J Psycholinguist Res (2013) 42:339347 DOI 10.

1007/s10936-012-9214-6

For a New Look at Lexical Errors: Evidence from Semantic Approximations with Verbs in Aphasia
Karine Duvignau Thi Mai Tran Mlanie Manchon

Published online: 10 May 2012 Springer Science+Business Media, LLC 2012

Abstract The ability to understand the similarity between two phenomena is fundamental for humans. Designated by the term analogy in psychology, this ability plays a role in the categorization of phenomena in the world and in the organisation of the linguistic system. The use of analogy in language often results in non-standard utterances, particularly in speakers with aphasia. These non-standard utterances are almost always studied in a nominal context and considered as errors. We propose a study of the verbal lexicon and present ndings that measure, by an action-video naming task, the importance of verb-based non-standard utterances made by 17 speakers with aphasia (la dame dshabille lorange/the lady undresses the orange, elle casse la tomate/she breaks the tomato). The rst results we have obtained allow us to consider these type of utterances from a new perspective: we propose to eliminate the label of error, suggesting that they may be viewed as semantic approximations based upon a relationship of inter-domain synonymy and are ingrained in the heart of the lexical system. Keywords Aphasia Categorization Lexicon Semantic approximations Verbs

Introduction Studies that have been conducted on the pathology of aphasia have principally focussed on lexical production difculties, due to their frequency and presence in all types of clinical
K. Duvignau (B ) Laboratoire CLLE-ERSS, Maison de la Recherche, Universit Toulouse 2, 5, Alles A. Machado, 31058 Toulouse, France e-mail: duvignau@univ-tlse2.fr T. M. Tran Laboratoire STL, Universit Lille, Nord de France, 59000 Lille, France e-mail: thimai.tran@univ-lille2.fr M. Manchon Laboratory for Cognitive and Neurological Sciences (LCNS), Neurology Department of Medicine, University of Fribourg & Hopital Cantonal, Chemin des pensionnats 2-6, CH-1700 Fribourg, Switzerland e-mail: melanie.manchon@gmail.com

123

340

J Psycholinguist Res (2013) 42:339347

aphasia. The terms typically used to designate these troubles are anomia, or lexical disorder (Goodglass and Wingeld 1997). Manifestations of this disorder are varied, and are often accompanied by modalized utterances that mark the awareness of a lexical difculty. The study of these lexical production disorders is principally done in a clinical context, by image-naming tasks (Khon and Goodglass 1985), and typically focus on the nominal lexicon (the naming of objects or natural categories). With regard to the semantic categorisation of the lexicon, two aspects are always privileged: the study of the categorisation of the nominal lexicon (Kim and Thompson 2004) and the analysis of non-standard utterances as errors (Kemmerer and Tranel 2000; Kemmerer 2003; Kemmerer and Gonzalez-Castillo 2010; Wells-Jensen 2007; Bormann et al. 2008; Hodgson and Lambon Ralph 2008). This theoretical viewpoint does not explain the dynamics of language production and does not take into account linguistic sources that may come into play in study of non-standard lexical productions (Le Dorze and Nespoulous 1989; Kittredge et al. 2006; Wunderlich 2006). Furthermore, although there is an increasing number of neuroanatomical or neurolinguistic studies exploring the structuring of the verbal lexicon (Piras and Marangolo 2007; Liljestrm et al. 2008; Bedny and Caramazza 2011), they still remain in a nominal or syntactic perspective (Deloche et al. 1996; Thompson et al. 2003; Thompson and Shapiro 2007; Hillis et al. 2006), as studies that have shed an innovative light on non-standard utterances by aphasics focus almost exclusively on the nominal lexicon and promote the notion of error (Caramazza and Hillis 1990; Kittredge et al. 2006; Hillis et al. 2006; Bormann et al. 2008; Cloutman et al. 2009). Therefore, very few studies consider the structuring of the mental lexicon with regard to the verbal system Raymer and Ellsworth (2002); Wambaugh and Ferguson (2007), and, presently, there are very few data available on verbal utterances such as: la dame casse la carrotte/the lady breaks the carrot, la dame dgraphe la carotte/the lady undoes the carrot, elle dshabille lorange/she undresses the orange. In this context, we propose a study of the production of these verb-based utterances made by aphasic speakers. In this paper, we dont consider status of these non-standard productions which can be: overextension (related to a lexical deciency) metaphor (no lexical deciency and deliberate production) or access problem to a conventionnal target word. We consider these utterances as semantic approximations and propose the following linguistic characteristics: (i) Intra-domain verbal semantic approximations : Clara elle secoue la soupe Clara shakes up the soup / context: stirring the soup with a spoon la dame coupe lorange the lady cuts the orange / context: a woman removes the peel from an orange

