Direct Object marking in Finno-Ugric and Turkic languages: verb semantics and definiteness of direct object NP Natalya Serdobolskaya, Svetlana Toldova [email protected], [email protected] 0. The paper is focused on the phenomenon of differential Direct Object marking (a term of Aissen 1998) in Finno-Ugric and Turkic languages. The presence/absence of the accusative marker on DO in these languages is usually attributed to definiteness / indefiniteness (specificity/non-specificity) of the corresponding NP (Nilsson 1978, Enç 1991 e.a.). Data: Det Tuvinian (Turkic), Muravyova 1992: 257 ACC 1)a.Men bo solun-nu nomçup kaapkan men. ‘I have already read this newspaper(-ACC).’ b. düün ( the DO NP includes the demonstrative this, it is definite and, thus, has the accusative marker 1. men solun nomçaan men ‘Yesterday I read a newspaper’. In (1b) DO NP is indefinite and, thus, has no overt marking Problem There are cases when there is no overt clue in the sentence or previous context of whether the DO has definite or indefinite reading (e.g. demonstrative, afore mentioning etc.): Translation from Russian into the target language (Comment: there is no overt marker of definiteness in Russian): Pre-experiment (1): KOMI-ZYRIAN ACC From Russian lit.: mother presoaked linen (2)a. mama mother kEtEd-Y-s kEluj. presoak-PST-3 linen Mother presoaked the linen. the accusative marker is absent From Russian lit.: mother dried linen b. mama koSt-Y-s kEluj-sE. mother dry-PST-3 linen-ACC Mother dried the linen (after washing). DO has the overt accusative marker The choice of DO marking in these sentences is the same for the majority of native speakers in spite of the fact that there is no direct instruction in the source sentence concerning the definiteness of the DO NP. Hence, there are some contextual features in (2a) and (b) that contribute to the definite/indefinite reading of the DO. The question is, what are these contextual features. What context properties impose restrictions on the definite / indefinite reading of DO NP (and, hence, its ability to trigger presupposition)? 2. Background: DO definiteness and accusative marking in Finno-Ugric and Turkic languages The accusative marker in Mari, Komi-Zyrian, Udmurt, Mordvin(Finno-Ugric); Chuvash, Uzbekh, Kazakh (Turkic) works as a presupposition trigger (see Appendix 2 for details): Example from Komi-Zyrian: afore-mentionedness and DO marking: (3) EtCid aG-i-s kaj once find-PST-3 …i dumajt-E: and think-PRS.3 zEr tuS’… sparrow mYj-la oat zEr corn tuS-sE what-COMPAR oat ACC GYnjav-nY? a new referent introduced into the discourse for the 1st time hence, no marking afore-mentioned referent corn-ACC part-INF ACC marking …Once the sparrow found an oat corn, and started thinking: “How can I part the oat corn?” Besides the mere familiarity of a referent there is a number of other factors that regulate the choice of DO marking. The factors vary through languages. Hence, the assumption about “definiteness properties” of the marker needs additional checking in each language. The present work is based on the assumption that the definiteness of an NP should be interpreted in terms of presupposition (Geurts 1998, Kamp 1993). Definite NPs are interpreted as requiring a presupposition: the presupposition inducing expression (c.f. specific NPs) should be bound or, if it is not bound, it can be accommodated (Geurts 1998). NB1: The distribution of the ACC / non-ACC DOs is different for animates and inanimates. For the sake of simplicity, here and below only inanimate DO are considered. NB2: In Komi and Udmurt, there is a cumulative marker for ACC and possessive 3d person singular. However, it is often used purely as a definiteness marker (see (1)); hence, it is glossed as ACC. NB3: For Mari, Komi, Udmurt, Mordvin(Finno-Ugric);Chuvash, Uzbekh, Kazakh (Turkic) definiteness is crucial for the DO marking. Cf. Turkish data: as argued by Enç 1991, it is the DO specificity that determines the DO marking in Turkish. 