the hill of the Lord ?
And who shall stand in his holy place?’

In this familiar sentence, let us suppose that one of the words in

the second part is of unknown meaning. The parallelism helps us,
but it helps us only a certain way. If the word ‘place’ were un-
known, we might surmise from the parallelism that it meant
‘temple’ or ‘house’ or ‘area’ or ‘ground’. If we really pressed the
idea of synonymous or identical parallelism so as to make it apply
to each word, we would guess that DIpD is another Hebrew word
meaning ‘hill’. There are places where the parallelism of two words
for ‘hill’ occurs, e.g. Isa. 10. 32:
‫ינפף ידו הר בית־ציון גבעת ירושלם‬
The same would be true if the meaning of Dip ‘stand’ were uncer-
tain. Parallelism might make us guess that it was another word for
‘go up’.
This is a central problem in all use of parallelism as an in-
terpretative key. The parallelism is a parallelism of the literary
units; it does not guarantee that the lexical units used within each
literary unit are synonyms.1 One element in a parallelism may use
‘walk’, while another uses ‘sit’ (e.g. Isa. 9. 1; Ps. 1. 1 has ‘walk’,
‘stand’, and ‘sit’ as the three verbs).
Let us see how this works in a real instance of philological
treatment. We have already looked at the interesting proposal to
1 This is already recognized by earlier works, e.g. Driver in J T S xxxiii (1932)
43: ‘parallel passages do not necessarily demand verbal identity’ and ‘close
strophic parallelism is not so required’ that words in parallelism ‘must mean the
same’.
 P A R A L L E L IS M 279
identify a Hebrew ‫‘ ע ש‬rottenness’ at Hos. 5.12. The sentence as a
whole is:
: ‫ואני כעש לאפרים וכרקב לבית יהודה‬
It is the close parallelism which gives strength to the suggestion
that ‫ ע ש‬means ‘rottenness’, thus being identical with ‫ ל ק ב‬, rather
than meaning ‘moth’. Both words then mean ‘rottenness’, giving
an absolute parallelism. But when one looks at this a second time,
one realizes that the absolute parallelism is not so great an ad-
vantage. Only if the parallelism is identical, not only in the two
thoughts expressed in parallel, but also in each word used in
each of the two thoughts, is the identification of two words both
meaning ‘rottenness’ entirely satisfying.
Conversely, even if the relation is between ‘moth’ and ‘rotten-
ness’, we still have a good parallelism. The two parallel phrases
are both about forces of decay and corruption. If the total parallel
sentence is an expression about decay or corruption in general,
it is possible that the effect is strengthened, and not weakened,
if we have words for two different corrupting agencies, i.e. moth
and rot, rather than two words for rot. Thus, even if the interpreta-
tion of ‫ ע ש‬as ‘rottenness’ gives a more exact synonymy for ‫ ל ק ב‬,
the understanding of ‫ ע ש‬as ‘moth’ is actually perhaps more true to
the general pattern of Hebrew parallelism. These considerations tell
on the whole against the acceptance of this particular philological
suggestion.
Parallelism, then, even when it seems to be of a ‘synonymous’
type, does not in itself provide any assurance that a particular
word will be similarly parallel or synonymous with the correspond-
ing word in the parallel clause. Or, to state the matter in another
way, even when we know that there is a synonymous parallelism,
we do not know in what aspects the two realities set in parallel will
be compared. In the case just discussed, if ‘rot’ is specifically the
point, then two words with this meaning will give a good effect;
but if the point is a more general one of ‘spoiling’ or ‘decay’ or
‘deterioration’, then ‘moth’ and ‘rot’ provide a good parallelism
even though they do not mean the same thing.
The argument from parallelism is rather stronger in Jer. 4. 31:
‫כ י קו ל ב ח ו ל ה ש מ ע תי צו־ ה כ ס ב כ י ך ה‬
Though it is possible to take ‫ צ ל ל‬as ‘anguish’, the parallelism with
280 L IN G U IS T IC , L IT E R A R Y , A N D C U L T U R A L PR O B L E M S
‫‘ ק ו ל‬voice’ was invoked by Driver in favour of an interpretation as
‘shrill cry’, implying another word ‫ צ ל ה‬, cognate with Arabic
sarra and sartr, and homonymic (at least in the consonants) with
‫‘ צ ר ה‬anguish’. The use of the verb ‘hear’ makes it a little more
likely that the second object will be a sound rather than a state of
distress. This argument, indeed, is not absolutely convincing.
Firstly, in Hebrew one could probably speak of ‘hearing’ distress
or anguish.1 Secondly, one could say that a phrase ‘anguish like
(that of) one bearing her first child’ is loosely attached to ‫ ל ה‬1‫ ח‬,
without implying logical dependence on ‫ ש מ ע ת י‬. Nevertheless the
point has some considerable strength, and may be further con-
firmed by an appeal to the LXX:
076‫ ־‬φω νήν ώ ς ώ δινονσης η κ ο ν σ α , ro d σ τε ν α γ μ ό ν σου ώ ς π ρ ώ το -
το κ ο υ σ η ς.

Further evaluation depends on consideration of the same

explanation at Jer. 48. 5 (LXX 31. 5). At 4. 31 the fact that the
same verb ‘hear’ seems to apply to both objects strengthens the
case for a closely similar sense. At 48. 5 there is a parallelism with
‫‘ ב כ י‬weeping’. On the whole I think that this example of philo-
logical treatment has been successful.
A still more striking example is Isa. 2. 16:
m t : ‫ו ע ל כ ל ־ א נ י ו ת ת ר שי ש ו ע ל כ ל ־ ש כ י ו ת ה ח מ ד ה‬
L X X : κ α ι 6776 π α ν π λο ΐο ν θ α λά σσ η ς κ α ί 6776 π ά σ α ν θ£αν πλοίω ν
κ ά λλο υς.

Driver argues that sense and parallelism ‘require’ ‫ ת‬1‫ ש כ י‬to be

the name of a kind of boat and ‫ ה ח מ ל ה‬the name of a country. The
identification of ‫ ש כ ת‬is widely accepted; that of Arabia Felix as
the country is less certain. ‘Ships of Tarshish’ is a set phrase; it is
less likely that another phrase existed relating the name of ships of
another region. The parallelism of cedars of Lebanon and oaks of
Bashan, on the other hand, is stereotyped (cf. Zech. 11. 2; Ezek.
27. 5f.). The passage may go as follows:
All that is high and lofty all that is uplifted and low
All cedars of Lebanon all oaks of Bashan

1 Cf. in jer. 41.11 (object ‫ ;) ר ע ה‬4 8 .29 (‫ ;)גאון מו א ב‬6 . ‫;) ח מ ס ו שד י שמע( ל‬
job 20.3 (object ‫) מו ס ר כ ל מ תי‬.
 P A R A L L E L IS M 281

All the high hills all the lofty mountains

Every high tower every fortified wall
All the ships of Tarshish all the lovely boats (views?)
‘Low’ (‫ ) ש פ ל‬in the first line, if this is the right text and sense,
implies that the parallelism is not absolute. In the LXX rendering,
it may be θέαν that translates ‫ ש כ י ו ת‬, and πλοίων may be an ex-
planatory addition to fit the parallel of the first hemistich; or, since
θαλάσσης is a midrashic-etymological rendering of ‫ ת ר שי ש‬, πλοίων
may be added to provide a complement.1 No other translation
perceives a word for ‘ship’ (cf. Aq., Sym. οψβις, θέας ; Syr. dawqe).
Thus, though there is a strong case for the identification of
‫ ש כ ת‬as the name of a boat, the argument from parallelism in itself
is not entirely decisive.
A similar argument arises at Isa. 44. 24-28; a considerable
chain of closely identical parallelisms may still leave uncertain the
position in parallelism of a single term:
He who says to Jerusalem, Thou shalt be inhabited
and to the cities of Judah, You will be built up;
and its ruins I shall set up again;
He who says to the deep, Be dried up;
and its streams I shall dry u p ;
He who says to Cyrus ‫ר ע י‬
and all my purpose he will complete;
Saying to Jerusalem, Thou shalt be built,
and to the temple, Thou shalt be refounded.
What is the sense of ‫ ? ר ע י‬Traditionally it is taken as ‘my shep-
herd’, parallel with ‘my anointed’ in 45. 1. But the parallelism in
44. 28 is kept very close if we follow the LXX 6 λέγω ν Κύρω φρονεΐν
and interpret as in Syriac r a y reyana, etc., ‘thought’:
He who tells to Cyrus my thought
and all my purpose he will complete.
The less rigid parallelism of ‘my shepherd’ and ‘my anointed’ is,
however, more probable, and the LXX analysed as the same root
which is found in ]!‫( ר ע י‬διαλογισμός Dan. 7. 28); cf. especially
Ps. 139 (138). 2:
33‫ת ה ל ר עי‬ συ σννηκας τούς διαλογισμούς μου .
1 On this treatment of ‫ תר שי ש‬see B JR L xxx (1967) 291 f. and J S S xii (1967)
117 f.
282 L IN G U IS T IC , L IT E R A R Y , A N D C U L T U R A L PR O B L E M S
The designation of Cyrus as ‘my shepherd’ is in fact a good
parallelism in thought, though not an exact one in form, with the
completion of the purpose of the Lord.
In general it is a good rule to interpret by observation of the
context. In the circumstances of our study, however, where one
linguistic element is obscure, the following of context is not
necessarily a good method. An unusual word may say just what one
would not expect on the ground of context alone. It is by following
context, where one linguistic element is uncertain, that interpreters
and translators are led to harmonizations and assimilations.
Parallelism may make clear the general bearing of a passage with­
out determining the exact information furnished by particular
words. There are different kinds of parallelism, and sometimes a
passage may turn out to be not parallelistic at all. It cannot be
invoked as a straightforward guide to the meaning of uncertain
words. It remains, however, a very important factor, which the
scholar cannot neglect. Finally, where Ugaritic words are used for
the elucidation of Hebrew, it should be remembered that the
meanings of many of these depend in the first place on paral­
lelisms in Ugaritic, and the same caution has to be used in any
reliance on these Ugaritic meanings.
(4) Religious Factors
The problems discussed in this book do not belong directly to
theology or the study of religion, and we have worked almost
entirely on a textual and linguistic level. There are, however, ways
in which judgements of a religious or theological nature may
affect our estimate of probabilities in a philological or a textual
treatment. These would seem to fall into two main categories:
(a) Judgements of fact about the religious situation of a par­
ticular time, which may affect our discernment of what a
text may be saying, and
(b) Judgements of style and taste, of religious logic and theo­
logical consistency, which may help us to know whether a
given utterance ‘makes sense’, whether it is credible that a
person of a given period could find it meaningful.
Turning to the first of these two categories, we should give some
particular mention to Nyberg. His discussion of the text of Hosea
was accompanied by the development of some rather original
 R E L IG IO U S F A C T O R S 283
views of the religious history of Palestine, and these views in turn
reacted on his textual and philological discussion. The combina-
tion of these two was characteristic of his method.
At the difficult place Hos. 7. 5:
: ‫יו ם מ ל כ נ ו ה ח ל ו עזרים ה מ ת מיין מ ש ך ; ד ו א ת ־ ל ^ צ י ם‬
Nyberg1took ‫ מ ל כ נ ו‬to refer to the god Melek and ‫ ש רי ם‬to refer to
his heavenly court; while of the words IT* ‫ מ ש ך‬he pronounced that
their meaning was unknown, but that they seemed to refer to
some ritual action, in which the god Melek was subject.
A similar element in Nyberg’s thought is the prominence of the
divine name fAl. Nyberg identifies this at several points in Hosea,
and its presence enables him to avoid emendation. Thus at Hos.
11. 7 (usually dealt with by emendation):
‫ ע ל י ק ר א הו‬- ‫ו א ל‬
Nyberg changes only the word-division and reads the text as
‫רו־דא‬.··‫ק‬: ‫ע ל ·י‬t ‫א ל ־‬v ‫ו‬:
*and therefore to *A1they cry’.2
The identification of the divine name eAl, this time in the form
‫( ע ל ו‬cf. Ugar. 'lw)y has been particularly well received at I Sam.
2. 10: ‫ם‬..‫ר ע‬: ‫ם _י‬. ‫מי‬ ‫ ע ל ו ב ש‬. If the first word is taken as the divine
_ T _ T T
name ‫ ע ל ו‬, we have good parallelism with the other half-verse, and
there is no need for the textual emendation to ]‫ ע ל י ו‬, proposed by
3 H 3f which gives us semantically the same result.
Such then are some examples of the possibility that information
about religious thought in the Near East may enable us to see
meaning in a biblical passage where it has formerly eluded us, and
thereby avoid emendation of the text.
Less frequently discussed and more subtle in nature is the
question of whether a given religious utterance ‘makes sense’ or
not.
There is a familiar verse in Proverbs:
Prov. 29.18 ‫? אי ן חזון י פ ר ע ע ם‬
AY *Where there is no vision the people perish*.
1 Nyberg, Hoseabuch, p. 50. Cf. Hos. 8. 10, ‫ מ ל ך שרי ם‬, Nyberg, Z A W , p. 251.
2 Nyberg, op. cit., p. 89; for his chief argument about ‘A1 see p. 58; and for
the divine name *Am cf. p. 27.
284 L IN G U IS T IC , L IT E R A R Y , A N D C U L T U R A L PR O B L E M S
RSV ‘Where there is no prophecy the people cast off restraint’.

Driver’s comment is: ‘Vision can hardly be said to restrain a

people from excesses.’ This conviction leads him to seek another
explanation; and, guided by the ε ξ η γ η τ ή ς of LXX, ού μ η ύ π α ρ ξη
ε ξ η γ η τ ή ς εθνει, π α ρ α νο μ ώ , he considers a word w meaning
‘magistrate’ and cognate with Accadian and Aramaic words.
This interpretation is weak, however, for two reasons. Firstly,
it fails to take into account the high originality of the Proverbs
translator. It is possible that his ε ξ η γ η τ ή ς 9 meaning vaguely a
guide or interpreter of the divine will, is only his interpretation of
‫‘ חזון‬vision’. He may also have seen a reference to the familiar
synagogue term ]|Π. The Hebrew word in any case is to him only
the source of his literary inspiration.
Secondly, in the context of Hebrew religion, with the importance
attached by it to prophetic ‘vision’, it is an entirely credible
sentiment that ‘vision’ should keep a people from excesses or from
disaster. The original doubt about the meaningfulness of the text
if the word was understood as ‘vision’ is thus misplaced.
The ‫ גי לו ב ר ע ל ה‬of M T at Ps. 2. π f. raises the same kind of
question. Is it indeed an improbable conjunction to construe as
‘rejoice with trembling’? Would the psychology of religion find
this lacking in congruity? The sense that some other meaning
must be required was present already in the Middle Ages, when
interpreters were already discussing the possibility of a ‫ גי ל‬meaning
‘tremble’ or the like. Reider thought that the sense was ‘worship’,
and compared Ps. 43. 4 ‫ ש מ ח ת גי לי‬, understood as ‘the joy of my
worship’; Driver identified a word meaning ‘be affrighted’ (see
Index, nos. 76-77). Is not the textual reorganization ‫ב ר ע ל ה נש קו‬
‫ ב ר ג ל י ו‬still the most convincing suggestion?1
Another interesting example is found in the Qumran Isaiah text
iQIsA at 52. 14. Here the M T reading is:

‫כ ך מ ק ז ח ת מ אי ש מ ך א ה ו‬

This has commonly been taken (with vocalization as ‫ ) מ ש ח ת‬to

mean:
‘His appearance was so marred, beyond human semblance.’
1 Cf. above, pp. 5, 175.
 R E L IG IO U S FA C T O R S 285
This is, of course, from the verb ‫‘ ש ח ת‬mar, spoil, destroy’. Now
the reading of the scroll is :
‫כן משחתי מאיש מראהו‬
Guillaume, discussing this, says that the verb cannot have been
the ‫‘ מ ש ח‬anoint’, which is ‘inconceivable without Christian ideas’.
He believes nevertheless that this is a verb ‫ ; מ ש ח‬it must be one
different from the familiar ‫‘ מ ש ח‬anoint’. Once again philological
treatment produces a new homonym. The true cognate in Arabic
is not masaha ‘smear’ but masaha, which ‘in its primitive sense’
meant ‘to gall the back of a camel and exhaust it’, and is also used
of the fraying of thread and of the transforming of men into
animals. The Arabic form masth is, according to the Qâmüs, ‘of
ugly form and without comeliness’. The meaning of the Isaiah
passage is thus probably ‘so did I mar his appearance’.
We do not need to go into the complicated series of considéra-
tions which are needed to evaluate this suggestion. The point for
the present is that we should note the starting-point in the assur-
ance that in this context the idea of anointing is impossible.
Such then are some examples of the way in which judgements of
theology and of religious history may enter into the determination
of probabilities in textual and philological matters.
(5) The Argument from Actuality
It may seem quaint that one should take a title such as this, since
purists might demand that all arguments should be based on some
kind of actuality. What I mean here is to notice comprehensively
all sorts of information which is not itself primarily linguistic or
literary in nature but which may affect our judgement of the
possible sense of linguistic items. Archaeology, the study of the
geography of the ancient world, the knowledge of social customs
and technological development may all by their increase help to
explain and clarify, and thereby justify, linguistic terms which
were formerly obscure, or may give suggestions for semantic
connexions and developments which were a cause of trouble.
Such arguments from actuality may thus form the start of a
dissatisfaction with the received meaning of a text, and so lead to
the search for new meaning through textual or philological treat-
ment. We have already mentioned how archaeological discovery
has validated the text of I Sam. 13.21 and given us clear evidence
286 L IN G U IS T IC , L IT E R A R Y , A N D C U L T U R A L PR O B L E M S
for the weight ‘pirn' or ‘payim’. There is nothing new in the kind
of argumentation we are describing here; it is one of the oldest
forms of exegetical thinking. All I want to do is to illustrate it for
modern philological treatments.
Pregnancy is esteemed as a great blessing by Oriental women.
Yet the speech of God to Eve in Gen. 3. 16 is, by the usual
interpretation, a statement which represents pregnancy as some-
thing punitive or disastrous. Through this observation, Rabin is
moved to seek a different meaning for the word ‫ ה ר נ ך‬. He is in-
fluenced also by the parallelism with ‫‘ ת שו ק ה‬desire', and he finds
the meaning to be probably ‘sexual desire', and possibly ‘whining',
with reference to Ugaritic and Arabic cognates.
The actuality observed may be a piece of fact known from out-
side the story, or it may be the congruity of the story itself. At
Judges 19. 2 Driver cleverly notices that the words ‫ו תזנה ע ל י ו‬
‫ פי לג שו‬should not mean ‘she prostituted herself', as the customary
understanding of the verb ‫ זנ ה‬would suggest. There is no hint of
such an element elsewhere in the story of the Levite’s concubine.
Indeed, the tragedy is substantially heightened if the woman had
been chaste right up to the time of the evil deed at Gibeah.
Moreover, several of the versions appear to agree in avoiding the
stain on the poor woman. The Targum has ‫‘ ו ב ס ל ת ע ל ו ה י‬she
despised him'; Jerome writes reliquit eum\ LXX has /cat (Lpyiadr]
avrcp (A text) ‘she was angry at him', or inopevOrj d7r5a vro v ‘she
departed from him' (B text). This is of interest because the normal
meaning of the verb ‫ זנ ה‬must have been sufficiently familiar to
these translators. The cognate Accadian zenu ‘be angry' may thus
give a solution, incidentally adding another homonym to Hebrew.
Place-names are another possible setting for this kind of problem;
and especially so when, as sometimes happens, a form can be
construed either as a place-name or as some other kind of word.
We saw above1 the possible place-name Izalla at Ezek. 27. 19,
which (if correct) serves to remove an enigmatic participial form
from the lists of exceptions in our grammars and lexica. Philological
treatments can cause place-names to disappear as well as to come
into existence. Some interpreters have removed Ophir from the
text of Isa. 13. 12:
‫ הם אופיר‬3‫ואךם מ‬: ‫ז‬$‫אוקיר אנוש מ‬
1 See p. 189.
 T H E A R G U M E N T FROM A C T U A L I T Y 287
arguing that it means:
T shall value a man more than gold;
more than fine gold shall I esteem a man.’
The word ‫ א ו פ י ר‬is then supposed to be from a verb ‫ י פ ל‬, cognate
with Arabic wafara ‘be plentiful’, Ugaritic ypr ‘make to abound’.
If this is right, one presumably has to argue that the other two cases
of the collocation ‫( כ ת ם א ו פ י ר‬Job 28. 16; Ps. 45. 10), plus the
other significant associations of Ophir with gold, have produced an
assimilation of the sense at Isa. 13. 12 and caused the verb to be
mistaken for the place-name. I myself think that ‘gold of Ophir’ is
a much more likely sense for Isa. 13. 12 also.
 XI I
SUMMING-UP
W e have sought in this study to work out the criteria by which we
may judge a suggested philological treatment to be successful or
unsuccessful. These criteria are not rules the simple observation
of which will certainly lead to a right result. They themselves are
probable rather than absolute; and sometimes they may seem
capable of working in either direction—as we have seen, a philo­
logical suggestion may be confirmed by the fact that a parallel
exists in post-biblical Hebrew, but may also be weakened thereby.
There may always be extreme cases which defy the probabilities
which generally apply; but it would be unfortunate if some such
extreme cases were to be taken as the rule and thereby allowed to
justify a long series of others equally extreme. Thus, though we
cannot set out rules which can be mechanically followed, the
following is a summary of the points which should normally come
into consideration when a philological treatment is suggested:
(1) How far the word in question lies within the normal
phonological correspondences with a cognate word considered for
its elucidation, and what are the long-range consequences for
comparative studies if this is ignored or treated lightly.
(2) How far the meaning given for the cognate word is a real
meaning, stated with adequate precision, and known with reason­
able probability as one which may go back to such time that a
cognate with a meaning semantically relatable may have survived
in Hebrew also.
(3) A critical consideration of the semantic connections pre­
sumed in the identification, in relation to acknowledged examples
and with a developed awareness of the difference between
theoretical constructions and historically controlled evidence.
(4) An awareness of the possibility that words are adoptions
from non-Semitic languages, or have been influenced in their
history by such adoptions; or that the information given about
them in dictionaries has been confused through contamination
with such adoptions.
 SU M M IN G -U P 289
(5) A consideration of the possibility of a textual error, i.e. that
the form in the text has arisen by an accident of the graphic
transmission.
(6) If the new identification produces a new homonym in
Hebrew, some consideration of how this is statistically related to
other homonyms and to the total extent of homonymy in Hebrew
and in Semitic languages generally.
(7) If the new identification produces a new synonym or near­
synonym in Hebrew, or, to put it another way, if it comes very
close to the semantic field of words already known, some con­
sideration of the change of balance this causes in the lexical stock.
(8) Some consideration of the statistical probabilities for the
assumption that a word in another Semitic language will be likely
to lead to the identification of a cognate in Hebrew.
(9) If the identification relies on versional evidence, a full
consideration of the versional reading for (a) the possibility that
the version had a different Hebrew text, (b) the style and transla­
tion techniques of the translator of this book, (*:) possible vicis­
situdes in the graphic transmission of the version itself, and, in
general, (d) the possibility that the rendering rests not upon exact
knowledge of the Hebrew word but upon other influences, such as
theological interpretation, paraphrasing, etymologizing, and
divination from parallelisms or from the general context.
(10) A consideration of post-biblical usage, including Jewish
Aramaic; and including also the possible effects of midrashic
expositions upon the history of understanding; and also, finally,
the views of the medieval Jewish lexicographers.
( n ) If the new identification involves an abandonment of the
Massoretic vocalization, a consideration of how the Massoretic
vocalization arose and what it means, and of how our abandonment
of it, seen in the long run, would affect (a) our view of the history
and nature of Massoretic activity, (b) our total knowledge of the
Hebrew language and meanings in it.
(12) Finally, it goes without saying that all new identifications,
even if all these requirements have been taken into account, have
still to be weighed for probability against the more traditional or
accepted meanings for the Hebrew words in question.
Of these various criteria, the one which deserves some further
elaboration at this point is the semantic one. As was noted earlier,1
1 Above, pp. 88 if.
290 SU M M IN G -U P
the semantic side in traditional philology was often naive, hasty,
and thoughtless. The following points are offered as suggestions
which may assist the student:
(1) The approach to the study of meaning must be strictly
linguistic, rather than logical, in character. It may seem strange to
insist on this: readers may ask whether, during the long reign of
philological method, this has not always been so. The answer is
that it has not. The older historical and comparative philology,
lacking an adequate approach to questions of meaning (especially
because of its concentration on the separate tracing of atomic
elements in their development), in its semantic notions very often
fell a prey to conceptions of a logical rather than of a linguistic
nature. This can be plentifully illustrated from the traditional
Semitic philology. Take a statement (admittedly, a fairly extreme
example) about ‫ נ ת ר‬. This verb is supposed to mean ‘tear’; but the
piel is used of insects which hop. In Arabic we have tarra ‘be
severed’, tarr ‘prancing, trotting’; thus, Driver writes:1
There can be no objection to connecting n-t-r as applied to locusts
with tarr as applied to light horses; and the underlying idea by which
verbs of rending or tearing asunder are linked to verbs denoting trotting
quickly or prancing or hopping seems to be that of separation from the
ground, whereby the beast appears to be now touching the earth and
now suspended in mid-air.

Though it is an extreme case, this passage typifies the way in

which many semantic relations were set up in the older philology.
The investigator, taking a series of forms which seem to be cognate
and of which the senses are known, works by thinking out a com-
mon factor attaching to them all, so that a more or less direct
relation from this common factor to the meaning of each form in
each of several languages can be seen. The procedure is logical
rather than linguistic in type, because it works by consideration of
common features in the referents rather than by working out a
process of meaning transfer within stages of linguistic usage. It is
true that Driver frequently detects ‘secondary’ elements in these
groups of meanings; but this is hardly enough to alter the basic
character of the method. The method is fundamentally one of
association of actual features of groups of referents. It is difficult
1 In the Robinson Festschrift (1946), pp. 70 ff.
 SU M M IN G -U P 291
to find any way of relating such an approach to that of modern
semantic studies.1
Linguistic history would seem to suggest rather that the mean­
ings of groups of cognates, collected across the lines of temporal
differences and from different languages, will tend to be haphazard
in their relation to common features of any group of referents
designated. Meanings are not derived from one basic idea, or
directly from a class of referents, but from the meanings of forms
already found. Form x, let us say, designates the referent^; and in
association with form x we find the forms y and z> which appear
to be cognate with x. It may be expected that, if all the history of
development is known, their meanings may be seen to have some
relation to that obtaining between x and its referent p ; but we do
not know which of the many aspects of this relation may be relevant
for the senses of y and z, nor do we know that the aspect relevant
for y will also be relevant for z. Semantic transfers will not general­
ly flow smoothly from a basic general idea, but may arise from
features which are quite minor, and which are thus logically
accidental and unpredictable, in relation to the meaning of a form
taken as original, or in relation to a previous change of meaning.
This criticism of a ‘logical’ approach does not mean that logic
is irrelevant, or that semantic work can be done without some
foundation of general theory. The logic implied in the process we
have just criticized would seem to be one of simple relations
between things and names, the names attaching to evident
qualities of the things.2 A more sophisticated logic of linguistic
usage may not necessarily have this damaging effect.
(2) Students may find it helpful to use the distinction between
reference and information.3 By ‘reference’ I mean that to which a
word refers, the actual or thought entity which is its referent.

1 As exemplified, for instance, in recent treatises such as the works of

Ullmann.
2 Such a logic seems to be implied in the very important ‘as* of the traditional
Hebrew dictionary, e.g. when it is explained that ‫‘ חנית‬spear’ is so called ‘as’
flexible etc.; so BDB, p. 333b, though this explanation is-feot necessarily ac-
cepted by BDB. The logic of this ‘as’ is all-pervasive in BDB; and no difference
is made by the simple observation that for this or that word a different derivation
can be suggested.
3 Already mentioned above, p. 118. The terms of the distinction were sug-
gested to me by that between ‘meaning’ and ‘reference’ in Quine, From a Logical
Point of View, e.g. pp. 9, 21 f., 47, etc. I have used ‘information’ because it has now
become fairly technical for the effect of choices within a recognized sign system.
292 SUMMING-UP
By ‘information’ I mean the difference which is conveyed, within
a known and recognized sign system (a language like Hebrew or
Arabic), by the fact that it is this sign and not another that is
used. The major linguistic interest, it would seem to me, lies in
the latter. Many arguments in which biblical scholars adduce
linguistic evidence appear to me, however, to involve some con­
fusion between the former and the latter.
For instance, there are certainly places where Canaanite and
Aramaic inscriptions use a word cognate with Hebrew □IpD, and
where the reference is to a place of burial. It does not follow that
this word communicates the specific information ‘tomb, grave,
place of burial’. Rather, the writers, referring to a tomb or the area
around it, called it a ‘place’. While it is interesting to note that
the word is used of a tomb, this fact does not entitle us to suppose
that ‘tomb’ (as information of a distinction from any other place) is
the meaning of this word, and then to transfer it, as Dahood wants
to do, to Hebrew passages like Job 16. 18, Qoh. 8. 10 (where it does
not improve the sense anyway).1 Though the place referred to in
some inscriptions is in fact a tomb, this does not make ‘tomb’ the
information specified by the choice of this word and thereby
transferable to other contexts.
(3) An increased emphasis should be laid upon the statement
of meaning in one language only. A Hebrew word has its meaning
only in Hebrew, an Arabic word only in Arabic. We have to
overcome the heritage of that supposedly comparative approach
(actually anti-comparative in its effects) which defines a Hebrew
word by thinking about what ‘it means’ in another language. The
resources used in philological treatments (for example, the mean­
ing of an Arabic word) are meanings in that language, while
the results to be reached are meanings in Hebrew. Meaning in
Hebrew is independent of meaning in Arabic, and depends on
choices within the Hebrew lexical stock of a given time. This is
an argument for, and not against, the comparative study of
Semitic languages; it is a source of confusion when scholars quote
Arabic or other meanings as if it were somehow ‘natural’ that they
would also be the meanings in Hebrew, and part of sophistication
in comparative philology is the learning to avoid this kind of
confusion.2
1 Index, no. 280.
2 One example out of many: Gray, Kings, p. 306, mentioning a quite normal
 SUMMING-UP 293

This point is also confirmed by the emphasis on ‘information’

which we have just discussed. The communication of meaning
in Hebrew was not determined at all by what the words meant in
Arabic, and was not primarily determined by what they meant in
proto-Semitic; it was primarily determined by the choice between
the words available in Hebrew. Because words operate in relation
to one another within Hebrew, all philological solutions to par­
ticular difficulties have to be considered not only in themselves,
for the satisfaction they give in a particular difficult context, but
also in their more general implications, for the effect they have on
the total balance of the available series of choices in Hebrew.
(4) One may on the whole expect that the results of our seman­
tic work, in the conditions which call for the use of philological
treatment from the cognate languages, will be probabilistic and
approximative rather than decisive and exact. This is not because
some philological suggestions are right and others wrong; rather,
it is something that is likely to attach to the method intrinsically
even if it can be substantially improved and refined. Only oc­
casionally will solutions achieve such certainty that they can
become given realities of the Hebrew lexical stock in the same way
as familiar and well-attested words with numerous known contexts
are given realities.
For one thing, our knowledge of the semantic developments
which have led to the various meanings of cognates in the ancient
Semitic languages is generally hypothetical and indirect; and, for
another, even if we can be sure of the rightness of a suggestion
inspired by a cognate word, it is unlikely that this will provide us
with an exact knowledge of the sense in Hebrew. Such a sug­
gestion is more likely to assure us of the rightness of the Hebrew
text and of the general field of meaning in which the word may lie;
this being done, the further determination of meaning will depend
mainly on the context. Precise references, nuances, and over­
tones, even when suggested within a philological treatment, will
generally be decided not by the evidence of cognates but by the
Hebrew context itself.

occurrence of ‫‘ ל ב ש‬honey’, tells us that it ‘might indicate Arabic dibs* (a liquid

preparation from grape-juice), and then goes on to agree that it probably means
honey (from bees). The repeated casual dragging in of the Arabic meanings, as
if they had some natural probability, is a source of confusion, and not of en-
lightenment.
294 SUMMING-UP
For instance, the identification of ]‫ ב צ ק ל‬on the basis of Ugaritic
bsql is as certain and successful a philological treatment as can ever
be expected.1 Yet it does not give us the exact sense in Hebrew.
We know that in Ugaritic bsql is a kind of vegetation, the growth
of which is highly esteemed and desired, and that it is of value for
food. Applied to the Hebrew example, this information is enough
to give certain negative precisions: it excludes the possibility that
the initial ‫ ב‬is the preposition ‘in’, and so it excludes all exegesis
in senses like fin his bag’. We are thus sure that some kind of
grain or vegetable is meant (or the plot in which such was grown);
but this was already clear as the general sense of the passage in
any case, and the uncertainty was its relation to this word. We
cannot be sure that there was a Hebrew ‫ ב צ ק ל‬which meant exactly
the same plant as was meant at Ugarit. There are ample instances
of words for animals, foods, commodities, and so on, where
different objects are designated by words formally cognate.2
Thus even a very successful philological treatment can give a good
deal of negative discrimination but does not necessarily furnish a
clear and positive precision of meaning.
These then are some ways in which we may expect the semantic
side of the assessment of philological treatments to develop.
Some further explanation should be added about the emphasis
just made on meaning within one language. The point is that the
meaning of words can be stated as the difference made by their
choice as against other words in the same language at the same
time; and negatively this implies that senses cannot be indis-
criminately imported from other languages just because words are
cognate. This does not mean, however, that the cognate languages
can be merely neglected, or that intra-Hebraic etymologizing is to
be encouraged. The latter has, indeed, historically been the source
of much trouble, from which the use of cognate languages has done
much to liberate us.
That ‫ ת ש ת ע‬- ‫ א ל‬at Isa. 41. 10 had the general sense of ‘do not
fear’ was perhaps always fairly clear; but it was the discovery of
Phoenician ‫ ש ת ע‬and the Ugaritic cognate which made it clear that
1 See above, p. 26.
2 Cf. dibs! ‫ ד ב ש‬, above, p. 292 n.; Hebrew ‫ ל ח ם‬is ‘bread’ or ‘food’ generally,
but Arabic lahm is ‘meat’ (Hebrew ‫ ) ב ש ר‬and Socotri lehem is ‘fish’; ‫ א רי‬is ‘lion*
in Hebrew, but Geez ’arwe ‘wild animal’ generally, Tigrinya yarawit ‘serpent*
(see Leslau, Harari, p. 31), and also Accadian eru ‘eagle’, etc.; see Ullendorff
in VT vi (1956) 192.
 SU M M IN G -U P 295

this was indeed the sense, and so overcame the necessity for a
dubious derivation within Hebrew through ‫ ש ע ה‬in a sense like
‘gaze about (in fear)’. Similarly, we have seen in the case of ‫חז ה‬
how the sense ‘agreement’ was known long ago, but this under-
standing came to be confused and darkened by attempts to derive
from the verb ‫‘ ח ז ה‬see’. In such cases the philological treatment
validates the form and meaning as Hebrew, where the tendency to
derive from other extant Hebrew words has obscured or confused
the morphological interpretation and the semantic tradition.
Again, it has been a common tendency to take the Hebrew
‫ מ הי ר‬in a phrase like ‫( סו פ ר מ ה י ל‬Ps. 45. 2) as meaning ‘quick’,
because the verb ‫ מ ה ר‬is ‘hasten’; but comparison with the Ethiopic
sense ‘train, teach’ and other cognate senses may suggest that the
sense is rather ‘skilled’. Similarly, in Isa. 16. 5 ‫ מ ה ל צ ל ק‬may be
‘skilled injustice’ rather than ‘swift injustice’. ‘Skilled’ is in fact
given as a gloss in the standard dictionaries; but in KB it is the
second gloss given, while in BDB it is the fourth. The effect of the
comparative information is to restore primacy to a sense which the
Hebrew context itself favours, but which has been somewhat
obscured by the attempt to show derivation from the sense of the
actual verb in Hebrew. Such an attempt, we may add, is favoured
by the tendency to look for a common root-meaning and to
identify it if possible with the sense of the verb.
When we talk of meaning within Hebrew, then, this is the
meaning of units as seen within the collocations in which they
occur and within the semantic fields of agreement and opposition
in which they function. Attempts to state meaning derivationally
cannot be intra-Hebraic in many cases, because the forms and
senses extant in biblical Hebrew may not be those through which
the derivation actually took place. One of the values of the com-
parative approach has been that it has set free senses, which are
likely in the Hebrew context, from domination by derivations
reached from within Hebrew alone.
Our arguments here have some effect on priorities in education
for biblical scholarship. The strong influence of comparative
philological method may have produced an unfortunate over-
emphasis on comparative study in the training of students. The
prestige and fashionableness of the philological approach often
cause students to study a larger number of Semitic languages
than they can master. These languages are not mastered properly
296 SUMMING-UP
and all the effort does not lead to a thorough knowledge of the
texts. This in turn leads to the cheapness and poor quality which
we have sometimes seen reason to deplore in philological treat-
ments of the Hebrew Bible, and especially to the phenomenon of
dictionary-searching.1 To observe this, unfortunately, is not
enough to put a stop to the tendency. The intellectual prestige of
the philological approach is reinforced by the apparent social
prestige of linguistic polymathy. It continues to be widely sup-
posed that study of a large number of Semitic languages is the
gateway to competence in biblical studies.2
All this would not matter if the effect of learning many languages
were accompanied by the acquisition of a thorough knowledge of
Hebrew. This, however, is not always so. Students may sometimes
be found to begin with Hebrew but, before they have really gained
wide experience in this language, to be directed to other languages,
on which their more developed and mature talents are then spent.
Thus the comparative emphasis, though intended for the elucida-
tion of the Hebrew Bible, in educational practice has sometimes
weighted the balance against a thorough knowledge of Hebrew.
Moreover, in the comparative approach there has been a certain
tendency to treat Hebrew as the distinctively unknown language,
while the other Semitic languages (with some exception for
Ugaritic, in which, as we have remarked, comparative insights are
particularly the basis of knowledge) tend to be treated as known
quantities, used as sources from which Hebrew can be elucidated.3
Thus the student, starting out to learn Hebrew, has to learn a
series of languages which are to serve as sources for the elucidation
of Hebrew. Within the study of these other languages, however,
1 Cf. the similar argument of Driver in VTS iv (1957) 5 etc.
2 I have sometimes advised students not to over-extend their mental energies
by attempting too many languages, and thus for example to avoid the effort of
learning Arabic when they had no intention of reading Arabic texts on any scale;
yet have been told that it would be a real disadvantage to them in an academic
career if they had only a small number of Semitic languages registered on their
records.
3 An occasional exception will prove this rule: Driver, discussing Ps. 22. 30
‫ כ ל ־ ד ש נ י ־ א ר ץ‬, says that the clue is in the Syriac rendering kpnh dV°; this
does not mean famelici terrae (Walton), but ‘those who are wrapped up’ in the
earth. This sense of k-p-n is not in the dictionaries but is suggested by cognate
words such as Hebrew ‫‘ כ פ ן‬be involved, intricate’. Note the parallelism with
‫ י ו ר ד י ע פ ר‬- ‫ כ ל‬. Thus the Hebrew is used as part of the evidence to explain
the Syriac, and the Syriac, once so explained, is used to explain the Hebrew. This
type of argument is, however, very exceptional in my experience.
 S U M M IN G -U P 297

he does not find to anything like the same degree a spirit of ex­
pectancy that problems will have to be solved by going outside of
the language being studied: Arabic is not taught as a language
which has to be elucidated through reference to Hebrew. More­
over, Arabic is now increasingly taught as a spoken language, and
this enhances its familiarity; sometimes it is the only Semitic
language known to the biblical student as a living mode of com­
munication.
Thus the proliferation of the comparative philological approach
has created a practical problem in the teaching of Hebrew. One
might expect the logical sequence to b e: first learn Hebrew; then
gain experience in the reading of Hebrew literature; finally, progress
to a study of cognate languages and their literatures. This sequence
was already made more difficult by the rise of textual emendation,
which greatly limited the ease with which biblical texts could be
used for the simple gaining of experience in reading. Now philo­
logical treatments appear to make it doubtful whether there
exists an agreed body of ‘Hebrew’ in the sense in which there is an
agreed body of ‘Greek’ or of ‘French’, and to suggest that there is
no stable basis until ‘Hebrew’ has first been reconstructed from
cognate sources. If such is the case, then the task of teaching is
indeed difficult.
Perhaps, indeed, there is no help for this, and biblical Hebrew
is simply a more corrupt and obscure entity than the other clas­
sical and Semitic languages. In its degree of being a consistent and
knowable linguistic entity, suitable for study in itself, Hebrew
may perhaps be more like Ugaritic than like Arabic or Syriac,
more like Sogdian or Tocharian than like Latin or Greek. I am
not sure, however, that this unhappy conclusion should be drawn,
and one doubts whether it has been meant even by those philo­
logists who have most devotedly added to the content of Hebrew
from cognate sources.
In spite of the multitude of difficulties and obscurities which have
provoked the recourse to philological treatment, it must be empha­
sized that Hebrew is a knowable and manageable linguistic corpus,
with a rich though diverse transmission of meaning from ancient
times on. The gaining of substantial experience within this corpus
is one of the most important ways in which the information yielded
by cognate sources can be controlled; and failure to study and
utilize the tradition of meaning has been one of the ways in which
298 S U M M IN G -U P
the philological approach has tended to impoverish scholarship.1
In spite of our debt to comparative philology, Hebrew does remain
a teachable subject in its own right; and, while the student must
now always be aware of the contributions of cognate languages,
he will, unless he is ready to study these languages thoroughly,
be best employed not in gaining a smattering of them but in
learning how to evaluate, in relation to his Hebrew knowledge, the
suggestions made on the basis of them‫ ״‬This means that eventually
adequate modes of communication and co-operation have to be
built up between two kinds of scholar: (a) those who really know
the cognate languages or some of them (can any now really know
them all?) and (b) those who only assimilate this knowledge
within their own grasp of Hebrew. But we can at least do something
to depreciate the false prestige which has attached to the polyglot
ideal, and rebuild the picture of the Hebraist. The polyglot ideal,
we may remind ourselves, by no means obtains in the Indo-
European field; no one supposes that to appreciate Greek literature
one must study all the Indo-European languages.
It would, on the other hand, be a hopelessly retrograde step if
one were to imagine that work within Hebrew might once again
proceed as if the pressure of comparative insights and methods
merely did not exist. This pressure is no new thing, but has been
acting upon biblical study for centuries; only lately, however, has
its presence begun to cause a kind of crisis in understanding. The
discussion of the comparative philological approach reveals many
new ways in which Hebrew requires to be studied, and makes it
impossible for us to contemplate a return to a traditionalist
Hebrew study, divorced from the framework of general linguistic
study (a framework even wider than that of Semitic comparative
philology).
We may now pass on to mention some particular practical tasks
in which our discussion may have some effect.
(1) Our discussion should have some effect on the work of
biblical translation, which is constantly going on. Sometimes the
production of philological suggestions has been connected with a
plea for the production of an absolutely accurate translation of the
Hebrew Bible. The approximative character of the information
1 In this respect philology, by going behind the tradition of meaning, has an
aspect similar to archaeology when the latter goes straight to the ancient sites
and cuts behind the literary historical traditions.
 S U M M IN G -U P 299

yielded by comparative approaches, however, means that this is by

no means easy to achieve. Translations have often been conceived
with the aim of furnishing the ‘right’ meaning. Where the philo­
logical information provides the exclusion of some impossibilities,
plus a range of relative probabilities, this is likely to force a
premature choice. Unless several alternative translations are
furnished, therefore, the act of translation is likely at times to give
a poor rather than an adequate representation of the result of
philological work.
This, we may add, is one main reason why the biblical student
has to study the original languages. It is sometimes argued that
improved modern translations are making linguistic study less
necessary than it used to be. Quite the contrary is true. Increased
modern philological knowledge, while it has enabled us to over­
come certain difficulties which the older translations had, has
also made the assessment of meaning more approximative in
character, has multiplied the factors which may bear upon it, and
may thus have made the production of definitive translations
more remote.
(2) Our discussion has shown the importance of certain
statistical information, such as comparative counts of homonyms
and of the percentages of words having cognates in groups of
related languages. The procedures used in this book for these
questions are extremely crude; the refinement of techniques for
this sort of problem should be undertaken in the years ahead.
(3) The desirability of an etymological dictionary of the Semitic
languages has often been asserted. Our discussion has shown that
such a work should not only register the existence of a word in a
certain language, but also negatively its non-existence. Registra­
tion of the non-occurrence of cognates is essential if we are to
assess the percentage of cognates normally found. Similarly, it has
to register the cases where a language has a certain root, but only
in a word-class or part of speech different from that of the other
languages; or, likewise, where it exists only in one or two particular
fixed forms, and is no longer free or productive.
(4) Dictionaries of Hebrew should be designed in such a way
as to distinguish those statements of meaning which are based
upon the actual collocations in Hebrew from those statements
which are based upon cognates and etymological information.
One might consider that a dictionary, instead of beginning with the
300 S U M M IN G -U P
mention of cognates, might restrict itself to stating meaning
within Hebrew, leaving etymological discussion to a separate
dictionary, as is done with modern languages like English; or it
might place the citation of cognates after the discussion of the
Hebrew contexts, so as to avoid the possibility that the statement
of meaning in Hebrew is distorted in order to show a plausible
line of development from the preferred etymology; again, typo­
graphical devices might be used in order to distinguish the different
kinds of source. The advantage of a separate etymological dic­
tionary would be that it could give full discussion instead of only
the brief citation of preferred examples.
If dictionaries were produced which gave less prominence to
comparative material than has lately been the custom, they might
give more space to contextual factors which indicate meaning, such
as parallelism (this is a most important indication, even though
we have shown that it is far from a simple or infallible one), and
to the oppositions or contrasts with other items which might be
expected to enter the same semantic field.
(5) The study of biblical Hebrew cannot be deemed complete
without a satisfactory follow-up into post-biblical Hebrew. The
present biblical dictionaries make only limited excursions as far
as Sirach and Qumran. It would not be an impossible undertaking
to construct in one volume a dictionary of the post-biblical
interpretation of biblical Hebrew words, in which would be dis­
cussed the treatment of words implied in the early Greek and
other versions, used in halachic and midrashic interpretation, and
stated in the Jewish lexicographers and commentators, up to
about a .d . 1300 or 1400. Such a work would be far from providing
the ‘right’ interpretation of biblical words; but it would provide
the setting within which linguistic meaning was transmitted and
might thus help us to assess ways in which such meaning had
been either preserved or distorted.
(6) Another practical problem is the design of a critical edition
of the Hebrew Bible. Classically, an edition has been conceived as
a representation of the textual evidence. But the significance of
this evidence depends on linguistic considerations. How can space
be found in a critical apparatus, however, for linguistic as well as
textual evidence? Yet to print only textual evidence, without
linguistic discussion, is to invite retrograde and conservative
constructions of it from many readers. Versional evidence in
 SUMMING-UP 30!
particular cannot be printed without some linguistic evaluation.
Perhaps no means other than compromise and good sense can be
found to solve this problem.1
It is clear that philological treatments have now come to
occupy the centre of interest which in the scholarship of the
nineteenth and early twentieth centuries was held by the practice
of emendation. Amid the general loss of confidence in these
emendations today, it is easy to forget that in their time they may
have served a significant purpose. They were an attempt to
rationalize the study of the text, in that they insisted that scholarly
solutions must make sense within the recognized body of know­
ledge of the Hebrew language: arbitrary exceptions to this could
not be made, and the artificial interpretative tricks of earlier periods
were to be rejected. Simply because this was tried so hard, and
because emendation explored and exhausted its own possibilities,
emendation in due course provided a certain important stimulus
to the new rise of philological treatments.
Nevertheless emendation produced in the end a deep depression
and scepticism, for the impression given was that the interpreter
in many cases was rewriting the text rather than explaining what
was written. The most serious criticism was both for arbitrariness
and for banality. The confidence that the grammar and lexicon of
Hebrew were already well known, a confidence which is part of the
rationalizing tendency just mentioned, had the result that many
emendations presented a much more familiar-looking Hebrew
than the existing text displayed. Emendations which result in an
easier text, however, break the familiar rule of difficilior lectio
potior (a rule which, indeed, is not universally and absolutely
valid). The process implied that an easier text was corrupted into a
more difficult, a corruption which is far from satisfactorily ex­
plained by an appeal to the irrational stupidity and incompetence
of scribes.
The philological treatment displays a different and a better
approach in these regards. It is just to claim with Winton Thomas
that it ventures out into the unknown; in a puzzling text it has
the chance of discovering something new, not a scrambled version
of familiar Hebrew but words and phrases which hitherto were not
known to be Hebrew at all. Such an expectation of expansion
1 On these problems see in particular my review of the sample edition of
Isaiah in JS S xii (1967) 113-22, and especially pp. i!9 ff.
302 S U M M IN G -U P
was only right in an era when the knowledge of Semitic languages,
and particularly those of the environment of the Old Testament,
was being greatly enlarged.
Nevertheless there is some danger that the present pursuit of
the philological treatment will end by being regarded with the
same deep scepticism which has been the fate of the fashion for
textual emendation. Firstly, many of the followers of the philo­
logical approach have also from time to time used drastic emenda­
tion of the consonantal text. As for the vowel text, the low opinion
of it has been so widespread, and the rewriting of it so drastic, as
to create an even worse impression of arbitrariness than was
produced by the older emendations of the consonantal text.
Emendation had generally treated the existing signs, i.e. the written
letters of the MT, as evidence which serves as a clue towards the
discovery of the true text, or which at least must be explained when
the discovery is made; but philological treatments have often
held themselves free to alter the vocalization without giving an
account of it, and have not accepted the existing vocalization signs
as presumptive evidence. Thus the treatment of the vocalization
in the philological approach has often been more arbitrary than
straight emendation of the consonantal text has been.
Secondly, and perhaps more fundamental, there is the sense
that philological treatments are often a semantic rewriting of the
text: just as emendation abandoned the graphic tradition of the
text in order to create a new or reconstructed writing, so philo­
logical treatments abandon the semantic tradition of meaning in
Hebrew in order to create a determination of meaning from else­
where. For these reasons philological treatments are not quite so
far removed from textual emendation as some claims would
suggest, and the danger of an ultimate scepticism hangs over them
both alike if their procedures are pressed too recklessly and too
uncritically. The bewildering number and variety of philological
treatments, and their lack of considered comprehensive theory
hitherto, are likely to make the student eventually feel that not
all of these suggestions can be right.
Such a judgement can by no means be attributed to an ante­
cedent hostility to the philological approach; for it is one which
may be confirmed also to some considerable extent by the judge­
ments of those who have themselves been foremost within it.
Even where scholars have committed themselves whole-heartedly
 S U M M IN G -U P 303

to the comparative philological approach, this fact in itself has not

necessarily led to any considerable measure of agreement among
them. Of the very numerous suggestions which have come from
Tur-Sinai, for instance, how many have received wide acceptance
even among devotees of the philological approach? Again, many
of the critical analyses used on philological treatments in this book
are represented also in Driver’s very strong strictures on Dahood.1
Once we allow for considerable crudity and extravagance in the
application of the method by Dahood, we must ask whether the
method strictly as a method is not only that which Driver himself,
and the movement for philological treatments in general, has
implied throughout. In other words, these strong criticisms of
Dahood, while striking at a lack of judgement in the application of
a method, do not appear to provide us with adequate means to
discriminate between a satisfactory philological method and one
which is not satisfactory.2 Thus, to sum up this point, philological
treatments have not as yet, in spite of a general agreement in
method, succeeded in presenting a very wide body of agreed
results; while, on the other hand, in discussing the method as a
whole we can claim to be doing no more than drawing out the im­
plications of types of critical evaluation which philological scholars
are already tacitly applying to their own and their colleagues’
work. Our investigation is thus not a criticism of the method of
philological treatments, but a making explicit of the criteria
implicit in the method.
It would, in my judgement, be a great misfortune if the move­
ment for philological treatments were to be received in the end
with unthinking scepticism. Faulty it has been in a number of
regards. It has prided itself on freedom from the faults of
textual conjecture, when it has in fact shared these faults and has
also obscured the fact that textual emendation is not at all an
impossible or invalid approach. Many suggestions have been
offered over-confidently, and the exultation of knowledge about
1 In yS S x (1965) 112-17; this review appeared when the argument of the
present book was already complete.
2 Cf. the judgement of D. N. Freedman in Hyatt, The Bible in Modern Scholar­
ship, p. 303. While far from committing himself to the rightness of Dahood’s
interpretations (he gives the cautious qualification ‘even if only a fraction stand
the test of scholarly criticism5) he maintains that Dahood’s work is ‘only a
logical extension of the pioneering effort of other Ugaritic specialists’. Quite so;
but some may judge that such an extension proves that there is something wrong
with the logic.
304 SU M M IN G -U P
new-found materials like Ugaritic has often encouraged extravagant
over-assertiveness and the ignoring of problems which are quite
plain to common sense. As I have argued several times, there has
been a failure to face the cumulative consequences which follow
when a method, quite valid in the single case, is applied over many
hundreds or even thousands of examples. There has been a failure
to consider the problem systematically and to give equal considera­
tion to all the criteria which are relevant.
If this has been so in the past, however, it does not mean that it
need be so in the future. If an adequate awareness of the wider
aspects of the problems, and especially of the relevant criteria, is
fostered, there is every reason to expect that the better solutions
will be well received and the poorer sifted out. The basic assump­
tion, that study of the relations between the Semitic languages
may further the understanding of the Hebrew Bible, is incon­
trovertible. The trouble has not lain in comparative scholarship,
but in poor judgement in its application, and in failure to see and
follow out some of the general linguistic questions which are
already implied in the primary use of comparative method.
Though it is not true that the comparative method is any longer
the major way to the understanding of language, and though it
has in the past sometimes shown an undesirable imperialism, it
remains highly desirable that its interests should be followed up.
This book is therefore consecrated to the deeper and fuller study
of that linguistic world in which the Hebrew Bible is set, and
against the background of which our future growth in under­
standing of its language must take place.
The time is past, however, when the primary creative and positive
contribution of comparative study was the production of in­
dividual philological solutions. That this was tried, and to some
extent accomplished, is entirely understandable, just as, at an
earlier stage of scholarship, it is understandable that every effort
was made to exploit the method of conjectural emendation. We
have now reached the stage, however, at which mere productivity
in philological suggestions has become negative and confusing in
effect, and the positive task is the exploration of the method in
itself. This present book is, I believe, the first to be wholly devoted
to this purpose.
 APPENDIX
SYRIAC VERBS AS IN BROCKELMANN,
SET AGA IN ST HEBREW VERBS AS IN BDB
(SLIGHTLY MODIFIED)
,
{see a b o v e p p , 1 6 2 /.)
A B C D
Hebrew, Hebrew,
with close Hebrew, but not not
Verb Sense sense sense remote verb Hebrew
Vs be evil x
bgh be delayed x
bgl be talkative x
bgn cry, seek x
bgr close, hold in X1
escape (?) X
bd mix X
be poured out X
be disturbed X
(etbadbad)
bd ’ pretend X
be talkative x(?)
bd l be mad X2
bdq scatter X
examine X
disclose X
b d q ll (denom.) repair x
bdr disperse x3
bhbh be anxious X4
bhl be quiet X
bhq shine X
bhr be glorious X
bht be ashamed x
b(w)d perish X
b(w)q rot X5
bzur be uncultivated X
bwt pass the night X
bz spoil x
bzh scorn X
bz' cleave X
b zq scatter X

