Getting to Know Your Fellow Researchers: Julia Hirschberg

October, 2017

What sparked your interest in speech and language processing?

I was originally going to get an MS in databases at UPenn. However, I took a course in NLP taught by Aravind Joshi, Bonnie Webber, and Barbara Grosz, who was visiting Penn that term. I thought it was the coolest topic I’d ever encountered and decided to come back and get a PhD in NLP. Luckily for me, Penn was very open to people coming from different backgrounds (mine being a PhD in 16th century Mexican history) and thought this was an advantage rather than a limitation. I was on leave from teaching at Smith at the time so went back and took a bunch of math courses to brush up before quitting at Smith and coming full time to Penn.

How do you think speech and language processing is changing the society for the new generation?

I think more and more corporations are realizing that expertise in this area is critical to the success of their businesses – in data mining, building dialogue systems, machine translation, every area we work in. This means more students want to take our courses and more of them will get jobs doing speech and NLP. This is really great for our field, since now people think what we work on is really important from a business as well as a research perspective.

What is your holy grail in speech and language processing? When will we achieve it?

I think we’ll always find things we want to do and can’t, or we can’t do as well as we like. So I think we’ll have jobs for many years to come.

Do you have any specific advice for students, junior faculty or others early in their careers?

I’d say, study as many aspects of speech and NLP as you can. If you take courses in NLP, take courses in speech too if you can, and vice versa. Also start building your academic network as soon as possible; it’ll be amazingly useful to you throughout your career. Join professional organizations too and keep up your membership. Those will be important roads to success in your career. And my major advice is: when you’re reviewing other people’s research, do your best to recognize the positive aspects and encourage people rather than focusing solely on the negative. It’s important to support your colleagues while being honest.

What development in the field has most surprised you? Was there a hard problem that turned out to be easy? An easy problem that proved surprisingly difficult?

I admit to being surprised at how well ASR has handled difficult problems of spontaneous speech in both Low and High Resource Languages with so much success. That’s really impressive. I remember what the state-of-the-art was in the 1980s when I don’t think anyone thought this kind of progress was possible. Or would even be useful for real applications.

Which of your publications is your favorite? Why?

After thinking about this a while, I think it’s some work Jacques Terken (then of IPO but visiting Bell Labs at the time) and I did, published in Language and Speech in 1994: Deaccentuation of words representing ‘given’information: Effects of persistence of grammatical function and surface position. It’s unusual in my experience to get really clear-cut results in an open-ended study like this one, where we had some ideas of what factors might influence how speakers produced “given” items but the results were surprisingly interesting and quite informative. It was also great fun working with Jacques because we came from different schools of modeling intonation and had very different views of how to talk about the results we got, but we had so much fun planning and running the experiments and agreeing on ways of assessing accentuation that it worked out extremely well. So this is one of my papers I think deserves more attention.

Publications & Resources

Conferences & Events

Professional Development

Community & Involvement

About IEEE SPS

For Volunteers