Sunteți pe pagina 1din 33

9

Structural semantics II: sense relations



g.l. Opp(Jsll~'I'J'" ,ami ,u:mbas:t

The nGtlon ~f silllSe'- (ilS di.stinct from dcm.otatitJ:n'" and T\efel"Cnce~ bas, meaJiy been introduced (chapter 7)." Our purpose in thiis chapter is to dC\'elop and reformulate what seem to be 'the basi{; prip&ivl.es of the

-theo!)' of semantic fields in t'CT'm ' ofsense-rcJations· (i.e.llelations of s,e'l'Ite l!,o~ding with:in sets of texemes) wi:thout P05:tu] atin!!:" my underlyi.n,g conceptual Of pe]'ll:::€p,tu:a.1 ~ub~ta_ncc I(rf. 8.4,). Th,e tJreMment wiU be rdatively 1il(fimil:d and at tiraes s.omewlMt silPc~ialti,~·c. ,iVe bt"g,in b}f' ~;i,i~au.ssing the aotion of amdi' m. ic 0' fillion.

. From its: 'lie be inrullgsstructural s,rlRanl'[cs, (and indeed structural lin l!Jistics in ,~n,er!l[', has emehasiZl~d the imp.ortance or relations of

-pa.rndi 'il~C oI!PositiQn·. 'T',ier himse.1f opens his major work (193 [) widll the IE-hall~nging statement, that eye!:)' \\"IOrd thaI: ispron ]['I~e..d

~foorth its o~poo~tejs.ejIl!f[l GegenteiJ} in 'th~ ecaseieusness of !h~ S ,ciker ant:! hearer: and thi s statement can be matdted with simi I;,u~9(:rtions,' • 'dl~t:rl.!~ural ~'~1ru)'1i\ticists. T:ri$,. i.t wiU be nOi~ell" cleims, .ll.l".,tJlers. hl!\/'e dOrJJI:1 thait dte opp{lsi~c is in !_0l_!lC: ;'~'31.pre$e .!!In ~lte mi nd 1J,r the in:ab-r and hCaII'cr dUII'Lng_ QJIl, Old of \I nierl'l.r.II;~. \'l/h'ether thicS is tma 'or milt j'S·.;1 'P'Syd!J.1:ilo'g!cal qUll'5t ion, aJI'Id one I!.hat is m.ol'1: re.le\tant to the COfi!ltrudlUfi of 3. dU'tlI'Y of Ial'lguage~l:lenllviol]r than it is ro the armdysLs of a l.:l1'Igungt-systcm {cr. 1.6). In l\'halt [oI1o\\'s we make :nil assun ptions abmit wha;t goes on in the mind of (he '&"peaker:md hearer durin,g 3:1'1 utterance, Trier's statement also lIfl:p-ea.n;. to imp']y tha.t C\'cry word in the yo cab ulilrJ h~!j an. opposite, and only (!IJ1eoPPQ!;ite. \\rhcther' this is t:;ue, 01!' n-ltt, is !! question with wh~ch we shall be concerned in this section.

.. The staIl,diarnl ,technical term fm" oppos~tencss of mCIUli;ng ]H:'t"llVcen

I I\,'[uch 0-( l;\'h~t '!I,flpe~ iTS he~ ~" ~h;'lph:r I) is ,Em ICl'q:JBnB:t[K[l I'JIf ch~:pter ~ Q !If Lynna (f t)68). 111m! Tll!ader is r<emimll:d. boW\wer, that. there :b~ eerta til, k~~ mimwklimali diffrc:unccs • .11'1 ~a'rtic~l~r. 'the term '~e:i'lgt~til1n:' Wl'I~ :Qat used ill Lyons. (l96Hj)<.

Lexemes is antonymy·'. But this is hardly more precise jn the usage of :;;ost authors tJilall! the word "~n'li'>I'i::>it_mCli15' which ir replaces, and!

~U'~ Il'EF

dichonllri,cs will elassi £y as. a~tOJlilms." Rll.irs of fexeme~, W~hllCh, as we

shall seC', ilf\C re:]ated in IlIraricty of 'ways eh~gh I ;~low". 'buy' ~ "sell",

• mille • ; ~ fcmale ", • mrrivc';' dC[lll.rt·., 'Ien:':· right". ~ front': t back'. etc.) .. 'Vbjll't aU th.es-e examples have i 1'1 CQrnmOflL it shol.:Il.!l be noted, ~9, thei r ut:pt:nd'l':rto,!,_ ~m'l di,~h~t[3'm'~1.at ion .l<\',e ~n leave to others to cl'lq'lII,i[~ whether the tendency: to dunk in urplJi!li'~es'l to ca.te rise 'e)! erienee in terms of bi.nary COlltr.aslS, is II universal hum~tenicm;:y whiC:h is but secondarHy reflected in 1.mg>uJlgc. as Cause producing ,effect" or whethef

it is the pre-em stence '1)[ II [a,r· It: number {l f _il'ppfised pairs, of lexemea ill:

our native 1:111' _Hle whil;h causes I.IS to d[,,,hatomizc, or larize, our jUdgement;""and experienees, h ill, however, 3. fact, (If whicb thcli~gu7sl

must. take cogni;;:am;:c,. that biD!! 0 osidon iii one (lithe mo-st impart!lnt - F I princi .-In ,Q'lfcfnunJ the structure of ial}IDJ,ages; and the most evident U1llru£'t:'St:1llti.on of this princi p~e. ~I$ far' a'51 the vecabu laTj" is. concerned,

is .antonym,..

But lexic:aI of'p.G!sites, as we have atTeady ~aid! are of se,\I'ctal .,l'iKctent kinds; and j't [10 -3, moot point just how rna.li}' dichotomous relations ~Ld be held to fall within the scope. of I anton~my '. L¢t us begin by drawing a dl~t.inct~oft bctweena4~~J ungrad;,\l:!l"" opposites. (cf" Sapir; 19#)' Cradin. '" im.'iOt ... Cf> OOIl!lar-ii!lCl,iii.When we OOlmpafl'e

two or moee objects, with respect to tAciT peesession of a certain pro-

pt:ny (this prOpCTty bring denoted typiroIly in Elilgli_sh by an Ildjc:cri!Sl" ...

it is usually, though not ~ln"ays" appropriate toenquire wh.ether they

]unrc tillS prop~irt r to the saIne degree 'QLnot. Fer exalnplc, we ,might

ask /, X <'u hat ;u "F? Th.e ('Itt tut we C_'llli ~~r X is as hot airY 01 X,:rIwitt'/' tlum Y depends upon the grnd:abiti'll' of ~]m~·. A lexeme like "fe-male' (un~ikc! feminlne '),.on t.he .odu~r hand., is un.gradable: we 'would

rIOt normally say X is tlI,/t!:l1IaJc (1.$ l' or X i~ '"((1Ft? ftm(11~ than Y (though

X is 'm~ u.s femim'm:(Js Y is a :pcncctly aoceptahle wterance). Each of' th~s:~ lC'~mcs iis 'l!llU:iI?d in thc:_ 'I,.'ocabuIM}· wiith. ",iha! 'wouM gen.cral.ly be de,..

scribed as j is o· osite: [cold' am] • nude '. rcspeetively. Now til' f:'ail:t

that [hot' a;i·cord' are grndable lexemes, whereas 'feml'de' and ·Ula1le • I

arc ungradable, is bound 1:],P with an hnporta.nt logical dillerenoe'-=-between the t"l'O pelrs,

Ufljmdable Q·pposites, 1I,1ll~n th~li' a~'C emrJoy:ed ~_EroediJ~tiv;c ex'P_l'essions, divide t.he unlversc~of-di9C(lll]~ (i.~" the '0 bjccts of \\'hich theyil:l'cpreclkable: d.1f!.3) into two s.Ql!!pleme_MOl.O'· .subcsets. It {oHclnvS from this., not o,nly that the' redicaeion 'Of eithe., one !Jl! the..,pai.: implies

<:.) 1.. • ~ .; . • C, .!.... .J r .,_,., ..... .. ,
,
r t'" 1 t. O~1-- (_ .., ~-. Ci~~
f+ "=- 'J _ ~l~ ~ ..... ~
(-'(.1 , (?:J A
... ~ ...., Fo- ~1:u:.JJ:[ooieat ion (If the negat~on (If lh£ oth~r: '. but niso that the procdiCl!tif!l1:\

(!if th.c ne<g'adol1! of either [mp~ilJs ~he ]:)r~t1f~~i~n of the other .... o~t:(ixamplc, the pfo'p®sitloll !' Xis female,' i m pliti! "'}["I's ~ot malt: t' : ) arid " X is not female ,j '(pmvi.ded that' !m:j1e ~ and 'female' are rredicab:le

(If X) i.m:.plics "X .us male".

'~ili._gp'I,dabtc ...:~n:u;msitcs, however ,rhe sit u.at·~o,n is Jiffct~trl·. The predietl!tlolt of the one implies the [lre~fic-;:tionor the Vile~tion of the' other: the ~roposrd.on uX. is hot·'- i.m pli es, " JL is not oom!~; !lll~ fif)[ Is cold" i mp~i,e,s "N; is no~ hot", But "X is nat h~it" J,@~ not goneraUy imply <OX .is ootd" (though 0[1 oarosions it nlaY be :~n~erpre'~ed ~n d'lis

~va.J!. and wc\¥lH ool:fi~ back to thLs poiru::),.

V\that ha:s- lJeCli1l :said! so far ~b~ut the distllil,ctiull.(lf ulilJ.~rn.da:Me and gra:clable ~_!it<~!.mn;g~ ;!in: fi~t s:ight, i'l2F ~al:'to b!;::. cQ1iCredslIitisfac-

tori I:; enough. by the 'lfilid!ti.onru IQg~ldjstinctiol'll of cO.IJJtfad.ictm~es· I- be; r",+ ,and cont(l1£ie;&fi. A p~po3itio'~,p ~~. thee~torY' ,of :inother pr"Qposition (jj.i f p' :md ,~ ca!'lnot both be trueer both Jalse: e.g. "This~5 a .male cat "; "This is a ~e:mA~C c!):C' ·(iIi:>. well ;I!'> sueh cQw:'5~ndi.ng a:ffi:m'ii.1Jtlve 1I.n(l nt:gatiw~ pro:posidon;1\ as "The ~jJee is ~Q]d ": "The ¢Offcc ls not oaM "} .. Il proposition p ~s, the.oontTd.ry1 ~f a:noth£r P-ITl,Pfb"'i.- c u tlen '1, ifcP and q cannot bo~h be Hue (though both m~.y be f8Jbe}. ,~.g"

j "The coffee<is. bot": "The coffee is. oold'" (as W(!;U a.s.suefu pairs as "AU fllCR :I:l.""C bald.":" No men .arc I:ml.d '")Q)\pplyi ng thi 5 disti nelio'it _to_lb.t:~tcJ1cca which c~;r.c~s sUC!..p_f!lH~sition~ aod 11'hcn ~cTl~~~\I~rYi in an

obvious '?I'll}',~ le~~,e~e~ 3Sl!fed i.cati'!.'e Cl;::pl'~i~~jJl.l.llem, w~ 0011](1 SllJr that .~ rnallf' and • :fie~~k: J ~r-:e op'l1ltrncl.ictodes Janli.duh.:I· 'hot' and "'oo~.d' areoo[lIraries; :!.lrn.d this. is O®rtoN:t. n~t ~hctc are man>, CmiiJ:~rn- ,4

'lies that w'OuJd not g.c:nc.ni!!Y__ be re~rd ~da.iLO,Pposi res (~g.,. J:.ru..i::

"iiiU'e', not lOlIli:~mioi'l. innumerable of her pairs $lId~ as 'tree':' clog",

J ~ squm·: ;. abstrai:f' ,. tl'l:c.):t!_sv ilIr"C not dtd~ot~mou~~l)o~ed ti.~ one.

~ . ~bc .d.ist~f:'Icdon of eorltradicwl'ies anA ,oontrari,~s ~n~sp:OlIldst .. ~) to t!"!~ ~~lnGl:loQriL of ungf~@able ar\d_g~.d:a:blo lexemcs wuhm~h.::. da_~s

of lexical <l'P posicc.s in a.language, out it ~I!I ues m.orewi.t.ieb' ~ and 'lhc

fact tbllit grndab]c anwnyms cai;gc[lcr:liHy be, t~kcn.~.s tlU![ltrll.TL~S, rnthci'

J thaI;! cOl1!trn.dilCtoriC~, i:S~l1!sequcm:::c Qf g!ad=1bility 1'L'liI:t: itg,-ca,~se ..

~ ::lThis st'iltem~~iL~~e~ ~~ n.{)tin.~nderl to cio n'!>lN thml £"ii\L 1 .mft 'Gne imp0rtan:t lind .. l'iele'fallt dcii1cr'i::l'lc-!C b~ t\~"CCIiI. ~o""tl'~~rc!i a[lQ O~tn'r~" I dlct@['J~~. As it ~tElntl!w "'"Q~rd3iluw a.s ~Vi'l!~['i:lT~~S. !;udii""ji;i:fS ~~ ., Th"l'!i c;}l'~

\ h;;:rc: is ~." atld "1'h.erl!J I~ no ~o:t:r'C~ 'ti!~r~ " : lliiJd nil,t; c:.,,~ ryjjjn~ would \/i·ish. to

t .,. ~.IOCl~pt th.L~ Q(}lli!eq!.l,gne~.

! ~ ~)u:ch ]ey,;Eme!l.~TI!! re,ga~,tled ~s :).nWnrnl~ ~, ~<,at~ (r9'64., I Q6~). Bu t .:hi:; i.1I!; ~i1 !.:I!Iilu~"U~ny 1i~ood .ii'ltetprel:3tllm i!!!f tim lunn an~un.yrny.

p

r

Gr~g' is made c~pl~cit in rorn~!!~iye scnt:en0(:ssl.Ich as ~OlJ.r house is 1IS big as YOUIi".!:'· :1T1d < O-u I" house ; R b~gg'~ ~ .~ h:t~, yours!. h. Eng~ish. there ;lin: :d~.cm:lth'or,,:', th:o'u,gh. less C(l'mrn,oIl.bnJ~ of cQmp;~n~

tive seraenees, enTpi'[lyi li1.g the verbs ., cqu.ld I, 'elj ffc[' and ' cX.(;Ccd ' eOur house e~ualsidLHcr5 fmm/c,.~cd~ YoClurs il'! s;iz;e·), rhe :1:tii;'lEti\'Cs A" '~ilnh~" and • d ifl'cn.::mt' CO ur h(HiiLSC i $ t1~ [! ~io:!rn~ as,fd iffo.lftnt [rCim.

pmrs in size"], ;IS wen j$ the co.rrchttiivoe !lnd _al'lmnyrno-u.s adverbs of { ~g:rcc 'rn.~ru' ' and j Iot.:ss~ C Ou I' house i ~ :!II.l.ore[tess oomfoufI(iJ!c than

you rs "} ; and there are varieus QLh.er construct i.ons pO$sibl~, some (If

them more 1Lttcpt;l:bh.: th>lll others in pa1'tiC:l.1la'll' i[lSctmClfl;. ilut there i:~ _ ,:gI[Jsid.e:r~blc~·lIFi;_ui(ln .~crl'l~s langu ~i}es in ·rrh.e Mea), Lnwhi~.!! g~d i n.g_ii1; li:~atlcaliZicil.h1 se~ltcnoCe~ ,~hichrll:lLbe ide nltiliied" in te.rms .of l.heir ~~.~m_ii,t~f.~II5t[]!l!sture, .a.o;.' cQmD<!~dv~ It is by fl!1l' means tn.1:) ca'~e thai!

all la[lguage~ wit.h aIlj~cti~"cs I)§C c::.::phcitly .gr;a,d~d .un the .most C1IDtnmO[l

kinds of' oomparnti'l.'e senrenees by I'HC<U'lS of li"lcadverbs of d~gT'ec correspond i~g t!J "more.' ani! ; I.nQst· (oo'.fl"€iaJtiy.~ wilh '!cli.!;"' and < ka.s~ , rnp~,cti,·dy). In '1l'M.:lt fol1ows. we wHt asSUN~1f lhat the gml]J1:m1ttllt::lli ctlnstnl.etions 'l1yaiiah~cror elip~icit :gradin.!l:. andthcie i)~llll~'~l('nec in

\.dHl.l C:lirl be .r.egardcd as paraphrasits of pa[liClIht[ sentences, can be ·~a11sfa.c:~~rily acoou atcd for ;1S, part of the gr~mrnatk:d a.natysi.s of an:!'

bn gUl!:gc \"h~B.(1 vccebularywe nreinlL ~"esl;gati ng. :a:n:a we will il1U1S:lrahl. thepai~nts Iil:r;: ~[]g made here @(U.H graJilI.bL~ a:n:wn¥ms by mt!!inS uf II limited St~tl!JfEl"1ghs~l. ,oomparnrtil;e :i1iCIlWrU::~5"

The firet po.n:t to be ~I)l~d us th3t the p.rop>Usit'iQl'ls expressed by comparnti\'e ~efltef:lces. like ·OUf hO'UiSit is b~KftCr 'lhan Y'01J!'s' or 'X's proof "of 'lac theorem.) us sLmpi.cr than 1"'5' imply, .m d :l,re. i1n.El~ed by, the prop~sitiQn1li e}l.pr~~s'cJ by sentenees Jike (a)' 'Your hOldse IS, sma:ll~[ than ours", • }"s 'Pl'oo£is more c~!'ltpl.e)1 than X's' (lIP (b) <"'{,QuI house is less

L. • h ' , V" ,11" 1 • l' t! ~"j ''''F'L ~t·· L' ~

Ujg.~ Olill.rs , JI S prOf)[ is c:~s· ~!umr C . n:m .. k s .• ~j].~11'

bel\'"'e>en. sentenees like • Our hD1~ se i.s .hlKgliT '~.han yours· ,and • 'Yo tJ:( hO~lse IS smllUt:~ OIJ~ ,. C""an- he hm!J.G.I.eci in ICITIl S of th~ C~llv,ef,o:;e..::· [less ~ 'off thc:;-I?rcd~~t,c;:s (d. 6.3) .a~ can ~~OO1din,g IliC~ ~'I'.~ ilifid pi!S~i "!'.~ ~C][U:tII~.c5 .li.kc '[aim kmed .niH' :II1J.' D.il:l was kill~d hy J~lnI>;. wewil] (";;:;;;- b;~ck teo this. The foint ;t:~J i}t not!;cl here ls l h.at ~hLsul:l&tLUn ion r ,~om:: o:f a p"~:r (ltig~a4able anto'.fiYrl~.Jf9r~he Olh~r and the t[)1LnlSllQ.!;ition. of the !'C1~"lHH f1oml!I1I! ...!!i~r~sslon:s. with.i-!_:I, oornp;'l{'olith~t: -sellt eflc.e .r"€;;;uits ~n a St:l11.ttfltk"Uy cgt~.hr.lJmtscnt!.::nc~~. Thi;s is ob\"~ous enough.

- R~~hJ;r less (lbV;'f:lLil~ is tile r~ct .that tl'te u'Sc of a\tlrnda~ an:tm::!Yrwi' II ~\li.'ay~irwoh"csc gradi,;!!:, im~li{; ~dJ; if fl.;! €X~ ci...ili~ This W(!$ sW;:ssoC d b~lIciF 1(.19#)' whQ s.e-cms too' have bt!en Ihc fir!>t lii'l;[jlli$t to e'mploy the.

~. term ·~r.ldil'lg' in ~hi~ sense\~Whe[l we ~ay, .liur e%amp~~, O~~r IJ·(JU'U·.I'& 4 Me «i.e. \\"hel'lwil uttee the senrence 'OU:I" house is big' in (!.'I'ti(:r to assert dIe propasjt ion "Ou r hCI1.1St is big ") we 3.Tr'C m'lt ascrih ~ng the fl rop.e·r~y big!'1'~s!'i. or s:iz:e .• to lhcr~ffrcm of • our house.', ~s we arc 3sCFihin;gt]l€ propCR:Y f~[I.ness t~, the n:£em:mt. ·af 'that' when "'1:: say :Tlll,~I~~ (t 7D book .

... W,c:lj\c~U1pFsi:~l,y !l:O!l:Iilpari.ng '~b~ h.ous.!.l Wilh sornet hi ng ds~ "fld ,fI..<;'$ert h:1g that it is: bigger. Tile :t;tJ,tld>liil'd olf (;Q:mptlrison m~y 1laa:\1e been eMpi~cidy,inlrl;ldl~lC:edl ttthe ~I:ltext ill whieh the se ~le:l]cel is ilIl tcrcd. O:~mmon~y, however. it win he some gen'C rdY~fXeptcd ~Qr:m. QUi' urmse ,:s big_ m~ght d~er"€fol'e: be uffilderstood! as meaning wmethifl~ like "'OUf hGU:S~

is biggf'r~hm'l the nOmlll[ hn ase • I' '[IF 'I Our hm]l~e i'S big for fI. hesse" ~ and the norm ,...:ioU. be \·ariaM.c ,across different languages, {or C'i.l~h:l!rcs) and a,crl;l.'&!; diffe:l'\l\ntgr!:lups VI,-it,fui.lil thesame ~(]dcty, Failure to reecgnise

~ he Iagical pflIlperties of grndlllh1e a:tlltOlnyms has g~\'en rise 'to !!! number of piScudo-[pw!J],cms. P]ato, for I.1_xampi.e'. w~s' pm:zled by she apparent ~ possib~~ ~ty ~ ~p-05Lt~ glJ~liti es (e.g. tallness and shortl:le~) co-e}1~sti rig in~hec same 0bijed: if we can ~!ty X ,'~ l'a./l'er than' Y tllfld jJlO1't(l' #'1)11, Z we ap~eall' ~o be aseri bi.ng both tallness and shartness to X., I\'lore u(!end).!. lo_gi.ciilrls and I,inguists heve discu ssed :sucl obvioU51.y fallllci:!]U~ ded!u13-' tiOIllS as «This i 5 l!. small e Iephaar, t]'),e',e;forr it is a smell animal" (in ;OIm~r:1l~t wit h "'.' This [S a r-ed' bOOlk" t h:erefore it is a red objoc~") ,,~' As,

_,. Sapir [:mts.;t~ '~:sucll C[lntl:mits as i$malP alIld ']arge", ~'.I~ulC' and 'mu.ch", • few' :lnd • m::ilily" give us ;)I ,dc-oept'lvc reding of absoluee vatueso within the fie~,d of qlli,antity' to:mparab~,t: to such qualitative di:ffereru;:es as ~ red,' and' gn:ciiI, ~ ~vithill the ficldof co.lor pcn:cptimTii. Th.; S~(leILng is an. iUIJsiaa, however; which i slargely dne to 'tht: li[lgu]sti.c fact thatthe gffad i~,g' whi.'ch is~mplkit in these terms is no,t .fo'i"fIl~lly illdka[cd, wb.crcas it is m<!!de eli:piicil i fI such ju dgmen:ts. ::IS "There were ]ie\7I'Cli people tl;u::re dlafll ht~ ,,' Of ,. He ha~ morcrll'il\ than I·' " (] 944: '~n}.

I~r.ld in~ ~ ,also b ~ Sc~!IiI~-ex[ll iciL ~:y ~("mi-exp]icit g rndil'!g is he1"c mC:!flt tlre !.Ise oK :rome m!l1Br;ative _!:OIl!!i~~tiQn. Wijl,lQllt "'~.I!i!til mention of Ul!IC $ta:ndard or coml?axiisQI'I. l'" 0:; ~exam ph:, [O~IJl.r;.JI!:tu~ is ~ ~er " lS _graded ~emi~expl,~l::it1y. ;md the srrii,ni!anl <!f OOU'lp~,~wJ1._ win I.lSlIaJ]y h~ve b~~mI 11r.t1,v~~~ir;m:oJ:uccd i!)l th~ CQ'fltext. SQ tOQ is • Our _

• The 'point S<lipIr WII~, nt~king j~, Wt!'I.I1t1l cnVIl til' Jogi~mJ~~ 'Do'! ck.aU~Ji;

_J,s rar As At!~I"1 ~ .{d. Ctl ~~.Ii:llfirl ,~(,).

·5 Kat~ (!g'7::!: 1ll54) dr~ws ;! distilliCtil'lI'l b~tu'G~l'Il;1e~~m'(l and 1iIJ~:s~~lutt adj~~tuv,~~ in this CQ(tllle!'!ion. ~!1i~r ('~g75} br~!1~ ~!.lll de.~~br I'h~ lugK:ai prou1:~m~ lha~ ·~~i~~ if ~dali",~ '8dj!lcti..,~~ 8 uc h:CBt~d. se.m~!1ti~~l~ll: ,Ii.ke oj'd·i.t'l~ry. uil.~.p1aee p~~dlmijte-s. Set: ~,ts(l,; ni~'I'wi&ICh {19:ti,,}, C!r!lliSe. (~9i6~, G'Yv,01l1l (~Q70). Lj'l..mg ([(milo.

boIDi~luoo bJL (th.el equiv.aIJenl of w:h.id!. in cerr:lun 13nguag~ is riot dI~$t.iml:rishcd f[\O'm. the cq!ilillllikiu of 'OUT hou.sc is blggcl"'" as the ,cl!Jui\<lllent. oJ ~ Ou r hO'lts~s the b~ggFs:e rn ayl'l~~ be cl~~t~!"IglJ.i.~hed from the eqlliva]el'ltof ·.our ho~se LS very big ').: the standard ,of comparioon= i.£!~ei!'~ ~ore 'CQmpreN' .s~IlCC it brilngs in t.he notion ~u rf![l;;'(!' and:!. whole Tru:Jge 0 f po.ss,~hlc critedll~nicltrn..1.y 'Of may Ill)t be ma~c IC:<i.pi icir C~ ... too bi~ [or us to n'ii1l:Lml:::!.ilil '"', "." .toe big for ~t9 $itt)'" cre.), In s,o [lIT as, a 'propos:itioHm.llst ha\i'C some determi nate truLit~'II:I]ue (d. 6.2), Wi! car.lIlot say .... ,hllrt ,rop.os~tiQrI. is e~p!:'eS;llcd by sentenees 'illf ['I1is kind. tlll~ess 1II'!-'C ·CIl!Ititst.111ifsJl, from ~he CQn~t or mhen:vi:se, t~ relev!lJflt s~afidMd of ~m!pi3:r.iWfi.

- Ilil mftiily ~[Ul.g1lagcs. including' English, t,hemQSI '~Ol1llm(lnlIYI!~(ld oppo>~ sHestc:m-l1 trO' be In()rnh,Q~Qgi>C'aJil'y unr-ela:ted (e.g. ~gQod ':' bad", 'higt~': ~ t <row' " beiMl'ti fy]' : 'ugly", • big! : !smllU '. • oidi. • : ' you:ng' j. But dt~sc are {:lUl'l:~11i.l m.hered in 'the l!oc:abllihry by suchl:lil~faHy related P-!lil'.!:

as ~ ma!"ri~d ": 'u ~ married', ~ fri~[Id]y·:;· uhfr.i.cnl~ly".' formal": 'i[lfo[:m :11\ <!egiti mate" :" i ~legit ima,te', etc, hi. each ease tire fulase~form· or one :member of the pair is d~ri'!ledjl from the base-form of the other OJ! the liIE~ition l'if the [I~~(ive ~fefixes un~ '0; in,- (cl';. Ij.~). :By virtue of this. mIi,rphoiQgi,eal. ror[C~i!Q_llq l;i'Ce" woros i ukie • unfr:ien.d'ly' • • ~Illiformal >.

~ ete., m>loY he described as _rn.0lWhq:logicaUy negll:l~\"e· wi,th. respeer to t.he - oorrespoflditl,g' rn(l)lrphologicarul,L_posit~e,fj words < fri,e[ld~y" "fmT!i'HtI',

etc. Now the fir.s( point te no6oe ~s that, ;a]t hJ)iil_W_W~t moroh,ok!g~1I;1l lln];d8:l:c4.~S.]itt:s. ]Fll Rnglish atieast:, are gn!daMe ~l1.d maRl" morphG~ logicalli'~la~!:d ~ositcs are ungr,adaMe, the dist E[lotiol'll between m.orrhllJ:~oglcal]y related :~IlLd: 1.1 IUl.lbted ~posit ItS ~s wndepcm:!~n,t of, aflJill:

docs not cerrelare :absolute'ly with. th.t:se~t'l:tic distrr:..ction of M,!_11Jll.J-,-

.. ab!c!l;[l~r.tdap!e. opp-osites~ 'married':' ~l ngle' , like • m;a;rri,ecl': 'U~m~ed ", is llClgr:ldabl.c, Wb.it:l1t::::1S' friend Iy , : ,! un ffknt.lly', like • friCli\d ~y': <hosti Ie j; is graGl3:ble. 'These !1:xamples n:i3.ve: been deliber~~e1y ehosen to' mUs:trnfiC; the fu,rthc]" point that the S.3!l1l1~ Loe,1{e.Fl."li: ~e_p~.ited ~'lhh b~ a mQ.fphol.ogi~]!y""!d;il~)ed and 11. Jl'lorplllDJng;icaUyu nrdakd_woLd in the ~'{l cabllt!my.,

\l\lb~ is ~haps mO'l'e irnportantis th1l!t e~1]mt ~hoL(J!i:iC3l1y I!fin::laMt'I oppo5th.:sj ]ike '~ood • 'Off' bad', can be dis.t:iflg~ishcd synt:<l:ctlcaUy 3:nd !1icma!'l~i:ea1ly ill! lerimll, of their If''Ositi~"c o-r l1I.ea,t:i'!.l'~olari.l.y·. We lend IOSll.Y thalt sm.al~ thiiil® laek sileo that wl1at .is roql.l[md i~, Tess ~cigh:l, Ill1lcli. ~,g Oil', rn.tbcrth:an thB:~ lar~ -lni!'J.gIl! IlI!ck smll'U n.e_!s and 'lhi.'lt w hllt i.s rerqlJlired il> rnO;N; ~o,w,,~:~;!,:,' HflW good ilS it:~ 'can!. be lIseilll \\"i~ho[lt 1my FH)elOUP(lCIISll:UO'[! or ~mp Hutiof:l t:hatthc rden;~l of! ie is g!iJod rather thal1

, ..

I

bi!!d '; blJt 'Row bad is it?' carries with it the p~u[JP~l>itll]!n thl1,t the....." :rcfil1:rcnt of 'it' is hadrarher than goad (in relation to some rclc.wftt fIiOnn). Tnt! (looitive apposite bli:nd~ W Er,e>c~de the li!_t:S?l!!e when

o . oS~h.:S are oo-ordinate<! in what MallHel (1959) ca.lIs irre"t'rsible binmni als "': ct. 'ggodf an,clI bad'. "high snd 'lr:lW~ , ~ great '01£ llmII.U·. This

. nincil!!_' flf pl"e:~crrccl scquenee .is in fact; of mu ·h w~d,f'r ;;tpplication. It enables lli!> to distifl£uish ,a positive and a negative mernberin such COntrast i I1.g F alffi ~: • man i and • woman', • parent' and 'child·, I noet h • and "south ", 'hea"cnj and "earth", • Iecd ' and! ell·ink·, "buy' and cs~lP. etc. As Maikiicl poi nts out, it seems 1'J~ O(ln~late quit e well wid'll what, on other g,munds too, '!lye mi.gbt de'~cribe as ,! h.ierru:chy fl)fsemantl-c preference.

Sapi; d is.cusses the polari ty IQ,f (tot onyms in terms of ,j'the tcndcliu:y to sJip kinacs'hetk implications into ,spel1c:h L";hichj sa o,frie-n renders .1. PUl"eliy ]ogka] IInaIY'Sls, of ~peech insufficient or e ven misl~::Idjng ,. (194,*: H14)· Lehrer ebserves that. ".it is tlu: negative cage which ap~ ~';oacbC's some lirni t or zam p(lint, wih]lc fhi s is, not true (If the positi ~'C CiJ,~e8, A tbi.'ilg ,can be 80m nan!)\\! or !IO shad or 5'0 small that it approaches 2!ei'\(I, illl extension I but there is no corresponding limit to nm .. · large. wide or tall ~omt:lhib\H can be" ([97.3;2,7 i cf,.also H. Clark, H)n). If We add to Lehrer's ObS.UVRtior! rhe fact: that mes morpholagicaUy neg:al:,i\'e gm(lab]e antonyms are also sernant i Olllly .!II~gati .. c (or marked) we can 'perhaps aOCQIJ lilt for Sari r'~ fedi rig that gradable Il!"l.ionym~· is suH'uscd with kinacsthetic im,Plicattiol:ls .... Thenoticn M a limit is, relevant to only asu hsBt (Jf t he antonymous pl1irs in :!1. ~W1gu;'J,ge, and most obviously 10 lexemes having to do with spatial. ,and u:nlporal extension, morphQlogical rdatio'nship again. is NJcvan.t to limly a subset of the afJJiorlymoUis p,airs; approxin'Cltio.~ to the ]imi,t or zero point and the prefixatien of r~tI'- o! itl,- are indcpendem of one another, hut 'they ()n.rrelate with tU:$l'Itivc polarity, whlch is, re1e~'lIlJ1ti'flI :til instances of gr,;\ti able antohym~'; and it is for this f' "ason verliaps thll't we interpret. as negative rullhc lexerne s wb.ich fu netion R},Ilt:ac.ti,cally like • small' or' • n ar['~Jow' and like • !.If! Fri c.ndl)'· or • [morma.1 ' .

_ h requires but a mommt.'s rd'lc){iol"l to sec thar there is no logk,I] nec;ssit ¥ l'Or lill'lgua~g to have 111oT'phoIQg;icaUy u f1rda~ed (Iprml!li~l':.5 (reganlle~s 'of whether the lan.guages in questiol] arc sueh that thcy can b ..... said to hal·c a level of morpholngic;1Il streeture or net: d. 10, I). English wauld b~ just as dJ:j'cic1'Dt a s.cntiotic :~Y'S"tem. one might think,

• _. h' dI .' . • JI," ,I' ,. 'd'" '.1' j f' , •

If we . a ;such: . ours as . grro!.l :' IlIngil()U , . WI C : .. lirrW.ue; 3:1':' \l1'I~

f3r"'. et\c. .~mie;ltd" there .j$ no 1~g:i.ca1 necC"551:Y for ~aJl1gU3igt:'s 'to h.we

lexical 0p'p'osites at all" Suppose wewere to amend the grnnl1TIatica[ structure of ~ngli.sh slightly, so that '·X Is not good", "X i~ bad". ... X is\ 1::"," bad" \'Vcrc expressed bY' 'mean!l nf the senrc noe-s ! X not is gOQrl' (or ·c X does, not be good 'J. ·X is nOI: gOne!', ex is very not good", The language would serve just a.~ we U fO],l'lu!king diistinctiilf!S, of deseriptlve mcuning as it (h:l(~S at r;escnt , .... jth a iargt"IC \'ocabybll:~'/; '\'hat thciI ,is tnt" 1t:35uifi for the existence of .1~x.iCllj opposites" and more ,patli,cula;]y motpho[ogn cally unrelated gn1.da~le ,oppasites:

_ We baH~ already [loled that anton. my reflect!! 'in' dctc]'~nil:les ,\1:ta.t ;a.prean;, HI be i ",cner:d huntfl.l'l '~encl:em:)~ catesoriz:c c;( pe;rience in 'h:rms~ of dlc.;;:;tom(li!l~ eentrases, Now it seems dear thaI: the Icxi,caU . 'tio!_!. (If polaTI,ry in t \~ I:llQFpholagi uJ] rl]n,~latl:'~ gradab1e antonyms {which ;f!uds to tn,earbhmrincl!<S and d~screi,encss ofthe sysW!'I:iL: d. !l.4J Cllhal'U;eS in some way the distinctness, or separation, (If the two oles, 50 rhat, as S~'r Sl]~ts: '~,amtrnsti'ng qU:ll.ities are f,ek a.~ lO'f a. n::latD~'ely alh$Oluh~ natu re • se ~{j1' spea,k. and • geod" and '!:lad'. for i nstance, ,anc! even 'far' -and 'i near", have ,1$ t~ea ps.ychologica'~ specificity as "green" and iyd., low~, Hence the ~ogical. norm between th~m is not fdt as a true ~onn. hut rn.ther as a bleadiarg-a h~whiCh qualitlee graded in lIpP(lsite directions meet, To the naive, c~'ery peFSUIi1 is eithee good or bacl; if he cannot be e~ily plaQed. he is rather pa,rt good and part bad ~han jLl.st . umanl)' norm~1 (Ill' nciwr good nnr had " (1944: ror), h is perhaps (~r t h:is - reason th~.t :mosL of the 9?~11;!Qn ~rradabLe ant!)" _IIJS i It EngLisll and other b:n u.agc~ an: rnorpho_!_o~ie~dl unrda;te..£!.; it refl.e~s ~ more cJ]irnpi~tc. 1€xicmbatiU'i'L ~f ~o~ariu:d conl:rm;ts .. ]n fact, ~t. ,can be argued I~ila~. carall~ka1izatj(ln ncccssn:ril)' implies morphologi'eal unrelatcdness.

