I hope I've got the right forum. I want to know about English specifically, although this is a linguistics question.

A common task in NLP and Computational Linguistics is to generate parse trees for various sentences. How these sentences are parsed, for one, depends upon POS tagging which itself depends upon how to enumerate the Part of Speeches of the language.

What I am interested in is sort of a reverse view of the process. Just as I can ask what are the most common words for the English Language, I wish to ask what are the most common parses for the English Language which are found, if in fact this can be measured or approximated? I am interested in the fully expanded representations, not something as basic as NP VP. For example: one might render:

The bat eats a cat

d    n   v   d  n

using one simplistic POS enumeration. Looking at this parse, I would ask what percentage of sentences in English follow this exact pattern?

  • I've flagged this non-prejudicially as being better suited to the Linguistics site. – StoneyB Aug 28 '12 at 18:25
  • 6
    Be careful what you wish for. As Abney (1995) demonstrates, far more sentences in English can be parsed in any particular way than ever would be parsed that way by a human. – John Lawler Aug 28 '12 at 18:26
  • @StoneyB That makes sense, but sometimes when you say a specific language, they frown on your question. – demongolem Aug 28 '12 at 18:26
  • 2
    maybe you should be asking that at the computational linguistics Q&A linguistics.stackexchange.com? – Mikhail Kozhevnikov Aug 28 '12 at 18:27
  • That's more general linguistics, though also some computational linguists hang out there, too. There ought to be a S.E for computational linguistics, though I suspect they've already got something equivalent of their own going. – John Lawler Aug 28 '12 at 18:29

I would utilize projects like

plus some programming skills to get such statistics.

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.