Lab9 Writeup Can: ---- One of the ways to express in Hebrew the modal "can" is through a rasing verb that takes an infinitival complement. I am labeling it a verb rather than an auxiliary since it seems to behave as other verbs that take infinitival complements do (e.g. want). #1000 Source: author Vetted: t Judgment: g Phenomena: modals ani ikwl lakl zkwkit ani ikl:PAAL.PRES akl:PAAL.INF zkwkit pron:NOM.1SG can:PAAL.PRES.SG.M eat:PAAL.INF glass:SG.F 'I can eat glass.' #1001 Source: author Vetted: t Judgment: g Phenomena: modals ani rwch lakl zkwkit ani rci:PAAL.PRES akl:PAAL.INF zkwkit pron:NOM.1SG want:PAAL.PRES.SG.M eat:PAAL.INF glass:SG.F 'I want to eat glass.' It is a somewhat irregular verb in that it is missing third person masculine singular past tense form. To form the past tense for the third person masculine singular past tense form, the third person masculine singular past tense form of the verb (aux) "be" is used and it is followed by the present tense third person masculine singular of "can" (#1004). However, many Hebrew speakers supply this missing form (#1005), but this is substandard. This is irrelevant to the grammar at this point but I thought I'd mention it since it might be relevant to its eventual labeling as a verb or aux or whatever, as I guess aux verbs tend to be irregular. Though at this point I do not see any reason to label it as anything other than a verb. The only true aux in Hebrew I think is the verb "be". #1002 Source: author Vetted: t Judgment: g Phenomena: modals hia iklh lakl zkwkit hia ikl:PAAL.PAST-h akl:PAAL.INF zkwkit pron:NOM.1SG.F can:PAAL.PAST-3SG.F eat:PAAL.INF glass:SG.F 'She could eat glass.' #1003 Source: author Vetted: t Judgment: g Phenomena: modals hwa ikwl lakl zkwkit hwa ikl:PAAL.PRES akl:PAAL.INF zkwkit pron:NOM.1SG.M can:PAAL.PRES.SG.M eat:PAAL.INF glass:SG.F 'He can eat glass.' #1004 Source: author Vetted: t Judgment: g Phenomena: modals hwa hih ikwl lakl zkwkit hwa hih:PAAL.PAST.3SG.M ikl:PAAL.PRES akl:PAAL.INF zkwkit pron:NOM.1SG.M be:PAAL.PAST.3SG.M can:PAAL.PRES.SG.M eat:PAAL.INF glass:SG.F 'He could eat glass.' #1005 Source: author Vetted: t Judgment: u Phenomena: modals hwa ikl lakl zkwkit hwa ikl:PAAL.PRES akl:PAAL.INF zkwkit pron:NOM.1SG.M can:PAAL.PRES.3SG.M eat:PAAL.INF glass:SG.F 'He could eat glass.' The canonical position of the verb "can" is immediately before the infinitive verb, but other ordering is possible given the right context. #1006 Source: author Vetted: t Judgment: g Phenomena: modals ikwl ani lakl zkwkit ikl:PAAL.PRES ani akl:PAAL.INF zkwkit can:PAAL.PRES.SG.M pron:NOM.1SG eat:PAAL.INF glass:SG.F 'I can eat glass.' #1007 Source: author Vetted: t Judgment: g Phenomena: modals ikwl lakl zkwkit ani, wla hwa ikl:PAAL.PRES akl:PAAL.INF zkwkit ani w la hwa can:PAAL.PRES.SG.M eat:PAAL.INF glass:SG.F pron:NOM.1SG and neg1 pron:NOM.3SG.M 'It is I who can eat glass, and not he.' #1008 Source: author Vetted: t Judgment: g Phenomena: modals ani lakl zkwkit ikwl, abl la rwch ani akl:PAAL.INF zkwkit ikl:PAAL.PRES abl la rci:PAAL.PRES pron:NOM.1SG eat:PAAL.INF glass:SG.F can:PAAL.PRES.SG.M but neg1 want:PAAL.PRES.SG.M 'I can eat glass but I don't want to.' The Hebrew verb "ikl.PAAL" is ambiguous similarly to English in that it can either express ability or permission. #1009 Source: author Vetted: t Judgment: g Phenomena: modals ani ikwl lakl zkwkit, ki ie li einiim xzqwt ani ikl:PAAL.PRES akl:PAAL.INF zkwkit ki ie li en-im xzq-wt pron:NOM.1SG can:PAAL.PRES.SG.M eat:PAAL.INF glass:SG.F because there:is pron:DAT.1SG tooth:F-PL strong-PL.F 'I can eat glass, because I have strong teeth.' #1010 Source: author Vetted: t Judgment: g Phenomena: modals ani ikwl lakl zkwkit, ki ama eli mreh li. ani ikl:PAAL.PRES akl:PAAL.INF zkwkit ki ama eli rei:HIFIL.PRES-h li pron:NOM.1SG