The present volume offers a selection of papers on current issues in Slavic languages. It takes stock of the past 20 yea
223 54 2MB
English Pages 394 Year 2017
Table of contents :
Contents
Introduction
On the interaction of P-stranding and Sluicing in Bulgarian
A Pilot Corpus Study of Non-verb Experiencer Predicates in Russian
Decomposing Prepositional Cases in Russian and Polish
Semantic and Morphosyntactic Features of Verbal Prefixes: A Case Study of the Russian Prefix pere- ‘over’
Zur Darstellung lexikalischer Eigenschaften belarussischer Prädikativa in einsprachigerklärenden Wörterbüchern
Parametric Variation of Slavic Accusative Impersonals
Russian namek and English hint as Ordinary Language Hyponyms of Grice’s Term Implicature
Ukrainische nominale Flexion – zur automatischen Generierung der Substantivformen in einem ukrainisch-deutschen Wörterbuch
Kommunikative Sprachmittlung: Jugendliche Herkunftssprecher des Russischen und ihre Eltern im Vergleich
Polish Perfective Generics
Prepositions as Category-neutral Roots
Branching Onsets in Old Czech
Aspects of Conativity in Russian: Towards a Linguistics of Attempt and Success
Interclausal Feature Relations with Subjunctives
Split Quantifier Phrases and Genitive of Negation in Russian
Aspects of Slavic Linguistics
Language, Context, and Cognition
Edited by Anita Steube Editorial Board Kai Alter, Newcastle and Oxford Ulrike Demske, Potsdam Ljudmila Geist, Stuttgart Rosemarie Lühr, Berlin Thomas Pechmann, Saarbrücken Richard Wiese, Marburg
Volume 16
Aspects of Slavic Linguistics Formal Grammar, Lexicon and Communication Edited by Olav Mueller-Reichau and Marcel Guhl
ISBN 978-3-11-051582-4 e-ISBN (PDF) 978-3-11-051787-3 e-ISBN (EPUB) 978-3-11-051585-5 ISSN 1866-8313 Library of Congress Cataloging-in-Publication Data A CIP catalog record for this book has been applied for at the Library of Congress. Bibliografic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutschen Nationalbibliografie; detailed bibliografic data are available on the Internet at http://dnb.dnb.de. © 2017 Walter de Gruyter GmbH, Berlin/Boston Printing and binding: CPI books GmbH, Leck ♾ Printed on acid-free paper Printed in Germany www.degruyter.com
Contents Introduction | vii Klaus Abels (London) On the Interaction of P-stranding and Sluicing in Bulgarian | 1 Loren A. Billings (Bishkek) A Pilot Corpus Study of Non-verb Experiencer Predicates in Russian | 29 Petr Biskup (Leipzig) Decomposing Prepositional Cases in Russian and Polish | 50 Kristina Gehring (Kaliningrad) Semantic and Morphosyntactic Features of Verbal Prefixes: A Case Study of the Russian Prefix pere- ‘over’ | 69 Claudia Hurtig (Leipzig) Zur Darstellung lexikalischer Eigenschaften belarussischer Prädikativa in einsprachig-erklärenden Wörterbüchern | 106 Uwe Junghanns (Göttingen), Denisa Lenertová (Berlin) and Dorothee Fehrmann (Göttingen) Parametric Variation of Slavic Accusative Impersonals | 140 Irina M. Kobozeva (Moskva) Russian namek and English hint as Ordinary Language Hyponyms of Grice’s Term Implicature | 166 Kersten Krüger (Leipzig) Ukrainische nominale Flexion – zur automatischen Generierung der Substantivformen in einem ukrainisch-deutschen Wörterbuch | 184 Grit Mehlhorn und Maria Yastrebova (Leipzig) Kommunikative Sprachmittlung: Jugendliche Herkunftssprecher des Russischen und ihre Eltern im Vergleich | 212 Olav Mueller-Reichau (Leipzig) Polish Perfective Generics | 235
vi | Contents
Hagen Pitsch (Göttingen) Prepositions as Category-neutral Roots | 259 Tobias Scheer (Nice) and Markéta Ziková (Brno) Branching Onsets in Old Czech | 285 Barbara Sonnenhauser (Zürich) Aspects of Conativity in Russian: Towards a Linguistics of Attempt and Success | 310 Luka Szucsich (Berlin) Interclausal Feature Relations with Subjunctives | 333 Yakov G. Testelets and Ekaterina A. Lyutikova (Moskva) Split Quantifier Phrases and Genitive of Negation in Russian | 356
Introduction This volume gives an overview of the last twenty years of Slavonic Studies at the University of Leipzig and calls special attention to the successful periods in the history of Slavic linguistics when it worked on the same methodological basis in close cooperation with general linguistics. While the methodological basis in the 19th and in the beginning 20th century was Indo-European historical comparison it is theoretical and formal linguistics now.
Historical background When the modern Indo-European languages were first represented by professors at the Philosophical Faculty of the University of Leipzig, German Studies became the first philological discipline. It was inaugurated in 1848 but Slavonic Studies followed already in 1876. With August Leskien, the University of Leipzig had the first professor of Slavic Languages in Germany. He and his Neo-Grammarian followers and students such as the first professor of Indo-European Languages in Leipzig, Karl Brugmann (professor in Leipzig since 1887), became world famous as the representatives of the “Leipzig School of Linguistics”. They created the theoretical ground for Historical Comparative Linguistics on an exact empirical basis and were the supervisors of the theses of Ferdinand de Saussure and Baudouin de Courtenay. The Russian disciples of the latter became founders of the Prague Linguistic Circle that directed the attention to synchronic linguistics and the system of language like de Saussure’s theory and paved the way for structural linguistics. It became the starting point for linguists in Leipzig after the Second World War. Since 1889, August Leskien, Karl Brugmann and a professor of Sanskrit were the three colleagues in the Indo-European Department in Leipzig. In 1922, Slavonic and Indo-European Studies became independent parts of the enlarged socalled “United Linguistic Department” that existed until 1946 when the Department of Slavic Studies became more relevant for the education of teachers and interpreters/translators of Slavic languages and was turned into a department of its own. In 1959, the Department of Indo-European was turned into the Department of Linguistics, and shortly afterwards, the slavist Rudolf Růžička became its director. As a linguist, he was interested in theory of grammar, and as a slavist, he used Slavic languages as the empirical basis for his formal generalizations. In this way, he became one of the first generative slavists worldwide. His main fields of expertise were syntax, semantics, and the comparison of Slavic languages.
DOI 10.1515/9783110517873-001
viii | Introduction
Gerhild Zybatow was an assistant of Rudolf Růžička and got all her higher academic qualifications in his working group. It is self-evident that she – as professor of East Slavic Languages within the Department of Slavonic Studies since 1995 – has always remained in close contact with the Department of Linguistics of her university as far as linguistic methodology and longlasting research projects are concerned. But much more has been done for Slavonic Studies in general in her sub-department: During the period of structuralism, successful groups of formal linguists existed in many Slavic countries as well as in the German Democratic Republic. But the political authorities in these countries supported Marxist-Leninist Linguistics (whatever this term may ever have meant) with different vehemence so that many linguists had to give up their methodology or even leave their countries. In the end, formal linguistics was no longer a strong discipline among slavists. Therefore, it has been an industrious and at the same time courageous work to start the “European Conferences on Formal Description of Slavic Languages (FDSL)” in Leipzig in 1995. This conference proving to be a great success, the organizers decided to repeat such conferences in Leipzig and Potsdam every second year. In the mean time, eleven conferences have been held and the papers have been published. What is more, beginning with conference Nr. 6, so-called halftime conferences have been organized in Slavic-speaking countries. This way, Slavonic Studies in Europe have regained a field of research underestimated in the discipline for decades.
