تفکیک اسم- فعل در زبان پریشی: نقش قابلیت تصویری و منبع کارکردی آسیب

Aphasic patients occasionally manifest a dissociated naming ability between objects and actions: this phenomenon has been interpreted as evidence of a separate organization for nouns and verbs in the mental lexicon. Nevertheless Bird et al. [Bird, H., Howard, D., Franklin, S. (2000). Why is a verb like an inanimate object? Grammatical category and semantic category deficits. BrainandLanguage, 72, 246–309], suggested that the damage underlying noun–verb dissociation affects the corresponding semantic concepts and not the lexical representation of words; moreover, they claimed that many dissociations reported in literature are caused merely by a strong imageability effect. In fact, most authors used a picture-naming task to assess patients’ naming ability, and due to the fact that this test involves the use of pictures to represent actions and objects, nouns were frequently more imageable than verbs [Luzzatti, C., & Chierchia, G. (2002). On the nature of selective deficit involving nouns and verbs. RivistadiLinguistica, 14, 43–71]. In order to overcome this drawback, we devised a new task – nouns and verbs retrieval in a sentence context (NVR-SC) – in which nouns and verbs have the same imageability rate. Patients’ performance on this task is compared with that obtained by the same patients on a standard picture-naming task. Of the 16 aphasic patients with a selective verb deficit, as revealed by the picture-naming task, two continued to show dissociation in the NVR-SC task, while 14 did not. The data indicate that at least some patients have an imageability-independent lexical deficit for verbs. The functional locus/i of the damage is also considered, with particular reference to the lemma/lexeme dichotomy suggested by Levelt et al. [Levelt, W. J. M., Roelofs, A., & Meyer, A. S. (1999). A theory of lexical access in speech production. BehavioralandBrainSciences, 22, 1–75].

Since the late 1960s, it has been widely accepted that cognitive models must explain pathological behaviour, as the latter is thought to reflect a normal cognitive system with specific modules partially or totally injured by cerebral damage. In this perspective, neuropsychological evidence is a crucial test for models based on data from normal subjects and is an important source of information about the human cognitive system; remarkable progress in understanding the mental organization of language has been made thanks to this methodology. Specific deficits of single linguistic processing abilities (e.g. phonological, lexical or syntactic) have been observed, revealing the functional independence of the mental linguistic modules. Lexical deficits may be even more selective: in particular, patients have been observed who suffered from a dissociated noun or verb impairment in tasks eliciting lexical retrieval (Caramazza & Hillis, 1991; McCarthy & Warrington, 1985; Miceli, Silveri, Villla, & Caramazza, 1984; Thompson, Shapiro, Li, & Schendel, 1994; Zingeser & Berndt, 1988). According to Caramazza and coworkers (Caramazza & Hillis, 1991; Hillis & Caramazza, 1995; Rapp & Caramazza, 2002), dissociated impairments may be caused by damage, which selectively affects verbs or nouns at a late lexical stage (phonological or orthographical output lexicons). This conclusion is drawn from the fact that a noun–verb dissociation may appear in some linguistic tasks, but not in others: for instance, patient SJD suffered from verb impairment only in written naming and spelling to dictation, but not in oral naming and in reading; on the contrary, patient HW suffered from verb impairment in a spoken naming task, but not in the written version of the same task (Caramazza & Hillis, 1991). An even more striking pattern emerged in patient EBA (Hillis & Caramazza, 1995): this patient performed better on verbs than nouns in spoken production, and on nouns than verbs in written comprehension. This dissociation between written and spoken output and between production and comprehension has been accounted for by hypothesizing a multiple representation of grammatical classes (i.e. noun versus verb) in all four lexicons (orthographic and phonological input and output lexicons). The interpretation of these findings offered by Caramazza and coworkers is arguably uneconomic. Why does the cognitive system need to represent information four times that might just as effectively be represented once? After all, we always use the same knowledge when we carry out syntactic processing, irrespective of whether we are speaking, understanding, reading or writing. In fact, several models of lexical access hypothesize unitary lexical–syntactic storage. Levelt and coworkers (Levelt, 1989; Levelt, Roelofs, & Meyer, 1999) proposed a lexicon model based on a first layer of representations storing/activating grammatical and conceptual information (thelemmalevel) and on a second more peripheral level where the phonological word form is represented (thelexemelevel). Shapiro and coworkers (Shapiro & Caramazza, 2003; Shapiro, Shelton, & Caramazza, 2000; Shapiro, Pascual-Leone, Mottaghy, Gangitano, & Caramazza, 2001) have suggested that selective damage of word forms is not the only cause of noun–verb dissociation. They observed some patients who were selectively impaired either in producing the third person of a verb (and of non-words used as verbs) or the plural form of a noun (and of non-words used as nouns). They concluded that these