<p>So I'm just starting my undergrad in the fall, majoring in linguistics. I'm thinking afterwords I might want to get an advanced degree in Computation Linguistics/NLP/whatever. My issue is this: I hadn't planned to focus on computer science while earning my BA. I haven't started seemingly endless sequence of math classes needed before advanced undergraduate study in CS, and I wasn't planning on doing so. I'm thinking of taking an Intro course and maybe a programing course but that's it.</p>
<p>I don't really understand how much undergrad preparation one needs for a Master's in something. Can you take beginning courses in Grad School? I've heard of people with BAs in English and MAs in Zoology, and the like.</p>
<p>Thanks,</p>
<p>STM</p>
<p>My background is in CS, so I may have a somewhat distorted perspective, but my impression is that different subfields of CL require a broad range competencies.</p>
<p>Far on the linguistics side are topics like linguistic analysis, discourse analysis, corpus linguistics, and lexicology. A bit of background in programming and perhaps statistics is handy, but there’s not much deep math/CS.</p>
<p>Then there’s natural language processing and generation, which overlaps quite a bit with inference engines and knowledge systems in general. These fields involve a lot of AI, and since efficiency tends to be a concern, data structures and algorithms are also important. There’s plenty of deep linguistics as well, particularly syntax and semantics. Translation more or less falls into this boat as well.</p>
<p>Then there are the slowly-dying fields of text-to-speech and speech-to-text. Both draw upon phonology, but in my view they’re deeper in the math/CS area, especially the latter.</p>
<p>And then there are those crazy folks who try to understand natural semantics through formal grammars… :o</p>
<p>With regard to math courses, you may want a bit of discrete math, graph theory, and stats, but topics like calculus, geometry, number theory, and differential equations likely won’t be useful to you. (That’s not to say that they won’t be required…)</p>
<p>I’ll be applying for CL/NLP PhD programs next year, so I’ll offer my knowledge into the matter.</p>
<p>CL and NLP both require a significant amount of computer science. CS will require some math, too, like linear algebra, but math shouldn’t be too much of a problem. CS theory is important in CL/NLP, as is programming.</p>
<p>For CL grad school, you have a choice: to go into a linguistics program or a CS program. There are only a handful of schools that have actual CL programs within these departments, and those that do (or are at least worth a master’s) are not easy to get into. You can take beginning classes once you get into the MS programs, but since these are so selective, you’ll probably need to have more preparation than that beforehand.</p>
<p>However, having advanced CS experience isn’t necessary. If you’re aiming for CS program, just get at least a basic background in CS (so programming, some discrete math, algorithms, maybe some systems). You’ll definitely need some math–calculus (generally a prereq for linear algebra), linear algebra, and probability theory will be immensely helpful. If you’re aiming for a linguistics program, having a strong background in theoretical/formal linguistics will be helpful, though even then you should have taken some basic programming and math. I also highly recommend taking some logic–it’s useful in both the CS and linguistics paths.</p>