Wh-questions in 567: wh-ques-phrase and SUBJ

(10-minute rule)

Here are the instructions on how to implement wh-questions for the 567 class:

basic-head-filler-phrase :+

wh-ques-phrase := basic-head-filler-phrase & interrogative-clause & 
		  head-final &
   [ SYNSEM.LOCAL.CAT [ MC bool,
			VAL #val,
			HEAD verb & [ FORM finite ] ],
				 VAL #val & [ SUBJ < >,
					      COMPS < > ] ],
     NON-HEAD-DTR.SYNSEM.NON-LOCAL.QUE <! ref-ind !> ].

extracted-comp-phrase := basic-extracted-comp-phrase &

extracted-subj-phrase := basic-extracted-subj-phrase &

Assuming you also add a pronoun:

wh-pronoun-noun-lex := norm-hook-lex-item & basic-icons-lex-item & 
  [ SYNSEM [ LOCAL [ CAT [ HEAD noun,
			   VAL [ SPR < >,
				 SUBJ < >,
				 COMPS < >,
				 SPEC < > ] ],
	                    RELS <! [ LBL #larg,
				       ARG0 #ind & ref-ind ],
				  [ PRED "wh_q_rel",
				    ARG0 #ind,
				    RSTR #harg ] !>,
			    HCONS <! [ HARG #harg,
				        LARG #larg ] !> ] ],
	     NON-LOCAL.QUE <! #ind !> ] ].
who := wh-pronoun-noun-lex &
  [ STEM < "who" >,
    SYNSEM.LKEYS.KEYREL.PRED "_person_n_rel" ].

– According to the instructions, this this should lead to Who chases the dog parsing as a question.

Here’s what I observe, starting with mini-English and making the above additions:

  1. No parse for Who chases the dog; here’s the parse chart:


  1. A unification failure in the SUBJ of the Head-Daughter:

  1. When I inspect the VP “chases the dog”, it says it is licensed by the head-comp rule, but I don’t see any daughters in the AVM! What happened to them, how do I find out?

I think the rule that’s meant to combine those two is the head-filler rule. Have you instantiated that yet?

As for where the daughters went – I think the ‘local AVM’ at any point in the tree doesn’t have them anymore. If you want to see the relationship to the daughters, you can look at the rule definition.

The wh-ques rule that I am inspecting there (which gives the failure at SUBJ) inherits from head-filler.

Hmm maybe when there is no parse it doesn’t? I think when I have a parse, then I see them even at the VP level.

Oh! That suggests that the subject extraction rule hasn’t applied then.

Sorry – too jetlagged just now to help further on this point. Ask me again later this week?

The daughters don’t show up in LUI windows, but they should be there in full in the LKB if you turn LUI off. The LKB (in common with all the DELPH-IN processors) discards daughter feature structures while parsing and generating, but reconstructs them when displaying the results in case the grammarian wants to inspect them. Select “Feature structure” from the menu at the top node of a single Parse Tree window.

1 Like

Oh duh! Thanks. Of course, extracted-subj-phrase is also a rule, so, should also go to rules.tdl (like wh-ques-phrase).