;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;;; Ginzburgh and Sag (2002) Implementation ;;; Copyright (c) 2001-2002, John Beavers, Chris Callison Burch, and Ivan Sag ;;; see license.txt (distributed with LKB) for details ;;; ;;; Filename: wh-words.tdl ;;; Purpose: Contains lexical definitions for wh-words and related things. ;;; This is a lexeme file not a type file. ;;; Last modified: 05/11/02 by John Beavers (JTB) ;;; Notes: ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; hell_0 := word & [ ORTH , SS [ LOC [ CAT [ HEAD noun & [ MOD < [ WH , LOC.CAT [ SUBJ < >, COMPS < >, SPR < > ] ] > ], SUBJ < >, COMPS < >, SPR < > ], CONT soa & [ QUANTS , NUCL emphasis-rel & [ EMPH-BY speaker ] ], Q-STORE , P-STORE ], SLASH , WH ], ARG-ST < >, DTRS < inflected-lexeme > ]. ;;; CCB - I'm guessing at the details of this from the tree in section ;;;5.2 for "Whose cousin's friend"... ;;;JTB - made all of these unfocused since that seemed to be the ;;;generalization they were grasping, and added a "what" determiner. ;;;JTB 07/23/01 - What the hell is DEG supposed to do again? inter_which-unfoc := wh-det-lxm-unfoc & [ ORTH , SS.LOC.Q-STORE ]. inter_which-foc := wh-det-lxm-foc & [ ORTH , SS.LOC.Q-STORE ]. ;;;JTB 07/25/01 - What to do about these? These are questioning the ;;;possessor. inter_whose-unfoc := wh-poss-det-lxm-unfoc & [ ORTH ]. inter_whose-foc := wh-poss-det-lxm-foc & [ ORTH ]. ;;;JTB 10/17/01 - Ok, here's a problem. These how's want their ;;;specified things to be adjectives with soas, but now we're assuming ;;;that attributive adjs have scope-objs. What if I took the extent ;;;coidentification to the adjective instead, and attribs can do one ;;;thing and preds can do another? That won't work because the extent ;;;thing is a rel property and not really refered to by adjs. I'd ;;;have to create two types of adjectives outright, gradable and ;;;non-gradable. This ruins the whole semantic distinction bit about ;;;these determiners. inter_how-unfoc := wh-det-lxm-unfoc & [ ORTH , SS.LOC [ CAT.HEAD [ DEG true, SPEC.NUCL.EXTENT #delta ], CONT [INDEX #delta, RESTR < >] ] ]. inter_how-foc := wh-det-lxm-foc & [ ORTH , SS.LOC [ CAT.HEAD [ DEG true, SPEC.NUCL.EXTENT #delta ], CONT [INDEX #delta, RESTR < >] ] ]. inter_what-unfoc := wh-det-lxm-unfoc & [ ORTH ]. inter_what-foc := wh-det-lxm-foc & [ ORTH ]. ;a_0 := det-lxm & ;[ ORTH , ; SS.LOC.CAT.SPR < [ LOC.CAT [ HEAD adj ] ] > ]. ;;; <<< >>> WHO ;;; 4 entries for each - one with a WH value and one without, times ;;;FOC vs not FOC.... CCB- These are messy large structures, containing ;;;a lot of redundant info. Reduce them when you get the chance ;;; CCB - taken from Ch 5 of G&S; should have a preson_rel in the ;;;STORE's RESTR once you figure out how to do that. ;;;JTB - 7/17/01 - Added focused but unmarked versions of all of these ;;;but left the spelled-out focused ones in. The ambiguity for each ;;;wh-word does have an effect on parsing, for instance, "who saw who" ;;;now how two possible readings: "who saw WHÓ?" and "WHÓ saw WHÓ?", ;;;both of which will appear when we enter "who saw who". This is ;;;predicted and fine but a bit undesirable for teasing out the two ;;;different sentences. interrogative_who := wh-n-lxm-unfoc & [ ORTH , SS.LOC.CONT.RESTR < [ PROP.SOA.NUCL person-rel ] > ]. interrogative_who-foc-2 := wh-n-lxm-foc & [ ORTH , SS.LOC.CONT.RESTR < [ PROP.SOA.NUCL person-rel ] > ]. interrogative_who-foc-1 := wh-n-lxm-foc & [ ORTH , SS.LOC.CONT.RESTR < [ PROP.SOA.NUCL person-rel ] > ]. ;;; <<< >>> WHOM interrogative_whom := wh-n-lxm-unfoc & [ ORTH , SS.LOC.CONT.RESTR < [ PROP.SOA.NUCL person-rel ] > ]. interrogative_whom-foc-2 := wh-n-lxm-foc & [ ORTH , SS.LOC.CONT.RESTR < [ PROP.SOA.NUCL person-rel ] > ]. interrogative_whom-foc-1 := wh-n-lxm-foc & [ ORTH , SS.LOC.CONT.RESTR < [ PROP.SOA.NUCL person-rel ] > ]. ;;; <<< >>> WHAT interrogative_what := wh-n-lxm-unfoc & [ ORTH , SS.LOC.CONT.RESTR < [ PROP.SOA.NUCL thing-rel ] > ]. interrogative_what-foc-2 := wh-n-lxm-foc & [ ORTH , SS.LOC.CONT.RESTR < [ PROP.SOA.NUCL thing-rel ] > ]. interrogative_what-foc-1 := wh-n-lxm-foc & [ ORTH , SS.LOC.CONT.RESTR < [ PROP.SOA.NUCL thing-rel ] > ]. ;;; CCB - add exist_rel to the WH valuse, as per ch 5. ;;; JTB 07/22/01 - Let's get the exclamatives going. exclamative_what := wh-lxm-excl & [ ORTH ]. exclamative_how := wh-lxm-excl & [ ORTH , SS.LOC [ CAT.HEAD.SPEC.NUCL.EXTENT #delta, CONT [INDEX #delta, RESTR < >] ] ]. ;;;All dead entries: ;;;JTB 7/10/01 - This may be ancient, let's kill it. ;interrogative_who-2 := wh-n-lxm & ;[ ORTH , ; SS [ LOC [ CONT [ INDEX #i ], ; STORE , ; FOC false ] !> ], ; WH *e-list-of-scope-objs* ] ]. ;;; JTB/IAS/CCB 07/10/01 - The constraint on WH is only applicable to ;;;fronted wh words, not accented ones. ;interrogative_who-foc := wh-n-lxm & ;[ ORTH , ; SS [ LOC [ CONT [ INDEX #i ], ; STORE ] !> ], ; WH ] ]. ;;;JTB 07/10/01 - Dead ;interrogative_whom-2 := wh-n-lxm & ;[ ORTH , ; SS [ LOC [ CONT [ INDEX #i ], ; STORE , ; FOC false ] !> ], ; WH *e-list-of-scope-objs* ] ]. ;;;JTB/IAS/CCB 07/10/01 - Also wrong, see above. ;interrogative_whom-foc := wh-n-lxm & ;[ ORTH , ; SS [ LOC [ CONT [ INDEX #i ], ; STORE ], ; WH ] ]. ;;;JTB 07/10/01 - Bye, bye, birdie ;interrogative_what-2 := wh-n-lxm & ;[ ORTH , ; SS [ LOC [ CONT [ INDEX #i ], ; STORE , ; FOC false ] !> ], ; WH *e-list-of-scope-objs* ] ]. ;;;JTB 07/10/01 - Same as above. ;interrogative_what-foc := wh-n-lxm & ;[ ORTH , ; SS [ LOC [ CONT [ INDEX #i ], ; STORE ] !> ], ; WH ] ].