• Author / Uploaded
• d_yee90

Citation preview

FrameStllWltics

Charles J. Fillmore llnivcrsllY

of Califomla,

Berkeley

1.101'-101I \

With the term 'frame semantica' r have in mind I research prolrtm in empirical &emantics Ind. descripllve framework for presenclal the mults or 5uch research. Fflme lemanliet oft"e,. I perdcular wly of lookln, •• word me.ninp, .. well •• I wly of chlracterizin. principles ror creal in, new worda and phri1eS, ror add in. new !DeIIRlnp to woro., and ror auemblina

Ihe meanings of elemenll iD 0 t.. 1 Into the total meanln. of the tell. By the term '(rame' I have in mind aay s)'Item O(COnceptl rellted in luch. wly that to undentand In)' one of them you have to undentand the who:

or

structure in which It IHI: when one the thinas in luch I ttNCture II in. troduced into I text, or into I converutlon,.n althe othen Ire lutomatally mado IVlil,ble. J intud the word 'frame' u used hera to be. ae~' M.er term rOt the let of eoncepll nriou.ly known, in the litenture on n.turol language undentandina. u 'schema', 'Kript\ 'aceRario', 'ideltional teaf. (oldin.', 'colnilive model'. or 'rotk theory'.1 Frame .emantica COmet out or tndltions of empirical semantica nthet than formal .. montiCi. III. moot akin to ethnolllphic .. miotic&, tb. work of Ihe onthropolo,ilt who move. Into an olien cultu,," ODdub .ud! q..... tions I', 'Whit Cltelorie. of experienee are encoded by the memben or thl, .pecch community throulh the IIn",lltic chom thot they moh whoa they .a1k r A frame .. mantico outlook I. not (or I. not n......nly) looompatiblo wi.h work and rCiull. ,In formal ..... ntico; but it ditr... ImpartlDlly (rom formal temantics in emphuizin, (hecantlnuilla, rather tbaD the dilCOtltinui· lies, between 10naUiae .nd "perie""". The Ide.. I will be _tlnllD thl, paper represent not .0 much. ,engine theory or empiricll temlDticl I •• set of waminp about the kinds of problema luch • theory will have to dell with. rr we wish, we can think or the remarks I make u 'prc-ronnlll' nther than 'non·rormalist'j I tlalm to be IIltin,. and al well.1 r can to be describ. ing, phenomena which mUll be well undentood ond ca,.,fully described berore serious formal theorizjnl about them can become poslible. [III)

Tho Linguistic Society of Korea (ed.), 1982, Lingui.lic. in the Morning Calm, Seoul: Han.hin Publishing Co .

.

,'--

!

.,

-,

113

112

or

In the view I am presenti"£!. words represent calc,orillitions experience, and each of these categorics is underlain by a motivating situation octurring dgainsl a background of knowledge and uperience. With respect to word meanings, frame semantic research can be thought or as the etrort to under-

he doesn't live .. lth her," The .ubstllution in Ihi. frame of BUT ond YET suggests thlt these two words han

stand what reason a speech community might have found ror creating the category represented by the word, and to explain the word's meaning hy

Q

e..:; t..:;-

(by thil diagnostic

at lcalt) very limilar

runclion.: Insertion or MOREOVER or HOWEVER ,uuest the .. loteoee of conjunction. functioning semantically ,Imllarly to BUT ond YET but .... quiring senlence boundarie •.The conJunclion. AND ond OR can ..... oloafull, be inserted into the frame, but in each case (and in each case with different effect) the logical or rhetorical 'point' of the whole utterance difren impor.

presenting and clarifying that reason. An"analoD Ihat 1 Hnd helpful in distinguishing the operation and the goab of frame semantin from those of standard views of rompositional semantics

one came to know about these words was the kind of Itructures

is between a lrammar and I set of tools-tools like hammers and kniv~, but abo like clock! and .hoes ,nd pencils. To know about tools is to know

,hey could Decur and .. hit funcllon they hod wi'hln those atruc1um. In the tilly ,i.,ie., together .. lth William SoY. Wan. Ind .. en,u.II, D.

what they look like and what the)' are made of-the phonology and morphology. so to speak-but it is also to know whit people use them for, why people Ire interested in doinl the thin~ that they use them for, and maybe even what kinds of people use them. In this analogy, it is po5sible to think of a linguistic text, not as a rKord or 'small meaninp' which give the interpreter the job of assemblinl th~e into a 'big meaning' (the meaning of the containinllut). but rather as a record of the tools Ihal somebody used in carrying out a particular aClivity. The job of interprelinl a text, then, is analogous to the job of figuring out what aetivhy the peorle had to be engaged in who used these tools in this order.

