Download as doc, pdf, or txt
Download as doc, pdf, or txt
You are on page 1of 18

THEORIES OF MEANING Odgen & Richards (1931) The Meaning of Meaning .

Conceptual theories of meaning maintain that meaning is all in the mind. Meaning is the specification of the cogniti e meaning as competent spea!ers of the language. These are closest to the commonsense ie" of meaning #ecause the$ ie" meaning as the structure idea (concept) of the person using the e%pression. &s children gro"ing in a culture "e ac'uire in the form of "ord meanings a huge amount of pre(pac!aged concepts shared #$ those around us. )ecause these underline concepts are shared communication #ecomes possi#le. This ie" of meaning is often associated "ith the 1*th c. philosopher +ohn ,oc!e. -.ords in their primar$ immediate signification stand for nothing #ut ideas in the mind of he "ho uses them. Conceptual theories ta!e arious forms/ especiall$ as regards the follo"ing important 'uestion0 1f meanings are concepts2ideas 3 ho" are the$ ac'uired4 ,oc!e #elie ed that the human mind is analogous to a #lan! state (ta#ula rasa) #ut since "e all share the same sense organs and therefore e%perience the same "a$ "e come to ha e influences/ the same #asic ideas. Our comple% ideas are then #uilt up form these comple% ideas/ deri ed from e%perience. This is the empiricist position #ecause it emphasi5es the importance of !no"ledge deri ed through the senses. Modern ersions of these positions are maintained #$ some ps$chologists and some linguists including 6. ,a!off and R. ,angader. Their theor$ 3 -The Cogniti e Theor$ of ,anguage7 3 a#stract concepts are formed ia a metaphorical/ conceptual transfer from the more concrete to the more a#stract domains of e%perience. The alternati e position to empiricism is rationalism "hich is held #$ contemporaries of ,oc!e8s such as 6. ,ei#ni5 & R. 9escartes. The$ claim that the er$ simplest concepts are a natural or innate propert$ of the human mind/ acti ated #$ e%perience #ut not "holl$ or solel$ deri ed from it. One argument in fa our of semantic rationalism is that it is hard to see ho" concepts as time causation and identit$ could e er #e deri ed from purel$ sensorial e%perience. &nother argument comes from the ast comple%it$ and intricac$ of ,anguage. 1t seems implausi#le that children can ac'uire the comple%it$ of language "ithout some innate pre(disposition. Rationalism0 our #asic concepts are innate. The present da$ linguists "ho adhere the rationalist ie" are :oah Choms!$/ ; . +ac!endoff and &. .ier5#ic!a. & !e$ concept0 <emantic primiti es. The$ claim that if meaning anal$sis is possi#le at all there must #e a set of #asic terms "hich cannot #e defined or reduced further and "e "ould reach as the end point of all anal$ses. <uch elementar$ meanings are called semantic primiti es (though the$ "ere !no"n as simple ideas to the 1*th c. philosophers). +ac!endoff -<emantics and Cognition7 (19=3) claims that meaning mirrors thought and is uni ersalist in nature. >e is a proponent of the idea of strong innateness and claims that general cogniti e a#ilities are sufficient to e%plain all linguistic #eha iour. >is "or! is immersed in the tradition of generati e linguistics and conse'uentl$ he maintains that !no"ledge is descri#ed in terms of a finite set of formal principles that collecti el$ descri#e the infinite set of "ell(formed structures and e%pressions. <emantics according to him should descri#e the mapping #et"een surface form and conceptual meaning. >e is preoccupied "ith larger than the "ord -linguistic elements7 "hich he claims are #uilt #ottom up from the com#ination of semantic primiti es/ according to the rules of semantic and s$ntactic "ell formedness. +ac!endoff proposes that among the innate concepts is a set of conceptual categories "hich are semantic parts of speech. These include thing/ e ent/ state/ action/ place/ path/ propert$ and amount. ?ach of these can #e ela#orated into a function 3 argument organi5ation #$ means of "hat he calls -conceptual functions7. Conceptual functions include 3 go/ sta$/ to/ from/ ia and cause among others. (a) e ent 6O (@T>1:6A/ @B&T>A) @?C?:TA (#) e ent <T&D (@T>1:6A/ @<T&DA) (a) )ill "ent to :e" Dor! (#) )ill sta$ed in the !itchen. & constituent "hich is an e ent can ta!e the form of either of the t"o e ent functions (1 st go/ motional pathE Fnd sta$(stasis o er a period of time).

?ach of these f( ) ta!es F arguments the 1 st of "hich in #oth cases is a thing and the Fnd a path in the case of go and sta$ in the case of sta$. This approach of a#stract metalanguage of conceptual functions is su#Gect to a serious criti'ue/ "hich emerges once "e as! the 'uestion -.hat is the relationship #et"een this metalanguage and ordinar$ ?nglish. & semantic metalanguage can ser e its functions onl$ if it is intelligi#le and the onl$ "a$ to understand a formula is ia our !no"ledge of the meanings of "ords. &n$ a#stract metalanguage is therefore a degenerate form of a natural language. To understand the formula "e ha e to turn it #ac! mentall$ into ordinar$ ?nglish undoing the deformation that has #een in ol ed in turning it into a technical formula. .ier5#ic!a8s theor$ is a!in to +ac!endoff8s theor$. <he claims that meaning can #e descri#ed #$ using the most natural !ind of metalanguage #ecause meaning is uni ersal in :ature. &nd the semantic primiti es are the same for all people due to the cogniti e and ps$chological unit$ of human !ind. Meaning in natural languages is a propert$ of "ords and is a com#ination of semantic primiti es "hich generate the o erall meaning of the "ord. <he proposes a ta#le of the semantic primiti es "hich constitute a list of le%ical uni ersals. <he proposes a num#er of a#out HI such primiti es "hich are the ultimate components that com#ine in a certain specifia#le "a$s to e%press meaning. The "a$s are descri#ed "ith the help of a natural semantic metalanguage using the method of reducti e paraphrase. &n ideal semantic metalanguage anal$sis is a paraphrase composed of the simplest possi#le terms thus a oiding circularit$ and o#scurit$. :o technical terms/ fanc$ "ords/ logical s$m#ols or a##re iations are allo"ed in the paraphrase "hich should contain onl$ simple e%pressions form ordinar$ natural languages. (1/ $ou/ someone/ do/ happen/ thing/ #ig/ #ecause). TYPES OF MEANING CONCEPTUAL M a central factor in ,inguistic comm.. &pplied to the conceptual content of an e%pression2"ord usuall$ the conceptual content is gi en in the form of semantic primiti es. .oman @humanE adultE femaleA 3 2uni ersal2 semantic primiti es. )achelor of &rts @human/ adult/ JmaleA. The aim of conceptual semantics is to pro ide/ for an$ interpretation of a sentence/ a configuration of a#stract s$m#ols "hich is its semantic representation -and "hich sho"s e%actl$ "hat "e need to !no" if "e are to distinguish that M form another possi#le sent. Ms in the ,/ and to match the M "ith the right s$ntactic and phonol.e%pression. ?ncoding the M. from phonol. 3s$nt. ( to semanticsE 9ecoding the Message0 <emantics 3 phonol. 3 s$nta%. CONNOTATIVE M. 3 the communicati e alue an e%pression has #$ irtue of "hat refers to/ o er and a#o e its purel$ conceptual conte%t. Connot. M can em#race the -putati e properties7 of the referent/ due to the ie"point adopted #$ an indi idual/ or a group of people or a "hole societ$. (#iped 3 ha ing a "om# ( Kps$chol. properties such as -e%periencing maternal instincts/ irrational/ fragileE social properties of "omen0 gregarious/ tal!ati e/ e%perienced in coo!ing). Connotations ar$ form indi idual to indi idual in a gi en speech communit$. 1t aries from age to age (L1L 3 LL1). Car$ from societ$ to societ$ in the same age2period. Conn.are relati el$ unsta#le. The$ R inindetermined and open ended/ in a sense "i "hich conc.M is not. REFLECTIVE M. ( in ol es an intercommunication on the le%ical le el of ,. 1t arises in cases of multiple conceptual M/ i.e. pol$sem$ 3 "hen one sense of the terms conditions or reacts to another sense of the "ord. (?.g. 1n a Church <er ice0 The >ol$ 6host/ The Comforter 3 refers to the 3 rd p of the Trinit$). The case "here refl.M intrudes through the sheer strength of emoti e suggestions is most stri!ing in ta#oo (e.g. se%ual terms -intercourse7/ -coc!7). The process of ta#oo contamination/ #rings a#out the d$ing out of the non(ta#oo sense of a "ord. AFFECTIVE M. 3 , can reflect the personal feelings of the spea!er/ including his attitude to the listener/ or his attitude to the su#Gect. &ff.M issometimes e%plicitl$ con e$ed #$ the conc. Or conn. M of the "ords used. Mactors such as intonation and oice pitch are also important here. &ff.M is largel$ a parasitic categor$ in a sense that to e%press our emotions "e rel$ upon the meditation of other categories of M 3 conc./ conn./ or st$listic. STYLISTIC M. 3 the M "hich a piece of , con e$s a#out the social circumstances of Ms use. .e recogni5e st.M through our recognition of diff.dimensions and le els of usage "ithin the same ,. &ccent can inform us of the geographical or social origin of the spea!erE other features tell us a#out the social relationship

