You are on page 1of 24

F

t * : i ; _ f

J:

*-

rF

2
LinguisticCategories, Shifters, and CulturalDescriptionr
M I C F I A E L SI L \ / E R S T E I N
Unirersityof Chicaglo

For Romanlakobson

INTRODUCTION

.F
of Ihis chapterwill iry to developconsequences the statementthat speech mcaningfulsocialbchavior.In itsclf, this statcrnentis onc of is those set nhrascsof pidgin sciencethat are used to cnsure minimal in trade relations the contact communitvof linguistsand socialanthro' porver.\Vhat i rvish to or pologists. givesus no anai,vtic descriptive It do here is demonstrate that rve do, in fact, alreadyhave a full, subtle of meaningstructures the elaborate "language"with rvhich to describe of speech is a languagcthat spcaks the "function" of signs, behavior. it their modes of signification,distinguishing frorn an'Iongthe types of '['hc of mearring tlris functional sign sign functionsshiftersor indexes. mode alwavsinvolvcssome aspcctof thc context in rvhich the sign occrlrs.In making thc nattrre of this involvcrnentclcarcr, I hopc to clcmoustratc lulll,sis of sllccchbchavior-irr tlrc that this "1>ragurlrtic"
11

the us frorn Peirceto Jakobson-allows to describe tradition extencling real linkage of languageto culture, and perhapsthe most important "the "melrniigl'-of speech. aspect of as At one level, languagehas long servedanthropologists a kind of cxcnrplar for drc. rraturc of things cultural. It scetnsto display these "cultural" propcrticsu,ith clarity in fhc tlngilllc rncclitrtuof articulatc phorreticspccch.Thns, and at anothcr lcvcl, could the analytic lcssons to arralogically othcr socialbehavior,giving of linguisticsbc trattsfcrrccl or, a kind of structuralizcdanthrop<llogy, l'norercmarkably,could thc actual linguistic (cspcciallylcxicographic) shucturcs of ltrnguagebe called culturc, I rvill bc clcvclopingtltc argttntcnt that this rcccivecl 'f!4 as12_cc!91 wrong. point of viov is cs.scrrtially l1lgyj_gJfylrclf lti!_

typesof connrunicativecvcntson the basis of the signalingmedr\g. the In the caseof language, signalingmedium is articulatespeech,{nd. eventscan be isolated on this basis. SpeechEvents a By such analysis, speechevalrt,endowed rvith an overt goal in a sociallysharcd systcrnof such purposivc functions, consistsof some or of sequencc spccchbchaviorsin rvhich sotuc s1>cakcr spcirkcrssignal of somehearcror hearcrs rneans a systcrnof phoneticsign vehiclcs by to into or callcdspccchlncssagcs trttcrirnccs. are J'hc uttcrauccs organiz.ccl of l>y a systcur tlrc participirnts virtrrcof tlrcir krrrts4cclgc a lingrristic for code,or grann'rar.'I'he specchcvcut takcs placc rvith t'hc participants 'Fhe roles of in givenpositions,or loci, and over a certain spanof time. and hearer may be taken by difierent individualsduring the speaker of courseof such an event. \,{any other characteristics such speech among them the other eventsmust also be takcn into consideration, of in ancl hearer, aspects the individuals the rolesof speaker sociological which are frequently salient in defining the event, the prior speech
cvpnic i r i f e -n1w \ n.
),

is uniqueamongthe phenomena iust the pa-rt_that functionally .of oJ lqlglqge !n !his,!radi'Hence- structuralcharacter_i9lr_c_sthe culture. as for cgllgrq, tionalviervcannotrealiyserve a,mode'l gthel aspects-of linguisiic(or lexicographical) Furtlter, nor can the metliodof analy'sii. grammaticai analysis must fron'rtire traditional stiucfirieithai emerge isomolphic,relationship rather than bear a problematic, of necessity
|n dra clrrrnhrrn nf nrr'llrrrn

4 -,s_qye$*as i1 iyaLyzg$_ 19{elr !5aditi91ally-_Eg1- !y l!c9i91',an _b$

t '- l- r p s c c f r r r n l o r l c i n e c i n c o n r r n r r n i e t t i o n . -

flrnf

invnrinhlv
-.-._----_-J

LINGUISTIC AND OTFIER COMMUNICATION necessarily implies bclnviorthrt it is meaningful of To .say social
a complexof signs tltlt is, that thc bchavi_o.r-.is tirat it is cotrttunrticativc, in (sign vchiclcs) that sigrtal,or stlncl for, sonrctlting somc rcspcct. participantsin a behavieralsignsare sigrriGcuttto solre Persons, Suelh event, and such bchavior is puq:osive,that is, goal communicative oriented in the senseof accomplishing(or in failing to accomplish) certain ends of communication, for example,indicating one's social and so efiecting a cure for a clisease, rank, reporting an occurrelrcc, peopie arc coustitutecias a forth. In general,tiren, rve can say tirat socicty with a certain anlture to the extent that they sharethe same mcans of social cott:urutticrttion.

accornpanies thc clistinctionbctq,ccnrolcs of hearcr spokcn languirgc, and audience, so forth. A description the specch of cvcnt must mini, and rnallytakeinto accountthesefunclamental definingvariables. Speech eventsso defincd, rroreover, cooccurrcnt are rvith eventsbasecl ou clistiuct sigrralingmcclia, ancl thc.sctogcthcr nrakc up largc-schlc cultural rontincs. Dcscril>tivcly, sinrplcst spccchevcnts s,oulcl be thc those which thcurselvcscorrstitrtccl thc cu[irc goa)=dircctccl social bc. havior.It is cloubtfulthat suchcventsexist. In our own cnlturc, rcading a scholarlypaper can come close to being a speechevent pure and simple, the purpose of which is expressiblein terms of informative of The possibilityof distinct discourse among socialcaiegories scholars. tirat can be invoivcd in tireseevcutsis not at issue. forrnsof syrnboiism on!,v_ ! gra_dqaligg_h_ere wit-hthe purposivenature of the speechevent in 'l'hc rnorc enrbccldcd specch cvcnts are those a systeur.of socialaction.
wrucl arg Part uI lrul:u ralEs-ssarc gulfural crrLErPrrSsS guurPrsx rlluars a5
-Lr^L -,--r -t -,,^L r--^ I^-i.---^,-rf----i ^-r---l---^^--r^-iL--1-

chaiactei'\ has as i,anguage a sysiemof eommunica'..on ihe saii-re analyticsubpartsI isticsas the restof culture.So in orderto distinguish distinguished among rve suchaslanguagc, havetraditionaliy of culturc,
t7

lclgding.speech, song, dance,dress,etc., u,herethe meaning of the behaviorin the speech eventsis usually integrnllylinked to the speech

r3

t . t

MIC}IAEL

SIT.VERSTEIN

Shifters,Linguistic Categoiw, and Cultural Description contributes cateSory "aspect"es a PuIereferential vs. grressive punctuai or as events continuous ongoing{w!er-ethey Io propositilnsdescribing In or conrplete. English,this necessarily vs.momentaneous so) "rroot is illustratedby The boy wasfumpingvs.The boy iumped,with seg' by be be(-ed) -inglv:0 (-ed)]v, * ed rcpresentcd uas' mentation this referentialmode, from in analysis Any forrn of grarnrnatical clcfincsthc signs, tlrc to Grcco-Roman transformational"gcncrativc, in and arrangernent this rules of combination and Categories, their g,d All fashion. of ouranalfllgrg-c-hP.lgugl,fglrye"1--dSssrip-tiv9"Igech&qfJ ho-. ho"i-f,*.iio'"a r"i t"f"t"ntial sisnswhichcontributeto rgfergdial j::'1::-:-::::E:
(We shall seebelorv that certain rg!-eleqltg!-cpeegb.,ete$s-. @eq11r a*ong tire referential catcgoriescauscdifficultics ivith this rvhoie apwe proach.) When we speak of linguistic categories, ,mean categories kind; hence one of the principal reasonssocial funclf ttrir referential tions of speechhave not been built into our analysesof language: thb
\'t;..n-(

Anatytically,the problemof mgd_i.a. s-ignaling presence these_ot-!-g of is verycomplex. in of Fing to giueihe meanings signals sucha situation EYents Referential Speech from of But the ultinrate justificationfor the segmentation speech to uscsthat seerns mcdia lies in <lrtcof the purposivc gthcr signalhrg the events, behaviorfrom all other communicative distitg!$sh speech or, in terms more culturallybound than function of.pure reference, nhilosonhical-the function of descriotionor "telling about." The -communication as can function of speech be lharacterized [i"r*i"f to of of clcscriptive states afiairs-subiect by propositions-predications asrepresentations taken and of in verification somecases obiects events, is so Reference characterized a communicative.event, of truth in others. are discourse rnadeup of signunits in of and the utterances referential the arrangcmcnts, mcaningof the wholc beinga ciescrip' grammatical fitnciioi:of speccli, Iiu" ot rcfcrringproposiiiur.it is iiris rciereriiinl sign, thc sigu its charactcristic uroclc, scurntttico'rcfcrcntial tlnl has ancl in linguisticanalysis the formedthe basisfor linguisticthcory and Westem tradition. erential LhryuisticCategories Ref olt Of AII liuguiSticarral,vsiS thc trralitioualSortProccc(h thc basis<rf or of to the contributionof elements utterances the refercntial denota' tive valueof the u;hole:And it is on this basisthat the traditionalseg' is nrentation, description,and definition of all lingUrsticcdteyories of segmentation utterideas abotrtthe significant made" Our standard in or all rest on sameness difierenceof utterances terms of ances Piuraivs. singuiar codedin speech. propositions, referringor describing can linguisticcategory' be as "number," for example, a purereferential dcscribing by analyzed the contributionof suchnurkersto propogitions cntity. In English,this is illus' vs. more-than-one not-nrore.than-o1e with -s trateclby The boysrun vs, The boy rur*, rvherenoun suffixed and subiect, of verb signalthe category plural-number and unsuffixed -ssignaitire eaiegory sing',:lai' of with noun and verbsuffixeci ' unsuffixe<i -sJr Thus we segment '9]v: '9]n 's]". Durativeor Pro' numberbubject,

qf cu-ct.rti of signmodes nlgstof u,h3!-ggl-9-ljl lbg-rirsjp.rity sB.gs-c}arc rro.trcfcrcntial.

t '

and Semantics Linguistic Analysis


'T'!,a cl,,r'1,,nf d.a urv vr I rrv JLuu)/ "mcnninr" nf Iirrgrristie sisns is usuallv Callgd -------J

in the of I them.For the purposes this chaptcr, tern $'ill be restricted mcaning, refercntial of is this lvay, thatsemantics the shrdy purc so
amhnrlierl in. .rnnncitionc . ' coded t hvE snecch- . T'his . . r . - r - . - _ - - . 1 oroocrtv of spccch,

semantics. it is clear fronr the rvay I havc charactcrizcd traditional howcvcr, that thc nctttrl obicct of sttrdy of , linguistic invcstigntion, of of has / sernantics bccu the refcrcntial rncariiri$S UttCrerreeS, ttre words f and categoriesand alrangements in terms of which we can analyze

can or abstractreference description, be calledits semanticity. of analysis the point of vieiv,all grammatical From an operational To on kaditional sort depends this semanticity. be able to analyze must be able to give evidenceabout the we linguistic categories, We of scmanticrelations parts of scrrtctrccs. must ultirnatelybe able is rvhetheror not a certainstretchof language to say,in otherwords, to rvithin the grammar, someother shetchof semantically equivalenf reiaiions,we can buiiti up sucir equivaience ianguage. <ietermining By of shetches a certaingranrmatical' a notion of defining,or glossing,

r4

r5

I^--..^--:r4rrbu46w rrr + - . * . Lvrrrro ^ f ^ r L ^ - ^ v! v!'!rJ,

D..! ^l^^^-'-rruL ErurJur6 is itsglf a reigrgnijal

speech

- ' t rvhicir can be exploitecl in speech events, that . r me commonalty cf language and many oiher cultural media lies.

i o

event. Metasenuntics Glossi'g specch e'ents take language itself, in particurar the semantics,.;i ianguage, the referen! or object of description.These as eventsuselanguage describe sernantics language, to the of and are thus metasemantic referential speech events. Such metasemanticspeech events are the basis of all grammaticalanalysis and description,and hence of all semantic descriptionas well. They are the basic activity of the traditional linguistics,which may be secnas the discovcryof the glosses a language, the classof all possible of of metasemantic speech evenis in iire ianguage.Lconard Bioonifreid's(rqil) "fundarnental postulate" is essentially about the semantic one and formal equivalence of ccrtain sentences that underlie utterances within a speechcommunity. similarly, such semanticequivalence the level of phrases ai and sentences has becomethe stock-in-trade the transformational of grammarian, rvho postulatcs common "nnderlying" structure for semana tically cctuivalent"surfacc" svntactic arrangcments. But it is intercsting that mqt:s_ema$.j--c_spccc.h eveg!g*41g.*a_:il,tpta!. occurreq-cc- everl'daylpeech, .a culturally-lcarnedspeech function. iu In our society,parentsarc constantly glossing rvords for crrildrenby using gramnraticallyconrplcxbut scmautically cquivalent expressions, exprcssions that makc ihc snrnccontribution to refcrcrrcc uttcranccs of as the glosscditcms. Thc mctascmanticproperty of la'g.age, the property that malies semanticanalysis(and hencesemantically basedgrammar) possible, is the one that is unique to language, and upon which rests the speechi function of pure reference. is rvhat makeslanguageunique among I It all the cultural codesfor social communication. Anihropologistshave I riiuai, n"ryth, other media oi sociaibehavioras making iong anaiyzed or about categories socialstructurc.But of wl'rat syrnbolic statemcnts of medium othcr than referentialspeechcan we say tliat the behavioral
oia-c ^^r' JaonriLa ll'^ -^^-;--. ^f J vr rL^ !rr9 ^.'-..^ Jr6rtJ !l-^-^^l-.^^. LttsrrtJtrrvcJf rn --I rrglg arg n()

