Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Language interference in trilingual speech production elicited by reading aloud
Poster Session D, Thursday, October 1, 4:30 - 6:30 pm, Wangari Maathai
Clara Martin1,2, Huiyi Hu1,3, Anne Neveu4, Tamar Gollan5; 1BCBL, Spain, 2Ikerbasque, Spain, 3University of the Basque country, Spain, 4Kean University, USA, 5UC San Diego, USA
Previous studies exploring the relative interference of a first (L1) and second (L2) language during paragraph reading have revealed that the L1 is more strongly inhibited—and therefore interferes less—when reading in the L2 than vice versa. In those studies, bilinguals were asked to read aloud paragraphs containing single-word code-switches, such as “I went to the market to buy apples y tomatoes,” where “y” is a code-switch from English to Spanish. When reading such paragraphs, bilinguals have been shown to produce more intrusion errors (i.e., failures to switch to the other language when encountering a code-switch; e.g., reading “apples and tomatoes”) in the L1 while reading in the L2 than vice versa, indicating weaker interference from the stronger language on the weaker language than the reverse. Using a similar design, we investigated the relative levels of language interference in trilingual speakers. In the present study, trilingual participants were presented with 18 paragraphs to read aloud. The paragraphs were written in English, French, or Spanish and contained code-switches into one of the other two (unattended) languages (e.g., a paragraph in English with code-switches into Spanish; 3 paragraphs for each of the 6 conditions). For each reading language, we measured the relative level of interference from the two unattended languages through the number of failures to switch. We aimed to test 60 participants, with 10 participants representing each of the 6 possible linguistic profiles (e.g., English–Spanish–French trilinguals). Because trilinguals exhibited different patterns of language dominance (e.g., English > Spanish > French or Spanish > French > English), we coded languages in terms of proficiency (i.e., L1–L2–L3) rather than nominally, in order to minimize any potential impact of language (dis)similarities. Our main proficiency-driven hypothesis predicted a graded pattern in which L1 would consistently interfere less than L2, and L2 less than L3. Preliminary results (N = 40) replicated previous findings from bilinguals and extended them to a trilingual context. Specifically, participants consistently produced more failures to switch into a more proficient language when reading in a less proficient language than vice versa (e.g., more failures to switch to L1 when reading in L2 than the reverse; more failures to switch to L2 when reading in L3 than the reverse, etc.). Participants also produced more failures to switch to the more dominant of the two unattended languages when reading in a given language (e.g., more failures to switch to L1 than to L3 when reading in L2). One exception, however, was observed when reading in L3: switching to L1 elicited fewer errors than switching to L2. This finding may suggest that L2 and L3 interfere more strongly with one another than either does with L1, although the full dataset is needed before firm conclusions can be drawn. We conclude that when trilinguals read in a given language, the weaker unattended language interferes more strongly than the stronger unattended language. A potentially greater degree of interference between the two weaker languages remains to be confirmed with further data.
Topic Areas: Multilingualism, Reading