CC BY
29
Лингвистика
DOI: 10.31862/2500-2953-2019-3-53-73 Е.А. Лютикова1, 2, П.В. Гращенков1, 3
1 Московский государственный университет имени М.В. Ломоносова, 119991 г. Москва, Российская Федерация
2 Государственный институт русского языка имени А.С. Пушкина, 117485 г. Москва, Российская Федерация
3 Институт востоковедения РАН, 107031 г. Москва, Российская Федерация
Синтаксис лично-числового согласования в татарском языке
В статье исследуются случаи необычного согласования с кванторами, интен-сификаторами и анафорами, засвидетельствованые в татарском языке в тех случаях, когда указанные элементы содержат лично-числовые показатели. Такие единицы могут демонстрировать не только согласование по третьему лицу, но также и согласование по лицу своего рестриктора или связывающего элемента. Мы считаем, что в таких конструкциях кванторы, интенсификаторы и анафоры обладают теми признаками, которые проявляются при согласовании. Мы предлагаем механизм, позволяющий указанным единицам приобретать признаки своего рестриктора или связывающего элемента. Мы исходим из идеи о том, что согласующиеся кванторы, интенсификаторы и анафоры содержат проекцию местоимения, обладающую своим набором неозначенных интерпретируемых признаков. Этот набор признаков оказывается означен вследствие идентификации признаков местоимения и связывающего его посессора. Ключевые слова: согласование, лицо, маркированные лично-числовыми показателями кванторы, интенсификаторы, анафоры, распространение признаков, связывание, татарский язык
Благодарности: Мы благодарны двум анонимным рецензентам, критика которых способствовала улучшению статьи.
Работа Е.А. Лютиковой поддержана проектом «Параметрическое описание языков Российской Федерации», реализуемым в Государственном институте русского языка им. А.С. Пушкина.
® I Контент доступен по лицензии Creative Commons Attribution 4.0 International License K^k^Hä^^B The content is licensed under a Creative Commons Attribution 4.0 International License
ДЛЯ ЦИТИРОВАНИЯ: Лютикова Е.А., Гращенков П.В. Синтаксис лично-числового согласования в татарском языке // Рема. Rhema. 2019. № 3. С. 53-73. DOI: 10.31862/2500-2953-2019-3-53-73
DOI: 10.31862/2500-2953-2019-3-53-73
E. Lyutikova1, 2, P. Grashchenkov1, 3
1 Lomonosov Moscow State University, Moscow, 119991, Russian Federation
2 Pushkin State Russian Language Institute, Moscow, 117485, Russian Federation
3 Institute of Oriental Studies, RAS, Moscow, 107031, Russian Federation
On the syntax of person agreement in Tatar
In this paper, we analyse the intricate agreement pattern attested with inflected quantifiers, intensifiers and anaphors in Tatar. These nominals can trigger not only a default 3rd person agreement, but also the marked person agreement reflecting the features of their restrictor or binder. We propose that in these constructions, inflected quantifiers, intensifiers and anaphors bear the features the agreement reveals, and propose a mechanism allowing inflected quantifiers, intensifiers and anaphors to aquire the features of the restrictor or binder. We build on the idea that agreeing inflected quantifiers, intensifiers and anaphors contain a minimal pronoun equipped with a set of unvalued interpretable features, and that this feature set gets valued via feature sharing.
Key words: agreement, person, inflected quantifiers, intensifiers, anaphors, feature sharing, binding, Tatar
Acknowledgments: We are grateful to two anonymous reviewers, from whose criticism the paper benefited greatly.
Ekaterina Lyutikova's work has been supported from the project "Parametric description of the languages of Russian Federation" realized at the Pushkin State 1 Russian Language Institute.
u s m
i FOR CITATION: Lyutikova E., Grashchenkov P. On the syntax of person agreement
^ in Tatar. Rhema. 2019. No. 3. Pp. 53-73. DOI: 10.31862/2500-2953-2019-3-53-73
Rhema. Рема. 2019. № 3 1. Agreement and formal features
Agreement is one of the basic and pervasive grammatical phenomena that has been in the focus of attention of various theoretical - both functional and formal - and typological linguistic research [e.g. Corbett, 2006; Baker, 2008; Preminger, 2014; Matasovic, 2018; Nichols, 2018; Haig, Forker, 2018, among many others]. The minimalist family of syntactic analyses of agreement goes back to the seminal paper by Noam Chomsky [Chomsky, 2000]. In this paper, the notion of agreement has been extended, so that agreement is conceived of as a generalized operation ensuring legibility of linguistic expressions at the interface levels. This view of agreement is based on the system of formal features that lexical items bear: the feature can be interpretable on the lexical item X (or its phrase XP), and then valued on X in the lexicon, or uninterpretable on the lexical item X (and its phrase XP), and then unvalued on X in the lexicon. Since uninterpretable features are not legible at the interface levels, they have to be "defused" in the course of syntactic derivation. Chomsky proposes that this can be done by valuing uninterpretable features; feature valuation, in its turn, is the result of agreement. Agreement is a process in which an unvalued uninterpretable feature on the probe matches the valued interpretable feature of the goal, copies this value and then can be erased, to the effect that no uninterpretable feature reaches the interface. Movement appears as a by-product of agreement, when the probe attracts the goal. In this way, agreement constitutes the core operation in the feature-driven syntax.
