5

I'm looking for an actually downloadable Chomsky grammar (preferably a context-free one) capturing as much of the German grammar (on a sentence level) as possible.

I am aware that there is no Chomsky grammar known which represents ALL of the valid German sentences (and I know how much debate is still going on on this topic), but I remember that during my studies in Computer Linguistics, 10 years ago or so, they had powerful context-free grammars which captured "fairly many" of the German sentences correctly. (These are probably not available for free - or even for purchasing.)

It is not really important to me, how big the fraction of correctly parseable sentences is (or even if occasionally incorrect sentences are successfully parsed). I am just searching for a practical way of generating parse trees for the "usual" German sentences.

Is there a way for a non-linguist (and a non-developer of MS-Word's syntax checker ;-) ) to gain access to such a grammar (or related tools)?

  • 1
    May I ask what you need the CFG for? The word order rules of German are a bit subtle; in principle, word order is free except for verbs, but if you deviate from neutral word order, you emphasize parts of the sentence, and this can become unnatural. This is quite difficult to capture in a CFG. Agreement on gender/number/case can be captured in a CFG, but it's unwieldy. You'd also need to put in most verb patterns from the dictionary in the CFG, because the verb determines the necessary cases and prepositions. So I doubt anyone has put in the work necessary to actually create such a CFG. – dirkt Jan 1 '15 at 9:57
  • It's a mixture of just being curious how such a grammar would look like (as I wrote, the linguists at the University of Stuttgart, where I studied, had a quite elaborate grammar - which I once saw in action parsing very nicely some random sentences from newspapers; back than they mostly had trouble parsing rhetoric figures which omit grammatical positions such as ellipses) and some personal research regarding the development of a "machine version" of German - meaning a subset of German which is easily understandable by today's computers. (Just to get an intuition, not to really implement!) – lukas.coenig Jan 1 '15 at 11:35
  • It doesn't have to be a CFG, btw. I'd just expect that most of German is context-free... – lukas.coenig Jan 1 '15 at 11:37
2

A bit of googling finds for example the following links to online-demos, software and other resources at FU Berlin:

These seem to be mostly for research, so I'd assume if you contact the respective owners they would probably agree to make the underlying grammars available to you for research/non-commercial purposes, if it's not already accessible via the links.

These don't seem to be CFGs, and I still think that a CFG would be unpractical as a description (though I'd indeed expect to be most of German to have a context-free representation).

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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