J79 1044 Metaforically Used Verbs

8/8/2019 J79 1044 Metaforically Used Verbs

1/74

American Journal of Computational Linguistics Mi c r o f i c h e 4 4

COMPUTER UNDERSIANDING OF

M E T A P H O R I C A L L Y U S E D V E R B S

SYLVIA EBER RUSSELLForest P a r k 9-E

Durham, New Hampshire 03824

Copyright o 1976Association for Computational LinguistiCs


2/74

A major problem conrront ing computer pro:lram:. drivenby nnturol-1 ngu ge inpu t consists of thc interprctntion of1 inguh . lic cxprossiobs f o r which the intendcd literalmeaning i s not e : q ~ l i c i t l ygiven by the l c ~ i o i ~ loml>onentso f the expresL;ion. ~'tnccxi~mpleis th c "extended useu of theverb ' l e a p t in 'the country l e a p t to prosperity'. buchextended usages--whether cuns i - a s inilntcd or original--czn be considered metaphorical t o th e cktcnt t h - , t t h e y arebased on analogies . Th i s paper establishes a framework f o rinterpretin? netaphoric,ll expres ionk by analysis o f under-l y i n g cbstrgct cor:~ponents-such 9s "trnns tio on" and "inten-

' i ' h ~ s i s i n c o n t r c s t t o pre-ity" for t h e nbovc example.viouc: epproaches which rely on 3 a~mbcr~f word sensesintended to represent me t n p oricL21usi ges d i r e c t l y .

An eexperimentcl r r o )ram f i n d , literal inter7rctationsf o r inyut rcprc~snting simyle scnt~l e m \.hit!. t h e"verb21 concepttf ( a c t i o n , s t a t e or c t t r i b u t c ) is used meta-phorically. This in ut h as t h c gc ne r 1 c o n A g u r a t i o n1 ~ u BCfl Vc,& OBJECT ;~UXCE/GO. IL or 2UBJLCT P iLD ICA TE-ADJECTIVEf. The i n t e r p r e t a t i o n a r e ~ i v e ? n the fo m ofprimitive ;;nglisFI pcraphrases . The-e parsphrases , whicha r e intended merely to illustrate the in formcr t i sn whichczn be extrrcted f r om metr.?:~orical inmt , dre b L ed on


3/74

scmr.ntic representations which are convertible to structuresspecif ied by SchL.nkt conceptual deaendency theory. Theinterpretation of mctaphori~nlly used verbs thus representsa particul..r case o f the general tusks of disambiguationand interpretation encountered by the conce-tual dependencyparser,

The hpprox im~t ionto th e l i t e r a l mc,ning of n metaphor-ical verb i s achieved through r e f e r e n c e t o s e m a n t i c d e s c r i p -tions based p r i m a r i l y on n srn.111number o f . conceptua l featuresand absrrrct structures. THe s e descriptors a re specif i cd Lorclasses of those concepts which ore c:qrcosed in dng l i sh bynouns, verbs, a d j e c t i v e s .:nd p r e p o s i t i o n a l phrases. Thecomplete set of values L-or the c l e sc r i p t o r s o r verbal conceptsis represented as a multi-dimensional matrix containing thedefined conceptd. This ma t r i x , which is only p ~ r t i a l l ydescribed in this paper, exhibits relatlonshipb and analogieswldch underlie metaphorically used verbs.

The relative independence and c b u t r c c t chnracter of theb a s i c sernuntic d e s c r i p t o r s render rhe system easily extensibleto f r t h c r c a p a b i l i t i e s , such a s more conclusive interpreta-tions or the tre tment or' more chnllenging expressions. Theemphasis on systematic uescriptions ,nd p r i m i t i v e concepts t oproduce sim~le raphrases is viewed as r e i l e c t i n g h w nunderstanding or novel linguistic expressions an& providinga model t o explore ques t ions related to such under-tanding.


4/74

Contents1. Approach

1.1 h a l o g i e s1.2 Conceptual dependency interpretations2. Characterization of Verbal Concepts

2.1 Levels2.2 States2.3 Structures2,4 Features3, Characterization of NOMLNALs

3.1 Features3.2 Function d escri pto rs4. Nethod of Interpretation

4.1 Conditions on metaphorical extension4.2 Operational context4.3 General procedure4.4 Operation of r o u t i n e4.5 Tests and c r i t e r i a

5 . Examples5.1 Le ve l s h i f t5.2 Category shift5 . 3 R-O switch5 . 4 Intra- leve l feature shift

5 .4 ,L Actor-feature s h i f t5.4.2 Object-feature shift5.5 ~ o u n ompounds

6. Conclusion


5/74

Metaphorical usages have often been regarded n s "specialccscou to vhich the particular language analysis mcthod underdiscussitm did not apply. This pnpcr prcsdnts n mcthod forcomsutez* undcistnnding of a class of phrases in which theverb i s used 9netnphorico l ly t t , but which ignorer: th e dis-tinction between rrextcndedlvnd "ass i~n i l e d t rusages. Thisapproach provides flexibility i n handling previously unseenusages, The assumjtion underlying t h i s approach is t h a tanalogies ore involved in language understanding t o a greaterextent than speakers consciously realize.1.1. -nalogies

Analogies arc t h e means by which we substitute, extendor borrow concepts. In the use of an analogy, a word isborro?+erl Jrom its usual con tex t t o exprcs3 some component ofmeaning sh'rrcd by t h e concept underlying the borrowed wordin its l i t e r a l sense and the concept which t h e borrowed wordi s t o reprclsent. Th i s results i n an extendeu o r netaphoricaluse or the word. The system t o be described i s intended toshow the analogy comprehension necessary for the interpreta-tion of metaphorical usages of verbs.

The problem of determining the meanins of a metaphoricalexpression is one of knowing the c r i t i c a l similarities anddifferences which a borrowed sense of a word has w i t h respect


6/74

to the o r i l g i n o l sensc. In somc case.; an es::cntic,lly meto-?horic,lL usage ccoses to be t l ~ o u g h t f a s bor ror~cd ,nnda c q u i r e s on id iomat ic scnsc of its own. r!owcvcr, i r thcsirr,il.;:rit.ies and dir 'r 'erencu which ontcr i n to : r t n p h o r i c . ; lusnses c.In be i d c n t i f i c d , vr? c:?n still ht dlc ~ u c h n ex-prcssian as wc uo those exnressions which :.rc p:encrallyviewed as iactn ~ ~ h o r i c n l Concid~r hc L xk3l.~ples

L) L'he ilousc 1;illeck t h e bill2) I scc w h ~ t ou mean

Here the f i r s t example oprlears to be ncraphorical, the secondnot. A langu,~ge nnaLy..er prepared to handle only non-meta-pl10ri.c~ input might achieve the c o r r e c t interpret2 i on o f'1 see1 i n the sensc of '1 un~crst'nd . :io,:evcr, it wouldsucccecl only if lsectswerei s ted' in t h e dictionall) as enuiv-a l e n t to tunaerut. . i~d' in one sensc. Such n so lu t ion ignoresthe cc:~pahilitieshich humrns hhve for correctly interpretingsuch sentences wi thou t h ving l ecrned t h i s s:.nonymity. Aparser which l acks t h i s a b i l i t y , i.e. to i n t e r p r e t withoutr e l y i n g on ad hoc aids, w i l l n..t have the f l e x i b i l ' i t y requiredto a,-proach similrr problems in r ~ h i c h uch nids are missingdue t o the prejudices of the aerson who deir'ines verbs forthe l e s i c o n .

In this sense, we should be able r o unucrst-nd mcta-phorical sentences on the basis of an analogy to the ord ina ryor l i t e r a l sensc of thc words involved. The excl~-nlea


7/74

1) The idea of growing t h e i r own radishes was born ,2) He h id his embarrassment about the honey pot3 ) He relinquished h is hopes4) Her painting s a i d something t o m e

are a l l metaphorical in d i f f e r en t ways w i t h respect to theordinary sense of the verb: th e l i t e r a l effect of ' h i d n ' isv isua l ; thnt of 'relinquisht has to do with con t r o l of aphyeical concept; thnt of say1 has to do with Linguisticallyexpressed information. But in each case there is an analogybetween the ord inary .mnti the metaphorical usage of the verb.The analop*.-d~onsistsf the similarity of the whichoccur in the mn-metaphorical and metaphorical usages:

I The idea (= to grow t h e i r own radishes) was bornThe baby (= Percy) was born

Effect = n ew idea (baby) can be related to2 ) He hid h i s embarrassi~~entbout the honey pot

Xe hid the honey p o tE f f e c t = Others a re n ot visually aw-ire of h i s emborrass-

ment (honey p o t )3) H e re l inquished his hopes

He relinquished the presidencyEffect = IIe no longer has a cer ta in o t tk ibu t e4) H e r painting m i d something t o m e

Her book araid somet1 ng to meShe s a i d something to me

Effect = I have a ~ n e w ental concept t o consider


8/74

Information derived f r om such metaphorical expressionsshould at l e a s t include analogous effects of t h i s kind,dhich represlmt the "rcsul t l t component of: tho meaning of theexpression. (A related problem of extracting conceptualinferences is d&scussed by Schank and Xieger (8).

This ta sk requires a verb description system whichcittcgorines verbs* y t w o c r i t e r i a :

1) t h e i d en t i t i c . i t i on of lm. underlying d t r u c t u r a lcomponent which is s imi lar f o r verbs which arc usedanalogously in linguistic expressi. ns , and

2) the identification of a certain level at which theverb a p p l i e s , such as "physicallt.

Each verb w i l l thus be classified, not i n t e r m s of a singlecategory, but in terms of two type9 of variables havingvalues according to these two criteria. Thus levels andstructure-concepts must be determined which can b e used asa basic form of description of verbs in the dictionary.1.2. Conceptual dependency interpretations

In addition to such verb descriptions, which serve theanalysis task, the form of "targetu representations, i . e . ofthe l i t e r a l interpretations must be considered. The ba s i ca s s t q t i o n underlying a choice of representation is that a"translation" from a metaphorical to a corresponding literalex:~ression cannot be achieved by manipulaf ion of components


9/74

at any syntactic level. What is needed is an tv inter l inguat t ,which deals wi th relationships between concepts a t thecognitive l e v e l . The conceptual represent tions which applyt o t h i e interlingua are not dependent on th e o r i g i n a l l ex i ca lform (or language) of the input, and can be used to generateparaphrases of the input i n t o the same or other languages,given the concept-to-syntax mapping rules f o r that ~ongu$ge.The c h d e a of a form of conceptual representation must beguided by the extent t o ~ h i c h t shows r e l , l t i on s h i p s brtweenconcepts at the cognitive level. The conceptual dependencytheory of Schank (6-9) provides such a- representation interms of predicntive and qualifying dependencies betweenconceptual categories and is assumed as th e context of themethod presented here.

In a dependency, according t o this theory, a concepto f one conceptual category is dependent on, q u a l i f i e s directlyor serves to describe a concept of another conceptual categoryaccording to rules of conceivability. These unambiguous,language-free dependencies are word-independent, although theconcept symbols on occasion map directly i n t o some l e x i c a lterm expressing these concepts. The nature o f conceptualdependency representations, as well as their s u i t a b i l i t y f o rmetaphor analyses, can be conveyed by a simple example. 'Theink stained the f l o o r ' can be represented conceptually as..nk color: X j . The significance of thls

colors x i


10/74

representation l i e s not i n the particular hotation adopted,but in the components of meaning which it r c v e ~ l s . Thedots (... ...) ndicate tht1t the ink is not necessnrily theagent, but is merely somehow involved in t h c action. The~ ~ d k @ t b nrrow!' (m) ndicates n causal reLntionship asopposed to thc eesclmple 'the ink h i t the floorv. 'Thenotation indicntcs n change of s t a t e of tfloorl, r mares p e c i f i c n l l y , an Inalienable P U T of the floor. The 'NEJ~ative'notation is a iconnotationtt 5 ) wl~ichis secondary to thepurely objective representdtion of 'stainT.