can:PAAL.PRES.SG.M eat:PAAL.INF glass:SG.F because mother:SG.F pron:POSS.1SG permit:HIFIL.PRES-SG.F pron:DAT.1SG 'I can eat glass, because my mother permits me (to do so).' I implemented the modal "can" beased on the "can as an auxiliary verb" instructions for lab9. raising-verb-with-infinitive-complement := verb-lex & trans-first-arg-raising-lex-item & [ SYNSEM.LOCAL.CAT.P-OR-Q -, SYNSEM.LOCAL [ CAT [ HEAD verb & [FORM fin ], VAL [ SUBJ < #subj & synsem & [ LOCAL.CAT [ HEAD noun & [ CASE nom ], VAL.SPR < > ]] >, COMPS < #comps & [ LOCAL [ CAT [ HEAD verb & [ FORM inf ], VAL.SUBJ < #subj >, VAL.SPR < >, VAL.SPEC < >, VAL.COMPS < > ], CONT.HOOK.LTOP #keyrel ]] >]]], SYNSEM.LKEYS.KEYREL.ARG1 #keyrel, ARG-ST < #subj, #comps > ]. Initially, when I added the keyrel constraint to my raising-verb-with-infinitive-complement, as suggested in the revised instruction for lab9, I did get the right semantics. But I also got three more parses that weren't there before the keyrel constraint was added. That was because I added the [CONT.HOOK.LTOP #keyrel] constraint in the wrong place and the LKB was looking for two comps for "can" rather than one, and since the second comp was uncosntrained as to case the subject "I" was quite happy to fill that slot and the sentence was being parsed with the pro-drop subject rule (two different ways, via HC and via CH), and the object "glass" was quite happy to fill that slot with the sentence being parsed with an indefinite null object for "eat". I fixed that and got one parse with the right semantics. Negation: --------- Hebrew has several negators. The one that is relevant to "It won't hurt me" is an adverb modifier that modifies the first tensed verb that appears in the sentence. It attaches to the left of the V. #1011 Source: author Vetted: t Judgment: g Phenomena: negation la hlk-ti la hlk:PAAL.PAST-ti neg1 go:PAAL.PAST-1SG 'I didn't go.' #1012 Source: author Vetted: t Judgment: u Phenomena: negation hlk-ti la hlk:PAAL.PAST-ti la go:PAAL.PAST-1SG neg1 'I didn't go.' #1013 Source: author Vetted: t Judgment: g Phenomena: modals ani la ikwl lakl zkwkit ani la ikl:PAAL.PRES akl:PAAL.INF zkwkit pron:NOM.1SG neg1 can:PAAL.PRES.SG.M eat:PAAL.INF glass:SG.F 'I can't eat glass.' #1014 - grammatical but not sentential negation Source: author Vetted: t Judgment: g Phenomena: modals ani ikwl la lakl zkwkit ani ikl:PAAL.PRES la akl:PAAL.INF zkwkit pron:NOM.1SG can:PAAL.PRES.SG.M neg1 eat:PAAL.INF glass:SG.F 'I can (choose to) not eat glass.' #1015 Source: author Vetted: t Judgment: g Phenomena: modals lakl zkwkit ani la ikwl akl:PAAL.INF zkwkit ani la ikl:PAAL.PRES eat:PAAL.INF glass:SG.F pron:NOM.1SG neg1 can:PAAL.PRES.SG.M 'I can't eat glass.' #1016 Source: author Vetted: t Judgment: g Phenomena: modals hwa la hih ikwl lakl zkwkit hwa la hih:PAAL.PAST.3SG.M ikl:PAAL.PRES akl:PAAL.INF zkwkit pron:NOM.1SG.M neg1 be:PAAL.PAST.3SG.M can:PAAL.PRES.SG.M eat:PAAL.INF glass:SG.F 'He couldn't eat glass.' #1017 Source: author Vetted: t Judgment: g Phenomena: modals hwa la ikwl hia lakl zkwkit hwa la ikl:PAAL.PRES hih:PAAL.PAST.3SG.M akl:PAAL.INF zkwkit pron:NOM.1SG.M neg1 can:PAAL.PRES.SG.M be:PAAL.PAST.3SG.M eat:PAAL.INF glass:SG.F 'He couldn't eat glass.' ;;The fact that this seems ungrammatical (rather than grammatical but with a non-sentential negation interpretation) is something to look at wrt word order - food for thought for later #1018 Source: author Vetted: t Judgment: u Phenomena: modals hwa hia la ikwl lakl zkwkit hwa hih:PAAL.PAST.3SG.M la ikl:PAAL.PRES akl:PAAL.INF zkwkit pron:NOM.1SG.M be:PAAL.PAST.3SG.M neg1 can:PAAL.PRES.SG.M eat:PAAL.INF glass:SG.F 'He couldn't eat glass.' To be able to handle these sentences I would have to add the auxiliary "be" which I will not do for this week's assignment. So the