Importantly, the system outlined above presupposes a strong correlation between interpretability and valuation: interpretable features enter the derivation valued, whereas uninterpretable features enter the derivation unvalued, and have to aquire values via agreement. Consequently, interpretable features cannot be valued in syntax, and uninterpretable features cannot be a source of features' values. Pesetsky and Torrego (2007) criticize this conjecture as too restrictive and propose that interpretability and valuation are independent second order features. Consequently, agreement is conceived of as feature sharing: an unvalued occurrence of the feature on the probe can match valued or unvalued and interpretable or uninterpretable occurrence of the same feature on the goal. Feature sharing forms a link between multiple occurences of the same feature; in this chain, only one valued and one interpretable occurrence suffice to produce a grammatical configuration.
In this paper we address theoretical challenges posed by agreement with inflected quantifiers in Tatar, one of the Turkic languages spoken in Russia. We show that the standard minimalist view of agreement cannot capture the peculiar agreement pattern exhibited by these constructions. By contrast,
I _
ISSN 2500-2953 Rhema. PeMa. 2019. № 3
a modified theoretical system along the lines of [Pesetsky, Torrego, 2007] allows us to provide a principled account of the phenomenon.
The paper is organized as follows. In section 2, we present the puzzle and outline possible ways of solving it. Then, in sections 3 and 4, we consider the two options proposed in the literature in accounting for similar phenomena, and show why these analyses are not suitable for Tatar. In section 5, we propose our own account for the peculiar agreement pattern attested in Tatar. Section 6 concludes.
2. The puzzle
Tatar has an elaborated system of person-number agreement which involves various structural configurations. Predicate agreement in Tatar is controlled by the nominative subject exclusively, as examples (1)-(2) demonstrate. So, no "trigger-happy" agreement [Comrie, 2003] or "omnivorous" agreement [Nevins, 2011] takes place.
(1) Min a-m kur-de-*(m). i.nom s(he)-Acc See-PST-ISG 'I saw her.'
(2) Ul mine kur-de-(*m). s(he).NOM I.acc See-PST-1SG 'She saw me.'
Agreement with a subject possessing the marked person feature is obligatory; thus, examples (3a-b) are ungrammatical without the agreement affix on the verb. Number agreement with a 3rd person subject, exemplified in (4), is optional, probably governed by the collective/distributive distinction, see [Lyutikova, 2017] for details.
(3) a. Min kil-de-*(m).
i.nom come-PST-1sG 'I came. b. Sin kil-de-*(q).
you.NOM come-PST-2sG 'You came.'
(4) Bala-lar kil-de-(lar). child-PL come-PST-PL '(The) children came.'
Surprisingly, inflected quantifiers, intensifiers and anaphors exhibit two patterns of predicate agreement. The non-agreeing pattern is what we would expect from these nominals, which normally bear the 3rd person
feature; it is exemplified in (5). However, there is another option, that is, the agreeing pattern shown in (6), where the predicate exhibits ^-features of the quantifier's restrictor.
(5) Jal-dan kajt-kac, tagin berar-egez kur-de-me a-ni? vacation-abl return-cNv again any-2pL See-PST-Q (s)he-Acc 'Did anyone of you see him again after returning from vacation?' [TNC]1
(6) A xazer, afande-lar, di-de Laplas, berar-egez susi and now sir-PL Say-PST Laplace any-2PL this kijem-ne kij-ep irkenlek-ka cig-ip kit-arga clothing-Acc put_on-cNv space-DAT exit-cNv go-iNF tela-mi-sez-me?
Want-NEG.IPF-2PL-Q
'And now, gentlemen, would anyone of you put on this clothing and go outside? - said Laplace.' [TNC]
Lexical items involved in the variation include universal quantifiers (e.g. har 'every', bari da 'all'), existential quantifiers (e.g. berar 'any', hicber 'no one'), adjectival interrogative pronouns (kajsi 'which') and anaphors -reflexives and reciprocals (uz 'self', ber-berse 'each other'). The inflected reflexive pronoun uz 'self' in Tatar functions as both a bound anaphor, as in (7a), and an intensifier, as in (7b); in both cases, the agreeing pattern is readily available.
(7) a. Marat uz-e-n kur-de.
Marat.NOM self-3-Acc see-PST 'Marat saw himself.' b. (Marat) uz-e a-ni kur-de.
Marat.NOM self-3 (s)he-Acc see-PST 'Marat himself / He himself saw her.'
Importantly, only inflected pronouns exhibit the agreeing pattern. Example (8) contains a nominal interrogative pronoun kem 'who' which does not take inflection characterizing the restrictor. In this case, predicate agreement with the restrictor is ungrammatical.
(8) (Sez-dan) bez-ne kem jakl-ij-(*siz), kem kil-er-(*sez) you-ABL we-Acc who defend-PRS-2PL who come-FUT-2PL bez-ga jardam-ga?
we-DAT help-DAT
'Who (of you) defends us, who will come to help us?'
1 Examples tagged as [TNC] come from the Tatar national corpus "Tugan tel" (http://www. tugantel.tatar/). Non-tagged examples are elicited.
This data immediately reminds us of several cross-linguistic parallels. First, inflected quantifiers and anaphors in Tatar are structurally equivalent to the possessive construction (the so called "3rd ezafe construction" in the descriptive grammars of Tatar). In this construction, the overt restrictor appears in genitive, and the head agrees with the genitive possessor. Therefore, agreement with inflected quantifiers can be regarded as agreement with internal possessors attested in a number of languages [Barany et al., 2019]. Secondly, predicate agreement with inflected quantifiers in the partitive construction is reported for Turkish [Ince, 2007; Aydin, 2008]. The authors claim that the predicate agrees with the silent pro in the subject position, the floating quantifier being stranded. Finally, agreeing inflected quantifiers are also found in Quechua [Muysken, 1989, 2013; Faller, Hastings, 2008] and several Bantu languages [Baker, 2008] including Kinyarwanda [Jerro, 2013; Jerro, Wechsler, 2015]. In particular, Jerro anf Wechsler claim for Kinyarwanda that at some stage of the diachronic development, the person/ number inflection on the quantifier is a cliticized referential pronoun, and this is why the predicate agreement reveals its features.
