ABSTRACT VERB-NOUN COMPOUNDS IN ITALIAN: A MULTIPLE SINGLE-CASE STUDY Sara Mondini1,2, Claudio Luzzatti3,4, Giusy Zonca, Caterina Pistarini4 and Carlo Semenza5 (1Università di Padova; 2Casa di Cura Figlie di San Camillo, Cremona; 3Università di Milano-Bicocca; 4Fondazione S. Maugeri, Montescano; 5Università di Trieste) Compound words constitute a rather complex system: how such a system is processed is, at present, poorly understood. Indeed, theoretical models of linguistic processing only marginally deal with this problem (a partial exception is to be found in Levelt et al., 1999). One pivotal and yet only partially answered question is whether compounds undergo (de)composition when being processed. The present investigation exploits the characteristics of the Italian compounding system and of the impaired lexical processing in aphasia in order to provide an answer based on empirical grounds. As with other complex words, also in the case of compounds three different hypotheses compete with each other. The first hypothesis maintains that compounds are stored as whole words (full listing model) and retrieved as such; the second suggests that compounds are generated by rule and have no single lexical entry (full decomposition model); the third claims that compounds while having a listed lexical entry, may also be generated by morphological rules (dual route model). Through neuropsychological methods, this study seeks information about the mental representation of Verb-Noun (VN) nominal compounds, the most productive type of composition in Italian. Semenza et al. (1997) found that in retrieving such compounds (picture naming), Broca’s aphasic patients, unlike Wernicke’s or anomic patients, show a much higher proportion of omissions of the verb component. A positional effect, namely, an impaired retrieval of the first component, was excluded, since it could not be found in NounNoun (NN) compounds. Indeed, Broca’s aphasic patients, as a group, are said to have a relative impairment on verbs with respect to nouns when dealing with simple words. However, since VN compounds are grammatically nouns, there should be no reason for Broca’s aphasics to omit the verb component unless, at some level of processing, the representation of VN compounds were decomposed into separate noun and verb forms. NN compounds and 25 long simple nouns (LSN). The three classes of randomly presented items, were matched for familiarity with the underlying concept. To check the ability to produce simple nouns and verbs in the same participants, a further picture naming task of objects (n = 30) and actions (n = 40) was used (Luzzatti et al., 2002). Data were analysed by means of a logistic regression procedure which allowed to compare the performance of each single patient when naming LSN, VN compounds and NN compounds. Furthermore, a qualitative analysis of the errors produced by each patient was done: in particular, omissions and substitutions of single components of VN and NN compounds were considered. RESULTS Thirteen patients were found to have a disproportionate verb deficit on simple words (four Broca’s, three Wernicke’s and four anomic aphasic patients; the remaining two patients could not be classified for any of the standard aphasic groups). Seventeen patients showed no significant difference between nouns and verbs (two Broca’s, three Wernicke’s and ten anomic aphasic patients; the remaining two patients could not be classified for any of the standard aphasic groups). Multiple single case analysis Table Ia summarises the distribution of the 30 patients according to their performance when naming VN and NN compounds with respect to LSN. In particular, three of the patients with prevalent impairment of simple verbs had also a prevalent impairment on VN compounds (group A), and three further patients failed with both VN and NN compounds (group B). Among the patients with no impairment with simple verbs, three failed with both types of compounds (group F). The category of aphasia is almost orthogonal to the patient’s subgroups. MATERIALS AND METHODS Qualitative analysis The lexical retrieval of compound nouns was tested in 30 aphasic patients using a picture naming task. The target names were 23 VN compounds, 21 Cortex, (2004) 40, 187-188 A qualitative analysis of the errors made by each patient was instrumental in order to 188 Sara Mondini and Others TABLE I (a) Distribution of patients according with their performance in naming simple and compound nouns. LSN = long simple nouns; VN = Verb-Noun compounds; NN = Noun-Noun compounds. LSN > VN Patients (n =) Simple words Verb Deficit 13 No verb Deficit 17 LSN ≈ VN LSN ≈ NN LSN > NN LSN ≈ NN LSN > NN Group A 3 Group E – Group B 3 Group F 3 Group C 5 Group G 10 Group D 2 Group H 4 (b) Omissions and substitutions on single components of VN compounds (out of 23 stimuli × 13 patients = 299) and NN compounds (out of 21 stimuli × 17 patients = 357). Om = omissions; Sub = substitutions VN compounds Verb Deficit No Verb Deficit NN compounds Patients (N =) Om/Sub verb Om/Sub noun Om/Sub 1st noun Om/Sub 2nd noun 13 17 30 25 11 30 21 23 25 23 discriminate between types of naming impairment (see Table Ib). While in the patients with prevalent naming impairment on simple verbs the verb component of VN compounds was significantly more omitted/substituted than the noun component [t (5) = 3.5; p < .05], no such difference was found among the patients with no prevalent impairment on simple verbs. The NN compounds showed no effect of component position. significantly more impaired in naming compounds than LSN. A similar case has been interpreted by Delazer and Semenza (1998) as a difficulty in retrieving two different lexemes with a single entry. The functional damage, in such cases, seems to affect either a morpho-syntactic lexical processor, necessary to combine different word components into a single lexical representation, or a specific working memory buffer, which becomes unable to process more than one component at the same time. DISCUSSION The present study confirms Semenza et al.’s (1997) results and shows that patients with disproportionate verb deficit (and problems with compounds) are usually also impaired in retrieving the verb component of VN compounds. This finding is consistent with the view of a decompositional processing of compound words. In line with several recent studies (Berndt et al., 1997; Luzzatti et al., 2002) the disproportionate verb deficit was found in Broca’s as well as fluent aphasic patients. Finally, there is a group of patients who are REFERENCES BERNDT RS, MITCHUM CC, HAENDIGES AN and SANDSON J. Verb retrieval in aphasia. 1. Characterizing single word impairments. Brain and Language, 56: 68-106,1997. DELAZER M and SEMENZA C. The processing of compound words. Brain and Language, 61: 54-62, 1998. LEVELT JM, ROELOFS A and MEYER A. A theory of lexical access in speech production. Behavioral and Brain Science, 22: 1-38, 1999. LUZZATTI C, RAGGI R, ZONCA G, PISTARINI C, CONTARDI C and PINNA GD. Verb-Noun double dissociation in aphasic lexical impairments: The role of word frequency and imageability. Brain and Language, 81: 432-444, 2002. SEMENZA C, LUZZATTI C and CARABELLI S. Morphological representation of compound nouns: A study on Italian aphasic patients. Journal of Neurolinguistics, 10: 33-43, 1997.