If i t i s assumed that the use of metaphor relies on somesimilarity of semantic components between an ordinary and anextended sense, it c.ln be seen that a representation of thistype, ref Lecting a conceptually-oriented sen-ntic theory , isndequcte to the task at hand. By reference to the abstractcomponents of causation, change of s ta te , pert vs. %,hole andnegativeness r e v e ~ l e d y the iibove c u n c e p t u ~ l tructure for*staint, paraphrase for the metaphorical 'h is businessactivities stained his reputationT is essily npproximated;'h is activities csused a neg~.tive hange in (part of) hisreputation1. There i s no dependence on complex trznsforrna-tions or mu k t i p l e word senses, which might in fact fail inthe: case of novel foms of expression, such as more rtcreative"metaphor.

The components of conceptual dependency representationcan be briefly described as follows. The conceptual categories


11/74

between which the various conceptdal dependencies exist areACT, PP ( l fpicture produceru) and PA ("picture assistcr~),At tho eyntaoti~ eve l , these categories are sometimes e mpremed in the English language by v e r b s , nouns and adjec-tives respectively. However, such correspondence does notalways occur . For example, many nouns can be expresseddirectly in terms of verba l o r attributive concepts ( 'the

' t h a t which,..'). Such nouns would not b et a t c of... ,mapped directly i n t o PPs .

The dependencies which hold be twe en the spec i f iedcategories o t the c ogn i t i v e level must ult mittely b e given*by a uconceptual grammarM which ref lects their conceivabilityand theref re their comprehensibility. Such n grammar,independent 6f actual word-construct usage, would includeinformation such a s what kind of concepts can be related bya specification of p o s i t i o n in time. Our concern here,however, w i l l be main l y with the lower-level and more detailedinformation contained i n a bfconceptuolizationl',or s i m p l econoeptual s tructure.

The general conceptual dependency format which has beenestablished for the conceptualizations which w i l l be referredto takes one of the f o l l o w i n g forms (semantic terms whichare irrelevsnt to t h e metaphor problem, such as tense,will be ignored):


12/74

PP (objec t ) PP ( objec t )ON

e,g. ink f l o o rThe ink is on the f l o o rThe ink is in contact with th e floorPP ( ob j e c t ) < PA (attribute)

e-g. ink COWR: blackThe ink is blackThe ink has a black co lo r

(attribute-value (new)> (change of s t a t e )(attribute-value (old))CCLOR: blacke.g. ink COLOR: ?

The ink turned blackThe ink changed to a black color

(goal>PP (actor)< $ -lm (action) +PP (object +- C source,I(continued) +Eiarye.g. John ATRANS +- C0NTROL:ink

I'32ohn"Cary ( IPART hand)(cont.) t ohn P T W +ink + John( IPART:hand)

John gave Siary the ink by handing it to herJohn handed Mary the inkCP(mrg1or J o h n e E P PRwS +$

?"-C P(John)John communicated the ink s t o r y to MaryMary heard about the ink from John


13/74

The a c tua l relevancc and character o f some of t h ecomponents of the latter t y p e depends on which ACT is present.The l i s t o f i&Ts i s : MOVE GRASP PTR hNSFROPEL SPEAK bXflCRiNSINGEST Am'ENL ATEtUSEXPEL E l 5 UILD'Ilhe source-goal component is irrelev nt to tile ACT GkLlSP,for exclmple. For FTk iNS (physic.11 t m n s i t i o n ) , the object ,source and goa l must be s p e c i f i e d and a r e physiccl. ForWCY\NS fmental tmns i t ion) , the object is i t s e l f n concep-t u a l i z a t i o n and t h e sourue and g o a l arc the rnentl-1 processorsof human o r a t least anirn:lte beings: - onscious -rocessor,Long Term Memory and Immediate iv!emory. For MTRANS (abstract-.I - - - -t r ans i t i on) the ob j e c t i s a f o rm of c o n t r o l and the sourceand goal re animate beings. Each of t h e s e three foms o ftransition involves a type of t ' c oncep tua l case t f : PTR.,NStakes the Directive or Locative case ( +"C and l+ll'&U?Sand A T W S t h e hecipient o r p o s s e s s i v e case ( +-RE.The object which is dependent on an A L ' ~ n t h a t i t i s

0"acted uponf1is in the Objective case ( t P (object)).

There are a number of other conceptual connectivesand modifiers which app l y to such conceptualizati ns. Thesecdn be referred to in (6). The r.ost important o f these asconcerns the representa tion of the concepts considered inthis paper i s the element of causation: 4 causing ccnceptualization>

m


14/74

orle e=> DOThis component underlies verbs such as 'make ' (one (=} 1)0 m 1 ( & c CDI,UK:


15/74

'forget ' and other 'verbs of mental transition. It is theseprimitive concept8 rather than :my snecif ic lexical verbwhich w i l l be retrieved from the ma t r i x as output of anoperational metaphor routine.2.1, Levels

The four levels postulated for vcrbnl concepts are:PIIYSJCAL (e.g. 'touch1)mNT& (em& 'think' )SENSOR' (e.g. ' seet )CONTROL (e.g. 'donate')

The PHYSICAL Level includes verbs which predicate theexistence, at tr ibutes or associations of objects w i t hsp at ia l (material) aspects,

The PENTAL-e v e l is distincf in that on t h i slevel ere representations of objects, or of other represen-ta t ions in a recursive manner. It is thus the l e v e l throughwhlh thought and communication take place. Verbs of thoughthave been analyzed in (8). Since l;Ch'T& objects are notreal-world objec ts or situations, but rather pointers tosuch objects, they cannot be concqtua l ly dependent onnon-3ENTAL concepts other than (usual ly human) mentalprocessors.


16/74

The SENSORY level includes V W s of perception, or thereception of %utgesl1. Concepts on this l e v e l prnvi.de thel i n k f r ~ mhe physical world t o the consciousness of alanguage user a s well a s to o t h e r animate beings . SEXMRYconcepts could be analyzed in PHYSICAL (spatial and temporal)terms. However, t h i s kind of dctail seems to have littlerelevance to the linguistic groblbms under considerht ion.

The CCNTRLL level refers t o relationships which expresspossession or cor l t ro l by an animate being. An object ont h i s l e v e l is a form of control, or a "potential for act ionfr .CONiiiOL mRBs basically consist of conditions attached to theactions of an animate being and are sometimes ewressedl ex ica l ly through modal auxiliaries, for example as 'can','may * or 'must *. Possession defined a s a CONTlWL r e la t ion-s h i p is thus distinguished from purely PIIYSIL iL or locativeralntionships. The verb 'have ( a phys i ca l ob j e c t ) ' , forexample, is defined in terms of t h e CON?R\L Level r a t h e rt b n he PHYsICAL l eve l .

Each of these levels has a few sublevels (e.g. SENSORY:eye, ear) which are sometimes s p e c i f i c a l l y referenced inmetaphorical extensions. These are described in (5).2 . 2 . States

Given the matrix format of the verb descriptors, thespecified levels (row components can best be c l sr i f i ed


17/74


18/74

cont, ACTUAL POTENTIAL

R en o y SUBJ : l i k eOBJ: appreciate0 be in CP p lease

VAL AV= : AV=+ : bc nice"be funtrVoli t ion

R0

VAL DV=+: be good (to do)The BLRTM, level is divided into two sublevels to correspondwith the faculties of perception and vulitioc. Perceptiani n turn has two forms--the predicction_ _ _ __ -f the existence ofthe -bject, and th e va lued perception of a p r e s u ~ o s e d bjec t .Only the former t y p e i s examined here.

It is first noted that this and each sublevcl a l lowsfor an R-, 0- and VU-form or' a VLB. 0 = Object , R =Recipient or experiencer ( "loc~tion" "sollrcelt or "goal t tof O ) , dnd VAL = VALW of 0. The labels R , 0 and VXL indicatewhether the lexical e n t r y which n ~ p s nto the slcts headedby these lzbels expresses a verbal concept from the p c i n t o fview of R or 0, o r expresses a va lue 0 2 0. Syntactically,the "point of view" of R o r 0 i s r e f l e c t e d b y a verb hav inga noun ~ i t h ' ~ ' r o l e " or 0 respect ive ly as s yn tac t i c s ubj e c t .For cxLlrnt?le,the R-role ' s ee corresponds t b the 0-role


19/74

' a pp e a r 1 . If no verb for o given slot comes to mind, aphrase is g i v e n which s mpl y ref lscts t h e conceptUcl represen-tation of t h i s STrdE . Thus the O-role entry corrcspondlngt o t b e l i e v c l i s tO


20/74

or wexpressedtt. 'Potentialt denotes that tllc object can ber e t r i e v e d i n order to create [In 'actual' relati nship. Thus'thinkt in the sense of m e n t d activity ('think about')involves an l1act iveW bject :md says


21/74

R need not be animate. The PHYbXCAL level reflects only thephysical nspcct of the r e l a t i on s h i p expressed by a VE1.B; Rmay h.,ppen to be animate, but the animate a s p c c t is irrel-evant t o khis l c v e l . This rnezins thLtt 'have in the genr;eJohn has t h e ncwspapert is nssicned to the CON'IXOL lcvel

rather t h i ~ n o th e P I T ~ ~ u A L .owever, a I J I IYs I t iAL relation-hip, as exoressed by ' John has t h e newspaper on h i s head1or '...in front of himt, could be derived 3 s a n inferenceof the CONTJiO L-level * have*.

Representative verb forxs for LtlA.:"Gs at t h e PHY , I L ~ALl e v e l follow:

ACTUAI, POTENTI&R have as partcontainhave on0 be connected tobe inbe onbe at be ne.tr

VAL be almost bebe -value>e,g. be redThe R- and 0-VERBS correspond to relctions between PPsi d e n t i f i e d in ( 5 ) (IN, ON, AT, PRDX), whereas the VAL VERBSare conceptual attributes--PA dependencies on IPS,

Th e ACTUAL/ l i Q TW I A L distinction as described abovedoes not strictly apply t o t he I J I I Y b I C ~& level, f o r in onesense all P:IYSIC;LL r e l a t i o n s h i p s ar e n a c t u e l l f . Fowever, ananalogy suggested by this analysis in comparison w i t h theother ltvels is discussed in (5).