- Mor hological n:iated !'LCSS between ~e!i:emes.12. r d criVl1ltO,fl,"} is tnld'Ll 101'1- aUy I'le'ga:rdeod. h_y ]ingui!1u as f.al:ling lil'idway between gnnnm.aticalization l'Ind lex i cal uzation.: 'good' ana I, i:ul,~I' are more obvio,usiy di:ffl:il'cnt ~ [~mes than ~n: ·.fric~(lIy' and' un£rie[!dly'. nut this is a more general point j, and \\'1: will take lit: up, in l\ later chapter: t].2.

Although gradllbfc and ungradi<lJble op posites may be djs~tngi[j isbed

., It 1_" mtereSdlilg eo note ~hid, C. ·K. Ol{d~llJ, tlu it'l'l'entor o~ Dru;k: E1ilj,l1 LSn. - he Wfis itl!in the author of II snn!"t., but lmli')tliU!t. book on the notion oJ opp-ilISHkm, (el. OI!:d~llJ, IQ'32) - B:~W fl~ tG indudc 8.rIum.1!l the 85.Q ~~Ncm.e~ of l)iJ.liic Im!!'if!lh :so ,ai!'! of mDTp];m'og[(;!lUy unrelated, opposites .. He i1J$<"> ,..n(l'> .... ed :~nr' ~lm fmmll ~ml 'W tthin. the S~'.9tc.m of Ii furl'!1j:r 50 Dpposih:-! b'~' prefucinl! W'j:~ ~o wba\ he eallGdI the n 11'!:I! '!If the qualitY'. thO-ligh he 1Idvi:!oed 'the ]~am~'r ttl ·1It1.ilke' 'IJ?C of '!!'IO,t' {cf.. Ogden, 1968: i. j r), "'ha.tc''l!:f DL~r crltidSJiI'il! one might milL..: of Ih:'lic English, on.!! must ~lijlel' C>llluJe in;!t Ogden's in~ti!'l~t p~ j'll(il~iilllent v.as :rJ,g'ltt in .i~ re$~t,

run terms of their' logical ~l'rope!trtie&, it must ~dso be home in rna rid lha'tgrli.dnb'lc lli'ltonym.s arc fn:qucmty emplf'lj"Cd t~ cverydacy languagebehaviour as conuadictories i!nh~i than oonl:Farie~. U we are ii.,sked If X II good t:Jjtt.r~play.eT? and We reply No, we may' 'en be held by t:lt questioner ~o have t'Clmmitted ours ] ... ~ ill'lplicidy to the pmposhiCl'[l ~hat )( is II bad u:;h,essFp:laye'f. This fa,~'~ .if! pe.rbps best hllncl led, l.ike m:;my ochers, by appcali,rtg, to a certain number (i f general !lllmioti e p:rincipJes, which govern the normal use of language. {Some !'If these haV',c been codified and discussed recently by Grice as what he callis oonversation:d implicatur.es·; d. 1""'3.) Fo,; most praotical purpOllesWIC Call. usually get a.tong quite well by describing th;ongs, ilii a first approximation ,:115 ,it wen.'!, in terms (:IF a yes/no dass]ficati.m:a., according to which things are 'either gO!ld or bad, big ot" small, etc, (r'dati\'c 'to some relevant norm], U tOre deIilY that s.omething is good! or assert that

.i~ is not good witihout qlJl.alifying our statements in any way 'or supplying any furtb<::E information rd.evll.llt w, this di.,t.hu'tll'ffiOUS ,Yes/no dM5.ification, it is reasonehle foI' 'the other partiel pant'S to a.5e1W me. 'thaI!: we are satisfied\'Ii-i.t,h a first IIppwximation to terms of which 1;I;rnd..\'ble antafl~'ms are interpretable as contradictories" The prepaskien "X is I:IOt. good ,,_ oh\rjousJy does not of itself imply .. X is had ", but under Ithe. eperatioa of this principle it may he held to dL'l' so 011 partiula:r occasions of the utterance (If a sentence e::'fpressin,g it . .H the speaker did not wish. to be {lOD1IDl .~ri (I the .nplicarion, he CQ],~I.d have hCCJl\ expected to m::&t .it dear that a.first a:pproxi'ltl3tiofl Wa insufficiently precise, by saying, for e;l:ampl~, X i~ "at good, bit" he's trot [}ad' I'ither: he'! fai'l prefly gfH.!dljtut. (Ji1out d~"I!rog~",

- .rnt is 31so a fa<:t: of norma] langl!ag:~beha\'iour that ungradab'le oppo-

sites can, 'on occasion, be expliditly graded. But the e:tplana!illl'll of this fact is uSl.m'lly of 9. dilferell,t order. If someone says to us lit X stm Qli'l:'~ thi'!i? and we reply Ycry tJlIlCh so 0[" Arul how!, we are not thereby ch"llellgi ng the IJI:lgradabi11 Ii' of {dead • ,: I ali VI(:' in. ehe ]aitlgu~e"system, What we are grading, ~;lF'esumab]y, are VllriOl1.5~ec6ndalFf implicat:LoFis ~ [II' OOrlTJ.uta:tiOfil;"* (d. 7.]), of "alive'. So too, if we say X is morl! rif a /Jaclndal'l'hall Y. we are pro ably comparing.X and Y in terms of certain mere or I,e. s gene.rnlly accepted mnnotntions of '. bachelor', Uut then.l :ue other nccasions when we will grndc II pa~r I)f 1'!000mally ungradable antonyms, h0C31!St: \'\;1: do "eject [heir Interprct~nil;J(n as i.\[.mlntd~~toti~s, JMa1l.c· and • rt:l1Ailh:;, are ohvious examples .. We normally operate llllder the assumption that any iu:bitc,arily sd~c-t,!:~ human being will be citac'l:' male or fern le (rather than n.;;ither male: nor ft::male., 0[' OOUl mile and

fc.en~~de). but 1i'O-C may well recognize lilalt c<I)fulin pf;opte Cill1llm he satisf~clori Ill' d~5<.<;i·ficd ~il terms of til; s ye~!lriIo o.ppo:sition (If e male" lI:nd i:fCU1;JJlc'. \\oTc can say. for example, X I~ tfC~ compleMfy male or' X is more P!l'tlll1t1rmljl1nlt.llif, Out in cases like this" we are modifying the languagesysk.m, it 'only tJempo~any_ Recognition of tn·c: possibility of gnlding normal I)' lIngradlible ,antonyms, in either ~d 'thll two "~F!.y~, ment Do!\ed here" does I'lfJit i rnply that th~rc itS not a. sh.iltJI dist unctionlo be dra WI'I between grad:abie and! ungradable an'tollrms in a Jllnglltage-!iystc.m.

So fa'l" we have been l1s'lng 'the terms 'alllwnymy' anc! 'opposition \ HlOre Oli less equivalently, for various kinas of OOlrntl'l'LSk bet"'fccn lexcmCSi and YO"!; llaV(l not explicitly 03 sti ngui.shed between 101'1' ·'S itions and ct;m:trnsts, Opiii'iLions wm differ about the advisability ~f Jrn.,.\ri~ a tenIliinologie:rl tlisilnct.ion 'n one way rarher than another. nut the folJowiHg dassifia:tionppe3!iS to be \\'orkabl'c.and~ convenient; and we will hencduI"th folIa\\' it. Contrast· \\'m be taken as the mlJst ~nerat term, Oli!1'Ying.!'1l i~nplrcgtiofiS as l~ the numher of elements in tJn.~ sd ~:f :P'arndigmJllt~cal.ly ro~l~rastlng ekmel11ts._ OppQiSit~n· wil~ be r~';S!ridmI to dichotomous, G. bln~'i ,oonu'asts; a:l'ld antQ'llym.y· wiJ1 be restricted still further. to ~ able (l P positu; S1Il:ch. as • big' : • small', Fhigb': I low'. ete, The reason for tlMs deliberate restriction in the scope of dle r,e.rms

• antlol'lymy' a~ d ~ antonym I lies in the fact that. ,II! w~ have· sec 11, gr.a:dahl e opposites manifest theproperty of polarity nlore ~t ~;k:i n,sJy rhan do other opposites" Unlgii",w.ahle opposites like ~mal,· ~..- _ n -': •• 1.[\lc" wiU be: termed. fur reasons whidl.. should now be dear, U)~.'l.pkn1'i.:nt3,~ ries e .. This leaveg the terms <'OO'll'adictor.!{· and 'OO'tl'tro.y· free {ar ,em.ployment -in 'tile !>eflse in '!i\-nicll thiey run e b.~n dcllnetl by logidalls.

-=- Curting' aeross the ~isti I1lGl[OIJl of antn ~mns and C~ t'll'plement:u:_ks is the distinct'iolll that many-structural s(:]na:l1.ticisl~. ~oUowbng TFlIlbetzkllilJr (1939')" draw between 'ri.vntivc· ilJl"id e ui ollent· oposjt~.s. This dis~il1ction (whicn bas bee exemplified above in oue diseassion of po];tI,rity) has been drawn, thougb mot m::ccssari]y in dle same temu, in ~1any ,g.eneri!l. l:reaJUnents of OppOffi ~t ion and (;0 nU':lst (e.g., Ogden~ I ~}J2;).

_ A 'l'lri\'lltil'~ o~siti.Ofi ill a contrastl II;; relation between tW!JJ' lcxcrnes one of which dCllotes some pa.si,tiv,e propcl1Y and the other of \\;Ill~ ,k:l'IOtrs UII: absence of r]lil.t p,ro~E!ny: e':I;', "animat ··:'ifl!Lnim;ne'. An equipoUent 'CJppcn,hion (0.1', R1.0r-C gcrw:raUy. 311 ,equipollellt cnatrast] is ,iI .. l"e.I:lIItion in wh.idh. Ci!cit of t'ke (:CIntraB;ting' k~emes UJenotes a WQ5]tivc p roPCTt)'; IC.g., 'male:' I emale ;'.

1'0 bed is:~i ~uis_hod from. _1IIItQRymy I!!ilcL~1)illlfil,I:.lne ntari t}C. is ,OOtlvers'enc:s,S", cxcrnplined by pairn Like 'hl.lsba.nd';"wi.fe' (which may he

regarded! as two~praC!e pre,diiC3t,C',s}, The sentence' X is the hmhancll Jof .. Y" ,a'lp'I\~SCS, a 'Pl'Opo:s.ili.arn wm!:ls<e converse is l~fl.rc8scJ. by • Xis the wi fe of Y' (c[ 6,3). A~ ... ~ I!oted abQ\I~ .c; C oomra,rill!ly~ [\orr'll!:. ofcx:plid1tiy ~ ~ecl. ~!Ilb!)n~lIlls ("'ilfgC'l":sm(fl/~r, etc_) and ~ o:ndifl~cxive and passive forms of trnllsit.i.ve "erb~ (ki]/ed: W(lJ? killfrJ)alsQ opc.:ratcy,'ithim s.'Ci'JItC]I eesas d,0 le j(;C:\IOOfH;'C~~es: < X kiUcd V' expeesses apropesitio i:I wbkh is dw ct>nvt:1"SC of the proposition e:.:pre.ssd ~y 'X was killed b)! Y'. N(lw'j by virt\iJ.~ of_~~ d!;nl)litioll ,of the ~~gi cal fi:~1I!tion OfOOJlJ'I'Cfll~ness, ]f R is a h\'o-phcc relatieu all1ld fl.' is its o"l1!ve.f,!l~, we ~mL slIab.st'lu.!tJe R,' {or .F? arid si IIllltllli.lallJU Lls~y lrnnSp(liS~ the: terms i ~ the rolatlo,1lI w,obtain an equiva.lelJlct:: R(;~', y} 0= .1~·(y, ;r) .. Prarvidlcd thllt '~lh.e appropriate g!<ln:!lIllatucal eha,ngcs arc earned. Oillt under the tram~[Jo5i,troll o:f tile nfJ<[!l.imal exp f€'8si:Clns, we tlln do the s;:[mc £0'1" JiI!Jlurs of sent~gs (o'l1lt3intng crl'rnlfcl's·e l e..l{emcs: er expressions anci[he rrnl'msil ions, eli:~ressifd by _!!Ie n'ilO m.!!mbc(s of ,caclt ra.i Ii' of scn~cn~~'i'ill be CQ1!ll valent: "X is, higg~r than I'~' == "Yis smaHer lh.1!l11 X" I "X r[el:l~dt'9, 'V'· == .. Y Fa~~(l,'i.\;S X' " '''X kiUcd 1"" == iN Y was kiJlt:d by .Y ", Con~ verne lIc1at ions between lex~rneswhl.rnmll:Y b~ed 3S twg~rhu::e pi:"edi.c;H~V€, ex~es~ ~o_!ls arc ,e~ed~lIy COJrtrrm(m in ~fcas of the ¥lutabtl,lary ~<Ji~ingw do w.uthrm;:ip.rnOlil socia] [0]C5, (·d'Vc.to:r": <l'JiiiticnC, 'mJilst~Ttr "m:is.tm"'ess·: "servant", ete.) and li;inship rda60ns (<fia:ther'j'mcnher':

i son j r daughter', de.), on the one hand. and tern poral ::I:!1d s.pat~al, n::ladoM r abevc': 'k]ow'~ ".ill! front of j : ' be h.i:nTl.d '" "before ~! ~ after", dC.)~, Ol'l the o.ner,

The~i tl!la.~n~r' ",,~th ~eS~tlel to lexernes J.ike ~ bu~ and ! 5t:H' iS~i'lthf::' !'fI:[r~ ~Ie". If ';;:·c tr;;t- them I1S thrlt:Jc~~£C _prcdlcatC'~ ~Qd correlate 'lh,e orn;f'[ of the terms In the l>yml;wli~ repr('Sel'lt:l.~[(m of thcrdatioos R(.'Jt:, )1, z) and R{x, y, F)' wLth &uch.grammll~ic:d funetions. assubj(;'ct, di,rc(;t ~;bjcct. iml~rect Obj~iC~, etc., ~n [he sentences c.Q,nta~uing ~buy>, We can say dUll 'buy· is the ] - 3 ennverse >0 f 'seJl' (d.. Da:r-lil inel, ~ 99'7<1}. K nOI,",,'i ng litint it is 'thcllrst andlhirJ Qflhe !l{)mi[u] exp resslo ns that must be tsansposed in' buy' {x, .1. :::) -::::l ~SJ~Il' (.t., Y. x) and kn~~wing also what grarn:If!l:ai.kaJ ,chal"!g~ musn he made i ITIl ~mtCI1CI1:S eontai rli ng '~buy' a!l~! '<"iell', \V,e' can relatestld. pa:ufll, ofscmh::nocsa:s 'X boughl Y fro'l]lJ. Z~:: 'Z ~ol~ ]I t!fI J[' .in terms of thccqui,v~h:l1ice 'of tbe propositions expressed by ~h,em" h us pussibic, .how~¥cr: th!l! the ~emafl[ic relation \.\·hid~ holds between mtl.l'IIy-pia.C( C~rII'1lCneS c.:!O in an 'ca.'Wli 'b~ amliysed :lis fhe prQcl1,jc~ 'of ~W(l 0 rmoreaimpkr :rcla:tions .( d. ] 2.t),

~'":Z:' Diur:rio'lllll, flr'lhogrmal tlnd l1l!1.lip(Jia,i f)pp(}~itim,~

001 fllif \\'e ~l~,ve lfemg,l'Ii.Zie'li thn~ckInd5 o[ It::KlCa.1 ~!E!.esitii{Dn.: 1I!nw~Y!TIY (n3~ro,\'~y dcfnncd in 'terms of grad:ltr~lity), cOl1llplcmentaruty aml.~I1~ ,'e!',se!1es~:" There is, yet ;1 fourth ~¥,e, wid'!. vag;ious sllbl11.l~,. 'Iovhicih, d~o'l.l;gb it ca~f1.('}t always be di~t.iflg!lllishcd {:rom the-se rtlwee, is 6uffidcn:dy impQrtt~t inlanJ;lil!""ge w, begiven ill sep<llmte ]abel. We will c~U it dirCi:e-

- ti.6:t1!:lJ jj 6;pp~it;~n~ It is seen m.ost dear]y .in fn~reI3!tions,hi p w~ k:h helds bet n"CGn • up' ~ < GI}W(l. ", ':i,rrivc I:: 'dcjp;;:u"t', <llld'~me':' go'" Wb,at these pru.I'S ~ntre in dQI1'Hll'!D:nTI., in 'what might biC'rcy.r~cd as thei:r most lypical usage", is lin ;mp~iC:3:tionflif m.ot!Qn in mae (If two< ol!l!.oscd durec~~Clns l\'lir:!l. fCS.P(K~~ t~, a ,g:ivel1_l!l;ace,. P. B\lt then: are Impo.rtant d:i:ffc'mrnCC:l< between Uiem.. - if we c[lI~:p1!!rc < up" ; , down' whh. • come": < go '" we ,=. see i iliI:ffiIlJedil:ltely that, whe:r~:ils 'OQMe' ": 'go' is based upon an opposLt.ioo belwOO:IJI m.{JtuollJ~o\-Ylllr(t$ P IQdrnotioll :lIU1!Y from P' (as, :.]SO ~s 'amvc'~ <g,t:'part '), IUp ":' down/ is basecll upon an O~(Isi~lion drawn \vithin motion 11\\;:I:¥ f:rom P. ~ R;,gh:t ' : • ~dt' and • front>: <D.i!Jok', when they areemp~.G>}"~i!lI in direction~] 0[" .ol'u'entati,o:naI expressions, are lil!:e 'up': • down' in thi5 respect. j3ut the ~irecli~ rl1i!l:1ly ,~f • up': • down' (i .e, in the \,,€,rt[cld. dimension) is aboo[ute •. in a way Ulal:: 'lhe: di:red ion:lIi~y of " right': " Jcft ~ i:lnd ;.[mn.J; !~~ck ~ms net. This lS211lJ impQl:ilaJlit peint; and we \,'Hl cemc back to it (cf ] 5.,5).,.

~ -If < COme ~ ~~~ ~ is, com[lifl.red with ':~1"I"b:e': '~de~:1in ':, ]t~'ill be seen tIun

~ here :!"~ \'ntlO!ll!$ differences. 'lh.e ma:s.t i m:p(1](FlOlnt, irr ;:.l I t n;~ <Jtnt:icaJ. point of view, has, w do with dre fact that the o~s~lion '!:'.'~'Yi:_.: "came" , ~nd ~~~!1ke dle QPpcnsitiQn of '. heR' ':!!Ind' theR:· aDd mao),olnei[

~ pairs, in'i'o]v,tlS dEtixis~ (1(:;5. 15-1), \l;hefea~ the oppositiea between

.~llrriye' and: <~~t' does net, WI!; can. say "X a:rrive.a in. Paris last nig!r~ Jt, J\ecgarnless {li5 whc'~he' we ;J.!1c O<lJfsdvCSo ]]1 P:rf.Is1lI lae t~mg (!If the uUeta:lI1loe, or '\~'e~c in Pi!!ru lit ~h.e tim.ci of tin: c"Vcnt b~ing described.

- Djr'(;cti,{!nlii!.O[lp-Os~t ion {:a:~:mot be d ~S<D1J$sed ~atisfac:~Qlri liye:-licept with; n 3 more ~ncral [!rruncwCI.f.k. which 3:!'laIX,lles IQ.ea.ti{l ~as, bdl'l!il i!l.

3. ce;t'!,ai ~ st1!t"~ ~mi ~lot ion ~as, SO~~ ki~.J of chan.g~ ~'E state. L()<[I~eJ \t

f rom chIs pO! rut Q£v.,ew J!!!.!!!!Ig ~ n rilirJ~ st~nds in th'~3~~i_f~!~911 ~ i!t'in_g in~. as ~~~ rnal'rtcd I~O hcfl1g nUU"dcd, ~)r ,M:tftil~'~ "~g_I,.~cAt~~ lo hal'il'K \~'~;Idt~ :ina d:~p~ding f.rO'II1l1L or i[e:!,lI':ing'Pa:r1s, is~n lh.~ same:

JldaltOfi [0 being 11'1 PI1:r:lS .1s.1!xiiJ!g i~tD ~iril!Ilg~_l" f~lrg!l.ttil1,&is ~,.lrno'i'd~"!g. It 'i,H.uld he di ffieLl,I:~ to el\1~ggel4l~,e ~ hl:iHlPO r~ancc of d i:recti£lfi.:al Qr po~~ tUOI1 hutb deic!i.e: _ and J1on~dei,e6c, :!L_:JI ~tl:'}~ et~rml rc1at[Cl!JI. ft lS allpen';u;b:e in bOI~h the g:rummlllt,ic.lli lInd the ](:Xicldstmctlllre (if 1.a~u:3gcs:

it is cent i! ec ramn:.l.t.::1'] catezories of tense, i!!2ect and case and the pe~ol1al and dcmonstratjve pTOnomil~, andl it: is the basis of mucbthat We might think of lIS metaphorical in the use of

artiCI:lla:r lesemes and exnresslons. urthermore, it may well be that our understanding, not ~-;;:Iy of dill!: ctiona] cp_positioll, bu~ oS o~osit ion - ~n g~eral.us bilL~ d upon some kind of Ilinilog-~cal c]l(t!i':n5~()n of disti nedone, whlch we f:irst lCilim, to apply w~th respect W our (111m eri e[ll'ati,aft and the lo cat iOf! or 'Im;)[ul1fnP (ill'! ~.r ,()thcr objects in the eXl'enul world.

_ This ws the thesis of Ioealisrrr", t)m ",hi,en we wiH rE!tum (d. 15.7).

~ _Motion from 3, place P n:·su'lts in 1!:IL~itig .... t :ii.ail:':-p1or not bdng at P); and motion tiD P resuhs in. being a~ P. Th£s glvea rise to two possible relatioDs (If oflRSeq~e[lcc· based 01'1 din:ctiona:li.t:', which may be dis~ t1nguished as 'pfl&ili:l'C 0'[ ncga:ti,,.c., Illcco'lfding to ""nether the rcsultarnt loeatiea is P I!!: not-P:_ P'ositi,'~ CllifiSCq_i.l!~IiCt: is c~mplifiotl in the ,impliciltianal relatienwhich holds betweea the~osidQfI e pressed, by ~X has, mmefgolJ:u: tJol I" and the pr'Opoohil}l"I e'xpressed by·X is, (now) ::l:~ P" ;.ne1;!!:t iv,t rolls.equmce i nthe rchti on Wlli.::h hoMs beh'l''C'C]I the 'Pl'lJ;IIposition ,exptt5,s,oed by 'X has ,omelgone r"-Orll P' and the propOSt~i.oil1. expressed by ,jX' is notlno l(llrn_gcr at pt. Positive or [Ie~ti"~_~n-::

S • Ilenee is rdeV:!ll'lt to the anahrsi's of the sense of p~i]'S of lexical. opposites in ma~_:_din:ererr~m:c3s ef thc \'Oclihular:Y:. ".Learn' lind ·know~ (:in certain ,oontm,.s) arc nd:ued by means of the implication that holds lUl!t\'i1!cnsuch pates, of 'proposition.s as "X has Iearned Y" -)"X (now) kno'w Y" (i,e, X ha.s gtIlne from not know1ng Y to kn.owi.~g 11iltnd 'forget" !lnd ; kni!l''Ili'''by means of the i mpH cation "X has fu.rgoltJCfi Y" -l> nx' does l!lDli1]c1) lOinge; knows Y" (i.e, X h~'!'i. gGr!e from knowirng to not kfiG,wifig Y I S· ~r~y f'Or the~os.i:~iveh~ }illed~f~~~' r' .1l£l~~ ") : ~ have' ('~ 'posse~(;"'). on II-m nll,~ hand, and the- 'O.~ti.,~~, rdat-c·[I ~ Jo.~.e ' : "have" i &C;";·(ll~)':'i!livilt. i (g",.tLdivo~!1': '(bc)_nl.!tt[ioo', On the otha. n,' virtue of thc~e rorat~IJI1S of ~l.scglJc I CC, 'learn'::' forgc't' (as 'well as • rem,embe,r": "f'!ug!rt" in ,other contexts and \\lith. &illme\!i'llal: differ,ent imp~ications). !,get married "; <get. divol"ced '. ete., may be I:"egardled as. dire,cti.onaJ 0;' esites, ]ike ·t.fi·;"fl"om!'.

- Let us noW dra'\_ another distinction between rthogon~" and-l'inti,.

I!odaljl, tlpposites .. If we comidicr the.EP.p-os~t]lln5 wltich hold within the

set {< north', ~ south i. "east", "west"} \'I~ see th at, thc;y ~re of tWQ k~n ~> ~

~ Each of th:e four mgmb~rS of the Sd ~5:._()~Sed a:~!ll!goIlilly (i,e. ~ 'i1endi'~uL~J ,~o t'\1!'(1 others ("north' i~ appo.scd in thi:llll\~' to ! Cl\Il;t." and twest '" <Cil9t~ is 0" 'P'~ed to "saurh ~ ilild "lUIrtbt·.J_ ,e~:L:md ~[ltipodalJy (t,e. dianu:tri<;alhl ,to 'one other (" no,rth· is thus oppased to 'sollth/ and

- "east' '~(",o ·wesl'}. TI:!~ antipodal OpposItions arc domiftant in hi .. set of f:oUI' le._cfIlcs)n the sense that native speake.s of E'il,g~i:sh will undoubted Iy say th at j, 'north' and 'solJlth' 01" • e-asct' andi I WCSt' are 0PJ'lI}-, sites rather than ' north'; and 'east' or • fHlI"tfl' lind • west .\Vht·n t1l,e,a!'ilupoJaJI opposites arc 'cl!!Iploy,c;d :as u"1!-f~il.CC prc~~Ci.ni,~"e ~l!: ressions (or wi th hI. such t"';o-pl;[cc pre dicative expressions :Is f M 'the south ,f') 'they are (If (ou~s<e eonverscs, ,Bad the m[llr.f sp,ecia\.eI<ltionshlp that hold.~ between j nort b ~ 0Ql d ! sou th' and bctw'~~[1 "e .. !,s;t" and "WCIS-t" d'c:ri\'e~ from the faot that aU foU.!' lexemes bdo[Dg to the same fie~d and c~cll. Iexemc is Ji,amctrDclIll.y opposed 'to its eanvcese in 11 tWfi-dimcllsional !lpace. Similarly. "abovc' ~s diametrical] opposed to ! below", "in front (If' toO 'I:H:bim]' and 'left' to' 'right' to • right' in a thre'f~dime!lsiomll "pap ~ _ We will USe the term anti~odal' for this kind of IJ 2p'Oc-s.i1ioll-

The cX!lJmpks that have just b«n given lllustratcthe nature of anti[!mlaI o~tioll in a. re'J't4ttvdy straig:htfl'lrw!1l!nl and ]]],tuutivdy Obl.jOUS manner _ ]:I u ~ ant i('!O(h I opp<lsit~oJllilS ~ no mmms (o-n.nned to, arC'..;I:S of the ,mcii::mII!ID'-!n'1!Y.ing 'w do w'ithJocatiQn _ OI.~ oriem.u jOell, in phystcil sp!!ce. his, uguabl-e that i'l (lp~mt,t:s, 'tu sorne degree ,I;!,t least, in the area of mlIJur. Any native speaker of English\Mould p:l'Obah~y agree, witnout hesitation: that !Mac:k' and 'wh~tc" are opPQsire~. Some speakers, thou gh pe rhaps a mi.nority! wou]d .claim tMt '. green' i 8' simitla rly

.I opposed to Ired' ~nd 'blue' to ')'cUO'\v: Nowit iis, !llIkrcsting to note that the fucal areas denoted b~' ehesc words are just those arc S \ 'hich are gillen t(:Xi~1 n~c~ notion in lalli' l!.!CS with a, six-term ~ol!!ur s tern a~tonHflg to the bypothcsis pur forl;l.'aru by D~rJio and K . (cl. 8.3) I an~_alstl ~~art __!h,e_y can b~rr~ 11~~ ~s pajred ant ip(1,,)';': t opposites

J / in a 'th'~-<dim.'~nsiQnal. s(l~ll~e fact d[u most. speai.ers l/ T~.li:{Tish-

.... ·1 k > -'~. b' • ''_ ( -. • • l,'1,

treat.' u :Jj.C· anu 'IWitt,C illi opposites, nur nm.'e!i:lC.ep~ In ccrrarn speCl.a

COlltC;XU;) "red ' aad 'greco', and. :stiU ,less "bt~e; and 'yellow'" would sl.Igge-St that the principle of anti:poda.l OPPO:&iliol1, illl English at least, is giNIC;u o.l!I.ly pal1:i:d reoogl1li,(iorl ill thc!l"oeabul'ary of oolQI,1I', though it may pl:!r}' some mJe .in the acquisidon.of ootullr terms by ohiMn~n.

I 1 'nest; Ihn:a= plllil'!' :J,~e in f;lctrc:~I,~11 iIil; QPPosJtcs in Ra~i!: E]lglis'h, d1('il,lsh~

.few .l!:n,gli'.lh apeak.!! IrS would ~h:ink that the enntrasr h~tw~e:!1 • blue' ~'!1d1 • i'l!lIow '. OF C'VCfj between r gn;-en' an d • red'; is (If the same orde r' a.~ the ootli:ra!l.l bdwc~1l ',black' snd 'white '. AroocdiTlR' till ORCC(l (.19'JJI: RS) "The SCluitive wlomist •.. will 00 lerliflph ~iq;: th~t ~,~d and !lf1~CI1 TD'!id~ the -. i'cml :\ndinlhd)it~!lii]c CIIS,G of ~Jlltiol'l. ''''IIcy p1J[~ hl.m BP.Bct. Bli ~t w~r~, ~mfltronIlLly, ml.d 'I.be, fact dUlt tbcy nnltmlize one ~l'IrilI1Ll!'r ~l! tumpl~m~ntuies

is, me ril~' :J. oor"lbry oi thci9:" .FUl"ld:Om~ntB[ IlIp~h;1 on ".

~ "There liS at 1~8!>t one thoory Df eelour which. '''Illl.hl ~~~1 to account for th~~c amapodyl n=rlflion~; nsmci)! Heril~G's h&74) dillDry (d .. Zollingu, ~07J).

Ki nshiJ? voeabulary in many langlll age-s also. mamifcsls the f,ri mtdplc of :mt!podll~ 'QP'l:msilEonin \'a:riOIJ!S, >t'Ay.s. Let us cons,ider, fn:r s~m!p.~i.c:ity> just a. t\'>'l}"dimf':lsiomlll SpliC~ 'stru C"l:J:I red: ~ n t erm ~ !If t.h~ s.1.m:II1I'lc'~ri c.'lI lfclatlol'ls biiing:m:Jlf'ri<edAo _~illg;tlln.e"'Sp(l1l5e-oQ ~nd being::OOrIJ.-OI.f~:lIg,e-!:;a!'l!'Jie-~J.larent5=as {blli[lg-t.he-;siMi'!tg~f). S\.lppo~e that ill [S the spf!!-u.se ©f ~iI UU!t " is the s;iblingaf a I' land ,d theSl,b~; ng -ofb. This m:mL be symbolized ':J;S: spouse (a, h) == SPOtlSC (b, a): :sibling (,a., c)s~bli[lg (c, {I) = &ibling (0, d), siMing {6j b). We call ~m~lllc :Fol1"l' memb!lr~ (If the set a, b,. c ,J in ,a :rretal"llP'liaf space a~ we ~..n arrange the ca:rdi nal poims ofthc t-(;lm'P'ilSS" such that ,a is orthogonal· (L.t. p~(p!,mUl~ cula:r) to b, b. to c, IIPcl; t. to d. N Q"W the cornpr~l<: r>elaltiol!'! b d~ g. th,e~siblfllig~of-dm s'p-olli.se-of,. whi~h may be symbolized <15 .:5ib~i11g- X -spouse (x, y) ~~. an antipodal opPQ$h ion which is d:u: p.lJocllact of~hc l\\'Q1 O"l~mg<l~ relasioas s,ib.ling (x,. !l:) aftcl spouse i(z, Y)i. 'The com;e:rse :llntLJ,¥Ida] tdatL~IJJ, sp8uge- x:sib~inig (yj x). is, i3J pmduct ofspollsc (y, z) and stblillg (.C', xl- These ~i1il1ti~dal rdatio:[l$,J hold i Il,g bct:,,,,e.enb aad eand b€t~"fena a~d din the prCSlf"Rt cxarn:pIe, may bereferred 'lo as ~11·law relations, We "~l1'IilO~...::; k~ i:!:l_!dvancl:, ~ f C!!lcursc. th!!J they win be le:x i M~ized ;:[1 any gi\'-e:n laJ:lguage. even th.ough. the language. ope.rates in ~ society whidJ. j nseitu~i(lnanzes mOIl~gIU1rous .m.a[[iage and structures its kinship system

in terms of it, Hut let us jw;t briefly 'c.gmp.ar-€ tWOI tii:ngJ]:Ige~ that (~O leRi-

_ IrnIi.~e thes.-e i:n'I~la.\\-re~,a,!i~n~nglt~h.:!ll'ld Rl!55i:~,IrnJ.