3. Context properties contributing to the definite/indefinite reading of the DO NP: previous approaches Examples (3)-(12) suggest that ACC marker in the languages under consideration is a presupposition trigger imposing certain restrictions on the context. However, example (2) manifests that the preferences of native speakers in DO encoding can be determined when there is no overt clue for the definiteness of an NP. (i) What context properties can impose restrictions on the definite/indefinite reading of the DO NP? The existing answers: Aspectual composition: The aspectual characteristics of the situation can interact with the (in)definiteness of its participants (cf. Krifka1989) Uzbek (11)Pete butqa-ni Pete porridge-ACC Pete ate up (all) the porridge (ACC marker) all the porridge that was in the plate / / / vs. Cumulative object + ACC-marker +perfective reading of the verb (12) Pete Pete har kuni every day stakan sindir-di. glass break butqa porridge ate some porridge some porridge / jet-ish-ga eat-NZR-DAT ket-di. go-PST and went off. cumulative object + no marker + imperfective reading of the verb Pete broke a glass (no marker) every day. (The habitual properties of the verb induces the indefinite reading of the DO no Accusative marker on the DO Possessive relation: If the speaker establishes the “possessive relation” between the Object NP and the Subject NP, definite reading is imposed: Uzbek (13) Opa-m pal’to-se-ni / sister-POSS.1 coat-POSS.3-ACC / (My) sister put on her coat (ACC-marker / *no ACC-marker). *pal’to-se coat-POSS.3 kij-di. put.on-PST NB4: The aforementioned phenomena are language specific, it is not the case that they are attested in all the languages under consideration. 4. Context properties contributing to the definite/indefinite reading of the DO NP: the lexical semantics of the verb (experimental data) Pre-experiment (2) Elicitation session in Udmurt: translation from Russian: ACC (14) a. so tEAl G’Et-i-z. he fire light-PST-3 He lit a fire (no marker). b. so kEs-i-z tEAl pu-ez he extinguish-PST-3 fire tree-ACC He extinguished the fire (ACC-marker). ? Is this an occasional coincidence in DO marking? ? If not, what discourse/sentence features regulate the speakers’ behavior: why do they encode DO as inducing the presupposition in (a), but not in (b)? ? What elements of the sentence force “specific” (cf. (b))/ ”nonspecific” (cf. (a))/ encoding? The distribution of the ACC is the same for different speakers Possible explanations: Different definiteness properties of the NPs in source language no NP internal structure - no difference no Difference in possessive relations (cf. ??) no Difference in aspectual characteristics of the verb – no Difference in semantic class of the verb – no (both are telic) Referent characteristics: quantized vs. cumulative - no Hypothesis: it is the lexical semantics of the verb that implies the definite/indefinite interpretation of the DO by the speakers Experiment elicitation session: 180 isolated simple sentences with different transitive verbs were given for translation from Russian into a target language (Mari, Komi, Udmurt, Mordvin (Finno-Ugric); Chuvash, Uzbekh, Kazakh (Turkic)) No clue about DO referential properties was given (no articles/definite markers in Russian). The verbs were taken from different aspectual classes and with different DO affectedness characteristics (according to Vendler, Dowty etc.). There was no difference in grammatical properties of the verbs (tense-aspect). e.g.: All the source sentences referred to the perfective past Results: we singled out a group of transitive verbs. These verbs manifest the same behavior concerning the DO encoding by the majority of speakers in five target languages*, e.g. Komi-Zyrian, Udmurt, Mordvin, Chuvash and Uzbek * The same experiment was conducted by Toldova for Adyghe language where the encoding of a DO NP also depends on its definiteness, this core group of the verbs manifesting the same behavior. The verbs from group 1 and group 2 form semantically correlated pairs (i.e. “opposite” situations: verbs of creation vs. verbs of destruction, etc.). Interpretation to mould to dig out to pour out to take off to unfasten to pour (into) to put on to knot to break to build to loose to return (smth. to smb.), to recommit to sell to let go to find to give (smth. to smb.) Subgro up (A) Group 2: Verbs occurring without overt accusative marker Subgro up (B) The resulting group can be divided into two groups of verbs with consistent preferences for the type of DO encoding (ACC-marker / no marker) Subgroup (C) Results Group 1: Verbs occurring with overt accusative marker Group 1 verbs Subgroup A verbs denote the annihilation of the state resulting from the corresponding member of the verb pair. Thus, it presupposes that DO referent be in this state beforehand, for the sentence to be felicitous (the clothes should have been already put on in order to take them off, the liquids poured in in order to pour them out etc.). Hence, the DO definiteness. Subgroup B verbs are “destruction verbs”, they denote the destruction of a DO referent (usually they denote the annihilation of the result of the “creation situation” denoted by the corresponding member of the verb pair). They presuppose that the DO referent is at the Subject’s disposal