The Impacts of Race on Speech Perception

PhD Dissertation

As the world becomes more globalized, the necessity to speak more than one language is becoming increasingly urgent. It has been estimated that more than 60% of the world's population is either bilingual or multilingual. One linguistic feature that is generally unavoidable, especially given the multilingual nature of our world, is accentedness. Foreign accented speech is speech that has so-called non-native features associated with it. For instance, if an individual pronounces the word think ([θɪnk]) as [tɪnk], they might be identified as a non-native or an accented speaker. What is crucial about accentedness is that accented speech goes beyond the linguistic features. Previously, it was shown that accented speakers are perceived to be less trustworthy. In our globalized world, the number of accented speakers is continually increasing. If these individuals are stigmatized in their every day life, the issue turns into a public health problem. Here, in my dissertation project, I tackle the factors that can affect listener’s judgments with a partial matched-guise technique in two different countries. Previous research shows that exposure to variable speech increases perceptual flexibility (Bradlow & Bent, 2008). In a bilingual environment, individuals are expected to have more perceptual flexibility, driven from the variability in that bilingual speech context. However, if bilingualism is perceived negatively or associated with foreignness in that context, the perceptual flexibility could be hindered by social cognition. While in the USA, being a bilingual or having an accent can be perceived negatively, being bilingual is the norm in many parts of Canada. I finished collecting data for the USA leg of my study in Spring 2019, and presented the preliminary results at the 12th International Symposium on Bilingualism conference in Edmonton, Alberta. You can find my poster here. Currently, the results from this study is under review. I found that speech paired with South Asian faces was less intelligible compared to the same speech when paired with white faces, regardless of the English variety presented. Currently, I am conducting the same experiment in Dr. Debra Titone’s Language and Multilingualism Lab at McGill University. I aim to show that variability in speech enhances perceptual openness, especially in places where bilingualism is not negatively perceived. I have been visiting Montreal regularly to collect data and to establish a scientific bridge between Florida and Montreal. I am expecting to finish my dissertation in Fall 2020.

The graphs below summarize some of the results. You can click on each to learn more about the results. You can access the R code from the button below.  

Intelligibility and accentedness judgments for American and Indian English
Intelligibility and accentedness judgments for American and Indian English

Intelligibility accuracy and accentedness judgments of American and Indian English when paired with a White face versus a South Asian face. Both American English and Indian English were transcribed less accurately when presented with a South Asian face. The same pattern was observed with the accentedness judgments. When paired with South Asian faces both American English and British English were judged as more accented.

Participants with more racially diverse social networks and their accentedness judgments
Participants with more racially diverse social networks and their accentedness judgments

Participants who had more racial diversity in their social networks judged both British English and Indian English less compared to those with less racial diversity in their social networks (next image)

Participants with less racially diverse social networks and their accentedness judgments
Participants with less racially diverse social networks and their accentedness judgments

Here, both Indian English and British English were judged as more accented compared to those who had more racial diversity in their social network. Moreover, Indian English was always judged as more accented compared to other two varieties.

Density of accentedness judgments (more diverse social network)
Density of accentedness judgments (more diverse social network)

Density of accentedness judgments with White and South Asian faces for participants who had more racial diversity in their social network. The graph shows that South Asian (I) faces were judged as more accented compared to White faces (W). However, the density of these judgments were significantly higher for participants with less racial diversity in their social networks (next image).

Density of accentedness judgments (less diverse social network)
Density of accentedness judgments (less diverse social network)

Here, the density of accentedness judgments are overall higher, though they are significantly higher for South Asian faces (I).

 

Outer-Circle Varieties and Bilingualism

Qualitative Work

Indian English is used widely as an official language, and is often acquired at home or early in education, resulting in highly proficient English bilinguals. However, Indian English is perceived negatively by many native speakers of other English varieties (Lindemann, 2005). In my work with Dr. Caroline Wiltshire, we measured the overt and covert evaluations of two different English varieties: Indian English and British English. We asked implicit and explicit questions regarding these two varieties in addition to participants’ language background and basic demographic information. Our results show that the majority of the negative stereotyping towards Indian English is coming from the media, while it is the opposite for British English. The data collection is still ongoing. Here is the published paper: https://osf.io/preprints/socarxiv/f9spu/ 


Here is the Amazon link for the documentary by Hari Kondabolu called "The Problem with Apu."

1_X8JozzBsYNssu0Dwpjb4Ig.png
 

Frequency Tagging and Visual Perception

EEG Work

I investigated if bilingualism as an experience could affect face and object perception in adults by using frequency tagging technique with EEGs. You can see the image below for a quick summary.

result.png
 

Theoretical and Emprical Investigations of Turkish

One of my all time favorite side projects to work on is Turkish. I have investigated Turkish-English bilinguals' syntax for my master's degree. I also worked on Turkish reduplication and liquid sounds. 


Ditransitive verbs in Turkish-English bilinguals' syntax: English has two different structures for ditransitive verbs ((1)I gave my book to Noam. (2) I gave Noam my book.). Turkish, on the other hand, only has a slight word order difference, as Turkish does not have prepositions ((1) I gave my book Noam. (2) I gave Noam my book.). In my Master's thesis, I tested Turkish-English bilinguals' representation of ditransitive verbs. You can find the poster here.

Turkish m-reduplication: My collaborators, Dr. Solveiga Armoskaite and Dr. Carrie Gillon, and I examined the syntax and semantics of m-reduplication in Turkish.

Turkish liquid sounds: Dr. Ratree Wayland and I looked at the frication patterns of Turkish liquid sounds.

jtpDj.jpg