What is the difference between Computational Linguistics and Natural Language Processing?

Question

On another site, https://cs.stackexchange.com/, I wondered why we had a tag for Computational Linguistics and another for Natural Language Processing? I suggested in meta a merge as I did not seem to me there was enough of a difference to justify two tags (I suggested making them synonyms), none of them being much used. To be honest, though I did some research work in the field, I never thought there was a difference. Some people objecting that these are two distinct scientific fields, I decided to check with the users of this site: https://linguistics.stackexchange.com/users/2129/babou.

Both tags existe here too:

• nlp × 93
  Natural Language Processing: Computer programming techniques designed to aid in processing human language.

• computational-linguistics × 112
  A branch of science that uses computers and mathematical methods to construct and investigate linguistic theory. Its technological and algorithmic implementation is called NLP.

Actually, 35 questions (i.e. about 1/3rd for each tag) carry both tags. Some are explicitly about one, and carry the other tag only.

The distinction between the two areas seems to be made, more or less clearly, in the definition of the Computational Linguistics tag. Still, when I look at the tagged questions, I do not really see much difference.

Is there a serious difference? Is it a difference in topics, or a difference in the background of speakers? How well is the difference enforced, whatever it may be? Is it worth maintaining a distinction that is not understood, and is misused, at the risk of creating even more confusion?

I do realize there was a previous non-meta question on this, apparently somewhat abstract questions: What are the fundamental differences between Natural Language Processing and Computational Linguistics?. Whatever the answers, are they supported by the actual use of tags on the site. Aren't they would be differences, rather than a useful classification, effectively perceived by practitionners?

Is the distinction useful, or is it just a nice way of confusing classification and searches with poorly enforced distinctions in terminology?

Answer 1

Personally, I would prefer to set nlp and computational-linguistics as synonyms of each other, in other words, merge the two tags, like how Stackoverflow handles it. Even though there are areas where one tag is more suitable than the other, the overlap between the two is so great that for tagging purposes, I'd rather not make this distinction.

If you disagree with this view, please provide your own rationalization as answers below (not comments). After a few weeks, if the "merge" side seems to be winning, we'll merge the two tags, as done on Stackoverflow.

Answer 2

In general, I agree with the merge of the tags due to the reason, that most people seem to use both terms as synonyms.

But I was quite surprised of the statistics in the thread opening question: There are ~200 questions with one tag or the other and only 35 of them carry both, those added, that are "explicitly about one, and carry the other tag only."

I have no overview of the questions but maybe the mayority of the users understands that there is a clear distinction between both? Although often related, I would not agree to say that every statistical measure that is useful for text processing is in any way "linguistic".

Nevertheless, I vote for merging as I guess that most searchers will profit from this.

Answer 3

I don't know a lot about the fields, but they seem distinct to me. Corpus studies would count as computational but aren't NLP.