The approaches outlined above exploit three different ideas of how to make agreement with inflected quantifiers and anaphors possible. The first idea is that the predicate agrees with the possessor of its subject. The second idea presupposes that the inflected quantifier is non-argumental and that the actual controller of the predicate agreement is the empty pronoun pro that bears the ^-features of the restrictor. The third and at first glance less plausible scenario is that inflected quantifiers and anaphors somehow aquire the ^-features of their restrictors. We will argue for the latter option, but first we are going to examine the alternatives.
3. Alternative 1: Agreement with possessor
Inflected quantifiers, intensifiers and anaphors in Tatar are built as a partitive construction exemplified in (9), and allow for the restrictor pronoun to surface as a genitive DP or be represented by pro, as shown in (10).
(9) baran-nar-neq kara-lar-i ram-PL-GEN black-PL-3 'the black rams' (lit. the black ones of rams)
(10) bez-neq / pro1pL kajsi-lar-ibiz we-GEN which-PL-1PL
'which (pl) of us'
Crucially, the partitive construction in Tatar is structurally and distributionally identical to the possessive construction with the referential
possessor, compare (10) to (11). The possessor receives the genitive encoding and controls person/number agreement on the possessee.
(11) bez-neq / pro 1p we-GEN 'our rams'
baran-nar-ebez ram-PL-lPL
This fact allows us to consider the agreeing pattern with inflected quantifiers, intensifiers and anaphors as an instantiation of agreement with the prominent internal possessor.
Agreement with the prominent possessor is attested in various linguistic families and areas. In the recent paper collection [Barany et al., 2019] examining the phenomenon multiple examples similar to Tatar are attested. Thus, in Maithili, an Indic language spoken in India and Nepal, the predicate agreement distinguishes between the standard agreement with the nominative subject (12a) and the special series of non-nominative agreement markers, which can be controlled by the possessor (12b).
(12) Maithili / Indo-European (Indic) [Yadava et al., 2019, p. 51]
a. [tohar nokar] ae-l-ai you.L.GEN servant.NOM come-PST-3 'Your (low respect) servant came.'
b. [tohar nokar] ae-l-au
you.L.GEN servant.NOM come-PST-2L.NON_NOM 'Your (low respect) servant came.'
An interesting example is provided by Tanty Dargwa. In this language, a number of lexical items like 'a waist' or 'a half lack their own nominal class specification, and copy it from the possessor, which is obligatory with such nominals. The predicate agreement thus reflects the nominal class of the overt or covert possessor.
(13) Tanty Dargwa / North-Caucasian [Lander, 2015]
a. hit-a-la cVrt-se
this-OBL-GEN thin-ATR
'She's got a thin waist.'
b. hit-a-la cVrt-se
this-OBL-GEN thin-ATR
'He's got a thin waist.'
c. zimizal-la cVrt-se ant-GEN thin-ATR
r-ag F-waist
w-ag M-waist
b-ag N-waist
te-r. be-F
te-w. be-M
te-b. be-N
'The ant has got a thin waist.'
I _
ISSN 2500-2953 Rhema. Рема. 2019. № 3
Moreover, examples of possessor's prominence are found within the Turkic family. It has been argued that in many Turkic languages (Bashkir, Shor, Turkish, Tuwan, Uyghur, Uzbek) the internal possessor of the subject acts as one of the pivots in the same-subject relation [Nikolaeva et al., 2019, 14ff]. This is illustrated with the Bashkir sentence (14).
(14) Bashkir [Say, 2019, p. 211]
Bolat [ [pro bas-a] awart-ap] kitap
Bulat head-poss.3 ache-ss.coNv book
uqa-w-a-n tuqta-t-ta.
read-NMLz-Poss.3-Acc stop-CAus-PST 'Bulat'st head started to ache and het stopped reading the book'
The -p converb is used for the same-subject clauses and generally cannot have an overt subject. However, this restriction is lifted if the converb's subject has a possessor coreferential with the matrix subject.
The linguistic evidence of this type suggests that possessors can be syntactically active at the clause level. In her encyclopedic paper Deal summarizes various mechanisms that make the possessor visible to the functional structure of the clause [Deal, 2017]. For the prominent internal possessor, several analytical options are available, including government or backward control or even long distance agreement. To sum up, indexing of the internal possessor on the verb is an empirically motivated and theoretically elaborated issue.
However, the prominent possessor analysis turns out to be a dead end for Tatar. Apart from the fact that possessor government, raising and long distance agreement are normally fed by internal arguments exclusively, which is not the case in Tatar, we observe that predicate agreement with true possessors is ungrammatical. Example (15) shows that when the standard possessive construction occupies the subject position, the predicate can only agree with the head noun.
(15) bez-neq / pro baran-nar-ebez kil-de / kil-de-lar / *kil-de-k. we-GEN ram-PL-1PL come-PST / come-PST-PL / come-PST-1PL 'Our rams came.'
