After hearing about the Linguistics and Computational Linguistics Program at Brandeis, the WHASC Editors had a chance to sit down with UCSC alumna Lotus Goldberg (BA 1993, MA 1998), who received her PhD from McGill University in 2005 and is currently Associate Professor at Brandeis.
WHASC: How have you been involved in Linguistics and Computational Linguistics at Brandeis?
Lotus: When I came to Brandeis in 2005, it was to work in the undergraduate Linguistics Program here, which is (and was) just focused on theoretical linguistics. James Pustejovsky, who is now the Chair both of the undergrad program and our Computational Linguistics (CL) MA Program, was here in the Computer Science Department and had PhD students doing CL with him. A couple of years later, he had the idea of starting an MA program focusing on CL that would take students who hadn’t necessarily studied Computer Science (CS) as undergraduates.
I was already working with James closely on a lot of the administrative work for our undergrad program and serving as its main advisor, and so when we started the CL MA Program, my roles sort of extended to include the same work for the MA program. The other main way I’m involved has to do with CL MA students either TAing for theoretical linguistics courses that I teach (since many of our MA students majored in linguistics as undergrads, and so enjoy getting to be involved in ling courses while beginning their study of CL), or taking those courses (either as electives, or as required background, for students who enter without having studied linguistics previously).
WHASC: How did your training in theoretical linguistics at UCSC prepare you to teach students in computational linguistics?
Lotus: This is a very good question! I don’t actually teach courses with CL content — but I do often have CL (and sometimes CS) students in my linguistics courses, and working with them is always a pleasure. Some have told me that the way I teach, especially courses like Syntax I (which is an obligatory course in the first term for all entering CL MA students who haven’t taken a theoretical syntax course before) seems very “computational” to them, in terms of how compatible it seems with the content of their other courses in computer programming and CL. But what’s funny about this is that I’m just teaching syntax! Since everything about linguistics at UCSC had such a big influence on me, my Syntax I course is pretty close to an earlier version of the UCSC Syntax A course that taught Principles and Parameters Theory. There’s no textbook, and we just start with a small and very inadequate grammar containing a small lexicon and a few PSRs, and then gradually build it up and improve it, working problem set after problem set, in class and on the homeworks.
I think, though, that the very formal, careful, and logical aspects of theoretical linguistics that are emphasized at UCSC — and the sort of linguist I became as a result of that — were all excellent preparation for working with CL and CS students. And I think the other thing I learned at UCSC that has been helpful has been a degree of theoretical open-mindedness, in the sense of learning a fundamental respect for any analysis that is formally solid, carefully worked out, and well-reasoned — even if that means acknowledging that there’s more than one good solution available, versus a single “right” answer. In CL, and especially in very applied areas of it, the goals can be so different from those in theoretical approaches that the methods used (and the thinking behind why they’re on track) can seem very counter-intuitive to someone trained in theoretical linguistics. But being grounded in the notion that there can be more than one coherent and effective way to approach a problem — and that a lot is determined by the particular assumptions you make — can be very helpful in understanding how even very basic assumptions from the theoretical tradition (like, say, constituency as a way of organizing the parts of a sentence) aren’t always used in CL.
WHASC: Why might the program be of interest to UCSC graduates?
Lotus: First of all, at just a practical level, we have an incredibly high rate of placement in CL jobs or PhD programs that graduates are happy with. As part of this, now that the program’s been going for a while, it’s been lovely to see that our graduates have continued to be employed to their satisfaction years later: so it’s a good decision to make, in terms of practicalities, for anyone who has studied and done well in linguistics at UCSC, and is interested in transitioning to CL. We require incoming students we admit to to be academically very solid, but it’s fine for us if students haven’t studied CS previously — and probably our most prototypical type of student is one who comes to us with a background in theoretical linguistics. So, in that sense, an undergrad or grad degree from UCSC in linguistics — with just a bit of introductory programming completed before entering — is probably an ideal background.
More substantively, though, I think one of the reasons our students have been so successful is that we really try to emphasize academic depth, and building up a full and solid foundation, methodically over each term in the program, in CL and its foundations from CS and linguistics. So this will be appealing to someone coming from the Linguistics Department at UCSC, since it’s also a place that emphasizes academic depth and rigor. There’s also a really nice community feel to our program, so it’s a friendly and comfortable environment that, again, will be familiar — even in its east coast way — to those coming from UCSC. I’ll just mention that we’ve actually had one alum of our program (she graduated last May — Amandalynne Paullada), who came to our program after finishing her undergrad degree in linguistics (and also econ) from UCSC, and really enjoyed it here. So she might also be a good person to talk to for thoughts on what the transition from UCSC to our program was like.