1 Cf. MH ‫ בג ר‬be of age*. 2 Hebrew sense = ‘divide*.

3 ‫ ב ז ל‬rare, usually 4 .‫ פ ז ר‬Unless cogn. Hebrew ‫‘ ב הו‬vanity*.
5 Or perhaps column D,
306 APPENDIX: COMPARISON OF SYRIAC AND HEBREW VERBS
A B c D
Hebrew, Hebrew,
with close Hebrew, but not not
Verb Sense sense sense remote verb Hebrew
bhn examine x
show X
bhr examine x(?)
choose (?) X
repair ( ?) X
bhs move, excite X
btl cease X1
btn (denom.) conceive X
by’ console X2
byn understand X
by* cease, delay X
byt (denom.) make X
familiar
bk’ weep X
bhr be ripe X
bl mix X
bl’ wear out X
blm silence, X3
bind m outh
bl* be struck X
bis flower X
blq appear, X
be directed
bU spoil X
bn’ build X
bs’ despise X4
bsm smell good, X
enjoy
bsr despise, refute X
bsr (denom.) X
be incarnate
b*’ desire X
b*d depart X
b*t kick X
b*k tread to dust X
b*l m arry; begin X
to ripen
b*q strike X
b*r seek out, glean X
b*t oppress X
b*t II rouse X*
bs (denom.) X
become thin
bs’ examine X
b§r be lacking, x (or B)
diminish

1 Qoh. only. * Cogn. 3 ? ‫ ביי‬Hap. leg.

4 But Hebr. /z/; cf. Grundriß, i. 153. 5 = ‫ ב ע ש‬if Brockelmann is right.
APPENDIX: COMPARISON OF SYRIAC AND HEBREW VERBS 307
A B C D
Hebrew, Hebrew,
with close Hebrew) but not not
Verb Sense sense sense remote verb Hebrew
bq rot X
bq examine, decree X1
bqr examine X
br* create X
brh make clear X
brk kneel, swear X
brm consume X
br$ penetrate, X
reveal
brq flash X
Ml ripen, cook X

1 Unless in personal name.

 ABBREVIATIONS

ABR Australian Biblical Review

A JSL American Journal of Semitic Languages
A nO r Analecta Orientalia
AO Archiv Orientdlni
ASTI Annual of the Swedish Theological Institute
B A S OR Bulletin of the American Schools for Oriental Research
BDB Brown, D river, and Briggs, Hebrew Lexicon
BH3 Biblia Hebraica (Kittel), 3rd or later edition
B JR L Bulletin of the John Rylands Library
BO Bibliotheca Orientalis
BSO AS Bulletin of the School of Oriental and African Studies
BW ANT Beiträge zur W issenschaft des A lten u n d N euen T estam ents
BW AT Beiträge zur W issenschaft vom A lten T estam en t
BZA W Beiträge zur Zeitschrift fü r die Alttestamentliche Wissenschaft
CBQ Catholic Biblical Quarterly
CC Corpus Christianorum
CML Canaanite Myths and Legends (Driver)
D LZ Deutsche Literaturzeitung
ESA Epigraphic S outh A rabian
ET Expository Times
ETL Ephemerides Theologicae Lovanienses
GB Gesenius, Handwörterbuch über das Alte Testament (17th ed.
by Buhl, 1915)
GK Gesenius’ Hebrew Grammary ed. Kautzsch, 2nd English ed.
by A. E. Cowley (Oxford, 1910)
HTR Harvard Theological Review
H U CA Hebrew Union College Annual
IC C International Critical Com m entary
ID B Interpreter's Bible Dictionary
IE J Israel Exploration Journal
IJ A L International Journal of American Linguistics
JA O S Journal of the American Oriental Society
JB L Journal of Biblical Literature
JJS Journal of Jewish Studies
JN E S Journal of Near Eastern Studies
JP O S Journal of the Palestine Oriental Society
JQ R Jewish Quarterly Review
JR A S Journal of the Royal Asiatic Society
JSS Journal of Semitic Studies
JTS Journal of Theological Studies
KB L. K oehler and W . Baum gartner, Lexicon in Veteris Testa ‫־‬
menti Libros (Leiden, 1953)
 A B B R E V IA T IO N S 309
LSJ Liddell and Scott, Greek-English Lexicon (rev. ed. by H.
Stuart Jones and R. Mackenzie, Oxford, 1940)
MH Middle Hebrew, Mishnaic Hebrew
MT Massoretic Text
O BL Orientalia et Biblica Lovaniensia (ed. G. Ryckmans, Louvain.
19 5 7 )
OED Oxford English Dictionary
O LZ Orientalistische Literaturzeitung
Or Suec Orientalia Suecana
OTM S The Old Testament and Modern Study (ed. H. H. Rowley,
Oxford, 1951).
OTS Oudtestamentische Studiën
PEQ Palestine Exploration Fund Quarterly
PL Patrologia Latina (Migne)
RB Revue biblique
ScrH Scripta Hierosolymitana
SJT Scottish Journal of Theology
TLS The Times Literary Supplement
UH Ugaritic Handbook (by C. H. Gordon, AnOr xxv, Rome,
1 9 4 7 )
VD Verbum Domini
VT Vetus Testamentum
VTS Vetus Testamentum Supplements
WO Die Welt des Orients
ZA Zeitschrift fü r die Assyriologie
ZAW Zeitschrift fü r die Alttestamentliche Wissenschaft
ZD M G Zeitschrift der deutschen Morgenländischen Gesellschaft
ZNW Zeitschrift fü r die Neutestamentliche Wissenschaft
 BIBLIOGRAPHY

T h i s bibliography lists works m entioned in the text and some others

w hich are relevant for the general questions discussed. It does not,
however, provide individual listing for periodical articles in w hich philo­
logical treatm ents have appeared; for these the reader is referred to the
Index of Examples.

A istleitner, J., Wörterbuch der ugaritischen Sprache (Berlin, 1963).

B acher, W ., ‘D ie hebräische Sprachw issenschaft vom 10. bis zum 16.

Ja h rh u n d e rt’, in W inter, J., and W ünsche, A., Die jüdische Literatur
seit dem Abschluß des Kanons, ii. 135-235 (T rier, 1894).
------ , ‘D ie Anfänge der hebräischen G ram m atik’, Z D M G xlix (1895)
1-62 , 33 5 ‫ ־‬9 2 .
B arr, J., The Semantics of Biblical Language (Oxford, 1961).
-----, Biblical Words for Time (London, 196z).
-----, Old and New in Interpretation (London, 1966).
----- , ‘St. Jerome’s Appreciation of Hebrew’, B JR L xlix (1966-7) 281-
302.
----- , ‘St. Jerome and the Sounds of Hebrew’, J S S xii (1967) 1-36.
------ , ‘Vocalization and the Analysis of H ebrew am ong the ancient
T ranslato rs’, in the B aum gartner Festschrift (Leiden, 1967), pp. 1-11.
B arth, J., Etymologische Studien (Leipzig, 1893).
B auer, H., and L eander, P., Historische Grammatik der hebräischen
Sprache des alten Testaments (Halle, 1922).
B eeston, A. F. L., A Descriptive Grammar of Epigraphic South Arabian
(London, 1962).
B en -Jehuda, E., Thesaurus totius hebraitatis et veteris et recentioris
(Berlin, later Jerusalem, 1908-59).
B ergsträsser, G., Einführung in die semitischen Sprachen (Munich, 1928).
Bertholet Festschrift: Festschrift fü r Alfred Bertholet (Tü bingen, 1950).
B irkeland, H., The Language of Jesus (Oslo, 1954).
B lau, J., ‘Etymologische U ntersuchungen auf G ru n d des palästinischen
A rabisch’, V T v (1955) 3 3 7 4 4 ‫· ־‬
------ , ‘Ü b er hom onym e u n d angeblich hom onym e W urzeln’, V T vi (1956)
242-8.
B loomfield, L., Language (London, 1935).
B lumenthal, D. R., ‘A Play on Words in the Nineteenth Chapter of
Job’, F T xvi (1966) 497-501.
 BIBLIOGRAPHY 3”
B oling , R. G., * “Synonymous” Parallelism in the Psalms’, J S S v
(i960) 221-55.
B orgen, P., Bread from Heaven (Leiden, 1965).
B otterweck, G. J., Der Triliterismus im Semitischen (Bonner Biblische
Beiträ ge iii, Bonn, 1952).
B rockelmann, C., Grundriß der vergleichenden Grammatik der semitischen
Sprachen (2 vols., Berlin, 1908-13).
---- , Lexicon Syriacum (2nd edition, Halle, 1928).
---- , Hebräische Syntax (Neukirchen, 1956).
B rüll, A., Fremdsprachliche Redensarten und ausdrücklich als fremd­
sprachlich bezeichnete Wörter in den Talmuden und Midraschim (Leipzig,
1869).
B uck, C. D ., A Dictionary of Selected Synonyms in the Principal Indo-
European Languages (Chicago, 1949).
B uxtorf, J., Tiberias sive commentarius masorethicus triplex (Basiliae,
1620).
C appellus, L., Arcanum punctationis revelatum sive de punctorum vocalium
et accentuum apud Hebraeos vera et germana antiquitate (Leiden, 1624).
Casey, R. P. Festschrift: Biblical and Patristic Studies in memory of
R. P. Casey (ed. B irdsall, J. N., and T homson, R. W., Freiburg,
1963)·
Caskel, W., Die Bedeutung der Beduinen in der Geschichte der Araber
(Kö ln, 1953 )·
---- , ‘Zur Beduinisierung Arabiens*, Z D M G ciii (1953) *28-36.
---- , ‘The Bedouinization of Arabia’, American Anthropological Society
Memoir no. 76 (1954).
Castell, E., Lexicon Heptaglotton (London, 1669).
Chicago, Oriental Institute of the University, The Assyrian Dictionary,
(Chicago, from 1956).
Cross, F. M., and F reedman, D . N., Early Hebrew Orthography (New
Haven, 1952).
---- , The Ancient Library of Qurnran (London, 1958).
D ahood, M., ‘The Value of Ugaritic for Textual Criticism’, Biblica xl
(1959) 160-70.
---- , ‘Qoheleth and Northwest Semitic Philology’, Biblica xliii (1962)
3 49 - 6 5 ·
---- , ‘Northwest Semitic Philology and Job’, in McKenzie, J. L.,
The Bible in Current Catholic Thought (New York, 1962), pp. 55-74.
---- , ‘Hebrew-Ugaritic Lexicography’, Biblica xliv (1963) 289-303, xlv
(1964) 3 93 - 4 1 2 , xlvi (1965) 311-32, xlvii (1966) 403-19, xlviii (1967)
421-38.
---- , Proverbs and Northwest Semitic Philology (Rome, 1963).
312 B IB L IO G R A P H Y
D ahood, M., Psalms 1-50 (Anchor Bible, New York, 1966).
------ , Ugaritic-Hebrew Philology, (Rome, 1965).
D elitzsch, F riedrich, Prolegomena eines neuen hebräisch-aramäischen
Wörterbuches zum Alten Testament (Leipzig, 1886).
D horme, P., Le Livre de Job (Paris, 1926).
D illmann , A., Lexicon linguae aethiopicae (Leipzig, 1865).
------ , Ethiopie Grammar (E.T., London, 1907).
D ozy, R. P. A., Supplément aux dictionnaires arabes (2 vols., Leiden, 1881).
D river, G. R. : for a full bibliography, including much material relevant
to this study, see the Driver Festschrift (below), pp. 191-206.
------ , Problems of the Hebrew Verbal System (Edinburgh, 1936).
------ , ‘Hebrew Poetic Diction’, V T S i (1953) 26-39.
------ , Canaanite Myths and Legends (Edinburgh, 1956).
------ , Aramaic Documents of the Fifth Century B .C . (abridged and
revised edition, Oxford, 1957).
------ , The Judaean Scrolls (Oxford, 1965).
Driver Festschrift: Hebrew and Semitic Studies presented to G. R. Driver
(ed. D. WiNTON T homas and W. D. M cH ardy, Oxford, 1963).
D river, S. R., Notes on the Hebrew Text and Topography of the Books of
Samuel (2nd ed., Oxford, 1913).

E fros, I. L, Maimonides’ Treatise on Logic (Proceedings of the American

Academy of Jewish Research, viii, New York, 1938).
E hrlich, A. B., Randglossen zur hebräischen Bibel (7 vols., Leipzig,
1908-14).
E issfeldt, O., The Old Testament: an Introduction (Oxford, 1965).
Eissfeldt Festschrift: Von Ugarit nach Qumran (ed. H empel, J., and
R ost, L., BZAW lxxvii, Berlin, 1958).
E itan , I., A Contribution to Biblical Lexicography (New York, 1924).
E llenbogen, M ., Foreign Words in the Old Testament: their origin and
etymology (London, 1962).
E merton, J. A., ‘The Purpose of the Second Column of the Hexapla’,
J T S N.s. vii (1956) 79 - 8 7 ·
------ , ‘Did Jesus speak Hebrew?’, J T S N.s. xii (1961) 189-202.
E ppenstein , S., Übersicht über die hebr.-arabische Sprachvergleichung bei
den jüdischen Autoren des Mittelalters (Frankfurt a. M., 1905).

F ield, F., Origenis hexaplorum quae supersunt (2 vols., Oxford, 1875).

F itzmyer, J. A., The Genesis Apocryphon from Qumran (Rome, 1966).
F lashar, M., ‘Exegetische Studien zum Septuagintapsalter’, Z A W xxxii
(1912) 81-116, 161-89, 241-68.
F raenkel, S., Die aramäischen Fremdwörter im Arabischen (Leiden, 1886).
 B IB L IO G R A P H Y 313
F reedman, D . N ., ‘T h e Biblical Languages* in H yatt, The Bible in
Modern Scholarship, pp. 294-312.
F reytag, G . W ., Lexicon Arabico-Latinum (Halle, 1830-7).
F ück , J., Die arabischen Studien in Europa (Leipzig, 1955).

G emser, B., Sprüche Salomos (2nd ed., T übingen, 1963).

G erleman, G ., Studies in the Septuagint: I. The Book of Job (L und,
1946); I II . Proverbs (L und, 1956).
G leason, H . A., A n Introduction to Descriptive Linguistics (revised ed.,
N ew York, 1961).
G olius, J., Lexicon Arabico-Latinum (Leiden, 1653).
G onda, J., ‘T h e Etymologies in the A ncient Indian B rähm anas’, Lingua
V ( ! 955 61-85 (6 ‫ ־‬.
G ordis, R ., The Biblical Text in the Making (Philadelphia, 1937).
------ , Koheleth— the M an and his World (New York, 1955).
G ordon, C. H ., Ugaritic Handbook (A nO r xxv, Rome, 1947).
G oshen- G ottstein, M . H ., ‘L inguistic S tructure and T rad itio n in the
Q um ran Documents*, ScrH iv (1958) 101-37.
------ , Text and Language in Bible and Qumran (Jerusalem , i960).
------ , ‘T h e T ib erian Bible Text*, in Biblical and other Studies (ed.
A ltm ann, H arvard, 1963), pp. 79-122.
------ (ed.), The Book of Isaiah : Sample Edition with Introduction (Jeru ­
salem, 1965); review by Barr in J S S xii (1967) 113-22.
G ray, G. B., Isaiah i-xxvii (IC C , E dinburgh, 1912).
G ray, John , I & I I Kings (London, 1964).
—, The Legacy of Canaan (V T S v, L eiden, 1957; 2nd edition, L eiden,
!965)·
G reenberg, J. H ., ‘T h e P atterning of R oot M orphem es in Sem itic’,
Word vi (1950) 162-81.
G rossmann, R., and S achs, H ., Compendious Hebrew-English Dictionary
(revised edition by M . H . Segal, Tel-A viv, 1949).
G udschinsky, S. C., ‘T h e A B C ’s of Lexicostatistics (Glottochronology)*,
Word xii (1956) 175-210; reprinted in shorter form in H ym es, Language
in Culture and Society , pp. 612-22.
G uillaume, A., ‘A C ontribution to H ebrew Lexicography*, B S O A S xvi
(1954) ! - ! a .
------ , ‘T h e A rabic B ackground of the Book of Job*, H ooke Festschrift
(Edinburgh, 1963), pp. 106-27.
------ , ‘Paronom asia in the O ld T estam en t’, J S S ix (1964) 282-90.
------ , Hebrew and Arabic Lexicography ( = Abr-Nahrain i. 3 -3 5 ; ii. 5 -3 5 ;
iii. 1-10; iv. 1-18), Leiden, 1965.
314 BIBLIOGRAPHY
H alkin, A. S., ‘The Mediaeval Jewish Attitude toward Hebrew’, in
Biblical and Other Studies (ed. Altmann, Harvard, 1963), pp. 233-48.
H alperin, H., Rashi and the Christian Scholars (Pittsburgh, 1964).
H amori, A., ‘Paronomasia in Abu Tammä m’s Style’, J S S xii (1967) 83-90.
H arris, Z. S., The Development of the Canaanite Dialects (New Haven,
1939 )·
H a t c h , E ., and R e d p a t h , H . A ., A Concordance to the Septuagint (2 vols.,
O xford, 1897).
H aywood, J. A., Arabic Lexicography (Leiden, i960).
H ockett, C. F., A Course in Modern Linguistics (New York, 1958).
H oenigswald, H . M., Language Change and Linguistic Reconstruction
(Chicago, i960).
Hooke Festschrift: Promise and Fulfilment (ed. by F. F. B ruce,
Edinburgh, 1963).
H uffmon, H. B., Amorite Personal Names in the M ari Texts (Baltimore,
1965).
H yatt, J. P., ed., The Bible in Modern Scholarship (Nashville, 1965).
H ymes, D., ‘Rate of Morpheme Decay in Arabic*, IJ A L xxv (1959) 267-9.
---- , ed., Language in Culture and Society (New York, 1964).

I bn K haldün , Muqaddimah (ed. F. Rosenthal, 3 vols., New York, 1958).

Jacob, B., ‘Das hebrä ische Sprachgut im Christlich-Palä stinischen’,

Z A W xxii (1902) 83-113.
Jastro w, M., A Dictionary of the Targumim , the Talmud Babli and
Yerushalmi, and the Midrashic Literature (London, 1903).
Jeffery, A., The Foreign Vocabulary of the Qur'än (Baroda, 1938).
---- , ‘Hebrew Language’, in ID B ii. 553-60.
Jespersen, O., Language (London, 1922).

K ahle, P., The Cairo Geniza (2nd ed., Oxford, 1959).

K atz, P., ‘Septuagintal Studies in the Mid-Century’, in The Background
of the New Testament and its Eschatology: Studies in Honour of C. H.
Dodd (ed. W. D. Davies and D. Daube, Cambridge, 1964), pp. 176-
208.
K autzsch, E. F., Die Aramäismen im Alten Testament (Halle, 1902).
K oehler, L., ‘Vom hebrä ischen Lexikon’, O T S viii (1950) 137-55.
K opf, L., ‘Religious Influences on Mediaeval Arabic Philology*, Studia
Islamica v (1956) 3 3 5 9 ‫· ־‬
---- , ‘Das arabische Wö rterbuch als Hilfsmittel fü r die hebrä ische
Lexicographie’, V T vi (1956) 286-302.
---- , ‘The Treatment of Foreign Words in Mediaeval Arabic Lexicology’,
ScrH ix (1961) 191-205.
 BIBLIOGRAPHY 315
K oskinen, K., ‘Kompatibilitä t in den dreikonsonantigen hebrä ischen
Wurzeln*, ZDM G cxiv (1964) 16-58.
K raus, H.-J., Geschichte der historisch-kritischen Erforschung des Alten
Testaments (Neukirchen, 1956).
---- , Psalmen (Biblischer Kommentar, Neukirchen, 2 vols., 1961).
K rauss, S., ‘Talmudische Nachrichten über Arabien’, ZDM G lxx (1916)
32153 ‫·־‬
K utscher , e . y ., ‫הלשון ו ה ר ק ע הלשוני של מגיל ת ישעיהו ה של מה מ מגילו ת‬
‫( ים ה מ ל ח‬Jerusalem, 1 9 5 9 )·
---- , ‘Contemporary Studies in North-western Semitic’, J S S x (1965)
21-51·
---- , ‘Yemenite Hebrew and Ancient Pronunciation’, J S S xi (1966)
217 ‫ ־‬25 .

L ambton, A. K. S., Persian Vocabulary (Cambridge, 1954).

L andau, E., Die gegensinnigen Wärter im A lt- und Neuhebräischen (Berlin,
1896).
L ane , E. W., Arabic-English Lexicon (8 vols., London, 1863-93).
L ehmann , W. P., Historical Linguistics: an Introduction (New York,
1963)·
L eslau, W., Ethiopie and South Arabic Contributions to the Hebrew
Lexicon (Berkeley, 1958).
---- , Etymological Dictionary of Harari (Berkeley, 1963).
Levi della Vida Festschrift: Studi orientalistici in onore di Giorgio Levi
della Vida (2 vols., Rome, 1956).
L evy, J., Chaldäisches Wörterbuch über die Targumim und einen grossen
Theil des rabbinischen Schriftthums (3rd ed., Leipzig, n.d.— 1867?).
Lö w, I., ‘Aramä ische Fischnamen’, in Orientalische Studien Th . Nöldeke
. . . gewidmet (Giessen, 1906), pp. 549-70.

M aimon , M oses ben , Treatise on Logic (ed. 1 . 1 . Efros, American Academy

for Jewish Research, New York, 1938).
M enner, R J., ‘Conflict of homonyms in English’, Language xii (1936)
229-44.
M etzger, B. M., The Text of the New Testament (New York, 1964).
M irsky, A., ‘Biblical Variants in Medieval Hebrew Poetry’, Textus iii
(1963)159-63.
M ontgomery, J. A., Kings (ed. H. S. Gehman, ICC, Edinburgh, 1951).
M oore, G. F., Judges (ICC, Edinburgh, 1895).
M orag, S., The Vocalization Systems of Arabic , Hebrew and Aramaic
(The Hague, 1962).
------ , ‫( ה ע ב רי ת שבפי י הודי תימן‬Jerusalem, 1963) (for an English review,
see under Kutscher).
3i 6 BIBLIOGRAPHY
M oran, W . L ., *Gen. 49. 10 and its U se in Ezek. 21. 32*, Biblica xxxix
(1958) 405-25.
M oscati, S., The Semites in Ancient History (Cardiff, 1959).
------ , ed., A n Introduction to the Comparative Grammar of the Semitic
Languages (W iesbaden, 1964).
M owinckel, S .,Han SomKommer (Copenhagen, 1951); English, He That
Cometh (Oxford, 1956).

N eubauer, A. (ed.), Book of Hebrew Roots by M arw än b. Janäh (Oxford,

1875).
N ida, E. A., Toward a Science of Translating (Leiden, 1964).
N öldeke, T ., 'W ö rter m it G egensinn (A ddad)’ in his Neue Beiträge zur
semitischen Sprachwissenschaft (Strassburg, 1910), pp. 67-108.
N ötscher F estschrift: Alttestamentliche Studien (ed. Junker, H ., and
Botterweck, J., Bonner Biblische Beiträge, i, Bonn, 1950).
N oth, M ., Die israelitischen Personennamen (B W A N T , iii Folge, 10,
Stuttgart, 1928).
N yberg, H . S., Hilfsbuch des Pehlevi (Uppsala, 2 vols., 1928 and 1931).
------ , 'D as textkritische Problem des A lten T estam ents am H oseabuche
dem onstrie rt,, Z A W lii (1934) 241-54.
------ , Studien zum Hoseabuche (Uppsala, 1935).

P alache, J. L ., Semantic Notes on the Hebrew Lexicon (Leiden, 1959).

P ayne, D . F., 'A m biguity in O ld T estam ent Exegesis’, A S T I , fo rth ­
coming.
P eake, A. S., ed., The People and the Book (Oxford, 1925).
Peake’s Commentary on the Bible (2nd ed. by M . B lack and H . H .
R owley, L ondon, 1962).
P erles, F ., Analekten zur Textkritik des Alten Testaments (M unich, is t
series 1895, 2nd series 1922).
P ope, M . H ., El in the Ugaritic Texts ( V T S ii, Leiden, 1955).
------ , Job (Anchor Bible, N ew York, 1965).
------ , ‘M arginalia to M . D ahood’s Ugaritic-Hebrew Philology', J B L lxxxv
( 1966) 455 ‫ ־‬66 .

Q uine , W . V. O., From a Logical Point of View (N ew Y ork, 1963).

R abin , C., Ancient West-Arabian (London, 1951).

-----, Qumran Studies (Oxford, 1957).
------ , ‘T h e H istorical Background of Q um ran H ebrew ’, ScrH iv (1958)
144-61.
 B IB L IO G R A P H Y 317
------ , ‘T h e O rigin of the Subdivisions of Sem itic’, in Hebrew and Semitic
Studies presented to G. R. Driver (Oxford, 1963), pp. 104-15.
------ , ‘H ittite W ords in H ebrew ’, Orientalia xxxii (1963) pp. 113-39.
------ , ‫ מ לי ם בו ד דו ת‬, in Encyclopaedia Miqrait.
R ankin , O. S., Jewish Religious Polemic (E dinburgh, 1956).
Record and Revelation (Essays by m em bers of the Society for O ld T esta-
m ent Study, ed. R obinson , H . W ., O xford, 1938).
R eichelt, H ., Awestisches Elementarbuch (Heidelberg, 1909).
R ichter, E., ‘Ü b er H om onym ie’, K retschm er Festschrift (Vienna, 1926),
pp. 167-201.
R oberts, B. J., The Old Testament Text and Versions (Cardiff, 1951).
R obins , R. H ., General Linguistics: An Introductory Survey (London,
1964).
R obinson F estschrift: Studies in Old Testament Prophecy (ed. R owley,
H . H ., E dinburgh, 1950).
R obinson , H . W . (ed.), Record and Revelation (Oxford, 1938).
R osenthal, E. I. J., ‘M edieval Jewish Exegesis: Its C haracter and
Significance’, J S S ix (1964) 265-81.
R osenthal, F ., A History of Muslim Historiography (Leiden, 1952).
R osenzweig, A., ‘D ie A l-T ikri D eutungen’, in Festschrift zu 7 . Lewys
70. Geburtstag (Breslau, 1911).
R ossini, C. C., Chrestomathia arabica meridionalis epigraphica (Rome,
Ï9 3 0 ·
R owley, H . H ., ed., The Old Testament and Modern Study (Oxford,
1951 )·
R udolph , W ., Jeremia (T übingen, 1947).
------ , Esra und Nehemia (T übingen, 1949).
R uzicka, R., ‘U eber die Existenz des g im H ebräischen’, ZA xxi (1908)
293 - 340 .
------ , ‘L a Q uestion de l’existence du g dans les langues sém itiques’, AO
xxii (1954) 176-237.

Saadia: S iddur R. Saadja G aon, Kitàb Gämi As-Salawät wat-Tasâbïh

(ed. D avidson, I., A ssaf, S., and Joel, B. L, Jerusalem , 1941).
S aussure, F. de, Course in General Linguistics (E. T ., N ew York, 1959).
S chindler, B., and M armorstein, A., Occident and Orient . . .: Gaster
Anniversary Volume (London, 1936).
S chultens, A., ‘D issertatio theologica-philologica de utilitate linguae
arabicae in interpretanda sacra lingua’, in his Opera Minora (Leiden,
1769), pp. 487-510.
S chulthess, F ., Homonyme Wurzeln im Syrischen (Berlin, 1900) (cf.
review by N öldeke in ZDM G liv [1900] 1 5 2 6 4 ‫) ־‬.
------ , Lexicon Syropalaestinum (Berlin, 1903).
318 B IB L IO G R A P H Y
S eeligmann, I. L ., The Septuagint Version of Isaiah: a discussion of its
problems (Leiden, 1948).
S egal, J. B., The Hebrew Passover (London, 1963).
S egal, M . H ., ‘M ishnaic H ebrew and its Relation to Biblical H ebrew
and to A ram aic’, JQ R xx (1908) pp. 647-737.
------ , A Grammar of Mishnaic Hebrew (Oxford, 1927).
S1DD1Q1, A., Studien über die persischen Fremdwörter im klassischen
Arabisch (G öttingen, 1919).
S kinner, J., Genesis (IC C , E dinburgh, 2nd ed., 1930).
S mend, R., Die Weisheit des Jesus Sir ach (Berlin, 1906).
S oden, W . von, Akkadisches Handwörterbuch (W iesbaden, from 1959).
S perber, A., ‘H ebrew based upon G reek and L atin T ransliterations’,
HUG A xii-xiii (1937-8) 103-274.
S tenning , J. F., The Targum of Isaiah (Oxford, 1949).
S utcliffe, E. F., ‘St. Jerom e’s P ronunciation of H ebrew ’, Biblica xxix
(1948) 112-25.
S wete, H . B., An Introduction to the Old Testament in Greek (Cam bridge,
1900).

T hieme, P., ‘T h e Com parative M ethod for R econstruction in L inguistics’,

in Hym es, Language in Culture and Society, pp. 585-97.
T humb , A., Handbook of the Modern Greek Vernacular (Edinburgh, 1912).
T orczyner, H ., ‘A 1 T ik re’, Encyclopaedia Judaica, ii, pp. 74-87 (Berlin,
1928).
T orrey, C. C., The Second Isaiah (Edinburgh, 1928).
T ur -S in a i , N . H ., 2) ‫ ה ל שון ו ה ס פ ר‬vols., Jerusalem , 1954-5).
------ , The Book of Job (Jerusalem, 1957).

U llendorff, E., ‘T h e C ontribution of S outh Semitics to H ebrew

Lexicography’, V T vi (1956) 190-98.
------ , ‘T h e Knowledge of Languages in the O ld T estam en t’, BJRL xliv
(1962) 455 - 65 ·
U llmann , S., Semantics (Oxford, 1962).

V ermes, G ., Scripture and Tradition in Judaism: Haggadic Studies

(Studia Post-Biblica, iv, Leiden, 1961).
V öllers, K ., ‘D as D odekapropheton der AJexandriner’, Z A W iii (1883)
219-72.

W agner, M ., Die lexikalischen und grammatikalischen Aramaismen im

alttestamentlichen Hebräisch (BZAW xcvi, Berlin, 1966).
W allenstein, M ., Some Unpublished Piyyutim from the Cairo Genizah
(M anchester, 1956).
 BIBLIOGRAPHY 319
W echter, P ., Ibn Barün’s Arabie Works on Hebrew Grammar and Lexico­
graphy (Philadelphia, 1964).
W ehr, H ., A Dictionary of Modern Written Arabic (ed. J. M . Cowan,
Cornell, 1961).
W eil, G ., ‘\Addad*, in The Encyclopaedia of Islam (2nd edition, L o ndon
and Leiden, i960, i, pp. 184-6).
W eil , G . E., ‘L a nouvelle édition de la M assorah (B H K iv) et Phistoire
de la Massorah*, V T S ix (1963) 266-84.
------ , Élie Lévita (Leiden, 1963).
W einreich, U ., Languages in Contact (New York, 1953).
W eiss, A. von , Hauptprobleme der Zweisprachigkeit (H eidelberg, 1959).
W ellhausen, J., Der Text der Bücher Samuelis (G öttingen, 1871).
------ , Die Kleinen Propheten (Berlin, 1898).
W ernberg-M oller, P., The Manual of Discipline (Leiden, 1957).
W ild , S., Das Kitäb al-A in und die arabische Lexikographie (Wiesbaden,
1965)·
W right, G. E., (ed.), The Bible and the Ancient Near East (New York,
1961).
W ürthwein, E., The Text of the Old Testament (Oxford, 1957).
W utz, F. X ., ‘D ie T ranskriptionen von der S eptuaginta bis zu H iero ­
nymus* (BWA T , ii Reihe, 9, S tuttgart, 1933).
------ , Onomastica Sacra .(T exte u n d U ntersuchungen, xli. 1-2, Leipzig,
1 9 1 4 - 1 5 ) .

Y ahuda, A. S., The Language of the Pentateuch in its Relation to Egyptian

(London, 1933).
Y asin , I zz ad -D in al-, The Lexical Relation between Ugaritic and Arabic
(New York, 1952).

Z iff, P ., Semantic Analysis (Ithaca, i960).

Z immermann, F ., ‘Folk Etym ology of Biblical Names*, V T S xv (1966)
311-26.
 INDEX OF EXAMPLES