Terence Lan&endoen and a number of other colle.sun,

2.

A Prlflte

Hblary of

thf

C~

or .yntlclic description by Oarlcs

Dnd played an important

role in the development

Frio. (Frie. 19S2)

of ·t.pnemic

formulas'

in the .. ork of Kenneth Pike (Pike 1967), the scholan who most dirctlly influenced my think ins durinl this period. Substitutability within the same 'slot' in such a 'frame' was subj«t to certain (poorly articulated) conditions of meaning-prncrntion or structure_preserVation, or sometimes merely meaningfulness-preservation. In this conception, the 'frame' (with its sin!!,lc open 'slot') was conside~d capoble of lending to thc discovery of imJ"Ortnnt functioning word clanes or grammaticltl categories- As In eumple of the workin~~ of such a procedure, we can take the frame conlisting of two comrlcte clauses and 8 gap hetween them, as in "John is Mary's husband-

wllh which

I WII IIlaclated

with

Ihe Project on Lin,ui"ic Analy.ls at 'he Ohio StIle Unlvenlty. My work on that project wa. 'arsely devoted to the classification of Enllflh verbs. but now not only accordinl to the lurface.syntactic frame! whicb were hotpltabfe to them, but also .ccordins to their «r1mmatlcal 'behavior" thoulhf of in terms of the lensitivlty of Itructure. conllinins them to ~!!iu~l~or~:::.-" ical 'transrormations." Thll project was whofe~heartedl 'tran.fomledsRalllt, basins its operations .t fint on the earliett work on Enlll sformad,

,rammar by Chomsky (19S7) ond Lecs (1961), ond in ill Iller 110,.. on within the theory lugested by the work of Peter Rosenbaum _ (Rosenbaum 1967) ond 'he book .. hich cstablish.d the Itond.rd workiD,

advances

'Frame'

Ilrace my own interest in semantic frames throu8h my career-long infertst in lexical structure and lexical semantics_ As a Jraduate student (at the Univenity of Michigan in the late fifties) J spent I lot of time exploring the cO-oC'Currence prlvilegel of words, and I Iried to develop distribution classes of English wordl usinl Itrings of words or strings of word classes as the 'framel' within which I could disco\ler appropriate classes of mutually substitutable elementl. This way of working, standard for a lonl time in phono· logical and morphologk:al innstisations, had been developed with particular

rigor ror purpotn

'anlly rrom thlt broulht about by BUT or YET, In each orthese ....... whIt

paradigm for transformationltist Itudiel of English. Chomlky (1965). What animated Ihis work W/IS the belief thai discoyeries in the 'behavior' of poIrticular classes or words led to discoveries in the structure of the ".mmar of

English. This ....

,.

so because It ....

believed that the diotribullonol

properties of individual words discovered by thil retearch could onl, be accommodated if the grammar of the 'anlUare operated under particular workins principles. My own work from this period Included I Iman monoVaph on Indirect object verbs (Fillmore 1961) and I paper which pointed to the c\lentual recognition of the tnnarormatioMl cycle at an openti",

prioeiple In a fonnlll'lmmar of En,llth (FilllIIOR 1963). The proj«t', .. ork on verbs .... ot ftnt completely ayntoctlc, IDtbe ...... that what was loulht was, (or elch verb, I fullacc:oual (elprnsed In IttmI or subcategorilition (fealures) of the deep Itructure Iyntactic fnmes wbicb were hospitlble to it, and a full account (expressed in terms of rule rttturn) of the \larious palhs or 'Iflnsformatlonll histories' by which ICtItenca COIl~.tning them could be transformed into lurfate senfences. The kind of wort. I hne in mind W8I carried on with much greater thoroulhneu by Fred '-fou!eholder and his collClfjues at Indiana Univenity (Householder et al

1964), and wllh ex'reme eore and aophi.tlcllion by Maurice Oros. aod hi, team in Paris on the verM and adjectives of F~nch

(Gross J9lS).

