31
DIFFERENTIAL OBJECT MARKING AND DIFFERENTIAL OBJECT INDEXATION: CROSS-LINGUISTIC AND LANGUAGE-SPECIFIC PERSPECTIVES Giorgio Iemmolo and Robert Schikowski University of Zürich Freitag, 7. Juni 2013

DIFFERENTIAL OBJECT MARKING AND DIFFERENTIAL … · DIFFERENTIAL OBJECT MARKING AND DIFFERENTIAL OBJECT INDEXATION: CROSS-LINGUISTIC AND LANGUAGE-SPECIFIC PERSPECTIVES Giorgio Iemmolo

  • Upload
    others

  • View
    57

  • Download
    0

Embed Size (px)

Citation preview

DIFFERENTIAL OBJECT MARKING AND DIFFERENTIAL OBJECT INDEXATION:

CROSS-LINGUISTIC ANDLANGUAGE-SPECIFIC PERSPECTIVES

Giorgio Iemmolo and Robert Schikowski University of Zürich

Freitag, 7. Juni 2013

OVERVIEW

Question: How different are differential object marking and differential object indexation?

Approach:

Large-scale typological study (Iemmolo 2011)

In-depth contrastive study of two languages (Schikowski 2013), Nepali and Chintang

Freitag, 7. Juni 2013

DOM AND DOI: DEFINITION

DOM and DOI: some direct objects are overtly marked or indexed on the verb depending on semantic and pragmatic properties of the NP referent (Aissen 2003, Bossong 1985, Comrie 1979).

Animacy: human > animate > inanimateDefiniteness (with specificity as subcategory): definite > specific indefinite > non-specific indefinitePerson: first, second > third Topicality: topical > non-topical

Freitag, 7. Juni 2013

AN EXAMPLE

Ləmma t’ərmus-u-n səbbər-ə(-w)

Lemma bottle-DEF-ACC break-PFV.3SG.M-OBJ.3SG.M

,Lemma broke the bottle.‘,Lemma broke the bottle.‘,Lemma broke the bottle.‘

Ləmma t’ərmus səbbər-ə

Lemma bottle break-PFV.3SG.M

,Lemma broke one bottle.‘,Lemma broke one bottle.‘,Lemma broke one bottle.‘

In Amharic (Semitic), only definite DOs receive case marking and are indexed on the verb (Amberber 2005: 299):

Freitag, 7. Juni 2013

APPROACHES TO DOM AND DOI

Distinguishing approach (Aissen 2003, Bossong 1985, Comrie 1989): DOM is used to distinguish between subject and DO when they resemble each other in terms of animacy, definiteness etc.

Indexing approach (1) (Hopper and Thompson 1980, Næss 2004, 2007): DOM and DOI index role or properties (individuation, affectedness) of the DO in a highly transitive event

Indexing approach (2) (Dalrymple and Nikolaeva 2011): DOM and DOI indicate the secondary topic status of the DO and bears the status of OBJ in LFG

Freitag, 7. Juni 2013

RESEARCH QUESTIONS

What is the role of information structure in the development and the distribution of DOM and DOI systems cross-linguistically?

Is there any difference between DOM and DOI as to their discourse function?

Freitag, 7. Juni 2013

METHODOLOGY

The present study is based on a convenience sample of 175 languages showing either DOM (133) or DOI (42) or both.

DOM and DOI systems have been grouped together based on the most relevant parameter(s) influencing the use of the constructions.

Freitag, 7. Juni 2013

RESULTS

In 115/175 languages (66%), DOM and DOI are mainly regulated by the topicality of the DO, i.e. topicality takes priority over the other parameters.

Although both systems are affected by the same parameter, their function is fundamentally different.

Freitag, 7. Juni 2013

FUNCTIONS

DOM marks DOs that have semantic and information-structural properties which are unexpected for DOs.

DOM is frequently associated with a topic-promotion or topic-shift function. 60/133 languages (45%) have a corre la t ion between DOM and dislocations or topicalisations

DOI, by contrast, is a strategy for reference-tracking and indicates high accessibility of the DO referent (Barlow 1988)

DOI is frequently associated with topic continuity

Freitag, 7. Juni 2013

DOM AND TOPIC-SHIFT

Álì-à kǝnâ-yè cítà

Ali-OBJ hunger-NOM grab.3SG.PST

,Ali was stricken by hunger.‘ (OSV, O topic),Ali was stricken by hunger.‘ (OSV, O topic),Ali was stricken by hunger.‘ (OSV, O topic)

Músà Álì(-à) bákcìn

Musa Ali-OBJ beat.3SG.IPFV

,Musa beats Ali.‘ (SOV),Musa beats Ali.‘ (SOV),Musa beats Ali.‘ (SOV)

In Kanuri (Nilo-Saharan), DOM becomes mandatory with dislocated DOs, even with those that would not be overtly coded in their canonical position (Cyffer 1983:195):