T h i s Index includes only a selection from the total number of philo-

logical treatments known to me. Almost all those referred to in the text
of this book are included, along with others which provide a representa-
tive collection.
Inform ation is presented in a very com pressed form. In order to
understand the argum ents advanced, readers should consult the source
itself. W here a suggestion has been applied to several different passages,
I generally cite only the m ore striking instances.
T h e order of presentation is as follows:
(1) Exam ples are listed in alphabetical order of roots, except in a few
cases, in w hich the familiar form is listed and a cross-reference
given.
(2) T h e H ebrew m eaning alleged by the suggestion.
(3) T h e chief language or languages used for cognate evidence, and, if
necessary, the form or sense therein. A question m ark is added
w here the existence or the sense of the cognate form cited seems
to be seriously doubtful in th at language.
(4) O ne or m ore of the biblical passages concerned.
(5) In parenthesis, relevant details such as parallelisms or versional
readings.
(6) A cross-reference, if the example has been discussed in the text of
this book.
(7) T h e source of the suggestion, not necessarily in its earliest form,
b u t in one w hich states the argum ent adequately and is reasonably
accessible (see p. 11).
T h is Index is n o t intended as a glossary of H ebrew ; it is for use in
conjunction w ith the discussion in this book. Inclusion of entries does
not im ply th at I consider the data or conclusions to be correct. I have not,
however, intentionally included anything w hich seems likely to mislead,
provided th at the argum ents of this book have been studied. O bvious
m isprints and m inor errors in the sources have been corrected.
1. ‫ א ב ח ה‬A. ‘slaughter’, Accad. ababu (?). Ezek. 21. 20 (cr^ayta).
D elitzsch, pp. 74 f. (b u t later abandoned?).
2. ‫ א ב ח ה‬B. ‘brightness’, Ar. 9ubbaha. Ezek. 21. 20. Reider, H U C A
ii (1925) 95·
3 · ‫‘ אב!־‬carry away’, Accad. abaku . Isa. 9. 17. D river, J T S xxix (1928)
390 .
4. ‫‘ א ב ל‬be d ry ’, Accad. Jer. 12. 4 (|| ‫)יב ש‬, etc. D river in Schindler,
P· 7 3 ·
5. ‫(‘ אגן‬moon) disc’. Ct. 7. 3. See text, p. 107. Reider, H U CA ii (1925) 94.
 IN D E X O F E X A M PL E S 321
6. ‫‘ א ד ב‬bind, oppress’, Accad. (?). I Sam. 2. 33. D river, jfT S xxiii
(1922) 70.
7. ‫ א ד ם‬A. ‘earth ’, as ‫ א ך מ ה‬. Prov. 30. 14 (|| ‫) א ל ץ‬, etc. D ahood, Proverbs,
PP· 57 f.
8. ‫ א ד ם‬B. ‘delightful’, E th. Prov. 12. 27, see text, pp. 28 f. Eitan,
p. 28. Cf. Ullendorff, F T vi (1956) 191 f.
9. ‫ א ד ם‬C. ‘skin’, Ar. H os. 11. 4, see text, p. 154. D river, JfT S xxxix
(1938) 161.
10. ‫‘ א ד ן‬father’, U gar. Prov. 27. 18. D ahood, Proverbs, p. 55.
11. ‫‘ א ה ב ה‬leather’, Ar. Hos. 11. 4, see text, pp. 144, 154. D river, C M L ,
P· 133; JfT S xxxix (1938) 161; H irschberg, V T xi (1961) 373.
12. ‫‘ ) הו ב( א ה בו הבל‬become guilty’, Μ Η ‫ חו ב‬, Ar. haba and cf. C D C
H os. 4. 18. Rabin, S c rH viii (1961) 389. Cf. no. 103.
13. ‫ א ה ל‬A. ‘have grazing rights’, Ar. G en. 13. 12, 18. Rabin, S c rH viii
(1961) 384.
14. ‫ א ה ל‬B. ‘count w orthy’, Ar. Job 25. 5. D river, A JfS L lii (1935-6)
161 f., following Ehrlich.
15. ‫‘ או ד‬be grievous, burdensom e’, Ar. ’ada. I I Sam. 13. 16 (M T * ‫ע ל‬
‫) א ו ד ת‬. Cf. Job 31. 23. See text, p. 266. D river, in N otscher, p. 48.
16. ‫‘ אי ד‬one b u rd en ed ’, participle of above. Prov. 17. 5 (|| ‫ ; ל ש‬άπολλνμένω).
D river, Biblica xxxii (1951) 182, cf. D ahood, Proverbs, p. 38.
17. ‫‘ מ א ד‬g rief’, noun from no. 15. Ps. 31. 12; poin t ‫ מ א ד‬. D river, JfT S
xxxii (1931) 256.
18. ‫ תארה‬A. ‘lam ent, plaint’, Ar. (?— 3-w -y ‘feel p ity ’ or taaw w aha ‘moan,
com plain’, Lane, pp. 130b, 129c?). Ps. 38. 10 (|| ‫) אנ ח תי‬, io . 17.
D river, J fT S xliii (1942) 153; Leveen, ibid, xlv (1944) 19.
19. ‫ ת או ה‬B. ‘ease, inactivity’, Ar. 3awa ‘tu rn aside, lodge’. Prov. 21. 25
etc. D ahood, Proverbs, p. 41 f., after Reider, no. 20.
20. ‫‘ או ה‬dw ell’, as above. N um . 34. 7. Reider, V T ii (1952) 113.
21. ‫ ת או ה‬C. ‘abode’, as above. G en. 49. 26. Reider, ibid.
22. ‫‘ או ל‬freem an’, Accad. awilum. I I K ings 24. 15 (Kethibh). D river, JfT S
XXXiv (1933) 33 f·; JfQR xxviii ( 1 9 3 7 1 1 6 (8‫ ־‬.
23. ‫‘ או ל‬rain, dew ’, Ar. 9ary. Isa. 18. 4, cf. ‫ צ ח‬, no. 268; Job 37. 11 and
medievals. Eitan, H U C A xii-xiii (1938) 65 f.
24. ‫‘ או ש‬gift’, U gar. 3sn, Ar. 3ws. Prov. 8. 21 (M T ‫ ;}יש‬see text, p. 181.
D ahood, Biblica, xxxiii (1952) 33 n.
25. ]!‫‘ א‬eq u ip m en t’ ESA 3dn. D eut. 23. 14. R abin, S c rH viii (1961) 387.
26. ‫ א ח ל‬A. ‘su b stitu te’, M H ‫ א ח ל א י ו ת‬. G en. 22. 13. Rabin, S c rH viii
(1961) 387.
27. ‫ א ח ל‬B. ‘w ith ’, n ot ‘after’, U gar. 3ahr || 'mn. Qoh. 12. 2. Scott, J fT S 1
(1949) 178; D ahood, Biblica xliii (1962) 363.
322 IN D E X O F E X A M P L E S
28. ‫‘ א מ ם‬walk forw ard’, Ar. 3mm. Isa. 60. 4, read ‫( תא^בה‬M T ‫) ת א מנ ה‬.
Eitan, H U C A xii-xiii (1938) 83 f.
29. ‫ א מ ר‬A. ‘lift up, be h igh’, cf. Accad. Am urru. Job 22. 28-9. T u r-S in ai,
30b, p. 349.
30. ‫ א מ ר‬B. ‘refuse, retract*, T alm . Aram . Job 3. 3. T u r-S in ai, Job, p. 50.
31. ‫ א מ ר‬C. ‘see’, Accad. Ps. 71. 10 (‘m y foes are on the look-out for m e’).
D ahood, Biblica xliv (1963) 295 f.; cf. ‘see’ as ‘original’ H ebrew
sense in A lbright, J A O S lxxiv (1954) 229 n. 47.
32. ‫ א ס ר‬A. ‘clinch, close’, not ‘begin’ (‘according to etym ology’). I Kings
20. 14. Gray, Kings, p. 377.
33. ‫ א ס ר‬B. ‘totality’, Ar. Hos. 10. 10. See text, p. 74. N yberg, Hosea,
p. 79.
34. ‫‘ א פ ת ה‬to the u tterm o st’, Accad. appuna ( = ‘m oreover’, von Soden,
p. 60). Ps. 88. 16. Delitzsch, pp. 135-7.
35. ‫‘ א פ ל‬be late’, Aram ., Accad. uppulu. Jer. 2. 31, here ‘(land) becom ing
parched late in season’ (|| ‫ ; מ ל ב ר‬Kexepoajfidvr]). See text, p. 266.
D river, JfT S xli (1940) 165 f.
36. ‫‘ א פ ס‬ocean’ as ultim ate sense, Accad. apsu, non-Sem itic. Pope, E ly
’ P· 72·
37. ‫‘ א ר א ל‬herald’, Accad. urullu. Isa. 33. 7 (M T ‫) א ך א ל ם‬. Oesterley,
Record and Revelation, p. 419.
38. ‫‘ א ר ב‬sexual desire’, A rab. 9irba. H os. 7. 6. N yberg, Hosea, pp. 52 f.
39. ‫‘ א רג‬smoke, sm ell’, Ar. 9arija. Job 7. 6. T ur-S in ai, jfob, pp. 137 f.
40. ‫‘ א ר ץ‬verm in’, Ar. 3arad. Job 12. 8, G en. 1. 26. T u r-S in ai, Joby
p. 210.
41. ‫‘ א של‬w arriors’, E SA 9sdy cf. Ar. 9asad ‘lion’. D eut. 33. 2 (M T
‫ ; א ש ד ת‬ayyeAoi). Beeston, jf T S N.S. ii (1951) 30 f.; M iller,
H T R lvii (1964) 241 ff.
42. ‫‘ א ש ר‬strong, strengthen’, A ram . Isa. 1. 17 etc. (cf. Isa. 47. 15 PorjOeia
for ‫) א ש ר‬. D river, W O i. 234, J fT S xxxviii (1937) 37, Biblica
xxxv (1954) 3°°·
43. ‫ אשש‬A. ‘be luxurious’, Ar. 9atta. Isa. 16. 7, see text, pp. 250, 254.
D river, in Eissfeldt Festschr., p. 43; B ertholet Festschr., p. 144.
44. ‫ אשש‬B. ‘be strong’, Accad. asisu (?). iQ p H ab. 6. 11. M ansoor, V T S
ix (1963) 320 f.
45. ‫‘ א תנן‬effigy’, Ar. tinny tanin (Lane, p. 318). M ic. 1. 7. See text, p. 145.
H alper, A JfS L xxxiv (1907-8) 366-8; D river, V T iv (1954) 242.
46. ‫‘ בג‬share, lo t’, scribal error for D , suggested by ‫ ב ג‬, Aram, from Iran.
baga-. Ezek. 25. 7. Perles, JfQR xviii (1906) 384. Cf. D river,
Aramaic Documents, p. 40 n.
47. ‫‘ ב ד‬diviner’, Accad. (M ari) baddum. Isa. 44. 25 (Vg. divinorum) etc.
D river, W O ii. 19.
 IN D E X O F E X A M P L E S 323
48. ‫‘ ) מ ך ( מ ב ך‬source’, U gar. mbk, npk. Prov. 8. 24 (M T ‫)! כ ב ד י‬. A lbright,
V T S iii (1955) 8. Job 28. 11, 38. 16. D river, C M L , p. 162 n.
49· ‫‘ בו ך‬squeeze, distress’ (not ‘w ander’), Ar. baka. Ex. 14. 3. Rabin,
S crH viii (1961) 388.
50. ‫‘ בו ץ‬ru n ahead’, Ar. basa. Ezek. 30.9 (read ‫ ; ב צי ם‬omevSovTes). D river,
Biblica xxxv (1954) 300.
51. ‫‘ ב ח ל‬jo in ’, U gar. p h r. Qoh. 9. 4, I Sam. 20. 30. D ahood, Biblica
xliii (1962) 361.
52. ‫‘ ב ט ח‬fall, lie’. Jer. 12. 5, Job 40. 23; see text, pp. 90 f. D river,
Robinson Festschr., pp. 59 f . ; E T L xxvi (1950) 341 f . ; K opf, V T
viii (1958) 165 ff.
53. ‫‘ בי ה‬rem em ber’, Ar. baha (?— see text, p. 166). Ps. 68. 5 (read as ‫) בי ה‬.
Guillaum e, J T S N.s. xiii (1962) 322 f.
54. ]‫‘ ? י‬region*, Ar. bin (not bayri). Isa. 44. 4 (‘field of grass’) etc. See
text, p. 165. G uillaum e, J T S N.s. xiii (1962) 109 ff.
55. ‫‘ ב ל י‬rottenness’, Syr. bly. Isa. 38. 17. D river, J T S xxxviii (1937) 47.
56. ‫‘ ב ל ע‬reach, slander’, Ar. big. Ps. 52. 6, Isa. 28. 7. D river, J T S xxxiii
(1932) 40; Z A W lii (1934) 52.
57. ‫‘ ב מ ה‬beast’, after relation of ‫ ב ה מ ה‬/‫ ב מ ה‬by A lbright, V T S iv (1957)
255 f. Isa. 2. 22 (‫‘ ב מ ה נח שב הו א‬m ust be considered a beast’).
D ahood, Biblica xliv (1963) 302.
58. ‫‘ ב צ ד‬price’, cf. A r. b'd ‘(far) b ehind’. Prov. 6. 26 (nfjLrjy pretium).
D river, V T iv ( 1 9 5 4 ) 2 4 3 ff·
59. ‫‘ ב עי‬beggar’, Syr. baaya. Job 30. 24 (1 ‫)? ע י‬. D river, A J S L lii (1935-
6) 164.
60. ‫‘ ב ע ל‬do, m ake’, U gar. Qoh. 8. 8 etc. See text, pp. 100 f. D ahood,
Biblica xliii (1962) 361 f.
61. ‫‘ ב צ ק ל ן‬fresh vegetation’, U gar. I I K ings 4. 42; see text, pp. 26, 294.
D river, C M L, p. 164 n.
62. ‫‘ ב ל ע‬go away’, Ar. baqaa (?; see text, p. 166). Isa. 7. 6 (‘make it go
over to ’, d7roarp€1jjo1JL€v). E itan, H U C A xii—xiii (1938) 58.
63. ‫‘ ב ק ש‬m agnify’, Accad. baqaiu (= rabu, synon. list). Prov. 29. 10.
D river, Biblica xxxii (1951) 194, J S S xii (1967) 108.
64. ‫‘ ב ל ל‬be cool’, Ar. barid. Isa. 32. 19. Reider, H U C A xxiv (1952-3)
88 f.
65. ‫‘ ב רי ת‬splendour’, Accad. bararu ‘shine’. Isa. 42. 6 (|| ‫) או ר‬. See text,
p. 144. T u r-S in ai, J P O S xiv (1936) 7.
66. ‫‘ ב ר ח‬w ound’, Ar. baraha ‘bruise’. Job 20. 24. D river, V T S iii
(1955) 81.
67. ‫‘ ב רי ח‬prim eval’, Ar. barih ‘past (of tim e)’, cf. Eg. b’h. U g. btn brhy
|| Is. 27. 1, Job 26. 13. A lbright, B A S O R lxxxiii (1941) 39 f., n. 5.
324 IN D E X OF E X A M P L E S
68. ‫‘ מ ב ש ר‬refu ter’, Syr. bsar. Isa. 41. 27, cf. Job 19. 26 (1 ‫) מ ב ש לי‬. See
text, pp. 119, 193. D river, N otscher Festschr., p. 46.
69. ‫‘ ב שן‬serpent’, Accad. basmu, U gar. btn. Ps. 68. 23. D river, C M L,
p. 164 n.
70. ‫‘ ג בו ל‬m ountain’, A r.jabal. Ps. 78. 54; see text, p. 248. D river, C M L ,
p. 146.
71. ‫‘ ג ד ל‬spin, weave’, cf. ‫ ג ך לי ם‬. Ps. 12. 4, 41. 10. D ahood, Psalms, p. 73.
72. ‫‘ ג ה ד‬act strenuously’, Ar. as in jihad. II K ings 4. 34 (tyAaaS etc., taken
as for ιγ ααδ). See text, p. 162. M ontgom ery, Kings, p. 372.
73. ‫‘ ג ה ה‬face’, Ar. wajh} jiha. Prov. 17. 22, cf. 15. 13 (‫) פני ם‬. Perles, J Q R
N.s. ii (1911-12) 102; D river, E T L xxvi (1950) 344.
74. ‫‘ גוי‬field, land’, Ar. jaw w. See text, p. 144. Zeph. 2. 14 (Tg. ‫חיו ת‬
‫ ; ב ר א‬τα θηρία τη ς γης). Eitan, ρ. 32 f.
75· ‫‘ גזל‬be great’, Ar. jazala. Job 24. 19. Guillaum e, Hooke Festschr.,
p. 116.
76. ‫ גי ל‬A. ‘w orship’, A v .ja la ‘move in circle,’ so ‘dance’. Ps. 2. 11. See
text, pp. 5, 175, 284. Reider, H U C A xxiv (1952-3) 98 f.
77. ‫גי ל‬/‫ גו ל‬B. ‘be affrighted’, Ar. wajila. Ps. 2. 11,18.16 (11| ‫)י ר א‬. See text,
pp. 5, 175, 284. D river, N otscher Festschr., p. 51.
78. ‫ גי ל‬C. ‘live’, ESA gyl ‘course of year’. Ps. 2. 11 etc. D ahood, Psalmsy
P· ! 3·
79. ‫‘ ג ר ם‬landing (of stairs)’, Ar. jaram a ‘cut off, com plete’ (?— see text,
pp. 165 f.). I I K ings 9. 13. G ray, Kings, p. 489.
80. ‫‘ ל ב א‬stren g th ’, U gar. dKat. D eut. 33. 25 (Tg. ‫ ; ת ו ק פ ך‬Ισχύς). So
Job 41. 14 (M T ‫) ך א ב ה‬. Cross, F T ii (1952) 162; Orlinsky, V T
vii (1957) 202 f.
81. ‫ ד ב ר‬A. ‘follow’, Ar. Cant. 5. 6. Eitan, pp. 34 f.
82. ‫ ד ב ל‬B. ‘be carried off’, A ram . Ps. 116. 10 (1 ‫) א ד ב ר‬. D river, J T S
xxxv (1934) 382.
83. ‫ ד ב ל‬C. ‘tu rn the back’, Ar. Job 19. 18 etc. Eitan, pp. 33 ff.; D river,
Z A W lii (1934) 55 f-
84. ‫ ד ב ל‬D . ‘overthrow ’, Accad. dabaru. II C hr. 22. 10 (|| ‫ ו ת א ב ד‬I I K ings
11. 1). W . T hom as, Record and Revelation, p. 393; D river, J T S
xxvii (1926) 159 f.
85. ‫‘ ד ג ל‬wait for’, Accad. Ps. 20. 6, see text, p. 262. D river, H T R xxix
(1936) 174 f. Cant. 6. 10 ‘brilliant’, G oitein, J S S x (1965) 220 f.
86. ‫‘ ד ה ה‬be astounded’, Ar. dhw, dhy ‘be stupefied’ (?). Isa. 44. 8; see
text, pp. 6 f., 166, 188, 231. D river, J T S xli (1940) 164.
87. ‫דון‬ a . ‘abound’, hiph. ‘provide abundance’, Ar. dana (w) (?— see
text, p. 166). Job 36. 31. D river, V T S iii (1955) 90. Cf. no. 90.
88. ‫ ד ץ‬B. ‘be close’, Ar. dana. G en. 6. 3. Rabin, S crH viii (1961) 388 f.
 IN D E X O F E X A M P L E S 325
89. ‫‘ לי‬soun d ’, Ar. dawiy. Job 39. 25 ( 1 ‫)ןנךי שו פ ל‬. Cf. no. 150. E ita
p. 36 f.
90. ]‫‘ די‬food’, Ar. zuw dn\ verb ‘feed’. Job 36. 31 (|| 17 .36 ,(‫ א כ ל‬.
Guillaum e, Hooke Festschr., pp. 122 f. Cf. no. 87.
91. ‫‘ ד כ ה‬acquit’, A ram. Prov. 26. 28; see text, p. 202. D river, J fR A S
(1948) 168 n.
92. ‫‘ ד ל ף‬oppress’ (not ‘d rip ’), Accad. daläpu. Ps. 119. 28. D river, JfT S
xxxi (1930) 277; G reenfield, H U C A xxix (1958) 207 ff.
93· ‫‘ ד ל ק‬p ursu e’ (not ‘b u rn ’), Ar. G en. 31. 36; see text, p. 144. Blau,
V T vi (1956) 246; K opf, V T viii (1958) 170 f.
94. ‫‘ ד מי‬half’, cf. Accad. mislu. Isa. 38. 10 (A' eV 7)[jLLcr€L} in dimidio, etc.).
See text, p. 259. D river, J T S xxxviii (1937) 46.
95. ‫‘ ד מ ע‬corn’ (not from ‫‘ ד מ ע‬tears’), Sam. dm ‘the b est’, Ar. dimdg
‘b rain ’. Exod. 22. 28. Blau, V T vi (1956) 246 f.
96. ‫‘ דן‬cask’, Accad., U gar. Ezek. 27. 19; see text, pp. 189 f. M illard,
J fS S vii (1962) 201 if.
97. ‫ ד ע ה‬A. ‘ask, desire’, Ar. d a d . Prov. 24. 14 etc.; see text, pp. 23 if.,
216. W inton T hom as, J fT S xxxviii (1937) 401.
98. ‫ ד ע ה‬B. ‘pull dow n’, Ar. dafd I I I (?— Lane, p. 883, b u t isolated).
Ezek. 19. 7, Ps. 74. 5; see text, p. 25. D river, JfBL lxviii
( 1 9 4 9 ) 5 7 f .

99· ‫ד י ו‬ ‘dom inion’, U gar. drkt. Prov. 8. 22. A lbright, JfBL lxiii (1944)
219 n .; V T S iii (1955) 7·
100. ‫· ד ר ך‬persecute’, Ar. Lam . 3. 11 (1 ‫ ; ד ר כ י‬KareSico^ev). D river, Z A W
lii (1934) 308.
101. ‫‘ ד שן‬be hidden, shrouded’, Ar. datara (but ‫‘ ד שן‬fat’ = Ar. dasam).
Ps. 22. 30; see text, p. 296 n. D river, H T R xxix (1936) 176 f.
102. ‫‘ ה ב אי ם‬flowers’, Syr. habtd. Isa. 27. 6, not from ‫‘ ברא‬com e’. T u r-
Sinai, Jfoby p. 196 n.
103. ‫ ה ב ה ב‬,‫‘ ה ב‬burn, b u rn w ith passion’, E th., A r., etc. Prov. 30. 1
Hos. 8. 13,4. 18. See text, pp. 233 f. G lück, V T xiv (1964) 367 fr.
Cf. nos. 12, 250.
104. ‫‘ ה ב ר‬w orship’, S. Sem. kbr (spirantization). Isa. 47. 13; see text,
p. 119. Ullendorff, J fS S vii (1962) 339 f.
105. ‫‘ ה ד ר ה‬divine appearance’, U gar. hdrt || him. Ps. 29. 2. Cross, B A S O R
cxvii (1950) 21.
‫ הרב‬see no. 12.
106. ‫‘ הרה‬w o rd ’, U gar. hw ty Accad. aivdtum. Job 6. 30. Pope, Jfob, p. 55.
107. ‫‘ הי ה‬fall’. I Sam. 1. 18 etc.; cf. ‫הרא‬, Job 37. 6. See text, pp. 250 f.
D river, W O i. 414.
108. ‫‘ ה מה‬behold’, Ugar. hm . Ps. 48. 6. Dahood, C B Q xvi (1954) 16.
326 IN D E X O F E X A M P L E S
109. ‫‘ ה רון‬sexual desire*, U gar. hrr. G en. 3. 16; see text, p. 286. Rabin,
S crH viii (1961) 390.
n o . ‫‘ ה רן‬pledge’, Ar. rahana, cf. M H . Job 34. 32 (1 T p ) . See text,
p. 166. T u r-S in ai, Job, pp. 486 f.
i n . ‫‘ ה ר‬stony tract*, Ar. harrat. Amos 1. 13. Reider, V T iv (1954) 279.
112. ‫ ז בו ל‬A. ‘princely estate’, U gar. Ps. 49. 15 (Sofa). D river, C M L ,
p. 149.
113. ‫ ז בו ל‬B. ‘throne-platform*, U gar. A lbright, Robinson F estsch r.,p . 16.
114. ‫( ז ר‬V^T) ‘enem y’, Accad. Ezek. 7. 21; Prov. 6. 1 ixOpco, see text,
p p . 256 f. D river, Biblica xxxv (1954) 148 f.
115. ‫‘ ז מ ר‬protect*, Ar. damara. Exod. 15. 2; see text, pp. 29 f., 182.
W inton T hom as, E T xlviii (1936-7) 478; Record and Revela-
tion, pp. 395 f.
116. ‫‘ זנה‬be angry’, Accad. zenu. Jud. 19. 2; see text, p. 286. D river,
E T L xxvi (1950) 348.
117. ‫‘ ח ב ב‬pure ones’, Accad. ebebum ‘be pure*. D eut. 33. 3, cf. no. 41.
M iller, H T R lvii (1964) 241 if.
118. ‫ ח ב ר‬a . ‘connect, inform*, Ar. habar (but ‘com panion’ = hbr).
Job 16. 4. R o sen th a l,Historiography, p. 10 n. 2. Cf. no. 119.
119. ‫ ח ב ר‬B. ‘heap u p ’, U gar. bth b r ‘store-house*. Job 16. 4. T u r-S in ai,
Job, pp. 262 f. Cf. no. 118.
‫ חגר‬see no. 140.
120. ‫‘ ח ד ל‬be p lu m p ’, Ar. hadula. I Sam. 2. 5. W inton T hom as, V T S
iv ( 1957) 14 f·
121. pin ‘gather*, Accad. hiaqum ‘mingle*. Prov. 8. 29. D river, Biblica
xxxii (1951) 178.
122. ‫‘ חו ש‬worry*, Accad. etc. Qoh. 2. 25. Ellerm aier, Z A W lxxv (1963)
197-217.
123. ‫ חזו ת‬,‫‘ חז ה‬agreem ent’, ESA hdyt. Isa. 28. 15, 18; see text, p. 23
D river, J T S xxxviii (1937) 44. Also verb, cf. A r. h d \ Job 8. 17,
J T S xl (1939) 391*
124. ‫‘ חז ץ‬magistrate*, Accad. hazannu. Prov. 29. 18; see text, pp. 283 f.
D river, W O i. 235.
125. ‫ ח ט א ת‬A. ‘step, walk’, Ar. hatwa. Job 14. 16; see text, pp. 142 n., 144.
E itan, pp. 38-42.
126. ‫ ח ט א ת‬B. ‘p en ury’, E th. h a tia t. Prov. 10. 16. See text, pp. 144, 166.
W inton T hom as, J T S N.s. xv (1964) 295 f.
127. ‫ ח ל ה‬A. ‘be sorry, think*, E th. halaya. I Sam. 22. 8 (7tovwv); Jer. 5. 3.
D river, J T S xxix (1928) 392, J Q R xxviii (1937-8) 101. Eitan,
H U G A xii-xiii (1937-8) 82 f.
128. ‫ ח ל ה‬B. ‘be alone*, Ar. hala. Qoh. 5. 12 (‘a singular evil*). Eitan,
ibid., p. 62.
 IN D E X O F E X A M P L E S 327
129. ‫ ח ל ה‬C. ‘adorn’, Ar. hala. Prov. 3. 35 (M T 18 .14 ,(‫( מ ח ל ו‬M T
‫) נ ח לו‬. D river, Biblica xxxii (1951) 177.
130. ‫ חי ל‬,‫‘ ח ל ל‬be troubled’. Ps. 55. 5, 109. 22, alternative forms, differen
sem antic fields, both L X X rapacraoj; cf. text, p. 252. K addary,
V T xiii (1963) 486-9.
131. ‫‘ ח ל ם‬be healthy’, A ram , him . Ps. 126. 1, so T g . Strugnell, J fT S
N.s. vii (1956) 2 3 9 4 3 ‫· ־‬
132. ‫ ח ל ק‬A. ‘create’, Ar. hlq. A m . 7. 4; see text, pp. 69, 26 0 f. M ont-
gomery, J B L xxiii (1904) 95 f . ; D river, V T S iii (1955) 91.
133· ‫ ח ל ק‬B. ‘die, p erish ’, U gar. hlq. Ps. 5. 10, 12. 4. D ahood, Psalms,
PP· 35, 73·
134. ‫חל ש‬ ‘reap ’, Pal. Ar. halasa. Isa. 14. 12; see text, p. 276. Eitan,
pp. 42-6.
135. ‫‘ מ חנ ה‬protective siege-w ork’ (not ‘cam p’), cf. M H ‫ חנו ת‬. Ezek. 4. 2.
Driver, Biblica xxxv (1954) 148.
136. ‫‘ ח פ ץ‬make straight, stretch ’. Job 40.17, Ps. 37. 23. Perles, Analekten,
p. 76.
137· ‫‘ ח ק ר‬despise’, Ar. haqara. Prov. 28. 11 (Karayvdjaera6, bsr), 25. 27;
see text, p. 258. W inton T hom as, J fT S xxxviii (1937) 402 f.,
after Perles; E itan, p. 7 n.
138. ‫‘ ח ר ב‬deceitfulness, vain speech’, Ar. haraba (Dozy, i. 356). Ps. 59. 8;
Jer. 25. 9 (ovclS lgijlos). See text, p. 116. D river,y T S x x x iii (1932)
42 f.
139. ‫‘ ח ר ב ה‬palace’, E SA rnhrb ‘castle’, Ar. mihrab ‘pavilion’ (contrast root
hrb ‘ru in ’). Job 3.14 etc. D aiches, J Q R xx (1908) 637 ff.; D river,
E T L xxvi (1950) 349; b u t cf. Pope, Job, p. 31.
140. ‫‘ ) חג ר( ח רג‬fear’, A ram . ‫ ח רג ת א‬, Ps. 18. 46. B ut versions ‘lim p’, read
‫וי חג רו‬, as I I Sam. 22. 46 = A ram ., Syr. ‘lim p’. G unkel, Psalmen,
p. 73; D river, H T R xxix (1936) 174.
141. ‫‘ ח ר ה‬be angry’, Ar. wahar (so not from ‘b u m ’). Rabin, S c rH viii
(1961) 390 f.
142. ‫‘ ח ר ם‬cut off’, Accad. haramu. Isa. 11. 15; see text, p. 119. D river,
J T S xxxii (1931) 251.
143. ‫‘ ח ת א‬sh atter’, U gar. h t \ H ab. 3. 6 -7 ; see text, p. 190. A lbright,
B A S O R lxxxii (1941) 47 n. 27; Robinson Festschr,, pp. 11 f., 15.
144. ‫‘ ט א ט א‬po u n d (m ud floor); annihilate’, Ar. w a tia . Isa. 14. 23; see
text, p. 56. K opf, V T viii (1958) 174 f.
145. ‫‘ ט ב ב‬know, announce’, Syr., A r., E th. Prov. 15. 2; see text, pp. 171 f.
Driver, Biblica xxxii (1951) 181.
146. ‫ טו ב‬A. ‘speech’. H os. 14. 3 etc.; see text, pp. 16 f. G ordis, V T v
(!955) 88-90.
328 IN D E X OF E X A M P L E S
147. ‫ טו ב‬B. ‘very m u ch ‫״‬, T alm . A ram. M ic. 1. 12 etc. See text, p. 17.
G ordis, JfT S xxxv (1934) 186 f.
148. ‫ טו ב‬C. ‘rain’. Ps. 4. 7 etc. D ahood, Psalms, p. 25.
149. ‫‘ יגע‬to rtu re’, Ar. waji a. I I Sam. 5. 8 (piel perf. cons.). G insberg,
Z A W li (!933) 308.
150. ‫‘ י ד ה‬voice’, Ar. dawiy w ith m etathesis, cf. no. 89. H ab. 3. 10.
G insberg, Z A W li (1933) 308.
151. ‫ י ד ע‬A. ‘make subm issive’, Ar. wada a. Jud. 8. 16 etc.; see text,
pp. 19 if. W inton T hom as, JfT S xxxv (1934) 298-306, xxxviii
(1937) 404 f.
152. ‫ י ד ע‬B. ‘care for, keep in m in d ’, Ar. wadi a. Exod. 2. 25; see text,
p. 22. W inton T hom as, J fT S xxxv (1934)300!., xlix (1948) 143 f.
153. ‫ י ד ע‬C. ‘be reconciled’, Ar. wadi a I I I . Am. 3. 3; see text, pp. 19 if.
W inton T hom as, JfT S N.s. vii (1956) 69.
154. ‫ י ד ע‬D . ‘dism iss’, Ar. wada a I II . I Sam. 21. 3; see text, pp. 21 f.
W inton T hom as, J fT S xxxv (1934) 298 if.; E itan, pp. 48 if.
155. ‫ י ד ע‬E. ‘w rap u p ’, Ar. wada a. H os. 7. 9; see text, p. 22. H irschberg,
V T xi (1961) 379.
156. ‫ י ד ע‬F. ‘sw eat’, dialect, cf. ‫יז ע‬. Isa. 53. 11; see text, p. 23. N öldeke,
Neue Beiträge, p. 194 f . ; D ahood, in M ackenzie, p. 72.
157. ‫‘ מ ד ע‬m essenger’, from ‫ ; י ד ע‬U gar. m nd'. Qoh. 10. 20; see text, p. 23.
D ahood, Biblica xxxix (1958) 312.
158. ‫‘ י סו ד‬tail-end, fundam ent, thigh’, Accad. isdu, same m eaning-range
b u t not cognate. H ab. 3. 13. A lbright, Robinson Festschr., p. 17.
159. ‫‘ י ס ר‬make strong’ (piel), Aramaism, A ram. ‫ א ש ר‬. Hos. 7. 15, Job 4. 3.
D river, J fT S xxxvi (1935) 295·
160. ‫‘ תו ע פ ת‬im petuosity’, Ar. wgf. N um . 23. 22. A lbright, JfBL lxiii
(1944) 215.
161. ‫ או פי ר‬,‫‘ י פ ר‬esteem ’, cf. Ar. wafara ‘abound’, U gar. ypr. Isa. 13.
see text, pp. 286 f. Eitan, H TJCA xii-xiii (1937-8) 61.
162. ‫‘ י צ א‬be clean, p u re’, Ar. w adua. Prov. 25. 4 (KadapLcrdrjaerai). See
text, p. 120. D river, Biblica xxxii (1951) 190.
163. ‫‘ י צ ד‬fasten’, Ar. wasada. Isa. 44. 12 (M T ‫) ל צ ך הו‬. Eitan, H U C A
xii-xiii (1938) 78.
164. ‫‘ י צ ה‬give last injunctions’, Ar. wasä I I ‘make a w ill’. II K ings 20. 1.
G ray, Kings, pp. 633 f.
165. ‫ י ק ה‬,‫‘ ת קו ה‬godly fear’, not ‘hope’; Ar. taqwä, root wqy. Job 17.
second ‫( ת קו ה‬Vg. patientia). G uillaum e, H ooke Festschr., p. 113.
166. ‫‘ י ק ה‬be insolent’, Ar. waqiha. N um . 16. 1, Job 15. 12, Prov. 6. 25;
see text, pp. 17 if., 271. Eitan, pp. 20 if.; D river, W O i. 235, 415.
167. ‫ י ק ר‬A. ‘be heavy, still; rest’, Ar. waqara. Prov. 25. 17. W inton
T hom as, JfT S xxxviii (1937) 402.
 IN D E X O F E X A M P L E S 329
168. ‫ י ק ל‬B. ‘split, hew ’, Ar. waqara. I K ings 5. 31. Gray, Kings, p. 148.
‫ י רי ב‬see no. 293.
169. ‫ ת ר ע ת‬,‫‘ י ר ע‬m ajesty’, lit. ‘terror-producin g ’, Ar. wari'a ‘be tim
ESA hwr ‘cause to fear’. N um . 23. 21 (1 ‫ ; ו ת ר ע ת‬eVSo£a).
A lbright, J B L lxiii (1944) 215.
170. ‫‘ י שח‬semen virile’, Ar. wasiha ‘be filthy’. Mic. 6. 14. Pope, J B L
lxxxiii (1964) 270, after Ehrlich, v. 288.
171. ‫‘ כ ב ת‬behum bled’, A r.kabata. Jer. 17. 13 (M T ‫) בו ש || ;י כ ת בו‬. Driver,
jfQ R xxviii ( 1937114 (8‫· ־‬
172. ‫‘ כ ד ם‬hold fast’, Ar. kadama (?— see text, p. 166). Ezek. 27. 32.
Guillaum e, J T S N.s. xiii (1962) 324 f.
173· ‫‘ ם י ון‬vagina’, Ar. kain. Am. 5. 26, see text, p. 277. H irschberg, V T xi
(1961) 37s f.
!74· ‫‘ כ כ ר‬gorge’, Yem enite kurkur. Rabin, W est-Arabian, p. 28. See text,
p. 100.
175. ‫‘ כ ל ב‬h u n te r’, as Syr. kallaba. Seen from transls. of Ps. 22. 17 (Aq.
Sym. Orjparaiy L at. venatores). D river, H T R xxix (1936) 176.
176. ‫‘ כ ל ם‬speak’, Ar. kallama. Judges 18. 7; M ic. 2. 6; see text, pp. 14 f.
Reider, V T iv (1954) 280.
177. ‫‘ כ ס ף‬be broken, depressed’, Accad. kasapu etc. Zeph. 2. 1. Driver,
A nO r xii (1935) 62 f . ; J T S xxxvi (1935) 404; V T iv (1954) 242;
cf. no. 178.
178. ‫‘ כ ס ף‬food’. Isa. 55. 1; see text, p. 153. References as for no. 177.
179. ‫‘ כ פי‬equal, like’, Ar. k ifa . Job 33. 6. T ur-S inai, Job, p. 465.
180. ‫‘ כ ר ה‬feast’, loan from Accad. karu. II K ings 6. 23; see text, p. 102.
Rabin, S c rH viii (1961) 399. Cf. no. 288.
181. ‫‘ פו ש רו ת‬song, m usic’, U gar. k tr t. Ps. 68. 7. G insberg, B A S O R lxxii
(1938) 13.
182. ‫‘ כ ת ף‬w eapon’, U gar. ktp. I Sam. 17. 6, D eut. 33. 12. O ’Callaghan,
Orientalia xxi (1952) 42 f., cites b u t does not accept for H eb r.;
D river, CML> p. 145.
183. ‫‘ ל א‬victor’, Accad. le u ‘overcom e’. Ps. 7. 13 (M T ‫) ל א‬. D ahood,
Psalms, pp. 46, xxxvi.
184. ‫‘ מ ל א ך‬priest, priestly m essenger’, cf. Phoen. parallel passages. Qoh.
5. 5, M ai. 2. 7 (|| ]‫)? כ ה‬. D ahood, Biblica xxxiii (1952) 207.
185. ‫‘ ל א ם‬ru ler’, U gar., Accad. limu. Ps. 7. 8, 148. 11; see text, pp. 133,
172, 254 f. D river, C M L , p. 158; E T L xxvi (1950) 346; Gray,
Legacy, p. 197.
186. ‫‘ ל ב ה‬w rath ’, Accad., A ram . Ezek. 16. 30. D river, J T S xxxii (1931)
T ’ 366.
187. ‫‘ ל ב ד‬stick together’, T alm . H ebr. Job 38. 30; see text, p. 234. T u r ‫״‬
Sinai, Job, pp. 530 f. O r ‘be m atted, felted’, as of cloth?
330 IN D E X OF E X A M P L E S
188. ‫ ל ב ש‬A. ‘linger’, Ar. labita. Isa. 14.19. Eitan, H U C A xn -xin (1 9 3 & ) 63.
189. # ‫ ל ב‬B. ‘draw near’, Ar. labisa ‘join closely\ Judges 6. 34 etc. Reider,
y j S iii (1952) 79·
190. ‫‘ לג‬waves’, Ar. lujj ‘depth of sea’. Job 12. 23 (M T ‫) ל ג ר ם‬. T u r-S in ai,
yob, p. 219.
191. ‫ ל ה ק ה‬A. ‘elder-com pany’, E th. Ihq. I Sam. 19. 20, see text, pp. 25 f.,
231 f., 267 n ., 270 f. D river, y T S xxix (1928) 394; Ullendorff,
V T vi (1956) 194; W inton T hom as, y T S xlii (1941) 154 (Prov.
30· 17)·
192. ‫ ל ה ק ה‬B. ‘com pany’, Ar. 5ilhäq ‘affiliation’, w ith criticism of no. 191.
Greenfield, H U C A xxix (1958) 212 if.
193. ‫‘ ל מ ךי ם‬strings’, M H (?). Isa. 8. 16. T ur-S in ai, yob, p. 240.
194. ‫‘ מ ק‬pow erful’, Ar. mäjin ‘bold’. Ps. 47. 10; see text, pp. 241, 251.
D river, y T S xxxiii (1932) 44. Cf. A lbright, V T S iii (195s) 10
(‘beggar’).
195· ‫‘ מ ד ר‬w et clay’, E th. mddr ‘earth’. Job 37. 17 (M T ‫ ;) מ ד ר ו ם‬see text,
p. 190. T u r-S inai, yob, p. 515.
196. ‫‘ מ הי ל‬skilled’, E th. mähärä ‘teach’. Ps. 45. 2, Isa. 16. 5. See text,
p. 295. Ullendorff, V T vi (1956) 195.
197. ‫‘ מ ח ץ‬d ip ’, Accad. mahäsu. Ps. 68. 24; see text, p. 192. D elitzsch,
Proleg., pp. 69 if. Cf. no. 205.
198. ‫‘ מ טו‬war, cam paign’, ESA m tw . H ab. 3. 9, 14. A lbright, Robinson
Festschr., p. 15.
199. ‫ מ ו ד ד‬,‫‘ מי ד‬be shaken’, Ar. mäda. H ab. 3. 6 (60‫־‬aXevdr]); see tex
p. 252. D river, Z A W Iii (1934) 54 f.
200. ‫‘ מ כ ר‬counsel’, E th. G en. 49. 5; see text, pp. 57, 270. W inton
T hom as, y T S x x x v ii (1936) 388 f.; Ullendorff, V T vi (1956) 194.
201. ‫‘ מ ל ח‬be dark’, Ar. maltha ‘be grey’. Isa. 51. 6; see text, p. 252.
D river, E T L xxvi (1950) 349 f.
202. ‫ מ ל ך‬A .‘advise’, A ram . Hos. 8 .4 ; see text, pp. 188 f. D river, N ötscher
Festschr., p. 50. Cf. no. 306.
203. ‫ מ ל ך‬B . ‘take possession’, Ar. N eh. 5. 7; see text, p. 188. K opf, V T
ix (1959) 261 f.
204. ‫‘ מ ע ה‬m u ltitu d e’, Ar. m a iy y a ‘com pany’. Isa. 48. 19. G ray, Legacy,
p. 192.
205. ‫‘ מ צ ח‬tread ’, U gar. msh. Ps. 68. 24 (1 ‫ ; ת מ צ ח‬M T ‫ ת מ ח ץ‬is graphic
error caused by ‫ י מ ח ץ‬in v. 22). Reider, H U C A xxiv (1952-3)
101. Cf. no. 197.
206. ‫ מ שח‬A. ‘m easure, extend’, M H . Ezek. 28. 14. D river, y T S xli
(1940) 169 f.
207. ‫ מ שח‬B. ‘m ar’, Ar. masahat masth *ugly’. Isa. 52.14 Q um ran; see text,
p. 285. G uillaum e, y B L lxxvi (1957) 41 f.
 IN D E X O F E X A M P L E S 331
208. ‫‘ מ תן‬stren g th ’, M H and Ar. matuna. Qoh. 7. 7. (M T ‫ מ תנ ה‬, εύτονίa,
robur); see text, p. 91 n. D river, V T iv (1954) 229 f.
209. ‫‘ מ ל‬blow ’, A ram. Job 10. 17. T ur-S inai, Job, pp. 181 f.
210. ‫‘ נ ד‬fire’, E th. nadda. Isa. 17. 11. Eitan, H U C A xii-xiii (1938) 65.
Cf. no. 212.
211. ‫‘ נ ח ל‬sift’, Ar. nahala, Accad., Syr. Ps. 82. 8. D river, H T R xxix
(1936) 187.
212. ‫‘ נ ח ל ה‬destruction’, E th. nahla ‘collapse’. Isa. 17.11. Eitan, cf. no. 210.
213. ‫‘ נטו ש‬clash (in battle)’, Ar. watasa, watts. Prov. 17. 14 (μάχη),
I Sam. 4. 2; see text, p. 257. D river, Biblica xxxii (1951) 182.
214. ‫‘ נ כ ר‬acquire, sell’, U gar. nkr. I Sam. 23. 7 (πεπρακεν); see text,
p. 267 n. Hos. 3. 2. G ray, Legacy, p. 190.
215. ‫‘ נ מ ה‬bring tidings’, Ar. nam a. Isa. 41. 27; see text, pp. 182, 193.
Guillaume, J B L lxxvi (1957) 40 f.
216. ‫‘ גסג‬forge’, Ar. nasaja. M ic. 2. 6; see text, p. 15. Reider, V T iv (1954)
280.
217. ‫‘ נ ע ל‬sparrow ’, Ar. nugar. Job 40. 29. W inton T hom as, V T xiv (1964)
1 1 4 ff.; cf. G ordis, ibid., 491-4.
218. ‫‘ נצה‬be joined’, Ar. nasa. Lam. 4. 15 (άνήφθησαν) ; see text, pp. 262 f.
D river, Z A W lii (1934) 308; W inton T hom as, Record and
Revelation, p. 396.
219. ‫‘ גשא‬u tter, pronounce’, Accad. nasu ‘take oath ’. Isa. 42. 2, 11, and
‫ מ ש א‬. T sevat, H U C A xxix (1958) 119.
220. ‫ נשר‬,‫‘ גשר‬herald’, Ar. nassar. H os. 8. 1, Job 39. 25; see text, pp. 26 f
T u r-S in ai, Job, p. 551.
221. ‫‘ נ ת ר‬tear; hop, leap’. Accad. nutturu. H ab. 3.6 etc. D river, Robinson
Festschr., pp. 70 if., V T iv (1954) 241; and full discussion in
E m erton, Z A W lxxvi (1964) 191 if. See text, p. 290.
222. ‫‘ ס ד ר‬rays of lig ht’, Ar. sadira. Job 10. 22 (φέγγος); see text, p. 242.
Driver, V T S iii (1955) 76 f.
223. ‫‘ סו ד‬chieftaincy’, Ar. sud. Ps. 25. 14; see text, p. 251. D river, J B L
lv (1936) 102, E T L xxvi ( 1 9 5 0 ) 345·
224. ‫‘ ס כו ת‬pole’ (cultic symbol), T alm . ‫ ס כ ת א‬. Am. 5. 26; see text, p. 277.
H irschberg, V T xi (1961) 375.
225· p o ‘care for’, not Accad. saknu ‘governor’ b u t rel. Ar. zakina ‘be
fam iliar’, M H ‫‘ ס כנ ה‬danger’. I K ings 1. 4, Qoh. 10. 9. Rabin,
S c rH viii (1961) 395.
226. ‫‘ ס פ סג‬glaze’, U gar., H ittite. Prov. 26. 23 (M T ‫ ;) כ ס ף סיגים‬see text,
pp. 219 f. G insberg, B A S O R xcviii (1945) 21.
227. ‫‘ מ ס פ ר‬lim it, boundary’, A r., A ram . ‫ ק פ ר‬. D eut. 32. 8 (|| ‫) ג ב ל ת‬.
Zim m erm ann, J Q R N.s. xxix (1938-9) 241 f.
332 IN D E X OF E X A M PL ES
228. ‫‘ צ ד‬tim e’. Kaipos at Jer. 11. 14, Isa. 30. 8, 64. 8, Ezek. 22. 4, 30; see
text, p. 247. D river, W O i. 412; C M L, p. 140.
229. ‫ ע ד ה‬A. ‘hostility’, Ar. 'dw. Job 10. 17 (M T ‫) ע ך י ף‬. Pope, Job, p. 79,
following Ehrlich.
230. ‫ ע ד ה‬B. ‘p rim e’, Ar. gdw ‘m orning, early’. Ps. 103. 5 (|| ‫) נ עו רי כי‬.
D river, J T S xxxvi (1935) 155.
231. ‫‘ ערד‬go ro u n d ’, E th. 'odd. Job 25. 5 (point ‫) ע ד‬. D river, A jfS L lii
( 1935161 (6‫ ־‬.
232. ‫ עז ב‬A. ‘help’, Eth. 'dzzdbd. Jer. 49. 25; see text, p. 140. D river, J Q R
xxviii (1937-8) 126.
233. ‫ עז ב‬B. ‘be agreeable’, A r .faduba. Job 9. 27; see text, p. 141. D river,
V T S iii (1955) 76.
234. ‫‘ עזי‬patience’, Ar. caza (razw). Ex. 15. 2; see text, pp. 29 f. Rabin,
S crH viii (1961) 387. O r ‘w arrior’, Ar. g azi\ G aster and W inton
T hom as, E T xlviii (1936-7) 45, 478; xlix (1937-8) 189.
235. ‫ עז ר‬A. ‘be valiant’, U gar. gzr. I Chr. 12. 1; see text, pp. 139 f. D river,
C M L , p. 142.
236. ‫ עז ר‬B. ‘ju stify ’, Ar. 'adara ‘excuse’. Isa. 50. 7, 9 (\\ ‫ ;) מ צ די קי‬see text,
pp. 139 f. Eitan, H U C A xii-xiii (1938) 81.
237. ‫ עז ר‬C. ‘h in d er’, Ar. 'azara ‘rebuke’. Job 30. 13; see text, p. 139‫״‬
D river, A J S L lii (1935-6) 163.
238. ‫ עז ר‬D . ‘be copious’, Ar. gazura. Zech. 1. 15; see text, p. 140. Eitan,
p. 8.
239. ‫‘ ע ט ה‬give’, Ar. ,apt. Ps. 84. ‫( ך‬Swaei); see text, p. 249. Rabin, West-
Arabian, p. 40, n. 5 and p. 32.
240. ‫‘ עי ט‬den ’, Ar. gayit. Jer. 12. 9; see text, pp. 128 f., 153, 235. D river,
P E Q , A pr. 1955, 139. Cf. no. 266.
241. ‫ ע ץ‬,‫‘ עין‬com pany, group’. Hos. 10. 10; see text, p. 226. W ernberg-
M oller, p. 59; D river, Judaean Scrollsf p. 435.
242. ‫ עי ר‬A. ‘revile’, E th. ta'ayyara. Job 3. 8; see text, p. 125. D river,
V T S iii (1955) 72.
243. ‫ עי ר‬B. ‘invasion’, Ar. gara. Jer. 15. 8; see text, p. 125. D river, J Q R
xxviii (1937-8) 113.
244. ‫ עי ר‬C. ‘fire’, Ar. wagara ‘be h o t’. Hos. 7. 4; see text, p. 126. W utz,
p. 312. Cf. N yberg, Hosea, p. 52.
245. ‫ עי ר‬D . ‘inm ost recesses’, U g ar.g r. II Kings 10. 25; see text, p. 126.
Gray, Kings, p. 507.
246. ( ‫ עי ר ) ע רי ם‬E. ‘protectors, gods’, U gar. gyr. M ic. 5. 13 etc.; see text,
pp. 126, 153 n. D ahood, Psalms, pp. 55 f. Cf. no. 253.
247. ‫ עי ר‬F. ‘bore’, Ar. gara ‘sink deep’, gawr ‘d ep th ’. Isa. 50. 4; see text,
p. 126. D river, J T S xli (1940) 164.
 I ND EX OF EXAMPLES 333
248. ‫‘ ע ל ה‬burn, boil’, Ar. gala (galy ). Ezek. 38. 18. D river, Biblica xxxv
(1954) 3°4* D ahood, Psalms, p. 74, com pares ‫‘ ע ל ה‬b u rn t
offering’ (Hom mel).
249. ‫‘ ע ל ה‬co-wife’, Ar. ealla. G en. 49. 4 (M T ‫) ע ל ה‬. Reider, V T iv (1954)
276.
250. ‫‘ ע לו ק ה‬erotic passion’, Ar. 'aliqa ‘hang’, b u t also ‘burn, love physi-
cally, conceive’ (?). Prov. 30. 15. G luck, V T xiv (1964) 367 ff.
Cf. nos. 12 and 103.
251. ‫‘ ענ ה‬stay in a place’, Ar. ganiya. M ai. 2. 12, Isa. 13. 22; see text,
pp. 165, 243, 250. Eitan, H U C A xii-xiii (1937-8) 62.
252. ‫‘ ע צ ל‬be fertile, am ple’, A r. gadira, U gar. gsr. Judges 18. 7, Prov. 30.
16. D ahood in M ackenzie, p. 73.
253. ‫‘ ע ל ה‬blood-daubed stone’, U g ar.g r, A r .gariy. Jer. 2. 28, M ic. 5. 13.
D river, CM Ly p. 142. Cf. no. 246.
254. ‫ ע ש ה‬A. ‘tu rn away’, U gar. 'sy, Ar. Job 23. 9, 1 Sam. 14. 32; see text,
pp. 67, 69, 98, 246 f. D river, Robinson Festschr., pp. 53-5.
255. ‫ ע ש ה‬B. ‘come to ’, Ar. gasa (w). Isa. 5. 4; Prov. 6. 32. K opf, V T ix
(1959) 270; D river, Robinson Festschr., pp. 53 ff.
256. ‫ ע שה‬C. ‘p ro tect’, Ar. gasiya ‘cover’. Ezek. 17. 17. D river, Biblica
xxxv ( 1954) 153· Cf. ‫‘ מ ע ש ה‬cloud’, Ps. 104. 13; Reider, V T iv
(1954) 284.· Cf. Isa. 59. 6; Eitan, H U C A xii-xiii (1938) 83.
257. ‫‘ מ ע ש ר‬libation’ (not ‘tith e’), U gar. 'sr. Am. 4. 4. Cazelles, V T i (1951)
131- 4·
258. ‫ ע ש‬A. ‘rot, p u s’, Ar. gatta (cf. 'utt ‘m o th ’). Hos. 5. 12 (6νρώς); see
text, pp. 144, 243 f., 252 n., 279. D river, Robinson Festschr.,
pp. 66 f.
259. ‫ ע ש‬B. ‘b ird ’s nest’, Ar. euss. Job 27. 18; see text, pp. 144, 244. As
no. 258.
260. ‫‘ פ‬and ’, A r .fa-. Ps. 48. 14, read as ‫ ס תו‬+ ‫ ; פ‬D ahood, CBQ xvi (1954)
17. Job 33. 24; G uillaum e, H ooke Festschr., p. 121. D ahood,
Proverbs, p. 53.
261. ‫‘ פ חז‬scattered’ (nomadic life), Ar. fahd. G en. 49. 4. Rabin, ScrH
viii (1961) 398.
262. ‫‘ פ ס ח‬passover’, Eg. p ’ §h ‘the blow ’. See text, p. 103. Couroyer, RB
lxii (1955) 481-96.
263. ‫‘ פ ר ה‬come o u t’, Eg. pry. Isa. 11. 1 (|| ‫)י צ א‬. Eitan, H U CA xii-xiii
(1937-8) 59.
264. ‫‘ פ ר ח‬be cheerful’, A r .fariha. Isa. 66. 14 (\\ 1 .35 ,(‫שש‬. Eitan, H U CA
xii-xiii (1937-8) 87 f.
265. ‫‘ צ א ל‬bough’, Eg. d*rt9 Copt. cal. Job 40. 21 f . ; see text, p. 105.
H u m b ert, Z A W lxii (1950) 206.
334 IN D E X OF E X A M P L E S
266. ‫‘ צ בו ע‬hyena*, A r.d a b u '. Jer. 12. 9 (ύαίνης); see text, pp. 128, 235.
D river, P E Q , A pr. 1955, p. 139. Cf. no. 240.
267. ‫‘ צו ר‬m id st’, Accad. surru ‘h eart’. Ezek. 21. 25 (eV μεσω αυτής).
D river, J T S xli (1940) 169. Cf. Jer. 49. 13 (L X X 30. 7).
268. ‫‘ צ ח‬sun ’. Ar. dihh, E th. dahay. Isa. 18. 4. Eitan, H U C A xii-xiii
(1937-8) 65 f. Cf. no. 23.
269. ‫‘ צי ץ‬salt’, Accad., U gar. gloss si-su-ma. Jer. 48. 9. M oran, Biblica
xxxix (1958) 69 ff.
270. ‫‘ צ ל צ ל‬b oat’, Aram. ‫ צ ל צ ל א‬. Isa. 18. I (II ‫ ;) כ ל י ג מ א‬Job 40. 31 {θ'
ττλοίοις). D river, Robinson Festschr., p. 56.
271. ‫‘ צ פ ה‬arrange’, Ar. saffa. Isa. 21. 5. Eitan, H U C A xii-xiii (1937-8)
67. Cf. no. 272.
272. ‫‘ צ פי ת‬guest’, Ar. dayf. Isa. 21 .5 , cf. no. 271. Eitan, as no. 271.
273. ‫ צ פון‬A. ‘island’, cf. ‫‘ צו ף‬float’, U gar. L and floats over void. Job 26. 7.
T u r-S in ai, Job, pp. 380 f.
274. ‫ צ פ ץ‬B. ‘hiding-place’. Job 37. 22. T ur-S inai, Job, p. 517.
275. ‫‘ צ ר ה‬shrill cry’, Ar. sarra, sartr. Jer. 4. 31 (|| ‫ ; ק ו ל‬στεναγμός), 48. 5
(L X X 31. 5); see text, pp. 279 f. D river, J B L lv (1936) 105, J Q R
xxviii (1937-8) 123.
276. ‫‘ ק בו ץ‬fixing; statue’, Syr. qbaa. Isa. 57. 13; see text, p. 122. D river,
J T S xxxvi (1935) 294.
277. ‫‘ ק ה ל‬forget’, Ar. jahila ‘be ignorant’, A m arna qalu ‘w ithdraw ,
neglect’. Job 11. 10; see text, p. 162. T ur-S in ai, Job, p. 194.
278. ‫ קי‬,‫‘ קו‬in fan t’ (‫‘ קי א‬vom it’). Job 22. 20 (1 ‫) קי מנו‬, cf. Isa. 28. 10. T u r-
Sinai, Job, p. 345·
279. ‫ מ קו ם‬A. ‘opposition’, Ar. maqama ‘com bat’. N ah. 1. 8. D river, J T S
xxxvi (1935) 300 f·
280. ‫ מ קו ם‬B. ‘grave’, Phoen. mqm. Qoh. 8. 10; see text, p. 292. D ahood,
Biblica xliii (1962) 360.
281. ‫‘ קו ר‬dig’, Ar. qara ‘cut a round hole’. Prov. 12. 27; see text, pp. 28 f.
Eitan, pp. 25 if.
282. ‫‘ ק טן‬household’, Ar. qatana ‘reside’. Isa. 22. 24. Eitan, H U C A
xii-xiii (1938) 68 f.
283. ‫‘ ק ל ע‬u p ro o t’, Ar. qala'a. Jer. 10. 18; see text, p. 108. D river, J Q R
xxviii (1937-8) 107.
284. ‫‘ קן‬strength’, Eg. qny ‘strong’, qnt ‘strength’. Job 29. 18. D river,
V T S iii (1955) 85.
285. ‫‘ קנ ה‬shoulder-joint’, U gar. qn. Job 31. 22. D river, C M L , p. 144.
286. ‫‘ ק צ ף‬be sad, vexed’, Syr. qsap. II Kings 3. 27; see text, p. 122.
D river, J T S xxxvi (1935) 293.
 IN D E X O F E X A M PL E S 335
287. ‫‘ ק ר א‬follow*, Ar. qara (<qrw). Isa. 41. 25 (|| ‫) א ת ה‬. Eitan, H U C A x ii-
xiii (1938) 77·
288. ‫‘ ק ל ה‬hold feast, invite to feast*, U gar. qry. Exod. 3. 18; see text,
pp. 102 f. Cf. no. 180.
289. ‫‘ ר א ה‬drink* ( = ‫) רו ה‬. Prov. 23. 31 etc.; see text, pp. 257 f. D river,
Biblica xxxii (1951) 187.
290. ‫ ר ב‬A. ‘showers*, U gar. rb, rbb. Job 36. 28. D river, C M L, p. 155.
291. ‫ ר ב‬B. ‘arrow* (‫‘ ר ב ה‬shoot’). Am. 7. 4 (1 ‫) ל ר נ ב י אעז‬. Cf. Job 16. 13
‫ ך ב י ו‬, Xoyxaig. D river, J T S xli (1940) 171 f.
292. ‫ ר ב‬C. ‘weak, afraid*, Ar. rwb. Job 4. 3; see text, pp. 134L T u r-S in ai,
Job, pp. 76 f.
293. ‫‘ י רי ב‬great*, S y r.yaributd ‘size*. Hos. 5. 13; see text, p. 123. D river,
J T S xxxvi (1935) 295.
294. ‫‘ ר ב ע‬dust*, Accad. turbuu ‘dust spiral*. N um . 23. 10; see text,
pp. 11 n ., 270. G insberg, Z A W li (1933) 309.
295. ‫‘ ת ר ב ע ת‬dust cloud’, Accad. As no. 294. A lbright, J B L lxiii (1944)
213.
296. ‫‘ רזי ל‬wickedness*, Ax. radii ‘base*. Isa. 24.16. Rabin, S crH v iii (1961)
386.
297. ‫‘ רי ב‬trembling*, Accad. rlbu. Job 33. 19. D horm e, Job, p. 454.
298. ‫‘ ר כ ב‬g ather’, cf. M H ‘graft, join, connect*. P s .68.5, cf. vecfreXrjyepera.
Ullendorff, V T vi (1956) 194 f.
299. ‫‘ ת ך מ ה‬treachery’, Accad. turmum. Judges 9. 31; see text, p. 216.
D ossin, O B L y i (1957) 163-7, b u t cf. Gevirtz, J N E S xvii (1958)
59 f- (7 ‫) ת ר ם‬.
300. ‫‘ ר מ ס‬rebuild, repair*, U gar. trmm. Ezr. 9. 9, Prov. 11. 11. D river,
CMLy p. 153.
301. ‫‘ ר ע ב‬be bew itched, fear’, Ar. raaba (cf. ‫‘ ר ע ב‬hu n g er’ = Ar. ragiba
‘desire’). Job 18. 12. D river, Z A W lxv (1958) 260.
302. ‫‘ ר ע ה‬th o u g h t’, Syr. r a. Isa. 44. 28; see text, pp. 281 f. D river, J T S
xxxvi (1935) 82.
303. ‫‘ ר שע‬rich ’, Ar. rassaga, rasig (Lane, p. 1080 c). Job 24. 6; see text,
p. 145. Guillaum e, H ooke Festschr., p. 116.
304. ‫‘ שאה‬wish*, Ar. sa a . G en. 4. 7 (M T ‫) ש א ת‬. Rabin, S c rH viii (1961)
399·
305. ‫‘ מ שה‬place for sheep’, analogy of Ar. mas ada ‘place to expect lions’
etc. Exod. 12. 4. Rabin, S c rH viii (1961) 394.
306. ‫ ה שי ר‬,‫‘ שור‬obtain advice*, Ar. swry ’aSara. Hos. 8 .4 ; see text, pp. 16
188 f. Source as no. 202.
307. ‫‘ שח ה‬wash, flood’ (not ‘sw im ’), Syr. Ps. 6. 7. D river, J T S xxxvi
(1935) 147· Am. 4. 13 (1 ‫‘ ) ו » י ר ל א ך ?ן ה מי שחו‬pours on the
earth the flood w aters’, T ur-S inai, Lashonf ii. 403 f.
336 IN D E X O F E X A M PL E S
308. ‫‘ ש כ ל‬form , beauty’, Ar. Sakl. I Sam. 25. 3; see text, pp. 244 f. Perles,
jfQ R N.s. xvii (1926-7) 233.
309. ‫‘ שפ ק‬abound’, M H . Isa. 2. 6 (|| ‫ ;) מ ל א‬see text, pp. 232 f. W inton
T hom as, Z A W lxxv (1963) 88 if.
310. ‫‘ ש פ ת‬decree, ordain’, denom . from ‫‘ ש פ ה‬lip ’. Ps. 22. 16, Isa. 26. 12;
see text, p. 85. Beeston, V T viii (1958) 216 f . ; cf. U llendorff,
V T vi (1956) 196 f.
311. ‫‘ ש ב ם‬m uzzle’, Ar. sabama, U gar. sbm. Ps. 68. 23 (1 ‫ ב ם מ צו ל ת ים‬$ ‫) א‬.
D ahood, J B L lxxx (1961) 270 f . ; M iller, H T R Ivii (1964) 240.
312. ‫ ש ב ר‬A. ‘attend to ’, Ar. tabara ‘apply oneself w ith zeal’. N eh. 2. 13,
15 (συντριβών). M T ‫ ש ב ר‬. D river, J T S xxxv (1934) 382 f.
313. ‫ ש ב ר‬B. ‘m easure by the span’, Ar. sabara. Job 38. 10. G uillaum e,
Hooke Festschr., p. 123.
314. ‫‘ ש ד ד‬expel’, E th. sadada. Prov. 19. 26. W inton T hom as, V T S iii
(1955) 289.
315. ‫ שדי‬, ‫‘ ש ד ה‬pour out, dow npour’, A ram . Ezek. 1. 24, II Sam. 1. 21
see text, p. 235. G ordis, J T S xli (1940) 34 if.; D river, ibid.,
168, and V T iv (1954) 239 f.
316. ‫* שיב‬run ab o u t’, Ar. saba(y); cf. ‫‘ עזוב‬re tu rn ’ = taba (w). Jer. 8. 6;
see text, pp. 242 f. D river, J Q R xxviii (1937-8) 105.
317. ‫‘ שיל‬governor, prince’. G en. 49. 10; see text, pp. 120 f. M oran,
Biblica xxxix (1958) 405-25, has full discussion.
318. ‫‘ שיל‬travel’, Syr. syartd ‘caravan’, Ar. sara. Ps. 138. 5; Ezek. 27. 25.
D river, J T S xxxv (1934) 388.
319. ‫ ש כ ב‬fp o u r’, Ar. sakaba. Job 38. 37; see text, p. 137. Orlinsky, J B L
lxiii (1944) 1 9 4 4 ‫· ־‬
320. ‫ ש כ ח‬A. ‘be paralysed’, Ar. kasiha by m etathesis. Ps. 137. 5; see
text, pp. 48, 152. Eitan, J B L xlvii (1928) 193 if.
321. ‫ ש כ ח‬B . ‘droop, w ilt’, U g a r .iM .P s . 137. 5; cf.no. 320. D river, C M L y
p. 151; b u t cf. Pope, J S S xi (1966) 240.
322. ‫‘ ש כ ת‬ship’, U gar. tk ty Eg. skti. Isa. 2. 16; see text, pp. 280 f. D river,
Robinson Festschr., pp. 52 f,
323. ‫‘ שלג‬soapw ort’, M H ‫ א ש לג‬. Job 9. 30; see text, p. n o . Pope, Job,
p. 72.
324. ‫‘ ש מח‬be kind, clem ent’, Ar. samuha. H os. 7. 3. N yberg, Hosea, p. 46.
325. ‫ ש מ ר‬A. ‘cultivate’. Hos. 4. 10 f.
326. ‫ שמר‬B. ‘rage’, Accad. samaru. Am. 1. 11.
327. ‫ ש מ ל‬C. ‘cast out, reject’, Syr. smr, Ps. 37. 28.
F or 325-7 see text, pp. 119 f., 141 f. D river, J T S xxxv (1934)
384 if. F or 327, also T ur-S in ai, Job, p. 240.
 IN D E X O F E X A M P L E S 337
328. ‫‘ שבה‬be em inent, of high rank’, Ar. sny, sana. Prov. 24. 21 etc.
Eitan, pp. 10 f . ; W inton T hom as, Z A W lii (1934) 236 fF., V T S
iii ( 1955) 286.
329. ‫‘ מ שנה‬equivalent’, Accad. mistannu (not ‘twice as m u ch ’). D eut. 15.
18; Jer. 16. 18. Tsevat, H U C A xxix (1958) 125 f.
330. ‫‘ ש ע ה‬betake oneself’, Ar. s a d ‘move quickly’. I Sam. 14. 32 (M T
‫ ;))וי ע ש‬see text, p. 98. Reider, H U C A xxiv (1952-3) 85.
331. ‫‘ שר ש‬be angry’, Ar. sarisa ‘be vicious’. Job 5. 3. T u r-S in ai, Joby
p. 94.
332. ‫‘ ש ת ע‬fear’, Phoen., U gar. t t \ Isa. 41. 10; see text, pp. 180, 233 n.,
294 f. Greenfield, H U C A xxix (1958) 226-8.
333· ^ 0% ‫ >י‬E th. tatahala ‘w ander’, D illm ann, col. 552. Job 4. 18.
Pope, Job, p. 37.
334. ‫‘ תוה‬w ander in m in d ’, Ar. taha. Isa. 44. 9, read as verb. Eitan,
H U C A xii-xiii (1938) 78.
‫ ״ ן‬see no. 45.
 INDEX OF BIBLICAL PASSAGES
Ordering and numbering are as in the Hebrew text. Arabic figures refer to
pages of the te x t; italic figures to numbers of the Index of Examples.

