Date of Award

Summer 8-2021

Author's School

Graduate School of Arts and Sciences

Author's Department


Degree Name

Master of Arts (AM/MA)

Degree Type



L2-accented speech recognition has typically been studied with monolingual listeners or late L2-learners, but simultaneous bilinguals may have a different experience: their two phonologies offer flexibility in phonological-lexical mapping (Samuel and Larraza, 2015), which may be advantageous. On the other hand, the two languages cause greater lexical competition (Marian & Spivey, 2003), which may impede successful L2-accented speech recognition. The competition between a bilinguals’ two languages is the oft-cited explanation, for example, as to why bilinguals underperform monolinguals in native-accented speech-in-noise tasks (Rogers et al., 2006).

To investigate the effect of bilingualism on L2-accented speech recognition, the current studies compare monolingual and simultaneous bilingual listeners in three separate experiments. In the first study, both groups repeated sentences produced by speakers of Mandarin-accented English whose English proficiencies varied. In the second study, the stimuli were presented in varying levels and types of noise, and a native-accented speaker was included. In each of these first two studies, the sentences were semantically anomalous (i.e., nonsensical). In the third study, the stimuli were meaningful sentences, presented in a single noise condition, and spoken by either a native speaker or an L2-accented speaker.

Mixed effects models revealed differences in L2-accented speech recognition measures driven by listeners’ language backgrounds only in Experiments 2 and 3; in Experiment 1, performance between groups was statistically identical. Results in Experiments 2 and 3 also replicated the prior finding that bilinguals perform worse for native-accented speech in noise.

We propose that neither a flexible phonological-lexical mapping system nor increased lexical competition can alone sufficiently explain the deficit (relative to monolinguals) that simultaneous bilinguals exhibit when faced with L2-accented speech in real-world listening conditions. We discuss the possible implications of processing capacity and cognitive load, and suggest that these two factors are more likely to contribute to experimental outcomes. Future studies with pupillometry to explore these hypotheses are also discussed.


English (en)

Chair and Committee

Kristin Van Engen

Committee Members

Mitchell Sommers, Joe Barcroft