The Role of Societal-Level Differences in Multilingualism on Speech Perception

• In Event: At the Intersection of Social Systems, Perception, and Representation: Going Beyond Perceiving Faces and Languages

Abstract

Listeners use visual cues in speech perception (Zheng & Samuel, 2017). While some showed that these cues helped the perception of different speakers (McGowan, 2015), others showed the opposite (Kutlu et al., 2021). For instance, speech paired with non-white faces was found to be more accented and less intelligible compared to when the same speech (same language, dialect, and accent) was presented with white faces (Kutlu, 2020). These different patterns are particularly important to unpack when the judgments are toward multilingual and multiracial individuals. According to the exemplar-based models, the exposure to variability increases the auditory perceptual space, allowing listeners to use cues from a diverse pool. Compared to monolingual locations, multilingual communities provide such exposure to diversity. However, it is not clear to what extent growing up in locations where societal level prejudice and stigma toward multilingual and multiracial individuals impact speech perception in the presence of visual cues.
American, British, and Indian English varieties were paired with white and South Asian faces to test whether listeners’ speech perception changes depending on the visual cue that they saw on a computer screen. These three varieties were picked to tap into linguistic prejudice and stigma associated with race and language. Two experiments were conducted in two locations: Gainesville, Florida (USA), and Montreal, Quebec (Canada). These two locations were chosen as both host multilingual communities. However, they differ in terms of societal values attested to multilingualism. While the USA devalues the multilingual experience and associates multilingualism with a deficiency in English (Kutlu & Kircher, 2021), Canada promotes multilingualism. We asked listeners to type each sentence and to judge whether the speaker had an accent (relative to their own accent) or not. We measured listeners’ racial diversity in their social network and their language entropy (Gullifer & Titone, 2020). All listeners self-identified as bilinguals in both locations with differing degrees of exposure to racial diversity. Results showed that listeners in Gainesville had lower transcription accuracy for each variety when the sentences were presented with a South Asian face. They also judged all varieties paired with South Asian faces as more accented compared to white faces. In Montreal, listeners did not show a face effect and a variety effect. All sentences were transcribed with high accuracy. Crucially, in Montreal, both British and Indian English were found to be equally accented as compared to American English.
These results show that not all multilingual spaces provide the same developmental outcomes for listeners. While growing up in multilingual and multicultural locations where multilingualism is embraced, listeners are able to take advantage of a larger exemplar pool. On the other hand, other multilinguals may not benefit from a large exemplar pool directly due to societal prejudice and stigma. We discuss the holistic impacts of higher-level societal values on perception and provide preliminary results from ongoing developmental research on how these are shaped during the first years of schooling experience in the USA.

Author

• Ethan Kutlu, University of Iowa
  Presenting Author