Shnultaneottsotve crentictlFwrctiotts N f speech everts trrat do not have refercntiar functionsaccomprish socially constitutcd endscomparable thoseof nonspeech to po, euents. example, is frequentry it throughspeecrr that rvc -.et.social boundaries on an interaction, ratherthanthrougrr physicar the separation parof ticipants. characterize To sucrrbehaviorabitractly,*. noi. that we canchoose language whichrvespeak o, the in ro io pr..tui" comprehcnsion the part of sonre on incrivicruars present; cauusea language \\,e all unelerstand, witil but iirai i:roiioiiiiiiali-,iarkcrs makctjre intenClc<l boundaries participatio'clear; ca!:usephrascology of we only sorne can understand; ca' spcllout the rvrittenrepresentations we of rvordsin thepresence thoseilliterate sonre of in writtenranguage; so forth. and Tlts*purp,qqrye_p4vacy.funq.tjq!_qp_ee-ch qf the b-eh;ioi ii_-simuttaneous.. rfith,-butanalytjcally.distiuct fi_om, ,.fcrentialfunction_therc_.y.lroi.u", is .iq tirc-c-vcut.,fr:r _sPc.,kpr intcnclccl hcarcr(s), onlv rhev 4l<] for --"'t particlpatc trrosc rorcs-in. rcr.icntiri io,.,mu,ii.rti;;;. !n trre ; one of the most aspccts ip.".tt bciravior, fact, is or in -interesting of whlt ,pp.r, to ic uttcranccs scquence, in { :T::i:l,frctiorrnlity ; 'r trachtlonalry rccog'ized rcfcrentill'aturcof sonrc partsof utteri arccsscciiiingto have ftcfC'fi44 rlarrv othcr furrctional clcrncnts j .sirnultauconsly. Frour th;lri,iut of"vicrvof trrc trrcritio'rrr ,",";l;j;;: -l referential linguistics, thcscotrrcrfunctiorrri,rro,t., oi l",f".,o*" ur" i:e "riding on" descriptivc propositio's. But trris is a rather ::rn' !o limitedpoint of vierv.For it takes considerabrc a'arysis the useof of such speech itselfto characterize is goingon in sLrch rvhat cases those as given above. onrybehavioral arethe-speech The data signals themserves.
just in the semantico-referential ,.nr., presupposes grammaticarcrea cnrirrlir." ^ rLrlpLrL'r f cacii oi thc disiinct refercntiai ur ^ ^ ^ l - - - r nrccria, on,r hanaa presup: poses isolationof the refercntialfunction ^f enoe.r.in r.,,^ r:-,. the
T O S aj r r . f o r- : : 1 _ :m r 'rr.lv , ii hiq fL f l i io . * - ^ 1 , ^ - : ^ c ^ -_ __ exa .: i a L , i i i j s i i ^ s r t s - -s i:n g u , r.r a olfferent..ianguager,,

naturally occurring"metamythic" events in the same way thai there nrc mctalinguistic oucs, nor "rnctlritualistic" cvcrrts rvith thc samc frrrrctiou:rl possibilitics, is iu othcr frructiorrnl It propcrtics hngulgc, of

r5

;;;:"Ti:T; speech bchlvior rccnrircs tlr:rt *,c carr co'trast .sig's,ail ritrrcr trrings rer:raining the same. From thc 1:oi.t of vicr' of furrctio':rl lrrnll,sis, t'ltcr, r'c rrrustrr*rkc srPhistiurtctr hyp<lrrrcscs is'/'rr,.,ri'rr,,utir7,r,". ,f u, r7

s,'stems ofsenantic signars.trrc So functron-ai ;;;ffi;i;

t ' .I 1I 'I

, , I

l SILVERSTEIN

MICHAEL

and Shifters,LinguisticCategories, Cultural Description of on squarely the manipulability this mode by the mctalinguistic a actualll' minor onc, itselfis iust one,pcrhaps But property. referencc of act" functions spceclt. or ,n ongthe "performative" "speech W"-dg other" qt5ugtgrc acc,o-mplisht9 basrcal]y "cic9-crrp!i,vql' not usc luiggbtip cnc to bc oncof thoscgQa]s, t,rmiu$c,rfi-cJgqlsj'tlescr\ttion Jrirp12c11,s cnds-. rvitlt othcrfunctional of in ttrafoverlapi folnal strrtcture sigrlals ns iatoryExtensio Abbrev ianrciercniiai of of tirc cascs, coursc, cxtcnsion ciescripiivc In certain is of class events uses to guage otherperformative is p*f?,t:,One such to A used abbreviations as ieqirelts. statement another conversational or suchaS"I'rn Cold" "It's cold with an openwindorv in a roOm person to is until in here"couldleadto a cliscussion theinterlocutor asked close in persons for so. and thewindorv does Or, ntorenaturally sophistiCated to iiselflcading the accom' thc we canabbrcviaic, statcmcnt oursocietl', of nlishment theaction. and suchexpericnce deduction require of subtypes staternents Several dcscrip' But sequences. suchsecmingly on based fui! formsof linguistic eventsare very circum' request usedas abbreviatory tive utterances signaling of a scribed and constitute level of delicacy manipulatory ihe to highlysusceptible failure.In general, point holdsthat d$9.4ptiv0 f$".allg!bg$ arnonglhqjpeeci is reference orre Jurdlgns, not thg.basis
PragmaticMeaningsof Linguistic Signs aPPgSLin-qp-e-g,.h The linguislic. sign*t-tbatunderlie utterances,-then, that servesmany sociaiiy consiituied funciioru. Tite ineanirrgsof such -they-Crnetge are from grammaticalanalysis, traditionally designs,-as speech, scribedin terms of their contribution to referringpropositional
^t vl -^^^^^:L-. ugug)JlLy

beforeany structural -of , comp-4rability function of $g sl$a-ling111tgi?, liscription is iustifiable


erenca and "P erformative" Speeclt Ref

otheruses tP::ill.,-gt!"i3rutlryiil!1 however, :f tust like rcference, somethingi oi "work" aon.; ifr.y accomplish "perfo'rm" coi.rstituted *gllg-thl above,-or as privacy, in the-example rvhether.-a-clieving or fgljgggol9,lg qitbe partiqlp..?tts,Tqlbg-a c-q$man{ statu! .socia'l for statu-s, cllqngcir1 social .a" lq"r.tlrirgl ot _.e".ting; pennancnt
example, marrying two people or knighting someone' o,f ffnucfrr.""rri analysish"s U"e' focusedon this performative aspect

pg,-ryosive fult:-t1jP9.eg trngu^g.use,in wttat t haveheretermed "-: distinguished upon the work of Auiiin, sorne-have .u.itr.-FoUorving ,'performative" content and of aspects speech the "semantic" beiween acts" "speech t"rty tt. phitosofherSearle(t969), havedistinguished as by represented utterances distinct from their propositional.content. has by unfortunate, the way,that "speechact" been 1Ii is somervliat since events, speech functional a term for the level of puqposive "srd.s usingit in anothersense below') I will be semanticn' startrvith a basicaiiy in aiproaches, otherwords, All these the eategodes, refetentialfrniuistic nnalpis {rom whjch ttre linguistie in etc., arrangements, emerge thc traditionalway'_fllgJ* .gramrnatical of horv thqsgsemantics'refereudal i a iack onto this analysis descriptron 'ioi.d" performatively^ approach entirelymisses This dtggqtigr,..n b. events, a fortiori, speech events referentiaispeech tf," loirt that -at_e, events. as with the samekind of purposefulness other speech endowed Th e lin' ce 4g -3mglg-nngsy' n Referen is -o e ki+d-q-fiiqeu-ilfi9-Pgif-9' 9-e speech s. semanjil from anali,sis-of tital emerge ;"1;;r.rt.iriilt i: rrerlr as the are modes not necessarily same tlrosetnat emerge ieferential of to spcak arrangcments ancl modes, it is prcsumptuotts otherfunctional in "usecl" othcrways. beiug nature propositional of a basically 'ray be 'roclcsof speech functional of signals clistinct The physical
.. 11 -r:r-^ ^:-^^ rt-o., aaam *n ha qrnarimnnsed parually allKtrr JlrlLs Luv)| Jevur !v uw -'\1l/v""'r----.-;

-:-^-^--^1-. ^:-^..--^-:L^.1 ^" il ic I'arn\ 'rv'v, (tne tgnn nof trEuruuJly l-rluulrr)Lrruvu 4r 'r 'o

Ohlrers- nartiCU-

curlular ot sense funcuonal in but utterrnaas, the meanings, thislarger The priority -..ningr,'"re ciifierent, h.n.. we havedistinctsigns. "r,d iinguisticcategories shucturerests and in of refeience establishing

i n t_ -e s a m e f O n n a l h ., ! - 1!_-_-r

tlt-elql4-I.-f,:g3ling." is uscs broacier of languagc to dcscribe sider actual the in iq orrly signs, pSrtof rr'Jr-ighsemnntic our nar' q! gopgtilrgqlllugqistic utter' rgyg{!-e_r!g!9]9g'.We must begin with thc factsof purposive The functions. linguistic tircir scveral and in ances sp&chevcnts, isolate
.:--.
Jr5rrJ

^ ---!:^l J^^^-:*!.1^a Pdr Lrdr uvJerrPLrvrr.

T'l.o r lrv

-.nl'la-' H^vvrerr

cal

fnr

,rc .rrlren

tue

ann-

Lr^ Y,v^ . r c

, 1J.l1a r$ ;L- ^ $L il r r's o . - ]" N ur r v

n l , - ro v - ir-r r rc 6 o . w r r q r .

.,rlrinh rln-anrl

nn llreir

nnntrihrr-

lve events canisolate. spcech kindsof functional tion to the several havesemantic signs linguistic in We cansee thiswaythat whilesome
19

rB

MICHAtrL :.^. ' r'iisI'[lN others have nonsenantic mean' meanings, contributing to reference, we functions.In general, can ings,contributing to other distinct speech of call the stndy of thc rneanings linguistic signsrelativeto their comnrunicative functious pragruttics,and thcse r:rorc broadly conceived m eanin gs are th en pragmatic meanin gs.Sqryaghigmg-gjtf&-is,--ol -ggursg in one sensea specialform of pragmatic mcaning,t\._.999-ql.sjpi!gg: tiori oF-sgns tliat contributcs to pure referential function. This fits eiitiiy iuittt ttte discoverythat gramrnatical analysisof the traditional metalinguistic or of the sort is equivalent to discovering class all possible ' events. refercntialspeech glossing ,l ,',"i' ...,,.r-.
t t '

Linguistic Slifters, Catcgorie.s, Cultural and Description

,,

controlling functional rnode of spcech,dictating our traditionrt ,.gmentationsand recognitionof categorics. can then conceutrate We on the nranifoldsocialpragmatics that are common to languageand every othcr fonn of sociallyc<lnstitntccl couulunication in socicty, TI.I.E NATURE OF LINGUISTIC SIGNS

Prtgnatic Categc,ries General pragmatic mcanittg of sigrrsand more particular semantic In in meaning are largcly superimposea th9_fel$l-lignals of speech. a classof signscallc&'ieJeylfiiaUndexesltobc characterized fact, thcrc'is scgarc bclol, in rvhich tltc trt'tl ttroclcs lirrkcd in thc setttccatcgorics, ttttxlcs, irr lrf sirrnrlt:rncorrsly lcltslttvo frtttctiorrtl ;rrrrl urcrrtrrl>lc isirl;rl.rlc
^--,. (JllU -..I^-^-.!:..1 lUlUlL;lffl.ll, ,..,,..'^r \rrlV ttr)t. I),, t)y ^..^...:..:,'!A.lllrlrrrlrS ^*1,' vrrl) ll'^"^ lrllrJv n,rlnanrino L.rrw6\rrrvrt $ .hra .S ( . L r " '.rrirl r n! 't

Flaving discussed the framework of function in terms of rvhich all meaningis constituiecl, shall tum norwto an examinationof the naI tglg_of_thg_ggges 9!;ign-i,fi91tlon-^qf-Iilguistiesigns in utteranccs.By meansof the analysis propositionalcontent in the referentialmode, of rvcwill be ablc to sccthc limitations in principleof pure semarrtie gram. nutical approachcs, tnd usc tlic critical overlappingof functions in rcfcrring inclexcs motiv:rtc a scparltion of thrce princ\>al classcs to of .signuodes. In particular, we can claboratcon the class of inclexcs, rvhichappcarto givc thc kcy to thc pr:rgnratic clcscriptiou languagc. of . U ttarurtcc Scrrlcrrcc, t!trd i\tlc.s.sagc ()o<lc uttcl

in at lcast hvo functiottal ntoclcs thc s:rnrcisohblc sPccchfrirction, for dcictic thi.sor t/trtt, vt'cmight gct thc mistakcnidcl cxaurlllcnn li,rrglislr interreferential categories is that the .superinrposition alwaysof discrete only of pure functional oncs.If spccchconsistccl calatcdrvith otherrvisc rcfcrcntial catcgorics(rvhich tr:rditiorrnllingiristic thcory postulitCs) have sernqntic rvoulcl thcn all isolablescgments and rcfcrcutial inclexcs, rvotrld lrave an additional nr*r'meanings, aud sorneresidtrtl segnrents

lior ptrrposcs scmautico-rcfcrcntial of dcscrilltion, ull rrlfcrrnrcc.s, or rrrc.\'.$rrgc.s, in s1>ccch cvcntslrrc arrtrlyzccl irrsfauccs sclrlcrrccs. lrs of Strch scntences coustrnctecl are from a finitc rcpcrtoircof clcrucntsaccorcling
ln rrrlnc nf .rr,"r-^,",rrl rrL, ."1 drru e^lJtvJJ -^f^.^"r:.'I tvrvrUrrLt.tl '--^,-^^:!:,.,,^ prVlrvJtLrWrtJ. 'Fl-,,^.. f ftg)g ^^-{-Ul!-

as shall r..,;;;.';;;.;;;;;l;ilffiThisis false, rve mode. matic


formal features. referentialand hencenonsemantic of to It is thus possible haveentirely distinct analyses the sameovert speechmaterial from ditreienlTuniiioriil pgints of yietV:The linguistic are signsthat have variouspragmaticrneanings only apparentlyrepre'
^^-t^-l JgllLgu ^l 4L r'l.o tttw " or, t r f ^ ^ n t v , rrowt ^f o-^^^1. innnlirrrrnrrc rrlletcnnec \-f,/c nrlrr

recallVictor Flugo'scouplet,"Gal, atnant de la Reine,alla (tour magi naninre!) lGaiianrmcrt dc i'arine i ia Tour iviagnc i'iimcs."nnaiysis strttctttralhctcrogeneity, lceds to this kind of supcrirtrposcd in gcncral of <lu dcpcnclirrg thc fuuctiorraluroclcof thc pragmatictucartirrgs uttcrd-is!i.rr-clttto.Utittp-yi-cl-<l Oncc n,c rcllizc thirt di.stirr,cl-;ru.gr.utie rurrccs. \1'e on as enr].rse! rrfter,lnees, ean scvcrorlr clcpendcnce rcfercnce the of
... 'q. t svv I

.stitrrcrrts thc mlcs togcthcr coustitutca cocleor gr(nnnrdrfor thc nncl languagc. \\/c scparatchcrc, thcn, thc scvcralincliviclualinstanccsor toften.s occurring nctirrrspccchfrom tlic sciriantico-grimnatibal in fypZi or clements sentcnces a language, of in rvhichtheseinstances said to are represent speech. in In a given speech event, an utteranceor message occurs in context. 'I'hc traditional grammaticalanalvsis such utterances, of horvever, deupon the hypotheses sameness difierenceof segments of and of i>cnds un<lcrlying scnicnccs iire code,otircr tokensof rviricirare nranipulable in in giossing speech eventsby the metalinguistic property of the rnedium. In othcr rvorcls, scrnarllic^o;g1a-rnuratigaJ-q.!a.1.,\,sis.c4r:onlllif-.sign-" function. tokcusprcscrve tlgjt*Jglgr.rrag,in-allthe spccch. cvcnts in rvhich they: <lccur. itcluding thc crucial glossiug cvcnt or its cquivalcnt.\\zc cx1:lain tltissurucncs.s of rcfcrcncc ll1,po.slrrl:rtirrgrrrrclcrlvirrg tvpc,rvitlr tlrc sigrr
c t - ^ ^ r ^ : I ^ ( r JnLrr. r1 .c^r^r r r r v v^-f r -u .l .s .r' ^ I; r r L r a-rl- l^u-i-u- l- l r r E . \ \ /v c r' ( t r -i:,. -^----! -I---: :---f-'t ! ,: illust always Dc aDlc ro qlsHngUlsn

ZT

".

