Third Annual Meeting of the Slavic Linguistic Society (SLS 3) Columbus, Ohio, June 10-12, Parallel Sessions IV. Clitics. 4: Anton Zimmerling (Moscow). The Emergence of 2 nd position Clitics in Slavic and the Order of Cliticization

Anton Zimmerling (Moscow State University of Humanities, MGGU/Russian State University of Humanities, RGGU)

Acknowledgements Research is a part of the project The typology of free word order languages funded by the Russian Foundation for Humanities (grant RGNF a), whose support is gratefully acknowledged. The participation of the author in the SLS 3 conference (Columbus, Ohio, 10-12, June 2008) was funded the Russian Foundation for Humanities, travel grant RGNF з. I am grateful to the audience of the SLS 3 conference and personally to Henning Andersen, Wayles Brown and Kyongjoon Kwon for valuable suggestions and comments. All mistakes are my own.

The main claims (1) Proto-Slavonic had 2P clitics. Proto-Slavonic had clitic clusters in clausal second position. Modern Slavic languages could have inherited 2P clitics from Proto-Slavonic, but not clitic clusters. Different Slavic languages have different clusterization rules, because some parts of clitics clusters were filled after the disintegration of Proto-Slavonic.

The main claims (2) Clitic clusters in Early Slavonic languages merged clitic pronouns with particles and auxiliary clitics. Old Novgorod Russian is a consistent 2P language. The sequence of clitics within each part of the cluster corresponds to the order of cliticization. Old Church Slavonic and Old South Russian were mixed systems, where short pronouns behaved as verbal clitics of the Romance/Balcanic type. Cliticization of the present forms of BE-auxiliaries took place in the dialect period. The 3d. person forms of BE- auxiliaries were cliticized later than 1-2 person forms.

Zalizniak 2008 Самые древние энклитики (же, ли, бо) относятся к начальным рангам, самые молодые (мя и т.д. есмь и т.д. былъ) к конечным…. Полноударные слова, переходящие в ходе исследования в категорию энклитик, попадают… в конец уже имеющихся блоков энклитик. Тем самым система рангов оказывается своего рода «окаменевшей историей» формирования энклитик. Но указанный общий признак все же не носит абсолютного характера. В частности, относительно молодая энклитика бы внедрилась в цепочку энклитик левее энклитик ранга 6 (ми и т.д.), имеющих древнее происхождение [16, 47].

Zimmerling 2002 Все частицы в древненовгородской системе стоят левее энклитических местоимений, а все энклитические местоимения левее связочных форм связки "быть". …Так как краткие местоименные формы вин.п. м, т, с, ны, вы, на, ва стали клитиками значительно позже форм. дат.п. и сохранили в древнейших славянских памятниках следы акцентной самостоятельности…. вся правая часть таблицы может отражать процесс поэтапного превращения энклиноменальных словоформ в клитики. Первую группу энклиноменов составляют краткие местоименные формы в вин.п.….Вторую, более позднюю, группу энклиноменов составляют формы связки "быть", занимающие крайне правое место в цепочке, при этом в разряд клитик переходят не все члены парадигмы, а лишь презентные формы 1-2 л. (i ) XP..... ClDat ] XP..... ClDat ] ClAcc ] (ii) XP..... ClDat + ClAcc ] XP..... ClDat + ClAcc ] Cl Aux] [18, 70], cf. [19, ].

What is 2P: prosody or syntax? Serbian/Croatian/Bosnian a) Dobar=se čovjek lako prepozna A good man is easily recognizable (Split Constituent Rule, SCR). b) [Dobar čovjek]= se lako prepozna (Complex Constituent Rule, CCR). c) [Dobar čovjek] || lako=se prepozna (Barrier Rule, BR). d) prepozna || =se lako dobar čovjek (prosodic inversion).

Clusters and splitting of a cluster A. Clusters are both prosodic and syntactic units at once. Clusterization and ordering of clitics is largely triggered by prosody (Anderson 1995; Zaliznjak 1993). B. Clusters are prosodic, but not syntactic units. Clitics in a cluster can be heterogeneous, some of them being true 2P clitics, the other ones being capable of moving out of the cluster and attaching to adjacent verbal heads (Franks 2007). C. If contact positioning of clitics is not obligatory, a language has no clitic clusters (Bošković 2002).

2P typology: Wackernagels Law as a Macroparameter Languages, where 100% of clitics/clitic cluster take clausal 2 nd position, do not exist. Languages with 2P clitics are similar and share a number of non-trivial syntactic features [18], [19], [20], [21], [22]. All or most systems with verbal clitics of the Romance/Balcanic type originate from systems with 2P clitics. Phrasal clitics tend to restrict their syntactic domain. The evolution from systems with verbal clitics to systems with 2P-phrasal clitics is not attested, contrary to [26].

