Sie sind auf Seite 1von 5


Verhagen, Pieter C., A History of Sanskrit Grammatical Literature in

Tibet. Volume Two: Assimilation into Indigenous Scholarship [Handbuch
der Orientalistik. 2. Abteilung: Indien, Band 8 Teil II]. Leiden – Boston –
Köln: Brill: 2001, pp. XI, 454. ISBN 90-04-11882-9.

In volume one1 the author had surveyed in great detail the Sanskrit
grammatical literature found in translation in one of the two large
Tibetan canonical text-collections, the Bstan-’gyur, and in manuscripts
found in Dun Huang. In this second volume he studies in equal detail
non-canonical translations, improved versions of canonical translations,
original Tibetan works on Sanskrit grammar, and finally the application of
Indian linguistics to the Tibetan language. The study of Sanskrit grammar
supported the extensive translation activity (many thousands of Indian
texts were translated into Tibetan), and Tibetan scholars went beyond a
passive reception of Indian texts and pursued the study of various aspects
of Sanskrit grammar in the tradition of Pān.ini’s, Śarvavarman’s, and
Candragomin’s works. They kept in touch with developments in India by
visits to Indian Buddhist centers (Nālandā and Vikramaśilā were the most
prominent) and later – when Buddhism had faded in India – with visits
to Buddhist centers in Nepal. One cannot but admire the industriousness
of the monastic community that from the seventh century on into modern
times not only translated such a large body of literature but through critique
and revisions improved the original translations and elucidated them in
An example is Bu-ston’s (seventh or eighth century?) detailed explana-
tion of the formation of te ’tra “these [men] here” in Durgasim.ha’s
commentary on the Kātantra I 2,17, where the formation of both words
and the sandhi between them are explained with references to the rules of
the Kātantra (pp. 83–87). Si-tu Pan.-chen (eighteenth century) occasion-
ally offered severe criticisms of translations by Bu-ston and others; his
new commentaries contain many improvements over these earlier works
(pp. 107f., 121f., 169–178). Though his magnum opus, a commentary on
the Cāndra-vyākaran. a, is primarily a commentary, it attempts a wide and
comparative overview including several Sanskrit grammatical systems.
Si-tu went to great lengths to gather all available literature, and even
traveled to Nepal to obtain manuscripts and personal instruction from
Indian pandits living there (pp. 121, 174, 177). That the connections
with India were not totally interrupted is clear from Si-tu’s references to
Rāmacandra’s Prakriyā-kaumudı̄ (fourteenth or fifteenth century). Either
this work or one similar to it was the source for Si-tu’s curious chapter “The
1 Cf. J.W. de Jong, IIJ 38 (1995), pp. 296–299.

Indo-Iranian Journal 46: 156–160, 2003.

© 2003 Kluwer Academic Publishers. Printed in the Netherlands.

formation of Vedic word-forms according to Pān.inian [grammar]” (p. 113).

Other late Indian works quoted are the Bhās.āvr.tti and the Rūpāvatāra
(p. 212). Remarkable is Si-tu’s caution in withholding judgement until a
better manuscript of a certain text is found (p. 119). Si-tu’s translation of
Candragomin’s grammar contains valuable material for a textual history of
this work (p. 131).
Besides grammatical texts proper, there were (from the twelfth century
onwards) pronunciation manuals (klog-thabs) that served to guarantee the
correct pronunciation of Sanskrit, especially the Buddhist mantras (pp. 58–
65, etc.). These texts call the vowels āli and the consonants kāli, terms
with a Tantric background but not known so far from Indian sources and
first attested in a manuscript from Dun Huang (pp. 30–36). In Buddhist
Tantric symbolism, the vowels are considered as feminine, the conson-
ants as masculine – exactly opposite to Hindu Tantrism; occasionally, the
nasals and vocalic /r./ and /l./ are considered as neuters (pp. 264–266). Some
authors of the eighteenth century refer to differences in the pronunciation
of Sanskrit in different provinces of India (pp. 167, 189); comparisons
with Tibetan sounds give occasionally clues to the history of Tibetan
pronunciation (p. 90).
Indian techniques were in time also applied to the study of the Tibetans’
own language. Since Tibetan has a very different structure from Sanskrit,
problems arose that were similar to those encountered by the Tamil
grammarians familiar with Sanskrit grammar or, in more recent times,
by the Christian missionaries who often applied the categories of Latin
grammar in compiling descriptions of exotic languages. The system of six
syntactical functions and seven or eight cases was ill-suited to Tibetan that
deployed a great number of particles with few one-on-one correspond-
ences. Tibetan grammarians creatively adapted the systems to the needs
of their language descriptions; but the interpretation of many details is
still under dispute (pp. 284–321), and I shall refrain from expressing any
opinions, lacking the requisite expertise in Tibetan. The Tibetan works
show Indian influence in their terse formulation. The (expanded) text of
the Kātantra may have inspired the metrical formulation of some texts and
an item-and-arrangement mode of description.
Candragomin’s grammar is probably the inspiration for the treatment
of case functions; for like Candragomin the Tibetan scholars work with
only two levels against the three levels best known from Pān.ini’s grammar.
Pān.ini and most grammarians after him progressed from a multitude of
interacting factors (kāraka) that the speaker wants to express to a set of six
syntactical concepts that could be expressed by case-suffixes, derivational
suffixes, and even verbforms. These three levels of “factors”, syntactical

