TABLE OF CONTENTS TABLE Or CONTENTS Testing the Psychological Reality of a Represcntatioh Model

A r e s e a r c h program is desc r ibed i n which a p a r t i c u l a r r e p r e s e n t a t t o n a l format f o r meaning is t e s t e d as broad ly as p o s s i b l e . I n t h i s format, developed by t h e LNR r e s e a r c h group a t The U n i v e r s i t y of C a l i f o r n i a a t San Diego, v e r b s arg r e p r e s e n t e d as i n t e r c o n n e c t e d sets o f s u b p r e d i c a t e s . t h e s e s u b p r e d i c a t e s may be thought of as t h e a lmost i n e v i t a b l e i n f e r e n c e s t h a t a l i s t e n e r makes when a verb is used I n a sen tence . They c o n f e r a meaning s t r u c t u r e on t h e sen tence i n which t h e v e r b is used. To be p s y c h o l o g i c a l l y v a l i d , thege r e p r e s e n t a t i o n s should c a p t u r e ( a t least) 1 S i m i l a r i t y o f meaning The more similar two verbs seem i n meaning t o people , t h e more t h e i r r e p r e s e n t a t i o n s should over lap , 2 C o n f u s a b U i t y The more confusab le two verb meanings are, t h e more t h e i r r e p r e s e n t a t i o n s should over lap . 3. Memory f o r s e n t e n c e s c a n t a i n l n g t h e ve rb ?he s e n t e n c e s t r u c t u r e s set up by t h e v e r b ' s meaning should i n p a r t de termine t h e way i n which s e n t e n c e s a r e remembered. 4. Semantic i n t e g r a t i o n The r e p r e s e n t a t i o n s should a l l o w for t h e i n t e g r a t i o n o f in fo rmat ion from d i f f e r e n t s e n t e n c e s i n t o d i s c o u r s e st ructure 5 Acquis i t ion p a t t e r n s l h e s t r u c t u r a l p a r t i t i o n s i n t h e r e p r e s e n t a t i o n s should correspond t o t h e s t r u c t u r e s c h i l d r e n a c q u i r e when they are l e a r n i n g t h e meanings of t h e ve rbs 6. P a t t e r n s o f e x t e n s i o n The r e p r e s e n t a t i o n s should be e x t e n d i b l e s o as t o ref lect t h e ways i n which people i n t e r p r e t verb meanings when t h e v e r b s are used o u t s i d e t h e i r normal context, 7. React ion times The t i m e t aken t o comprehend a sen tence u s i n g a g iven v e r b should reflect t h e s t r u c t u r a l complexi ty of t h e v e r b meaning. Experiments concerned w i t h p r e d i c t i o n s 1-5 are desc r ibed here . The r e s u l t s are promising f o r a g e n e r a l approaoh of r e p r e s e n t a t i o n o f meaning i n t e rms of i n t e r r e l a t e d s u b p r e d i c a t e s , b u t do n o t c l e a r l y d i s t i n g u i s h between s e v e r a l similar r e p r e s e n t a t i o n s . For example, t o test p r e d i c t i o n (21, I read people s e n t e n ~ e s c o n t a i n i n g v e r b s w i t h similar meanings, and asked them t o recall t h e sen tences . The deb, =e of o v e r l a p i n t h e semant ic s t r u c t u r e s was a good p r e d i c t o r of t h e number of confus ions between sen tences . I n a n o t h e r sentence-memory experiment ( p r e d i c t i o n ( 3 ) ) , s e m a n t i c a l l y compl& v e r b s t h a t provided more under ly ing i n t e r c o n n e c t i o n s bqtween t h e nouns i n a sentence lea t o b e t t e r memory f o r t h e nouns i n t h e s e n t e n c e t h a n s imple g e n e r e v e r b s , o r than o t h e r complex v e r b s t h a t d i d no t provide such e x t r a i n t e r c o n n e c t i o n s , To test p r e d i c t i o n ( 5 ) , I tested c h i l d r e n ' s oomprehension of a set o f p o s s e s s i o n ve rbs . Both t h e o r d e r o f a c q u i s i t i o n among t h e v e r b s and t h e k inds of e r r o r s f i t t e d well wsth an account of t h e a c q u i s i t i o n o f v e r b meaning i n t e rms o f i n t e r c o n n e c t e d subpred ica tes . T h i s