In this category, the approximate nature of the verb stems from a non-correspondence between the verb that is used and the reality to which it refers. This is the case when the use of the verb in another context in which it would be appropriate (respectively, shaking up a bowl of soup and cutting an orange in two with a knife) does not provoke any semantic tension between the two terms in question (to shake up can be said in reference to a soup / to cut can be said in reference to an orange), but refers to, in the context in question, a way of carrying out the action that does not specifically correspond, in the real world, to the action that is being performed. This is what we call a pragmatic approximation.

123

J Psycholinguist Res (2013) 42:339347

341

(ii)

Inter-domain verbal semantic approximations: elle dshabille lorange , she undresses the orange / about a whoman peeling an orange Elle casse le livre , she breaks the book / about a book that has been torn

In this category, the utterance may have the status of an overextension or a metaphorical gure. The utterance presents the following linguistic characteristics: the approximative verb (underlined) used by the speaker is in an inter-domain synonymic relationship with possible conventional verbs with which it shares the same hyperonym. Therefore, for example, the verb above to undress, is an approximate verb that is in a synonymous inter-domain relationship with the conventional verb to peel in the action video [To peel_orange]. The verb to undress is in a co-hyponymous relationship with to peel because both refer to the hyperonym /to take off/ and in an inter-domain relationship with to peel as it belongs to a different semantic domain [DOMAIN OF TEXTILECLOTHING] than that which is demonstrated in the context of the action [DOMAIN OF FRUITORANGE]. Because of this, inter-domain semantic approximation can be spotted independently of the context of its utterance and its detection is made on the semantic and linguistic levels. Given the nature of these data, we wish to challenge the label of error for these utterances and demonstrate their importance for the aphasic speaker.

Methods Participants 17 monolingual, native French-speaking aphasics, from 40 to 82 years old (average: 61 years old) Level of education: average

The aphasic population for the study was chosen by the following criteria: We selected aphasic patients, currently being treated in language rehabilitation, who present varied clinical pattern following central cerebral lesions. All patients demonstrate difculties with lexicalisation. The patients who were selected demonstrated the abilities required to carry out the experimental task (sufcient expression and comprehension of the instructions, absence of visual or gnostic impairment, absence of attention deficit disorders). We excluded: patients demonstrating troubles with comprehension that rendered instruction-following difcult; patients demonstrating major arthritic troubles that affected the intelligibility of their speech, which rendered transcription problematic; patients with visual or gnostic impairment, with whom we could not use the video aid; patients demonstrating secondary language troubles due to a neurodegenerative disorder; and nally, patients demonstrating troubles with consciousness (mental confusion). The aphasic population consists of 5 men and 12 women. The clinical tableaux represented in the study correspond (by order of frequency) to the Brocas aphasia aphasic scale of Broca (8), mixed aphasia (4), subcortical aphasia (3), anomic aphasia (1), and sensitory transcortical aphasia (1). Aphasia tableaux related to phonological troubles (conduction aphasia and Wernickes aphasia), as well as frontal aphasias are not represented by our population. The aphasic population was classied by order of the degree of severity, which allows us to consider two groups: Group /severe aphasia/: 8 speakers who correspond, on the scale of severity issued by the BDAE (Goodglass and Kaplan 1972), to levels 1 and 2.

123

342

J Psycholinguist Res (2013) 42:339347

Group /moderate aphasia/: 9 speakers who correspond, on the scale of severity issued by the BDAE (Goodglass and Kaplan 1972), to levels 3 and 4. 17 monolingual native French-speaking adults, comparable to the aphasic population in age, gender, and level of education.

