;;; -*- Mode: TDL; Coding: utf-8 -*- ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;;; file: japgram ;;; module: ;;; version: %I% ;;; written by: Melanie Siegel/Emily Bender ;;; last update: %G% --- %U% ;;; updated by: ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;;; author | date | modification ;;; ------------------|-------------|------------------------------------------ ;;;Melanie Siegel (MS)| | Emily Bender (ERB), Francis Bond (FCB), ;;; | | Chikara Hashimoto (CH), ;;; | | Takaaki Tanaka (TT), Akira Ohtani (AO) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;;; ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;Rules for sentence chains or single sentences ;declarative sentence, finite verb ; (do-parse-tty "食べる") utterance_rule-decl-finite := utterance-sf-type & [SYNSEM.LOCAL [CAT.HEAD.SMOD decl], C-CONT.HOOK.INDEX.SF prop, ARGS.FIRST.SYNSEM [LOCAL.CAT.HEAD [MODUS uttmodus, FIN +], NON-LOCAL.QUE ]]. ;wh-question without ka ; (do-parse-tty "何 を 食べる") utterance_rule-wh-without-ka := utterance-sf-type & [SYNSEM.LOCAL.CAT.HEAD.SMOD quest, C-CONT.HOOK.INDEX.SF ques, ARGS.FIRST.SYNSEM [LOCAL.CAT.HEAD [MODUS uttmodus, FIN +], NON-LOCAL.QUE ]]. ; (do-parse-tty "食べ て") utterance_rule-imp := utterance-sf-type & [SYNSEM.LOCAL.CAT.HEAD.SMOD imp, C-CONT.HOOK.INDEX.SF comm, ARGS.FIRST.SYNSEM [LOCAL.CAT.HEAD [MODUS uttmodus, H-TENSE te, VN -, COP -, J-SPEC 1-list], NON-LOCAL.QUE ]]. ; (ERB 2002-01-14) Don't want VN+kudasai to use this one anymore, ; so adding SMOD imp. ; (MS 2002-01-16) introduces simple-imperative vs. polite-imperative to divide utterance_rule10 and utterance_rule2e. ; (MS 2003-12-17) Adapting to Matrix 0.6 ; (MS 2005-12-08) We don't want the NONLOCAL.QUE condition on the argument, as embedded sentences; could have some QUE left in their storage: "誰が行くか教えて下さい。" ; EX: (do-parse-tty "食べ て 下さい") ; EX: (do-parse-tty "食べ て なさい") FIXME (one bad parse) utterance_rule-command := utterance-sf-type & [SYNSEM.LOCAL.CAT.HEAD.SMOD imp, C-CONT.HOOK.INDEX.SF comm, ARGS.FIRST [SYNSEM [LOCAL [CAT.HEAD [MODUS simple-imperative, H-TENSE present, FIN +], CONT.HOOK.INDEX.E.MOOD imperative], NON-LOCAL.QUE ], RMORPH-BIND-TYPE imp-morph]]. ; (MS 2004-12-14) an utterance rule for sentences with an infinite ending, like "偽りなく", ; at the moment restricted to adjectives. ;;; FIXME --- move to Fragments utterance_rule-decl-infinite := utterance-sf-type & [SYNSEM.LOCAL.CAT.HEAD.SMOD decl, C-CONT.HOOK.INDEX.SF prop, ARGS.FIRST.SYNSEM [LOCAL.CAT.HEAD [MODUS uttmodus, H-TENSE te, FIN -, ADJ +, VN -, COP -], NON-LOCAL [QUE , ROBUST +]]]. ;sentence with sentence end particle, finite verb ; (FCB 2006-05-09) Pass up SMOD, assume particle sets the mood ; this allows many rules to be collapsed utterance_rule-sep := utt-sep-type & [SYNSEM.LOCAL.CAT.HEAD.SMOD #1, ARGS.FIRST.SYNSEM.LOCAL.CAT.HEAD.SMOD #1]. ; (ERB 2001-11-05) These idioms probably shouldn't get the prpstn_rel. ; Also, this was turing up with unbound HCONS lists. ;;; FIXME: doesn't scope, harmonize with ERG (FCB 2006-05-08) ;;; (do-parse-tty "お早う") utterance_rule-idiom := utt-idiom-type & [SYNSEM.LOCAL.CAT.HEAD.SMOD decl]. ;; end of utterances ; (ERB 2001-10-03) Rule that adds unspec-loc-rel to temporal ; nouns, and makes them modifiers. adv_np_rule := adv_np_rule-type. ; (FCB 2005-06-30) moved supertype into rule-types. ;compound-noun-rule := compound-noun-rule-type & ; [SYNSEM.LOCAL.CAT.HEAD.MOD < [LOCAL.CAT.HEAD ordinary-or-date-noun_head] > ]. ;expanded from ordinary_noun_head to noun_head: dore kurai. ; (ERB 2001-10-10) For constructing k2ys it's helpful to have ; the unspecified relation between nouns in an nn compound