MICI{AEL

SILVERSTEIN

and Cultural D escription Shifters,Linguistic Cate gories, Speech Referential Indexesin P'ropositional more complex for propositionalanaly' Holvever,the situation becomes 'tense" suchassignsfor indexes, referential that sisof sentences inclucle abovcthirt thc vcrb drirrftin thc cxamplcwas"tcnsclcss."But I spccificcl by on consider the other hand an utteranccsuch as that rcprcscntcd the gramnratical commonexatnpleThe boy l.|ltthe ball. By a similarsort of here rvith of we analysis, c:rn say that a sentence English is represented the boy and verb. The noun phrases agentand patient and transitive tie baltare both "definite" (a term the analysisof which I do not wish' tensc" to takeup here). But when suchan utteranceis rnadcwith "1>ast give the meaning (and hence analyze) the verb token, how are rve to underlyingcategorialtyPes? qdrly;the form hft is to be segmented hft]v* Past (: hif]v+ as walklv * Past : walklv* Present).Under such an analysis, Presentz: we can glossthe stem hit and give its "senses"as gramnraticallycomplex
-^r.^J,racpc
P4r4l'urocvr.

contypesfrom utterance-bound, context-independent sentence-bound, analysisof lan' textualized tokens in this PuIe senantico'rcfercntial guage.Where this property of speechsignalsis not found, the iradibrcaksdorvn. tional fonn of gramruaticalnnalvsis Pr op o sitionaI Anallsis we using ihe traditionai grammaticalapproach, can analyzeanv sentence the signs of rvhich are purely referential, that is, rvheretokens in mctaiinguisticusagccan bc saiciio rcprcsciiilireeisclyilic samciiiidera elements greatnumber as lying tvpe.we can analyze distinct sentence chair, man (in several of ti" ttoonr of a language,such as English table, great number of verbs,suchas stand,run, eat; and a num"senses")1a 'number' and 'aspect', such as ber of apparentgrammaticalcategories, which I discussedabove. So predications of timeless truths coded by rvith such elcncnts are readily analyzablcas such, e.g., Uni' sentences corns drink anbrosid. (T'hc verb here is "tenseless";that is, does not refer to the prcscntbut to all tintc.) subiect ancl a transitive predi- i This examplehas a plural noun.phrase proposi' catervith verb and massobiect noun, and it codesthe universal We mighi represeni this propotion that ali unicorns <irinicambrosia. way, rvithout logical quantification, sition, in a kind of rough.and-ready 'eirink(uuicglns,q4brosia)', sllowi4g that 'dlir:k' is a "transitivc" as prcclicatcof trvo plrrccsthat nrakcsa claiur abottt art "agcllt" (rcprc' sented by the subject in grammaticalconstruction) and a "patient" by (representecl the object). For each of the sign types that make up we of the constituents thc sentence, cAnglossanothertoken of the form they repre. about the grammaiicalcategories to . . .-under hypothcses anaiysisto give the sent. (It would iequire a treatisein grammatical mode be' Languagein the sernantico-referentjal heurisiicsof discovery. to must be cornpleted justify a , ing a loosesystem, much of the analysis such as grammatical categories, hypothesis,) For the residual ' pa-rticular object-of[*ass] , fcount] nouns,and subject-of-verb-representing-agent, we can shorvthe proportionality of meanings verb-rlpresetting-patient,
....-lo. lrnnclnmrolinnrl r^s^ror Ltlluvt rnnnirrrrlrtiont " As orlr (oost-)Saussurcan Drin-- --L \I t rvitlr a ParaPnrase-h unlcorn ls . . ., t r r r l u t v t t L ur r '
^ ^--L-^^: :^

Tkrf rrrhat nf the nrnmholooiccl '-'*o----

sepment --o-----

- Past --

that

rve wish

.',

'f

sl' I v^ )'inl'r ' \

io

Whilc it is pcr' tlte undcrlying utterancc? lo aitributeto thc scntence of a resiclual thc grammati' as sucha category to fectlyfeasible segment ' as cal analysis, we c:ln seein the proportion just above,to givc a and is in meaning terms of glossing impossible; semantico-referential by explicable gramnot to ),etthereis clearlya contribution reference tqns3, o$gr-wo1$' is-nqt The inaticalarrangement. g4tqgory p-q5-t U -gf arxl.l'rcncea token-s, ; rcprggclted inttle6rye9l-fu?:qfq::e{er-qr+ti+l -qjgrr this to qqalyliS sUcf t6rnantigo-relereniial h.pqiy-erl,q$ describe obvious the of , category lrngurgr. (That this fact has not hindered description t u r of languages do theories not the natives' the merelyattests kuth that prevent are always us what the natives really doing,nor do they teil bounds') that obvious solutions arestrictlvout of theoretical of Presuq|osition Beferenee !r'.d.exics! we of the meaning this kind of categor,v, have to "l , In orderto describe ,l .tf I make of of aboutthe class tokens "tense"in utterobservations certain t r._.r\ances. the by to contribute propositions describing time of an T'hese '' tO ciainr be verifiable some makes event; that is, the whoieproposition ,, But are suchsign tokens referential. for a particulartime. In this sense

dernand. ciples
na

lvl ICI-IAEL

SILVERSTEIN

Shifters, tin5i,stic Categories, and CulturalDescription

' ,

the more specifically, past tensetokensrefer to a time t, that is as' them is coniaiuing prior to the time t"nat which the utterance sertedly in tcnrporal catcgorics, pasttense particuand In spoken. othern'orcls, of speech the lar, compare time for rvhiclithe proposition a referential speech is asserting something with the time of the referential event 'tense' of type meaning any categorial to eventitself.So the referential depends a comparison the tol<ens whichwe want to assign several upon of in event of the time referredto rvith the-.time utterance eachspeech incorporating token. the or of tokenof the pasttense uttcrance interpretation each The proper of thc then, prcsnpposcs knowlcdge thc time at which the catcgory,
SPCL-CIr gvgl|l
--- ---L

Lar\(.:) Ptiruu. J\

-l----

,.1.-^^

r^---..

LUIIJG uiltcSgry

!^1-^^

til.l\l-J lllu

!l-^

!:----,

Llluu

ur

^I

rl-^

Lllu )Ps(.ull

^---.^^1-

in cvcnt as thc fixcclpoint of comparison rcfcrring to anothertime, t.. "cxistcncc" of t*u, just as t*,, clcntlnclscognitivc cogrritivc It assumes or such a tcusccatcgory, its cquivalcntoccurs. "cxistcncc"onll' 11,hsn ()ttlr.'grtrics l,ittgrtisti<: l)tnrlilt:Alo<ft,

Ii

in speaker a refcrcntialevent of the propositionencodcdby the semantico-grammaticalelements; deixis, rvhich indicates the spatio-temporal rclationsof somepresupposed rcferent in thc speechcvcnt to spcaker, hearer,or othcr referent; and so forth. A verylrrrgepart of the Wlrorfian oeuvre (1956), in fact, can now be seenas a first attempt to draw out the Boasianimplicationsof how pure referential(semantic) categories and duplex (referential-indexical) ones combine differently from language to languageto accomplish ultimately isofunctional referential la,!&Uagg acgomplishes.inu-tt9-J1ng_e_g.y!!h speechevents.Q{_ope a f ( ulg]9;gt"_tSgljql ildgl _(f_o-r e-xample, tens.e),.another..accompJishes \, rvith plus ,qgrlanliccategor.y leferentialindex (for example, f 1 c9m!-inat!.gn--qf
a.rna| ol^s,.c\ \X/1.,'-t r Yv rrvrt ' :_t__"Jr""t-_l*.-'],.::.::,1:_.1 _ l.:..,.,.1f rlrrrrJwtr 1..^1.,,.1 r.tLAL\r +1.^ Lrrv +1,^,.-^!:^..I lrrevrvLl!dl ]^-...:-^l^-., lLlrltlrlvt"by l

with which to nrakethis clear,and his rvritings have had thc slcl fate of bcing nrisrcprcscntccl thc "poptrlar" anthropological in litcraturc for a gcncratiou,unr'lerthe guisc of sonrcvagrrc"relativity" takcn literally, rathcr than asthc mctal:horicrrl icliornof the thcn-bcgirrrring afornic agi. Itu/rrs [-]s:cr ol

'i

"l'lriskirrclof rcfcrcrrtill intlcxlr:rsllso bccrrclrllccl ^6if'r"r,ltcmttsc on thc rcfcrcncc"shift.s" rcgularly,clcpcncliug tltc fact<il*ofltrc s1>ceclt rcfcrctttill iusituation. It is vcry hrtcrcstingthat thcsc prcsr.rl>posing, opcrat' or shiftcrs,arc what fakobson(tqlZ) etlls "c1t1pl-cx_IgJlS," dexcs, The mcssagc sirnuitarrcous-l-y. segnrentation irrg at-the lcvclso-f-cc-rgiq-q!_d r"rai to thc iccoguition of scntcnccsin tlic r";r;ti.";.f;a;i"lr--a; on by unanalyzable the methodsdepending thg of this semanticresiclue, but constituting distinct kind of superimposed a property, metalinguistic rvith pure semanticcategories. linguistic type that fits tongue-in-groove as sign vehicle a Such categories tense unite in a single segmentable or quasi-semanticmeaning and an indexical or pragmatic referential
^-^ vlls. 'T'L- -^f--^-*f^I l rre rerervrrLr4r .'.I.,^f ^ ,l-^a-,{. n- }},a t9a.u-Y,,Yl-.1.:_rr:r.!vrr_t-lvrug-Yerr-svye-r.ruJ^-"*._.-::-:-ta-.-:9Y'y___ "1,llro-

,.1 "

A considcration such cluplcxsignsbrings up thc rgucstion horv of of thc indcxical of urrxlcof suchscgnrcntablc clcurcnts uttcranccs to bc is clcscribcd, is, to bc given a systcnratic that accountin tcrms of sign types anri meauir:gs. We irave secrrtirat tire particuiariyiuciexicaiaspcct of. thc rrrcaningof sucltj]Iiitcrs ir_rvolvcs prysyp.gqsjtiqn a *q"f*t}-c--"cxis:i t9tl9j_l'_oj,jl.cog11Jiyg valqe in the domain o!) qp-qqi6q .fogug=on,;or3g gf situation. On the onc i:and, the referential cori- i .variab-les tlg^qpgegh tribution of a shifterdepends the specific on valueof one or more of the variables being realized;on the other hand, the specificvalue being rcalized during somespecificutterance permits the categoryto occur as
c chifter nF tlro l cnaniEn onr* J PVWUTW JVr t.

posiqJn iir pi' g"i.i19-ia!i-., o-i


incorporate theseduplex signs,referentialindexes. Ail languages Th"y whieh anchor, as it were- the semanticoare pervasivecntcgories, rcfcrcntial nloclc of signs, those rvhich rcprcsent pure propositional cvcnt of rcfcrcncc,l>y capabiliticsof hngulgc, in thc actull s1>ccclt ou rcfcrcuccclcpcnclcr:t tlrc snitableirrclexirrg propositiorrnl making tlrc but of thc spccchsituirtiorr.Not ouly is tcusc such a cluplcxcategory, also.sfatus-, which, follorving Whorf. indicates the truth value for the

We can summarizethese conversepropertiesof implication between contextualvariableand indexical token by a generalfunction we can call a rule of useor n$ellrn$$ealin We e'.an that a n:le*o:l use*is_ s-ay a_gcrlgglr9!f!ra111!q11!r_c ci4gs_tf_irctua! shifter tokcns occurring.in the, c.l,1gs*9f_1c,!g.11__q2cc_c.l1gg1[-c1t-s. thc spccificnlly inclexical In this scnsc, aspcctof a .shiftcrtoken can be saiclto rcprcsentsomc irrclcxical t'ypc, tlrat is, sourc unclcrlying gcncralsign thlt strndsiu thc srmc rclatiopto its tokens-permittinc us to analvzethem as "the same"-as the usual

z1

z5

MIC}IAEL SILVERSTEIN

iptiott and CulturaI D escr Shifters,Linguistic C ategor ies, can of But tive case. in both modes a shifter,the description Proceed mode, referential In tokens. the signtypesfor occu:ring onlyby defining basedlinguistic plirli.d iiltough traditionalreferentially tfris is "..on lsrvcsshiftcrs rcsicluals, tlrc irrclcxical it nroclc, is Irr irs which analysis, trsc' of rrrlcs of tbroughthc constihrtion gcncral lccornplishc<l
TrichotomY of Signs Peirce's These two modes of signification combined in the classicalshifter illusirate z of the 3 elemettarv sign tvpesgiven by onc semiotic analysis ''''1, of of C. S. Peirce (t-glr). Altogcther,hc prcsentedthrec trichotonries on the ), on eachone classified a distinctbasis.The 6rst wasbased signs, ,', I nitoi, of the sign vehicle, the secondon the nature of the entity sig' naled,and the third, the most important, on the natureof the relation' ,),,1., t"t ship betrveenentity signaledand signalingentit/, !hq! i1 oir th3 nature of ihe meaningthat iJcommunicated. (Of the z7 iogically possiblesign
tyPes, only 1o occur, uroug[
r r r - :1r -^r

in sign. sort of generalsemantico-referential It is clearthat the senses a fact are which ,i" h"u" signtypesin thesetwo modes quite difierent, on for grasp, the onc cleperrcls rulesof usefor defini easyto not ahvays of opcrations glossing tion of tire typc,tlrc otlrcrou thc rnctalinguistic or cvcllts thccclttivllctrt. spccch cription of Indexicality Formal D es be' thg function that describes relationship A rule of useis a general someoi rvhicirmusi tweenspecchcontext,givcnas a set of variabies, somemessage and someportion of the utterance, havespecificvalucs, eventgivenat of the fraction.Recalling minimaldescription thespeech z, y to hearer aboutreferent x the outse! we cansaythat speaker speaks of in itself 0), analyzable -terms fraction do (message gsing mCIsage grammarG, at time t, in spatial configuration semintico-referential of l,J'tl-)-the referentneednot be presentindependent its -creation wiil the variables of Some ly tft. rp.".fr eventitself-plus other factors. as bl presentin a clescription such, while for othcrsrve rvill have to use thc orderto characterize appropriate of in ,p..ify particularvalucs theshifter. evcntto rvc rvlierc rcfcrin tlc spccclt past'tcnsc', Thtts,for English thc wc candcscribc iuclcxical a time bcfore thc tiruc of the ttttcrattcc, sf aspeet t}is shiftcrby thc sclrcura: t"u,l, sp(x,y,tr(t-u,{-od},Gs, . . .)+Plst tcusc value valueof the rcfcrentz, andt o is the specific wheret, is the specific 'status', asserts wherethe speaker For assertive of the time of utterance. -being in Engiishwe usea heavily uttered, tie buth of the proposition
JLlAJgs

I wlll IIUL ut;v$rvy

-:^-.^t^--

r1.i^ D-i*^nan.lerlttnlinn

ru'

r e'wsr

here.) by The threesigntypes,eachcharacte{2g$ its owtttyPe o.tmeanrng a-rq$-i9,iry!$)a.{g'illi?.-fcorts arc tlosc signsrvherc for t5e .rsers,
rl.o ,.arnpirrohla nrnnerfics of the sign vC[iClC itSClf }avC iSOmOqrhiSnr tO ----- --q rr.w
rw.vv.rsr,.l---r--"--

o|=aceerlinf,pr^ferl ueib- such as -auxrliarv or modal, in the general case. ------l..rvvlvs J

We can describethis bY: 'sp(x,y,T(f (2,,,, .,zn)),[Ari*],Gt,t,l, .,) -+Assertive . and T(f ) arguments f wherethe proposition (2,, . . ., zo) may take several is the truth-valueindicator. indexical to Sueh rules of use for shifters are necessary describetheir
. ' moog

arC frtcleXes thosc signsrvbcretltc OCCur' SCnSc. in arC,'liker:CSses" SomC -' tc q:l*io' ol a stgnvclllcl tokcn bclrs r ccnrncctiotrof utclerstooel rcncc is, <lf tcurporll coritiguitv to thc occtrrrcucc the cntity sigilirlcd'T'lrat in thc-context of to oflome entity is perccivecl bc signaled the presence incorporoiing t6. sign vehicle. Symbolsarc the residual communication elassof signs, where neither physical similarity nor contcxtual con' They form the tiguity holi betrveen sign vehicle and entity signaled-' spokenof as the fundamentalkind .rr" ir.,arbitrary" signs-traditionaily of linguistic entity. Sign vehicle and entity signaled,arerclated lhroug! meaning in the scnseelaborate<l the bond of a semantico-referential carlier. and in Everylinguistic sign token is an icon of the linguistic lign typ-e, an iconic. mo{e. this senseJvery linguistic sign trivially incorporates
F .! f urtnef, . . svgry --.*r--.! symuul r-!,^rt ^- uruv^ vt rrrv L(,rssrr i. dll i-,-fao nf the s'.rmho! tgoe- sinCg itS --' 1r-,

T"lat-is,t1c cntitics of (up to i4c'tityivitir) thosc t1c c'tity sig':rlccl.

rvheretwo meaning.In thesecases, scribetheir semantico-referential w8 havc a fortunateillustra' modesare united in the samecategoYr z6

-. -^--:-()I lucaurrr!,

-..^1. llluerl

^ D r tw vr d ^ , .r,u- . o ^ f d d r a n m . r

(-1 c-r-e n-e - -e s- a r v t o d g - -c -- s J -

of "existence" that part oi the upon cognitive usein conte*tdepends value.In explainsits rcferential which grammar scmantico-refereniial 77