InEng~i;sh, being the sibling of the husband is, ideM~ifi£'cllcxi~ny wi til ~'ing the si bJ,iflg (If the wi f(l ,,~brotiTter Isi:nel"-U~l.li'I.w '),.F.:u:'therrnol'\e, ~rb]ing. X -spmlse {j,.,)') is icle'l1t~Jill:d with SpCi!JSC- X -sibUng {.\', y), There ~s~ h(J~,,'e\'er. in slan(hud lil1,gli5h no lexeme ':!Oib~ing-in-b.w~ (or alTIlY $l ... gh: k:«:rnc that i~, ,synonymous. with this) .. W,e must eboosc between "bmtheli-i[l-la.w" ;]!I]d 'sistH-.in-law; acooroi~,i5 to th~ SCI( of the I.~cts,on ~~~l1grdl':rreJ. !~;. 'i\?h:i,tw~ find idel1dfie(CIks:ic.1Uy i~ Engh5,f!,thc.c~ fore" :are .. brother-ef-spouse" and 'j hl!l~band~of~sibling" (;1$ 'tl"'!U as u h'i;]5b;'lind..aF..!5ill~; Ilg~oJ -spouae ") by m cans o:f < brether- ln-Ia w· OIlnd .t:si Mer~Or""'SpoUise "and or wi fe-o:f .. si.hlxl1g j I' {aslh'eH :lis. " wi fe-IYE-ISi blill;g-of~ sp~u;sc") by means of ·si:ster~]]1L-law~. ]n other wordS, a fur~lhcr-in~.llIw is; a ma]e~bling~lfi~li'!w :a:nd a. sls~er'--i[l=raw is :1. ft:maLe sib.lillg~in=[1I:w,

11'1 Rus:siafJI,.there :are six: :~cxeme"S., :n.ct [?rc, '~~ be cons:idC'f\C!ti.As in En!!,Ji5:n~ lhese .... of du: person being reft.n::d 10< lsi 1'1 all cases relev:ln to l]iut being th:f' ~ib~:in:g of~he' husb;a:~dI is, dist u~g-1Ll.ished lexI.caU y fWIl'il bd:ng the ~,ihling of the w.ife: :1Ind bmh relat.ions am a:syml:l"lim;rlic~l Gi\ .. c~ that ~ IS the husb>!m:i (If b,thel~~ 'tht rdati~fI. between ll' ~fld c r~hroth~r'sW~rif") ;,8 lel!.icaiized ft$ "ne'Vest.b' a.ffiLd. the :re~alion ~twe¢n

d R1:ld iii C'si5t~r's bW!bamP') as ·zjatl,. The CQmre:r5'C of , neves lb.' (~'t c)- iskxi·Cldi.Zied in. the Jisj,ul'lClio[l '0'[ "dlu'fc-rj' (, ~h1iBb:m-dl's broth.:er ") and " 2:010\'10:..'1 , (" h1l15h~ntl'$ G;is.t-er "); liUl d th.;!~ of I zjl!lW {d, Ii"). itrl, th.,c (:J:[sJuoetion of 'shurin' C"wiife:'~ brother") allid's\'~j~cil1jca.' C' wIfe's sis~er "}, .It "wiU be Qhvi~us that t!1il se'ns~ of the$cRu~siaf! le),;-1;:Jiltcs (as h ha.s, been represllnteaso E[tr:), .i~ more .readi.ly lI.Cloou:rtlted rIOt" ~ha!'l is the sense of "brother-in-lass" and 's:is:~cr--~n-bw" inE1'lgti$i} 2st~e ]uoduct of the senses of an ordered pair of asym:m:t'~dcal rd;ltions tali:CIJI from ;lI s~t mCilnln!l: {" husband.", "wife ", !. brother j', •• siseer"}, t[l.ch. of wnich is, in faet leldcaH~ed ill Ril SSi~fl, as 'i~ Engl ish, The f!l;c~ t);ud t W-o of thc preduets Co sister's wife'" and .,' brother'.s nlJ1Sband''') are not~exi·caiized requires nBc~pianarri(!ln.. But it w(lub:l h,we oor;n qulre. ,e.OIlcdv<lbh:" 8. priori, ltiat there shnuld be a distinct ennverse iDr 'shuria i! and 'sv-oj:3cci'~,i-r;:i"lI!, QT1I the Oll~ hand .!!f!d for 'de\·c'l"j' and • lliQIQvka'. on the other, just as i.t is ql.i itc oonceiJ.'13.hle, a priori, ~hat 'j male sib]i ng: or male" Mid ,. ml1ll1 5ihling of female·' s]wuM be dls~ t ung,uish~cI l~:ically. Both t!~e Russian and the English systems are internally oonsistc:I1It ilndisomorphic as- fa, as. the erthogenal rdatiO!)$< arc ceneemcd, hut they di {fer with ~espcct to the l·e.l!ic!l~ iZlltion of the ai'litlPQdaI Ollpositions. ~

- The d,i'5t~ni::tjo~ ~e!\\;liCll orthogp~al. and iln~iReq,l'Il QP-l!9l1>it;saba,Lflas j~~ been i1~t~te~ is nOL!C's~r!cl£.dJ in [lrirt~pI~. to"'ll;;o!l\'crnes,_' Man" is 0pp0$.ed to its lC.omr]ernent:ary ~ ","Omalll " en one dimension. anili toiss ,complementary' boy', on a[lbt~er di[ncnsion, ;]S.' gid' is opposed oril.hogonally '~(I 'bay· aoo' womao ', By virtu!! oJ ~his fact., 'man i : ' gi:r.I ' aF1c1 ~ woman" ,:. bOly' are al1!t:ipod~1 oppO~ttf$. "thol1gh. pres~n.fI1Lably ~n:r nea-Iinguistic reasons, they are less commonl}' ItD:pposecl in use, :But it is important to realize ~h:lt ahere i s n(jl~ necessarily ;JJ :~irLglc a",SWt· r t!{} tne qucsrr.!QI'!. "~Vhat is the 'Dpposiu~ of sucb-nnd-euch a lexrmd!' The t, Th= are also other diffc~FlC~~ bctv;,~c~ th~ ·t';\'"o :~:)'"!lt~rn~ whidh. wt wCluld.

neer! t;1 ~~1te :lJ~-CQUI1t of illl n !"nDT,-C ~!h==t~1l~d! 1Il]I8J~i·~. " \Vile'·s ~is~~~'~ ~'b~nd " is rg~"'~1"1 ~F!i3Jr\lIJt~ k,.lc~l mDogn iti(]Il i.n .Rwsaiiirl, Ihollgl:i fi<:i'l!l: lur dlj:: ®th~f thf(l~ 'Poa!tlhi'~iti{!~ Il ~dcr o<.gp'[n~~~"'()f4.;jblunJ::.lJf;'~pll!u:sc" .i!. 'iii,i'~vC$ ,of 1:l[1cthers ~ ~c ~i)mm~tric~~ ~~, ~dRl~~. ro Dne ~rludt~t in the ' •• I1cY~lklilt' ·;r,~l!llti!1f1is;litip. H Wi'b~:nd~ 0-1 sis;~~l's do not st'.a.nd [Ill, !li'lV lelli~n:i nOO$J;n [zed. rcl~tiollishjp. Funh~ ~rnnmi i t ]~n.o~ JUGt "i5;['G:tce~ husbfi~ld", btU ~ts® ,,, Jau!lhwr's husb~fld "', ~hM. d~ l!C:l!;cali,2;~d a!; j ... j 3tj " ~nd "~QTl'~ wife·j 'i~ in.eluded. in dle ,s.ense of·n.~\I'i:!~tb" Sin~~ "'hti~"hlmu:1'!; ~~t.h.cc" lmd, "',,!~f~"~, fath~r";"I!'e UJ~tiflJ;lI,Jj__~:hed ~c;c:'icany. ,~~ alliso ~.~~ "htJ!~bam:l.'s ~iat!tef" <'lind'" wire':r mother", the ~tllt~mcnt tlf th.c \,.B~t~US rum'~r~~ n:t~tIDtl~ 11iI:h".~efl ~~i~~ i;,r 1~"ftl-eS :lR qll ut~ llompit'J[: it is. hCi~\lgtrer. u"te!"flally ~jn~i~ten~,a~ th~ TGI.d~:r can ~"C'Lfy :fiH' hirl'!:setf. ''h~r~' i~IiI'" rc~~on W:;JtY' th~t • ;o.jlUj', ror tl!jlIDplc, ha~ tw~ ,ths{l~ut ~IlSe;!! ., SOf:l'S hu~~nJLI" mad Hbrol:her"s, h.wQ.sj[uJi.".

~rihog(lfl!l!i op'posi'~~ol'lls ilirC dominant i III the set f" man "1 '~'ffi'11lI!iltfl' ~

• b0y'" 'g;lI'r"l,. as the a.m:~i Il<ld~! oppo~it;();ru;arc domi.n~[It in the aet fJitoJ.h'" "scHluhl', I ,east j and 'I~,.est''}. II'!" {spring', :'s!JfilmCr~.," autumn", "\.':Inter' ~, • wi nu:~' ics mere s'lrml,gly 0ppOS1.."Il l;,u its a:ll!ti[}~dal !:lpP'Qsite

C s'li.mm,elr· ;than il: is M its. Qmhfl.g'Onai !lPPQ~~tlCS 'spwll1g'and • :lU'lU~1W. " b1J'~ !rsP:li~[Ig" .is no, more strongly o~:P'Qf:c<d Lo Lt s antipodJc ~ aut"! rn Ill' ~rum. h is to !!lumm~]"" :a:IiLJ < wi IlJt€'ff~. Nor ; s the d isl~tlot ~n lb.e1:we-en;: ortholl'Ql'lal ,<Ll'ld <m:ll[md~ru oppm~:liI ~!ays as cleae .i!:l the ~'Ocabl[j]ilry as might appearrto be the use from, th~ pcrhap6 r,::tthcF S p~da~, ~1'I~la~ces ~hs:t hav,f,l ~e~ nllenn!lllJed; arld:tbe interpret:tttio1ill of rome .of dle&e -;. mi.gih:t well bcrnaH~TI:g;e:d. We aJte less co.nel"l'l~ art ehis pOi.lxd 'lo ddend Iffil dis~lf.lCtf(jifi 1:1'1 par:ti.cul:lt rnn5~il:llOtl'S t]HlllJ. w~ :ire to' e]udd~e viH'itru;s up~C!:s; of thiC li'lotio:t'l of' k:.x-:icll.l opposit~C!IJI.

_ The ~,crm • ~~.y'\'i.'~; coined. inthe nil]l;!~ccnth ,centlJry~o des~ri~ ;Il pheacmenon, o'pp(l~:itleugs of ~'llea~irJ(g. whiie!) w,as its,cli OO]1;~ ceived as be~flg l1he C!i,pp@Si~e of synQD::4my; and, d:mrc has been a lot of cO!1lrllUiQn in i'iic[!1~nticsca;!!.sed :by the cammon p.!'llotire of trea~ung the terms ~synonym/ and ~:lIn:r~,~" then/Is-elves as ()p'posit.~. "~in;~onj'm:yi - (~n t.he hmadur sense 'of '~oppositenes[l; of m;ea.ning:') has 0 ften been tfuoug~1:t cf as reJerri rig to. the opposi.e ,~XifIfeme from udeflt:i~.1f of meaning:

i .e, to, thCi~i m.~m dc~c{) f di lft:ta1Ice ~n m.can~ng. D LIt this is obv.[ousl}! '''TOO~ in so :far iilS most of the c~rnp~,es of ,a:llJtOIly.ny eited i 111 dictinR~ ..aiics 3:n.cli h.:l[ndbooks, of Se:rn.aniltCSar~ ct)~"emed .W.hcn~7c_ CO'm.p.lIIl"C - ;and! cnn~t two obj(l~ts witH [es]l'fct to the! r [)03$!C S5tOJI or fmik of ,uric Of mote pro'pertic~. we do 00 gmcraHy OFl, the basis of their ~jm.i]il..ity

ill other rellpeots, WII! can My tnM )( ismarried m:l.d. Y ,is .'lingle" bu.t, in all Qlner respects slrni!'l!I:'. Ma ID'J'!'!2:., we eannot predicate: the wnrd

~ manied· ;il.Il!('I. "si Illgle' or X al'l d Y. unless il CC'i'tial!'!. nlillmbcr of O'the.r words are :Iho pl1cclJicll,blc of X arid 1', TIlLS nolt]s fur most, if [lint an, l.eJ.iical onposi tcs, ~[lm.itiD!le~ are d r:J;~\'I"t a1, .1g some dirncnsluQI of ~ simi lilrit~.

Now sonie semanticists naive r,o~osed, as ';.VI: (ljhall see in nur ~'I"eat'ment of CillmF"2:n!~ !lit unLE..[Im)~~, that the sense of ~I [ lexernesi n th~ '!"ocal.:fu lacy should. be dt:'5cribaMe in ~icrms~f ~t of bi nary oontrallt.s (cr. 9-9). This immJ'~s dull!t: cv~ kli.em_e~mn jle: oompam(lwith. e¥!ry mher le~",eme 1n tillie voCitliu,brv i [I a fI'Ill[~tdi mens~®in.al s~acl: li:tnjdi]nlcl in term~ ~Q~kiQfI.~- W,~;hi~ ~h;/1 spa!:.!!: ~ere wm be flymt:oous

L ~ ins,t:llloes fir (]I!!h.oglOn:d 0PH~_s:ition (h~M]~g hetwocn lexemes wn~ch are iIRO:l:lposi!ti!cm, 0.0. a sim,g!tc dimension); a.ndi wi ~hifi "·~fi(lll~' 511b!9pa.oe~ there wili be iR'~tances of 3.l:ltip:{Ida1~[lpositi(l[! or the kind mcmloned

iIlOO'l,'C. Rut "iI'lti~~! oP.E!s ~t!C8 like '~~th ' ... fld 'sortrth· ina ~WjN! imel'l~ SlOOllIl,_I_sulJ:spam_:il.llJd 're~ 'amI • green' ~un aJ nll~ee-dim.c nsio nal iSiubsJlare fif indeed ~hcy are 'r 0 l~e analysed kl thi5 war) wiU of eeu r&~ be identical wutn Fe&perct eo t ~ I.e P()~~.e-"'s'i[J'n or lack oi aU tbe ot!'!.e:r corn.~ uaellits of mca.fli:llg tlcli~ Eldi withi'Ul the total sI'~,:lJCe., ~~PPQS~i Jw,,",'C\·.11.j, that therewere ~ertaill! pairs or ]e::ilit."MC:S \-'!'ni.clt ~iff,er,ed ~mt f':.'cry climtm.io.n.

__ These m.;tht. tncn b [l d escri bed a5Mi ~ ~i.maUy d i ffiClJe.lU in meaning or absoluue~HOS;ltg.. (with:ilrll til.!! \iocabu]8iry taker] as a. wh~le ]exie.d !lys~em), and~:1l~v WG~JJlI:l(l in a. malol:lmai :!ntJpgdJI. ,oJ!'P~.siti:an. C~~'t:~ such a. d:l.5si[i~tion (If ]e:tem,es,lr wClllIhll aha be p<I~si bl~ Ofthe;[~ seemed to be any poi nt in. do]n,g 50.) 110 pose such questions I1JS 'the foUowi!i!g: I. Is • m.:a:rn· ItljO;.e s~mil .. r in meaning to ~ ilIshtiiaY' dum it is Ito

f boouty' r .. Now 'thcft: aJ"e~¥llsthat ha".~ been made flO:; =. ~u.ri:!!g sirnH:ari'tf of _!!lieanil'lg. (along it s>ca1c goins; fI'ClID. identity- to Inax.~m!kl d ilfern rJ p~) whkh\l'Ou.!d, i n It rincipic" answer ;such qucst:iorls. Rut none of tbie:nl, .in MI fiill" a9~hey ha"e been ilipplied" has l'iclded ;liny usefu I results; and .. it fS, dioobtill1 \dtethecr there is .afiy validity in till: :n.otion of a scale {iII simubr,ityal'ld d~ff'~ren~e of m.canmg!!.pplied~Q- the

vo ~bd;ary alO :!who,i~ :mclh;lnring' ag, ~t:!; t heeretieal cnd-poi.ms s.yl'llon.ymy ~nd 3~Iute all'tip(ldll~ opposition.

However that may be, tnc;; dist i,ii[t:tiQ;mt _between ,anl'i.PQd1!] and ortll:O~:_ ;gona[ oP'p'Qsltl_0n. seems to be 1Ip-p.Ji#bk;, and, u~fiillY. so i.n[I\:e amlJ~.ys;~s of pa<ticLllali h:lo!.ica~ nelds {snd\ as ~hc ileLid of kinsh!:p-:b il.nd t.he rewg;nl,i~imit offill'lti.~odal opposi~ion$ within muH~Jim.erl!lona&fidCls migh:t bti:ng t~e a~:dy.sj s of oert~n cyclically ordeeed gets. (lf~e}l>emes such as {'I'I(lrdc~', • south It' e~e.'~·¢5el witnil1!th.e soope of ~om.po.""~llda'l semantics, and cy,en slIlcn 5!C~ as • hlack '\ < whi:1e I > ~ Nd', • green '. Jydl:tiw', "blue ', which is _par!!y_c.,n:_liC.1'.

". "._._ ,

9· 3. iV(m~bbillry ron~rast~'

l.essnt:ltd bt)~a!id a~bO·iJ t n®n-bln3!r)' cenerasts o.f SCllS\; (naUil a~'l.u oppoS'llion. It seems clear lila! thc:re aee such eontras s, c.v~ni[ m~ny l!J?E:art:['Idy fiQ.!!-~binary cngtr~ts (for cX<'lmpl.e, the CQlllfas~ wnidl hold bc<tw,~efl eseh y,f the set of co kll.l~ w",[ds, il>n.G:ll'VCry other mcmbciI" of the Sd) u[~m~tdy [li!'OV':: to bc al1at~abi~' in rerrns OF scnral hiaary d:~s1:inc.-

- ti:a~lS.~t, is ham. t~im<'!gine tl:i:lt e SMllld:ay" "Mo:ll!clay'~ .. '" ·Sill!tu:rd:af~}. ~ {' J anl]<lIry', 'Ft:bma:ry'" .". l ~ .lDecenibcr' '} '0'. ,even {' rose '., '. peony>! <tu]ip', ~del,!:ilh.inium'T me.} wil] be Sll:tisfact:OF][Y ;:r!laJY:llabIe in this wily. Nor dot,S t~-c'l'\e, ~r]1ear to be an}' go04 :reaooJ1 lor bd:il::'''lOg 'tba~ m.al'l}= member h:xLcal se!s like this ncc~ssa:r.i.~y b~lonJl:~ to lloom'Qsr~cidiZi~

~88

9.,3. J!lml-binmy C()lluaS!ts

['C'Cnnica] or dCi1Iti:ti.c subvoeabulary rather than I'!III> the ~neral vocabu~ilry of ill langu:age. The most that can be ssid perhaps is that the Jl!ibo[:l:-tiofi (if mao)"-mcmber lexical sets (e.g" the words uenoti.llg the elements

i [I chemistry or different species in 'botany) i!.!_llotc _typical dlhpr.dal.izcd t~tln()mi es t~ it is of la1i'l,gulI:ge in g£n~£!Il ~ an d it is n.otc\\'OrthYI hat

_ _!Eeciaht:,::d .~'< llll-."lie$. even when they nn ''''' I1S.e of ,everyday words, will orfte iii ire mtd! w,imposJli!' upon Ie)ol~cal !idsi 111 the ~a"'l1Im:amy a. mlH\e rigid ,elructl!1~~ than ,is Il;haracteristi,c tDf the vocMn:llary o:f cver!i'day u~age.

_ The n:kUioushi(! (If S!':1l5C wi1i,eh holds between lexernes in manyrnern her sets such as (' Sun !la}", • Momla),', "_ ~ ~ SauJl\day "} Rlay be described a.~ i:ncom ,atibilitj'&. This notion is diffil:::lIlt.to m:ab as precise as the notion !:If ~ o<siriQl'l. R ha~ been pointed out, fo.r example, that

a dlen lIIition of ~n.com..[!at i1llili t ' in terms of contradicteejneas run,~ into pr'Qblems,: ·~X is arose" implies "X is n~!__ 11 peon.Y/tulipJddphiniuml (tlc!'; but ""X , .. rent there ooSat.urday " ~o~~ not imp~y "X d4d Dot gl:F-" there on SU[lclayblF:l1id[tY" tine! j,j mn p~mched Mary'" Joe.s not ~mply "Bill did not kickr~l;&p II.'lrny" (cr. Lehrer, ~973: ;as). This is true; lind yet it is evicl t:ll't thll t thcr,e .is 111 reIalil:lllsh ip of ~nmrtll!atibi litr-

- holdL'!fl: within the lexical sets in lIluestion. Given that X went on only - one da (ill' rhar we are enqlliril'l,g. iilbOIJ't one occasion of his going); if we say It waf m'l Saturday thtll X 'weill .fJU:Tt we wjU nQtrn.all}~ he ReM to b,,1I\'C said ,som~thi1flil" which imp,l;,cs "X did not go there on Slling.ay". Si.mi'11II.[1!l·, gi.n::n that I1il] st ruck or hit I\la'!y i Iil one \\"3y rather than aMther", ~f VI\C say !!m.l~ll&he!!_}!aIX (""jEll heavy 5tH;!>S on the verb, marked, here with an ;;l,C:U~C IICCCI~t)< we \"I'iIJl norm:).lIy be hdd to haall'e said something which implies ". Bin diol'l{fit !idrJsia:p, MflfY. ", X'~ ilDing_trn - Sfl.'turday i~oomEati.b~ with hi.!! g,a Lng on Bunda)' or any m her day, not in the sense (ku he could !lot have gone both Dn Si!.lurday and ab~ on Sund:l,y~ but that he could not have gCHIm on botb.:-Saturrla;,--and-Sun-

.:;:. day. 'There i~ [he 'funi-le(' diffit:lllt)' that. inc:orr.?lltibilit~\ ,as 3. st C ural relation is,' r.ot .&!.W.l.VS dlla.rly d.isl[l'ignisia:!bk pr,c-lbe{Jlrcticilly at .Iea~f, from. whaJt\l'C ,,"ouhi be inclin it to describe a~nrel;n,etl~C5s of mean=

~. nlwt we will not go further into this quesLiDI'.i here, The important point is that ii'!loQfflP~tihililylL~ II Icxiai rda:tlo:n, ti~e op~tionis based Oil oontr~";t \~thin silmiiarity: • I'O'SC , and j'pLg' aJre contradieteries, l:m'~ there if, liu'~e po~nt i Ii d~5'C1I5Si fig t ~H1:i'r stllhlS as: i nO~!Hnpat~ bles, SlnO[l the one denetes a flo\'~~r ,lind the other an O!.nim;lll; and tile sense tlf the

one c;]n bardly be said to delil1l]~t the sense, of the other. •

V'iI'riOUS kinds Qf o:rdl;ring3.fC found :LIt man),-member sets of incom=

j.?3.t'ibles; by a many-member set, ,in thi,s co'nrext. is meant II set which. contain!l more than two lexemes Such sets rna ~ seriallv 'or c}rcilica!y"

- orde reil_ T J'l a ;cril1Uy ordered set there aretwe oillte~m(l~ t rnern hers (if - tht? set is deterrnl nate), iItJ.d all other' Iesemcs in the s.el are on:len:d bClwC'C'Il two otbe~:, in ~ gel ieaU.., ordered set c\'cl1'tr t. lC is ordered between t w~ others. An;wng s.c;"r.iaUy ordered sets; sE:Y"~~' ;~ h.? dis~

t i rlg~!ishcd from nn.ks ~ accordi fig; t.owhet her- the cOl'ls~ill.lcnt k:xem('s

aT'!! graJahk Or not (cr .. Lehrer, [973: 29)., Th'e o:rderif'l.g in ~cs

in terms of ineom atib3lity is cha:r:achrristkaUy less strict than ~t ,is

in ranks. Conside-fl for example, the ~t {'exl:,CUcnl', '. gond J, 'fair-'. _ • poor ", • had'. 'f atrocious '}. -:. - . ~ of aU. i ~ is somcwh:U W[ determinarer Sh01UM ,,"€ add to it 'euperb I,," ;W';'!llJ', etc.? Secondly, although,we '~'ouldl probably agree that they can be arrlll1g\."11 ~n a ~c' in the oider in which they have just been listed, it is only when two Of' mere of them are. '12 ~plichl}' conrrasted in some ~:l<ll:rtkQ.il:arooIrte:'i.t (cf. Slte'& not {j!lst} 8DaJ - Bhl)',~ (!;w'eUrn:t}, that. they are tak~,I'1!!$ inwmp<atib~!!? i'urt~lennorc', \dthi n the set we can i.u entify the anlO[!!!!!p~[" < 00&': j b~~ ~ bti& stvlistically rnnrc neulrnl and '!!_~rha!p~ more me.1l in ~pI1cab]H~y than the Qther.s; ~~ tI'ds is ty'p'lcal of lexi ca:!sCilks. '\'hat:lre hequemly f,egnl"ded as stylistically II::$s neutral, more 'cm{lth,'C, lexemes, c.g., "e ',cdh:m! 01[ fatrociOlJs'. 3<t'lC p rhaps descriptivdy e'lIl1f\'a,lellt ,to exp]iddy grndctl exressions like "'fry gQl}d' Q![ "very bad'. The!;cak [,blllt" c''''ann 'J <cool' 'col,(PJi'S Tather unusual in English In that lit: contains an {lute. and an i finer pa~:r of anton.,lflns, "hot': ,I cold ' and 'Wo'lIntI "; 'cool', Howe .... er, as Lehrer points out, when they al'eiRl~ licitly - grall'cd wilt h F'e:spect to some temperature nerm for' f'Ood Of' we.:a:thc:r

.:;;::-t11Cj'· or centrast in a way simila.:r to iflct]!!1~~~ble terms.' (~973~ 28). The 'I)nte'rm(l~t me mbers m II seale (e.g." • fn:~z{f1g' and < h[)tJ ing ~ i~ the set {' boi Ii i'lgo; • Lwt I. • warm i, • CO(l t', 'cold" 'ttee;r;i[lg '} may be described as scalar l~l!1!Q_sit,~ .

Ranks exhibiL tilt: nrinci Ie. of 'Serial ordering i!a :)l s~r.ch, e f!ln'~; but the}' arc I~&'l cha~;lctt'ristic of the nee-technical us!.: ' ... i '"In' _ ''{r·_ Onc (If Triers e:;;m~plcs of a lCKical fielt1fa11s into this, 'CJ.tegoty: sets of lexemcs used f{lr gt"ouping c'Xamimttion candidates acoonJj'l1g to their performance. If the eonventien accepted by the examif'ler~, i~ that every ~ caml uttate ",'i.1l be cIass,ified in terms (If the !!!!!.J' excellent ~ I • good • if

·aver,:!~'e'. ":fair', )loot'], tlil:,t?s!:_lell.emllcs wiU hc c~strucd as a ~uilill'y "

I] rdered s~t 'Of in C\Jm2~;ibkarni Lmgradable '~erms '; ~d the sense () f any lexeme w.ill be determined by' ~ts posltion in theraek. TIle set of le:x:emes used 'to dcsesibe dilfel'em;es ofJr~ilitary clinkJlmv:ides another example:

within the sci: C field maHtllll "~ • generaill:', .. oj '~()rp~r,al'. • pri vate 'J" QS 'th.cy !!Ire i!ppHcd with r~pe~t tQthe tlritish lIIm1Y, the outermost members,'liJeld marsh-al' (a phra~d ~eliem.c) i1.od ·priYl'Itej• are; rank Ci'p,l:looi'~es" Num:cn~5> ill JE,n;gli.!ih fon,e~, I two'!", , • t 'Ctwclvc~, ... , ~ l1u [!dn~d •. j • '~housn:ndl \ < milli~H~ ",' billion ", ... }. also ron$,ti.~1Jtc . .aranic, which ~ the .i.IlIu:!\c.s1.~ng' propen.y tha:~ h h<l:~, it& ~\~n ~ bgr~1ii[m:~ lIClcDrd:ung tllJi ,,1lk!h.lUI infinite set of lexically 4Orn,F[exe:orpfcS~iQL15 Ulii:y bl) oo.llstru,ctt':J" it Inay be mentiened inpass urJg that. th~l:e :I;!'C i nh!'re'd ing fo·rman dilerenccs in the subgm:rmllar:s of ntlmern~s, [ound ~[l diil~renn h!:"gtl(!"gcs;. and these havc{ bee mt IItlraCling the attcnti~n OF Jingoists recently (cf- H1IJrfon:l, 1975).

The most obvious examples of ~ical sets, o'rcyclcs,~., arc to be round! ~1I1lpng: woms that den~t~ HHits er p~r~Qds of time! [,'!pring., ,_

~ 5ummer'. ",autUl1Iln', • wi I:!l~1" '1; f Jal'l.uary '. . . '> 'D~~ mbl!l'r "J ~ e Sun- "L 3 d<llY~, .•. !I Satyro'ay~}. 'Ihe3c :J!\e 'iifOrdered in t errns of stu::cess~

l'llBna~ the :LIlI:JIIJ";J:i!C;it1(ce~~r3~lmcdi;lteJy precedes sl~n~m~r ",

'j Satu:rday imm.ediately rol!o\Y~ Friday"., " October is, between ,septem-

ber aad NQvem her' (c;f. u~C'h, ] 969: IIO) ,Un !ike sca~e;s :linG! ranks,

cJ!""ol~s d],Q [lot navl!.1 ~1JI,tctmQS't members 1 or e'NCtTemes: ~"cen: mcrn b.c,_:_of

'the set b erdered bet\vecfII_t\"G QJIllie<rs. The fact: that ~hcill'c is a. conven.: -= ",ilonal] fjlrKland t~t: ~nemiber In. m~y (:If~h~51: ~,~ts, {th."l:t 11lnu~ry ~s the

finl,t mo.l'lth '!,!liT d'le y~av! Sa~urday the b:st day :of die week.ctc.] does not. diet'r,a:ct ~rom thei r eye! iC:l~ it}':' I J oh~ 'came on S,lu'!lr(j:a.y atl~iIPir~

on I[be foU'm~,ng day~' irnp~nes ;·Pcte:r C¥IlI! {Ill Sunday'" and "][0]10

Ume On Monday, butPc'~c~ came on [he p:r~cr d i!li~ day" :t]SD impbes

"" ~cteit alIn£ on SUfid3Y"" On the 10thci': hand, it must be .re~~gnized

that alternative lruerpretarlons of phrases like 'ncxtF~day', "last Thursday' Co Fr'Id3y ,(If next \lI"ef'.k'·: "the next Friday fullowing ~)(H:i:1yj',

•• 1"lU;Lfsday of lalit '",.eel.:: ':;; ,. tile: rn.Qst recent ThuHdilY·'). which" a~rding. to the diary f,lf the week on whrd~ they ;]J.~ u~ttrn:d. may di ffcr 1

in re:Fe1"encejre~t upO'!'I!1l potentia] eon m,el between tm.t:iing d~~ s:~ I: as t

a. ~]_~_al1d !!lk~[!g ~~:I!_ ~ seeies,

F~rmHr,i IlJ ~hi:s br] ef d lseussion ()fseEal afl:~. t"fdic<ll sets of lC:XCl"ll,cs,

it may be [l~~r .. ed that ~oth I!rillciple-~m~y be o]er<lttve "vithi_n the

, S{llIIIC lcxieal .. fidd~" Witfuin.~hl:l ~t of basic ~(l']i:}Uf" \, ... oeds in E;'~g!u~h 1- 'S (' Mack ", e :grey', ! white' j, oons6t Ia~C a_ seale ; and {! red', ~ yd~'o'w '\ "<=.

• gre,c'11I " r blue" 'pu:rpk'} ~le. It has air{''11ay been mem iOllJcd that _ both • black j ~ , '!o'dli'le' (~.t IheeN.tremi t.i~~ (:If a 50l!h':}i!fld 'ret.l~: j groen.'" !j1cllmv': Ib~ue~ 'with.i~ 3, c)'d~) calli perhaps be n'!gllrcll~'Cl asl anlipod:!~.

opposites. -

9· 4·HYPf1~ymy

NO' less ~mport::mt dt:all [lppositiol'l a~d! w!!!!..ast as a parad.~g!!l!l\'l:i'c relation of SCilse ~s the :reiati.{l:t! \\'h:~ch hol,ds between ,it more sped]i{;, 0]" sub'llroi~ate. lexeme a!'ld a more gcm:ro_1 'llilf 'St!!'JeroL~liil~,_ .Ifxs::m~, as_ exemplified bY:ll'l_u:;;lll pairs as, "cow": "iltJilirud', 'roo,e~: <Flower',. 'honesty); t~'irtllc:J" fbuyi;fSl~~\' jcrirn~fl'n·;'red['. There is no gefler.aUy aec.cp1tcd. term fer this rdtuio[l (Oil" .its, conveToSe).]h:cl;lltljl'.how,c~e~,lh~term hyp,oOO'm· (corn:cd by amd.ugy with • ;J:nt:l:mymy' 1ILIld • symnyrn,y!) has been gQiO~flg currency; aaditwoeld seem to he mere !Ii~pr(Jprii!,~e tk:l11l such all!frll atives 11.$ II ~~dU!llOJiL' Of • 5u.borclinlltion ! "wh.j,ch. aJile also< used ill o~ht[)r S~'Im'.lC~ i 1'1 linguistiesnnd ]egk Let Us say. fh.el'l.tt:iat, • row' ii~ a l'!iYp'~lTIym· of 'anum:!!l', 'rose' is a hypu.nllijl'iIin of ~ti'O'lnlr\ ;m,(i &0' on; and further, th!Jt (rnsQ \ <tulip .,' daffodi.l', etc., s.in,ce ,ea!c:h I:S ii, h)'ponym ~r '~flower~. :rre cO'~hyponyrns'· ~of lhe ~lm:l_g lexeme], fie ebvious Groek-bas~d torrelativ'e term for 'the {;ty.l1¥'CP.H~ ~b'l:i{!lln, "hyperon}'rn:y" (ct, Mulder' &, HeNey. ] 972'),~S unfortunately l'();f) sirni]al[' in EOml to ~ hyponry.m.y' and ]iket;r to' CS,tlM: COl'lfIJSlOJ:it.WC'i"i]l use, instesd super ... o~ i narien II, ,.'hiclJ., 1Ilntike ~ subordination ", is .not widdy 'Emp1ilyOO: as atcclmicd ttetmi u lingtllsfi,es with a oonfii,oti og sense,

HnlOnyroyis frequently distlussro by Iogieo[1I.flsiin terms of elaseindus.ionat (cf 6-f); it!~cl,u:p to a point, this is ;a~Macto:ry enough. For c:i!;ample,if Xis the eiassof f1hwcrs and Y ~s tbe class of ndi:p&, then it is in fat'!: thc~c tlIllJt X proper])' inelndes Y I~X:::l Y &y :p~.B~n there are probhmls ntt:1u:::hrung ~Ol[lI.'f def.inilkl mt 10 f hYPoflymy i n ten~s of thelogic of classes. First of aU, it ]S 1)[f1deal"' whel'h,e~ we !!;houlJ. G.:l!ytliHllt ahYffilfl[lym 1$ i flcb.lJcd ~nLts supererdinate or as.ul::terorcl!i naw in its hyponyrn(s). U \"'CJ ~~H~i der 'lhiccxtensiQn" ef l~N.CIrr1cs, we wnuld say tJ~at the :supe'rornilllJlte leserne ismere [lldm;IVe i but as fa:r as the intens1:Il'IlJ-OJf [exernes is oDlilce:r,ncd the hypo:llynl is mnre inelusivn (b~iip& hllve an the defini rIg p~1fl perl ies tC)f .fljoweno, and certain :additionlll pro·, pert:ies which. tclistin,guish them. EI'~nl reses, traITo-dih, t:~c.). This fae~, i rn ltsdf,is i:1ot part~tularIy trou bk$[lme: it is, 'i "deed a:l. tor atir h logi.c tl1a.t eztenston \ln~ 1l:ltenslM. sh~mk~ be 1"~lalt,edJ in this, W~.~' 'J.IL~" _ .]fldli~ sian. More se.nioll~, howc"re.., h n~~ probl.e,m. that d~s"lo,gic do-es no~ seem to b e ~1!J~tabl,e for the fQrrrLaU6.1tioll ~'f S!t."m.:mties, unless we make ratru:]' eonrrovcrsial assumr'~i!Il'll1s about ehe !!'IJterddin:ab.iJity (!IF deMta,'lion. and sense: s.~ying, J)~r example, tk:u d.e dell(1tat~fliI of "flo,wer:i is a class of obj,ec~s and its senseis the defini'UIIg prClP"'1'I:'~[es of th.e dass" This is lk~ 3ppn;;ac~ follol'l'cd by such. schollll5 B5 Car.Mp (1'956)., .Du~

we Sire usi ng (he term '~denotation ~ fur the leb.ti.DrII which. hnl1d:!l hetw€.e!1 I(.'..)OOmes and ei.th.!::r clasSc\<; ,of i IJ.di viduals O'li"pf()pe:rtic5, Ilctu'litcies, pl"l:leesses and rci:a;tions, as sec~~ !!pr·roJ·!l'taf~: ~n pu~icubr Cl!;l;CS [cf .. 7.,4): and we have ~;eded th~ $S!lltlp,uioTl tn :tt" ; f al){]xem~ hu eense, it must IILsC! have denotation.

H yp~l1iymy is ~.cfinaMc in term.!! of uniilttcrnl. imnilication. Fol" exantple') -(crirn.g~-i~ta!Mishcd a'S '~ fuyp'Onym. of ~[~d· ~1li]d (buy' as i h:yp[)~ym of ! g'L:"~ 'I by "droll: 'l'Jf ~he impUca:tkm:s .. She \,\'3$ wearing a. ,crimson dress." _;." She was wearing J. red d ress", n:.IT bO~lghti't from :i! fl'iend~'_" '~I ~t it from a friend" (i.e. betweenthe proposi.tiocfiS ex.pressed by 'the 'sentences. j She w.:l~ we.a:dng a erimsen dress' and 'She \\'lIS 'wdnng: :1:. rOO Jr,c55 ", ete., ",11c'l1I theae sentences are une roo. 'tio fllak!!: ;:!.f1 assertion], The: denfllitl(lll. Qf hy[lQlloymy ~n terms of unilateral implj~ ,caJt:ionerl;libI~s us to dc.fi ftl:syno.[Iymy" as bilateral, or sym:m.cIl trieal , h1'POlmymy~ if xis a hYPO'IlJ"I1l1 ,~f y a.ndy is [!.I'ryponym of x, then x aIDId, are :5yn®nymQUs,. If hY!Xlnyony is, defi~e" as non"'iS;"nUl1t'trical {~sit must be if synom:ym),! il'il~at;oo as symmetrical hyp-onymY)ltfien proper hYPOfiynlY· mi'l:y be d ist[l"Igtii shed frO:l:ll synonymya~ being asymmetri'Cit] (for the du.ti IIIctiol"l between nen-symrnetrieal and 1!symm.etri cal relat~ofl.s. ef, 6.3). This distinction (If asymmetrical hy~nymy a.s iii !lpecia1 case-of non.-:s:y.llIIme:tric:il. hypon~!,m.y isanal.ogoU5 with 'the: standard. distinD1ill:li'l ~I propC'if i1itdusion from i uelusion in 'lite logic of d.1SSCS (cf.. 6.4). Thee IJ gnou.t this seerion we sh~l be CO'ltcerncd pdm~tily with ptoptr hj"P'ln.ymy, Mid we wilt usc the tcrm~h)'pl)[lymy' with.out qUOIH ficat[n tl in tb.l S sense,

Hwonyrntis a transitive' rda,tiOI1. If . .,.i~ 81. hyponym. of )1 ~nt]y is m hypm:tym of;:", then .x, is n hltponyrn of z (€Of. 6.3). Fot exal'Iilp.!C, <en,w' is :a hY]l<lIlym '(!,E 'i mammal ~afid ~ m:immal· is a hyponym ,0£ • :lnima]" ; therefore • coW" I ;:s 111. hyponyrn u.f" ,animal'.

