1 introduction - uzh · mass/countandlevelsofquantificationinchintang...

6
Mass/count and levels of quantification in Chintang Robert Schikowski (UZH, Department of General Linguistics) Mass vs. Count im Sprachvergleich 16 December 2013 1 Introduction Chintang (Tibeto-Burman > Kiranti) is spoken in Eastern Nepal (Dhanakuṭā district, Chintāṅa and Āhāle VDC). crucial syntactical phenomenon: S/A detransitivisation (Schikowski 2013). Defined as alternation between transitive default construction: A-ERG O-NOM 1 V-a(A).o(O) detransitivised construction: A-NOM O-NOM V-s(A) e transitive construction is used when the referent in O is at least spe- cific (1a). When it is non-specific (1b), the detransitivised construction is used. (1) a. Debi-ŋa Debi-ERG seu apple kond-o-ko. look.for-3[s]O-IND.NPST[.3sA] ‘Debi is looking for the/an apple.’ b. Debi Debi seu apple kon-no. look.for-IND.NPST[.3sS] ‘Debi is looking for apples.’ (elicitation PRAR 2010) Topic for this talk: interaction of S/A detransitivisation and other phe- nomena with mass/count in Chintang 2 S/A detransitivisation and quantifiability A crucial prerequisite for specificity in Chintang is quantifiability (Schikowski 2013) → link to mass/count questions. 1 O is a special label used here to cover all P-T-G that can have O agreement. Apart from this, the role system used in this paper is “Bickelian” in the sense of Haspelmath 2011 and is based on Dowty (1991), Primus (1999), and Bickel and Nichols (2009). Observation with “mass nouns”: transitive construction is used with quan- tified O referents, detransitivised construction with non-quantified refer- ents: (2) a. Kok rice a-ca-no? 2[s]S-eat-IND.NPST ‘Are you having rice?’ (CLC:CLLDCh1R02S01.0292) b. Ek one mana-iʔ-ko m.-LOC-NMLZ kok rice a-c-o-kko 2[s]A-eat-3[s]O-IND.NPST mo CIT para… COND ‘Suppose you eat one mana (≈ one pound) of rice…’ (CLC:CLLDCh1R02S01.0292) (3) a. Kappe, K. ba-iʔ PROX-LOC cuwa water thu-i. drink-[SUBJ.NPST.]1p[i]S ‘Kalpana, let’s have (some) water here.’ (CLC:CLLDCh3R10S05.412) b. Cuwa water ek one gilas glass thuŋ-c-o. drink-[1]d[iA]-[SUBJ.NPST.]3[s]O ‘Let’s have one glass of water.’ (CLC:CLLDCh1R05S01.800) But overt quantification is only an indicator of quantifiability. In (4a), S/A detransitivisation marks non-quantifiability in the absence of any quanti- fiers. In (4b), the transitive frame indicates quantifiability. (4) a. Abo now sa meat tac-ce. bring-[SUBJ.NPST.1]d[iS] ‘Now let’s bring (some) meat.’ b. Abo now sa meat tac-c-o. bring-[1]d[iA]-[SUBJ.]3[s]O ‘Now let’s bring a specific amount of/the meat.’ (elicitation PRAR 2010) 3 Definitions of mass/count Point of departure: there is no categorical lexical distinction between mass and count nouns (cf. e.g. Bunt 2006, Pelletier 2012): 1

Upload: others

Post on 10-Sep-2019

5 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: 1 Introduction - UZH · Mass/countandlevelsofquantificationinChintang RobertSchikowski(UZH,DepartmentofGeneralLinguistics) Massvs.CountimSprachvergleich …

Mass/count and levels of quantification in ChintangRobert Schikowski (UZH, Department of General Linguistics)Mass vs. Count im Sprachvergleich16 December 2013

1 Introduction• Chintang (Tibeto-Burman > Kiranti) is spoken in Eastern Nepal

(Dhanakuṭā district, Chintāṅa and Āhāle VDC).

• crucial syntactical phenomenon: S/A detransitivisation (Schikowski2013). Defined as alternation between

◦ transitive default construction: A-ERG O-NOM1 V-a(A).o(O)◦ detransitivised construction: A-NOM O-NOM V-s(A)

• e transitive construction is used when the referent in O is at least spe-cific (1a). When it is non-specific (1b), the detransitivised construction isused.

