Tibetan Corpus Linguistics: present obstacles and future prospects

Tibetan Corpus Linguistics: present obstacles and future prospects Nathan W. Hill and Edward Garrett Tibetan in Digital Communication Overview 1. Current analytic obstacles Tibetan in Digital Communication Overview 1. Current analytic obstacles 2. Software development desiderata Tibetan in Digital Communication Overview 1. Current analytic obstacles 2. Software development desiderata 3. Future research questions Tibetan in Digital Communication Overview 1. Current analytic obstacles 2. Software development desiderata 3. Future research questions 4. Future collaborations Tibetan in Digital Communication Overview 1. Current analytic obstacles 2. Software development desiderata 3. Future research questions 4. Future collaborations 5. Possible projects Current analytical obstacles • Demonstrative uses of relator nouns Current analytical obstacles • Demonstrative uses of relator nouns • Zero nominalization of verbs Current analytical obstacles • Demonstrative uses of relator nouns • Zero nominalization of verbs • A few tricky words raṅ, re, etc. Demonstrative uses of relator nouns Demonstratives are sometimes used without an antecedent. • der tshogs-paḥi lha-mi The gods and men who had gathered there Demonstrative uses of relator nouns Demonstratives are sometimes used without an antecedent. • der tshogs-paḥi lha-mi The gods and men who had gathered there We consequently treat some spatial words as demonstratives. Demonstrative uses of relator nouns Demonstratives are sometimes used without an antecedent. • der tshogs-paḥi lha-mi The gods and men who had gathered there We consequently treat some spatial words as demonstratives. • phug-paḥi mdun-na mar log ḥgro-źiṅ ḥdug-pa-la While they were returning back down, passing in front of the cave. Demonstrative uses of relator nouns Demonstratives are sometimes used without an antecedent. • der tshogs-paḥi lha-mi The gods and men who had gathered there We consequently treat some spatial words as demonstratives. • phug-paḥi mdun-na mar log ḥgro-źiṅ ḥdug-pa-la While they were returning back down, passing in front of the cave. We consequently treat relator nouns, when they are not relating, as demonstratives. Demonstrative uses of relator nouns Demonstratives are sometimes used without an antecedent. • der tshogs-paḥi lha-mi The gods and men who had gathered there We consequently treat some spatial words as demonstratives. • phug-paḥi mdun-na mar log ḥgro-źiṅ ḥdug-pa-la While they were returning back down, passing in front of the cave. We consequently treat relator nouns, when they are not relating, as demonstratives. • phi-na gos-daṅ / naṅ-na ḥtsho-ba phye-daṅ bcud ye med-paḥi stobs-kyis / lus keṅ-rus ltar soṅ-ba / Because neither clothes outside nor nourishment inside, whether food or drink, had he any, his body became like a skeleton. Demonstrative uses of relator nouns Demonstratives are sometimes used without an antecedent. • der tshogs-paḥi lha-mi The gods and men who had gathered there We consequently treat some spatial words as demonstratives. • phug-paḥi mdun-na mar log ḥgro-źiṅ ḥdug-pa-la While they were returning back down, passing in front of the cave. We consequently treat relator nouns, when they are not relating, as demonstratives. • phi-na gos-daṅ / naṅ-na ḥtsho-ba phye-daṅ bcud ye med-paḥi stobs-kyis / luskeṅ-rus ltar soṅ-ba / Because neither clothes outside nor nourishment inside, whether food or drink, had he any, his body became like a skeleton. Maybe the spatial (temporal) uses of relator nouns and demonstratives should get a special tag? Zero nominalization of verbs In general Tibetan requires a suffix to change a verbal stem to a verbal noun, but not always, especially not in poetry. • bla-maḥi gsuṅ // ma-rig min-pa dbyiṅs-su dag / The words of the guru, which are not ignorant, are as pure as space. Zero nominalization of verbs In general Tibetan requires a suffix to change a verbal stem to a verbal noun, but not always, especially not in poetry. • bla-maḥi gsuṅ // ma-rig min-pa dbyiṅs-su dag / The words of the guru, which are not ignorant, are as pure as space. • ḥdir bźugs gsan cig 'listen, O you who sit here!' Zero nominalization of verbs In general Tibetan requires a suffix to change a verbal stem to a verbal noun, but not always, especially not in poetry. • bla-maḥi gsuṅ // ma-rig min-pa dbyiṅs-su dag / The words of the guru, which