'Cetn,e_rnlly _seeakir~g, i~E[I!;;J~sh, when. thll r_~I.1J~{):i1J 0:1' mtj·p.olif1ymy h,oldJs bctwc:.c:n nO 11115, it as possible ttl i;fiS€lil5Yl'ltac'l:~eaUy a.ppmprJ;lIt¢ el(:pr~sioru! rnntll:~n~[1g them in p:~lICg of :~ and y in the fo~iowing fQ:r~· !Til.!la '!.:t is a .k~nd of J' ~ (whcT<!~ ,~ 1511 hypaM:Yil110F):') ~nd this will yi.cM a sentenee whieb expresses ;1, mcta_~lngllist!tc or i'eflie1'!:i,,'c prop~hlon \\"hid[ (w t.h.c d('gree. tbn.t !I.m)' metruinguLS'tk pl1opositkm rehui!'lgwi natural liln.,g:u ~ges i 8 S!naly'!:i c} i ~afl:;j:lydc (cr. 6.S). Thus :Ehe pn:'POll'LtiOI'l eN.pl)f~ed by ' . .1 cCiwis a kmm of lllnimal", ; A tuLip is .:11 kind of!1Gwcrj', eh::.! may be trut.'C:n!I to hCaD.I!lyrt:i.c. Under 1110['(; :£I::strlctt.:tl co.nd Lt iOrls, < sort' and < type' may be s~bsthytedl for • kind' i!"l coJ]f;lqlJfaIEllJgl~ ~h ;

• Aoo';v is, ill rort (If 'a~im::!.] \ 'A iI;U~ip is,a type of [low;e,r ~. ':rhe;;ea,re m,arny

other mere slliIDcifi.c kxem.C'S (whi eh llf\c,themselves, hyponyrns of 'kind' wh~c:h mny be cmv,lay.ed, f~r~e'rti1iiI11 vi1Jlnc$ of;'I; and y: e.g., "shade" i: • erirn~on: u~ a ~l~ i!.de~,f mO!. j j' ' m:d;,e· .~!i\ 'An A'~lfIi'! Mati! i n is a. .lnikC U[ca:r I a:o·(:1: :E.~ on. WhJi;11 a fUILlil ,"I: ,j s supClrnrd i1'1,a:te tn more t~'!;'I'o' one hypol"lr~ J', =, etc., sUich e".:1ipre:Ss.ion5 as the folIu1wi:rng lI\rill bel aceepted ali meaning f1.11: 'CQ ..... s afilJ ,other. (kil'lJs of) ili£lirflai5", • bJ.l~p..l! and m~cr:r fki n(ls or iQower:s "I\,'h;ich mtlty be oont[,lls~C'd with th.e sim1=tica.Uy anom:liloQ'U ·C'DWS' and other (Id~~d!l of) lSi . c ..... ers ' an.d ·tuli.i~S aml other' {k.i.Il(b; ,QI animals ' ,. h ma.y b~ assumed ili:M the fre'q:llcl'I t eccurrence .:J.f 'sud cxpressi.ons plays an impor~altl!: Pll{tt in d:l~ ~~t~Mi silm~nl of hyponf~ end oo~h3T{)nyrny in. b!n;gIJJtge -acquisi l~m"l; ~nd it; s imp(lr:tll.l'lt {o lfI.!Jt that <OilliW c:ln.l~r~" in, rhis way, llt!!:t one: lexerne is a hY'poil'l!~.un of anothe or '~han: two ~exem.~ arc ~Cli'"hYP9'nytms Whl'lOilt if! pnf!ciple knowllil; a[jylru~ Il'II':i'rC Qf t.h.ci. m.e.an~ ng. lndecd,rnuch of our kJl;{f,,\;JeJg~ 'of th meanI]Qg of words in cur native hflg~age m~y be ,oF tb~s IdnGl.We m~gh know, for Ie'Namp.le, that: "baIliyafl' ]S 11 hyp'o'~ym of ; tree> or' m.pl'ey > 01 third "and yet he unable to Siiy how ba:f1},a:.l'ls dilfier from other tree or O'spreys from other b[rds.

NO\l' ;,t is ,,1$"0' ~t"l be observed t \:!.~a q uestion I iliie What klnd 0'/ anbnc; :wag £r? (put, shall ,~7C say, by a parent to a ehild after a visit. tothe 2.QO may be answered appiropriaJtelywlth ,~it;her An .e,lry)umt ,l!IIr A W:U Oil (where <QtlC' m~,y be th01.lghlt of as a prnnom:iilli.l. ~l.]bsti1i:l1te fo • anim~ l ~) .. Tbis would snggest dlat,. in m~ny (JII~C5 ilt lcas~, 11 Thtyp I:!lmy-.. C'nru:pswilIt,C,., * the [{:1ilSC of SOMe adjeC:lh-il mod (fier i'lilld cnrnbines i \v.tt h the se.I"I~.e of the. supeMNlin1t~e lexeme (d.!L 5). This, d(l~~ no mean that Ihehy poe nym is a:]W1lYSe eq ~hmlerlt to,c:J.f S },'lm.(I1'I}'mous l>\'ith ,8; pht"'.!sc i 1"l\1!'1iII.i,ch ~he 8(1fcrotd ina~c Ic~ern~ is modifi,cd by means Q one or rome :l:djroti~·cs. In so.m.c instaaces tnis may be w:1 tyrant" i a hyponym of' ruler " arid' de1.put.i.c ruler' ~:r 'en.wl .mk:r" .is perhap equivaklllL[ (IIlI many CQ;lIILh;:>;t~) to 'tyrnrl't'; and itma,y Wof]t be that th, S<::i!JSc of 'tyrnnl' ~!I oftc·n ]cRmc.d hy 'l,'liltllC of i ~5 cqukalcnce~(J one I1iI other of these p~r;tSes" the sel'llseof < Cnlel .' er 'l'iesp[)tle" and ,~J ~rul~ [ being known in .uh;mce. nUll thesense of a word like "(OW' ~s M.I:rd: not learned on the basis of irr~,~qui1,'akll1ce~o 11 phTa~ellke <bw;'in' :lTlI.imam j. The pro~e:5S is much more Jikdy to be t h.efevef'S~ Our U rider sta.ndililg of 'bo,,-inet win be depel'ui(1OlJit, upon ollie priee kno\ldeclge (I what kin dI of i!.fllInds eow& al'Ie. 'The point l~hat is being m~de hel:t i s:~ n:I ply that, for·rfl:'!LI~y ·mtntm~ at least~ the' Se<ll'~e of ~ hYPQ~~'1liL'I can b, 'I'cgard.ed as the ~r'ndnct of the sense of flf S U p~m]"di nate' nOl1[n an.d. (I lume actual me potm[~j,11 .adj~cri\o'iI~ modifier. The ilIPP ropriatet1ess 0

I

F,'!.

... '!i

~. 1i,

, ,

A' tyraP:il or A, cruel Ol1e' .~Iil reply t~ the q Eltti:ol'l Wlia~ Ainil oJ ruler WQS ,x? . refki;t!>t~ IS fact.

"Vc~b&, Ildject~~"es.. adverbs and olher p;J.r~~ of llpeech cannot be i 11- 6erti~cl i:nto the fOilnutm ':It ~s a kind @r y ~ withollt prier nOfl'li fla]]Zilt ion 3 (ct, 10.3). and even d1!e[l the resultant semel'u;e is generallyra:~hc]" UITf;lil'lIJr.t]" :~r not ;Jbs{dutely uaaeceptable (e_g.'Uuying is, 111 kind of gett:~ng>}_ 'But tJ-H':F~ ru-e' od:ter\lI'Oros and phrasIJII"'" lm:ich serve '~~ stt'LI!C:~lJire (I;u~ v,ocatn.~Iary'~n t,erml!! of hyponymy fOf ~he 0ther parts- of ~pe;eck. as 'wilLilt kind or ... ' l!ioe-gbr nON!m1l1. Comp;J!!lI,bJ~ whh th.e Cjuestions HIJi~l kinti Dfammai !)YlS it?, lind, Hf'a~:i"i ~'eO'W aT Srl!l:liUJtl!~F kiJtdojan.l:mal? arc How did !reget it - by buymC 11, (11' stealirrg i~? and! Did Ire; O!fY ~'II(JT g€~ .it lin $i9mt other way? Sinllb.rlyfor :tcljccl~l{~'S like ~ fri.enJly· aad • nke' :

T¥h,ffli }"'Oll say ru' & m:a., dIJl Jll.'m: ml!!-l!lIlI th~t he. I~ /rli!n dty £1.1' w:u. J)f so,me QII!t!r U![ij' '? hs W"C 'UIJI SiiY A CtlW' i& m,1 lam/mil of (iI c~tllin ki~uI" 00 W~ ci!ln Si\y though. perllaJps less id!OIDOI!Jticany, T(l btlY S_~#Ji~lg l~ ,IQ get iti!l ,~ ctJ'tait,lK'tij! a.nd'l'o fJ,~ friendly {to 5!l!.l:IWm)i i$ to be 11.iu (tQtO'Nl'CQIW) rlt a c:~r-tain WdJI,

In gen.e1<!l as adjm::t.~Y'1lL1 modification;s tl(} R(lU I'IS, so advt:rhiiiilmo&fic:ltl('lrl 1~5 tiOl.crbs and adjcodve5, and .iI$ 'what kind af, . ,,' i 51 aaswered by an adjenivaU;· roottified :mlllllJl . or a hypa{[Iyr!l.lolls nO~rl, so <hm,,·' 'Ol" '~ilil \.4n:at w:t'Y 'is answered by an 1iIdverh~allr 'modified or hypOl1ymQ:llS verb {~rr adject~ve. There are, however, many dliffl:rcnl st~bdasses l(;If athrt:fbs; ;l;l'Id • how' or < in ''1;'h~t wail' !is. [lot ,ihmys 3_P'propri,a:~e t"!$ the 'type of a~;h'e:fbia1 rned ~ficati'iln dm! is inn Ived"i't wou~.cI be t;f)i1[OUS, eve n, ; f it wem f~a;si'bl,e, to :tttl:mp'~tD I ~Sit here ~I the \vay-s in which hypoillymy is raani feslt for the dilIcliffiJt P:l!l"ts i()f &pe,ech. and SltD clas~ of them~ lor the use of int,er[loga[~\',e w{I:rds ~l1d phrases an.d of ph rases oollllmifli.ng ~50me atb<~r~ or',11 certain' (oompa.rahlc willh ~'SOI~, nther 1k:ind~ and <of a certai [I l;,ind '1; and lhe fuaeticn and dismibu;~i an of these words iLild phrnses o(]1L1ld nO't be ~a~isfllctorilyacooll'lil1:ed for exoepJ within tne frsmework of .:1, com:prehen$.ivc gr:ammj1Jucal dcsL:.1'ipt~o]ll of the ]:liJ!:11!lillage. We. wiUassum.e, howe"~r ~ that 'the :gen~r.rlprind~le is clear, HYPQrlYm.y is apar;:lldi @:~QJtilC relati®'fi of &eflse\<\;hicb '~&ls: npolll the 'I:tlc~ps'llla .. ion lnthe hypony'flil of .some s:ynt:a:g,mllttic m{ldi~cati'DI1I ru the SCl'ls{to£ Ih,~

sup€mrd~nate Icx"m.e.

G \>_(i\;l'illjy_5~~;M._0)~h;'!i'p-orlY!ITS of the aame sup c'rordi nateWiHw_!ltr.ls,t tn gell~ (we win u~mp!lr.lllly disregllird til!; prn;sibil ity af I'IO!'l~ClQIltra!ii~Dng: syn~fJ:}1'QOUS C:r:!-ilrponyms); and the natur't: (if the Contrast can be r;:J.lp~icatedilii terms, ,of :I, difference in the encapsulat~d syn.tagma~ ti,o mooificali®:lU (I.r thee super@!f\di[latc, ,POl:' e'.'I;lImp]c, 'buy' <In;d',teaJ;'

arc in 'COlltrast .• as co-hyponym.s of ',g(:~" in sucllst:nt.enu[t"S as 'x bought th~ book from y' :'Ind '-v stole the b(JQ}: fm~tJ y' .

..... H~wiflg ~sl;l!bjish1l;;d that two ]e""'em.-e~ an oo·hYiI20Pj'ffi& of the same s:u~crord im.atc lI]]1,d that they are i neontrast, we- cal:! go on to determi ne th~ nature ~f t heir co nl~l5t by sp.ecifying the n:l;!tio[1Swlhich noh-1 betwsen tMe:IH :ilfid-o~h(;r ]C))efnCS under :implJc:r"lfon. [[I some eases, but not aU, their contrast ~Il seiilM: can be aesoelare d wit h 'iiI c~~trast between ~ \\"0 sYl'ltagmat ic mod ifiers of 1!:he superQni ~~ate ~e~me. t:a[ eX:J,r'l,p~e! 'huy· ]s, j,n COIllt:r::IslW'~ith 'steal " IL'j, :I. hyponym. of 'l!of1i:~ arid the contrns:~ between lh.e two eo-hyponyms can be a5sodat~A w~th Uli!: eonL~t bctwel;:l'1 the a(]yerbia~ p'futrase:s • hy P~I'C:na.s'E" and "by thef[" used as syn:tagm.a~i C modifiess of ' get'. :ilu t~he5ense of t nee nouns • pllrchase'

Figmc ,. A lffiod:e:l of a hi~n.~chk:3J1y orga niscd ,'QC;'jjbil.huy

and • theft' eould lurdly be anal,~ sa:~;s!iI!oto'ri~y eNiCe,pt 1]1! b_"rm.s of "[m}, • r s..eIP a:n.d • saeal"; and it seems that lhil;re is, [Ie single pai r of I;Qlltrnsti ng cKpressiOl1s i niE!Iil.g]i£h ""]l~ch eould ~el"l;~ as .~'ynJtagmlltic ~lod~.fims of ! ,get ~ in a l'l.on-,t::ircu~:J:r .metali~guLS.t:i)i!: glO'ss on '. buy' . amI " st'oCat ', Thi~ is by ncmeaas tm:l:ypieal The eKp.fcssirms 'by purchase' ::IJiiJd < by thefl:', wihen dler ate crnplo)i'¢u illl this\'i'lllY, are h~!St thought of ~Sl 5urrmlari:2i~ a whole SCI of more paui.cula<: m.{)dific:a,tion~ of the 5f"I:~e of < get .' en.~,r~ll iat>ed in ' buy' and "~.~ eal ',

9.~' Jliert]T~II~"t:a.t .dYuel'!lI"C 1:11, the MCd 1m 10. ry

=:. The tcliltion of hYf!0l'lymyim~'osi:~ a hk .... !3.r~h~ CJl stnl ,·ur" l1"';-ll the ..:.Tf:abliJll~rII Eaftietll.a_!_~~lds wi.rlliml the YQ~l:nl~;nd .he hie:rareni,t::al m:-d~ing ~f lexernes can he renreS£lt.e4'l JJ;Jr~:llil~3, tree- 4!l!grnm" ;liS. iihilst:ra:tedsch~a'l:k.d~y in :fi;gure '7. In the diagram.,

Q, ,OJ C, ••• j k, .. 1, etc. $ta:!l.d for indfvidllal. Iexerscs ; andthe poi n:ttlf ora gin, or root, af l_~e tree is labelled Wil~l zero «@). Twe branches i!.t'\C ahown ~SSUi£lg from e:\~h node; but this is ef'eeu rse ~pprop[] ate oniy fOil" oo':hyponyl]lJ;S :rebit!e,di by o[]!lositi,['m. (d. 8.].). The blloken rifJ.es lndieate further branches. On the tree. As \,w~ n.:we f:Jt.ei'l, Iit}1llQ[]IY!m.Y W~, t:ransi'tlvc; so that lIny ]ex>c:me is a nyponyro ~.f any '(Idler lcxemc that domi~te$.·· it QtlL [IH: teee (~.~.~h:llt is h]gher on the tree :lind ()onncct:m;1 hy a path oonsisting ooldy of desulli.ding bra[lches). For example, in terms or Jigu_f.q, H(l. b), 8(/, . .fil), H(l, 4, Ii(g, a), and f,rJ Of! ~\l,'llc:re 'W stands for tbc relation 0 f hypanrmy)" If l,'C [lOW irlllmJ LIce ~:be n~lim~ of direct, ~([ ~Inmfldiat!;, dQfI1lll:lat]o.n. (bt'ing eennected ~y j!..lst a 5~[lg~'e de£ccnd:i~g br.il:fllch), we ca:tI, I..oloy that a irn:rn:e'liiJtdy II dominates both c and d,. but melt g, f~,I, eee.: thilt (: imm~itIia:tcly d{.i!:Hrin.3Jtj~-S!J 11ln,d .li; tnat b imrncd ua.~ te]y do:rrunates 'eIDd f: a 1IItd: so orl,,,lJyv~I't.1!l.~ 0'( ~b~.s fllct a ~s the: j Inlm~d i:!tc Sl[~pefOfdifl3ll!e i!)f c anA d, !lin.d e and {lam immed~~tc hyponyms of ~.:. there is [10 (proper' hYlPonym o·f ~ such that c or Ii is a (propl:r) hyponyrn of lr.

~Ve m!ly n{lwooitu~idef' \JL'Jle'dler a tree~d.lltgTam such lIS ~he one pn;;._ se rn:ed_ ;,n figure, does in £ia.ct :reflecr rifle structure of the vucahu 1:iI~y, (ll" part,'l; l1II<f the vGcabulary. 'Th~s gcmend question splits 1ll,P ]~toa number of more pan; C1!li:a,jf quest ions ~ and it is one of 111,e prin t;ipiflIJiHi:ljr~5ti.c advantages of co'nstm~til1:g such diag,rams (which can be interpreted ~ stltlple rnat]l>em.aitiod models .®.f some empirical dornain) that thq ~~,~ us to cO!J:!'!l!~ler' PQin~!'; whi ch m.igh·~ otherwise escape OUF artentina.

Onc!>l.I;cl\prunlt may he mCl1tinncti arid diamuS5cd without a gre1!:t de!!l of d1r~Uf;Silll~ _~t would be filllU ral, !(lOki ng at figu1ie 7, to 1l.lllJ.quu[e wherher the ~en~·reI3tiQnwb ich holds he ~w-e en JJ and Ii Is noccssllri!y [if ever the lSl:lrnl:: &.. [hat which holds between ~' and i: more ge lIler.;dly., if fi) Y and zure iJI.rmediate CI1],.hypOilYms Qif :N:" fI~) q .and r' are imru~d]~tll1 co':hyrpoi:l:y:ms ofp" and {i]~) x .andp an: iml1lil~,diate oo-hypiunYlTI5 of n, is it r:ll!wa}l's or ever the ~e tM!: R,{y, .:;) = R iq, r)? Tl!t€lan~wer seems to he: uh!l!t ~!:lmeti:m~si,t ill ;'ind enmetienes .~t: is not, f Of example, .i~>el,l:rr. day .English {in whicfu! !SWell wMcl~ a~ 'maOllrnal', "!"eTtcli:l.mt~·. ~lC. wotlii.cl n,gt, b t: ~s.cd) ! h.ol'SC' (~> IU'Ld' sh~~p~ (p') .~~ lmr:m.::d iah: m.hY[lQ!lY ms '0 f <;mimal' (J~); <'s.tallion' (y)':lnd ~11! .. :rel (:>:} arc i!nm~di,ne co~bJtpon}'in'i> of' horse ., and ~ mIll. i (q) and' l"WC' (1') rlr~ immediate 0;1- hypOli}rmS ,of ~ sh~l]p ";: 'arid 'lhc 5e'ilsl,;-[!'c"1i{t~ Oll h(ild i t1~ ht:·~wct:n • 5la1licm' and '!ll:'!.rl"· i~~he same. we. nay asS,UJrne, as the reia:ti{l[II h.._~" een 'm.1l1' aJiil.d '~'''';:', On the ethel:: lund, • 'Oi rd 'and "n;sh" (a.~ \,n:U as ~ an umal' ~~ d pcr:har~ ! !iJi::rlion 'j arc imffil:diilll!e co-hYPQnJ"~ of 'cr~atme~; but the oortems:ts

wh.hin the se.t C eagle jj ~!h:;l:lSh "~I tst:rrling", "curlew", ~t,~rmL', toystereatcher '., CH: •. } and f.eat', 'horse'.' shcl:'p \ '·em.\" ", "welf", • elephant", 'e~c.l ;liU; hctc'rog(;'m:o,u 5 .and I di<nsyn~rati c. At bC-,St''''''emighl be able to find \'a'l".ious. kindso.f scalar {lIpposi~es with,in ,eou;h set, based Oil! such c.r.iteri a a.: size, f riencl.l~i1l12~!'i .or utili.tyw rll:t!l, and so on. But it \.\!ould1 be hope]g.~$ t~, .~ttCrtlr~ :I. C01'11Lplc ae QIUl]YGis ~f the sense of ·the r:L1Le:mliJ>cfS of each s1;:1!i:n terms gf CGnt~"st'~ r,g:lenfit to bnth sets, The theorctieal !ilgnificancc ofth ~s: ,,·ill beo(m~!~ dearer' ~~;bJ~ 1'1 we move OIl to consider the prob ~el'n5 of oDm.ron~rlit:ial analysi,~. The ,~H ~s meanwhile invited to ooIlsicier various sets of c(ll.hYPQ n;y,l1jlS i~ ]ElIl:gli~h and other ];mguagc:siw the' i~ght of the point rais~d i n thus, p:lr:l.grn.plit.

In figill"e 'Jt we have put j} zero rather than a i.cltCt: stdit'l.dlog for some ~ct:liI;l:i. lexeme at the rom of the tree. Thia is :~nteru:l,~-d tQ5Ugg~ st that the voc.:toubrJl\ and indeed anyp:tr~icu:~aF put o.f .~t. m ay fu~ Sh'!iJ:ehm::J h,iElrnI1~h iCIIUy (rom a poin.! whiQh iis~~f is; not lIssm::iated with Maclua] Iexerrrc ... ~.[1 SID fal' a.s£igum 'I represents the hil:nlrchiea] s:~mCt1!11'll of the l'U ~bl1l.ary as 11. \\'horc! .ut seems d= that there l~·i~ ~ be rno ~~N.(lmc at the point o·f <liil'rigln" T1Je faet thll!t lcxernes, .i n most tangu a;ges !l:t leilst., f<l.U into 111 number of di:st][u.::t parts of speech "muM of -itself p;redude the Wernrchieal o.rclel'lng of the vo cabtllary i [I. terms (If byp~nym.y under a si f1g:le le~e[l~c.F (It' a lexeme bdO!'lgi fIg to one part iJlf s:pOOtih. eannet hea hyp~;lflym of :ill!exern.c bc[of'l;g:ifl~ 'to all;rl!i-..cr part ,@f speech" [ffigure; rep:resents the MrUicture of the ,ror.ahuia.ry as a. whoie,llllien, ,(l and b are sur~ mrcli nUe;., which :!,l"E;: nor hYP[I'!l!ym.Ul of any oth,er lexeme ;:li~~ 't~c br:am~c·s d£~c:cnding N;I thcmfrmn the f'tJlot ott" the tree muM: b e dlIl'ilIrta:~cd~)

But. is it I·~n.c case lhal the v(lt::]hu:i ~ry of a 1~Ji]gu;:!gcis Mrl.lct~uedl hi!erard~icdly ill! tierms of hlPo~m1 under SJl;';Vc"al diffe~ent _;points of o:r!glri, e<l.ch one ass'ociat!:t1vdthaparticular part of s,p{)ed~ !ursornc major slJ~dass (If one ofth~ pat'ts 0:£ s:PC(;ciht?' This is,at fi~S'E sig;ht at leas~, a I~re pbn~~blc fiodon; and it rc!,atc,sw, t h(! Aristoldiafl. doctri ilLt.: of '~he 'C'.lteggfie~ 0 f pred~ C~~lrl tl !lind &ub.5l~q ucnt d cvelopl'lu::nts of .it. Let us ti rst elf :111] co!),sidef' the nO\! !'I.s in ~n:gli5h. ,[h.~rcis fit) :~c~~rn~ iIlE:Ll:g~ish '.:h:ich issuperc·,,~J.il'latc tiO ",u ·!10l!Ul~. E\''fl'lthe mo"e or ~ess !ech~utal word

lDl~gists li~.~·l:p~Id .1 !:lood dtill.llfMtentlo:tll ~<:l th~, n.t!~i(}Ii. of

~ ht~r~:rehi. _'letUIn: :in the 'v{i!Ci!bul~T"e~r!!<:;ian,>, in ow!i,~!!i·.~J'1 ,..-1il1 th~ study-

of st;:;;;;l1 - t~g=~ (d_ ntr[in, ~~d~(J\lI!! &. ~a"l!n. ft)M. tQ,.+;

.conklin, 1'9'6~" 1<;\1721, F,-.il""~" 1(;/12; Sturt~:\"~n~. ]9(4). {}n~ pdn!: ~ha~9.~rn~ ,!" b~ w~lI ~(8hl~n.mJ b thMrnlk-l~,/I~l=,unlik..-- mO",.rj. ;'k..:~ntifiic ~a:l:!lJInomles.~1\e .:not ~xhausti ... !: o~ tih.~ .a.()1n.tirl dlillt th~y dl!:ssif~ ~. fi'" Ja they n::~ih dw S:>IllC dJq~ ree·of $pecificity illl ~.U~IIea:s.

'~~:~tiry'f~iJs in this l">I;l.spcct~ ~irJre it eovcrs '(I'I~ly ccmn:ta.ble noU'n~'; 'illnd its nearest. C<fil ivaknts i 1'1 cV(:Tyday .En,glish, < thi ng' and .' objcot: ~. are still :IJlJClr-e restr-kte d. 'There is, l1!O le1'lcme whi eh is sllliper:or~1 tm1ltc to all abs.tr,act m}l1l'lls, @.f all eO'ilCrctc nouns, OK aU mll~S nou IliS" 01" aU the members of any of the m;lJj OF su bd.asse~ of [lm.l1T1:Slh3;~ arc t:'uslOm!lrH}!' recognizee! for E-lilgli sh (d. ] I. 3). 'V\'ha:~. we Ii [Id instead is as!lt of very :gen0.ra! lexemes ~ 'perM)],!!' (Qr' 'hum.:11l being'), 'animal~. ·!l:sh ", <bird 'j ! insc~_t I', '~th ~flg '. 'pl,lH:e j. • stllJfl"", "m atc:r:lal >, , qu~Jit:y ~" • pifOpl':rty !!' ,j stal'!e·, etc, - wrnd~ are supe'midin;l,tc to br~r g:r~rnal]er' subsets of these subdalO~es Oflil~1iiInS,,~1 'ftc ~r~ditiona~ dcflllitian of I1Ir;r~fi!S" ~t .may be tlbsen:ed', as words .ihid~ denote pevsons,. pleees and (hlngs is dllfi,cu~m~ (~part foom its other i nadequadcs) in. that it makes no mentiiQrl ,~J' .anirnal~, fishes, :IIIlil!! b ~rds, On 'l:ne nne h:and!" and uf qual! ~ics., 11l:atcs" feeling'S. ese., JUt" tne other. .h is abo, I n'lpot:"lIIl1'l. t Ci note nUl 'anhnal. ~ il'l onl:tnaryE:ng,~i;shi s, not s:upemrdi nate ~O'. bus in conlTItSt both whh ~p~rsofl' . aXld with ,j fish', "bird' and 'i,n~~~~~ ;lind that, ahhough ",~ea:turc' is supC[ordi~a'te to aH five lexemes, it is sty[i:sticllUy restrieted ]l] variQus",'ays. There is I'm SUppdirl: im! lh~ ]exiea~ ,&tmoture(l,f English, therefore, for thevi,e~' that ,~.I] lIl:ijUnS denoting a"i~n:l,te bdngs 311c hicrnrchkaUy o,dered .• ~n terms of nyp(mymy, as a. sirlfl]e class,

& .his .~,,;e~ mQf:e s;~ri killg~y the ease' fol" ether F:lrts" 0 f S~t;h thalli it 1S

for 1'1.01:11:1 s tn,;u thee)' are !'lOC ].ienlf'clii e-~I1}" (lrgani~~d u [] dlel:" 111 si ngle' supcro:rdi natereN)CfIle. There are certai n vcry gc.1:i.crnl. verbs like ; lICt". <move "~ 'oorome', ·ltIllIk:t:,~, 'glCl' <lind' be", which have ]a;rgc l1lL1mbcrs of hypQnyms. Vi.' C ~tl. ~'e ~linady!;ct:n, [0[" example, rha.t • gl.!t' has as. by~mnym.s 0: buy ~ amel < Meal. '. :1:0111 tn lhc:;~ we H'IiI)" add ! borrQW'. 0: wi n " • e:3irn ", "'>Catch', ~ fi ad '. 'gra:ap", etc. But no one oJ the most general ¥~.rbs in En,g'Jish is .stJpe[lrnminalc 'to :~JI the traMutivc or in~["allsjti\~c v~rblll to all 'th.e v~rbs of sta~c 01" ,·erbs of' aeli\'~~y, '10 ,aU the verbs of motion, or teall the members of lilly of tlrn.t: ~T:lI,rl ufianJ U y ,lrccogn,imd s\i,bclasses of' verbs, If we tak~ the most common adjectil't:'S ,UliI, Engl ish wew.in see l~'iI;~ thereare no supt'[IlJ.rdinatc adjct::ri\'If;s, at lIU of i;~1lidll parti.Cllbr :ulbsets! art hy-pen_YfflS. Thera ~re no lcxcmes o( whldT! adSeri:ti ves, den{)'~i ng' din'rtc'i'Joe"& flf oolou r :IF'~ aU It },PQfI),l]lS. We do not s~yWar i~ tttl qF colo!U&i i·~~ s.om~ ()llL~ way?, bllt radle. Wtl'~i~ ret! VI" fa!) .l'Qme I'JthU!mloul'? SimHariy (01" Sl!bcl<li5S~S ofad!j ec~~ yes (k"IlU'~i nc diffc[\lllllces of $h~p~. textul\C, til;ste, sgulld,. age, ~;i~\:, st1Ltc of mil'ld, ,Nc..

n .1 n man}' J;D;ngu~!l~;S, ~Q~»~ of th~ .m~5.t ~n~.I'~,t JIlQI.mS.. 'c(JrnIJ'81'~bii!: S~T1'1rJ~i t31~y

wj'!cn •. En!,:H,sh 'pj::n;.on " '3f1/i;ma'~ '" • th~.nil1'·, ~tc-, fiIJht,t ior~. ~ynt~CJicrul r ~~ d=ificr~ (ct, ~,. ,4"

There ru~ rtQ p.a:radi:gma:tLc su~;u~,rordiliL3:tll of u,n.lcih • round ~, l'sq11!a:re' c ob~ollg', eB:.are hJ,!lOFl}"m5: wlhAlt we find :il'!5tcnd is 'w~t m.~ght b. ea[ted aql.liA5i~]:lar~d:i,g1:[lat ie relation between these ~Qre spedfic.a~j.ec:, t~""e!> an d tlile mi:Jire gene~rnil, :absttttt I1.0lUn "shafe' (d. ~rl7lal shape waS! it 'rtmll,(l 0'1' jq!uJ'wtd''). AdJi~ctives JfktE: 'Slo'I."ClCt" '.sour' and "bitter' stand it iii. s.im~ls:r kil'ld of ql;lllai.~pafild:igl:llatic I1clatiQfI widt til!.: "S!It!pcfO'rdi.[Iatl verb ! taste ~ i( cf. Wlra't diles it tW~f! ll:ke.f}, ~nd derivatively with 'till cm;rcsJH,nij]ingMun • tast~" (d. Whal /lind of'li18te; 1JQ8 it g(lj ?). further· moJ'tc, t~.e nouns • shape', • :'.i:ze" and • CQIQur" are in. qUIll$i-parn.cl.igmllt:i1 rdati.an with tht:H;~-Ti 'lank (,!ike)', as h is emp1Qj;'ed in SlJ.lclJi s~l[le'L1!ce: as, What dOl1S i/t look fike? alitd in. pa;rndlgmaHcrdrat:ion with. lhcoor'u" s~:I!I!dingMI!Jin. ~ ilpp~ar:anee • (d. J)e5CmUt rls apJtMr'tM(;t·~ w!iatwliOw ,ami skape WIllS it, (:ed ,and Sq"lilQJ't', oJ' /J1',e.erlaJ:lJrolmd.:?).

'H.c 3mlstlin\~'hich 'till: tertu 'quasi.-,para.diigm:!!:t ie' is bt:1 ng used hen should be dear enough from the examples: "'rid it em: easity be 'm!tdl :p.rec!s!;; wjthin the framew-o:d:; of a reas.o~ab]y ~Qmprehef;!sive teans~o'rflUt~~md!l' gnul1Im!l.1' or English ,(d. ~o.j), \~.fe maydJe5tcriib~ t:h~ ~.bl~o.l'I~hip llet.~~!Cef!J, f(lf e"<I!mpie,. "round' and ",sili!,a:pe> Of' • sweet" an, ~ taste !, as being eneof quasi~hypO'J'Iym:y!t, .~ r we i:ndud:e qUllisi":hyponym} w.ilih hnon}tlfliy as :I. relation ;,n tams orwhi.~1it "oOcainllar:i:es ar~ simi c'~1lJJred hi,eral"JC:.hicalily~ the hypotlrn:e!'si:s that the yocabub:ry :iI'J_ 'aU [a[lgu!ilgc~ is- st:ruct:u:I'OC! hitraffh.~c1lJly under a ruath't::ly Mnill.l set of lC:i:icme-s 01 'verygen.en'l& sense is rather more: plausible ... It is ;1 Jm.YP'Ofhe~i S, however; nrhidlt is diticU:l~to eva! uate on the basis 'Clfthe e""idc[lcel~~t is at pn~sel'll :avail~ble'.

O:m,vcntiol'la'l dh::tionarles, inwh_k,h, ]exern~ (in ~beil" citation f:mm; arc ]ist'l1d in alp~abedcal order and their [l:r:incipal S-!:<I'HlCS ~iifilled. an.~ 'el'<_-~mpli.fj;edJ widt qu.otll:~i:o'][S (uslllally from. wrin~n5(J~6),.\'\'i~i fre' quel1ltly append to sorn~, though Hot all, of uhci r enrrles a list of S~ ctdkcl syn(lfiy:rns and antonyms. The beteer and! ]]lore oomp'rehensiv'~ alphabelical di.ctio~:arics w~U WV(;so!rnc hlldication. of ·the lCQrlltexts iF. which :roMgMy equi vldent lex:emes lire i~ten;hange!l:ble <UI d wi!! dr.l\\ alttllTl:1!:illl'! t.Q di fferen'c~ in thei'l': (:Q!'In:ot:a:~i:ons or ~ma1tiVlli'm~rt; blblt n~ dl~tlO'I>:try systematical Iy disLltlguishes the d~lfferellt ki.ndls· of fexical 'OppOSlt~O!fl found in hmgtfn:agc (d. 8.3). As fill' .;)'5, ~bttofls of h)~il!!Jymy and <l!fl'wnymyaft ,oo'fl~ml,;d.thct:Se al"C raNly made exp:~icit; anti they cm.l1lJil always, bl': wnfcrrcd frQm dw def.in:iti.ons.