before the situation denoted by the subgroup B verb takes place. Subgroup C verbs denote the possessor change the Subject being the possessor before the event takes place. In order for the sentence to be felicitous, the DO is to be at the Subject’s possession before the possessor change takes place. to buy to catch Group 2 verbs The verb by default denotes the situation in which a new referent appears in the discourse in case there is no explicit presupposition inducer in the NP (e.g. verbs of creation such as build, knot introduce a new referent into the discourse etc.). Hence, the DO is (by default) indefinite, and occurs without ACC marker. Group1 verbs: the DO referent must be at the Subject’s disposal before the situation denoted by a verb of group 1 occurs. Hence, they entail the presence of the DO referent in the main DRS. Group2 verbs: the DO referent is introduced into the scene for the first time. Hence, they imply the first introducing of the DO referent into the DRS. More examples on the verb pairs Komi-Zyrian (15) a. Sija kErtal-i-s gErEd. he knot-PST-3 knot He made a knot (no marker) (lit. knotted a knot). ACC b. Sija raG-i-s gErEd-sE. he unfasten-PST-3 knot-ACC He unfastened (the) knot (ACC-marker). Uzbekh ACC (16) a. sengelem samsa jop-ti. b. sengelem samsa-ni je-di. sister-POSS.1 cake bake-PST cake-ACC eat-PST Some more examples from different languagessister-POSS.1 My sister baked a cake (no marker). My sister ate (the) cake (ACC-marker). Chuvash (17) a. aCa-sem tilĕduri tytnă. child-PL fox catch The children caught a fox (no marker). ACC b. aCa-sem tilĕduri-ne jană. child-PL fox-ACC set.free The children set (the) fox (ACC-marker) free. NB5: The verbs of both groups permit the opposite DO encoding in broader context: e.g., a verb of the second group can bear ACC if the referent of the DO is afore-mentioned, or if NP has other overt markers of definiteness. On the contrary, the DO NP in the sentences with a verb of the first group can occur without the ACC marker if the verb has other temporal or aspectual characteristics. Kazakh The adverb žii (‘often’) contributes to the habitual reading, (18) ül bala portfel ži:i žoγalta-dy. son child schoolbag often lose-PST (My) son often loses his schoobag. hence, the schoolbag is interpreted as generic (he always loses another schoolbag, not the same one), hence no ACC marking, although the verb to loose belongs to the ‘presupposition triggering’ group of verbs •Taken into account NB5, the distribution observed above does not work as a strict rule. in ‘default’ cases the “presuppositional” properties of the verb are crucial for the definiteness of the DO (if the context does not block the definite reading (as habitual context does, cf. ??)). Hence, lexical semantics of the verb define the ability of the DO to trigger the presupposition. The presupposition triggering for an NP in the languages under discussion can be a compositional characteristic of both NP and the verb (apart from the interpretation arising from the • However, aspectual composition). References Aissen J.1998. Differential Object Marking: Iconicity vs. Economy // CSLI Workshop ‘Is Syntax Different?’, Stanford December 1998. Comrie B. 1975. Subjects and direct objects in Uralic languages: a functional explanation of case-marking systems. ÉFou 12, pp. 5-17. Comrie B. 1979. Definite and animate direct objects: A natural class. Linguistica silesiana 3: 13-21. Enç M. 1991. The semantics of specificity // Linguistic Inquiry: 22.1, Winter 1991. Fodor J. D., Sag I. A. 1982. Referential and quantificational indefinites // Linguistics and Philosophy 5, 355-398. Geurts B. 1998. Presupposition and anaphors // Linguistics and Philosophy 21: 545-601. Kluwer Academic Publishers. Geurts B. 2002. Specific indefinites, presupposition, and scope // To appear in: R. Bäuerle, U. Reyle, and T.E. Zimmermann (eds.) Presupposition and Discourse. Elsevier, Oxford, 2002. Geurts B. 1999. Presuppositions and pronouns. Amsterdam – Lausanne – New York: Elsevier. Givón T. 1990. Syntax: a functional typological introduction. Amsterdam. Heim, I. 1982. The semantics of definite and indefinite noun phrases. Amherst, MA: University of Massachusetts doctoral dissertation. Johanson L. 1977. Bestimmtheit und Mitteilungsperspective im türkischen Satz // Zeitschrift der Deutschen Morgenländischen Gesellschaft. Supplement. III/2, 1186-1203. Kamp H., Reyle U. 1993. From Discourse to Logic. Introduction to Modeltheoretic Semantics of Natural Language, Formal Logic and Discourse Representation Theory. (Studies in Linguistics and Philosophy, v. 42). Dordrecht, Boston, London: Kluwer