When is gender accessed? A study of paraphasias in Hebrew anomia.

This study explored access to grammatical gender during naming in Hebrew. Studies of anomia and tip-of-the-tongue states (TOT) found that speakers of various languages (Italian, Spanish, German, Dutch) have information about the grammatical gender of words they fail to retrieve. In Hebrew, on the other hand, a TOT study found that Hebrew speakers could not provide gender information. To test access to gender in single words in Hebrew we used an implicit measure--the analysis of paraphasias of anomic patients with respect to whether or not they preserved the grammatical gender of the target word. The rationale behind this measure was that when a paraphasia is created, it generally conforms to the partial knowledge the speaker has on the target word. If speakers have gender knowledge when they fail to name, they should produce paraphasias that match their partial information, and thus match the gender of the target. Such gender preservation in paraphasias was found in German for individuals with anomia, and in Arabic, French and German for slips of the tongue. Participants were 22 Hebrew-speaking aphasic patients with phonological, semantic or conceptual anomia, who produced 532 paraphasias. None of the participants showed gender preservation in their paraphasias. Even phonological anomics, who have access to semantic information, did not preserve grammatical gender in a single-word naming task. We suggest that this difference between Hebrew and previously studied languages relates to the fact that in Hebrew bare nouns are allowed, and therefore gender is not accessed in single-word naming, whereas in languages in which a noun should be produced as a full NP (with a determiner or case-marking for example) gender has to be accessed even in single-word tasks. We propose a hypothesis according to which gender is accessed if and only if the noun is incorporated into a syntactic tree (or a chunk of a tree) that includes an agreement phrase.