Thel''t:is, however. :mother ki.tnd of dktionary,whil'h is oomm.only descr~blld as. bt~flgooncep'tu:a'I.r:i!,ther r]:Ulfi ::dphabcrica.1 {d, mlrn:ii!ln, ~957:1[JjJ~ 196~~ :ilj4fi'}. The best ImU\'i'Il oIthes.¢, .~d, Uil uLicst 11'1

modern ri mes., is 'Rog,e1"t Tlit!8rzm'l'IS Qf lJ,igb·sli nlQrti, (1M P'llrlMe~ (~8 5.2). The princi:ple llndcrl}ing a oonceptual dicti~!:lna.ry. or thesllunls· (to employ 11: di,st:im:ti.'I.<"e mm and O[ic",h:~ch deus not (,U'C!MIpPQse any ,oommitment to C(lnCep't\lalism· in semalltics: d,"'.3) is, as RO!l:,ethim.self put ,it in his ~ntr(irluction ~C' tbe:6rst. cdi'tion"w 'cla!>s.iiy '~the ",,'Oros ,allld phrases of the .language, frOt ~c()rdllfig to thek sound 01" their rl'l'thography, but mi'tidy lIIccording to their sigflin,cuion ". Ahho1tl'gh Roget's work "las designed! primarily ",to facl]jtatc tbe expression of idea andto ass;isr in litcFm')' composition '\. i~ was strong])' infliHmce:d by sevcutcmnth-century philosoj:lhical spec;ulatj'Gn (deriving from ideas, of F .. mds lla!Cont Ues~t&s, and Lei~:Ilil) aoo'u,t Lhc pus8ibilil]" of canstmcting IIIII iueal language fo:r the systematization and developlllill'lt of &cicnti fie .Immdedgc and UllQ.C particlI.lady by the famous fInlY 01iil th~s ,subject (:1668) by John Wilkins (d. Robins, 1967: 1 1zff; SaJmoll, .I966}. Thesauri ,compaJt"llb]c'ivit:h Rogel's forElliglish, but drawing: upon advances made in descriptive sem"'ntics in the intervening, per.iod~ also c;\:ist for Gl:ml:liO ([)oms~ifi. 1'933) il:fid Spanish (C::lsa:re~, 19.~2): and t~el'e is an outline thesaurus. fQr Frcnclt. Q,lI'crlly modelled on Rog~. in the:i1ppendi_x to Dally (1909). AlS(!, to be mentioned in this cornnexion is Buck's DictionaTj' of Select'ea SylIouynu in ,ti,t PrincipalltldD-EMrnpecm Laulfuageof (] 949). NOil e oK these works, however, \·allJi able In<l1!gh they are.prmfides the' ;infQrrrunion \yewoll!d need in order to l'esohre the qucsticn whether the vocabularies of the lallguag:es ~ hey dC[l.1 with ar,e org,anized. on ·~triiC-tly h~erlU'cbicai pfCinciple5.

The mast arabi tious scheme for the ccnstruetian of a. tb.e...aurus in 'lemlS of an allegod~y I.lniver~aL fra:mcwork ()r semantic ca.te,gorits and soi:H::;:lt<:gmie:s that has eo fllc been produced is Ha1l.Lg and. ,Vanburg's, i(1iJ5~) so-called conceptuai s}"St,em (Dcgriffssyste-rn). Despite its claan to be I'!.n empirically based and unkeli"aa1~y ::Il~plicSiblc dassiih:at!o1lY sJ'S~I.?:mj which rcilclCts «the intelligent Q\"Cmgc penon's picture -of the world {Wellhihl] as, 'th.is is, dotermineu by the pre:-;sdelltufic genera!: cOrlce pts established i~ Janguage [ du [lC]~ die sp'ra,ell! ich bedingten ,'onvissc[ls-ctiaftiichcn All~.rn(;]nh-c~fl'e besrimmt] ", it is epen t~ the cnti cism t hat il is 3JO much an a pri.olF.i s)'Stem as Rogct's and P'l'o:i).a.hil b~ascd" in SO far as it .is gf\oundeJ, ln descriptive semantics, in f:n'Ollf of the ni'ili"'~lieali:&1fI of speakers of wh!l:t "WhorE {19S6} called Standard A""e:m_geEul:'!J'!lt'an.. It can also' be criticized ()nmhergro\Jnds(cf~ Ullmann, uJ51; 3 [41; G~eke[er, 1911; 99f) , It is difficult to justify. for English at least, even the l1ighe$t~l~'el tripartite diVision of ~hli: v0CIlhllhWy into leeeraes relating totae universe, to H1a1I, afiiJ to man and the ul1iv'e'rae;

as it is difficuh to j1l!!stiFYt in hmns 00,£ hypol\}''D\y and qUal>i,.hyp-onymy, Roget"s siN: main classes IO[ Iexemes, (~) absh<l!ct relatinns] (ii) sp~e; (iii) nutttJCr; (iv) intellect: (v) volition: (v~) SCR'ricnt an~~ mura1 powers,

So fa1!' rdatively li:tdc is knownabone the: lelfical. structure of dle 'Illst majo:rity of the world's laoguage~,; :llIldl. :as we .htlV.C seen, i't ill as yet impossible to evaluate, even. for well llu,u:Hcd and: ~al;:it-1 acccssibJ.c European lrnngua.,g-cs, the hyp-mhesis fh!lt the 'IIfocab:ula;ry is hier:uihically ordered, IliS l!. whole, in t'e'rm.s of hyponymy and quasi-hyp£Hlymy. The lhoo'rutili::d sem~flticist IIhQuld 1iJ_~ ,romsponJingly oeamim,lSablllUt p!:!t:tll'lg fa:rwaro gencml hypotheses of this kind. It .is undeniable, however, thl2lt there i.iI, some de-groc e of hicmf(:ioimil cmrgJlilllzatiolJl In all arells of the \l"O'calr~larie5 of languages dm,t ha.¥c been investigated. Indeed, it- ~s hard to conceive of any ]ilnguag,c of't::nuillig Si!lt:i&fact~ri Iy in ,limy culture witiu:Dtlt l·ts, vtJ.!::3bulary being structured in terms of the ,complementary prm,dples of hypOfiYro)' and COflU'lIst ; and the descriptive WIiIl\k that has bee 11 dene On various reas of the vocabU!I~' in paruCl.!lar 1a:!IIguag II!Jlpean to supportthis oonclIjJS[O:ll.

9.6 • .Lexical gapl

L-et, u now i[lterpre~ figure ';' nll)lt as a repres~ntation {If tile hiernrchic;;al le1rganization 1)£ the' whole "mClll:11!ila~ of 111 Jangullge-, but of particula.r lexical fields- within 1I.1t;Iocabu]lIIry (d. 8.2). One of the qucstions which ariSiCS in this conl'ltx~on ~s. whetlH1 there am. be w.hat have been celled ]exlcfl] gaps·. lNt- are not eoneeraed here Wiih the absence 6£ 2. ]eX!e:me dcnotin.g an object which h~ppel'ls not to 'cx:ist wn the culture in ''r'!h:iC:b a bnguage operate!>; stilll,l:s!!, with the. absence 4f)f lcxtm~ ~'hich wouM eneap::mb~e:' the Soens~ af colJtt:l'adi,ei,ofY' lIyntagmatic modifiers (e ,g. the non-cxisten.ce of III lexctne mc-aning .. married bsehelee" IJ[ ,. Quare c.ircle ~'}. By a lexical ~ iis here mci)i1l~ wh(l;t !;:tfuct!.1lmlists often describe, metaphorically, as a hate ~n the . aUem: that is to saYI the absence of ,a lexerae at a pal'ttcubl"pl:u:'e in the stnlcturi!! ofa re'.Jllcal field.

Tt will be' wcaliled tllllt, acoording to TfiieF; this is theoretically inecnocinble. hut the ~sumptiotls, which determine hisrej oct ion of the possibility of lexical gaps are qllestioaablc (cf. 8.2:). Lexical gaps (If the kind that concem Us here faU into the ci!legol')' of \~;I:i;l!.tLehrel·" 1973 : (7) calls matrix ga:ps. As she says "3. rnat'rix gap sho\vs up 'when re[ah:d lexieal iteras are ttnalysed into semmt:Lc featuees an.d] placed (l.],[ II n:hart 0[" matlFix". \Ve w:ill diaeasa tile qucst:io.i'i, however. !"\'Iib particulaF reference to hierarchiul structure, Leoked at from. tlm,il! poill~ ~f vi(;w it re&Olves itself into two more specific qucstlom.: (i) Can lI'I'e have

Structural semtmtU:s II: smsc TtlaliOM

cohY:PORrymy 'I'II~thol!lta[! existl[lg superordinate Iexerne f {ii) Can we eve. uf' l~t there is ii. mooJt;ld gap at !!: place in the hi eri!l!l"'Chictll .stmcru.re where we would expect a hyponym (l,r an e:,os:ting s-uperordinatc to be?

Tile answer to the first question, in, 'ilie terms in wllich it hss just been plllt~ ls decidli:d in advance by QUI' uefiniti.oTl of hYp'cmymY1l.nd collsequ~ntJy of co~tlypoJi'l.ymy. Hut it can be rdo,rrn.!!Ilated in t h~ fuUowing way: ]1'8 ,it ever the case dH'it t WI[! or' more lcxernes are iT! CO.flt:rus~ without there beilng 8Jny ,sIJP~rordifiah: lexcme of whleh they are inlfm;~~ diatc byPon)'I1UI? \lVe ha,f,e ,already tooL:dat II number of ~xamplcs of ~exieal gaps of this kind. If fWlre 7 represenred the neld of eoleur adjeetives in English; there would lei e no lexerne at the roOol of the tree. O:n: 'th~ other' hand, if we gwuped tog¢t:hcr ill the fie1d lexernes 'belonging to dilJenmt parts of speech (assuming some sati~fact<lry invt;gr'ation or grnmmatical and le:xi~ stmcture) W~ cou[cl ,say that the noun! IXIlour' is imml!!diatety s'Up~mrd inate to r fled '. '. grecn " •.• }. - nder this i nterpr,etati~'n of hierarelucsl struenirc t.he number of lexical gaps among superosdinetes will be much reduced. Ih.u i.t wifl ;'lot disapputr ej[jtir~]y, There ls no im:mediate :superordiilJI'U: for 'g'",' and "eome" iii Englishj .for • teacher" :md 'I pupil ~ i. 'Ct)l" 'buy' aad • sell': or- for fiUJlY othe!" pairs, of oppos:ites [1'1 English. The nen-esistcace of pertain superordinates (arid the relative infrequem:y Of' restrieted 3[iiplica.tiio[l of eehers: e.g .• c S,potJU3e t s,upefio.rdi nate to 'husband': ·!\'l"ife'.·' parl;eD~' superord inate ~o "f'iltherJ,:'Jmotile.'" etc.) i.s relat.ed to the im:pO'rt!lnt notion of cadllbility (d. Brown,. 19.58: ~1s:1i)'.

The second question is rather more diffi!:uh to answer in general terms, To tak.e one {If Chomsky's, examples (196 S': 23 r : cf .. iLehrel', 1973' 97): in Eng1ish there Is a word' oo]'pSfe' me<lll'ling rou,gWr '~body of !l dead human bein,g'" and 3 '\,,"Ord. ~cnl'c:as.'l!~ mC!lning ,,' body of a dead animal". but no worn whicl!. is app]i.ed to dead pla.~u&. [hit thlS example is, !lot as s'traightforward as it might appear at first .!>\ght. First of all, it sh01:illd be noted thM the sense of t corpse" lind • carcass • is not a simple produ'CI: ~d lhe S(! nse of C d,ead ~ 3:lld1pers1)[l'. em the ·Qne hand" lind of

• dead ~ and f al'limal", ~m. the orher, '~Corpse' is not a hyponym of

• pcroon t I and • carcass' ls not a hyponyrll of • animal '. Once we i m:lude the Sense of ! body' in the gloss, then the all,cgctl parallelism between r corpse", 'eareaS!' aadl 3. poteotiall, but unactuali ~e4j lexeme ap pi i~ble to dell.(] pla:nu is dC:illQYlld. Furthermore, it might be arg\ied. lThtat I oorpse" a.nd "earcass" do nm ,COntrast ill the way suggested by the glt:<J;.<;,ses 'that We have attached to them. If C;\[JRihilism wC'!"e insd'[Utim"l:aliu'!:l in Engthili .. spcaking oonIltrie~ and human beings were slaughtered with

sheep, csttleand pi g:s :fQr food, lit is pred ietable that 'carcBSs'would be applied to the dead bed ics of human bcin,~ ddivered to' the butcher's silal,'l, On the other hand, if Nl;uy's. Jiltl!: lamb dies and, being ver)' fond of it. sne.. deeidea til inter i,t in the garden and vc:ihaps to read. the buri:d service over it during Ute ceremony Il! intcrmen[,she will certai.nly nor describe wh:u she is, doing as bl!lryUng its, carcass: i:twi]l be the bodle, or corpS!e, of the Jamb 'tha.t is ineerred.

h is not 3. frivolous or f.acetious paint that: is being made .il1l COl1lnc.xiaA with tbis erwnpJe, Nor tan it be di'smtsscd wilh an appeil to the differel'lt:e between the cognJth.e meaJning I!.fld the emorive m.eruUng ef words. b sinlp1y is not Clear (bat • CJ)fPS,c· and' J~9S' are rc]attodl in ,oogfliti1.re me:ll'lil1l,g, or sense, in th,e way that hl'l.ll been stlggest~d. The point [s that our dealings with dead hu:man bdlDgs and dead animals are characteristic .. U}' different and these are institutionalized in fllneraJs and inquests, on the one hand, and in slaughterhouses, butchers' shops lind the preparatlon of foad, on the other ~ and lit ~s, in re latio.n to SQC~ cllh:urlll ~l'Isti.'l1.ni.orm tbt the dist'umctiotl between • corpse" and ~ carcass" is, te~icaLized. The faet lIl.ln we do not have a word mean mg If dead 'plant" is 'preslIlna'bly to 'be a.ccounted for by the fact that dcsd pl:llilts, las a class of object'll have Ilocuhmalily recognized tole in the socktk:f.! in which Engl.ush has evolved. Lch'I"CI"'s (J973) discussion of dlLe l'Dcaill.J[a:ry of oooki!iig in v,lI:tiou$ Lill'lguagc$ !.hows bom tlu:~ irnport~ QIlCC of culJtIJI'~ cQ!'Isid,e-ra.tion ~ and the difficulty of deeidi n,g: what is ood what i!s nOI a srructur,ally definable lexi.cllgap under !I com:nwn 5U pernrd iaate,

Let us, nnw tllke another example, lis we saw shove (d. '9,2')1 there :rre

,- , . f·· ,< -t'· 1.., .i,. " '!'h b -..I' broth r'"

s£]"Ill'r!l.te wnrds no R 1J ~S,Lan nr '. 'Whe ,R iP[t'l'~ne:l"'. us am.lI S .. rne ,

"""'if'll:·s sister"; "husband's sister", "brother"s wife" find nsi~tef·!,; husband~', but (not. surprisingly) no words for "brothi:f'S lmsband" and" sisti r's wife". Theile 3.r~ undeubtedly two holes in the !Il,aUern :!It these po,ints; but there arc not l W'(iJ oo.I"[lI.:sPQndifll!: k'o!.~cal gaps ~n EngHsilj siacerhe in~law 'Voc3b'Y1~uo.y af English. ;EI:S we have ~~r[l, is, structured ]0 terms of d.iffelfenl oppositions. One might be :inc~i.ned 'l'@' say that there muldn'[ possibly bewOIHls in any lal'igll~ meaning "brother's husband" er "sister's ,'!fle:"', nut this is, surely Rot so, Supptil'Se .i~ became more CQi'llJmo~, and mere socially acceptable than i.~ is at presene, For two pmple of the same sex W emer into a permanent r,ellltioRship, invokim1lg not only ,oohabita't'ion, but an overtly reoogni.za:hl.e distinceion cfro]ie~ C!Hlli"1pm.r:lble w-irn the distinction of roles 1that c:msts in a. conventional marriage. Such a relationship milghl: well

30S

be illlitlatcd with a wedding .:erem.QIlY and the exchange of "OW~; 3]'I,d tile couple in question might de eribe themselves, iUld be d~scTibed bJl others, ~ bdng married, one being the husband and the otl'u."r the wi.fc. Whal c.lect l'I'l(iIuldlil!s have Oil tb,1;l ICJol~ca~. slruc:turc?The products of the seases of 'sister~ and' \ i.lic" and! of'br-mher" and' husband I resuldng in "sbtcf's \viI!ic'" and" bn:u'l1er's husband! ,i are rcatl~ly tntC'rpretahle under the circumstances that we hav't: ,envi:&aged. So tbe lexical gaps in rite strucrure of the R1!!),ssiliH yocabulMY 'Df kinship WQu.M cerrespond to ,Potcntial, I:nu liLon~eKiste]l~, lesemes, If the g3i£'& were fi l1e~ by new words. the lexieal structure of the Iangu.'\gc would he, to this Cldent, unaffected, The enses 'of the at presem non-exi stent wcrds are rurcady th,e:I'e:, as, it were, ready '~iiI' be IcxicaJi~d. Suppose, hQ\,\',ey-er. thlH they w,er'e nat fiUed by flew words,. but that ,instcad two of the existing in-law laxemes were extend ed l[l sense tii:l' fill t he gl'l~iS = JI'Ol!l.Sibl:y, ':tjatj" ~ncl ~nel'~tla 'I' the former ooming to m.ean "siblin,g's {mi' cbild~s} husba.nd ", anti the Iatter "sibHng" (or child's) wire j", This would be a steuetural change; and it would :skew the Current pllltem of relations that hoM in this. lexical neM. J\s. f3.r as E:l'lg~ish is concern d, however, there are no gap~ that, tu:~ed to ne fined, wh~Jher byc:reatlng new worils or '~lttend,i!;lll: th!e sense of alreadj tJdst~ng words, One's 'brother's husband ,is p'i"e.;\umablTthe mate spoase of one's sibling and is, therefore readily referred t:1;I by meens of the te:rm'br~thc';-in-Iaw·. Altehn'ltq\f~l}r, if the sex of the referera Wen: nat heLd to be decisive, but .rather the social role that he/she ~s playing, the male S[!OU5e: of one's bwtherrn.ight be referred to, by means of <brother-in-law" ur 's'Lste~~in-bw' ;u:QGFdng to whcthtlr he is the. husband Of the 'wife in the relation. h.i p,

The exaen pl,e discussed in lhe [In'''''!(Dl!I!S p~ [,ag rnp h 'L'luy seerrt rather fanciful: but ch:LIlge. s dl1ll take pl.acc in social.insti1.udoIllS and praette.es; :'lild Jilngu:ag~s C3r1 Ildap,t. 1I10fiJl:' ill..: lines S'!I],gges.ted abiJi\fc. to changed circumsteaccs, Anthropologiul diseessions ofkinsl~ip should sene to warn us again~t nssllming in advance lbal even :ao universal a Feature as b iologica1 sex rnUU [lCC~;SsMLt.r be reAe\!t~(d ~rul be darn ~na.rn i fi thiis area (If t he vocabu lilli''}'. The m;!,j n reason for i ntred [I r..:rng thi.s bYllotn.eltCal example at this pnim, ~'ro\ ... ever, was tOI 11.lusll:':llte the notion of le~ical gaps w~t:hi"1 the framework IJIf srruetural semantics arul, iI,'t the sa:llne time. to ,gi"f,e some iniJic::uio'.IJJ of the difficulties which arise when we begin to cons;uder :seri[l~9]y the dillc~llt~ between possible aml~r[Jp05,sible le: x emes, The exprcs!;lo!'ls 'femsle husband> and 'mllie wife' wOlJ1d prob~bly be regarded as semantically unacceptable (like' square circle") by most spe:rkcl's of' English. But, as we ha\~e seen, it does not require

much imflgi natien tOE:l'lviSJIge a wor[dI in which. s~:ch expressions would not be b'~ld to be cofitmdi.etory. Evcn in the. w'Odd. as we know it,. the propositions expressed by such sentences, as 'stu: is. the father' of five children' (;rr ~ She still Loves her' wife" ate in. no 'Wl1i;> scmanti'ti:aJly anomalous. The journalist J,am~s Morns" for esample, did not cease to be the father of hi s chi Idrel;'l, when he became' 3. woman (all.dI wo.k the name '1i111 Morris'''; d. Morris, 1974). The t'edl'llique (If C'liLv'iS;Jg,ing possibfe worlds other dun our OWfi ad then oollsidel;ing the applieabililty 'Ilf existing kcxemes, or t'lilUocations of ~l(~S'!illg l\;1!cmes, 'in re]:litian to lhe:-m. is a tii!:hniqllc dlat is fri1l.ught wi h difficulties. It is easy l&nul1gn to ~ubml t " qu.est ionnalre to II group of native ,spcllker$, 'lski n,g them what 'lhey ""{mId say in !!uclt...,'lm;l~5IjJch circumstances (and it is even easier for t he lin pillt to Ol)fls\dr. his own intuition}. nut the lnterpretatiolll of the' results, obta:in~d by such methods is always subjecl to Austin's, C<L\ut; t. ol'dinafy Jru:Jguage b[lelLk.s down ill extrnord'illary cases ., (1970: 6S).

There ~rc aeme clear cases of lexica] gaps ,in bt!IJguage..s which, \tnLite the examples J~(.I[I~scd here at: ..some l,cl'Igtll, present no pmblems with respect to the :pos3ib~lity' of tIiu,~rc being cirC:llllnsta:nees jr'~ wnich one might usc a wor;(i Wilh 3. partlclJlat sense. In French., i05 m English and other ];1Ingllillg~s, there are mal'll' pairs of al'ltipocla1 '!lppobites used in the defied ptlon '9£ spatial cXi:en:sI1JII1I or IlOut ion: '~.g. ~ haut I : j bas ' ("high Ir: "low"}. "1.o1illg·:·c~urt· C~l(lng'·:"sbl!l.l"t"). There is a :Iexe.me ~ p"rofund", meani fl,g •• deep", but it has no anumpoga1. oppo.site (d. [shalLow' in Eng[i;sh): either • ilnflfond" is negated or the expression r peu prruol'ld' ( .. deepto a small degree oj) i5 used to fin t he kxicalgap. If French i a du~d such ]Cl>:\cmc:s as ~ bas \ \:ou!"t >, etc, itnd f'l1gulad}' made 1Il.Se of the cNpressi()fl~ • peu haul', ~ peu ,long', etc." \ 'c w()uM not: of course talk of leslcal ,gaps; as we ha,vt: seen :!ntonymry is th,e.(ln:ti~lly di~pcn:o.ablc (d. I). I ).

9./. M:arkcJ' and 1U11ml1rked ,terl'fl.!:'

1I.'larki ng· {or marir,edness "}. whi cllI derives fr'om the work (If th~ PlilI.gJJe Sch001 rd. Vac.!:te:k) 19(\4, ~9M),is 3m e"trcme~impoTt~ronct. t ~ 9'tructuml ~ingUistics. UI'I(ort't:!n1!t,ely. however, ~t is a conoept which COlVer), a numb<:l' of d isparate and independent phenomena. Tn what foHm~'S, we shan be eoaccmed ,vitip marki I1Ig Oldy in Sf! fiLr as it lis rele-, nt to the analY!'Iis 'Il.f lexical structure; wI:! we shall di~lifigl..lish three !>M,SCS in whj cit lexemes may be cilc5>CJf1beu as m1I:rhd· or un !illarllied II ,

W'e win be~in ,'!·ith what mal)' be called ionnjd markin~ The ,,;vord:s

..

L ,~

"h .. ""t··"ho"·e""· ·" ... ---t'·"-- te > '1' • <'I" .' I'

, oj» ,. ... "", .... "un • ,coun ess j 101:1: 1(ln,~S,. ete., aremarp )0-

l,ogi,eaUy. lor formally, relared ccmplementaries (d. 9-]). The: forms of the second member of each 'pair (e.g. lio!te5!, .!w.,Uesse.) colltain a suffix, -~.ur" wlilich the f[\i:i:"iiiU: of the first member' (IJost, lmus) l,ad. This suffix ls the fermal m[l~k ,0;£ the opposit ion, as t~ pn:fixcs 1111=, in-, ii.i- arc'ilie !'Orma~ marks )0:£ the 0 pposhi~,[I, in "!l'iC:fldty ~: • u [If:~Le~dly ", 'COla sisrent ": "i IltorJsi~stc nJ 'II' • 'Ilf'spec,tfut" '; 'disrespe.ct£l!J,1 ',I' de. (d. 'j nne m,e:mber of lile 'P!I!ir of opposites is charac:terited by 'the pn:s,t:flI:.'C and the other by the ab~no~ or a mar'k [Merkmal] J' ,(Trubetd:.oy, 1939; 67)). In eases 1ike this! the, notion of marking is based on the presence 0,[ absence of some partiC'lllaT' dem!m'i: of form; ami tbe lexemes whose forms co:ntrun this, element arc said to be ('for.m.:lIIl}l) marked for- '~he eppesition, in contrast .... 1:th the lllniD8I1~t"d mcmbcnof ciiq;:h PM!:, """hieh lack the elemeat in qUeI1ition. Not an ef tilesco('lpo!'<ltinlls, it should be noted, are such. t.hilt the forfAAllr marked term would he ~~I;~(iribed as ne,g~dve QIJi semantic gWlIlnds, No x- is .it the ea'5e that 'in 'Fm,,]lIlaliy related opposites one must he fm'maUy marked ami the 'Other fo rrnl!l]y unlnarked; cf. 'useful.': "useless!. ,! fTtl.itful' : ! firu,itLess'.

Now £orm,al marking com.m.only, thCiugh not illvulably, ,ooIT(1l1att:'S W:lm. Ii difference 11\ disuibuti.,o~: the formal.!}' marked member of the opp~hi{ln tend! 'tQ be mQr-e restricted inks distribUlthHl (i.e, in the range of contexts in which it occurs) than the' formaUV' UlIllulrked member, But this criterion of distribut.iona] restrjction is j~dcpcmJ:enl of form:d marking as such and m.a:y be applied equally ",dl to fOlfJ:lla.]ly Ul'lre]atc:d lexemes, As we haveseen (9',]), what m~y be l",~)!:das the negative - members of such; ,o,ppos:it ions as I hi gh ;: <low', • go(!-d ';. bad'. 'bappy':

I unhappy'. etc." do fiot rlOrman), OCCIU' i I1i sueh se!'lte_'i"lce~ ,a"s "How ... wu X?". In eearexts of this kind the Oppo;sililLln is. sa:id to be suspended or neutraJiz:cd ". It is an important fact, about the SUl.1 eture of languages" at a,U levels, thu, 'w:lien an opposit'ion is characterized by Ioeraal mal'kin_g, it ls the fOl'mally mded member thQt ls t:xcllld~J. from the neutralizing' contexts; and Te.cogniti(lU of '~his genelil'll COITcb.ti:olll between fo:rmal :m.arkin~ and distribution has been responsible lo'!' the extension of the terms c:ma'l'kcdi' and 'unmarked t onpu.rdy di:~tJ'ibl.!tionl gl'Oumh to pai'rs of form.illy u IiIIfclated lexemes, Hut ~t must he e 00 ph!l~izcd that two distinguishable pr,upcf"cics aTC' invelvcd ~~H', and the use flf the term 'ffiarldllg' for both c~!'l lead to canfu sl 01'1.

Let us eonsid cr the pairs, ~ count": ~ couarcss" and 'I:io[l' ': • H()rn.::ss ", In each case, rhe second member is fonnallj' marked and the ~rst member is formally unm.;!."k:ed. But the t\i\"'Q p,airs differ wieh respect to

the criterion of dietributinaa] restriction 'or l1eUloozattOn. '[.ion' has 3! wider distribut ion til an 'I,ioness':' m.ale Ilim' and 'female lien" lire acceptab Ie eolloeations, but • mllle Jionos.$ I and i female lioness' arc not, (the one be~ng conrradieto ry and the other taI1l.Lological). The oppf.l~j t urifl betw-CCl'l 'oount' and' countess' (Of' between 'pnnee ' and 'pri:n,c€ss'}" hO\~71~\',c:r, i 5 not !'Ie~tnti uzcd in B~miLbr contexts: the oo]Jocatt'O!l'i!i , fema~ (; en unr' !lind • Male co Ilntl:Ss' al'C EO[l'l:r:ad ido!')" whe'l"eas 'mAle DQIJIlt' arid • female Oi'l'Yr!le!l$' lire tllutolog!c31. We can draw a dj'stiJlcriol'l. therefore b-etwee.nf 0 rrnal marking and distributional marking. "rh~1'I both kinds (If marki:lDg are 'relevanf, 'they tend '~(l, coineide (as with , Jion ' : ' lioness '. 'happy';' unhappy;., ete.], llu~ rhere are lnan formlllily rmlIk:~dI lexamcs that are not d:istri buti onalty marked ( .g, "eouatess" in relation to e coua t '], Amil there are many distri butiQnally marked kli!ei1~es t11!U arc n(!t formally marked: notahly, the negati vernembers rtf such {'Omlilly II nrelaied ,arutonyrrurus pai I'~ as ' goodi'.: ' bad' j '~:igh':' ]ow' j etc.

DiJst:r:~hl!1t~l ma:rkiFa,g' ,oor;reta.tell with, :;11\0 in ma[l· cases. 'can be p~allsibI,. ,m.,-plai:ru:d as being delennined by •. :se[ll\l:n;l.~C mark;i'n,·:~ !lind, oru::c again, this is pri.nciple lfldt:-pend,ent of fonn!t~ marking. A &e-11l1ntieally marked lexeme is one dlll.t is more specific in sense dlan th.e corresponding s.e:I'(l:a:luical.ly unmarked lexeme, I Llones!>' i.s, IlIJ,m: '5pcci.i'ic in sense than j Iion I. as ' bitch • is more speeific in sense tll an the formally unre] ted' dog." For' lioness' and I bitodh' denote only femaleswhereas 'li'oQ;)" and' dog' can he applied, in. m.an~· ,oon:~exts" eo either males or females i and i.t is t:l1lr this reason that the C;OH0C31tions 'male llon' and • female l,iOJi' ,and "m_a]~ dog' and "female d1ilg " arc ®ql1atly acceptable. 111' such cont!axt$ th,t~ oom.3.Jlt ic oontmst het.\lreCIII 'li.on' and 'lioness,' and between 'dog' amI • bitch r is f1.cmrnUz.e.d. In other contexts, ho,wever, artd most o!)v:iOlJlsly \,vhel'l the .opposites are empLllyed in a disjilnc.tiv,e ques.tion (Is it II ,deg 1)1' t1 bitcli?) or a SClt'CD1:e-nt in whiCh ont: is predicated and, the ether negated (It's 0.. oog, .rml a f11:tC'h}. the l1'llma.rked lcxeme assumes a raore spceifie sense whi.ch ~s inoompatible with the Lnhercn:(ly specific sense of the marked lcxcme, It shoukl be ]totod, however, t:hl1it, ,\ ... hereas all sematlti.cal1y marked lexernes IU··C (by virtue of (heir mere specific SCllS!:) d1~tribution.iny JIta:l!kc~. Ute eanversc does not: hold. X /rllS a (/fJ.!J can be uuere ~ to make a true sta~rr!en't \vheth,el!' the ani mal f'L'l-lerred to is male o:r female. DliH. the P mp<ls,iJtion cXpitt:!lscd by \II ttr.:ring X has 1Jo.riJ1~d tl 6ig lIQU'Ui' \\lC1'tll~ be g,t:'m::mHy rtgarflled as fals€: if the hoase WatS in bet small rather than laJrgc ~n relation to the rdcn.nl norm.

So far we have dl$CUssed hn1uliiymy {and quasi-hypolllymy) under the 3ssuitnption that it i:& necessuiJy ,Bill. irreflie)\ivc n:b.t~Q[I (d. 9.4). R'li:lt thi.s assumption is questionable in the light of wbt has now been sai.d about SCHLMUie marking: tll',:u the unmarked member Dr the opposition. bas both a me lie gene.1'a:1 arid a. mo,re ~peci'fic sense ;Jicoo:rding to context, Sjnc~ 'dog' is< .~Qrn.d~mes in ro!1Lf~t with' birch" and! sometimes SUpf1'a",dli!l!ale t'lli it, it fOniil'~"S that in ,Wl:ai!ll c:~n:ulIlSla .. ces ~ dog' can be :ilL hyp 1JIl'l]r'm of .itself. ;. Is that dog 11 dog 0:1" bitch? j,. is a mean] ngftd" tbough perhaps, ra.ther odd. acnreaec. If this wen: an i.solated pheno!'Qe.IHl'll in langillage. One might be inclined to 5;1..)' that • dog' had two distinct ~Cn5CS "dog t and .. dogt" :tii1l,d, th:ilt in One sense, "dog-ll'" it wa't supererdiasre to 'bitch .' arid in the Qtber sense, .< d{lg~n, oo~h}':ponym.ou:!l with it. Ilut dle phellllJimenml, is 'i!"ldespre:ad throughout lilt: vocabulary or English and other ]ang11ag·e:!'1. It is, a. direct ennseqsrenee I)f semantic marking and should not be treated as an instance of polysemy· {d .. l3.4-). As, far 33 the .re:l:at iOIJMlip ~hv'C£n ;, dog' and • bitch' is concerned, it; s aiS ,Lf the l~ldcal sUuc:lUre of English does nat expect us toO be eonceened ahout the sex of dogs ~nle$s thc)' 3in: female. and then noot aJw,aY!I.

It is worth emphasizing ~hat, when there are tWo[l lexemes .for a patticubit species of ani mal, one Iexeme being semantitaUy marked and other unmark.ed for sex, ~t i:i net alWaJi"SEhe lexerne which denotesthe female ~\!hi,ch: is markl:ld, 1119 it is ~n Efig~~ sh for' ]j,on I : ' I~ eaess ' •• t~gcr' : '~ig;l'ess ' , "de:cr":'dac', an' l.1 general lor aU pa,ifS 'err sem;1t1'tkally mad::ed words deaetlng 0 -dOifl'.;.stkated animals and bird. The word 'bull' is

~.- '" . I· " '. i." l' '} . i·· 'I '

'rnilon<!e .. 11'1 re anon to row. ,COCK. ,or reoster In re anon to ~en

and (fOf' those 'Speakers who would not 1lI0rmal.ly employ the .. "on1 , ew,e') 'ram' in rdatio·.11 to ~ ,sheep'. The relSOll. fot this wnuW scr'm to be tbat males ofthese s:rt:t:ics are [Iorma.lly kept in smilLh.=f numbers by farmers. than females, arid pun:ly Em' breeding.: the main stock is female, and this is t.rf'atecl by thc' lexical structureof Em,glish as the unmarked norm. 'l,vh3ltel"Cr the reason, the thceretieallyimp ortant point is that" in the le'J<ical~zali[oiO ofa distif'lction of SC'!\:, for somespecies it isthe lcxcrne !J.llnotin,g males, and for other: ~pccics the Iexeme defloting fc Ili:Uih:s., tba.t is scrnaaticallj' marked. Tho· imfllic.~tion for comptlll\ ncial, alla~rl>i'r; is that ,:1 slrlgtc l\\·o-vll!lycd feature of [lluS-Qr-minu& male or plus-or-miaus female. cannot be generaliliedl over the \'Qcabulary as ~ \\"hole (d. 9.9).

r .~~ON!' we! ul ormsl (leratLon of tl:U:SI~ and IDt her cxamph:s shm.'"S, lh:ll~

whether a lexeme i~ semanticaJl~' uamarked or !lot is ill matter 01 degree . . Dog", for many speakers or English at least; i ~ complctdy u nrnarked semlmti.ca.lly with. respect to' 'bitch' in dru:t it (;30 functi.on without

restrietien H a superordinate in :rclation 'to ~ts. m.a:!'iKcd h,·p®ol'li~'m. 'Cow' is less unmarked than "dflg'; ,andlho: same .is true of • hen' t( or , chid,cit. '). and perhaps als() of '. sl'tecp '. One mIgIJ.c "'tTy wel] refe r 'to a group flf animal& by means of' an expl'ess.ion like ··tttose COWlS. (over there)' without thereb}- implying that the group contain(!;u, M bulls. 11m the same: expression W(JUJ ld pl'ubably not: be used 00 refer to " grou r cOll!sistiliD,g' so'~el.y of bulls. }\ga_iiil" oac rnigh~ 'In::!] employ the e~prc:ss:ion • mille eow j as iI. r·eflexjye O[ metallngloll.stic gloss .fOT • bull": hut "male co'!.'" (unlike • fc-malc dog ') is [lOt 211:1 aceeprable colloc;.ui01l .in nonreflexive UKS of the expression (thougb it was apparently used in. the niileteenth century as a ,e:lIpnk']tJ.ism. flil! 'bull.". It ill ~df~mntradktol'Y, jun :I.); 'female bull' is, Nor cai'l one s.ay clIITccdy T"ka.t Cli'filI ,'& II hull {as nne can s.~:y TIIf:4t d.og if a .bitch), except of course in sitlla:t'il!llu in ,vhlch 'that COWl is to be construed :LS meaning somethillg like" thaI: anima] whi.ch you have (inoorre~tly), described as 11 cow".