concepts, and cases, etc. (and ultimately the actual case/derivational/verbal

suffixes on a fourth level) allowed elegant adjustments between the levels.
Such adjustments were necessary, since there are no perfect one-on-one
relations between physical relations and grammatical forms in a natural
language. Candragomin (and the Pali grammarian Moggallāna) omitted
the middle level and handled the adjustments differently.
In I 4 32 Pān.ini had given the beneficiary of an action (or indirect
object) the label sampradāna (“bestowal, giving”) and added in I 4 35
that also the creditor in conjunction with the verb dhāraya- is so labeled –
and is expressed consequently in the dative by rule II 3 13 (that demands
the dative ending for sampradāna). Candragomin who has no middle level,
ruled that the dative ending (which is ruled in for indirect objects in II 1
73) is also used to denote the creditor: Devadattāya gām . dhārayati “he
owes Devadatta a cow.” In other instances Pān.ini made the adjustment
on the next level; the accusative suffix that generally denotes the direct
object (karman), is also required in connection with adverbs like
“south of . . .” – grāmam “south of the village.” Candragomin
in II 1 53 made a similar adjustment. The results of Pān.ini’s and Candrag-
omin’s rules are generally the same as M. Deshpande has pointed out, but
Verhagen (p. 292)2 misreads Deshpande’s argument when he says that the
systems work exactly in the same way. Instead of the three-tiered system
of Pān.ini (which Whitney3 had deplored as “the vastly more difficult and
dangerous method”), Candragomin has only two tiers: external factors and
In I 4 42 Pān.ini had defined karan.a (“instrument”) as “what is most
efficacious” and had added in the following rule that it is also called
karman (“object”) with the root DIV “gamble.” Hence the dice can be
expressed either in the instrumental (as “instruments”) or in the accus-
ative (as “objects”): aks.air dı̄vyati or aks.ān dı̄vyati “He plays with dice.”
Candragomin apparently recognized also a construction dı̄vyati as
correct. But he gave us no indication of this at all in his rules; it is only in
the commentary (that is now generally credited to a certain Dharmadāsa,
not Candragomin himself) on rule II 1 49 that we are told that the speaker’s
intention (vivaks.ā) decides if he wants to stress the aspect of being an
object (use of accusative), instrument (use of instrumental) or location (use
of locative) in denoting the dice.
Verhagen (p. 292f.) challenges the prevailing opinion that Candragomin
dropped in his system the middle level (the six conceptual definitions)

2 Also P.C. Verhagen, in Tibetan Studies. Proceedings of the 5th Seminar of the
International Association for Tibetan Studies, Narita 1989, Narita 1992, p. 838.
3 W.D. Whitney, American Journal of Philology 14 (1893), p. 171.