not ; be an instance of prep_mod_rel. ; (ERB 2001-10-29) Trying to keep integers promoted to NumClPs ; promoted to NPs out of this rule. Trying the hack of giving them a ; REL value that is not a string. So, constraining this guy to ; be REL string. ; (ERB 2001-10-29) The REL names, written 'relation, are actually ; treated as atoms, not strings, by cheap. This means I need to ; move this contrast one level up. ; (MS 2001-11-26) Changed the modification from ordinary-noun_head to noun_head in order to allow compounds with exceptional nouns (mitai). ; (MS 2001-12-19) This solution gave too many ambiguities. I'll introduce a special rule for exceptional nouns. ; (ERB 2001-11-26) Like other modifiers, this should grab the MOD-IND ; and MOD-HAND. ; (ERB 2001-11-26) But it should pass up the LTOP and INDEX of its MOD value. ; (ERB 2001-11-26) Experimenting (as we so desire) with using ; a prep-mod-rel again here to allow compounds with nominalizations ; as the head. ; (ERB 2001-11-28) I don't think we need to worry about the INDEX and ; LTOP on the mother here. These only show up as non-head daughters of the ; compounds-rule, which is an intersective head-adj rule and therefore ; takes the head daugther as the semantic head. ; (ERB 2001-12-03) ARG3 should be the INDEX of the daughter, not ; the MOD-IND. ; (ERB 2002-01-28) MOD...HEAD is ordinary-or-date-noun_head to allow ; nouns like TOKI as head of compounds. (Was just ordinary_noun_head) ; (MS 2002-06-11) The POSTHEAD is unified with the POSTHEAD of the argument. Thus, I can prevent ; words like 間 from being the first argument in a compound. ; (MS 2003-09-26) Changed no_nmod to unspec-compound-relation, as defined in Matrix. ; (MS 2004-12-22) no MOD-IND/MOD-HAND ; (MS 2005-04-11) If the LKEYS.KEYREL.PRED is atom, we cannot parse more than two nouns in a compound. ; So, I'll take this restriction out. ;compound-noun-rule := empty-head-type & ;[SYNSEM [LOCAL [CAT [HEAD empty-nmod-p_head & ; [MOD < [LOCAL [CAT [HEAD ordinary-or-date-noun_head, ; VAL.SPR 1-list], ; CONT.HOOK [LTOP #hhand, ; INDEX #hinst]], ; MODIFIED.PERIPH -]>, ; J-POSTHEAD #ph], ; VAL saturated], ; BAR -], ; MODIFIED.PERIPH +], ; C-CONT [RELS , ; HCONS ], ; HEAD-DTR.SYNSEM [LOCAL [CONT nncompound-sem-type & ; [RELS , ; HCONS ]], ; LKEYS.KEYREL #key & unspec-compound-relation & ; [LBL #hhand, ; PRED "no", ; ARG0.E.TENSE no_tense, ; ARG2 #hinst, ; ARG1 #nhinst]], ; NON-HEAD-DTR.SYNSEM [LOCAL [CAT [HEAD noun_head & [J-POSTHEAD #ph], ; VAL saturated], ; CONT [ HOOK.INDEX #nhinst], ; BAR -]]]. ; (MS 2003-09-26) Changed no_nmod to unspec-compound-relation, as defined in Matrix. ; (MS 2004-12-22) no MOD-IND/MOD-HAND ;compound-exceptional-noun-rule := empty-head-type & ;[SYNSEM [LOCAL [CAT [HEAD empty-nmod-p_head & [MOD < [LOCAL [CAT [HEAD exceptional_noun_head, ; VAL.SPR 1-list], ; CONT.HOOK [LTOP #hhand, ; INDEX #hinst]], ; MODIFIED.PERIPH -] > ], ; VAL saturated], ; BAR -], ; MODIFIED.PERIPH +], ; C-CONT [RELS , ; HCONS ], ; HEAD-DTR.SYNSEM [LOCAL [CONT nncompound-sem-type & ; [RELS , ; HCONS ]], ; LKEYS.KEYREL #key & unspec-compound-relation & ; [LBL #hhand, ; PRED "no", ; ARG0.E.TENSE no_tense, ; ARG2 #hinst, ; ARG1 #nhinst]], ; NON-HEAD-DTR.SYNSEM [LOCAL [CAT [HEAD noun_head, ; VAL saturated], ; CONT [ HOOK.INDEX #nhinst], ; BAR -], ; LKEYS.KEYREL.PRED atom]]. ; (ERB 2001-11-28) Grab MOD-IND and MOD-HAND. ; (MS 2004-12-22) no MOD-IND/MOD-HAND ; (MS 2005-03-15) Grab CONT.HOOK.INDEX instead of LKEYS.KEYREL.ARG0! ;compound-name-rule := empty-head-type & ; [SYNSEM [LOCAL [CAT [HEAD empty-nmod-p_head & [MOD < [LOCAL [CAT [HEAD name_head, ; VAL.SPR 1-list], ; CONT.HOOK [LTOP #hhand, ; INDEX #hinst]], ; MODIFIED.PERIPH -] > ], ; VAL saturated], ; BAR -], ; MODIFIED.PERIPH +], ; C-CONT [RELS , ; HCONS ], ; HEAD-DTR.SYNSEM [LOCAL [CONT appos-id-sem-type & ; [RELS , ; HCONS ]], ; LKEYS.KEYREL #key & ; [LBL #hhand, ; ARG0 event, ; ARG1 #nhinst, ; ARG2 #hinst]], ; NON-HEAD-DTR.SYNSEM [LOCAL [CAT [HEAD name_head, ; VAL saturated], ; CONT.HOOK.INDEX #nhinst, ; BAR -]]]. ; (ERB 2001-08-06) Disallowing numbers from compounds. ; (ERB 2001-08-21) Changed type nonint_head to final_head. ; (ERB 2001-09-13) Changed type from head-adjunct-rule1-type to head-adjunct-rule-final, ; since I'm trying to get rid of head-adjunct-rule1-type. This looses the constraint ; that the mother's INDEX is identified with the non-head dtr's INDEX, so I'm adding ; that in here. ; (MS 2001-10-17) Taking the saturated restriction on the head-dtr out, such that it is possible to ; add a determiner to the compound. ; (ERB 2001-11-26) I don't want the INDEX/LTOP to be identified ; with MOD-IND/MOD-HAND on the mother of this rule, because it ; may be headed by a nominalization. As a first stab, say that ; INDEX and LTOP come from NON-HEAD-DTR and MOD-IND/MOD-HAND come ; from HEAD-DTR. ; (ERB 2001-11-28) See notes at head-adjunct-rule-type. ; (TT 2003-08-05) add "J-POSTHEAD right" ; (TT 2003-08-27) delete "J-POSTHEAD right" because it disabled lkb to parse NNN. ; (MS 2005-07-04) so, let's try with J-POSTHEAD compound ; (FCB 2006-06-06) moved the meat into rule-types. compounds-rule := compound-rule. ; (TT 2003-08-29) experimental rules for dealing with prefix and suffix correctly ; compounds-rule-nn := hadj-final-i & ; [C-CONT [RELS , ; HCONS ], ; ARGS <[SYNSEM [LOCAL [CAT [HEAD final_head & ; [ ; MAIN-PRD -, ; MOD < [MODIFIED.PERIPH #periph] > ], ; VAL saturated], ; BAR -], ; LEX -]], ; [SYNSEM [LOCAL [CAT [HEAD final_head ], ; BAR -], ; LEX -, ; MODIFIED.PERIPH #periph]]>]. ; compounds-rule-prefix-n := hadj-final-i & ; [C-CONT [RELS , ; HCONS ], ; ARGS <[SYNSEM [LOCAL [CAT [HEAD final_head & ; [J-POSTHEAD left, ; MAIN-PRD -, ; MOD < [MODIFIED.PERIPH #periph] > ], ; VAL saturated], ; BAR -], ; LEX +]], ; [SYNSEM [LOCAL [CAT [HEAD final_head ], ; BAR -], ; LEX -, ; MODIFIED.PERIPH #periph]]>]. ; compounds-rule-n-suffix := hadj-final-i & ; [SYNSEM.LOCAL.CAT.HEAD.J-POSTHEAD posthead, ; C-CONT [RELS , ; HCONS ], ; ARGS <[SYNSEM [LOCAL [CAT [HEAD final_head & ; [ ; MAIN-PRD -, ; MOD < [MODIFIED.PERIPH #periph] > ], ; VAL saturated], ; BAR -], ; LEX -]], ; [SYNSEM [LOCAL [CAT [HEAD final_head & [J-POSTHEAD right]], ; BAR -], ; LEX +, ; MODIFIED.PERIPH #periph]]>]. head_subj_rule := head-subject-rule & [SYNSEM.LOCAL.BAR +]. ; (ERB 2001-08-06) Splitting this in two for Smith-style number names. ; (MS 2001-11-14) Took the BAR + restriction out in order to make compounds with subcategorizing nouns possible (振込先指定). Let's see if that overgenerates. ; (MS 2001-11-19) It does. Let's take the BAR from the head-dtr. ; (MS 2001-11-20) Still too much ambiguity. Everything is BAR + again. ; (ERB 2002-02-11) Making embedded-question-rule a binary rule, akin ; to head-complement rules because unary version was relying on LKB ; cycle detection to keep from overgenerating. As a result, need to ; block comp-int-lex heads from head-complement-hf-rule. Will do this ; with a hack: require MESSAGE.LIST e-list on head daughter. This hack ; should cease to work round about when it becomes unnecessary: when ; we overhaul the analysis of clauses in this grammar. ; (MS 2003-07-11) Changed the MESSAGE hack now. hf-complement-rule := head-complement-hf-type & [SYNSEM [LOCAL.BAR +, NON-LOCAL.AFFIX #afflist], HEAD-DTR.SYNSEM [LOCAL [CAT.HEAD final_head, BAR +], NON-LOCAL.AFFIX ], NON-HEAD-DTR.SYNSEM [NON-LOCAL.AFFIX #afflist, LOCAL.BAR +]]. hi-complement-rule := head-complement-hi-type & [SYNSEM.LOCAL.BAR +, HEAD-DTR.SYNSEM.LOCAL.CAT.HEAD init_head]. head-complement-affixbind-rule := head-complement-hf-type & [SYNSEM [LOCAL.BAR +, NON-LOCAL.AFFIX 0-dlist ], ARGS < [SYNSEM.NON-LOCAL.AFFIX 1-dlist & [LIST [FIRST #aff, REST < > & #last], LAST #last]], [SYNSEM.LOCAL.CAT.HEAD p_head & [PTYPE #aff]] >]. head-complement2-rule := head-complement2-type & [SYNSEM.LOCAL.BAR +]. ; (ERB 2001-09-29) I need a 2obl complements rule that works for ; date expressions, so I'm making one general one. Since elements with ; two obligatory complements are barred from the ordinary head-complements ; rules, I don't think there will be any ambiguity introduced. ;kara-made-rule := head-2obl-complements-type & ; [SYNSEM.LOCAL.BAR +, ; ARGS <[SYNSEM.LOCAL.CAT.HEAD noun_head], ; [SYNSEM.LOCAL.CAT.HEAD postp_head & [PTYPE made]]>]. ;parenthesis-rule := head-2obl-complements-type & ; [SYNSEM.LOCAL.BAR +, ; ARGS <[SYNSEM.LOCAL.CAT.HEAD noun_head], ; [SYNSEM.LOCAL.CAT.HEAD parent-end_head]>]. ;head-2obl-complements-rule := head-2obl-complements-type & ; [ SYNSEM [LOCAL.BAR +, ; NON-LOCAL.QUE ]]. ;this is a test: can one say that the modifier must always be saturated? ; (ERB 2001-09-13) Trying to move to a system where there are only two ; HMRs, one for prehead and one for posthead modifiers. The intersective ; modifiers themselves take care of the necessary handle identities. ; (ERB 2001-11-28) See notes at head-adjunct-rule-type. ; (ERB 2002-01-02) [NUCL -] on mother - see notes at ; head-adjunct-rule-type. hf-adj-i-rule := hadj-final-i & [SYNSEM.LOCAL [BAR +], C-CONT [RELS , HCONS ], ARGS <[SYNSEM.LOCAL [CAT [HEAD [J-POSTHEAD left, ADJ -, MAIN-PRD -], VAL saturated], BAR +]], [SYNSEM synsem & [LOCAL [BAR +]]]>]. ; (ERB 2001-11-07) I think this is part of analysis of low attachment ; of adverbs wrt ba/tara/etc. ; (ERB 2001-11-28) I think we don't need this anymore with MOD-IND and ; MOD-HAND and separate rules for scopal and intersective modifiers. ; (ERB 2001-09-21) I want to restrict NumClPs made out of intergers only ; (i.e., mother nodes of the empty-cardinal-numcl-rule) to modify plain ; nouns or compound nouns, but not NPs. This is particularly crucial because ; they are posthead modifiers. However, the head-adjunct-rule-2 is requiring ; both daughters to be BAR +. I'm going to try removing the BAR + ; constraint on the head daughter and see what happens. ; (ERB 2001-11-28) See notes at head-adjunct-rule-type. ; (MS 2005-09-30) The BAR + status on this gave us problems when having nominal suffixes ; in compounds, such as: "汎用機用" ; (MS 2005-10-18) So, we need prefix and suffix rules. ; (MS 2005-10-06) Playing around with VAL saturated on the second argument, to avoid ; spurious ambiguity in 前年三月 hi-adj-i-rule := hadj-first-i & [SYNSEM.LOCAL.BAR +, ARGS <[SYNSEM synsem], [SYNSEM.LOCAL [CAT [HEAD [J-POSTHEAD right, MAIN-PRD -], VAL saturated], BAR +]]>]. ; (ERB 2001-11-28) Rule for scopal modifiers. ; For some reason, 新しい could be the modifier here until I put ; in the constraint [VAL saturated]. I don't see why that should ; change things. (If I type the MOD.LOCAL of everything, it will ; be doubly ruled out, but that's a separate issue.) hf-adj-s-rule := hadj-final-s & [SYNSEM.LOCAL [BAR +], C-CONT [ RELS , HCONS ], ARGS <[SYNSEM.LOCAL [CAT [ HEAD [J-POSTHEAD left, MAIN-PRD -], VAL saturated ], BAR +]], [SYNSEM synsem & [LOCAL.BAR +]]>]. ; (MS 2003-06-11) Rule for scopal modifiers - head first ; (MS 2005-12-22) Let's try to