22/74

2.3, StructuresStructures reflect abstractions of verbal concepts, i , ~

clerncntn of c o n c e p t u a l s t a t e s o r act ions which humansrecognire independently of whether any matter or objectinvolved is visible, A V E d structure cons i s t s of an" e f ~ c c t orn)onentl1 :md, i f the concept o f chnn~


23/74

Thus there are three t y p e s of change-of-state effects:a ) TUNs . tT ION (complete) mu 1

I--*b) start TRiNSITION ( l e a v e STA'IE) (TR-L) 'I--) f i n i s h TILINSITIUN (enter STATE) (TR-E) ,I-, .$These structures underlie verbs such as a) 'give1, ' pa s s t o ' ;

Features might be thought of either 1) as-achproviding an additional dimension of t h c matrix in terms ofits set of vnlues or 2 ) as applying t o structures in variouscombinations of vnlues to f m onfigurations of featurevalues. In the l a t t e r case, the ~~nfigurntionsrovide thevalues (columns of one (horizontal dimension. In eithercase, the vnlues of the following binary features indicatewhether a certain conceptual element is present in n VERB.The above structures implicitly presume n negntive value forall Eeatures except CONTINUOUS, which is positive (c.f.Fillmoret l-bbmentary (1)). Explicitly stat& featurevalues expand the i do rmd t ion given by these structures ina WkB descriprion. The fentures,witk '+ ' and - * examples, arc:

+/- AGENTive e.g. break (vases)/(vases) breakHypothetical believe/contemplateSHARED agree/believeCONTINUOUS i i v d d i eREP eated bea t /n i tVOLitional (voluntary) look/seeTRY (tried without o f fer/give

implied success)


24/74

it is relatcd t o the ro le spec i f i qn t i on of tlw vcrb a; dc-scribed in s e c t i o n 2 . 2 . The syn tac t ic subjec t o f +.IcYLNI'rcrb hns role L\C;:;NT , ~ n d hc ob cu t r o l e ;i or O. The featurei t s e l f is clciincu in a r cb t r i c t i v e sense; +/&G ,i':ll r e fe r s to,d~ethcr:n q p n t which is external, i.e. o t h e r th..n L or 'ir,

I - Y l y\i.4 i nvo lvhkb . 'thus 'tell ,mu 'give nrc + h (one whotolls = .\GAZPT; r ec ip i en t of inEonlre-sent nct~phorical hr, s e s . In the verb de i ' i n i t i ons , t h i si n f oma t i on is given in t e rm of s ~ecificotions n the


25/74

NOMIN4Ls which a!)penr i n thc d i c t i o n t : r y d e f i n i t i o n o r theverb. The t w o dcgrres of r e t j t r i c t i ~ n re rnitrkcd 'B (Broad) 'and 'N (Narrow) ; t h e specifications the~nselvcscon,:ist ofe i the r s g e c i i i c Ncjla.IN.Ls or fe.ltures of NOhLNALs. T h c ~ ed e s c r i p t o r s a rc illustrated i n ~ e c t i v n s ~ n d .3.1. kieatures

f * urc-oriented system of clcscrl ption for NOl!lKi~sis described in ( 5 ) . Hcre th e de f in ing elements LIL' NOIIINALsare presented without olaborai ion, merely t o show the t cn i~si n whicl~NOT41K'iL de?endencicct on verb 1 e o n c c ~ ~ t sre-specified. ii conf igur : . t ion of levels f c r SLIPEJILLs h t f s beendevise^ ~(r:nchis not i . :cntical t o b u t i s rc l l i tcd to t h a t

b , t f l lY~ IXL , T IM , , . Ilowevcr , f o r c h i s Limitea dis-cussion, VLU3 l ~ v e l s r i l l be ns.;umrd f o r NQbIIN,Ls , with

The fei i twes arc presented in three g r o q x , altholrghthis clividion i s n o t s i g n i f i c a n t t o the i imlementnt ion of th etheory. The first groun expresses t opo1o:~icnl o r b .sicphysicd propert ies :

+/- PART r o o f , s t e p / house, proofbWPE rainbow, i d e a / l og , geograql~yCONTAIN shoe / penci lFIXED f ie ld , tree / b i r a , ball1 - U I I ~W b i O N A L f ence , streak / ball, f l a s h2 - U I I Z N S I UN . ~ , ocean, tL:ble/ pole, st t u eF I ;U ID trpLurr:llr o n c c ~ t , iver , (some) time/ ice, moment-


26/74

It might be seen by t h e examples given t h a t these fet:turcsarc cons ide red : I , rbb s t rn c t ~rcopcrticswhich i ?nb c:;ten.~e\.to l e v e l s o t h e r Lhl1n t l ~ c 'IIYdIbJiLr

Thc u ~ c o n t l r o u p c~nsi~t:; f:

+~~mt iNi l l bc c'lns idcrod t o itnpl:; +.'iXliAi11'2.1 ' h ~third group iocu , ,cs on t h c "n~~::.ninC;"of a conce;-t

r s t h e r t han on an, ob j ec t i v e p r o p e r t i e s :+/- L I E k . y , motor, s t o r y / boy, stoneU~TNAI~IC oy, motor, story / key, stom

T h e uYNLUAI(: featurc refe-*; n o t to t h e r c s enc c or ~bsenceol;: n ~ ing par t ; , bu t rather to whether t h e c o n c c ~ t as somek in d of l~con t inuous xistence" by ~ t a e l f , ther than meres p a t i a l presence. The d i f fcrence between 'story and 'motort

ton one hdnd and k e y on t h e ot 'lcr is t h ~ t key i s a n i n e r tobjec t nr71ich is used ~ - , ~ s s i v c l yor D s ing l e o . e r n t i o n , . f t e rwhich it .iguin becviacs mcrel a * ~ i e c ef nc t a 2 . A motor( l i k e an cnimcte be in g ) , once stnrtcc;, r3pe:rs t o f u n c t i o nby i t s e l f . Likewise, o story cnu i n fact m - s t mcntnl conceptscan be thought Q a s having an efeect or l t con t inuous functionnf o r those peo;,le who come in contact w i t h these conce-ts .ThL significance of this f citure i s suggcstca by t h e m ~ n ycases in which peoale s ~ e a k f +UYN.tUiIC concept s as being"al;Lvetlor e f f ec t ive i n themselvea,

These f eatures are a l l essentially binary ( 1 + 1 , t - t )w i t h a possible v a r i a b l e vnlue ( ' ? ' ) f o r s u m features.


27/74

The ' F U E L t f ea ture , for example, is ltv~riab.bleu. A floweris +F lXEU in its natural s t a t e , b u t -FIXEU when in a vase orin many other circumstnnccs.3.2, Function descr iptor8

In addition to c ~ n c c p t u a l eatures which aetern~inethe

thcre are spec i f i c non-conceptual function assoc i ; \ t i ons whichapply to ma y NObiI&sIs , e s p e c i a ~ l y b L i i ~ b E nes, which servea s dcrining elements. Also ''sizeM c r i t e r i a f o r dependenciesarc recognided in the form of a 0-5 scale vL.lue for physicalobjects. Although these descriptors ore more i l nno r t on t forproblems not dealt w i t h here ( s e e ( 5 ) , they also enter thequestion of metcphorical interpretations. iior example, theknowledge thL : t th e funct ions of both a ship o t r a c t o r in-cluue the not ion of 'goingt or 'movingt is of use in recog-nizing the substitution of i p ' i o r ' t r a c t o r t i n ' the s h i pplowed the seat vs. the L i t e r n l 'tile t r a c t o r plowed the f i e l d ' .The function can then be incorporated into 2n approximation ofthe mekning of the former exarn~le.

aeveral t y p e s of function have been identified, accordingto th e conceptual roles which the objec t plays in the actionwhich represents the realization of t h i s func t i on. Thetype which is probably referred to most extensively in meta-phorical intetpretations is l-GXTMN L1, eaning that thefunctional object appears as an qxternal (to the actor)object in the conceptual representation of an ac t i on which


28/74

serves as an instrument t o some ra su l t . I n th i s experimentalinq~lcmcntnt ion n clbbrcviated functidn r c ~ r a s c n t n t f o n sused: 'kni~e EN: T ( cu t ) ) ' .

Method of Interpretationr\s imintcd out n t the bc::innin=, of t h i s paper, if n

dofinition of rnetb~p!lor t: restricted t o i n c l u d e only t h s eusages which strike the speaker of the , i v c n 1nngun;:e a sp , > e t i c o r c o l o r f u l , t h a t u c f i n i t i o n w i l l be uous , f o rlansuage i s constant ly changing w i t h r l a p e c t xo what i s -cons ide red "originr.1" vs . ghat is an established wordsense o r idiom. T h i s phen.menon c ~ u l d r o v e to be aquandary f o r anyone de f in ing verbs or other l e x i c c l itemsf o r e n t ry i n t o the dict ionary . 'rhc L - u a s t i o naf 16a: s ~ n s eof il verb is l i t e r a l m d wt1.d i s m e t a p h ~ r i c ~ ~ lan be cmectecto vary not only from one i n d i v i d u a l to another, but alsoover time,

In order to alleviate this problem, it is suggestedthat a d e f i n i t i o n 0 2 a metaphorical us~:geinclude any verbwhich i s "borrowed" f rom arrot\er level, whether o r notspeakers are still consc ious o f this borrowing. Forinstance, the word 'destroy' is easily conceived of asapplying t c all levels ('destroy house, image, idea,p r i v i l e g e ). I IoweVer , tFlis s y s t e m a s s i p s i t t o th ePHYSIWL l e v e l , f r o m which it can b e borrowed by extension


29/74

to other levels. A verb is simply always deLined as applyingltnoru&llyltonly t o a certain base Level (which in case o fdbubt con be considered to be the PHYSICAL l e v e l , if t h a tl e v e l is one of the alternatives1. A human e d i t o r w ed notworry about whether usages of the verb at other levels Brametaphorical.

Thus t h e proposed procedures rest on the. essmptionthat the ltmetaphorical senseM of a verb i s not in thelexicon as such1 the semantic component should exhibit theanalogy comprehension of humans, who do not need to hovesuch senses explained to them. If we accept that analogiesrefhr t o the sharing of a conceptual component, and aretherefore reflected in our t l l ev e l sv , which share one or morecolumns of our matrix, then the most significant way inwhich the verb description system can be app l i ed is ev ident :glven e verb which is defined in h e rn'itris by ~ ? nn t r y ig agiven column (structure) and row [ l e v e l ) , a metaphoricalsense of t h i s verb is represented by a V m i t h the samestrwture but a d i f f eren t l e v e l . This type of extensioncan be referred to as "leve l s h i f t u . A second t ype ofextensioqwhich abstracts the effects of animate a c t i ~ n sand app l i e s them t o inanimate objec t s is aescribed inSection 5.4,4.1. Cond-itions on metaphorical extension

Identification of a meta;>horical usage rec uires the


30/74

knowledge t h a t semantic restrictions od che dependencycontext of' the verb as used on the base l eve l re beingviolnteiL,but that t h i s v io l a t i on re!>rcc;cnts o. conlprchcnsiblemetaphoricLllsubstitution rot!ler than &In vanomalnusM casewhich must bc p a s bcd to some subsequent routine f o r i n t e r -pre t a t i on . In other words, i n t e r p r e t a t i o n s r o r mrtnphoricalex!~reslsionsmust sa t i s fy ccrtclin notions o f conce i vL*b i l i t y ,just as conccpts underlying l i t e r . * lushgea do. In terms ofdepqndcncies hctween an ob jec t , i t s l o c a t i o n and/or i t sattributes as described above, the most im.\ort


31/74


32/74

F I W D , on t h e other hand, is l e ss i m : ~ o r t a n tthan CoNTAlNin the dete mination of a met:r:)hcrical expression.

In juclg ~ n g he consec,uences of t h e unccr tc~in ty h i c hmay arise in the d e f i n i t i o n of these criteria, tlnc s h o u l dkeep in mind t h a t t h e ais t lnct ion between c r n e t a p h ~ r i c ~andan incornprehcnsible ex l r e s s ion i s also vnguc i:nd in Iborder-line1' n.:cs mcy vary from onc inu iv iduol t o another. T h eproblcn in language understanding is more oftcn to l'ind ani n t e r p r e t a t i o n rather t h a n t o e s c l u d e t l s tmngew onstructs.h lexicon e d i t o r , theref r e , may In cese cf doubt rezm nablyadopt a p o l i cy cf minimizing t h e 'Broadt r e s t r i c t i o n s ont h e NOh IN t i L s potev t i a l l y dc .endent on r he verb which isbeing ue~ined.

A related probl-ern of d e f i n i t m n i s t'7.e inker-3rc ta t ionof the feeturos in terms of which r h e f o r c g o i n ~ e s t r i c t i o n s=ire ciefined. 'l'he meaning 01 , .e prcscntcd features has beenb r i e r l y describecl f o r t h e PtIYalCj\L l c v c l ; a m ~ r e omdc t einterpretation of a t ese fe, u r e s for o t h e r lev 1s s h o u l deventual ly be concisely ddscribed. For exam le , at t -ePHYSICAL l e v e l ttlere i s a L i s t i n c t i o n b e t w e e n 'contain'i n the sense of 'surround' and 'containt in the sense of' cons i s t of'. At t h e IEhPTAL l e v e l these senses merge, orratiyer th e former sense seems t o l o s e i s elevnncc .