The most important works of Gerhild Zybatow Gerhild Zybatow is one of the most prominent representatives of Generative Grammar in the field of Slavic linguistics in Germany. She has been concerned with using Slavic data for the development of linguistic theory, pursuing a cognitive approach to the study of Slavic languages. Issues she has been working on include theory of grammar (morpho/syntax, semantics, lexicon theory), psycholinguistics, comparison of languages, contrastive linguistics, and translation theory. More recently her research has focused on the changes of East Slavic from the perspectives of text theory, sociolinguistics, and psycholinguistics. She has been interested in the emergence of Ukrainian and Belarusian as national languages of two independent states as well as in the language of Russians living in Germany. As mentioned above, Gerhild Zybatow’s career is closely connected with theoretical linguistics at the University of Leipzig. She completed her dissertation project under the supervision of Rudolf Růžička. Anita Steube acted as supervisor
Introduction | ix
of her habilitation thesis. This background set her for a strictly formal approach to the study of language with foci on syntax and semantics. In 1978 she submitted her doctoral dissertation (cf. Zybatow 1978) titled “Syntaktische und semantische Eigenschaften kognitiver Verben des modernen Russischen” [Syntactic and semantic properties of cognitive verbs of modern Russian]. Right at the beginning she presents her credo (cf. Zybatow 1978: 1), translated): “The language system is one of those scientific subjects that cannot be investigated directly. It can, however, be objectivized [through linguistic theorizing].” Included in the investigation were selected classes of cognitive verbs of Russian (thinking, understanding, knowing). In her dissertation, she developed the view that semantic features constituting verb classes determine particular syntactic structures and rule application. This accounts for related syntactic surface structures of a certain type and helps to explain the interplay of syntax and semantics. She distinguished between mental achievements [+inchoative], e.g. osoznat’ ‘realize’, and mental states [+durative], e.g. sčitat’ ‘be of an opinion’, each subdivided according to semantic and syntactic characteristics. One whole chapter was devoted to the analysis of Raising with cognitive verbs. Three major publications on verbs appeared in the following years – Zybatow (1983): Syntaktische und semantische Eigenschaften der Komplementsätze kognitiver Verben des modernen Russischen [Syntactic and semantic properties of the complement clauses of cognitive verbs in modern Russian], Zybatow (1987): Epistemische und intentionale Einstellungsverben im modernen Russischen [Russian epistemic and intentional propositional attitude verbs], and Zybatow (1988): Über Bewertungen epistemischer Einstellungen [On evaluations of epistemic attitudes]. This period marked the transition from the topic of her dissertation to the topic of the habilitation thesis. Her habilitation thesis (Zybatow 1989) was titled “Untersuchungen zu der semantischen Repräsentation ausgewählter russischer Einstellungsprädikate und ihrer Interpretation im Satz“ [Investigations into the semantic representation of selected Russian predicates of attitude and their interpretation within the clause]. This work was focused on the semantics of epistemic, intentional, emotive, and evaluative attitudinal predicates of modern Russian using the Restricted Extended Standard Theory of Generative grammar. Issues dealt with: Syntactic analysis of clauses with attitudinal predicates, formalization of meaning structures of particular predicates, interpretation of the predicates within the clauses they appear in, and aspects of the pragmatics of attitudinal expressions (e.g., description: sožalet’ ‘regret’ vs. signaling: k sožaleniju ‘unfortunately’, etc.). Zybatow (1990): Die deutschen Modalverben und ihre Wiedergabe im Russischen (German modal verbs and their Russian equivalents] proceeds from a
x | Introduction
survey of modal verbs in German (dürfen ‘may’, können ‚can‘, mögen ‚may’, müssen ‚must‘, nicht brauchen ,need not‘, sollen ‚shall‘, wollen ‚want‘) and a characterization of types of modality (deontic, alethic, epistemic). Russian has various equivalents for the German verbs in their different uses. These possibilities are discussed and illustrated with authentic examples. Zybatow (1994): Infinitive, kleine und große Pros und (in)kohärente Konstruktionen im Russischen [Infinitives, small pros, big PROs, and (in)coherent constructions in Russian] discusses the traditional approach to infinitives as instantiations of sentence elements like subject, object, adverbials and opposes it with a structural syntax account suggesting conditions imposed by matrix predicates (subcategorization, selection). Around 1995 Gerhild Zybatow became interested in phenomena accompanying information structuring. The relative freedom of surface structure in Slavic languages combined with the claim that there is an articulated underlying hierarchical syntax seemed to constitute a hard challenge. The investigation yielded a number of single and co-authored papers – Zybatow (1997): Determinanty informacionnoj struktury [Determinants of information structure]; Junghanns and Zybatow (1997): Syntax and information structure of Russian clauses; Zybatow and Junghanns (1998): Topics im Russischen [Topics in Russian]; Zybatow (1999): Informationsstruktur im Russischen [Information structure in Russian]; Zybatow and Mehlhorn (2000): Experimental Evidence for Focus Structure in Russian; Junghanns and Zybatow (2009): Grammatik und Informationsstruktur [Grammar and Information Structure]. Gerhild Zybatow and her co-authors claim that communicative weight is reflected in syntax by features that impose restrictions on surface placement of clausal constructions and determine prosodic realization and conceptual interpretation. Zybatow and Mueller-Reichau (2011): Typ oder Token? Zum Auftreten von est‘ in russischen Possessivsätzen [Type or Token? On the occurrence of est‘ in Russian sentences with possessive meaning] is an investigation into the semantic composition of Russian possessive constructions. Syntactic evidence is provided to show that the explicit appearance of the existential verb est’ ‘be’ signals that the possessum merely denotes a type introducing a property in the discourse. If the predicate est’ is present, the nominal does not have existential force on its own but rather functions as a type term, whereas if est’ is absent, the nominal is interpreted as having existential force and, thus, functions as a token term. Biskup and Zybatow (2015): Verbal prefixation in Slavic: a minimalist approach. The authors discuss effects of prefixation (e.g., perfectivity and telicity, case assignment, meaning and argument structure of the verb). A morphosyntac-
Introduction | xi
tic minimalist approach is suggested according to which prefixes are incorporated prepositions projecting a result predicate in the complement position of the verb. The syntactic and semantic properties of the incorporated preposition serve to account for the various derivational effects. Gerhild Zybatow has co-edited an impressive number of volumes. Steube & Zybatow (1994): Zur Satzwertigkeit von Infinitiven und Small Clauses [On the clausal nature of infinitives and small clauses] contains the papers of a conference that took place in Leipzig in 1991 and was attended by linguists from both parts of Germany – the first joint conference after the German reunification. Other volumes bear witness of the contribution to Slavic linguistics by the papers presented at FDSL conferences 1, 3, 5, 7, 7.5, 10 (see the volumes edited between 1997 and 2015 in References ii. below). DGfS – the German Linguistic Society – elected Gerhild Zybatow second chairman for the period 1993–1995. She was involved in the Graduate School “Universalität und Diversität: Sprachliche Strukturen und Prozesse“ [Universality and Diversity: Linguistic Structures and Processes] in Leipzig as well as in various funded research projects. She has been supervisor for numerous doctoral dissertations and habilitations and has been elected dean and vice-dean of the Philological Faculty of the University of Leipzig.
