Bilingualism: Language and Cognition 6 (2), 2003, 97–115 C  2003 Cambridge University Press DOI: 10.1017/S1366728903001068 97 Competing activation in bilingual language processing: Within- and between-language competition ∗ VIORICA MARIAN Northwestern University MICHAEL SPIVEY Cornell University Two eye-tracking experiments examined spoken language processing in Russian-English bilinguals. The proportion of looks to objects whose names were phonologically similar to the name of a target object in either the same language (within-language competition), the other language (between-language competition), or both languages at the same time (simultaneous competition) was compared to the proportion of looks in a control condition in which no objects overlapped phonologically with the target. Results support previous findings of parallel activation of lexical items within and between languages, but suggest that the magnitude of the between-language competition effect may vary across first and second languages and may be mediated by a number of factors such as stimuli, language background, and language mode. Bilingualism presents a useful setting for exploring fundamental questions about the cognitive architecture of language and cognition. One of the most important questions in the bilingualism literature revolves around the nature of bilingual language processing. Is bilingual lexical processing language-specific, or is there overlap and interaction between lexical processing in the two languages? Early research suggests selective processing of the two languages in bilinguals, an idea that is not only plausible, but also intuitively attractive and cognitively efficient. Such a mechanism would protect a bilingual from spurious mappings onto the inappropriate lexicon. After all, it seems inefficient to activate words from a second language when the circumstances do not require it. Evidence in support of selective processing of independent lexicons comes from repetition priming in tasks such as lexical decision or word-fragment completion. Although significantly less time is required to make a lexical decision task following a same- language repetition, most studies fail to find any repetition advantage when the stimuli are repeated in different languages (Kirsner, Brown, Abrol, Chadha and Sharma, 1980; Gerard and Scarborough, 1989). Word-fragment completion studies led to similar results – performance was better only when the languages at study and test were * The authors wish to thank Ton Dijkstra and Judy Kroll for helpful comments on this work, as well as the editors and reviewers for valuable suggestions and comments on the manuscript. We also thank our students, Nora Chan and Alex Raichew for assisting with Experiment 1, Eugene Shildkrot and Olga Kats for assisting with Experiment 2, and Margarita Kaushanskaya for help with phonetic transcriptions and with tables. Address for correspondence Dr. Viorica Marian, Department of Communication Sciences and Disorders, Northwestern University, 2299 North Campus Drive, Evanston, IL 60208-3570, USA E-mail: v-marian@northwestern.edu the same (Watkins and Peynircioglu, 1983; Durgunoglu and Roediger, 1987), but no facilitation was found across languages. These and other studies (e.g., Ransdell and Fischler, 1987; Scarborough, Gerard and Cortese, 1984) have been interpreted to suggest that bilinguals can access each language independently without interference from the other language and that only one lexicon can be used at any given time. However, more recent evidence seriously challenges this account. The first set of data supporting parallel activation of both languages in bilinguals comes from performance on a bilingual version of the Stroop task. The basic Stroop task consists of naming the ink colors of words that spell color names. Monolingual speakers usually show interference when the ink color and the word color are incongruent. This basic task was adapted for use with bilinguals by Preston and Lambert (1969). Would bilinguals show any Stroop interference if the printed words are in one language and color naming is performed in another language? The reasoning was that if color naming in language A restricts activation to that lexicon only, then there will be no lexical activation from the printed words in language B, and no interference in color naming should take place. Preston and Lambert (1969), Chen and Ho (1986), and Tzelgov, Henik and Leiser (1990), among others, found that interference does take place when color naming and the printed words were in different languages. This pattern of results held true not only when the two languages shared similar orthographic representations (e.g., Preston and Lambert, 1969, with English–French bilinguals), but also when the orthographies were completely different