UC Berkeley Linguistics Expert Explains How ‘Language Is Everywhere’

Whale whisperer. Time traveler. Robot wrangler.

For Gašper Beguš, associate professor of linguistics at UC Berkeley, it’s all in a day’s work.

Beguš’ love of languages started early. In elementary school in Slovenia, his education included learning the language of his home country, of course. But it also offered a hearty helping of linguistics, the scientific study of words, their meanings, and their past lives. He remembers coming across an etymological dictionary as a teenager, enraptured by the hidden world revealed between its covers.

“By the time I was in high school, I already knew that I wanted to do linguistics,” said Beguš, who can speak more than a half-dozen modern languages, and has studied over 10 that are long extinct. “It’s one of those things you’re basically almost born with.”

That curiosity has sent Beguš on manifold journeys, from uncovering lost sounds of ancient Sanskrit to carefully examining the clicks and codas that marine mammals use to communicate. At the Berkeley Speech and Computation Lab, Beguš has led efforts to create realistic artificial intelligence models that provide rare insights into how babies learn a language.

We talked to Beguš about time travel, interspecies communication, the power of libraries, and what linguistics can reveal about the world around us.

What inspired you to study languages in the first place? In particular, what drives your interest in dead languages?

I always knew that I wanted to study linguistics, and it was kind of a silly question that I had as a kid that got me interested: I wanted to learn which language is the most difficult. Soon thereafter, I realized this is a bad question because there’s no such thing as the most difficult language.

There’s something so beautiful about old languages because, in a sense, they allow you to time-travel. You read amazing literature and legal texts that are 3,500 years old, and you realize some things were done surprisingly similarly then, and some things were done surprisingly differently. You get into this world of authors and writers and poets and legal scholars who lived so long ago.

Then there’s this absolutely fascinating thing, which is that you can reconstruct how a language sounded 6,000 years ago based only on a knowledge of linguistics. And that is so mind blowing.

How do you use the UC Berkeley Library in your research, and how important are libraries to your work?

Well, I always use the UC Berkeley Library.

I think it’s great that we have digitized all these languages, but the value of print is that you see and understand things in a different way, especially when it’s a dead language. Sometimes you spend hours trying to translate just a single line. It’s just so different to look at them on screen versus in the form they were actually written in.

Writing was one of the biggest revolutions of human history because it allows us to transmit ideas in space and time, and formalize our thoughts. To me, a library is a sacred place. I think libraries are monuments to this invention of humanity, which allow us to access so many ideas and worldviews and time periods.

Can you talk about your work with artificial intelligence, and how your work with languages informs or intersects with that?

On the surface, there seems to be no connection. One thing that the study of dead languages teaches you is to pay attention to the unusual forms. You can learn more about past stages of English from the plural “oxen” than from the regular plural. The odd one out is the one that usually carries the most precious information.

In the age of big data, the majority trends tend to be more important than the exceptions. The attention to detail that I learnt studying ancient languages offers a unique perspective. This approach also allowed me to develop new techniques to better understand the inner workings of AI.

In turn, I use those techniques to help answer some of the deep questions: What is language? How did it evolve? How does it behave? What is uniquely human about it? 

Can you talk about your eye-opening research into the Rigveda, the ancient Indian collection of hymns, and its poetic meter?

In poetry, oftentimes we have a metrical structure, which means that we alternate between stressed and unstressed syllables, providing an aesthetic poetic effect. For example, William Wordsworth wrote a poem: “I wandered lonely as a cloud / That floats on high o’er vales and hills, / When all at once I saw a crowd, / A host, of golden daffodils.”

The Rigveda was created probably around 3,500 years ago by poets who were composing hymns to various early Indic gods. Those hymns have a metrical structure, although it’s slightly different from the one that Wordsworth used. The fascinating thing about those hymns is that they were preserved not in written form but orally across generations of priests who learnt the entire corpus by heart.

In the Rigveda, it appears that not all lines, not all hymns, conform to these rules. In my paper, I show that if we reverse-engineer how language sounded slightly earlier than what is written, we can repair these lines. They become regular. They become as expected.

In English, you have such rules as well. When we speak, “over” has two syllables. But oftentimes in poetry, it has one — “o’er.” It’s a similar rule that allows you to read the lines more regularly.

You are the linguistics lead of Project CETI (Cetacean Translation Initiative), which is dedicated to listening to and understanding sperm whales. Do you envision a time when we will actually be able to translate — into English, for example — what whales are saying? I remember hearing this idea that even if lions could speak, we wouldn't be able to understand them anyway because their context is so different.

With lions, we share the concept of a tree and air and drinking and sitting and lying.

And none of these things exist in whales. There are no trees. There’s no drinking. There’s no sitting (laughs). It’s a very different world.

Of course, we share things. We share the concept of a mother, the concept of sleeping. But if what you said is true for lions, it’s even more true for whales.

If it were possible to communicate fully with a whale, what would you ask them? What would you want to know?

I think I would just listen for a while before I ask any questions.

What can the study of languages teach us about the world around us?

Linguistics is an unsung hero. People realized that Greek, Sanskrit, and Latin are so closely related that they most likely go back to a common ancestor that no longer exists. Decades later, Charles Darwin uses that example when he’s arguing the theory of evolution. When Darwin argued for evolutionary theory, it was considered outrageous. But by that time, everyone already acknowledged that Latin, Greek, and Sanskrit came from Proto-Indo-European.

This notion of a common ancestor that no longer exists — and that there’s a mother tongue and daughter languages — really introduced the notion of evolution and helped Darwin convince people.

By reading about your work and your research, it made me see linguistics in a different way. It’s not always about words and their meanings — it’s about something bigger.

When I was in grad school and before that, linguistics was not a very hyped field. I was always intrigued by that. You practice law with language. You express your thoughts with language. You learn everything — or a lot of things — through language. So I always wondered why linguistics is not a way more important field.

And now I think we’re finally at a time when linguistics is getting its due attention. Language is everywhere, and studying language is so important.

This Q&A was edited for brevity and clarity. Earlier this year, Beguš’ paper exploring the peculiar way certain sounds behave at the end of words in the dead languages of Vedic Sanskrit and Avestan appeared in Transactions of the Philological Society, the world’s oldest linguistics journal in continuous publication. His paper proposing a new rule of Vedic poetic meter appeared in JAOS.