Example (15) shows convincingly that possessive inflection on the nominal head is an exponent of possessive agreement and cannot be a controller of the predicate agreement. Therefore, neither the genitive possessor itself nor its features copied onto the head can serve as a goal for the predicate's ^-probe.
4. Alternative 2: Agreement with pro
Tatar examples (1)-(6) align directly with the predicate agreement pattern found in Turkish as described in [Ince, 2007; Aydin, 2008]. The personal pronoun as a subject triggers the obligatory person/number agreement (16), possessive noun phrases only trigger the standard predicate agreement with the head (17), and inflected quantifiers are compatible with both construals (18).
(16) Biz ev-e git-ti-k. we.NOM home-DAT go-PST-lPL 'We went home'
(17) Arkada§-imiz ev-e git-ti-0. friend-lPL.NOM home-DAT go-PST-3sG 'Our friend went home'
(18) Hep-imiz ev-e git-ti-k / git-ti-0. all-1PL.NOM home-DAT go-PST-1PL / go-PST-3sG 'All of us went home'
The Turkish agreeing construction with inflected quantifiers has a number of characteristic properties that prompt the researchers towards the floating quantifier analysis. First, it is attested in finite clauses exclusively, that is, clauses with a nominative subject. Example (19) shows the agreeing inflected quantifier within the finite embedded clause. In (20), the nominalized embedded clause is exemplified. In this case, the predicate person agreement is ungrammatical. Since feeding of the quantifier float is often restricted to the (finite) subject, this pattern is expected.
(19) Ali [hep-imiz ev-e git-sin/-elim] isti-yor. Ali all-1PL.NOM home-DAT go-iMP.3sG/-OPT.1PL want-PRS 'Ali wants all of us to go home.'
(20) Ali [hep-miz-in ev-e git-tig-in/*-imiz]-i bil-iyor. Ali all-1PL-GEN home-DAT go-NMLz-3sG/-1PL-ACC know-PRS 'Ali knows that all of us went home.'
Similarly, no agreement with inflected quantifiers in the possessive construction is attested, cf. (21). If the inflected quantifier itself occupies the possessor's position in the embedding noun phrase, its head cannot bear the restrictor's features.
(21) hep-imiz-in / ik-imiz-in araba-si/*miz all-1PL-GEN / two-1PL-GEN car-3/1PL 'all of our/two of our's car'
J
Finally, examples (22)-(23) demonstrate that the overt restrictor of the agreeing quantifier bears no case marker, and can therefore be analyzed as a nominative subject.
(22) Siz / *siz-in kag-iniz on-u hakli bul-uyor-sunuz? you.NOM / *you-GEN how.many-2PL he-ACC right find-PRS-2PL 'How many of you think that he is right?'
(23) *Siz / siz-in kag-iniz on-u hakli bul-uyor? *you.NOM / you-GEN how.many-2PL he-ACC right find-PRS 'How many of you think that he is right?'
To sum up, the agreeing pattern in Turkish has got all the hallmarks of the subject-related floating quantifier construction. This is essentially the idea underlying the analyses proposed in the literature; the account in [Aydin, 2008] differs minimally from that of [Ince, 2007] with respect to the exact position of the stranded and agreeing elements. The researchers claim that in the agreeing construction, the subject DP has a complex internal structure involving pro-doubling, as shown in (24).
(24) [DP1 [DP2 hep-imiz] |pro1PJ] / [DP1 [pro1PL] [DP2 hep-imiz]]
all-1PL all-1PL
On its way to the subject position, the DPj splits, producing the floating quantifier configuration. At that point, the analyses diverge. Ince (2007) proposes that pro reaches Spec, AGRP from where it controls predicate agreement, whereas the DP2 containing the floating quantifier lands in Spec, EP, which is assumed to be a canonical subject position. The account is then based on the idea that subject properties are split between the two structural positions. Aydin (2008) claims that pro is a bona fide subject situated in Spec, TP and controlling the predicate agreement, whereas the quantifier is stranded inside the verbal domain, supposedly in Spec, vP or Spec, VP. Crucially, in both analyses it is pro that enters the agreement relation with the functional head responsible for the person/number agreement and nominative case assignment; the stranded quantifier remains caseless or bears the default case. It is important to note that the no-case or default-case analysis of the floating quantifier is tenable due to the fact that nominative case has no overt marker in Turkic languages. Consequently, the nominal lacking a case affix can in principle be analysed as nominative or caseless.
Despite the striking similarities of Turkish and Tatar agreeing patterns in finite clauses, they differ significanty in other configurations.
In Tatar, inflected quantifiers, intensifiers and anaphors trigger ^-agreement in all the agreement configurations, that is, in the possessive construction,
embedded nominalized clause and postpositional construction. This is demonstrated below with the corpus hits. Example (25) shows the agreeing pattern in the possessive construction; example (26) demonstrates agreement with the inflected quantifier as a subject in the nominalized clause; in (27), the agreeing pattern within the postpositional construction is exemplified.
(25) Ugtslt xezmattaslek ocen ber-ber-ebez-neq
beneficial cooperation for rec-rec-1pl-gen momkinlek-lar-ebez-ne ham ixttja^-lar-ibiz-ni capacity-PL-1PL-Acc and interest-PL-1PL-Acc
ojran-erga kirak. study-iNF need
'For a mutually beneficial cooperation, we have to study capacities and interests of each other.' [TNC]
satuci-dan produkcija seller-ABL production bul-u-bul-ma-u
be-NMN-be-NEG-NMN
sostav-t-nda GMO content-3-Loc GMO turinda sora-gan-ibiz about ask-PART-1PL
(26) Kajsi-biz-niq
which-1PL-GEN komponent-lar-i component-PL-3 bar? be.ExisT
'Which of us asks the seller about the presence of GMO components in the products?' [TNC]
(27) Üz-eq-neq self-2sG-GEN küz-eq-a eye-2sG-DAT
jan-iq-da
near-2sG-Loc genâ
EMPH
bit, here kara-p look-coNv
tins, diligent
tijnak,
modest
tora.
