What is it about?

We investigated the effects of perceived race (Black, white), region (southern or non-southern), and gender on memory for spoken language. We found a strong bias toward remembering speech produced by white talkers better than that for Black talkers - independent of listener race. Non-southern white voices tend to be favored in memory.

Featured Image

Why is it important?

Age, gender, race, region, mood - these are all traits that we might infer when we talk with someone. The pervasive belief that there is a standard form of American English, and the idea that this (non-existent) standard is strongly associated with the idea of "unaccentedness" and and whiteness, contribute to vast differences in social capital across American society. We found that these ideologies are pervasive at the level of memory for spoken words - with words produced by white talkers being remembered better than those produced by Black talkers, for both Black and white listener groups. Discovering that these biases exist and reflect asymmetries that exist in our society, hopes to lead to awareness and subsequently change in both scientific and organizational practices.


Research in the speech sciences often approaches speech by talkers who do not fit our idea of some prescriptive idealized American English as noisy or difficult. In doing so, we inadvertently promote a particular way of speaking as the norm of comparison in our studies. I think what excites us about this work, is that we clearly identify the value of looking rigorously across variable talker populations, and find for better or worse, that doing so mimics a lot of what we see in society, at the point of encoding spoken language into memories.

Meghan Sumner
Stanford University

Read the Original

This page is a summary of: The episodic encoding of talker voice attributes across diverse voices, Journal of Memory and Language, February 2023, Elsevier,
DOI: 10.1016/j.jml.2022.104376.
You can read the full text:



The following have contributed to this page