and assumes that he rather eliminated the descriptive references to the

outside world. He points out that three of the expressions used by Candrag-
omin are the same as Pān.ini’s defined terms (kartr., karan.a, sampradāna),
whereas only one (ādhāra) is identical to Pān.ini’s external definition of
the concept term adhikaran. a (“location”). Two others (āpya “object” and
avadhi “point of departure”) are altogether different. Verhagen in an earlier
publication4 conceded the awkwardness of his position when he admitted
that Candragomin has throughout avoided defined terms: we find no gun.a,
vr.ddhi, sārvadhātuka, etc. A closer inspection shows his position to be
Pān.ini had deployed only six conceptual terms for all the factors
involved in an action. They are introduced by the heading kārake (“when
a factor is concerned”) and consist of a definition in the nominative
(ādhāro ’ or through a sentence (karman.ā yam abhipraiti, sa
sampradānam); supplemental inclusions use various forms. In Candrag-
omin’s grammar the second case ending (i.e., the accusative ending) is
ruled in for “what is to be reached by an action” (kriyâpye dvitı̄yā), for
“prodded [agent]” (prayojye [kartari]) of certain verbs, i.e. when they are
put into the causative. The seventh case (i.e., locative) ending is used to
denote ādhikye “excess” and svāmye “domination,” the third (instrumental)
kartari “agent,” karan.e “instrument,” “characterization,” vyāpye
“object,” and hetau “cause” (II 1 43–68) – all in the locative!
Candragomin did know and use the word kāraka (in II 2,16; cf. also
the commentary) to denote factors of the physical action; and there is no
limitation of the number of these functions. Dharmadāsa’s commentary on
the Cāndra-vyākaran. a makes this very clear. On II 1 62 it paraphrases the
rule with kartari kārake tr.tı̄yā bhavati “When the agent is a factor, the
third case ending comes into play” (similarly on rules 63 and 64). On II
1 74 it explains ruci-yukte kārake caturthı̄ bhavati: rocate Devadattāya
“When the factor is connected with the root RUC, the fourth case ending
comes into play: it pleases Devadatta.” Here Devadatta is the kāraka! On
II 1 75 it explains dhārayater uttamarn.e kārake caturthı̄ bhavati “After
the verb dhāraya-, if the creditor is a factor, the fourth case ending comes
into play.” Here the factors of the action (kāraka) are obviously elements
of the outside world as in Pān.ini’s system, not the equals of Pān.ini’s
six abstract notions. The number “six” attaches itself to the term kāraka
only gradually: Vākyapadı̄ya III 7,44 considered kāraka divided six- or
sevenfold, and Kaiyat.a on Mahābhās.ya on I 4 23 (beginning) spoke of
the [kāraka]. But apparently only in recent texts there are clear

4 See note 2.

statements about the six kāraka-s: The Nyāyakośa5 refers to the Śabda-
śakti-prakāśikā, and Carudeva Śastri speaks of the six kāraka-s in his
Pān.ini: Re-interpreted:6 “There are six Kārakas.”
One may ask what motivated Candragomin to make this change. The
motive may lie in Buddhist philosophy. Nāgasena denied in a famous
dispute in the Milindapaňha that there is something like a chariot: “Chariot
is a mere empty sound. It is on account of having all these things – the
pole, and the axle, the wheels, and the framework, the ropes, the yoke,
the spokes, and the goad – that it comes under the generally understood
term, the designation in common use, of ‘chariot’ ” (II 1,1). One might also
consider the later distinction of nirvikalpaka-jñāna and savikalpaka-jñāna
of which only the former, the first sense perception, puts us in touch with
“absolute reality,” while the latter contains mental constructs of uncertain
value. I have to leave it to experts in Buddhist thought, if such concepts
could explain Candragomin’s innovation which would mark his grammar
truly as a “Buddhist” grammar. We cannot, perhaps, totally dismiss the
possibility that Candragomin made his changes for the sake of a simpler
grammatical description.
Other influences of Buddhist thought have been identified by Verhagen
(pp. 240–251 and 330–333) in the early treatises on Tibetan grammar:
instead of the build-up from roots with the help of suffixes, to words
and sentences, these texts follow a concept inspired by philosophical
Abhidharma texts that take phonemes as their starting point, proceeding
to words and sentences (Note the similar notions in Nyāya texts!). The
common linkage with Tantrism invited mystical speculations on the values
of various sounds. While the complicated derivation of verbs and nouns
from roots in Sanskrit has thus no counterpart in these Tibetan grammars,
they copied features of case grammar and the descriptive technique of
terse formulations, dittoing of terms into following rules, use of headings,
etc. (p. 329). In several appendices Verhagen gives excerpts of relevant
sections from Tibetan grammatical texts. A detailed bibliography and
several indices (regrettably no topical index) complete the book. Verhagen
has produced a very learned book, full of valuable insights, that raises
expectations of further investigations in an interesting topic.

634 Breckenridge Place HARTMUT SCHARFE

Simi Valley, CA 93065

5 Nyāyakośa ed. Bhı̄mācārya Jhalakı̄kar, revised ed. Vāsudev Shāstrı̄ Abhyankar, Poona
1978, p. 221.
6 Charu Deva Shastri, Pānini: Re-interpreted, Delhi 1990, p. 2.