References i. Selected works by Gerhild Zybatow (single or joint authorship) in chronological order Zybatow, Gerhild (1978): Syntaktische und semantische Eigenschaften kognitiver Verben des modernen Russischen. Doctoral dissertation. University of Leipzig. Zybatow, Gerhild (1983): Syntaktische und semantische Eigenschaften der Komplementsätze kognitiver Verben des modernen Russischen. In: Rudolf Růžička, Wolfgang Motsch (eds): Untersuchungen zur Semantik (= studia grammatica 22). Berlin: Akademie-Verlag, 169– 200. Zybatow, Gerhild (1987): Epistemische und intentionale Einstellungsverben im Russischen. Zeitschrift für Phonetik, Sprachwissenschaft und Kommununikationsforschung 40.5, 637– 651. Zybatow, Gerhild (1988): Über Bewertungen epistemischer Einstellungen. In: Bierwisch, Manfred (ed.): Syntax, Semantik und Lexikon: Rudolf Růžička zum 65. Geburtstag (= studia grammatica 29). Berlin: Akademie-Verlag, 313–322. Zybatow, Gerhild (1989): Untersuchungen zu der semantischen Repräsentation ausgewählter russischer Einstellungsprädikate und ihrer Interpretationen im Satz. Habilitation thesis. Universität Leipzig.
xii | Introduction
Zybatow, Gerhild (1990): Die deutschen Modalverben und ihre Wiedergabe im Russischen. Universität Leipzig. Sektion Theoretische und angewandte Sprachwissenschaft. Zybatow, Gerhild (1994): Infinitive, kleine und große Pros und (in)kohärente Konstruktionen. In: Steube, Anita and Gerhild Zybatow (eds), 141–154. Zybatow, Gerhild (1997): Determinanty informacionnoj struktury [Determinants of information structure]. In: Junghanns, Uwe and Gerhild Zybatow (eds), 359–370. Junghanns, Uwe and Gerhild Zybatow (1997): Syntax and Information Structure of Russian Clauses. In: Browne, Wayles, Ewa Dornisch, Natasha Kondrashova, and Draga Zec (eds): Annual Workshop on Formal Approaches to Slavic Linguistics. The Cornell Meeting 1995 (= Michigan Slavic Materials 39). Ann Arbor: Michigan Slavic Publications, 289–319. Zybatow, Gerhild and Uwe Junghanns (1998); Topiks im Russischen (= Sprache und Pragmatik 47), Lund: Germanistisches Institut der Universität Lund. Zybatow, Gerhild (1999): Informationsstruktur im Russischen. In: Doherty, Monika (ed.): Sprachspezifische Aspekte der Informationsverteilung (= studia grammatica 47). Berlin: Akademie-Verlag, 61–89. Zybatow, Gerhild and Grit Mehlhorn (2000): Experimental Evidence for Focus Structure in Russian. In: King, Tracy H. and Irina Sekerina (eds): Annual Workshop on Formal Approaches to Slavic Linguistics. The Philadelphia Meeting 1999 (= Michigan Slavic Materials 45): Ann Arbor: Michigan Slavic Publications, 414–434. Junghanns, Uwe and Gerhild Zybatow (2009): Grammatik und Informationsstruktur. In: Kempgen, Sebastian, Peter Kosta, Tilman Berger, and Karl Gutschmidt (eds): The Slavic Languages. An International Handbook of their Structure, their History and their Investigation (= Handbook of Linguistics and Communication Science 32.1): Berlin/New York: Mouton de Gruyter, 684–707. Zybatow, Gerhild and Olav Mueller-Reichau (2011): Typ oder Token? Zum Auftreten von est‘ in russischen Possessivsätzen. Zeitschrift für Slawistik 56.3, 305–318. Biskup, Petr and Gerhild Zybatow (2015): Verbal prefixation in Slavic: a minimalist approach. In: Müller, Peter O., Ingeborg Ohnheiser, Susan Olsen, and Franz Rainer (eds): Word-Formation. An International Handbook of the Languages of Europe, vol. 2 (= Handbook of Linguistics and Communication Science 40.2). Berlin/Boston; de Gruyter Mouton, 1492–1515.
ii. Edited volumes Steube, Anita and Gerhild Zybatow (eds) (1994): Zur Satzwertigkeit von Infinitiven und Small Clauses (= Linguistische Arbeiten 315). Tübingen: Niemeyer. Junghanns, Uwe and Gerhild Zybatow (eds) (1997): Formale Slavistik (= Leipziger Schriften zur Kultur-, Literatur-, Sprach- und Übersetzungswissenschaft 7). Frankfurt/Main: Vervuert Verlag. Zybatow, Gerhild, Uwe Junghanns, Grit Mehlhorn, and Luka Szucsich (eds) (2000): 3. Europäische Konferenz „Formale Beschreibung slavischer Sprachen“, Leipzig 1999 (= Linguistische Arbeitsberichte 75). Leipzig: Institut für Linguistik der Universität Leipzig. Zybatow, Gerhild, Uwe Junghanns, Grit Mehlhorn, and Luka Szucsich (eds) (2001): Current Issues in Formal Slavic Linguistics (= Linguistik International 5). Frankfurt/Main: Lang.
Introduction | xiii
Zybatow, Gerhild, Luka Szucsich, Uwe Junghanns, and Roland Meyer (eds) (2008): Formal Description of Slavic Languages. The Fifth Conference, Leipzig 2003 (= Linguistik International 20). Frankfurt/Main: Lang. Zybatow, Gerhild, Denisa Lenertová, Uwe Junghanns, and Petr Biskup (eds) (2009): Studies in Formal Slavic Phonology, Syntax, Semantics and Information Structure. Proceedings of FDSL 7, Leipzig 2007. Frankfurt/Main: Lang. Zybatow, Gerhild, Philip Dudchuk, Serge Minor, Ekaterina Pshekhotskaya (eds) (2010): Formal Studies in Slavic Linguistics. Proceedings of Formal Description of Slavic Languages 7.5 (= Linguistik International 25). Frankfurt/Main: Lang. Zybatow, Gerhild, Petr Biskup, Marcel Guhl, Claudia Hurtig, Olav Mueller-Reichau, and Maria Yastrebova (eds) (2015): Slavic Grammar from a Formal Perspective. The tenth Anniversary FDSL Conference, Leipzig 2013 (= Linguistik International 35). Frankfurt/Main: Lang.
Papers of the Leipzig circle of Slavic linguistics The authors of this volume are former or present colleagues in the East-Slavic Sub-department, former or present graduates and academic friends of Gerhild Zybatow. They mostly write on problems of formal grammar and the lexicon as well as on fields of education in the students’ program. West- and South-Slavic languages are included as well.
Syntactic problems Klaus Abels (London) presents an improved theory of Sluicing. Sluicing is a case of underspecification i.e. a derivational process for incomplete wh-questions in which only the wh-phrase has an overt representation. The theory of Fit applied demands semantic identity between antecedent and sluice but allows syntactic mismatches. This compensates the shortcomings of older theories. The availability of a well-formed recoverable pre-sluice even explains the Bulgarian sluices of wh-phrases with stranded prepositions which – however – require further tests because of the regional and individual variations noticed. Petr Biskup (Leipzig) is concerned with prepositional cases on a par with structural cases as a reflection of the operation Agree between ɸ-features and Tense-features. The type of the assigned prepositional case is determined by semantic properties of particular heads of the decomposed prepositions. Specifically, there is a correspondence between semantic properties of particular heads and their syntactic features. Syntactic features incorporated into the case assigning head are copied on the prepositional complement by Agree at the level of PF. These features are spelled out as a case by means of a specific vocabulary insertion rule.