Experimental Design This experimental design, conceived in 2001, is comprised in its latest version (Approx, 2004) of 17 action videos demonstrating the alteration of an object, caused by an action, and consists of a naming task then a reformulation of the action being performed in the video. Here, only the naming task is taken into account. 17 standardized action video sequences lasting 45 s each: 1to burst a balloon 2to crumple up a piece of paper 3to smash a glass with a hammer 4to crush a tomato with ones hands 5to tear up a newspaper 6to peel a carrot 7to peel an orange 8to bark a log 9to undress a doll 10to take apart a structure made of Legos 11to peel a banana 12to saw through a plank of wood 13to cut bread with a knife 14to tear bread with ones hands 15to chop parsley with a knife 16to unstitch the sleeve of a shirt 17to crumble up bread with ones hands Task Each participant is placed in an oral situation, where they must name the action being performed in the video. The instruction given once the result of the action is visible is: What did the lady just do?

Results Procedure of the Analysis of Answers The answers of each of the participants are analysed in respect to the three following criteria, for which we provide a cursory definition: Valid or Excluded Answer An answer is considered valid when it contains a verb, no matter the completeness of the utterance that is produced, which refers precisely or vaguely to the target-action. This is why an utterance that consists of only a verb that is related to the target-action will be considered as valid, while an utterance that only contains a nominal element (= naming of the object) will be excluded with regard to the action-naming task. For example: for the action video [To peel_orange], enlev/removed, elle a pluch lorange/she peeled the orange, elle dchir lorange, dchir cest pas le mot je crois/she ripped the orange, ripped isnt the right word, I think are considered valid answers. Let us note that in our study we feature target-actions in our videos, but do not expect a particular target-verb. Our theoretical position leads us to consider that there are several target-verbs possible, even if a specic verb (for example, the verb that is the most typically used by the control group) could be expected in the context. Therefore, we nd ourselves, with respect to the analysis of the answers, that, for a given action, there is not one, but several, valid answers possible.

123

J Psycholinguist Res (2013) 42:339347

343

A response is excluded when it does not contain a verb that has a relevant link to the target-action. For example: for the action video [To peel_orange], la dame elle va manger une orange/the lady is going to eat an orange, jai soif/Im thirsty, je sais pas/ I dont know, orange/orange, are excluded. Conventional or Approximate Answer A valid response is conventional when it contains a conventional verb, that is to say, that belongs to the same lexico-conceptual range of application as the noun with which it is combined in the utterance. The use of the conventional verb does not provoke any semantic tension in the utterance and therefore its use can be considered as lacking in strangeness in the linguistic community in question. For example: for the action video [To peel_orange], la dame elle a pel lorange/the lady peeled an orange, la dame elle a enlev la peau de lorange/the lady removed the peel of the orange, enlev/removed, are considered valid and conventional answers. A valid response is approximate when it contains a verb that, with regard to the action is question, provokes a semantic tension, and for this, can be considered as a mark of strangeness in the linguistic community in question. The following section provides a definition of approximate answers: Intra- or Inter-domains Approximative Answer An intra-domain approximation is a valid answer that is characterized by the presence of a tension between the verb and the designated reality. The intra-domain approximative verb belongs to the same semantic domain as the object/noun in the lm but does not designate the action being shown in the lm. For example: for the action video [To peel_orange], la dame elle a coup/the lady cut the orange, coup/cut, elle a touch la peau/she touched the peel, are considered intra-domain approximations. The use of this verb in respect to the reality of the action in question is imprecise and provokes a semantico-pragmatic tension. In the lm /to peel the orange/, it is inexact to use the verb couper/to cut, which could be used in a conventional manner in another action video: the situation in which the orange would actually be cut. In this case, the approximation is intra-domain because only one semantic domain is involved: the tension is of a pragmatic nature. An inter-domains approximation is a valid answer that is characterized by the presence of a semantic tension between the verb and the object/noun. The verb does not belong to the same semantic domain as the object/noun in question in the lm. For example: for the action video [To peel_orange]: la dame elle a dshabill lorange/the lady undressed the orange, elle a plum lorange/she plucked the orange, elle a cass a/she broke that, are considered inter-domain approximations. Results in Severe Aphasia Versus Moderate Aphasia Our analysis concerns the three following populations: Group A: 9 moderate aphasias / Group B: 8 severe aphasias / Group C: 17 control adults. We observe in Table 1 (Bonferroni test) that none of the answers presenting significant differences ( p < .05) between the different groups concern the speakers that have moderate aphasia and speakers that have severe aphasia (respectively GROUP A and B).

