Structural prominence and agrammatic theta-role assignment: a reconsideration of linear strategies

In this paper we examine the tendency for agrammatic aphasics to make thematic reversal errors in comprehension, e.g., a tendency for English-speaking agrammatics to assign a preposed object the subject role. Although this tendency has been argued to follow from either a linear (Grodzinsky, 1995) or a directionality (Hagiwara & Caplan, 1990) strategy, we show that such proposals can, at best, function as language-particular strategies. We examine data from English, Japanese, German and Dutch, and propose a Structural Prominence Hypothesis which captures the following cross-linguistic generalization: thematic reversal errors result from a tendency to assign thematic roles based on the relative structural prominence of the candidate NPs.