Genesis Numbers
I. I 59 5· 13 137
i. 8 45 14· 37 16
1« 26 40 16. 1 17 f., 271, 166
3· 16 286, 109 21. 14 46
3· 20 47 21. 15 235
4-6 251 23. 10 11 n., 270, 294
4· 7 304 23. 21 169
4· 15 211 23. 22 160
5· 29 47 24. 17 53
6· 3 88 34· 7 20
10. 13 204
13· 12 *3 Deuteronomy
13· 18 13 1· 13, 15 20
22. 7 f. 57 11. 14 249
22. 13 26 II . 30 SB
25· 3 254 f· 15. 18 3 2 9

27. 29 254 23· 14 2 5

30. 20 I82 25.19 213

31· 36 93 32. 8 227
37· 2 16 33· 2 59, 41
41· 43 105 33· 3 230, 117
43· i i 30 33· 12 182
49· 4 249, 2ÖJ 33· 25 80
49· 5 57, 270, 200 34· 3 125
49. 10 120 f., 234, 3^7
49· 22 30 f. Judges
49· 26 21 5· 5 114
50. 5 57 5· 16 258
6. 34 j #9
Exodus 7· 3 55
2. 3 152 n. 8. 16 19 f·, J5J
2. 25 22, J52 9· 3i 299
3· 18 102 f., 233 · 37
1 1 174
4. I I 255 n. 12. 6 129
5· 3 102 f. 14. 18 152
12. 4 305 15. 16 152
14· 3 49 16. 9 21
15· 2 29 E, J-T5, 234 18. 7 14 f., J76, 252
15. 10 136 19. 2 286, j j 6
16. 13 137 n .
17· 13 276 I Samuel
22. 28 95 1. 6 67, 69
23· 5 140 1. 18 250 f., J07
 IN D E X O F B IB L IC A L P A S S A G E S 339
I. 20 47 20. I 1 6 4

2. 5 1 2 0 21. 12 137 n .

2. 10 283 24· 15 2 2

2. 33 6

3· i i 137 n. a ia h

4. 2 257, 2 1 3 i. 17 4 2

4· 15 3 0 I. 22 2 1 9

7· 2 231, 264 f. 2. 6 2 3 2 f., 3 0 9

10. 2 248 n. 2. 12 255 n .
12. 14 264 2. l 6 2 8 0 , 3 2 2

13· 21 225, 285 f. 2. 22 57

14· 25 f. 144 5· 4 255
14· 27 144 n. 5· 7 4 8 , 1 5 2 f.

14· 32 67, 69, 98, 246 f., 254, 5 -8 23 4

330 7· 6 6 2

15. 19 247 7· U 133 n .

15. 23 58 8. 16 193
17. 6 1 8 2 9. i 2 5 0 , 2 7 8
19. 20 25 f., 231, 270 f., 1 9 1 9 · 17 3
20. 13 265 io . 32 2 7 8
20. 30 51 I I . I 2 6 3

21. 3 21, J54 i i . 8 182

22. 8 I 2J i i . 15 119, X 42

23. 7 2 .6 7 , 2 1 4 12. 2 30
24. I 175 ! 3 .8 2 0 4

25· 3 244, J 03 13 · 12 2 8 6 f., x 6 x

25. 14 247 1 3 ‫ ״‬21 58

13. 2 2 2 4 3 , 2 5 0 , 2 5 J
II Samuel 14. 4 55
I. 21 235, 3-*‫־‬5 14. 12 134
5.8 1 4 9 14. 19 1 8 8

13. 16 15 14. 23 2 5 0 , 1 ‫־‬4 4

22 191 16. 5 2 9 5 , J 9 6

22. I 192 16. 7 2 5 0 , 254, 43

22. II 193, 218 17. II 2 1 0 , 2X 2

22. 46 1 4 0 18. I 2 7 0

18. 4 2 3 , 2 6 3

I Kings 2 1 . 5 27X , 2 7 2

I. 4 2 2 5 2 2 . 2 4 2 8 2

5· 31 1 6 8 2 4 . 16 2 9 6

I I . 16 213 26. 12 3 1 0

20. IO 232 27. I 6 7

20. 14 32 27. 6 1 0 2

20. 40 246 27 . 10 2 5 0

22. 53 71 n. 28. 7 5*
28. 10 273
II Kings 28. 15, 18 2 3 0 , X 23

3· 27 122, ^36 2 8 . 2 0 20
4· 34 162, 7 2 30. 8 2 4 7 , 223
4· 42 26, 6x 32. 6 153
6. 23 102, x3o 32. 19 6 4

9· 13 79 33· 7 37
10. 25 126, 245 34· i 2 5 4 f.

II. I 8 4 35· i 2 6 4
340 IN D E X OF B IB L IC A L P A S S A G E S
Isaiah (cont.): 64. 8 247, 2 2 8

35· 7 193 n. 66. 14 2 6 4

37· 25 28
38. 10 259, 9 4 Jeremiah
38. 17 55 I. II 49
39· 2 17 2. 28 153 n., 253
40. 1-3 52 f. 2. 31 266, 35
41. I 254 f· 3· 17 264
41. 10 233 n., 294 f·, 332 4· 31 279 f-> 2 75
41· 25 2 8 7 5· 3 1 2 7

41. 27 182, 193, 63, 2x5 5· 24 249

42. 1-4 153 6. 2 212
42. 2 2 1 9 6· 3 212
42. 5 252 6. 7 280 n.
42. 6 65 6. 20 17
42. II 2 i g 8. 6 242 f., 3 1 6
43· 4 254 f· 9. 21 211 f.
43· 9 254 f· 10. 13 204 n.
44· 4 54 10. 18 108, 2 8 3
44· 8 6 f., 188, 231, 36 II. 12 247
44* 9 334 II . 14 247, 2 2 8
44· 12 1 6 3 12. 4 4
44· 24 252 12. 5 90 f., 5 2
44. 24-28 281 f. 12. 9 128 f., 153, 235, 240, 266
44· 25 47 15.8 125, 2 4 3
44· 28 281, 302 16. 18 329
45· i 281 17· 13 1 7 1

45· 12 252 19· 3 137 n·

47· 13 119, 1‫־‬04 19· 15 153 n.
47· 15 4 2
25· 9 1 3 8

48. 13 252 30. 21 264

48. 19 2 0 4 31· 19 21
50. 4 126, 247 31· 22 264
50. 7 2 3 6 32. 41 254
50. 9 244, 236 33· 13 264
51· i 204 41. II 280 n.
51· 4 254 48 (31)· 5 280, 2 7 5
51· 6 252 f., 20J 48. 9 2 6 9

51.8 58 48. 29 280 n.

52. 14 284 f., 207 49. 10 £. 140
53· 3 20 49· 13 2 6 7

53· 5 264 49· 25 140, 2 3 2

53· 10 54 51· 16 204 n.
53· i i 20, 23, J5Ö
54· 5 100 Ezekiel
54· 13 46 1· 13 259
55· i 153, 178 i. 24 235» 3 1 5
57· 13 122, 276 4· 2 135
57· 17 253 7. 21 1 1 4

59· 6 256 16. 30 1 8 7

60. 4 23 16. 31 62
60. 13 211 17· 17 2 5 6

61‫ ־‬10 254 19· 7 98

62. s 250 21. 20 L 2
 I N D E X OF BIBLICAL PASSAGES 34i
21. 25 267 4. 4 257
21. 32 120 4· 13 3 0 7

22. 4 247, 2 2 8 5. 26 277, J 73> 224

22. l8 219 6. 12 236 n.
22. 30 247, 2 2 8 7· 4 69, 260 f., 129 ,32‫־‬J
25. 7 4 6 8. 2 48 f.
27. 5 f· 280
27· 19 189 f., 286, 96 Micah
27. 25 3 1 8 1· 4 235
27. 32 I J 2 i. 7 153, 45
28. 14 2 0 6 I. 12 147
SO* 9 50 2. 6 15, J76, 2ZÖ
31. 3 137 5· 13 246, 253
38. 18 2 4 8 6. 14 I J O

Hosea Nahum
2. 17 264 i. 5 252
3· 2 2 1 4 i. 8 279
3· 5 17 I. 10 33
4. 4 74
4. 10 f. 1 4 1 , 3 25 Habakkuk
4. 18 J 2, JOJ 2. 16 252
5· 12 243 252 n., 279, 25S 3· 5 212 n.
5· 13 123, 2 9 3 3· 6 252, J 99, 22J
6. 3 24 3· 6 f. 8, 190, J43
7· 3 524 3· 9 1 9 8

7· 4 2 4 4 3. 10 1 5 0

7· 5 283 3· 13 158
7 .6 35 3· 14 1 9 8

7· 9 22, *55 3. 16 137 n.

7· 15 !59
7. 16 69, 74 Zephaniah
8. i 26 f., 220 2. I 177
8. 4 188 f., 202, 306 2. 14 74
8 .9 74 3· i 148
8. 10 283 n.
8. 13 233 ■2‫־‬03 Zechariah
10. 6 123 i. 15 140, 2 3 8

10. 10 74, 3 3 , 2 4 1 II. 2 280

M
O

71 n.
Ui

II. 4 154, 9, n Malachi

11. 7 283 2. 7 1 8 4

11. 9 125 f. 2. 12 243, *5-r

12. 2 107
14· 3 (2) 16, *46 Psalms
I. I 278
Joel 2. I l f . 5. I7S. 284, 76, 77,
2. 23 249 78
4. II 182 4· 7 I 4 8

5· 4 153
Amos 5· 10 · rjj
i. 13 III 6. 7 on
on'-‫׳‬
2. 13 53 7 .8
3· 3 153 7· 9
342 IN D E X O F B IB L IC A L P A S S A G E S

P s a l m s («cont . ) : 68. 32 107, 210

7 · 13 183 71. 10 31
10 8 10, 14
. , 228 f . 7 1 · 13 249
10. 17 18 72. I I 255 n .
12. 4 7 *, 133 74· 5 9 *
18 191 74· 14 236 f .
18. I 192 78. 54 248, 70
18. 8 252, 273 82. 8 211
18. I I 192 f . , 218 84. 7 2 4 9 f · , 23 9
18. 16 77 88. 16 34
18. 46 140 89. 20 13911.
19. 6 210 89. 46 249
20. 6 262, 85 89·. 47 253
22. 16 310 90. 8 130 f.
22. 17 175 102. 24 f . 213
22. 30 296 n . , 101 103. 5 230
24. 3 278 104. 13 256
25. 14 251 f . , 223 104. 18 230 n .
29. 2 105 108. 10 5 4 f-
29. 6 32 f . 109. 19 249
31· n 244 109. 22 130
31· 12 *7 109. 29 249
33· 6 252 n o . 3 210
3 7 · 23 136 116. 10 82
37· 28 141 f . , 327 n s ( 117) . 14 30
37· 39 142 n . 119. 28 92
38. 10 j 3 119. 74 154
39· 3 17 126. I 131
39· 12 244 135· 7 204 n .
40. 4 154 135· 10 255 n .
41. 10 7 * 137· 5 48, 152, 320, 321
43· 4 284 138. s 318
45· 2 98, 295, 196 139. 2 281
45· 10 287 139· 8 51
46. 2 230 n . 148. I I 254 f · , 185
46. 1-3 211
46. 4 252 Job
47· 10 241, 251, 194 3· 3 30
48. 3 57 3 -8 125, 242
48. 6 108 3· 14 69 n . , 139
48. 14 260 4· 3 1 3 4 , * 5 9 , 292
4 9 · 15 112 4. 14 134
52. 6 56 4. 18 333
55· 5 130 5· 3 331
55· 23 56 6. 21 154
58. II 192 6. 30 J 06
59· 8 * 3 * 7. 6 39
60. 10 5 4 f- 9. 21 22
68. 5 53, 293 9· 27 141, 235
68. 7 181 9· 30 323
n o ,
68. 9 114, 252 9 · 31 96
68. 19 175 10 17
. 209, 229
68. 23 69, 3 * * 10. 22 242, 222
68. 24 192, 197, 205 I I . 10 162, 277
 IN D E X OF BIBLICAL PASSAG ES 343
1 2 . 8 4 0
37- i i 23
1 2 . 2 3 1 9 0
3 7 - 17 190, 195
1 4 . l 6 1 4 2 , 1 2 5 , 3 2 1 37. 2 i 6 9 n .
1 5 . 1 2 1 8 f ., 6 9 n . , 1 6 6 3 7 - 22 2 74
15- 3 0 6 9 n. 3 8 . 10 313
1 6 . 4 1 1 8 , 1 1 9 38. 16 48
1 6 . 1 3 2 9 1 38. 2 4 2 6 0
1 6 . 1 8 2 9 2 38. 3 0 2 3 4 , 186

17- 1 5 1 6 5 38. 36 57
1 8 . 3 6 9 n. 38. 37 1 37 , 319
1 8 . 1 2 3 0 1 39- 25 27 f., 8 9 , 2 2 0
1 9 . 1 8 83 40. 17 7 1, 1 3 6
1 9 . 2 6 6 9 n . , 6 8 40 . 21 f. 105, 2 65
2 0 . 3 2 8 0 n . 40 . 23 9 1, 5 2
2 0 . 1 8 6 9 n . 40 . 29 2 1 1
2 0 . 2 0 2 1 40. 31 2 10
2 0 . 2 2 2 3 2 41. 9 2 3 4 , 1 5 6
2 0 . 2 4 6 6
4 1. 14 80
2 1 . 2 4 6 9 n . 41. 18 6 9 n .

2 1 . 2 7 6 9 n.
2 2 . 2 0 2 1 8 Proverbs
2 2 . 2 8 f. 2 9 3- 2 7 100

2 2 . 3 0 147 3 - 35 129

23- 9 254 5- 3 2 57
2 4 . 6 303 6. 1 2 5 6 f., 114
2 4 . 1 9 75 6. 18 2 1 0

25- 5 * 4 , 2 5 * 6. 25 166
2 6 . 2 139 6. 26 58
2 6 . 3 6 9 n . , 1 3 4 6. 32 255
2 6 . 7 275 8. 21 24
2 6 . 1 3 6 7 8. 22 99
2 7 . 1 8 2 4 4 ) 2 59 8. 2 4 48
2 8 . 1 1 4 8 8. 29 1 21
2 8 . 1 6 2 8 7 10. 9 2 1 , 23

2 9 . 1 2 139 10. 16 126

2 9 . 1 8 2 8 4 10. 18 16

3 0 . 1 3 139) 2 3 7 10. 32 2 4
3 0 . 2 4 59 11. 11 300
3 1 . 2 2 2 8 5 11. 31 2 5 8

3 1 * 2 3
2 6 6 , 1 5 12. 27 28, S, 2 S 1

3 1 - 39 1 0 0 14. 18 129

33- 6 *79 I 4 * 33 21 , 23

3 3 - 19 2 9 7 15- 2 i 7 L *45
33- 2 4
2 6 0
15 - 13 75
33* 2 5 6 9 n . 17 - 5 16

34- 4 17 17. 14 257, **5

34 * 3 2
110 17. 22 I 7 L 75

35- 9 134 19. 26 5*4

3 6 . 1 7 9 0 21. 2 0 29
3 6 . 1 8 6 9 n . 21. 25 *9
3 6 . 2 8 2 9 0
23- 3i 2 5 7 f., 2 S 9

36. 3 1 8 7 , 90 24. 14 24, 2 1 6 , 9 7

3 6 . 33 6 9 n . 2 4 . 21 3 2 8

3 7 * 6
2 5 1 , X 0 7 25- 4 2 1 9 , 1 0 2
3 4 4 IN D E X O F B IB L IC A L P A S S A G E S
Proverbs ( cont .): 10. 9 225
25. 14 204 10. 20 20, 23, 157
25. 17 167 12. 2 27
25- 27 258, 1 3 7 12. 6 210
26. 23 219 f., 2 2 6
26. 28 20 2, 9 1 Lamentations
27. 18 10 2. 3 263
28. II 258, 1 1 7 2. 16 234
29. 7 24 3. 11 100
29. 10 63 4. 11 263
29. 18 283 f., 1 2 4 4- 14 263
30. 1 4 7 4. 15 262 f., 2 1 8
30. 1 5 234, 1 0 3 , 25 0
30. 16 25 2 Esther
30. 17 234, 271 n ., 1 9 1 1. 13 109

Canticles D aniel
3- 8 153 5. 26 ff. 49
3- 9 103 7. 28 281
3- i o 154
4. 1 161 Ezra
4- 3 147 9. 9 300
4. 4 45
5- i 144 n. Nehem iah
5. 6 81 2. 13 312
6. 5 161 2. 15 312
6. 10 85 3- 8 141
7. 3 107, 5 5- 7 188, 2 0 3
7. 10 16 f. 6. 19 16, 1 4 6
8. 2 -8 39
Q oheleth 13- 10 30
2. 25 122 i 3 - 19 137
3. 11 162 13- 24 39
3- 17 101
5- 5 184 Chronicles
5- 6 81 1. 11 204
5. 12 128 7- 23 181
7- 6 153 12. 1 139 f., 235
7. 7 20S
8. 8 100, 6 0 II Chronicles
8. 10 292, 2 8 0 1 4 .1 0 134 f.
9. 4 51 22. 10 84
 INDEX OF SUBJECTS
Arabic figures refer to pages of the te x t; italic figures to numbers of the Index
of Examples.

Accadian, general, 36 f., 95, 102 if., lexical idiosyncrasies of, 114, 116,
110-13, 123, i 58> 178- 119, 165 f., 169 n.
homonyms in, 149, 155. synonyms in, 117.
lexical tablets, 119 ff. homonyms in, 106, 126.
lexical idiosyncrasies, 255. script, 198.
script, 198. Job read as, 34.
discovery of, 70 f., 92. supposed close to proto-Semitic,
loan-words in, 103, n o , 149. 113 f·
words cited from, passim. predominance attached to, 67,
Accents, Hebrew, 33, 59, 196. 112 ff., 297.
Acoustics, 81, 275. words and features cited, passim.
Actuality, argument from, 285 ff. Arabism, 15, 121-4, 241.
*Addady 173-7. Aramaic, of Old Testament, 31,73,104.
Aeschylus, 152 n. replacement of Hebrew by, 38-43.
Aistleitner, J., 149. of Talmud, 43.
*Al, deity, 283. influence on other versions, 54 f.
A lb righ t, W . F ., 34, 75, 112, i i 4 n . , similarities to Hebrew, 50-6, 58.
139 n., 188 n., 190, 225 , 240 n., in Israelite names, 184 η.
3 1 , 48, 5 7 , 67, 9 9 , I 1 3 , 1 4 3 , 15 8 , influence on late Hebrew, 228, 235,
i6 o f i6 g f i g 4 , i g 8 , 295. 262, 289.
Aleppo Codex, 204. Aramaized Hebrew, 40 f.
Allegory, 3, 44, 50; linguistic-form words in biblical Hebrew, 121 n.,
allegory, 44. 178·
Allophones, 215. words and features cited, passim.
Alphabet, 198 f. Aramaism, 121-4.
Al-tiqre, 45 f., 212, 214. Archaeology, 285, 298 n.
Al-Yasln, 119. Archaizing in style, 40.
*Am, deity, 283 n. Area preferences, m - 1 4 .
Amarna, 33, 76. Aristophanes, 72.
Amorite, 33, 36, 184. ‘Artificial’ language, 41.
Analogy, semantic, 90 f., 109, 161. ‘As’ in derivations, 104, 291 n.
Animal names, 10, 128 n., 235. Assimilation, phonological, 96 f., 101,
Antiochus Epiphanes, 39. 178 f .; of texts, 247, 282, 287.
Apodeictic assertions, 80, 93. Assonance, 48, 152 f.
Apollonius Rhodius, 40. Asterisk, not used, 11.
Aquila, 59, 174, 210 ff., 244, 253, Atomistic study, 8, 94, 134, 290, 302,
258 f., 263, 270 f., 281. 304·
Arab philology, 62, 116-19, 126 n., Avestan, 131 f.
143 n.
Arabic, cited in Talmud, 56. Babel, Tower of, 185.
used by Jews, 60-4, 217. Babylonian punctuation, 200, 217.
improved knowledge of in West, Bacher, W., 61, 203.
67 f. Bailey, H. W., 108 n.
words without Hebrew cognate, Barr, J., 44 η., 50 η., 91η., 2 ii n.,
162. 215 n., 226 n., 239 n., 281 n., 301 n.
346 IN D E X O F S U B J E C T S
Barth. J., 97, 165 n. Chicago Assyrian Dictionary, 115,
Basic meanings, 172, 197. 143 n.
Bauer, H., 82, 200 n. Chinese, 77, 151.
B e ‘from’, 175 if. Christianity, 285.
Bedarshi, A., 64. Classification of Semitic languages,
Bedouin, 113. 113-
Beer, G., 69. Codex B 19a, 65, 204.
Beeston, A. F., 85, 176 n., 41, 310. Cognate words, proportions of, 156-
Belardi, W., 108. 64, 305- 7 -
Ben Asher, 61, 65, 207. Communication, 134-45, 150 n.
Ben-Jehuda, 236 f. Comparative literature, 10.
Ben Naphtali, 207. Comparative philology, 76-94 and
Bergsträsser, G., H 3 f., 157 if., 164, passim.
168, 203, 217 n. ancient rudiments of, 50-60.
Bgdkpt, 215. Compatibility of consonants, 106 n.,
Biblia Hebraica (B H 3), 18, 22, 25, 178-81.
27 f., 30 ff., 62, 65, 69 ff., 105 n., Componential analysis, 142 n.
139 f-, 189, i92f·, 210, 230, 239, Concrete sense in words, 276.
283. Conjecture, see Emendation.
Bilingualism, 41 f. Consonants, varying frequency of, 87,
Biliteral theory, 99, 166-70, 273-6. and see Patterning, Compatibility,
Binyanim, 130, 149. as basis for ideas, 197 f.
Birds, names of, 273. and character of writing system,
Birkeland, H., 39, 114. 198 f.
Blau, J., 91, 93, 95. Consonantal text, fallibility of, 191-4.
Bloomfield, L., 138, 142, 274 n. Context, dubious guidance of, 282.
Boling, R. G., 254 n. Coptic, 92, 105.
Borgen, P., 46 n. Correspondences, phonological, 29,
Botterweck, G. J., 167 n., 174 n. 64, 78, 81-5 , 93, 95‫ ־‬102, 169,
Brockelmann, C., 31, 82, 86, 93, 180 f., 288.
96 n., 97, 150, 160, 162 f., 176 n., Hebrew-Aramaic lexical, 50 if.
217 n., 234 n., 241 n., 265 n. Semitic lexical, 156-64.
Brown-Driver-Briggs (BDB), 19 f., Couroyer, B., 262.
22, 26, 27, 57, 69, 97, 102, 105, Cowley, A. E., 104 n.
123 m, 125, 128 n., 132, 145-8, 150, Criteria, 8 f., 11 f., 288-304.
182, 229 f., 233 m, 235 m, 246, Critical edition, problems of, 300 f.
291 n ., 295· Cross, F. M., 231 n., 246 n., 80, 105.
Brüll, A., 57 n. Cushitic, 106.
Buck, C. D., 164.
Buxtorf, J., 66. Dahood, M., 23, 100 f., 112, 126,
B‫־‬words in parallelism, 183, 254 f. 153 n., 188 n., 240 n., 253, 268 n.,
2 9 2 , 3 0 3 , 7 , i o y 16, 19, 24, 27, 31,
Caesura, 32. 5J , 57, 60, 71, 78, 108, 133, 148,
Cain, 47, 48 n. J 5 6 , J 5 7 , 183 f., 246, 248, 252,
Cairo Codex, 65. 260, 280, 311.
Caique, 264 f. Daiches, S., 13g.
Canaanite, 100 f., 229, 292. Daniel, Book of, 39 f., 49, 220 f.
Cappellus, L., 66 f. Date, as criterion for preference, 112 f.
Case endings, 78. Dead Sea Scrolls, see Qumran.
Caskel, W., 113 n. Decipherment, 92.
Cassuto, M. D., 26. Delitzsch, Franz, 70.
Castell, E., 67. Delitzsch, Friedrich, 70 f., 158, 192,
Cazelles, H., 257. 1, 34, 197-
Checkmate, 107 f. Derash, 44, 46.
 IN D E X O F S U B J E C T S 347
Deutero-Isaiah, 146, 148, 155 n. Efros, I. I., 126 n., 143 n.
Dhorme, E., 19η., 2 9 7 . Egypt» Jews in, 54, 208, 241, 268 f.,
Diachronic linguistics, 94. and see Septuagint.
Dialect, general, 83 f., 98-101. Egyptian language, 92, 103 if., 111,
in Hebrew, 23, 25, 73. 178.
in Canaanite, 100 f. Ehrlich, A., 71, 266, 14, 170, 229.
in Arabic, 99 f., 174, 186. Eitan, I., 9, 18 f., 22, 28, 75, 140,
Dictionaries, problems of, 8, 95, 105, 142 n., 153 n., 276, 8, 23, 28, 62,
115-21, 145, 161, 163, 224, 291 n., 7 4 , 8 i , 8 3 , 8 9 , 1 2 5 , J 2 7 f . , 2 3 4 ‫־‬,
296, 299 f. 2 3 7 ‫־‬, J 54 , J 6J> 16 3 , 2 33 ‫־‬6 6 , 2‫־‬,
Dillmann, A., 57, 70, 139 ff., 158, 166, 2 j ‫־‬o , 2 1 2 y 2 3 6 , 2 3 8 y 2 5 J , 2 5 6 ,
333· 263 f., 265, 271 f.y 2 8 l i.y 287 y
Discontinuous morphemes, 199. 320y 328y 334.
Dissimilation, 96 f., 101, 178 f. Elephantine, 39, 181.
Dossin, G., 216, 299. Ellenbogen, M., 103 n., 104 n.
Dozy, R. P. A., 116, 265 n., 1 3 8 . Ellermaier, F., 122.
Drift, phonetic, 89, 201 f. Ellipse, 3, 48, 63.
Driver, G. R., general statements of, Emendation of text, 3 f., 6 f., 30-5,
3 3 7 5 ,72 ,7‫ ־‬, m f · , 116, 165 η., 69-75, 188-222, 239 f., 245 f., 289,
185 n., 225, 296 n. 297, 301 if., and in detail passim.
on the versions, 72, 240 n. Emerton, J. A., 39 n., 221.
and semantic method, 290. Enclitic memy 31 ff.
on Aramaisms, 121 ff. English, 135 f., 138, 155, 195,274,300.
on parallelism, 278 n. Enoch, 4.
on onomatopoeia, 273 f. Eppenstein, S., 98 n., 175 n.
on biliteral bases, 168. Erroneous translations, evidence from,
and Dahood, 111 f., 303. 240, 244 f., 247, 254, 256, 258 f.,
particular observations, 26, 28, 91, 272.
105 η., io8 f., ii9 f ., 125, 128, ESA, see South Arabian.
i3 3 n ., 1 3 9 1 5 3 ,149 , 4 3 ‫ ־‬nEthiopian
·, 1 7 * , languages, modern, 105 f.,
174 f., 188 f., 225 f., 230, 235, 112, 176, 196.
241-4, 247, 2 5 4 2 6 2 ,9‫ ־‬n., 265 f.,Ethiopic, general, 36, 67, 78, 97, 112 f.
280, 284, 286, 3y 4y 6y 9, I i y 1 4 - loan-words in, 102, 105 f.
l8 y 22y 35y 42y ^ ‫ ^ ־‬, ^ Ο, homonyms in, 149.
5 2 , 5 5 f., 5 * f · , 6 r , 6 3 , 6 6 , 6 8 - 7 0 , gender in verbs, 30.
73, 77, 82-7y 91 i.y 94y 98, writing system of, 198.
100 f., 107y 112y ii4 y 116, 121, words tabulated against Hebrew,
1 2 3 i.y 1 2 7 y I29y I32y 1 3 ^y 158 f., 164.
1 3 8 - 4 0 y 1 4 2 , *45, 1 5 9 , *62, semantic field of ‘know’ in, 171 f.
166, 171y 175, *77 f·, 1 8 2 , words cited, passim.
1 8 5 f.y jp r , I94y 22 0 1 ,99‫ ־‬f.y Etymology, general, 62, 85, 103 f.,
2 o 6 y 2 o 8 y 2 I I y 2 I 3 y 2 l 8 y 2 2 1 ff., 115 f., 129, 143 f., 160 f., 229.
2 2 8 y 2 3 0 - 3 , 2 3 5 y 2 3 7 y 2 4o - 3 , medieval, 62, 64.
247 E, 2 5 3 -6 , 2 5 8 f.y 2 6 6 f.y popular, and etymologizing, 45-50,
2 7 0 y 2 7 3 f.y 279, 2 8 3 ~ 6 y 2 8 9 - 9 1 y 52, 59, 62, 108 f., 118, 174, 210,
2 9 3 , 3 0 0 ~ 2 y 3 0 6 f .y 3 l 2 y 3 1 5 f .y 251, 253 if., 266 f., 269, 281, 289,
3 l 8 y 3 2 1 f.y 3 2 5 ff. 294 f·
Driver, S. R., 69, 144 n., 264. etymological dictionary, 299 f.
Ducange, C. D., 270 n. Eve, 47.
Duhm, B., 75, 242.
Dunash ibn Labrat, 62 f. Face, hiding of the, 253.
Favourite words of translators, 251-3,
Economy of hypotheses, 80. 261.
Education, linguistic, 295 f. Field, F., 58 n., 271 n.
348 IN D E X O F S U B J E C T S
Fish, names of, 165, 236 f.
Fitzmyer, J. A., 39 n.
Fraenkel, M., 82.
Fraenkel, S., 108 n.
Freedman, D. N., 231 n., 303 n.
French language, 136,138,155,274 n.,
297.
Freytag, G. W., 7 n., 72 n., 116, 117 n.,
165 f., 181, 263.
Gaster, T. H., 234.
Geiger, A., 16.
Gemser, B., 24 n., 28.
Gender, 30 f.
Geographic proximity as criterion,
113·
Gerleman, G., 256 n.
German, Germanic, 81, 83 f., 164.
Gesenius-Buhl (dictionary = GB),
19, 57, 128 n., 137 n., 148, 152,
189 n., 229 n., 230, 233 n.
Gesenius-Kautzsch (grammar = GK),
24 n., 31, 216.
Gevirtz, S., 299.
Gillieron, J., 138 f.
Ginsberg, H. L., 33, 219, 270, 149 f.,
181, 226, 294.
Gleason, H. L., 83 n., 186, 195 n.,
196 n., 199 n.
Glottochronology, ii4 n ., 184-7.
Glück, J. J., 103, 250.
Goetze, A., 121 n.
Goitein, S. D., 85.
Goldman, M. D., 23 n.
Golius, J., 117 n.
Gonda, J., 47 n.
Gordis, R., 16 f., 45 n., 153 n.,
162 n., 213 n., 146 f., 217, 315.
Gordon, C. H., 26 n., 149, 175 ff.,
194 n.
Goshen-Gottstein, M. H., 42 n., 94 n.,
165 n., 216 n., 239 n.
Grabe, J. E., 248.
Grammar, generally, 3, 30-3.
in analysis by versions, 209.
standardization of Hebrew, 73, 206.
anomalies in Massoretic, 216 f.
Grapheme, 196, 198, 218 n.
Gray, G. B., 233.
Gray, J., 85 n., 107 n., 126, 292 n.,
32, 79, 164, 168, 185, 204, 214,
245-
Greek language, 12, 40, 43, 77, 86, 91,
109, 152 n., 155, 195, 297·
modem, 201.
words in Hebrew, 40, 57 f., 62, 104.
adaptation of Phoenician script,
199-
words cited, passim.
Greenberg, J. H., 106 n., 178 ff.
Greenfield, J. C., 41 n., 271 n., 273 n.,
92, J92, 332.
Grimm, J., 81 f., 84.
Grossmann-Segal (dictionary), 236 f.
Gudschinsky, S. C., 184 η., 185.
Guillaume, A., 34 b, 99, 123 η., 125,
173, i 82, 53 f; 75> 9<>, 165, 172‫־‬,
207, 2x5, 260, 303, 313.
Gunkel, H., 75, 140.
Gutturals, 214 f.
Ha-Levi, Isaac b. Eleazar, 64.
Hailperin, H., 66 n.
Halper, B., 45.
Hamitic, 175.
Hapax legomena, 6 f., 61, 70, 102, 119,
147, 204 f., 230 f., 263.
Harari, 294 n.
Harmonization, 282.
Harris, Z. S., 165 n.
Hatch and Redpath (concordance),
252, 257, 263.
Haywood, J. A., 116 n.
Hayyug, Yehuda b. David, 63.
ο *Εβραίος, 18, 271.
Hebrew language, disuse of by Jews,
38- 43·
a human language, 67.
uniformity of, 73, 100, 206.
regional pronunciations of, 200 f.
knowledge of in Egypt, 208, 240 n.,
268 f.
dialects in, 23, 25, 73.
general intelligibility of, 297 f.
mode of reading unpointed, 197.
size of vocabulary, 223-7.
Hebrews, Epistle to the, 49.
Hempel, J., 217 n.
Hexapla, 209 n., 211 f., 230, 243.
Hieroglyphic, 92.
Himyarite, 174.
Hirschberg, Η. H., 22, 277, 11, 155,
173, 224.
Historical criticism, 68, 77.
History, as element in philology,
76-81, 89.
Hittite, 92, 104, 219.
Hitzig, F., 162 n.
 IN D E X O F S U B J E C T S
Hockett, C. F., 274 n.
Hoenigswald, H. M., 77, 80 n.
Hommel, F., 248.
Homonyms, general, 17 f., 23, 125-
55, 243, 289.
complete and partial, 18, 23 f.,
127, 129 f.
and homographs, 18, 24, 131 n.,
2 4 9 ·
in roots, 130 f., 145, 150.
in verbs, 131 ff.
and style, 135, 151-5.
and semantic fields, 136-42, 146.
and polysemy, 142 ff., 147.
and textual corruption, 144.
and riddles, 152.
and communication, 134-45, 150 n ·
and phoneme mergers, 126-9, 149 f.
produced by semantic change, 174.
in Syriac, 109, 149 ff.
in Ugaritic, 149.
in Arabic, 106, 128.
of like and unlike function, 135 f.,
146.
separation by text or period, 146 f.
word-formation and, 131 ff., 148.
medieval terms for, 126 n.
and Massorah, 204 f.
contamination between, 108 f.
counts of, 145-51, 299.
newly identified, passim.
Hugo of St. Victor, 66.
Humbert, P., 105, 265.
Hummel, H. D., 33.
Hyatt, J. P., 303 n.
Hymes, D., 80 n., 184 n., 186 n.,
275 n.
Ibn Barun, 64 n., 98 n., 174 n.
Ibn Ezra, 17, 63 f.
Ibn Janah, Abulwalid Merwan, 63, 98.
Ibn Koreish, Yehuda, 62.
Ibn Khaldun, 126 n.
Ibn Tibbon, 60.
Idiosyncrasies, lexical, 114, 119, 224.
Indian etymologies, 47 n.
Indo-European, 86, 164, 185, 298.
Inflexional affixes, 86 f., 221.
Information, 118, 291 ff.
Internal reconstruction, 80.
Interpretation, linguistic elements in
Jewish, 44-50.
Islam, 59 f., 63, 74, 276.
Isoglosses, 83 f., 113.
349
Izalla, 189, 286.
Jacob, B., 11 n.
Jastrow, M., 234, 235 n., 244 n.,
257, 263 n., 264 n.
Jerome, 5011., 65, 67, 20911., 211 ff.,
215, 231, 233, 249, 253, 255, 258,
262, 270.
Jespersen, O., 86 n., 143, 274 n.
Jesus, language spoken by, 38 f.
Job, Book of, read as Arabic, 34.
treatment in B H 2, 69 f.
Greek version, 256-9.
Jouon, P., 119.
Josiah, 181, 220.
Kaddary, M. Z., 1 3 0 .
Kahle, P., 203 n., 214-17.
Katz, P., 268 n.
Kautzsch, E., 121 n.
Kennicott, B., 1.
Keywords in prophecy, 48 f.
Kimchi family, 55, 64, 67.
Kitab a l-A in , 116 n., 118.
Koehler, L., 91 n., 236.
Koehler-Baumgartner (dictionary =
KB), 17, 19 f., 26, 30, 57, 91 n.,
128 n., 146 n., 147 f., 174 n., 229 f.,
233 n., 236, 241, 259, 265, 295.
Komlos, O., 229.
Kopf, L., 106 f., n 6 f ., 175, 188, 52,
93, 144, 203.
Koskinen, K., 178 f.
Kraus, H.-J., 67 n., 75, H 4 n ., 192.
Krauss, S., 56 n.
Kretschmer, P., 138 n.
Kutscher, E. Y., 42 n., 216 n,, 217 n.
Lagarde, P. A. de, 26, 82.
Lambton, A. K. S., 106 n.
Landau, E., 173 n.
Landsberger, B., 121.
Lane, E. W., 107 n., 116, 117 n., 118,
165 f., 236, i 8 y g8.
Laryngals, 214 f.
Late Hebrew, 69 h, h i f., 151, 222-
37, 262, 289, and see Mishnaic
Hebrew.
influence of Aramaic on, 228.
poor training in, 224.
Latin, 78, 86, 91, 130 n., 131, 155, 175,
274 n., 297; ecclesiastical, as ana­
logy to Hebrew, 38, 42.
Laws in linguistic change, 81 ff.
350 IN D E X O F S U B JE C T S
Leander, P., 82, 200 n.
Lebanon, 32, 49.
Lehmann, W. P., 80 n., 83 n.
Leslau, W., 97, 141, 159, 174 n.,
294 n.
‘Let’, 138.
Leveen, J., 18.
Levi della Vida, G., 105 n.
Levita, E., 66.
Levy, J., 264 n.
Lexical tablets, Accadian, ii9 ff.
Lexicography, Arabic, 106, 115-19.
medieval Jewish, 58, 174, 289, 300.
general problems, see Dictionaries.
Lexicostatistics, 184-7.
Linguistic-cultural relations, 276 f.
Linguistics, modem general, as di­
stinct from philology, 94.
Liturgical reading, 217.
Loan-words, 17, 101-11, 226, 288.
Logic, 126 n., 164 n., 290.
Low, I., 16, 165, 236 f.
Ludolf, H., 26, 57.
LXX, see Septuagint.
Maimonides, 60, 126 n., 143 n.
Mandaic, 78, 142, 149.
Mandaism, 74.
Mansoor, M., 229, 44.
Mari, 114 n.
Massorah, 61, 65 f., 204 f., 208.
Massoretes, general, 33, 59 f., 188-
222, 289 and passim.
traditionalism of, 203.
double sense of the term, 206 f.
compared with LXX, 209.
supposed failure to understand text,
31, 190, 202 f., 240 n.
supposed innovations of, 214-17.
semantic element in work of, 202-7.
anomalies in work of, 216 f.
and pre-Massoretic vocalization,
207-17.
Massoretic text, antiquity of, 1, 73.
pre-Massoretic date of errors in,
219 ff.
Matres lectionisy 195 f., 199.
Meaning, in comparative philology,
76, 86-91, 289 ff.
in one language, 292, 294 f.
Mechanical preservation of text, 73.
Medieval comparative philology, 56-
6 5 ,7 6 ,1 7 4 ·
Melek, deity, 283.
Mem, enclitic, 31 ff.
Menahem b. Saruk, 62, 277 n.
Menasce, P. J. de, 226 n.
Menner, R. J., 135 n., 138 n., 143.
Mesopotamian institutions, 172, 255.
Merger of phonemes, 126-9, 174.
Metaphor, 3, 242, 276.
Metathesis, 25, 97 ff., 169, 178 f.
Metzger, B. M., 4.
Michaelis, J. D., 68.
Midianites, 181 n.
Midrash, 41, 45 h, 57 ff., 214, 263,
281, 300.
Millard, A., 189 f., 96.
Miller, P. D., 41, 117, 311.
Mirsky, A., 231 n.
Mishnaic Hebrew, 38-42, 62, 69 n.,
76, 122, 163, 217, 305 n., and see
Late Hebrew.
Mixed language, Hebrew as a, 165.
Moabite, 122, 225.
Modern (Israeli) Hebrew, 195, 236 f.
Montgomery, J., 162 n., 260, 72, 132.
Moore, G. F., 14 f., 55 n., 216 n.
Morag, S., 198 n., 215 n., 217 m
Moran, W. L., 33, 36, 114, 269, 317.
Morphology, 206, 295, and see
Grammar.
Moscati, S., 98 n., 113, 129 n., 167 n.,
168 n., 178, 197 n.
Motivated terms, 275.
Mowinckel, S., 120.
Multiplicity of meanings, 44 ff.
Murtonen, A., 79.
Nabataean, 166.
‘Naked’, words for, 169 n.
Names, symbolic use of, 47 ff., and
see Personal names, Etymology.
Nehemiah, 39.
Neofiti Targum, 254 n.
Nestle, E., 4.
New Testament, 4, 39 n., 46 n., 49.
Nicholas of Lyra, 66.
Nida, E. A., 142 n.
Noah, 47, 48 n.
Nöldeke, Th., 82, 150 n., 17311.,
261 n., 268 n., 156.
Nötscher, F., 120, 230.
Non-Semitic words, 101-11.
Noth, M., 106, 181 ff., 184 n.
Nyberg, H. S., 72-5, 100 n., 107
123 n., 126, 194 f., 282, 33, 38,
244, 324·
 IN D E X O F S U B JE C T S
O’Callaghan, R. T ., 182.
Oesterley, W. Ο. E., 37.
Onomatopoeia, 273-6.
Ophir, 286 f.
Opposite senses, words with, 173-7.
Origen, 243.
Orlinsky, Η. M., 137, 80, 3x9.
Orthography, 214, 219.
Oxford English Dictionary, 108, n o .
Palestinian Arabic, 276.
Palmyrene, 133 n.
Pan-Ugaritism, i n .
Parallelism, 32, 62, 65, 183, 243 f.,
248, 254 f., 267, 277-82, 286, 289,
300.
homogeneity in, 255.
differences between parallel pass­
ages, 191 ff.
Paronomasia, 48, 154.
Patterning of roots, 178-81.
Patterns in word-formation, 132, 136.
Payne, D. F., 155 n.
Peake, A. S., 188 n., 225 n., 240 n.
Perles, F., 71, 244, Φ , 73f 136, 137,
308.
Persian, general, 67, 72.
words in Arabic, 106 ff.
words in Hebrew, 104, 109.
Personal names, 22 f., 29, 47, 105 f.,
145 f., 181-4.
Peshat, 60, 65.
Peshitta, 16, 18, 21, 53 n., 74, 123,
174, 230, 234, 243, 249, 258 f.,
265, 270, 281, 296.
caiques in, 263 ff.
Philo, 49, 208.
Phoenician, 36, 92, 101, 112, 196.
Phonemes, 78, 83, 98, 178, 195, 210,
215, 218 n., 275.
merger of, 78, 126-9, 149 f., 174.
varying frequency of, 87.
marked in writing, 195 f., 198 f., 210.
phonemicization, 102.
Phonetics, 81, 101, 180, 200 ff., 206,
208, 211.
Pirn, 225, 285 f.
Pitch accent, Greek, 195, 201.
Piyyutim, 152, 228, 230 f.
Place-names, 286 f.
Polyglot ideal, 296, 298.
Polysemy, 23, 142 ff., 147.
Pope, Μ. H., 85 n., 96, 36, 106, 139,
170, 229, 321, 323, 333.
351
Porter, J. R., 152 n.
Praetorius, F., 97 n.
Pregnancy, 286.
Prepositions, 176 f.
Probability, 191, 292.
Productive and non-productive roots,
160 f.
Pronoun suffix, 2nd person, 215.
Pronunciation of Hebrew, 199-203,
214 f.
Proto-Semitic, 27, 37, 78 f., 81 f.,
96η., ιο ί, ii3 , 128, 167, 185, 293·
Proverbs, character of Greek version
of, 256-9, 284.
Psalterium Gallicanum, 213, 249, 253.
Psalterium iuxta Hebraeos, 213, 249,
253, 262.
Punctuation, see Vocalization.
Qalir, 231.
Qàmüs, 117, 119, 285.
Qaraism, 59 f., 206.
Qere and kethibhy 30, 228 n., 246 f.
Qoheleth, 100, 123, 147 f., 153, 206.
Quine, W. V. O., 291 n.
Qumran, 1, 7, 39-42, 112, 120 n., 182,
191, 193 ff., 199, 207, 246, 249,
262, 284, 300.
new words detected in, 225 f.,
284 f.
Quran, 59, 117, 224.
Rabbanites, 206.
Rabbi, 56.
Rabbis citing other languages, 56 ff.
Rabbinic interpretation, 44-50.
Rabin, Ch., 30, 41 n., 42 n., 100, 102,
104 n., 205 n., 226 n., 232, 249, 286,
12 f., 25 f., 49, 88y io g y 141, 174,
i8 o y 225, 234, 261y 288y 296,
304 f.
Rabinowitz, I., 39 n.
Rahlfs, A., 248.
Rankin, O. S., 120 n.
Rashi, 55, 59, 66, 174, 234.
Rask, R., 86 n.
Reading, modes of, 196 f.
Reference and information, 118, 291 f.
Reichelt, H., 132 n.
Reider, J., 15, 33, 98, 107, 284, 2, 5,
19 ff., 64, 76, h i , 176, 189, 203,
216, 239, 249, 256, 330.
Religious factors, 43, 282-5.
Restorations in later Hebrew, 227-32.
352 IN D E X OF SUBJECTS
Retroversion, 238 f. knowledge of Hebrew, 208, 240 n.,
Reuchlin, J., 66. 267 ff.
Richter, E., 138 n. favourite words in, 251 if.
Roberts, B. J., 8, 211 n., 213 n., 214 n. obscurity of sense of, 262 f.
Robins, R. H., 83 n. use of Greek words similar to
Robinson, T. H., 290 n. Hebrew, 58.
Romance languages, 36, 62, 77 f., examples cited, passim.
185 n. Sexual reference, 20 f., 277, 286.
Root as sign of basic meaning, 160 f., Shaphel, 233.
197 ff. Shiloh, 120.
Rosenthal, E. I. J., 60 n. Sibilants, 78, 129.
Rosenthal, F., 126 n., 118. Siddiqi, A., 108 n.
Rossi, G. B. de, 1. Sign systems, 292.
Rudolph, W., 16, 140. Siloam inscription, 195.
Rûzicka, R., 129 n. Simon bar Kosba, 40.
Sirach, 39 f., 219, 224 if., 232, 241,
Saadia, 60 ff., 175, 231, 264. 260 f., 300.
Sacra philologia, 68. Skinner, J., 31, 57 n., 270 n.
‘Sailor’, 109 f. Smend, R., 261 n.
Samaritan, 67, 258. Social systems, terminology for, 172.
ostraca, 182. Socotri, 294 n.
Samson, 21, 152. Soden, W. von, 115, 121 n., 153,
Samuel, 47 f. 192 n., 34.
Sanskrit, 12, 47 n., 131 f. Sogdian, 297.
Saul, 48. Sound, sound changes, 76, 81-7, 89.
Saussure, F. de, 83, 274 n. constancy and exceptions, 82 f.,
Scepticism, 301 f. 201 f.
Scholarship and transmission of Source analysis, 68 f., 76.
meanings, 43. South Arabian (including ESA), 22,
Schultens, A., 67 ff. 29 f·, 78, 85, 95, 114, 143, 158,
Schulthess, F., 109 n., 150. 176 n., 182, 230, 234-
Scott, R. B. Y., 27. Spelling, 214.
Scripture, status of, 44, 50, 220. Statistics, 83, 98, 105, 113, i 2 i f . ,
Seeligmann, I. L., 55 n. 130, 269, 299.
Segal, J. B., 103 n. Stenning, J. F., 52, 264.
Segal, M. H., 40 n., 236. Stoicism, 49.
Semantics, 86-91, 143, 172 f., 289-95. Strugnell, J., 131.
a weakness in philology, 88 ff., Stummer, F., 230.
289 ff. Style, general, 29, 49, 63.
semantic rewriting, 302. post-biblical, 232.
semantic development, 109 f., 115, and homonymy, 151-5.
118, 143, 274, 288, and passim. Suffix, 2nd person, 215.
semantic fields, 90, 136-42, 165, Sumerian, 48 n., 92, 103, n o , 119 if.,
170 ff., 180, 227, 260 f., 289, 300. 149.
semantic and phonological com­ Swete, H. B., 55 n., 58 n., 59 n., 208,
ponents in comparisons, 83 f., 211 n., 268.
86 f., 99. Symmachus, 212 f., 233, 244, 254,
Septuagint, date and origin of, 268 f. 259, 262, 271, 281.
influence of Aramaic on, 54 f. Synchronic linguistics, 94, 148.
varying techniques used, 2, 15,53 f., Synonyms, 64, 289.
208 f. Syntax, 31 f., 74, 114, 266.
in Isaiah, 250. Syriac language, general, 59, 67, 95,
modes of analysis of Hebrew, 71 f., 119, 149 if., 160, 234.
208 f. homonyms in, 109, 149 if.
 IN D E X O F S U B J E C T S
verbs tabulated against Hebrew,
162 f., 305 ff-
words cited, 69 n., 86, 108, 122 f.,
141 f. and passim.
for Syriac Bible text see Peshitta.
6 Evpos, 243 n.
Talmud, 43, 45 b, 56 f., 60, 196,
197 n., 213 f., 220, 226, 230, 234 f.
Targum, general, 16, 18, 30, 39, 52-
60, 62, 174, 208, 230, 234f., 243 f.,
252 f., 254 f., 258 f., 262-5, 270 f.,
286.
paraphrastic nature of, 52-5, 270.
influence on LXX, 54 f., 259.
importance for medieval philology,
62 ff.
and caiques or etymologies, 52 ff.,
263 ff.
Textual criticism, textual treatment,
1-13, 68-72, 301-4 and. passim.
textual errors, date of origin of,
218-21.
in Hebrew, 144, 245 ff.
in versions, 247 ff.
Theodotion, 212, 258 f., 271.
Thieme, P., 275 n.
Thomas, D. W., 19, 21 f., 26, 34, 72,
75, 165 n., 233 m, 240 f., 301, 84,
97, n 5, 120, 126, 137, 151-4,
l6?y 190 f., 200, 217 f., 234, 309,
314, 328.
Thumb, A., 201 n.
Tigre, 159.
Tigrinya, 113, 294 m
Tocharian, 92, 297.
Torczyner, H., 45 n., and see T ur-
Sinai.
Torrey, C. C., 146 n.
Transcriptions, 18, 71 f., 129.
Translations in Bible from other
languages, 123.
Translations, modern, 298 f.
Triliteral root, 62 ff., 166, and see
Biliteral theory.
Tsevat, M., 2x9, 329.
Turkish, 77.
Tur-Sinai, N. H., 27, 75, 123, 134^·,
142, 162, 190, 234, 242 n., 303,
29 f., 39 f., 65, io 2 t n o , 119, J79,
187, 193, 195, 209, 220, 273 f.,
277 f., 292, 307, 327, 331.
353
Ugaritic, general, 36, 38, 75, 85, 92 f.,
11 fy 114m, i58f.,222,22cf, 240n.,
268 n., 286 f., 294, 296 f., 303 n.,
304-
homonyms in, 149.
parallelism in, 282.
script, 198.
graphic errors in, 191, 194.
b ‘from’, 175 if.
words cited, passim.
Ullendorff, E., 2 6 , 2 9 n . , 5 7 , 8 5 , 1 0 7 n.,
1 1 3 , 1 1 9 , 1 3 9 n., 1 5 8 , 1 9 7 n., 2 9 4 n.,
5, 1 0 4 ‫־‬, J 9 J , J 9 6 , 2 0 0 , 2 9 5 , 310.
Ullmann, S., 2 7 4 n., 2 7 5 n., 2 9 1 n.
Verb, homonymy in the, 131 ff.
verb-classes, 131 f.
verbs as loan-words, 104 n.
Vermes, G., 49 n.
Vemer, K., 84.
Versions, general, 2, 5, 73, 144, 238-
72, 289.
their mode of *reading’, 238 f.
influence on each other, 259 f.
stereotypes in, 252 f.
and grammar of original, 265 f.
imprecise renderings, 52-5, 249-
5 3 , 2 5 5 ‫ ־‬8 .
uncertainty of sense, 262-5.
and see Septuagint etc., and
Erroneous translations.
Vision, Prophetic, 283 f.
Vocalization, general, 188-222, 289.
existed before marked in writing,
195 ff., 207-17, 221.
evidence for pre-Massoretic, 207-17,
233 n.
age and reliability of, 73, 220 f.
interdependent with textual history,
218 f.
not based on semantic decisions,
202-7.
and analysis of Hebrew grammar,
36, 63, 65, 216 f., 221 f., 289.
alleged dispensability of, 35 f., 67,
73, 188 ff.
inspiration of, 66 f.
ambiguity of term, 196.
examples of emendation, passim.
Vollers, K., 55 n.
Vowels, as modifiers of basic sense, 197.
Vulgate, 18, 3 0 , 59, 140, 174, 230,
2 3 3 f·, 2 4 3 n., 247, 258 f., 262, 270,
286, and see Jerome.
354 IN D E X O F S U B JE C T S
Wagner, M., 121. Wright, G. E., 33 n.
Wallenstein, M., 231. Writing systems, Semitic, 198 f.
Walton, B., 296 n. Latin and Greek, 199, 215.
Waw consecutive, 217. Wurthwein, E., 8.
Weak consonants, 63, 96, 189. Wutz, F. X., 55 n., 71 f., 107, 248 n
Wechter, P., 64 n., 98 n., 174 n. 244.
Wehr, H., 175 n.
Yahuda, A. S., i n .
Weil, G., 173.
Yalon, H., 226 n.
Weil, G. E., 66 n., 204 n., 205 n.
Yannai, 230 f.
Weinreich, U., 42 n.
Yemenite Arabic, 100.
Weiss, A. von, 42 n.
Yemenite Jews, 217 f.
Wellhausen, J., 68 f., i n , 144.
Wernberg-Moller, P., 225 n., 226 n., Ziff, P., 274 n.
241. Zimmermann, F., 227.
Wild, S., 116 ff. Zimmern, H., 121.
Word divisions, 189 f., 209, 267. Ziudsudra, 48 n.
Word-formation, 63, n o , 131 ff., 275. Zoroastrianism, 74.
 POSTSCRIPT