xiv | Introduction
Kristina Gehring (Kaliningrad) writes about Russian prefixation as a part of a syntactic derivational process. The Prefix-Split-Theory devides the semantically and syntactically heterogeneous class of prefixes into two or three subclasses: the lexical or internal prefixes are generated within the verbal phrase, the superlexical prefixes higher up in the syntactic tree. A third group under discussion are the intermediate prefixes. Gehring gives a detailed analysis of most semantic subgroups of the prefix pere- ‘over’ and decides on their positions in derivations. Uwe Junghanns, Dorothee Fehrmann (Göttingen), Denisa Lenertová (Berlin) apply a Chomskian syntax supplemented by a two-level semantics to Slavic Accusative Impersonals (AIs). The outcome of the paper is: 1. A survey on the crossSlavic variation with respect to the availability of AI-types; 2. the statement that morphologically marked AIs and Adversative AIs are different constructions. 3.1 Adversative AIs can be described with an external pronoun (evidence: overt FORCE-pronoun to in Cz, control into gerunds and infinitivals). 3.2 the semantic realization of the indefinite pronoun is twofold, either as the existentially quantified FORCE-pronoun 1 for Slovenian, Czech, Upper Sorbian (just like Czech to, but without lexical realization), or as the FORCE-pronoun 2 realized as variable x that is not existentially bound at SF and, therefore, can be interpreted via co-indexation at CS (for Russian, Belarusian, Ukrainian, Polish, Slovak). Assuming two pronouns accounts for the possibility vs. exclusion of overt FORCE-phrases with Adversity AIs (cross-Slavic variation). Testelec and Lyutikova’s (Moskva) paper successfully uncovers the dissimilarities between Russian non-split and split quantifier constructions. In the literature for a long time, the latter were treated as mere topicalized versions of nonsplit QPs. Chapter three, however, shoes that many SplQPs do not have a nonsplit counterpart. On the other hand, the empirical tests in chapter four prove that SplQPs and constructions with genitive of negation (GenNeg) are very similar: 1. Both constructions are available only in the context of the two structural cases nominative and accusative; 2. both show the same behavior of the subject of transitive and intransitive verbs and of the existential verb byt’. The authors explain these striking similarities by a common syntactic structure for SplQP and GenNeg: GenNeg is an instance of genitive of quantity with the non-overt quantifier Q that belongs to the same class as the polarity items ni, nikakoj, etc. requiring sentential negation. Because of its incompatibility with the DP, Q is placed in VP from where it can license case and allow for several instances of genitive to occur in the same clause, like e.g. Ni odnogo fonarja ne osveščalo ulicy. – No lantern lightened the street.
Introduction | xv
Luka Szucsich (Berlin) provides syntactic derivations and different feature makeups for indicative and subjunctive complement clauses of volitional/directive matrix verbs in North Slavic languages in comparison to the complement clauses e.g. of verbs of saying. In contrast with earlier explanations by domain extension or cross-clausal head movement analyses, the author works with enlarged selectional requirements that the volitional/directive matrix verbs impose on the temporal features in complement clauses, so that co-indexation can occur when the relevant features are in maximally local relation.
Semantic and pragmatic problems Based on Gricean theory, Irina M. Kobozeva (Moskva) writes on the hyponyms hint – to hint (in English) and namek – namekat’ (in Russian) which she uses as representatives of the families of nouns and verbs designing ‘conversational implicature’. The paper defines ‘hinting’, lists the different types of ‘hinting’ with illustrations as well as with the semantic differences of the nouns and verbs in their respective word families. Olav Mueller-Reichau (Konstanz/Leipzig) addresses the issue of Polish generics with perfective verbs. In discussion with the results of Greenberg, Krifka, and Klimek-Jankowska, he proposes to analyze perfective morphology in generics as an indication that a statement is being made about the language being used – in contrast to imperfective generics, which are habitual statements about the world being talked about. It is shown that this kind of approach yields correct results even for Polish perfective generics with proper name subjects, once the obligatory retrieval of background knowledge with these is properly taken into consideration. Barbara Sonnenhauser (Zürich) approaches conativity in Russian from a broad perspective considering at least its philosophical, modal, and grammatical aspects which include – beside Slavic verbal aspect itself – the meaning of the respective verb together with its event-semantic specifications, the verbal arguments, negation as well as grammatical and pragmatic features of success or ineffectiveness.
Lexical categories Loren Billings (Seoul/Bishkek) presents a case study on three classes of non-verbal experiencer taking predicates that – like psychverbs – have their experiencers
xvi | Introduction
in Dative case. They are: pora – ‘[X is] time’ (an etymological noun); žal /žalko – ‘[X is (a)] pity/pitiful’ (an etymological noun/adjective); vid#n – [X is] ‘visible’, slyš#n – [X is] ‘audible’, zamet#n – [X is] ‘noticeable’ (morphologically adjectival predicates). The paper is based on a corpus. In all occurrences of the respective sentences, the referentiality of the nominal expressions, their intonation and the corresponding word-order of the sentence occurrences are documented and correlated in order to derive the information structural behavior of Russian sentences with non-verbal experience taking predicates, hitherto an untouched field of grammar. Supported by a thorough analysis of sentences containing the Belarusian predicatives trèba and patrèbna, Claudia Hurtig (Leipzig) shows theoretical and practical possibilities for Belarusian dictionaries to adapt to modern lexicography in general and to modern word-classes in particular. She analyses the DPs (mostly in Dative case) and the complements of the predicatives as well as the semantic characteristics of the partial synonyms. On this basis, she can propose a new lemmatization of these two Belarusian predicatives as an example. Hagen Pitsch (Göttingen) reconstructs the classification of lexical categories since Chomsky, Wunderlich, and others: Category P is dispensed with. As opposed to N, V and Adj, P becomes a category-neutral head. No pains must be taken any longer to differentiate its lexical and functional properties because PPs adjoin to whatever they modify in terms of sortal adequacy and semantic type. The solution given naturally includes the Slavic ‘semantic cases’: Both are modifiers, both are defined by their Argument Structure (AS), Predicate-ArgumentStructure (PAS), and their missing Grammatical Features (GF). The only difference between them is that overt prepositions have Phonological Forms (PFs) whereas the mere morphological forms of ‘semantic cases’ have none.
Phonology – Phonetics – Graphematics Kersten Krüger (Leipzig) writes on the automatic generation of the nominal wordforms in the future Ukrainian-German Online Dictionary by presenting and illustrating the necessary rules: 1. rules for deriving the stems from the corresponding lemmata, 2. rules for setting up the inflection classes of the corresponding stems, and 3. rules for forming the accent classes to which the word-forms belong. The inflection classes are the basis for deriving the case endings added to the stems. In this process, phonological, phonetic and the corresponding graphemic alternations have to be observed. Finally, the inflected nouns of an inflection class receive their corresponding stress patterns.
Introduction | xvii
Tobias Scheer, Markéta Ziková (Nice, Brno) explain the size of consonant clusters (plosive plus sonorant) of Old Czech (OCz). The empirical basis consists of a list of OCz nominal and adjectival word forms (Nom Sg, Gen Pl; adjective in its short form) with these consonant clusters. Government Phonology explains their different behavior by principles and parameters which attribute empty nuclei to trapped consonants (křtit – ‘to baptize’) in the CVCV skeleton that are not visible from above the skeleton, so that trapped consonants may be preceded and followed by consonants and cannot be stressed. The theory can also explain which clusters not appearing in the written record are accidental – and which ones are systematic gaps of OCz.
Communicative competence Grit Mehlhorn, Maria Yastrebova (Leipzig) write on the communicative competence of Russian adolescent heritage speakers in language mediation. Although this competence has been included in the German education standards of schooltaught modern languages for several years, the respective potential of heritage speakers has not yet been investigated. The paper introduces an instrument for mediation from as well as into the heritage language Russian in which the respondents have to switch between formal and informal registers and use culturespecific terms. The evaluation grid given pays attention to the information units mediated, their linguistic appropriateness, the reaction ability of the speakers as well as the suitability for recipients and their comprehension. The data from bilingual adolescents and their parents have been collected twice in a longitudinal test and the changes are documented. Uwe Junghanns Anita Steube
Leipzig, July 2016
Klaus Abels (London)
On the interaction of P-stranding and Sluicing in Bulgarian 1 Introduction This paper is about the grammar of ellipsis, specifically, sluicing (Ross 1969). Sluicing is the process deriving incomplete wh-questions where only the whphrase is pronounced, (1). (1)
a. John bought a car, but I don’t know which one. b. Nino talked with Andrea, but I don’t know with who else. c. A: Hans left. — B: When?