115

CMrta J. Fill*"

114

In the late li.'ies r belan to btl inc that cer.ain kinds of ,roupings or -verbs and classifications or clause types could be Ilaled more meaninlfully if the IlrUclurti with which verbs were initially luocialcd were described in terms the semlntlc roles of their Issociated .,sumen ... I hid become IWlre Clfterliin American and European work on dependency .rammar and \'.I~nc:e theory, Ind it seemed clear to me that whit wu really importlnt .bout • ~.rbwiliu 'oemantic y.Ie...,· (II on. miaht .. 11it)•• description of the .. m.Dtic role oflu .,...menu. V.lence theory .nd depcndeDcy anmmor . did not auip. tbe lime clauifk:atory role to the 'predicate' (or 'VP') that

Of

ODefound IDtnllwormationalilt work (.... e.•.• Tnniere 1959); the kind of .. m.ntic CluaificotioDl that I needed could be mode more compl.,e .nd sensible, J believed. If, instead of relyin. on thtorcdcally separate kinds of distributional .tatementa luch II 'strici lubcatelorization

feature.' and

'Kteclionll features,' one could take into account the acmlntic roles of all arauments of I predication, that oC the 'subject' beinl.imply one oC them, QUHlionin., ultimately, the reJevlnce or the I ..umed be.le immediateconstituency cut between .ubject and predicate, I proposed that nrbs could be seen ubuiCilly hl~ln. two kind. oC (eaturH relevant to their di.tribution In ICntenoes: the fint a deep-ttnlClure valence description elpreued in terms o( what I called 'cue (ramea', the second I description in tennl oC rule features. What I called 'cue frames' amounted to description. oepredictting words that communicated such inCormation a. the Collowin,: 'Such ....nd·such I verb OCCUR in expressionl contain in, three nominlls, one delilnating an actor who performl the act delfanated by the verb, one desianatin, an object on which the Ictor'. act hu a I.altOse a schematitallon or human judsment and behavior involvlnl notions of worth, rnponslbility,. judgment, etc., such that one would want to say that nobod)' ean reall)' understand the meanings or the words in that domain who does not understand the social Institutions or tKe structures or uperienee which they presuppose. A second domlln In which I Illemptcd to chlnclerize I cognitive 'scene' wilh the same function was that or the tcommercial event' (see Fillmore J977b). In particular, I tried to .how thlt • larle and important set of Enllilh ver'" could be Iftn .. aemantieall)' related to each other b)' virtue or the different WI,.. In which the)' tindexed' or tevoked' the lame ,eneral 'rane" The element. or this schematic S('tne included a penon interested in uchlnafnl money for loods (the Buyer), 0 penon inlere.led in uehanglng good. for mOllty (the Seller), the JOO4. which IheJIuyd'did or C-- .

118

119

Itood il. f.irly huge sltcc of the lurroundi"1 cullure, and this backlround undentlndinl il best undentood I' • tprololype' 'Ither than II • aenuine body or assumptions about what the world II like. II il frequently useful. when tryi", to Itlte truth conditions tor the .ppropriateneu or predicating the word or IOmtthin,. to conltruct Ilimp~ definition or'he word, allowing

the complexity of ftt between usn of the word Ind real world .ilultionl to be .ttributod to the det.ils of tbe prototype bockaround frame rather than to the de'lila of the word'. meanin,. nUl we could define an ORPHAN II • child whole parenti Ire no lonaer Ii~in .. and then understand the Cllc,ory II mOlivalcd Ipinlt I background or • partkulu kin~: in thl, .ssumed baclcllound world, children depend on their plrentl (or care and ,uidance .nd pa .. nt. a,"pI the responsibility of provldin, thl. ca .. Ind ... ido""" wilhout question; I penoD without plRntl hiS I lpecillslltw, (or soeiety, onl)' up to I pinkul., lse. becaUICI durina this period a lOCiely needs to provide lOme special way o( providin, are and inuNCtion. The alcoay ORPHAN doa not have 'built into It' any .peeift .. tlon of the .10 an.rwhich it is no lonser relevant to ,peak of somebody U In orphln. because that undentlodins il • put of the backlf'Dund prototype; I boy in hil twentin i... n.rally .. prded 01 beinl.bI.,o toke .... ofhinuclflnd to hay. passed the a.. where the main ... idlnc:e i. expected to come from hi' family. It I. tb.t backlfOUDd I.formatlon which delmnlnn tho f.CI th.l·tho word OR· PHAN would no! be Ippropri.tely used of .uch • boy, rather th.n Infonna. tion that II to be septu.tel)' built into I dHCriptlon of the word's meanlnl. Jn the prototype ,ituation, In orphln is teen .. aomebody dtservin. of pity and concem; hence Ih. point of the joke about the youn, m.n on trial for the murder of hi. ~renll who laked the court Cormercy on the arounds that he WI. In orphan: the prototype Kene 1.. lnst which sociel)' hUI realOn 10 clleaonu lOme children I' orphln. does not take into Iccount the cue in which. child orphans hinuclf. A, I teCOnd eumpte oC a alelor)' Ihat hIS to be fined onlo I back,round of Inltitutiono and pro