1.

Introduction
1.1. Definition and the beginnings of semantics
Semantics is the major branch of linguistics which studies meaning communicated through
language (words and sentences). The term 'semantics' is of relatively recent origin, being coined
in the late nineteenth century from a ree! verb meaning 'to signify'. This does not mean that
scholars first turned their attention to the investigation of meaning of words less than a hundred
years ago. "n the contrary, from the earliest times down to the #resent day grammarians have
been interested in the meaning of words and fre$uently more interested in what words mean than
in their syntactic function. %yons (1&'()*++) argues that the #ractical manifestation of this
interest is the #roduction of innumerable dictionaries throughout ages not only in the west but in
all #arts of the world where language has been studied.
In s#ite of the interest in meaning manifested by #hiloso#hers, logicians and
#sychologists, linguists doubted that meaning could be studied as objectively and as rigorously as
grammar and #honology and thus semantics came to be neglected and received #ro#er attention
only since the 1&'+s.
The beginnings of semantics as an inde#endent linguistic disci#line go as far bac! as early
1&
th
century, to the wor!s of the erman linguists ,h. ,. -eisig and .ermann /aul. -eisig was
the first to formulate the object of study of the new science of meaning which he called
semasiology and conceived the new linguistic branch of study as a historical science studying the
#rinci#les governing the evolution of meaning. .ermann /aul also dealt e0tensively with the
issue of change of meaning.
The 1birth date2 of semantics as a modern linguistic disci#line was mar!ed by the
#ublication of Essai de smantique (1(&3) where the 4rench linguist, 5ichel 6real, defines
semantics as 1the science of meanings of words and of the changes in their meanings2. .owever,
in 1((3, that is ten years ahead of 5ichel 6real, %a78r 98ineanu #ublished a remar!able boo!
called ncercare asupra semasiologiei limbei romne. Studii istorice despre tranziiunea
sensurilor (:ssay on -omanian Semasiology. .istorical Studies on the Transition of 5eanings).
This is one of the first wor!s on semantics to have a##eared anywhere. 98ineanu am#ly used the
contributions of #sychology in his attem#ts at identifying the semantic associations established
among words and the logical laws and affinities governing the evolution of words in #articular
and of languages in general.
4erdinand de Saussure's distinction between the two basically different ways in which
language may be viewed, the synchronic or descri#tive and the diachronic or historical a##roach
introduced a new #rinci#le of classification of linguistic theories. The ne0t section will ma!e an
overview of the major theoretical trends in semantics.

1.; An overvie of semantic studies
Diachronic or historical semantics
<iachronic or historical semantics develo#ed through the literature on semantic change
which had a golden #eriod between the last twenty years of the 1&
th
century and the 1&*+s (the
1((+s and the 1&*+s). "ne of the longest treaties on semantic change is ustaf Stern's boo!
!eaning and "hange of !eaning# #ublished in 1&=1. Stern's #rinci#al aim was to establish a
theoretically tenable and #ractically wor!able system of classification com#rising all !nown
ty#es of sense change.
Views of semantic changes
Sterns classification
Stern (1&'()1';) defines change of meaning as >the habitual modification, among a
com#aratively large number of s#ea!ers, of the traditional semantic range of a word...to denote
one or more referents which it has not #reviously denoted or to e0#ress a new manner of
a##rehending one or more of its referents?. Stern starts by classifying a large number of authentic
sense changes and then formulates a theory to account for the e0istence of the different classes. In
other words, the classes were established inductively rather than deductively. .e analyses
historical instances of sense change mainly with regard to the #sychic #rocesses involved and
identifies seven main classes of change) substitution, analogy, shortening, nomination, (regular
transfer), #ermutation and ade$uation.
Substitution is a change of meaning due to an e0ternal, non@linguistic cause. 4or instance,
alterations in the design of shi#s have brought about changes of meaning in the word ship. It once
meant only a sailing vesselA now it can mean a steam@driven vessel of $uite different a##earance.
Therefore the referents of a word undergo some change so that new referents are added to or
substitute old ones.
Analogy occurs when a word assumes a new meaning on the analogy of some other word with
which it is connected derivationally (e.g. the adjective fast has borrowed the sense >$uic!? from
the middle :nglish adverb faste), semantically (e.g. the s#ecial meaning of lo# >non@dogmatic?
in $o "hurch on the analogy of %igh "hurch where high means >dogmatic?.
Shortening is he omission of a word from a com#ound e0#ression, the remaining words carrying
the total meaning that formerly belonged to the whole e0#ression) e.g. private is a shortening of
private soldier (common soldier), periodical is a shortening of periodical paper& revie.
