Across-task variability in agrammatic performance

Abstract

Background: As agrammatism is, at least partly, an adaptive behaviour, we investigated how compensation strategies manifest themselves in agrammatic performance.

Aims: Within a functional theoretical framework of language use we conducted an in-depth exploration of across-task variability in agrammatic patients' oral production, submitting the morphosyntactic properties of their utterances to quantitative and qualitative analyses. Methods & Procedures: We designed an original data collection protocol comprising four tasks with increasing situational constraints (gradual manipulation of two external factors: use of instructions and presence or absence of pictures), in order to prompt the production of (i) spontaneous speech, (ii) narrative speech, (iii) descriptive speech and, finally, (iv) isolated sentences. We administered the tasks to six French-speaking agrammatic patients and nine normal controls, yielding the equivalent of 9 hours of speech data. We then conducted a multi-level and fine-grained analysis of the agrammatic and control corpora to assess oral production, entailing both morphological (open- vs closed-class word distribution, frequency of determiners, verb inflections) and syntactic aspects (sententials, non-sententials, well-formedness).

Outcomes & Results: Results revealed across-task variability, suggesting that participants tended to adjust the morphosyntactic aspects of their speech according to task-dependent factors. Moreover, trade-offs were observed between morphosyntactic accuracy and oral fluency (i.e., speech rate), further pointing to the agrammatic patients' ability to gradually vary grammatical accuracy according to task constraints, rather than as a function of the limited online processing resources available to them. Results showed that agrammatic speakers used a variety of strategies to improve or reduce their grammatical accuracy according to task constraints.

Conclusions: Agrammatic speakers rely excessively on ellipsis in spontaneous speech, and on corrective or monitoring strategies in elicited speech. Thus adaptation strategies vary from one task to another, depending on the type of speech to be produced (connected vs disconnected) and monitoring factors (attention allocated to formal encoding). Finally this study confirms the usefulness of functional and compensation-oriented therapies in aiding recovery from agrammatic aphasia.

Keywords:

• Aphasia
• Agrammatism
• Adaptation
• Task variability

Notes

¹In the control corpora the slight decline in the frequency of well-formed sentences was influenced to some extent by our use of the [Subject-Noun Phrase + Verb Phrase] criterion for scoring. This result may therefore reflect the use of a normal elliptical register, rather than genuinely incorrect grammar.

²The speech rate is provided for connected speech production (Tasks 1, 2 and 3) but not for sentence production (Task 4). In actual fact we did not find any reliable way of accurately calculating the speech rate for the sentence elicitation task because of the numerous false starts, rephrasings, long pauses, failed formulations/trailings-off, etc.