MrUll.all!L

.Jl L

Y li,l\J

^ !lrr

t gO Ar tLC J I Iq C S, Lrllg? tS V LLe r rct.,d I lu Lt t LtUfdLu e SCr tlo 1 4)

relationtlre as this sense, Peircenotecl,thcre is exemPlificcl progrcssivc of thc thrcc signmoclcs. ship of inclusion lcons since,inter' with iconismin language, I do not deal here .cxtensivclv with to estingthough the subjectbe, it is largelyperipl'reral our concern At of the cultural contextualization language. the formal levei of single which contain onotnatopoeias, are units,however,all languages seento duplicatethe thing signaled in the physicalmedium of sound.Thus, 'noisethat that means of bziz,to a speaker English, is an onomatopoeia bees'flight, gouncls like thc sign vchiclc" uscd particularlyto describe and so forth. It is usuaiiyassimhigh*pecd sawscutting through woocl, Since pattem of thc langtrage' ilatcclas a lcxical itcru to thc phorrctnic rcquirc a volel, it is writtcn out as buzi, irr rnonosyllables Iir-rglish found for is prono,i,rccd [b.(z(:)] or [baz:].'fhisassimilation frequently lrrrr' vlricty to thosc in cliffcrcnt git'irtgrt rcrtrrtrkablc orrouutoltoci:ts, rroLtll:sctlrc llttt this s'horrld flrc slrnc rroisc. to s:ri<l rcprcscrrt glrilgcs it,^ r,,^rr'lr.rrtn ll'r' rrqlrq flrc iconicrnorlcof rncltttittg thc orlc that is itr significatrcc spccch. thc givcs sign Thcrc arc l)liuty kirrcls of icotts itt httgttltgcs,rarrging ftont reliicas of rvhcrc tlc physicalproperties signal and thing signaled and,bnages, rvhercthe pcror totally alikc, throtrgh diagrants, arc inclistinguishablc arc c'liagrrn'rs spcechparts are strltctttraiiy isomorphic.Many ceivecl are for of internai.UnivcrsallarVsof sequencing morphemes, example, of sr inverse diag+ams syntactic units, and so forth. frequently djreet
LlrU l.lvl lrrs(r Lv O

This proposi' the that typesin ccrtainarrangcments ur:clerlie tokcr:s. rclatiorrs be ana' crn of in tionalvalue thc sigus termsof equivalcnce then,are rvhatrve Suchsymbols, n:an\rulatiorr. lyzcdby urctascltlalrtic analysis. on abovc propositional in describcd tircscction vs. Syrnbols Shiffers
that the svmbolicmode of signsis one mechanism It is to be observed events.The imple' in for achievingrcference actual rcfercntial speech of the symbolsby tokens dependson-presuPPoses-the mentation of knorvledge the grr*ntar G in a pu." ,.f"r.ntial event. In contrast,I are tirc shiftels,-refercntial -iuclpxes, a Fechanism in rvhich there is no I rclations, but ouly thcl of ptol>psitionalccluivalcnce lbstract-rystc11rulcsof nscrvhichspccifythc rclationshipof actualrcfcrent of the sigrl I of !q!gn !o thc othervariablcs thc contcxt,amongthcm the sign vchiclc: I c.11en! is.ggnstituicd by thc spge.ch vaiue_pf-"a-5.1!fter 1.119_1c_fcrgntial varirblcsof thc spccclr cvcttt,irrcltrdnl,ly l)rc.supl)osc al)y itsclf;slriftcrs
" g l, i n g f l r e s c r r t l r r t t i c u lllltr' t s c < l r : l u l n l : l rC ( f t l r c x r t t t r l : l c t r r : t l l l r o r i cs n ' i t c l t
.,,f,'.,'.-^.''t\ en.,,n 'li"l;'.'.'.i"1. l.z'1.',,.r.rr or.rrrrrrlin.rllrr l u l ! | u l r L ! , l . \ ) \ , l v u r r r ( r n l \ " 'r. J"($r r | 5 ( l l J l l r , v L l v w v l r J w | | l | . | . l r w . l r ' r.rrrrclilrrt'r.rl

\ s1'nrbol.s, lbstrirct propositiorulvalucsof rvhicharc intplcnrcntcclin thc I rctwrl rcfcrcrrtiiri cvcuts,aucl thc slriftcrs,or rcfcrcutill iudcxcs,thc

I
J

lxoposiiionalvalucsof rvhichare linkccl to the unfolding of thc spccch cvent itsclf. Thcsc are two distinct tvpcs that rlerge in thc at)l)arent but stmcturc of uttcranccs arc analyticallyscparablc. lndexes in We have seenindexicaireferenceexernplified shifters.But it renrainsto observethatindexicalit2, the property of sign vehicle signaling
nnrrl'avtrrel (tevicfpnnett nf an anlifv ic i+self o cio .* "'b"n marle inrlarrr"nr:lcnf

SYmbols is language most "languageTnfhe s,..mbolic rnofle of sien meehanism, of negativecharacterization no From the like', in the traditional sense. tOken betrveenSignvehiCle or nc..qqjlrvn!'r..,sical contcxtual connection signalecl,the syrnbolic mode of communication clepends and entity
CntlIClV Olf an t , al)SfraCt -^,--- --^rl^-cullllUcLlL'llr ---^!:.,^r^l rltutrvdLurt rl.-^.,-L Lrrrv(I6rr .n-.nlinnrvr.'r'rLrvv

of the other two. In the duplcx categories illustratedabove,the referentirl wr'lrrn rlprrnnrln.l rr!\nn lhF inrler.inr'l rnlrra Of nnrrrcp
vv u. vv,

ilrprr

it

r'c -^c-

sign types ancl their rules of combination.This kind of graurpiatical of s)'stctuof chssical cliscussious larrf:urc rcfcrcncc fonns tlrc closccl in rlf sigrrtokcrrs irtlygivcncvent vnlrrc gurg.*",t,t,ttics.'l'lrcrcfcrcrttirtl contributionsof the sign a.pendsonly upon the scncral proDositional 28

siblc to conceive indexicalsigns of languagewhich do not overlap of rviiir rcicrcniiai caicgorics, iirai is, cio not coniributc io aciricvingreicrcncc.Sucl'rnonreferential indexes,or "pure" indcxcs,arc featurcs of spccchu,hich, inclcl>cn<lcut irny rcfcrcntial spccchcvcnts that nray of llc occurrirrg, sigrurl sorncltlrficrrl:rrvlluc of onc or l'norccorrtcxtual v:rriables.

z9

MIC}IAEL

SILVERSTEIN

Shifters, Linguistic Categories,and Cultutal D escription the opiration of phonological changesin the one form and not in the other. And these pairs of related fonns can function referentially in containingthem in the sameway; the only difierence utterances exactly of is in ihe pragmaticsuitability for ccrtain classes spcakcrand hearcr. '['hcsc morpltologicalancl phonologicalruccluttisnrsof sex csscntially and described from utterances indexingmust be functionallyabslracted Koasatisp(9 (x),y,z,V]+s,G6,t,1,...)tI by rulesof use,for example, will take up the characterizationof the Yana casefurther below, in disrule mechanisms. cussing
llxacuv tfts samg sorf
^ -'- ^t

we From the point of viervof pragmaticanalysis, have to recognize regardbehavior, of speech contributions indexical suchnonreferential indq5lg3l.eventoccuning. The$ YaIigusspeech lessof the ciominant gygqtq' distinct--s-pCech ftgy go languagc it'ttoconstituting of clcmcnts !yjt! bqt t!:cJ rrie fnuction,rtty:.tirct.tc, b,qlnvlQrilIly lrqlcit-tj,tlly--qYqllqu r q-b-o:s' relgrellalsPgggn tl-1ggUif'gq!'of,d f!!gl?ng9!, 9! !e !e!94 pure inae=C*t T"rt"*r uttiiancei are describablervith rules of use, "f gut the rules'of-ost do-"oi ip*ify ;"tai:* ;'reTerential-ina.*.r. " referent independent of those createdby other elementsof the utter' ance, for thcsc inclcxesarc not rcfcreniial. The "meaning" of these indexes is purely pragrnaticand does not intersectrvith semanticoreferential meaning exemplifiedin s1'rnbols. i --

uf, uullftrItrlcrlllal

---^l^-^-r:-t

:-

ulut^lLdr

J^--:^^1

-^J-

lrrvsv

:"

^..-l ro frvurru

i-

rrr

s N onreferenti al I ndexe speechevent i i Su.l indcxcs as do r:ot contribute to the refercntial context.Someof the most interesting isignal the structureof the speecir J ioithese indexcs,certainlyior the sociai anthropologist, those that' are inclex feattrrcsof thc personscof the spccchcvent. For cxample,sex or categories othcr are for indexes somc languages fornrally systernatic languagcsof the southeastcrn obvious featurcs. In ihc Muskogcan Uniteil Statcs,such asKoasati (Flaas1944), therewasa suffix* (or its ctlnnologieal eqr+ivalerrt)tjurt appcarcd (rvith charactcristicpJrOnolOg: vcrb fontts of cvcry tton' orr in icnl altcrrr:rtions shnpcs) tlrc inflcctccl 'quotative Irr uttcrancc spokcnby a socially fcmalc incliviclual. dircct'

is speaker index' as quotation, wc' !might.*peci,thesexof ihe original +


:--!!-.
lcarly

*=.-.---,^t

or is utteiance, and of the virb especially, exactiythe same,whether makesnc referentialcontnbution, The suffix not the form has ihe s'uffix. providesthe categorialinformation or but rather its presence absence Noi oniy 1'firstpersorr"r'orms sex about the sociological of the speaker. of verbs, in utterancesreferring to spcaker,but verb forms of all "pertake this suffix,and the rcfcrential contcnt of the speechin both sons,, fonns is unaffccted. and suffix-bearing suffixless b;r is reporterJ sapir (1929) for Yana,a language A more cornplexcase of California, in which there is one form of aii maior woids in utteranother rnale, male to sociological ancesspoken by sociological -and truo forms are typically relatedby form for all otirer combinations.,The

Pftistrlvqjll'

lL

i J i n rn n rrlrc r rrL a rlr Pw dr l l

+v^ evp e r r r 4 lt rhn o v r

the referential

value

Of the

indexes, where speechsignalsinequalitiesof status, rank, agc, cleference rePorted sex,and the like. For example,we may take those of |avanese, by Geertz (tq6o) and more lucidly by Uhlenbeck (r97o) and Home a (ry67; +B), whereone of the modesof contrastis betrveen vocabu(the variety calledkrtmc) restrictions lary set and certaingrammatical rvhile useclbasicallyby lower-to-higheror high,to-high on these scales, other, "unmarked" vocabularyitems and all constructions(r3rkr) are items It uscdin the oppositecases. is intcrcstingthat most vocabulary forms, 1'et do and virtually all constructions not have these alternate the power of the alternation was apparcntlyvery great in traditional conteni oi iire utterances society. Fiereagain,the propositionai iavanese krrmr/jckr vocabularyis iust the same, while the rvith corresponding and defcrcnce lhey inclexbetwecn spea-ker ltcarcr cliffers'The rulcs of itrc for clcfCrcncC nlrval,sof t'hc forrn rusc blscd ott tltc parrttttctcrs l s p ( H ( x ) , L ( y ) , . . . ) ,p ( L ( x ) , I { ( y ) , . . . a n c s o o n . s ) frequently and especiallyintersect with the These deferenceindexes referentialindexescalled "first and secondPersonPronouns" in the
stanqard lltelature, glvmg, as lor cxamPls llr lllal
, .----l-:T'L^: --l tl.---^^^ ta'^^l,^ alru, rrLrrrrrsJe \vuw^v

"pronouns" for useas 19Zo),upwards a scoreof setsof segnrentable of inder combined into one referentialpersonalindex plus pure deference
nrrnaranr sry4rvrrL ..,-r^^ourr4vv a^t oaa* vdre6vr/. Tr-r r *^n,, J u.4r. lono"ooec /cea Rrnwn qnd Cilmen

in markingof socialdcference pronominal r9fu), functionallyanalogous indexesis accomplishedby skewing otherwisereferential catcgoriesof 'person' 'number'. and These special eftects,pragnutic ntetaplrcrs (to from a distinct indexical be dealt with below), are to be distinguished
n.'nrFccinn nf cnaial .{efercnnc rvith .'"
"

nniorre -----l--

forme]

sisnals-

of bifurcation lexicalitems into comple.\ distinctnonreferentiai speech Aboriginal in mentary indexical waswidespread Aushalian sets
ll

3o

MICHAEL SILVERSTEIN

Shifters,LinguisticCategories,andCulturalDescription

.l

by communities.As described Dixon (ry74 ry72) for Dyirbal, a lan' there is an Rain Forest in Northern Queensland, gtagc of the Cairns set, "evcrl'day"sct of lexicalitcrns,aud a "mothcr-in-larv" rvhichhad to rnother-inof only in the presence his classificatory be uscdby a speaker vocabulary law or equivalentaffine.In other rvords,the mother-in-law totally distinct fronr the everydayone, indexesthe specifiedaffinal rela' speaker(x) and some"audience"-not the sociallydefined tion bctrveen (f)-in the spcechsituation.As such,the srvitchin vocabu' addressee situation,maintaining as lary serves an affinal taboo index in the speech distance. and creatingsociological structurein the traditionalsense It is interestingthat the grammatical What changes remainsexactlythe samein thesetwo kindsof situations. lexical items. Moreover,since the is the entire set of nongrammatical is to vocabulary approximately ratio of everyday nrother-in-larv 4:r, the vocabulary codedin everyday strictly scmantic contcnt of propositions and many more n'ill rcquirc rnore claboratcgrammaticalconstructions Scnranticcontentwas lcxicalitcms to coclcin urotlrcr-inllnvvocabttlary.
^,-,.'.'r.,rll.r c/ri/^r.,lv rr,rlrrr.r.rl irr:rnltt:tl lnrlrrrrrtrir':rlirrrt lirrrilrcr tlrrl

refercnt (that is, sp(x,y,y,...)).Sincc the lexical alternants focused havethe samebasicpropositional valle in krcmc iggyelor plain styles, understrict semanticanalysis should rvant to dcscribcthis speakerwe referentdeference switch as a pure indexicalone. But especiallyin the of about the addressee, case speaking where the reference perforce of is krcmc : r3cko, an indexicalsort, the two systems krcmo ilgel : fplain] seenr merge.The actual facts of any given instanceprobably rest ulti. to mately on the distinction betweenindexicallypresupposed and indexi callycreatedreferent anotheraxis of classification.
1

