The pronoun bias effect strikes again: a reply to Pynte and Colonna.

The gender of an initial adjectival phase (AP) influences the analysis of a phrase which is a potential main clause subject in French. Pynte and Colonna (this issue) establish this generalization but analyze it without considering critical aspects of sentence structure, specifically, the fact that the AP has a null pronominal subject. If grammars contain a violable constraint requiring coreference between the pronominal subject of an initial subordinate phrase and the subject of the main clause, then in Pynte and Colonna's study perceivers must choose on-line between an analysis which satisfies the same subject constraint and one which doesn't. From this perspective, it is expected that the subject analysis of the ambiguous phrase predominates. Further, on this view, the preference fits with independent generalizations about language such as the existence of switch-reference languages which obligatorily mark the coreference relation between the initial subordinate subject and the main clause subject. It also fits with already established facts about language processing, such as Cowart and Cairns' (1987) Pronoun Bias Effect.