AUX
'Here is he near you, diligent and modest, keeps looking you in the eye.' [TNC]
Importantly, in configurations other than the finite subject, the restrictor surfaces as a genitive, not nominative, pronoun. This is shown in (29)—(31), as opposed to (28). This means that if the quantifier in these constructions were stranded, both pro and the quantifier phrase would receive marked genitive case, and no caseless nominal could arise.
(28) Sez / sez-neq berâr-egez cig-ip kit-ârgâ telâ-mi-sez-me? you.NOM / you-GEN any-2pL exit-CNv go-iNF want-NEG.iPF-2pL-Q 'Would anyone of you go outside?'
(29) *bez / bez-neq ber-ber-ebez-neq ixtija^-lar-ibiz
we.NOM / we-GEN REc-REc-lPL-GEN interest-PL-1PL 'interests of each other of us'
J
(30) *Bez / bez-neq kajsi-biz-niq aniq turinda sora-gan-ibiz We.NOM / we-GEN which-1PL-GEN it. GEN about ask-PART-1PL bar?
be.ExisT
'Which of us asks about it?'
(31) *sin / sineq uz-eq-neq jan-iq-da you.NOM / you. gen self-2sG-GEN near-2sG-LOC 'near you'
Therefore, the stranding analysis proposed for Turkish can only be extended to cover Tatar finite predicate agreement with inflected quantifiers. Crucially, it cannot explain agreement with inflected quantifiers and intensifiers within possessive, nominalized or postpositional configurations. The reason for it is that in possessive, nominalized and postpositional constructions, only one structural case is licensed, as evidenced by the ungrammaticality of (32) and (33).
(32) *minem Marat-niq surat-i/-im
I.GEN Marat-GEN picture-3/-1sG
intended: 'my picture of Marat'
(33) *minem Marat-niq kur-gan-e/-em
I.gen Marat-GEN see-PART-3/-1sG
intended: 'my seeing of Marat'
Moreover, no case copying or case agreement takes place in appositive constructions shown in (34)-(35).
(34) kurse-(*ga) abi-ga neighbor-DAT grandma-DAT 'to the neighbor old lady'
(35) Marat-(*niq) ukituci-niq surat-i Marat-GEN teacher-GEN picture-3 'the picture of Marat the teacher'
This means that with non-nominative DPs, stranding should be disallowed, since no additional source for the genitive encoding of the stranded quantifier or intensifier is available. As for the inflected anaphors, the stranding analysis is odd irrespective of the particular agreement configuration, since anaphors do not float.
Therefore, we conclude that the genitive restrictor noun phrase cannot be coargumental or adjoined to the inflected quantifier, intensifier or anaphor,
as the stranding analysis requires. On the contrary, the genitive restrictor DP is embedded under the inflected quantifier/intensifier/anaphor DP. The corresponding structural analyses are represented in (36)-(37).
(36) Turkish inflected quantifiers
[Dpi [DP2 biz / pro1pL] [DP3 hep-imiz]] we.NOM all-1PL
'we all'
(37) Tatar inflected quantifiers
[DP1 [DP2 bez-ne4 /^°!PJGEN [D' har-ebez]]
we-GEN all-1PL
'all of us'
Let's take a stock of what we have established so far. Agreement with inflected quantifiers, intensifiers and anaphors in Tatar has a number of peculiar properties which make analyses proposed for similar phenomena in other languages not suitable for Tatar. More specifically, this agreement is fed by derived pronouns exclusively, hence agreement with prominent possessor is excluded. This agreement is manifested in whatever agreement configurations, hence agreement with the subject pro under stranding analysis is not sufficient. Finally, this agreement varies systematically with the 3rd person (default) agreement, hence the two options should be structurally represented. In the rest of the talk we develop the analysis which is based on the assumption that it is the inflected quantifier (intensifier, anaphor) that controls agreement and that it can aquire ^-features of its restrictor.
5. Analysis
We propose that person agreement with inflected quantifiers, intensifiers and anaphors is an instance of the standard agreement, and that inflected quantifiers, intensifiers and anaphors bear the ^-features this agreement reveals.
In order to develop this analysis we first identify the set of pronominal elements available in Tatar. We adopt Dechaine and Wiltschko's proposal that pronouns come in various structural size [Dechaine, Wiltschko, 2002, 2010]. Specifically, we distinguish between DP-pronouns (which are indexical, cannot shift and cannot be bound) and ^P-pronouns (which are non-indexical, can shift and can be bound). DP-pronouns embed the ^P-layer and inherit the ^-features of the 9 head, as represented in (38).