take the BAR status from the head: ; "探しばかりする" hi-adj-s-rule := hadj-first-s & [SYNSEM.LOCAL [BAR #bar], C-CONT [ RELS , HCONS ], ARGS <[SYNSEM synsem & [LOCAL.BAR #bar]], [SYNSEM.LOCAL [CAT [ HEAD [J-POSTHEAD right, MAIN-PRD -], VAL saturated ], BAR +]]>]. ; (MS 2005-10-18) Prefixes and suffixes have a different BAR status. When we want a prefixed noun ; to be in a compound, it should not be BAR +: "政界再編成" prefix-rule := hadj-final-i & [C-CONT [RELS , HCONS ], ARGS <[SYNSEM.LOCAL [CAT [HEAD noun_mod_head & [J-POSTHEAD left, ADJ -, MAIN-PRD -], VAL saturated], BAR -]], [SYNSEM synsem & [LOCAL [BAR +]]]>]. suffix-rule := hadj-first-i & [ARGS <[SYNSEM synsem], [SYNSEM.LOCAL [CAT [HEAD noun_mod_head & [J-POSTHEAD right, MAIN-PRD -], VAL saturated], BAR -]]>]. ; (ERB 2001-11-26) The hallucinated topic-rel should take the MOD-IND ; of the head noun as its ARG3, not the INDEX. ; (ERB 2001-11-28) See notes at head-adjunct-rule-type. ; (ERB 2002-01-02) Allowing relative clauses to attach to either ; NP or N'. Need to constrain the quantify-n-infl-rule so that ; it only applies to unmodified elements, to avoid spurious ambiguity. ; (ERB 2002-01-02) [NUCL -] on mother - see notes at ; head-adjunct-rule-type. ; (ERB 2002-01-14) It may be desirable to split the relative clause rule in two ; to prefer low attachment of relative clauses: rel-cl-rule-1 requires a [NUCL ; +] head daughter, while rel-cl-rule-2 requires a [NUCL -] head ; daughter. ;(CH 2003-08-15) 主名詞を関係節のどこにも関連づけない (non gapped) ; *** 2003-08-14 の Melanie Siegel さんの情報に基づく修正 *** ; there are three thing to change on the first daughter: ; 1) MAIN-PRD bool ; 2) NUCL not_fin_conj ; 3) The co-indexing of #adjevent to CONT.HOOK.INDEX instead of ARG0. ; ;(CH 2003-08-22) ; "ARGS.REST.FIRST.SYNSEM.LOCAL.CAT.HEAD noun_head" を追加 ; (MS 2004-12-13) This had [ADJ -] on its argument, preventing the correct ; reading for 目が美しい猫. ; (FCB 2005-06-15) Added ARG0 to messages relative-clause-rule := hadj-final-for-rel-cl-topic & [SYNSEM.LOCAL [BAR +, NUCL nucl_minus], C-CONT [RELS , HCONS ], ARGS <[SYNSEM [LOCAL [CAT [HEAD [ADJ bool, J-POSTHEAD rels, MAIN-PRD bool], VAL saturated], CONT.HOOK [ INDEX #adjevent, LTOP #adjhand], BAR +, NUCL not_fin_conj]]], [SYNSEM synsem & [LOCAL [CAT.HEAD noun_head, BAR +, CONT.HOOK.INDEX #ind]]]>]. ; (ERB 2002-01-02) [NUCL +] on head daughter seems overly ; restrictive (esp. since I'm trying to make all head-adjunct rules ; [NUCL -] on the mother). I'm guessing it was here to reduce ; ambiguity in sentences with multiple te clauses. Perhaps something ; similar could be achieved by distinguishing te-coordination on ; finite clauses and te-coordination on other te-clauses. ; (ERB 2002-01-14) Allow comma-headed conjuncts in this rule by ; generalizing to verb_head-super. ; (ERB 2002-01-22) NUCL conj now on supertype. ; (MS 2003-12-19) Matrix 0.6 ; (MS 2004-06-25) [HEAD.EMPTY -] to prevent spuriou ambiguity on 食べているご飯 sentence-te-coordination-rule := sentence-coord-type & [SYNSEM [LOCAL [BAR +, CAT.VAL saturated, CONT.HOOK.XARG #xarg]], C-CONT.RELS.LIST.FIRST.PRED "coord", ARGS <[SYNSEM.LOCAL [CAT [HEAD verb_head-super & [J-POSTHEAD coord, C-MOD < synsem & [LOCAL [CAT #1, CONT #3, BAR #5, NUCL #6], LEX #2, NON-LOCAL #4] > ], VAL saturated], CONT.HOOK.XARG #xarg, BAR +], J-NEEDS-AFFIX -], [SYNSEM synsem & [LOCAL [BAR +, CAT #1 & [VAL saturated, HEAD.EMPTY -], CONT #3, BAR #5, NUCL #6, NUCL nucl_plus], LEX #2, NON-LOCAL #4], J-NEEDS-AFFIX -] > ]. ; (MS 2004-06-25) We need something for the XARG. Try the XARG of the second conjunct. ; (FCB 2006-03-12) FIXME: I am pretty sure it shouldn't be that, but I don't know what it should