Grammar without sentence structure: A conversation analytic investigation of agrammatism

Abstract

Background: Although research into agrammatism has done much to characterise the nature of the underlying disorder, most studies have analysed elicited, task‐based data. As a result, little is known about the grammar that people with agrammatism use in everyday talk with habitual conversational partners. There is evidence in the Conversation Analysis (CA) literature to suggest that conversational grammar may not mirror the grammar of elicited language samples. The data presented in this paper were collected as part of a larger study funded by the Economic and Social Research Council (ESRC), entitled “An investigation of aphasic syntax‐for‐conversation” (ESRC R000222754, Wilkinson, Maxim, & Beeke, 2001). Aims: To explore the notion that conversation and task‐based data do not necessarily reveal the same grammatical phenomena, addressing the following questions: (1) What resources does a speaker with agrammatism make use of in order to construct a turn at talk? (2) Is the conversational grammar of a speaker with agrammatism organised in a systematic way? (3) What is the relationship between patterns of turn construction in conversation and the grammatical characteristics of output elicited by decontextualised language tests? Methods & Procedures: A videotaped conversation between an agrammatic speaker and his adult daughter is analysed using CA. Four recurring turn construction formats are described and illustrated with extracts. Background information on the client presents the results of picture‐naming and sentence production tests. Outcomes & Results: There is great variation between the grammar of conversation and test data. Test results reveal a severe problem with verb access and sentence construction, with ability declining sharply as the number of verb arguments increases. However, the speaker deploys interactional alternatives to standard grammatical structures, and it is possible for him to recount events without explicit articulation of verbs and argument structures, using a combination of talk and mime. Only a minority of his conversational utterances are concerned with recounting events—commenting, assessing, and reasoning are highly prevalent. Conclusions: Conversation and sentence‐level tests provide complementary but essentially different information about grammatical ability. This implies that assessment of conversational grammar should become a routine part of any investigation of agrammatism in order to gain a more complete picture of an individual's ability to impose structural order on their talk, and to explore implications for successful interaction with others. Currently, approaches to assessment and intervention over‐emphasise events. In conversation, other actions such as giving an opinion are just as prevalent. Findings suggest a mismatch between what appears problematic on testing and what is treated as problematic by the interactants in conversation, and that intervention might profitably seek to address grammatical difficulties that have a basis in interaction.