ISSN 2500-2953 J
(38) a. DP-pronoun DP
D
b. ^P-pronoun 9P
9P
9
NP
N
NP
N
The properties of various pronouns in Tatar have been studied extensively in the recent PhD thesis by Podobryaev [Podobryaev, 2014]. Podobryaev shows, in particular, that overt personal pronouns and covert pro differ systematically. Overt 1st / 2nd person pronouns cannot undergo indexical shift, whereas 1st or 2nd person pro<1/2> can (but does not have to). Similarly, overt 1st / 2nd person pronouns are referential expressions, subject to condition C of the binding theory, whereas 1st or 2nd person pro<1/2> is ambiguous between pronominal and anaphoric construals. The corresponding examples can be found in (39)-(40).
ajt-ti?
say-pst
ajt-ti? say-pst
(39) a. Alsu [min kaja kit-te-m diep]
AIsu.nom I.nom where go.out-PST-1sG that 'Which place did Alsu say I went?' <non-shifted> *'Which place did Alsu say she went?' <shifted> b. Alsu [pro kaja kit-te-m diep]
AIsu.nom pro<1sG> where go.out-PST-1sG that 'Which place did Alsu say I went?' <non-shifted> 'Which place did Alsu say she went?' <shifted> [Podobryaev, 2014, p. 84]
(40) a. *Min mine kur-de-m.
I.nom I.acc see-PST-1sG intended: 'I saw myself.'
b. Min pro ata-m kur-de-m. I.nom pro<1SG> father-1sG see-PST-1sG 'I saw my father.'
c. Marat pro ata-m kur-de-m. Marat.NOM pro<1sG> father-1sG see-PST-1sG 'Marat saw my father.'
Based on these diagnostics we conclude that overt 1st / 2nd person pronouns in Tatar are DPs, whereas pro is ambiguous between DP and 9P construals.
Let us further investigate the ^-feature set of ^P-pronouns. If the pronoun is free, its ^-feature set is valued. But if the pronoun is bound, it can enter the derivation with unvalued ^-features and aquire features' values via binding, as [Kratzer, 2009] suggests. Therefore, it is plausible that Tatar may possess a minimal pronoun 9 with unvalued interpretable person and number features.
We believe that this minimal pronoun is the source of derived personal pronouns available in Tatar. We propose that agreeing inflected quantifiers, intensifiers and anaphors differ from the non-agreeing ones in that they contain an additional 9P layer between D and (substantivized) nominal structure, as shown in (41)-(42).
(41) agreeing inflected quantifier DP
DP D'
bezney / pro 1PL we.gen ii>: val
9P
xp x 9
berar Xpron
'any'
D
-bez -1pl
(42) non-agreeing inflected quantifier DP
DP
bezney / pro 1PL
we.gen
ii>: val
D'
XP
berar 'any'
D
-bez -1pl
The trees in (41)-(42) represent the internal structure of agreeing and non-agreeing inflected quantifiers berar-ebez 'any of us'. Non-agreeing inflected quantifier in (42) is a 3rd person DP, it is non-pronominal and behaves like a referential expression. The additional ^P-layer of the agreeing inflected quantifier in (41) converts it into a derived personal pronoun.
As a next step, the 9P in (41) gets bound by the c-commanding restrictor in Spec, DP. The ^-sets of 9 and the restrictor match, and the unvalued features on 9 are identified with the ^-set of the restrictor. Technically,
I _
ISSN 2500-2953 Rhema. Рема. 2019. № 3
this operation may be conceived as a reverse agree [Wurmbrand, 2017] or as feature sharing [Pesetsky, Torrego, 2007]. Crucially, the restrictor is the nearest source of ^-features for the minimal pronoun, and this is why inflected quantifiers and intensifiers aquire the ^-set of their restrictor.
Finally, transmission of features from 9 to D takes place. Valued features on 9P are further inherited by D in the standard manner, producing 1st / 2nd person inflected quantifiers and intensifiers, which are definite R-expressions, analogous to overt personal pronouns.
The derivation of bound inflected anaphors is slightly different. As a starting point, let's observe that bound 1st / 2nd person anaphors never contain an overt genitive possessor. Consider examples (43) and (44).
(43) Min gena uze-m makta-di-m.
I.nom only self-lsG praise-PST-1sG
'Only I praised myself.'
1. strict reading: nobody else praised me
2. sloppy reading: nobody else praised herself
(44) Min gena minem uze-m makta-di-m.
I.nom only I.gen self-lsG praise-PST-1sG
'Only I praised myself.'
1. strict reading: nobody else praised me
2. *sloppy reading: nobody else praised herself
In (43), the pronoun uzem 'myself' lacks an overt possessor and can be interpreted as free, giving rise to the strict reading (43.1), or as bound, producing the sloppy reading (43.2). In (44), the pronoun contains an overt possessor, and the bound interpretation is excluded.
We conclude that agreeing bound anaphors are DPs hosting a ^P-pro in their specifier. The ^-set of pro is identified with the ^-set of the minimal pronoun 9 via feature sharing [Pesetsky, Torrego, 2007], creating a chain of occurrences of the same features. The ^-set of pro is valued by its external syntactic binder, along the lines of [Kratzer, 2009]. In this way, the external binder, the bound pro in Spec, DP, and the minimal pronoun 9 of the derived anaphor end up by bearing identical values of ^-features. These features are further reflected by the agreeing external probe. Importantly, agreement with the external probe can take place before binding and feature valuation.
Suppose that the agreeing anaphor is embedded under a postposition. Then feature sharing would proceed as follows. First, the unvalued ^-sets of pro and minimal pronoun 9 are identified as two occurrences of the same ^-set. Then, the ^-set on D is coindexed with them, giving rise to a three-link chain. Then, the ^-probe on P gets coindexed with the chain, producing
the fourth occurrence of the same ^-set. Finally, when the binder (presumably the subject) enters the derivation, its valued ^-set is identified as yet another occurence of the same ^-set, and the values of the ^-features are being copied to all the occurrences of the coindexed ^-set.