In a d d i t i o n t o the abovc c r i t e r i a , there are s e n ~ a n t l ccr i terL1 governing t ! ~ e" t a r ~ c i r e - > r e s c n t z t i o n w ?b.lchensure


33/74

that the i n t e r p r e t a t i o n given a s o u t p u t satisfies the generalrequirements of conccptual dctjendcncy for any conceptualstructure. Sincc these conditiuns are not p ecu l i a r to t h eproblem of metaphor i t s z l f , it is noted here on l y t h n t twolabels exist w111ch i n i ~ i c a t e thow seriously such criteriamus t be taken t t . The s o - t i s fnctiun of tunc~nditi~nalriteriaind cntec thnt the resulting i n t e rp r c ntion should bcaccepted in any cas& 'Condition.11 r a ie ru to c r i t e r i a whichsupport a " l a s t resort" in te r l ) rc to t ion- -an interpretationto be considered if no better alternativ: s ,Ire ava i l ab l e toth e parser. The in lemented ~~ r o c e d u r c s ~o n,?t yet exhibitthis discriminntioh in their output.

4.2. Operotiunal context%he parser wi th vrhlch t h e n~c t n p ho rinterpretation

procedure is intendcd to f unction (Riesbcck (4) ) operateson tl-e basis of scmantic expectations. To a large extent,the.;c expectations are concerncu w i t h f i nd ing in the sentencebeing pdrsed an object which c o n f ~ m s o ba s i c sem: nr i cre uirements ~ovcrning ' - e depenuenc of that ob jec t on averb which has appeared in the sentence. If there is morethan one pos s ib l e scnse o f t .e verb ~ ~ h i c has b een found,the choice of sense depends on what kind of cn ob j ec t isfounc. This object is described by a few fer-tur:.s such asPHTISZCAL and ANILTE. As t h e perser presently is uphysically


34/74

orientedM, expecting physical objects for verbj whichordinarily are interpreted in a physical sense, it is notable t o f i n d nn i n t e rp re t id t ion for e::tondcd us.il;cs in whichthe only candidate f o r on object is non-PHYaICAL.

l ~ o r e mcif i c n l l y , a p p o s e t h a t the parser finds theverb 'dropt i.2 t h e course of a sentence analys is ; t h a t onlyone sensc Q L the verb is given in t h e dic t ion : r y n p n r t fromidiomatic usngcs such as 'drop someone o line' ; and t h a t therniniwl requirements f o r its o b j e c t incluLLchc. specification'I?HYSIG.LL~. If this restriction is not satisfied, t h e porsermust turn t o the metaphor routine f o r an interpretation.Thus if 'ideai vere t h e on ly candidate f o r an object of' d rop ' , the parser w ould note t h c t c PHYSICAL- l eve l specif -cation (which could be represented as a +I)IIYhICAL feature-value) is missing from the definition of *idcat. It wouldthen chcckwith th e metaphor rou t ine , passing as informationthe c'indidnte for an object ('idcat), the v e r b sense of' d ropt which would have been selected, had th e object pos-sessed o +PIIYSICJ, feature, and any potential dro~per, ourceand/or goal .

As output t h e metaphor r o u t i n e returns a representationf o r each level a t which th e verb can be interpreted. 13isrepresentation, w ' i i c h is based on the serncntic components

o =n t r o d u c e d in Section 2 ( 'TR-L (0 AT R) (VOL +) ... ,' idea' , R i s +ANITWTE, e.g. 'het), p r o v i d e s the information


35/74

t o b u i l d the corrcc t conceptua l structure or t o f o rm imo?pro::im te pa r a ph r a s e according t o t * o c r o p m ucscribcdbolo;:J. ih:lt i , this informxtion cont: ims mc tr ix ilimcndionl ,ointer; \: ic l i lcitd Lo he catc' ory oi the inv lvcd oci:ionor bTAit~'t:.nd t o t h c .rJT or conccptucil n o t i i t i o n vhich under-lies thi:; action or S'l.rTS rr:;pccLivcly. For < ur ex: mnplc,ij\c und, ?lying conceptu; l l information nssociatcd wit11 thenbovc scm ntic coml3oncnt s at the IblA:li'i~L l c v c l u detcrmined

he (9 ~ ~ ~ ' I L L N S idea t Or,the c l iminc3ton of a cotnpuncnt of the $1 ;WJ'AL STLiL' of :ini n u i v i ~ u a l . ( Ad d i t i o n a l n o t a t i o n reprc.;entinz she concc7tunder lying '+VOL1 i s discussed i n ( 7 ) . )4.3. --~ n e r a l rocedureThe g e n e r n l . n c t l ~ o d .E t h c metaphcr routine f o r under-stondin:, 1 .1e t~ lph~~r i ca lxpres ions c n be s p c c i f i c d asf l ~ o w s . Thc r o u t i n e cxamlnes t h ~;emL~nt-icescr i l~ tor : ;of t h ei l l is scrnantic i n f o m c t i o n c-:nb e o b t a i n ed d i r ec t l y f ro m t h ed i c t i o n .ry entry f o r t h a t verb, o r i n d i r e c r l y in case t h eentry i s re!~resented i n t e r m o f another verb and c e r t a i nf ature va1uc.s. I t no t e s h e s ; ~ e c i fed NOBIINAL dependencies,including the 'Narrowt s~ecificationson these NOIIIR .Ls,if any. The sat isf ction-of t ' lcso specif i c a t i s n s by the


36/74

NuPI~NALs which ac tua l ly occur in Lhe i n i ~ u twould i n d i c d et h , - t C bcl;e interpretation is a v a i l a b l e . ;'he r ou t i nc thusconta ins the capability of determining such intcrprctations;howcvor, i n a c tua l 017cri1tion i t w i l l be nlisurWd that t h ep a r s e r 1) h as unsucccssiully checked f o r thc : ? o s s i b i l i t yof base i n t c rp rc t a t i cmz bcf ore tu rn ins to the mt.,phorroutine, or 2 ) has found n b.lsc rctprc.;cntntion, but isintcres ted in ,mss b l e t ~ l e t n p h o r i c ~ l lnterprct:rt ions.

Case (2) re f l ec t$ the f d c t rh, l t the i d c n t i f i c x i o n of abase intc; .pretct ion ~rccludes nonaly buc not t h e p o s s i b i l i t ythat a m ~ t n p h o r i c ~ l ln t e r p r e t a t i o n was ;c.ually-intended.This i s pL~ r t i c u l c : r l yi k e l y i n the case th,::: t h e NZ l l IN Lsinvolved have f e a t u r e s which ?lace than on n o r 2 than onel e v e l , w i t h the metaphoricnl l e v d l being more lrusunlllthant h e base l e v e l . rm ex i lmp le o f 1 type to bc consideredis tGuropC and h c r i c o o r e d r i f t i n g apart(.

In e i t h e r C D S C , the t,:sk of t h e r o u t i n e 1s to determine,on the bnsls 02 the z u i d e l i n c s of hecCion 4.1, whether t herea r e rL~e tap ' i .~r icz lnternretat ons Lor t h e g i v :n inl>ulr, and ,if s o , to r e t u r n r e 7 r c s ~ - n t a t i o n s t en. I n t e n r e t a t i o n sf o r a l l p o s s i b l e levels should u l t i r x t e l y be z iv-n 7 3 r i o r i t i e s .No del ' in i tc met ' l od has been s t a b l l s h e ~ ~o r detcrrniningpriorities i n i ~ o l z t i o n rom t h e context of discourse .Presumably such context vould bc th L dommating f c c t o r mestablishing the l e v 2 1 of the expression. T -us if the


37/74

n c t i ns of h m n s are bein discussed, t+.uropet ould beinterpreted in its i n s t i t u t i o n a l o r AN1M;X sense rathexthan its geographical P H Y s I C A L - ~ ~ V ~ ~ense.

If the expression is accepted as metap'1oric.11, i t smeaning remains to be repre.mntcd. In orde r to a rrivc a tt h e verbal c onccpt which expresses ihe tteefect I underlyingthe analogy amploycd , the ramgram uses t ~ c r u c t u r a l olomentsunderlying t h e input verb as a ltroadmnpi'through the tnlltrixt o ob t a i n the corresponding target verbal concept at thet~esired e v c l . That is, the structural elements or featurevalues cim b~ thought of as vclluc; of dimcnuions of thematrix which sp e c i f y an entry. T h i s en t ry , which may consis tof a 7rimit ive AC1i', for exzmple, con then be inserted i n t othe representation which g ive s an zpproxi rn t i~ ion f h emeaning of the phrase.

d o n g with s t ructurc l l elements, any magnitude descriptorspresent, i.e. rULUNT or I NA .UaI1lY:> , ore carricd l ~ v e r othe targelrepresentation, since it is i r e uen t ly these com-panents which ;re Socused on in a mt:taphoric.J expression( 'he jumped ( I i d L I Z j I TY : > ) to conclusionst). However, theprogram reterred to here doe; not y e t i nc lude this mec!~anism.4.4, Operation of routine

The procedure to b e described has been irwlemented inan extended version o i F&TRHN IV, which wal the onlylanguage conveniently accessible at t h e t i m e . me outline


38/74

given here represents the rocedure nc~ually ollowed inthe implemcntntion, which was dcsi,gned only L u r test

a ) Input: Yhc in,>ut: con~ists l: tw o or t h m c l ex ica l itemsin Lheir " r o o t t t forms in the order 'noun verb (noun)'. Thisgroup regrescnts a syntncL i ~onr'igurat ion d~ tc rmincdt c n t o t i v c l y by t h e pnrsor a s 'subject verb1 p r 'subjectverb object '. In terms of roles, the f i r s t case mayre2resent 'AGENT VV.LGtor '0 (OBJECT) VLABt ; the second

V B or '0 VUG R ( S ~ W C r GOAL)*. Theoret-ically, then, the entire role conf $swa t ion 'AGENT WRBOD CCP &3;11R08 and/or GO&' need not explicitly b e providedfo r in the fnput , since t ' l is configuration is covered by thetwo component configurations just given.b ) Dic~ionary e f i n i t i o n s : F i r s t , t h e ucmnnric d e f i n i t i o mo f a l l i tems a r e retrieved f rom the uict ionary . Examples :(noun) ship ( (PIES) T -1 (GONT +) ( 1 1 (XU +) (2D -1(S&iPL +) ( S I L Z 3) ( F L U I U - ) (AN111 - )(>a1+) (FN: EXT ( s a i l ) 1))(verb) plow ( (PIIYS) TR-E (bTATZ (0 BE bTJ E: ) ) (.G.,NT +)(WLE 0) I?'STR: TLIXST.&sTE (0 AT K)))(0 (hBW land) (BRU (2D +) (FIXED + ) I ) ) )Control i s then gassed t o t h e ' subject verb1 (SV) or'subject verb objectt(SV0) routine for determinction of roles .C ) Roles : A t t h i s poin t of t h e procedure, r o l e s to b eassigned are only temporary; a t e s t f o r the lfR-O switchTrtype of metaphor (Section 5.3), for mnstance, may determine


39/74

thct the role configuration ex,)ected on the b a s i s of syntacticinformi~dionhas been altered in the extended use.