I will suggest that the interaction between preposition stranding, case morphology, and sluicing in the Bulgarian of – at least – some speakers strongly supports an account of sluicing under which (a) the intuitively missing part of the question is syntactically represented, (b) the missing part of the sentence is elided under semantic rather than syntactic identity with the antecedent, and (c) the pronounced wh-phrase has to fit in a specific sense to be discussed into the antecedent. Assumptions (a) and (b) constitute Merchant’s (1999, 2001) approach to sluicing. As pointed out by Lasnik (2001, 2005), (a) and (b) by themselves do not derive important, well-established, central properties of sluicing. Assumption (c) is intended to fix this gap in Merchant’s account. The conjunction of (a), (b), and (c) makes novel predictions not shared by competing accounts of sluicing like a Lasnik-style syntactic identity account or a Culicover-and-Jackendoff-style account with no syntax at the ellipsis site. The Bulgarian data presented here show
|| Note: I dedicate this paper in gratitude to Gerhild Zybatow. You took me under your wing when I had just finished my Ph.D. and gave me the space to follow my instincts as a postdoc in your project on the information structure of non-declarative clauses in Russian. The papers I wrote were about exclamatives in English and German. They had little to do with information structure and didn't mention Russian. Here, then, is another contribution to the same project. Thank you Gerhild! This paper would not have been possible without the observations gathered by Andriana Koumbarou and Dagmara Grabska on Bulgarian sluicing during my 2014 ACTL course. I also benefited from comments by the audiences at a colloquium talk in Newcastle in 2015, at CECIL’S 5, and at the 2015 LAGB meeting.
DOI 10.1515/9783110517873-002
2 | Klaus Abels (London)
that the specific expectations of the present account are borne out, giving it the empirical edge over its competitors. It will be useful to settle some terminology before proceeding. (2)
[John bought [ a car ]] but I don’t know [[which one] ] [correlate] [ remnant ] ellipsis site [ antecedent ] [ sluice ]
I will refer to which one as the remnant. The clause intuitively providing the meaning of the elliptical question, here John bought a car, will be called the antecedent. The indefinite, a car, whose identity is queried in sluicing will be called the correlate. The gap immediately following the remnant, where the remainder of the question would come in a canonical wh-question, will be called the ellipsis site, and the clausal structure containing remnant and ellipsis site form the sluice. 1 Example (2) has an indefinite correlate whose identity is queried by the sluice. This type of example is called the “merger type” in Chung, Ladusaw, and McCloskey (1995). When there is no overt correlate in the antecedent, we speak of sprouting, (1c). When the correlate is definite and the identity of a different referent satisfying the antecedent is queried, we speak of contrast sluicing, (1b). Finally, when discussing theories of sluicing that assume the presence of unpronounced syntactic structure at the ellipsis site, I will borrow the term “pre-sluice” from Dayal, and Schwarzschild (2010) to refer to the fully pronounced version of the sentence that gives rise to the sluice. A range of plausible pre-sluices for (2) is given in (3). (3) a. which car he bought b. which car it is The question of how ellipsis works has implications far beyond descriptive adequacy. These implications justify the surge of interest in ellipsis in recent years. The central concern of cognitive science is the correct characterization of the initial and final states of a given information processing device, the characterization of the mapping between the two states, and of the extent to which this mapping is modulated by data from the organism’s environment. In linguistics, poverty of the stimulus arguments have played a major role in this investigation. Poverty of
|| 1 The distinction between remnant and sluice is important even under theories where there is no syntactic representation of the ellipsis site, because the category of the sluice and the category of the remnant do not match. This fact will be discussed below.
On the interaction of P-stranding and Sluicing in Bulgarian | 3
the stimulus arguments purport to reveal properties of the final state that cannot plausibly be derived from environmental input by general purpose reasoning or pattern extraction devices and that must, therefore, be attributed to the initial state or to specific biases in the mapping. For recent discussion see Pullum, and Scholz (2002), Legate, and Yang (2002), Lidz, and Waxman (2004), Berwick et al. (2011), Chomsky (2013), Perfors, Tenenbaum, and Regier (2011), Reali, and Christiansen (2005), Kam et al. (2008). Elliptical constructions have full, regular interpretations supported by superficially incomplete structures. Since the missing material is manifested in the input only as an absence, it raises the poverty of the stimulus problem in a particularly acute way. If the arguments in this paper are correct, then neither of the two traditional, pleasingly trivial theories of ellipsis are correct: Neither does the ellipsis site lack syntactic structure nor is the structure simply identical to that of the antecedent. Since the correctness of the present, more complex account is revealed only by arcane negative data, we have a genuine poverty of the stimulus problem on our hands. Whatever the correct solution, it must ultimately reveal properties of the initial state or biases of the language acquisition device fairly directly. The findings reported here reveal their full significance in light of the considerations just discussed. The relatively complex grammar of ellipsis forced by the present data is acquired substantially without relevant evidence. It must therefore constitute the default or one of several alternative defaults (Nevins 2010) of the initial state or the language acquisition device. The answer to the question of whether there is one or several defaults depends on the range of between-language and between-speaker variation. The data reported below are superficially at variance with those in Merchant (1999, 2001), who reports that preposition stranding is categorically impossible under sluicing in Bulgarian. Taken at face value this would point to the possible existence of several different idiolects and, correspondingly, several alternative defaults. However, it is possible that the judgments reported by Merchant are intended as a mere preference of one structure over the other, which would be consistent with my own findings and would not lead to the conclusion that there are several defaults. One of my informants has produced the opposite pattern to what is reported here. Whether this is just noise or a real divergence in grammars will have to be determined by broader, carefully controlled data collection with many subjects and a methodology that can distinguish stylistic preferences from more or less categorical effects. This task is not undertaken here. Instead, I will take the judgments I report as true and explore the theoretical consequences.
4 | Klaus Abels (London)
The remainder of this paper is structured as follows. Section 2.1. introduces four main facts about sluicing. Together they constitute what I call Ross’ dilemma. Section 2.2. introduces the three main approaches to sluicing from the literature and proposes an amendment to Merchant’s (1999, 2001) approach that allows it to resolve Ross’ dilemma. This amendment takes the form of the fit condition, which stipulates that the remnant has to fit into the antecedent replacing the correlate if there is one. Section 2.3. demonstrates some empirical payoffs of the fit condition concerning sprouting, voice alternations, and spray-load alternations – notoriously difficult problems for Merchant’s approach without the fit condition. Section 3. discusses the Bulgarian data. I show how the data are accounted for under the current theory and why they are difficult from the perspective of competing approaches that either posit no syntax at the ellipsis site or demand syntactic identity between antecedent and ellipsis site. Section 4. concludes with the discussion of some open issues.
2 On Sluicing This section briefly reviews a few important generalizations about sluicing, 2.1. It then introduces the three main approaches to the phenomenon and suggests an amendment to Merchant’s theory of syntactic ellipsis under semantic identity, 2.2. Finally, I demonstrate that the suggested amendment has empirical payoffs substantially beyond the facts immediately motivating it, 2.3.
