Does syntactic transfer across languages modulate the bilingual speech production disadvantage?

Runnqvist, E. 1 , Gollan, T. 2 , Costa, A. 3 & Ferreira, V. 2

1 Universitat de Barcelona
2 University of California San Diego
3 Universitat Pompeu Fabra

Syntactic structures may be sufficiently abstract that they can be shared across languages in bilingual speakers. This possibility is supported by cross-linguistic syntactic transfer in structural priming studies (e.g. Loebell & Bock, 2003; Hartsuiker et al., 2004). The current study further tested this hypothesis in a different paradigm, using the bilingual disadvantage as a litmus test for shared syntax. We compared the performance of English-Spanish bilinguals to English monolinguals in a sentence production task. Participants produced English sentences by combining a verb (“hold”) or adjectival phrase (“is cute”) with two nouns (“girl doll”) in the order they appeared on a computer screen (e.g., “the girl holds the doll”). The target syntactic structures could be low (passive and prepositional genitive construction; e.g. “the doll is held by the girl”; “the doll of the girl is cute”) or high (active and synthetic genitive constructions; e.g. “the girl holds the doll”; “the girl’s doll is cute”) frequent in English and, critically, also low (passive and synthetic genitive construction) or high frequent (active and prepositional genitive construction) in Spanish. Replicating previous single-word production studies (e.g., Gollan et al., 2008; Ivanova & Costa, 2008), bilinguals began speaking more slowly than monolinguals. Importantly, this bilingual disadvantage was biggest for the passive construction (980 ms.), which is low-frequent in both languages. For the prepositional genitive –which is low frequent only in English- the bilingual disadvantage was of a similar size as the high frequent conditions (around 400 ms.) These data demonstrate (a) that there is cross-linguistic syntactic transfer during more natural sentence production even when the other language is not tested at all; and (b) the speech production disadvantage for bilinguals, rather than diminishing in connected speech, increases dramatically – at least under some circumstances.