Computational Linguistics in San Diego

Where We Have Been, Where We Are Now, and Where We Are Going

It wasn't too long ago when computational linguistics was taught and pursued almost exclusively within computer science departments, particularly ones with an artificial intelligence component to their research orientation. More recently, however, the relevance of computational linguistics to theoretical linguistics has become apparent, with respect to both basic theory development and empirical testing of those theories. Ironically, as this change to theoretical linguistics was getting underway, mainstream computational work had shifted dramatically toward engineering-oriented and linguistically-shallow approaches, in which much of the interest in language as a scientific object of study was lost.

There are a variety of reasons for this trend in computational circles -- too many to go into on a web page -- but the reasons for the shift away from linguistic methods are not without merit. Suffice it to say that the linguistically-rich, knowledge-intensive approaches commonly pursued in the 1980's failed to deliver robust, fieldable applications systems. Nonetheless, there is reason to suspect that the statistical revolution is also reaching its plateau, and many important problems that are not directly amenable to such techniques have languished. It is unlikely that statistical methods based solely on linguistically superficial cues will continue to be sufficient to make significant advances in these areas. New theoretical infrastructure needs to be developed that is informed in part by linguistic results that shed light on the structure and meaning of language, and in part by empirical modeling concerns that need to be conquered if sophisticated applications are to be realized.

Why San Diego?

San Diego is quickly becoming a premier place in the United States to pursue computational linguistics, and particularly so for those who want to complement their study of computationally-oriented methods with the perspective and intensive training in the core properties of language provided by a linguistics department. There are currently three full-time faculty who specialize in computational linguistics in the linguistics departments at the University of California San Diego (UCSD) and San Diego State University (SDSU):

The interaction between UCSD and SDSU is significant: our students take courses in each other's departments (we have a reciprocity agreement that allows this), we are close research colleagues and collaborators, we attend each other's colloquia, and our faculty serve on the committees of each other's students. Courses in both symbolic and statistical paradigms are commonly taught, and the research grants that we pursue and have been awarded generally incorporate both.

Language Research at UCSD

UCSD also has one of the broadest programs of research in language available in the country. We have top researchers in formal linguistics, psycholinguistics, cognitive linguistics, functional linguistics, connectionist approaches to language, neurolinguistics, anthropological linguistics, and philosophy of language. For links to some of the faculty doing work in these areas, see the web page for the Center for Research in Language. Mining insights from related disciplines will almost certainly prove to be the key to revolutionary advances in computational linguistics, and no other place offers the broad range of perspectives and techniques that UCSD does.

Industrial Activity

San Diego also has a strong industrial base to support computational linguistics. Links for companies in the area that are actively involved in computational linguistics research and development are provided below.

Interested?

Potential students who would like to hear more are always welcome to contact me.


Andy Kehler 2002-07-10