Modeling the cognitive challenge of accented speech

Washington University, Saint Louis MO

Investigators

Abstract

In a globalized world, accents are a persistent hindrance to clear communication. Everyone has an accent, of course, but not all accents are alike, even when we speak the same language. “English” in St. Louis, for example, is not the same as “English” in Singapore or Glasgow. Relatively little is known about how we adjust our perceptions to each other's accents or what kind of effort that requires. In order to understand how such communication works, this project investigates why some speakers are more difficult to understand than others and how listeners adjust to unfamiliar accents. The research team will partner with the St. Louis Science Center, the local university, and other community groups to help others learn about accents and the rich world of linguistic diversity. Through its experiments and its educational outreach, the project ultimately aims to promote tolerance and better communication among speakers and listeners of all different backgrounds. Research on accent perception has typically relied on measuring how many words a listener can correctly identify. However, some listeners may have to work much harder than others to understand an accented speaker. They may achieve the same outcome, but the processing involved may be very different. This project shifts the focus of accent research to the measurement of cognitive processing during listening, employing a combination of experimental methods from the listening effort literature and the field of behavioral economics. To identify the features of accented speech that most contribute to listening effort, the proposal also relates acoustic-phonetic analyses of the speech to experimental outcomes. Finally, it uses eye-tracking to test the hypothesis that listeners relax their linguistic categories to process unfamiliar accents. This also has important scientific implications for automatic speech recognition, language teaching and language proficiency assessment. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.