r e s e a r c h I l l u s t r a t e s a b r e a d t h f i r s t approach t o t e s t i n g a r e p r e s e n t a t i o n . I n t h e b r e a d t h f i r s t approach, many d i f f e r e n t psycho log ica l p r e d i c t i o n s are made. Each d i f f e r e n t a r e a o f p r e d i c t i o n r e q u i r e s a set of' proceus gsaumptions, and i n each case t h e p r o c e s s assumptions used are t h o s e t h a t seem most p l a u s i b l e g i v e n p rev ious r e s e a r c h i n t h e f i e l d . If one r e p r e s e n t a t i o n a l format can make c o r r e c t p r e d i c t i o n s about a number of d i f f e ~ e n t k i n d s o f paychological phenomena, t h e n t h a t r e p r e s e n t a t i o n s t a n d s a g r e a t e r chance of being g e n e r a l l y u s e f u l t h a n one which was t e s t e d i n on ly one d e p t h f i r s t way. T h i s paper d e s c r i b e s a p r o g r m of r e sea rch t h a t b e s t s a r e p r e s e n t a t i o n a l format for ve rb meaning. T h i s reseacoh grew ou t of t h e LNR (Footnote 1 ) a t tempt fa t h e r e p r e s e n t the meanings of wdrds i n a psychologica l ly s a t i s f y i n g way. Verb meaning seemed a n a t u r a l p l ace t o start for t w ~ reasons: (1 ) verbs a r e important: it is a rguab le Chat" they provide t h e c e n t r a l b rganiz ing semant ic s t r y c t u ~ e s i n sen tence meanings; and (21 ve rbs a r e t r a c t a b l e : their meanings are more e a s i l y analyzed than thoo f , for example, common nOURs. S i n c e d i f f e r e n t d i s c i p l i n e s look af! meaning i n d i f f e r e n t ways, it may be w ~ r t h w h i l e t o d e s c r i b e the s t a n c e we took, What we wanted was a system of r e p m s e n t a t i o n i n which we could c a p t u r e ou r i n t u i t i o n s about what a word t y p i c a l l y conveys; o r more s p e c i f i q a l l y ab6ut t h e i n f e r e n c e s a person normally makes (or b e l i e v e s should be m e) "f: when a word is used. The assumption is t a t t h e same r e p r e s e n t a t i o n s o p e r a t e when a person u s e s t h e word i n speech as when t h e person comprehends it; however t h e methodology o f experiments). psychology makes it n a t u r q l to spend more time pondering t h e i n p u t process than t h e ou tput process . T h i s approach d i f f e r s from th ink ing o f meaning i n terms of necessary and s u f f i c i e n t t ru th -cond i t i ons , as many p h i b s o p h e r s have done, o r from th ink ing about meaning i n gene ra t ion r a t h e r than i n comprehension, as many l i n g u i s t s have done. Each of those s t a n c e s l e a d s t o u s e f u l i n t u i t i o n s . Ove ra l l , t h e r e has been a r e a s s u r i n g degree of convergence between f h e r e p r e s e n t a t i o n s proposed. of There are many n o t a t i o n a l systems f o r r e p r e s e n t a t i o n , of verb meaning ( e .g . , Abrapamson, 1975, Chafe, 1970; F i l lmore , 1971, Gentner, 1975, Lakoff , 1970; McCawley, 1968, Rumelhart & Levin, 1975; Schank, 1972, 1975, Talmy, 1975). These models of verb meaning d i f f e r from One ano the r in d e t a i l , bu t t h e r e is widespread agreement on t h e i d e a t h a t vero n~eadings can be represen ted i n terms of i n t e r p a a t e d sets of. subpred ica t e s , such a s CAUSE, cm CHANGE. These subpred ica t e s are n o t merely ooncatenated wi th in a word * s rep resen ta t ion . Rather, they are i n t e r r e l a t e d , i n s p e c i f i c ways. Representa t ions of verb meaning include n o t a t i o n f o r spec i fy ing t h e r e l a t i o n s h i p s among t h e subpred ica t e s t h a t make up a wordts meaning. The no ta t ion developed by t h e LNR Group is a network format, I n t h i s system o f r e p r e s e n t a t i o n , verb meanings a r e expressed i n