123

344 Table 1 Moderate aphasia and severe aphasia Answer Conventional Approximative

J Psycholinguist Res (2013) 42:339347 Groups A B A B C C C C Signicance .023 .000 .050 .043

Therefore, there is no significant difference in performance between the 17 aphasics in terms of their productions, whether they are valid or excluded, conventional or approximate. Results in Aphasia Versus Non-disordered Adult Speakers As we had not observed significant differences between the two types of aphasia, we regrouped the aphasic participants into one group and will now present the results concerning the following two populations: Group 1: 17 aphasics (moderate and severe, 4082 years old: average 60 years old) Group 2: 17 control adults (4082 years old: average 60 years old)

Result1: Valid or Excluded Answer Concerning the production of valid answers (vs. excluded answers), we do not observe a significant difference between the aphasic group (1) and the control group (2): respectively 89 and 98 % valid answers. Result2: Conventional or Approximate Answer Concerning the production of conventional or approximate answers, see Table 2. Concerning these data, we observe that the aphasics produce significantly fewer conventional answers ( p = .000) and more approximations ( p = .003) than non-disordered adults. Result3: Intra-domain Versus Inter-domain Approximations Concerning the production of intra-domain versus inter-domains approximations, see Table 3. Concerning these data, we observe that the aphasics produce significantly more approximations of an intra-domain nature than the adult control group (respectively p = .048) but that there is no significant difference in inter-domain semantic approximations.
Table 2 Aphasia controlconventional and approximate answers Answer Conventional Approximate Group 1 2 1 2 Average 10.71 15.24 4.53 1.47 Standard deviation 2.469 1.522 1.663 1.375 % 63 91 27 8 2 1 .003 Groups 2 1 Signicance ( p ) .000

123

J Psycholinguist Res (2013) 42:339347

345

Table 3 Aphasia controlintra-domain semantic approximations and inter-domain semantic approximations Answer Group Average 3.41 1.53 1.12 .41 Standard deviation 1.734 1.546 .993 .618 % Groups Signicance ( p ) .048

Intra-domain approximations Inter-domain approximatons

1 2 1 2

20 9 6.5 2 .5

Discussion In the context of this rst study, we were able to establish that adults with aphasia produce a considerable amount of verbal semantic approximations, of the kind: la dame broye le pain/the lady grinds the bread (for the action video To crumble up_bread), elle plie le papier/she folds the paper (for the action video To crumple_piece of paper), elle dshabille lorange/she undresses the orange (for the action video To peel_orange), elle casse le journal/she breaks the paper (for the action video To tear up_newspaper), elle pluche larbre/she peels the tree (for the action video To bark_log). The production of this type of utterance shows that the participants possess a exible mental lexicon (Hofstadter 1995) in which verbs are linked together by a relationship of inter-domain synonymy (and not only intra-domain synonymy or antonymy), a structuring that allows them to communicate despite their lexical limitations. Given this considerable production of verbal semantic approximations, we propose an innovative perspective that differs from past perspectives on non-standard utterances. We propose to avoid the label of error to qualify utterances such as undress an orange, rip up bread, break a tomato and suggest that they be integrated into the lexical system. In fact, if we analyse the verbal semantic approximations made by aphasic speakers, we observe that they feature a lexico-semantic cohyponymic relationship between verbs Duvignau et al. (2004, 2005, 2007); Mligne et al. (2011). If we reconsider the utterances made by people with aphasia with this perspective, we realize that they feature this kind of lexico-semantic relationship that makes up one element of the organization of the lexicon. The possibility to explain overextensions in terms of co-hyponymy or inter-domain synonymy between verbs conrms the idea that their nature, far from being erroneous, is rather ingrained within the heart of the linguistic system. This change of viewpoint allows us to explain the importance of aphasic (and child) communication, as discussed by Jakobson (1956, 1968) and Clark (1993, 2003) or Tomasello (1992, 2003): speakers who lack a word do not develop random or irrelevant strategies. We consider that, in place of terms to which they no longer have access, they will often employ words that are inter-domain synonyms of the missing verb, which compels us to call them verbal (vs. nominal) semantic approximations, echoing the notion of approached identication, introduced by Jakobson and Halle (Jakobson and Halle 1956; Jakobson 1963, 1968). In the continuation of this study, we plan to establish the fact that semantic approximations with a verbal pivot hold a metaphor status later than those that bring into play nouns that have a referential function (such as, la dame est une girafe /the lady is a giraffe). This last point will lead to the articulation of verbal metaphor and its categorisation in the tradition of certain theoretical positions that underline the importance of semantic or conceptual proximity in the