Nomination is a change of meaning in which a name is intentionally transferred from one
referent to another. Stern gives as e0am#le of nomination the convention of using #ro#er names
for units of measurement , inventions, or discoveries(e.g. volt# sandich). "ther e0am#les include
#lace names for #roducts (e.g.champagne# a 'erse(), article of dress for #erson (e.g. mac)intosh*#
habitual e0#ressions for #ersons (e.g.'ingoes >music@hall #atriots who sing jingo songs?
(Regular) transfer is the unintentional transfer of a word from one ty#e of referent to another
one resembling it. :0am#les are root as in root of hair and bed as in river bed.
Permutation is the unintentional shift from a referent to another brought about by the #ossibility
of inter#reting a word in two ways in some conte0t. +eads in %e is counting his beads can mean
either >#rayers?(the original sense) or >little balls on a rosary?.
Adequation is the change of meaning resulting from the ada#tation of the meaning of a word to
the actual characteristics of the referents.
Stern2s main e0am#le is horn, which, in order of historical develo#ment of meaning, denotes (i)
>animal horn?, (ii) >animal2s horn used for music?, (iii) >musical instrument made from animal2s
horn? and finally (iv) ?instrument for #roducing a certain !ind of sound?. The change from (ii) to
(iii) is an instance of ade$uation.
Bde$uation differs from substitution in that the immediate shift does not lie in the referent ,as
in the change from (i) to (ii) or in the change from (iii) to (iv) but in the s#ea!er2s a##rehension
of the referent. Bs can be noticed, ade$uation occurs after other sense changes (e.g. substitution)
have ta!en #lace.

Ullmanns classification
Ste#hen Cllmann (1&';) #ro#oses a DbetterD version of Stern's classification of semantic changes.
,oncerning the causes of semantic changes, Cllmann distinguishes two main a##roaches) B.
5eillet's theory and S#erber' s theory. In his article "omment les mots changent de sens (1&+*@
1&+E) Bntoine 5eillet maintains that there are three main causes of semantic change, vi7.
linguistic, e0tralinguistic and social. S#erber's a##roach is different from other a##roaches in
that he em#hasi7es the role of emotion. 6y see!ing in emotive forces the clue to changes in
meaning, S#erber (1&;=) focuses e0actly on what the 4rench #hilologist had disregarded.
Blthough S#erber neglected the non@e0#ressive functions of language he introduced a new
#ers#ective for the understanding of changes of meaning and their s#read. 4ollowing Cllmann
(1&'=) we conclude that the two theories mentioned above are mutually com#lementary rather
than e0clusive.
Cllmann (1&'=) distinguishes between semantic changes due to linguistic conservatism and
linguistic innovation. Fhen we !ee# a word, in s#ite of the the fact that the character of its
referents has changed, we have @ in Cllmann2s terminology @ an instance of linguistic
conservatism. Farren (1&&;)&) rightly notices that Cllmann2s linguistic conservatism corres#onds
to Stern2s substitution.
The semantic changes due to linguistic innovations are grou#ed into three main subclasses)
transfers of names, transfers of senses and com#osite changes. ,onsidering the word a union of
name (form) and sense (content), Cllmann assumes that there are two #ossibilities) either the
name or the sense of the word may change or be transferred. 6oth transfers of names and
transfers of senses occur due to contiguity or similarity relations.
B case of name transfer through sense similarity is overloo) which is related to the sense of
oversee. Instances of sense transfer through sense similarity are antro#omor#hic transfers li!e
leg of a table# e(e of a needle# bridge head# etc. Sense transfers through contiguity are sail
meaning?shi#?, ton meaning >its inhabitants?. Such sense transfers based on similarity and
contiguity corres#ond to the ancient categories of meta#hor and metonymy.
The third subgrou#, composite changes, includes all asssociative lin!s that can be conceived)
com#osite name transfers, com#osite sense transfers and sense@name transfers.

In general, more recent a##roaches to sense change list generali7ation (.ughes1&(&, 6erndt
1&(&, Cngerer and Schmid 1&&'), s#eciali7ation (.ughes 1&(&, Farren 1&&;, Cngerer and
Schmid 1&&'), figurative use (6erndt 1&(&, Farren1&&;) and substitution Gsemantic shift (6erndt
1&(&).
Specialiation of meaning can be illustrated by the "ld :nglish word fugol which referred to
all !inds of birds. radually fugol was re#laced by the word bird whose meaning underwent the
#rocess of generaliation(": br(d 1young bird2 @H5od. : bird 1any bird2)
,oncerning figurati!e use we further consider the category 6I-< and the attribute 1loc!ed in
a cage2 characteri7ing #arrots, budgerigars, and the attribute 1e0otic a##earance2 which a##lies to
ostriches, flamingoes, #eacoc!s.