I ndexi cal Presuppo sition of In all cases indexes, have conshuctedindexicalsign types by we rulesof use.These rules of use state the relationshipof mutually implied existence sign vehicletoken and certain aspects tire context of of of cliscourse. all of the shifterswe cxaminedin the section on referFor cntial inclcxcs,rvc coulcl furthcmrorc say that thc asp,ect _tlt-c*lp_p,.clt of 'l'hat is, ,r.Si-tgl-:11'..f!,"+ ltyl {llqtiqrl tlil! -!t-!g_lyp!trt;od" thc sign tokclr. tokcrq.iq gnintqrprcgblc rcf_crcntially knowlcdgc 9_{so1.n9.rvithout-thc aspectof the.silu:rtion. A plrticularly clcar cascof such prcsul>position thc opcration of is deictics, English, for example,fhis and tlrat in the singular. Wl'rcn in wc usea token of the full noun phrase this table ot that fable (rvith strcssccl vowel in both words), pointing out therebysorne particufull lar object the referent of the token of table must be identifiable, must "exist" cognitively,for the deictic itself to be interpretable, The proper useof the token of the deictic presupposes physicalexistenie of in the

l>ackancl fortlt, unclcrthc santc grantlltar,can bc possibilitics glossirrg But tire ns cxploitccl, wastlortcby Dixon, to iustifysemanticdescription. pfinciptc of this bcirrg a purc inclcxicaldcvicc, inclcpcr:dcntof tire form of the rule of use coutent,makesthe general semantico-refercntial the sp( [rx.][Af (x,y') f ,y,2,{LrJ,. . .), whereAf (x,y') expresses relation' the and "auclicnce" and Lr(:Lt) reprcsents disship bctwcenspcakcr junet set of lexieal itenrsSo therc is a distinction betrveenlefe-{glijial_ilde.. such ,as tepse,. 9s,. ones,such- lhe dfsjulct.sgts"*qf-fo-rmsjo-.codQ.-s-o:. as and nonreferential situation. Some pheciologicalrelations of pasolae in-the-.speech. between nomena, however,appear to bc interestingborderline cases basic ciistinctionoi In where the shifters and pure inciexcs. Javanese, index, there is into krrrnc ancl rlrkr sctsis a pure deferencc vocabulary distinctionbctwcena sct of lcxical fonns (krcure pcrvasive anothcr,lcss
rygyei) snowrng ccrerelrcc ur sPcaKsr LU sulll.c sxalLsu lltllrlarr rsrtrlslr|..
. r \ t^t---,.--l ---l ^ ----l!^-l L-----^-C--^-r

actu-I o6leci ;hich- aa; propeity -igga*ea- Uytabla,or"if iiii:sup U9 t9-

poscs priori6gment of referentialdGiourse *trict, has specified such a a rciereni.Oiherwise ihe use of the rieictic token is inappropriate;it is uninterpretable confusing. ancl (There is a relatednoun phraseincoqporlting reduccd-vowcl "deictic" form, with rcduccd strcssand distinct
inln'raliaaalla* yarrvrrr, ..^^l uJeu f^tvl ^^-J^:^!:^ rrvttucrullu -l^C-:!^ ugltrrlLg -^c^-^-^^ lgttrIgttL;g, llu plgsup-

This sct of forms, codedin stemshaving to do with parts of the body, pcrsonalactivitics,arrclso forth, occttrsin both krrmr nrrdlrkr stylcs. but one which interaxis arr It constitutcs inclcpcrtclcrrt of lcxicalcltoice, dcfcrctrccwhcn the hcarcr is also thc scctsrvith thc sr:cakcr-ltcarcr

positionof referent being involved,and no prior discourse necessary.) If rvcusethc wrong cleicticfor the rcfercnt, or riscthe dcictic rvith the tu'rong lcxical noun (onc that cloesnot properly clcscribc obiect in arr corrcctpositiou for thc dcictic), again coufusiourcsults,or corrcction

72

\
l: !sl.r .. j,.(.:.1'
,

'33
t
4I -, tti,t t.

3
L.i

);

;.

./

,/
l

.:,1' :)., :. ,.-.. 't-.t.-ri-r 1., .t.". . ,.., ru i , l,{r<-ct-t,i.

'l\

i
9 t !1L'ti,ri*i,Lr;t,-+, n,; .r. .

MICI{AEL SILVERSTEIN

cription goies, and Cultural D es Shifters,LinguisticCate insult, or go into the llg]{g-ll,corstitute a porverfulrebuftor -bCA-rqq of creation iioqy aq4-hqiliglof Classification Index Types/ T okens inrlexical tokensmngeon a sliding scaleof creativityor pcrformative by displavccl <lcicticsto thc value from the extremeof presupposition extreme of creativity displayedby subtle social indexes.Ttre particular on placementof any given indexical token dependsto a great extent of its use: horv many eventsare i1e factors of the individual context occuning;horv many independentnredia are signaiing simultaneously havetaken place;how many the factorsof ihe context;what prior events sort are occurring in speech. cooccurrentindexesof a given lunctional As rvc have seen,the different kinds of indexical types have inherent Underlying all these rangeson the functionai scale of presupposition. are horvever, the rulcs of 'se, nortns as it *'cre, for thc qr.iifi" usages,
relauonsnrPormu[uaIexNtclrUgugLlvgUrIlUtrlvAlLr4|
r :-r^-^^ L^!.-.^^.^^..!a-l,.al 'r.ti.hlac rnrl

"Oh, youmeanthat othertablel"or "This is not a by the interlocutor: table,it's a chair!" in The useof the deictic,then, is maximallypresupposing, that the configuration in required someappropriate are conditions contextual rvith a deictictoken.f'h3eel"eld-P1!!9Ig for properindexicalreference pf 'sut-in' spcllcd is Somcaspcct tlre cqtrtcxt 91otl t!S-lififj-e-r-qsirniJ-cr. gl 94 iu-pldgl lo-r-fte19felcd4'qgn-upp .tfi ig,9f pie is fi*e.l and pres itself is oncemore reference And t iu"tior tq be m-_ade. in this sense, basis the of seento be an act of creation, Changing contextual-ilie for preRecaiiihai one of ihe waysitr w'irich events. further speecir have referredto the is of supposition the deictic can be satisfied to in question. entity
Indexical Creativity But tlrere is a general qeative or petformafive aspectto the use of pure inclcxicaltokcns of ccrtain kincls,which cdn be said not so mucl't cxplicit and overt io cl',"nge tire context, as to r:qakc By stqUg-$lg i]19gngQltl-glcnJs. thc vcry usc of arr i'clcxical token, -of value from the rules of use setting up the ,"hi;h d"ti*; its indexical indexicaltypes,rve have brought into sharp cognitivereiief part of the -speech. signalis of the In somecases, occurrence the speech contextof pcrhaps,by vcrifiable, t[p onl]r ovcrt sign Of t[e colrtcxtlal p4rameter, thc in bcfiavi<lrs <lt|cr tttcclil, but ncvcrthclcss ttrost Other,cooccurrirtg the in' salient inclex of the specificvalue.Under these circumstances, work, seemingto dexical token in tpr.ch performs its greatestapparent is be the very medium through which the relevant aspectof the context indexicai Pronounsiiwc ar'{' rnade to "exist." certainly, the Engiish creativefunction in bounding off 7ou (vs. he/she/it/ttrey) ierform this such as ih. p"rronne of the speJch event itself; in those languages, 'prl.i-^1..ai'rat
UhmooK (UolumDla
h.

--/

r....^i.]*..-

sigrrals. spccch of thc a nafur-c , Tlle rcfcrcntialvs.tlonrcf-c,rcntia-l -Qf'indcxcs,nlcaslrc modeof com' / i"f;o.ra.;." oi ind.*.s fronl the se'rantico-rc[c-rcntial T'hepresup' gf tynes' 1 ,,lonication. oneaxisof classificatinn, indexical is
of a nature of inclexes, measure the independence lrosiiionalvs. creative rnccliumand nrodc in spccch uf i,,.1.*", front cvcry othcr sigr:aliug tlrc of cvcrrts,is an<ttilcrlxis oF clirssificrrtirxr; itrtkxicri trrkctrs.llee.ulsc' cxist. Thc spcaker'refcrcnt interact,borclerlinecascs two classifications for vocabularyof Javanese, example,seemsto be used refer' cleference uPon the presupposi' in a way that depends entially or.nonreferentialiy given token in context' This tional or performative nature of the ------- -r----

r\lverr

ir--rl,

t\ortrl

Allrsrruail

A -^i^^

--.irl.

\ryrL'

ruwruor

anrl

evenmore the clusive'pronominal indexes, boundaryfunction becomes

finely diawn. Soc'{- tndgl.t-lgQ .tlgJ&t.!99-Y-o!?-bub49y:f{ fg:gb*qv.g, examples--ol-sa.xllualh-qrati-Ys.-e-L are sligg*ligls..g!$!on-J-{ m"?t,q g-oqlei *hi9!, by llrSil vgg -u!e! Pe1glle:devices, ttry perlormative
l---e---:-r----: r-^^-^-"^....t:^ir A,ll.aranap [ers oI $PEaKtrr aIl(l llsartr s^PrrvrL. rrurrvrv^rvv ln fhe nrrryne -r.,:.:.:.:,--:-----l. cneeified bv -:: :.',-':..::::::::

so95r] peiceived iilo o-flS.- reinfoiieC-Thej1e!ti9"l q! 1petkgla!-d


t't

the discourse reference, actuaj unfoldingreferentiaispeechevent,is propositional (semantic) oncemoreseen be riistinctfrom abstract to of type,characteristic The in reference, implemented discourse. former to with rulesof use,responds such and described referential indexes rvhile the latter, indexical propertia as presupposition/performance, does analysis, not. based semantico-grammaticsl on by arrayis thus generated thesefunctiorul charac' A kind of four-celf and discussed, in ierisiics indexes, rvhichwe g:ln placeihe examples of provide furtherexamples: for 5>

NlICHAEL SILVERSTEIN

Jllllrcri,

IJIILBUISLLUvcttugiv'l rldt v''v

v*'v*'

-' -

-''

PresuPP0stng
teuse locativcdeictics,

( creativePerformative)
pcrson second Pronominals

-values' have nottreferential o.td to descrlp-tion, on thc other hand, they that structurc the tactors of tlre speechsituation' Ifgj-$ljtSled'Sai

referential

-rvl1i!-c-titg'!1991{ sf ;So:.f-f?Orffii*tU.*'cpri'iti'ui't yoclc q1ec9l1' cqualig ineeltlali$6 sr ;;;o!Gjqr,_"f tf* i.iii-,node o-f-'rarking scgmcnttirc

firstperson lronominals

nonreferential

of indexes sPeaker' defdrence

we nright of By analysis the ,urf".. t"itgoii"t oi ryceclt' risiduals,and then attempt to as pronominals semantico.grarimatical as of meaning the forms'But inasmuch trvo modes pragmatic ipecifytl're more considerably it category, would take areunitedherein onesurface on two bascd use ruies'of areinvolved' that two clistinct to analysis see distinctfunctionsofthefornrs.Atafunctionallevel,then,thereare indexi' surface by whichhappento be represented the-same two indexes

a1i O:y::-lf: not' ..i*eoty, on"" ,hjfitr, one Thisfunctional sHrflngPorrrlrur Lrrs be mustalways tne of the two modes

descrip' contributeto propositional indexes Referential,presupposing of value some pointthe but tion in discourse, onli by trking asa starting in of tinrercfercnce tense u"rinft., as for il,c corirp'tatio' ;;;*l the rcflcctin spcech inclcxcs prcsupposiug Nonrcicrcntiai suchas the "r"g;ri.t. of sonre ualoei-ofco'textualvariables, specific existcnce texicalitcms.Referential, ,,1,1i.n.. in motrrer-i'-law Jtnlif il;;;; description to coDtribute propositional inclexes Lr*tiu"ry pcrfornrativc
ln

distinctness and must rest ultiin isolationof the pragmaticcategories language' functionof utterances' of ,*,.ty on a se'siiive-analysisihe speech-event anthropological' social a taskwhichis essentially
l\ef are ntial ltnalo gv it t D iscoursc Tlresituationisevennroreirrtercstirrgirrtlrccaseofpragrlaticmeta. rcfcrring to thc hearer phor, .onn.ctecl with pronourinal shifters of found in many languagcs'.Instead (rp(*,r,r,...)), a pllenomcnon social rclations, dlistinetforms inclexingihc quality of speaker'hearer skcrving of otlrcrrviscscthc "sccortdpcrsolt" ixonuutt incorporatc data' r95o; 1956)' (see Beuveniste 'exteniion To analyzc,the.sc mantic categories of rejerential or analogy we have to iistinguish tws kinds of in categories discourse. categoriesof The so'calledpronouns frequently seem to incolporate 'nu'mber',so thai we iend to speakol'lt* and second 'person'an.l ,,Third pcrson" proprononrinalfornrs. i.rron singularand plural" for tirat obvirrtcthe necd can r)ourls bc true sn5stitutes,iupinric devices fnr renplition of a f,Jl, lexicallycomplexreferring noun phrase (thus' rwl/vrrr.v^ rvr as negative Theman sat down.He . . ' ) . In ihe referentialmode, they act
--r--- ^- L^^,^- .-^-*inr'nnnlc i. cnpech events|ndexesrnneverrnoexlngsPeaKtrluIlI9i1lgIyallrvrre..w

ilrc rvlticlr assign as sp.e.h.eucntfeaturcs, irr tlrc cltoicc of 1:ronornitlals, and referent to certaiu indi'1 hearer,aucliencc, ,otes of speaker, e'uerrt nonreferentiat rel+tively pe| uiau"t, in the maximd} case. FinallR formativeindexessemeasindepenrJentspeechsignalsestablishingth{ as in deferenceforms' whicf{ prrn*","tt of the interaction themselves, ineffectestab]islrovertlytliesocialrelationsoftheindividua]sinthd or auclience, speakerand roles of speakcr aDclhlarer, spcakcr and referent. e F unctional Aggr gation in Indexical Forms TheTlraiexamplecitedabovcinwlrichsocialdeferenceindexesare po11! up.the^fact that eve!united with prono4irylr.-eJerqnliellldgle! On tlre onc illSlcxicr! .rrcguii.t tttiiitt pilgtluiigt'liy---ryulqifurrctional' valuesthat contribute proiominals trauc,liscouist-referential ir*f,,fr"

:- (-!j^-.-...-..+t l r r i,r.r'lrlitiort fttnction as thc sigrul for the existenceof ^,'.I r r l d llscotrls(;r

the inclexes, contriare pJrrtn" fornrs refcrential But ,'firstanclsecond functionally of reference which comesaboutby bution to cliscourse hlvc tlo arraPltoric fontrs 1:ropcrtics' rulcsof usclsuch distinct

36

"t

MICHAEL

SILVERSTEIN

and Categories, CulturalD escription Linguistic Shifters,


a pronoun, either replaces third singular anaphoric,or nonpersonal or referSto and indexesa singular addressce feminine semanticnoun of rvhile indexing deference speakerto hearer. some languages,such as Germanand Worora (Northirn Kirnberley,WesternAustralia), srvitch this defcrence, both person(r- 3) and nnmber (rg.- pl.) to exPrcsl 'clual' morc high)y rnarkcd numbcr catcgory, a In thoselanguagcs-with 'plural', d.l"t.n.* is indicatcdalong the axisof tlttrnbcrby srvitch than index. This occurs,for example,in Yokuts to duil-number adclressee of of south centralCalifornia (Nervrran 1944) and Nyangurfrata north' ern west Australia (o'Grady tg6+).(curiousiv, in both these ianspccificd' must bc accordcdto a gcncalogically the rleference guages, in the first, mother's brother and l.r*n", mother-in-larvand equivalent equivalentin the second.) is What unites all of theseseeminglyisofunctionalusages the uni' "second person Plonoun" directionality, in every case,of switch fron
r^ .'+},.i-il " lrrrlur !v f.nlav1ar

When we use a "third person" pronoun,the singularor plural num'number' ber is derived by the rules of anaphora from the semantic mode, of In specification the noun phraseit replaces. this pure semantic 'number' signalsmore-than-one rvhatever of entity is referredto plural of Bnt somc occttrrences by the lexical stemsof the noun in question. only irt thc lcvcl of discoursc plural numbcr catcgorydcrivc scconclarily established scrrrantically by referencc a kind of sumrningup of inrliviclual andlillwentup ahill.They. . . ).It is at entities(ihus,English, /aclc this secondlevel of refercnce that the first and second person pronominal inriexesget iheir apparent "singuiar" and "piurai" iorms. For 'I'; 'we' +'I' the form rveis an inclexthat refersto anclpresup+ English speech and at leastone other individualin the referential poses speaker a pluralrefers G8,t,1,... ). Similarly, person second event,sp(x,y,x&w,we, Rushearerand at least one other persona, example, for to and indexes sian sp(x,y,y&v,vy,Gn,t,l,...).Oniy by the summation of the individual referentsin discourse.rvhich are referentiallyindexedby such 'plnral'rCfCrential valirC emeige. pionominais,doci theii tingulii' or With tiris analysisof the distinction bctwccn anaphoric ancl uonanaphoricprouonrinalindcxcs ("pcrsou"), aucl of scmlnticllly birsccl (summccl "nuurbcr") rcfcr(cardinal "uumbcr") vs. cliscoursc$ased in encc to cluantity,it is possiblcto scc the nrture of thc skcrvings so"ironorificsccortd pcrson1)rorrouus." callccl

av-anlcrl vilt,vv::vs

ttcinorrlrr
"l::b:l:'

rrrrnhertt ':-':'-::

to
::

ttnltrraltt
rl --

(or
\

ttdualtt).