Academic Publishers. Lambrecht K. 1994. Topic, focus, and the Mental Representations of Discourse Referents. Cambridge, Cambridge University Press. Moravcsik E. 1978. On the Case Marking of Objects // Greenberg J.H. ed. Universals of human language, v.4: Syntax. Stanford University Press, Stanford, California, 1978, pp. 249-289. Muravyova I. 1992. Unmarked noun form in Turkic languages: a typological point of view // Altaic Religious Beliefs and Practices. Proceedings of the 33rd Meeting of the Permanent International Altaistic Conference, Budapest. Nikolaeva I. Secondary topic in information structure // Linguistics 2001, #39-1, pp. 1-49. Nilsson B. 1978. Definiteness and reference in relation to the Turkish accusative. // Orientalia suecana, v. 27-28, 1978, pp. 118-131. Wickman B. 1955. The Form of the Object in the Uralic Languages // Uppsala Universitets Årsskrift, 6. Acta Universitatis Uppsaliensis. Uppsala. Appendix 1: Languages under discussion: the genetic classification data Uralic languages Samoyedic languages Northern Samoyedic Enets, Nenets, Nganasan, Yurats Southern Samoyedic Selkup Finno-Ugric languages Ugric Turkic languages West Turkic Bolgar group Chuvash Oghuz (Southwestern) group Turkish, Afshar, Azeri, Turkmen e.a. Kypchak (Northwestern) group Kypchak-Bolgar languages Hungarian Hungarian Tatar, Bashkir, Baraba Kypchak-Cuman languages Ob Ugric Khanty, Mansi Crimean Tatar, Urum, Karachay-Balkar e.a. Kypchak-Nogay languages Finno-Permic Permic Kazakh, Karakalpak, Nogay Chagatay (Southeastern, Karluk) group Komi-Zyrian, Komi-Permyak, Udmurt Finno-Volgaic Mari, Mordvin Uzbek, Uyghur, Aini. Lop, Ili Turki East Turkic Kyrgyz-Kypchak group Kyrgyz, Altay Uyghur (Northeastern) group Yakut, Tuvinian, Khakas, Shor e.a. Khalaj Khalaj Appendix 2: Background on the ACC marker in Finno-Ugric and Turkic languages as a definiteness marker Finno-Ugric and Turkic languages allow co-existent strategies of Direct Object encoding. The direct object can be encoded by the accusative marker, or it can be unmarked. Reference grammars of these languages argue that this choice is determined by the referential status of the DO NP. Hence, the accusative marker in Mari, Komi-Zyrian, Udmurt, Mordvin(Finno-Ugric); Chuvash, Uzbekh, Kazakh (Turkic) works as a presupposition trigger, c.f. Komi-Zyrian: Presupposition binding: (A) afore-mentioning: (3) EtCid aG-i-s once find-PST-3 kaj zEr tuS’i sparrow oat dumajt-E: mYj-la zEr corn and think-PRS.3 tuS-sE what-COMPAR GYnjav-nY?. oat …Once the sparrow found an oat corn, and started thinking: “How can I part the oat corn?” (B) pronouns, proper names, demonstratives etc.: (4) vok pErEd-Y-s tonE brother cut-PST-3 sijE DEM pu-sE this / tree-ACC *pu. / *tree (His) brother cut this tree (ACC marker / *no marking). (C) complex NPs with restrictive relatives: (5) me aG-i-m bat’-lYS’ lYj-Em dEzmEr-sE. we see-PST-1PL father-GEN shoot-PART great.grouse-ACC We saw the great grouse (ACC marker) that father has shot yesterday. (D) superlatives: (6) peTa Peter Su-E mYj verm-a-s med<a say-3 that can-FUT-3 <EkYd most zadaCa-sE complicated reSit-nY. task-ACC solve-INF Peter says that he can solve the most complicated task (ACC marker). corn-ACC part-INF Presupposition accommodation: (7) sija mYrdYn SuE mYj sijEs he obstinately say that no puLa-sE siGi egi me lYjl-Y-s-nY he.ACC shoot.at-PST-3-PL aFEj. but bullet-ACC we PTCL NEG.1PL find He insists that he had been shot at, but we haven’t found the bullet (ACC marker). Tests on presupposition (‘family of tests’): definites under the scope of negation: (8) taja maSa VaSka-sE ZugEd-i-s? – this Mary ez, maSa VaSka-sE cup-ACC break-PST-3 taja peTa ZugEd-i-s. this Peter break-PST-3 no ez ZugEd, Mary cup-ACC NEG-3 break (The speaker enters the kitchen and sees shatters of a cup.) – Is it Mary who broke the cup? – No, it’s not Mary (lit. Marry did not break the cup (ACC marker)), it’s Peter who did it. Cf.: (9) taja maSa this duki-sE Mary ki<t-Ema? – perfume-ACC eg ki<t, taja NEG.1 spill this eg, me duki spill-EVID.PST vol-Y-s no I perfume nina-EVin. come-PST-3 Nina-aunt (The speaker enters the room and smells the heavy odor of a perfume) – Is it Mary who spilled out the perfume? – No, I haven’t spilled out any perfume (no marker), it’s only that aunt Nina came to see us. definites under the scope of modal verbs, in if-clauses etc. (vs. (4)): (10) pukt-Ema zev ydZyd klad no put-EVID.PST very big treasure n’ekod o-z nobody NEG.PRS-3SG sij-Es boSt-nY-sE but this-ACC take-INF-EMPH vermi. can The treasure is hidden there, but no one ever will be able to take it. Hence, ACC marker in Komi-Zyrian does trigger presupposition.