Reconstructing Proto-Slavonic clusterization The data from the early period (XI-XIII centuries) are incomplete. We are left with 2-3 well-attested syntactic systems – Old Novgorod Russian (ONR), Old Church Slavonic (OCL) and, probably, Old South Russian (OSR). It is necessary to go beyond the templates of clitics in a cluster and analyze the behaviour of clitics, especially the factors, that trigger late placement of clitics and splitting of clusters (=Barrier rules [17], [18]). A reconstruction must explain how either the ONR system can be derived from the OCL/OSR system, or the OCL/OSR system can be derived from the ONR systems. A reconstruction must explain how modern Slavic systems with 2P clitics (cf. Czech, Serbian/Croation/Bosnian, Bulgarian) shouln be are derived either from the ONR type or from the OCL/OSR type.

Two caveats The term OSR is not used here in a pure geographical sense. It refers to a body of bookish texts, which have been written down in Old South Russia, and not to a common ancestor of all present-day dialects, which have a South Russian/Ukrainian localization. The term ONR is used both in a geographical and textological sense. It refers to a body of vernacular texts, which have a North-West Russian localization. We lack vernacular OSR texts and cannot tell for sure, which features in the ONR clitic system are dialect- bound and which are not. Slavic clitics have different prosodic features, which are either observable or reconstructable, cf. [6], [11], [12], [13], [14]. 2P cliticization is about syntactic clitics, not prosodic clitics, since it takes effect on clitics with widely different prosodic features, cf. the ability of clitics to take stress over from adjacent enclinomena.

Migdalski contra Zalizniak Migdalski 2007 [8]: Proto-Slavonic was not a 2P language. It was close to OCL, where short pronouns are verbal clitics. 2P phenonema in Modern Slavic languages are a recent development. Zalizniak 2008 [16] basing on Jakobson Proto- Slavonic was a 2P language, but only a part of Early Slavic languages, incl. ONR conform to this principle. The system of OCL/OSR lithurgic/bookish texts is a reduction of an ONR-type system. Zalizniak 2008: The latest layer of Slavic clitics (pronouns, BE-auxiliaries) behave as 2P clitics in ONR, but as non-clitic words in OSR.

Old Novgorod Russian (1) Rigid order of 2P clitics. Long sequences (up to 5-6 elements in a cluster). Mixed syntactic/prosodic ordering: the particles come first, the pronouns precede auxiliary clitics. The order of clitics in each fragment of the cluster corresponds to their relative age and may reflect the chronology of cliticization. (Zimmerling 2002 [18]), (Zalizniak 2008 [16]). 3d person auxiliary clitics are absent from the system.

Old Novgorod Russian (2) Complex Constituent Rule is lacking. 2P clitics attach to the first phonological word.1 st word ~ 1 st constituent variation is absent from the system. Two non-clitic words can form a clitic host only if one of them is an enclinomen, cf. сего дне, оже далъ. Limited possibilities to form clitic words consisting of a proclitic + an enclitic. Cf. da=li, but not *ne=li. Barrier Rules can move clitic clusters to the right. (~ Skipping effect).

Old Novgorod Russian (3) Sentence categories, which act as Barriers, can split clitic clusters. Barriers can be blind: in this case they move the whole clitic cluster n steps to the right. Barriers can be sensitive to a particular type of clitics: in this case, splitting of a cluster takes place. Usually, splitting takes effect on reflexive clitic ся: ся moves to the right and ends up after the verb. Occasionally, clitic copying may occur. The left copy is in the 2P position, the right copy is in a verb-adjacent position.

Old South Russian Strong clitics = clitic particles. Usually are not moved by any Barriers. In terms of Migdalsky [8] strong Slavic clitics are 2P particles, which mark the Illocututionary Force/Clause type. Weak clitics = clitic pronouns. Are frequently moved by Barriers, which leads to cluster splitting. Accusative clitics move out from clausal 2P position more frequently than dative clitics. Auxiliary clitics move out from clausal 2P position even more frequently than accusative clitics.

Barrier Rules and late placement of clitics in ONR На молодогъ // далъ=есемь рубель. (BB) As for the malt, I have given a rouble, lit. On malt // gave=CL.Aux. rouble. А [дроужина моя] изнемогла=ся [1187] л. 227 об. And my army fource broke down=Refl. Того=ся каю I repent that, I am sorry for that. [Того всего] каю=ся [1151] л. 151 об. I am sorry for all that. *[Того всего] =ся каю. NOT ATTESTED. Fronted Vocatives, fronted Adresses, fronted Topicalized & Focalized constituents and fronted long initial constituents consisting of more than one accented word form act as Barriers in ONR.