Even less unmarked than 'cow' i 11 relation to 'bull" is r man' in rela'rioTl, t·,go ',,,,omm]'. j f!.1an' :!Ipd 'wo:m.a:m' are. n.J l1i que among counrahle COillmOn nnune ill English in thaUh,ey cen be used in the singul!!if, \\fith~ out i'!; dete rm.in.er. !'Is. generic II rderri~g'pTessi(lns (cr .. 7.2.); and. • I'l\lIn ' is. more com:mnniy used in this way than 'wemarr', Now j man' when it is. employed in the singular in l!l generic rderring expressien i.S1l0- marked; d. l~ is "ta,~ ,thot ,~. ~.es-Pfm!ib.le /Qr enf-'1r(mmtn~'1l1 jllJli.[{HQ1!, to which. the r,eference of the expressien 'man' r1l:ty be construed "g" indud~ ing 0[" excluding women. Silmilady, for the pJlJlrnl (If 'mae" "S t 1>rrr:.rj,c reJcn1llg Cxpn::ssioiii : cf. Mm ha,t:1! lit;,t;d all rhi!1 uifma Ivr telJthmIS(md' J'Nr,~. But iii! most, if not ell, other ki nds 'of expressions, whether nJere foulal Or' P mdi.ca:~hrc, • ~nii:lfi. ' is not heM to be ,~u,emrdinilite to • woman '. Not Qidy can' om: no't :Sll}' oor,ret:'dy 'Thul 1lI-(ltI i. (l WO"man (e!<!cept uru,kr the circumstances noted above For' Th:tt cow is. a. bull '). but one would not normally emp,Joy the expression "those men [over there)" bur • 'those people (over there) " in rd~rring to a gmu p ~ontain~ng' one or more WOmJf 11. If' man." is said to. be unmar;ked in rel atinu [0 "weman", i.t mus1t be recogl1ized ~hat this is so (li).ly in higl1!y n~sl:li'icn::d eireumstaaeea.

As we have seen, the sense of :II. hypoonym caa gcncmlly be analysed as. the p mallet ohh.c sense IJ.Hts superordinare and of snme S)"Ha.,grnat ie modifier oithe sl!pemrdl fl.at,e.L!'I.flglil~ previde t iaf: m&.'ln.s. of constl1i.q;~flg:m i ndennitely ~a~ge set ef l:iy~mll)'!flo'us expressions by- explicit syntagmat ic l1I.CdibcalilloJl (' book' J ' ]:JIfg'~ boo1.; , t r largered beok' I 'etc.); thar hmguage;(! enahle U!I ro do this, and to be as specific and .a:s precise

jiO

JU

in dcseri'bing persons, objects, activities, ere., as the CiWllrll$hUl~S demand. depends upon the ~ign feature of rrodl~c:tivi,t:y" (cf. J-.j,l. lVlanYClF these phrases, by ,,':innc ·of their Irequeru cmploy,mcl'lt in OO!'l'~eli;,ts whleh neutralize, m render ill.1Ipp$icabLc, ,ccrtain 'of 'theirimp lieatieas can, 'i~l the eo;!] rse of time! acquire a m~,r·e .s~ ecial;ized sense. 3S, can single words uader I,be Same clilmlitions. Vl,rhel1 this happens they are well on the wa:y 00' l!.chi,el'ing the. status of plli.r,asal .Ic:>;"'f'mes or ~Vien WIl'rtl~fe.x(m.cs. \i\Te wil] discussthis question more fuJty Jater h3.::=). HCfiU we !Itt eencerncd to relate i,t to the rhCR!Jmcnon .or kman'li,c mark-, iflg.

Ollf e~~rn:plcs will be 'marse'" 'fem.ate nurse.' and 'TIIulc nurse", on ~he one hA'~ d. and I student " 'n'l;:de sluden't' and 'f~mllle student'! on the other, The reratl~n9hip' bet ween t he lexeme • st udenr' and 'the hjponymous expressions • male studcnr" and 'female st udent", Ct:.lna:tl.'uct.ed, from it by syntll,girnati C mod i Iicat~Qjn ar;lt::Qrdimlg to the' PfoDduClh'C ,'[UJles of tbi; I alllgu age-system , i.s s~mt:ghtfo'l"ward enough. F:rom a s,tatcmcnt like My CDtuin. it a ~tlldellt mnhi l'I,g can be inferred ah()ut the sex ohhe Kferent. of • my cousin '.and there i . no' reason f;ol!' us, I~O' thiink (If j male 'stmlcnf' or ~ female student' as single phrasal.. kxeme~. From the statement 1111' ~o:uritl i$' a l!Iurn!, hewever, moot speakers of E:n,g,"sb w~n lWeI." that 'the -person bc'llg referred til is female. Is t.l\is ilruerence based UpOIll an ~mpUcation \'\rhich belongs to the sense of iI'l1JfSC" ~ And, i'f 810, JOt:'S 'nun:.e" imp.l)' I female' in the way 'that 'cow' does, by virtue (If being wmantilCl!.Iay IlnDlMkeci in the laltlguage-$]'st&m in relation to <male nurse'? Or i!1; rt-,C: infe:Jiencoe: pr~hi!.bil~s.tlr ~ill'lg d'ct:Clr.rnine-di by our k.nowrBdg:~ that mo I:. nurses, like most secretanes lind rnos; sludcn~s of damesti,c seience Or' speeoh tnif:l'.mpY. hap]Jel'l t.o be ~cln:Llcf

It is mguab~e tllai • nurse'. by virtue of its seosll:illJ the li!.flgUl~gcsystem lit the prese.nt lime, implies 'female' (or the dis j uactina of • gi r'l ' and '\\'oman')1 Hld that it is unmarked in relation to 'male nurse'. first of aU, ~t. shnald be noted that il"fy (muin U. a ~flale ttl.ltt(!: is a perfectly 1IIIu'ln"~ \l tteranee (,."h~re;as l1il;;V c(lMsh~ it a /'em't11e i!I.!fl'U ig dec~(kdly' odd}. Furtlilermol'c, Ilol ,oruly is the phrase • mal- muse' of Clompa..rati.vely fir!]qu~l'It eceureenee in 'C"I'cryd'ay discourse', bUl, ~whm it is !.Isei.l as a predlutt \"e expression l rl spoken EIlglish, each ef it!'l co nstituent words is ,gi\~C'n equal stress, 'T'hi ~ of it ' If is, il:1'i ~nd,i c.ati'OIl that • male" is not be Lng 1J5ed W modify • nUI'S£':l~ a strllightforwarJ aU:ributiYe aJjootivein im'plici r oofltrast wit h 'f&malc'. The ml cs of nurses and male fl'I,USCS in a haspi~aJ are, tt}, some extent, distinct. Ttl say that snrnenne LS a: tn_ale nurse (when. 'male nurse" is stressed in the normal way) is not to imp.ly

'that be is a nurse whll n.:a;ppcns to be mare, As we le:n;v isa!l:f,cl, aehange in the implications of' husband' and 'wlfe' conscqtl.cnti.alllJpoB the posslbl.e ~i'!~titufiQI1a1i.zai: len 0.f nomo,scxtlare malTi:lge wi th the d:isti netion of the; husband' role and the \\,ife's mil! held more, or less constant (d. S.l). :;0 we can ~n!l LSllg'C !II change ill the i mpiicati{lI'Is of 'nurse" and ~ male nurse" sueh t.hal persons fulJilliQg one mle would be descri bed as [JU rses and persons fi.llnl1ing the other- role would be dlescribed as millenlll[ses., lIe:ganHess {;If th,eir &~x. At the: prcsii:!'It 't~mc, howe\'er.. • nurse' and 'm.ale nurse 'should perhaps be f,egilrded as le:remts whi eh are related in sense a11 the IJl:lm-arlred and marliled 'I1i.l;mber C)f an oppOs~lion i!1l the !!'Mabu.LIt'Y·; and th~y are mow Wi ke '00\.,' and • bun' t]Uln • do,£,' and "lJitch" mterms of semantic m~rki.ng . .II may be pointed out" ii!l. this oonw.ooOl]; rb,at at the' turn Qr the ceftE1:H1 in Beitai n the cxpressin IlL i Jady typist' was quite commonly employed. in contexts (c.g.; III ad,'cl!'ti!'icrtl¢'rlt's} ~n whick I typist' W(illild mily,' be Ils,e!li.

In our discussion [if semantfe marking y,"C have done l.ittle more than paint out some of the d'istinctians that would need to be drawn in a f'IJiHer lreatm,enl of ""'.h;a,t ~s, a cOfIlple'x and oOllitrovfr&ial subje~t. Some of the statemcn'ts 'I:ha, hA\I'e been made abmd:. ]il:arti.cular examples might: !be cih.aUe>nged on factl,lal grounds. Tb,ere can be IlO' doubt. however, that semllntic ma.rking iss. matter of dc,g:I'iBe; and that it is, an important feature of the l~xical, slna ctU]"IC: of Jallguagcs. The fact t hat an the e:xamp les used abo,n hal1tl had to do wit In. thc lexi e;a]i z;a.tlall n r the d.istinct ion o.F sell: in hu.man beings and animals sholild not, be lak'~11 tt' imply that senm.alltic marki,ng is :p (i.'C'L1I~i1r' W this distinct ion. W'Il have c"nct..1l.-,:l'ted upon this distinction pmly because i~ is :rdatin~ly str9ii,g',ntfQr\\'iU'd and part[y hcceuse it is &0 ·ofwn i'l1\I'Qked: in dlseussions of semantic m;ldillg. It is rare for- authors to discuss, differences of de,gr~~ in semantic markIll,g;. iHl[l a oonsiderablc ammmi: of JC~Cfipli"'c work on \'''-l'i!'lliS bngll<l!g,es w]u be reqJuired, bt;fof'C .anythi!'Lg like a oompre.herls~~rc tL'\e9t~ ment of lhe subjed can be \lI;nttw.

9-8. Par,t ..... wllole n!l'ati(}l1S

Mention. should now be made lOr a stlrl'lC'whl'lt diffe:nmt hie!'ilf~hical re atlnm;lUr' from hrP?1l,Ym. ,: ~hG ~m-wh'Q]el reIatioflshi • 'lllis is €xemjl,lified by" arm J ! ~ hady'. • wheel • : " bi c -01 e', ere, I n cases I ike til is tbe d:ist.incti'On be:t\\'t"tll hyE'S!l),!!\Ji' mild part-whole r~lat ioJ1.s 15 _ _cleM ~QtI&h. An iIDn is n_ot a kirul QI body but lI. Jlart of a body; .alld pll.ra$c$ li~e "annsallcl other kinds of body' are nQflsensical. As il number of ,a;utMno have podn.t!~d out (cJ. Bicrwiisch. ~96S; Kiefer, ]966),

9.8. P4rt-e:hol't relations

3~J

pil!rt-1tl!holerdlation:s kLW(!'€:Il lexcmcs are ~ r"lHtd up with a particular' sub~dass (If P'Pss~SS,iVif: col'lstn:u;:t(ol'lS, cXlcmp.uila;i by s u:ch s-emant Lca!!y, and perhaps ,gr,arn_m;u,icllll:!" :related phfil;Se~ arrd sentences as, ~ John'S - d ;nt arm' llind ~ Iobn has a right ann '. .POSs.eSSll;'(: construe [ions of this k..ind are j,ll,many la.nguages, though Dot in English, distinguished grammatially from such phsases and sentences as'John'~ book' and 'JQhfi, ba;s ;;I b!J'Qk', the fo'nnoer 'bei 1111; de~l:iribcd as i:'!..:diernl,bl~' ,an.d the ~au:er

as alimil!l.bl,e· pClssessi ~·CS.

- P,art-whole lexica:! relations are at least as dlverse as the variQUS Kinds of hy.E{mymy f(ll.ll~d in Ja!l8uage, ;!Inti! we ;'iU not attempt to' discliss them In. detai t nne ql!estta,n that has been debilUd i n several recent trea:ll1lcnts ltJ<f the !lubj eel i s wh~~t her the [!:J"t'1t-~~!hole rdatiinn; !Lice - hnmnl'!'lY, is tnm:siri\'c, The f.act that di Ilerent aut Imni disa~n:c aboue this point is perlUlp an imlicat.iml '~"at hc'[c a:fa "':lrrio\IS kind, (if ru-t-whole r~lal:io[lS in ~aiiJfllag,e and that the logi1c,"d dift'eretlC:~5 between them aTe g;,eater t]t:tn the_pirr~flc_~s between _v~,lii(ltl3 kinds (if' h)'p~ny.my' {and qm~5;;-hypol1l~rmy}. It may :I1sa, ere II refiC:1(iQh of a failiu'e to maintain a consistenc dist i nction between the plIrtwhole relation:as it hoMs between tile refcrc[I\S o,E ,el;;pr~fis:i(ln~ (i.e. !'is

a relation whioh holds, between the pa.rate l)o~ separable components

I) f a th.ing lInd the whole [hi rig of 'lirh Dell t h~y' are componcms) 3,nd ii, srrUlctllm'LI relation of sense in the "ocabul:a.ries of lan,g;uiL,gcs .. The ~~whole re1adoflship which holds betw-ecn ph~i~nr !fscre'tl: referents

; is elesrl ~siti"e: jf some dung ,I( i iii palt_gf some dring'S which is

a> part of sam tlling_~. t~cn x :is ah,nys describeblc liS a Jlart. of ~. ~sitivitl also holds, due allowance being made (or a certain degree ofindlC'tC'tmimu:.] irr the referenec (it expressions in such C;ISe:s, whell the referentsir; question arc n(j~ physkal tlbjt::clS, but points Of regiol1s in physical spncc (or space-time). I(.lI is at point or- regicm. \\'hich is j:!:aft - of a re io~ y which is {l'll'ft ,of a I\egioll :t, then ;1( is a part. of e.

The F::u:;t th;at O!'lC enti ty may be de;::.e1i"ihod as a part of :arlm he; em Lt:y dees net il:llp~y> hOl;.\l\C\I',er, thlllt there is 011 'f'I:ltt~wh(lle ,rd:uion. holding ~n the vccsbulaey betweea lltt lexemes used in expressiene \,·J-rich ref(~r 'w these entities, Foil" exall1pk" ij, certain object.v may be referred t!> as • rhe Jl3_fidl~' and be a part of anoth~r ob,kct: j1, ,~-bich n'il)' be referred 10 a ' 'the door ami be a _part ,of a 'thirJ obj~ct ;;;, which rna be ~reilrgl,to as • t11C hoyst::'. ;'i; is 11 part 0'0£ ,0;: (by vi rh~ (1 f t lie t mnsi t i'l'itr 0 r the partwhole relationship holding between physical entitles], But sentences like 'The house has, a./no handle' or 'There's a/no handle on this house" are" ,to say the IB3st!, peculiar: a!l.d: such phrases as • the house- hallJ le '

of " he htlndJe of the ho,lI,1se' :lire dlefil:li:tdy ul'I~C(e"~3b'c. Sucn phrnsesas 'the dCi.f!I r-w.m:lIe' !lind 'the hilnd]e of the daor ', as weU .._s [he sen~ I renee "The door has, a/no Imndlc' arc perfectly 3.c.c.cptabte. So too are 'the door of the. llaulOc· (a.nd possibly 'the hcuse-dcer"] 'and "The house has, IIfno door'. '\\'~ might therefere be inclined to set, up !l.parlwhole rclatl{iIllirup o( !Efise b.ctwcCri "handle" 3'111 d 'door" and bet\l'~en "door' ~;I'id ',house', but not bil1'1:wun "handle' and' hnuse",

'-(here are, however, many probkms attaching to 'the nQtion of part!",·hole rda:!ions h@td'~g between I' remes, If we sa • that ~hey' areby definition intransjrjve, we shall be forced to recognize anenorrneus number ,00fp:ll'it-''i,"hole lexical pairs, m2lrw of\:\'hlch eeuld be eliminated in the an~lysis of the '''ocl!butal1' by mean5, of general n:diJirulancy [uh:s,· but!d on trnn5.1I:ilrity., as suggested by Blerwlseh (~9,65). For examrde,

• CIlji[':' sleeve" and "sleeve";' jacket' !lre part-waole paira, 50 too is

• cuff I ; j j acket '; d, ! These sleeves, have no cuffs i. • The sleeves or thi s ja,ck,ct have no cuffs', f This j:ae.kd has no cufls·. ] n ord 1:[ to account 5j1S~cmilt oc:llly far the acceptabi I~ty of these dl.f'ce sentences and for t.hci r semantic relatedness, it would seem 'to be essential to In'vo-k.e the no'tlcm of trnnsit;"ity. For the part-whg]e celation holding between "euff" and • jacket" is surely to bercgaf.[lcd as the produetof the part-whole relariens holding between 'culf' and 's]ec'o'e' and, between 'sleeve I and

, jadl.(t \ The pi:01bkm, then, is that W,e have examples like "handle "; - , door • ~ , house'. on the one hand, and 'i:uff':' sleeve ' : ' jacket', on 'the ether, The readeris i vi ted to censtruer and consider other ,exa:mples of both kinds fClT himself. If he does this, he will scum get some idea of' the Mt1L!rC of the problem, T~ _ th::tt11Irt-'wh.o.le Iexica], reIations ~re- 1'I0[l~t ransIdve, rathell' 't han being all u':lIJiJsit i\'oC: or in" .• rn, 'v',~.s true ,enough; but 1~ hardly adeances our under~tandjd1g of ~he ~.=·J('t.urc of tn~ voc-abulai'ies of languages, Wilat. i required, if it an be fouad, is some senera! prif!c;p~e which wauld enable us. to decide, with reference to the sense of particular sets (If lexemes; whf'd'l.er' they eeasilnne whillt B~~ C::IUS l!art-''1ao11il~ains (Td-von-Kencn) in thie vocsbulsry, withou,tspccifyil'lg for each. k1'l1:'mc, as part af its sense, the phu;c j,t (jocupi,~s in. a part~\ .... hnle chain. Nonc of the Jl~CIU treatments of the slJbj.e&t, m!Jmin-ating though they may have been iii'! their discussion of Ila'l'tiC1!lIla r sets of lexemes, has revealed a1l1Y 'II Iablc general priuci pII:9 of the :kindr1e-qllircd.

It c(lU'I.d be argucd that the l;l'hole question ~s llTelcnnt fOf [ill,truist'ic sema1l\ti:cs: that it ,is aU ;!J. matter of OU.I: gcncrral l:mQwkdge of the relatiDns which n(l,]d between entities, in the exte nal Wfi,l':ld_ But this will not do.

We might wei say, for example, 3!1I1cl plal1silbly enough. it might: 3.]'pcar, that • door ,. 11:'!.S II particular meaning lind 'hl)!.!~e' a particular m-e.i\Ii,ill:g (ailruy:sab[c :im terms of M:Jil5C and denotation) Jlndi that toe part-whole l7ielat ion which was ,assi.gned abov'l! to the .exicaipair • door • ; i hlluse' should be a.w:ibuted. insteadte OUl" klllowl'tl:d;ge of the purdy ,contingent fact that allh,oml;es. (Ol' all normel houses] have deors; There are. how!.iv,er, numerous Ieli:em~s :~n the 1.-oca,1)1;li3.'l"ies of l:.mgu0.gcs whose meaning cannnt be- specified ill.depa'Ldicntly of some part-whole relation. of iSCifiSoC. r low lOOuld we hope tn analyse the mean Lng of ,r sleeve' o r I 13 pel • W'itlioutinvnking a PlllI'lt-'I,\'110Ie' n:lat:iOiIJl between these lexemes ilnJ • toot '. 'j ackct'. 'ga.rment.·, etc. (as weI.! as ~ h di iIe(en.t relation whi.eh he]ds bl!hn~t"n 's]C€l',e' and 'ilrm]? Even meee cofl\',ifi'(:lng: are sets of wa:rd s I ike I seecnd " '.mi flute', "ho ur", 'da,. \ • week " ete, 'nile meaai ~g of' d.a}!'. I month' and I yeali' (and perhaps "week ") could be 'c1tplained, at m~ast pardy. wi.thou. men~ioni]lg aHY part-whole :l'd~tiol'ls, holding \\'.ithin the set; and it. eould be 'regarded as a mal'lel' of CQl:ltingc'flt fact tlilat there are i1l'1pr'Q,ximatety thirty days in :I. hUlai" month and between t\Ji'elw and thirteen (lu!rlar) months in. ii, year. nm .it is in principle impossible ro 'exp]ain the mcaniltlg of '~!!('ond'. 'nlinute' and 'hour' wi~bm.lt. s.1)eci~yit'l:g the lParl:-\'I·J:u;)]e rdation~ h.oldilllg w,ithilJl tJH~ set; and ,~'e co~W1 not d iSlinguish between sola r months .( or c~lend3!r months) and hHlar muntiu wi'tJ:iollJtmcntioni I'lg the [la.rt-wihok :l'ielat ions wiithin this set {If ]ex!emes.

.. TllC difftrt:'rlcc between 11]" ot'lym f and ~art- .... ·hQkrdations~ iit was

'saiicl, is clear enough in cages m:c "arm";' bodly " 'whce:l':' hicy,c1e ' ; l.e. " h.ell th lie emes in question are nouns d.el'lOti,l'lg discrete physical ob·ccts.M.ost gf the ~fi.5<:ussim:l. 'Oof pa,n-w~l(lTc lexical rda:tLIl<I"i$ by .I~n~ guists, has been l~.&l:liicl)ed to such eao;;,es_ J'tis :.IrguaWe. however, 'that" other parts (I. sp .. ~h heeides concrete I\GIJ.03, dcnming diserere pltysical IJbjects may stand in a part-whole rebtion; aJnd the distinction between. the two "elations ill sneh e..iScs is often far from obviatl.S. For example,

~ ~Id is b~]'tb a kil'ld of rg_atle.r al]fl_E_par1_l)Lm51,!UL We' can sa)' equally well .Tlrll mbstafll.€ }laS lJold in if er Tbis substance ,,;mmlS (if/is ca!flpDS~ of gold (ami! atlifT 111~fa!f) and 1~M5 .mmt{1m:~ is cold. We C3I'iIilQt sellsib~y say Tliis ,tfnimal cOllsist1 oj a t:CJ1),.' (~IId' other 71rammiJ!:S) or This boJJI' i~ lin ann. Abs:tratt nouns" lib concrete mass .110111:15, with ~:hich. th Y Rave a ~:erurin ,1;icllJl affini,ty (d. r ~ <d. m_ay a~&no berdliJtl!:cl: both as hyponyms to -;;: slI.<erordillate and ~~ piiTts t:Q a ""hole. Hone sly may be regarded as ~ ::1. kind of lID rtue :lind 3 lSi;! a part of virt ue, SO H]O for many verbs dcmn [ns - ~e~. Fur example, the proposition •• X can sew" mar be held to

imply :II cO'l'Iju.nct ion of .. X ~Il t~u;:k' '" "X can hem "'. "'X,~ baste.", etc, Eachofthe\.'e bs in'tlu~~ct Clack', 'hetTI'., ilbgstc'.et~.J I. ail, po""ym of • sew 'and may yet he :~a,i ~ to denote ,u, ,i!.C'ti,,~ ty WhlCh ~s t" :'tt o:.r tile activhy denoted by 'sew', Thelli' few cX:llmpl~ win serve to illustrate the ''''ay ~n which the ho.crarchical relatig.llsh~F" between [cxemes may. for lcxernes oth.er' thMl counrablenouns q,Jcnoting diseretc objc<ct.s, be treated by lang'",a '(; as b - ny:rny 01:' a part-whole relatiun: Of pCJi'hapS as a rdation which is ineermediate bet\\''l,;tQn t:iKm :md shsres certain characteristics with ~hem both. Furthc" complexities and interrelations~ips CJn~rg'C whtm we cOI'l~td·er partieular types ofPiLrt-\\,hlll]e rei" ions sueh as that of lleirlg~a-t!.:mp(ltill~~slioCC-o£ (I:f. the put-whol.'i!' rehltiom between Lchildho'ld!' and ' life' find the h)'l,onymuus relation which !toMs bt:h;;;"cn 'c~d' and d PCf50!iI ')". \'il,c wiil not disDUSS the-sc"

Ment.ion shouldalsc be ma.de in thls secti.on of 'variolls kinds 'of coUe~tilres· > such liS. 'cau1e' i' ---C-1ergy' r 'furnitu re "1 ' her~' • Hock ", 'f:imil y' , "li bra ft". CQilective nQUlil s may be defined, sernJllTht1ical1y a.~ I~cs wll:icb denote colleetions or g:rouf!s, I)f percSorls or objects. III English, ~hey iaJ] into n. numhe r 0 f ,d i Ifcr·cnt grammatical clrtSs~s. • Cattle; atl di ·dergy', fOj' eJC3imr·!c. are treated as plurnl. l:lUt.'furniturc' as singular (d. . These cn.ut~ lire ... ' ; ~ Thi~. furnittm: ~s .... "], Oth~s are :singular with respect toecncord \,-ithin the: noun-phrase, 'bout (in Hrhish English at least) may be C(!inst:ru&di as either singular or p1uml for tile f!Il,riPOsc of c:oncord wi.th the verb or verb-phrase in the sentence (cf. 'this fmlmil)' ':

"The famiIv has decided .. .' or 'The family have decided. _. '). The

- ,gra~mmatiod ilmhi\'alence of truill.f coiilecti\lc~ wi't:i'lres.rect ~o the distilll~ rion 'OF sin,gular and plural is, to be .explained of couse by tile fact thai a [;(i~lec:!ion @'fobject';marbe r'egarded frem one poin't or view 3!1. a single I!:Illity. but. from Ilillotlilt r poj nt of vi ~\\', or f Q • tither plUr. ,OlW~. as Ili pllJLrali.ty. \Ve ha\"'C already mentioned tilll.l f,lurul 110U .. ,::nii"ah.et te.g. "those men') flJn~ti'uni l'ig as g'C'R~ral refeml:kg' ,t:1'!.p,,~sii®ns are SQOH:ld mCS employed if! order t'o 3S~ribe a c:e.rta.in property til each of toe members of ii. -class, but that they m8.y also be used t'tll assert i>ometh~[Jg of ,thl;! class as II ""haJJ.: (cf. 7 .~). N oun- hrases collt:aining coUectiVC5 areli~ noun-phsases i.n this I:'espt;:ct· _~ntl it i~, interesting till flOte than when such nOlm:~pihrases, r'C:f~r to gtOtlps or human be:ings disu1buliset.r. th~y nm:cs!!arily seft:tt the re];i.E,i .. 'C' PII'0l10UR 'who' (r.tther than • which ') ;J'I;)d plural QD.Ill:)ord.Both of the foUo\~~~ngare ]XIss!!JicJ.i1l British English},. the fgrlner with distributin:and the latter \,:idl <;Qlh.:ctiv· reference to the Govcmmll'nt: "The Go'!{cmmenl, ,,,ilo have .. " -.H: • '0 'Th~ GOll''eml'l!1;fit, which has, _', is .. "' Ullut neidu~.r 'The G~:r"eromcllf.

3,:r6

9 ·9· Componential antIlysis

who has.... is ... • nor . The G evemment, which ha~'e.... ate .•. is grammatically :!JctlCptabie.

\Ve are here eonecrned with the place' eeeupled by rl(lUccti,res in the structun: or the l.oca.bohry. Many of them cerve as soperordinates in relation Ito :3J set nf qJua.si~hrponyrm. his" howew.1t", quasi-hyponjrny o{ ,il dlfi"cFent kind f:rl!ln:) d1:!J.t noted Ilb'{l'\'e in eermcxion w,ith such c.xam,p.l~s u ., round": • 'shape , or "blue I:' oo~o'lJlr '. For example, 'rc.1ttle' [·s s'u,petO'rdi:nate to {' cow', ,I buH'. I steer i" etc.}. a~ is s hewn by ~he I'egular usc of such ex-pressions as '00\'"8;, bulls ii!nd odlcr C3u],e; and 'clergy' js superordiuate to r bishop'. "prlesr', etc.]. nut there are d.ii:rerenc.es, between these twe exa mples, AI tnCll IgIl • priest '1I.n(l 'hi&bop' are qt1asi-hyponyl"llS of • cI,~rgy" :Ii_S • co,w' and 'l:n.d1· l!.l'C of • catde' (or • mall' and • ""OmliTi' of" people '). "priest' and • bishop, " also stand in a particular kind of j)!lIr(,-whole rdilitiollJ with respect to 'dergy': d. ," priests, bishops and ethee mernbers of the: clergy'. 'Furni.lu:re' dift'ers r~om • c~'ergy" grammatllClllly. but it is sem Il.l1tlcaHy parn]kl wi lh it: cf. • ta,b.~e~. cl"C.l~rs, and o~her kinds/itl."IRS I]If fng mtt!:l I'C ", There axe Illany :such ~oUcct""fS in the ll0-Cil.bulary of Englls.h uKI other languages g'hic.h are superordinate to sets of lcxe mes III a h ie:r:a.rd[icni 1"elat iOl.'lshi p dblt is ambivalent with 'respect to the distinction I]If l1yponymy and the partwhole :re1at~on. The fa.c!: l"hal thaI: is ambiv,a:lcl1cc (If this kind correlates ""'mth 'the fact th.at mrn 00'11 ectives, wbethc,f tn,ey are grammatically singular lOr (dural, 1I1:"e. Yell' similar, s,f'mantically. to mass nOl1llnS,; and we ha'l",C arrell!d'ys.een that the distinction between hyponymy and the part-wholc iFc1iati:Dn is Jess clear-cut with superc rdinate mass nou 1"15 than it. is withsu pen:rrdi.rmtc (:ouml.tab] e nouns de-flmi Ilg di~.crcte physical ohj.e c ts, It !hmddiilsi:J' he nated that the fU'liti:;~n.o'!ll. of such wo"ds aa < ki nd • , I, part ", • memher", "item'. etc. (in expressil]lns l.i.ke • kl rids I]If ani mitis> 'members of th~ clergy', ip:lJrts of the body', 'irCfl1S of fUirniture') is comparable with that of the 50~called clasei crse ill 1a:1II~i!lages which draw no grammancal dis.tine.ion betwl.'cllsinglliar and I"hmll (cc. 1,1 • .+).

.. 'nat her kind of c[lUecel\,'c:is 'c~'!cm ,lined by • ftm: 'lu:rd I .. 'librlll1" ami • forest'. The rdlltionsh ip between 'slH":'_cp ~!!~_ :.tJock' i I cow' and , herd', etc., is eleaely not OIlC of !:!rJ!o 1'1 Im'\': BY c·h_ pheases as's heel' ani! other kim:!.s o( H;ock' are nonsensical. Nor is it apiiirt-\l~hllie relationsiltip £If the ,samerype as th~t holding: bgM1;f!11 'arm' all(l 'ood.t_:. Co.l~~cti yes i i~,e 'fl,ocic' serve mw::h, the sam e i rid ividuad [Ill' FliInc~ ion ali wl'lI,nlslJike < PQ~I.' ar ' 'pound' i n ~ two pools of \,\",atet' or 't hree peu ntis" I]If butter' (d. 7.6). Tht:1Jt: is ':I, diffen:rmc of eourse: 'water' anti' bulter' are mass nouns, whereas "sheep ' ,is i countab.(e nOlin, bch sh "P in

th [! fl ock is an i,l'Ini"i.ilual. '\,Tbat a eolleet i\lle Hill: ., fl Delt' dot'S is to i ndivi t~uac 11 set flf u ndj fferent iated i ncli'!iid uals in th I:: way that "pool" Or 'pound" individuates III quantity.of Wi!UtJ" I),];" butter. I £lock mny be; c(!mrw~ed of sheep ana lambs, a'S the e]crgy is composed of brshops, priesas, etc." and a b~tlJ is, oomJiii~se-d of !irm~.. le~~" e~", fJoc~s" the c1~l'gy and bodies flla.y all be oo:osit]ered , from l~US. [lOint '0'1 VI ew, .:IS coUec:tu.oIlS ,of ~Ij'l:it ies, nut • the flock of s~~p·. unlike 'the clergy of prirosts' lInd. "the body (if legs'. ~s an acceptable phrase. 'FJoC:k· •• hero ~ , 'forese, 'Jlbr.1ry', etc, are like the rnore generel words 'sc~··. 'collection', 'group'. ctc., c~t~pl. lhal 'th!.:)' are SYl;Itagmatic:lIIl}' IfCstrictcd (:u'Id (hi'S also is characteristic of ml'1I y, I:mt not all" of the class.incrsil ill. various Jalll,g1!lagcs: cf, :I 1.,4). llcings)'lltagmaticaHy restricted, tlu~ mal" encapsulatc· the sense of the lexemes which denote members of the collaet"ions in question (d. 8.2). Thephrsses • a he-rd of eattle ' and' a suite of {llrni,turc'inusrratc the difference between the hw dillt'H'[Il types, (If IZtlllccti \'(:5.

9 ·9· C(Jmpmle!1 hal (lfJaly:sis

-It is probably true to !l!!y that themlljority !;If St1'1LlC.tllrnl s>el1iulluici~ts subscribe nowadays to some 'fcrsi()ilI._ or ~her of componential analysis·, This approach tl]l the description (If It\e 'lUc3.ni:t1g 'of words and nhrascs ;CStS~llpOI[l the thesi that the se1l..se of every lexeme can be'

" . -. .

;:Jllaly:sed in terms 01 :I set of more general sense-cemponents . or

semantie $e::JtUiliCS·). so'mt: or all or which , .. iU be common to se\Terali ditfereni"lcxemea in rl:;; ';f.I~:Jbulfl,ry. 11'1 So £a'l"3;S ~mpoflcntia1 :m3l~'5is is assl]lciat~,dl wLth COflCieptl!l~I.im;1 (cf..4,.;l)" t~~ 5eilUe-cc'mpommts (fo.:r which there i;s so fllr' no gcncrnUy a,ocep'tcd l(:ITIl)~a.~' he thOlqth~ c:f ;jSc atomic, 3i!I d the senses of 'PlIrtieu lsr Ie 'eme-g as, mnlecelar, cO l'iCC'p~9~

- For example, the sense of 'man' ((,OtlS;~n.li;:J as the CQl!I;picln!.:'iiHUY o!f '~\' an ": ('J. 9,1' might be beld to comblne (in the molecui -r ",Of Cl:pt .. IIlan ") tbe atomic ceneepts "m le ", .. adult j" and "~uman . ~ IUld l ne sensccf ,r woman" (Yi;ll ... woman") might he held to (hlT'er irom that of '1fI1I1l' soleI.): in that hoo):nbinc§ .. female;' ~r ., !1ot=~.i1Iie "). ra~hof!ro than !' male.", with "adult " and" human". £om£<!netl'tl'l anal 'SIS Iflt 1;"-. preted ~rI t hie!'; ",mry. Gift be. rel ated to tim id ~ (If L::~b n,iz ~.d \¥itk ins 'which, :IS, we saw earl ier, served ,as 3n i llilpil'1Q,ti~n re R.ogd In ~ he rorn~

pi laction of iri!5. thcS<lurus (d. ~;p). . .' "

. The ell.rli!:st, ,i!,nd most hillluentaa.1 pmponmts of oompmumtml, !lrqalysl~ in the ~{lst·S9,l!5SUl:e!!.n strucrurelisa h"adillilllii \.\~re .Bjdmsle\' a.l'ld Jakobw sOn. ~[~heir' views are not idcn'ti,cal. but they are similar enough as far

;jl!l

as their ad,'ocac:y of Olmponential anaIysi~ is. concerned. tb.ey boU\ believed th;u. theprindples '!.h t TmbetzkllY (~931J) had introduced into plumo.logy eould, an.dI should, be extended imo bnth granUlll'lr lInd: semantics, FIl'~ma5t_3mong_thl! rc~cn.tatlvc;s ohl'lis charnderistkaHy European versiol\ I!f oomporlt::ntial: :J:ml~y"Sis 1tr~ Gn:i~ (I96S., J~'7o}., Pouier {l9'41. Priem {~964, (966) .and Ceseriu ,(cf, C08eriu & C·edu~lcr • . ~ 974J. ~en!l:.i31 .analysis an America 'i!.~aril" U;l have developed md~ml!.:fitly. It 1.Ya,.'j. finit propooSoed, U(lt by lingui~ts, as a geaeral theory of seman'ticstruetlJl\e! but by MI.thro(!o·lo_gists ss ~ tt:chnrqlle for descrihin.g and ,comparing the ,!loc:abula_ry of kinship in va.riou!I tangn3g~s (d. Goodenough, J9S6~ Lo_unsbury, U)56; "I:\'aIJace ,& Atkins, 1960). On1)' some years later 'I\M it taken up and gcncn]i.zled 'lby sueh 5'choI, rs as Lamb (uJ6.f);. Nid.i1. (J9%. m9iS) a.nd\Vt:inn~ich (~963. u)6(i,), as w ell as by I{au:. and Fodor ,( [gil;)}, in their seminal papr.:[. w]~ich led to' the i [!teg~ati,,1'1 of semantirs :md! Gynliil{ witth~n the fmmn'11J1k, of transforrnatia .. aal grammar!! (d. TO.S').

