Carpet or Cárcel: The effect of age of acquisition and language mode on bilingual lexical access

Abstract

Lexical access was examined in English–Spanish bilinguals by monitoring eye fixations on target and lexical competitors as participants followed spoken instructions in English to click on one of the objects presented on a computer (e.g., ‘Click on the beans’). Within-language lexical competitors had a phoneme onset in English that was shared with the target (e.g., ‘beetle’). Between-language lexical competitors had a phoneme onset in Spanish that was shared with the target (‘ bi gote’, ‘mustache’ in English). Participant groups varied in their age-of-acquisition of English and Spanish, and were examined in one of three language modes (Grosjean, 1998, 2001). A strong within-language (English) lexical competition (or cohort effect) was modulated by language mode and age of second language acquisition. A weaker between-language (Spanish) cohort effect was influenced primarily by the age-of-acquisition of Spanish. These results highlight the role of age-of-acquisition and mode in language processing. They are discussed in comparison to previous studies addressing the role of these two variables and in terms of existing models of bilingual word recognition.

Keywords:

• Age of 2nd language acquisition
• Bilingual lexical access
• Bilingual word recognition
• Interlingual competition
• Language mode

Acknowledgements

The authors would like to thank Viorica Marian for helpful advice during the initial stages of the study. This research was supported by a NSF grant (MCAA award 0345950 to E. Canseco-Gonzalez) and by the Sherman-Fairchild research fund. Thanks as well to the Whiteley Center at the University of Washington for support with grant proposal preparation.

Notes

¹All auditory word stimuli will be placed in single quotation marks. We will identify when we are referring to a picture by placing the word in all-caps. All foreign words will be placed in italics.

²Throughout this paper, we will refer to all sounds using the International Phonetic Alphabet (IPA).

³Within the early bilingual group, 12 participants reported learning both English and Spanish from birth, 27 reported learning Spanish shortly before English, and 10 reported learning English shortly before Spanish.

⁴We thank an anonymous reviewer for pointing out that this manipulation may have served to pre-activate the Spanish cohorts. However, if this was the case, all AoA groups should have been equally affected, which is inconsistent with our findings.

⁵We included these control trials in order to compare our coding procedure (mentioned below in the eye-tracking analysis section) with an alternative technique used by Spivey and Marian (1999), and Marian and Spivey (2003a, 2003b). This method yielded an identical pattern of results. Therefore, we will restrict our report to the analysis using our own coding procedure.

⁶We thank an anonymous reviewer for suggesting that just because a participant can not explicitly name a picture in L2, does not mean that its name was not co-activated during the task. Therefore, we carried out an extra analysis including all of the data. The results in the mixed and bilingual modes were quite similar to those reported here. In the monolingual mode, and collapsing across the three AoA groups, we just found a slightly smaller inter-lingual activation, suggesting that our participants indeed lacked the Spanish label of some of the objects.

⁷In order to address a possible concern regarding violation of independence in the contrast analysis reported below, we also report complementary binomial tests based on the absolute number of fixations in each epoch. That is, whenever we carried out contrast analysis comparing the proportion of fixations on any two objects, we also tested whether the absolute number of fixations on those two objects differed reliably from a uniform 50:50 distribution. In the Results section we point out any differences between the two analyses, and address the implications of any such differences in the discussion section.

⁸We restricted this comparison to those groups of participants and testing conditions that are similar between the two studies.

⁹The monolingual group tested with these materials lacked any evidence of inter-lingual activation, ruling out the possibility of our inter-lingual cohorts being associated with more salient pictures.