Selective (Discriminating) Barriers ONR: A Barrier moves particle ли: А оу королева=еси мужа // слышал=ли о томъ честномъ крест ѣ 166об. Did(nt) you hear about this worthy cross from the kings man?, Lit. And from kings = CL.2Aux man // hear=CL.Q about that worthy cross. Macedonian: A Barrier moves reflexive clitic =se: Ti (1)=ne-Neg (2)=si-Aux2Sg (3)=mu-DatSgm (4) pomognal (5) ? Ne-Neg (1)=se-Refl (2) voznemiruvajte=Imp2Sg ! |[Ne]| (1) voznemiruvajte(2) se (3) [3, 103]. OSR: A Barries moves reflexive clitic =ся: б ѣ =бо оуже лице йего // изменило=ся 32в; * б ѣ =бо=ся оуже лице йего изменил. NOT ATTESTED

Clitic Copying: Birch Bark letters. SLOVO & ONR Chronicles Мн ѣ =ся // не может=ся (BB 124п) [И съ нимъ молодая м ѣ сяца, ОЛегъ и Святъславъ] // тьмою=ся поволокоста и в мор ѣ погрузиста (Slovo 103) Тамо=ся нал ѣ зе=ся моужъ родомъ Половчинъ именемь Лавор [1185] л. 226 об. И н ѣ льз ѣ =бы= ны=ся с ними тою р ѣ кою бить=ся полком [1148] л. 132 об. И пакы како=ся по нас йала=сь Руская земля вся [1151] л. 152 об. (NB: the second =сь has been scraped off in the manusript).

From 2P clitics to verbal clitics 1) И н ѣ льз ѣ =бы= ны=ся с ними тою р ѣ кою бить полком. (Clitic Cluster, base variant). 2) И н ѣ льз ѣ =бы= ны= i с ними тою р ѣ кою бить=ся i полком. (Discriminating Barrier, late placement of a reflexive clitic after the verbal head). 3) И н ѣ льз ѣ =бы= ны= i с ними тою р ѣ кою бить=ся i полком. (Clitic Copying, left copy acting as a 2P clitic, right copy acting as a verbal clitic).

Weak clitics in OSR: verbal clitics or non-clitic words? Descriptive statement. In OSR, weak clitics (clitic pronouns and 1-2 BE-auxiliaries) are not placed at random, but show up after verbal heads. They can also sporadically occur in the 2P position. Reconstruction. If one assumes that weak clitics were treated as non-clitic words in OSR, the distribution in ONR and Modern Slavic languages is a mystery. Since these clitics represent the latest layer of cliticization, one has to assume that in a group of Slavic dialects they developed as non-clitic words, while in the other group of Slavic dialects these most recent clitics suddenly lost their clitic status during a short period of time. Conclusion. Zalizniaks explanation can be movitaved only by a wish to explain the distribution of 2P clitics by pure prosodic factors. This approach is not tenable.

Conclusions The ONR system of clitic placement could not be derived from the OSR/OCL system, but the latter could be derived from a ONR-like system. Modern Slavic systems with 2P clitics originate from a ONR- like system. Zalizniaks claim that weak clitics (dat., acc. pronouns, BE- auxiliaries) behaved as non-clitic words in OSR/OCL is not tenable. In the OSR/OCR pronouns and auxiliaries behaved as verbal clitics. This fact implies that after the disintegration of Proto- Slavonic clitics belonging to the latest cliticization layer, did not have 2P-properties any longer. The BE-auxiliries could still cliticize and develop in the direction of 2P clitics in a group of conservative Slavic dialects (ONR, Slovene, Carpatian Ukrainian).

References (1) [1] Anderson, Stephen P Toward an Optimal Account of Second-Position Phenomena. // Optimality Theory: Phonology, Syntax, and Acquisition /Dekkers J., F. van der Leeuw & J. van de Weijer (eds.). Oxford, OUP, [2] Bošković, Željko Clitics as nonbranching elements and the linear correspondence axiom // Linguistic Inquiry 33.2: [3] EuroClitics Clitics in the languages of Europe. Eurotyp Ed. by Henk van Riemsdijk. Mouton de Gruyter. Berlin- New York. [4] Franks, Steven South Slavic Clitics: Balcanization or Unification // SLS 2 conference, Berlin [5] Franks, Steven & Tracy Holloway King A handbook of Slavic Clitics. Oxford: OUP.