terms o f aubpredioa tes t h a t s tand f o r states, changes of s t a t e , a c t i o n a l s , etc, E l v t s of Verb -. Verbs provide a system i n whiah people can t a l k about happenings i n t h e world, i m p l i c i t l y d i s t i n g u i s h i n g s e v e r a l t y p e s o f conceptual p o s s i b i l j t i e s . The s imp les t of t hese is the w. k s t a t i v e p r e d i c a t e convqys a r e l a t i o n s h i p t h a t endures f o r a period of time between two arguments, normally an o b j e c t ( o r person) and an o b j e c t o r value wzthin t h e conceptua l f i e l d speczf ied by the s t a t i v e . For example, cons ide r t h e s en tence shown i n F igu re 1. I d a owned a Cadillac from 1970 t o 1977. The ve rb phm conveys t h a t a r e l a t i onah4p of p e s s e s s i o s kxisted between Ida and t h e Cadillac far some d ~ r a t i o n . Bes ides s t a t i v e s f o r possess ioq , t h e r e a r e a large number of o t h e r s t a t i v e s , inc lud ing l a c a t i o n ($0 be &, , etc.) and emotion ($0 hat&, I n a d d i t i o n t o s imple s t a t i v e r e l a t i o n s h i p s , ve rbs can be used .to co vey changes bP s t a t e . Following Chafe ( 1910y 1 w i l l refer t o a change of state a s a m, For example, the sen tence I d a r e q s i v e s $1 0.00. t e l l s u s ( 1 ) that Ida now has $10.00 ( 2 ) t h a t someone else had the $10.00 before , ( 3 ) t h a t a change has taken p l a c e from t h i s p rev ious state o f possess ion t o t h e p r e s e n t s t a t e . More commonly, verbs e x p r e s s not s imple changes o f state but c a u s a l changes of state. We seem t o be very i n t e r e s t e d i n p roces ses That are v o l i t i o n a l l y caused by humans and cther s e n t i e n t be in s. Figure 2 shows t h e r e p r e s e n t a t i o n of t ! e sen tence I d a g i v e s Sam a rose. An a g e n t nay cause a change of s t a t e t h a t re la tes t o ano the r ob jec t . O r t he same person may aot on both agent and evpe r i ence r of t h e change o f state. The l o e a t i o n a l verb move can be used i n either way, as i n t h e fo l lowing examples a . I d a moved t h e c a r , b. I d a moved t o the f r o n t seat. I n both these c a s e s the a c t i o n taken by Ida 1s unspec i f i ed . We o f t e n dont t c a r e exac t ly what someone d i d t o cause some p roces s t o ocqur, However, t h e r e are also ve rbs i n which t h e c a u s a l a c t i o n is p a r t r a l l y or wholly s p e c i f i e d egg-, walk? saunter, lllamkL, strrde, ua, sarint., taGa, iCrat, m. (See Miller ( 1972) and Mlller & Jonnson-Laird (1976) f o r a more ex t ens ive d i s c u s s i o n of t h e v e r b s of l o c a t i o n , ) Thus, t h i s system a l lows f o r t h e r e p r e s e n t a t i o n o f verbs as states, changes o f s ta te , c a u s a l changes of st%&e. simple a c t i o n s , and complex oases i n w m s p e c i f i c a c t i o n s cause changes of s t a t e . F u r t h e r d i s c u s s i o n of t h e LNR s y s b m of verb semant ics can be found i n the articles by Abrahamson, Gentner, Munro, Rumelhart & Levin, and Rumelhart & Norman i n t h e Norman & Rumelhart (1975) volume. There aPe c e r t a i n l y g a p s i n t h e system, and a s p e c t s of ve rb meaning that are no t e x p r e s s i b l e i n t h i s s imple vocabulary. Some unresolved i s s u e s are d i scussed l a t e r i n t h e paper. However, t he system seems p l a u s i b l e p t the fLrst level, and allows a fa i r range of ve rb meanings t o be cap tu red at l e a s t roughly. A t t h i s point i n the research it amfed appropriate to begin testing the psychological rightness of the system as so far stated before going on Lo refine it. One advantage of psychological experimentat ion ( or of cbmpu t er implementation) is that it fbrcos one to make expl ici t the