We win I'IO'~ deli s%,~ematic.~Uy with the simHarides 'and diHl.:rcnces between the several vern ions Q£ cQmponcnti al 3JIl;;l~j'sis mentioned ~i'l the prev;ious paragraph. we \Vm OOIjIC1:rnttate instead u;pOlIJ some of tlae more __ gt!ltcnd lhe-Dl'eti~~ and methodological ql.lc:stiall.~ that ilny v!.'rsi,on of componential analysis must face;. and we will begin by frii;oducing 3. ;n.o~ationa1oon"cntion l\rhicb will. ellllbt~ us to formubte rhese questions more deady. Om:r'.clrlvcl"ltion will be 1101 lJs.c smal1capita:11l for the = represeruatlon of' scnse-eemponents, Instead of saying Ithat: !, raan " is the product of .. mll~e''', ,. adult" and "human ", we VI"".iU s~ t:hi!l. "rnan .. ~the m.~.ni!'lg. er more precisely the sense, of the lexeme "man" ~ d. 7,,3)

19 the product Illf~R, ADm.:]' and IHJt\!IJIJ'\I .. ,",rj-mt .is. :melnt by' ~'n.ldllct.' h'C:lil! is onl! of the quest lOllS that we mu.st c1iscus~. .Another i.s the rcb.'tiJOnsho.p betweea ~'I"'t:1l and" male", between A1.lt!l.T and" adulr", het.\"i"ccn HlJlI.1_~N and "human", and so on. For, as "rnsn" is the meaniog: of the English Ie.;"eme • man '. so "rnalc " is the n~.eanillg 'of the English lexerne 'mde;' lind "l:!ulIllall" is tho mc:ming of the Englis.h lexerne • hU]iflU '.

One am. we. ~o the_:; uestinn wln~thc.1f !If AI_I! is to he id ent uf,i.c,d \t' ~lJru

.. _I." ." .. lL " ..!I ~" d . L •

male "I\.I)[ li.. T wlhl 31J1Uj l ,an sri on. ! S mar the rc IS a sharp dis-

t il1Jction to· be d rawn, i nprinciple, between t he m~ni figs ,I)f lexemes and 'the I1!t,oml.c ,concept!>, .O'[ st:llse·com~rJHl.ents, into which ~l'Il!s·~· mea:;ings can be fadorii.:!'jed i and ·that, CQI'lScquen t.1~r, JI.nAl£ lind ADUl... T !I~e not te be idenrl ned wwth .. male" and ,. adiJi It ". MALE. Ant.Jl. T ,. etc!" are 1M:I!.! ~o belong to a set of universal. i'ltomi.c OOItlcep15 whllch ffi:ly Of mil.y Rat be

lexical.ized* in particular languages; and. Icxicalizatiofl j·s held to .oonsisl

in provid.ing a ]r.:i'1ieme.. w.hosc meaning contains :at ]east one of these atomic s.eu5!l:-c{lro:ponenls. It follows that diffe,eflt iall agct_,,,'iIIn,fltneeessarilv ]eli:icali~c the same sensc~ci)m~IlE'l1ts ami, ~nJ>a· fnr' ~~ do l.exica.Ii.m thessme ~elll~c-com .. 'lOI:!itS,_lMt tht."1 "",'iU not ncccssltrily cOl'l'lbme them in till: same ,my, \Ve wilt prn·vi:siorullliv acc"pt this point

of v ~tW I \,,'h~ch. 115 we h[llve al rC"~d:' seen, enables dl.e· sti'1Uctnral semanticisl

to aW!liid the more extreme kind! or re~ati"'~!offi (d. 8.3»'

""'Y,e .must nO'" alii kwhat is meant by tile rerrn e. raduet 'when it is said

f(l'r ,c:x;rr;p Ie, tl~iH "man" i~ the 'Iuodutt ef ;\tALE, AD IIL. T a.nd tIlIMAl'l.

En thi~ case, it is lausibk ~o .interpret < pr'!l(hJ:cL' in terms of the COnjl.lllClion of scnse~cQmpQl'lent&: th.e ~x!~CfiS10H of • man' (coli§.trllcd as the complementary of • woman ') .is the tfltersectwn Dr the classes 1\'1, A and. H, whose .nten:;;ion~ are th;z I.ltmnic CO!'lCept:SIUAI.E, ADULT and }lUMA~, respecrlvely (er. 6.4). It is this l.nternret:tion otl!..roduct ,(though - it. is rnre1y madie e:>iplidl) which scem~o~fld~d,:e mOS~ of th~ earliest wO"rk f'I]\ ~®'nl".l]'Qne[ltinl anaI)i!is, bod,. iEl!IJ'OB~m :;tnJ Arn.eri·CaTI. For example, Potti-e.r's (~96.J) wdl.known analysis ofthe French [exemes

i ehslse', 'fauteuil', '~.IIHlpe and 'tabouret·· (roughly equ ivalent to English 'cl~air·.·arm-<thair·. "sofa' and 'stool,) :in 'tcrms.nF the sensecomponerns FOR sn:n" G UPON., WHH LEGS, 'IlntH A BiKK, wnn .-.m.l~ llnd FOR ONE PIERSON is presumably to be understood in this way, So too

. H' I lcv's ] \ I' c. ,,-.' •• - - . ; "1> •

IS :!:l,e msicv 5> U)591 ana. yats. {I:! . ram, ewe, - man , woman I OJ.

, gUFI'. ,. smmon ", • marc' > and Katz ;lind Fader's (u)63J analysis 1ilf what

they t(lok to be the four distinct senses of 'bachd.or·. _

AnaiYlles, oJ kinE)'Jitip \'~'H::;3I:rlda:ry,. on the other hand, t,ypically aH.ow f,Oii

- b(ll~h the di:sj unction and the cortjunct LOri of smse~or:Hnfl.[lnc'l1t.s,_ F,o:r l.;!\atl!E1e, on the ilSiSnmption that t.his is in fact t'~~e cerreet analysis lind .~ [hill; [l.ot onb,r tl:IAUi, but 1I~_s,1) ·~lle~\,.(j"pLacc relationa'!___p:redicat~ 51\01J5€ ex J') ami SlpU. G (x, y), a're atomic ~n6eFts. the sense ~{ "brother-in-lnw ' Inl lit be r·c)resenh:d (],[I -ut at l~~) ~. 1I1M.£ x' &_ (SI>O\J!;[..oF.sml..l"'I:-oF ('\', y) VSml.lN(l-OI'~~;OIJSE';()f l .rl~ \s this example shows, {moe we .combine both c!)njl!l.ncriol'l ~nd tlis.jiun.otiol'i iN':' - must introduce into the rcpresentntiofll'lf the llCn...."-C of h:'xe~s (by mclInE

or braekets or otherwise :-cr. 6.4) ill distinction between suh C1as5Cs <Jr~ (X,(Y +Z}) and (X.Y) I-Z)I. Fo~' I(X.(Y +Z}), but not ((X.'!t)+Z), 'i~

l! )o:Jt.~fiJ~j f:,Ii'laUy i thmti ~ ~-ith (X. Y) -+- (X.Z): e.g., if s: is y's brother~in.

law, then .'1: is dt her both. mate and the spouse !'If the sibling of)' 0:1: both male. MIld the s~hl;ng of dle SpOUSiC o,f y. OLl.f ulm;mplc al'so sho\.l'S: (i~ ·that, ;if 're]ational pn1di.cates like S!~'IOI:l£E (.':fl. y) and S~IiILING (.~, y) an:

admitb:d into ~he stock of .:l'U>Ill,ic concepts" there l'IlIust he- some way (\vlucthcr by lIsing VllJnibl.es Iii!!!: ;'( aDd y or m:herwis.e) III,indi,catillg the diTu:.tiooality of the relalion; and (ii) that, jjf oii'mple:l; relarious like. S.l'lll.lSE-OF-,$~BLlN.G~OF (.r, y) arid 5lUU.NG-Dl;'~llr"OtiSE:wOF (x, y) are emplo)re~~, they must be de:lincd in such a l;"il!)' that they an: not nccessarLly eq,l1ii.\ml~l:lJt ]t w:iU nat de, therefurc, bj say tl-i(!t tIle sense of SI lc.:>!cme is an unsmsctured set of scnse-eetnponems: that" brf.l!hi! r-in .. law ", for instilnre~ is ~hc proouct of lItAL~; Si'OLSr: and Sll:lUNG ... \s We h. \"C aLlJead !>CI.'I1 (in nur brid ron. id~'[a'lion, om EngLish aud Russian: d. 9,2.') SII!I,Ll'(;.oF-S:E'OUS:~-OF and SI'OUSE-O[.'~~UJLL G-OF mayor:' III 'Y not be lexicaliaed by mea s of the arne kxema; and the sex of )'; in' tead of or in addition to the sex of ;0;:, Ill:l}' be I::l"itcri<!I'.M,l\l~ (X)I &. MALI! (y) &: srIlUNG-QF-liI'OUSE:-Of.' (X',y). l\W.~ (.~) & l'Ei)'I,1I.1: (:v),", SIDLING-OF-::wOUSl!-OF (x, y), MALE (x} & ~tAU: (y) & SP()lJ~ll~OF-SI!lUNGm .. (.1', y) and ~tJll.E (x) &; J'J~:IHIU.E (Y) & SI'OUSI:-(.W-'sUIUNG-OIf (.\"" y) are all, .. in princi~r~1 k:xioa.l~y i!listinJ;fllillha:blc; ami ....... hcther ~he~r are ~Ii~~ ti'fI,gtli,shed Of' fltFt in di:fl'crcfI{!M1guages is 1Il19iiUiC1" of romii1lgcn~ fact.

Indeedl it req'll i reg but a mOIIl~I1ft"S n:flt:xj'tJ'n to sec 'th.at eertai 1:1 more cQmpkx com'f)i]l<l:toons of Sl'O'[j3F. and uU .. [:<\G are possible and thall: whether, ana how, they are 1e)('Lcaliz;cd in particular languages, is :I(SO a matter ofoorningel'rt faet. M.an}' speakers of English I(though apparently nO c all) subsume S!'()USE~F-.sIBUZ'iG-OF'-SPOIJ'llE~OF' (,\:, ,). but not SIDL[NG·OF·'5roUS:E-OF~s~m.]NG~OF (or '). 0'1- KPOUSE-OF-sm l;.rO.-OFKI'IOUSE-OF-S!fll.ING-OF (.v, y), (It SI.flLING-O -&l~Ol!lSE-nF-S'l!lI.I;.lG-OFSPOtlSE-O~ (.:\., y), eee .• mid!:f "brother-in-Iaw' and! '~,isler-in-Jaw'. AU 'Ilf these 'relations, are l,e!l~calbzabk. <lind it must beposslbte to specify \"11'1 tel'! of [hem are grmipc:d ~logethcJl' i'~ the same lexeme an d wlti cr;h are not. h lnllL~t, also be PQ5sible, ; n princi plc, '0 han die ccrlai nf,ec,y r:sil'e. combinations of Sl'OUSE (x. y) and SIIlU~G (.~. y). o.l'J£S ~ib~ilD.g'~ sibling is either oll.r-self Or' one's sibling. But in II nlln-mllllogl\nuJUiI sociery one's sptlu!;e's spouse is not ncceilillanly on~~elf. n foll.:;I\\'_S l'h,lI,t a simple relarion like SI'OU:iE: (:1', y) is illfillill.:ly rcculI'$il" , .andi, unli.ke the rnuch more obviom:ly recursive (rulll preslIImllbly nen-atemic) relation of being the anc[!:!;;tor O'f, it is!lt'ln-tnmsitiv!: (d. Lt:, Nom:: of tin: kLTl.",hip terms of Engli sh 0( ~pilrt from "ancestor' a III d 'dr-sccndan.t ') W~\! Ii I seem ro jn""()I\'~ recursion. There are other Ilanguage-5, hnn,\",en:or.in ,,:hich f~ u~ n-cu rSl,,'e app.~icatio f[ '0 f ~ he same tl~:(lm.i,c; reI;'lt ion is funliamel'lta:l to the componential :IJII,llly&is of the "o(1cal:II:llary of J;;Jn~,h.ip (d. Lou.nsbuty, 1'96",). As far as ]ECIlglish is eoncerned, we must specify whi,dll or the il'lfilllitd)i many preducrs (If such .Pl.dtllivc at-omic lI't:.!alil)n~ as SPOUSE

(.~. y) and SIBLING (x, y) arc lexicaiizcd andwbich of them are 11lClt;, a.nCl" as we' have $een. this cannot he dQneimply by I.iMing the sense-eamponents that !Ire combined Not t:. err relation ,coFltaining both ,Sf"()'lJS£ (x,y) and SIIlLING (.,:, y) is I~~caljze(l as 'brother' in-Jaw' or ,j sister-inlaw'.

I't is ilrglllable that t:~t' notion of product \\·ith \!,Illicit 'weop~rate when \q; :!i;::1JY t hat the S~IlI5e 01 a lexeme is, (he 'prod uct of ~ set of .1t'Dmic concept S must be- even rich er tha [I the one tha.t \?II1! have cabo rated so far. According to Weinreich I 66, , lexernes hal'e !lfI internal. stru.eturS,:" wh ich ~li ;;:;'rn l~lC s ~n,tacl[c st ructure ,of ~(mtllnces and -hrases; and this potl'lt (If ~;ic.w hots been adopted by the so-celled gem:rati,'e sernanticists· (c " IO.s'-:-~ example, '1'\: 'c-Ca\\'lruJ971) h~s sugges.too that the ~ of the verb 'kill' can be analvsed into C!\US'Il, n:~ME, NOT and A.LIVE and that these elernenrs are ~t simply conjoined {as, ret LIS :SIIYI 1I1.!U.)!, all LT lind fIU1iI.M>: are C0njoi ned i n '[he sense 0'£ • man '), Dl.lt are combined .in ~. hiera.rchi'ca] §,tmctlltt. wh,]ch :ma}' be represented here (\~·i.~h ehe o<mis~ion 0 f eertaln vari ab~.,t:iS) as (CI\USIl (smoM l! 1(l'{OT (A.tnrll))i)J. .As.sClc~ated l"l'ith this. 03l1al 'Si5 is 'l~c Eurth~~JI_r~(J.!~:~J lhi!t, illl this ,-:~_c. .English h~);;il::1Llhe5 not enl the wJl~ie complex, bu. each of the constituent olilmhirla.!_ians.:: thi\l /lUVE IS k.>;[c.1Tized in, • alive", (NOT ALIIo'.E) as ' dead I. and {BECO~'IE (NOT I\I.!.\,:£» as '. die'.

\\'c wilillot here go i'Dlto the details of this ~nl1ilysis" It is suffident .fall' the present pUFp05Ie to point out that, (In the assumptiGli'I that CAUSE; lH£CME, NCH and ALIVE ecrnbine til yield as their preduct the sense '0.[ th e verb • ki ll", they must be combi ned i 11 ~ hi'C:11I.I"(h ical structure of (he ki nd that .is marl! fcs.t I n tlrte oompiGX '~'Np'rE:.'!!s'iD'~ 'cause to become. n~t a.live ~, nther than tha;t wWeh is m:lnu£'e:Sot :in, let us Sjl)', 'CrlUl!(l notto blilwme :I[i,;r' or • not ('tO~ cause to bcciJmC' alh'e'. A~ we 'shall sec Later, a 'smncwh:Jt diffcre~t vi ew of the ,in~cma] s~ructu[,e or Jcxcmts is taken by theec sehelars whe base 'their theory of ,gr-arnn\ar on the not Lon of valency" (cf.12.Z). Hilt they '100 would aTgue that the principles or operations, by means {If which sense-components are embined ill. lcxemes in the process of lexici:lii:;lation aTe essmtillHy 11 .. s.111le as the principles O{i" eperations whel'eby words anti CJ<prc:'lsioC1s 31 .. r-""nll;;im:d in S)':rlltnCltcatly well-Jcrmcd s~ntenc~.

Enu1lJ gb has been :53i d 'to slflow that m!ltriC!~.s of the kindthatare Qft~fi em ployed in I is~s 'of sense-eempnacnts must be ,supplement,ed, fO'l" SOl'lle ICj(~ul(,s lIl. ~~a.,,:~, with a r.pecUicatiofl of the way in whicia the Sensecomponents 1I,n: COm bincd ; and f .utherrn~re that the ir c~mbinadon cannot in all instances he acceunred for in terms of he simple operaaiens of

conjnnctiol'l and disjuillction (with or without Fec'lm;ion).We ma,:y ROW take 1111' a. number of 'olh!:f points.

The first has to do wjili th,e question (If bina.lri15mii and the use .of fe8tnre-·nQtattO"fi'_' (which is commQnly therugn not Ilcce~!lril~'a.'i%Oci.1ted with bin.arism). As \!'-t; have seen, the prin~i[il~ of dicho'wmollS contl:'a:!>t i:s: of consid,t::rnble imporu.m:c in the lexical structure of lafl!fuage.s: !'nany pairs of Ie'Kernes cal'l be decscr;ibcJ ,as antorWms, (11': oomplmne:n~ tar.ies (cr. 9. [). F urthermore, many I)f dl.l~ (lppos.itioml that hoM betwee II. antonl'HlS, :allQd oompbmerllarlcs can be regarded .BS havrng a marked and an \H'Im3rli!oo member (ef, 9.j'). The thesis of bin:uisrn·', all We shall .intllT,prd tile term, SllYS that ~LIJ lexica] contra.sts are 1bnth <liclmtmm:.lIIS l!IH~ privlltiv!:1. _1\8 the pftu:m.:mcs 'p/ and [b/in FreD'1.cn st:tndin opposi.tiDn to one anotl:u;r on the pn.(.lfIological dirn\:n:;ion ,of voice, so dUI • man' lind. ~ womaa", 'boy' and e gi r.1 i. et e, on the semant ic di.l'I1cns~Q III of selt (d. Greimas, 1965: 2011 ;,Pottier I l!l7 4: 6 I ff).ancl as fbI CiI,n be ~J to conlain the phQoologi,ca1 ,feature of voice, wl'r!icfi /p/ll.cks, so (i'l might be a_rgtl.ccl,) I,' mal'!" and! .. c-oy" contain. the stnse-oompill,ne-n.t MALE, whj cb !~ \\'Qman" :m.d "g:irl."· Jsek,

~IJ.t why, j~. m.ay HOW be ~~. do \\'C s!ly that .. woma~" and '"girl" Jack. dl~mponem MA:r.:r., rather than th1l:t .. man ,. and " boy" lil.Ck tile oompom::nt ffJIl""t..EE As \\'(: h;Jxe seen, it is • i'tl;_'UIll' rather than I \Y'(D!'n1Ln I that is thc- unmarked membC'1l of the f1pposit i6l'1 j thOlloS'h I man" .is by no' mei!l't~ :1.$ comph,rcly unmarked as <dog' is in ·re,lation to. 'bitch.' (d. 'I.l, 7). If we apv:'- to semant ies the same ki rid <If co.nrsi(]cra:tionSrthat Trubctz.koy ,. 93~ ~~rlt rood uced illl (II pho [lolag),. It is clearly pl"t:fi:rnbte to i5a)' t.nat the scnee-cornpoucm \\'hose presence o\f absence di s~ t'il!lgtJ i:sh s .. woman ,~ from ., m an " lind .. bitch. ". fmm .. dog" is I'IlI!IlI.LE. But thcn~is lUI ~lJch reason to say th;a'~ • boy' ~!I st)mal'ltkally urnnarked i III relat III n to • gi.rl \ • ram' in relar ion. to 'ewe'l)f st;tUion i in rda~j,()n to "mare'. It is equally !'!.pproprb,te to &'1y that "boy" lind .. Jam" 'lI;ol1ttai fir MA.l.'F~, which I~ :Iac'ki IlS' in .. g~r'l." and ",eW,1: ", ael; it l~ to say that ~'gid" and ",ewe" contain JFEM,i\Ui, wJJiic:h is !lIc:Kin,g in, "1:l{I~';" and! ;. :!'am " .. This would run be a '!lMMem" 'p;erhap'll, i f i~ ','!'er,e not (01'" the f.J.ct rha tin. other pairs e f cemplernentarics, Ii ke • eow :. < bu'll ~ an.d • dud:. ' : ' drake ", it is thl1: orte d e.not ing Hac !Ilite that is sernan ti caUy marked,

If we b'lke the view toot ther iii:! UlI.i,rl:i'Sa.! set of atomic OOfU::Cpt.s whiehare ]c)oiiCt1lizc~ ~n ,;1jrt~cuJa'rhngLl"g.cs, the rar:t tJiat, as ,rID" as d~e distinction tlf sex "11> CO neeened, :it is sometimes. 1>H'ILE illnd 5Om.edmcs ~ that ~Eears tiD be pn.'Sc.nt in the meanin.g of ~he marked 'Illemb.er

of a pair of oom.plemeJItlllries ool:lfmnu us with a dilemma, We can a:rbitrarH}! select either MKLI< Of FEDo.1AL!E as ,;I rlBtaliit~ universal at,omic con c:ept an.tI define the other nC,gati1l'dy in terms (1£ it [i,e. We can (ll,'leJ:~lec:ith<..'T \~;itl1 +r1'tltUI.j,: anel ~r.El'\tUI1 or with +!vIAI_E, and -MA.I...E); and, as we na\'e j~ is t seen,. this will k3d tnthe UI'I$<!li sfil.ctoryall:d1l>is, d;lf certain pain of compleraensartcs, A'1t,ema,livciy. we call :lliow that bo:~1JJ MALE and IilIlMALE are alll"mic OOI"lOC,Pt~ i but 'this Wi"ll!"Id rum enunter to the w'iwle spirit of bi'lilarism, since it would leave +MJlU and! ..LI'F.MAVF. lU! (beflrrl~tiC3ll}' lIlnrdated" and pGte!'ltially oo-cNist:t:n~. comronent8 and (ill default I(I~ rol1'J.~ 5 u:pplcnumt:ilry s,t:J.tc-melrltOr'r;tll'e tOI thc' c:ifeJCJt: thal +l!L!'ILE il"llplics -fEMAl.E and that +I'EM'Al.E implies -:o,.1ALE) it would not cml~ribute i fI iltiyw.l'Iyw an. expl icat.io,n of the rd.atiol"l.&hi,p of en'tal Lmenl which holds bet ... ~I!:1;11 sueh pmpo:;itjon5 as .. Th3!t horse is 3. srallion" aad .," That horse ~s not a mare >, •

~ \V c have just iWl.tr,od II eed a :fUJFtnJ!!l!" notatiQ,na] eonventian, This, is the ('1 use 'of ,iI. ·plus-sign. or 3! m~tu,s--sign tu distinguish be ~~'ecn the I!_ooici .... c ,![lid-:the negatll'(; v,~Jues of wfultffi, meNd to tcci'l[licalll, in liHlJi:Uli$ti:es.,

as " fea.t\J1re~. Thill" term 'feature'" it may be add.ed, is, aho ,emlllo'fed

wllh r"e$[lcct to the values ef tile variable ; i.e .. nat onLy is the ~"3l'iiable ±.1>lM.I! (or ±FEMIllLE) d:es..:ri'bcd as a featll.rc, but sa also are its two values, +~II!.L."R and -M,\LE ('Olf +FEJ, .... L~ end -F,t:"tAut). We will. COI:I,~

tim.u: tu L!:lse th~ term • cempo nent.' :fot 'the values, I;'cserv.iog • feature '

for the v,arillb[1;; of whick they are \'allle~.

'Fhe U~!1: of featllre-fmUtion raises a fmth~r question, Ilf -Mt.,U,E i~ held tifireprcscnt, [lot a Qompo!'lcflt ')(J,wvalcnt to FEI!.1.AL.;, DU. the ibsctl,cc Q[ + ~UJ . .E. hQW do we ,t:a{ltIJ.Fe the differeno;. between i'horse" and .. mare" r For "horse" also lacks t'n.e component + ]l1:IIL! (on the ilSSu.mrti en til at the J eat ure in terms of which .. sta'lliun" alll.a ,. mailc " are dist~ogtlishlld is ± MALli;, father than ±,~EMAL£)~ and ri:t tbis point tt .may be adJel..1 that it is not uncommon for he term ;'unntal'ik!oo> to be used by ~ illg(li ~ts i 0 a> Wl"l y tin at ~ b5~lIres the dil~en:tlce ~tw(1lel1 • d!lg' or j duek", 01'1 the onc hand, and 'herse PI' "child', on. the ~ther. The words 'horse' ami 'child', in this "Ilsa,ge of the term 'I.Hl.m.:ttked·, lire said to hI: u~mi!idtcd for the feature ±M&I..F. ~u<r ±fEIIULF:). Hut 'horse· and "clrild.· are [lot the 11li'llllll1rked mem be'li"s (If a pri''itllJtivc !l~,poFoi ti an, as "dog' ami 'du'cl!:' lire. We must lm cil.n:ful, therefore. ND draw :iI distlru:tiol'l between the minllS-'1'a1l1c alldi the eere-value of a feature: ~.IC'. between - MALE (c.f. ,L. dUl>:k" ·a.nd '~hli'l1.E (d. "horse" er ",chllcl") ami between -FEMAU (d. "d,jlg") and PMi\t.E (cf. "·hl.l~c" 'or ''',ehild.j'). Unless this dIstinction is drnwliI. ,a pmp{ls~ttonJikc "That's a '!i.Ol'iiC Q,n:r

there" \'lj'iU be wnmgily held t,!] be 'll'qui\'3~ent [0 eilJiter "That'~ a stallion over' there" or "That's a mare ~1\'e:1" f'lte.c:", according [0 ",;llelhe.l"±f£M}I.,1,J! or ±M4~R,F.: is; seLe~ued :is tbe feature whose values, d i6t ingl.llsb "stallion '" and ., mare ".

Indeed, there is y,et a rlltther distinction that one might wish 'ti'!' draw: between thaw lexemes 'whose meanill~ i~ oomparible with. both the negllt'iveiro d the' pOS-Itiv,e valu e of'1 bina ry fc~h,Wil and t hOSJf Jexemes whose meaning: is compatib,[c widl neither the Ilega_tiv,e ,~or th.t:jlOsitive vallue. FOT example, .' horse ' i<!l compatible wjlli both ,'l!iIJI.LJ. and -~1I'.LE (or both +r-E~tM.E and -I'SJ,.tAL£), whereas" ho'Utle "., it might be ['Il""g'led. ns .'compatible wit_h IlIdth r. If tile disti;nnion between t!t~ minus~Vli.lue ~Ild. t:l1.I~ scm-value of II. binary featllre IS aecepted, the further distinction, bet.wecn the z;~ro="1!I11iJled • horse I and the non-valued ~ house' is rellrli Iy r'CIlr'cscl'ltoe..Q by layi Ilg that. WhCFcH the meani fig of • hoJ'S£' eentains P1i1lAl.E «(1:1' jJITloI:AL.E} as a o!lH11pOneflt" tin: rneani rig of 'House' eontains nil] ~'01lue of ±1>L"L,LF.: (or ±f'F-1>h\U.}. nut by adopt,ng this \\':!ly of representi ng t:he di.sti netien ~twecll "horse" "imd ;. h~1J!Se" with n:S,pect to the Ieature ±MALE (m' ±f&.)'I1ILE) W~ are obviousJy making the: feature .n questif.)'11 three-valuet] rather dUlIl b'ollry. An altun3lj;"oe ~~ 'tn licj"_l tile distinction between zero-valued and no(j,~ value<d lesernes and to S2lY that I'Ichhcr • horse" MI" 'hQ use' is spc(il1cd for any value of ±1\.L\1.E {or :::;:FOIA1.E). This is more in keeping '!;\'1ith rhe spirit of binarism; and it is arguable that it i.s suffici·e·I!l.1 for the ]!lurp03C f llmp(ll"leflri~~ i'lnI1JJy~,19 to ,ui:Mingllish. between minIll3-~<a[ueJ and Inl= i:noo. The fact that hloth ; male' harse I and • female horse • are normal read ily inter. prctii b le expressians wh ''fC~S ~ male, heuse" :m_d • :femll,1e c:' arc nat (0.11I [he aS$1J m.pnon that d'll:!'; ! ~ to be IICCi:i'u l'Ued rur in tJe~ of the meaning of 'horse' !toil "house ') is e-l\.j.lli~ab~e ill rerras of the- pres.;noe in 'Ii horse" of 'CQmpOllcnt like T&"IIIMATE and its absenee from .. ho use", 0 nly animate entiti es may be male 01;" femAle: a nd th.e fact that filere are 601"l'lJe ~pj,'!cif$ of 5~1'ik~ ot hcrma:piu®ditic. ereatuses is perhaps reasonably held. to be II. matter nr OOi1'tirlgcIlOY. rather than of lilgical neces::;ily. ;mod to be indcvam 'to 'Ihe deseriptiun uf English or 3ny ~th~r !all,iluas:e.

There ,(1,1'(., o·r ooun;~. m;tI'lY 1e.JO;ica~ c~tmsts which. rio mH illppe:! r to bedidhatil!IllIlU'S (cf_ 9-l}: and. as we ha .. re seen heth. here and earlier, eVt;'rI Itll iiJpllFiil1I'Cflt I)' !SLraightJ:or ward dich(1ltvll1'lou~ comrast sueh all, that whi.dt hoMs between 'man' lind • woman' I 'mill' and "ewe'. 'slalli!>n' and ' mare'. el:!:" rrl!sent6 t !t,e ,iUlllJ'yst w:ith vari [IllS rroblcm8 if.he \Va shes to treat it as ll! privative opposition. ecrnparable v;ith the phonol,ogical

opposition that holds lIetweClil the phonemes lbl and j'p/. Idl and Jt/. etc.

A further difficujty wit Iil. felllure-no[alionis, lh~u it c!lnmH nsru t'i:l:lly Fepresent the dj.r.tlncLiorl between cemplcmeneariey and antDnrmy W~(Rout fai Ii 1'1 II til represent the si.m:ib:r,ity be'~'\II"eCI'I these I;wo kinds of d ichi'l ~Qmeus contrast (d. 9.1). Bieewiseh (~969), fQ)r example, draWs a djstincti On between ","' !lait he calls s;~ngu~.ar rnark-eno like H I.I"'I<\J:", whose logical ne,gatliml, NOi:iI·UUMAN, LS, sim'ply its clln;tl1!.ciictOl,ry and dnca [lot denote a positive pror~ny, and ,",'hat Me refers 10. as. 3ntorl}'U)oQJ]s 1NlIptes (il'l an ulIIUJ~uaHy btooid sense of the ~,~rm ';!{~ilco:ra.ymous' deri\'ing lrom Katz, ~964. £966). lih::aml"l~~ of these so-called aruonymlllH. n-tl1ples il.'i"1; {1I1Au;. F&"'\lIi.LE} and {IH .... CK. warre, RIED, GREEl'l" ••• }: and it is suggest~d that each mcnlber n,f ~he set is 11 pos.itive value of an. n-'li"aln,cd feature (whe're it = Z). Thus, MALl! wOlddbe one of the ewo po.s5ibk equipolleet values of llle fca:llln:.sEX; lli:.Ae~ \'i'ould be one of the possible 'Values of [he feature COLOUR; and so on. (Morc precisely, wal!l.t We have jl.ls~ represented as MALE and BUCK: WO'u Id b~ bipartite seflS~,-CI)iI!ill:ponel'lts consisting ,of (i) a supererdinate marker taken from tne set !I.I = [SEll. iCUU:I\;;~. AGE, SPEC I ii.<;. • •• } and (iH 1I slIhotiliniitc marker, ~ .• specifying whi ch parttcular loC3.t ion withi fII t he domain denot zd OJ t~ • .c :supcror.din:atc il:larkt:]' is denoted by the subordinate marker. But we [Iced (lin go in't,1l! the details of Uienti Ch' formalism ~in the pJi'\:!icnt conn.e.Ni:CFIIL)\'Vithin this framcwm"k. it is casy enough 'till forrnuilltea ge:1!'lernl rule aI;'CO~ !lliRg for thcrdm:tiollship ~hat hold3 between comp:lt::~ Ht€nlary eqll~pollcfit values of 11 two-valued £eatlH'c, s:illce two-member n-tuplcs are merely a partil.:lIlllllr case of n-valllled n~hJpres. An}" l'ogicliJ relationshi p t na:t is defined to hold berwccn ;31l. arb; trll,I"Y member of iUiI tt-mcm'bt:r set and dthc.:r 'the OOUljUl'lctlOI'l Or the di.sjllllU;tion elf 'the .remaillitlg ,,- f members of tbt set win hohl between each member (If ;'I t~'Iil-memlb.el" sd H'l1d the remainin sili.,g!le member, BU'~ the imporLlInre (JIf die1li.otom:o(ls ~:fxi.cai epposision in language i~ such tI:mt it is rounlc.= intuitive, to ~I!ly the k~ast, '~O U·cal oomr1c-mcnt::uity as being [10 dilfcr~lu in kimA! from multiple equipollent contrast, even iheugh it rna}' be satisfactory £I'lollgh fMm ,:I, purdy fOrfllli!~1 pui,lll of ... i,~\1!;' to do rhis, Furthermere, a.lr~o1il}rmy (in the narrower sense thaI we l;un'!: given ~o the b!'fnl: cf. 9.1) cannot be batlJdlctl \\;thil'l thifi f:mmework \ itlholit int:l'odu:cing some s,lJ!ppll:fI'lCntary '1'I(ltatiDoa.1 eonventinn or S{U]!JC additional COlnpOl'lent. {such as the p(l~it'i\'€ value of lhG two-valued £e-a,u'Uf ±I'OLM! or the t\\"oKI'II,il~erdatiomd component GRE.,\TF.K (x, )'): d. ilicn,~sch. 196,. 1970} to distinguish IIIllto[lyms from cornplemenrarics.

'We win m.~c nO furthc:r usc of' fcature-aotetinn j i~ this seet ion; lind vn.; '''''til s;;w no more about binarism lind til. formal. or notalioul, proh lcms ~hilllt: It gwes rii se to, There an: various more' gefl.cralpui nts to be made ablHIl ComPlJ.l'lecll:tial analysis.

Compom:nt'ual31'131ysis calli, be seen as 1m cxeenslon of ficJ,d,-theo'ry and, more panicularly. as an :I!ucmp~ 'hi' put fidd-thooory ena &IilJ<tIIiI!h:r theoretiw lUlLd ,mct-nodo-Io,gi&a] footiU1g (cf. 8.4); and thlS is the, wa}' ,that com~ pliifl£ntilli ,3:nalYl'iis is cllmmonly prtsCflt!ed by ~UI"opcanstructura:!ists

- (d- G eakder, m 97 I).. Tltat it should be 8.0 iIl'tcrpreted i s lia~lIral Illlmlgh.

It ,is important tll rell.lj,~.c. howc\l)(: r. ~bat irae i~h.er p resupposes fieMlhcorJ nor is i'L presuppnscd ""i' it, On the on' hand, it is possiblc to hold the \',ic\'h' that certal n subs~t:5of the 'mtalil}' of ]cx(;m~ in 2. rangullge: constitute .:1. fielda.!ld con,tucta. . 'arid}! of sense .. relatlens with on'li; IInO'thtr 'ilmIdmt d:m same time to rejec~ com:IlDltl£ntia] i:m.alys:i~ a$ !I method of idclIti f}iflg thelii:.Id andi stat Lrlg the' ~cnsc~.~latiolls hiding amollg the members of the field!. Of! the 'I>th..cf hand. one might equally well ad(lpt cQ)'l'npommtial analysis as II rnea1'lSofstlllting Ithe saase-relatiorts that hijlcl1 among scls (If ic emes, but rdllsc :bo reeogniae that the notion of ~ lexic:lol field has 311'1 rol~ to 'play.

- Only one :!Speer of ,,~t. some scholar'S see as tlile, tn~erdcpcnd~,ce of 'lieM-t'filoor)' an~ componential !l!rudys;~ need ,oonOcI"I'l us he:re; a.nd this Jj s the distineti onth~t h_as been ,dlrown en the basis: {If tl:!; sanegt'rli interdependen,ce between twa kinds of semantic eorapenems: berween Sl:me,s,· and dassemesj, (d. P(lui!;'r .• ~974; Coseriu, 1907}.1I..oeootding to Cos~ri\!, semcs are the: mil'lunal d,istillOtiV~ features of me:JIning {hILt !Ii!'!! ope-nit ive ''I.'i:m tlltl'l a ~.hlgic le);ical 'ntM, and they serve ~O' s.tructure. the ficlcl i'll terms ,of v3rious kin.ds of cp]}ilsitiol'l (cf. Coserill &. Geckder, U,lJ4: r,!-g). E:l.t..Jl}ples o'f ~m s n'O'lJld be rhe sense-cornpencats that Ponici:' moo,gnizesas b~i~g distinCli\',ein the ~cNical fieJ,d CIlnsist~ng of • chaise", • iautcuil', etc. Class.clIDcs, in ecatrasr with sernes, are very ,gJ.:l:lcral ~~'nlS<tl>-oCO:m'p<lllentl raat are COmmH'i1I to lex'em,es be'longi Itg tr}' severa] different IC)1ical fid&;,; lllild thcy tend to be, riol only lexicalized, but:tlso gm"iimm(!llica:liorecl (cr. Cos-e:rDl.II &; Geckel~I', HJ74-: 15:;:)1. E!ffimrl~~ of elassemes wuuld be .ltK];\t. ... .TF./[NA;)·UM!!:U, J\lI\LIl/FEIIIJlI.r; 3n.i:! possibly CAllS!! ,all d, HII. a H jel:mslev's analysis ohhe mea oi !'I,g of· man'. • woman '" I ,~~~l1iQ[i I" 'm:JII:'c j, etc. \.\"QLDld pre~Ulmi!bly i nvolvc boot II I!em~!i and elasscmes.

