This paper addresses the problem of the Role-Argument Uniqueness. More specifically, it is concerned with the way this constraint (hereafter RAUP) might be expressed within HPSG. Its starting point represents Davis’ pioneering work. In section 2, I describe Davis’ solution to the problem of the Role-Argument Uniqueness and I show that his proposal has a certain empirical drawback: the filter devised to do the job of the RAUP is not able to rule out some classes of ill-formed constructions which appear to be RAUP violations. I put forward the hypothesis that this happens because the real locus of the RAUP violations is elsewhere, presumably at the level of the representation of semantic roles. In order to check out this hypothesis, I examine two classes of lexical semantic representations which might be held responsible for RAUP violations. One of these classes involves certain cases of concatenations, while the other exhibits instances of failed unifications. I conclude that the latter one may be invoked to explain ill-formed constructions considered RAUP violations. This leads to the last (and the central) claim of the paper: unlike the GB theory, no specific principle ruling the role-argument uniqueness is needed in HPSG.
The citations/references list is based on indexed publications only, and may therefore be incomplete. For any and all inquiries related to the database, please contact us at [Please enable javascript to view.].
pdf 
html 