!'

of Metaphors Cramnntical Categories Pragmatic


Russian for cxample,or French, wc can index For somelanguages, deference that is indicated eisewhere the samekind of speaker-hearer by vocabularysrvitch (see Friedrich 1966;Ervin-Tripp r97r), when by a addressing singleaddressee, using the "secondpersonplural" (vy, ,
. yousJ
.\

or both concomitantly.T'hcrc is a kincl of mctlrphor bascd on tlte it valuc cliscoursc-rcfcrcntial of tlrc catcgorics, would sccm. Itr--tltc onc rvhcrc an casc,it is shift out of tlrc rcalur of scconcl Pcrsolluclclrcss, situation flcc'to-face,to thc rcalm in indiviclualis inclcxecl thc spcech ns cntity is un<lcrstood thc estrrblishccl of ariaphora,rvhercan alreacly 'l'his largcr tlrln lifc by rulkcs thc addrcsscc rcfcrcritof thc srrbstitutc, impliecl lry lhe face-toJacc ad. personhooel lrkirtg away tlre individual it case, is a shift out of the realnr of the singular, clress. the second In to, where an inclividualis refcrreci and inio thc rcalm of nonsingular, rvhere, as it were, the surnmed nurnber of individuals referred to is count for more than one greater than one. This makes the aciclressee socialindividual; to his pcrsonaaccruesthc socialrveight of many, as 'lve"' .Joesthe inverse, we compared with the speaker-(The "royal shouidnote.) exampiesof isofunciionai inriexing rviih in such universals expression, are r.eiling metaphoricalplays upon semantic categories, an important demonstrationof the inteqplaybetrveenthe semanticmode of language through speech.The a-n{ the ffagrnatic coistitution of social categories even rvith "fudges" to permit slmantically basedanalysisof categories, incolporation of the anaiyuc residueoi shifters,rioesnoi capture ihese generalizations. The perspectivemust be one that frees us from de'

ralngr

rl

^-

rL--

Lrlalr Lus

rL^

aa^:----l--r,

Surgural

\L/, Lu), lll ullrsl

/r--

a--\'T-

^!L^-

---^-l^

!L^ ^^-^-!:^ l \Y(,lus, LrrE rsurarrlru r/

discourse eitherrefersto I use, numbercategoryin its summed cardinal indexing r, or addressee, concomitantly j to pluraladdresseesrefers a single of speaker hearer. somelanguages, example' for to In i the deference the indexuses what is otherwise second person Italian,the deferential
rr.r. -\rr --^---!--l l^-t^-^-L:-^J -^l^-^-t:^l 'tnrrol , r , t - - t t ^ - t - t ' slngular (rcmtnlnc/ ..rErjEr-sllua-t Pronomrnal rorm rur s:ulll,1?}ltrq

In iq4.lilg and-gocialindeaiSrg. other words,third singularfeminine

3$

39

MICHAEL SILVERSTEiN

-.,. tfters, Linguistic Categoies, Cultural esuiption and D


t !

on or pendence semanticcatcgorics, evcn refcrentialones,as the definof rcquiring analysis. ing segmentation speech gues Functiornl and F ormdl Analo I Irr cachof thcsc cascs, hlvc bcen claiming a kinclof functional cornp:rrallilify of thc 1>rrrrrllcl fornratious. Rulcs of usc arc nonns bctwccn 'fo and somc fonnal fcatureof thc message. be thc contcxtualvariablcs 'functiondlly dnalogous,then, indexes must be describedby rules of use that specify analogouscontexts under analogous speech events. (Obviously,the utterancefractionsin differentlanguages hardly be can
a-^aalaj ln l.a alil.a \ \I/l---;"1^..^" wr eaa'rvvtrr ln rv L^ uv -^^^.-'.1;ol.i--.-^ln, quwurrrrJrrurb

referential frameworksin the semanticmode, linked as hypothcsesthat serveto justify a particularanalysis. F ormal Distinctions SignalingFunctional O nes of The parallclfonnal-functional analogy all thc pragmaticmctaphors is for spcakcrJrcarcr clefcrcncc au cxccptional casc. Ordirlarily,givcnsonrc grounds and indcxedin somc social parameters constitutcdou nonspcech language,rve might want to ask rvhether or not these are indexed in of American someotherlanguage and, if so,ho!v.The sexindexes several mentionedabove somewhat are diverse functionally, but Indianlanguages at tbrmaiiy overiapin ai:parcnt pironoiogicaiciranges tire encisof aii The Thai pronominal systcmfor words,in particularof inflectedverbs. first and seconcl pcrsons-inclcpcnclcnt rvords that indcx in complexways thc various incqualitics rcsultiug in dcfcrencc-incluclcsanrong thc of of scx. Amcrigrourrcls rlcfcrcncc distinction sociologicnl So thc scvcrirl can Irrcliun systcnls sccurto isol:rtctlrc socill vnrinblcof scx,indcxingit (T'hc ethnographicrccorclcshbrvith a unique fornral set of changcs. lishcsthe grcrt salicncc thc distiuction,at lcast in thc societics of spcaking Muskogeanlanguagcs,but its cultural position has not bcen Asian) systcms assirnilatc established.) The Thai (and othcr Southeast of thc socialvarilblc of scx to tltc functionalcharactcrizatiorr incquality nlore generaily, making a pointed ethnographicstatcrnent on equivaas lencesof stratification.It is alwaysnecessary, this exampledemonstrates, to take the functional perspectivein tsms ol rules of use to be items fit into systematic socioable to seein what way such pragnratic logicalpatterns,of which linguistic onesare a maior part. F ornul Characterizationof Indexes From the formal point of vierv,the sign vehiclesthat function in an indcxicaluroclc cxtrcmclyvariccl. wc notcclrvhcrrdealirrg witl-rthc are As
.t-:.'^^''C'''.^ri^'.^rr^.-'.^.'^.'''il^1.^f^^'.^.r:^^-^f^.^''li^l
[Jtlv.r!) ft]rlUlI(rIl Ut lalrrEtl.ISV, J!YrLLlr vl J!lIl<lrrllW_l!lWlUrr!r.ll 6rr{rlrrrrdf

gous socially constituted tasls, we can speak of cross-linguisticcomparability.So the claim of functional analogyfrom a heuristicpoint of vicrv mrkcs hypothcscsabout thc social parameters s1>eech of evcnts. Iirorn a thcorcticalpoirrt of viclv it clcpcnds upon thc rcsultsof social :tutlrrollologyfor a friunovorkof clcscription socill cirtcgorics, thc of for stnrctural signifioruccof tirc pattcnr of iuclcxical spcechnomls iu the givcu socicty.Urrivcrsals functional significatiouthus arc thc trcccsof sary meansfor creatinga rcal scienceof language pragmatics-that is, for cstablishingthe ctlu:ography speech-just as witliin the semantic of modc, univcrsalhy1>othcscs aboutphoncticsancircfcrcncc thc ncccs. arc saryempiriealeorrela of scnraniieo-grammaiiCai tes auaiysis. indexicals dependupon crossOn the other hand, formallyanalogous of systemiclpecification of equiya]ence meslqgefractionq,In the carsg of the pragmatic metaphorsillustrated above,there is formal comparability of in the expression deference through the pronominal categories themselves, which can be isolatedin the referential mode in everyone of the in languages question. (Note horv the formal analysis one mode dein
-enrlc nn icnfrrnnlinnrlitrr in nlhere
,

re rrr4e nrenlinnprl
gu ,,ge

ohmrp

T$p

lon-

guages use formally similar categorialsubstitutions,definable in the all urodc,to indcx dcfcrcntialacldrcss. From a formal scmartico-rcfcrcntiirl as Doint of vicrv. thcu, wc sccksolnewav of charactcrizing structurallv fractions inclcxcs. But auy such strucanllogousthc rncssrrgc scrvinga.s tlcllcrtcls upon lnllysis of fonns, rvhichitsclf rcstsou tural spccifica,tiou basisfor comparison. findingsonreisofirnctior:al Ultimately,then, crosslinguistic fonlal analogyand fuuctional analogyare,likc phonetic and

carritsclfscrvc an inclcx. as Wc havcsccnvocabulary, phonological affixcs, rnlcs,:rncl prouoruirutl catcgorics scrviuginclcxical fturctioirs syucrctistic. rvithin uttcrances. Indcxical dcviccssuch as anaphoric pronoul'ts, melltionedabove, rcfercnce certainsurfrcestruc. in rvhichrnaintaindiscourse

4o

4r

I\TICIIAEL

SILVBRSTEIN

and Slifters, Linguistic Categories, Cultural D escription by constitutecl rulcsof uscat t]rclcvcl of discoursc of aspect meaning, ritherfunctional sening functions, ,.Lr"n.", and (3i the pureinrlexical for of independent reference, w'hichno'rcferentialrules of use modes behveen(r) and havea point of overlap (z) w-e In areconstituted. "pragmatic ir"t.,r.. B't a formal descriptivc l3);hence thcir cluplcx t:rode, spccializ.ecl into ln valid sernantics, asa gi;io*or" mustintegratc systctn. inclusivc
PRAGMATICS AND CULTURAL DESCRIPTION behavior' I have analytically separatedfunctional modes of speech in linguistic signs' furthershowingthe modesof meaningso constitutcd briefly the integration of these rnodes in a I want now to characterize of pragmatics languagi,indicatinghow this purports to be a systematic *or. ,a"guatJ descriptiveparadigm for speechanti othcr communica' rela' tive behavior. This will lead naturaily to a considerationof the or iionship of such pragmatic description to broader ethnographic "cultural" description. in Futtctionol Alternatives ltlrles of Use indcxesshorv Rulesof use for both shiftcrsanclother, uonreferential utterthc existeotialrcl4tioqshipof contextualvariablcsto somg ovcrt specifythe refcrent (z) as ancefraction. fi,e tJ.s oi ur. iot s1'tifiets to contributC_ rcfcr' well, consonantrvith the fact that such catcgorics the third ential spcechevcnts.It rvoulclsccm that formall,v, _variablc rules of useshouldbe the func- i to (correspondlng z) in nonreferential

tural configurations,are formally defined only over at least two noun sentences. phrases,frequently found in distinct semantico-referential are further kinds of featuresthat are Intonation pattems and stressshifts characteristicallyindexes,though I have not dealt rvith them directly by such as the distinction bctrveen example.And syntacticconstructions, or "passive"forms of utterances, thc English "1>etforma' "active" and tive" constructionI [V] 7ou [X], are virtually ahvayspragmaticunits, formally isolableon functional grounds.In other words, the range of possibleformal elements that can serveas speechindexes,accordingto understanding,inciudes the enour h'aditional semantico-grammatical level, sentence indexing, discourse-sequence tire range of languageJevei affix level, and phonological alternations that can be level, word and characterizedby rules, including intonation and other nonsegmental gradient devices. The description of all these occurring pragrnatic formal features of a presents vastproblem for our traditional ideasof what a grammar speech grammar,it (G) is. From the point of vierv of a semantico-referential would appearthat everypragmatic index is a kind of "shuctura! id-igA,ll where the constructionscannot be analyzedaccording to semantico. rnakcby far the grcatcr regularities, J'his rvoulcl rcfcrcntial cornbinatory of bulk of a clcscription spccchinto a list of sucir "icliotns."Tirc undegiventhe kinds of rcgularities of suchan alternativeis manifest, sirability above. of pragmaticfunction excmplifiecl So some attempt to patch up haditional grammarcannot serveas a of principled descriptionof the pragmatics languagc-a fact that nost For linguistic theoristshavc not yet aPPrcciatccl. tlte charcontemporary analy' acterizationof pragmaticsas dependenton semantico-referential riiscusseci above-becomes ioiaiiy sis-the "performative" approach hopelessoncewe considerthat only a portion of the indexicalsin speecb are shifters, with connection to the semanticallybasedgrammer in the
I I -1.,,

spgecn

runcuoll

(Jr lclcrglrutr.

-c

-^t^,-

--

.rL-

Illtr

-^--:-J^-

lculalrlLtsl

vl

^f

!L^:..J^.'-.

LIls luusaeD

^-o.i'.cl

4rv

f"na-

,uJL

ruuv

of tionallyindependent reference such. as as speech the one then, becomes of how to represent The question, it is, while at the sametime apparentlycontinuousformal medium the preserving pragmaticdistinction among (r ) the pure referential
tuncuon
r r.-

rank;somespatialcon-'' suchaskinship,sex, domain, sociological some in I figuration, of configuration persons ihe speech , ,,i.h the"proxemic" ", i e.J.eni; so forih. Intther words, only referential speech events, not and I I the Uur other typesaswell haverulesof usethat specify functional "it ou., *hi.h the particular is modeof meaning being pragmatic domain ' ) where '-' rulesof usefor sp(x,y,z', realized. not onlycloivespccify So .,referent," alsosp(x,y,2i,...) n'herezr is a variable func' of bui z. is
ri^-^
LlullJ

bonally

' dgtermmec

r t , ! grr ^ - r : L - - - . 1 - : ^ L : . L o i - inrlevprl Ellrlty wururr rr uvrrr6 'rvv'rvst Klng

fnr

evcmnlC

; i

or sellrautrc

--!:^

asPegL uI lrrsarrruE,, llvur

-----r

^f

-^--:--

f-^-

...L:^L

wrlrerr Jwur4rruw-rw^wae.r

...i.-nfiaa.afotao-

-^-o
ltlvls

---^-^11.,
blrrvrqu/r

Jv uGr . a l o s ^ vu

h ' , l h e t-o tn-c r e o f- sf n e e c-h e v e n t s . - - ' " r

and derivetheir analyzability, on which one facetof refertial systems or (z) acts entialspeech rests; the shifterfunction, indexical-referential 42

for the underru.h generalization, "rulesof use"wespecify shifters, .,duplex"" into mustbe furtheranalyzed what are indeed catigories, the

47

MICHAEL SrLV!,.

:ErN

Onesuch.functiqnal' ryode' iefer' two elementary-functiona-l-mgdqs' mode.is non'-" another functionai z' specihed; ."iirf,';tft-tioti"ui" suchas zt "tegpgla] zt-to be specified, with somevariable referential, so forth' In shifters'an ele' and sn Darameter," "privacy'boundai1t," func' indexical fu'ction ancla disti'ct elementary Ir;"a* ,tr"r.nti"t if rveexamine but category' speech surface ;;i;"t i' thc sarne ;;;; of_a z'is frequently difierent .-rT reierent that thc rve 1 "' themcarefully, cansee the as zr. """'' PresuPPose referent indexed Deictics, we saw, Jo,orin from'ihe . " ' i: I ot as for exarnple, well as the sgeaker hearer i*t pr"uiousdiscourse, referentrelativeto location,and refer to the locusof the presupposed or'freirer' thatof sPeaker
'l'\rc Cotrstittiion of SpecchActs Wecancalleachorreoftlreseelcmcntaryfunctionallyspccificrules mean' \Are can notc that such norns for pragmatic of usea speech,act. of speechat the trg tat,il"t .l.p.nd upon the functional specification

' above.we can speak of the "referential functionz" already discussed to achieving of of actual signsin the sense the contribution they make similarly, we can speakof of a valid inst-ance function, of describing. ,.socially seriating function2" of actual signs in the sense of the . the hierarchies rvithin contribution they maie to the function, of def ing Spcechis multifunctionaJ,jn-,tlre sense-.!!1!-it-can" social categories. simul-tanco-uslybeusedtogo*1$tq!cclistinctkindsofevents.Speechis of- surface-form. nrultifunctionrl, in 1|'g s.nrJ ttrat aBg-arent-elernentr This ,.u.rrt*_4iqthct in-dexical_tyBes. -of. ,.iu^uu incorporate_ryeanings grammatical function, an in' a;""6i n6tft *i iuditionai notions of, terms of the contri' stanceof function3, alwaysultimately specifiedin system' bution of elementsto the semantico-referential Theanalysisofspeeclractsisthusageneraiizationoftheanalysisof and lan' for proviclirrg mcaning rclatio-ns scnuutico-rcfcrcntillsystcrns, analyzcdsort' In a guagcuscsclistinct from those of thc traclitioDally structttre of rvhich the preanalogy,it is thc more gcneral ,loaihcroatical particularly' viously explored tyPe'tums out to bc a specialcasc' More signs functionl exclusively in the speechacts foi semantico-referential an<lsuch aside-and exiensions evenis-abbreviaiory referentiaispeech the gram' ,r. ur.oo,ti spccifieil,with the exception of presupposing mar G. ncis The"Grammar" of SPeech permits us to seeat leastthe nature of a more such a characterization "inclusivekind of "grammiir," which includes the traditional ssrt as a is a finite' component.if grammarG, as in our presentunderstanding' replesentations rccursiveset oirules which reiate semantico-referential recall to utterance types (or "sentences" in "surface form"), rve can from the first part oi this paper that the meaningsrvithin G are defined are and the sentences segin terms of the function, of pure reference, pure referentia!sentences, on meniabieinio constituents ihis basis.In so functioningl form a continuoussequence. tlre iurface elements

tPuon Shifters, Linguistic Categories,dttd I'ulturdL D escr

i , ,. --^^^r- ^-.a..!o n. iF ic o! thic hioher leve1 of analvsis that one level or sPcEull Evsrru, rvr rr 'u.-s!