#2n spea!er and hearer. .e rarel$ find "ords "hich ha e #oth the same conceptual and the same st$listic M.NOno true s$non$ms e%ist. COLLOCATIVE M. 3 consists of the associations a "ord ac'uires on account of the Ms of "ords "hich tend to occur in its en ironment. @prett$0 ( girl/ #o$/ "oman/ flo"erA/ @handsome0 ( man/ #o$/ essel/ carA/ @rancid0 ( #utterA/ @heinous0 ( crimeA/ @noisome0 smellA. @Collco./ affecti e./ reflected K st$listic M are associati e.A. THEMATIC M. 3 1t is a com#ination #2n the "a$ in "hich a sp. organi5es a message in terms of ordering/ focus/ emphasis. @e.g. -Mrs <mith donated the 1 st pri5e7 or -The 1st pri5e "as donated #$ Mrs <mith7A. Them.M us a matter of choice #2n alternati e gram constructions. &nother classification0 1:T?:9?9 @"hat the listener infers "hen he recei es the messageA s. 1:T?RBR?T?9 @the meaning of the spea!erA Mrom theoretical point of ie" "e recogni5e all these t$pes of meaning and tr$ to anal$5e them/ ho" ordiner$ spea!ers recogni5e all these t$pes of meanings4 The$ all ha e !no"ledge of the language. This illustrates that language is innate/ "e are #orn "ith it. Mirst "e ha e the ps$cholog$/ "hich deals "ith e%ternal realit$E second/ "e thin! in specific structures and models/ and third "e #eha e in specific models. &ll these sho" that "e are social animals and part of the societ$. .e e%press our conception of the "orld in the language. There are as man$ conceptions of the "orld as man$ persons. This is a fundamental 'uestion/ common for semantics/ culture/ cognition/ and ps$cholog$. 1s it me as a part of the "orld or is it me against the "orld4 Thus/ "e "ill #e a#le to understand the meaning of some "ords. &t first it seems that it is me as a part of the "orld/ if 1 am ali e/ e er$thing is ali eE it gro"s/ gets old/ and dies. <ometimes "e use euphemisms instead of ta#oos as an e%pression of our attitude to"ards ocean/ lightening in case not to ma!e them angr$. This is a product of primiti e societ$. .ater and earth are gi en female names/ #ecause the$ are associated "ith #irth and death. >ea en is associated "ith male/ so e er$thing that comes from there is gi en a male name. Man #ecame more po"erful/ determiming his o"n position in the "orld and formed sertain oppositios of thin!ing/ "hich are uni ersal for all cultures and languages/ li!e0 $oung(old/ "arm(cold/ up(do"n/ etcE usuall$ the first element is the positi e one. .hen "e deal "ith connotations in the language/ "e ha e these ( e and K e oppositions ( #asic stucture of human thin!ing. Man is in the centre/ in the domestic/ K e side/ he is the !no"n/ "here the nature is the un!no"n. The language as a li ing organism "hich reacts and changes "ith the en ironment. &ll languages differ in meaning/ or ha e the same structure of meaning ( it is a maGor pro#lem of linguistics ( the uni ersalism ( all languages ha e the same semantical structure/ #ecause "e are all human #eings. On the other hand/ languages differ in their semantic structures/ #ecause the$ e%press different ie" of the "orld. &t the #eginning of the centur$/ <pair and .horf started in estigating the culture and customs of &merican 1ndians. &fter the$ anali5ed these languages/ the$ came up "ith the idea that the$ differ from ?uropean ones #ecause the "a$ of thin!ing is different. The$ ha e different idea of space and time. <?M&:T1C R?,&T1O:< There are relationships in semantics. The most general t$pes must correspond to the t"o #asic t$pes of relation in language. 6oing #ac! to the theor$ of le els/ "e come to the syntagmatic and the pa a!igmatic relations 3 also alid for the relations among le%ical units. & "ord is related paradigmaticall$ to other "ords "ith "hich it appears in the same conte%t. E"# a ------------- of milk ( in the gap 3 "ords denoting containers). &ll the "ords in the gap are in paradigmatic relation. & "ord is in s$ntagmatic relation "ith all the "ords/ "hich it colours0 all the "ords denoting containers are in s$ntagmatic relation. .e can anal$se the "ords paradigmaticall$ and loo! for relations/ "hich organi5e them in such a "a$/ and loo! for linear relations (O s$ntagmatic relations. S$%$cti&na% $st icti&ns the position of the er# in the sentence. The t$pe of er# "e choose usuall$ selects the other "ords that appear in the sentence. E"# S (obligatory) make O! (obligatory) InO A!' Inst .e are a#solutel$ sure for the o#ligator$ and the optional items. S personal pronouns 2 human nouns/ #ecause the er# ma($ means volitional action.

O! er$ indefinite categor$. <emantic theor$ is interested in the Fnd t$pe of selection "here the appearance of the "ords is not found so sensi#le/ and anomalous. The pro#lem of semantics is why is it anomalous? What makes it anomalous? E"# a stallion got pregnant 3 semantic clash #et"een "ords. The silence was pregnant with meaning - no semantic clash. There is no pro#lem "hen the "ord is used metaphoricall$ or figurati el$. The paradigmatic semantic relations group "ords into smaller su#di isions (smaller than the semantic fields) and then e%press er$ specific relations among or #et"een the "ords 3 sometimes the$ concern t"o/ sometimes more than t"o "ords. .e ha e0 syn&nymy) *yp&nymy) ant&nymy) inc&mpati+i%ity) c&mp%$m$nta ity) c&n'$ s$n$ss) pa t &, $%ati&ns) cyc%$s) an(s and so on. .ords are related in man$ different "a$s. The relation can #e +ina y 3 oppose or relate t"o "ords onl$. E"# boy girl a #inar$ opposition O se% O pri ati eE man girl opposed in more than one dimension0 sex age! P i'ati'$ '-s $./ip&%%$nt &pp&siti&n# P i'ati'$# implies that one of the mem#ers is characteri5ed #$ the presence of a feature0 man" woman. The first mem#er is al"a$s a strong one and carries the feature characteri5ing the opposition0 man (#maleE ( female)0 woman ((maleE K female) 3 the features al"a$s appear in this order. <ocial or cultural agreement is to come first. E./ip&%%$nt# a #inar$ opposition/ "hich consists of t"o mem#ers #ut the$ are e'ual 3 the$ mutuall$ e%clude each other0 alive" $ea$. .e can use sentences li!e0 %s he still alive? or %s he alrea$y $ea$? / #ut "e cannot as! such 'uestions for he 1st t$pe of opposition. M/%tip%$ &pp&siti&ns# in ol e more than t"o "ords and the relation among such "ords is of m/t/a% $"c%/si&n# the $ays of the week al"a$s in opposition. Syn&nymy# e%plains the difference #2n s$non$ms. <$non$ms are "ords "ith similar meaning. .ords "ith different meaning are0 tree" &niversal. 'ot v(s col$ are ant&nyms 3 the$ ha e 1 feature in common. &nton$ms ha e er$ much the same meaning. The s$non$ms usuall$ refer to one and the same phenomenon or o#Gect. The additional elements of meaning can e%press sta#le connotations/ the attitude of societ$ to"ards a gi en o#Gect0 unpleasant" ugly" $isgusting. 1f it is a er# 3 the$ can differ in the "a$ of performing the action in the goal of doing it or differ in the "a$ of performing it. E"# 0& !s ,& (i%%ing# kill mur$er slay slash assassinate electrocute stab hatchet guillotine hang $ecapitate behea$ poison $rown stifle gun shoot choke suffocate exterminate massacre mutilate crucify rape assault $estroy. POLYSEMY AN1 HOMONYMY 3 <ame form different meaning. Criteria for distinction0 1.>istorical deri ation of the "ords. >omon$ms0 9e eloped from formall$ distinct le%emes. e.g. #an! 3 1 . of a ri erE F.. financial institution Psuall$ in the dictionar$ these are gi en in separate entries/ #ut it is not entirel$ decisi e "ith respect to nati e spea!ers. F.On the s$nchronic plane0 criterion of relatedness of meaning. This criterion correlates "ith the nati e spea!er8s feeling of connection #et"een certain "ords0 Mouth 1. of a ri erE F.of a #ottleE 3. of the human #od$ >ead0 1. of a #od$E F. of a famil$E etc. This criterion is not al"a$s relia#le due to fol! et$molog$0