2.1 Ross’ dilemma Ross (1969) was the first generative paper to discuss sluicing. There are four basic facts, all of them explicit or implicit in Ross’ discussion, that continue to set the terms of the debate on sluicing to this day. These facts, in the order in which I will discuss them, pertain to the category of the sluice, its interpretation, the case of the remnant, and the apparent scopal properties of the remnant. Together, these four facts constitute a bit of a dilemma, the satisfying resolution of which is the main analytical challenge for theories of sluicing. In terms of their category, sluices are interrogative clauses no matter what the category of the remnant is. This conclusion rests on commonalities in the distribution of sluices, wh-questions, and declaratives on the one hand and the differences between the distribution of sluices and that expected (on the basis of
On the interaction of P-stranding and Sluicing in Bulgarian | 5
category) of the remnant. Starting with canonical, full wh-questions, their external distribution is not determined by the category of the wh-phrase but arises instead from the clausal syntax of questions and from their semantics. The following examples illustrate the clause-like behavior of questions. (4) a. *Fritz denkt (dar-) über die Frage, welches Gericht er Fritz thinks there about the question which dish he bestellen soll, nach. order should after ‘Fritz thinks about the question of which dish he should order.’ b. Fritz denkt (*dar-) über nach die Frage, welches Gericht Fritz thinks there about after the question which dish er bestellen soll. he order should Example (4a) shows that the preposition über takes a noun phrase as its complement but that the prepositional adverb darüber does not. Example (4b) shows that the nominal complement of the preposition cannot extrapose. | (5) a. Fritz denkt *(dar-) über, {welches Gericht er bestellen soll Fritz thinks there about which dish he order should dass er bezahlen muss}, nach. that he pay must after ‘Fritz thinks about {which dish he should order | the fact that he must pay}.’ b. Fritz denkt *(dar-) über, nach, {welches Gericht er bestellen Fritz thinks there about after which dish he order soll |dass er bezahlen muss}. should that he pay must ‘Fritz thinks about {which dish he should order | the fact that he must pay}.’ Example (5) illustrates the contrasting behavior of clauses. They cannot function as the complement of prepositions but only of prepositional adverbs, (5a), and they do extrapose, (5b). The interrogative forms of the prepositional phrase and the prepositional adverb are shown in (6), which are consequently interpreted as wh-questions.
6 | Klaus Abels (London)
(6) a. Über was denkt er nach? about what thinks he after ‘What is he thinking about?’ b. Wo-r-über denkt er nach? where-R-about thinks he after ‘What is he thinking about?’ Embedding an indirect question under a prepositional adverb does not form a wh-constituent, which is why (7) is interpreted as a statement. (7)
Da-r-über, welches Gericht er bestellen soll, denkt Fritz nach. there-R-about which dish he order should thinks Fritz after ‘Fritz is thinking about which dish he should order.’
Sluices behave like interrogative clauses and unlike wh-phrases in all respects. They cannot be embedded by prepositions but only by prepositional adverbs, (8a). Like clauses and unlike noun phrases sluices extrapose, (8b). (8)
Fritz weiss, dass er noch ein Gericht bestellen muss,… Fritz knows that he still a dish order must ‘Fritz knows that he should still order a dish… a. und denkt jetzt *(da-r-) über, welches Gericht, nach. and thinks now there-R about which dish after and is thinking about which dish’ b. und denkt jetzt *(da-r-) über nach, welches Gericht. and thinks now there-R about after which dish and is thinking about which dish’
Furthermore, a prepositional adverb combining with a sluice does not form a whphrase, hence (9) is interpreted as a declarative clause rather than a question. (9)
Fritz wird ein Gericht bestellen, aber darüber, welches Fritz will a dish order but there.R.about which Gericht, denkt er noch nach. dish thinks he still after ‘Fritz will order a dish, but which dish he is still thinking about.’
On the interaction of P-stranding and Sluicing in Bulgarian | 7
All of this leads to the conclusion that, in terms of their category, sluices are interrogative CPs.2 Further arguments for this conclusion are given in Merchant (1999, 2001: chapter 2) and it seems uncontroversial (Culicover, and Jackendoff 2005). The second fundamental fact about sluicing is that sluices are interpreted like full interrogative clauses. Any example of sluicing can be used to illustrate this conclusion, but the fact that (9) immediately above can only be interpreted as a declarative is particularly striking in this regard. The third discovery was that in sluicing (Ross essentially only looked at merger type sluices) the case of the correlate determines the case of the remnant. For nominal remnants this entails that the case on the remnant is not assigned by the predicate embedding the sluice, if there is one. The basic observation is illustrated in (10). (10) a. Er hat jemandem geholfen, aber ich weiss nicht {wem he has someone.DAT helped but I know not who.DAT |* wen |*wer}. who.ACC who.NOM ‘He helped someone, but I don’t know who.’ b. Er hat jemanden unterstützt, aber ich weiss nicht {*wem he has someone.ACC supported but I know not who.DAT | wen |*wer}. who.ACC who.NOM ‘He supported someone, but I don’t know who.’ c. Jemand ist von ihm unterstützt worden, aber ich weiss nicht someone.NOM is by him supported been but I know not { *wem |*wen |wer}. who.DAT who.ACC who.NOM ‘Someone was supported by him, but I don’t know who.’ The case on the remnant in (10) changes with the case of the correlate. When the correlate is in the dative, as demanded by the verb helfen, the remnant must be dative. When the correlate is in the accusative, the case governed by the verb unterstützen, the remnant must be in the accusative. Finally, when the correlate is
|| 2 The conclusion that sluices are clauses does not entail the presence of a (silent) complementizer in the syntax. This conclusion only follows given certain additional assumptions about the endo-centricity of all phrases, an issue I do not discuss here.
8 | Klaus Abels (London)
in the nominative carried by subjects, the remnant is in the nominative. The matrix verb wissen itself governs accusative case on its object: (11)
Fritz weiss {den Weg |*dem Weg Fritz knows the.ACC way the.DAT way ‘Fritz knows the way.’
|*der Weg}. the.NOM way
I will refer to this observation as case connectivity. The forth and final basic discovery about sluicing concerns the apparent scope of the remnant. If we take the antecedent as our guide in constructing a paraphrase for the sluice, we find that sluices can have interpretations corresponding to paraphrases that violate constraints on movement, in particular, island constrains. Thus, example (12) is grammatical and has an interpretation which can be paraphrased as in (13). Example (13) however is ungrammatical, since it violates the complex noun phrase constraint. (12) (13)
They want to hire someone who speaks a Balkan language, but I don’t know which (Balkan language). *They want to hire someone who speaks a Balkan language, but I don’t know which Balkan language they want to hire someone who speaks.
I will refer to this class of facts as the island insensitivity of sluicing. Theories of sluicing face the challenge of accounting for all four of these observations simultaneously. We will see in the next subsection what analytical options there are and why they struggle to meet the challenge.
2.2 Theories of sluicing The four properties just introduced characterize sluicing across a broad range of languages (Merchant 2001, Merchant, and Simpson 2012) and theories of sluicing must account for them. Theories of sluicing must also take a stand on the following two questions: What, if any, is the syntactic representation of the ellipsis site? Is the recoverability condition on ellipsis syntactic or semantic/pragmatic?3 Accounts of sluicing (and ellipsis more generally) can profitably be cross-classified
|| 3 As discussed, the lack of direct evidence about these two questions poses a particularly sharp poverty of the stimulus problem for the learner. Essentially, the answer to the questions must be
On the interaction of P-stranding and Sluicing in Bulgarian | 9
by the answers to these two questions, as in the following table, adapted from Merchant (to appear). Tab. 1: Cross-classification of approaches, adapted from Merchant (to appear)
ellipsis site contains structure
contains no structure
syntactic identity
Syntactic identity: Ross (1969), Chomsky (1972), Lakoff (1970), Sag (1976), Fiengo, and May (1994), Chung, Ladusaw, and McCloskey (1995), Lasnik (2005), Merchant (2008)
N/A (incoherent)
semantic identity
Semantic identity +: Abels (2011), Baker, and Brame (1972), Merchant (2001), van Craenenbroeck (2010a,b), Barros, Elliott, and Thoms (2014), Barros (2014), Abels (to appear)
No syntax at ellipsis site: Dalrymple, Sheiber, and Pereira (1991), Ginzburg, and Sag (2000), Culicover, and Jackendoff (2005)
The two simplest accounts of sluicing are the ones in the top left and the bottom right corner of the table. The top left represents theories according to which there is syntactic structure at the ellipsis site and the recoverability condition is a syntactic one. On these accounts, the recoverability condition demands, roughly, syntactic identity (or isomorphism) between antecedent and ellipsis site. I will refer to this class of accounts as syntactic identity accounts. Syntactic identity accounts contrast with a class of theories, in the bottom right of the table, that deny the existence of (context dependent) structure at the ellipsis site and that posit semantic/pragmatic recoverability conditions. I will refer to these accounts as the no-syntax accounts. No theories are listed in the top right corner of the
|| provided more or less directly by defaults of universal grammar. Speakers converge across languages and evidence that would override the defaults is lacking in the primary linguistic data. The table below indicates that questions about the nature of the ellipsis site and the recoverability condition have not been settled in linguistic theory despite substantial attention to the issues. This reinforces, in my view, the poverty of the stimulus argument: since syntacticians have access to more (including negative) and more controlled data than learners with the added advantage that the syntacticans’ data can be annotated for acceptability and interpretation. Despite all of these advantages, syntacticians have not been able to agree on an answer to the question whether there is syntactic structure present at the ellipsis site.
