Bidirectional transfer: The effect of sharing a translation

Abstract

This study investigated reciprocal influences between the first and second languages of bilingual speakers. Participants were monolingual English speakers and bilingual speakers of English and Hebrew who learned Hebrew either as a first language or as a second language. Participants rated the semantic similarity of English word pairs that either shared a Hebrew translation or did not, and that varied in their baseline relatedness in English. Shared-translation pairs (e.g., tool and dish are both translated as “kli” in Hebrew) were rated as more similar in meaning than different-translation pairs by both bilingual groups, but not by the monolinguals. Knowledge of Hebrew influenced the way bilinguals processed words in English not only when Hebrew was the native language but also when it was learned as a second language later in life. These findings provide evidence for bidirectional transfer, and emphasise the dynamic nature of the bilingual lexicon.

Keywords:

• Bilingualism
• Translation ambiguity
• Semantic transfer
• Semantic similarity
• Shared-translation effect

Acknowledgements

We thank the University of Pittsburgh's PLUM Lab for assistance, and Marnie Arkenberg, Dana Basnight-Brown, Jon Andoni Duñabeitia, Peter Gianaros, Nan Jiang, Erik Reichle, Leida Tolentino, and Tessa Warren for comments on an earlier version of this manuscript. During the writing of this manuscript, AP was funded by NRSA-F32HD049255 and NT and TD were funded by NSF-BCS-0745372. A version of this research was presented at the Sixth International Symposium on Bilingualism in Hamburg, Germany.

Notes

¹Data from nine additional bilinguals were replaced because they learned English and Hebrew simultaneously, or had an L1 other than Hebrew or English. To equate the number of participants in each version, we randomly excluded two Hebrew-English bilinguals. Data from two additional monolingual English speakers were replaced because they did not meet language background criteria.

²This finding is not due to overall higher similarity ratings in bilinguals, because the overall average rating of the bilinguals was lower than that of the monolinguals.