Interviews

Listen or read the following collection of Scott’s interviews with leading experts and learn from their studies.

Dr. John Goldsmith

Dr. John Goldsmith is Edward Carson Waller Distinguished Service Professor at the University of Chicago Departments of Linguistics, Computer Science and Physical Science Collegiate Division, and Humanities Collegiate Division.

What is your background?

I’m a student of language and linguistics—most of my official training was in linguistics. I got a Ph.D. in 1976 for work in phonology, at Massachusetts Institute of Technology. I’m interested in how language works, and even more in how we come to better understand things (like language) in general.

That latter question involves both how each of us, as individuals, learn things and how all of us, as a community, come to learn things (which could be described as how our knowledge advances, step by step).

One of the most important components of learning involves simplicity—that is, we are able to learn when we have developed a sense of what simplicity is and when we begin to recognize it. Awareness of simplicity is a crucial element of learning.

What is your profession?

I’m a professor, and I teach linguistics and computer science (mostly computational linguistics).

(Followup) Can you explain your work with Linguistica? What is the goal and how does linguistics play a role with this? What are the practical implications upon its completion?

I’ve been working on Linguistica for nearly twenty years. On the face of it, it’s a computer program whose goal is to take in a large sample of a language—any language—and then be able to figure out the morphology of the language, by which linguists mean the way that words are put together in the language. This includes figuring out what are the roots and the stems, the prefixes and the suffixes, things like that. My goal is to understand what linguists do well enough that I can turn that knowledge into a computer program.

That really only scratches the surface of what interests me, though. The deeper question is this: what kind of knowledge—let’s say, what kind of smarts—do we need to endow the program with in order to accomplish an impressive piece of linguistic analysis? At risk of oversimplifying, there are two principal views out there regarding what kind of knowledge a language learner needs to learn. The perspective that [Noam] Chomsky championed over several decades was that there was a rich human endowment, an innate endowment, that was certainly not obvious and certainly not just a matter of common sense; it had unsuspected structures, hidden from our view. The linguist, in Chomsky’s view, would find that the correct description of any particular language was a fixing of a relatively small number of ways in which languages may differ. I don’t want to make that sound silly; far from it, and I was myself committed to that perspective for a good long time. But there is a different perspective that has driven my work for quite some time, which has grown out of the Linguistica project.

It is all derived from the notion that the simplicity of a grammatical analysis can serve as the best guidepost for the learner of a language, regardless of whether that learner is a person or a computer. I was very struck by work by Carl de Marcken and Michael Brent in the mid-1990s on how the words of a language could be learned, if there is no overt indication of where one word ends and the next begins. They explored computational complexity as the central tool for a learner seeking to find both repeated sequences of sounds, and hierarchical structure as well.

I decided to work on morphology where they had looked at word discovery. Morphology has more interesting stuff in it for a linguist. And I’ve been working on the project on and off for quite some time. It has changed how I understand the fundamental questions of linguistics, in several ways. More than anything else it gives me a different sort of sensitivity to what my colleagues do when they posit some aspect of universal grammar as an explanation of something that they observe in a language. I find that they have little idea of how difficult it is to design a learner who can use a bit of innate knowledge to solve an actual problem.

What are your personal/professional experiences with language?

I’m engaged with language in some way or other in almost everything I do, each and every day. Linguistics has given me some of the tools, or skills, that are helpful in understanding what some of the questions are, when we gaze at what we say and write, and when we gaze at what others have said or written.

I’ve been working on writing since I was a kid, and I’m still working on the skills it takes to write. I’ve also enjoyed working on learning other languages. I’m not very good at learning languages in a formal environment, like a classroom, but when real life experience provides an opportunity to learn another language, I very much enjoy, and get into, the process. I’m sitting in France right now, and I’ve been here pretty much every summer for nearly thirty years. I’m very grateful to the speakers of French to allow me to experience a linguistic culture different from my own — even though francophony has been formative, in centuries gone by, for my language.

Where can we find out more about your work?

The best place is http://people.cs.uchicago.edu/~jagoldsm/