Freitag, 7. Juni 2013

DOI AND TOPIC-CONTINUITY

towøl.wojøl jål-pøsøwl-øs-tø

then PREF-blow-PRET.SUBJ.SG-OBJ.3SG

,Then he blew it (i.e. the fire) out.‘,Then he blew it (i.e. the fire) out.‘

näär sons-i sågrøp-mø sons-i

what look-PRS.SG axe-OBJ look-PRS.3SG

,Wherever he looks, he sees the axe.‘,Wherever he looks, he sees the axe.‘,Wherever he looks, he sees the axe.‘,Wherever he looks, he sees the axe.‘

In Eastern Mansi (Uralic), DOI is restricted to the most accessible and continuous DOs, while DOM alone signals a contrastive topic that serves to shift the current topic of discourse from one item to another (Virtanen accepted):

Freitag, 7. Juni 2013

SOME DIACHRONIC FACTS

The relevance of topicality is corroborated by the frequent polysemy of object markers with topic markers (Iemmolo under review).

For instance, in Kanuri, the marker -(k)à is also used to introduce conditional and temporal clauses, as well as frame-setting topics (spatio-temporal expressions).

The extension of topic markers to (some) direct objects is triggered by the topical nature of such direct objects which, like topics, tend to occur in sentence-initial position.

Freitag, 7. Juni 2013

SOME DIACHRONIC FACTS

Similarly, the link with the dislocated/topicalised DO position, gets lost over time, once those DOs are reanalysed as neutral from an information-structural point of view.

However, the role of information structure can be still seen in “optional” contexts, where DOM becomes obligatory only when the DO occupies a certain topical position.

Freitag, 7. Juni 2013

FROM TYPOLOGY TO SINGLE LANGUAGES

Do the typological claims made in part 1 stand closer scrutiny?

Look at two languages (one with DOM, one with DOI) in an in-depth study

Freitag, 7. Juni 2013

TWO NEIGHBOURING LANGUAGES

Nepali

Indo-Aryan

ca. 20,000,000 speakers in Nepal, India, Bhutan + diaspora (Gulf states, Korea)

national language of Nepal and lingua franca

Chintang

Tibeto-Burman > Kiranti

ca. 5,000 speakers in Chintang VDC, a group of villages in the hills of Eastern Nepal

seriously endangered

Freitag, 7. Juni 2013

NEPAL AND CHINTANG

Meilenkm

7001000

Freitag, 7. Juni 2013

DOM IN NEPALI

Nepali has DOM, like many Indo-Aryan languages

For example:

U yo manche*(-lai) mʌnpʌraũ-ch-ʌ.

3s PROX person-DAT like-NPST-3s

,He likes this guy.‘,He likes this guy.‘,He likes this guy.‘,He likes this guy.‘

U yo pustʌkalʌe(*-lai) mʌnpʌraũ-ch-ʌ.

3s PROX library-DAT like-NPST-3s

,He likes this library.‘,He likes this library.‘,He likes this library.‘,He likes this library.‘

Freitag, 7. Juni 2013

FACTORS IN DOM I: ANIMACYSathi?(-lai) bheʈ-nubhʌyo?

friend-DAT meet-PST.2HH

,Have you found (your) friend?‘,Have you found (your) friend?‘

Gai(?-lai) bheʈ-nubhʌyo?

cow-DAT meet-PST.2HH

,Have you found a/the cow?‘,Have you found a/the cow?‘

Ghʌr(*-lai) bheʈ-nubhʌyo?

house-DAT meet-PST.2HH

,Have you found a/the house?‘,Have you found a/the house?‘

Freitag, 7. Juni 2013

FACTORS IN DOM II: SPECIFICITY

Mʌnoj manche bheʈ-nʌ ja-nch-ʌ.

Manoj person meet-INF go-NPST-3s

,Manoj goes to meet someone/people.‘,Manoj goes to meet someone/people.‘,Manoj goes to meet someone/people.‘,Manoj goes to meet someone/people.‘

Mʌnoj manche-lai bheʈ-nʌ ja-nch-ʌ.

Manoj person-DAT meet-INF go-NPST-3s

,Manoj goes to meet someone/a certain person.‘,Manoj goes to meet someone/a certain person.‘,Manoj goes to meet someone/a certain person.‘,Manoj goes to meet someone/a certain person.‘

Freitag, 7. Juni 2013

FACTORS IN DOM III: TOPICALITY

Tyo pa-erʌ saed u khusi ch-ʌ.

MED get-CVB probably 3s happy be.NPST-3s

,Receiving it he is probably happy.‘,Receiving it he is probably happy.‘,Receiving it he is probably happy.‘,Receiving it he is probably happy.‘,Receiving it he is probably happy.‘,Receiving it he is probably happy.‘

Ʈebʌl-mathi-ko tes akaswãɖi-lai pheri pʌɖ-yo.

table-on-GEN MED telegram-DAT again read-PST.3s

,He read again that telegram on the table.‘,He read again that telegram on the table.‘,He read again that telegram on the table.‘,He read again that telegram on the table.‘,He read again that telegram on the table.‘

Freitag, 7. Juni 2013

MISSING THE FOREST FOR THE FACTORS?