constraint that "negation has to follow the FIRST tensed verb" cannot be implemented for now. Negation was implemented by the LKB based on the customization script I filled out. The only changes I made to the original lexical type definition was adding the constraint that the verb modified must have [FORM fin]. neg-adv-lex := basic-scopal-adverb-lex & [ SYNSEM.LOCAL.CAT [ POSTHEAD -, VAL [ SPR < >, COMPS < >, SUBJ < > ], HEAD.MOD < [ LOCAL.CAT [ HEAD verb & [ FORM fin ] ], LIGHT + ] > ]]. "It doesn't hurt me." is ambiguous between the reading "The glass doesn't hurt me." and "Eating the glass doesn't hurt me." In Hebrew these two sentences are expressed differently. In the former case, the pronoun agrees with "glass" and as such it must be feminine and singular as the Hebrew word for glass, "zkwkit" is feminine. #1019 Source: author Vetted: t Judgment: g Phenomena: negation hia la mziqh li hia la nzq:HIFIL.PRES-h li pron:NOM.1SG.F neg1 hurt:HIFIL.PRES-SG.F pron:DAT.1SG 'It doesn't hurt me.' In the latter case, the pronoun to be used is not a personal pronoun but a demonstrative pronoun (which makes the Hebrew construction akin to "this doesn't hurt me"). #1020 Source: author Vetted: t Judgment: g Phenomena: negation zh la mziqh li zh la nzq:HIFIL.PRES li pron:SG.M neg1 hurt:HIFIL.PRES-SG pron:DAT.1SG 'It doesn't hurt me.' The demonstrative pronoun "zh" is always used in its masculine singular form when used to refer to event or state (a previous sentence) rather than an individual (NP in a previous sentence).I think that "zh" needs two lexical entries, one as the demonstrative pronoun "this one" that is listed along with its feminine and plural variants and one as the pronoun "it" when used to refer to a previous sentence. Only in the latter sense when in the accusative it must it be preceded by the accusative marker "at" without the "h" preceding it. #1021 Source: author Vetted: t Judgment: g Phenomena: pronouns ani awhb at zh ani ahb:PAAL.PRES at zh pron:NOM.1SG ahb:PAAL.PRES.SG.M ACC pron:SG.M 'I like (doing) it.' #1022 Source: author Vetted: t Judgment: u Phenomena: ronouns ani awhb zh ani ahb:PAAL.PRES zh pron:NOM.1SG ahb:PAAL.PRES.SG.M pron:SG.M 'I like (doing) it.' #1023 Source: author Vetted: t Judgment: g Phenomena: pronouns ani awhb at hzh ani ahb:PAAL.PRES at h-zh pron:NOM.1SG ahb:PAAL.PRES.SG.M ACC the-pron:SG.M 'I like this one.' But because "zh" requires the accusative marker without "h" it doesn't behave as either a personal pronoun (the accusative marker is incorporated in the accusative pronoun) nor as a common noun (which requires the "h" to attach to it if preceded by the accusative marker). So it is labeled a pronoun, which means that it is DEF+ and that it doesn't go through the indef rule, and it also means that it won't go through the rule that attached "h" as only common nouns go through that rule. It can be the complement of the accusative marker because it is nominative and DEF+, which are the only constraints on the complement of the accusative marker. And come to think of it what is preventing nominative personal pronouns from being the complement of the accusative marker? Nothing, actually. This problem has somehow slipped through the radar until now. I guess I didn't have any relevant examples in the test suite. So this needs to be fixed. I adjusted the noun types for that: noun :+ [ PROPER_PRONOUN_COMMON ppc ]. ppc := *top*. proper_or_pronoun := ppc. pronoun := proper_or_pronoun. proper := proper_or_pronoun. personal_pronoun := pronoun. common_or_it := ppc. common := common_or_it. it_pronoun := common_or_it & pronoun. zh_pronoun := pronoun-lex & [ STEM < "zh" >, SYNSEM.LOCAL.CAT.HEAD.PROPER_PRONOUN_COMMON it_pronoun, SYNSEM.LOCAL [CAT.HEAD.CASE nom, CONT.HOOK.INDEX.PNG [ GEND masc, PER third, NUM