assuroycions ' underlying representation and process. A t l eas t some of the ahoices made can then be tested as hypotheses. Some important assumptiorls are ( 1 ) a verb's representation captures the set of immediate inferences that people normally make when they hear or read a m t e n c e containing t h e verb; [Z) i n general, one verbleads to many inferences ( 3 ) these networks of meaning components are accessed during comprehension, by an immediate and largely automatic process (4.) the se t of components assocLated w i t h a given word is reasonably s table across task$ and cbntexts (5) surface memory for exact words fades quite rapidly, so that a f te r a short time, o n l y the representational network remains. In test ing these representations, I took a very l i t e r a l f nterpretation of -the not ion of representation -namely that the nodes and arrows i n a r ep rescn ta t i~n correspond to the concepts and relatibnships that are stored when a person comprehends a sentence containing a verb, The mare ferociously l i t e r a l the intebpretation, the better the chances o f discovering counter-evidence. -One paychologica1 cr i ter ion i g t h a t the representations should agree w i t h peoplets intui t ive notions of synonymity and Bimilarity i n meaning. One straightforward measure of t h i s overlap is the degree to which people rate verbs as similar i n meaning. I n a study of about 60 selected verbs, I found that people's average rating of the semantic similari ty between two verbs agreed very closely with the degree of semantic overlap between the i r representations, A more subtle measure of psychological similarity is the degree to which people unconsciously confuse things i n memory. People i n a sentence-memory experiment probably try t o keep their sentence traces clear. B u t , suppose that w i t h i n a short time a f t e r hearing a verb i n a sentence, a person has only the representational network of concepts and relationships, and not the surface verb Assume further that some pieces of the memory representation may be los t or unaccessible a t any time 6the t ' f a l l ib i l i ty of human memoryft assumption) . Then the more two Verb representat ions overlap, the more like1 y it is that sentences containing the two ver6d w i l l be confused i n demory, despite people's attempts to keep them straight . In an experiment i n sentence memory, using verbs of varying semantic overlap, I found that subjects d i d indeed confuse the verbs i n exactly the way predicted by the theory (Gentner , 1974 ) . The correlation between the number of confuaions aubjeota made .between two verbs and the semantic overlap between the verbs, as predicated from the representations, was quite high. In faot , the correlation between representational overlap and number of confusions was s l ight ly higher ( though not significantly so) than the c o r ~ e l a t i o n between the nymber of confuadons and the rated similari ty between the vorbs. (The siaiiarity ratings were taken from the first-mentioned s t u d y , wi th a different se t of subjects). . Semantic oorppl exi t y refers to the number of underlying sub predicates and intercohnec t ion8 t h & e up the basic meadlng of a verb. More complex meaqings correspond to more specific actions or events. For exgmple, &j&+ is more specific than gp, Its meaning contains more subpredicates, We know more havirlg heard sentence ( a ) than sgntenoe ( b) . ( a ) Ida strode across the f i e ld . ( b 3 Ida went across the f ie ld, Various researchers have looked for evidence t h a t semantic complexity may affect comprehensibility ,, generally on the a sumption e that more complet semantic s t ruc t re8 are harder to process (Kintsch I Thorndyke, 1977) a However, the reaul ts have been negative. There is no evidence that more complex