Phonetic transfer in mixed-language speech

This is a collaborative project with Prof. Indranil Dutta, Jadavpur University, Kolkata. The study was conducted at the Speech and Language Processing Lab, EFL University, Hyderabad, during 2019-20.

In the speech of multilingual individuals, sound categories of the different languages influence one another while listening and speaking. However, this kind of cross-language interaction is often complex. Understanding these patterns can tell us something about how these language systems are represented and processed in the brain.

In this project, we wanted to understand how the langage context – the situation in which a multilingual person is speaking – affects their speech. To do this, we asked proficient bilingual participants to read English phrases in two contexts: one where they were using only English, and another where they were seeing both Bengali and English text, and using both langauges.

We found that in the mixed-language condition, English vowels became more ‘Bengali-like’, i.e. showed increased cross-language interaction. This showed that even in a lab environment (where participants were not interacting with another person) and without consciously deciding to do so (participants were not aware of what the study was about), speakers change their speech to ‘fit the context’. This suggests that the way language systems interact is conditioned by the context of use.

Another aim of this project was to compare two experimental paradigms that are often used to elicit cross-language interaction, to understand if this independently affects the behavior we are interested in. This was important because existing studies often use different tasks to elicit mixed-language speech, making it difficult to directly compare findings across studies.

The materials, data, and code for this study can be found on OSF here.