Additional evidence supporting the analysis comes from case assignment in postpositional constructions. Denominal postpositions in Tatar involve differential argument marking: personal pronouns receive genitive encoding and combine with the agreeing form of the postposition, whereas other nominals appear in the nominative (unmarked) case and combine with the default 3rd person form of the postposition. This is illustrated in (45). We see that the 1st person personal pronoun min 'I' differs consistently from other nominals including possessive phrases, proper names and non-personal pronouns in case marking and agreement pattern.
(45) a. minem / *min
I.GEN / I.NOM
'instead of me'
b. ata-m / *ata-m-mq father-lsG / father-1sG-GEN 'instead of my father'
c. Marat / *Marat-mq Marat.NOM / Marat-GEN 'instead of Marat'
d. kem / *kem-neq who.NOM / who-GEN 'instead of whom'
urin-im-da instead-1sG-Loc
urin-in-da instead-3-Loc
urin-in-da instead-3-Loc
urin-in-da instead-3-Loc
The crucial data comes from the experimental study of denominal postpositional constructions in Tatar [Lyutikova, Gerasimova, 2019]. The aim of the study was to estimate variation in differential argument marking in the postpositional construction. 119 native speakers of Tatar participated in the study; the respondent fulfilled production and rating tasks. Among the stimuli, there were 1st and 2nd person inflected quantifiers kajsibiz 'which of us', kajsigiz 'which of you' and 1st and 2nd person reflexive pronouns uzem 'myself' and uzeq 'yourself'.
The study revealed strong correlation between agreement and case in both production (fill in the blank task) and perception (acceptability rating task): agreeing inflected quantifiers and anaphors are preferably genitive, and non-agreeing inflected quantifiers and anaphors are preferably nominative. The most popular patterns are presented in (46) for inflected quantifiers and (47) for inflected anaphors.
J
(46) a. kajsi-biz-nig / ?*kajsi-biz urin-ibiz-da
which-1 PL-gen / which-1 pl instead-1 pl-loc 'instead of which of us'
b. ?*kajsi-biz-nig / kajsi-biz urin-in-da
which-1 pl-gen / which-1PL instead-3-Loc 'instead of which of us'
(47) a. üz-em-neg / ?*üz-em urin-im-da
self-1sG-GEN / self-1sG instead-1sG-Loc 'instead of which of us'
b. ?*üz-em-neg / üz-em urin-in-da
self-1sG-GEN / self-1sG instead-3-Loc 'instead of which of us'
Since genitive encoding occurs with personal pronouns exclusively, we consider this data as evidence that inflected quantifiers and anaphors convert into personal pronouns, and this is what our analysis essentially claims.
6. Conclusions
In this paper we have discussed agreement with inflected quantifiers, intensifiers and anaphors in Tatar, which allows for both non-agreeing and agreeing pattern. We have considered three options of accomodating the unexpected person/number agreement: agreement with the prominent possessor; agreement with pro (inflected quantifiers and intensifiers being stranded); standard agreement with the inflected quantifier, intensifier and anaphor, which have aquired ^-features of their restrictors or binders. We have opted for the last variant and provided an account of how exactly ^-features of the restrictor or binder could be transferred to the quantifier, intensifier or anaphor. We have based the account on the idea that agreeing inflected quantifiers, intensifiers and anaphors contain a minimal pronoun equipped with a set of unvalued interpretable features, and these features get valued by the restrictor (for quantifiers and intensifiers) or by the binder (for anaphors).
Importantly, the analysis proposed in this paper requires a less restrictive view of feature valuation and agreement. First, the interpretable ^-features of the minimal pronoun get valued in the course of derivation. Secondly, the c-command relation between the probe and the goal are reversed: the constituent with the valued ^-set c-commands the constituent with the unvalued ^-set. Finally, the derivation of constructions involving agreeing inflected anaphors requires that unvalued feature sharing take place before valuation. Putting it all together, we conclude that the theoretical
system based on feature sharing is superior to the standard minimalist view of agreement in accounting for yet another intricate agreement pattern exhibited by inflected quantifiers, intensifiers and anaphors in Tatar.
References
Aydin, 2008 - Aydm Ö. Agreement with partitive quantifier in Turkish. 14th International Conference on Turkish Linguistics, 6-8 August, 2008, Ankara University, Manavgat, Turkey.
Baker, 2008 - Baker M. The syntax of agreement and concord. Cambridge, 2008.
Barany et al., 2019 - Prominent internal possessors. Barany A., O. Bond, I. Nikolaeva (eds.). Oxford, 2019.
Corbett, 2006 - Corbett, G. Agreement. Cambridge, 2006.
Comrie, 2003 - Comrie B. 'When agreement gets trigger-happy'. Transactions of the Philological Society. 2003. No. 101. Pp. 313-337.
Deal, 2017 - Deal A. External possession and possessor raising. The Wiley Blackwell Companion to Syntax. 2017.
Dechaine, Wiltschko, 2002 - Dechaine R.M., Wiltschko M. Decomposing pronouns. Linguistic Inquiry. 2002. Vol. 33. No. 3. Pp. 409-442.
Dechaine, Wiltschko, 2010 - Dechaine R.M., Wiltschko M. When and why can 1st and 2nd person pronouns be bound variables? Ms. 2010. URL: https://ling.auf.net/ lingbuzz/001079
Faller, Hastings, 2008 - Faller M., Hastings R. Cuzco Quechua quantifiers. Quantification: A Cross-Linguistic Perspective. L. Matthewson (ed.). North Holland Linguistic Series: Linguistic Variations. Vol. 64. Emerald Group Publishing, 2008.