words lead ei ther to longer reaction-times or ts greater processing loads than do simpler words. . I believe' t%at itt s incorrect to assume accoss the Wrd that complexity is psycholsgically hard. Some research o f mine suggests that the ef fec ts of semantic complex1 t y i memory are mare particular. l e w d Co. Although the view that semantic complexity leads to diff icul ty has not been suppor4ted, there i s another side t o the complexity-issue. The , additional semantic components in a complex verb may sct up additional connections among the nouns in the sentence. I n t h i s case, more complex verbs should lead to a richer and more highly interwoven sentence representation, and thus to he_t_ter. memory for the nouns i n the sentence. Notice that t h i s prediction derives fFom a fanatically l i t e r a l interpretation of the verb representations: more paths i n the representation means more conceptual paths in memory. This prediction is quite apecifio. It is not simply a question of certain complex versus simple verbs having some overall effect , but rather of complex verbs providing extra connections between the particular noung i n question. T h i s is clearly true for Ida and her tenants i n the case of versus prive, as can be seen in Fig 3a and 3b. L tested for t h i s kind of improvement i n connectivity in a ser ies of experiments in sentence memory (Gentner, 1977). 1 read people skntenaes that differed i n the semantie o o n n e c t i v i t y o f t h e i r v e r b s , such as t h e fo l lowing p a i r o f s e n t e n c e s * I d a gave h e r t e n b n t s a c lock . ( s i m p l e ) I d a s o l d h e r t e n a n t b a c lock . (complqt connec t ive ) Then I gave t h e people t h e names o f t h e c h a r a c t e r s and asked thdm to recall we s e n t e n c e s . A s p r e d i c t e d , t h e y wer@ b e t t e r a b l e t o recall t h e noun tenants when t h e complex c o n n e c t i v e v e r b was used then when t h e s i m p l e v e r b aiPe was used. More semant ic c o n n e c t b n s between tKe two nouns l e d t o s t r o n g e r memory connec t ions . To see t h e s p e c i f i c i t y of t h e p r e d i c t i o n , c o n s i d e r a complex v e r b f h a t merelry a m p l i f i e r t h e s imple v e r b and dues I1P;rc. add c o n r l e c t i o n ~ between t h e ke nquns. For example, t he ve r l ( F i g 3cr adds t h e infopmat ion t h a t t h e method of t r a n s f e r was,& m a i l i n g o r some such long-das tance t r a n s f e r . Using mail l e a d s t o more i n f e r e n c e s ( a more s p e c i f i c even t d e s c r i p t i o n ) t h a n u s i n g a, However, the knowledge t h a t t h e o b j e c t was mai led l e a d s U few, i f any, a d d i t i o n a l c o n n e c t i o n s betwpexv t h e a g e n t , m, and t h e r e c i p i e n t , tenants, T h e r e f o r e , t h e p r e d i c b t o q was t h a t use o f such non-connecting s p e c i f i c v s r b s would l e a d t o no improvement over u s e or g e n e r a l v e r b s i n rnemgry between thenourfs, The r e s u l t s were e x a c t l y as p r e d i c t e d The o b j e c t nouns of complex c o n n e c t i v e v e r b s were recalled b e t t e r t h a n W o s e of g e n e r a l v e r b s and non-connecting complex ve rbs . These d i f f e r e p o e s were n o t t r a c e a b l e t o d i f f e r e n c e s i n imagery o r wordf requency . Thus c o n n e c t i v i t y is b e n e f i c i d l t o s e n t e n c e metilory i n a ve ry s p e o i f i c way. . There may be a more d i r e c t r e l a t i o n s h i p between complexi ty and d i f f i c u l t y i n c h i l d r e n t h a n i n a d u l t s . Young c h i l d r e n o f t e n f a i l t o comprehend the f u l l meanings of s e m a n t i c a l l y complex terms ( e ,g . , Bowerman, 1975, C la rk , 1973, Gen tner , 1975, i n p r e s s ) . Working w i t h t h e v e r b s o f p o s s e s s i o n , I have observed t h a t c h i l d r g n act o u t t h e s imple v e r b s &gg and tc\ke c o r r e c t l y b e f o r e they a c t o u t t h e more complex v e r b s SLYJL and trade. S t i l l later they l e a r n t h e y e t more complex v e r b s &, sell and .The o r d e r i n which t h e v e r b s are l e a r n e d is e x a c t l y t h e o r d e r o f i n c r e a s i n g semant ic complexi ty . T h i s complex i t o r d e r i n g can be made q u i t e p r e c i s e , s i n c e t b e v e r b s are c l o s e l y r e l a t e d i n meaning. The r e p r e s e n t a t i o n of a v e r b a t t h e n t h l e v e l o f s i m p l i c i t y Is p r o p e r l y n e s t e d w i t h i n t h e r e p r e s e n t a t i o n of a v e r b a t t h e ( n + l ) t h l e v e l . F u r t h e r , when c h i l d r e n around 4-6 y e a r s are agked t o act o u t (as i n *Make E r n i e se l l Bert a boat.") they act o ~ t nive. i n s t e a d ( A boa t is t r a n s f e r r e d from Ern& t o B e r t ) . S i m i l a r l y , ~JUY is a c t e d o u t ad w. They s y s t e m a t i c a l l y act o u t complex v e r b s l i k e s imple v e r b s ; and more s u r p r i s i n g l y , t h e y chpose t h e a p p r o p r i a t e s imple ve rb . My i n t e r p r e t a t i o n , c o n s i s t e n t wi th C l a r k ' s ( 1973) semant ic f e a t u r e s a n a l y s i s , is t h a t ' c h i l d r e n l e a r n t h e s e complex v e r b meanings g r a d u a l l y , by add ing components to t h e i r p a r t i a l l y correct r e p r e s e n t a t i o n s . k t any g i v e n M e , t h e ch i ld comprehends Language i n terms o f t h e oomponents t h a t h% has so far acqu i red . t ic * Another impor tant p s y c h o l o g i c a l requirement is combinab i l i ty . The b a s i c no t ions 'o f s t a t e , change o f s ta te , bauae, and so on must be combinable i n t o networks larger than t h e i n d i v i d u a l sentence . When two v e r b s s h a r e p a r t s o f t h e i r under ly ing s t r u c t u r e . t h i s redundanpy shou ld be u t i l i z e d t o comhane t h e two r e p r e s e n t a t i o n s i n t o one d i a c o b r s e s t r u c t u r e . How can we test whether t h i s happens? One way is t o a r r a n g e t h i n g s s o t h a t c o l l a p s i n g t h e redundenc ies between two v e r b s shou ld create t h e r e p r e s e n t a t i o n o f a t h i r d verb . Then t h e p r e d i c t i o n is t h a t peop le shou ld u s a t h i s t h i r d v e r b i n recall. I n a s t u d y of semant ic i n t e g r a t i o n , I r e a d peop le s h o r t passages and t e s t e d t h e i r memory by having them f i l l i n b l a n k s (Gentner , 1978). Every passage c o n t a i n e d a general v e r b , sbch as g&g. Hal f t h e passages a l s o c o n t g i n e d a d d i t i o n a l semant ic In fo rmat ion , such a s t h e f a c t t h a t t h e g i v e r a c t u a l l y t h e money he was g i v i n g . A c ~ o r d i n g t o t h e r e p r e s e c l t a t i o n a l model, t h e i n t e g r a t i o n of t h e r e p r e s e n t a t i o n of a w i t h t h a t o r a shouJd have created t h e s t r u c t u r e o f u. If what peop le havb 1a -t.nei,r minds af ter h e a r i n g t h e v e r b s is t h e network r e p r e s e n t a t i o n s , and if t h e s e r e p r e s e n t a t i o n s are r n t e g r a t e d d u r i n g d i s c o u r s e cojnprehension, t h e n people who heard nF\re and should end up w i t h t h e r e p r e s e n t a t i o n of w. A s p r e d i c t e d , s u b j e c t s F e a r i n g t h e e x t r a material f a l s e l y r e c a l l e d t h e v e r b which best f i t t h e aompos i t e s t r u c t u r e (e .g . & r a t h e r t h a n t h e v e r b a c t u a l l y p resen ted . I have a d e t h e assumpt ion t h a t a v e r b carries w i t h it a se