The production and comprehension of verbs with alternating transitivity by patients with non-fluent aphasia

Abstract

Background: Recent studies revealed that aphasic speakers have difficulties with the production of the intransitive (unaccusative) variant of verbs entering transitivity alternations. A key point of the current interpretations of these difficulties concerns the movement operations taking place at surface syntax, namely, the A-movement operation (Bastiaanse & van Zonneveld, 2005; Bastiaanse, 2008; Thompson, 2003).

Aim: The present study revisits the issue of processing verbs with alternating transitivity in non-fluent aphasia in Greek, a language with rich morphology and relatively free word order, which lacks A-movement. In addition, in Greek, unaccusative verbs appear with different voice morphology: One class of intransitive variants of alternating verbs bears active morphology, another one non-active morphology and a third one can surface with both. The presence of non-active voice has been argued to correspond to the presence of a voice projection in syntax of these variants, while the variants that bear active morphology are not associated with a voice projection at the level of syntax. This study investigates the ability of non-fluent aphasic speakers to produce and comprehend verbs entering transitivity alterations and explores the role of active vs. non-active morphology and word order in the performance of aphasic speakers.

Methods & Procedures: We tested five non-fluent patients and fifteen control participants. We used two tasks supported by pictures: an elicited production task and a comprehension task. The experimental material consisted of fifteen transitive and fifteen unaccusative verbs (marked for active, and/or non-active voice morphology) in sentence contexts.

Outcomes & Results: The results indicated that (i) the aphasic speakers performed better on the production and comprehension of transitives than of unaccusatives, (ii) they showed significantly lower performance on the comprehension of unaccusatives with active morphology than on unaccusatives with non-active morphology, and finally (iii) they produced transitive (S)VO structures instead of the unaccusative ones.

Conclusion: We suggest, in agreement with other researchers (for example, Schwartz, Linebarger, Saffran, & Pate, 1987) that aphasic individuals overuse a mapping strategy that associates the theta roles of agent and theme with syntactic subject and object respectively, as they produce transitive (S)VO structures, to a large extent, instead of unaccusatives. In addition, as they had difficulties with unaccusative verbs marked for active voice, we suggest that they could not successfully interpret unaccusative verbs with active voice morphology as non-agentive structures.

Keywords:

• Aphasia
• Transitive verbs
• Unaccusative verbs
• Syntax
• Morphology

Acknowledgments

We thank all participants in this study for collaborating with us. We also thank three anonymous reviewers and Christina Manouilidou for comments and suggestions.

Notes

¹Inflectional Phrase is the phrase in which the verb has its finite forms; that is, it is inflected for tense and agreement (Radford, 1997, p. 263).

²This is so because Greek is a relatively free word order language.

³Greek has unergative verbs (e.g., O Giannis traguda: John sings). While these verbs attracted much attention in the field of aphasiology (e.g., Lee & Thompson, 2004) they do not constitute the focus of our study, which deals with verbs entering into transitivity alternations and the different subclasses of unaccusatives; see Koukoulioti, Stavrakaki, Ioannidis, and Tsirka (2010) and Koukoulioti (2011) for current studies on unergative verbs in aphasia in Greek.

⁴For example, the verb ending in (a) below is an active voice ending while the one in (b) is a non-active voice ending: a. ta louloudia skorpis-an The-flowers-nom-scattered-ACTIVEb. Ta louloudia skorpist-ikan The-flowers-nom-scattered-NON-ACTIVE“The flowers scattered”.

⁵In all three patterns of unaccusatives, the by itself phrase is licensed irrespectively of the morphology of the unaccusative verb. As Levin and Rappaport Hovav (1995) discuss in detail, this PP (by itself) is out in the passive.

⁶Every phrase has a head category that determines the properties of the phrase. In a Voice Phrase the head is the functional category of Voice (Radford, 1997, p. 261).

⁷By contrast, in passive structures an agentive voice head appears, as an Agent is implicit in the passive (see Alexiadou et al., 2006).

⁸While in the previous sections we discussed the linguistic properties of verbs entering into transitivity alternations, in this section we focus on how speakers deal with them. In our view, models of linguistic competence and psycholinguistic performance models are not incompatible. While the competence models focus on the abstract level of linguistic representation, the performance models deal with how this knowledge is implemented.

⁹For example, the grammatical encoder while accessing the lemma of a lexical item is taking into account the morphophonology of the previous word and other related information (e.g., prosody).

¹⁰All patients (AP2, AP3, AP4, AP5) who showed moderate or significant impairment in naming had difficulties with both noun and verb naming.

¹¹Verbs were presented in the past and not in the present tense due to the specific properties of the intransitive verbs in Greek. Specifically, intransitive verbs in the present tense marked for third person singular are often interpreted as generic or as having a particular property that the verb shows: (1) O tihos leroniThe-wall-nom dirty-3sThe wall gets dirtyThe sentence in (1) denotes rather a potential property of the wall, that is, “to get dirty” interpreted as “the wall can get dirty” rather than an eventive situation.

¹²Since current research has shown that Greek aphasic speakers do not have problems with subject-verb agreement marking (see Stavrakaki, 2005, for a review of findings from Greek) we assumed that this morphological requirement would not cause problems to patients.

¹³Although verbs were presented (visually and orally) in past tense, in some cases aphasic participants repeated those verbs in the present tense, presumably due to their significant problems with past tense marking (see Stavrakaki, 2005, for a review of findings from Greek). In a few cases, aphasic participants did not inflect the verb for the third person singular but used the first person singular for the transitive forms.

¹⁴We do not comment here on the examples containing a null subject, as researchers disagree as to the exact location of pro, pre-verbal or post-verbal (see Cardinaletti, 1997, for some discussion). Others even suggest that no pro is present in the structure (see Alexiadou & Anagnostopoulou, 1998, for discussion).

¹⁵Post-verbal subjects are not allowed in non-null languages. In particular, Lee and Thompson (2004) found that the English-speaking agrammatic participants of their study produced phrases like “bounce the ball”. Apparently, this is not an error in null-subject languages.

¹⁶In this respect the theories proposed by Lee and Thompson (2004) and Bastiaanse and van Zonneveld (2005) can make predictions extended to non-agrammatic non-fluent patients, as is the case for most of the participants in the current study. However, we cannot characterise these theories as correct or incorrect on the basis of the present study. These theories were developed to account for data coming from agrammatism (Bastiaanse & van Zonneveld, 2005; Lee & Thompson, 2004) as well as from anomia and Wernicke's aphasia (Bastiaanse & van Zonneveld, 2005).

¹⁷We thank an anonymous reviewer for suggesting this point to us.

¹⁸An anonymous reviewer adds that this is in line with the discourse configuration properties of many languages. S/he suggests that the SVO canonicity may be predicted by pragmatic constraints. An alternative explanation of why the canonical word order is the simplest one was proposed by Menn (2000). According to her, the most frequent frames for a verb are very easy to retrieve and provide the correct slots for the retrieved nouns.