10 | Klaus Abels (London)
table because, for a theory to fall into that cell, it would have to make the incoherent demand that there is no (contextually variable) structure at the ellipsis site and that the recoverability condition on ellipsis is syntactic in nature. Finally, the bottom left hand cell is occupied by accounts that assume the presence of contextually varying syntactic structure at the ellipsis site but hold that the recoverability condition is semantic/pragmatic rather than syntactic. They are designated as “semantic identity +” because to reach descriptive adequacy they have to add further constraints to the simple semantic identity condition. The present paper endorses this type of approach. As we will see promptly, what answer a theory gives to these two questions strongly guides its account of the four observations from the previous subsection. Turning to syntactic identity accounts first, they posit that there is full fledged syntactic structure at the ellipsis site at the point of the derivation/level of representation relevant to interpretation. The structure is either silenced or absent at the point of the derivation/level of representation relevant to pronunciation. These accounts also assume that, modulo wh-movement of the remnant, the ellipsis site is structurally identical to the antecedent or very nearly so. Such accounts have a straightforward explanation for the observation that sluices behave like clauses (The sluice is a clause under such approaches.), that the sluice is interpreted as a full interrogative clause (That’s what it is.), and for case connectivity. Case connectivity follows from the assumption that antecedent and ellipsis site have identical structure, which, of course, entails identity of case governors.4 The principal problems for syntactic identity are the island insensitivity of sluicing and a host of cases that do not lend themselves to an analysis in terms of strict syntactic identity, because strictly identical paraphrases of the sluice are either impossible or have the wrong interpretation (see Merchant 2001: chapter 1 for review). The island repair problem has given rise to a substantial literature (for review and references see Barros, Elliott, and Thoms 2014, Abels to appear), but the proposed solutions are often stipulative (see Culicover, and Jackendoff 2005 for this point) or fail to address the further wrinkle that locality sensitivity returns in multiple sluicing (see for example Takahashi (1994), Lasnik (2014), Nishigauchi (1998), Marušič, and Žaucer (2013)) and contrast sluicing (see for example Fukaya 2012, 2007, Griffiths, and Lipták 2011, Winkler 2013).
|| 4 Van Craenenbroeck (2010a) proposes that a looser version of syntactic congruence should be used with the choice between alternatives regulated by last resort (see also Kim et al. 2011). The main problem here is to define the conditions under which non-identical structures can be allowed. The discussion here ignores such refinements and concentrates on the basic syntactic identity approach.
On the interaction of P-stranding and Sluicing in Bulgarian | 11
Despite the problems in accounting for island insensitivity, syntactic identity accounts have enjoyed enduring popularity, probably because the analysis of the first three properties of sluicing is simple and convincing. In particular, case connectivity is often viewed all by itself as convincing evidence for the existence of identical syntactic structure at the ellipsis site. The argument is based on the prior conviction that case is a syntactic phenomenon and that it follows from properties of universal grammar that there must be an inaudible case licenser for the remnant in sluicing. This fact is tied elegantly to the semantic congruence between antecedent and sluice by syntactic identity accounts. The only otherwise unmotivated assumption that needs to be added to universal grammar is that ellipsis is recovered under syntactic identity.5 It is this trivialization of ellipsis that makes syntactic identity accounts so appealing. Approaches where there is no (contextually variable) syntactic structure at the ellipsis site typically take a more surface-oriented, less abstract approach to syntactic structure generally. Sluicing is then viewed as a subtype of interrogative constructions. This classification can be learned from semantic evidence. The classification endows sluices by default inheritance mechanism with clausal syntactic structure. Island insensitivity of sluicing follows directly from the assumption that there is no structure at the ellipsis site: structural constraints on displacement cannot play a role when there is no structure. This leaves case connectivity in need of an account. Here, no-syntax approaches invoke a relation between correlate and remnant whereby the remnant or the remnant’s case is anaphorically related to the correlate’s. The principal problem here is that the account is too loose. Language internally, this plays out as an overgeneration issue. A relevant case is given in (14). (14) a. Mary is proud of someone, but she won’t tell us (of) who. b. Mary is proud, but she won’t tell us *(of) who. Semantically, the version of (14b) with and without the preposition of are fully synonymous, yet, the preposition of is required, which must be attributed to syntax. The only syntactic consideration entering the account has to do with anaphoric case on the remnant. In (14b), the remnant has no correlate, hence, nothing to be anaphoric to. However, this alone cannot be the explanation for why the preposition is required, since in many other cases a remnant without a casemarked correlate is possible without a preposition as part of the remnant, (15).
|| 5 I am ignoring here and throughout questions of ellipsis licensing distinct from ellipsis recoverability. For work on ellipsis licensing see Aelbrecht (2010), Lobeck (1987), Johnson (1996).
12 | Klaus Abels (London)
Thus, there seems to be no non-stipulative way to account for the obligatory presence of of in (14b): As it stands, the account is too loose. (15)
John will eat, but I don’t know what.
The looseness of the account manifests itself in a different form cross-linguistically. It is unclear why case connectivity should hold across languages. A language where, for example, default case instead of anaphoric case is used on the remnant seems plausible, but we don’t seem to encounter such languages. Finally, the island insensitivity of sluicing is derived in such a deep way in no-syntax accounts, that they fail to offer any approach to contrast and multiple sluicing, where we do find locality effects. Merchant (1999, 2001) develops an account that combines a semantic condition of recoverability of deletion with the assumption that there is syntactic structure at the ellipsis site. This explains the clausal categorial and interpretive behavior of sluicing directly. This way of approaching the phenomenon has the additional advantage of allowing for cases where syntactic identity is too strict: as long as the structure at the ellipsis site gives rise to an interpretation that can be recovered, its syntax may deviate from that in the antecedent. Merchant suggests that this possibility of a syntactic mismatch between antecedent and ellipsis site explains island insensitivity (at least in part); when island effects are avoided, there is a synonymous non-island violating paraphrase acting as presluice. This logic can then be used to argue that when island sensitivity is observed, this is due to the absence of a non-island violating paraphrase (Fukaya 2007, 2012, Marušič, and Žaucer 2013, Abels to appear, Barros, Elliott, and Thoms 2014). While the idea is an attractive one, it is unclear how it accounts for case connectivity. In example (10a) above the case governor in the antecedent is the verb helfen, which assigns dative case. One might expect violations of case connectivity if any of the following paraphrases were allowed at the ellipsis site as presluices: (16) a. … wen er unterstützt hat. who.ACC he supported has ‘…who he supported.’ b. … wer es ist, dem er geholfen hat. who.NOM it is that.DAT he helped has ‘...who it is that he helped.’