be. conj-rule := conj-rule-type & [SYNSEM [LOCAL [BAR +, CONT.HOOK.XARG #xarg], MODIFIED.PERIPH +], ARGS <[SYNSEM.LOCAL [CAT [HEAD [J-POSTHEAD left, C-MOD < [MODIFIED.PERIPH #periph] > ], VAL saturated], BAR +]], [SYNSEM synsem & [LOCAL [BAR +, CONT.HOOK.XARG #xarg], MODIFIED.PERIPH #periph]]>]. ; ERB (2001-09-03) Redefining the vn-light-rule to be a kind of a head-specifier rule. ; This is so that the light verbs can have some say in the valence of the whole ; combination. In particular, I want dekiru to be able to ga-mark the object. ; ERB (2001-09-03) Turns out we don't need a special subtype of the h-spr rule, as ; the ordinary one will do the work (of course!). ; (MS 2003-12-19) Matrix 0.6 ; (MS 2004-01-09) grab the KEYS from the VN. vn-light-rule := head-marker-rule & [SYNSEM [LOCAL [BAR bool, CAT.HEAD.KEYS #keys, CONT.HOOK #hook], LEX +], J-NEEDS-AFFIX -, RMORPH-BIND-TYPE #morphbind, ARGS <[SYNSEM synsem & [LOCAL [CAT.HEAD vn_head & [KEYS #keys], CONT.HOOK #hook, BAR -]]], [SYNSEM synsem & [LOCAL [CAT.HEAD light-verb_head & [AUX aux_minus, LIGHT +]]], INFLECTED +, J-NEEDS-AFFIX -, RMORPH-BIND-TYPE #morphbind]>]. ; (ERB 2002-01-02) Adding a rule for VN+vends (esp. VN + kudasai). ; Pulled out similarities between new rule and vstem-vend-rule to ; supertype vstem-vend-rule-type. ; (MS 2005-11-14) MAIN-PRD should come from the stem, otherwise we get ; spurious ambiguity with the ga-no-conversion rule. ;;; CH+FCB+TK added LEX + ;;(lkb::do-parse-tty "投げ た") vstem-vend-rule := stem-end-rule-type & [ SYNSEM [LOCAL.CAT.VAL.SPR null, MODIFIED.PERIPH #per], ARGS < [ SYNSEM [LOCAL.CAT.HEAD ordinary-verb-stem_head & [LIGHT #light, MAIN-PRD #main], MODIFIED.PERIPH #per, LEX +], J-NEEDS-AFFIX + ], [SYNSEM.LOCAL.CAT.HEAD verbend_head & [LIGHT #light, MAIN-PRD #main]] >]. ;;(lkb::do-parse-tty "投げ そこなっ た") ;;; CH+FCB+TK added LEX + ;;;「た」が「投げ そこなっ」と「そこなっ」のどちらに結び付くかの曖昧性をなくすため aux-vend-rule := stem-end-rule-type & [ SYNSEM [LOCAL.CAT.VAL.SPR #spr, MODIFIED.PERIPH #per], J-NEEDS-AFFIX #aff, ARGS < [ SYNSEM [LOCAL.CAT [HEAD other-verb-stem_head & [COP -], VAL.SPR #spr], LEX +], J-NEEDS-AFFIX + ], [J-NEEDS-AFFIX #aff, SYNSEM [LOCAL.CAT.HEAD verb_head, MODIFIED.PERIPH #per]] >]. na-vend-rule := stem-end-rule-type & [ SYNSEM.LOCAL.CAT.VAL.SPR null, J-NEEDS-AFFIX #aff, ARGS < [ SYNSEM.LOCAL.CAT.HEAD na-adj_head], [J-NEEDS-AFFIX #aff, SYNSEM.LOCAL.CAT.HEAD verbend_head] >]. ; (MS 2004-11-08) This one should keep the LMORPH-BIND-TPYE: 来やすい ; (MS 2004-12-21) This doesn't necessarily have to be FIN +: 偽りなく ; (S 2005-01-06) Copy up ARG-S. unary-vstem-vend-rule := unary-type-super & [SYNSEM [LOCAL [CAT #cat, CONT #cont, CTXT #ctxt, ARG-S #args, BAR +]], J-NEEDS-AFFIX -, LMORPH-BIND-TYPE #lmorph, ARGS < [SYNSEM [LOCAL [CAT #cat & [HEAD verb-stem_head & [FIN bool, VN -, LIGHT -, EMPTY -]], CONT #cont, CTXT #ctxt, ARG-S #args, BAR -], LEX +], LMORPH-BIND-TYPE #lmorph, J-NEEDS-AFFIX -] >]. ; (ERB 2002-01-02) Getting double parses for "お送りいただきたい" ; because いただきたい is compatible with the head of hsr and the ; vend of vn-vend-rule. (Similar problems don't arise with ; 食べていただきたい because 食べて is also [HEAD v-end_head] and ; therefore incompatible with the vstem-vend rule.) This is a ; bit of a hack, but keep the vn-vend-rule from appying in these ; cases by requiring vend to be [AUX -]. ; (MS 2002-11-29) Do we need this rule??? ;vn-vend-rule := stem-end-rule-type & ;[ ARGS < [SYNSEM.LOCAL.CAT.HEAD vn_head], ; [SYNSEM.LOCAL.CAT.HEAD.AUX -] >]. ; (MS 2003-12-19) Matrix 0.6 vend-vend-rule := head-marker-rule & [SYNSEM [LOCAL [NUCL nucl_plus, CONT.HOOK #hook, BAR #bar], LEX +, MODIFIED.PERIPH #per], J-NEEDS-AFFIX #aff, LMORPH-BIND-TYPE #lmorph, RMORPH-BIND-TYPE #rmorph, ARGS <[SYNSEM synsem & [LOCAL [CAT [HEAD verbend_head & [MODUS #mod], VAL.SPR 1-list], BAR #bar]], J-NEEDS-AFFIX +, RMORPH-BIND-TYPE #mt, LMORPH-BIND-TYPE #lmorph, INFLECTED +], [SYNSEM synsem & [LOCAL [CAT.HEAD verbend_head & [MODUS #mod], CONT.HOOK #hook], MODIFIED.PERIPH #per], J-NEEDS-AFFIX #aff, LMORPH-BIND-TYPE #mt, RMORPH-BIND-TYPE #rmorph, INFLECTED +]>]. ;seminaa ga ichinichijuu haitte orimashite ; (ERB 2002-01-02) For some reason, we weren't copying up the ; MOD-IND and MOD-HAND in this rule. It might be a more general ; problem, but fixing it here for now. ; (MS 2004-12-22) no MOD-IND/MOD-HAND head-specifier-rule := head-specifier-rule-type & adjacent-spr-check & [SYNSEM [LOCAL [BAR #bar, CAT.VAL [UNSAT -, SPR null], CONT.HOOK [XARG #xarg, INDEX #mind, LTOP #mhand]], LEX -, MODIFIED.PERIPH #per], ARGS <[SYNSEM synsem & [MODIFIED.PERIPH #per], J-NEEDS-AFFIX -], [SYNSEM synsem & [LOCAL [BAR #bar, CONT.HOOK [XARG #xarg, INDEX #mind, LTOP #mhand]]]]>]. ; (MS 2005-06-20) AHON needs to be copied, because the prefixes are often honorific. prefix-attach-rule := prefix-rule & [SYNSEM.LOCAL.CAT.HEAD.FORMAL.AHON #ahon, ARGS < [SYNSEM.LOCAL [BAR -, CAT.HEAD hon-prefix_head & [FORMAL.AHON #ahon] ], J-NEEDS-AFFIX -], [SYNSEM.LOCAL.BAR bool, J-NEEDS-AFFIX -] >]. ; (MS 2003-12-19) Matrix 0.6 ; (do-parse-tty "ビール が 飲み たい") tai-obj-change-rule := obj-casechange-rule & [SYNSEM.LOCAL.CAT.VAL.COMPS.FIRST [OPT - , LOCAL.CAT.HEAD overt-case-p_head & [CASE ga]], ARGS < [SYNSEM [LOCAL [CAT [HEAD v-op-end_head & [KEYS.KEY tai_v_want_rel], VAL [SPR null, SUBJ null]]]]] >]. ;; 機関の変更したい vn-light-obj-change-rule-case-p := obj-casechange-rule & [SYNSEM.LOCAL.CAT.VAL.COMPS.FIRST [OPT - , LOCAL.CAT.HEAD overt-case-p_head & [CASE no-case]], RMORPH-BIND-TYPE #rmorph, ARGS < [SYNSEM.LOCAL.CAT [HEAD light-verb_head, VAL.SPR null], RMORPH-BIND-TYPE #rmorph] >]. vn-light-obj-change-rule-noun := obj-casechange-rule & [SYNSEM.LOCAL.CAT.VAL.COMPS.FIRST [OPT - , LOCAL.CAT [HEAD noun_head, VAL saturated]], RMORPH-BIND-TYPE #rmorph, ARGS < [SYNSEM.LOCAL.CAT [HEAD light-verb_head, VAL.SPR null], RMORPH-BIND-TYPE #rmorph] >]. ; (ERB 2001-09-26) Constraining head-dtr to be [HEAD verb_head-avm] ; so that this rule can't apply after the vstem-vend-rule. ; (FCB 2006-08-09) Constraining things to be BAR - for the same reason ; FIXME why do this and tai-obj-change-rule come on opposite sides of the vend-rule ; (do-parse-tty "ビール が 飲める") ; (do-parse-tty "犬 が 食べ られる") rareru-obj-change-rule := obj-casechange-rule & [SYNSEM.LOCAL [CAT.VAL [COMPS.FIRST.LOCAL [ CAT.HEAD overt-case-p_head & [CASE ga]]], BAR +], ARGS < [SYNSEM [LOCAL [ CAT [HEAD verb_head & [KEYS.KEY rareru_v_can_rel, MODUS potential], VAL.SUBJ null], BAR +]]] >]. ; (MS 2004-12-10) I don't want this to apply to copula verbs (generation). ; FIXME doesn't apply now ;(do-parse-tty "犬 を 欲し がる") garu-sbj-change-rule := subj-casechange-rule & [SYNSEM.LOCAL.CAT.VAL [SUBJ.FIRST [LOCAL.CAT.HEAD overt-case-p_head & [CASE wo]]], ARGS < [SYNSEM [LOCAL.CAT [HEAD [KEYS.KEY garu_v_seem_rel, COP -], VAL.SUBJ.FIRST.LOCAL.CAT.HEAD.CASE ga]]] >]. ; (ERB 2001-09-20) We want currency NumClPs at least to be possible ; left daughters of compounds for things like 百円ノート or ; 二万ドル車. That these are actually compounds is shown by the ; pronunciation of the latter (niman doru guruma), which involves rendaku. ; (MS 2001-10-10) But if you restrict them to be BAR -, they can (e.g.) not be modified. Let them be BAR bool and everything is possible? ; (ERB 2001-11-01) Breaking this up into three rules. See notes ; in ruletypes.tdl. ; (ERB 