L o o k in g back over nearly twenty years, the author can hardly fail
to be satisfied with the effect that this book had. For it can scarcely
be doubted that it succeeded in its central aim, which was to
introduce an element of systematic and critical reflection into the
proliferation of novel identifications of Hebrew words, supposedly
based upon the methods of comparative philology. That such
identifications could be right, that the method could indeed work
successfully, remains true, and I myself never doubted it. But the
number of such identifications suggested, their constantly increasing
proliferation, and the frequent contradictions between one such
solution and another could not but lead to a deep scepticism in the
end. The inventive virtuosity of those who made the discoveries
could not prevail against a cool and methodical evaluation.
Comparative Philology and the Text o f the Old Testament was
published, as it turned out, near the crest of the wave of philo­
logical innovations, just a short time before that wave began to
break and disintegrate. The rise and fall of the fashion can be well
traced, among other places, in the modern translations of the
Bible. It was very odd that G. E. Mendenhall criticized my book
for failure to consider whether the examples quoted had been
accepted in recent translations.1 I could not refer to the New
English Bible, for its Old Testament had not then been published
(it appeared in 1970). But everyone who was in contact with
scholarship at the time knew that the NEB would be full of exactly
the sort of examples I reviewed in this volume, and so it was.2 The
launching of new philological proposals upon the public through
translations—often before the scholarly world had accepted them,
or had even heard of them—was equally manifest in the work of

1 George E. Mendenhall, ‘Debate over Linguistic Methodology’. Review of

Comparative Philology and the Text o f the Old Testament, by James Barr.
Interpretation 25 (1971) 358- 62.
2 For my own discussion of this version, some years after it appeared, see my
article, ‘After Five Years: A Retrospect on Two Major Translations of the Bible’,
Heythrop Journal 15 (1974) 381- 405, in which I discuss it jointly with the New
American Bible, which appeared in the same year.
356 POSTSCRIPT
M. Dahood, most notably in his translation of and commentary
on the Psalms in the Anchor Bible.3 Translation and the innovative
philological method seemed then to be closely linked in practice.
But this was the highwater mark. Since then, if the production
of new translations forms an index of scholarly opinion, scholar­
ship has generally become much more cautious in the acceptance
of novel philological identifications. The New American Bible, in
spite of the very considerable textual and philological expertise
that went into it, was much less a prey to the semantic emendation
of words than the New English Bible. And the more startlingly
innovative renderings of NEB found few defenders: if at some
future time a revised edition is published, one may surmise that
many of these renderings will prove to have been eliminated.
Confirmation of this trend comes from the fine version of the
Jewish Publication Society of America.4 The translators included
some fine philologists, who in principle probably fully supported
the claim that the comparative philological method worked, that it
could produce effective and convincing results. Nevertheless in this
respect the version, as it turns out, is distinctly semantically con­
servative: it tends to look for a meaning within Hebrew, rather
than one that can be traced from another Semitic language. Look,
for instance, at its rendering of Cant. 6. 4:
You are beautiful, my darling, as Tirzah,
Comely as Jerusalem,
Awesome as bannered hosts.
—not so very distant, after all, from the King James Version of the
same. But another feature of the JPS version is its modesty about
its ability to solve the problems. Its footnotes are peppered with
the telling phrase: Meaning o f Heb. uncertain. It shows no quick
confidence that some source like Ugaritic will let the meaning be
read off; in spite of high philological expertise, the meaning of the
Hebrew remained uncertain.

3 Mitchell Dahood, Psalms: Introduction, Translation, and Notes, 3 vols. Anchor

Bible, vols. 16, 17, 17A (Garden City, N.Y.: Doubleday, 1965, 1968, 1970).
4 A New Translation o f the Holy Scriptures according to the Masoretic Text,
3 vols. (Philadelphia: Jewish Publication Society of America, 1962, 1978, 1982;
2d ed. of vol. 1 appeared in 1967). All three volumes were revised and combined in
a single volume entitled Tanakh: A New Translation o f the Holy Scriptures
according to the Traditional Hebrew Text (Philadelphia: Jewish Publication
Society, 1985).
 POSTSCRIPT 357
And the change of mood among scholars concerned with Bible
translation on the highest standards was obvious, as we come
closer to the present day, in the international committee (chaired
by D. Barthélemy) that produced a report for the United Bible
Societies.5 The report was prepared by a group of scholars, most
of whom had been actively concerned in the major modern biblical
translations in several languages. But from the beginning they lay
down their conviction that comparative philology will be useful
only occasionally and with great reserve. There are indeed possible
parallels unknown till now; but these evidences are rare and
dispersed. Our grasp on the material of languages recently dis­
covered is itself dependent on our knowledge of biblical Hebrew.
The committee therefore resolved to make only a very circumspect
use of comparative philology in the interpretation of the Old
Testament.6
This shift of mood has another side. One of the claims of the
comparative philological method, as it had been exercised in Old
Testament studies, was that it preserved the text and enabled the
scholar to avoid emendation; indeed, this was a methodological
necessity for it, for only if the text was right did it have a starting
point in the search for a parallel in some other Semitic language.
Philological discovery was thus to be much more important than
textual criticism. But the philological pioneers seldom adhered to
this principle with much consistency. The New English Bible con­
tained a very large number of departures from the Masoretic
Text.7 The New American Bible had a similar and extensive list.
The Jewish Publication Society version, though for understand­
able reasons claiming to translate ‘according to the traditional
Hebrew text’,8 gives at times the impression that the translators
know quite well that they can make good sense only by implying
or preferring some different reading. In fact, therefore, some of the

5 To date only two volumes of a projected five have appeared (covering Joshua,
Judges, Ruth, Samuel, Kings, Chronicles, Ezra, Nehemiah, Esther, Isaiah, Jeremiah,
and Lamentations): Dominique Barthélemy, ed., Critique textuelle de lAncien
Testament. Orbis Biblicus et Orientalis, vols. 50/ 1, 2 (Fribourg Suisse: Editions
Universitaires; Gottingen: Vandenhoeck and Ruprecht, 1982-).
6 Barthélemy, Critique textuelle, vol. 1, xiv.
7 They are listed in L. H. Brockington, The Hebrew Text o f the Old Testament:
The Readings Adopted by the Translators o f the New English Bible (Oxford:
Oxford University Press; Cambridge: Cambridge University Press, 1973).
8 Note the use of this phrase in the subtitle of the combined edition (see n. 4).
358 POSTSCRIPT
major translations produced in the sixties and seventies, however
much they professed opposition to the recognition of textual varia­
tion as the cause of problems, may be said to have been, despite
themselves, monuments to the need for serious textual criticism
even among those who would prefer to bypass it through philo­
logical discovery. But any really critical attitude to the text is
likely to make more precarious the point of departure toward
philological solutions. The work of Barthélemy’s committee is
solidly textual in its approach.
Various aspects of these issues come to expression in the three
articles which are printed below and which form a supplement to
the original book. The article ‘Philology and Exegesis: Some
General Remarks, with Illustrations from Job’ was originally a
paper presented at the Journées Bibliques of Louvain in 1972. The
meeting took a series of current problems in Old Testament studies
in which a confrontation of contrary opinions might take place,
and one of these formed a session in which Professor Dahood and
I read papers, both relating to the general approach of compara­
tive philology and referring this to the particular example of Job
3.9 In my own argument I seek to explain, among other things,
why my original book had concentrated less on Dahood’s work—
and less on Ugaritic—than some might have expected. I also show,
in the case of the Punic name Magon, that the evidence alleged by
Dahood from the Phoenician area was actually a complete mis­
understanding of the source. Along with this goes another point:
The position taken by Dahood, far from being well based in
comparative philology, was destructive of comparative philology,
for he went beyond using Ugaritic as a source from which Hebrew
could be elucidated and was interested rather in proving that
Ugaritic and Hebrew were identical; in his own words, it was a
method of ‘virtually equating Ugaritic, Phoenician and Hebrew’.10
I also, with examples from Job 3, look at another aspect: the
problem of the internal relations and nuances that obtained within
Hebrew itself, as distinct from the ‘original’ associations of the
same terms even when these can be determined. These thoughts

9 Both papers were published in the conference transactions: Dahood’s article,

‘Northwest Semitic Texts and Textual Criticism of the Hebrew Bible’, is found on
pp. 11- 37; mine is reprinted on pp. 362-87 below (the source note contains the full
bibliographic data of the original publications).
10 Dahood, ‘Northwest Semitic Texts’, 19.
 POSTSCRIPT 359

come near, in some ways, to what has in recent years come to

be thought of as a ‘canonical’, as distinct from a ‘historical’,
perspective.11
Insufficient attention to Ugaritic? Well, perhaps. I felt, in writing
Comparative Philology and the Text o f the Old Testament, that
the waters of Ugarit had been so muddied by the trampling of
those seeking solutions for the Old Testament that it could not, for
the present and without extensive and intricate argument, be easily
used as secure evidence. My article ‘Ugaritic and Hebrew SBMT
shows some reason why.12 The identification of a word sbm as
‘muzzle’ in Ugaritic went back to the earliest days of Ugaritic
studies, and by the sixties the same was being found in Hebrew.
Far from wishing to dispute this, I began by accepting it as an
excellent example, and intended to use it as a favoured illustration
in this book. Only gradually did I realize that the identification
(based on a rare Arabic word) within Ugaritic itself was almost
certainly wrong. The article shows the complicated series of con­
siderations that led to this realization. Within Ugaritic studies
themselves the article has not been without effect. In J. C. L.
Gibson’s revision of G. R. Driver’s Canaanite Myths and Legends,
the offending sbm ‘muzzle’ has disappeared from the vocabulary,
and the form is taken as from sby ‘captured’ (which brings us back
rather closer to the traditional Hebrew of Ps. 68. 23).13 Incidentally,
the article on this word exemplifies something that I came to feel
more and more strongly: namely, that a suggestion of a new
meaning for a word, on the basis of cognate relationships, was not
something that could be established in a note of four or five lines.
On the contrary, in many cases a rather full article of twenty pages
or more would be needed if all the relevant considerations were to
be taken into account.
Behind all these questions there lies the more theoretical question:
How far are we justified in expecting that lexemes and their
semantic indications will be shared through the various members
of the Semitic language family? For, obviously, the higher the

11 On these problems in general, as related to the text of Job, see also the more
recent work of Lester L. Grabbe, Comparative Philology and the Text o f Job: A
Study in Methodology (Missoula, Mont.: Scholars Press, 1977).
12 Reprinted on pp. 388-411 below.
13 J. C. L. Gibson, Canaanite Myths and Legends (Edinburgh: T. & T. Clark,
1978), 50 n. 5.
360 POSTSCRIPT
degree of this sharing, the more likely it is that comparative phi­
lology will produce good suggestions for particular problems in the
Hebrew text; and, conversely, the lower the degree of this sharing,
the less likely it will be that resources for such suggestions will
exist. I had already addressed some aspects of this question in
chapter 7 above, and I returned to it in the paper on ‘Limitations
of Etymology as a Lexicographical Instrument in Biblical Hebrew’,
which was read in various forms to various groups during the
seventies, such as the Cambridge University Oriental Society, and
finally to the Philological Society in 1979.14 Though I have
described the problem as a theoretical one, the handling of it is
based very largely within practical operations. There are certain
aspects of Semitic languages that encourage a fascination with
etymology and similar relations, and this fascination undoubtedly
fosters the idea that comparative philological solutions to prob­
lems of vocabulary are likely to succeed. My own practical experi­
ence as a lexicographer of Hebrew, however, has led me in the
opposite direction. As against those who follow a more recent
trend and want to discount comparative information altogether, I
gave reasons, based on the actual functioning of the languages and
upon the practical needs of scholars, why this information con­
tinues to be of positive importance. But on the other hand I had to
register some of the many cases where, even given the best will in
the world for the pursuit of philological research, it seemed simply
to provide no answer, and the lexicographer had to acknowledge
that he had no relevant comparative information at all, or that,
even if he had it, it simply failed to illuminate the relations between
words in the way in which it has commonly been expected to
illuminate them. That this need not be surprising was confirmed
through reference to D. Cohen’s Dictionnaire des racines
sémitiques.15 Another recent publication that should be mentioned
is L. Kopf’s Studies in Arabic and Hebrew Lexicography.16

14 Reprinted on pp. 412-36 below.

15 David Cohen, Dictionnaire des racines sémitiques ou attestées dans les langues
sémitiques, 2 fascicles to date [through GLD1] (Paris: Mouton: 1970-).
16 Lothar Kopf, Studies in Arabie and Hebrew Lexicography, edited by Moshe H.
Goshen-Gottstein (Jerusalem: Magnes, 1976). I had known of Kopf’s works and
used them in my own writing, but only limited parts were available to me. This
collection of Kopf’s previously published articles presents us with a much larger
range of Kopf’s material and thus gives us a much better purchase on the problems
of using indigenous Arab lexicography as a guide to Hebrew vocabulary.
 POSTSCRIPT 361
The text of Comparative Philology and the Text o f the Old
Testament, and of the three other articles, is here reproduced
without any change, apart from the correction of a very few minor
typographical errors.

Oxford, May 1986

 PHILOLOGY AND EXEGESIS*
SOME GENERAL REMARKS, WITH ILLUSTRATIONS
FROM JOB

I do not propose to repeat here the arguments of my book Com­

parative Philology and the Text o f the Old Testament (1968).
What I shall do is the following: First, I shall make some remarks
about the discussion in general, taking into account some points
made by reviewers.1 Secondly, since this is the introduction to a
joint discussion with Professor Dahood, I shall say something
about his position in particular. Thirdly, since he and I have
agreed on a common text for discussion, I shall refer to some
examples from Job 3. Fourthly, I shall try to say something about
the philosophy of the matter and place this debate about biblical
Hebrew within the wider framework of the general modern discus­
sion of language and semantics.
Since the reception of the book has been generally favourable
and encouraging, and since the ensuing discussion has given me
little reason to change my mind about major points, it may be
good to start with one or two questions of general aim and empha­
sis in the work as a whole. The first such points are taken up in
Father Dahood’s detailed review.2

* Reprinted with permission from Questions disputées d ’Ancien Testament:

Méthode et Théologie, ed. C. Brekelmans, pp. 39- 61. Bibliotheca Ephemeridum
Theologicarum Lovaniensium, vol. 33 (Gembloux, Belgium: J. Duculot, 1974).
1 The fuller reviews of Comparative Philology (title thus abbreviated) so far
include: F. I. A n d e r s e n , JBL 88 (1969) 345 f.; J. B l a u , Kiryat Sepher 44 (1968-
1969) 223- 25; F. P é r e z C a s t r o , Sefarad 28 (1968) 321- 26; M . J. D a h o o d ,
“Comparative Philology Yesterday and Today,” Biblica 50 (1969) 70- 79; T. N. D.
M e t t i n g e r , Svensk Teologisk Kvartalskrift 45 (1970) 129- 33; W . L. M o r a n , CBQ
31 (1969) 239- 43; E. U l l e n d o r f f , BSOAS 32 (1969) 143- 48; S. D. W a l t e r s ,
JAOS 89 (1969) 777- 81; P. W e r n b e r g - M o l l e r , JThS 20 (1969) 558- 62. Citations
of these scholars by name in the following will refer to these reviews. I do not take
as serious comment the wild ranting of G. E. M e n d e n h a l l in Interpretation 25
(1971) 358- 62, although points of interest are occasionally touched upon by him.
2 I would like to acknowledge the generally pleasant tone of Father Dahood’s
review; some points where he becomes more scornful, and where this scorn may be
misplaced, will be mentioned in the course of this article.
 PHILOLOGY AND EXEGESIS 363

The first question is about the scope of the investigation. Dahood

remarks, and quite rightly, that my study is devoted mainly to
lexical elements, such as nouns and verbs, for which new meanings
have been identified in modern times; it says much less about
pronouns, suffixes, particles, and grammatical categories such as
tenses. He thinks that if one writes about “philology” one has to
include full treatments of comparative phonetics, syntax, etc. I
recognize that there is a valid point here, and that Dahood’s own
work contains a great deal that consists not of new lexical identi­
fications but rather of interpretations of suffixes and the like. But
there is no issue of principle here. I was not writing a full treatise
on comparative philology, but a study of one type of operation
which is an application of comparative philology to the biblical
text. Within the total impact of this operation, it is the lexical
examples which seem to have the most striking and drastic effect.
Questions of suffixes, tenses and the like are commonly much more
marginal; it is the lexical instances which have made the question
critical for our generation. Dahood’s own work would be very
different in its impact if it offered no new lexical identifications but
only new interpretations of suffixes, syntactical elements, prosodies
and so on. In any case, I did make it clear in the book that non-
lexical instances did occur, and discussed a few of them.3 If it had
been my aim to discuss not the general impact of this kind of
applied philology, but Dahood’s work in particular, the propor­
tions might have had to be different; but this was not my intention.
In itself then there is no real issue here, but it leads on to
something which may perhaps be a real issue, namely Dahood’s
distinction between “the traditional philological approach and the
new method and criteria introduced by Northwest Semitic dis­
coveries of the past forty years.”4 Dahood seems to imply three
things: firstly, that the older scholars worked almost only by lexical
means—the point discussed in the last paragraph; secondly, that
any faults which existed in the older method have been left behind

3 Comparative Philology, pp. 30-33 [above]. The remarks about enclitic mem
have excited some comment; a “perverse attack”, says Andersen. Perverse perhaps,
but not an attack: I do not in fact express any opinion of my own about the
existence of enclitic mem in Hebrew, but only about the character of certain argu­
ments offered.
4 D a h o o d , p. 71.
364 PHILOLOGY AND EXEGESIS
by the newer; thirdly, that the discussion in my book was out of
due time, since the situation described was one now left far behind.5
Of these latter two points, neither seems to me to have sub­
stance. As for timeliness, although the method under discussion is
not a new one, its cumulative effect has quite suddenly become
very much greater, both through the amount of material published
and through the particular fact that so much has been made public
in the form of popular Bible translations.6 As for quality, I do not
share the view that the “new method and criteria”, if we are to
judge by Dahood’s own publications and those of his pupils, stand
at a higher level than the work of an older generation such as
Tur-Sinai, Driver and Winton Thomas. On the contrary, that
work of older scholars has seemed to me to have a much better
and more responsible character, and it is for this reason that most
of the examples I studied in the book were taken from this group.
Though in the end I came to regard many of them as uncon­
vincing, the study of them was a real intellectual stimulus; few of
them were just obviously wrong, and only a careful analysis of the
sources enabled one to form a judgement. The study of them was
of real value. That Dahood should look down on this work of an
older generation as deficient, and that it should often be simply
ignored or disregarded in works inspired or guided by him,7 is very
difficult to understand. It is true that my study paid proportion­
ately more attention to the work of older scholars, and that it cited
less of Dahood’s suggestions than it might have done; but a main
reason for this, I must in frankness say, is that if I had cited more
of the latter I would have felt liable to the charge of picking out
quite obviously fantastic suggestions in order to bring ridicule on
the method as a whole. The fact that this accusation was indeed
made8 confirms that I was right in being sensitive to the possibility
of it. In fact, in my selection of examples for discussion I leant
over backwards to avoid, wherever possible, the citation of sug­
gestions which were obviously worthless, which did not teach some

5 D a h o o d , pp. 71, 79 and passim‘, also the title of his review, “Comparative
Philology Yesterday and Today”.
6 On this see further below, pp. 385 f.
7 Cf. for instance W. McKane’s remarks in his review of W. A. v a n d e r W e i d e n ,
Le Livre des Proverbes, in JSS 16 (1971) 222- 36, especially pp. 232- 34.
8 A n d e r s e n , p. 345.
 PHILOLOGY AND EXEGESIS 365

useful lessons through the study of them, or which might seem

designed simply to bring the method into ridicule»9
Thus, if my book gives less attention to the work of Professor
Dahood than some would expect, this does not have its cause in
any special hostility towards his work.10 On the contrary, I have no
consciousness of ascribing to his suggestions any value different
from that which is ascribed to them by the vast majority of Old
Testament scholars. This is a powerful fact which has to be faced
by Dahood—not because majority opinion determines what is
true, but because the final court of appeal for so many of his
suggestions lies not in Ugaritic or in Phoenician, but within the
Hebrew text o f the Bible itself the final criterion is their alleged
concinnity, their alleged improvement of sense, in the Hebrew
Bible, and it is of this that Old Testament scholars are the proper
judges. If scholars are sceptical of Dahood’s suggestions, as I
believe the great majority of them are, this is not to be attributed
to any opposition to the use of Ugaritic or any distrust of “the
Northwest Semitic method”, but to their experience of the way in
which these suggestions have been made and supported: their
experience of wildness in the use of evidence, extreme bad taste in
respect of that which constitutes a probable understanding of the
text, obsessive attachment to the discovery of a Ugaritizing solu­
tion at any price, absence of proper justification,11 and exaggera­
tion of the difficulty of the present text and/or meaning in order
at all costs to secure a new understanding, said to be based on
Phoenician or Ugaritic.12 For—and this must be emphasized—the
logically decisive step in operations like those of Dahood, and in
the process of evaluating them, often lies not in the Phoenician or
Ugaritic facts but in the fitness of the proposed interpretation for

9 Plenty of such suggestions, however, can be cited; cf. below, pp. 369 if.
10 A n d e r s e n , ibid., uses the phrase “ill-concealed hostility towards the work of
Fr. Mitchell Dahood.”
11 See recently for instance C. J. Labuschagne’s review of A. C. M. B l o m m e r d e ’s
Northwest Semitic Grammar and Job, in Ugarit-Forschungen 3 (1971) 373- 74.
12 On all these points McKane’s remarks in the review quoted seem to me to be
right in themselves and also to express what is held by the central current or
scholarship. For a view which stands at the absolutely opposite extreme from
Dahood’s, cf. J. F. A. Sawyer’s remarks on the generally lucid, intelligible and
meaningful character of the Masoretic Text, Semantics in Biblical Research (1972),
366 PHILOLOGY AND EXEGESIS
its context within the Hebrew biblical text. These points will be
illustrated below.
There is however one particular point about Dahood’s approach
which I had not sufficiently taken into account in the writing of my
book, and which to me seems even more unclear now than it was
then: is Dahood’s method really intended to be a comparative
approach at all? I have devoted my investigation to those explana­
tions of Hebrew where meanings are derived from the meanings of
forms in a cognate language, i.e. a language related but different.
Dahood’s position, at least sometimes, seems to be rather that
Ugaritic is Hebrew and Hebrew is Ugaritic; the two are the same
language in somewhat different temporal and local manifestations.13
His method then is not really comparative, in the sense in which I
have used the word, but consists of internal elucidation within the
unitary Ugaritic-Canaanite-Hebrew world. The demonstration of
this unitary (though diversified) linguistic (and also cultural)
mélange is Dahood’s real interest in the matter. When he contrasts
his method with that of the older scholars, it is at least possible
that he is trying to say that their method was really a comparative
one, working with languages which are recognized to be cognate
but different, while his is a non-comparative one, working with
internal relations which at most are only dialectal.
If this is indeed Dahood’s intention, then two things follow:
firstly, I failed to give proper attention to this (reasonably enough,
since Dahood himself does not make it clear); and secondly my
arguments do not deal, and were not intended to deal, with that
segment of his work which is intended to be not comparative but
internal in its method. But, on the other hand, even if Dahood’s
work with Canaanite languages is not comparative but internal,

13 It is, of course, perfectly conceivable that Ugaritic and Hebrew should be “the
same language” in this sense; and if this is so, then of course it is so. Dahood’s work
can be represented as an exploration of the hypothesis that this is so; but the
exploration is carried out in such a way as to override as far as is possible all the
evidence that might indicate that it is not so. His presentation of evidence is not
designed to assist discussion of the question whether Ugaritic is thus related to
Hebrew or not; it does not leave such questions open. Any real comparative
discussion is difficult, since there is very little in Ugaritic of which Dahood will
grant, even for the purpose of discussion, that it does not exist in Hebrew, and vice
versa. This is, incidentally, one reason why I published a comparative table of
Syriac and Hebrew verbs rather than one of Ugaritic and Hebrew (cf. M o r a n ,
op. cit., p. 241): under the present circumstances, one would not have an adequate
agreed basis for the setting up of the latter table.
 PHILOLOGY AND EXEGESIS 367

there is no doubt that he in fact shifts the whole time back and
forward between comparative and internal operations, and that the
comparative segment of his work is affected by, and subject to, my
arguments just as before.
This brings us back conveniently to a point of principle, namely
that the validity of comparative philology as a discipline is not at
all in question, at least from my side. It is the logic of comparative
philology that is taken for granted, with only slight modifications,
throughout my investigation; and if solutions are found to be
deficient it is because comparative philological method, when
properly examined, shows that there is evidence against them. My
purpose was to state properly the methods required by compara­
tive philology, so that readers might be better able to judge whether
suggestions, allegedly thrown up through the use of this method,
were in fact validated by it. The notion, pursued in Wernberg-
Moller’s review, that I want to “dismiss” comparative philological
method or to erect a “new edifice” to take its place, has very little
to do with my purpose.14
Our subject here however is not comparative philology itself but
a sort of applied comparative philology. The basic work of com­
parative philology is, let us say, to establish a picture of the
Semitic languages and their interrelations through the joint use of
data from the various branches of the family. This basic philo­
logical work provides the fundamental logic for the task which we
are discussing. But what we are talking about is not itself com­
parative philology in that sense: rather, it is an applied and heuristic
operation, dependent on (or at worst parasitic upon) comparative
philology. It is an operation which appears to use the insights of
comparative philology but can very easily override them. Those
whose prime interest has been to reconstruct a hitherto unknown
stage of Hebrew or its prehistoric relations with Ugaritic have not
necessarily observed the proper needs and rules of comparative

14 See W e r n b e r g - M o l l e r , p. 559. His picture (especially on pp. 560 and 562 of

his review) of my ideas about the relation between philology and linguistics is wildly
remote from my actual opinions and often attributes to me the opposite of what I
think; many of his notions are quite expressly excluded by clear statements of mine.
Cf. also my “The Ancient Semitic Languages—the Conflict between Philology and
Linguistics”, Transactions o f the Philological Society, 1968, 37- 55; also, in general,
my paper “Hebrew Lexicography”, to be published in the proceedings of the
Florence Colloquium on Semitic Lexicography, held at Easter 1972.
368 PHILOLOGY AND EXEGESIS
philology (or historical linguistics, as it might be better called); and
the principles of that discipline need to be reasserted and to be
protected against being pushed aside in the impetus of the rush to
identify and to reconstruct. Several sections of my book were
intended to do just that; one instance is the treatment of the
phonological correspondences.15
One of the aspects of comparative philology which needs to be
emphasized is the need for each language to be seen and under­
stood for itself. The study of Semitic languages was long damaged
by the enslavement of each one to the ancillary service of explain­
ing the text of the Bible—Arabic was a principal case in point. The
present interest in Ugaritic is something which in itself should be
approved and supported in every possible way. But the frenetic
desire to exploit its resources as a means of explaining the Hebrew
Bible threatens to hold Ugaritic study within the same framework
of slavery to sacra philologia from which other Semitic language
studies, such as Arabic or Akkadian studies, have been liberated.
While Akkadian studies have long established their freedom from
a status ancillary to biblical study, Ugaritic studies still suffer
damage from the Entdeckerfreude of those who have exploited
them for the elucidation of the Old Testament. I would set it forth
as a good principle: if you want to know something reliable about
Ugaritic, ask an Ugaritologist who is interested in looking at the

15 F . I. A n d e r s e n , op. cit., p. 345 f., regards this as a falling back on neo­

grammarian dogma. No “dogma” is involved. If there are correspondences other
than those which have been recognized as normal, then of course there are, and that
is all there is to it. Andersen must be willing to use his terms very loosely if he
classes the linguists cited on p. 83 n. of Comparative Philology [above] as “neo­
grammarians”. I make it entirely clear in the book, pp. 83 f., that there is a
difference between accepting the existence of such abnormalities and taking them as
a basis for identification where ex hypothesi the semantic component is the quantity
to be discovered. Cf. also Transactions o f the Philological Society, ibid., pp. 48 f.
Meanwhile D a h o o d , op. cit., p. 75 waxes scornful over the discovery of nbs for
nps in a Hebrew ostracon. The scribe, he says, mocks my warnings on this matter.
What is mocked is Dahood,s ability to understand an argument. Though the
empirical finding of nbs had not occurred when my text was written, the entire
argument was written in order to provide for this sort of possibility, and only
factual details, but nothing in the structure of the argument, require to be altered.
From another side, cf. the comments of Mettinger, p. 130b. Dahood’s argument
becomes totally unintelligible to me when he alleges with some heat that I,
“hamstrung by textbook theories” (what textbook?), “cannot bring myself to admit”
that pairs like lbs and Ips, nbs and nps “are non-phonemic variations.” I
never supposed they were anything else.
 PHILOLOGY AND EXEGESIS 369
material for itself, and not at the material as a quarry from which
new quick identifications of Hebrew meanings can be dug. The
position I take, far from seeking to ignore or diminish the impor­
tance of Ugaritic, is interested in the valuation of Ugaritic as a
quantity in itself. The haste to establish the maximum possible
equation between Ugaritic and Hebrew can be criticized not only
for creating Hebrew in the image of Ugaritic but also for creating
Ugaritic in the image of Hebrew.
The same is true of Phoenician, and an example from it (actually
from Punic, but this makes no difference) will illustrate some of
the points which have been made above. Dahood complains that I
do not discuss any Phoenician form;16 let us see what happens
when we do discuss one.
One of the words identified by Dahood is a Hebrew noun magan
“suzerain, sovereign”. It is discussed on p. XXXVII of the Intro­
duction to his Psalms /, and more fully on p. 16, with reference to
Ps. 3. 4. He translates this as:
But you, O Yahweh, are my Suzerain as long as I live . . .
This involves, of course, a departure from the traditional under­
standing of Hebrew magen as “shield”. Thereafter the same sug­
gestion is repeated several times: at Ps. 7. 11 (“My Suzerain is the
Most High God”; see note on p. 45, there qualified with “perhaps”);
at 18. 31 (note on p. 114), 47. 10 (pp. 286 f.), 59. 12 (but here the
understanding “shield” is retained by Dahood, with only the pos­
sibility of “suzerain”, see Psalms //, p. 72), 84. 10, 12 (ibid.,
pp. 282 f., with note alleging that this rendering demolishes the
interpretation of “the anointed” in The Jerusalem Bible). All of
these go back to the same starting point, the notes on pp. XXXVII
and 16 f. On p. XXXVII we hear that the reading as magan is
“based on the Punic name for 6emperor’, magori”; and on p. 17 we
read the following remarkable assertion:
In Punic the Carthaginian generals are given the title m a g o n , which
Latin inscriptions reproduce by im p e ra to r , “suzerain”, or d u x ; see Louis