The t e n t a t i v e ro1.e~ re assigned nkcording to roleinformat &on ;hen in the d e f i n i t i o n OX the verb:

F O ~ V: Role oi verb (F; or 0) s assigned t o subject.- .L I)For SVO: If verb is +AGXNZ':-

AGdNI' is assigned to subject and role o fverb is assigned to oF j o c t .- -If verb is -AGENlt:

Role of verb (K or 0) is assigned to subjectand the 'tither role (O or B respectively)is assigned to -bject .d) ~nterpretations: Control is then passed to o t h e rroutines, depending on which role configurat ion is present:

RVovRVOOVRA mAVO

These routines return any i n t e r p r e t a t i ons found, a ccodin; ,to the criteria to follow. In this version t h e interpretat ionsare expressed as pseudo-paraphrases, i . e . p.lraphrases whichignore ce r t z i n syntact ic details such a s word suf f ixes andtenses, in o r d e r to a l l o w for some measure d judgment as to;he extent to rh i ch the meaning of t h e metaphcricel phrasei s captured. However, i n actuz l operation, the targetrepresentation will be a c-0ncc~:tualone, which c o ~ l d eoperated on by a dialogue program or by a paraphrase program


40/74

4.5. Tests and criteriaThe following test:^ wi th corrosnoncung criteria for appli-

c,ition ern{ f o r success represent procedures which hme been in -plernented. Each test (b through d ) refers t o a certuin type ofmetaphor 0 3 shown. Tho discussion o f relev nt exkmnles in thenext section complements these speci~ications y b d ic a t in g therationt:lc used in the oprjroech to finding mctnphor ic 1 in te r -prctations. .\ e a t for n b~se-lcvclintcrpr~t~ctiona) has beeninclurred f o r ilurposes of cort~parisonwith examples seen as eithermetaphorical or (with respect to the given toot anornolous.a) -ase Level (alrunys t r i e d )

1) A l l EU'CHIEAU are consistent with bnse l e v e l of verb, i e . :l e v e l of 0 is base l e v e l of the verb;R for any -PHYSIC.iL verb is + i B I i -.Ti3 orhas an .WIE,L~TC function (e.g. co~~uter')R for any +pW6IC.L verb is +,FYoIC.rL;

2 A11 NOELINHLS f u l f i l l Narrow sp ec i f i ca t io n s foundin the dcf n i t i o n OF' the v e r b .In te rpre ted : IIe drank the inkThe sh ip LisintegrntcuNot interpreted: The chzir arsnk t h e inkHe closcd h is mind

b) Intra-"level (PWalCAL) Feature Shift ( t r i c ~f cll itemshave PZIY,ICld, l e v e l , but base interpret t i o n fL: i l s ) :Actor-fecture s h i f t :1) Verb spec i f ies +ANIU.TL fe~ture o r R;2 ) R is +PHYaICsL but not +ANIMATE;3 ) 0 Eulr' i l ls *Narrowt pecif ioa t ions f ound in verb

de f i n i t i on , or 0 is absent.


41/74

Interpreted : The c h ~ : i r rank th e inkNot i n t e r ~ ~ r e t e d : 'he s h i p plowed th e sea~b eer- f aeure sh i f t :

~ubject nd ob jec t fulfill t h e 'Broad' but notnecessarily %he 'Narrowt spoc i f i c r t i o n s byverb def in i t ion ,Interpreted : The s h i p plowed thc seaThe s k i e r plowed the sea

(Thc 'Broad1 s p e c i f i c a t i o n for t h e subject hereis ltsometkiw which goes", i. e . "somethingwhich changes l oca t ion : 'TR (0 AT R ) ' .interpreted: The c h ~ . i r ',owed th e sea

C ) Levcl S h i f t (from M I Y ~ I C A L e v e l only, at present)(tr ied f o r each p o s s i b l e l e v e l of t h e o b j e c t when nobase interpretation is found or icr a l l levels wheninpu t f rmht is (subject ( + A N IM d L ) verb)) :1) R i s e i th er absent, +ANIB;'iTE oran ANIbL.ZdE (Xnalieneble PART ,' i. a.4 PP 7ANIM T E ~IilART:


42/74

Interpreted: ~ o s p e r i t y isintegratedH i s ind i f ferenco col lapsedPro.qm5.ty came to the country

Not interpreted: r r o s p e r i t y was occupi~dP r o s p e r i t y cimo to :-hecho i rd) Level h i f t w i t h R-0 Switch ( t r i ed when nc>base--

interpretation is found ur when implied source (goal)is not expl ic i t ly present):1) Source or goo1 (temporarily assigned r o l c kt)

has level bLNT"U,, sF:Kx?&Y or CQKti'RGL;2 ) Temporary 0 is + ~ U L S E ;3) Source or goal i u l f ills 'BroL.dvspecificationsfor 0 given in verb de f i n i t i on .1nterpretp.d: The count ry l e a p t t~ ?ros;\erityNot interpreted: Thc cha i r l e a p t t o ;>rosperii.y

b m e S C ~ I F ~ ~ Sf i n t e rp r e t a t i ons a r e $ h e n i n Figure 1,which rei)resents a c t u a l output . In ut datci is ~ i v c n nFigure 2. Pa r t i cu l a r d e t a i l s or' the procedures used ire givenalong w i t h ciiscussion of these cnd o the r ex, ples . theyoccur i n t h e following expositian cf the various types ofmetaphor.

5.1. Level shiftNot 211 extensions a r e made from the PHYSIC& t o t h e

non-PHYSIC;& levels. The KNTAL, s ,NSOl.Y and CQXrRCIL levelssometimes serve n j a base irom which metaphorical extensions


43/74

can be made. The examples which rollow indicnte certainextensions (eotnc of which have evolved into idioms) whichcan b e made between l eve l s . Some types of extension areobviously more frequent or interesting than others.Examples for s p e c i f i e d extensions are:PHYSICAL - $EN?!&: He closed h l s mind,Protests rained upon the govemmcnt,Euro!~oand America are dri~ting part.Kohoutek's tail points to i t s origin.

(.lmbiguous between PIIYSICAL andMENTAL levels . On M:NTALI l e v e l ,a refers to 'inorrnc9tion 2. boutKohoutekls tail1 and ' o r ig in* tot informt ion about o r i g i n t . )PHYSICAL - SENSORY: Music f loodcd the room.PHYdfCA - CONTROL: The pr i v i l e ge of cleaning t h e erdserslanded i n his lap .Control of the si tuat ion slipped away,MLFlT& - PHYSIW: That chocolate didnt t agrce with me.SENSORY - M E W & : I searched for an cnswer.Let us x-ray t h i s political party.SEIQSORY - CONTROL: Their rights disappeared one by one.CONl'ROL - PENT&: She offered him an idea.CONTROL - SENSORY: H e r hat usurped h is view.

'He closed h i s mind! appears i n the output o f Figure 1.The base--i.e. PEIYsICAL--definitionof fclosetis one of themore complicated verb definitions, s nce the syntact ic ob jec tis either a space - an object containing the space (which isf i l l e d or e l iminated) , and i t s complete representation w i l lnot be discussed here. However, the "effectwportion of


44/74

the semantic representation ior tim scnsc i n w h i c h the ~b ec tis a space is th-t nothing can pass i n t o or cut of the objectcontaining the space. The rclev.~nt:por t ion o z the "pass i n t oHinternret i i t ion is renresented by t h e nested ueLini tLn:( c lo se ( (PHYs) (TR-L) (STdL; (( iIYP +) Tk-E (S'I'JL'TE (0 IN R)))){ROLE R) (riGdNT %) .(R (NR\T ( 0 ) (BKD C (CONY + I ) ) )).It is noted t h a t t h e *HY2 Value refers to the i~otentiol harnc-ter of the outemnost ST.iTE; the HYP vl\lue fen. the innermostSTATE is negative, honsistont w i t h the obscrv~itionin Section 2.2th'.t all pt~ysic i j lrelationships (excluding separdt ion) can beconsidered ltactuallr.

The r o l e routine determines th\ , since *close1 is R-role,the u i r e c t object of 'c lose , i.e. ' h i s mind l , maps i n to R, Butt h e ba se - i n t e rp r e t< ! t i on r o u t i n e then discovers that ii is +MENTALand not +WIa lC .L as r equ i r eu by ' c l o s e ' . The vroSr;im thereforeattempts an interpretation c t the l A d X ~ 'L lcvcl, she l e v e l of'mindt. R = 'mindt is an ANIId1'L P (*he CI:.d'I': nind =+ ) n fulfillment oE c r i t e r i o n - 1 It can thereforeserve as a t l locat ion' ' a t the 2, :NTIIL l e v e l . 0 is unspecifiedin the sentence and thus does not impose any l e v e l - or otherrestrictions. In checking t o see til-t the ' E roau ' s l , e c i f i c a -tions by * c ~ o s e ' r e satisfied by *mindt, h e p r o g rm findsthat 'mind does h;rve the +CGSI'-;IN f edture 2 s require~ l .

' f ie erb can therefore be i n t e r x e t ed st t h e XENTAL l e v e l .The innermost structure--TR-E (STA." ( 0 .Z R ) ) - - i s extractedand the ~.:ZNTIIL level is s u b s t i t u t e d for th e Y"HY,>IC&* Theabsence of 8 V E ~ ~ U Cndication is interprerea a s + The


45/74

4sttroodmapwthrough the matrix ort ti on given i n ~~iigure thenconsists of the dimensions 'EUNTiiL ( l cvc l ) P, (sublevel)R (kol*) SGZI'E (structure) -1HP (feotura] + (vLLlue). For pur-poses ~f p i~r .~nhms ingi r ec t ly out of t h e matr ix , the entry re-sultins f rom this so,.rch is the hnglish verb v t h i n k 1 , whichx~ouldcorres!~ond t o a concq) tuc~ l tructure 0=> ~ , L ~ L O C :Cp(K) ),i.e. be (mentally) located in the conscious processor of K t .k i t h closer attention to the sublevel(s) of vmind', a more spe-cif ic e%yression could b e deternitled. For instance, an asso-ciation of both the P and V sub levr l s with !tnin(J y i c ld s' think about- the t r u t h of.. . and ' hinlc about doing.. . .

In order t o complete the parilphrase, the progrcm assumesthi i t ther ds the silme r e f e r e n t as and notes f rom theclef i n i t i o n of 'mind1 that 'mind1 is an IPAR'J! of t h i s r e fe r en t .I t then picks up those conceptual elements uf tclo set ther thanthose describing the innermost underlying aT4%TE--(TH-L aT .bE( IHP +) 'SR-E)--and submits the entire l i s t of elements (he(IPART: mind) Tit-L STATE (HYP +) T L E t h i n k ) t o a simple mapoutroutine. This gives the rlparaphroselr'he ( I i ' A IRT : mind) STOPWSSIBUXTY-OF START th inkt . An a c t u a l genera tor could arriveat paraphrases such a s 'he stopped thinking '. 21 correspondtng0-role interpretation would be 'nothing ccn s t c r t to b e in(i.e. enter) h is mindt, which is also a reasonable approximation.

The aspect of tclosel hieh implies t h ~ t cthing can l eavcrather than enter would y i e l d 'he s tar ted to keep everything inhis mindT. This might be uwerstood- in terms of not forgettingor not expressing oneself. The inability of the described


46/74


47/74


48/74

smilingt. Yet the : ~ b i l i t - y o thus r e l a t e these expressionsmust be b . ~ s e ~n some underlying similarity with intuitiveappea l . This tnslc rcrluires a verb def n i t i . c n procedure suchas t h e one presented here, r:hich rcsts on n sm 1 1 number ofconceptual elements. The p r i m i t i v e element of TRiNS IT IuNunderlies 'he s t o ~ p e d miling , 'his smile d i s i n t c g m ~d tand l his smile left himt, even though on t h o s u r f ncc itappears o n l y t o underlie t h e latter ( t h i r d ) form of ex-pren \> ion . Phis element i s expressed i n our semnntic repre-sentation as TR-L, From there the TR-L structure could beincorporated into a conceptual diagram i n a number of ways,ee.g. by n t r a n s i t i on arrow I _I , by a ttcause-to-notnstructure 9 or by a "f n i s h - A C T t t n o t a t i o n

L ,

Lf , Thc l a t t e r notat ion i s the one a c t u a l l y used i nconceptual dependency for exampleaof the type @en. Thisnota t ion does not express any relationship between the threeforms o f the exxn7le given above. Howevef, it is rnnpyed outoL TR-L, which does show this relatianship and is referredt o in the f o l l o w i n g analysis of 'h i s indifference disintegratedv

Briefly, 'dis integrate ' ~s efined as changing fromexistence to non-existence of an object, on the PHYSICALl e v e l : ((PH) TTt-L (STATE 0 BE)...). since ' indifferencevis not consxstent wit11 the PHYSICAL l e v e l , a base-levelinterpretation fails. Since 'he ' is +ANIIL?LTE,the IEmAL-l e v e l noun ' indifferencet can serve as an a t t r i b u t e of 'het


49/74

a s ex,eriencer A. There are no feature specifications whichmu ; t be f u l f i l l e d by the a t t r i b u t e * indiffercncc ; theref r ethere crc no o b s ta c le s t o an intc?rprat, i t ion on the PiJY'l'ALlevel. Thc delfinition of 'indiffcrcncet yields the descriptors'((1.C A) (sT.:C& (0 r f R) (VAL + - ) ) ) ' , where 'A' representsiVAL, i.e. the LGNTd , sublevel ' A t t i t ude r , and +- is thevalue ..or 'neither p o s i t i v e nor negnt ive'.

