What your voice gives away

When Kamala Harris speaks, Nicole Holliday hears her multicultural background and distinctly California roots. When a voice assistant answers a question, Holliday hears a set of choices engineered to sound human enough to trust.

Holliday, acting associate professor of linguistics, studies the social information carried in speech: the subtle cues of region, age, race and gender that listeners decode automatically, often without knowing they’re doing it. Her work has taken on new stakes as AI-generated voices move into everyday life, raising questions about what it means when the thing talking to us isn’t a person at all.

After Office Hours is a new UC Berkeley News series where faculty explain, in their own words, the “aha” moments powering their research. Holliday, our first faculty member featured, talked with us about how she explains her work at a dinner party, what changed her thinking and what she wants Californians to know about the public value of basic science.

What was your Fiat Lux moment — the breakthrough or new insight that changed your work?

My linguistic research has always come back to one question: “When we hear speech, what properties do we attune to that enable us to make social judgments about each other?”. It has major implications for how we think about society, identity, and inequality. We now live in a world where humans are speaking and interacting not only with other humans, but also with technological systems designed to give us the illusion of humanity. My fundamental linguistic question now has serious consequences for our psychological and sociological well-being, as well as how we design the type of society we want for ourselves.

What’s the simplest way you’d explain your research to someone at a dinner party?

You know how you can get a mental picture of someone on the phone without seeing them at all? I study how people do that. When you hear a particular word, or a vowel, or even a tone that gives you an idea of where someone is from or how old they are, you’re using linguistic information to understand the social world around you. Humans have evolved the ability to do this because sociological information is very useful to us, and because language is an important part of what makes us human in the first place.

What would you want people to know about the work you do?

Basic research, in all fields, including linguistics, helps us understand the world around us so that we can tackle big problems. When taxpayers fund work that seems obscure — like a project that focuses on the sentence structure of a language with 10 remaining speakers, or how airflow differences in the mouth affect what sounds we hear — they are actually paying for the foundation that allows us to understand human language itself. And that’s necessary if we want to build structures and systems that make the world better for all of us.