".tasks" constituted various canrecognizc ixog*atic .odcs, Lhesocially R9-fq1e,'ce or ts-g,1e.uf,i.f, rp'...h behavioia-ccornplishes "performs'" acts are of trvo kinds' and 'eierentialspeech ;;;i p;;g*atic mode, the the naiure of the referringutterancefraction.on ,"ni.t' *t' "".-"pl.inshifters motivate elementary referential on, iond, the Tltt. :f sPeech:ev.qnt-variable(s)' to rvith lespect some .*,here is referent specified qf ootivate-r-uleg . ,e_gi!!es other hani, the semane|*eferential OO1Uu the-sr-1[which mcely specify'teriablaz'and prest-opos9 ,(i!d;x-) use by th_g_sgma,ntico-_ 1i valueis de_tgrmllgd ir,ii,_i refcrential ;;r'6$;i referentialsPee.ch enliilg-{ - in ; - f 9t1gtionqily g;r"r*tri*f lules itlpf modes u6inettq;irqd' Othgr pragmatic !u.ng no futth; specification oi rvhich'as we haveseen' many acts, i.n* ait,itct t in& of speech For indexes. examsurface multifunctional the same, ;;;;i;p; precisely
--ur sonal referent, as rvell as a soclal-rl-ct-ql ,us1:er.'-Y:' use for cleference otherwisereferential categories pog.rti. metaphors equality/inequality' in i',ultiple functions expressing *="1-oii" meanins or function2of spcechiigns' Prd6rrralrv "':"';".
tl're framervork . - . .- r,.-^L:^-- ^r .^^;tt', of purposlve TutLvLLvt'1vr Jvvrsrr

pr.n"i'pronoung".represent.in:i:?i?'-:,r{^9T:"u.fi}iilll;1i,t"llildexicli f?'*g"Iff tor-sP+iffncthe actsarethe elementar)' .Itrey


operatewithin
^nnctitrrted he-hnvior

i;;;;i

Toconstruc|agrammar(G,)ofspeechdcts'theanaiogousgener. alization is a set of rules which relate pragmatic meanings-functions2 specifiedby functionallyr indexed variables-to the "surface form" of - utterances. Utterances,r., tue have severaltimes remarked, continuOus
:- -^L---lll llatulg. m-^ -^^!L..ll, I lltr Ergirl uulN ^f vr o,,nl' Jswrr r rrt!lv 'rs ^n a p c s e c r r r o r e o v e- - t 'r o i V e- t-h e oaOOeafI I

15
41

MICIIAEL

SILVERSTEIN

Shifters, Linguistic Cate goies, and Cultutal D escriptiott Multifunctionality and Pt agtndticSfi ategy There is a structureto a pragmaticglammar so constitttted,the de' tails of which are now ohly partially clear. speech acts are ordered, for relatiorx among funcexample,a reflection of pragmatic markedness tionai, meaningsof utterance fractions. For exanrPle,therc is a hierarchical relation among all the kinds of sociologicalvadables leading rvhich can be formally dcscribedby intrinsic ordcr' inclexes, to cleference ing of the speechacts characterizingtheir usc (see Ervin'Tripp t97t
I IOf d-t -.r -r-- -^r--.:--!:^-\ gllaraclgllzitLI9lr llgw'cnarf

ance of formal integration in terms of phrase,lvord, and affix structure, the for espeCially referentialsegments, shifters and semanticelements' rules (G) Tiis would seemto indicati that the traditional grammatical into pragmatics(G'), that is, that at least some. must be incorporated of speechacts consist of rules showing the contextual dependence tra-, surfaceforms. ditional grammatical rules for generating by hvo pragmatic examplesI have already This is further confirmed which langnage, mentioned.One is the srvitchof semantico-referential ern seweas an index. I{ere the whole set of rules of the traditional sort
:^ l) ^ 4 r..- ^L:^lulrwLrvtl ^C . i-'tro.'o.-v^-uruv^ rl.a ornmmolinol nnmnaienna' nf snenker-

hearer, anci audience. so obviouslyour pragmatic descriptionshould show the selection of rules GL, not just an infinite set of messages is case the Yana The second {0L,}, as a function of contextualvariables. mdle vs. female indexing. Flere phonological rules, which shorv the ' of regularities shapein pairs of forms for the maiority of rvordsin the language, characterize the context'sensitiveindexing, rather than any affixation or other segrnentalmaterial. We u,ould want to characterize of the indexing here as the depcndence the implementationof certain sex. sociological of upon the variablcs spcaker-hcarcr phonologicalrules gcogmplriof Any phonologicrl inclcxcs this sort, such ns tlroscrutrkirrg
r r, r , . of cal Qldlect t1 tnc - -.- -l ,-Sl)cilKcr, --, -l -^^ --mI:--!-.-t uf ct:t55-itulilat(.(r ^..-.'-.r.. ..-..^! 1.,, .:.'-:1.'-l', l'(rr rvL)r, rrrLrrL rr! rrrrrrr.rrr/

trcatcci. of So a grammar of spcechactsG' consists rulesof usethat map the \Mlth tliia vaiiablci of speechevents into rules generatingutterances. rvc have moved frorn thc hcuristic clcviceof direCtiy characterization, to relating contextualvariables "surfacc" uttcranccfractions-detailing, a irr othcr worcls,thc dcfinitiou of an irrdcx-to constrtrcting p-ragmatlc
-COtfr..,rh::: J/JLLrrl !!.^tl trtcr r'.:,rlrirre v^lrrurrrJ t!ra Lrrv rn!::finrr ."...':-r: !r{rlr-r-'.tl!r nr-'rt:trcttt strrtctrrfal

so fractions, that features function,of utterance tionsof speech-event somefunctional, mode,lessapcontributenormallyto of utterances mentioned metaPhors Pragmatic to thoughpossibly, others. propriateiy, beingcxtended, categorics semantic in case point,basically .bou. "t.-.filling out "holes" in the pragmatic structure. asit were, meansthat there is a , utterances of The muitifunctionaiity apparent and takenout of context, f of indeterminacy utterances kind sf pragmatic in of language thc contextof speech uscs for the posiiUitlty strategic I vq rcLcen-tialI te cven Qqt -qlegntgLLwg- rqg onlv*!ta J*c-qaurrg---ft-cts. w.'!r:S-'lli't I "'oig tl-Slgjglltbirt&lltg-llul-llilg-ot l1!!,"-tr.LgSr"..tiricll itioriri.!\r,i'i{ \ trcrs,;;Jcir rifiiy-"coul"iiii".." /rcici ior crccl>tions siri tjyg ively crc,l or pcrforma-iriclcxci tc re.laf iri tqxg .a csp,c-cially cgn -WhcretYLf 1atl gm t*e, tk{g g-!-b g-pr4 ati=q rglgt=ggnu!y !1,,u-ttcr4nces This leadsto such rnanipqlqtdfu d-e.!9d!vidual"s,.an-inJerastiqn. in adiclion; o1 "dr$lg!gd-'. be!a$-o1s, as phenomena . pi a{matic contr . contra' pedia q9-mnrunicatiYQ signSling o_I_J!_o_r-e wlriehplay upon,1ry9 ot_{tc cot19q51911t .{iet-ory {rdexic;ilmsanings !!ry-{g,c_c!v-e,r !o - - - ^ ^ - - - - . - ^ ^ : . - - : . - - l ^ . , ^ . . . ' i l l ' ; - lqq$-$g-e-q' r r'-r -t-:-l-I-lra coma
or uPgn cont_raqlctury lllgllly wrrrr'r' PlssuPPv)rtl5 .r_f.{!l::\_Y:

A-l / | .clllu

f..-!L61L--o ^.n lur lrrw^r Lrrvrw q'v

*rrLe.jlnacc rrrq^'\vs'

rpln-

nature. any given Tor fornrand its rnultifunctionalr,z tinuity of surfacc aets theremaybe many.sPeech whichmotivateits fraction, utterance
raracpnnp i. o cneech pvent- r -' -_ihrt- _-_ is, -nnv -"J trfterance fraction ' mavr bg a

the senanticity, semantico'Iefelenrnedium. thereis residual Similarly, can tial meaning which a speaker claim after the fact for potentiaily
L:-LI-. ru6,llly

or modes, a constituentin in constituentof pragmaticstructures several in and not a constituentin others. Reference, particular, some mode ideasabout utter' motivated all our the functionl which has heretofore ance constituenclr motivates only one such Pragmaticstructure, at the
^n'. nf ..,lrinh arc eccenlirllv rrrleq of rrse se,lectinp G- The shifters reouire

to denialof speech ofiensive the hearertakesthe form gf "All I said or paraphrase repetitionof the was. . . " with a semantico-referential a can A referential of the originalutterance. speaker create so' content of im' cial persona himself, playingupon the hearer's for Perspective
a.,t^,| ysLvw :-)^-.'--ti&. LaLw^LvsLLLf, ..,L^*^ lYlrw^v 1L^ rlr! .^-^L-' oPL4^!r 1,.o ^1'.'onlprictinc ottrihrrted tn

--^---!:^^ll-, Pra6ulalllLdrlJ

^t^-^^A err4rSeq

.--^^L Jywu.r.

1r' 1 ' , , c rrsu

*ra !.rv

nharantericfin

<nellrerts -r------

rulesof G', thoughthey function,alsoin referen" distinctfunctional2 tial events. 46

him on the basisof the rules of usefor certain utterance fractions.Thus the phenomenonunderlying the plot of My Fait Lady' Finally, diplo'

47

I\,IICIIAEL

SILVERSTEiN

andCulturalDescription Categories, Shifters,Linguistic


Limits to M etapragrnatie Avarvness

mdtic nonindexicdlity,on the analogy of diplomatic nonreCognitionin foreign policy, allorvs the hearer to respondto speechas though it con' ,l comevenf all the while understanding stituted a semantico-referential which overlap in surface function2 of the indexes pletely the distinct form. pragmaticcontradictionand imputed indexicality alike in dependare Resid:* high,functional2potential of utterances. ing on the unavoidably in .depending are ual semanticity and diplomatic .Fonindgxical_ity alike whereby the s.emanof-p,gop!e, universalmetasemantlcqwaqen,e,s-s on the or of spgechis t!9 9ffi9ia11y ove*ly recogniieil tico-referentialfunctionl one, the one io which actorsmay reireat with full social approval. (This

8."!!_g-[ry-P!!WsLc* timcs several by Sapir,) made pointwas olP"'.1gT+g$q911ij[8 manipulatiol itratcgics, l+Vll-!r-.ehlu.,.9r''d.c:
of thqlharccl undcrstanding no.rmg.-fo.2 pcrid iifi:tre-Fit analysis,r.rPg-Lr somc indivicltiili arc bctter eiern.ntsin qrccqlihlti, O6viriiiSly, inclcxicai ltave tltan othcrs,iust as sourci:rdiviclttals at thcsepragmatic_stratcgics of a more explicit and accuratcconccptionof the pragmatics their own
l^-'---^-^ r ..,^-,1^. l:ll!6Llil5,u. L wvrrrrwr .,,I'nllrnr drn l.rrn'cl,illc rrn rnlnlnrl

But more importantly, it would appeaxthat the nature of the indexical elements themselves,along form'al.tunctionall dimensions, limits of metapragmaticawareness languageusers.Indexesrvere characterized that is affix, rvord, phrasevs. some other vs. nonsegmental, assegmental feature of utterances;as rg1felgntialjs,n'q4referefiial-fhaljs*slrft-qf-y-Svs. relativ_e1y-_p_fe.g!i_v_g*nonshifter ind-e6 and as re_lqtiv-ely--pr-es]lppg-sjpg o5pedgrmative. It is very easyto obtain accuratepragmatic informa: tion in the form of metapragrnaticreferential speech for segmental, referential, relatively presupposingin'dexes.It is extremely difficult, if nonto not impossible, make a native sptaker aware of nonsegmcntal, rvhichlrave no.metapragrcfcrential,rclativclycreativcformal fcatures, of matic reality for hi,ur.Inclcxes the first type, wlrich arc susccptibleof a.rc, of cottrsc, closcsl in thcir accuratc conscionscluractcriz.ation, to foruml-furrctional,1>ro1>crtics scilantico-rcfcrcntial scgnrcnts, for manipulation is possiblc.Noticc oncc again that rvhich metasemantic
lllcLilsglllilllLl(;
:^

SP(;g(-ll uvgilL)

^-^^^l-

\Jr.;g a{rruvG/

/^^^

-.t{^-..^\

a{l(j LttLr) a JlJUUlar, sqtlatlurl:{r

^-^

!l-.-^

^--^^:^l

^^.-^!:^-^l

Metopragnrutics If strategyrequirespurposivemanipulation of pragmatic rules, then it of may alsorequire an overt conceptualization speecheventsand conqtructure -of of acts.Suchcharacterfzatiqn the pSqgpatiq stitueni speech of as the eharaeterjzation semaltico. -mueh languageis-metapragna_f1;5 these The structureis metasenrantics. distinction betrveen grammatical as two realmsis vast, howcver.Whilc languagc a Purc rcfercntialrnereferential spe.ech in as its own metalanguage metasemantic dium sewes no metapragmaticspeechevents in which use of events,there can be the in speech a given functional mode e:.piicates pragmaticsEuctureof that 'rery. functional urode. !-re -49t4p-ragrnatic.--ebaractcriz,atio-n-oJ
' i l

cvcnt.Thc cxtcnt to rvhich signshevc propcrtics sort of mctapragnratic ollcs,in fact, is akin to thoseof strictly scgnrcrrtal, scinailtico-rcfcrcntial rvhich lve can gct accuratemetaobviously lneasurc the eascwith a of characterizations them from native speakers. of Inyeslgqtio"_ pragmatic of the triply distinct fqJq4-fuggtiglek_slegren-lg ,speech,-o-p._the,qt!g5_ 9f ,-,--LJglserv.stl_ox .a f-unstionall framervo-rk. liq!4_fqggll-c_s_-lg!9_{pJe-!ati-ye*in I tlrinl that everyfieldworkerhas lrad such experiences, whre a careful sortingout of kinclsof pragmaticeffectsultimately just cannot rely on the metapragnratic testimonyof native participants.(That so-called gcnerative semanticists insist on the validity of their orvn f intuitions" aboutpragmatics Cedankenforschungen in simply atteststo the unfortunatenaivetdand narrowness most co-ntemporary of linguists on mat|^.. rrrJ ^6 vr ^^^^^L Jlrvevrl ^L^^*,^ri^vuJurY4Llvrt ^-J crru ^C vr ^,,rra-^tl^ DyJrvrrrdrt! ^-^^,-^l:^ PrdSrrrdrr9 $1,^^*, LtLwty,I \ Tttl rL^. LrrE

