Bilingual access to interlingual homographs: An examination of effects of sentential context, word frequency, and proficiency.

摘要

Support for the exhaustive activation of multiple meanings of interlingual homographs like coin (meaning money in English and corner in French) comes primarily from studies presenting homographs in isolation (e.g., Beauvillain & Grainger, 1987; Dijkstra & Van Heuven, 1998; De Groot et al., 2000). I investigated whether exhaustive activation occurs when interlingual homographs are presented in sentential contexts. My studies examine the influence of word frequency and sentence context, which have been shown to influence the activation of multiple meanings of lexically ambiguous words in monolingual studies (e.g., Tabossi, 1988; Duffy et al. 1988; Martin et al. 1999).; I investigated the role of sentence context (Experiment 1), language of processing (Experiment 2), word frequency (Experiments 3--4), and proficiency (Experiments 3--4) on exhaustive activation of interlingual homographs. French-English bilinguals made lexical decisions to homograph translations (e.g., "corner" for coin) or control words following sentences ending in interlingual homographs. Experiment 1 showed slower lexical decisions to homograph translations relative to control words following neutral sentences but not following sentences strongly supporting only one homograph meaning, when processing was in L2 English. Experiment 2 revealed slower lexical decisions to homograph translations relative to control words following sentences in a second language (L2), but not a first language (L1). In Experiments 3--4, lexical decisions to L2 homograph translations were slower following homographs having a high L1 frequency (e.g., French coin 129 occurrences in a million) than following homographs having a low L1 frequency (e.g., French four 10 occurrences in a million).; These results show that lexical access is exhaustive, but that it is constrained by sentential context and influenced by L1 word frequency. Asymmetrical inhibition effects are hypothesized to be due to weaker representation of L2 words which permits more strongly represented L1 meanings to be activated more quickly and influence processing. These results parallel those in the monolingual literature and indicate that processing in L2 is subject to the same mechanisms as L1. These findings can be explained in terms of the Bilingual Interactive Activation+ Model (Dijkstra & Van Heuven, 2002) and the Context Sensitive Model (e.g., Martin et al. 1999).