sg ] ] ]. The case marker type definition has been adjusted as follows: case-marker-p-lex := basic-one-arg & raise-sem-lex-item & [ SYNSEM.LOCAL.CAT [ HEAD adp & [ MOD < > ], VAL [ SPR < >, SUBJ < >, COMPS < #comps >, SPEC < > ]], ARG-ST < #comps & [ LOCAL.CAT [ HEAD noun & [ CASE nom ] & [ PROPER_PRONOUN_COMMON common_or_it ], VAL.SPR < > ] ] > ]. The indexed MRS for "It doesn't hurt me" is as follows: < h1,e2:SEMSORT:TENSE:ASPECT:MOOD, {h3:pronoun_n(x4:SEMSORT:THIRD:SG:MASC:+), h5:pronoun_q(x4,h6,h7), h8:_neg_r(e2,h9), h10:_hurt_v(e12:SEMSORT:TENSE:ASPECT:MOOD,x4,x11:SEMSORT:FIRST:SG:GENDER:+), h13:pronoun_n(x11), h14:pronoun_q(x11,h15,h16), h1:prop-or-ques_m(h17)}, {h6 qeq h3, h9 qeq h10, h15 qeq h13, h17 qeq h8} > This looks correct. However, I was getting two parses for this sentence, both of them leading to the same semantics. One resulted from the adj-head-scop-phrase rule applying before the head-comp rule which unifies "eat" with "glass" (i.e. modifying the verb) and the other with the adj-head-scop-phrase rule applying after the head-comp rule (i.e. modifying the VP). Since I want the negation adverb to modify the verb and not the VP, I added the following constraint: adj-head-scop-phrase :+ [ HEAD-DTR lex-rule ]. This correctly eliminated the extra parse. However, it created or exposed other problems. Some sentences involving a transitive verb with indefinite null instantiation stopped parsing. For example, the sentence "He/It (hwa) is not eating" stopped parsing. The constraint that the adj-head-scop-phrase rule follow a lex-rule is too stringent. Here is the revision: adj-head-follows := avm. basic-head-opt-comp-phrase :+ adj-head-follows. lexeme-to-lexeme-rule :+ adj-head-follows. adj-head-scop-phrase :+ [ HEAD-DTR adj-head-follows ]. TSDB: coverage 43.8, overgeneration: 3.6 I looked at the sentences that were still not parsing and decided to add a bunch of words to the lexicon and lexical rules to form the future tense. This increased the coverage to 45.8 but also increased the overgeneration to 4.5. ------------------------------- One unresolved issue: 1. Adverbs - I am getting three parses for: hia la ikl:PAAL.PRES-h qra:PAAL.INF mhr pron:NOM.3SG.F neg1 can:PAAL.PRES-SG.F read:PAAL.INF fast 'She can't read fast.' This is because in the current grammar the adverb "mhr" can modify "read" or "can't read" or "She can't read". Something needs to be done about this as in cases such as this, there should only be one parse. But how to block the other two parses without blocking other possible word orders? #1801 Source: author Vetted: t Judgment: g Phenomena: adverbs hia qrah mhr at cl hsprim w hxwbrwt halh hia qra:PAAL.PAST-h mhr at cl h-spr-im w h-xwbrt-wt h-alh pron:NOM.3SG.F read:PAAL.PAST-3SG.F fast ACC all the-book:M.NON-SG w the-pamphlet:F.NON-SG the-these 'She read all those books and pamphlets fast.' #1802 Source: author Vetted: t Judgment: g Phenomena: adverbs hia qrah at cl hsprim w hxwbrwt halh mhr hia qra:PAAL.PAST-h at cl h-spr-im w h-xwbrt-wt h-alh mhr pron:NOM.3SG.F read:PAAL.PAST-3SG.F ACC all the-book:M.NON-SG w the-pamphlet:F.NON-SG the-these fast 'She read all those books and pamphlets fast.' #1803 Source: author Vetted: t Judgment: g Phenomena: adverbs mhr hia qrah at cl hsprim w hxwbrwt halh mhr hia qra:PAAL.PAST-h at cl h-spr-im w h-xwbrt-wt h-alh fast pron:NOM.3SG.F read:PAAL.PAST-3SG.F ACC all the-book:M.NON-SG w the-pamphlet:F.NON-SG the-these 'She read all those books and pamphlets fast.' #1804 Source: author Vetted: t Judgment: g Phenomena: adverbs hia mhr qrah at cl hsprim w hxwbrwt halh hia mhr qra:PAAL.PAST-h at cl h-spr-im w h-xwbrt-wt h-alh pron:NOM.3SG.F fast read:PAAL.PAST-3SG.F ACC all the-book:M.NON-SG w the-pamphlet:F.NON-SG the-these 'She read all those books and pamphlets fast.'