.ji,'1"1* al p qh us p !i tc. rc g qc .ni ! qo g !sr a- fcrcnti cy;q! lJl-ylrr"l fo ot#f a raug,uaBu f wurcll Illc )o o,e!-cllPll-o,ll-. oDvrously cxtclrt
the oD,lects _o_t has semanticlexical items which accuratelyrefer to the indexed variables, to the constituents of speech,and to puqposivefunction is one by of measure the limits of metapragmaticdiscussion a speakerof that
l q n o r r no e

course invcstigatingWasco-Wishram (Chinookan), for example, I of attcurpteci systcma with infonnautsthc diuriuntivc-nugrncutativc to tizc consonantisms rvhich are ubiquitous in speechacts of endearment/ repulsion felt by speakers toward referent,without referentialcontribution. They form a pragmatic metaphor on the more "physical" speaket
estii.netinn nf cize relafirrp ln o clonrjlorrl-rhc alaccin crrnaolpanramolia

48

49

, MICHAEL SILVERSTEIN

Shifters, Linguistic C ate gories,lnd Cultur dI D esctiption above). In English, for example,theseitems fit into thc schemal/we prcscut,nonprogrcssive [V] you [X], where the verb V is inflectcd for narne thc socially constituted (punctual) tense-aspect. They actually Jecclr cvent of rvhich thcy fonn an nttcranccfraction: clvistcn, dub, and s;entence, so forth, particularly sociallysalient routines u'hich are primarily linguistic eventsbehaviorally.Th.y are referential,creative rvhich are most inrportanf to ethnographic (or perfonnativc) in<lcxcs ccrtain ongoing ftrrrctionsl of description,since thcy incliviclualize lrapircning."l'lrcy constittttca lncssagcabout tlrc spccchas thcy arc functionl of the mccliunr,iunctionaliy! a pfagmatic act. Tire crosscultural investigation of these nrctapragmaticslifters is a very urgent desideratum and important anthropological Lexibal ltents itr General

are and problemof small elephants the like. Theseeftects entirelypho' rvhich in participating pairs (or n-tuples) most consonants nological, of by phonologicalrulesregardless their positionin lexical alternate in items.A lexical item thus appears overt form with trvo or moresets for for the of consonants, example, nominal adjectives size,the parai' iexample, -qbaix(superaugmentative), -g(')aix elaborated digm -k"uaic' -k'uaic (diminutive) (quasi-diminutive), ii(augmentative), , of for Upon request rcpctition a lexical i- -k"orit'|(super-dirninutive). ininformattts in item with suchcffectsthat had occurred discourse, "unmarkcd" variabiygavea lexicailynonnaiform-the pragmaticaiiy 'herbelly of a form. Sorequesting repetition iia-muqbal [whichI think by is hugeand repulsive, the way]',one getsi'td-muq'al. "But you just said'-muqbal'didn't you; that meansgreatbig one, no?" "No, it's 'her greatbig belly'?""Iagait, itanruq'al." . . . "W'ell, horv cloyou say is itantuq*al['It's large,her belly']."Notice that the last qucstion in' the equation. metasemantic for ss te.ryreted- -areo.uest an interlinguistic being consonantism beyond markerof rulesfor augmentative pragmatic manipulation. studied
Irtfo.fahrnomnliaf .o*irdl flems

i i i i i i i l

, , :

at to A certainamount of reference pragmatics the levelof speech of language rvhich in functionr) is accomplished every events(purposive of by quotationframingverbs,the equivalents English I am arvare (of such phrases ashe said(to him), he told (if ) to him, he asked him), have how him, andsoforth.It is remarkable manylanguages heordered to rvhichserve name fervof these, the expressing first only constructions ,l by Framed suchverbs,p'1'' events. set the entire,undifterentiated of speech
,,rhinh ipqorihe eertein sneeclr events. end the inflections of rv.hich de- i' r serv^ wlrlvrr
. t r r t, .,, f-, ---r:---^^^L

T'he-merenrnqmatic content of certain lexical items brings up the -'-'- ----r-questionof the pragmaticcontent of lexical items. AS complementary evcntsthat cquate rncaningsof above,metascmantic have discussccl I forrus of languagcarc thc basis for scgmcntal,scmantico-refcrcntial grammaticalanalysis,and vice versa. Obviopsly, in the semantico' rvouldconsistof elementary referential mode alone,the ideal language and their ruleS of hierarChicalcom' referring grammaticalcategories bination. But, as many linguists, particularly BloomfieJd (r93) and lexicaliterns-the elements Chafemore recently(e.g,,r97o) haveseen, equations-form a kind of irreducibleset that enter into metasemantic of of "idioms" or "basic irregularities,"the existence n'hich is really theory.True lexical inexplicable the basisof semantico'grammatical on or quality of specialization extensionor items have that unpredictable multiple senses their referentialfunctionsi which makes them rvhat in
l

scribe the participants, we find reported speech, the messages Pur' ror usec. I nere N a wnole range or qeYlces rcPurulrguPsssu Poftedly events,from exact quotation through indirect quotation through pseudoof reference, subtleties which the and quotation,paraphrase, descriptive I cannot explorehere. Additionall;r, languagesincorporatelexical items which in certain con'
r ,r! r f--^ rl-^----^L ^-.^^! ^l .-,L:^L ^ r^1,^strucuons rerer to, Illat ls, naulg, ttrg JPsccrr svulll ul vylrleu.{ LU\su

they are,referential primesof somesort, ihat iire coding of seenr' Bui ii is preciseiy ihe ievei of pragmatics ai clear.For lexical ingly arbitrarychunksof referential"realitv" becomes made up of semanticoitemsare abbreviations semanticcomplexes f.or referentialprimes in grammaticalconstructions (Wcinreich 1966;Sil' verstein r97-z and refs. there), together rvith all of the indcxical
nlOCLAnUe$ Ot meanlng
, ,t 1t.. . .

tnat

rt

maKe tllg fUnCtlt)llilll

iJ

-^rl---l

-----1r

lsulL uucxPst;l'cu,

---^---^-r^,1

their dcscrip' adumJ:rated fonns an uttcrancc fraction. I have alreacly tion above (see Reference and "Performative lSpeech,"pp. r8-r9
/;' .i I

nnalvsiscan ncvcr In othcr worcls,traditional scmantico-gr:tmr:tntical hope to specify meaningsfor lexical items finer than the grammatical

51

MICHAEL S^- .

JfLqLVlJ,

)rLtLBULnLLt rJllte$v[vu,

s'|ru v*]ee' "

.n allotvs,for everylexi: structure of implicit referentialcategories pragmaticresidue-an indexicalcomponentmotiva' .. .rnly includes a being only one reference at the level Of speechacts, actual discourse such mode. (Certain kinds of lexical content in the discourse-r'eference by mode have been characterized linguists as ad hoc "selectional" of restrictionson thc cooccurrence lexicalitems') namesin kinship terms or personal lexicalitcms as so-called So such It by be characterized a "semantic" analysis. is any society can hardly begin with; to that makcstl:em lexicaliten'rs the pragrnaticcon:ponetrt imit iJ the pragmatic functionssthat make them anthropologically pointing out among others, have never ceased portant, as Schneider, and chapter8 in this volume). Further, solsee sehneider1965;ig68: called folk taxonomiesof nominal lexical items, again "semantically" ripped from essentially reference, of analyzedby a proccclure ostensive For the rvhole pragthe tontexi of speech,give us no cultural insight. matic problem of why these lexical abbreviations form a euitural domain, rather than someother collection, rvhy theselexical items occur remainsentirely at all, rather than someother semanticcornbinations, siruciure of ihesevocabuetirnoscieniific The so-caiic<i to be expioreri. out to bc, from thc point of vicrv of a functional lary items turns linguistics, a restatcmentof the fact that these semantico-referential ratltcr than others,in fact occur. abbrcviations, Pragmatic Structureand Culttnal Function The linkage bctrvecn tlrc pragmatic grammar subsuming the tradi' tional sort anclthc rcst of "ctllttlrc" is through the tu'o typesof function of spccch.On the onc hancl, the cultural functionl of spccch colrcs nature, rvhich is to accomplishsome kind of froni its goal-directed rvork.Frcquently,as rve have seen,thcre are explicit eommunicational lexical items rvhich are shifters referringto such functionsl in overtly for But iireseiabeisare not necessary ceriain events. speech recognizcti On rccogniz.ecl. thc otlrcr ltand, the cultgral social functionsl to bc by functionl is the rvholemeaningstructuredescribed the specchacts grammar.As I have mentioned,all but a part of this of a pragmatic and in functione is not susceptible, general,to consciousness accurate as rules of semantico-gramtestimony by native participants, much .

are not. But thesespeechelements,rvhiChrepresent* :' matical systems of recurrences behavior,have such indexicalmodes of mcaning as Pre' of and createthe very categories society rvhich form the para' suppose event. metersof the speech then, to take naive native participant testimony, It is unreasonable, as interpretation, anythingmore than an ethnosociology mctabelraviorai rvith a tme funcfionall,2 (anclproblematically)ovcrlaps wlrich partially basedon indexicalmeanin$, in socioligy termsof a pragmaticgrammar of For the'investigation the latter must proceedrvith all the difficulties and comparativc that areat orrccdcscriptive hypothcses of iuterpretativc above PP. 4r4r, and Good' Formal Analogues, (seeFunctional atid enough 1g7o).Ancl the interestingrcsult is to see the rvaysin rvhich linguisticmeansto constituteand maintain use socielies specifically ccrtain socialcategoriei,one societymerging some of those given by keeping thcm distinct.with a anothersociety perspective, comparative give insighi stricily linguisticfocus, the pragmaticstructuresof speech "surface" material in distinct func' into the useof the sameapparent on And we can study the universalconstraints this rich tional modes.
---rr^t--rr-^r'lulll l)illlglrllllB.

^ L r v d^s,w l ^ -^ 1 u
d

arrrrr.vrvrvb

^-tl"n,.nlnoinol

nercnnnlivp

the nrnqmnfic

b-::::-j-.-

is of i1.'stcm speecli part of culturc-in fact,Pgfbap,s-the-rnos't-+ignif,cant and functionl,3of rvhich is structure ;rnn1mir6_;a[ililriit'iirc the probabty realmodelfor the restof culture,when thc term is a conbehavior. of system socialized structfor the meaning
CulturalMeaning constructto explain the mcaningfulness is Languaga the systernatic and bchavior.Wc hitvc sccn tlrat i3oll,", ir5.\-c5.ic{, uni919]1 of spccch accrueto speechbchavior.l.l}us any of rFiU_qliqttodes meaningfulness thcscdistinct enoughto conrprise has tintion of language to be iirclusive tire and elai:oratcci, indexicai rnodcs,in particular, as I havc strcssccl i::oclcs that link specchto the wider systemof sociallife. The investiga' of the uniqueness a real symbolic tion hcre lias claimed for language signS.I uscciior Purc sclri:rlliic rutotic, t'irattcntr catt'Dcjustifiabiy lts havelinked this property to the possibilityof the traditionalsemantico' and granrnaticalanalysisin terms of rnetasemantics, have found the distinct. The pragmaticaspect otlrerlinguisticmodesto be categoricaily \{

5z

1a 'i;*

MICHAEL

SILVERSTEIN

Shifters, Linguistic Categoies, and Cultural D escription' NOTE


at a r. This study replaccs longeroneof the sametitle discussed thc Schoolof Amcri' "Meaning in Cultural Anthropology." That rvork was a clraft Researchi.-inrr can on for tbe operringscctionsof a hrger work in progrcss the anthropology.oflanguagc. of som-e that, incoqporatingmaterial f-rom four This wori<,narLwcd in focus,refashions lecturesgiven during r974 since the conference: "Pragmatic Functional Analogucs in univerJity of'Nerv Mexico, Ir,lnrch z5; "Metascmenticsand MctapragLlrnguag-c," ior nrati-cs,-Implicatious Coguitivc llcscarch," Univcrsity of Chicago, N{ay 8; "1'hc of and ihe Pragmatics Meaning," University of Chicago, Mry oi lvleaning Prag,natics .7, nnd-R.rc"rch School of PacificStudics, Australian Natioual Univcrsity, Scptenr' like to thank Carol ber r8. For comnents on thc draft ch:rptcrs,I rvould cspccially Fcldn:an,Pat! Fricdrich, and Ncrn:rn McQ':own, !n addition to the particioants at of scminar.Iror plrticularly trscfuldisctrssion points tlrp Schoolof American Research lectures, aiding my attcmpts at clarification, I am indebted to PhiliP raiscd in the Bock, Carol Feldman, lvlaishalt Sahlins, David Schneider,Milton Singer, |ames Talvitie, Anthony Forge, Roger Keesing,and Anna Wierzbicka.The 6na1draft has and I beg the conditions of ficldrvorl<, becn completed-undei the iess-than-ideal reader's indulgence of my bibliograpbicl'lxity. The grlley proofs berrefitcd from a by carcful reading of thc minuscript graciouslycomrnunicated Rodney Huddleston,

by for that of language, example, rvhichis constituted its inilexical


mpde, can similarly depend upon metapragmaticusesof speechitself observauPon sensitive Othenviseit depends in only very limited areas. tion and comparativeillumination of functionall,2 speechacts and eventsfor the indexicalmode to be understood. speech is If language unique in having a true symbolicmode, then obviously other cultural media must be rnoreakin to ihe combinediconic and inthat of In then, rvecan conclude dexicalmocles meaningfulncss. gencral, and see "cultural meaning" of behavioris so limited, exceptfor speech,
a cuLLuTClL SY$L;ItPLLU|L A5 a lllas5lvg,
--.7L-.-1 s-^--:LL:-^----:--^ ..---lr:-l--

UrUltrPry

Pri.r6rrraLrs usJurllrLrvrr

-----^L:^

l^^^-:^!i^-

vt

^l

of horv thc social categories groupsof peopleare constitutedin a crisscrossing,frequently contradictory, ambiguous,and confusing set of pragmaticmeaningsof many kinds of behavior. about pragmatic meaning, If there can be such apparentvagueness one might be tempted to seein actual behaviorthe oniy level of then in integration,of orderliness, culture.But for the socialanthropologist, as for the linguist, regularitiesof pragmaticform and functionz will and integrationof such meaningsysultimately define the orderliness tems. \\/e need invoke "symbolism" for a certain modality of speech of alone;thc vast rcsiclue languagcis culturc,anclculturc is pragrnati

ore SlmkeWelIBef Using(L'envoi) "semanlike We mustbe carcful how rveuseterms "sign,""symbol," "functionr"andothcrlcxical to itcrnsrcfcrring entitics ticr" "meaning," in in theory. havctricd to bc consistent usagc this chapter, I of serniotic This ott ncccssitatccl, cx;rurplc, for usiug subscripts ccrtniutcrnrs. rvhich
:-!^-J^-l llllLttLlLll ^^-^f,.1 udlLrtll ^^...-.-!:^^ -^f,.-^.-r:.,1 JLlllatrlLIUlr'lululL!ltr(ll t,...^!:^.. ttlrlLtl\rrrl ^C vr ..-,,^^ \rJ(l5v ...,,"s lltllJL l.a trL +l'a Lrre

solccriterionof judgnrentof the argumcnthere that culture is, with the of of but exception a smail part of ianguage, a congeries iconic'indexicai of of sJstems meaningfulness behavior. Usageof the same terms by others should be similarly scrutinized for actual referentiai content, rvhich may differ considerablyin terms of the underlying theory. We must not be carriedawayby the rhetorical -that is, pragmatic-force of scientific argumentation, wherein, conthis natural eommunicationtradietorilyenough,lies its sole power -as chapter,alas,being another token of the type.

54

t.

z
cn -i a
rJi

F-|{

z z
b Fr{
F-t!i bErd ,* \./ -

i -f$

.ix
d

z
7
)i

r.
H

F-J , )

\)

:i--ii

X
l-l

t F
H

7 r

\
L-J 1

H \r'

K
Frd l - I

L-J

tn o

ar

v
\/

q)
H

U
J

i
F

;n

r- z
F+
-G

v')
a

\4,,,

F:

* b E
\ 7

l \,\.i

u ) v
(h t*

\1 J

zl q

You might also like