Leveraging African American English Knowledge: Cognition and Multidialectal Processing

Abstract

This dissertation uses theories and methodologies from sociolinguistics, neurolinguistics, and psycholinguistics to investigate how American English-speaking listeners cognitively interact with voices from Black and White individuals. For so long, social and cognitive subfields in linguistics have been pursued independently. Sociolinguistics focuses on where language comes from; living, breathing, diverse individuals. Sociolinguistic methods span a variety of modes, providing nuanced insight into communities, variation, and change. However, most sociolinguistic methods are offline, and this can be limiting when considering not only how language exists within individuals and interpersonally, but how language is processed in real time. Psycholinguistics provides a range of methods and theories that evidence how speakers and listeners process language in real time. However, psycholinguistic methods have historically looked at standard or colonizing languages, rather than minoritized ones. Incorporating study of minoritized language varieties is paramount to broaden and refine our knowledge of how human language is processed.

In this dissertation, I investigate how different parts of grammar modulate variation in perception from a sociolinguistic frame of reference. The data presented come from over 90 sociolinguistic interviews, four online surveys, a virtual eye-tracking study, and two EEG neurolinguistic experiments. In Chapter 2, I investigate the relationship between perception of race and perception of emotion by operationalizing the Angry Black Woman Trope through a survey including stimuli from one Black woman and one White woman. Results showed that the Black voice was most correctly identified racially in the Angry and Neutral conditions, while the white voice was correctly identified in the Happy condition. However, low base rate correctness overall in race identification did not coincide with free-write responses from participants, indicating their perception of “lax” voices sounding Black and “enunciated” voices sounding White.

This disagreement between the audio identification and the free-write responses formed the impetus for Chapter 3, in which virtual eye-tracking was incorporated to better understand listener’s implicit perceptions of emotional speech from Black and White women. Participant experiential linguistic knowledge was measured through exposure and familiarity surveys, and usage through virtual sociolinguistic interviews. It was hypothesized that listeners with higher experiential linguistic knowledge of African American English (AAE) would show less bias, determined by identifying emotional speech with emotional and racialized image stimuli correctly, as recorded through the virtual eye-tracker. The results from this study indicate that participants have a broad range of experiential linguistic knowledge with AAE, and trends in the data suggest that higher usage can predict less bias.

Chapter 3 shows the benefit of using implicit processing models to probe cognition and variation. Chapter 4 extends the findings of emotional prosody perception to syntactic prediction during processing. Two electroencephalography (EEG) experiments focus on syntactic variation between AAE and Standardized American English (SdAE), probing variation in listener expectations. Results show that AAE and SdAE are processed differently when produced by a so-called Black bidialectal speaker and a white speaker.

This dissertation contributes to further understanding how social information interfaces with online processing, and expectations that may be formed depending on the perceived identity of a voice. Future research will build upon these findings to investigate broader claims about languages as they exist and vary in context, from person to person, further contributing to a multidialectal cognitive model of language.