are not ignorant, are as pure as space. • ḥdir bźugs gsan cig 'listen, O you who sit here!' • gzuṅ-daṅ ḥdzin-paḥi sgrib gñis bral free from the two obscurations of 'taken' and 'taker' Zero nominalization of verbs In general Tibetan requires a suffix to change a verbal stem to a verbal noun, but not always, especially not in poetry. • bla-maḥi gsuṅ // ma-rig min-pa dbyiṅs-su dag / The words of the guru, which are not ignorant, are as pure as space. • ḥdir bźugs gsan cig 'listen, O you who sit here!' • gzuṅ-daṅ ḥdzin-paḥi sgrib gñis bral free from the two obscurations of 'taken' and 'taker' • ṅa-rgyal-daṅ ni ma-dad daṅ // don-du gñer-ba med ñid daṅ // phyi-rol-rnamg.yeṅ-naṅ-bsdus daṅ // skyo-ba-ñan-paḥi dri-ma yin // Pride and lack of faith, lack of interest and being distracted outward, being withdrawn inward and dejection, (these) are flaws of listening. A few tricky words • raṅ • [p.refl] raṅ gi bu-mo rnams sad-du btaṅ-ste '(he was) sent to wake his own daughters’ khyed raṅ ‘you’, ṅa raṅ ‘I’, etc. A few tricky words • raṅ • [p.refl] raṅ gi bu-mo rnams sad-du btaṅ-ste '(he was) sent to wake his own daughters’ khyed raṅ ‘you’, ṅa raṅ ‘I’, etc. • [p.pers] raṅ gis raṅ la lcag btab-nas 'they each hit one another' A few tricky words • raṅ • [p.refl] raṅ gi bu-mo rnams sad-du btaṅ-ste '(he was) sent to wake his own daughters’ khyed raṅ ‘you’, ṅa raṅ ‘I’, etc. • [p.pers] raṅ gis raṅ la lcag btab-nas 'they each hit one another' • [d.det] khyod las rem-po raṅ źig 'one more assiduous even than thou' A few tricky words • raṅ • [p.refl] raṅ gi bu-mo rnams sad-du btaṅ-ste '(he was) sent to wake his own daughters’ khyed raṅ ‘you’, ṅa raṅ ‘I’, etc. • [p.pers] raṅ gis raṅ la lcag btab-nas 'they each hit one another' • [d.det] khyod las rem-po raṅ źig 'one more assiduous even than thou' • [???] la-la na-re Mar-pa raṅ smyos-nas 'some said Marpa (himself) had gone (quite) crazy'. A few tricky words • re • [p.indef] re raṅ 'each' (we use this tag inconsistently) A few tricky words • re • [p.indef] re raṅ 'each' (we use this tag inconsistently) • [d.det] yum-gyis nas khal re yod-paḥi chaṅ tshod chen-po gsum btsos-pa-la 'mother cooked up each load of barley that she had into three great measures of beer' A few tricky words • re • [p.indef] re raṅ 'each' (we use this tag inconsistently) • [d.det] yum-gyis nas khal re yod-paḥi chaṅ tshod chen-po gsum btsos-pa-la 'mother cooked up each load of barley that she had into three great measures of beer' • [n.count] re źig 'one time', re 'hope' A few tricky words • re • [p.indef] re raṅ 'each' (we use this tag inconsistently) • [d.det] yum-gyis nas khal re yod-paḥi chaṅ tshod chen-po gsum btsos-pa-la 'mother cooked up each load of barley that she had into three great measures of beer' • [n.count] re źig 'one time', re 'hope' • [num.card] źo re źo do 'one or two ounces' Software development desiderata • Automation of recompiling and retagging Software development desiderata • Automation of recompiling and retagging • Incorporation of Pablo's consistency tester Software development desiderata • Automation of recompiling and retagging • Incorporation of Pablo's consistency tester • Post hoc rule based consistency checker Software development desiderata • Automation of recompiling and retagging • Incorporation of Pablo's consistency tester • Post hoc rule based consistency checker • Seamless web interface Post hoc rule based consistency checker • Negation and verb stems mi [neg] dgaḥ-ba [n.v.fut.n.v.pres] 'not happy' mi [n.count] dgaḥ-ba [n.v.invar] 'happy person' Post hoc rule based consistency checker • Negation and verb stems mi [neg] dgaḥ-ba [n.v.fut.n.v.pres] 'not happy' mi [n.count] dgaḥ-ba [n.v.invar] 'happy person' *mi [n.count] dgaḥ-ba [n.v.fut.n.v.pres] *mi [neg] dgaḥ-ba [n.v.invar] Post hoc rule based consistency checker • Negation and verb stems mi [neg] dgaḥ-ba [n.v.fut.n.v.pres] 'not happy' mi [n.count] dgaḥ-ba [n.v.invar] 'happy person’ *mi [n.count] dgaḥ-ba [n.v.fut.n.v.pres] *mi [neg] dgaḥ-ba [n.v.invar] • If mi is [n.count] then change dgaḥ-ba [n.v.fut.n.v.pres] to dgaḥ-ba [n.v.invar] Post hoc rule based consistency checker • Negation and verb stems mi [neg] dgaḥ-ba [n.v.fut.n.v.pres] 'not happy' mi [n.count] dgaḥ-ba [n.v.invar] 'happy person’ *mi [n.count] dgaḥ-ba [n.v.fut.n.v.pres] *mi [neg] dgaḥ-ba [n.v.invar] • If mi is [n.count] then change dgaḥ-ba [n.v.fut.n.v.pres] to dgaḥ-ba [n.v.invar] • If mi is [neg] then change dgaḥ-ba [n.v.invar] to dgaḥ-ba [n.v.fut.n.v.pres] Post hoc rule based consistency checker de [adv.proclausal] ḥi [case.gen] tshe [n.rel] 'at that time' de [d.dem] ḥi [case.gen] tshe [n.count] 'his life' Post hoc rule based consistency checker de [adv.proclausal] ḥi [case.gen] tshe [n.rel] 'at that time' de [d.dem] ḥi [case.gen] tshe [n.count] 'his life’ *de [d.dem] ḥi [case.gen] tshe [n.rel] *de [adv.proclausal] ḥi [case.gen] tshe [n.count] Post hoc rule based consistency checker de [adv.proclausal] ḥi [case.gen] tshe [n.rel] 'at that time' de [d.dem] ḥi [case.gen] tshe [n.count] 'his life’ *de [d.dem] ḥi [case.gen] tshe [n.rel] *de [adv.proclausal] ḥi [case.gen] tshe [n.count] • If tshe is [n.count] then change de [adv.proclausal] to de [d.dem] Research questions we can address with our Corpus 1. Conflicts between da-drag rules and syntactic cues to verb stem disambiguation. Research questions we can address with our Corpus 1. Conflicts between da-drag rules and syntactic cues to verb stem disambiguation. ku-śu ḥdi ni ḥbras-bu las skyes-pa ma lags te ། chab-mig cig gi naṅ nas rñed-pa s slan-cad ni bdag gis mi rñed de ། mi ḥbyor [v.past] ~ [v.pres] to "This apple was not born from fruit, but I found it from inside a spring, so I cannot find it hereafter. It will not be encountered." Research questions we can address with our Corpus 1. Conflicts between da-drag rules and syntactic cues to verb stem disambiguation. ku-śu ḥdi ni ḥbras-bu las skyes-pa ma lags te ། chab-mig cig gi naṅ nas rñed-pa s slan-cad ni bdag gis mi rñed de ། mi ḥbyor [v.past] ~ [v.pres] to "This apple was not born from fruit, but I found it from inside a spring, so I cannot find it hereafter. It will not be encountered.” If da-drag rules are ordered before the rule that prevents the past stem after mi, then the suffix -to would have triggered [v.past]. Research questions we can address with our Corpus 1. Conflicts between da-drag rules and syntactic cues to verb stem disambiguation. ku-śu ḥdi ni ḥbras-bu las skyes-pa ma lags te ། chab-mig cig gi naṅ nas rñed-pa s slan-cad ni bdag gis mi rñed de ། mi ḥbyor [v.past] ~ [v.pres] to "This apple was not born from fruit, but I found it from inside a spring, so I cannot find it hereafter. It will not be encountered.” If da-drag rules are ordered before the rule that prevents the past stem after mi, then the suffix -to would have triggered [v.past]. (We order negation rules before da-drag rules, so ḥbyor is tagged [v.pres].) Research questions we can address with our Corpus 1. Conflicts between da-drag rules and syntactic cues to verb stem disambiguation. 2. See whether any converbs imply things about what verb stems can be coordinated (e.g. V-pa-daṅ … V-pa-daṅ, do they have to be the same stem?) Research questions we can address with our Corpus 1. Conflicts between da-drag rules and syntactic cues to verb stem disambiguation. 2. See whether any converbs imply things about what verb stems can be coordinated (e.g. V-pa-daṅ … V-pa-daṅ, do they have to be the same stem?) 3. Can we identify different uses of śad by counting how many words occur between them? Research questions we can address with our Corpus 1. Conflicts between da-drag rules and syntactic cues to verb stem disambiguation. 2. See whether any converbs imply things about what verb stems can be coordinated (e.g. V-pa-daṅ … V-pa-daṅ, do they have to be the same stem?) 3. Can we identify different uses of śad by counting how many words occur between them? 4. How do conditional clauses work (cf. Greek). Research questions we can address with our Corpus 1. Conflicts between da-drag rules and syntactic cues to verb stem disambiguation. 2. See whether any converbs imply things about what verb stems can be coordinated (e.g. V-pa-daṅ … V-pa-daṅ, do they have to be the same stem?) 3. Can we identify different uses of śad by counting how many words occur between them? 4. How do conditional clauses work (cf. Greek). 5. Functions of la [cv.all] other than to coordinate imperatives. Future Collaborations • TBRC • Berkeley • CASS • You! Possible Projects • Tibet Daily Corpus (e.g. for keywords in current affairs) Possible Projects • Tibet Daily Corpus (e.g. for keywords in current affairs) • Continuous crawling of Tibetan blogs (e.g. for new terminology) Possible Projects • Tibet Daily Corpus (e.g. for keywords in current affairs) • Continuous crawling of Tibetan blogs (e.g. for new terminology) • Yours? Thank you