(1) a. Debi-ŋaDebi-ERG

seuapple

kond-o-ko.look.for-3[s]O-IND.NPST[.3sA]

‘Debi is looking for the/an apple.’b. Debi

Debiseuapple

kon-no.look.for-IND.NPST[.3sS]

‘Debi is looking for apples.’ (elicitation PRAR 2010)

• Topic for this talk: interaction of S/A detransitivisation and other phe-nomena with mass/count in Chintang

2 S/A detransitivisation and quantifiability• A crucial prerequisite for specificity in Chintang is quantifiability

(Schikowski 2013) → link to mass/count questions.1O is a special label used here to cover all P-T-G that can have O agreement. Apart from this,

the role system used in this paper is “Bickelian” in the sense of Haspelmath 2011 and is based onDowty (1991), Primus (1999), and Bickel and Nichols (2009).

• Observationwith “mass nouns”: transitive construction is usedwith quan-tified O referents, detransitivised construction with non-quantified refer-ents:

(2) a. Kokrice

a-ca-no?2[s]S-eat-IND.NPST

‘Are you having rice?’ (CLC:CLLDCh1R02S01.0292)

b. Ekone

mana-iʔ-kom.-LOC-NMLZ

kokrice

a-c-o-kko2[s]A-eat-3[s]O-IND.NPST

moCIT

para…COND

‘Suppose you eat one mana (≈ one pound) of rice…’(CLC:CLLDCh1R02S01.0292)

(3) a. Kappe,K.

ba-iʔPROX-LOC

cuwawater

thu-i.drink-[SUBJ.NPST.]1p[i]S

‘Kalpana, let’s have (some) water here.’(CLC:CLLDCh3R10S05.412)

b. Cuwawater

ekone

gilasglass

thuŋ-c-o.drink-[1]d[iA]-[SUBJ.NPST.]3[s]O

‘Let’s have one glass of water.’(CLC:CLLDCh1R05S01.800)

• But overt quantification is only an indicator of quantifiability. In (4a), S/Adetransitivisation marks non-quantifiability in the absence of any quanti-fiers. In (4b), the transitive frame indicates quantifiability.

(4) a. Abonow

sameat

tac-ce.bring-[SUBJ.NPST.1]d[iS]

‘Now let’s bring (some) meat.’b. Abo

nowsameat

tac-c-o.bring-[1]d[iA]-[SUBJ.]3[s]O

‘Now let’s bring a specific amount of/the meat.’(elicitation PRAR 2010)

3 Definitions of mass/count• Point of departure: there is no categorical lexical distinction betweenmass

and count nouns (cf. e.g. Bunt 2006, Pelletier 2012):

1

Page 2: 1 Introduction - UZH · Mass/countandlevelsofquantificationinChintang RobertSchikowski(UZH,DepartmentofGeneralLinguistics) Massvs.CountimSprachvergleich …

◦ “count” used as “mass”: Seeing too much table is a candy buffet no-no.◦ “mass” used as “count”: He had five beers last night.

• But: nouns may be used in different ways –they may be construed andpresented as quantifiable (≈ count) or non-quantifiable (≈ mass).

• emost common definitions ofmass/countmake reference to homogene-ity, either via divisibility or cumulativity (Kria 1989). But both have in-herent weaknesses (Rothstein 2010). Important point for Chintang: wordslike seu ‘apple’ may (in object position!) refer to arbitrary amounts ofthings without a plural marker → divisibility+cumulativity can also beapplied to “count” nouns

• Alternative suggestion (similar to Rothstein 2010, but less formal andmoreflexible): nouns (i.e. nominal lexemes) may have a clear base level or not.A noun may be said to have a base level if combining it with the numeral‘one’ evokes a clear mental image 2

• is is a continuous and language-specific distinction. at is,

◦ Nouns may have a more or less clear base level. Compare cat (clear)with cheese (less clear, possibly a wedge or loa) and soil (completelyunclear).

