Violating canonicity in Spanish agrammatism / Silvia Martínez-Ferreiro

Violating canonicity in Spanish agrammatism / Silvia Martínez-Ferreiro. In "Perspectives neuropsycholinguistiques sur l'aphasie - NeuroPsychoLinguistic Perspectives on Aphasia", colloque international organisé par l'Unité de Recherche Interdisciplinaire Octogone de l'Université Toulouse II-Le Mirail (France). Toulouse, 21-23 juin 2012.

The present contributions aims at unravelling the repercussions of constructions that do not correspond to the Subject (agent)–Verb–Object (patient) pattern in the performance of mildSpanish agrammatic subjects. To fulfil this aim, experiments analysing unaccusatives and psych verbs, passives, and interrogatives are reviewed to purse a unified explanation.
Production of unergatives, unaccusatives and verbs of alternating transitivity in 6 agrammatic subjects reveal that unergatives are the better preserved entries (naming: 63.3% correct; sentence completion: 56.7% correct), being unaccusatives the more severely affected group (naming: 37.6% correct; sentence completion: 28.7% correct). In comprehension of alternating verbs, the transitive reading was easier (84.4% correct) while the intransitive reading was only identified at chance level (Martínez-Ferreiro et al. 2011).
The results confirm previous studies in Spanish unaccusatives indicating low percentages of correct responses in the production of these forms (48.7%) (Sánchez Alonso et al.(2011).
The results of an elicitation task and a forced-choice task to assess production and comprehension of psych verbs in 5 mild agrammatic subjects (Martínez-Ferreiro etal. 2011) show that, even if comprehension is better preserved than production, the pattern of errors is constant across modalities. In production, transitives and alternating constructions are impaired 42.4% of the time vs. 88.2% for intransitive entries. In comprehension, intransitives also appear as the more severely affected group (62% errors) while patients performed better for both the transitive (20% errors) and the alternating conditions (14% errors). The results from 6 mild agrammatic subjects in an auditory sentence-picture matching task aimed to test reversible passives show that actives are better preserved (71.7% correct) than passives, that were identified slightly above chance level (55% correct) (SánchezAlonso et al. 2011).
Martínez-Ferreiro (2010) includes results of an elicitation and a sentence-picturematching task run to assess the production of yes/no and wh-questions, and wh-comprehension in 5 Spanish mild agrammatics. Production results show 52.31% sparement for wh- (34/65) and 80% sparement for yes/no questions (48/60) and a preference for substitution by yes/no questions (15/38 errors) in the case of wh- and by declaratives (7/12) in yes/no questions. In the sentence-picture matching task judging the comprehension of subject/object questions with animate subjects and wh-words, subjects performed virtually perfect (6.1% errors for wh-questions; 3.3% errors for wh-words). Additionally, the results of the yes-no question production task reveal that correct yes/no questions including an overt subject preceding the verb were 96.4%, indicating a tendency towards the avoidance of VOS structures.
The overall results show that, despite the wide margin of variation in constituent ordering allowed by Spanish, the violation of the SVO or the agent-patient pattern leads to high error rates in all the tested constructions, thus confirming the role of movement as a predictor ofthe deficit across constructions.