123

346

J Psycholinguist Res (2013) 42:339347

linguistic or cognitive system or that defend a differentiation in the structuring of the verbal system vs. the nominal system (Jakobson 1941, 1963; Jakobson and Halle 1956; Barsalou 1987, 1993; Gentner 1978, 1981; Gentner and Boroditsky 2001; Kauschke and Stenneken 2008). This perspective will lead us to not accept lexical choice as the only argument in determining the status of metaphor but rather the phenomena of movement to clause-nal position and modalisation as markers of the condition of the metaphor: a (de)categorisation of words in distinct categories. In fact, with regard to the inter-domain approximations made by non-disordered adults as well as adults with aphasia, they can be described as the use of an unusual term that is semantically close to another more conventional term, based on a lexicon with constituted categories, which therefore assumes a process of decategorisation, that is to say, a limited erasure of conceptual boundaries. Such semantic approximations could, to this effect, be considered as metaphors when made by non-disordered adults and adults with aphasia, however, with a slight difference: the metaphor would be of an intentional nature when made by the non-disordered adult, whereas it would very often be involuntary and provoked by an access problem to a conventional term when made by the aphasic speaker. References
Barsalou, L. (1987). The instability of graded structure: Implications for the nature of concepts. In U. Neisser (dir.), Concepts and conceptual development. Cambridge: Cambridge University Press. Barsalou, L. (1993). Flexibility, structure, and linguistic vagary in concepts. In A. F. Collins et al. (dir.), Theories of memory. London: Lawrence Erlbaum Bedny, M., & Caramazza, A. (2011). Perception, action and word meanings in the human brain: The case from action verbs. Annals of the New York Academy of Sciences. Bormann, T., Kulke, F., Wallesch, C.-W., & Blanken, G. (2008). Omissions and semantic errors in aphasic naming: Is there a link? . Brain and Language, 104(1), 124132. Clark, E. V. (1993). The lexicon in acquisition. Cambridge: Cambridge University Press. Clark, E. V. (2003). First language acquisition. Cambridge: Cambridge University Press. Cloutman, L. L., Gottesman, R., Chaudhry, P., Davis, C. L., Kleinman, J. T., Pawlak, M., Herskovits, E., Kannan, V. C., Lee, A., Newhart, M., Heidler-Gary, J., & Hillis, A. E. (2009). Where (in the brain) do semantic errors come from? . Cortex, 45(5), 641649. Deloche, G., et al. (1996). Picture confrontation oral naming: Performance differences between aphasics and normals. Brain and Language, 53(1), 105120. Duvignau, K., Gaume, B., & Nespoulous, J.-L. (2004). Proximit smantique et stratgies palliatives chez le jeune enfant et laphasique. In Parole J.-L. Nespoulous, & J. Virbel (Coord.) (Vol. 3132, pp. 219255). Belgium: UMH Duvignau, K., & Gaume B. (2005). Linguistic, Psycholinguistic and Computational Approaches to the Lexicon: For Early Verb-Learning. A special issue on learning. ESSCS Journal, Journal of the European Society for the study of cognitive systems. March, 6-2 (3), 255269. Duvignau, K., Gaume, B., Pimenta, M.-A., & Elie, J. (2007). Semantic approximations and exibility in the dynamic construction and deconstruction of meaning. Linguagem em Discurso 7(3). In H. Moura, J. Vieira & M. I. A Nardi (Eds.), Metaphor and Context (pp. 371389). Gentner, D. (1978). On relational meaning: The acquisition of verb meaning. Child Development, 49, 988998. Gentner, D. (1981). Some interesting differences between verbs and nouns. Cognition and Brain Theory, 4(2), 161177. Gentner, D., & Boroditsky, L. (2001). Individuation, relativity and early word learning. In M. Bowerman & S. Levinson (Eds.), Language acquisition and conceptual development (pp. 215256). New York: Cambridge University Press. Goodglass, H., & Kaplan, E. (1972). Boston diagnostic aphasia examination (BDAE). New York: The Psychological Corporation. Goodglass, H., & Wingeld, A. (1997). Anomia: Neuronatomical and cognitive correlates. San Diego: Academic Press. Hillis, A. E., Heidler-Gary, J., Newhart, M., Chang, S., Ken, L., & Bak, T. (2006). Naming and comprehension in primary progressive aphasia: the inuence of grammatical word class. Aphasiology, 20, 246256.