2004-1-20) Trying to conflate the two noncards. ;nominal-numcl-rule-nomod := nominal-numcl-rule-nomod & ; [ SYNSEM.LOCAL.BAR bool ]. ;nominal-numcl-rule-mod := nominal-numcl-rule-mod & ; [ SYNSEM.LOCAL.BAR bool ]. nominal-numcl-rule := nominal-numcl-rule-type-noncard. ; (ERB 2004-1-20) Try getting rid of this one in favor of ; always jusing the noncard variant. ;nominal-numcl-rule-card := nominal-numcl-rule-card & ; [ SYNSEM.LOCAL.BAR + ]. ; ERB (2001-09-24) Overhauling the date expressions. They will be ; nouns to start with, so we don't need this. ;nominal-date-rule := nominal-date-rule-type & ; [ SYNSEM.LOCAL.BAR + ]. ; (ERB 2004-1-20) This seems to be working now, but it's creating ; huge amounts of ambiguity. I suggest leaving it out until we ; have a means of putting in sortal constraints to keep these guys ; from serving as modifiers of other number names which are ; serving as nouns. ; empty-cardinal-numcl-rule := empty-cardinal-numcl-type. ; (ERB 2001-11-14) Unary rule to bind off QUE values in embedded ; questions. ; (ERB 2002-02-11) Changed to binary rule see notes at embedded-rule-type. ; If and when we make the embedded-question-rule a subtype of hcomp-rule-type, ; the constraints on AFFIX could probably be moved up. embedded-question-rule := embedded-question-rule-type & [SYNSEM.NON-LOCAL [QUE , AFFIX #affixlist], HEAD-DTR comp-int-lex & [SYNSEM.NON-LOCAL.AFFIX ], NON-HEAD-DTR.SYNSEM.NON-LOCAL.AFFIX #affixlist]. quasi-morph-rule := binary-basic & [SYNSEM [LOCAL [CAT [HEAD #head, VAL #sc], CONT #cont], NON-LOCAL.QUE [ LIST #firstque, LAST #lastque ]], ARGS < [SYNSEM [LOCAL [CAT [HEAD ordinary_noun_head, VAL.UNSAT +], BAR -], NON-LOCAL.QUE [ LIST #firstque, LAST #middleque ]]], [SYNSEM [LOCAL [CONT #cont, CAT [HEAD na-adj-basic_head & #head, VAL #sc & [UNSAT +]]], NON-LOCAL.QUE [ LIST #middleque, LAST #lastque ]]] >]. ;(CH 2003-08-15) 動詞語幹+動詞語幹 ;(FIXME) semantics がおかしい場合がある。 ; ex) V2がV1のイベントを取る場合の解釈ができない。 ; almost lost the ambiguity with VAL.UNSAT + ; but for some reason subj-zpro-ins-lrule won't run after the vstem-vstem rule ; because of a conflict of SUBJ NUL and sat-arg-list ; (MS and FCB 2003-10-29) With UNSAT - one gets rid of the ambiguity, but the ; identification of VAL is not correct; and one cannot get GA-WO. ; The valence composition must be re-worked here. ; (MS 2004-02-25) No, UNSAT - is also not right. One doesn't get readings for both ; zero pronouns and subjects, playing around with UNSAT. ; Does the valence come from V1? Identifying the agent? ; (MS 2005-02-14) Moved the definitions to rule-types.tdl. ; (FCB 2005-05-24) replaced by richer rules *-lex-vv-rule ;vstem-vstem-rule := vstem-vstem-rule-type. ga-wo_transitive-lex-vv-rule := ga-wo_transitive-lex-vv-rule-type. intransitive-lex-vv-rule := intransitive-lex-vv-rule-type. ;(CH 2003-08-27) ; 形容詞を名詞に変える。 ;「古 さ」などを解析するため。 ; (MS 2005-02-14) Moved the definitions to rule-types.tdl. adj-nominal-suffix-rule := adj-nominal-suffix-type. ; ADV -> NOUN ; 「重要 さ」 ; adv-nominal-suffix-rule := head-final-type & ; [SYNSEM.LOCAL.CAT.VAL.UNSAT -, ; NON-HEAD-DTR [SYNSEM [ LOCAL [CAT [HEAD na-adj_head, ; VAL.UNSAT -], ; BAR -], ; LEX +], ; RMORPH-BIND-TYPE adv_head], ; HEAD-DTR [SYNSEM [LOCAL [CAT [HEAD ordinary_noun_head-avm & ; [J-POSTHEAD right], ; VAL.UNSAT +], ; BAR -]]]]. ;;; Fragment rules ;;; unary rules ;犬 frg-np := frag-np-rule. ;東京 へ frg-pp := frag-pp-rule. ;たぶん frg-s-adv := frag-s-adv-rule. ;少し ;とて frg-i-adv := frag-i-adv-rule. ;;; binary fragment rules ;東京 へ 彼 frg-pp-np := frag-pp-np-rule. ;少し 彼 frg-i-adv-np := frag-i-adv-np-rule. ;そこ へ 2 frg-pp-int := frag-pp-int-rule. ;犬 が 動物 frg-cop := frag-cop-rule.