◦ Nouns with identical translations may have different base levels indifferent languages. Compare English ginger (no clear base level)with Chintang phidaŋ (base level = one rhizome).

◦ For some languages, base levels may not be relevant at all (cf. thetypology of nominal “Seinsarten” in Rijkhoff 2002:54ff.)

• Definitions related to homogeneity presuppose the notion of base levelsbecause dividing and cumulating have to start from some level.

2Presently this is an intuitive notion. However, since it can be grounded in language use, itwould be possible to operationalise it on the base of annotated corpora. e idea is that the baselevel of a noun becomes the clearer the more oen a speaker/hearer experiences quantifiable con-struals with identical selected levels (see below).

4 Base levels in Chintang• Chintang is highly flexible with respect to base levels. ere are appar-

ently no nouns which only allow for one construal. Here are two examplesfor nouns which would normally be construed as non-quantifiable (bhuja‘fried rice’, kham ‘soil’) combined with thia ‘one’:

(5) iaone

bhuja=yaŋfried.rice=ADD

a-ham-c-o-ko=kha=lo2A-divide-d-3[s]O-IND.NPST=NMLZ2=SURP

naŋ?but

‘So you even divide a single grain of fried rice?’(CLC:CLLDCh2R07S01.1098)

(6) iaone

khamsoil

u-pid-a-ŋs-a-ŋ=kha3[s]A-give-PST-PRF-[SUBJ.]PST-1sO=NMLZ2

budhi-ŋa.old.woman-ERG

‘e old woman had given me one piece of soil.’(CLC:nangsuba A.120)

• Other well-known tests for the mass/count distinction (see e.g. Ghomeshiand Massam 2012) so far also have not yielded two clear-cut classes:

◦ pluralisation: all equivalents to English mass nouns can be pluralisedvia “Universal Packing” (Bunt 2006): kham-ce ‘pieces of soil’

◦ quantifiers: small and big quantities are denoted by the same quan-tifiers for all nouns: miʔmuŋ ‘lile, few’, badde ‘much, many’

◦ classifiers: Chintang has two numeral classifiers (+ vs );both co-occur with all nouns

◦ determiners: Chintang does not have articles; specificity of O can beindicated via the transitive frame for all nouns:

(7) Kham=losoil=SURP

neg-o-ko!bite-3[s]O-IND.NPST[.3sA]

‘He bites off a piece of soil!’ (CLC:CLLDCh2R03S06b.0577)

• As a result, Chintang does not fall into any of the three mass/count lan-guage types proposed by Chierchia (1998, 2010):

◦ number is in most contexts obligatorily marked (differently from“classifier languages” and “number neutral languages”)

2

Page 3: 1 Introduction - UZH · Mass/countandlevelsofquantificationinChintang RobertSchikowski(UZH,DepartmentofGeneralLinguistics) Massvs.CountimSprachvergleich …

◦ there are obligatory numeral classifiers (differently from “numbermarking languages” and “number neutral languages”)

◦ classifiers regularly co-occur with the nominal number marker

• So ismass/count relevant in Chintang at all? e distinction does not seemto be grammaticalised. But:

◦ Differences of a continuous kind can be elicited –some nouns haveclear base levels, some don’t.

◦ is may serve as the base for predicting grammatical phenomena.Annotation data are available for S/A detransitivisation from thestudy in Schikowski (2013). As shown in figure 2, this phenomenoncorrelates more or less directly with the tendency of objects to beconstrued as quantifiable.

◦ Pluralisation behaves quite differently from this, as shown in figure1. is is expected, though, since having a clear base level is not asufficient criterion for frequent pluralisation.

◦ Base levels are also necessary to interpret the semantics of the non-singular marker -ce: if there is a base level, -ce operates on that level.If there is no base level, a referent first needs to be “packed” beforethe corresponding noun can be used with-ce.

5 Selected levels in Chintang• Apart from the base level, many more levels are possible.

• Such levels can be flexibly selected by a speaker in a concrete NP and serveas the base for numerals such as thia as well as for the non-singular -ce.

• Selection may take place ad hoc (e.g. more or less water) or may be guidedby natural distinctions. For instance, makkai ‘maize’ may refer to

◦ a single grain of maize (8a),◦ a specific amount of maize grains (8b),◦ a maize cob (8c) (= base level), or◦ a maize plant (8d).

