Toes in Pies: Exploring the Processing of Spanish –\ud English Interlingual Homographs

The interplay of bilinguals' languages during visual written word recognition was investigated using a novel set of Spanish (L1) – English (L2) interlingual homographs, which are words ambiguous cross-linguistically. Differently from previous literature, both identical (e.g., PIE, meaning foot in Spanish) and near-identical (e.g., CARPET, with CARPETA meaning folder in Spanish) interlingual homographs were \ud considered and a database was created to allow for an easy-accessible list of stimuli including their associated linguistic characteristics in both languages. A series of studies were conducted in the participants' L2 at both the lexical and the semantic level to explore the extent of the non-target L1's activation and the role of cross-linguistic \ud orthographic overlap. Study 1’s findings showed that L1 linguistic properties significantly predicted bilinguals' performance in lexical decision task. In Study 2 \ud interlingual homographs were used as primes and inserted at the end of sentences biased to the L2 reading; and were followed by targets related to either the Spanish, English, or unrelated to either meaning. Prime duration was manipulated to explore late (500ms; Study 2a) and early (200ms; Study 2b) stages of processing. Bilinguals showed \ud significant negative priming for the Spanish-related meaning in the 500ms prime condition but not in the 200ms. Overall, Study 1 and 2 found no significant processing differences between identical and near-identical homographs. \ud Study 3, instead, investigated within-language ambiguity by using homonyms, which, like interlingual homographs, have the same word form but different meanings. Like Study 2, homonyms were placed at the end of sentences biased to the non-dominant meaning and stimuli were created to mimic near-identical interlingual homographs. Native English speakers produced significant priming for the sentence-relevant meaning only, irrespective of whether meaning activation was probed at early or later processing stages, thus suggesting a swift resolution of ambiguity in L1.\ud Findings are discussed with reference to existing models of the bilingual lexicon (e.g., BIA+ model, Dijkstra & van Heuven, 2002). In particular, evidence of the activation of the non-target L1 was found (i.e., Studies 1b and 2a) which supports the idea of non-selective access.