patients had a selective “deficit in retrieving or manipulating syntactic features” of nouns or verbs (Shapiro et al., 2000). However, while these findings are per se very interesting, they do not directly account for noun–verb dissociation. The selective impairment of number features (which are generally held to be significant on nouns) versus person features (significant on verbs) or possibly of the corresponding rule (“inflect for number” versus “inflect for person”) might in principle leave the corresponding lexical categories unaffected. It is conceivable, in other words that a deficit might affect a morpheme (or a morphological rule) without impacting the lexical–syntactic category typically associated with that morpheme. It is also unclear how the deficit identified by Shapiro et al. could explain the difficulty encountered on the picture-naming task, in which the relevant morphemes are apparently not called upon. The existence of a lexical–syntactic representation of grammatical class has been claimed by Berndt and coworkers (Berndt, Mitchum, Haendiges, & Sandson, 1997a, 1997b; Berndt, Haendiges, Burton, & Mitchum, 2002). They tested 10 aphasic patients using several tasks involving isolated words or sentences and deduced that the deficit causing noun–verb dissociation would concern a lexical device, either at an orthographic–phonological modality-specific level (i.e. the lexeme level in Levelt's model) or at a unitary lexical–syntactic device (i.e. the lemma level in Levelt's model). This claim is based on three major outcomes of the study. A qualitative error analysis brought to light a great number of semantic errors in some patients and an absence of such errors in others; some patients showed an important word frequency effect, while the imageability effect was significant in others; some patients had considerable deficits both in the production of well-structured sentences and in the comprehension of reversible sentences (two deficits typically related to lemma damage), while others did not. Taken as a whole, these results seem to indicate two different breakdown loci, with some patients having a lemma deficit, and others a lexeme deficit. Bird and coworkers (Bird et al., 2000a, Bird et al., 2001 and Bird et al., 2002; Bird, Lambon Ralph, Patterson, & Hodges, 2000b), on the other hand, argued that noun–verb dissociation might be a semantic, rather than lexical phenomenon. Moreover, they suggested that many dissociations might be generated by an increased level of sensibility in aphasic patients to a number of semantic differences and imageability in particular. In fact, since nouns refer to concrete objects, they usually have a higher imageability rate than verbs and tests used to assess noun–verb dissociation were frequently not matched for this variable. Furthermore, many studies showed imageability to be an important predictor of patients’ ability to retrieve words (e.g. Bates, Burani, D’Amico, & Barca, 2001; Berndt et al., 2002; Luzzatti, Raggi, et al., 2002). In spite of Bird et al.'s interesting attempt, strong evidence has emerged in recent studies to support the position that imageability is not necessarily the main cause of grammatical class effects (Rapp & Caramazza, 2002), and even when arguably a causal relation exists, it does not suffice to explain the entire noun–verb dissociation phenomenon (Berndt et al., 2002; Luzzatti, Raggi, et al., 2002); however, these findings, though very consistent, do not yet appear to be conclusive (Berndt et al., 2002). As it is well known, predominant noun impairment is also attested (Berndt et al., 1997a and Berndt et al., 1997b; Luzzatti, Raggi, et al., 2002; Luzzatti, Zonca, et al., 2002). This phenomenon cannot be explained in terms of imageability, since verbs are systematically less imaginable than nouns. At the same time, the idea that verb impairment may be caused (in total or in part) by an increased imageability effect remains plausible. According to Bird and coworkers, even patients who continue to manifest dissociation after balancing for imageability are not suffering from a lexical deficit. Their naming impairment would be the result of selective damages to either their sensory or functional semantic features. Since verbs mainly denote actions and action concepts are mainly defined by functional properties, selective damage to functional knowledge may result in verb impairment. Analogously, selective damage to sensory knowledge will result in verb superiority, given that nouns frequently denote concrete objects and that concrete objects are defined above all by their sensory properties. This interpretation, appealing though it may seem, has some drawbacks. The assumption that object concepts are mostly defined by sensory features and action concepts by functional features is rather vague: it lacks a precise definition of the terms “sensory” and “functional” and a clear-cut experimental basis (Caramazza & Shelton, 1998; Farah & McClelland, 1991). Moreover, Bird et al.'s explanation predicts that verb impairment should co-occur with a better performance on natural objects rather than on artificial objects and, symmetrically, noun impairment should co-occur with a better performance on artificial objects rather than on natural objects. However, this prediction is not fully verified empirically (but see Bird et al., 2000a and Bird et al., 2000b). Finally, noun–verb dissociation has been explained as a consequence of syntactic damage (Friedmann, 