(8) a. Uncu-makai-ce=ta2dPOR-maize-ns=FOC

u-lo-u-ku-ce.3[pA]-bring.out-3O-IND.NPST-3nsO

‘ey bring out their maize (grains).’(CLLDCh3R01S01a.682)

b. Etithis.much

chup-lokgrab-SIM

makkaimaize

cemd-u-ŋ=go.grind-3[s]O-1sA=NMLZ

‘It’s only one handful of maize that I have ground.’(CLDLCh3R03S02.1226)

c. Makkai-cemaize-ns

naCTOP

tak-madegrain-INF

a-hid-u-m-cu-mh-e?2A-finish-3O-2nsA-3nsO-2nsA-IND.PST

‘Have you finished degraining the maize cobs?’(CLLDCh2R05S01.080)

d. Hiccetwo

makkaimaize

lis-a-s-e.become-PST-PRF-IND.PST[.3sS]

‘Two maize plants have grown.’ (CLLDCh4R05S05.1192)

• Selected levels are relevant for S/A detransitivisation because there areoen quantifiability mismates between levels:

◦ one level quantifiable, subordinate level not.◦ one level quantifiable, superordinate level not

• Non-quantifiable subamount: onewhole tangerine vs a part without clearborders that has been eaten by a mouse

(9) BaPROX

sencakmouse

ci-a-ŋs-e.eat-PST-PRF-IND.PST[.3sS]

‘A mouse has eaten from this (tangerine).’ (field notes 2011)

• Non-quantifiable superamount: one jug of water vs virtually infinitesource from which it is taken

(10) Abonow

ba-iʔPROX-LOC1

cuwawater

ɨg-ɨ-kɨ-ŋ.scoop-3[s]O-IND.NPST-1sA

‘Now I take out some water.’ (CLLDCh3R07S03.602)

3

Page 4: 1 Introduction - UZH · Mass/countandlevelsofquantificationinChintang RobertSchikowski(UZH,DepartmentofGeneralLinguistics) Massvs.CountimSprachvergleich …

• antifiability mismatches can lead to structural mismatches across lan-guages. For instance, normally the O referent in the Chintang transitiveconstruction will have an article in the English translation, whereas the Oreferent of the detransitivised construction will have no article. But…

◦ Chintang conventionally selects the level of a complex referent thatis actually affected by an action

◦ English sometimes seems to prefer to select any quantifiable level(conditions not clear)

◦ In (11), Chintang selects the affected level (non-quantifiable sub-amount→S/A detransitivisation), whereas English selects the quan-tifiable superamount (→ definite article):

(11) ca-saŋa=taeat-CVB.FGR=FOC

numd-a-kt-a-lokdo-PST-IPFV-[SUBJ.]PST[.3sS]-CVB.BGR

‘while he was still taking (the medicine)’(CLC:appa katha talk.021)

• In general, English seems to be more flexible than Chintang with respectto level selection. For instance, (9) could also be translated as

◦ A mouse has eaten the tangerine. (select superamount even thoughnot fully affected)

◦ A mouse has eaten from the tangerine. (select superamount and in-dicate partial affection by preposition)

◦ A mouse has nibbled at the tangerine. (select superamount and indi-cate partial affection by preposition and verb)

◦ A mouse has eaten a piece of the tangerine. (select subamount)◦ But interestingly ⁇A mouse has eaten some of the tangerine. (select

subamount)

• Selected levels can also account for apparent number/construction mis-matches within Chintang. Normally, O NPs in the S/A detransitivised donot carry number markers and the number marker is only used with def-inite quantities. However, examples such as (12) are possible:

(12) Kholakhi-cewild.yam-ns

tus-i-ki-ŋa.dig.out-1pS-IND.NPST-e

‘We dig out wild yam roots.’ (CLC:phidang talk.045 + elicitation RBK2012)

• Khi denotes a single yam root, khi-ce a small, quantifiable group of yamroots. e construction in (12) selects a group of such groups (i.e. a non-quantifiable superamount).

