3

I'm getting ready to enter graduate school, and I would like to do PhD-level research in computational linguistics and natural language processing. I know that these two fields are very interdisciplinary and draw on various subfields of linguistics, computer science, math, and statistics. I also know that research in comp ling and NLP is done in different departments at different schools, with some schools having it in the CS department, and some in the linguistics department.

Ideally, what background knowledge should someone have in order to do research in computational linguistics or natural language processing? Which areas of linguistics, computer science, math, and statistics are necessary or helpful in studying comp ling and NLP, and is there one field among those four which is overall more necessary than the others?

(Note: per this question from Linguistics.SE, the distinction between comp ling and NLP is pretty blurry, which is why I mention both in my question.)

  • This question is an overhaul of my original, closed question. I put the current version on meta for feedback on whether it was on-topic before posting. – tsleyson Oct 1 '14 at 8:01
  • Conceptually speaking, I have a minimal amount of background with Comp Ling. That being said, I would feel that computer science is pretty high on the list of importance. The translation of languages to other languages is a major part of computer science itself, for example Java to Bytecode to Machine Code. – Compass Oct 1 '14 at 13:45
  • Multidisciplinary areas benefit from people with very different backgrounds providing their own point of view. Take it from a theoretical physicist undergrad doing a PhD in Bioinformatics. – Davidmh Oct 1 '14 at 18:03
  • @Compass My undergrad is in CS and I made sure to take compilers, programming languages, and whatever else would involve translation, and AFAIK you are absolutely right. I actually was curious how important linguistics and math were, but I phrased the question generally to get it reopened. – tsleyson Oct 1 '14 at 21:03
  • @tsleyson Every subject is arguably as important as you want it to be. I graduated with a BS in Bioinformatics, tried to get into medical school, and failed repeatedly. I eventually received an MS in CS, but I used my medical knowledge while doing software development at a research firm. That being said, you probably won't need more than basic statistics or math. Linguistics, I would think is more valuable due to how it covers the 'grammar' that holds things together, whether they be speech patterns or types of loops. – Compass Oct 1 '14 at 21:15
1

Linguistics Ph.D. admissions is unusual in that no particular background is necessarily required. The reason is that many institutions do not offer a linguistics major, so the pool of linguistics majors is small. In addition, linguistics is highly interdisciplinary.

To answer your question more specifically, if you want to enter computational linguistics, some programming ability is essential. A bachelor's degree in any of the disciplines you mention may be sufficient. I would also add psychology, anthropology, and foreign language backgrounds as potential entry points for a linguistics Ph.D. You must also be prepared to learn statistics.

  • Based on your closed question, you already have an appropriate background for a Ph.D. program in linguistics, you just lack the appropriate grades. So you should pick a master's program based on your interests and the program's placement history. – Anonymous Physicist Oct 4 '14 at 20:44
  • Thanks for your answer. So I should decide whether I prefer CS, stats, or linguistics, then pick a master's program and work hard to get the right grades, experiences, and recommendations for admission to a Ph.D. program. (I was already turned down by one Ph.D. program, but from what you say that was over my grades.) – tsleyson Oct 5 '14 at 22:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.