?ar 1. Qof a corn 3 (RSTUV WU XUYZ[\XU )E of a human organ. The$ come from different stems/ #ut nati e spea!ers "ould interpret them as related. &#solute and partial homon$m$. &#solute0 #elonging to the same part of speech/ all forms are identical and the$ are s$ntacticall$ e'ui alent0 #an! 3 ri erE mone$. Bartial0 the "ords do not coincide in all forms/ or #elong to different parts of speech. e.g. found (Nesta#lish)/ (Npast t. of ]find8) This ma$ gi e rise to am#iguities0 -The$ found the hospital and charita#le institutions7. )ut usuall$ the "ords occur in different s$ntactical en ironment0 last (adG)/ to last ( er#). Bol$sem$0 Cruse defines the relations e%isting #2n pol$semes as linear and non(linear. ,inear relations0 1f one thing (&) is more #asic than another thing(#) and (#) is more speciali5ed than & then ) is a speciali5ation of & A/t&2*yp&nymy) 1t occurs "hen a "ord has a default/ general sense and a conte%tuall$ restricted sense "hich is more specific. (9ogs and cats are animalsE That8s not a dog/ it8s a #itch) A/t&2m$ &nymy0 9eals "ith the part("hole relation (finger 3 hand 3 arm). .e ha e automeron$m$ "hen the more specific reading denotes su#part (6o through that door and ta!e the door off its hinges). A/t&2s/p$ & !inati&n. The generali5ation of a su#ordinate term (-men7 for the "hole of humanit$) A/t&2*&%&nymy 3 & scratch on the arm means that the scratch is on the non(hand part (<he "as "a ing arm 3 hand included). N&n2%in$a 0 Morphos$ntactic "ords "hich ha e one and the same phonic representation are called homophones. Morphos$ntactic "ords "hich ha e one and the same orthographic representation "hich is formall$ e'ui alent in the graphic medium are called homographs. >omophones ma$ #e defined as a#solute or partial. N&n2%in$a p&%ys$my0 metaphor 3 #ased on resem#lance of "ordsE meton$m$ can #e characteri5ed for the use #ased on associations ^Most often Migurati e , especiall$ metaphorical e%tension of the primar$ meaning. ( -The #ed of a ri er7/ -hood of a car7/ -coc! of a gun7). &lso meton$m$ 3 surger$ 3 the art of a surgeonE the room "here patients are e%amined. <emantic #orro"ing0 the #orro"ing of meaning. (e.g. ,ord meant master. :o"ada$s it also refers to 6od/ due to the influence of the 6ree! "ord for master "hich set the pattern. The same "ith Mrench (<egneur).Creation of technical terms0 e.g. action0 in la"/ in militar$ language 3 has more specific technical meaning. Commerce0 compan$/ interest/ share <ports0 score/ goal/ #ac!/ centre._ H&m&nymy 's. P&%ys$my 1t could #e argued for man$ :,B applications that a coarse differentiation #et"een the different meanings associated "ith a single le%ical form is ade'uate to esta#lish a #asic interpretation "hich can guide su#se'uent processing. This might #e the case for some MT s$stems or :,P s$stems "hich onl$ need to esta#lish the general conte%t of discourse. This coarse differentiation might #e at the le el of homon$ms/ lea ing pol$semous "ords to #e associated "ith an underspecified le%ical semantics "hich is ne er made full$ precise. <o "hat is the homon$m$(pol$sem$ distinction and on "hat grounds is it made4 1 "ill sho" that the distinction is not clear and therefore not a useful #asis for deciding "hat "ords2senses to include in the le%icon. & "ord "ith (at least) t"o entirel$ distinct meanings $et sharing a le%ical form is said to #e homonymous (e.g. mogul/ an emperor/ or mogul/ a #ump on a s!i piste)/ "hile a "ord "ith se eral related senses is said to #e polysemous (e.g. mouth/ an organ of the #od$/ the entrance of a ca e/ etc.) l$ons0**. .hile these definitions are intuiti el$ clear/ it has #een pointed out man$ times in the literature on le%ical semantics that a clear operational distinction #et"een homon$m$ and pol$sem$ is lac!ing. 1 "ill re ie" some of the criticisms #elo"/ #ut "ill #egin #$ introducing an e%ample "hich emphasises the difficulties of esta#lishing criteria for distinguishing #et"een homon$m$ and pol$sem$. One of the most commonl$ cited e%amples of a homon$mous "ord is bank/ "hich has a financial institution sense and a e$ge of a river sense. These senses seem clearl$ unrelated/ and the fact that the$ are associated "ith the same "ord form seems purel$ accidental. >o"e er/ historical linguistics research on 1talian has re ealed that at some point in the de elopment of the 1talian language/ these t"o senses of bank actuall$ coincided #$ irtue of the fact that #an!ers (lenders of mone$) sat on the ri er#an!s "hile doing their #usiness. <o going to the financial institution meant going to the edge of the ri er/ hence to the bank. Thus a connection #et"een the t"o modern senses of bank can #e esta#lished. The relationship #et"een these t"o senses should presuma#l$ not #e considered strong enough to esta#lish a relation #et"een them/ and therefore

to consider bank pol$semous rather than homon$mous/ #ut "hat criteria for pol$sem$ do the$ iolate4 On "hat criteria do "e decide that senses are related4 Lyons ,$ons l$ons0** de otes a section to the discussion of the homon$m$(pol$sem$ distinction (p. ``I( `H9). >e identifies the follo"ing criteria as those traditionall$ applied in ma!ing the distinction0 1.et$mological information (( homon$mous le%emes aashould #e !no"n to ha e de eloped from "hat "ere formall$ distinct le%emes in some earlier stage of the languagebb F.relatedness of meaning (( homon$mous le%emes ha e unconnected meanings ,$ons correctl$ criticises the first criterion as #eing irrele ant to the s$nchronic anal$sis of language/ since nati e spea!ers are largel$ una"are of the et$molog$ of the "ords of their language $et the$ are a#le to assign meanings to them. :otice that this criterion "ill not e%clude bank as discussed a#o e from #eing pol$semous/ gi en that the t"o senses historicall$ deri e from a single le%eme and in fact a single sense. The second he identifies as an important consideration/ #ut points out that relatedness of meaning is a su#Gecti e measure for "hich intuitions ma$ ar$ among indi iduals. ,$ons considers t"o alternati es to circum enting the homon$m$(pol$sem$ issue0 1. Ma"imis$ *&m&nymy (( associate e er$ meaning of a "ord "ith a distinct le%eme. ,$ons sho"s that this "ill lead to considera#le redundanc$ in the le%icon/ as much morphological/ s$ntactic/ and e en semantic information "ill #e repeated in the le%ical entries for the distinct le%emes. >o"e er/ this redundanc$ can #e greatl$ reduced gi en current inheritance(#ased approaches to le%icon construction. More pro#lematic is the o#ser ation that this approach depends on the a#ilit$ to spell out in ad ance all of the possi#le senses in "hich a "ord "ill #e used. ,$ons suggests that sense distinctions can #e aamultiplied indefinitel$bb (19**0``c) and that therefore this tac! is hopeless. 1t "ill ne er #e possi#le to decide in ad ance the full range of possi#le senses a particular "ord might #e associated "ith/ and furthermore it ma!es the computational tas! of selecting the appropriate le%eme daunting gi en the num#er of le%emes "hich might #e associated "ith a particular "ord form (see <ection H.c #elo"). F. Ma"imis$ p&%ys$my (( adopt the notion that no t"o le%emes can #e entirel$ distinct "hen the$ are s$ntacticall$ e'ui alent and "hen the set of "ord forms the$ are associated "ith are identical. On this ie"/ there are onl$ arious !inds of partial homon$m$ (i.e. "hen there e%ist s$ntactic differences among uses of a "ord). This remo es the ague concept of aasemantic relatednessbb from the le%icon. >o"e er/ it "ould result in an e%tremel$ underspecified le%icon from "hich er$ little information a#out the meaning of "ords could #e gleaned. 1t suffers from the pro#lem of an ina#ilit$ to e%plain the intuitions that underl$ the notion of homon$m$/ and/ more rele ant to computation/ from a complete ina#ilit$ to identif$ the normal conte%t of use of a particular "ord and no #asis for esta#lishing s$non$m classes or other semanticall$(#ased groupings. >o" an$ useful interpretation could #e accomplished "ithout some sense differentiation is difficult to see. Cruse Cruse (19=H) distinguishes lexemes from lexical units. The former are aathe items listed in the le%icon/ or aideal dictionar$b of a language.bb & le%eme corresponds to a particular "ord or "ord form/ and can #e associated "ith indefinitel$ man$ senses. The latter are aaform(meaning comple%es "ith (relati el$) sta#le and discrete semantic propertiesbb (p. c9)/ and the meaning component is called a sense/ corresponding to the intuiti e notion of sense 1 ha e #een using. <o bank is a le%eme/ "hile bank-financial institution and banke$ge of a river are le%ical units. 1n discussing the semantic contri#ution of a "ord to a sentence/ Cruse differentiates #et"een contextual selection of a sense and contextual mo$ulation of a sense. <election refers to acti ation of a particular sense of an am#iguous "ord form due to the conte%t/ "hile modulation refers to ariation induced #$ the conte%t in terms of emphasising or de(emphasising arious aspects of the sense. The sentences in cc1 e%emplif$ modulation in that the t"o sentences highlight different parts of the car (the engine and the #od$/ respecti el$) rather than re'uiring that car refer to different entities in each case. The sentences in ccF/ on the other hand/ are instances of conte%tual selection0 a different sense of light is selected in each case.