16 D a h o o d , p. 71. Dahood’s assertion is entirely incorrect in point of fact; he

seems to have counted only the five mentioned under “Phoenician” in my index,
which refer to general statements about that language; actual words discussed are
entered in the index of examples, and cf. also under “Canaanite” Some Phoenician
370 PHILOLOGY AND EXEGESIS
Maurin, “Himilcon le Magonide. Crises et mutations à Carthage au
dé but du IVe siè cle avant J.-C.”, S e m itic a 12 ( 1962 ) 5 - 4 3 .
Let us then look at the article of Maurin, from which such
drastic effects upon the Hebrew Bible have been derived. The
article concerns an important family who are known as the
Magonids, so known because they are descended from one whose
name was Magon—the normal usage, of course, for noun forma­
tions in -id. The genealogy is set out by Maurin on p. 13. Magon is
the first name, the descendants have equally familiar names like
Hasdrubal, Hamilcar, Hannibal.17 In about the fifth century, as
Maurin puts it, the known members of this family “seem to have
been regularly invested with the generalate at Carthage” (p. 13).
The passage upon which Dahood appears to be relying (for there
is no other in the article which is to his purpose) is on p. 16:
Les titres des Magonides se retrouvent eux aussi de Malchus à la fin
du IVe siè cle. Les textes latins emploient avec constance ceux d H m p e ra-
to r ou de d u x pour qualifier les gé né raux carthaginois, et une seule fois
le mot “dictatures” est utilisé par Justin à propos d’Hasdrubal, le fils
de Magon.
Maurin then goes on to discuss the titles used in Greek texts,
where for instance Diodorus often distinguishes between the title
of king and that of general. This last is the point that Maurin is
discussing. The question is whether a man, whose name might be
Hasdrubal or Hannibal, has the title (and office) of king, or that of
general. With this question in mind Maurin considers the name
Malchus. Does this word mean “king” and function as a title, or is
it just the personal name of the man involved? Maurin thinks it is
the latter:
Le nom de Malchus ne repré sente pas forcé ment le titre royal lui-
mê me, comme on le croit ordinairement, mais il peut en ê tre un
composé ; d’ailleurs les transcriptions antiques de ce nom sont variables;
on pourrait peut-ê tre le comparer avec le nom de Magon, qui se rat­
tacherait à une racine signifiant “protecteur”, et qui n’est pas un titre.18

instances discussed, such as stc on pp. 180, 294 etc., are among the suggestions
treated most favourably.
17 The persons concerned as well known and are fully described in any standard
history.
18 M a u r i n , op. cit., p. 16, n. 2; I have omitted the references to other literature
which are included in parentheses.
 PHILOLOGY AND EXEGESIS 371
In fact one sees that everything said about Punic magon by
Dahood is wrong. The word is a personal name, and not a title.
Carthaginian generals were not given magon as a title; either it was
their personal name before they became general, or it was not.
Latin imperator or dux is not a reproduction of magon, but a
designation of the office of general held by people with names like
Hannibal or Hasdrubal, who happened to belong to the family of
the Magonids. The passage has nothing to do with a reproduction
of Punic words in Latin inscriptions; it is a matter of terms used by
Latin (or Greek) historians to designate the office held by certain
Carthaginian leaders. The idea that magon is a title, or that it is
“the Punic name for ‘emperor’”, is not only not supported, but is
expressly ruled out as self-evidently untrue, by the source quoted
by Dahood in favour of his interpretation.19
In fact the sense of the name Magon is well known: it comes
from a very common word meaning “give”, perhaps “offer, deliver”.
This occurs in Hebrew but is rather rare. In Punic on the other
hand it is very common. The overwhelmingly probable meaning of
the name Magon is “he has given (a son)”.20 It belongs to the same
common type as the Hebrew name Nathan. Far from being ren­
dered by Latin and other bilingual inscriptions as “emperor” (!), it
is “rendered” simply as the name Magon, and very numerous
examples can be found.
Thus Dahood’s entire construction of a sense “suzerain” for
m-g-n in Hebrew lacks any basis in Phoenician-Punic evidence,
and is contradicted by the sources to which Dahood appeals. This
is not a matter of evidence in existence, which might be interpreted

19 The only mistake in Maurin (whom I take to be a classicist rather than a

Semitist) is that he renders Magon as protecteur, implying the root g-n-n “protect”,
present in the familiar Hebrew magen “shield”. This does not make any difference
for our purpose: he rightly takes it as obvious that the familiar personal name
Magon is a personal name and not a title, and argues from this that the term
Malchus is also a personal name, i.e. that it does not imply that its bearer was
called king. On Magon, Maurin refers to the article of J. G. F é v r i e r , “Para-
lipomena Punica”, Cahiers de Byrsa 6 (1956) 13- 27, see p. 21; but this makes no
difference to our question: Février is simply discussing whether a writing apparently
mgnm should be read as the name Magonam or as a noun “protector”, with the
reading mgnn.
20 For instance, D o n n e r - R ô l l i g , K A I ii. p. 67: “Er hat (einen Sohn) geschenkt”.
For an instance of the rendering of this name as the same name in another
language, cf. the Numidian text of K A I no. 101, and the Latin forms as exemplified
in K A I ii. p. 67.
372 PHILOLOGY AND EXEGESIS
one way or another: the entire construction built up by Dahood,
and repeated at several places in the Psalter, rests on mere mis­
understandings and misconstructions. It looks as if some one has
misinterpreted material registered on a card index; and this of
course can happen to anyone, and is nothing to worry about. But
there is a more serious side to the matter: considering the extremely
high frequency of the name Magon in Punic texts and the plentiful
opportunity given to students to observe its usage, I find it impos­
sible to understand how a misconstruction such as this could have
been believed and accepted by anyone actually at home in the
study of Punic texts. So perhaps we may hope to hear less from
Professor Dahood about the ignorance and non-appreciation of
Ugaritic and Phoenician on the part of those who do not share his
approach.
To sum up, this instance illustrates how the study of the cognate
sources (in this case Punic) is damaged and corrupted by the zeal
to make them into material for a novel understanding of the
Hebrew Bible. For surely no one can suppose that this whole
construction of magon as meaning “suzerain” and acting as a title
for Carthaginian generals would ever have arisen out of the Punic
sources studied for themselves. It is the presence of the sequence
m-g-n in Hebrew, and it alone, which initiated the whole opera­
tion. Yet in Hebrew itself there is no support for it, since the idea
that God is shield fits perfectly well in all the instances concerned.
The instance illustrates also the method of numerical accumu­
lation which is so prominent in Dahood’s work: the suggestion,
once made, is repeated again and again, and when one asks for the
justification of any single example one is told that numerous other
examples of the same thing have been detected. In this case all the
examples are equally nugatory, the basis for all of them being nil.
Here we see a contrast with the majority of the examples dis­
cussed in my book, in which something is to be learned from the
analysis of the example even if it turns out to be unconvincing. By
contrast, examples like the magan just discussed contribute nothing;
they teach no new lessons in philology, and nothing is learned
from the analysis of them. The process of analysis can only be
negative. Unhappily, it can be repeated over many instances.
As I have said, the final decision in many cases rests with the
taste of the scholar dealing with Hebrew, rather than with Ugaritic
or other cognate evidence. Dahood, for example, thinks it impos-
 PHILOLOGY AND EXEGESIS 373
sible to take Ps. 58. 11 as “he washes his feet in the blood of the
wicked”.21 According to him, one cannot wash one’s feet in blood;
“no clear visual image emerges” from it. Well, I am delighted to
leave it to the judgement of Old Testament scholars. But Dahood
goes on: “One schooled in Ugaritic” would have known that it
meant not washing in blood, but washing from blood: thus “he will
wash his feet of the blood of the wicked”.
So let us look at the Ugaritic text CTA 3, ii. 34 (Gordon cnt,
II, 34), one of those which Dahood cites. The text as a whole
seems to me to be interpretable as follows. Firstly, cAnat’s slaughter
has its climax in her wading in blood (thus lines 13- 14, repeated
more or less in 27- 28). This slaughter goes on until she is satisfied
(line 29). Now (lines 31 f.) something is poured into a bowl,
apparently dm “blood”. After this it goes straight on to say that
cAnat washed her hands bdm dmr. It seems natural to suppose
that this is a ceremonial washing in blood. There follows a scene
with some movement of furniture; and after this, line 38, cAnat
draws water and washes. It is at least feasible that there are two
washings here, one in blood and one in water with dew, and the
lack of repetition of phraseology supports the view that these are
different (contrast the double reporting, in similar words, of her
earlier wading in blood, above). I do not insist that this inter­
pretation is right, but it seems entirely reasonable; and in any case
the view that cAnat washes in the blood of dmr in this passage is
no novelty of mine, but can be found in a number of experienced
workers on Ugaritic,22 though the contrary view (that she washed
her hands from blood) is also found. We may hope therefore to
hear rather less from Professor Dahood about what will be thought
by those “schooled in Ugaritic”.23
It is of course absurd that one should have to appeal to a
Ugaritic text on such a matter, and I do not in fact so appeal; I
simply point out that the Ugaritic evidence itself is very like a good

21 D a h o o d , p . 77.
22 Cf. C. H. G o r d o n , Ugaritic Literature (1949), p. 18: “She washes her hands in
the blood of soldiery”; G . R. D r i v e r , CML, p. 85: “washed . . . her hand(s) in the
blood of the guards”; recently A . S. K a p e l r u d , The Violent Goddess (Oslo, 1969),
p. 50: “She washed her hands in the blood of soldiers”; and not least W. F.
A l b r i g h t , HUCA 23 (1950- 51) 20, 38: “with whose blood she washed her hands”.
23 On the point at issue in all this, i.e. the recognition of b meaning “from”, my
position has the welcome agreement of W. L. M o r a n , pp. 239 f.—he being doubt­
less also unschooled in Ugaritic.
374 P H IL O L O G Y AND EXEGESIS

deal more ambivalent than Dahood supposes. Another such point

may be added: in so far as I understand Dahood’s objection to the
idea of washing one’s feet in blood, he seems to mean that “wash”
means “cleanse”; blood however is not a cleansing medium but a
dirty substance, so that you wash not in blood, but to get rid o f
blood. If this is intended, it is a crassly literalist point of view,
ignoring both the metaphorical nature of the Psalms passage,
which by no means maintains that blood is a good substance for
washing the feet, and also the familiar religious conceptions of
blood as a cleansing agent and of the shedding of blood as a form
of retribution. But, apart from these obvious points, the semantics
of rahas do not necessarily imply cleansing, in the sense of remov­
ing a dirty object; they may imply soaking, splashing, bespattering.
This can be seen in Hebrew (Song o f Songs 5. 12: “splashed by the
milky water”, NEB; Job 29. 6); it applies in Ugaritic (if cAnat
“washes” in blood, it does not imply that she uses blood to cleanse
away some other matter, but that she wets or splashes herself in
blood), and in Akkadian, e.g.:
“[Just as] this chariot with its base-board
is spattered [ r a - a h - s a - t u - u - n i ] [with blood]; just so, [in battle with]
your [enemy], may they spatter your chariots
with your own [blood].”24
These examples must suffice as illustrations of some of the
problems one must feel about Dahood’s work; but it certainly does
not exhaust them. I leave aside such matters as Dahood’s attitude
to the consonantal text, which to me seems to be a compound of a
traditional religious-superstitious reverence for the signs on paper
on one hand, and the pragmatic necessities of his own method on
the other. I leave aside his mode of semantic discussion, which
commonly seems to me to work from the semantics of the English
words used in his renderings, rather than from the Hebrew or the
Ugaritic. These and many other issues must be left aside here.
Basically, my idea of language is entirely different from his, and
also my idea of what constitutes care in the handling of philologi­
cal evidence. Some further general points will be added later in this
article. The following remarks about some words in Job 3 are not

24 Iraq 20 (1958) 75, line 612-15 (D. J. W i s e m a n ). I am grateful to my colleague

M. E. J. Richardson for very helpful discussion of these questions.
 PHILOLOGY AND EXEGESIS 375

of a very controversial character: their main purpose is to illustrate

the complexity of strata with which we have to deal in the analysis
of a Hebrew text, and the unlikeliness that simple philological
parallels will prove to be in themselves decisive.
First of all let us look at the well-known case of the word
salmawet, which occurs in Job 3. 5.25 As everyone knows, the older
tradition took this as “the shadow of death”; but more recent
scholarly tradition has taken a different turn, holding that it is a
word salmut (or ‫־‬o t) of the root s-l-m, with the sense “obscurity,
darkness”. This has now come to be so completely accepted that
some works have ceased to mention that the older tradition of
meaning ever existed. This is surprising, for even in modern times
there has been a current of very significant opinion in favour of the
older understanding. Th. Nöldeke, after all—and whose opinion
could be more weighty in a matter like this?—wrote in 1896 a short
article26 defending both the traditional pronunciation and the sense
“shadow of death”. He pointed out, though he cautiously refrained
from pressing the point, that an early-Islamic poem included the
phrase zill al-mawt, with the sense of “thick shadow”27—in this
case not unpleasant gloom but refreshing shade of trees; he denied
the existence in Hebrew and Aramaic of the root s-l-m “dark”. The
reading as salmawet was again defended by J. Hehn in an article in
1918, which carried farther the line developed by Nöldeke;28 and
the arguments of Hehn proved sufficiently strong to convince Bauer
and Leander, whose detailed grammar records their judgement in
his favour.29 Certainly not all of the arguments of Nöldeke and of
Hehn could stand unchanged today;30 some of the terms in which
the question was discussed in the nineteenth century might seem

25 Cf. also J. F. A. S a w y e r in JSS 17 (1972) 257 if. (review of Holladay’s

dictionary), and his Semantics in Biblical Research (1972), pp. 14, 40, 90.
26 ZA W 17 (1897) 183- 87.
27 Ibid., p. 184; the source is Yäqüt 4, 566, 21. Nöldeke refrained from pressing
the point because he thought that the Arabic phraseology might have undergone
indirect influence from the Old Testament. Some counter-arguments against
Nöldeke,s general position were offered by S. R. D r i v e r {Job, ICC, part ii,
pp. 18 f.), but they scarcely constitute a refutation, as D h o r m e {Job, p. 27) thought.
28 In Orientalistische Studien, Fritz Hommel zum 60. Geburtstage . . . dargebracht
(Leipzig, 2 vols., 1917 and 1918 = Mitteilungen der Vorderasiatisch-aegyptischen
Gesellschaft, 21 & 22 (1916 and 1917), vol. ii, pp. 79- 90.
29 Historische Grammatik, §61t/z, p. 506.
30 My own studies in Hebr. selem “image” {BJRL 51 [1968- 69] 18 and 21) led to
conclusions about this word which differed from Nöldeke’s arguments: I doubt its
376 PHILOLOGY AND EXEGESIS
out of date by now; but, in spite of modifications which would
have to be made, the weight of opinion, especially with Noldeke’s
judgement involved, is still imposing, and it is surprising that these
opinions seem to have been very much neglected.31
It is not my purpose here, however, to argue that the older
tradition is “right” and that the word should indeed be read as
salmawet and understood as “the shadow of death”. My purpose is
rather to show the importance of this tradition for the under­
standing of the word, and thus to show that the philological facts,
even if salmut is the right reading and “darkness” the “right”
meaning, cannot short-circuit the complicated task of unravelling
the tradition of understanding.
Even if we accept that “darkness” is the right meaning, this can
hardly be the end of the matter. The understanding as “the shadow
of death” is very old: out of the 18 or so cases in the Hebrew,
about 10 or 11 are rendered with σκιά θανάτου in Greek. These
include not only some in the Writings, such as in Job itself and in
the Psalms, but also two in the Major Prophets (Isa 9. 1 and Jer
13. 16). Since, on quite other grounds, Job may be a somewhat late
book, at least in its final form, no very great distance in time may
separate its completion from the Septuagintal interpretation of this
word. It is thus quite possible that the sense “the shadow of death”
was already understood by the final redactor of the canonical text
of Job, or was known to him and influenced his thinking, even if it
was not intended by the earliest composer of ch. 3.
This in turn only leads us to further questions. As we have seen,
it is usual to say that the word should be read as salmut, with the
abstract noun-ending -ut. But is it really probable that the tradi­
tion took a word which had previously always been pronounced as
salmut and altered its pronunciation to salmawet, purely in order
to support a midrashic-etymological explanation, and without any
justification whatever in the current phonetics of the word? This
seems to me unparalleled in what we know of the transmission of
the text. Midrashic-etymological explanations were of course many,

connection with an Arabic s-l-m “cut off”—a suggestion made by him only hesitat­
ingly in any case; and I give hesitant favour to the derivation from s-l-m “dark” of
the two instances in the Psalms (39. 7; 73. 20), so that I do not share his confidence
that the root s-l-m with the sense “dark” did not exist in Hebrew and Aramaic
at all.
31 Hehn’s article was published during the war and is somewhat inaccessible.
 PHILOLOGY AND EXEGESIS 377

but were not generally accompanied by a systematic change of the

pronunciation of words for the sake of conformity to the explana­
tion. The case of names including heathen divinities or apparent
obscenities is a different matter from such a change of pronuncia­
tion, if the original form was salmut. One reason, therefore, against
the modern explanation of salmawet is that it makes the phe­
nomenon into something quite isolated and peculiar.
It may be apposite to mention here some places where the
generally accepted sense “darkness” does not fit very well; some of
these were already mentioned by Nöldeke or Hehn. (a) In Job
38. 17 the parallel is with mawet “death”. Hehn maintained that
“darkness” was impossible here. Even if this is going too far, the
sense “shadow of death” seems a strong contender. Note also that
salmawet is in this case a place, a place with doors (or door­
keepers—so NEB with its “the door-keepers of the place of dark­
ness”). (b) In Ps. 44. 20 there seems to be a parallel between “in the
place of tannim” (probably to be read, or understood, as referring
to the tannin or dragon) and besalmawet. “You have covered us
with darkness” seems rather lame and thin after a previous half-
verse which talks of crushing “in the place of dragons”. It is
perhaps because of this consideration that NEB renders with:
Yet thou hast crushed us as the sea-serpent was crushed and covered us
with the darkness of death
—in other words, the recognition of death as a semantic element in
the sentence gives it a proper balance.32
(c) In Jer 2. 6 our word occurs in the description of the desert,
parallel with terms suggesting its quality of waste and dryness.
“Darkness” does not fit well here, at least if one takes it in the
more natural sense, and this would seem best in view of the
other terms in the passage, which are used in very normal ways.
That “darkness” is difficult here seems to have been felt by the
NEB translators, who for MT be-Deres siyya wesalmawet write:
“a country barren and ill-omened”. I do not know precisely
what philological evidence was taken into consideration in ren­
dering “ill-omened”.

32 It is possible, however, that NEB presumed a textual error and read the text as
salmut mawet. The corruption would be easily explained as haplography; but there
is no other example, so far as I know, of such a collocation.
378 PHILOLOGY AND EXEGESIS
I do not wish to claim too much for these examples; but it is
only right that they should be set alongside the well-known cases
where salmawet stands in parallel with words for “darkness” like
hosek or in opposition to words for “light” such as Dor.
Two further aspects seem worthy of attention. One is the inves­
tigation of locutions in which a pattern like “the x of God” is taken
to mean “a very great x ’\ Professor D. Winton Thomas in an
article on “Unusual Ways of expressing the Superlative in Hebrew”
devoted several pages to combinations with mut and mawet.33 He
did not there mention, however, the case of salmawet.
In 1962 Professor Winton Thomas published a full study of our
word, in which he resumed some of the early arguments and
brought to bear also the force of his own studies in mut/mawet as
an expression for the “superlative”.34 He also concluded in favour
of the sense literally “shadow of death”, but in effect “deep, thick
darkness”, and in favour of the form salmawet. The case of this
word is a curious one, in that the majority opinion in modern
times has certainly been in favour of the form salmut and the sense
“darkness” (root s-l-m), but among scholars who have devoted full
independent studies to the word the trend has been in the opposite
direction.
He provided some further discussion in another article in 1968.35
Meanwhile S. Rin had published a note on “The mwt of Grandeur”,
which widened the subject by making reference to the deity Mot
and the work of Cassuto.36 These contributions appear to give
some support to the idea that salmawet, if analysed as “shadow”
plus “death”, nevertheless might—in some cases at any rate—have
had a sense like “very deep shadow”, independently of any referen­
tial component specifying death. If this were so, then the under­
standing as “the shadow of death”, as represented for instance in
the LXX renderings, would be no more than a literal representa­
tion of the original idiomatic phrase. Such a representation leaves

33 VT 3 (1953) 209- 24, especially pp. 219- 22. Quite incidentally, it seems a
little unfortunate that this construction has come to be categorized as a mode
of “expressing the superlative”—this is hardly what is meant.
34 In JSS 7 (1962) 191- 200. I developed the arguments of this present paper
independently of Winton Thomas’s article, which might have made this section
superfluous; in any case my position is not quite the same as his, and I develop the
facts in a different way.
‫ ״‬VT 18 (1968) 120- 24, especially pp. 122 f.
36 VT 9 (1959) 324 f.
 PHILOLOGY AND EXEGESIS 379
it uncertain how far the total sense of the idiom was still appre­
ciated, or how far on the other hand that total sense had now been
replaced by the sum of the independent senses of the two com­
ponent parts. The questions then remaining would be: (a) why the
vowel of the first element is a, instead of the e customary in the
free form of the construct sel “the shadow of . . and (b) why the
whole idiom was written as one word, without word division.
Neither of these is a very serious difficulty.37
Another relevant consideration, which can hardly be ignored
even if it is not clear in what direction it leads, is that salmawet has
a certain similarity to a personal or place name. There are in
Hebrew at least two comparable forms, and these are both proper
names: Ifsarmawet “Hadramaut”, and cazmawet “Azmaveth” or
(spelling of NEB) “Azmoth”. This latter is the name of several
persons in early Israelite history; there was also a place called
Beth-azmaveth. There is good reason to suppose that these names
were compounds including the divine name Mot.38 It is a reason­
able surmise that salmawet could have had some association with
this type of name, (a) This would account for the “compound”
word form and the anomaly of the vowel a\ phenomena of this
kind are frequent in names, (b) The root s-l-l with the sense
“shelter” is of course common in Semitic names; cf. e.g. Tallquist,
Assyrian Personal Names, p. 303b, and in Hebrew the familiar
Bezaleel (and very likely Zelophehad).39 (c) In this case the analysis
of the term into “the shadow of death” would be not only a
linguistic process but also part of the demythologizing, so to speak,
of a name associated with an alien deity and the “shelter” given by
him—the deity becomes ordinary human death, the shelter becomes
shadow.40

37 The argument sometimes made against the form salmawet, namely that “com­
pounds of this kind do not exist in Hebrew”, is a weak one; for no more is
necessarily involved than a graphic peculiarity, i.e. the writing of the entire phrase
without a word-dividing space. But cf. also another aspect, to be mentioned below.
38 On Hadramaut see already G. R. D r i v e r in PEQ 1945, pp. 13 f.; on Azmaveth,
see C assuto, Ha-elah cAnat (Jerusalem, 1953), pp. 28 f., 47 if.
39 Cf. N o t h , Die israelitischen Personennamen, p. 152; Noth seems wrong in
rejecting a connection between s-l-l and the name Zelophehad; a sense like “may the
deity (pahad) shelter, protect” would seem very probable; cf. N o t h , p. 256,
no. 1204. ’
40 I merely mention, for the sake of the information, the pair of apparent names
glmt and zlmt in CTA 4, vii, 54 f. (p. 30), Gordon UT 51, VII, 54. These may be
380 PHILOLOGY AND EXEGESIS
That there was a place with the name salmawet cannot, of
course, be demonstrated. We have already noted, however, that the
word does occur in spatial contexts on some occasions. One piece
of possible support lies in the similarity between “the valley of
salmawet” in Ps. 23. 4 and “the vale of the bakaD” in Ps. 84. 7. It is
commonly held, and with quite good reason, that this latter was
the name of a particular real place. But, whatever the “original”
meaning of the name, it is very probable that from an early date it
was understood to have the suggestion of “Vale of Weeping”, and
this provides the universality required for use in the Psalm context.
This understanding therefore, though found for example in the
LXX with its έν τη κοιλάδι του κλαυθμώνος, is probably already
intended in the Hebrew text of the Psalm. In general, it is likely
that the Ps. 23 passage has played a central part in spreading the
“shadow of death” understanding of our term.
I do not claim to have solved the problem with these considera­
tions, but only to have shown that it is a more complicated one
than is generally supposed, and one in which several different levels
of explanation can be and must be held together in the mind at one
time. It is at least possible that the total history involves: (a) forms
from a root s-l-m “dark”; (b) an idiom where “shadow” plus
“death” meant “deep shadow”; (c) a name of the type “may
Mot give protection”. The eventual standardization of “shadow of
death” would then be not an artificial invention, but a universali­
zation over the entire usage of that which had earlier belonged
only to a part. The phonetic form preserved in MT is not an
artificial invention but has real foundations in the history of the
term.
The case is an example, then, of the complexity of the strata of
tradition and understanding with which we have to deal in the
analysis of a Hebrew text, and the improbability that simple
philological parallels will prove to be in themselves decisive. In
particular, philological information from cognate languages is
something that is to be taken into account, but does not in
itself provide a decision; and the provision of a “correct transla-

relevant but I am not sure how. zl “shadow” is well established in Ugaritic, but zlm
“darkness” is not; this may be relevant to the question whether it ever existed in
Hebrew.
 PHILOLOGY AND EXEGESIS 381

tion” cannot be a correct representation of the dimensions of

the tradition.
From the same verse we can speak more briefly about another
illustration, MT kimerire yom. It has long been customary to
identify here a word “darkness” from a root k-m-r (Syriac kmir
“gloomy, sad; black, dark”)41 and this of course would make good
sense; but is it what the poet wrote? There is evidence in early
post-biblical Hebrew which points in another direction. Sir. II. 4
has Dl tqls bmryry ywm, and in the Qumran Hodayot and Pesharim
we have several phrases like wnpsy bmryry y w m A1 Perhaps it
would be hasty to conclude with Mansoor on this evidence that the
text in Job must be read as “as the bitterness of the day”; it would
still be possible that the Sirach and Qumran texts are restorations
or restitutions, formed within late Hebrew on the basis of the MT
in Job.43 Nevertheless the probability lies with Mansoor’s position:
even if the root k-m-r appears to make better sense, and even if it
“lay behind” an early stage of the Job poem, which is itself unlikely,
it is likely that the final stage of the Hebrew intended the phrase to
be understood as k plus ra-r-r.44 There is, in fact, very little trace of
k-m-r in a sense like “gloomy, dark” anywhere in Hebrew, either
before Job or after. May k-m-r in this sense be an isolated devel­
opment of Syriac?45
Our next instance concerns the importance of parallelisms and
word-pairs, whether traditional or otherwise. At Job 3. 8 it is an
easy suggestion, and one commonly adopted, that instead of MT
yom we should read yam “sea”, giving the parallelism of sea/
Leviathan. If we do this, we may be tempted to go a step farther
and make the verbs parallel also: haying D-r-r “curse” as the first,

41 The Hebrew dictionaries tend to give an impression that the sense “black” is
basic in Syriac. My impression is that the sense “be sad” (of personal emotion) is
much more central than that of colour; for the colour black Syriac uses primarily
other words, especially the root Dkm.
42 See M. M ansoor, VTS 9 (1963) 316 f. The Qumran readings support for
Sirach the reading as cited above, as against the variant kmryry (= MT in Job).
The citation bmryry, above, is the form as given by Mansoor; for our present pur­
pose it makes no difference if it is read as bmrwry.
43 On such formations, see my discussion in Comparative Philology, pp. 227 ff.
[above].
44 For the syntax, cf., perhaps kimeribe kohen, Hos. 4. 4.
45 Akkadian kamaru is “heap up”, and not “overshadow, darken”, as is said by
Dhorme, Job, p. 27; Ethiopic kdmr is also “heap”.
382 PHILOLOGY AND EXEGESIS
we may go on and say that the second is cognate with Ethiopic
tacayyara, giving a sense “revile”.46 Parallelism in the verbs is
achieved also by the NEB rendering:47
Cursed be it by those whose magic binds even the monster of the deep,
who are ready to tame Leviathan himself with spells.
We shall concentrate, however, on the question of parallelism in
the nouns, and bring the verbs in only incidentally.
We recognise the intrinsic probability of a connection between
Leviathan and the sea; since Leviathan was a sea-monster, that can
go without saying. It does not decide however what the poet of Job
was saying. In his context the basic subject matter is a day or a
night, the day of Job’s begetting or of his birth, his “day” (3. 1). In
verse 7 he is still talking about this same thing, about a “night”. In
verse 9 also we are still dealing with features of this day—its
twilight, its unsuccessful longing for light, its relation to the twi­
light or dawn. The probability then is that the MT is right with its
yom: the poet is talking about the “day-cursers”, and there is
nothing about the sea in the passage.48 A traditional parallelism
between sea and Leviathan is so likely that it hardly needs to be
proved, and one need not doubt its presence in the “background”
here; but in the actual poem as produced, this relationship has now
become otiose. Because the poet is concerned with day, he now has
a significant poetical connection between Dorere and the waiting for
light, Dor, in the next line; it may perhaps have elements of ambi­
guity, suggesting both “those who curse” and “those who give
light”. As for Dorer, it is most probably the familiar Hebrew with
the sense “stir up”. The primary need of the exegete is not to
“identify” the mythological background, in the sense of stating
exactly what pre-existing myth is presupposed; what is more
important is the myth as it is reconstituted by the poet for his own

46 This is example no. 242 in my Comparative Philology; cf. text on p. 125

[above].
47 I do not know on what evidence NEB based the rendering with “to tame with
spells”; surely hardly Arabic tacwfdh, cudha “charm, spell, amulet”.
48 I do not feel quite certain about this view; one could of course go the opposite
way and say that these considerations are the reason why the tradition altered an
original yam into yom. If “sea” is correct, then the verse introduces ideas and
images which are quite extraneous to the remaining structure of the chapter. This,
and the fact that the reading as yam is an extremely facile surface emendation, seem
to tip the scales in favour of the position I state above.
 PHILOLOGY AND EXEGESIS 383
purposes. There are, he hints, powers that can or may or do curse
the day, just as there are powers that give light; these powers can
also give trouble by stirring up Leviathan, a bad-tempered monster
who is opposed to the cosmic order. The pre-Israelite background
is interesting information, but is not more than ancillary to the
explanation of the passage.
*
* *

I would like to conclude with some remarks of a more philo­

sophical character, if one may call it so: some remarks which take
the discussion of Hebrew philology, the use of Ugaritic, etc., and
bring it into relation with the other main problem which has
concerned me, the discussion of biblical semantics. When my book
on semantics was published, one of the main criticisms made
against it was that it belonged to an extreme current of empiricism
or positivism.49 In itself, of course, such a remark is nothing to
worry about: the giving of names such as empiricism and posi­
tivism does not constitute argument. But, I ask myself: supposing
that there is some real issue here, and supposing that there are
really empiricist or over-empiricist, positivist or over-positivist,
attitudes to be found in the treatment of biblical language, then
among whom are they to be found? The direction in which I can
see such viewpoints lies among the scholars whose work I investi­
gated in the book on Comparative Philology. It is among them
that I, as I view the matter, see the outworking of a fairly extreme
empirical positivism. I would not admit that empiricism and posi­
tivism are in themselves in any way terms of reproach; but it is
here, among the Drivers and the Dahoods of scholarship working
on the detailed biblical text, that I find not just an empiricism or
positivism but an almost entirely uncritical and unanalytical empiri­
cism and positivism. And I do not use these terms as mere labels,
but will explain the characteristics which I have in mind: the
concentration on a method which has simple, obvious and practi­
cal outlines; the emphasis on details in black and white, which
details form the units of demonstration in the method; the almost

49 On this see my Biblical Words for Time (2nd edition, London, 1969), pp. 194-
207, “The Philosophical Background”. One reviewer, there quoted (p. 194 n.),
identified “an extreme positivist and formalist attitude, which detests any introduc­
tion of philosophical content.”
384 PHILOLOGY AND EXEGESIS
complete absence of critical and thoughtful analysis of the basis of
the method; an alienation from the sophisticated discussion of
ideas, of entities such as theologies; the use of purely pragmatic
criteria in argument; and, because no properly argued foundation
is provided, the corroboration of each detail and its function not
by any theoretical foundation but by the mere addition of yet more
details of the same kind and on the same level50—a principle which
at its worst deteriorates into that of validation through claims to
multiple success.
Out of all this I shall pick one or two special cases for further
mention. It is hardly disputable that the scholars who have been
most productive with philological suggestions, who have produced
not just a few but hundreds or even thousands, have at the same
time been people rather alienated from the main currents of exe­
gesis. Form criticism, for instance, which is one of the major
modern approaches to biblical literature, seems to have passed
them by. Not only this, but even fundamental literary criticism
seems often to have passed them by; some of the points at which
philological explanations have been offered were difficulties requir­
ing new interpretations only if widely accepted literary-critical
solutions were ignored. I shall illustrate this from only one instance,
a case which I mentioned in Comparative Philology, without how­
ever touching on this aspect of it.51 At Num. 16. 1 the text begins
with wayyiqqah qorah, at first sight apparently “and Korah
took”—but there is no object; and hence the suggestion that this is
a completely other verb, meaning “was insolent, was defiant,
rebelled”. The NEB, I notice, has “Now Korah . . . challenged the
authority of Moses”, which I take to be the same interpretation in
a more generalized form. This is a place, however, where tradi­
tional literary criticism had already provided an alternative expla­
nation, by suggesting that two different sources were compounded.
In one source, let us say, Korah “took” the “men” who appear as
the fourth word in verse 2; some others out of the long list of
names, now in verse 1, form the subject of the other verb “and they
arose” at the beginning of verse 2. Thus we have two sources:

50 Cf. the contrast I make between a logic dominated by discovery and a logic
dominated by analysis, Transactions o f the Philological Society, ibid., p. 54.
51 Comparative Philology, pp. 17 f. [above].
 PHILOLOGY AND EXEGESIS 385
(a) And Korah took men of the children of Israel (P?)
(b) Dathan and Abiram, etc., arose before Moses (JE?)
I do not say that this analysis is necessarily correct;52 but if it is
even taken into consideration as a possibility, the case for finding a
difficulty in the present text with the sense “and he took” is likely
to disappear. In my distinction of two modes of dealing with a
difficult text, the textual through emendation and the philologi­
cal through new identification of meaning on the basis of a cognate
language, I should have added for some cases the source-critical,
as here.
Something similar can be said on the level of theological exe­
gesis. Proponents of the philological identification from cognate
sources seem often to suppose that the provision of a right transla­
tion of the words constitutes more or less a complete exegesis; on
the other hand, the handling of the ideas and the theologies of the
texts is often wild and irresponsible. It ranges from a kind of
reductionism, which seeks to diminish as far as possible the
theological dimensions of the text—a tendency very visible in the
work of Driver—at the one end, to a blind traditionalism, at the
other end, which discovers in the Psalms traditional beliefs like
that in immortality, in defiance of all that has been learned through
patient study in the religion and theology of the Old Testament. In
Dahood’s case it seems likely that the values and terms of a highly
traditional Christian theology are attached to the Hebrew-IJgaritic
cultural-religious mélange which accompanies in his eyes the lin­
guistic indistinguishability of the two.
The aspect of translation has already been mentioned, and must
be further stressed. It is unfortunate that many of the philological
suggestions now being mooted have been aired for the first time, or
almost the first time, in translations; and not only in translations
but in translations intended for the general public and as such
magnified by the full publicity of modern advertising. The work of
translation is not in my judgement at all a good way of making
these suggestions available for discussion. It is of course only right
that the general public should have made available to it the results

52 It seems to have the support of S. R. D river, Introduction to the Literature o f

the Old Testament (1891), p. 59; more recently, R. P feiffer, Introduction ( 1941),
pp. 171, 189.
386 PHILOLOGY AND EXEGESIS
of the latest scholarly opinion. But it is only in our generation that
for the first time interpretations of a quite novel and drastic type
have been made known to the general public before they were
known to—and, still more, before they were accepted by—scholarly
opinion. With the publication of the Old Testament of the NEB it
has become something of a sport among scholars to try to work
out the basis upon which some of the renderings have been pro­
duced. Even those who have studied hundreds of modern philo­
logical suggestions can be left open-mouthed, wondering how the
translators obtained from the Hebrew what they did. Just to cite
one case, at a well-known place in Zech., 3. 2 the phrase ha-boher
birusalayim has been detached from God and attached to Satan, so
that it is no longer God who “chooses Jerusalem” but Satan who
“is venting his spite on Jerusalem”. Thus NEB prints: “The LORD
rebuke you, Satan, the LORD rebuke you who are venting your
spite on Jerusalem.” Only by pure luck had I happened to come
across the suggestion on which this was presumably based53—the
identification of a b-h-r, different from the familiar word “choose”,
and related to Arabic words “to steam” and the like; this new word
is found in no less than seven places in the Old Testament.
My own belief is that translations, at least such as are supposed
to be a considered interpretation, presented to the public, can and
should be produced only after a full consideration by the world of
scholarship, and only after a full process of exegesis, including
literary-critical, form-critical, redaktionsgeschichtlich and theologi­
cal consideration.
To return then to the general philosophy of the study of biblical
language, I believe that the position I have taken in the two main
discussions, though often classed as an extreme and negative one
in the first impression made, is in fact a moderate one. As I see it,
the positions investigated in Semantics and in Comparative
Philology are the two opposed extreme positions in the study of
biblical language: in very rough terms, one was idealist, the other
positivist, if such terms have meaning. To sum up the case about
comparative philology, there is nothing wrong with the method of
using cognate languages in order to derive novel senses for Hebrew
words; but of all the products said to have been derived from this

53 D river, in Mélanges Marcel Cohen (1970), pp. 236 f. See my remarks in

JSS 17 (1972) 133.
 PHILOLOGY AND EXEGESIS 387

method in the last decades, only a small proportion are satisfactory.

Those who think this judgement too negative have a simple course
open to them: they can print the list of such suggestions which they
personally consider to have been established beyond doubt. The
judgement which I passed upon the products of this method in my
book, though subject to some modification in proportions, is, I
believe, in its basic structure a right and fair one; and, far from
taking any extreme position, I believe that in making this judge­
ment I have spoken for the central current of Old Testament
scholarship, in which I have confidence.
 UGARITIC AND HEBREW “SBM” ? 1

A t an early stage in Ugaritic studies a word sbm was identified,

and the meaning ascribed to it was “to muzzle”. The main text
originally involved includes the following:
Imhst. m dd
Dil ym . Iklt. nhr. Dil rbm
Pistbm. tnn. Distmlh
mhst. btn. cqltn
slyt. d. sb ct. rDasm

A representative translation is:

Did I not crush 3IPs Darling, Sea?
Nor destroy River, the great god?
Nor muzzle Tannin full well?
I crushed the writhing serpent
The accursed one of seven heads.2
This and other Ugaritic passages will be considered more fully
later; for the present this is set out simply in order to make clear
the context of the form sbm. Most of the other terms of the
passage were then already quite clear and well identified, and they
can still be taken as fairly certain. The word here printed as Distmlh
was of uncertain reading; some thought it should be read as
Disbmnh, i.e. as repeating our term sbm. The context, then, is a
destruction of great cosmic dragons or monsters; the new identifi-

1 I have to express thanks to many who have assisted me with their opinions and
advice, especially in their specialized fields, and most of all Professors A. F. L.
Beeston, W. G. Lambert and R. B. Serjeant, and my Manchester colleagues
Professor C. E. Bosworth, Dr J. D. Latham, Dr M. A. A. el-Kafrawy, Mr M. E. J.
Richardson and Mr S. Strelcyn. None of these, naturally, is responsible for the
results I have reached, but I am grateful for their willingness that I should quote
their opinion where this has been done. [Reprinted with permission from Journal o f
Semitic Studies 18 (1973) 17- 39.]
2 C. H. Gordon, Ugaritic Literature (Rome, 1949), pp. 19 f.
 UGARITIC AND HEBREW “SBM”? 389
cation of sbm added to this the image of the muzzling of such a
monster. No other context in Ugaritic, such as the muzzling of
actual dogs or other animals, has been identified. The sense of “to
muzzle” seems to have been very widely accepted and passed on
from one scholar to another; only occasional doubts and qualifica­
tions, which will be discussed below, have been expressed; and the
identification may be said to have become part of the received
tradition of meanings in Ugaritic.
On the basis of the Ugaritic sbm with this meaning, such a
word has been identified in Hebrew also. At Ps. lxviii. 23, where
MT has:
‫אשיב ממצלוזז ים‬ ‫מבשן אשיב‬
it has been suggested that the latter part of the line should be
read as
‫אשבם מצלות ים‬
with the sense something like “I muzzled the abysses of the Sea”.3
This view is strengthened by the parallelism with ‫בשן‬, this word
being understood as “Serpent”.4
The suggestion is at first sight an attractive one, and has received
considerable recognition; it was described as “brilliant” by M. H.
Pope, who regards as corrupt and “meaningless” the text when
taken in its more general previous understanding as something like
“I will bring back from the depths of the sea”.5 Further elabora-
tions on the suggestion have been made, and we may quote that of
P. D. Miller, Jr., who proposed to read the text as
‫אשבם מצלות ים‬ ‫אשבם בשן‬
something like: “I muzzled the Serpent, I muzzled the abysses of
the Sea”.6

3 So M. J. Dahood, “mismar ‘muzzle’ in Job 7, 12”, J.B.L. lxxx (1961), 270 f.,
and again in Psalms I I( 1968), pp. 131, 145 f. The example is included by me in the
list at the end of Comparative Philology and the Text o f the Old Testament, no.
311 [above], but is not further discussed in that book.
4 Dahood, Psalms //, pp. 145 f., reaffirms this, criticizing the contrary view
previously expressed by A. R. Johnson, Sacral Kingship in Ancient Israel (Cardiff,
1955), p. 73 n., who said that this view was against the context.
5 Pope, Job (1965), p. 60.
6 Harvard Theological Review lvii (1964), 240.
390 UGARITIC AND HEBREW ‫״‬SBM‫?״‬
Miller at the same time reported a reading suggested by Pro‫־‬
fessor F. M. Cross and “virtually identical” with the Ugaritic
passage:
‫>מחצת< מצלות ים‬ (‫אשבם בשן אשב)מנה‬
“I muzzled the Serpent, I muzzled him;
I smashed the Deep Sea”

II

What is the evidence for the meaning of Ugaritic sbml

From the beginning, Ugaritic sbm was identified on the basis of
Arabic sabama; Virolleaud wrote: “Nous expliquons estbm ( lre
p. impft. ifteal) et esbmn ( lre p. énerg. I) par l’ar. ^ bâillonner
(plus particulièrement: d’un chevreau, pour le sevrer).”7
Most publications cite no other evidence than this, and none at
all that I have seen cites any Semitic evidence which might lead in
another direction; the only note I have seen which works from
Akkadian evidence (to be discussed below) accepts this Arabic
evidence entirely. Thus the sufficiency of this Arabic evidence
seems to have been almost entirely accepted, and the view that sbm
meant “muzzle”, and was used of the muzzling of a tnn or cosmic
dragon, has become part of the received tradition of meanings in
Ugaritic.
It has not been generally noticed that the Arabic evidence8 for
this meaning is extremely thin. Though it does appear in tradi‫־‬
tional dictionaries such as Freytag and Lane, from which Virolleaud
himself doubtless procured it, and which themselves got it from
indigenous dictionaries, I find that specialists in Arabic have little
awareness at all of the existence of this word with this meaning,
and little evidence of productive use of it can be obtained. I doubt
whether any of the Ugaritologists who cited the word had seen it
used in any Arabic text.
First of all, the sense of s-b-m which is really in use and
productive in classical Arabic is quite another one, namely “to be
cold” (of water, of the heart, etc.). It would be too much to say

7 La déesse cAnat (1938), p. 53; his transcription differs somewhat from that now
usual, and used by me in the rest of this article.
8 “Arabic” means North Arabic, as distinct from South Arabian.
 UGARITIC AND HEBREW “SBM”? 391

that this is common; but at least it can easily be found in texts, e.g.
the famous poem of Kacb b. Zuhair, B a n a t S u ca d , which talks of
wine “mixed with a limpid cool stream of water” (b i- d h T s a b a m i n
m i n m a Di , literally “with that from water which possesses cool­
ness”).9 In MutanabbI we find a phrase m a n q a l b u h u s a b im a “one
whose heart is cold.” 10 As far as I have been able to discover,
“coolness” must be regarded as the main extant and productive
sense in classical Arabic.
What then of the sense from which the meaning “to muzzle” for
Ugaritic has been derived? It seems clear that the basic form in
question is the noun s ib a m . Of it Lane says:
A piece of wood which is put crosswise in the mouth of a kid [and tied
behind its head . . .] in order that it may not suck its mother.11
The verb forms s a b a m a or s a b b a m a seem to be clearly denomi­
native from this noun: they mean “he put the s ib a m upon the kid”.
The dictionaries offer also another sense for the noun s i b a m . To
quote Lane again, it is used of:
Two threads, or strings, attached to the [kind of face veil called]
by which the woman [draws and] binds [the two upper corners
b u r q u c,
of] it to the back of her head.

As soon as this is taken into account, the semantics of s ib a m

appear in a very different light. The sense in Arabic, from which
the sense “muzzle” has been derived, even assuming that it is both
genuine Arabic and correctly reported, does not mean anything
like “muzzle” in the sense required for the Ugaritic and Hebrew
passages, i.e. a device to prevent a large and dangerous animal
from biting, and particularly a cosmic dragon. In the case of the
woman’s veil absolutely no element of muzzling, fettering, or limit­
ing the movement of the woman is involved. If one takes the sense
“device to prevent a kid from sucking” together with the sense
“threads to hold a veil in place”, the likely common element is
nothing near “muzzle”; what is more likely is that s ib a m on the
basis of these two meant a “string” or “thread”. As we have seen,

9 This poem is conveniently available in Nöldeke, Delectus, p. 110, 4th line.

10 Edition Dar Sädir (Beirut), 11, 118.
11 Lane, p. 1499.
392 UGARITIC AND HEBREW “SBM”?
Virolleaud himself properly noted12 that the Arabic sense con­
cerned a device to assist in the weaning of a kid; but this qualifica­
tion has been little noticed.
It is true that the participle musabbam is said to be used of a
lion: “Applied to a lion, it means having its mouth tied” (Lane,
ibid.). A proverbial verse cited in this connection is:

S - j

I ·X—*v j I 6—

“She is frightened at the cry of the crow

and breaks the neck of the lion w hose m o u th is t ie d ”.

