Currently, I working on my application for the Bachelor of Philosophy program at Penn State to major in cognitive linguistics, but at I am also a computer science major, my advisor wants me to incorporate computational linguistics into my curriculum. I understand what the field of cognitive linguistics is, but what would I do as a computational linguist?

What are some similarities and differences between cognitive linguistics and computational linguistics?


2 Answers 2


There are a few different things computational linguists do, broadly speaking:

computational work for linguistic knowledge: this includes, eg, statistical analysis of corpora to confirm or explore linguistic phenomena, the explorations of the computational complexity of human language

language work for computing applications: a lot of actual or potential computing applications need human language capabilities. Some examples dealt with by computational linguists include:

  • machine translation
  • conversational agents (computer programs that can talk to you, eg, there's a "virtual nurse" application in testing) (this field overlaps with HCI)
  • automatic question answering

language work required for the workflow of computing applications like parsing and so on: generally speaking a tagged sentence or parse tree isn't the end result of an application but may be required as an intermediate step in order to achieve the best results.

There are also quite a few problems in computational linguistics that are to some extent in search of an application, or considered interesting in and of themselves. Examples typically given include word sense disambiguation (eg does "bank" in a given context mean a river bank or a money bank?) and textual entailment recognition. Likewise, it may not always be clear that there is an application that requires a .05% improvement in parsing reliability, but parsing researchers will pursue it (as researchers should!)

Like any research field it has developed its own conversations. It's important not to think of computational linguists as either linguists who can program, or software developers who know some linguistics. Computational linguists spend a lot of time on problems and tasks they have developed within their community. Familiarity with those problems may help you get an idea of what you are in for. Have a look at the Association for Computational Linguistics's Special Interest Groups for an idea of active subfields, likewise if you look at the front matter of the 2011 ACL conference proceedings (look for the actual schedule of presentations, it will tell you which track each paper was assigned to).


I graduated as a computational linguist. The cognitive sciences really weren't alluded to much at all at my uni. so can't help with the compare and contrast.

So, what a computational linguist does: Not being exhaustive here... computational linguists work with corpora, work to make corpora, design tools for corpora, like POS-taggers, word sense disambiguation; and design natural language parsers, work towards machine translation and text categorization eg. "is this spam or not", and I'd say the first word net sorts under CL.

I have a feeling computational linguistics (CL) might be to linguistics what AI is to computer science, developing methods, which when they are commonly in use, are no longer considered to be part of CL/AI. Which is okay, since CL/AI is already working on the next big thing.

So, basically whatever is bleeding edge language data crunching at the moment.

I'm assuming the reason your advisor wants you to do some CL is because you'll need to know the tools, corpora and methods. Then there's getting a better understanding of what's easy and not easy for a computer vs. what's easy and not easy for humans, which should be relevant to cog. ling. indeed, and finally, it's an obvious suggestion since CL is basically going all CS on poor, unsuspecting language data: it's fun! :)

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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