The car nee$s servicing. The car nee$s washing. The room "as painted in light colours. (cf. $ark()heavy) &rthur has rather a light teaching load. (cf. heavy()$ark) Conte%tual selection corresponds to the !ind of "ord sense disam#iguation underta!en in most :,B s$stems (( there is a pre(e%isting set of senses for a le%eme ("ord) and the rele ant sense must #e identified #ased on the conte%t. Conte%tual modulation refers to one creati e aspect of language use/ and points to the need for comple% representation of !no"ledge a#out a "ord. This is #ecause multiple aspects of a "ord can #e acti ated simultaneousl$0 The car nee$s servicing an$ washing is completel$ felicitous despite the fact that different facets of the car are referred to #$ each of the er#s (cf. ??*rthur has light teaching loa$s an$ rooms in his house "hich indicates that multiple senses of light cannot #e acti e at once). This !ind of !no"ledge is directl$ rele ant for discourse processing (e.g. anaphor resolution/ for e%ample in a discourse such as aaThe car nee$s servicing. %t also nee$s washing. bb "here the anaphor refers #ac! to the car as a "hole/ not Gust the engine) and generation of coherent sentences2discourses. Cruse (19=H0H=) introduces the concept of a gra$ient of establishment of senses. )$ this he means that a le%ical form can #e associated "ith some senses "hich are potential rather than e%plicitl$ represented in the le%icon. Conte%t can stimulate rules "hich generate an appropriate sense. This idea ser es as the foundation of recent "or! on the 6enerati e ,e%icon (BusteGo s!$ 1991/ pusteGo s!$09`a0 see #elo"). Cruse furthermore argues for the e%istence of sense-spectra/ in "hich the senses of a le%ical form lie along a continuum/ "ith no clear #oundaries #et"een them/ and in some cases "ithout an encompassing sense. copesta!ed#riscoe09` pro ide an e%ample/ sho"n in cc3. That book is full of metaphorical language. That book is full of long sentences. That book is full of spelling mistakes. That book is full of typographic errors. That book has an unrea$able font. That book has lots of smu$ge$ type. That book is covere$ with coffee. &s the$ point out/ co(predication of the first and the last properties seems odd ccca "hile co(predication of adGacent pairs seems natural/ e.g. ccc#. 4That #oo! is full of metaphorical language and is co ered "ith coffee/ so itbs er$ hard to read. That #oo! is full of t$pographical errors and has an unreada#le font. Cruse (19=H0*3) suggests that the description of sense(spectra is pro#lematic since a full sense( spectrum does not function as a single le%ical unit. Det he proposes to treat them as a le%ical unit/ "ith recognition of the senses along the continuum as local senses. & lexeme/ corresponding to a le%ical entr$/ is proposed #$ Cruse (19=H0*H) to #e a famil$ of le%ical units. This famil$ can either correspond to a sense(spectrum/ or to a set of senses "hich can #e related to one another ia regular le%ical semantic relationships (captured #$ le%ical rules). Thus the structure of the le%icon on Crusebs ie" essentiall$ reflects onl$ producti e relationships and groups of senses capturing different aspects of a single entit$ "hich cannot #e consistentl$ delineated. 1n proposing this/ Cruse focuses on the le%ical unit as the primar$ semantic unit and on the distinction #et"een le%ical units/ de(emphasising the importance of the "ord. &s such/ he s!irts the homon$m$(pol$sem$ distinction issue/ "hich in ol es the relationship #et"een le%emes and le%ical units. &lthough the difference #et"een conte%tual selection and conte%tual modulation points to phenomena affected #$ the distinction/ he does not propose clear criteria for esta#lishing the distinction. <o the computational le%icographer is left "ith $et another reason for ma!ing the distinction #ut still no #asis for ma!ing it. Kilgarriff &dam ;ilgarriff (199F) de otes his thesis to a discussion of pol$sem$. The essential conclusion "hich he dra"s/ on the #asis of considerations of the traditional distinction dra"n #et"een homon$m$ and pol$sem$ and of in estigation of le%icographic techni'ues for delimiting dictionar$ senses/ is that aaBol$sem$ does not form an$ !ind of anatural !indbbb (;ilgarriff 199F0c). 1nstead/ pol$sem$ descri#es a aacrossroadsbb #et"een homon$m$/ alternations/ collocations and analog$ #ased on general !no"ledge and reasoning.

Bol$semous "ords can #e characterised #$ at least one of these four methodologies/ and ;ilgarriff argues that all four must #e allo"ed for in order to capture the full ariet$ of pol$sem$. ;ilgarriff o#ser es/ ho"e er/ that collocations and analog$ depend on fre'uenc$ information and are su#Gect to conte%tual ariation "hile description of homon$m$ and alternations relies on rules/ and that Goining the t"o approaches in ol es augmenting formal le%ical structure "ith fre'uenc$ data. This ie" is supported #$ the results of m$ in estigation of logical meton$m$ (Chapter `)/ "ahich suggests that con entionalit$ pla$s an important role in predicting language use. The computational le%icon must therefore #oth reflect linguistic generalisations and pro ide information on con entional language usage. The main implication of this "or! is that pol$sem$ is not a term "hich can #e applied to characterise "ord senses in an entirel$ precise "a$. There cannot #e clear(cut tests for identif$ing pol$sem$ due to its multi(faceted nature. >omon$m$ is not orthogonal to pol$sem$/ #ut rather an endpoint of one of the dimensions along "hich pol$sem$ can #e descri#ed (full$ predicta#le sense ariation (( unpredicta#le sense ariation). Murthermore/ most "ords displa$ some ariation in the meaning the$ e%press and the criteria for pinning do"n senses are often dependent on 'uestions of fre'uenc$ and predicta#ilit$ rather than on clearl$ delineated distinctions. Mor the purposes of designing a le%icon for an :,P2:,6 s$stem/ that means distinguishing senses of a "ord "hen there are s$ntactic differences in the "a$ that "ord is used/ and "hen there are ariations in meaning "hich seem to follo" from general/ producti e relationships. HYPONYMY 2 A pa a!igmatic $%ati&n &, s$ns$ 0*ic* *&%!s +$t0$$n a m& $ sp$ci,ic) & s/+& !inat$) %$"$m$ Co"0 animal Rose0 flo"er >onest$0 irtue )u$0 get Crimson0 red There is no generall$ accepted term for this relation (or its con erse). >o"e er/ the term h$pon$m$ has #een gaining currenc$E and it is more appropriate than such alternati es as ]inclusion8 or ]su#ordination8/ "hich are also used in other senses in linguistics and logic. ,et us sa$ that ]co"8 is a h$pon$m of ]animal8/ ]rose8 is a h$pon$m of ]flo"er8/ etc./ and further/ that ]rose8/ ]tulip8/ ]daffodil8/ etc./ since each is a h$pon$m of ]flo"er8/ are co(h$pon$ms of the same le%eme. >$pon$m$ is fre'uentl$ discussed #$ logicians in terms of class( inclusion and up to a point/ this is satisfactor$ enough. Mor e%ample/ 1f L is the class of flo"ers and D is the class of tulips/ then it is in fact the case that L properl$ includes D (LD & D L)/ #ut there are pro#lems attaching to the definition of h$pon$m$ in terms of the logic of the su#classes. Mirst of all/ it is unclear "hether "e should sa$ that a h$pon$m is included in its superordinate or a superordinate in its h$pon$m(s). 1f "e consider the e%tension of le%emes/ "e "ould sa$ that the superordinate le%eme is more inclusi eE #ut as far as the intension of le%emes is concerned the h$pon$m is more inclusi e (tulips ha e all the defining properties of flo"ers/ and certain additional properties "hich distinguish them from roses/ daffodils/ etc.) >$pon$m$ is defina#le in terms of unilateral implication. Mor e%ample/ ]crimson8 is esta#lished as a h$pon$m of ]red8 and ]#u$8 as a h$pon$m of ]get8 #$ irtue of the implications -<he "as "earing a crimson dress7 -<he "as "earing a red/ dress7 -1 #ought it from a friend7 - 1 got it from a friend7 (That is/ #et"een the propositions e%pressed #$ the sentences -<he "as "earing a crimson dress7 and -<he "as "earing a red/ dress7/ etc./ "hen these sentences are uttered to ma!e an assertion). The definition of h$pon$m$ in terms of unilateral implication ena#les us to define s$non$m$ as #ilateral/ or s$mmetrical/ h$pon$m$0 1f % is a h$pon$m of $ and $ is a h$pon$m of %/ then % and $ are s$non$ms. 1f h$pon$m$ is defined as non(s$mmetrical (as it must #e if s$non$m$ is treated as s$mmetrical h$pon$m$)/ then proper h$pon$m$ ma$ #e distinguished from s$non$m$ as #eing as$mmetrical.

