Question production in Dutch agrammatism

Abstract

It is generally acknowledged that question production is impaired in agrammatism. Yet, questions have not received as much attention as other aspects of sentence production, such as verbs and finiteness. The study of questions is, however, both clinically relevant, and linguistically interesting, as has been illustrated by studies by Thompson and Shapiro (1994) and Friedmann (2001, 2002). Thompson, Shapiro, Tait, Jacobs, and Schneider, (1996) demonstrated the importance of distinguishing between argument and adjunct questions. They found that training one type of argument questions resulted in improvement of other argument questions, whereas training of adjunct questions improved other adjunct questions. Friedmann (2002) showed that Hebrew and Arabic speakers with agrammatism exhibited a difference in producing Whand yes/no-questions, performing much worse on Whquestions than on yes/no-questions, whereas an English-speaking patient was shown to be equally impaired on both question types. These data are explained with the Tree Pruning Hypothesis (Friedmann & Grodzinsky, 1997; Friedmann, 2001) that states that the highest nodes of the syntactic tree are not accessible in agrammatism. In Hebrew and Arabic, Wh-questions require movement of the Wh-element to the highest node on the tree ([Spec,CP]), whereas in yes/no-questions there is no such movement. In English, both yes/noand Wh-questions require movement to the highest nodes of the syntactic tree, and therefore both are impaired. This study examines the abilities of Dutch agrammatic speakers on different types of questions. In Dutch, the Wh-element moves to [Spec,CP] and in yes/no-questions the verb or auxiliary has to move to C. On the basis of the Tree Pruning Hypothesis, we expect that Dutch agrammatic speakers will be impaired on both question types, patterning with English and not with Hebrew and Arabic. Both argument and adjunct questions are included to explore whether the agrammatic patients show any differences between these questions without explicit training.