Work at SourceForge, help us to make it a better place! We have an immediate need for a Support Technician in our San Francisco or Denver office.

Close

Referring to "phonetic form"...

Help
2010-07-09
2013-04-08
  • Is there a way to refer to the "phonetic form", that is to the string corresponding to a sign, in openccg?

    It would be most useful in the treatment of idioms. For instance, 'go ahead' (with a special, non-compositional meaning) could have the category VP/NP (where ph is the feature encoding this fictitious "phonetic form").

    If in such a simple case one could obtain the same result by adding a feature similar to pform in verb subcategorization (e.g., pform=to in the second argument of ditransitive verbs), more complex idioms like 'kick the bucket' would be much more difficult to treat this way: instead than a relatively simple category like S/NP we should that the NP is determined ad includes a singular N with the nform 'bucket'.

    Does the ph feature really exist?

    Thank you,

    Giuliano Lancioni

     
  • Michael White
    Michael White
    2010-07-12

    Note that conventionally a 'lex' feature is often used instead of 'pform', as in 'pick up' with cat s\np/prt/np.

    Another option is to have special lex items like kick_the_bucket or just the_bucket, where the latter would have no semantics (as with particles).

    Beyond that, I don't see anything that's easy to do in openccg (or ccg for that matter), such as making the word sequence generally available as a syntactic feature.  But perhaps nothing really should be done in CCG.  Though I'm not up to date on the idioms literature, I believe a case can be made for re-interpreting the compositional meaning of 'kick the bucket' as an idiomatic meaning at the semantic level.  For example, to interpret a statement like 'a bucket list is a list of things to do before kicking the bucket', I'd think you'd need to represent 'kick the bucket' compositionally as well as idiomatically.