'mdifferencel can be defined in terms OL e i t h e r theLTM or th c dP, i.e. either as +HYP or -1IYP; +IiYP is arb i t rar i l yassumed f t i r non-PlNSIQi'L concepts. T h e program thus entersthe matrix with dimension informarion (b& A K bTL!1'E +H +- ) toobtain a corres1,onding K-role VEI\D. It finds 'be-indif rent-to1, uses the R nlrenriy determined as subject and adds struc-ture element TR-L given by the ver,b to obtain 'he STOP be-indifferent-to...', leaving n slot for the object of thei nd i f iercnce.

The procedure is s i m i l n r for ' h i s smile disintegrated '.The mogrcm determines o cdtegory ~ h i f t nd nccet*s the TR-Lstructure f o r 'disintegratef with the I t lo ss of existcncell of0 (0 BE) interpreted as che t t l o s s of s ta te1 ' of ii (0 AT R).Thus the resulting R-role representation is the sane as t h d tfor 'he stopped smil ingt . (It might be noted t h s t the sub-stitution of a concept sucha s 'smilet for n physical objectcould be represented as a P H Y S I U - t o - A C T I V C , s b i f t , if anACTIVE level ispostulated f o r HO3lIN~iLs snd VERBS ( 5 ) .pursuit of t h i s approach would designate this example, like


50/74

the o t h e r exmples of t h i s sec t ion , as n case of l e v L1 s h i f t . )T ~ Qxc in~p l c ' t r u t h burned up ' is dismissed by Katz ( 3 )

as som.,ntic:llly l o u s . However, if h w n s ctm unders endsentences involving verbs which e:-lpnrently v iolnte selectionalrestrictions, tlxen such expresstons are i:lw z u b je c t tocompu te r understanding. In t e rms oC th e matr ix, ' t r u t h ',a conceptual attribute, represents a p o s i t i v e v.:lue of ona t t r i b u t e of n (lexically abs en t ) EC4N'i'tU ub oct , which isin tu rn dominated by a ( l e x i c a l l y absent) R. kince 'burn uptdif fers from Tdisintegratet only in t h e me.nz or manner ofthc aation, ehe endysis of t h i s e x ~ + m p l es s i m i l c j r to t h a tof 'h i s indifference disintegrated*, w i t h a s h i f t to theTVAL- rather than to the sublevel. By L l l o ~ i n gor theassumed R and 0, the o u t p u t r o u t i n e c o n o b t a i n t h e opproxi-motion 'one STO: know.. . ( tpeople storlped knowing ) or

one STOP i a l r iTE-TO t r u e inf rrnation ( 'people s oppedh'iving or telling the t r u t h ). Thus lilthough contex tualr e s t r i c t i ons on 'burn upt would inuicnte a +PiTYSlC=rLNOMINAL as a c t o r , the program st ill lluricierst ndstr he usagewhile recognizing t h a t it is not i' base or ltnormalv usage.T h l s is p o s s i b l e because t h e system i s 0 1 t e s rhe ~ r i ~ i t i v estructure of a verb from its ordir.sry selectional restrict ions .5.3. R-0 switch

B-0 switch isexemplif e d by 't le country l e a pt t oprosperi ty' in t hat 'prosperi ty ' rather than 'countryt


51/74

sppeors to be the goal :>nd is thus i n i t i a l l y oadigned role Rr. ther t h ~ n. This k i nd or metaphor may a c t u a l l y i n c l u d cB category s h i f t (which itself may include n levcl s h i f t ) ,and is unad to express a change o f s t a t e (of 'countryt)nsa transitioh (of lcountry').

Look ing 1,113 prosperity in t h e d ic t otlLbry, tho programfind (prosperity ((GO I3 W) (hOLC R) (ST.;I'E (t) (&*IT 3 ) )(R (Hum + I ) (BRL (?IUCLN+ ) 1 ) 1).That is, 'prosperity1 maps into on a t t r i bu t e on the CL.NTH)Llevcl (ExtrinsicL control of -Hysicol concepts) , is positivelyva lued , of o great MiOUNT, and dependent on any 4HULW conc ep t .'Country ' has tho f liture refuired for an R on t h e CUNTIULlevel (+ANIF!), and fu r t he r , it satisfies the +HWJ;Nspecification demimded by 'prosperity ' ; ' l eap to specif iesno pa r t i c u l a r restriction &or li other than +PHY,ICAL.

\.e thereiore wish t o take ove r the structure for ( l e a p t o t ,but t o indicate the CONTRPL rather than the FHY: ICAL level .The structurc essentially is TR-C as found in the def in i t ionof the 0-role varb 'leap-to '. The conce:~tof 'start to b e qor 'become' which underlies TR-E at any l eve l is transformedt o 'start to have' in m R-role e p ression. The object ofthe control involved,in ' p r oq erity , which is given as'PHYJICAL', can b e mapped in to t h e word ' rnater iz l t f o rpurposes of generation. Thus our represent tion y i e ld sthe R-role Tcountry SPART have-materialT in the implementation,


52/74

and c o u l d be the basis for other non-mctnphorical paraphrnsessuch a t l ~ c - r o l e 'the country 13: s bcsorning prosncrous .

.;nothcr esrmplc, intcrestin:, bccLruc;cp o t c n t i n l l y alll c v c l ; ~ l r c nvo lvcd in i t s mctaphoricol in te rn ic t :~ ton , isliurope ancl Ancrica arc d r i f t i n g a p a r t 1 . Subject to t h econtest of t h e ~iiscour;c, thc ~nct,t ,)horical ntcrnrr tit ionin t h i s cnsc 111.y turn out to bc n more likely interprctationthan thc bt se (:IIYoI~. '~L) one. ' d r i f t apa r t1 is defined inthe ~ictionnry s a symmetric, i.e. + & W U erb on theI? YbI&iL level:( ( P IES ) T&L (bT TE (0 AT R ) ) (NIL 0 ) (::Gtlrn' -)...(AKSZIJ + I ) .Si-.ce the syntactic joint actors, tl;uro!>e anSl '2merica1, areboth deiined ~ 1 3 av ing a +Pi;YdIGiiL comnonent, i . e . t h e i rgeographical areas, we have t h e PINaI(;& interprctction t h o tthe continents of Lurope and mc r i c ' : .are in the process ofgoing away iron one another. That i s , ,uro?e o r America orboth o r e l o a l n g t he loCation t h e y once ;ih:)red.

Since ' d r i f t t n o t e n t i a l l y takes a source or @a1 asindicatca by 'r-iT R t , th e excmple satisfies thc condition f o rtesting f o r an R-0 switch. The mis~ing OSiINAL or a t t r i b u t e(which rrould correspond to 'prosperityt in t h e previousexzmple) imn l i e r i in the sentcnce czn hzve any l e v e l , sincei t i s not e m l i c i t l y given. '2uropet and ' mcricaf asinstitutions f u l f i l l the +ANINATE condition f o r R:


53/74

(~uro;~e/lbrerica ,.. (ANIII + ) ... ) ) The level of thenissin~ OhlINAT, or a t t r i b u t e f rom wh-lch they are driftingis mknom. Thus the. program determines t h a t interpret t onson the ILNTAL, LGN;N~CJRYnd CONi'ROL levels arc also possible.On the P d N T A leve l , the abovc structurc f o r 'drift apar t 'is the structure which underlies a possible poratjhrase gen-crntion of 'Europe and iuncrica no Longer zqgrcet; on theSLNSORY love1 it is the structurc for 'Lurope clnrl Americano lon~er erceive the same things' onti on the LONTIiOL l e v e lit i s the structurc l o r 'Curope and .dnerico no longer havethe same rights, responsibilities or types of control!.5.4.. Intra-level feature shift

In the level shifts descrised above, a verb is usual lyborrowed frbm bne level znd aapl ied a t the level of the ob jec twith which it w i l l be used. In intra- lcvei ieature shifts,all components conform to the same level, usual ly the PINSIC&,but o apccification(s) or t'eature(s ) of the ob e p t is v i o l ~ t e d .h%en the +HNLIvGiTE feature of an ac t o r is v io la ted , n kind ofpersonif ica t ion or anthropomorphic behavior results, as in'the chair drank the ink1. This ccm be referred to as anItactor-eature s h i f t u . (1% corresponding ex-le on the IENTkUl e v e l might be 'that painting s s y s something t o m e 1 , wherethe psinting does not l i t e r a l l y say anything, but the resu l t oflooking at the painting i s the same as if something had been


54/74

said. ) If, however, it is the ob j ec t which docs not meetthe snecif i c z t ons of t h e verb def i n i t i on m d yet the 13hrnseis ~ c ~ r n ~ ~ ~ e h c n s i b l e ~ ,hcrc is on 'tabjoct-ieature +if t".An cxcmple is the s h i p plowed t he e eo r .5.4,l. kictor-Eenturc s h i f t

As s t l ~ t c d bove, there is no change in l c v c l t h i st ypc of mct t>p3~or ,u t t he +.it!.:IPI.T,: c s t r . i c t i \m on tho actoris vio la t ed . Thus ' t l ~ c ha i r drcnk.tb:c i n kT i. .n exampleo f intra-lcvcl h i l t , b ut ' ' ! ~ eboy uranul: i n the ->oetry isnot, as it involves an es t ens ion to a d1r"ierent l c v c l . Ingenera l , the semantic ret uirements on t h e object of such ane q ~ r c s s i o n rc th e scme as tho.;c in r n~n-mctap:~oricalsage.m ' t i bechair dz.nk the inkt, oth the ' c :n i r t 2nd the *inktarc ordins ry phy- ical concepts , a l t housh the use of 'drinkis not q u i t c the ordire ry one. m e::;.~.~~nation f t h i sexa~xpleby Lhc jcta nt ic corn-oncnt r c v c n l s nothin unusual.~bou; 'drank t h e inkt ; inkt is +P!;oibd~L and +FLUId asre< ireu by t h c NarTowt s pec i i i c a t i on s of 'drinkt. t clIc* 9though + + A ~ Y ~ ~ C L I L ,s noted t o lack the +LiNIiiiTZ feature valuesp e c i i i e d by 'drinkt, so the ordinary sense is r e j e c t ed ,while th e condi t ions ior an actor-feature shiCt cre s a t i s f i ed .