>$pon$m$ is a transiti e relation. 1f % is a h$pon$m of $ and $ is a h$pon$m of 5/ then % is a h$pon$m of 5. Mor e%ample/ ]co"8 is a h$pon$m of ]mammal8 and ]mammal8 is ]a h$pon$m of ]animal8E therefore ]co"8 is a h$pon$m of ]animal8. .hen the relation of h$pon$m$ holds #et"een nouns/ it is possi#le to insert s$ntacticall$ appropriate e%pressions containing then in place of % and $ in the follo"ing formula +x is a kin$ of y, ("here x is a h$pon$m of y) and this "ill $ield a sentence "hich e%presses a metalinguistic or refle%i e proposition "hich is anal$tic. Thus/ a proposition e%pressed #$ -& co" is a !ind of animal8/ ]& tulip is a !ind of flo"er8/ etc./ ma$ #e ta!en to #e anal$tic. Pnder more restricted conditions/ ]sort8 and ]t$pe8 ma$ #e su#stituted for ]!ind8 in collo'uial ?nglish70 ]& co" is a sort of animal8/ ]& tulip is a t$pe of flo"er8. There are man$ other more specific le%emes "hich ma$ #e emplo$ed/ for certain alues of % and $0 e.g./ ]shade8 in ]Crimson is a shade of red8/ ]ma!e8 in ]&ston Martin is a ma!e of car8/ and so on. .hen a noun % is superordinate to more than one h$pon$m/ $/ 5/ etc./ such e%pressions as the follo"ing "ill #e accepted as meaningful0 ]co"s and other (!inds of) animals8/ ]tulips and other (!inds of) flo"ers8/ "hich ma$ #e contrasted "ith the semanticall$ anomalous ]co"s and other (!inds of) flo"ers8 and ]tulips and other (!inds of) animals8. The concept of h$pon$m$ can #e e%pressed in ordinar$ language/ as - is a type(kin$(sort of . . 1t is interesting that some pairs of "ords that satisf$ the logical definition of h$pon$m$ collocate more accepta#l$ in this frame than others0 & horse is a t$pe of animal. 4 * kitten is a sort of cat. (* kitten is a young cat.) ? * stallion is a type of horse. (* stallion is a male horse.) ? * /ueen is a ki$ of woman. (* /ueen is a woman.) 1n Cruse (19=H) the relation e%emplified #$ horse0 animal #ut not stallion0 horse "as la#eled ta"&nymy/ #ecause of its rele ance to classificator$ s$stems. Ta%on$ms t$picall$ resist (genuine) anal$sis in componential terms and do not ha e o# ious definitions0 * stallion is a male horse. * horse is a animal. Cer#s/ adGecti es and other parts of speech cannot #e inserted into the formula ]% is a !ind of $8 "ithout prior nominali5ation/ and e en the resultant sentence is generall$ rather unnatural/ #ut not a#solutel$ unaccepta#le (])u$ing is a !ind of getting8). The general principle0 >DBO:DMD 1< & B&R&916M&T1C R?,&T1O: OM <?:<?/ .>1C> R?<T< PBO: T>? ?:C&B<P,&T1O: 1: T>? >DBO:DM OM <OM? <D:T&6M&T1C MO91M1C&T1O: OM T>? <?:<? OM T>? <PB?ROR91:&T? ,?L?M?. SYNONYMY <$non$m$/ Gust li!e man$ other linguistic issues/ has #een ie"ed in a different "a$ in the different semantic theories. )ut "hat the$ ha e in common is the large tas! to collect together the collect together/ e%plain and s$stemati5e the properties of a language that a semantic theor$ should e%plain/ and linguists agree that in order to #e ade'uate/ a semantic theor$ must fulfill at least three conditions0 3i4 it m/st capt/ $ ,& any %ang/ag$ t*$ nat/ $ &, 0& ! an! s$nt$nc$ m$aning) an! $"p%ain t*$ nat/ $ &, t*$ an!&m $%ati&n +$t0$$n t*$m5 3ii4 it m/st +$ a+%$ t& p $!ict t*$ am+ig/iti$s in t*$ ,& ms &, a %ang/ag$) 0*$t*$ in 0& !s & s$nt$nc$s i.$. it m/st gi'$ s&m$ $"p%icit acc&/nt &, t*$ $%ati&ns &, synonymy, logical inclusion, entailment, contradiction, etc. 5 3iii4 it m/st gi'$ t*$s$ c*a act$ i6ati&ns in t*$ ,& m &, a ,init$ s$t &, /%$s capt/ ing t*$ $g/%a iti$s c&ntain$! in s&m$ sp$ci,ic in,init$ s$t &, s$nt$nc$s. .e !no" that all languages ha e "ords "ith meanings associated to them/ and these "ords can #e arranged in different semantic structures to form sentences . <o/ "e ha e t"o le els of generalisation in the form of t"o different components0 (i) a syntactic c&mp&n$nt 3 containing all the general statements a#out the principle of sentence formation and (ii) a s$mantic c&mp&n$nt 3 containing a statement of "ord meanings and a set of rules predicting sentence meanings. &nd of course/ there is an interaction #et"een syntactic an$ semantic properties of sentences. 1n ?nglish0 1. 0ever before ha$ the opera house been close$ same meaning 3 onl$ the place of the ad er#s never an$ before is changed The opera house ha$ never been close$ before F. 1ats chase $ogs

3nb sp2

the su#Gect and the o#Gect are interchanged 3 #ig difference in meaning 3nbsp223nb sp23nbsp2 2

4ogs chase cats Mrom this follo"s that sometimes re(ordering affects the meaning of the sentence and the prediction of sameness or difference (a matter of semantics) depends on grammatical la#els such as ad er#/ su#Gect and o#Gect (a matter of s$nta%). 7&*n Li&ns ma!es a t"ofold distinction #et"een -sense7 and - reference7 or -meaning7 and -reference7. -Reference7 is often related "ith the relation #et"een "ord and o#Gect. &s for -meaning7/ t"o linguists Ogden and Richards proposed that -meaning7 can actuall$ #e approached as 0 -emoti e7 and -referential7. Mor e%ample the "ords - horse5 and -stee$5 ha e the same referential meaning/ #ut differ in emoti e meaning/ so the opposition #et"een a more central/ or st$listicall$ neutral/ component of meaning and a more su#Gecti e or peripheral component of meaning is a commonplace of discussions of s$non$m$ and this is also alid for the distinction #et"een descripti e/ social and e%pressi e meaning. &s far as semantic information and semantic fields in general are concerned/ the possi#ilit$ of selecting one unit rather than another (and in most cases com#ining it/ according to the rules to the signaling(s$stem/ "ith other meaningful units) is a precondition of #eing a#le to transmit different messages "ithin the signaling(s$stem in 'uestion. Baradigmaticall$ related units are not necessaril$ different in meaning/ ho"e er/ the selection of one le%eme rather than another ma$ ha e no effect upon the message that is transmitted. 1n this case/ "e can sa$ that intersu#stituta#le le%emes are c&mp%$t$%y syn&nym&/s. The selection of one rather than another ma$ change the social or e%pressi e meaning of the utterance/ #ut hold constant its descripti e meaning ( is an$) in "hich case "e can sa$ that the intersu#stituta#le le%emes are !$sc ipti'$%y syn&nym&/s (i.e. ha ing the same sense). Mrege8s classic e%ample used in discussions of sense and reference is0 (i) &n #spE The 6orning 7tar is the 8vening 7tar

&s F $g$ pointed out/ the t"o e%pressions -the 6orning 7tar7 and -the 8vening 7tar7 had the same reference ()edeutung)/ since the$ each referred to the same planet/ #ut the$ could not #e said to ha e the same sense (<inn)/ for if the$ did (i) "ould #e tautologous and anal$tic/ as is0 (ii) &n #spE The 6orning 7tar is the 6orning 7tar

)ut (i) unli!e (ii)/ is potentiall$ informati e0 it can ma!e the hearer a"are of some fact of "hich he "as not pre iousl$ a"are and "hich he could not deri e simpl$ from his understanding of the meaning of the sentence. <o/ it follo"s that -the 6orning 7tar5 3 9the evening 7tar5 are not s$non$mous/ that is the$ do not ha e the same sense/ as runs the standard argument. 1t ma$ #e o#ser ed in connection "ith (i) and (ii) that e%pressions such as - the 6orning 7tar7 and -the 8vening 7tar7 might #e regarded as falling some"here #et"een proper names and definite descriptions. 1n so far as the$ appro%imate proper names "e can 'uestion the assertion that the$ ha e sense/ for it is "idel$/ though not uni ersall$/ accepted that proper names do not ha e sense. On the other hand/ if the t"o e%pressions are treated li!e definite descriptions/ "hich differ in sense in a "a$ o# ious to an$ spea!er of ?nglish #$ irtue his or her !no"ledge of the language/ there is the pro#lem that0 (iii) &n #spE The 6orning 7tar is not a star (but a planet) This e%pression is not onl$ contradictor$/ #ut also potentiall$ informati e. Of course/ as a matter of historical fact/ it "as !no"n to the astronomers that neither the Morning <tar nor the ? ening <tar "ere actuall$ fi%ed stars/ #ut planets/ long #efore it "as disco ered that the t"o "ere identical. :onetheless/ the rather uncertain status of the t"o e%pressions ma!es them less than identical for the purpose for "hich the$ "ere used #$ Mrege. One might e en argue that the$ differ not onl$ in sense/ #ut also in reference/ the conditions under "hich the planet Cenus is isi#le from ?arth/ rather than its spatiotemporal continuit$/ #eing in this case more rele ant to the notion of referential identit$. )ut still Mrege8s e%ample has #een introduced simpl$ to illustrate in a general "a$ the nature of his distinction #et"een sense and reference. E"p $ssi&ns may !i,,$ in s$ns$) +/t *a'$ t*$ sam$ $,$ $nc$5 an! 8syn&nym&/s9 m$ans 8 *a'ing t*$ sam$ s$ns$9) n&t 8*a'ing t*$ sam$ $,$ $nc$9. & rather #etter e%ample than Mrege8s is >usserl8s/ - the victor at :ena7 and -the looser at Waterloo7/ #oth of "hich e%pressions ma$ #e used to refer to :apoleon. The criterion of sameness and difference is made more directl$ dependent upon the descripti e meaning of utterancesE t0& & m& $ $"p $ssi&ns 0i%% +$ !$,in$! t& *a'$ t*$ sam$ s$ns$ 3t& +$ syn&nym&/s4 &'$ a

