Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Behavioral identification and neural representations of talkers' voices vary with lifelong language experience
Poster A78 in Poster Session A, Friday, September 12, 11:00 am - 12:30 pm, Field House
Ja Young Choi1, Kevin Sitek1, Louis-Phillipe Langlois2, Stephanie Deschamps2, Jen-Kai Chen2, Casey Roark3, Shari Baum2, Bharath Chandrasekaran1, Denise Klein2; 1Northwestern University, 2McGill University, 3University of New Hampshire
Listeners identify talkers by the sound of their voice less accurately in a foreign language than in their native language, a phenomenon known as the language familiarity effect. Previous studies on language familiarity effect have shown that listeners’ prior knowledge of phonology and lexicon facilitates talker identification in their native language, and that greater lifelong immersion in a second language can improve the listeners’ ability to identify voices speaking that language. While some studies have shown that international adoptees with early exposure to a specific language retain trace memory of their birth language without conscious awareness, it is unclear whether the brief exposure to their birth language very early in their development can give them some advantage to their birth-language talker identification. Building upon previous neuroimaging studies showing functional integration of processing linguistic and talker information, we aimed to identify how the neural representation of talkers varies between people with different language backgrounds. We recruited three groups of adult participants in Montreal -- native French speakers (French), native Mandarin speakers who also spoke French (Mandarin), and French speakers who were adopted from China early in their lives (Adoptee; mean age of adoption = 10.29 months). Participants listened to ten Mandarin sentences spoken by four male native Mandarin speakers and learned to identify talkers over three blocks. While they were performing the task, their brain scans were acquired using structural and functional magnetic resonance imaging (fMRI). All participant groups learned to identify talkers more accurately by the last block of the study compared to the first block, and the results showed a clear language familiarity effect such that the Mandarin group identified Mandarin-speaking voices significantly more accurately than the French group. Interestingly, the Adoptee group was significantly less accurate than the Mandarin group in the second block, and was significantly more accurate than the French group in the third block, suggesting a slower trajectory of voice learning relative to the Mandarin group and a subtle advantage relative to the French group. Univariate analysis of the fMRI data showed a general increase in the activity of auditory, speech related brain regions including bilateral superior temporal regions when listeners are identifying voices, without significant group differences. To further investigate how the voices are represented in the brain, we ran a representational similarity analysis (RSA) by computing the representational dissimilarity matrices (RDMs) of the brain activity patterns and comparing themato talker model RDMs. Results from RSA showed that talker identity was most robustly represented in the Mandarin group among the three groups across the auditory speech associated brain regions bilaterally, and talker representation in the Adoptee group fell between the Mandarin and French groups. Multivariate analysis of our fMRI data revealed that listeners’ language experience defined by their native language fundamentally shapes how talkers’ voices are represented in the brain depending on the language of the talker. Moreover, this study provides us with a unique insight into how trace memory of birth language may impact international adoptees’ ability to identify talkers in that language.
Topic Areas: Speech Perception,