The determination of a metaphorical i n t e r p r e t a d o niup1i.e~ h a t the effects or Linguistic inr'hrencc; d e r i v ~ b l efrom the underlying concep tua l i za t ion a r e sirnil& to thosed e r i v ~ b l e ro n n conccptu a l i z a t on containin; t h e l i t a r a l

4


55/74

sense oi tarink', which is 'to 1KGG;SST a +PHYSIC.Ly d?LUII)substancet, Since the in13ut: example is already in R-roleform, i.e. with tho Recipient a s subiect , an 0-ro le formis given as pmaphrase in the output. Becausc th c structurei s a T R 3 one, w i t h 0 = ' i n k ' , i t is known t h a t the ink wasremoved Errlrn somewhere and is now in the chair. The inf r-mot ion given a s o result is 'ink STAiXT BR in chairt. Con-sidering other vari:~tionson th i s i n pu t , we note t h . l t we couldnot readily interpret 'the blotter drank the c h r i r ' , since'chair is - r 'ZUIU,5.4.2. Object-feature sh i f t

The examp l e 'the sh ip plowed the sea' f a i l s n literalinterpretation on the basis crf the def in i t ion of t p l ~ w t(PLOW ((1~3X3) TR-I: ( b T l b E (0 BE SHAPE: ) ) (ROLE 0) AGENT +).., (IasTRt TR (STATE (O L ~ T)))(O ( N R ~ and) (B!U (212 +) (~1ml.1) I ) I ) ,since 'scat is not a synonym for a n d However, 'seatful r i l l s the '8rond1 spec i f ica t ions of 'plow' for 0. 'Shipt,the syntactic subject, is assigned the r o l e c . f AGXNT. In(S ) , i t is explained t h a t an AGENT, i.e. n NCEIINAX which hassome r o l e in a causative ac t ion , either 1) i s +;th'lbL.TE,2) i t s e l f represents an action and therefore has the ACTIVEl ev e l , or 3) hac a spec i f i c func'iisn w h i c i l entor:. into thecausation. Since 'shipf isneither +.rNIB2~'l'i: nor ACTIVE-level,it is assumed to have a functional role in the causingconceptualization. The program checks to see that certain


56/74

ranuirements oi an instrumental involvement of 'ship1 inlplowing' n ro fulfilled. The function Qr' vshipl is givenas ' sa i l . '. The strucfure ufidcrlying J s . ? i l is TK (:iT.;TE(0 AT R)). Although the noun vnlo~vtight be g i w n a s thecx ) l i c i t instrument of tho verb .'olow , the program ignoresthe fai lure to cgree w i t h such specif ic i n f om : a~ t i o n , j u s t ;ASit ignores Narrow restrict ons on' ob o c t s when consideringmetnphoricnl cx 3ressions. On examinlbg the gcncral structuregiven for the instrumental concoptunlizotian of '?low', the

4program finds TR (bT.SE: (0 AT R) , which aI;rees v i t h th everb s t r u c t u r e of the Punction of s . In othsr words,although only o 'plow' can truly '?low1, in a r n e t a ~ h o r i c a li n t e r o r e t a t i o n anything' xhich tt~hysicallyoesu can con-ceivably h..ve a l lp low-l ike" effcct. i'hc ?rosrarn the= o r earrives at the roq:h intcrprctotion 'ship DO (sen 3 dT(BE SIIAPE: ) . Conceptual dcpen..ency rules would thentransform ' sh ip DO ' into a structure corresnonding to'one o7ernte sh ip ' .

~ ~ n s i d e row a vzrb- - 'kill - - w ? ~ i c h is subject: t ometaphorical use. but in a non- straight rword way, sincea l eve l s h i f t and/or-an object - feature sh -ft may k involved.Iq the exam2Les 'John Bilged the cat' and 'the House k i l l e dthe bill , the ordinary object of 1 ( c c t ) is no morea mere ;d~ysical b & c t than t h e metcpY.oricc1 o b j e c t (tbilltis a mere mental object . The P~?YSI~-to-i~iiW.~Lxtension


57/74

in the aecond example is obscured by the simultaneouspresence of an object-feature shift.

To c l a r i f y the r o l e which each type of s h i f t plays, asimilar example is f irst presented which i n vo l v e s onlyob ect-feat- s hift: 'he killed thcmotor . This examplecould be more explicitly paraphrased as 'he d i d somethingwhich caused the motar to die* , The interpretation dependson what it mewrr-fot the cbject. *motdrt, o ' d i e 1 or 'bedead'. It would be deo%rablc to obtain the interpretation'he stopped the opekathn or r n i n g of the motorv, whilerejecting a s b - s intcrplrtation f o r 'he k i l l e a the s tone1 .*Motor*and 'stone * are,both RIYaICAL NObIINC\Ls; no l e v e lshift is involved. Rather, th e +ANII~ IYZspecification onthe object is vio lated, yielding a mctaphoriccl, interpreta-t ion in the f irs t case and no interpretation in the second.?Motort is a medningful object of 'kill1because it is a~ ~ I COMIN L-, its funct ion k in g to r u n . ( It i s r e c a l l e dfrom Section 3.1 that the +DYNAMIC feature va lue spec i f i e s afunction which can be ident iEied with th e part i cu lar meaningof a NOMINAL.) When a motor is ?killeda, h i s f u n c t i o nattribute i s eliminated-- a consequence which. d i f f er s frome.g. the disintegration of the motor which might representi t s being 'destroyed'.

The proceaure of the program operat ing on the f i r s t twoexamples can b e outl ined as follows. The semantic repre-sentation for 'kill is:


58/74

(kill ( (PHYS) TR-L (STATE (0 )) (ROLL 0) ( IGD IT +)(TRY-1 . a .CO (N1W (AN'IM t))) (BRD (UYN + ) ) )).

For the exmple 'John k i l l e d t h e cat I , the progrim w i l l f indt h a t all specifications on the NOMIN &s by the definition o fthe verb ere met by the wrds of th e in3ut. In p.lrt icular ,'u Pttrd'n'ttis + '.~TIBk'~TE;h d t is, in terms of DYNAMIC FUNCfIOK,it 'livest in a literal sense, Thus th e Liter,til sense of'kill' 'is accepted. If the inbuttJohn killed the stonet isencountered, the program notes t h a t ' tone has no r . '~NIEtiTEfeature vCllueand therefore f a i l s a base interpretation.Furthermore, 'stone isnot +DYNAMIC and theref ore does notsa t i s fy t h e 'Broadt spec i f i ca t ions necessdry f o r a mcta-phor ica l interpretation.

The input ' t h e House k i l l e d the bill presents a moreinteresting case, '~ouse' n the sense of ' ~ o u s e f Repre-sentative~~r 'Lower HouseT h:*s th e +ANIPLiTX featurepreferred by the target representat ion condi t ions on anAGE;NT as sp ec i f i ed above-with respect t o ' h i p t ; however,(bill1 does not hov e the +ANINATE feature value as requiredby 'killt. Thus a base- level in terpretat ion i s rejected.IIowever, t b i l l does have the +DYNAPlIC v ~ l u e , orrespondingt o the observation t h a t it has a ucontinuous effectly on people.Thus th e bas ic components are s a t i s f i e d for an interpretation.since the dispensable 'Narrow + A N I ~ $ ~ T Ee s c r i p t o r , i.e. thel i t e r a l live' function, is v i o l ~ t e d , ut the minimal, i .e.


59/74

'Brood1 requirements are fulfilled, the employment of 'killtis considered a metaphoricill extens ion from the l+IIYSfCAL tothe EZ1:NTAL level, Th e structure TR-L ( C L " 0 N O )(AGENT +) then yields the paraphrase lHouse &TOP bill becomelaw', where 'lawt is n COXI'WL concept r e ~ ~ r e a c n t c dn termsof lone mustt and lone nmyt.

Gy no t i = t h a t which il; co\~-unon o both the base senseand metnphorical senses o f ' k i l l ' , w e c.ln cotl-tp rc the m&nn-ings of thcoc srnscu. The underlying s t r u c t u r c oE t h e verbi t s e l f s p e ~ i z i e s n $11 cases that on 2ction was successfullytL4ken o e l b i n ~ t e he 1)YNUIIC: func t ion o r effectivaness ofthe object . 'fie c f f e c t component o i this structurc sayst h a t the Object no longer ~ x i s t s n its previous State, f o rth i s i s t71e in t e r ur e t nt ion assigned to the TR-L structure.Thus the ccit no longer lives; the o t u t o r no longer runs;cons iderat ion of the bill stops, and the ntended result,

A.r ; Idefined conceptual ly a s i s prevented--thus the p roh i b i -tion, o r de r or permission contained in the b i l l is neverr e a l i z e d ,

5 .5 , Koun compoundsThis type of metaphor a n d y s i s can also b e a p p l i e d to

noun compounds i n which the nouns are defined in terms ofv e r b ~ l oncepts. The further ~.evelopmento f t h e above rnech-a n im sm s must precede an implementation of the more coclplexnow-compound metaphor analysis ; however, t h e =.p,n-oach t ointerpretation of such constructs can be ind ic at e d . An


60/74

example is given by the noun compound idea factoryt, \.hichis close in nleoning to t h e verb-noun cdmnound ' t h i n k tankt.If the +I' l lfolU~L bjects or mc:tter u su L. l lyasvociatcd withf:;ctory or 'tank1 arc ignored, noun-coml~ound i n t c r ~ x x t , ~ t i o n

pr0cedurc.s (5) can be used to arrive at ' i n s t i t u t i o n whichm ~ k e s deasl or lcnvironment in which one thinksq e s p e c t i v e l y .Ncre tllc verb ' th ink and the noun ' i de a ' , which is an o b j e c tot thought , retain t h e i r l i t c r c l l srnsa, 'c~!~ereushe funct ionsunderlying 'factory ( tmalcc ) and 'tnnk ( be in undergo ann b ~ ; t r , ~ c t i o nrqcess similar to t h a t involved in l e v e l shift.

Consider a l s o the esarnple ' t h e foreign-born may hold t h e\nz i te IIousc key s o c n t . It: is p o s s i b l e t o uncierstnnd *I..% iteHouse key1 in i t 8 metaphor ical sense because: 'keyt 5s aNOBIINAL described ,is having th e iunction ' o w n ' ; 'opening

ti m p l i e s the possibility of enteringv (cT. ' c l o s e ', sec t ion5 . 1 and ' b i t e Hou;c * is not only a A I T ~ L & &bui ld ing , b u tis a l s o defined w i t h the fectures of $n institution, whichincludes ANIbdfE beings. Thus the f c lmework exists i o rhandling some m e t a p i ~ o r i c h l l y sed noun constructs withunderlying verbal and/or ettribvtive concepts.

The examples of Sect ions 5.1-5.4 are representative oft h e v ~ r i o u smetaphor mech:tnisms which have been identified.The (;uestion arises as to the extent to which such mechanismsh o l d for any metaphorical use of a verbal or a t t r i b u t i v e


61/74

concept. An assessment of the v a l i d i t y of the nnClysismethod f o r such metaphorical uses depends upon 1) the com-pleteness of t h o identified c a t e g o r i e s , i.e. whether suchcategories cover all types of verbs in t h e c l a s s undercons idera t ion; 2 ) whether ~ u c h a t e g o r i e s cre based on themost "importantH component which enters into me t a p h o r i c a lextensions; and 3 ) the extent t o ~ h i c h : . r i o t i o n s within acategory af ~ e c t h e plausibility of a n i n t e r p r e t c t i c n . Thefirst two condi t ions a r e concerned.r:itk thc q u o o t i o n of 2"rninirn4lMinterpret'.tion, i.e. th c exclusion of n :!lscflinterpretation, the t h i r d with an "adequate" i n t e r p r e t ation.

With tespect t o th e first point, the Verb descriptionsystem presented has intenticnnlly focused on the breauthor scope of the categorization r d h e r t h m on 2. more d e t a i l e dil~ustr~tionf any one category. Such an ovcrview must havep r i o r consideration bccciuse t h e trensl; i o n of a mctaphoriccllverb requires c o m p z r i s ~ nwi t h o t h e r vrrbs, which themselvesmust be assigned a ~ t l o c c t i o n ' lwithin th c s y s t e m bclr'orc ~ n yrefinement of inter~retatons can b enin. The ~ i v c n ystemout l ines t is cc7tegorization in terms of three r e l ~ t e d rim-i t i v e structures-= &TATE, ENTER-LT-bE and L&iVL-STATE, whichare subject tc. embeduing, as in t h c cuse o 'close' (Sec t ion5.1). Roles define the ap l i c a t i o n of rhcL s t r u c t u r e s t o an"objectw, a "locat iont t of an object and an '?a&entftf anychange, the result of 1:hich is represented by such a s t r u c t u r e .


