Take a look at Jason's slides here: http://www.cs.indiana.edu/~alexr/nlpclass_2012/slides/eisner-lect01-intro.pdf
A language is a dialect with an army and a navy. It's largely a political question...
Bug filed against Chromium: https://code.google.com/p/chromium/issues/detail?id=48088
what does it mean to "know" a language?
Mikes slides about this are quite good
Quite a lot of things!
- If you know a language, you know how to understand novel text (spoken and perhaps written). You know how to produce novel text (again, spoken and perhaps written).
- You might not explicitly know what you know, but in practice, you have these abilities. Especially if you haven't formally studied these things, you might not know what you know.
- "Knowledge of language (competence) is often distinguished from the use that that knowledge is put to (performance). But these may not be clearly distinguished in the minds of language users, and the distinction may not be helpful in computational applications."
Kinds of knowledge. Humans definitely have all of these, but it's interesting to think about which NLP applications need which ones of these...
- phonetic knowledge (how to synthesize and recognize individual phonemes...)
- phonological knowledge (which phonemes can follow which other ones, what the possible words could be in a given language...)
- morphological knowledge
- syntactic knowledge
- semantic
- pragmatic...
how many languages are there?
Good question!
Thousands, at least. There are a bunch of large ones, these days, and there are fewer all the time. There are projects to document the
Mike Gasser's lecture notes:
http://www.cs.indiana.edu/classes/b651/Notes/ling_know.html
http://www.cs.indiana.edu/classes/b651/Notes/hard.html
Mike's book!