Bs can be noticed, the meta#horical uses of bird 1#risoner2 and rare bird 1strange #erson2 rely
#redominantly on #eri#heral attibutes rather than on central attributes such as 1can fly2 and 1has
wings2 which are at the basis of bird 1aero#lane, missile, s#acecraft2.
The major attributes of good e0am#les of the 6I-< category (1can fly2 and 1has wings2) in
Bnglo@Sa0on times were #robably similar to what they are today. .owever, there are instances
where central attributes of a category are re#laced, normally as a result of e0tralinguistic changes.
This ty#e of meaning change is traditionally called substitution or semantic shift or, in cognitive
linguistics terms, prototype shift.
Synchronic semantics or descripti!e semantics
6esides diachronic or historical semantics, we also have to consider synchronic or
descri#tive semantics which dis#lays the a##lications of the #rinci#les of structural linguistics to
the study of meaning. This ty#e of semantics is also called 1#aradigmatic2 (6idu@IrJnceanu et al.
1&&3) *=E) because it studies meaning through #ragmatic sense relations such as synonymy,
antonymy, homonymy, #olysemy and hy#onymy. Synchronic semanticists also study meaning
through semantic components# distinctive semantic features or semes that are can be e0#ressed
with the hel# of a binary feature notation using 1K2 and 1@1. The major re#resentatives of le0ical
structural semantics are B. L. reimas (1&''), 6ernard /ottier (1&3*) and :ugen ,oMeriu (1&(1).
In ,oMeriu 2s a##roach, le0ical items are o##osed to each other and this o##ositional
contrast yields s#ecific distinctive features or semantic com#onents. ,om#onential analysis (,B)
or the decom#osition of le0emes into semantic elements or com#onents goes as far bac! as 1&*=
to the <anish linguist %uis .jemslev and has a long tradition both in :uro#ean structuralism and
Bmerican anthro#ology.
,oMeriu (1&(1) stresses the im#ortance of the distinction between signification
(>6edeutung?) and designation (>6esiechnung?). <esignation, for him denotes the relationshi#
between the full linguistic sign (combining significant and signifie) and the e0tralinguistic object
or referent. Bs signification (meaning or 6edeutung) alone is believed to be significant for
structural semantics ,oMeriu 2s theory, e0cluding e0tralinguistic objects and relations and being
therefore restricted to language itself can be characteri7ed as a >language@intrinsic? or >language@
immanent a##roach to semantics?. (%i#!a, 1&&+) &&).
"nterpreti!e or syntagmatic semantics
Bs in their conce#t of meaning they do not normally loo! beyond language, the schools of
inter#retive and generative semantics must also be considered language@immanent a##roaches to
semantics. The main focus of inter#retive semantics is syntagmatic semantic relations, i.e.
relations that hold between members of different grammatical categories which are
simultaneously #resent in a single syntactic structure. This is why inter#retive semantics is also
called syntagmatic semantics. -e#resentatives of inter#retiveGsyntagmatic semantics such as
Lerrold L. Nat7 and Lerry B. 4odor (1&'=), Ooam ,homs!y (1&'E) and -ay Lac!endof (1&(') are
concerned with #ossible combinations of #articular words and with restrictions on #ossible
combinations of meaning, the so@called >selection restrictions?, i.e. semantic restrictions on the
choice of individual le0ical units in construction with other le0ical units (e.g. pregnant will
ty#ically ?select? a subject referring to someone or some animal that is female). Syntagmatic
semantics also deals with the meaning of com#le0 linguistic e0#ressions, including sentences.
This e0#lains why some scholars refer to this ty#e of semantics as 1sentence semantics2.
#enerati!e semantics
enerative semantics is that ty#e of semantics that is concerned with le0ical
decom#osition, i.e. the analysis of word meanings into smaller units which are seen as standing to
one another in constructions li!e those of synta0. enerative semanticists such as Lames <.
5c,awley (1&'(, 1&3=), and eorge %a!off (1&3+, 1&31) argue that le0emes have an internal
structure li!e the syntactic structure of sentences and #hrases. 4or e0am#le, the sense of )ill can
be analysed into ,BCS:, 6:,"5: , O"T and B%II:A these elements are not sim#ly conjoined
but are combined in a hierarchical structure which may be re#resented as (,BCS: (6:,"5:
(O"T B%II:))).