Haing, Forker, 2018 - Haig G., Forker D. Agreement in grammar and discourse: A research overview. Linguistics. 2018. Vol. 56. No. 4. Pp. 715-734.
Ince, 2007 - Ince A. On default agreement in Turkish. WAFL. Harvard University, Boston, MA, 2007.
Jerro, 2013 - Jerro K. When quantifiers agree in person: Anomalous agreement in Bantu. Studies in Linguistic Sciences: Illinois Working Papers. 2013. Pp. 21-36.
Jerro, Wechsler, 2015 - Jerro K., Wechsler S. Person-marked quantifiers in Kinyarwanda. Agreement from a Diachronic Perspective. J. Fleischer, E. Rieken, P. Widmer (eds.). Trends in Linguistics. Studies and Monographs [TiLSM], 287. Mouton De Gruyter, 2015. Pp. 147-164.
Kratzer, 2009 - Kratzer A. Making a pronoun: Fake indexicals as windows into the properties of pronouns. Linguistic Inquiry. 2009. Vol. 40. No. 2. Pp. 187-237.
Lander, 2015 - Lander Yu. Dargwa contribution to the typology of agreement. Ms. IL RAS, 2015.
Lyutikova, 2017 - Lyutikova E. Agreement, case and licensing: Evidence from Tatar. Ural-Altaic studies. 2017. Vol. 25. No. 2. Pp. 25-45.
Lyutikova, Gerasimova, 2019 - Lyutikova E., Gerasimova A. Postpositional constructions in Tatar: Methodologies for measuring intralingual variation. Computational Linguistics and Intellectual Technologies. Papers from the Annual International Conference "Dialogue" (2019). 2019. Vol. 18. Pp. 412-435.
Matasovic, 2018 - Matasovic R. An areal typology of agreement systems. Cambridge, 2018.
I _
ISSN 2500-2953 Rhema. Рема. 2019. № 3
Muysken, 1989 - Muysken P. Predication chains: Case and argument status in Quechua and Turkish. Linguistic Inquiry. 1989. Vol. 20. No. 4. Pp. 627-645.
Muysken 2013 - Muysken, P. A note on inflected quantifiers in Quechua. Anuario del Seminario de Filología Vasca 'Julio de Urquijo'. 2013. Pp. 265-272.
Nevins, 2011 - Nevins A. Multiple agree with clitics: Person complementarity vs. omnivorous number. Natural Language and Linguistic Theory. 2011. Vol. 29. Pp. 939-971.
Nichols, 2018 - Nichols J. Agreement with overt and null arguments in Ingush. Linguistics. 2018. Vol. 56. No. 4. Pp. 845-863.
Nikolaeva et al., 2019 - Nikolaeva, I., Bárány A., Bond O. Towards a typology of prominent internal possessors. Prominent internal possessors. Bárány A., O. Bond, I. Nikolaeva (eds.). Oxford, 2019. Pp. 1-38.
Pesetsky, Torrego, 2007 - Pesetsky D., Torrego E. The Syntax of valuation and the interpretability of features. Phrasal and Clausal Architecture. Syntactic Derivation and interpretation. Amsterdam, 2007. Pp. 262-294.
Podobryaev, 2014 - Podobryaev A. Persons, imposters, and monsters. MIT diss., 2014.
Preminger, 2014 - Preminger O. Agreement and its failures. Cambridge, MA, 2014.
Say, 2019 - Say S. Prominent internal possessors in Bashkir. Prominent internal possessors. Bárány A., O. Bond, I. Nikolaeva (eds.). Oxford, 2019. Pp. 196-227.
Wurmbrand, 2017 - Wurmbrand S. Feature sharing or how I value my son. The Pesky Set: Papers for David Pesetsky. MIT WPL, 2017. Pp. 173-182.
Yadava et al., 2019 - Yadava Y., Bond O., Nikolaeva I., Ritchie S. The syntax of possessor prominence in Maithili. Prominent internal possessors. Bárány A., O. Bond, I. Nikolaeva (eds.). Oxford, 2019. Pp. 39-79.
Статья поступила в редакцию 09.08.2019 The article was received on 09.08.2019
Об авторах / About the authors
Лютикова Екатерина Анатольевна - доктор филологических наук, доцент; доцент кафедры теоретической и прикладной лингвистики филологического факультета, Московский государственный университет имени М.В. Ломоносова Lyutikova Ekaterina A. - Dr. Phil. Hab.; associate professor at the Department of Theoretical and Applied Linguistics of the Faculty of Philology, Lomonosov Moscow State University
ORCID: https://orcid.org/0000-0003-4439-0613 E-mail: lyutikova2008@gmail.com
72
J
Гращенков Павел Валерьевич - доктор филологических наук; доцент кафедры теоретической и прикладной лингвистики филологического факультета, Московский государственный университет имени М.В. Ломоносова; ведущий научный сотрудник отдела языков Азии и Африки, Институт востоковедения РАН
Grashchenkov Pavel V. - Dr. Phil. Hab.; associate professor at the Department of Theoretical and Applied Linguistics of the Faculty of Philology, Lomonosov Moscow State University; senior researcher at the Department of Asian and African Languages, Institute of Oriental Studies of RAS
ORCID: https://orcid.org/0000-0001-9754-2452
E-mail: pavel.gra@gmail.com
