Now that I fixed a few broken ICON appends, I have the following regression for clausal complements (open in new tab to make it bigger):
(1) Do I think that the cats chase the dogs?
Look at the very last line, there is new ICONS on the current result.
Does it make any sense? I don’t think so, seeing as it is an underspecified information structure relation? So, probably need to get rid of it? (Not sure how yet.)
(Edited the original post a bit to not have unnecessary information.)
This happens because of the following in matrix.tdl:
; Sandy believes that Kim sleeps.
; Sandy ga Kim ga neru to shinjita.
; Sandy croit que Kim dors.
; ERB 2007-01-22 Inserting qeqs now that we no longer have messages
; intervening.
; changing: added cat-sat (PASSES all reg tests)
clausal-second-arg-trans-lex-item := basic-two-arg & one-icons-lex-item &
[ ARG-ST < [ LOCAL [ CAT cat-sat,
CONT.HOOK [ INDEX ref-ind & #ind,
ICONS-KEY.IARG1 #clause ] ] ],
[ LOCAL.CONT.HOOK [ LTOP #larg,
INDEX #target ] ] >,
SYNSEM [ LOCAL.CONT [ HOOK.CLAUSE-KEY #clause,
HCONS.LIST < qeq & [ HARG #harg,
LARG #larg ] >,
ICONS.LIST < [ IARG1 #clause, IARG2 #target ] > ],
LKEYS.KEYREL [ ARG1 #ind,
ARG2 #harg ] ] ].
See particularly the ICONS.LIST.
Does that seem wrong, then? I didn’t put it there so I am not sure…
…Furthermore, the gold MRS used to have such ICONS for sentences without auxiliaries (where append was not broken). So, should normalize to either one or the other. I think by default, I will normalize to what it was before ( < e2 info-str e25 > ), for all sentences. Unless I hear that it should in fact be the opposite.