$onceptual semantics
The decom#ositional theory of meaning has been develo#ed by -ay Lac!endoff (1&3;,
1&(=, 1&(', 1&&+) who identifies a number of structural categories, including) :vent, State,
Thing (or "bject), /ath, /lace and /ro#erty. %oosely s#ea!ing, 'event' and 'state' tend to be
categories #resent in verbsA 'thingGobject', in nounsA '#ath' and '#lace', in #re#ositional
(/-:/"SITI"O) and B<I:-6IB% constructionsA and '#ro#erty', in B<L:,TII:S. These categories can
all be sub@categorised by reference to s#ecific semantic com#onents. The event category, for
instance, can be bro!en down to include features li!e cause, motion, change and contact.
Similarly 'thing' can be sub@categorised in terms of the features PQboundedR. This will distinguish
between count nouns such as table and chair and mass nouns li!e music and ater. Oouns which
are bounded are basically conceived of as units. If we dismantle a chair we can't call the
individual #ieces a chair. 5ass nouns, however, are thought of as substances. If we only hear a
few bars of a sonata we have still heard music. This is reflected in the grammar so that mass
nouns, for e0am#le, cannot go into the #lural, e.g. ,musics, whereas count nouns can, e.g. chairs.
,once#tual semantics is a com#le0 and so#histicated attem#t to identify universal semantic
categories and ma# them onto syntactic o#erations and structuresR.
Philosophical semantics
/hiloso#hical semantics e0amines the relations between linguistic e0#ressions and
#henomena in the world to which they refer, and considers the conditions under which such
e0#ressions can be said to be true or false, and the factors which affect the inter#retation of the
language as used (,rystal, 1&&1)=1+). /hiloso#hical semantics is also called #ure, formal or
logical semantics (because it is based on logical #ositivism). It studies the meaning of e0#ressions
in terms of logical systems of analysis or calculi, and is thus more a!in to formal logic or
mathematics than to linguistics. 4ormal semantics a##roaches meaning using the notion of truth
(inherited from logic) and attem#ts to formali7e the meaning of sentences and the relations that
hold between them. These relations are synonymy, #ara#hrase, tautology, contradiction, anomaly,
entailment and #resu##osition. :ntailment is the relation between sentences such that the truth of
the second sentence necessarily follows from the truth of the first sentence, but the falsity of the
second sentence does not necessarily follows from the falsity of the first sentence. 4or e0am#le,
!ar( is running entails, among other things, that !ar( is not standing still. /resu##osition is the
relation between sentences such that the truth of the second sentence (the #resu##osed sentence)
is im#lied (#resu##osed) by the truth and by the falsity of the first sentence (the #resu##osing
sentence). 4or instance, the statement -.m sorr( it.s raining #resu##oses that it is raining. The
#resu##osition also holds if the statement is negated/ -.m not sorr( it.s rainin# also #resu##oses
that it is raining.
The #hiloso#hers and logicians whose wor!s had a strong im#act on the study of
linguistic meaning are ,harles /eirce (1&=1), ,harles 5orris (1&=(, 1&*') and -udolf ,arna#
(1&*;). Bs the study of language can be included within the more general theory of signs called
semiotics or semiology, it is im#ortant to mention the three areas distinguished by the three
#hiloso#hers within the field of semiotics ) semantics, synta0 and #ragmatics.
In his earlier wor!, 5orris (1&=() ') defines semantics as the study of the 1relations of
signs to the objects to which they are a##licable2 and syntactics @ or synta0 as the study of 1the
formal relations of signs to each other2.
%inguistic semantics
%inguistic semantics studies meaning in natural languages and is concerned #rimarily
with le0ical meaning, grammatical meaning and sentence. It is generally recogni7ed that one
cannot account for le0ical meaning without accounting for sentence meaning and viceversa. Thus
the meaning of a sentence de#ends u#on the meaning of its constituent le0emes and the meaning
of some, if not all le0emes de#ends u#on the sentence in which they occur. %yons (1&&E) 1**)
discusses the im#ortance of grammatical meaning as a further com#onent of sentence meaning.
Cnli!e language@ intrinsic or language S immanent a##roaches to semantics that e0clude e0tra
linguistic objects (referents) and relations, referential or denotational (language@e0trinsic)
a##roaches to semantics focus on the #ro#erties of the referents denoted by the linguistic signs.
In contrast to ,oMeriu, for e0am#le, %eisi (1&(E), in his analysis of the content of words
ma!es e0#licit reference to s#ecific characteristics of concrete e0tra linguistic referents.
-eferential a##roaches to semantics must be related to the notions of #rototy#e and gestalt in
#sychology and linguistics.