- The di'Shi'tOlion between scmes ii1l.d classemes hilS been mentioned here Ibeea:l!lsein c:eftlJinr respects it correspofid::;., at k!l.5t I'CIl.lgh Iy. ",-it hi i1,~ cqlil :llly colItl'O'\'ers,ial, bUl fIlon: fllHliHll.I', d Lst.lnct Lon bNWCCIl dis~

y

./

ti ngll ish~rs"4t and m.atkcT,s ~ ,whi£h. as .it was origill'l'.llly d [lI,WIl by K.ill" and Fc.10!.lr (I9!l'3), was held to j",e.f'ieICt tfle dislinctiQ.I1 between "h: t ~\~SI I!YJ!t~n~al ie ['or' ~he language in the mC~!IIl [Ill: oJ i1. tex,e'IDC and .... I,,,. """3):1 not.

- The lJ:ut of the ml:lIl!lil1lg of \l lexem that was s.ystematic WIIS to be rCo1t'i'"C~ S'l::TltOO hy It sd of markers an.] the residue by 3. d istingl~ishcr. The neees&;),'1')' lind sufficien't cond i iions for decidi ng wIn: ther a pm:-tiCUIi!f earn. ~ ponent is Ii marker 011:' a liI.is'tinguisher we're m'lot precisely spe.cified in

- Katz and fGdof 0( 196,3.); :Ind the- WIID~C basis of du: di~tlf!QiirlD'! between the: [\\;0 kinds of QI!lIDponefltS was challenged lily Bolinge!.' (]9fl,~), \\fl;in~ reida. (u)66 J. mer-wisch ( r 9~9) and f1thcn .. What eencerns us here is lhcsimilui!.t liiety,,'een the ~1I'nefclaseme distinction drawn br eertain Euccpean nructur-a1ists and th~ ,di~fi!lgllisherirm'LikC'r dlisti.nCliQfl 'that has been postu.lated by some tfllnJDrmation:itI gramm3rians, and ~ejectei;] by etmc!I"s •

Alii rst sight, these lliiL'(i d ist i [J)ctioflS, w~m Id ap p~ar to ]l aVe nothing if! 'common: the IIl'LtC is ov'e:rtiy based Olil the prior !le]uiI1itation or lcxil:~1 fields: l!J.t ,~ther is to~ll~ independene 'of 'lidd.theory and is llcl~ to rest prirnarily on the,notim:! of systclil::tticity within the !Mgu:a_ge. Bill there

is :Ii s:~nlti lru-ity betwccn the E:!.l[opean structll.ra1l8.t$' c-{JIm::eptlclIR of ciassernes a~d K.atz's, eonception of markers, \oVhen 1(a11:1: S;!I[Ys Lhu lltis markers are syst'emat[c for the la:liIg~n\gc (that i'!;HI.,l!.~, for example, is sfstcmatlc for Engllish}. wha~ he hag in min~ is the rotc that lll~ mtlTk,e,rs play, acoording m hit; thlrory, in du: s:ta:temcnt of se]ect(Ofl!"Cstr.uctllJn:s.* (c[ ~O.5). F,@:!" example, the tUl(;!ma~}' of a !'<(!lltcnCle like "That man i pregnill'1t' [on the assurnptios that ir is 5cmnntic~Uy anomalous) could he nceounted for by ensQring that" pregnam' ,cannot combine ,(\lith ~ny OQlJ;1'i whose meaning' contai I~S the oomponent ~~!\u,; and t rus implies that. :i.JAU is systematic for E'!Iglish- SimiJ~dYI when P'Jtlh·t" omJ - Cos,eri'u d'~vidl!. semaraie c[Jmpo~enl5 into semes and e1as.:.:::.L~, ~ncy emphasize that it is the classemes that deterrnine die semantically b:'l.~cJ .spltagm:!.'li,c interdependences Iilclwcrm nouns and ,a;ljjcc "'1:5 or nouns, and verbs; thllt it is thl!' classenie MAlI'; liar' C~·;Jmple, that dctJenlli,i'le~ tile s<~[ecti(l11 jjf halillil. r 3!111ffiOgloi91\<;i' (rather than I r1iwlliLr.>i 'J, Rumani ;ut • a Sif ,ifl!i1J ra.' (r.a{h~r than ',~ 5e m ari.ta 'J, R ussian ~ ijmit~j:J, , (rat" r than • -vyxlJditj zam~!~ ') in sentei'lCles c{ln;csponJing: W Elliglish sentences 'c:onta[nimlg' th!;: ~~li1b • marry', Th~re is perhaps less similarity

b ·t.\'een • eemes" and 'di&tingllisb.I."D"S', since t he former an: held to dep-end lipan minimal fUIlc:tiaDnl oppositioru. (whetht:'r priY;;Lti'l'-e Of equipollent). ",hcrC'as the b,tl~T lire mcrdy the residu:e '01' lexical tnea,tl,a lfig th..1:t ~s not ;;u:;cou illJ!:d fOf'iill 'terms of Inar'kcliS .• .\t the S<1IfnJ: ti.IIIc.;, i.t .is

evident that slIlcn. ecmponents as Ponier's ron SITIn-'H:l VIION (which he classifies as a sem.el would be treated as disllflglJ[shers, by Klltr..

NoW', systcmaticnty w,itlilir.i ~he language, ,as Katz: construes this .nOtiOO1, tends, to. tom.:I'atc with s(:\'eTaI. otber cliaraeteristi,cs. The semantic disti[lotioTl;'S lind equhfaknc.es b~twccllilll~em~-.<;that an: -aDC()Julltlld fl}r in terms of d3S$f'!'iles or m.arllicrs, in the examples that are given by ~he ,atnhor;s ~'ho operi;lte "lid. .'\\'0 'ki nds of semantic r.>Omponents, - re 6,uc:l1. that: (i) they lICC readily idmtifiaMe IIC[()'SS l.lIngll;1igeS!!Thci are less Ilb~iollsJ1 langu.a_gc-depelldcilt 0.1" Cu,lhuc-de-pend:ent than are 'Ihie djs'cilli::t'iol'l$ !Uid! equlvalcnces a~Ql!lnted for in rerrus 0'( senn~~ Or d'i:s~ tillgui.shers; (Ii) they are.sJl1tactic:diy ,l"elevl'iILhm.d may he grilffintatiC<lli zed as, wdl <'IS fe!(tCiLln zed.; (i iI) they ar,~ not TeS~rkt.cd till <1 few lexemes, !but are wi.Jdy distllribut~J tnwUg]lollt LlM~ vucahularj, All these criteria are indcpmnlcnt of one anether. It fu1l0ws, t1.l1lt, if any ene of them ''I'3S, made defil"lliliv~ (01'1 tl1i1l assnrnpticn that. it eould be specified ,pro[1'Ciseiy ennugh Eo:r the. purpose); it mi.ght be ilil. conflict with the: others, For CN:ample, 1i1."oLIl: iSlWt syntactically relmra.llt in Turki~ (th<:rc ltD;: no disti nctiolll5 of gender. the pt:::i"50nal pmnQ L11lS are not disti oglll ished wi tit respect to '~he ~CN: of the reL'erent, ote.).: 'b~t lit seems to be as wi,Jely distributed throughout the vocll!:mlary as it docs in El'Ig,li~h,. French, Russian, ,etc.; ami ir is, of OOlll"Se, 00. especially pbl"sibJe case of what might he heh:l to be !II ~allg1JlIIg~-ifU;lepc[lderu, and eultu re.;ind,e'petftdent a,~omi·c concept.

This discuss~OIl of the scme/clas~eM.c distinction, 01:1 tll!1 one hand, and (If the distinguisherimarkei"'ilistillct.iol1. on the oth~t, 'hal!; not ,only st:rve'd 'til, cJl1p:Ei2lr£iie; .~w· act diat there ar~ dlfficul'tles.i'l"lVoh·ci.I in making these distinctions precise. It has!u.d the man: POSLtlV{t, purpose of r 'lati:ng the notion (If cemponentlal a.mdysis lo the dIL!1CWlSioll, of -uni,,·ers.a1ism ;li(ldreJ~tivisrn, iJ;[ the prcvieus oh:1Jrt~r (8'31. So far" in our exposition of componential analysis, we have nat explicidy ealled into quesrr(lT\ die ilIs~'It!]nption tbilt scnse~CO'l1pon~lli[s must be un\~i'~at atomic cancepts, Bu.t tl:li~ assumption can be challenged.

e- As far ~ the ronc'phml staHIS of sense ... oompOflJ';IUS is cencerned, it

• must be emphasized that tk~~nll neDC!s!;ary ~'I'l!"J:e1\ion bC1twCCTI com oeential :mlltysis anrl oollceprual ism 'II in the sense that W~ ha"e gi:ven to this term (d. +.J). It wfl<Lild be- quite possible trn flidorize the sensl:-relatioiiS that hold hetwcclli lexemes and t!} treut these rar.'Wr5 as

}dteort!'licnl ecnssruets, wl'](aS(l pQ;!;.~u,[~doll _ imriafies the J~-l!cripttfln of I t he language, but do ·s Hot IXIIT1_mlt the Ii ngLllst '~O the existence of any tClrrespo~dirJg mental entities. For example" as \~'C can extract from the

all"idlLmctirnl, prllJ'p Drltiol"l Z I : .a: +: : ] ; ; ~ (;I the factors 7, 5" :1 and 2, slao (7X3):(PO:: ~)::(5 X3):(5X2), so we calli extract from the sem!l;f!lttl 'PnllpDn~OI'l 'mtl:~':' Wl(Jma:n >;:" st1tllioll,": "mare" the factilrS til b, C, I (however these factors might be' libeUcd or sylllbQ.liz.ed if! the linguist': medel of the lall;~uag{';-s.ygtem). The fa ceers would derh'e their I iogli.isti. signUi,cam::e from llle fllct tlmt f;]JchofdteITI enables the Jinguisttoaccol.U'I1 for the semantic acceptabillt e 0.1:' UIliB.DCl:ptllibHity of sets of'Sen'tf'IKts; du presence of t'l in" m~n "\"I'OU ld !lie-count f~r ehe :acceptability of 'Thai man cuts !lis, own hair' and the unacecptah:ility of "That mAn cuts bitl:1 '~'wn I\t~~r' 011' • Th;!~ 'IiiIlan IS l:!'regnant'; 'tIw prcscnre of both b .~,t\d d i~ If mare" (and of d, in Icombin;niofl with. one ,01" nwrespceitif:d oomponents in .j fool") '\\"[add,u;count: fur the acceptability of 'TfuI~ mare has i ust gi ven birt h to a beaut ifi<JJ J:ittlc foal' alllrl the ~~:rcscncc 'Gf ,eithe1 rlHlre (or both) in "<man", '''woman'' and "sta:ll.ion" would aCCGliInt fo]? the L1il:Jiccep ~ib ilit}r of 'Thn mall/w~mai'l{st'llllil:Jn has j ust gi,r;en birth toa beautifui li'tde foal ", Whe-thc!:' th~ c sentences are .ru fact $(;'tmmticilly anomalous 0'[ not is, of cou~. a se-flamte question (el. 10.5). The point is t hat the ex~raetiol1 of scnsewca:mpol'lents ea II. be ,C<!!.ri l)cl' out (In the basis of such proportions as 'mat\':: "'\,\,oman ;; : < s.taLli [III '. : ' mare' i and theY'Illid'ity of the fa0wr-s stan.d~, or FalLs by ~ h,eiiiC e . .j"·plaflatolY po,"~er in rdlltion to the 1iJSC of lallg!Jlligc.

~ A·lost propo'l1(?ms ... of ~~nent'!;!! ~alla.l,ysis. it ~ true, VI'Oul d. net b,c cC'n'tent to. say that sense-cnmponent» are to 'b~ defined sakll}' in turns, (If the acCt'P~abiliry ,(lfsei'lt~lJcesaI'lJ d~e l"eb.tiQl'Is oi C<l.llivah:m:eandi implieatiou th.a.t hold between sentences wi~biJ'i 3, single Is.!\lgu:yge ~nu't this, il may be obs.erYei:l, \\"<1.5 UjeImslc'/$ "iew; and, to this ,ext.., our prese nHlll<m of iii is ~llJaly3is of 'milD.':' werman'. "5t11Ui0n';' IYIMli: " ete., was rath'E'1" miJeading.) They would wish to s.ay that tbe labela chosen to i(kntify t he com ponents ha~·emore oontent titan our aJgcbJ'lIlC ta.ctOi'S 11, b, e aUld rl; and rl1cy would "d3h to :rela~eat least some of the f:<l.ct{l.1"S to lhc external \\.·o.l'.ld i II terms of the rela tion of dt:iil.otil'tl0n, saying that !HALF.: (o~ I r a) lfeftQtCr> the class (If aLI. entities lhat have IlU~snd-auch a IlropITty. that UlTllT.!IN {our c) denotes the t:hss Df all entities that h~wc' iii diff,~rtrrt (but cornfhltiblc) property, a:11J: so Oil. There can be no qjuarrel, with 'this. It is obvious, however, that, lmlcss one provides some c:>Itcn5iomil, d,efinition of M"LE; IruMAN. etc., Of soml! ilnll~nSlDI:I:d definition that does not m~ktl .m.,cta:lil'lgLri~tic use of tIn: En\gJish worn 'male', 'hrunan', etc, (or tbe Frecncfu. words 'millc', 'humai!l'" ete.:

Of the Rus.sim:'1 words • rn:ull:7.koj·, '~et(!frl!~kij'; or the ,V'(Icd's ,of same other flahit~ ]a:llguage), no explamuwll has been gin:n of the meaning

330

of " mile' •• hum~!'I' or of the sense- components in th.e meaning of • m:ln • dmt adds anything ,0 an analy~is dlat mikes ]S~~.f pll,n;ly algebraic symbots (cr. U'wiSI ]97~). Furthermore, in \'iew of the looseness \ hh which the term • concept' tends to be employct:l by semanticists arullhe cri'ticisms that have been d ~rec;tell agai n1st ,oonc~pHl!ll ism. by both philosoplrLer& and psydl;oI@gi!;:t~, ~t must be ,etl'lphll.s hZiml:tha.tthe extensiQilIll or intcnSLOJil.~~ deoo:ution of MAU:;;, UUlllA.r'l, etc, does !'ltd n,e-cesslIrily imvoh'c the ]lil$tl!Jlation of any co'rretil:tet'! mental entities.

Let 'liS n()W tum tto the Guesl]1l11 of atllrrllcity. ,E\~f'r since Lc'ibIl]z put ~. D l"()17wa.nis pmpi).5;ilil for tile CDJ1:struCtlilfi o,f II. til:riveorsal &yrnoolic. ,Iin,-

gil, e the princ~pfe n atomicity has been p~~ljien:t ill philosophical dis..s..ussiotls of the ""ay in whkh the me.luing.> of words might be

allal ed into smaller, and presumably inCire basic, components, 1].11'

• basic', ~n this context, j,t is ill!p]icd that theoompooeri,l:t;, in question constitute the po,ints of attachment bct\'\-een ]angu;:I. e and th,!; external world: i, e. that th.ey ~an be defined by rda:ti rig them <II i red]y 1i!!IJ entities

o uhi,melangtlil.;gc. ] t was l..eibniz's ll'ltcnt·io[l that th.c sym.bo ~s in his universal b:nguag€'~h.Ql,lld e.:lI;P:~~ 5.imp[~ (.i. c. atomic) ideas; and this has generally 'been held to i.mply, illl the empiricisr tradition ,a.~ ,lc;l.S'~, that

they should h£ acquired] by virtue of i.mJ!i.led.iatc~!;1J5{I[y experieaee, Whate\re:r rl'repl'iilosophieill me:rirs of the empiricist. principlc of atomicity, it is ob"i.ous itha.t J;rll)st IOf d~e sCllsB-cmnponellts that have

been pOiSitula:tdl by Zinguists (e.g,. i:l!IALE, 1I.LlVE, FOR SITTI;NC UPON) U"'1i:

not atomic in this sense,

It .m.ay be that rhil:lr~ an: ex: rtain le",e,mt:s whesc denetatlor, can be a.coom.t~a for 1[1 hYl'liU (l pellCt:ptuaraistincti:ol1s tl-tat are phys,io~(Jgic31ly atomic in thar dn.ey carl be shm"L'JII, to depend upen an [l1,I-or.n,odling' r'l:$pIlnK hJ' il5cn5~,y stimulus, for ~xampl.c'. the ~cc[)~niti(lfl of a reddish or greenish nue uligiu be phrsiologically atnmic: itappears to be lhc ease t.hat du:tc arc specifie ,~!ls lfl. tilt retina th:il:t eishcr respond ,err fail [1:1 respond to a stimulus aooording to whether it is (It" is, nnt Q,r the bue til' which the 'cell's an: tuned: and we have already seen baw thi.s Fact rniglu 'be rele..-ant to the hypothesis, af Ih:·1in nnd Kay (d. 8.,3) .. As fat as- most sense-enmpeaems an: concerned, however, it is, hartili till' see hOlw om: might decide, even. in principl,c, whether they Be ~tQm.ic 11:1" nonatomic, The notion of re,n:eptlll:d atoraicir !!.eems irrelevant to' them. (c.g., SMllS!::, SJ.Bt.lN[;)o; =m d there' i~ no other notion of :a:~O'lil~ citY - that is not opm to the erit'icism that its \-a:tid.~t)' is, llJnv~l"ifiab~i:. Hut, alll[lIni~m~ like ronccp'tual ism, is dearly flot tan c.sScnt iill ~[Igmdil:I1t of componential ,ana.lysis. W,e need s.ay no HllIre about it.

3J,:t

Th~exi,(lIl between com o'llel1l'tjal analysis and uI.!i.1_'er8 ,Ii flLis. rather more cmnf!!elti: i since there an: scv,c:rnI! versions !IE lll1iw· .,'131" s m i and much. of the 3.'ttrnction of component lal MillYSls unrlou bt dly derives from jts assccianon, whether lhi ' is cCii:U'ingcmt. 01'" essential, wlith (m~ or mher of thCtlc versiens, ThC' most exrreme form or the thes,us Qf universalism WQulcl conibine at ~:east the f'QUowing three dist:i:n,ll;u!sh= able s;u b-taeses: (~) t~l!It 'there is II !1i.1ted !!e:~ of -st'R1aflt.ic cO'.rl1poricnt~" which are UD i versal in ,that they arc ~~xir:arizcd in all 1~J]gllages ,; (i i) that the formal pri.llcip]cs by whi.ela these EH~I1s!!:-romp[ll'leuts, are cQmbincd 'to yidd as their pmducts thoe me:a:lUI1iIfiS! o'f lesemes ~rc universal (and, p res1Jmably i nnau:) ; !lfld (i i i) that tin e sense of alJllexemes ,in "llla.nguages is decomposable, wit] ou.t residue, into variable C[lmbiilllltions of (homegC[lCOlLIs) eense-components. Th_ distinction. between (i.), and (ii) has to do with the disti fiction t hat Chomsky .( ] 9!}5) dra WB betV.'l;l!fI substantive and foml.ai universals, Let us 'th.eil'cfQi\t:: ~f4lJl to (i,) and (iii) as the llle&e8 of S'U bsUuiltlrveam:l forma] unh'ersalr~t.y, Fespecliveiy ~ ,;mdLo [iii] as, 'tbe withom-msidue tile&is,

.'\s we hll\~f:' seen, Katz docs not subscribe to the em'Jl pO$he tho;is ,uK extt'Gmc univeI'sa1ism: he doe>': not no[d to ~he ~~'lth(lut-nt:Sldue pad i and his "ilew of su!;;sf3nti;f! uoiversels :is not ,hat Ian languages 'Illust !exical~ iZ(l (ot' grnmmaticafure) them, but rather d:t!ll al] lite sense-components (other limn dlstingu~hers) whkh arc Lcxi.c.a1~Z!cd in any language are taken from _ fil{cliiflJ"'t:lltiory (the klilowlcd,gc of which is iirlnll~Je). Thus is, the view that Chomsky and all his follo\\'ers t<lk.e of ilil substaetive universals, semant ic, !i.ynt.a.ct DC ~I'\d! phollln:J lilt:;i cd. It is 'llerta:iI'lI~1 Bierwischt"s view> who critieiees Katz's [IotiOI'l of distinguishers 'Of.i the :grnumll thai: the}' calli be >Ilnalysed ~nto 1l1.OfC ~ru;ic elements and in doing 5(1 e~plid tly decl!.fot:S .h.i~ adherence to what ::i:~ t11'&~ S'ight iooks ~ike esreme universalisra ; and this wo"dd nl!'lJ'J seem t1D be the • 'LeV" ot t h~ so-called generative. se.manti.cists*' (,cr. 10,5). The Chonlskyan form (If the. thesis of S1!lbSM'ItIvc I.lniversality is tn,e-rdOfC milch weaker tbw wha.t W~ have' presented abo,,!! as (~}.

_ None of the :European structurnlisrs have been or are esrreme universali sts, H - elmslev -.m:aintained his own, .. da.ti 'I/~ly ","(;;lk, version of the' thesiso,f forrna~ji.h'C~lir .' but Cl:xplicitly rejected the thesis of 81Ub~ stanti \Ie univcrsal[ty in any fermwharecever, MOo.e recent Wftters in the -

\ same p-Qst-SauSl'lll!!'\e,U'l lr:!lUlhio]l (hJiOtllMy P,[mr;.:, Cos.e:riu, rGr,f"imasj M've !lIst> ma~1: It d~~ dnLthe_y reject~t: h.'ill;1It~h~~l!;ol'l fOl':lll ~f the \ thesis of sUMEtnti\le untversdit- ; iil1il:! their adherence to the widlOQJJt,~

n:si.dur.: thesis is weakened partly by their methodological pri ndplL!

th:tl the ;<n:al},sis should not be C1irrioo. beyond the P(}LfI~ aJt which r;vc~ lexeme Is dill~i I'I~~ tp,I;.'>d frorn 'C\'cty mnoe:r' I'Ul n -synonymous Iexeme :I!'Id partly by thdr rc,i;og;t1itio!l of two kiluls of sense-coraponeats - c:I:a;s:sernes and: semes, Although some of the semes might be [Illi\'cwlJal, .and iCven .a.tomic [e.g., the semcs which fli~t'i.nguish "red" from" gfc:CIiI';)" m.o.st: of them are pa1iimtly non~11I'Iiilter9aJ; arul scme ,of them are Qomp]~a!'!d! 're~iduil!J in the way ·that Kau's dislifil,g1Jisnel'"S are. Finiliy ~t might be merU]ened that Russian Ichol:a:rs [ike Mel~~'Ll k. Zolkovskij. A l'ielljafl, whose work wnl be refereed te later (cf. n.3) ~Qu!d Il~t !!ppear ·t[l 9U'&~ scribe 10 t.b,e thesis of e;l:treme uni,,-crsa.lism, des slre their .intel'ie'St. iin the "Com;tnuiioll of II. UllWCrM~ semantie m.elaJafiguag~.L WM;~' lf~cabulary (~ik)e that OiF ,L~ ib;n iz"s s}'mbo~ ic Jllirlgu ~ge: cf. Apreaj [11, m ~n4: ]11) is to be CQil'Ipo$J;!cl (If atomi,e M:1I5e·romp0l'lenlS; nor would Leech [cf, J974: 231-(2), or Lehrer Yi"974). or '\fi,e'~Eiicl!:a ([972.), who has, devoted ~ whole boek 'I) thl; ·establimment ·of an, ~n\'(;ntory af univers I senseoomponents, and to ane~~mplmt:.;'}till]\ oJ how such seusc-components are k'."!h:'Jlil,j~d .. ~il shnrt, .il is .!l.ot clea~r that there ,is alii)' representatIve of e:Ktl'eme IJni,rersmis'mlIQ be found among, til1l .. guis't$ whmJ cuncndy IKivo,eaee or pnu:ti5lc oomponentia.1 anal)'Sis •.

On present evidence, the most :p]nusible versien ,of the universalist thesis wou M ili:letri to !it: the olle that \~'a$ (ludialc<d abo\'e in 0111 r djseussirm

of the hypothesi,s put fOf",\'ar(! b.) ~efli n and Ka!1:f. R 1 )" H some (but by no means ail) of the stnl<lDtic disti 1'I.'~1i~~t;_!Irnll'n_ in la:tlg~~aMs .arc determined I:lj' a~eticall~' tr!.I!SI1l~red dis()ositi:Qu 10 ~C:Sf!'(mli _!:,Il biologi,c,"l'lly and clJllturaJly sa)·i.ent stimuli. languages will tend t€llexiC!!lize. I(and perhaps also 1.0 gl'lIfllmatica]itc) theE.e sellil~ntic distinctions: e.g.,-'the difference bt.'1\VCCIl"", .. I1::ut is ~'ert icill~· exten.d ed and what ~~, not, between what \S ~Il]i d lind wl1.'i.t' is ml!t, ~eh~'ccfi what D,S. animate and what is not. Co~e-ql!ll:l'I.lIJI the analysis of many, if no 31l, languag.e!:lySlemS wiUr·t: ... ·cil dIa:t tbetc are scn~e-rellltt!lnS holding in mi!.lly .Ulc3S of lJ~voca:bulary which can be accounted for by pos·tlllating such oom:ptJJIIc;nts as \"ERTlCAL, sou D, A~F;"1A,!~; anll theS':_SCfUlIr.·compon.ent'S (\,\"MSe dTstn1iutiol'l thr·ougllou t the ]anglJa,g:~~ of dlif wOiild will pr;e,· :~lImab'J)' vary in proportionwith the. relarive uli ence of the dislincti (1'I1S ·they encode) \VII] be such 'that they 'wO\dd be treated as milrkc~ (rather than dlst'inguishllmfby Katz. [a so far as dIe m.!lrike:rtdisting1J.isn:eJ' distinction eoreesponds with the dassemelscrIl,c (Iustinction (and l'I.eiithe. .1'1.[ thc~e ~WO di:&t~nctJLQins •. as n'C .h.lI,"'·~ 5'~<S'.IlI. :hu ~'(lt be-en made pn:ci!>c). tl:u:y WOlll ld i!so be d;!l,s9-eDlt:s (rnt]Hrrthiiln semes); f'Of tlley would tend to be syntactLcally relevant aneJi ormrin: 'in 5C~lt, -(11 lesic I. fields, Not

an the mllrkers or elassemes rcoo.gni:zed in the 3fIalysis of part icuJaI ]angL!Lages, however. wl,Hlld b-c ;ubiiiiilti~univerfia!R. since there ~~ nMhlng to prevent a 1angllagc frmm. ]e1l Lc,a]iz~ng a nen-universa] dj~· tifictiolill and mai!:ingits~"tlt3!cticallyrele1. .. lult.II: f01h:"",'s {.i) that the distinction between di!ls!!itl'l'les and 5t:mes, On the 'l!llIlfl hand, ilind bct","ce'll :mall'kle:rs and distilliguisherst on the other, CalH'l.ot be e."Ipcctcd 'to eorrespondl,. o~hel" lhai1J approximately, with !t:hc distinet ion: bet'~\~eenU_l'I.hl:r5al and nOfil-lIniv~rSd senae-compenents af! d (Ii) tb~ the jnv€'l!Itmy or semc-oompllnrnts ~n 'terms of \'Iimch the \'o~ulary of any parLi,cu~at 1:ingnage is lIeGCfi'il.oo might contain bo'th universal and aon-uaiversal sense-components,

-W'I:: ha:ve nuw mm'p'h:!u;d our e~~sitioL'l (If t!le g!;,m::'Iml p~ri<nlcivle~, of IOOmpol1lmtial ll.fl:_i!ly~is. The recent luerature of lil'lgttis~:~ s("rr.=llltics is f!.lll ofpmgll'ammatic statements te 'tile d[eet tha:t the mean!" Q iJ[ lexemes in aU ],arn.m.e'S can, and must, be acc!)1.lIl'ted -(orin terms of the ce'Jtnbi nation of 2!!~~dlJ." more 'ba.sJ.c;andi _ os:sibJ:yu Dliversal" :sen se~mpOinerlls. So· faJr,. h01p\"!::'lfu;thc analyses lhat have be~!1l plibl~~hcd

-are inoomplch: and, for the mlllst eart, lJI!\cotl\"in~qDg; and they have been <:qfillll1cd to relativdy few are 5 of he vO>C3bulary in !l'clati~el Fe Ian .. ~ages._For this reason alone one should be cautious about aectptiillg as vall d the claims that are mad e on behalf of camponeatial analys,is by illS more ~:ll:llJ5:]asti,c advcestes. But them are mher reasons tiOQ.

It is now \\cidcl}' Kcogniz.ed that in cC'rtainanms of the vocalna'ia '

1 ill which ool'l~£~,nentiJJl anal~fSis has heen pri!cJt.ised, arid 'most, notab~>f in the field of kinship votaJ~ la£)~ (d. ROInDey &; U' Andrade, Ill!'i4-). it ]::1 possible Ui provide s,everal cq1l1a.Hy pl ausible aO.illyses, ro," the ~am,e set of lcxemes. Given that thms. is ~i)i, how do 1;\-e decide dtat 'ORC, >!naly.si s is COfI1CC~ a~d th!lit'tln: (nhcr.s !lire Iwtl So far ·this questioo remains \IIn:lr'l~ .... 'Cn:d • Indeed, il is, lIlot INell dear that it is answerable. For it has yet 10 be demonstrated that sense-cempenents of the kind tita:t linguists, have f,r,:n,cieo tra iflVOklC in theilr ::Imdysis of the mearung of lexemes play :lJlY part 1'vh!itOOe'!i'f'f in the productiofl .. nd interpretation of langl.l'~:ge· IliUet'ilIlCeS, and. if the aUew::d:ly mofC basic :~ense-oompone,nt's ca.nnot be shown ·tlll 'hit',!: any psycb,o[OIgical. 'I<"3lidi,t}'; much or th·t: 'initial attractii)R of ,t:.()mpIlnc-ntilil anal},&is disapperu's ..

- The ]lSj·choh;tgicalreality of scnse-componeets lus orten been calkd intn doubt, So too hilS. thl"ir 1:.lni,,,er~alit)'. "'I'1'tat is .l]fl'l usu 1&1' mentioned, however, d I'll gem:ral discl,Ission~ of tll~ lJIlcnts of component:La'l .iOl.!lalysis Is the fact that. even [IL those MilliS in whieb,~t ~"oks reb.tively Clln· lfincil1!g. it lea,>'Cs ulll.~Kflained ;a:t least as, much as it sll.u::eoos in

ais

explai ni-Jug.F'Of examplE', if the m£afl1i'lg oJ the lexemcs r raan", e woman ", j adu]t '" 'giirl'. • boy' ,and] • chi Id' are analysed in terms. of the sensecomponcnts H U~ UA.,"', A.L1 U'L'J' il]lJ ymtLU.:E:, we can read i Iy cxpl~itl ~ he folild 'that phrases like' ad lilt child' !illi' 'null!' girl' Me wmantically anemaleus, :10 do:ing: soo; w~ 'm~st assume ,(anll rl i:~ more often ;IsS'nuned than ,Iua:~cd el!plidtly illl tr,eatmeots of componcntial analysis.) that •• male." [i.e, the 5Cru.-e of the Engltsh kx.'e~nc 'male '} colHains and is cxbaestcd by the ~eMe=oo'mponellit. -FEM,", E, that "adult ,. comains and is 'l;:}th:1ltl5tcd. by IIDUJ..l'. and so' ~. On this .:i.SBUlm~l~iofi, ho,w,c!i'er, • male dlitih:l' sheuld be synonymlln.ls with • boy'. But, it is not. An ef.,ghtcen-year old boy is eerta lnly not a (h~d. F urthermere; if fO boy" differs, from, "gi;rL" ~(ll'lely in tbat ,it contains -liI::MAI..F., rather than fEMALE, how do we accoui'lt fur the f:tct that the lesemes "boy" and' gi'ri" sim(l~Y ;Ire 1'I00l used ,~n sueh II, way dillt the allegedly C.OmmQ[I-All1ltl.f can be gi\'en a \lliLi~al)' interpreta,tl,tlIn? By any of th e most {)b\'iIJlll~ cfutel'ia (.sr.::1Ii; ~ al maturity, etc.], girls reach what \\'Q'!.ddl normally be described as adllidmild earlier, rather thanl,u1cr than, boys; and yet ll1cy are described as girls far longer LI.a.n bf:l}'S ,a,n.: described as 'OOi'S. 'f'he propositioll "X i!li now a. rna n " maywell imply" X is 11.0 ~o:llger a boy" ; but ... X i~ [lOW a woman" does l'I,ot imply' "){ is :110 lollgc;r a gid", It might he argued, 0f course, th:tt this differ!1:nc:e in th~ use of the lexemes • boy'~ 'gid' and "child '" ,IU;COru]l'Ig to ",hiClJ. the alleged]y COmmon oomponent -AIWLT is giv~n a different interlPreta~io]l. in the theee cases, involves soulctliing over and abo\',e their lireral meanil'lg. BlIt what iSlruS additional ~ngr,t:dient? It is easy lenough to !'WI",; 3DY hypothesis b~' postulating uni,denlutied additioll'l!l dlem'mts or by ,i~nr!]king too :swiftl):· the distillictiQfI between Iiteral lind non-literal mcaning. Hinee cornponcntial analysis promotes the Ilieafch fQr gtncnilitllctiml. (i. e. far t.lu: idE'rn:i fkat iou 0 r the same sensemrnp<ments, over the J'argC'st number of lext'mcs) iris always liable to f~ll ·.~ictiliD to rathee F1lCile (lVc r- ,gt:tll.:rniiza.tio!'ls. Wh~ne~'er we ,I1PP'Cal to such allegedly ~mmtlH sense-ccmponents as JlI!t\IAN" ADULT and FEMAU:, we ~must 2l'ik IJoIUSe!.,-,e-s what their crl)s:s~ltxcmic ~'t,a.ttll~ is, how they are 11'11' be iden.tificcillimi whattlu.:i, explanatory powe:r is,

Itis importaillt alw t.o k~c:p curu;tmdy i11 m~!'Id the dLffe ... enec between a lexcmc, tht: meaning o,fa lexeme 31iC:lI some h}'potl'l·l.ical sensecmnponenl whidl is in COITcspomlence' with the InciJDing Qf a kx~me (e,g., between • human", "hillman" anJ IIUM.'1N}. There is flO reason, j,n pnnci'p[e,,,,,,,'hy the citati;)l:l-fornt~ (D{ lcxemes f~om E'ngJish ,or an~' nlh{;r langllage should be used to label. the eense-eompenems that firu: postu]a:tes in the SEIOMlt:iC ,au1ys,is of English; aIid if some (,u 1lL~ r system oaf

i:dl!'rllificati'o:!lwere aclo:p't!:d (e.g, II. f:lrumerical system according to th.e sense-eompenenr'« position in 'orne s.t:and;aWizedl mMtcl' I'ist), there would be less rikclihoodl of eur assum;i.l'I,g thas, becaus-e """C know the meani rig lof, let us lii:ty, I human '\ we ~SC' kilO,.,,' the til,caning' of H{)MAJ.""" Sin~ the meaning of "hillman 'is !ilIPPillsed to be H!l:p'laJRcd ,in 'lefms of the pmltu~ilu:dI theareeical entity })'UM_JiN (i.e. ," hum~I'I" i:s he'ld lO' 00[1- taLI1 and be exhanl~ted hy HllMilK)' '[he thenret·ic:tl'ell.tity i~sdf must be defincd etherwise than in terms of 'human'. Unless 'this is done, eom~oel1cndai 3o::ilysi:s ]s rcducdll, mU only in pracli,cc. 'Ib~.t al~~ in 'Principle, to the highly iJluestionabie pI'ocedUl:c of treadng as basiceenseoompo.fi~nt!l i iii d}€ !LMI)''Sis £Iof any :li1Inguage thiJt the ~~nguist iii de-scribin Lhli meanings of certain lexemes, like • human', I ,adult' or • f:emale • [mm his, iYLvn l;Ial:i".~ laJ'lgl~a;g'e or fmln seme Other l~n!l'lI.age thai is wmmonly employed as ill mellilanguflge in lheorctiea.ll and dl:seriptiv,e i,i'ng lIi~t ies,

S-ar putea să vă placă și