2

In constituent analysis, we break down sentences into their main parts of speech (Det., N, Adj, V, etc.). English sentences have recursive syntactic patterns. For example, a simple sentence in English has several syntactic combinations:

  • The students finished their homework. [NP+V+NP]
  • She is beautiful [Sub. Pron. + V + Adj.]

Same thing apply to all other types of English sentences. I am looking for a comprehensive list of all the possible syntactic combinations of English sentences. Is there any reference that contains these rules? I am sure that such a resource exists somewhere because these rules were used in primitive Natural Language Processing applications under the name of symbolic approach or rule-based approach (as opposed to statistical approach).

  • 4
    There is no such list, since it would be infinite. Admissible syntactic constructions are finitely described by rules that generate them. For an adequate description one needs something more sophisticated than context-free rules though. – Atamiri Jul 24 '16 at 20:59
  • It looks like what you're after is a list of English phrase structure rules, not a list of "all possible syntactic combinations". For example your first sentence can be described with the phrase structure rules S -> NP VP, NP -> D N, VP -> V NP. – TKR Jul 24 '16 at 23:08
3

As Atamiri says, a complete list is likely to be impossible. You are right in observing that most work in NLP is no longer based on hand-crafted rules but on statistics obtained from large amounts of data.

However, projects which attempt to create hand-built broad-coverage grammars still exist: see the LinGO project by Dan Flickinger for an example of a broad-coverage English grammar. The term broad-coverage means that the grammar aims to include production rules that assign an analysis to as much English text as possible.

Here's the list of production rules for this particular grammar.

Another possibility is to extract production rules automatically from something like Penn Treebank. Those production rules, however, are quite sparse, since the Penn Treebank doesn't conform strictly to X-bar principles.

| 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.