6 Conclusions

• Mass/count distinction not grammaticalised in Chintang: all nouns can beconstrued as quantifiable/non-quantifiable

• But: very different frequencies of construals –nouns do have preferences

• Preferences are reflected in base levels, which can be elicited and correlatewith morphosyntactic phenomena (in particular, S/A detransitivisation)

• Other levels may be flexibly selected as required

• Level selection is particularly relevant for explaining grammar in the pres-ence of quantifiability mismatches

References

Bickel, Balthasar, and Johanna Nichols. 2009. Case marking and alignment. Ine Handbook of Case, ed. Andrej Malchukov and Andrew Spencer, 304--321.Oxford University Press.

Bunt, Harry C. 2006. Mass Expressions. In Encyclopedia of Language and Lin-guistics, ed. Keith Brown, volume 7, 530--534. Oxford: Elsevier.

Chierchia, Gennaro. 1998. Reference to kinds across languages. Natural Lan-guage Semantics 6:339--405.

Chierchia, Gennaro. 2010. Mass nouns, vagueness and semantic variation. Syn-thèse 174:99--149.

Dowty, David. 1991. ematic proto-roles and argument selection. Language67:547--619.

4

Page 5: 1 Introduction - UZH · Mass/countandlevelsofquantificationinChintang RobertSchikowski(UZH,DepartmentofGeneralLinguistics) Massvs.CountimSprachvergleich …

Ghomeshi, Jila, and Diane Massam. 2012. e count mass distinction: Issuesand perspectives. In Count and Mass Across Languages, ed. Diane Massam,1--8. Oxford: Oxford University Press.

Haspelmath, Martin. 2011. On S, A, P, T, and R as comparative concepts foralignment typology. Linguistic Typology 15:535--689.

Kria, Manfred. 1989. Nominalreferenz und Zeitkonstitution. München: Wil-helm Fink.

Pelletier, Francis Jeffry. 2012. Lexical nouns are both and neither + and+. In Count and Mass Across Languages, ed. Diane Massam, 9--26.Oxford: Oxford University Press.

Primus, Beatrice. 1999. Cases and ematic Roles. Tübingen: Niemeyer.

Rijkhoff, Jan. 2002. e Noun Phrase. Oxford: Oxford University Press.

Rothstein, Susan. 2010. Counting and the mass/count distinction. Journal ofSemantics 27:343--397.

Schikowski, Robert. 2013. Object-conditioned differential marking in Chintangand Nepali. Doctoral Dissertation, University of Zurich.

Abbreviations1 1st person, 2 2nd person, 3 3rd person, ADD additive, A agent, CIT citation,COMPL completive, COND conditional, CTOP contrastive topic, CVB.FGRforegrounding converb, d dual, ERG ergative, e exclusive, FOC focus, G goal,i inclusive, IND indicative, INF infinitive, IPFV imperfective, ITR intransitive,LOC locative, NMLZ nominaliser, NPST nonpast, ns non-singular (= dual orplural), O object (= argument triggering O-AGR), POR possessor, PRF perfect,PROX proximative, PST past, p plural, P patient, SIM simultaneous, SUBJ sub-junctive, SURP surprise, s singular, S intransitive subject, T theme

Prop

ortio

n of

non

-sin

gula

rs

0.00

0.02

0.04

0.06

0.08

cat

doghousemoney

wood

water gr

ass

alcohol

Figure 1: Pluralisation of some nouns in the Chintang Language Corpus

5

Page 6: 1 Introduction - UZH · Mass/countandlevelsofquantificationinChintang RobertSchikowski(UZH,DepartmentofGeneralLinguistics) Massvs.CountimSprachvergleich …

TR 100% DETR 0%

menuwa ‘cat’

TR 86%

DETR 14%

kocuwa ‘dog’

TR 76%

DETR 24%

khim ‘house’

TR 68%

DETR 32%

paisa ‘money’

TR 47%

DETR 53%

sɨŋ ‘wood’

TR 40%

DETR 60%

cuwa ‘water’

TR 24%

DETR 76%

ghãsa ‘grass’

TR 11%

DETR 89%

arkha ‘alcohol’

Figure 2: Framing for some nouns in the Chintang Language Corpus

6