Q. Should I take this class if I have never studied linguistics?
A.
Ideally you should take LING1101 first. This is in fact a prerequisite---but may not be enforced this year.
If you have not taken a linguistics course, you will need to spend extra effort to fill in the gaps.
Q. What CS background do I need?
A. The most relevant thing is sophistication in the discrete-mathematical CS curriculum, for instance finite state machines,
context free languages, propositional logic, and predicate calculus. Familiarity with functional programming is relevant
in the modules on semantics. In addition you need the skill of working efficiently at a command-line interface.
If you have this background, you will need to spend extra effort to fill in the gaps.
Q. I'm a graduate student in Linguistics. Does this course teach things that will be relevant to my research? Does it connect with linguistic theory?
A. Yes. The course includes computational material on minimalist grammar, categorial grammar, rule-based phonology, and formal semantics. In the past,
the technique that graduate students have ended up using most is methodology for finding phenomena in corpora.
Q. What is the relationship between computational linguistics and natural language processing?
A. Classically, the label "NLP" suggested an engineering focus whereas "comp ling" suggested a science focus. Nowadays it's clear that both are mutually informing. Consider rounding out your education in this area by also taking CS4740. The classes are coordinated to minimize overlap.
Q. Will this course focus on historical linguistics?
A. Time permitting (and this is questionable), there will be a module on computational approaches to historical linguistics.
Q. Is this course about machine learning?
A. This course is about processing algorithms such as parsing, representations like grammars, trees, feature structures and formulas. Machine learning is used in the module on phonetics, and we will use ML-based parsers in the modules on syntax.
Q. What does this course have to do with cognition? Why is it cross-listed with Cognitive Science?
A. Cognitive Science cuts across Psychology, Linguistics, Neuroscience, Computer Science, Philosophy, and more. Of these the
grounding of the course is mainly in Linguistics and CS. Some of the formalisms and algorithms we will look at are used as
models in computational psycholinguistics.
Q. Is there a textbook? Should I buy it?
A. As much as possible, we will use chapters from Jurafsky and Martin's Speech and Language Processing.
Chapters from the third edition are online at web.stanford.edu/~jurafsky/slp3.
We will use some chapters from the 2nd edition which probably will not be available soon enough in the 3rd edition.
Copies of the 2nd edition are in the Cornell store, and modestly priced used ones can be found online.
There will be independent readings on topics not covered by J+M.
Q. When can I meet, one-on-one with the instructor?
A. The best time for that is office hours: Wednesdays 3-5pm in Morrill Hall 203A. It's next to the Linguistics main office.