t of i n f e r e n c e s t h a t a r e normal ly made d u r i n g comprehension, a s well as s e v e r a l s u p p o r t i n g assumpt ions . T h i s view h a s been f a i r l y well suppor ted by t h e r e s e a r c h p r e s e n t e d h e r e , b u t n e v e r t h e l e s s it seems to me a n ' overs impl i f i ca t ion . There remain a great many q u e s t i o n s , some l a r g e and some smal l . ( 1 ) Where shou ld t h e l i n e be drawn around a word's meaning? A s C la rk and Cla rk (1977) have p u t it, is word meaning more l i k e a d i c t i o n a r y o r an encyclopedia? The extreme o f t h e d i c t i o n a r y approach would be t o t a k e a minimal c o n t r a s t approach, s t o r i n g wi th a word o n l y enough to d i s t i n g a s h 4 t from a l l o t h e r words. Tmhe extreme o f t h e encyc loped ia approach would be to access t h e e n t i r e long-term memory whenewer any word is used. The q u e s t i o n is, how t o d e f i n e a r e a s o n a b l e middle ground. ( 2 ) What $a t h e p r o c e s s o f expans ion i n t o a semant ic r e p r e s e n t a t i o n d u r i n g comprehension? a) Are t h e r e i n v a r i a b l e inferences?. When a n incoming w o ~ d is processed , is t h e r e a set o f i n f e r e n c e s (auch a s t h e set I have c a l l e d t h e w a l m o s t i n e v i t a b l e i n f e r e n c e s n t h a t is a lways made duri .ng comprehension, o r i g t h e r e v a r i a t i o n i n which i n f e r e n c e s get made? b) I f there is v a r i a t i o n , is it q u a n t i t a t i v e o r q u a l i t a t i v e ? Do con tex t and t h e pe r son ' s i n t e r e s t s and a t t e n t i o n determine which in fe rences g e t made, so t h a t t h e r e are q u a l i t a t i v e * d i f f e r e n c e s i n what i n f e r e n c e s get made? O r is t h e d i f f e r e n c e merely q u a n t i t a t i v e , w i t h t h e r a d i u s o f expansion varying wi th t h e amount o f a t t e n t i o n (o_r energy, o r i n t e r e s t ) t h a t the persorf b r ings t o bear? The not ion of at least q u a n t i t a t i v e v a r i a t i o n a seems haFd t o avoid. It is a f a i r l y s t r o n g i n t u i t i o n t h a t we process word meanings with varying degrees of energy Fur the r , t h e phenomenon of instantiation (Anderson, R,C., Stevens , K.C., S h i f r i n , Z., & m o r n , J . , 1977 ) makes it clear t h a t a model o f sentence comprehension must a l low f o r q u a l i t a t i v e d i f f e r e n c e s i n t h e f i n a l set of in fe rences s t o r e d . For example, compare t h e sen tences Rover a t e h i s d inner , Mr. P r i t c h w d a t e his dinner , T%e verb P;Bf; conveys v a e t l y d i f f e r e n t a c t i o n sequences when used with d i f f e r e n t q e n t s , though its c a u s a l change-of-state s t r u c t u r e remains more-or-less cons tant . It is poss ib le t h a t t h i s q u a l i t a t i v e v a r i a t i o n can be accounted for by simpZe underlying q u a n t i t a t i v e processes spreading a c t i v a t i o n . We may have t o settle for a more complex model, i n which some p a r t s of a ve rb ' s meaning are almost always accessed while other in fe rences dcvelop o u t o f the i n t e r a c t i o n of t h e verb wi th its c o n t e x t , inc lud ing its pragmatic context . Id H e w i t t t s (197b) terms, there may be both if-added i n f e r e n c e s and if-needed inferences . Where i n t h i s model (and whether) we want t o draw a l i n e b e t ~ e e n meaning and knowledge-of-the-world is not a t a l l clear t o me, ( 3 ) Carrying t h e noticul of v a r i a b l e y e r b meanbng sti&l f u r t h e r , how does metaphorical extens ion wark? Most common verbs can be used i n s e v e r a l r e l a t e d ways. For example, cons ider t h e range bf meanings t h a t nlv_e can convey depending on t h e