123

J Psycholinguist Res (2013) 42:339347

347

Hodgson, C., & Lambon Ralph, M.-A. (2008). Mimicking aphasic semantic errors in normal speech production: Evidence from a novel experimental paradigm. Brain and Language, 104, 89101. Hofstadter, D. (1995). Fluid concepts and creative analogies. New York: Basic Books. Jakobson, R. (1963). Essais de linguistique gnrale. Minuit: Ruwet. Jakobson, R. (1968). Child language, aphasia and phonological universals. The Hague: Mouton. Jakobson, R., & Halle, M. (1956). Fundamentals of language. Berlin: La Haye. Kauschke, C., & Stenneken, P. (2008). Differences in noun and verb processing in lexical decision cannot be attributed to word form and morphological complexity alone. Journal of Psycholinguistic Research, 37, 443452. Kemmerer, D. (2003). Why can you hit someone on the arm but not break someone on the arm? A neuropsychological investigation of the English bodypart prssessor ascension construction. Journal of Neurolinguistics, 16, 1336. Kemmerer, D., & Gonzalez-Castillo, J. (2010). The two-level theory of verb meaning: An approach to integrating the semantics of action with the mirror neuron system. Brain and Language, 112, 5476. Kemmerer, D., & Tranel, D. (2000). Verb retrieval inn brain-damaged subjects: Analysis of erros. Brain and Language, 73, 393420. Khon, S. E., & Goodglass, H. (1985). Picture naming in aphasia. Brain in Language, 24, 266283. Kim, M., & Thompson, C. K. (2004). Verb deficits in Alzheimers disease and agrammatism: Implications for lexical organization. Brain and Language, 88, 120. Kittredge, A. K., Dell, G. S., & Schwartz, M. F. (2006). Aphasic picture-naming errors reveal the inuence of lexical variables on production stages. Brain in Language, 99, 8119. Le Dorze, G., & Nespoulous, J.-L. (1989). Anomia in moderate aphasia: Problems in accessing the lexical representation. Brain and Language, 37, 381400. Liljestrm, M., Tarkiainen, A., Parviainen, T., Kujala, J., Numminen, J., Hiltunen, J., Laine, M., & Salmelin, R. (2008). Perceiving and naming actions and objects. NeuroImage, 41(3), 11321141. Piras, F., & Marangolo, P. (2007). Noun-verb naming in aphasia: A voxel-based lesion-symptom mapping study. Neuroreport, 18(14), 14551458. Mligne, D., Fossard, M., Belliard, S., Moreaud, O., Duvignau, K., & Dmonet, J.-F. (2011). Verbs production during action naming in semantic dementia. Journal of Communication Disorders, 44, 379391. Thompson, C. K., Shapiro, L. P., Kiran, S., & Sobecks, J. (2003). The role of syntactic complexity in treatment of sentence decits in agrammatic aphasia: The complexity account of treatment efcacy (CATE). Journal of Speech, Language, and Hearing Research, 46, 591607. Raymer, A. M., & Ellsworth, T. A. (2002). Response to contrasting verb treatments: A case study. Aphasiology, 16, 10311045. Thompson, C. K., & Shapiro, L. P. (2007). Syntactic complexity in treatment of sentence decits. American Journal of Speech-Language Pathology, 16, 3042. Tomasello, M. (1992) First verbs: A case study of early grammatical development. Cambridge: Cambridge University Press. Tomasello, M. (2003) Constructing a language. A Usage-Based Theory of Language Acquisition, Harvard. Wambaugh, J. L., & Ferguson, M. (2007). Application of semantic feature analysis to retrieval of action names in aphasia. Journal of Rehabilitation Research and Development, 44, 381394. Wells-Jensen, S. (2007). A cross-linguistic speech error investigation of functional complexity. Journal of Psycholinguistic Research, 36(2), 107157. Wunderlich, D. (2006). Towards a structural typology of verb classes. In D. Wunderlich (Ed.), Advances in the theory of the lexicon (pp. 58166). Berlin: Mouton de Gruyter.

123