2

Formally representing constituency based structures, using brackets to denote constituents, is straight-forward - until discontinuities are thrown in the mix.

Consider this example, adapted from this Wikipedia article:

Sarah tutors John

Which could, for the sake of argument, be bracketed as follows:

{ {Sarah} {tutors {John } } }

By means of WH-fronting, we could easily turn this sentence into:

Who has Sarah been tutoring?

Visually representing this is quite easy - you simply cross some lines on a piece of paper (example, again, pilfered from Wikipeda):

enter image description here

How would you go about representing it via a bracketing system, though?

Improve this question
10
  • Why not represent 'movement' as actual movement? MIles Shang's Syntree does this to some extent: try entering [? [text [S [np [Who_A]] [aux.inv [aux [has_B]] [vp [np.su Sarah][vp.prf [auxHAVE.3s _<B>][vp.prg [auxBE.pa.ppl been][vp [v.pr.ppl tutoring][np.do _<A>]]]]]]]] [punc ?]].
    – StoneyB on hiatus
    Commented Apr 4, 2017 at 1:33
  • 1
    There is no movement to speak of in my case. I'm coming from a computational perspective - generating training data for sentence parsers. I need to formally represent discontinuous surface structures.
    – player.mdl
    Commented Apr 5, 2017 at 10:57
  • 1
    Bracketing presupposes that there is no discontinuity. To accommodate discontinuity you need some sort of anaphoric device which correlates an existing remote constituent and a local 'empty' constituent. After that it's just a matter of telling your parser how to represent that, whether as line crossing or movement arrows or some other convention.
    – StoneyB on hiatus
    Commented Apr 5, 2017 at 11:56
  • 1
    Thought as much :) Are you aware of any literature that explicitly discusses this limitation? Would love to read up more. Also please add your comments as answer. I'd like to accept it.
    – player.mdl
    Commented Apr 5, 2017 at 12:48
  • 1
    I'm strictly an amateur in this, winging it for my own purposes, so most of the literature I've encountered is too technically formalized for me to judge its validity; and the notations and formalizations are tied to specific theories/approaches I'm usually unfamiliar with. (Your notation, for instance, where a word acts as its own parent, is very odd to me!) Consequently I'm not comfortable putting up an Answer, as if I actually knew what I was talking about.
    – StoneyB on hiatus
    Commented Apr 5, 2017 at 13:03

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.