Prototype semantics
Bccording to #rototy#e semantics, word meanings contain the #ro#erties of cognitive
categories, i.e. we can distinguish central and more #eri#heral meanings of a le0emeA word
meanings are not rigid, there are often gradual transitions between word meanings. /rototy#e
semantics is closer to #sychological reality than traditional feature semantics (or all or nothing
semantics).The advantages of #rototy#e semantics over feature semantics does not diminish the
usefulness of the later for the descri#tion and com#arison of word meaning es#ecially for
identifying semantic structures li!e le0ical fields and sense relations. /rototy#e theory and the
family resemblance model cannot do without a feature@based classification. Cltimately, #rototy#e
and feature semantics com#lement each other, in the sense that feature semantics receives a
sounder #sychological basis.
Blthough #rototy#e semantics is #articularly ade$uate for the descri#tion of concrete
(e0tra linguistic) objects, es#ecially those in which sha#e and si7e are relevant, it cannot ca#ture
connotative feature or deal with deictics, relational words and syntagmatic relations (restrictions
or transfer of features). Oevertheless it has clear advantages in com#arison with feature
(Bristotelian) semantics. Thus, the #rototy#e a##roach can e0#lain) (1) vague, fu77y category
boundaries (;) gradual category membershi# (=) categories with #rototy#ical !ernels (*) the
different im#ortance attributes.
Fhat is ultimately needed in semantic theory is an integration of both language@intrinsic
and denotational Greferential a##roaches. It is only in this way that the limits and boundaries of
either traditional structuralist semantics or #rototy#e semantics can be transcended.
$ogniti!e semantics
,ognitive semantics has develo#ed in the 1&(+2s on the basis of findings in cognitive
#sychology. The main difference between structuralist semantics and cognitive semantics is that
the former analy7es meaning from a #urely language@internal #ers#ective (i.e. on the basis of
semantic networ!s connecting le0emes) whereas cognitive semantics e0#lains meaning #rimarily
in terms of categori7ation (i.e. the grou#ing of similar #henomena into one class).
In cognitive semantics, meaning is considered to be ine0tricably lin!ed to human
cognition, to the way we #erceive the world and grou# #henomena into conce#tual categories.
%anguage and cognition are considered to be inse#arable) the structure of linguistic categories is
held to reflect the structure of conce#tual categories (e.g. in the sense that the meaning of a word
is the cognitive category connected with it). "n the cognitive view, word meaning is not
determined by the language system itself, but reflects how #eo#le interact with, #erceive and
conce#tuali7e the world. Blthough on the classical view word meaning was attributed a
conce#tual status (Saussure, 1&1') 1E'A Trier, 1&=1A Nat7, 1&3;) =(), it is only the cognitive
#aradigm in linguistics that the mental (i.e. conce#tual) status of word meaning has been so
clearly brought to the fore.
,oncerning the structure of word meaning, the classical a##roach advocates that le0ical
conce#ts are well deliniated entities whose definitions are e0#ressed in terms of an invariable set
of necessary and sufficient features a##licable to all instances in that conce#t. Fhile cognitive
semantics does not e0clude that some le0ical conce#ts may be analysed in terms of necessary and
sufficient features (e.g. >odd number?, >even number?, >#lane geometry figure?), it maintains
that on the whole, they are not amenable to this ty#e of analysis. "n the cognitive view instances
of a conce#t may not be lin!ed because they all share the same features, but because they share
different sets of features with each other. The features lin!ing the various instances of a le0ical
conce#t have been called 1the family resemblance relationshi#2.
,ognitive semantics suggests that all conce#tual information associated with a le0ical
item is broadly encyclo#edic, that is, it is #art of and needs to be understood against broader
cognitive structures. These cognitive structures have been labeled by using a diverse range of
terms) schema (6arlet, 1&=;), scri#t (Schan! and Bbelson, 1&33), #rototy#e (-osch, 1&3(),
e0#eriential gestalt (%a!off and Lonhson, 1&(+), global #attern (de 6eaugrande and <ressler,
1&(1), frame (4illmore, 1&(;), ideali7ed cognitive model (%a!off, 1&(=), mental model (Lonhson@
%aird, 1&(=), cognitive domain (%angac!er,1&(3).
,ognitive semantics is the semantic a##roach of linguists who see no se#aration between
linguistic !nowledge and general thin!ing, or cognition. ,ognitive linguists tend to ado#t a
functional view of language, as o##osed to the more formal accounts favoured by ,homs!y and
similar generative linguists. They argue that no ade$uate account of grammatical rules is #ossible
without considering the meaning of elements. Bs such, the difference between language and other
mental #rocesses is viewed as one of degree rather than !ind.