62/74

These r o l e s , which ore f e w i n nut.iber and r e l i l t i v e l y simple t oidenti~y or ilny given verb, are adec ucte t o r c l c t e ilny lex-i c a l verb rorm t o i:n undcrlyin s t ruc tw c . This structure-r o l c d e s c r i p t i o n d iv id e s the c l a s s o r prcd icc : i v c concepts

b

with t h e c s c c n t i o n of " l o g i c c l t t terms such a s ' i lnplyT o r' equate '. Thus th e field of verb(11and attributive conceptsi s covered by this m i n i m i l l c l a s s i f i c a t i o n b r o s p e r i t y t i s notobvious ly metaphorical, t h e r e i s c " t r ~ n s l ~i:ntt between it


63/74

and the phrasc 'he became prosperous which is similar tothe translat ion between th e more coLor iu l 'he l eap t t oprosperityt and 'he became (suddenly) prosperoust. Ineach c n s c , the pr imi t ive TK-E representing 'start t o 1relhtes th e two forms of the expression. 'l'his s imi1: l r i tyrests on the f a c t that-l l l i n gu i s t i c expressions whichtreat abstractions (lprosperous') as objects ( 'prosper i ty ' )might in a sense be considered metaphorical. It seemsreasonable to ap roach the problem of metaphor with ananelysis valid f o r the simplest form of such expressions.The analysis represented by the structural descriptors ist r i v i a l but bas ic in th'lt it is a prerequi s i t e to any morecomplete interpretation, and in t h a t it relates expressionsexhibiting varying degrees of metaphor withou t resorting t oad hoc def in i t ions or rules.Although they provide a bas ic in te rp re t~ t - i i rn , h e

structures and features wh i c 7 ? render an extension meaningfulare not necessarily, the focus of a metaphoric21 expression.The focus moy be an attribute which, while provided for and


64/74

broauly c la s s i f i ed by t h e structure-lcvcl d c i i n i t i o n of t h everb, i t t ielf rul~nins o bc dof n e d . To tr:kc n r. h e r d i f f i c u l texan~,?le,h e verb t b l e a c h l can h . i incu :I!; '+AGJNT ll'&,LaLdi"I'L; 0 2 1 : o ) ' , o r , ' t o Ci7udc ,m bject toV g 1 d E of Qkc~i:'. If 'blench l i.,u;ea ~ n ~ t u p h o r i c i ~ l l y ,as in :,he bleached t h e storyt., t h i s dciinition gives themini!~mL informilt ion that solnc L : t t r i b u t e of the tory dis-appears. This is the n o d btli;ic ,or nccc.-.,:*ry;I. rt of t h einterpretation, but is not'very interestins. It would alsobe desircble t o know how the attribure itself enters intothe metaphor, i.e. what the color or loss or c ~ l o r ignifies.

For quan t i t , tive attributes, i . e . those ~ c i t h agnitudesas values, t he p r i i ~ i t i v e s. . and I + r c in-cluded in the deiinition at the PI':YAILAL l e v c l rnd &easily ostcndcc to other l cvc l s . The ch~racterizotion fqun1it:tivc a t t r i b u t e s , such CIS 'wi th or without ( a certnin)c o l o r t , i s more d i P i i c u l t . A 3ug;cstcd d s p r o ~ c h 5 ) assignsP C ~ I " ~ ~ ~ / N , ~ G. ' I ' I ~alue lVconnot i o n sg to ttributes wI-.erethey suggest themselves ; f o r exarqle .'bri;ht : 202,tflt7t: N S G ~ but 'even: POS' j. ~ h c s c ssignncnts can beexpected t o rcelcct c u l t u r r l d i f f ~ ? ? c n c c d n uncresstandingmetaphor. or t h e resent essmL*le, ven tb; Q .ninFi.mL def ini-tion is d i f l i c u l t , because literal bleaching can bc uone r o rd i f f e r e n t purposes : bleaching ni,ht bc qerccived as rdG.\TIVCin the sense of lrepuvin~ olort, but ,lOaI'.iIVE in th e aense


65/74

of 'launuering or Irer.~ovingstains ' ( 1 - I 3 ( 0 ( IP AT: )i3Z COLuR: G IIowcvcr, t h c rcsulfin~ ~ m b i g u i t yn anymcta;~horicnlintcrnretntion l m g c l y rcflcxtk the ambiguity o fany l i t e r a l use of the verb. In b o t h cases n lcnowlcdge of- -he l i n g u i s t i c or situational context is requ i red 2or ecorrect unc~erstandingof the use of 'blench ( thc bleachedtho repor t of t h c war cnsualticci, blcnchcd th e anecclotes').

Thi.: example points out the acco~aplishmentsand limitsof the ny;tcm in defining components :signif icnnt to metaphor.V h u t i t uoes is to s - - e c i f y structural Er l l s~cwor l cin terms ofwhich those p r o p c r t i eu o f pred ic . t i ve concepts re lev . .n t t oi n e t o p h o r i c ~ lusages ct.n b e n\cthodiccl l ly dcr'incd. In o t h e rwords, the sys tern riistin.guishes the conceptual structurecornponcnt underlying a vcrb form i'rom t h d s e s ernantic a t t r i -butes vhich a r e ;lnon- s ructureu". Tl~us the s tructure unucr-lying ' b l each ' is autmtticnlly specified, as obovc, and3roviJes l ls lo:s la such as PU ~ / KGG f o r the a t t r i b u t e repre-sented by i w l d t e l or 'w i thou t c o l o r ' . The s-,eciEicationsLor quc l i t a t i v e g t t r i b u t e s nust reraain f e : t ib le , subject tothe associations ~rhich c u l t u r c or subculture tlssitns tos'uch a t t r c v u t e s ; the symbolic value of dn ab:traction suchas 'black * i; difE i c u l t t~ .define in i3 gcnerblln nncr, a p a r tf r o m cny c5ntext. In thLs sense t h e r c u u c r i ~ nOF a verb uset o q u a n t i t ~ t i v e r i m i t i v e s IU clualitative attribute-valuesmay.re?re;ent the limit t o which metaphorical imelyses canbe generalized,


66/74

Given t h C l t this system of reproscntntion pro\.uces minimalin te rpre ta t ions wl~crevcrp o s s ib l e , ;is opnosed to the nltcrna-t i v e oL disminsin:: an cxprcssion n s ..nom.~lou::, t r r l . \ n i ~ ~ soconsiuck thc t h i r d c o n d i t i o n l i s t c ~ l bnvc, n mcly tile ade-quocy of the mct l~odas anpl:.:d t~ v \ r bt l l cancc-3ts which .ire- 4clni:.~cd ti- f n l l within- category, i . c . which . ~ c i : ;~icdt h c sal,\e c onL i $ u r . ~ t i u r tof ~ l c s c r i l ~ t o r o . t i~!ssu\\\ed t h a tthe p r i m i t i v e structure unuorlying o verb is ;lw,lys carriedovdr i n LI m~ t nphor i c . i lusage, , : l thou h i t nIy bc magnitude(which is l l o ~ ~ c do r in t e r n ~ s f AWQUKT and IKJENa lTY) orsome other a s ~ e c t f ~ s t $ l e t twhich is not- woviucd f o r ) whichIs cmphcisized. T'*us l e a p t o ' , ' d r i f t m t o t . 'land on1, h i t '

cnd 'plow into o r 'plow through t o (which share the s ~ .m estructure '2nd f e t ~ t u r c ulues except EoP VOL) a l l lend t os i m i l a r i n t e r p r e t t i o n s , given a common g o@ , c.p. prozperity '.All y i e l d ' the beginning ol: a prospcrdus state1;with incor-porati.cn or' the INTEKITY d e s c r i p t o r , ' l eap t o 1 and ' h i t tyield 'sudden beginningt. Trea t i ng (one sense of) ' h i t ' and' leap t o t a s ne;rly synonymous (TR-E & T A X (0 ON R f o r ' h i t ' ,0 AT R for ' l e a p to ) (INTENaITY: > ) (&OLE 0) ), which theyar e not, e n t d i l s some l o s s of information, of course, but theresu l t ing approximat ion i s useLul. In the case of r ~ L o wthrough to , on the other hand, the l a c k of th e in rmationthat a wlaboriows effart" is involvcd weakens the in te rp re -t a t i o n to L~ greater degree; t h i s k in d oi s t y l e , which depends


67/74

on the specific mechanics of the a c t i o n o r the a t t i t u d e oft h e ac tor , is d i f t i c u l t t o i nco rpo ra t e in to o sys temat icchornctcricntion. IIowc\ler, sucfi i n topma tion, independentlydetermined, could be added to t h e ve rb dhsc r i t~ t ion . Forexample, the desc r ip to r tINTIIMbITY:>t appcnclcd t o thefeature v t L l u e +VOLt could be assigned t o thc verb tplow'.These dcocriptors would be cnrr iod over t o the incom>letebut more iMormnt i~e netaphorical ' l i n t c rp re t a t ion t t : 'hevery consciously d i d something t o become prosperous t .

Thus it can be concluded t h a t tlte method prcsentcdcovers a major c lass o f p r e d i c a t i v c concepts, 2nd t h a t th eresult ing opproximi. t ion to the meaning of on expression i sreason..ble but varies i n the amount of information conveyed.It i s oc' sibmificance t h a t t h e emphasis on inclusive classestoge ther with t sp ec i f ic suggested format allows f o r extensionof thc system. Inter~rctationsproduced n the basis of rela-tively rninimd informat ion will not ~ l w a y s e completelysatisfactory, nor will they 7rovide all t h e nuances of l ing-uistic expression. However, the possibility t h a t in terpre-tations of L lerge c l a s s of m e t a p ~ o r i c o l xjressions can beapproximated by a systematic analysis of t l ze concepts i n v o l v e densures f u r t h e r opportunities t o aevelop computer understanciingof novel expressions.


68/74

F1GU"iXE 1OUTPUT INTERPRETATXOXS

Fa mat : < I N 1 )UT 131111ASE~< T Y P E OF hIErrAP1.IOR> ~INI 'SSIZLROLE COSFI(7URATL'OH >r OUTPUT P l S R A l ~ l ~ ~ l AE 3

( 1IF DR INK INK )13A SE RVO( HE DR I N K I N K )

( EIE C I - O SE I N K )( N O I N TGRPRE J ' A T I ON )( H E CLOSB 311ND )I . EVEL-S I I IF I ' R VR( IIE ( IPART: Y I X D ) STOP POSSIl3ILII'Y-OF STAHI' I'HINIC )( s l r IP PLOW SE-4 )F EATURE - S I I I F T AVO( SH I P DO S E A START ( BE S lLZPE: ) )

( SIIIP PLOW CHAIR )( NO I N T E R P R E TA T I ON )

( C I t 4 I R PLO\q SEA )( NO I N T E R P R E TA T I ON )


69/74

F IGURE 1- Continued

( saIr 1nsI~rsam'rERASE OV( SHIP D I S I N T BG lWT E )

( C O U N r R Y LEAP-TO PRO S PER I T Y )1 1 0 - S I IXF I ' 0 11( COUNTRY START 1IAVE-?LATERIAL )

( P RO S P E R I ry D T S I N T E G R A T E )CATEGORY-SITIFI ' OV{ HE STOP I fAVIS -> !ATERIAL )


70/74


71/74

FIGURE 2- Cont inucd

(DISINTEGRATE (PI{) R-L (STATE (0 BE)) (ROLE 0 ) (AGBNT )(0 N R W (BRD 1 ) 1 )

(LEAP-TO ( ( P I I ) TR-E (STATE '0 A T R ) (INTNS ) ( ~ 0 1 . ~)

Date post:	10-Apr-2018
Category:	Documents
Upload:	vikinger
View:	232 times
Download:	0 times

J79 1044 Metaforically Used Verbs

Documents