Word frequency effects in speech production

gender representation in the production lexicon will be reported in Chapter 6. How should one conceive the connection between a lemma and its gender node? One can ask whether these links are one way or two way. It is obvious that when a lemma is selected, its syntactic information, such as gender and word category information, becomes available. But can, inversely, syntactic information be used as a cue for the retrieval of a lemma? In this case gender would help to narrow down the number of candidate lemmas. If information about feminine gender would be present, nouns of feminine gender would gain a privileged status in the cohort of activated lexical items since these nouns would receive additional activation from that intralexical source. At a theoretical level, we are faced with a problem of convergence. Once a lemma has been accessed, its gender can be unambiguously determined it only has to be looked up. On the other hand, there is no such way to get from gender to one specific lemma, simply because the lexicon contains thousands of noun entries but only a few gender classes (and in that respect gender priming is definitely different from semantic or phonological priming in which a more or less restricted set of words will be preactivated). That is, knowing that, for 1 9 Indirect retrieval holds for a number of closed-class elements. For example, some, but not all, prepositions are retrieved indirectly upon selection of another lemma. The lemma for the auxiliary is not directly conceptually accessed, but determined by a verb lemma's diacritic features, in particular tense, aspect, and mood. In contrast, models, though closed class elements, are accessed conceptually (see Levelt, 1ΘΒ9 for details).