What is the value of this testimony? First, even if the tradition as

printed above is correct, the phrase al-Dasad al-musabbam is surely
a literary extension, in a paradoxical sense, of the usage as applied
to a kid: it means a lion to which, if one can so imagine it, the
device for weaning a kid has been applied. In other words, it does
not mean that the device was used for lions or other major ani­
mals, or that the word designated any muzzle or other device used
upon lions or other major animals.
Secondly, however, the reading itself may be questioned.
Kazimirski13 suspects an error in the transmission of the proverb;
and this would amount to no more than an error in diacritical
points, between b and t. This idea appears to be not a mere
conjectural emendation, but to represent a variant form found in
the written tradition.14 “Au lieu de musabbam dans ce proverbe, il
serait plus naturel de lire musattam.” The strength of this is that
is in fact used of lions: thus “ —grogner, être grognon (se
dit d’un lion)” and —Sinistre, au visage rébarbatif et ren­
frogné, austère. De là ^ILJI Lion.” 15 In fact the proverb seems to
make more sense as a contrast between the woman’s fright at the
cry of the crow and her ability to deal with a fierce-looking lion;
less good sense is made by a lion to which the sibâm of a kid has

12 Indeed, he emphasized it beyond what has been evidenced in the quotation

above, adding in a later paragraph that the exact sense was “mettre le bâton appelé
sibamm [jic] dans la bouche (d’un agneau)”.
13 A. de B. Kazimirski, Dictionnaire arabe-français (Paris, 1960), p. 1187.
14 Lane, ibid., already records this as a variant by another “relation”, which I
take to mean another tradition, written or oral.
15 Kazimirski, op. cit. p. 1190.
 UGARITIC AND HEBREW “SBM”? 393
been applied. If, then, this alternative reading in the proverb is
accepted, this Arabic text concerning the lion has nothing to do
with the root s-b-m and can throw no light upon it.
There is, then, no evidence of any use of s-b-m in Arabic for
anything like the muzzling of a dangerous animal. To this we must
add the second major difficulty of the Arabic evidence, namely the
fact that the tradition of sibam/sabama, even when meaning a
device for the weaning of a kid, appears to be extremely thin. I
suspect that the dictionary entries, as found in Lane and compar­
able works, may all have arisen as explanations of the one verse
of the early poet cAdI b. Zaid (Christian, from Al‫־‬HTra, sixth
century a.d .):16
*xjI *1*•‫)■»׳^־‬
~ 0 J o

'jib ^ ^

« J

“Man does not have any refuge from the assault of fate
which will avail him against the s ib a m of a she-kid.”
My colleague Dr J. D. Latham kindly suggests to me that the
last phrase might perhaps also be taken as “avail him as much as
the sibam of a she-kid”, i.e. “avail him to the slightest extent”. This
last would be a use analogous to the QurDanic use oji “t0
the extent of a grain”.17 ‫׳‬
This is an early and somewhat out of the way source. Further
usage in quotable sources I have not as yet been able to discover.
On the basis of the evidence here presented, it seems reasonable to
conclude that s-b-m with the sense of using a device to tie a kid’s
mouth belongs to a very narrow line indeed. It may indeed not be
a genuine Semitic word at all. It is certainly unquestionable that
the sense “coolness”, of water or of the heart, is much better
established. Moreover, as we have seen, there is no evidence to
support the idea that the semantics of s-b-m came close to muz­
zling, i.e. that the semantic element of preventing a dangerous
animal from biting was basic to the meaning. On the contrary, if
one takes together the use for weaning a kid and that for the

16 See Lisän al-carab, xv, 210, line 2; on the poet see Brockelmann, Geschichte,
1. 21 f. On this aspect of Arabic indigenous lexicography see L. Kopf in V.T. vi
(1956), 296, and the writer’s Comparative Philology and the Text o f the Old
Testament, p. 118 [above], with citations from Kopf and Wild on this point.
17 QurDan iv. 44/ 40.
394 U G A R IT IC AND H EBREW “ SBM” ?

attachment of a woman’s veil, the main discernible semantic ele-

ment is a meaning like that of the English noun “string”, “thread”.
To this we may add another argument: Arabic has a good
supply of words which do mean “muzzle”, and sibam is not among
them. The most prominent no doubt is the root k-m-m, and in
addition there is k-c-m and g-m-m}%In the Arabic Bible (at least in
the printed texts, which I have consulted), it is k-m-m that is used
in the classic instance of the “muzzled ox” (Greek cpipouv, Hebrew
‫ )חסם‬and at Ps. xxxix. 3 (Hebrew ‫ אשמרה לפי מחסום‬, Arabic
kimama). The grouping of the Arabic vocabulary in this semantic
field thus leaves no reason to expect the presence of an de‫־‬
ment s-b-m.
It will be relevant at this point to mention a Hebrew passage
which might seem to be parallel and which was already cited as
such by Virolleaud, namely the tying down (?) of the tongue of
Leviathan in Job xl. 25 f. (EV xli. 1 f.):19
‫תמשך לויתן בחכה ובחבל תשקיצ לשנו‬
‫התשים אגמון באפו ובחוח תקוב לחיו‬
At first sight this mention of doing something to Leviathan’s
tongue with a cord seems parallel; it is not indeed the “muzzling”
of a sea-monster, but at least it is some kind of tying. The key
phrase of our purpose is the second part of the first line: what is
intended when the text asks (literally) “Can you cause his tongue
to sink with a line?” It is not clear exactly what operation is
intended. NEB writes: “or can you slip a noose round his tongue?”
This would seem to suggest the use of a noose which would catch
over the lower jaw and trap the fish or animal, meanwhile inci-
dentally holding down the tongue. Another possibility is that the
hook is lodged in the gorge or stomach of the animal. The line
(‫ )חבל‬is tight with the strain and thus it presses down the tongue,
as the fisher pulls against the strength of the beast. In any case, if
either of these is right, and indeed whatever the detailed meaning
here, the whole context is very remote from that of doing battle
with a cosmic dragon and muzzling or fettering it. That mythologi-
cal scene is not the point here: the problem is the human problem

18 For the first two of these see M. Ullmann, etc., Wörterbuch der klassischen
arabischen Sprache: k-c-m, p. 235; kicäm “muzzle, strap”, p. 236; k-m-m, p. 343.
19 Virolleaud, ibid.
 U G A R IT IC AND H EBREW “SBM”? 395

of fishing on so grandiose a scale. The genre of the Hebrew

passage is such that it mentions with some scorn the different kinds
of instruments (hook, line, etc.) with which one can catch ordinary
fish but not this one (whether whale, crocodile, or something more
like the Loch Ness monster). The context is a fishing one, not a
mythological combat as in the Ugaritic references from which we
began; there is no question of “smiting”, but only the fact that
man’s power and knowledge are not sufficient to catch this fish.
Brief reference may also be made to Job vii. 12:
MT: ‫כי־־תשים עלי משמר‬ ‫הים־אני אם־תנין‬
“Am I the Sea, or a Tannin, that you put upon me a m is m a r V ’
It has been suggested of this also that the mismar might be a
muzzle.20 God is silencing Job as he has silenced Yamm and
Tannin. Against this we may bear in mind the considerations
advanced by Pope: “The consideration which deters the present
writer from accepting this interpretation here is the impression that
it does not suit the context. There is nothing to suggest that God is
attempting to silence Job. What Job complains of is the constant
harassment and surveillance which God maintains (cf. vss. 18-20)
and this accords perfectly with the normal meaning of mismar as
‘guard, watch’.”21
This argument, however, does not in any way diminish, or
conflict with, the idea that God in the Old Testament holds the sea
within bounds or limits, from which it would otherwise break out.
This is entirely adequately evidenced. The question is only: {a)
whether the image used for this keeping within bounds is that of
muzzling; (b) whether Job vii. 12 concerns constraint or rather
watching, surveillance. The matter is mentioned only as a guard
against the supposition that, muzzling or other constraint of the
sea being accepted for Hebrew, this would constitute a preference
in favour of the sense “muzzle” for Ugaritic sbm. It is, however,
primarily the sense of this Ugaritic word that is at present in
question. To this may be added the observation that the Hebrew
word which unquestionably does mean “muzzle”, namely ‫חסם‬, is
not in fact found in use as an image for the holding of the sea
within bounds.

20 Dahood, J.B.L. lxxx (1961), 270 f.

21 Pope, Job, p. 60.
396 UGARITIC AND HEBREW “SBM”?
Thus, to sum up the argument so far, the Arabic evidence upon
which the original identification in Ugaritic was made is very weak,
so weak indeed that an identification made purely on the basis of
the context in Ugaritic, and thus depending on no cognate evi­
dence whatever, could hardly be said to be weaker.
Yet, if it had not been for the Arabic, it is doubtful if the
Ugaritic context itself would have led to the identification of the
sense as “to muzzle”. On the contrary, this sense does not fit very
well. The other verbs in the Ugaritic passage (mhs, kly, mbs)
suggest rather destruction or annihilation. Does one muzzle a dead
dragon? Quite in abstract, a muzzled Tannin could make good
sense, suggesting a dragon kept under constraint, which might
nevertheless be let loose again; but this, though true and even
probable of dragons in the abstract, does not fit well in our text.
Again, in theory muzzling could be a preliminary to killing, but
this does not seem to be the sequence of events in the Ugaritic text.

in

These intrinsic difficulties with the now accepted understanding

were noted in a short study by S. E. Loewenstamm in 1959.22
Pointing out that most of the verbs mean something like “to
destroy” or generally “to beat, fight”, he then says, “The only verb
which does not seem to fit into this picture is sbm . . . The com­
monly accepted explanation is ‘to muzzle’ . . . It may be doubted,
however, whether this meaning is acceptable in a context where
another Ugaritic synonym of ‘destroy’ would be expected.”
As Loewenstamm suggests, it is doubtless because of an aware­
ness of these points that Gordon in his Ugaritic Handbook (1947),
p. 271, no. 1909, gave the sense of sbm as “to check, annihilate”; in
this entry he did not cite the Arabic sib dm as a clue.23 In the later
Ugaritic Textbook (1965), however, Gordon cites the Arabic par­
allel and gives the sense as “muzzle”; this may be an effect in part
of Loewenstamm’s note, which, as we shall see, returns in the end
to the sense “muzzle”. It is interesting that another lexicographer

22 “The Muzzling of the Tannin in Ugaritic Myth”, Israel Exploration Journal ix

(1959), 260 f.
23 Also Ugaritic Manual (1955), no. 1801, cited by Loewenstamm. The later
reference is Ugaritic Textbook, p. 487, no. 2378. Gordon had offered only “annihi­
late” in his still earlier Ugaritic Grammar (1940), no. 799.
 UGARITIC AND HEBREW “SBM”? 397
of Ugaritic, Aistleitner, in his Wörterbuch, no. 2576, gives the
sense indeed as “Knebel in den Mund stecken”, but only with a
question mark.
This difficulty led Loewenstamm to consider a piece of Akkadian
evidence.
This Akkadian evidence concerns the word napsamu, and
Loewenstamm cited a passage from Ludlul bei nemeqi as discussed
by Landsberger.24 Landsberger reconstructs a sequence like this:
im-ha-as rit-ti ma-hi-si-ia
u-sad-di gl§kakka-su dmarduk
i-na pi-i gir-ra äkili-ia
id-di nap-sa-ma dmarduk
This would mean something like:
Marduk smote the hand of the smiter,
caused his weapon to fall down;
Marduk put a muzzle on the mouth
of the lion who was eating me.
This evidence convinced Loewenstamm, in spite of his recogni­
tion of the difficulties in the sense “to muzzle”, that that sense must
be correct. The decisive point in his argument is the observation
that the Akkadian passage and the Ugaritic have not only one but
two related words in common: mhs “fight” is found in both, and
napsamu, derived from the stem psm , “is most probably cognate
with Arabic and Ugaritic sbm”.
There are several points to be remarked on in this account of the
matter. A minor one, to begin with, is that the mhs of the Ugaritic
passage is not in close proximity with the sbm; it occurs only in
line 43, six lines later.25 The verbs in immediate contiguity with
sbm are kly and mhs.26 By the time we come to the mhs of Dimths
in line 43, the immediate sequence in which sbm is found has been
passed and a new sequence is beginning. The contiguity of sbm and
mhs is thus less than immediate. If one were to consider Ugaritic
mhs to be more or less the same thing as mhs, then the position
might be different; but that is not the view held by Loewenstamm.27

24 A.f.O. xvm (1957- 58), 378 f.

25 Loewenstamm by a slip says “1. 41” in his note.
26 smt, with a very similar meaning, also appears, but not until 1. 41.
27 See his footnote, loc. cit. p. 261.
398 UGARITIC AND HEBREW “SBM”?
The second point of interest is that Landsberger’s reading, to
which Loewenstamm appealed, is a reconstruction: two lines in an
Assur fragment are identified with two lines cited in the Com-
mentary to Ludlul. This very reconstruction was minutely examined
by Professor W. G. Lambert in his edition28 and pronounced to be
erroneous. If this had been right, then the main point used in
Loewenstamm’s argument, namely the co-occurrence of both of
two terms in both passages, would have been lacking in force.
Professor Lambert, however, tells me by letter that he has indeed
convinced Landsberger that he, Landsberger, was wrong in this;
but he, Lambert, had subsequently found new evidence, as yet
unpublished, which shows that the reconstruction proposed by
Landsberger was in fact right.
We must then accept the validity of the reconstruction upon
which Loewenstamm’s argument from Akkadian was based. There
remain, nevertheless, some other sides of his argument which
deserve to be discussed.
The correlation of Akkadian napsamu and Ugaritic sbm involves
abnormal phonological correspondences, and this was recognized
by Loewenstamm himself; I shall cite the relevant passage from his
note:
The Accadian noun n a p s a m u , derived from the stem p s m , is most
probably cognate with Arabic and Ugaritic s b m , notwithstanding the
irregular correspondence between Arabic ^ and Accadian s which is,
however, attested several times, especially near b /p . Cf. e.g. Accadian
b (p )u s s u r u with A r a b i c H e b r e w ‫ ;בשר‬Accadian k a b s u with Arabic
Hebrew ‫כבש‬. The interchange between b and p is a well-attested
phenomenon as for example in the Hebrew ‫דבש‬, Accadian d ip su
“honey”. It appears, therefore, that the Accadian root p s m developed
from a metathesis of the proto-Semitic root sb m in a way that reminds
us of the well-attested change between Hebrew ‫ כבש‬and ‫כשב‬.
Thus Loewenstamm’s explanation involves an account in one
breath of three peculiarities: the alternation p /b , the alternation
s/s, and the change of order ps/sb. Though parallels can be put
forward for each such peculiarity, one would not suppose that
Professor Loewenstamm thinks that such irregular correspondences
can be admitted lightly or easily; indeed, his own “notwithstand-

28 Babylonian Wisdom Literature (1960), pp. 24 f.; texts on pp. 56, 58. Professor
Lambert tells me that his arguments as then set out still seem to be convincing on
the basis of the evidence then available.
 UGARITIC AND HEBREW “SBM”? 399

ing” is evidence of the opposite. It is likely that the semantic

probability of the correlation is, to his mind, so great as to con­
stitute a reason for accepting some abnormality on the phono­
logical side. I shall not in this paper further discuss the matter of
the peculiarities in the phonological correspondences, beyond
observing that they are there.29
We return, therefore, to the passage cited from L u d l u l ; and here
we cannot but observe that, though the two pairs of lines are in
juxtaposition, they present two very different pictures: the first is
the disarming of a physical assailant, the second is the muzzling of
a lion. The two complement each other, but units within them do
not seem to be integrally related with the kind of integral relation
posited for (let us say) m h s and s b m in our Ugaritic passage. In the
Ugaritic, if you m h s a Tannin, you will also within the same
universe of discourse k l y a Tannin, and, if s b m means “muzzle”,
you may s b m a Tannin; but there are no such relations within the
elements of the Akkadian text. The application of the n a p s a m u
belongs to Marduk’s work as protector; it does not belong to the
same conceptual field as the smiting or beating, which is the work
being done a g a in s t the afflicted writer. Or, in other words, we do
not have a parallelism between Marduk as a fighter (m a h i s u ) and
Marduk as the applier of the n a p s a m u . The text does not neces­
sarily imply that the one being smitten is himself engaged in
fighting; it is equally likely, or more likely, to mean that the
complainer was defenceless, and under assault from an armed
man. The real parallel for m a - h i - s i - i a , taken syntactically, is a k i l i - i a ;
or, to put it in another way, that in the Akkadian verses about the
beater which corresponds to the application of the n a p s a m u is not
the m a - h i - s i - i a but the u - s a d - d i , the causing of the weapon to fall.
This very obvious point considerably diminishes the relationship
seen by Loewenstamm between the n a p s a m u and the verb m - h - s \
and the loss of this relationship considerably reduces the force of
his argument for the identification of Ugaritic s b m as “muzzle”.
Further, some consideration must be given to the semantic
development of n a p s a m u itself and the related verb p a s a m u (vari­
ants p e s e n u , p a s a m u , p a s a n u ) . n a p s a m u corresponds to Sumerian

29 The question is not whether such anomalies occur, but whether their occur-
rence can be presumed when, as is here the case, the identification of the words in
question is exactly what is in dispute; this makes a world of difference from familiar
cases like the Hebrew ‫כב ש‬/‫כ שב‬.
400 UGARITIC AND HEBREW “SBM”?
su (= k a t a m u “cover”), and synonym lists equate n a p s a m u with
k u lu lu , a head covering. The commentary to L u d l u l explicitly
explains n a p s a m u as m a - a k - s a - r u s a p i sTsi, a binding or covering
of the mouth of a horse.30 What Marduk did to the lion seems to
be something like putting a bag over its head.
This is important, because for any conclusions about the sense
of Ugaritic s b m we will probably have to start not from the derived
noun n a p s a m u in Akkadian but from the underlying verb, p a s a m u ,
etc. The idea of “muzzle”, i.e. of an action intended to prevent a
dangerous animal from biting, is not at all intrinsic to the verb.
Even of the noun n a p s a m u , as we have seen, this is also true: used
of a horse, which from the annotation of the commentary to
L u d l u l would seem to be the normal application, it means a cover,
which might have many other purposes than that of preventing
biting; only when applied, specifically and doubtless in a trans­
ferred sense, to the lion does it explicitly come to be what we call a
muzzle. Of the verbs this is all the more true: their semantics
appear to contain no essential element of muzzling (to prevent
biting) or of fettering (to prevent movement); their essential seman­
tic element is that of covering (to put out of sight), p u s s u m u is
used of a veiled woman—a goddess, a bride—and von Soden
considers that the verb p a s a m u might be a denominative from
p u s s u m u . p u s u m m u is a veil.31 p e s e n u also is “conceal, make
secret”.32
The possibility of establishing a close correspondence between
the root of Akkadian n a p s a m u and Ugaritic s b m has thus become
considerably less. Loewenstamm’s study neglected to consider the
probable semantic history of the Akkadian words; the striking

30 Von Soden, AHw, p. 7406, suggests a horse’s bit, “Gebiß des Pferdezaums”,
but his earlier explanation, “vielleicht eine Art ‘Maulkorb’”, Orientalia xx (1951),
268, is surely better, as Professor Lambert assures me. In general, the constraint of
the mouth of animals in the ancient world seems mostly to have had the purpose
not of muzzling in our sense, i.e. preventing the biting of people, but rather the
purpose of preventing the eating of grass, etc., and the concomitant distraction of
the animal from its work. Thus in the Hebrew law, Deut. xxv. 4, the owner would
naturally wish to constrain the ox because (a) it would eat the very product it is
being used to produce, (6) this eating would reduce its efficiency in this work. Such
constraint is however forbidden by the biblical law.
31 AHw, pp. 8826, 883a, 838a. Cf. the note in Orientalia xx, 267 ff., which takes
it as manifest that the basic sense is “veil”, verschleiern.
32 AHw, p. 8566. Cf. Landsberger, Z.A. xli (1933), 220: “psn, das auch in den
Spielformen psm und psn erscheint, bedeutet eigentlich ‘mit einem Schleier
bedecken’ this note is on pasTru “secret” and related words.
 UGARITIC AND HEBREW “SBM”? 401
similarity of the end product n a p s a m u , as applied in the L u d l u l
text, to the sense already conjectured for Ugaritic s b m (on the
basis of Arabic) was taken as sufficient. But as soon as it is seen
that the Akkadian root means something different, the correla­
tion becomes more precarious; and the phonological peculiarities
involved in it, which might have been brushed aside when the
semantic equation was a close one, come to appear more serious.
Moreover, we have incidentally seen that the Akkadian word-
group in question occurs with a considerable variety of root letters,
and therefore it is by no means simple to search for the true
Semitic cognates: we have to look at the sequences p s m 9p s n 9p s n ,
p s m . This brings us to the question, what forms can be cited from
any other Semitic language that correspond to the Akkadian
p a s a m u (or variants)—apart from the isolated case of Ugaritic
s b m , which is just what is now in question. The nearest I have
found is the Ethiopian (Gecez) f a d a m a (Dillmann, 1391), meaning
“to obstruct, block up (a road)” and used in a transferred sense of
“to reduce to silence”. It is in fact used—surely by a great
coincidence—of Jesus’ “stopping up the mouths of” the Sadducees
in Matt. xxii. 34, where the Greek has ecpipcoae, originally of
course “he muzzled, he stopped up”. But I do not seriously think
that this is closely related, except in so far as we come to depend
on hypotheses of how all· words beginning with p have some
meaning element in common, and the like. Is there real evidence
that any cognates of p - s - m existed in Semitic outside of Akkadian?
The Akkadian evidence, then, does not after all provide any
corroboration for the application to Ugaritic s b m of the sense
“muzzle”, itself based as we have seen on very weak evidence from
Arabic. And Loewenstamm himself, as we have seen, began his
note by remarking on the apparent intrinsic weakness of this sense
for the Ugaritic context itself.

IV

Is there then any other direction in which we may look tor the
identification of Ugaritic s b m l
First, a phonological point: Ugaritic s has a normal correspon­
dence with both s in and s h in in Hebrew, s h in and s in respectively
in Arabic, so that cognates with either could be considered.33 It

33 Cf. the table in Moscati, Comparative Grammar, p. 44.

402 U G A R IT IC AND H EBREW “SBM”?
was only the correlation with the Arabic s - b - m that previously
seemed to favour one choice rather than the other. Incidentally, if
the correlation with Arabic s - b - m were correct for the Ugaritic,
then one would expect that any Hebrew cognate would be ‫שבם‬
and not ‫ ;שבם‬this point seems to have been unnoticed by most
proponents of this correlation, and if there were a Hebrew cognate
meaning “muzzle” it should in all probability be entered as ‫שבם‬.
This however does not much concern us now, since we are now
considering the possibility of a fresh identification; I merely point
out that two consonants both in Hebrew and in Arabic can be
considered, so long as no evidence from yet a third source is
forthcoming.
Next, a small point within Hebrew itself: there is a place called
Sibmah, ‫ שבמה‬or ‫שבם‬. This name could possibly be cognate, and
it would give us certain evidence of a formally equivalent con-
sonant sequence, offering normal correspondence with an Arabic
or Ugaritic s b m . I shall say something more about it in a moment.
Now, the only Semitic area in which I have found a lively
representation of a root s - b - m is South Arabia. Even here it is
found basically in the names of places and persons, but in these it
is quite common. A glance at the glossary of Conti Rossini’s
C h r e s t o m a t h i a shows that the place name s - b - m is frequent,34 and
in modern Hadramaut there are several places with the name
Shibam; this same name seems to exist in (North) Arabic as the
name of a tribe. Among personal names in South Arabian Conti
Rossini lists: Db s b m , d d s b m , cm s b m , m cd s b m , h ls b m . For a form
like Jb s b m he suggests the reading Abl-sabima, but he considers
also the possibility that the second element might be read s ib a m ,
following the modern pronunciation of the place name. Other
names, such as Y a s b u m , 3A s b a m , etc., are listed by Ryckmans;35
and from the Thamudic inscriptions van den Branden lists s b m
and s b m t \ the latter he reads as s i b a m a t . 36
What then is the sense of the element s - b - m in these South
Arabian names? Ryckmans mentioned the North Arabic sense “be
cold”, which we have seen to be the dominant sense in that lan-

34 Chrestomathia, pp. 246 f.

35 Noms propres sud-sémitiques (1934), i. 205a, 2546.
36 Inscriptions thamoudéennes (1950): HU 251 (p. 134); Jsa. 129, 157 (pp. 400,
405); index, p. 545.
 UGARITIC AND HEBREW “SBM”? 403

guage; but it is not clear what semantic connection he had in mind,

or what such connection there could be. For a place name like
Sibâm the sense “cold” might fit, but it is hard to see what sense it
could have in personal names like Dbsbm. The other North Arabic
sense, that of the device to prevent a kid from sucking, is of course
of no help.
A much more probable suggestion is one which comes from
Landberg and is accepted or at least reported by Conti Rossini.
For the verb s-b-m in the Hadramaut dialect Landberg reports the
meaning “être haut”, and for sibâm he gives “hauteur, nom des
villes”.37 For the Sabaean personal name Dbsbm he offers the very
probable explanation “mon père est grand, haut”.38 The first ele­
ment would, of course, be a term for a deity; and this interpreta­
tion would be entirely in accord with well-known tendencies of
Semitic name-giving, producing the same structures as common
types like Abram, South Arabian DAbacaliy.39 The town name
Shibâm would be of a type similar to familiar names like Ramah.
Incidentally, it is not unreasonable at the same time to go back
to the verse of cAdT b. Zaid, cited above (p. 393) as the early text the
explanation of which has perhaps caused the transmission of the
interpretation of sibâm as the device for the weaning of a kid. It
may be questioned, however, whether in that verse itself the under­
standing as such a weaning-device is absolutely necessary for the
sense. Clearly, the verse requires the context of something to do
with a kid; but it might be possible to think of something like the
“rising up”, the “leaping up” of a kid (cf. the parallel, “the assault
of fate”), and this would seem to give quite good sense.40
Returning to the Hebrew place name Sibmah, an interpretation
as “high place” would also be a very easy and natural one, and

37 Landberg, Glossaire Datinois, p. 2018.

38 Landberg, Dialectes de Varabie méridionale, 1 (Hadramaut), 616; cf. also 11
(Dathina), 317 n. 2, 701, 1298 n. 2. Landberg makes further suggestions in that he
sees relationships between s-b-b and s-b-m (cf. groups like sadda and sadama,
satama “close”; saqqa, saqama “break”); on p. 1298 he says Mehri has two verbs,
sôbb “lever les pieds de devant et sauter (cheval), ce qui est !^<*2‫” ׳‬, and sebù “être
élevé, ce qui est l’ar. L-i ; ,fry* en est un développement”. Professor Serjeant con­
firms to me that he remembers shaba, yashbu as meaning something like “to go up”
in Hadramaut.
39 On this name see recently M. Hôfner in H. Gese, M. Hôfner and K. Rudolph,
Die Religionen Altsyriens, Altarabiens und der Mandder (1971), p. 379.
40 Note, for what it is worth, the use for the prancing of a horse mentioned by
Landberg, above, n. 38.
404 UGARITIC AND HEBREW “SBM”?
would fall into the same class as names like Ramah. An interpre-
tation after the Arabic (North Arabic) sense of s-b -m , i.e. some-
thing like “place of cool (waters)” is no doubt possible,41 but is
rather less likely. It is an amusing coincidence, but of no greater
evidential value, that Jerome, who knew a place close to Heshbon
which was understood to be Sibmah,42 interpreted the name as
Sabam a tollens altitudinem .43 This is a manifest explanation on the
basis of Hebrew ‫שא‬, i.e. a part of ‫“ נשא‬lift”, and ‫“ במה‬high place”,
and has no relevance for historical etymology.44
To sum up our results so far, the South Arabian s-b-m is by far
the best-established Semitic term that we have thus far found
which might present a cognate for Ugaritic sbm . It is clearly and
repeatedly evidenced in ancient sources, and has a far stronger
claim to consideration than the thinly evidenced North Arabic
sense “apply a device to prevent a kid from sucking”.
Two general points may be made here. First of all, a sequence
having b in second position and m in third position is abnormal
for Semitic triliteral verb roots. Such a sequence is indeed a stan-
dard example for consonantal incompatibility; and in Greenberg’s
well-known article, which registers the combinations found in
Arabic triliteral roots, only one case of -b-m is registered (pre-
sumably our s-b-m ).45 For at least some of the words which include
the sequence -b-m , therefore, one is tempted at times to consider
the possibility of extra-Semitic origin; this is true particularly of
the Arabic sibam , the device to keep a kid from sucking. The fact
that the verb s-b-m “to use the sib a m ” is denominative is sig-
nificant, since the compatibility rules are known to be less rigorous
for nouns and by consequence for verbs formed from them. The
one Hebrew verb with -b-m , i.e. ‫יבם‬, is denominative from the

41 Aharoni, The Land o f the Bible, p. 98: “names related to water sources
are . . . widespread”.
42 Comm, on Isa. xvi. 8 (Corpus Christianorum, lxxiii; Turnhout, 1963), p. 181.
43 Interp. Hebr. Nom. (Corpus Christianorum, lxxii; Turnhout, 1959), p. 122.
44 I add in a note this additional point: Sibmah in the Bible is mentioned as a
place of vines (Isa. xvi. 8 f.; Jer. xlviii. 32), indeed this is the only association of it
which we know. Now S. Fraenkel, Die aramäischen Fremdwörter im Arabischen,
p. 154 n., says that the Arabic place name probably originally meant
“vine” = Syr. sbmD. If this information about the Syriac is correct, then the Hebrew
place name may simply mean “(place of) vines”. But I have not found any trace of a
Syriac sbmDwith the sense of “vines”.
45 J. H. Greenberg, “The Patterning of Root Morphemes in Semitic”, Word vi
(1950), 162-81.
 UGARITIC AND HEBREW “SBM”? 405
noun ‫ ;יבם‬and the place name ‫ שבמה‬may be considered in the same
way, as a basic noun. But it is hard to consider the South Arabian
s-b-m as other than a true original verb.
A second consideration is the possibility that a form like the
Ugaritic sbm might be a s-causative; in the form Pistbm one would
then have either a root t-b-m or more probably a verb with prefix
Dist -. The root might be b-m , b-w -m or b-m -m , or something like
these. One thinks, for instance, of the well-known Hebrew ‫במה‬
“high place” and cognates, apparently from a root b-w -m but with
no verb existing.46 (The argument now has nothing to do with
Jerome’s etymology of Sibmah, mentioned above.) Could sbm
conceivably be such a form?
The question may perhaps fall into two parts. One possibility is
what we may call a productive s-causative, i.e. a word where a
causative may be formed by a standard grammatical transforma-
tion, like the formation of a hiphil in Hebrew. The other is the
possibility that at a previous stage of the linguistic history a root
something like b-m has had a s-causative and that this has eventu-
ally passed over into the state of a normal qal with the three
radicals s-b-m , being thus transferred from the grammatical to the
lexical side of the language. This would be like the case of (say)
‫שקץ‬, if such a theory of this and like words is accepted.47
One implication of any idea involving a s-causative should be
observed. Theoretically, the s-causative is perfectly possible for the
Ugaritic sbm . It would of course rule out completely any con-
nection with Akkadian pa sd m u , napsam u ; but we have already
more or less ruled that out on other grounds. More important, it
would rule out the connection with South Arabian s-b -m : the
sibilant in this South Arabian word is s 2, which corresponds
etymologically to Hebrew sin , while the causative prefix is 51‫׳‬,
which corresponds to Hebrew sin.A%Thus, while the Ugaritic sbm
could, so far as is determined by the sibilant, be a causative, this is
not a possible interpretation of the South Arabian. The same

46 Driver, C anaanite M yth s a n d L egen ds , p. 165a n. 5, thought of a root b-w -m

“to swell” or the like, which would unite the senses “high place” and “belly, waist,
back”, comparing Deut, xxxiii. 29 (NEB “trample their bodies under foot”).
47 Cf. J. Alberto Soggin, “Tracce di antichi causativi in $‫־‬realizzati come radici
autonome in ebraico biblico”, A n n a li of the Istituto Universitario Orientale
di Napoli, xv (1965), 17-30.
48 Cf. A. F. L. Beeston, E pigraphic S ou th A ra b ia n , pp. 4, 14, 19.
406 UGARITIC AND HEBREW “SBM‫? ״‬
argument would imply likewise the abandonment of any connec­
tion with the Hebrew place name Sibmah (with IP).
One small point: in one of the Ugaritic passages where our word
sbm was recognized, not the passage so far cited, the verb was read
as p s tb m , and in the very next line the word y b m occurs. Because
of this circumstance I long considered the possibility that the two
were connected; if this were so, it would support the idea that the
i, and indeed the s t , were a prefix. If, however, the text should not
be so read (see below), this argument loses its importance.

v
We are now in a position to look again at the Ugaritic evidence
itself. The relevant passages appear to be three, since we shall leave
aside the passage CTA 3, B, col. ii, line 16, which does include the
sequence sb m , but where it may probably be taken as from sb “old
man” or the like.49 The text there is:
mtm. tgrs
sbm. bksl. qsth. mdnt.
If this may be left aside, we turn to our three passages:
A. CTA 3, D, col. iii, line 37 (p. 17); UT, cnt III: 37
This is the passage already cited at the beginning of our present
article: I cite it again with a slightly fuller text.
34 mn. Dib. y p c. lbcl. srt
Irkb. crpt. Imhst. m dd
Dil ym. Iklt. nhr. Dil. rbm
37 Pistbm. tnn. Distmlh
mhst. bin. cqltn
slyt. d. sbct. rDasm
The important line 37 has several textual questions in it:
(a) The second n of tnn is not clear; but the reading tnn is
hardly to be doubted.
(b) The reading of the second verb, here printed as Distmlh
following Herdner, is pronounced by her as “très incertaine”.
Virolleaud had suggested a tentative Dis b m (l)n (l)h .50

49 So for instance Aistleitner, no. 2573.

50 Cf. Herdner, CTA, p. 17 n. 2.
 U G A R IT IC AND H EBREW “ SBM”? 407
(c) In place of Pistbm Aistleitner proposed to read P i < s t > s t
b m .51 The reasoning behind this is unfortunately not given in the
source, nor elsewhere so far as I know. It may have a connection
with the question mark which he, as we saw, set against the sense
“muzzle” for sbm; and it may have something to do with the
similar reading to be mentioned below under our Passage B.
B. CTA 6, col. i, line 29 f. (p. 39); UT, 49: I: 2

29 ]n. Dal[Di]y n b [cl

]hh. tst bm. c[
]zrh. ybm. Pilm
Here again there is a textual point. Where CTA prints tst b m ,
the first publication had p stb m ; see Herdner’s note 3 in CTA, ad
loc. The reading as here offered, and printed in CTA, is that
proposed by Virolleaud, Syria xv (1934), 236.52 On the assumption
that the reading is p s tb m , this instance has been entered in glos­
saries as a case of our sbm (e.g. in Gordon, UT, no. 2378); but if
the reading as printed above is correct, then there is no case of sbm
here at all. The context appears to contain none of the terms of
fighting and destruction, of sea and dragon, which formed the
environment in Passage A, and there is good reason for accepting
the view that no sbm was here.
C. PRU 11. 3 (p. 12),
especially the section:
5 Isnm. tlhk.
smm. ttrp
ym. dnbtm.
tnn. Isbm
tst. trks
10 Imrym. Ibt[
The tablet is a small piece and much is damaged. The reading
tnn in line 8 is widely accepted, but the text itself has Pan.
Since this passage has not previously been discussed here, we
shall consider it now. It is part of a short mythological tablet, of

51 Aistleitner, Untersuchungen , p. 83, no. 190, cited by Herdner, ibid. p. 17 n. 1.

52 He there wrote: “On peut proposer . . . ‘eile (cAnat) met son . . . dans le . .
408 U G A R IT IC AND HEBREW “ SBM”?
considerable obscurity. In the E xposé prélim inaire to PRU 11
(p. x), C. F.-A. Schaeffer writes of this text:
Il appartient également au cycle de la déesse Anat et contient cette
image poétique: “Les langues lèchent les deux; la lèvre (ou la langue)
tournée vers les deux.” On y trouve aussi une allusion aux deux queues
de Tannin, qu’Anat attache ou enchaîne au mrym de sa maison.
Auraitelle dompté Tannin et fait de ce monstre un gardien de son
temple?
It is interesting that Virolleaud in his own annotations on this
passage does not venture to attach a clear sense to s b m :
Les deux queues de Tannin (lire tnn), pour . . . , elle (cAnat) les dispose
et elle (les) attache au m rym , à la maison de cA[nat?]. Pour sbm , dans
un passage ou il s’agit également de cAnat aux prises avec Tannin, voir
V AB, 3, 37 [= our Passage A].
Thus Virolleaud, though himself a pioneer in the identification
of sbm in our passage A, does not leap to make the same identifi­
cation here.
For the understanding of sbm here two or three different pos­
sibilities come to mind, (i) Gordon, UT no. 2378, takes Isbm as a
negative plus passive participle, “unfettered, unmuzzled”, (ii) The
strongest case for a sense like “tie” would be made by taking Isbm
tst as parallel with trk s : “she puts in a tie, she binds”. We have seen
that the Arabic sense, from which the original identification in
Ugaritic was made, came closer to “thread” or “thin cord” than to
“muzzle”, (iii) But we have also seen that South Arabian evidence
suggested a possible sense “high, be high, go up”. If we follow this
up, we then could have a good parallelism also, taking Isbm in
parallel with Im rym : “she puts the tannin on high, she ties her to
the height, to the house o f . . .” (iv) There are yet other possibili­
ties, such as the retaining of the reading fa n instead of tnn
(improbable), or the reading of Isnm (“tongues”, as in line 5) for
Isbm. These will be left aside. At any rate, our passage C seems to
be consistent with more than one possible interpretation of the
sequence sb m . The third possibility here mentioned, giving the
parallelism between sbm and m rym , has the advantage of stronger
evidence from other Semitic languages than is available for other
senses one might suggest for sbm .
We return then to our passage A, which remains the main
passage for the discussion of our word. As has been pointed out,
 UGARITIC AND HEBREW “SBM”? 409
the sense “muzzled” does not fit particular well. The fact that the
word we have printed as Distm lh is uncertain decreases in its turn
the assurance with which we can explain Pistbm ; for the similarity
of form between the two words suggests that they have a close
connection. As we have seen, the environing words are distinctly
verbs of annihilation and destruction. If the sense of sbm were
basically something like “be high”, then it is at least possible that
the meaning in this line would be something like “lift up”, “remove”,
“get rid o f”. This would be comparable with familiar usages of ‫סלק‬
“go up”, developing into ‫“ הסתלק‬be removed, taken away”, or
Akkadian elum “go up” but *etlum “go up and away from; get rid
of (a thing)”.53 Such a suggestion for our Ugaritic text may seem
both speculative and indecisive; but in fact it has much deeper
rootage in evidence from the Semitic languages than the “muzzle”
sense recently in vogue.54

VI

In conclusion, then, it seems that only rather tentative ap-

proaches to the sense of Ugaritic sbm can be made; and the sense
“muzzle”, which has widely been attributed to it, is quite pre-
carious, depending as it does not on the Ugaritic material itself but
on an Arabic comparison, a comparison which has been shown to
have a particularly thin line of attestation behind it. It follows that
attempts to reconstruct a ‫ שבם‬in Hebrew also, on the basis of this
Ugaritic, must be deemed precarious. This will be true a fo rtio ri of
attempts to carry out further reconstructions on this basis, e.g. to
read the obscure word, here printed after CTA as Distm lh , as
Disbhnh 55—an attempt which at the same time uses the Hebrew
‫“ שבח‬to still, silence” to restore the obscure Ugaritic text, its
suitability there being supported by the parallelism with the Ugaritic
sb m , taken to mean “to muzzle”.
The fact is that the Hebrew Ps. lxviii. 23 is not a particularly
obscure or difficult verse, and it is certainly very exaggerated to

53 Von Soden, G rundriß der akkadisch en G ram m atik , §92e, p. 121; cf. A H w ,
p. 208a.
54 As has been seen, if one is really to work from North Arabic the search must
basically be directed towards a sense connected with coolness of water or the like;
but it is difficult to make sense of the Ugaritic on this basis.
55 Dahood, U garitic-H ebrew P h ilology (Rome, 1965), p. 20.
410 UGARITIC AND HEBREW “SBM”?
call it “meaningless”. The problems of its meaning are not lin-
guistic ones in the strict sense; all the terms used are familiar and
normal ones, which furnish acceptable sense when taken as they
are customarily taken in Hebrew. The problem is not an intrinsic
linguistic difficulty, such as can be found in many biblical passages.
It is rather one of {a) ambiguity, and (b ) context. Does “bring
back” refer to a saving action, a bringing back from danger, or to a
punishing action, a bringing back from escape in order to face
retribution? Is basan the place Bashan, or is it the word for a snake
or dragon? If the place Bashan, what are the associations intended
here, and what are the relations then with the “depths of the sea”?
Who or what is the object of the “bringing back”?
It is at least a probable filling out of the linguistic bones of the
verse if we suppose that the persons brought back are the enemies
of God, mentioned in the previous verse, their blood to be spilt in
the following one; the basan is the mountain of Bashan, quite
likely to be identical with Hermon, and in any case twice spe‫־‬
cifically mentioned in the nearby v. 16 as “the mountain of God”.
The two places in the verse represent the utmost height and the
deepest abyss.56 The thought has some analogies with Ps. cxxxix. 8,
where also it concerns a flight from God, though not the flight of
an enemy of God: God’s hand will grasp him wherever he is. It is
not necessary however to insist on this interpretation, which is one
among several that could be given.
No more, in any case, needs to be said about the Psalms passage.
Let it be observed only that this argument has nothing to do with a
defence of the traditional understanding of the Hebrew as against a
meaning suggested on the basis of the Ugaritic. On the contrary, it
is the sense of the Ugaritic itself that is in question. The sense
“muzzle”, though widely accepted for Ugaritic, was, it seems, never
established on a thoroughly sound intra-Ugaritic basis and never
thoroughly researched for its probable traces throughout the Semi-
tic languages. The sense “muzzle”, far from being established on

56 Cf. already H.-J. Kraus, Psalm en /(Biblischer Kommentar), p. 476. He makes

the very apt comparison with Amos ix. 3, where the contrast is between Sheol and
Carmel, as well as the depths of the sea, where the dragon is ready to bite the
enemies of God. The verbs there, it can be added, provide very close correspondents
of ‫“( ^ו ב‬from there” God will “bring them down”, ‫“ ;ירד‬from there” he will “seek
and take them”, ‫“ ; ל ק ח‬from there” he will “command the serpent and it will bite
them”).
 UGARITIC AND HEBREW “SBM”? 411
the basis of Ugaritic, was identified for Ugaritic itself through
Arabic—the very language which some today are contrasting very
unfavourably with Ugaritic as a source of suggestions for meanings
in Hebrew!—and within Arabic through a hasty identification on a
very weak base.
But what is striking about this identification of the sense of
Ugaritic sbm is the strength which it has already, over a few
decades, generated in our minds. Meanings in ancient languages
were passed on in a tradition; and there is some uncertainty today
about the validity of that ancient tradition. But even in languages
like Ugaritic, of which we have no ancient tradition of meanings, a
tradition still very quickly establishes itself; and we believe that
sbm means “muzzle”, because this is part of the scholarly lore in
which we have grown up. Even when we are forced, as I think we
are forced, to consider other possible meanings, the thought that
sbm must after all mean “muzzle” intrudes upon the mind and is
difficult to get rid of. This present writer, if a personal note will be
permitted, did not begin with any antecedent scepticism towards
the “muzzle” sense of sbm ; on the contrary, the example seemed a
striking one, and one deserving to be set out as an instance of what
can be achieved. In the writing of Comparative Philology and the
Text o f the Old Testament I originally proposed to use it as a
favoured illustration. It was the peculiarity of the form as a Semitic
form, under the rules of patterning and compatibility, that most
induced me to investigate it more fully, although in the end this
aspect has been given only a brief mention and has not been taken
as a major issue. In the end it is the thinness, and the oddity, of the
North Arabic evidence, that does most to damage the now tradi­
tional identification, and unless stronger Arabic evidence can be
brought forward in its support this identification must probably be
abandoned. The fact that the much stronger evidence in South
Arabia, well evidenced and in ancient sources, leads towards a
quite different meaning (even if we cannot be precise what that
meaning was) must force upon us a change of mind.
 LIMITATIONS OF ETYMOLOGY AS A
LEXICOGRAPHICAL INSTRUMENT
IN BIBLICAL HEBREW1

“Etymologists, a race not wanting in effrontery . . . ”

(Tylor, Primitive Culture, i. 397)

T h ispaper is not a highly theoretical exercise, but rather a discus­

sion of practical questions that arise in Hebrew lexicography. It
consists to a large degree of reflections occasioned by two pieces of
research. The first is my own experience as editor of the Oxford
Hebrew Dictionary} The second is the interesting and useful
Dictionnaire des racines sémitiques of David Cohen,3 a work which
began to appear in 1970 but has as yet covered only a very small
segment of the relevant vocabulary (in fact, at the time of writing,
only the first two letters of the alphabet and a portion of the third).
In the modern tradition of biblical Hebrew lexicography, com­
parative and etymological considerations have been accorded a
prominent place. The policy of separating the comparative and
etymological matter from the description of actual forms, usages
and meanings, by writing a completely separate etymological dic­
tionary quite distinct from the normal registration of the language—
in other words, the course that has been taken as normal in many
western languages—has not been seriously attempted and, though
it remains theoretically possible, looks unlikely as a practical
eventuality. The current and accepted practice is that comparative
and etymological material is integrated into the entries of the main
dictionary of the language. Not only so, but articles are so struc-

1 This is a revised version of a paper read to the Philological Society at Worcester

College, Oxford, on 9 June 1979. [Reprinted with permission from Transactions o f
the Philological Society o f Great Britain, 1983 (Oxford: Basil Blackwell) 41-65.]
2 Since this paper was originally read, it now appears unlikely that the dictionary
as planned will be completed; but it is hoped that some aspects of the research done
for it will be published separately.
3 David Cohen, Dictionnaire des racines sémitiques (Paris and The Hague:
Mouton. Fascicule 1, 1970: Fascicule 2, 1976).
 L IM IT A T IO N S O F ETY M O L O G Y 413
tured that this material has a prominent position. It is commonly
the first main item in an entry, after the lem m a itself, and some­
times after other very basic information such as the gender of
nouns or the basic gloss indicating meaning; in the German-
language dictionaries like WB it generally precedes even these.4
Thus a representative entry from WB begins thus:5
D-k-l\ mhe., ug. kan. ph. jaud. aam. pehl. aga. u. iam. (DISO 12), ba. ja.
cp. sam. sy. md. (MdD 16b) ar., akk. (akalu)\ ath. u. tigr. (Wb. 376a)
Dekel Korn.
The Hebrew word in question is the common verb ‘eat’. The
meaning of the paragraph quoted is: the following is a list of
languages and dialects which have a cognate form comparable with
the Hebrew word under consideration. In a few cases the actual
forms in these languages are cited, and in one case there is a note
on the meaning of the cognate word because it does not cor­
respond exactly with the Hebrew word: this is noted for Ethiopic
and Tigre because in these languages the verb ‘eat’ does not come
from this stem but from another, and only the noun meaning ‘corn’
survives from this root. Our point, however, is: all this rather
diverse and possibly confusing information appears in the Hebrew
dictionary at the very beginning of the entry, before the reader has
even learned the rather basic information that the Hebrew word
itself is the common verb meaning ‘eat’.
If we were to ask for the rationale of this order of presentation,
it might well be answered that it is, at least in part, a comparative-
historical one: an entry presents first of all the comparative-
etymological matter, which gives a sort of picture of the prehistory
of a word (actually, a picture of its distribution and meaning in