c$ tain ang$ &, /tt$ anc$s i, an! &n%y i, t*$y a $ s/+stit/ta+%$ in t*$ /tt$ anc$s 0it*&/t a,,$cting t*$i !$sc ipti'$ m$aning. 1f the utterances are such that the$ ha e a determinate truth( alue/ constanc$ of descripti e meaning "ill guarantee constanc$ of truth( alue. The con erse/ ho"e er/ is not alidE for the su#stitution of one e%pression for another ma$ change the descripti e meaning of a statement "ithout there#$ altering the truth( alue. Mor e%ample/ let8s ta!e the sentences0 (a) :ohn is a fool (b) :ohn is a linguist 1f "e ta!e for granted that :ohn is #oth a fool and a linguist then the propositions e%pressed in statements made #$ uttering these sentences 3 ha e the same truth( alue/ #ut the$ do not ha e the same descripti e meaning. >o" do "e !no" that the$ differ in descripti e meaning4 .here the difference is as gross as this/ our intuiti e or pre(theoretical response to the 'uestion - 4oes (a) mean the same as (b)?5 is relia#le enoughE and it should not #e forgotten that part of "hat "e are doing in descripti e semantics is e%plaining such intuiti e Gudgments. ,inguists claim that t"o statements "ill #e descripti el$ e'ui alent (i.e. ha e the same descripti e meaning) if there is nothing that is entailed #$ the one that is not entailed #$ the other. :/in$ sa$s that0s$nt$nc$s a $ syn&nym&/s i, an! &n%y i, t*$i +ic&n!iti&na% (formed #$ Goining them #$ - if and onl$ if7) is ana%ytic.9 This formulation #rings out/ as it is intended to do so/ the interdefina#ilit$ of - s$non$mous7 and anal$tic7. (The 7aussurean $ichotomies) The definition of s$non$m$ can #e dra"n on the #asis of contrast and opposition #et"een t"o features. Mor e%ample/ the term - ant&nymy9 "as coined in the L1Lth centur$ to descri#e a phenomenon/ oppositeness in meaning/ "hich "as itself concei ed as #eing the opposite of s$non$m$ 3 i.e. sameness/ similarit$ in meaning. >$pon$m$ is defina#le in terms of unilateral implication/ Mor e%ample/ -crimson7 is a h$pon$m of -red7 and -#u$7 is a h$pon$m of -get7 ("hen the sentences are uttered to ma!e an assumption)0 (i) (ii) &n #spE &n #spE 7he was wearing a crimson $ress -; 7he was wearing a re$ $ress. % bought the car from :ohn -; % got the car from :ohn.

The definition of h$pon$m$ in terms of unilateral implication ena#les us to define s$non$m$ as bilateral or symmetrical hyponymy0 i, x is a *yp&nym &, y an! y is a *yp&nym &, x) t*$n x an! y a $ syn&nym&/s. 1f h$pon$m$ is defined as non(s$mmetrical (as it must #e if s$non$m$ is treated as s$mmetrical h$pon$m$)/ then proper h$pon$m$ ma$ #e distinguished from s$non$m$ as #eing as$mmetrical. C /is$ distinguishes #et"een three #asic t$pes of s$non$m$0 a+s&%/t$) p &p&siti&na% an! n$a 2 syn&nymy. Cruise accepts the interpretation of s$non$m$ as sameness in meaning/ #ut rather/ he prefers to ie" s$non$ms as "ords "hose semantic similarities are more salient than their differences/ "hich opens a potential area of interest. (1) A+s&%/t$ syn&nymy &#solute s$non$m$ refers to complete identit$ of meaning/ and so for the notion to ha e an$ content "e must specif$ "hat is to count as meaning. >ere a conte%tual approach is adopted/ according to "hich is an$thing "hich affects the conte%tual normalit$ in grammaticall$ "ell(formed sentential conte%ts. On this #asis/ a#solute s$non$ms can #e defined as it$ms) 0*ic* a $ $./in& ma% in a%% c&nt$"ts) that is to sa$/ for t"o le%ical items ; and Y/ if the$ are to #e recogni5ed as a#solute s$non$ms/ in an$ conte%t in "hich ; is full$ normal/ Y is tooE and in an$ conte%t in "hich ; is totall$ anomalous/ the same is true of Y. This is a er$ se ere re'uirement and actuall$ er$ fe" pairs 'ualif$ (if an$). The follo"ing e%amples illustrate the difficult$ of finding uncontro ersial pairs of a#solute s$non$ms (-K7 indicates -relati el$ more normal7 and -(- indicates -relati el$ less normal7)0 (i) &n #spE brave v(s courageous

,ittle )ill$ "as so #ra e at the dentist8s this morning. (K) ,ittle )ill$ "as so courageous at the dentist8s this morning. (() (ii) calm v(s placi$

<he "as 'uite calm a fe" minutes ago. (K) <he "as 'uite placid a fe" minutes ago.(() (iii) big v(s large

>e8s a #ig #a#$/ isn8t he4 (K)>e8s a large #a#$/ isn8t he4 (() (i ) almost v(s nearly kick the bucket <he loo!s almost Chinese. (K) <he loo!s nearl$ Chinese. (() #uc!et(() E &n# spE 3nbsp2 23nb sp2(v) $ie v(s &pparentl$ he died in considera#le pain. (K) &n#spE E&n# spE&pparentl$ he !ic!ed the

<till/ there are items that fit in the definition of a#solute s$non$m$0 sofa" settee 3 pullover" sweater. >o"e er/ e en for these items/ in a t$pical class of students/ a si5ea#le minorit$ "ill find conte%ts for them as discriminator$. One thing is clear/ under this description a#solute s$non$ms are anishingl$ rare/ and do not form a significant feature of natural oca#ularies. The usefulness of this notion lies in its status as a reference point on a putati e scale of s$non$mit$. (F) P &p&siti&na% syn&nymy Bropositional s$non$m$ can #e defined/ as its name suggests/ in terms of entailment. 1f t"o le%ical items are propositional s$non$ms/ then the$ can #e su#stituted in an$ e%pression "ith truth(conditional properties "ithout effect on those properties. To put it in other "ords/ t"o sentences "hich differ onl$ in that one has one mem#er of a pair of propositional s$non$ms "here the other has the other mem#er of the pair are mutuall$ entailing0 :ohn bought a violin entails and is entailed #$ :ohn bought a fi$$le2 % hear$ him tuning his fi$$le entails and is entailed #$ % hear$ him tuning his violin2 7he is going to play a violin concerto entails and is entailed #$ 7he is going to play a fi$$le concerto though fi$$le is less normal in the last e%ample. 9ifferences in meanings of propositional s$non$ms/ #$ definition/ in ol e one or more aspects of non( propositional meaning/ the most important #eing0 (i) (ii) (iii) &n #spE 4ifferences in expressive meaning &n #spE 4ifferences on stylistic level &n #spE 4ifferences of presuppose$ fiel$ of $iscourse

Psuall$/ more than one of these comes to pla$ at an$ one time. 1n the case of violin" fi$$le the difference depends on certain characteristics of the spea!er. 1f the spea!er is not a musician/ or an outsider to iolinistic culture/ then fi$$le is more collo'uial. 1f the spea!er is a professional iolinist tal!ing to a colleague of his/ then fi$$le "ill #e the more neutral term "hile violin is used mainl$ to outsiders. 1f "e ta!e the pair shin" fibula/ the difference is almost one of field of discourse0 shin is the e er$da$ term/ "ith no special or e%pressi e loading/ "hile fibula is used in the field of medicine. Bropositional s$non$ms seem to #e commonest in the areas of special emoti e significance/ especiall$ ta#oo areas/ "here a finel$ graded set of terms is often a aila#le occup$ing different points on the euphemism 3 d$sphemism scale. The$ also seem to #e pre alent in connection "ith concepts that are applica#le in distinct conte%ts/ "ith differing significance and implication in those conte%ts. (3) N$a syn&nymy There is a clear #orderline #et"een propositional and near 3 s$non$m$/ "hich is not the case "ith the #orderline #et"een the near ( s$non$m$ and non(s$non$m$ is much less straightfor"ard and it is not o# ious "hat principle underlies the distinction. T"o points should #e made at the outset0 (i) <anguage users $o have intuitions as to which pairs of wor$s are synonyms an$ which are not.

(ii) %t is not a$e/uate to say simply that there is a scale of semantic $istance an$ that synonyms are wor$s whose meanings are relatively close. (This "ould e%plain the some"hat uncertain #oundar$ of near( s$non$m$0 people are t$picall$ ague as to "hat constitutes/ sa$/ an ol$ woman or a tall man.) The reason this is not ade'uate is that there is no simple correlation #et"een semantic closeness and degree of s$non$m$. The follo"ing items are semanticall$ closer as "e go do"n the list/ #ut the$ do not #ecome more s$non$mous0 -; entity -; living thing -; animal -; animal -; $og -; spaniel etc. 3nbsp2-; process 3nbsp2-; ob=ect 3nbsp2-; plant 3nbsp2-; bir$ 3nbsp2 2-; cat 3nbsp2-; poo$le