Some formal factors (passive, co-argument case)

Many more functional factors (part of speech, quantifia-bility, modifiers, focus, disambiguation, affectedness)

Common denominator: DAT marks objects with unex-pected properties, unexpectedness being mostly centred around topicworthiness/topicality

Because of its high complexity, DOM is better predicted by a probabilistic model (logistic regression in Schikowski 2013) than by rules

Freitag, 7. Juni 2013

DOM AND UNEXPECTEDNESS

The unexpectedness of values associated with DOM can be directly observed in corpora.

Basis: 9558 annotated sentences from the Nepali National Corpus; 3183 objects eligible for DOM

animacy: the value most strongly associated with DOM is +human (68% of all human objects get DAT) - this value is only found on 9% of all objects

part of speech: pronouns are obligatorily marked by DAT - but only 2% of all objects are pronouns

givenness: 26% of all given objects get DAT (vs. only 8% of all new objects) - but given objects are a minority (74% are new)

Freitag, 7. Juni 2013

DOI IN CHINTANG I

Chintang has DOI: either A and P are indexed, or A is indexed by the same pattern that would normally be used by S.

Side effect: A is marked by NOM when it has S-AGR ➞ two distinct frames:

fully transitive: A-ERG P-NOM V-a(A).p(P)

S/A detransitivised: A-NOM P-NOM V-s(A)

S/A detransitivisation is not as well known as other alternations but by no means rare - cf. other Eastern Kiranti language, „paratransitive“ construction in Algonquian, „transitivity discord“, „semitransitivity“ in Oceanic.

Freitag, 7. Juni 2013

DOI IN CHINTANG II

Here is an example:

Debi-ŋa seu kond-o-ko.

Debi-ERG apple search-3[s]P-IND.NPST[.3sA]

,Debi is looking for a/the apple.‘,Debi is looking for a/the apple.‘,Debi is looking for a/the apple.‘

Debi seu kon-no.

Debi apple search-IND.NPST[.3sS]

,Debi is looking for apples/any apple.‘,Debi is looking for apples/any apple.‘,Debi is looking for apples/any apple.‘

Freitag, 7. Juni 2013

FACTORS IN DOI I: QUANTIFIABILITY

A-nicha-ce-ŋa bisauli sa

1sPOSS-younger.sibling-ns-ERG 1.25kg meat

u-c-o-hatt-e!

3pA-eat-3[s]P-COMPL.TR-IND.PST

,My younger siblings ate 1.25kg of meat.‘

Pi-ssa u-ca-no-kha.

cow-meat 3pS-eat-IND.NPST=NMLZ

,They eat beef.‘,They eat beef.‘

Freitag, 7. Juni 2013

FACTORS IN DOI II: SPECIFICITYAkka asinda sum-bhaŋ u-tiy-a=go

1s yesterday three-CLF 3pS-come-[SUBJ]PST=NMLZ

,They eat beef.‘,They eat beef.‘,They eat beef.‘,They eat beef.‘maʔmi-ce kond-u-ku-cu-ŋ-ta.

person-ns search-3P-IND.NPST-3nsP-1sA-IPFV

,I’m looking for the three people who came here yesterday.‘,I’m looking for the three people who came here yesterday.‘

Akka sum-bhaŋ ka-pha-pa maʔmi

1s three-CLF ACT.PTCP-help-REF person

,They eat beef.‘,They eat beef.‘,They eat beef.‘,They eat beef.‘koĩ-yã-ʔã-ta, jo=go nusayaŋ yaŋso.

search-1sS-IND.NPST-IPFV whoever=NMLZ CONCS is.okay

,I’m looking for three helpers, anyone is okay.‘,I’m looking for three helpers, anyone is okay.‘,I’m looking for three helpers, anyone is okay.‘,I’m looking for three helpers, anyone is okay.‘

Freitag, 7. Juni 2013

FUNCTION OF CHINTANG DOI

Quantifiability and specificity are both related to reference tracking:

only (theoretically) quantifiable referents can be tracked - instances of masses or groups of countable referents without a clear boundary cannot be identified wrt each other

various types of non-specific „referents“ do not get tracked (independently of quantifiability) because they are deemed unimportant

Freitag, 7. Juni 2013

DOI AND UNEXPECTEDNESS

In contrast to most factors associated with DOM, quanti-fiability and specificity are not unexpected on objects

Basis: 6606 annotated sentences from the Chintang Language Corpus; 1912 transitive, non-ambiguous sentences

97% of all quantifiable objects are indexed - but 76% of all objects are quantifiable, so the default gets marked!

parallel: 91% of all specific (incl. definite) objects are indexed - 75% of all objects are specific or definite

Freitag, 7. Juni 2013

NEPALI VS. CHINTANG

Nepali DOM Chintang DOI

marking object case object indexation (+ A case)

function“high”/unexpected objects

in terms of animacy, specificity, topicality...

reference tracking

determination flexible rigid

Freitag, 7. Juni 2013

CONCLUSIONS

Both typology and language comparison show that:

DOM and DOI have a common base (information structure), but

DOM serves to mark lexical objects with unexpected properties (wrt topicality and related properties): topic-shift

DOI serves to keep track of referents: topic-continuity

Freitag, 7. Juni 2013

Thanks for your attention!

Freitag, 7. Juni 2013