References (2) [6] Jakobson, Roman Les enclitique slaves // Selected Writings, vol. II. Word and language. The Hague - Paris: [7] Marušic, Frank Positioning Slovenian clitics // SLS 2 conference, Berlin [8] Migdalski, Krszystof On the emergence of the second position cliticization in Slavic // Formal Description of Slavic Languages FDSL-7. Leipzig, [9] Mihaljevic, Milan The Interaction of li and Negation in Croatian Church Slavonic // Formale Slavistik / U.Junghanns, G.Zybatow (hrsg.), Frankfurt- am-Main: Vervuert,

References (3) [10] Radanović-Kocić, Vesna The Placement of Serbo- Croatian Clitics: a prosodic approach // Approaching Second/ A.Halpern, A. Zwicky (eds.). CSLI Publications. Stanford, California, ] Dybo, Wladimir A Дыбо В.А. О фразовых модификациях ударения в праславянском. (On Phrasal Modifications of Stress in Proto-Slavonic). // Советское cлавяноведение, 6, [12] Dybo, Wladimir A Дыбо В.А. Закон Васильева- Долобко в древнерусском языке на материале Чудовского Завета. (Wassiliew-Dolobkos Law in Old Russian on the basis of the Chudov Testament). // International journal of linguistics and poetics, vol. 18/ 1, [13] Dybo, Wladimir A Дыбо В.А. Славянская акцентология. Опыт реконструкции системы акцентных парадигм в праславянском. (Slavic Accentology; a Reconstruction of Accent Paradigms in Proto-Slavonic). Moscow.

References (4) [14] Zaliznjak, Andrej A – Зализняк А.А. От праславянской акцентуации к русской. (From Proto- Slavonic Accentuation to Modern Russian Accentuation). Moscow. [15] Zaliznjak, Andrej A – Зализняк А.А. К изучению языка берестяных грамот. (The language of Birch Bark letters). // В.Л. Янин, А.А. Зализняк. Новгородские грамоты на бересте; из раскопок гг. (Novgorod Birch Bark Letters from excavations ). Moscow. [16] Zaliznjak, Andrej A – Зализняк А.А. Древнерусские энклитики. (Old Russian Enclitics). Moscow : Yazyki slavjanskoj kultury.

References (5) [17] Zimmerling, Anton W – Циммерлинг А.В. Порядок слов и синтаксические позиции. (Word Order and Syntactic Positions). // Dialogue Proceedings of the international seminar on computer linguistics and its applications. Moscow, [18] Zimmerling, Anton W Циммерлинг А.В. Типологический синтаксис скандинавских языков. (Typological Scandinavian Syntax). Moscow : Yazyki slavjanskoj kultury. [19] Zimmerling, Anton W – Порядок слов в славянских, германских и романских языках. (Word order in Slavic, Germanic and Romance languages). // Славяно-германские исследования, том 3 (Slavo- Germanic studies, vol.3). Sankt-Petersburg : Aletheia,

References (6) [20] Zwicky A.M. On Clitics. Bloomington. Indiana University Linguistic Club [21] Aikhenvald, Alexandra Y Typological parameters for the study of clitics, with special reference to Tariana.// Word: a cross-linguistic typology. / R.M.W. Dixon and A. Y. Aikhenvald (eds.). Cambridge: Cambridge UP. [22] Billings, Lorren & Abigail Konopasky The role of morphology in ordering verb-adjacent clitics: from syntax to prosody in Bulgarian and Tagalog. [23] Halpern, Aaron Lars Topics in the Placement and Morphology of Clitics. Stanford, CA. Doctoral dissertation, Stanford University.

References (7) [24] Klavans, Judith The independence of syntax and phonology in cliticization. Language 61, [25] Jelinek, Eloise Predicate Raising in Lummi, Straits Salish. // The syntax of Verb Initial languages. / Andrew Carnie and Eithne Guilfoyle, eds [26] McConvell, Peter The functions of Split- Wackernagel Clitic Systems: Pronominal Clitics in the Ngumpin languages (Pama-Nyungan family, Northern Australia). // Approaching Second/ A.Halpern, A. Zwicky (eds.). CSLI Publications. Stanford, California,

References (8) [27] Nash, David G Topics in Warlpiri Grammar. (Outstanding dissertations in linguistics). MIT. [28] Nespor, Marina, and Irene Vogel Prosodic phonology. Dordrecht: Foris. [29] Roberts, Taylor The optimal second position in Pashto. MIT. [30] Tsunoda, Tasaku The Djaru language of Kimberly, Western Australia. (Pacific Linguistics, Series B - No. 78). The Australian National University.

References (9) [31] Wackernagel, Jakob Über ein Gesetz der indogermanischen Wortstellung. // Indogermanische Forschungen 1: = Wackernagel Jacob. Kleine Schriften. Bd. I, Basel, 1953, [32] Werle, Adam Sonority-determined clitic order. // WECOL MIT, Amherst. [33] Browne, Wayles Word Order in Burgenland Croatian: clitics. Talk at the Third Southeast European Studies Association Conference, April 26-28, 2007, Columbus = Talks Commemorating the 10 th Anniversary of the Naylor Professorship.