This list ma$ e entuall$ continue indefinitel$ "ithout e er producing s$non$ms. The point is that these "ords function primaril$ to contrast "ith other "ords at the same hierarchical le el. 1n other "ords a maGor function of $og is to indicate - not cat(mouse(camel(etc.5 that is to signal a contrast. <$non$ms/ on the other hand/ do not function primaril$ to contrast "ith one another. :e ertheless/ the$ ma$ contrast in certain conte%ts/ and this is speciall$ true of near 3 s$non$ms0 'e was kille$ but % assure you that he was not mur$ere$ ma$am.5 Characterising the sorts of difference "hich do not destro$ s$non$m$ in no eas$ matter. &s a rough and read$/ #ut still not er$ e%plicit/ generali5ation it ma$ #e said that permissi#le differences #et"een near( s$non$ms must #e either minor/ or #ac!grounded/ or #oth. &mong minor differences ma$ #e counted the follo"ing0 (i) &n #spE A!<ac$nt p&siti&n &n t*$ sca%$ &, -degree70 fog " mist laugh" chuckle hot2 scorching big" huge $isaster" catastrophe pull" heave weep" sob etc. (ii) &n #spE C$ tain a!'$ +ia% sp$cia%i6ati&ns &, '$ #s0 amble" stroll chuckle" giggle $rink" /uaff2 (iii) &n #spE Asp$ct/a% !istincti&ns0 calm" placi$ (state 2s disposition)E (i ) &n #spE 1i,,$ $nc$ &, p &t&typ$ c$nt$ 0 brave (protot$picall$ ph$sical)0 courageous (protot$picall$ in ol es intellectual and moral factors). &n e%ample of a #ac!ground maGor distinction "ould #e pretty (9female7 presupposed) 2s han$some (-male7 presupposed)/ the propositional meaning of #oth of "hich ma$ #e glossed as -good 3loo!ing7. .hen the gender distinction is foregrounded/ as in man" woman the resulting terms are not s$non$mous. <a$ing "h$ "e get near 3 s$non$ms in a particular instance/ rather than full$ contrasti e terms/ is also difficult. & possi#ilit$ is that contrasti e terms appear "hen the conceptual differences ha e concrete #eha ioural conse'uences/ as in technical and -e%pert7 fields. Much research remains to #e done in the field of s$non$m$ and sense relations. 1i,,$ $nc$s =$t0$$n Syn&nyms

1n contemporar$ linguistics it has #ecome almost a%iomatic that complete s$non$m$ does not e%ist. -.ords7/ 9. +ones once remar!ed/ -are seldom e%actl$ s$non$mous.7 Macaula$ has e%pressed the same idea in terms of "hich "ill commend themsel es to the modern linguistics0 -Change the structure of a sentenceE su#stitute one s$non$m for anotherE and the "hole effect is destro$ed.7 1n the "ords of )loomfield/ - each linguistic form has a constant and specific meaning. 1f the forms are phonemicall$ different "e suppose that their meanings are also different "e suppose in short that there are no actual s$non$ms.7 :e ertheless/ there are cases "hen se eral s$non$ms arise around a ne" in ention/ technical or medical/ or scientific term/ until the$ are e entuall$ sorted out. <uch s$non$m$ persists for indefinite period and there is no actual difference #et"een the "ords #ut in terms of register. Mor instance/ in medicine cranium and scull are s$non$mous/ in phonetics consonants li!e s and > are !no"n as spirants and as fricatives and so on. 1n ordinar$ language/ one can rarel$ #e so positi e a#out identit$ of meaning/ since that matter is complicated #$ agueness/ am#iguities/ emoti e o ertones and e ocati e effectsE #ut e en there one can occasionall$ find "ords "hich are for all intents and purposes interchangea#le. Mor e%ample almost and nearly are such integral s$non$ms.

Brof. .. ?. Collinson has made an interesting attempt at ta#ulating the most t$pical differences #et"een s$non$ms. >e distinguishes #2n nine possi#ilities0 1. One term is more general than the other0 refuse re=ect F. One term is more intense than another0 repu$iate refuse 3. One term is more emoti e than the other0 re=ect $ecline c. One term ma$ impl$ moral appro#ation or censure "here another is neutral0 thrifty economical `. One term is more professional than the other0 $ecease $eath H. One term is more literar$ (poetic/ archaic/ or other) than another0 passing $eath *. One term is more collo'uial (familiar or slang$ speech) than another0 turn $own refuse =. One term is more local or dialectical than another0 7cots" flesher butcher 9. One of the s$non$ms #elongs to child 3 tal!0 $a$$y father 1f "e loo! more closel$ at these series "e8ll notice that all of them fall into se eral distinct groups. Mor instance/ (1) refers to o#Gecti e differences #2n s$non$msE (F) com#ines o#Gecti e and emoti e factorsE (3) & (c) are emoti e/ "hile (`)/ (H)/ & (*) in ol e e ocati e effects/ "hich are a special t$pe of emoti e meaning. (=) & (9) actuall$ stand apart from the rest since dialect and child 3 tal! are outside or at #est on the fringes of <tandard ?nglish. The #est method for the -delimitation of s$non$ms7 is the su#stitution test of Macaula$. 1t re eals to "hat e%tent are the s$non$ms interchangea#le. 1f the difference is predominantl$ o#Gecti e/ one "ill often find a certain o erlap in meaning0 the terms in ol ed ma$ #e interchanged in some conte%ts #ut not in others. Thus broa$ and wi$e are s$non$mous in some of their uses0 - the broa$est sense7 of a "ord N ]the wi$est sense7. 1n other conte%ts onl$ one of the t"o terms can #e used0 "e sa$ - five foot wi$e7/ not7broa$7/ and a -broa$ accent7/ not a wi$e one etc. 1f/ on the other hand/ the difference is mainl$ emoti e or st$listic/ there ma$ #e no o erlap at all/ ho"e er close in o#Gecti e meaning/ the$ ma$ #elong to totall$ different registers or le els of st$le and cannot normall$ #e interchanged. 1t is difficult to imagine an$ conte%t/ e%cept a comical or ironical one/ "here mingy could replace avaricious or "here pop off "ill #e su#stituted for pass away. One can also distinguish #et"een s$non$ms #$ finding their anton$ms. Mor e%ample/ the er# $ecline is more or less s$non$mous "ith re=ect "hen it means the opposite of accept/ #ut not "hen it is opposed to rise. 4eep "ill o erlap "ith profoun$ in -$eep sympathy -/ "here its anton$m "ould #e superficial/ #ut not in -$eep water7 "here its anton$m "ould #e -shallow7. &nother "a$ of differentiating #2n s$non$ms is to arrange them into series "here their distincti e meanings and o ertones "ill stand out #$ contrast/ as for instance the arious adGecti es denoting s"iftness0 /uick swift fast nimble fleet rapi$ spee$y. Syn&nymyc Patt$ ns

The s$non$m$c resources of a language tend to form certain characteristic and fairl$ consistent patterns. 1n ?nglish s$non$ms are organi5ed according to t"o #asic principles/ one of them in ol ing a dou#le/ the other a triple scale. The dou#le scale/ also called -<a%on7 ersus -,atin7/ is "ell !no"n. 1n ?nglish there are countless pairs of s$non$ms "here a nati e term is opposed to one #orro"ed from Mrench/ ,atin or 6ree!. 1n most cases/ the nati e "ord is more spontaneous/ more informal and unpretentious/ "hereas the foreign one has a learned/ a#stract/ or e en a#struse air. There ma$ also #e emoti e differences0 the -<a%on7 term is apt to #e "armer and homelier than its foreign counterpart. Bhoneticall$ too/ the latter "ill sometimes ha e an alien/ unassimilated appearanceE it "ill also tend to #e longer than the nati e "ord "hich has #een su#Gected to the erosi e effect of sound(change through time. There are man$ e%ceptions to this pattern $et it occurs so persistentl$ that it is o# iousl$ fundamental to the structure of language. 1t ma$ #e noted that the term -nati e7 in narro"l$ et$mological sense #ecause it ma$ include "ords of foreign origin that ha e #ecome thoroughl$ &nglici5ed in form as "ell as in meaning. 1t "ill #e sufficient to 'uote a fe" e%amples of this s$non$m$c pattern in "hich all maGor parts of speech are in ol ed in the process0 a!<$cti'$s# >n+sp5?n&/ns# bo$ily ( corporeal 2 >n+sp5 answer ( reply ? '$ +s# 2 5 3nb sp2 fi$$le ( violin

brotherly ( fraternal heavenly ( celestial actor sharp( acute 2

buy ( purchase 3nbsp2 rea$ ( peruse tire(weary) ( fatigue

3nbsp2 2 2

help ( ai$ player (

3nb sp2

worl$ ( universe

1n a fe" cases/ these s$non$m$c alues are re ersed and the nati e term is rarer or more literar$ than the foreign# dale - valley; deed - action; foe enemy; meed - reward; to heed - to take notice of . 2 3nb sp2 The pattern ena#les one to a oid nati e (or 'uasi( nati e) terms "hich for one or another reason ha e #ecome tainted and might call up undesira#le associations # ? bloody sanguinary; blooming flourishing; devilish diabolical; hell inferno; popish papal. <ide #$ side this main pattern there e%ists in ?nglish a su#sidiar$ one #ased on a triple scale of s$non$ms0 nati'$) F $nc*) an! Latin & G $$(# ; nati'$ begin en$ foo$ kingly rise time ? F $nc* >n+sp5? Latin - G $$(