4 I use WB as the siglum for the third edition, edited by W. Baumgartner, of

L. Koehler and W. Baumgartner, Hebräisches und aramäisches Lexikon zum Alten
Testament. The original work was entitled Lexicon in Veteris Testamenti Libros
(Leiden: Brill, 1953), and is often styled KB: Koehler did the Hebrew, the larger
portion, and Baumgartner the Aramaic. The third edition was begun by Baum­
gartner himself, after Koehler’s death; also published by Brill of Leiden, its first
Lieferung appeared in 1967 and its second in 1974. About one half of the work has
thus appeared in the third edition. The other standard academic dictionaries of
biblical Hebrew are considerably older. The standard English-language work,
Brown-Driver-Briggs, was completed in 1907, and the corresponding German-
language work, Gesenius-Buhl, in 1915. These are commonly indicated by BDB and
GB respectively.
5 WB, p. 44b.
414 LIMITATIONS OF ETYMOLOGY
cognate languages and dialects, from which, however, some histori­
cal projections may perhaps be made), and then it goes on from
that to the usage and meaning in extant Hebrew texts.
But this basic philosophy of the matter has, of course, not gone
unchallenged in recent decades, and quite contrary philosophies of
the question are to be heard. An extreme contrast is provided by
the assertion that etymology should be excluded from dictionaries
altogether: it may be an interesting sort of speculation, one hears it
said, but the dictionary is not the place for it to be carried on. Less
extremely, it may be argued that, while etymology can and should
be practised within the dictionary, it should be done as a footnote
or as an optional extra. In that case its subsidiary place should be
recognized by the consignment of it to the end of the entry, there
to represent an additional remark kept well out of the way, so that
the basic analysis of forms and meanings in the texts is done first
of all, without any interference from etymological or comparative
considerations, and these matters come in, if at all, only as an
afterthought. Among major academic dictionaries of ancient Semi­
tic languages, the one which has most fully exemplified this mode
of operation is the imposing many-volumed Chicago Assyrian
D ictionary. (The other contemporary dictionary of Akkadian, by
von Soden, gives brief and modest etymological indications, placed
at the beginning of the article in the manner of the Hebrew
dictionaries, but as a rule less conspicuously than in the latter.)
There are many temptations for the editor of a Hebrew dictionary
to follow the Chicago example. To place the comparative material
at the end of the article is to reduce its prominence, to make it
more optional, to make it subsidiary to the analysis of the Hebrew
itself, and to avoid the danger, to which many articles in the older
Hebrew dictionaries certainly fell a prey, of allowing the semantic
analysis of the Hebrew to be biased by the comparative informa­
tion already given.
Before we decide immediately to adopt this newer fashion,
however, it is good to consider the complications of the subject
and ask just what are the considerations which have in the past
moved lexicographers of Hebrew to pay the attention they did to
considerations which appear to be comparative and etymological.
It is not necessarily the case that biblical Hebrew is in the same
position as Akkadian, and it is not necessarily true that the ‘etymo­
logical’ interests of dictionaries have had no other basis or function
 LIMITATIONS OF ETYMOLOGY 415
than historicist and comparativist curiosity. Behind the interest in
evidence from cognate languages and dialects, and other such
material, there lie, I suggest, questions of inner-Hebrew relations
which have to be clarified; and it is because they are thought,
rightly or wrongly, to help with the clarification of these inner-
Hebrew problems that comparative and etymological materials
have been so constantly introduced into prominent parts of the
entries of the typical dictionary of Hebrew.
One of the questions here involved is that of the relation between
root lexemes and actual words. As is well known, it may be said
that in Hebrew and in many other Semitic languages a word is
constituted by the compounding of a discontinuous root lexeme,
e.g. k-t-b ‘write’, and a pattern of affixes and vowels, e.g. m iC\ C 2
aC$, which generates the actual word found, i.e. m iktab ‘letter’,
‘writing’; or, with another pattern, the verb form katab ‘he wrote’.
The existence of these root lexemes remains rather conspicuous in
the structure of the language and is given additional emphasis
through the script and the spelling conventions. Therefore, it may
be argued, part of the process of essential identification and analy­
sis of m iktab is to perceive that it belongs to the root lexeme
k-t-b. Would we say that anyone really knew Hebrew if he knew
that m iktab meant ‘letter’ but did not know that it contained the
same basic lexeme as the verb ‘write’? In other words, in asking, in
a case like this, what is the root lexeme of a word, we are not
necessarily, and not in fact, indulging in mere historicist or
comparativist speculations but we are recognizing the fact, neces­
sary for the normal operation of the language, that the lexicon
works on two levels, that of the root lexeme and that of the word
itself.
But the difficulty comes along as soon as the semantic path
between the root lexeme and the word becomes cloudy, within
Hebrew itself. There is a well-known m aPak which means ‘angel,
messenger’ and a well-known m eldDka which means ‘work’ or the
like. The lexicographer and the dictionary user perceive that these
must contain a common root lexeme l-D-k\ but no such lexeme
appears to exist in Hebrew other than in these two forms, and
there is no recognizable semantic content which obviously attaches
to such a lexeme within Hebrew. Therefore, unless the lexicog­
rapher is to confine himself to the two disparate words and offer
no account of the lexeme common to them and yet unknown as
416 LIMITATIONS OF ETYMOLOGY
such within Hebrew, he is forced to consider the cognate lan­
guages. The existence of words like the Ethiopic la^aka ‘send’ then
appears to confirm the existence of a common lexeme and to give
it a meaning which does something to make sense of the Hebrew
phenomena. (The sense ‘messenger, angel’ is fairly easily derived
from ‘send’, probably through an intermediate stage ‘something
sent’; the sense ‘work’ is by no means so obvious, but within
Hebrew itself the phrase mislah y a d ‘sending out of the hand >
undertaking > work’ might provide a reasonably close analogy.)
Thus the fundamental fact is that some of the root lexemes of
Hebrew are known and identifiable in meaning from within Hebrew
but others are not. The function of comparative and etymological
material quoted within dictionaries has often been not to depict
hypothetical historical origins or prehistoric processes of change
but to make some sort of sense of relations that appear to exist
within the Hebrew lexicon but seem not to make sense on the basis
of its obvious internal relations. Its prime function therefore is not
comparative in direction; rather, it strives to provide answers to
questions of what might be called the inner etymology of Hebrew,
the relations between its root lexemes, their meanings, and the
actual words. This problem arises for the lexicographer whether he
lists the root lexemes as actual items, in addition to the word-
entries, or lists only the latter, including within them references to
the root lexemes.
If one function, then, is to bring together root lexemes and
words and to establish significant relations between them, another
is to separate them. Hebrew has a large number of homophonic
and/or homographic words, and this is so both at the word level
and at the level of the root lexemes. The making of a distinction
between pairs, triplets or even quadruplets of such homonyms is
often an essential for the provision of an intelligible account of the
lexical stock. Dictionary makers do this by the accepted conven­
tion of numbering the homonymic lexemes. I hebel is said to be
‘cord’, etc., from the root lexeme I h-b-l ‘tie’; II hebel is supposed
to be ‘destruction’ from the root lexeme II h-b-l ‘act corruptly’. But
how do we know that these are really two separate root lexemes
and/or two different words, and not just one big word which
means both ‘bind’ and ‘destroy’? Working strictly from within
Hebrew we might have difficulty in deciding. Arabic and other
cognate terms suggest that the homonymy was produced by pho-
 L IM IT A T IO N S O F ETY M O L O G Y 417
neme merger: two phonemes which were distinct in a pre-Hebrew
stage, and which in Arabic are still distinct, were merged, produc­
ing eventually homophony and homography in Hebrew. The first
term is cognate with Arabic habl ‘cord’, the second with khabala
‘confound, confuse’. The same process enables us to distinguish
I c-n-h ‘answer’ from (say) IV c-n-h ‘sing’ (cf. Arabic cognates
cana and ghanna respectively), when the Hebrew contexts them­
selves could quite easily have allowed confusion to continue (in the
contexts it remains quite plausible to imagine that Miriam
‘answered’, i.e. spoke antiphonally or the like, rather than that she
‘sang’, with no component of ‘answer’ in the meaning at all, at
Exod. 15. 21, and indeed many people on reading the text consider
this to be the obvious meaning). Thus, to sum up, in cases of this
kind it seems that assistance from comparative studies does not
mean a departure from the strict study of Hebrew for itself but
rather constitutes a means of sorting out the Hebrew material and
making clear its own inner contours.
Let it be supposed, then, that these considerations in some
measure justify the involvement of the Hebrew lexicographer in the
use of comparative and etymological material; and, if this is so,
since this work involves the basic recognition of relations between
root lexemes and word forms, and the separations between one root
lexeme (and/or word) and another, with semantic consequences—
all of which might have to be modified if a full account of the
problems were to be given, but let it be granted for the purposes of
the present paper—then let us look at some of the practical con­
sequences that seem to follow. In the past I have at times been very
critical6 of the traditional emphasis on comparative and etymologi­
cal study because of its unfortunate semantic effects; that is, I have
warned against attempts to take the etymology as if it indicated the
‘origin’ or the ‘basic meaning’ of a word or dominated its actual
usage and semantic functioning. But that is not the point of the
present article. My point is this: granting that the incorporation of
comparative-etymological material within the Hebrew dictionary
may well be justified through the functions which I have sketched

6 See in particular my S em an tics o f B iblical Language (London: Oxford Uni­

versity Press, 1961); B iblical W ords f o r Tim e (London: SCM Press, 2nd edi­
tion, 1969): ‘Etymology and the Old Testament’, O udtestam en tisch e Stu dien 19,
1974, 1- 28.
418 L IM IT A T IO N S O F ETY M O L O G Y
out above, it remains a question whether these functions can be
adequately discharged in anything like a satisfactory degree, simply
because suitable information is not available. For the fact remains
that the etymological relations of many words remain very uncer­
tain. As I look back over the articles that were prepared for the
Oxford Hebrew Dictionary, I am myself surprised at the number
of cases in which I had to enter ‘etymology unknown’, ‘apparently
no significant cognates in Semitic’, and other such negative con­
clusions. Most of this present article will be a series of illustrations
of this theme. I am sometimes tempted to the gloomy conclusion
that the only etymologies that are reliable are the ones that tell us
nothing useful.
For instance, no one will doubt the etymological validity of the
example with which we started: the Hebrew Dakal has cognate
verbs, or other related forms, in most languages in the Semitic
family, and all of these mean ‘eat’, which only goes to confirm that
the Hebrew word itself means ‘eat’, which we knew all the time in
any case. The etymology is unquestionable and unquestioned but
we learn nothing from it, or nothing that applies particularly to
this Hebrew word. The same is true of Dozen ‘ear’. For this there
are precise cognates throughout the Semitic family: Akkadian uznu,
Arabic Dudhn and so on. This is how it is set out in the WB entry
(WB, p. 27a):
Dozen (188 x): Lks, mhe., ag. ydn (EG 1, 154); ug. Ddn, aga. Ddn , pehl.
Dwn (D ISO 5 ), ja. md. cudnaD, M dD 342 b, akk. uznu, ar. Dudn, iraq-ar.
Didn (BzA 5 , 112), ath. Dezen, tigr. Dazzan langohrig (Wb. 381 b); aram.
Grdf. Didn (Friedrich Or. 12, 20 ): sy. cp. nam. Dedna . . .7

All this adds up to the same thing: though there are differences
in form, agreeing with the normal correspondences between vari­
ous languages of the Semitic family, plenty of them have a word
that provides a more or less complete formal correspondence with
Hebrew Dozen, and all of them, amazingly, mean ‘ear’. Once again

7 The legend DISO 5 in the second line should be DISO 6. The reference is to
Jean and Hoftijzer’s standard Dictionnaire des inscriptions sémitiques de l’ouest
(Leiden: Brill, 1965). As in the previous citation from WB, the abbreviations stand
mainly for various Semitic languages and dialects. The only one that may surprise
the reader is ‘pehl.’. This refers to the use of an Aramaic form as an ideogram in the
writing of Pehlevi: though such ideograms function within Pehlevi language, their
existence gives evidence of usage and meaning within Aramaic, from which they
were taken.
 L IM IT A T IO N S O F ETY M O L O G Y 419
the etymological correspondence is totally convincing, but it
remains otiose: there is no gap in our organization and under­
standing of the Hebrew lexical stock that is filled up by it. Actually,
the most valuable service that this information furnishes is not
lexical at all but phonological: precisely because the meaning of all
these words is so clearly identical, so that there is no doubt that in
that sense they are ‘all the same word’, they—along with a host of
other such pieces of evidence—provide good demonstration of the
normal phonological correspondences between members of the
Semitic family, for instance, good exemplification of the fact that,
where Arabic has dh (d ), Hebrew normally has z and Aramaic has
d. This is important and has certain lexical implications, and
indeed the Hebrew lexicon cannot be intelligently used by the
reader who does not know that these correspondences are normal;
but in itself it is not lexical information, applicable to the word
‘ear’ or indeed to any word any more than any other word. About
the word ‘ear’ in particular it tells us nothing.
Why then is all this information here at all? Partly it comes from
a desire for consistency: perhaps in some other word something
important and necessary will be disclosed through the citation of
the Mandaean form or of the form implied by the Pehlevi ideo­
grams, and therefore these forms must be cited in every case where
they are known. In part it comes from a compilation concept of
scholarship: the task is to gather the material from other dic­
tionaries and put it into this one. Instead of classifying the material
and separating out what is significant, everything is put in. No one
can seriously suppose that the citation of half a dozen or so
different forms of Aramaic/Syriac makes any difference to our
understanding of this Hebrew word. Moreover, the compilatory
conception of lexicography has the further disadvantage that it is
insatiable and unfulfillable. If Iraqi Arabic is to be specially men­
tioned, what about Egyptian, and if it, what about Moroccan? If
among Ethiopian languages Tigre is to be specially cited, then
what about Tigrinya and Amharic?8 What about Socotri and the

8 In fact the reasons for the preferences shown in WB are often practical: that is,
where languages (or dialects) have a modern dictionary or glossary, easily accessible
and usable—and especially if in the German language—then they are more likely to
receive attention. Thus Tigre is favoured because of the excellent and convenient
work of E. Littmann and Maria Höfner, Wörterbuch der Tigre-Sprache (Wies­
baden: Steiner, 1962).
420 L IM IT A T IO N S O F E TY M O LO G Y
South Arabian languages? Moreover, among the many dialects and
languages listed as having a cognate to the word for ‘ear’, are there
any which, in addition to it, have a term of some other meaning
which nevertheless appears to belong to the same root lexeme and
which therefore might suggest something about pre-Hebrew devel­
opments of meaning?9 However, our purpose here is not to criticize
this carefully compiled list for its inadequacy or for omissions of
material which ought to be in it if it is to be so lengthy anyway.
Our purpose is rather to indicate that, assuming the adequacy and
correctness of the impression given by this section of the dictionary
entry, it still does not tell us anything about the Hebrew word
for ‘ear’.
Now it can easily be said that, while etymology is rather unim­
portant for the understanding of well-known and well-established
words such as this one, there are many words that are rare or
poorly evidenced for which etymology may be our only way of
finding a meaning. And as a general statement I do not dispute
this; but as a matter of practical experience I am more doubtful
about it than I used to be. What surprised me in my work on the
Hebrew dictionary was the considerable number of words, some of
them rare words, or doubtful words, some of them hapax legomena,
but also some of them quite basic and frequent words, for which
we gain very little guidance, or only very ambiguous guidance,
from the consideration of possibly cognate words in other Semitic
languages; and yet few of the words I have in mind are words of
which one might reasonably postulate origin as loanwords from
outside Semitic. Again and again we find that either there is no
word formally cognate, or that the senses are so remote that no
useful etymological contact can be made.

9 The most obvious such term, as applied to this particular entry, is the wide­
spread group of words like Arabic Didhn ‘permission, authorization’, with which is
connected the form familiar in English as muezzin ‘one who calls to prayer’. The
question whether this group forms the same basic lexeme with the words meaning
‘ear’ is obviously relevant for the lexical understanding of both; but the WB entry
gives the dictionary user no hint of these facts. Contrast Cohen, p. 10. The
relevance of this for ancient times and the neighbourhood of ancient Hebrew
culture is indicated by the presence in ancient South Arabic of meanings like
‘obedientia, iussum’ and ‘permisit, licitum fecit’ attaching to this root: see. K. Conti
Rossini, Chrestomathia Arabica Meridionalis Epigraphica (Rome: Istituto per
l’Oriente, 1931), p. 100.
 L IM IT A T IO N S O F E T Y M O L O G Y 421

The first example has a certain loanword element in it, or may

have. It is Ded, which appears in the familiar verse Gen. 2. 6: an Ded
used to go up from the earth and water all the face of the earth. It
has become common opinion that this word is derived from, or is
cognate with, the Akkadian word edu usually said to mean ‘flood’,
and itself said to be a loanword from Sumerian: it is thus the
welling up of an underground stream, of groundwater or some­
thing of the sort (contrast the traditional ‘mist’ of the AV). This is
given as the etymology, and therefore ‘Susswasserstrom’ or
‘Grundwasser’ as the Hebrew meaning, by the recent dictionary
WB, and so also many commentaries on the passage.
Deeper research into the Akkadian, however, seems to make this
comparison doubtful. As CAD (E, p. 35 f.) makes clear with
numerous examples, edu means something far more violent and
catastrophic than can be related to the Hebrew passage. CAD
gives as its main gloss onrush o f water, high water: it is something
like a huge wave that may sink a ship, or again it is ‘the high tide
of the sea’ which can overwhelm a camp. Used of rivers, edu may
be its high flooding, but CAD emphasizes that this is a rare and
catastrophic phenomenon (by contrast with rriilu which is the
normal annual high water). When a hymn to Marduk refers to that
deity as bel kuppT naqbT e-di-e u tamati ‘lord of sources, springs,
high waters and seas’, it is the kuppu and the naqbu, the source or
spring, that might have fitted the Hebrew passage, while the edu is
a phenomenon of the high seas. Thus all in all it is far from clear
that edu gives a suitable parallel which will in turn make it into a
good etymology for the enigmatic Hebrew Ded of Gen. 2. 6. But
if that is so, and if the word really means some sort of water
rising from the ground, then the lexicographer has no choice but
to enter under this word the legend ‘etymology unknown’, for
there is apparently no Semitic evidence other than that of the
Akkadian edu.10

10 Another option, indeed, is to say that Ded is indeed derived from the Meso­
potamian culture but represents not the Akkadian edu, for the reasons just out­
lined, but the Sumerian id from which it is said to be derived and which commonly
means ‘river’, being translated into Akkadian as naru with this meaning. The
Hebrew would then be derived directly from the Sumerian. This, while quite
possible purely linguistically, runs into the difficulty which will be mentioned below:
if it was a river, why did it not irrigate the soil?
422 L IM IT A T IO N S O F ETY M O L O G Y
It is indeed possible to go in another direction and understand
the word in the traditional sense as ‘mist’, the sense already familiar
to readers of the Authorized Version. But this would mean a
departure from all connection with Akkadian edu. The advantage
of this course is that it provides a good connection with another
known Hebrew word, the homonymic Ded of Job 36. 27, yazoqqu
mdtar le-Dedd. Though the exact relation of the elements here is not
clear, it is certain that the context is meteorological: we have terms
for ‘drops of water’, ‘rain’, ‘clouds’ and ‘drip’ all in the same pair of
lines. Perhaps the reference is, as Dhorme suggested, to water
droplets volatilized (verb z-q-q) ‘to become his mist’. It is, indeed,
precisely this passage that caused traditional sources to understand
the Gen. 2. 6 passage as ‘mist’ from the beginning.11
If we look at it in this way, we have only the one word for these
two passages, for it is hardly possible to suppose with Speiser that
the Akkadian edu is the parent also of the instance in Job.12 But if
our one word is the word ‘mist’ known from Job, it has no known
Semitic or other etymology anyway. It is probably better to have
two examples of one word, both meaning ‘mist’, and to admit that
there is no etymological information at all, than to have two
different Hebrew words Ded, each occurring only once, and one of
them related to the Akkadian word and meaning floods of water
from under the earth while the other means ‘mist’ but has no
etymology at all.
Even, however, if we count this as only one word, occurring
twice, we have also to reckon with the homophonic Ded, spelt with
yod, D-y-d, the meaning of which is well exampled and well known:
it means ‘calamity’. It is a question whether for this we have any
convincing etymology either. Traditional works cite the Arabic
root D-w-d with the sense either of ‘bend’ or of ‘burden, oppress’,
but the most that could be said for this is that it is a conceivable
speculation. The fact that the word, which occurs over twenty

11 Among the ancient translations of the Bible, the Greek (Septuagint) translated
the word at Gen. 2. 6 as πηγή ‘fountain, spring’, and the same was done by the
Syriac and the Vulgate; the Targums (Jewish versions in Aramaic) translated rather
as ‘cloud, mist’.
12 Speiser, Genesis (Anchor Bible: New York: Doubleday, 1964), p. 16. He seems
to agree that the case in Job does mean mist ‘or the like’, but thinks that this ‘need
signify no more than the eventual literary application of’ the rare word for ‘an
underground swell’. This is very difficult to accept.
 L IM IT A T IO N S O F E TY M O LO G Y 423

times in the Bible, is always spelled with yod may probably indi­
cate that it does indeed come from a root like D-y-d or D-w-d, and
thus belongs very likely to a different word type from our previous
Ded ‘mist’ (or whatever it is). Cohen relates it to an Arabic word
meaning ‘calamity’ and cited by him as muDyTd.u If this is valid, it
means there is a good Arabic parallel to our word, even if the
derivation from a root lexeme is as vague and uncertain in the
Arabic as it would have been in the Hebrew if we had had no
Arabic. In any case, though we do (if these facts are correct) have
some comparative and etymological information about our word,
we do not as a result know anything more about it than was
already manifest from the contexts in the biblical literature.
Another remark should be added about our Ded which came up
from the earth and moistened the surface of the ground. In a
language like biblical Hebrew, and in the case of rare words of this
kind, literary questions are relevant and one cannot proceed purely
linguistically. What is the sequence of thought in Gen. 2. 5- 6?
There was, we are told, no vegetation in the world: it was a dry
world, in which God had as yet sent no rain. But then, it tells us,
an Ded used to go up and used to wet the surface of the ground. It
is not easy to make good sense of this in the context. If the Ded was
some kind of spring or groundwater, welling up from underground
sources, it would surely have provided some irrigation, enough for
plants to grow. But according to the story it did not. Might this
perhaps fit with the idea that it was after all a mist? There was a
vapour which rose from the ground, it damped the surface, but it
did not provide enough water for the plants to grow. It only
damped the surface, perhaps thus making the earth pliable for God
to fashion man out of the soil (Ccipar, traditionally glossed as ‘dust’,
is not necessarily dry like our ‘dust’, it is often more like soil).
Admittedly this explanation is far from perfect: even the damping
of the earth by mist would probably be enough to sustain plant
life. Perhaps the writer discounted the irrigative value of mist: for
him only rain was enough to sustain proper plant life, and espe­
cially a garden. If, however, we cannot go this way, and if the
explanation as ‘mist’ is impossible, then perhaps we have to say
that the piece about the Ded is a fragment, perhaps from some
previous legend, quite out of context in the present story. If this
13 Cohen, p. 16.
424 L IM IT A T IO N S O F ETY M O L O G Y
were so, then of course it could be a great wave or onrush of water
like the Akkadian edit. Indeed some of those who have followed
that explanation and considered that it was a flood of groundwater
or the like have—when they have been careful and observant
enough—come to the conclusion that this is so, and that the
sentence in question has no proper context in Genesis at all.14
Then, of course, it could mean anything. But that must, I think,
count as a very negative and desperate conclusion.
Let us move to another example: take the Hebrew word Degrop.
Exod. 21. 18 is a law about quarrelling. If a man strikes another
man with a stone or with a/the Degrop, then certain consequences
follow. The traditional sense is ‘fist’; but there is also a long
tradition of interpretation as an implement: a shovel, spade or the
like, even a broom. The NEB goes in this direction: ‘i f . . . one hits
another with a stone or with a spade’ (adding ‘or fist’ in the
margin). One reason for this preference lay in inner-Hebrew
etymology: the verb garap means ‘sweep’ or the like, and the noun
magrepa, surely derived from it, is a shovel. The etymological
sense, seeking to bring these together, tended to make Degrop into
the same sort of thing, hence a broom, shovel or spade.15 But lately
opinion has swung, perhaps decisively, back towards the fist: for
the Zadokite Document, discovered early this century, includes a
law that a man must not raise his hand to strike a beast with a/the
Degrop on the Sabbath. Since it was forbidden to carry any imple­
ment on the Sabbath in any case, the law must mean that it was
forbidden to strike without an implement, i.e. with the fist. This is

14 The careful and detailed commentary of C. Westermann, Genesis (Neukirchen-

Vluyn: Neukirchener Verlag, 1974) accepts that the original context of the sentence
is unknown but was certainly different from that of the present Gen. 2. 4- 6: see his
p. 273. Many other commentators seem, like Speiser, to proceed as if this question
did not exist at all. When he tells us, innocently, that ‘the biblical verse might have
been lifted verbatim from an Akkadian lexical work’ (Bulletin o f the American
Schools o f Oriental Research 140, 1955, 9- 11), he is really openly proclaiming that
his interpretation, however well founded in Mesopotamian research, can make no
sense in the context of the Book of Genesis. But he goes on as if this constituted no
difficulty. Cf. Westermann, ibid., p. 274.
15 BDB, p. 175b said clearly ‘fist’, in spite of its tendency, visible in many other
words, to be over-persuaded by etymological considerations; GB, p. 8b was
decidedly vague and went no farther than to say that it was an ‘uncertain word’.
Koehler in KB, p. 10a came down firmly on the side of ‘broom, shovel, rake’; but
Baumgartner, WB, p. 11, clearly considers that the Zadokite Document swings the
meaning decisively round to ‘fist’ (just at about the same time as the NEB preferred
‘spade’!).
 L IM IT A T IO N S O F ETY M O L O G Y 425

CD 11. 6; cf. Rabin’s note.16 But, if the term really means a fist,
then there is no clear or meaningful connection with g-r-p ‘sweep’.
The distance one has to go to find an etymology is well shown by
Rabin himself, who, deciding for the sense ‘fist’, offers as compari­
sons the Socotri grff ‘courber’ (i.e. ‘to bend, to crook’) and the
Amharic gdrfaffa ‘obese’. Such comparisons, however, are so
extremely remote—in time, in linguistic geography, and in mean­
ing (plus, in the case of the second of the two words, the fact that
the word seems not to exist at all)17—as to constitute a clear
admission that we know of no etymology at all. If Degrop means
‘fist’, as seems very probable, there is no more to be done than to
say that this is the meaning and that no connections of significance
with words anywhere else are known.
These are not isolated examples: this happens again and again.
Take the words Dakzar, Dakzdrl, DakzarTyut, tolerably well-evidenced
words, understood to mean ‘cruel’ and ‘cruelty’. There are a dozen
or so cases in the Bible plus some in the Dead Sea Scrolls. Usage
and meaning are fairly clear; but the fact is that we have no
etymology to offer for this word, there seems to be just no sub­
stantial cognate evidence.
The same happens with the fairly common word Deden, which
designated a sort of base or socket, upon which were set the planks
or pillars of the Mosaic tabernacle. It has been customary to
connect this with Akkadian adattu which, in turn, was supposed to
mean ‘nest’ or ‘resting-place’.18 But the CAD (A, p. 110) now
makes it clear that adattu does not mean ‘nest’ or anything similar:
it is a succulent part of a reed, used for fodder, and it is irrelevant
to our Hebrew word on semantic grounds. The meaning is clear
but there is simply no significant etymological material.
I stressed in my preliminary remarks that much etymological
study in Hebrew is essentially linked with inner-Hebrew lexical
relations, which it is used to clarify. Often however this does not

16 Ch. Rabin, The Zadokite Documents (Oxford: Clarendon Press, 1954), p. 55,
note 2 on verse 6.
17 The Socotri word is quoted after W. Leslau, Lexique soqotri, p. 116, but I
think the supposed Amharic word is quoted in error. It does not occur in any
dictionary, including the Amharic-Amharic ones. I consulted the late Stefan
Strelcyn, F.B.A., about this and he wrote to me to say ‘I do not know this word’
(letter of 12 June 1979).
18 E.g. WB, p. 16a.
426 L IM IT A T IO N S O F ETY M O L O G Y
work. A good example lies in the word (or words) Dob. On the one
hand we have a clear case meaning ‘wineskins’ in Job 32. 19; it
seems to be agreed that this is some sort of skin or leather bottle. It
occurs only here. But we also have the familiar Dob that brings us
into the world of ghosts and necromancy. The word appears to be
used both for the ghost and for the necromancer, the one who
divined or consulted through the ghosts of the dead. The famous
Witch of Endor was such a one, an Deset bacalat Dob, a woman who
possessed an Dob. Now does this pair of words belong together or
not? And, in either case, what sort of comparative information can
be found? For the first Dob, the skin bottle, no one seems to have
any cognate evidence to put forward. For the second word, the one
in the world of ghosts, several comparative suggestions have been
put forward, of which I will at this point mention three. Firstly,
some have thought of a connection with an Arabic Daba ‘return’;
but I find no probability in this, for it seems to be suggested by the
modern French revenant ‘ghost’ which, if confirmed, would be
an interesting parallel but in itself hardly constitutes evidence.
Secondly, Rabin has pointed to an api ‘sacrificial pit’ in Hittite,
surely an extremely remote suggestion, being phonologically not
very close, semantically quite distant, and unlikely in principle
since there are no features of the relevant passages that suggest
reasons why a Hittite term should be used. Thirdly, some have
pointed to the Ugaritic name or term DelDeb; for instance, so
Albright. But this again is unlikely: it has nothing necessarily to do
with ghosts, it is more likely to be connected with the moon, cf. the
Ugaritic nkl wDeb (also Akkadian epithets of the god Sin). Thus
the extra-Hebrew evidence suggests the conclusion that the ety­
mology is unknown.
Can we, however, see a sort of inner-Hebrew etymology, in
other words can Dob ‘ghost’ be somehow explained on the basis of
Dob ‘leathern bottle’? One such argument came from the distin­
guished Semitist Nöldeke, who pointed out the similarity between
two words in different Semitic languages, a Syriac zakkürä ‘ghost
called up from the dead’ and an Arabic zukra ‘small bag, wine
skin’.19 The argument goes thus: the bag of skin has a common

19 Nöldeke pointed out this similarity but I am not sure just how much of the
following argument was intended by him. His note on the matter, in Zeitschrift der
Deutschen Morgenländischen Gesellschaft 28, 1874, 667 n., is really a marginal note
and principally on another subject.
 L IM IT A T IO N S O F ETY M O L O G Y 427

semantic element with the stomach, and the calling up of the ghost
is connected with ventriloquism; the Septuagint quite frequently
used έγγαστρίμυθος for Hebrew Dob. If there is a connection
between two words with radical consonants z-k-r, the one meaning
a skin bottle or bag, the other a ghost, even if these are not in the
same language, we can therefore suppose that the two meanings of
Hebrew Dob are similarly related.
Philologically, however, this line of reasoning will not hold
together. There is no reason on the basis of the Hebrew, as apart
from the Septuagint, to suppose that ventriloquism came into the
matter at all; and even the LXX έγγαστρίμυθος probably did not
specify ventriloquism but meant only the general sense of ‘person
who delivers oracles through supernatural and magical means’.20
Philologically also, Syriac zakkura surely came from the Akkadian
zakaru with the sense ‘call upon, call up’, but Arabic zukra ‘bag’
can hardly come from the same root, since that would have had
the consonants dh-k-r in Arabic.21 In the end, therefore, we find
that this argument does not give us an analogy that will establish a
relation between the two senses in Hebrew as if they came from
one common meaning. We therefore have to register two separate
words Dob in Hebrew and accept that the etymology of both is
unknown.
It would be tedious to multiply examples, but here is another of
the same general type. Hebrew has a word DagappTm, plural, used
only by Ezekiel; I think it means ‘troops’. We also have, once, in
the Dead Sea Hodayot an instance which seems quite clearly to
mean the banks of a river: ‘the streams of Belial go over all the
lofty banks’ (1QH 3. 29). Here then are two similar, yet sharply
differentiated, words in Hebrew: does etymology help us to dis­
criminate between the two? No, it does not; on the contrary, far
from helping us to discriminate between the two, it does not tell us
anything about either. For the word meaning ‘troops’ the classical
etymological approach has been simple: the word comes from, or
is cognate with, the Akkadian agappu, Aramaic D-g-p ‘wing’; and
originally this was the ‘wing’ of an army, hence troops in general,

20 The term was used in Greek primarily of persons who delivered oracles
through this form of speech; the spread of its use in the Greek Bible suggests that
the reference was not to the mode of speech but to the general associations of
theurgy and necromancy.
21 See Brockelmann, Lexicon Syriacum, p. 196b.
428 L IM IT A T IO N S O F ETY M O LO G Y

and cf. Latin ala as a unit of cavalry. But all along the line this
fails to work. The Akkadian word is kappu and agappu is a less
common variant spelling. Moreover, kappu is the wing of a bird, it
is the arm or hand, the lobe of a lung, a part of a horse’s bit, and
the region of the eyebrow and eyelids, but nowhere does it have the
slightest thing to do with armies, although armies are very well
represented in Akkadian texts. There is not the slightest hint that
Akkadian kappu is relevant: of course it could be relevant, but
there is not the slightest positive evidence for it. Similarly, on the
Hebrew side there is no trace of any meaning like ‘wings’ or ‘flanks’
or ‘different sides’ of an army: the Ezekiel text gives us no ground
to think of any other than the general meaning ‘troops’. As for the
other case in the Hodayot, the sense ‘river bank’ is clear there and
also later, in the Talmud, but there is no trace of a connection with
the Akkadian kappu nor indeed, so far as I know, with any other
term.22
Here is a case of another kind, where an etymology can easily be
found but there is reason to suppose that the word will not fit into
it. At I Sam. 2. 36 one speaks of doing something for an Dagdra of
money/silver and a piece of bread. It has been customary to attach
this word Dagdra, which occurs only here in the Bible, to the
familiar Akkadian agaru and related Aramaic words, all meaning
‘hire’. Thus we have a line of authorities who say that it is a ‘fee’, a
‘payment’, of money. The other opinion, which seems to be better
based, is that this is not a word for a ‘fee’ but a word for a tiny
monetary unit. This is the way in which the ancient versions took
it: LXX οβολός, Targum meca, Vulgate nummus. The point is
lost if we take it as ‘fee’: what goes with the small piece of bread is
the very small monetary unit. Modern Israel, in reusing Dagora for
a very small coin of its currency, is correctly following the sense.
But the consequence is that, having gained a sense, we have lost
our etymology; for, if the word means a particular and very small
monetary unit, we cannot so easily derive it from a word meaning
‘hire’. Thus, though we have available a good Semitic root, it may
be that the etymology is unknown. Or, on the other hand, if we

22 WB, p. 1la rightly observes the absence of a military use comparable with that
of Latin ala, and prefers therefore to look in the direction of an affinity with Arabic
juff, jaff \ said to mean a company or body of people. One must, however, be wary
of this also, since the common meaning of this Arabic root is ‘dry’.
 L IM IT A T IO N S O F ETY M O L O G Y 429

were to be adventurous, we might suppose that the noun derives

from Dagar ‘gather’, which is the verb of this root actually active in
Hebrew, and that an Dagora was a small handful of tiny monetary
tokens, as in Greek δραχμή was derived from δράσσομαι, being
what a hand might hold of small tokens such as spits or nails (this
was the original sense of οβολός). This is possible but it remains
speculative: in particular, though Dagar is a verb in active use in the
Bible, it is used only of the gathering of food. It may be that once
again we should admit that no etymology is known for our word.
Not only minor and rare elements of the Hebrew lexicon, but
also quite prominent and well-evidenced ones, may be without
substantial cognate evidence such as would constitute a thorough
etymological base. Take for instance the root lexeme b-g-d, from
which we have the common noun beged, the most frequent general
term for clothes or garments, and the verb ‘be treacherous’ with a
number of associated formations meaning ‘treachery’, ‘treacherous’
and so on. But large areas of the Semitic languages seem simply to
have no comparable term: Akkadian, it seems, has none, Ethiopic
has none, Aramaic does not have it in any of its numerous
branches, Ugaritic does not have it either, and, as for Arabic, we
have only a faint echo, if a real one at all, in bijad, said to be some
kind of beduin garment, and a verb bajad ‘deceive’ registered only
for the remote dialect of Dathina. Within the main areas of the
ancient Semitic languages Hebrew b-g-d, in either of its two very
different senses, has scarcely any echo.
The case of b-g-d brings us to another of the functions of
etymology, as it has been used in the lexicography of Hebrew,
namely that of providing a bridge to cover the semantic gaps
between sets of actual and existing Hebrew words. We already saw
an example of this, above, in the words meaning ‘angel, messenger’
on the one hand and ‘work’ on the other. But, though it is pleasant
when this works, it often does not work; and the relation between
beged ‘garment’ and bagad ‘be treacherous’ is a good example of
the latter. Even if we accept the tenuous etymological comparison
with a South Arabian dialect, this does nothing to bridge the gap,
for the gap is there in the comparative material just as it is in the
Hebrew. One can of course say on purely logical grounds, as
various people have long said, that there could be a relation
between ‘covering’ or ‘clothing’ on the one hand and ‘deceiving’
and the like on the other; but the point is that this is a purely
430 L IM IT A T IO N S O F E TY M O LO G Y
logical speculation, there is no philological evidence of it at all.23
There is no Hebrew verb b-g-d that means ‘cover’ or anything like
that. The idea that there was a term meaning ‘cover’, and that it
developed in one direction to mean ‘clothes’ (but lost its sense
‘cover’) and in another direction to mean ‘treacherous’ (but lost its
sense ‘deceptive’, for ‘deceptive’ is a good deal less than ‘treach­
erous’) is of course logically possible but there is really no lin­
guistic evidence in the languages to support it. The existence in
Arabic of forms derived from labisa ‘clothe’, ‘wear’ and meaning
‘deceive’ (as well as ‘be obscure’, ‘be dubious’, etc.) is probably a
clear case of metaphorical extension from a well-known term,
which retains its full range of meanings including those from which
the extension has started out. With Hebrew b-g-d we have nothing
like that. If there is a connection between the two Hebrew terms,
we have to leave it open that it may have taken place by a chain of
connections that logically are entirely adventitious and that are
quite unpredictable and unknowable from the linguistic evidence
that we have. It is more probable that the two Hebrew terms have
simply no etymological connection between them.
This situation is not uncommon, but recurs in several important
sets of terms, of which I will mention two. The first is the group of
words of the root z-k-r. The typical verb of this group, in Hebrew
zakar, means ‘remember’; but the noun form, zakar, is ‘male,
masculine’. We move to Arabic, and the corresponding words
display exactly the same meanings. In Akkadian the verb is more
‘call, speak’ than ‘remember’, but this can be accounted for through
a middle term like ‘mention’; the noun again is ‘male’. Interestingly,
in spite of their strong tendency to over-etymologize, the major
Hebrew dictionaries hardly even tried to bridge this gap in their
etymological sections. BDB, p. 271a, in the entry on zakar ‘male’,
said ‘relation to above yj obscure’, though it went on to mention,
faint-heartedly, one or two attempts; GB, p. 197b, mentioned the
guess that the basic sense might be stechen ‘prick’, but there is of
course no word of this root which actually means this in any
of the relevant languages; and the most recent lexicographer,

23 The series of supposedly analogical cases set out by Palache, Semantic Notes
on the Hebrew Lexicon (Leiden: Brill, 1959), p. 10, and requoted by WB, p. 104a,
Cohen, p. 42b, is a good illustration of the quite dubious etymologies that can be
conjured up in the endeavour to demonstrate such a logical connection on a lin­
guistic level.
 L IM IT A T IO N S O F E TY M O LO G Y 431
Baumgartner, WB, p. 259b, is doubtless wise in entering ‘Etym.
ign\ Schottroff similarly says that there is absolutely no Anhalts-
punkt for the establishment of a connection between the two
words, quoting also Fronzaroli.24
Another prominent example of the same kind is expressed in
Hebrew by the verb gâlâ, which has two conspicuously separated
meanings: firstly, ‘uncover’ or ‘reveal’, and, secondly, ‘go into exile’.
If we turn to comparative methods to elucidate this, what do we
find? In Arabic, exactly the same thing as in Hebrew; and in other
Semitic languages the meanings appear to group themselves either
with one or with the other or with both, but nowhere furnish a
bridge between the two. In fact, as far as etymological guidance
takes us, the two are so sharply distinct that it would be perfectly
reasonable to distinguish them as two quite distinct (though
homophonous and homographie) verbs: semantically, for instance,
they are much farther apart than the cânâ ‘answer’ and the cânâ
‘sing’ which we distinguished before. It is very doubtful whether
lexicographers have done wisely in maintaining gâlâ as one single
but polysémie verb. There is no possibility of making a unitary
statement about it on this basis.25 The fact that they have tried to
do so is evidence, if any were needed, that formal and comparative
evidence has counted for more in the lexicographical tradition than
semantic indications for determining the organization of material
in the lexicon.
Thus, whether with common and central terms like beged and
gâlâ, or with rare terms and hapax legomena, the intransigent fact
remains that many words, and relations between words, which in
the past have been obscure remain so today in spite of the atten­
tion given to etymological explanations. In Ezek. 21. 20 God says:
T have put the Dibhâ of the sword against all their gates’. Con­
textually all sorts of things could fit here: it could be the flash of
the sword, the whirr or noise of the sword, the face of the sword,
the edge of the sword, the violence of the sword, the danger of the

24 Schottroff in Jenni and Westermann’s Theologisches H an dw örterbu ch zu m

A lten T estam ent (Munich: Kaiser, 1971), i. 508; Fronzaroli in the A tti della
viii/19, 1964, 244.
A ccadem ia N azionale d ei L incei , Rendiconti,
25 An explanation such as that ‘Emigration or exile can be understood as an
uncovering of the land’ (Zobel, in Botterweck and Ringgren, Theological D ic­
tio n a ry o f the O ld Testam ent (Grand Rapids: Eerdmans, 1975), ii. 478, can surely
not be taken seriously.
432 L IM IT A T IO N S OF ETY M O L O G Y

sword. There is nothing in the context that will tell us specifically

more than that. But there is no really good comparative illumina­
tion to be found anywhere. NEB says ‘the threat of the sword’, and
it is probable that this is based on the Arabic verb wabbakha,
which however does not really mean ‘threaten’ but rather ‘reproach’.
In any case the Hebrew form begins with Daleph. The variant
spelling in Arabic with Dalif instead of wa is found in the Lisan but
there is no supporting evidence from texts, and even in the Lisan
there is a note that replacement of wa by Dalif is rare when a
follows. In fact the suggestion is no more than a remote guess, and
comparative study leaves us without any real clue to the meaning
of this word. Either it is a textual error, and the real text should be
the familiar tibhat ‘slaughter (by)’, implying that the first letter of
the word had been mistakenly copied; and this is the way in which
Baumgartner, very justifiably, still takes it; or else it is a Hebrew
word, with no parallels or cognates known, the meaning of which
is anything in the variety of contextually possible terms.
To sum up, then, up to this point: it is true that comparative and
etymological interests are interlinked with the inner-Hebrew rela­
tions between root lexemes and actual words and the meaning of
both; and it is true that in certain cases the comparative infor­
mation serves to help in organizing the Hebrew material, separating
out the components of it and making their relations intelligible. It
has been a mistake, however, if this function has been overempha­
sized, as it sometimes has been: for what is sometimes true is not
always true. Practical experience suggests that for many relations
which are problematic in the Hebrew lexicon comparative infor­
mation should not be expected to help. This is not because
comparative information may be semantically misleading, though
that is also true, but because there is no comparative information
that is to the point, or because it will only lead to a duplication of
the situation which is already apparent in Hebrew itself.
For the basic reasons behind all this we can turn to the work of
David Cohen. In the past it is probably true that most etymo­
logical work in the Semitic languages has been done from the point
of view of one particular language or another: at least within
dictionaries this has been so, for a Hebrew dictionary has sought
to present the cognates which are relevant for a particular Hebrew
word. It therefore by the nature of its undertaking filters out the
irrelevant matter. But the extent of the irrelevant matter is pre­
cisely the most striking fact of the entire question. Cohen’s die-
 L IM IT A T IO N S O F ET Y M O L O G Y 433

tionary brings together all the Semitic material on an equal basis,

and all words with a particular root lexeme, i.e. the sequence x-y-z,
are brought together whatever the degree of their semantic rela­
tionship or absence of same. The importance of this for the future
of Semitic etymological research is enormous. For, in particular,
his work demonstrates systematically what was already manifest to
those who cared to think about it, namely that for many given
Semitic roots, formally defined, i.e. defined as the root b-r-h or the
root g-d-l, one does not find a single concept or meaning from
which there ramify various derived or related meanings, but on the
contrary one finds a list of a plurality of terms, perhaps three or
four, perhaps as many as ten, to which words in this or that
Semitic language are attached, but for which in most cases there is
no overarching conceptual unity and no prospect that further
etymological research will create such a unity. Take for instance
the root b-r-h as displayed by Cohen (p. 83):
1. ‘to flee’: so Hebrew, Ugaritic, etc.
2. a name of various animals, chamois, gazelle, etc.
3. ‘to shine, make clear’, and also various colours
4. barh ‘torment’ in Arabic, to which Hebrew barTah and the
Ugaritic parallel are (questionably) ascribed
5. ‘bar’ of a door, familiar in Hebrew and elsewhere
6. ‘good’, ‘exciting the admiration’, so in Arabic and perhaps
Harari
7. ‘to draw away the seal from a well’, in South Arabian
8. a Tigre word for a mixture of water and milk
This is by no means an exceptional instance. Another good
example is furnished by D-d-m (Cohen, p. 9):
1. ‘red’, familiar in Hebrew
2. ‘man’, also familiar in Hebrew
3. ‘land, surface of earth’, also familiar in Hebrew
4. Arabic Didam9 ‘something eaten along with bread, enrichment
of bread’
5. Akkadian adamu ‘procure, obtain’
6. Amharic ddme ‘age’
7. Amharic ddmo ‘large hut with terrace’
Clearly, future research might provide us with semantic paths
that would bridge over the gaps between one and another of these
units in either set; and Cohen himself discusses these possibilities
434 L IM IT A T IO N S O F ETY M O L O G Y
where appropriate. But it is most unlikely that they will ever be
reduced to a unity. Thus, contrary to the idea rather vaguely held
in much Semitic study, the typical Semitic root, formally defined,
does not lead us back to a conceptual unity but rather to a variety
of unconnected semantic possibilities which can be listed but can­
not be explained through derivation from one another or from a
putative common ancestor. These different possibilities are taken
up and realized in different ways in different words in various
groups of the language family.
In the future much more of the comparative and etymological
work that has hitherto been done by the lexicographer of individual
languages such as Hebrew will have to be passed over to the work
of pan-Semitic lexicography of the type initiated by Cohen. This
will reduce the burden on the Hebrew lexicographer and enable the
student of Hebrew to obtain a much better overall view of the
shape and character of the Semitic lexicon than could ever be
provided in even the best etymological work starting from Hebrew
in particular.
It remains obvious, indeed, that the work of the pan-Semitic
etymological lexicographer will be reciprocally dependent on that
of the lexicographers of the individual languages. Take the Hebrew
phrase nahas bariah, mentioned just above. Is this a ‘wicked ser­
pent’ as Cohen thinks, or is it a fleeing serpent, a tawny serpent or
something else? If it is one of these latter, then this instance will
have to move from Cohen’s no. 4 and be attached to one of his
other numbers. He is of course well prepared to allow for this, and
it makes no difference to the validity of his method.
Sometimes the lexicographical tradition of the individual lan­
guages is at fault, not seldom as a result of its own etymological
speculations. A good example can be found in the lexeme s-p-d
which means ‘mourn, lament’ and is well evidenced in a long series
of languages with this sense. Various dictionaries, for instance in
Ugaritic and Akkadian, suggest that this derives from an etymo­
logical sense ‘beat (the breast)’,26 which of course theoretically

26 J. Aistleitner, Wörterbuch der ugaritischen Sprache (Berlin: Akademie-Verlag,

1967), p. 222, no. 1944, gives: ‘die Brüste schlagen (in der Trauer)’, thus seeming to
make the beating the ultimate component and the mourning incidental; simi­
larly von Soden, Akkadisches Handwörterbuch (Wiesbaden: Harassowitz, 1965),
p. 1024, gives for Ugaritic, Hebrew and Aramaic the indication ‘aus Trauer die
Brust schlagen’, although for Akkadian itself he gives simply the meaning ‘trauern’.
 L IM IT A T IO N S O F E T Y M O LO G Y 435

makes good sense and has partial analogy with common terms like
the Greek κόπτεσθαι. But so far as I can see there is no evidence in
any of these languages that the beating of the breast was the
specific meaning. In Greek κόπτω was a common term for any
kind of beating, but this is imply not true of s-p-d. The idea that
this was the meaning may well have arisen from Hebrew itself, and
there from a rather obscure passage, Isa. 32. 12, which has the
collocation cal-sadayim sopedTm, commonly taken as ‘beat the
breast’. But it is very doubtful if this is a right construction here,
and whether it can really suffice to validate supposition that s-p-d
ever means ‘strike’ in any sense.27 The only major evidence which
shows it to be other than merely ‘mourn, lament’ is the Syriac, and
here it seems to mean something more like ‘fear, tremble’; and this,
if right, suggests a quite different etymological direction.
The uncertainty of etymological research, which is implied in
much of the foregoing, is of course no new observation. Meillet in
his work on Greek remarks that ‘non-specialists are often not
sufficiently aware that, for every certain etymology, the dictionaries
offer ten that are doubtful’, and he adds that ‘there is only a small
number of Greek words of which Indo-European offers a certain
etymology’.281 do not know whether this is still the opinion among
Greek linguists. In any case it is worthy of note that Meillet
connected these remarks with observations on the importance of
loanwords from outside Greek, i.e. from non-Indo-European lan­
guages of the environment. In Hebrew also there are such loan­
words, and indeed there may be more of them than we are aware.

27 Wildberger is right in his recent commentary (Jesaia, Biblischer Kommentar,

Neukirchen, 1978, p. 1263) in seeing that the preposition cal will here probably
indicate the thing that has to be mourned, as in the two phrases that immediately
follow, and that therefore the word for ‘breasts’ here is probably a textual error.
The phrase should therefore mean something like ‘mourn for the fields’ and not
‘beat upon the breasts’. This is indeed no new idea, and similar suggestions were
made long ago. But it has not always been observed that, if this is agreed, the small
degree of evidence for ‘beat’ as basic meaning of s-p-d disappears. In any case the
Syriac meaning ‘tremble’ points in a very different direction.
28 Meillet, Aperçu d ’une histoire de la langue grecque (7th edn., 1930), p. 62. His
own words are worth quoting: ‘Les personnes qui ne sont pas du métier ne savent
pas assez que, pour une étymologie sûre, les dictionnaires en offrent dix qui sont
douteuses et dont, en appliquant une méthode rigoureuse, on ne saurait faire la
preuve . . . Les étymologistes visent donc à interpréter chaque mot grec par com­
paraison avec quelque autre langue indo-européenne. Mais, en fait, il n’y a qu’un
petit nombre de mots grecs dont l’indo-européen fournisse une étymologie certaine.’
436 L IM IT A T IO N S O F ETY M O L O G Y
But it does not seem that their existence is a major cause for our
etymological uncertainties. In this paper attention has been directed
almost entirely to the problems of inner-Semitic etymology itself,
for it is there that the major problems assuredly lie.