I am reproducing the discussion https://github.com/delph-in/pydelphin/issues/200 here since the same output was obtained with ACE and LKB_FOS.
In the sentence below, see that
x3 variable is not the
_tie_v_1 as expected. The
i9 variable was introduced and the fragment
a man is disconnected from the rest of the graph. Is that a bug in the parsers or a problem in the grammar ?? Or maybe none of them ??!
SENT: A man is untying a shoe.
[ LTOP: h0
INDEX: e2 [ e SF: prop TENSE: pres MOOD: indicative PROG: + PERF: - ]
RELS: < [ _a_q<0:1> LBL: h4 ARG0: x3 [ x PERS: 3 NUM: sg IND: + ] RSTR: h5 BODY: h6 ]
[ _man_n_1<2:5> LBL: h7 ARG0: x3 ]
[ _tie_v_1<9:16> LBL: h1 ARG0: e8 [ e SF: prop TENSE: pres MOOD: indicative PROG: + PERF: - ] ARG1: i9 ARG2: x10 [ x PERS: 3 NUM: sg IND: + ] ]
[ _un-_a_rvrs<9:16> LBL: h1 ARG0: e2 ARG1: e8 ]
[ _a_q<17:18> LBL: h11 ARG0: x10 RSTR: h12 BODY: h13 ]
[ _shoe_n_1<19:24> LBL: h14 ARG0: x10 ] >
HCONS: < h0 qeq h1 h5 qeq h7 h12 qeq h14 >
ICONS: < > ]
Would it make sense for you to try and see if there is a simpler parse in such cases (e.g. one without analyzing untie as un- and tie)? But this suggestion is of course off topic.
sure, that is one approach, but nevertheless, it is a bug in ERG, right? That analysis should not be one possible analysis anyway. Am I right?
Following @goodmami suggestion, I have also reported the issue at https://github.com/delph-in/erg/issues/9.
Yesterday I was reading again the slides from the LREC 2016 workshop and if I understood it right, this
un- prefix case is something recent to ERG: http://moin.delph-in.net/ErsTutorial.
Alexandre and others,
You’re right that this is due to a bug in the ERG’s treatment of verb-prefixing rules; somehow, the grammar was failing to preserve the CONT.HOOK.XARG from daughter to mother. I have fixed this, and will check it in to the trunk ERG by the end of this week, once I’ve done some further testing. I’ll follow up on the github issue tracking as well - should have seen it earlier!