nbuns it is used w i t h a r o s e a job. an h e i r . Ida gave Sam an excuse a t a l k i n g to . a l l h i s b e s t i d e a s , t h e time o f his l ife. Clea r ly t h e subpredica te s t r u c t u r e varres aoross thew sentences , so much s o t h a t some might want to desc r ibe t h i s as a w l l e c t i o n of e n t i r e l y d i f f e r e n t aenises of t h e same word. This misses t h e s t r u c t u r a l similarities. Some kind -of metaphorical extens ion of meaning seems a necessary p e r t of a theory o f verb meaning, s i n c e it is g e n e r a l l y t h e verb t h a t does most of t h e a d j w t i n g . A series of s t u d i e s by Albert Stevens and me sugges t8 tha t people faced with an odd aentence assume t h a t some of t h e subpred ica tes normally conveyed by the verb are not meant t o apply i n t h e sen tence at hand A c u r r e n t projeot i a t o m o W t h e r u l e s for which subprediodtes apply i n d i f f e r e n t contexts . 4 X have so f a r t r e a t e d nouns as nodes i n the semantic r ep resen ta t ion . C lea r ly i n o r d e r t o analyze sen tedce i n t e r a c t i o n s i t is necessary to have a r e p r e s e n t a t i o n of noun meaning. Some progress been made with a b s t r a c t nouns, such a s k inahip terms. But t h e t r u l y nounlike nouns ---basic-level nouns--resist analysis. 1 b e l i e v e t h a t t h e s e d i f f e r e n c e s i n amendabil i ty to a n a l y s i s r e f l e c t d i f f e r e n c e s i n t h e kind of meaning t h a t verbs and nouns have, and t h a t a us&fu l r e p r e s e n t a t i o n of concre te noun meaning may be q u i t e d i f f e r e n t from t h a t used f o r verbs , p repos i t ions and even a b s t r a c t nouns. (5) There are s e v e r a l a s p e c t s of t h e r e p r e s e n t a t i o n a l scheme t h a t need f u r t h e r thought. To s i n g l e ou t one i s s u e , cons ider t h e not ion of change of s t a t e . The LNR r e p r e s e n t a t i o n r e p r e s e n t s a verb l i k e gat as conveying a change froas an i n i t i a l s t a t e of possession to a f i n a l s t a t e of possession. Schank's Conceptual Dependency theory would r e p r e s e n t t h e e n t i r e sequence 8s a p r i m i t i v e a c t . Many g e n e r a t i v e s e m a n t i c i s t s have represented only t h e inchoa t ive p a r t of t h e chain ( t h e change to t h e f i n a l s t a t e ) as belonging t o the a s s e r t i o n of t h e verb , consFdering t h e i n i t i a l s t a t e to be more i n t h e na tu re of a @resupposi t ign (e.g. Filbmore, 1966). All these p o s i t i o n s seem to me t o have merit. The LNR use of change from i n i t i a l t o f i na l s t a t e allows a chang;e-of-state verb t o hook a u t o q a t i c a l l p with r e l e v a n t state information. The use of acts as p r i m i t i v e s c a p t u r e s t h e psychological wholeness of change. Thq use of t h e inchoa t ive captrrres t h e i n t u i t i o n t h a t people seem more i n t e r e g t e d i n t h e r e s u l t s o f an even t --i.e. i n t h e f i n a l state-than i n t h e s e t t i n g state. The e x p l i c i t change-of-state formats (LNR format and inchoa t ive format) have a n a t u r a l way of capturing some k inds o f metaphorical ex tens ion by substFtuLLng a d i f f e r e n t s t a t i v e rese rv ing t t e rest of t h e v e r b ' s s t r u c t u L . Summarv This work is j u s t beginning. Nei ther t h e r e p r e e e n t a t i o n s nor t h e p rocesses t h a t a r e assumed t b o p e r a t e on them come very c l o s e to c a p t u r i p t h e s u b t l e t y of human language use. S t i l l , t h e r e s u l t s of t h e experimental inves t i .ga t ion a r e promising some kind of decompositional model along t h e s e l i n e s . Ido mailed her tenanls a clack ................ Ida Cadi l lac 197