2000; Saffran, Schwartz, & Marin, 1980). An early account considered selective verb deficits to be a more general aspect of agrammatic morphosyntactic impairment (Saffran et al., 1980), while a more recent account specifies the syntactic hypothesis in greater detail (Friedmann, 2000). Verb deficits would result from a pathological pruning of the syntactic tree: in the presence of such damage, verbs cannot move to the relevant functional categories and be inflected. Syntactic explanations share two major drawbacks: they do not account for predominant noun impairment nor do they supply a reason why non-agrammatic patients may suffer from selective verb damage. Many other aspects of noun–verb dissociation, and in particular the role of argument structure underlying verb lexical entries, have been discussed in literature (Jonkers & Bastiaanse, 1997; Kim and Thompson, 2000 and Kim and Thompson, 2004; Thompson, 2003). However, the functional account of word-class effect – and its interaction with imageability – still seems to be poorly understood. If the noun–verb dissociation occurs as a consequence of unmatched imageability rates instead of as an effect of lexical or syntactic damage, the major neurolinguistic evidence that grammatical classes are mental entities and not only purely theoretical constructs would no longer be valid. If, on the other hand, new findings were to demonstrate that noun–verb dissociation does not merely follow from an imageability effect and that grammatical classes are represented somewhere in the cognitive system, then it would be important to understand at which psycholinguistic level/levels this happens. In conclusion, there is still little consensus on noun–verb dissociation; in particular, the functional locus/i of the lesion causing the dissociation and the role of the imageability effect have still to be clarified. This study aims at providing a contribution to the recent debate on these two topics, based on experimental data. More specifically, the following questions will be addressed: (i) Does imageability play a role in determining predominant verb impairments? (ii) If so, is the imageability effect the unique cause of this dissociation? (iii) If additional damage occurs somewhere in the linguistic system, at which level of processing does it take place? In addressing these questions, the present study describes a task in which grammatical class effects are disentangled from imageability effects (i.e. nouns and verbs are matched for imageability). In this task, the target word is not triggered by a picture, but by morphologically and semantically related words belonging to the opposite grammatical class: for instance, esplodere (to explode) is triggered by esplosione (explosion). This task is similar to that used by Shapiro and coworkers ( Shapiro & Caramazza, 2003; Shapiro et al., 2000 and Shapiro et al., 2001), exploiting English inflectional morphology: however, the English plural and third person singular inflection (s) is completely regular and allows the lexical retrieval to be by-passed by using a simple sub-word-level inflectional rule. Our targets on the contrary entertain a derivational relationship with the stimulus. Since several suffixes can in principle enter the derivational process (though only one yields the correct output), we can be sure that patients will not be able to fall back on a sub-lexical strategy to solve the task, but will have to retrieve the appropriate lexical entry.

The results which emerge from this study provide new evidence of a strong association between disproportionate verb impairment and imageability effect (Bates et al., 2001 and Berndt et al., 2002; Luzzatti, Raggi, et al., 2002; Luzzatti, Zonca, et al., 2002). We have argued that this association may reflect a causal relationship: since nouns are generally more imageable than verbs (at least in standard picture-naming tasks), imageability causes predominant verb impairment. However, it must be hypothesized that an additional, imageability-independent damage can make nouns easier to retrieve for some patients than verbs. This is indicated (i) by the fact that some patients presented a predominant verb impairment also in an imageability-matched task, such as the NVR-SC task and (ii) by the outcome of the multivariate LRA, where grammatical class effect was evaluated after being disentangled from the effect of some concomitant variables and, in particular, from the imageability effect. Moreover, the presence of additional damage predominantly involving verb lexical representations becomes evident in the group analysis: in fact, insofar as the imageability effect is concerned, the better performance on verbs in the NVR-SC task than in the picture-naming task cannot be accounted for. Indeed, we have argued that these findings can only be explained by locating additional damage at lexical–syntactic level, i.e. the verb argument structure. It is also possible that the co-occurrence of imageability and verb impairment arises from extensive damage to the left hemisphere language areas. This would induce the emergence of right hemisphere lexical abilities, which are limited to high-frequency, concrete nouns (Coltheart, 2000 and Zaidel, 1990). Therefore, imageability effect and noun-superiority are expected to co-occur. We suggest, considering also the site of lesions causing predominant verb impairments (Luzzatti et al., 2002b), that both accounts may be valid, possibly in interaction with each other.