3nbsp2commence 3nbsp2initiate 3nbsp2finish 3nbsp2 2conclu$e 3nbsp2nourishment 3nbsp2nutrition 3nbsp2royal 3nbsp2 2regal 3nbsp2mount 3nbsp2 2ascen$ 3nbsp2age 3nbsp2 2epoch 3nbsp2 2 1n most of these com#inations/ the nati e s$non$m is the simplest and most ordinar$ of the three terms/ the ,atin or 6ree! one is learned/ a#stract/ "ith an air of cold and impersonal precision/ "hereas the Mrench one stands #et"een the t"o e%tremes. 1t should #e noted/ ho"e er/ that here too there is an occasional re ersal of alues0 in the series0 kingly royal regal/ for instance/ the Mrench term is the ordinar$ one/ "hile the nati e kingly is comparati el$ rare and most at home in literar$ conte%ts li!e Mar! &nton$8s - 1 thrice present him "ith a kingly cro"n.7 & third t$pe of s$non$m$c pattern arises "hen t"o or more s$non$ms de elop on parallel lines. <ince "ords "ith similar meanings are closel$ associated "ith each other/ a change in one of them ma$ set off an analogous change in another or in se eral others. .hen/ for e%ample/ in earl$ Modern ?nglish/ the er# overlook ac'uired the meaning of 9 $eceive5 its s$non$m oversee/ soon #egan to #e used in the same transferred sense. <imilarl$/ "hen cram #egan to mean -deceive7/ its s$non$m/ stuff under"ent the same change of meaning. Syn&nymy an! Sty%$

1n his Rhetoric &ristotle made an interesting remar! on the difference #2n s$non$m$ and am#iguit$. <$non$ms/ according to him/ are - useful to the poet7 "hereas - "ords of am#iguous meaning are chiefl$ useful to ena#le the sophist to mislead his hearers7. <$non$m$ is indeed an in alua#le st$listic resource not onl$ to the poet #ut also to an$ "riter and it lends itself to a ariet$ of uses. .e distinguish #2n F #road categories according to "hether the spea!er has to choose #2n s$non$ms or prefers to com#ine them for some specific purpose0 c*&ic$ +-n syn&nyms (emotion/ emphasis)E c&m+inati&ns &, syn&nyms including 'a iati&n (to a oid repetition) and c&%%&cati&n (to pro ide an outlet of strong emotionsE to ma!e one8s meaning clearer and more emphatic/ help to produce a contrast effect/ etc.)
OBBO<1T1O:< Opp&siti&ns a $ +as$! &n dichotomisation. =ina ity ant&nymy mani,$sts =ina ity. L$"ica% &pp&siti&ns# g a!a+%$ '-s /ng a!a+%$ &pp&sit$s. @4 >n +sp5 G a!a+%$ &pp&siti&ns "e ha e some !ind of comparison and "e compare and contrast #2n F things0 hot" col$. A4 >n +sp5 Ung a!a+%$ &pp&siti&ns# male" female -; L cannot #e more female than D (Othis results in different ungrada#le oppositions/ "hen implied as F 3 the predication of 1 of them implies the predication of the other.

- is female implies that - is not male2 - is not male - is female The proposition2 predication of one of the opposite implies the negati e implication of the other. The re erse is also true. .ith grada#le oppositions there is difference the predication of one of the opposites implies the predication of the negation of the other. E"# The coffee is hot the coffee is not col$. The coffee is not hot does not mean that 3 the coffee is col$. There is a similar distinction #2n c&nt a!ict& i$s (ungrada#le oppositions) > c&nt a i$s (grada#le oppositions). ,et8s ta!e the t"o propositions P > : the$ cannot #e #oth true and false. .e cannot sa$ this is neither a female nor a male cat. .hile "ith contraries the$ can #e #oth false0 The coffee is neither hot nor col$. The distinction is e'ual to that of grada#le and ungrada#le oppositions. Tree -; $og -; not dichotomousl$ opposed to each other. 6rading is made e%plicit in different "a$s. C&mpa is&n can #e e%pressed #$ an adG in comparati e degree (more an$ less ( same an$ $ifferent ( $iffers e/uals) this t$pe of grammaticalisation of comparison is different in the different languages. E"# ?ur house is bigger than yours .our house is smaller than mine( ours. 1n (1) "e su#stitute one mem#er for another and produce an e'ui alent sentence. The use of grada#le oppositions implies comparison. ?ur house is big "e do not state e%plicitl$ "hat "e compare our house to. There should #e a generall$ stated norm to "hich to compare it 3 a generall$ accepted norm. .e compare the house "ith the general norm. .e thin!0 ?ur house is bigger than the generally accepte$ norm. The pro#lem of Blato0 ho" a single o#Gect can com#ine different oppositional 'ualities4 This is a small elephant it is a small elephant - the standard norm for comparison is not clearl$ stated "hich misleads us. 6rading ma$ also #e semi 3 e%plicit0 ?ur hope is bigger -; "ithout e%plicitl$ including the standard f comparison 3 modification is needed. 1i,,$ $nc$ +-n m& p*&%&gica%%y $%at$! an! m& p*&%&gica%%y /n $%at$! &pp&sit$s# @4 >n +sp5 M& p*&%&gica%%y /n $%at$!# goo$ ba$2 high low2 ol$ young2 -;the$ are more fre'uent in language. 2) n bsp; M& p*&%&gica%%y $%at$!# frien$ly unfrien$ly -; "e get these #$ attaching to the #ase form the pre(fi% un-, il-, im- and so "e get the other opposite. The #ase form is the (K ). T*$ m& p*&%&gica%%y /n $%at$! g a!a+%$ &n#spEE&n# spE&n#spE E&n#spE&n#spEThis is not al"a$s so. T*$ m& p*&%&gica%%y $%at$! /ng a!a+%$ E".# marrie$ single ( -unmarrie$ unrelated ungrada#leE ;frien$ly @) hostile grada#le 3nbsp2 A) unfrien$ly ungrada#leE One and the same "ord can ha e different opposites in morphological terms. Morphologicall$ unrelated opposites goo$ ba$ can #e differentiated in terms of semantic and s$ntactic polarit$.

Ma%(i$%# tal!s of he so(called irre ersi#le #inominals 3 "hat "e get "hen coordinating F opposites. >e notices some hierarch$ of semantic opposites0 goo$ ba$2 high low irre ersi#leE (#) la$ies 3 gentlemen (-)2 (#) men 3 women (-). )ut if "e as! ho" lo" or ho" #ad is it it could onl$ #e #ad. There is no logical necessit$ to ha e morphologicall$ negati e oppositions/ "e can su#stitute -#ad7 "ith -ungood7 #ut the fact that "e ha e a goo$ ba$ opposition leads us to the conclusion that human #eings need to characteri5e things in dichotomous contrast %$"ica% p&%a ity. .e use different "ords t*$ %$"ica% !isp&%a ity not morphologicall$ related enhances the difference #2n the different "ords. Pngrada#le oppositions can sometimes #e e%plicitl$ graded0 - is more of a bachelor than .. .hen "e reGect or den$ to accept them as e%cluding each other0 male" female -; L is more male .e grade e%plicitl$ opposites 3 "e do not 'uestion their ungrada#ilit$ #ut arious secondar$ connotations and implications.. Li&ns# c&nt ast &pp&siti&n ant&nymy c&mp%$m$nta ity. @4 A4 B4 C4 C&nt ast# no implication as to the num#er of terms in ol ed in the opposition. Opp&siti&n# dichotom$ or #inar$ contrast. Ant&nymy# opposition of grada#le le%emes. C&mp%$m$nta ity# opposition of ungrada#le le%emes.

T /+$t6(&i# p i'ati'$ an! $./ip&%%$nt &pp&siti&ns. @4 >n +sp5 P i'ati'$ &pp&siti&n# "e ha e t"o mem#ers and one of them is characteri5ed #$ a propert$ and the second #$ the lac! of this propert$0 animate" inanimate. A4 >n +sp5 E./ip&%%$nt &pp&siti&n# t"o e'ual mem#ers0 male" female. Li&ns# %$"ica% &pp&siti&ns &, c&n'$ s$n$ss# - is .,s wife & . is -,s husban$ -; con erse to each other. ,ions tal!s a#out three t$pes of oppositions0 ant&nymy) c&mp%$m$nt) c&n'$ s$n$ss. The cth t$pe he calls !i $cti&na% &pp&siti&ns 2? up $own2 come go2 arrive $epart. .e ha e some place and a mo ement a"a$ or to"ards it. O t*&g&na% an! antip&!a% &pp&siti&ns# >n+sp5 :orth antipodal opposition. orthogonal. ?ast &n#spE <outh N&n +ina y c&nt asts occur in sets "ith more than t"o opposing mem#ers. The sense relations are ones of inc&mpati+i%ity. - went on 7un$ay <unda$ is opposed to an$ other da$ of the "ee! as "ell as "ith an$ other mem#er of the opposition. S$ ia% an! cyc%ic a ang$m$nt. .ith the serial arrangement "e ha e t"o outermost mem#ers scales and ran!s/ "hile "ith the c$clic arrangement there is no end of the c$cle. S$ ia% a ang$m$nt# i4 >n +sp5 sca%$# excellent goo$ fair poor ba$ atrocious -; "e ha e a pair of anton$ms. ii4 an(# consists of mem#ers incompati#le and restricted ungrada#le terms0 fiel$ marshal general- corporal private. Ran!s are more strict than scales. .est The antipodal or diametrical opposition is more fre'uent. &n#spEE&n# spEnorth and south are in &n#spEE&n# spE&n#spE E:orth/ east/ and "est 3 in

The mem#ers in a set can #e arranged seriall$ and c$clicall$ (arranged successi el$)0 winter-spring summer autumn 6on$ay Tues$ay - ! - 7un$ay2 :anuary - ! 4ecember. Within one and the same lexical field we can have 2 arrangements- scale and cycle as with colours. and a more general/ or superordinate/ le%eme/ as e%emplified #$ such pairs as# winter-spring summer autumn, Monday Tuesday - - Sunday; January - December.

You might also like