Examining the empirical and linguistic bases of current theories of agrammatism.

This paper, which is organized into five sections, critically examines the empirical support and linguistic assumptions underlying several current accounts of language disturbances in Broca's aphasia. In the first section, following a discussion of the use of signal detection methodologies in investigating grammatical sensitivity, the reliability of results from two studies that suggest that Broca's aphasic patients are differentially sensitive to grammatical constraints is examined. It is concluded that in some cases, claims of intact sensitivity are not supported. The second section examines the empirical support for the hypothesis that agrammatic patients are unable to compute syntactic dependency relationships because of slowed lexical processing. It is argued that the statistical treatment of the data and interpretive problems associated with the lexical decision paradigm undermine this hypothesis. In the third section, some of the linguistic assumptions underlying criticisms of chain-disruption hypotheses are examined. It is concluded that these criticisms are based on arguable linguistic assumptions. In the fourth section, it is argued that the linguistic and empirical support for both earlier and revised versions of Grodzinsky's default interpretive strategy is lacking. Methodological and conceptual shortcomings arising from this proposal are also discussed. In the final section, potential relationships between disordered language and currently developing models of normal language processing are discussed.