Search Abstracts | Symposia | Slide Sessions | Poster Sessions
How are Phonological Categories Represented in the Bilingual Brain?
Poster Session E, Friday, October 2, 11:00 am - 1:00 pm, Wangari Maathai
This poster is part of the Sandbox Series.
Aditi Singh1, Xuanyi Jessica Chen1, Esti Blanco-Elorrieta1,2; 1Department of Psychology, New York University, 2Department of Neural Science, New York University
Each language partitions its phonological landscape uniquely, defining separate phonemes only when sound distinctions (e.g., /r/ vs. /l/ in English) mark differences in meaning (rock vs. lock). Bilinguals provide an especially revealing test case of flexible phonological categorization during speech comprehension: they must resolve misaligned phonological boundaries across their two languages. Past work has yielded mixed results on whether bilinguals do this by maintaining separate language-specific phonological systems, or one shared system that adapts based on context. Prior studies reporting separate—or “double”—systems in bilinguals, however, often confounded language and accent context. Considering that all listeners adapt in real-time to talker-specific idiosyncrasies, including unfamiliar accents, here we disentangle these factors and ask whether monolinguals and bilinguals adjust their category boundaries based on language context or talkers’ accents. We leveraged a /d/-/t/ contrast in English and Mandarin, where both languages use voice onset time (VOT) as the distinguishing cue between /d/ and /t/, but place the category boundary at different points along the continuum. In a two-alternative forced-choice task, English monolingual, Mandarin monolingual, and Mandarin-English bilingual listeners were asked to label syllables and words that morphed from /d/ to /t/ across increasing VOT values. In the first behavioral experiment, listeners heard syllables (“t” or “d”) and words—English minimal pairs (“to/do,” “toe/dough”) and their Mandarin phonological counterparts (“秃/督” [tū/dū], “偷/兜” [tōu/dōu])—produced by one English-native and one Mandarin-native talker. A short narration preceded each set of word trials to familiarize listeners with each talker's accent. The language of the response labels (task language) was orthogonalized with each talker’s stimuli and narration (accent); monolinguals only saw labels in their native language. At the syllable level and for the to-do/秃-督 word continua, English and Mandarin monolinguals showed clear population-level differences in their /d/-/t/ category boundary, replicating well-established cross-language differences in VOT categorization. At the word level, both monolingual groups consistently shifted their boundary to an earlier VOT for Mandarin-accented speech and a later VOT for English-accented speech. Bilinguals replicated this same pattern. Critically, we replicated previous work showing a “double boundary” in bilinguals under matched language-accent pairings (English task with English-accented stimuli; Mandarin task with Mandarin-accented stimuli). However, by further testing mismatched pairings (English task with Mandarin-accented stimuli; Mandarin task with English-accented stimuli), we found that this double boundary was induced by accent differences rather than task language. To test generalizability of our results and determine whether listeners adjusted based on accent or talker-specific acoustic idiosyncrasies, we added a second Mandarin-native and English-native talker and had all four talkers produce words in both their native and non-native languages. We replicated monolingual population-level differences. Moreover, listeners showed closer category boundaries within accents than across accents, while still exhibiting acoustic-driven categorization differences within the same talker across word continua. Taken together, these findings suggest a consistent speech comprehension strategy for bilinguals and monolinguals alike is not to maintain rigid language-specific boundaries, but to rely on a single, flexible system.
Topic Areas: Multilingualism, Phonology