On the interaction of P-stranding and Sluicing in Bulgarian | 13
c. … wer von ihm geholfen gekriegt hat. who.NOM by him helped become has ‘...who was helped by him.’ In addition, examples like (14) are problematic under Merchant’s account, because both she is proud of tNP and she is proud tPP appear to be semantically recoverable in the same way from the antecedent she is proud. These considerations are at the heart of Lasnik’s (2005) objections to Merchant’s theory. To solve the problem of case connectivity for Merchant’s approach, I would like to propose a condition that I call fit.6 (17)
Fit condition Modulo agreement in the antecedent and wh-movement, replacing the correlate by the remnant in the antecedent must lead to a syntactically well-formed structure with the right meaning or – for sprouting – adding the correlate into the antecedent and making no further changes must lead to a syntactically well-formed structure with the intended thematic interpretation. (Compare also Barker 2013.)
The fit condition is intended to express the fact that, ignoring the need for the remnant to undergo wh-movement, the remnant must be able to grammatically replace the correlate in the antecedent under the relevant thematic interpretation. By assumption, there is also structure at the ellipsis site and the remnant is moved from it. Therefore, the remnant also fits into the structure at the ellipsis site. Under this approach, the remnant is like a piece of a puzzle that fits in two places: It fits into the antecedent (by the fit condition) and it fits into the structure at the ellipsis site (by normal assumptions about structure and movement). As formulated above, the fit condition is clearly not precise enough. We will want to say that in German a noun phrase in the nominative does not fit into a slot for a dative noun phrase when the two are morphologically distinct. But there is a real question about examples where there is case syncretism. It is known that some processes such as free relatives are sensitive to purely morphological fitting. Other processes require identity of syntactic case. We will see some evidence
|| 6 There are, of course, other avenues open. We could try to explicitly limit the syntactic differences between antecedent and ellipsis site (Chung, Ladusaw, and McCloskey 2011, Chung 2013) or stipulate case connectivity. It should also be noted that the fit condition can in principle be bolted on to any of the other approaches to sluicing. It should become clear why I opt for the specific combination of assumptions I do.
14 | Klaus Abels (London)
bearing on the question of how to make the fit condition precise in the next section, but will have to leave aside the task of formulating fit with full precision. Note also that, in order to reach explanatory adequacy, it needs to be shown how the fit generalization follows from principles of universal grammar as a default. Again, I will not undertake this task here and will make only the briefest of comments in the conclusion to the paper. The three approaches to sluicing can be summarized by the following three diagrams that show the overall logic. In a syntactic identity approach, case connectivity follows from syntactic identity between antecedent and pre-sluice, but this raises the problem of why sluicing is island insensitive and why certain mismatches are allowed. No-syntax approaches can explain the lack of island effects directly, but need to stipulate case connectivity via a special case copying mechanism that lies outside of normal case assignment mechanisms. Finally, the current approach demands semantic identity between antecedent and sluice while allowing syntactic mismatches. This explains the island insensitivity of sluicing. Case connectivity is explained in terms of the fit condition, which guarantees that the remnant can locally fit not only into the pre-sluice but also (modulo movement) into the antecedent. (18) a. Syntactic identity: sluice
3 remnant 2 antecedent
E site
2 syntactic identity 2 2
2 2 2
case assigner
correlate
case assigner
tremnant
On the interaction of P-stranding and Sluicing in Bulgarian | 15
b. No syntax at ellipsis site: antecedent 2 2 2 case
sluice
correlate
assigner
remnant case copying
c. Semantic identity with syntax at ellipsis site and fit condition: sluice 2 remnant antecedent
E site
2
semantic identity
2
2 tremnant 2
3 case
2
Fit
correlate
case
2
assignerE
assignerA
2.3 Fit at work In the previous subsection we considered the paradigm in (14), repeated below. (19) a. Mary is proud of someone, but she won’t tell us (i) who. (ii) of who. b. Mary is proud, but she won’t tell us (i) *who. (ii) of who. The examples are notoriously difficult for accounts positing a semantic identity condition, essentially, because of is semantically vacuous syntactic glue, which can be stranded under wh-movement:
16 | Klaus Abels (London)
(20) a. Who is she proud of? b. Of who is she proud? The problem is solved by the fit condition. The remnant in (19a-i) is who, the correlate is someone. Who fits into the position of someone modulo movement. In (19a-ii) the correlate is of someone. Again, modulo movement, the well-formed prepositional phrase of who fits in the position of the correlate. In (19b) there is no correlate and who does not fit into the antecedent clause. The sprouted argument of proud fits into the antecedent only when of is present: (21) a. She is proud *(of) someone. b. Who is she proud *(of)? This is how the fit condition solves the problem of why preposition stranding appears to be disallowed under sprouting. The same essential logic applies to voice alternations, which are notoriously troublesome for semantic identity accounts. The problem can be illustrated by the following examples: (22) (23)
John was arrested, but the report doesn’t say which officer. *Some officer arrested John, but the report doesn’t say by which officer.
The sluice in (22) cannot be interpreted to mean which officer arrested John.7 Given that the antecedent in (22) entails that someone or some officer arrested John, the following pre-sluices should be recoverable: (24) a. …which officer John was arrested by. b. …which officer arrested John. Thus, sluicing should be allowed under the relevant interpretation. We can account for the fact that it is not by invoking the fit condition: an agent noun phrase does not fit into the passive antecedent without the addition of by, which is therefore required for a well-formed sluice: (25)
John was arrested, but I don’t know by which officer.
|| 7 The interpretation that (22) has is a contrastive one suggesting that John and an unknown officer were arrested. This is unproblematic and irrelevant here.
On the interaction of P-stranding and Sluicing in Bulgarian | 17
Example (23) suffers from the opposite problem: the superfluous presence of by. Again there is a well-formed, recoverable pre-sluice: (26)
…by which officer John was arrested.
The ungrammaticality follows from the fit condition, because the agent cannot be expressed as a by-phrase in the antecedent. The same solution works for the difficult case of spray-load alternations. Again, semantic identity accounts run into issues of over-generation. The problem can be illustrated by the following examples: (27) (28)
He loaded something onto the wagon, but I don’t know with what. *He loaded some wagon with hay, but I don’t know onto which wagon.
The question is why the sluice in (27) can only be interpreted as a question about the instrument of loading but not as the object being loaded and why (28) is ungrammatical, given the in-principle availability and recoverability of the following pre-sluices: (29) (30)
…with what he loaded the wagon. …onto which wagon he loaded the hay.
The fit condition provides the answer to this problem. A prepositional phrase headed by with fits into the antecedent in (27), but only with an instrument reading, so that is the reading we derive. The prepositional phrase headed by onto does not fit into the antecedent in (28). The discussion in this subsection was meant neither as a justification of the fit condition nor as an exhaustive demonstration of its analytic value and certainly not as a discussion of its problems and limitations. Instead, it was intended as a brief illustration of its power and potential value. Most of all I hope to have created a sense for the style of analysis necessary under the current approach, where a syntactic well-formed pre-sluice undergoes ellipsis under semantic identity with the antecedent and the remnant is subject to the fit condition. This style of analysis requires the remnant to be part of (and therefore fit into) a well-formed pre-sluice and into the antecedent. In the next section, I will apply this style of analysis to unfamiliar data involving the interaction of Bulgarian preposition stranding, case morphology, and sluicing.
18 | Klaus Abels (London)
3 Bulgarian In this section I present Bulgarian data which fall out quite directly from the setup here but which are difficult to explain under alternative accounts, in particular the syntactic identity and the no-syntax accounts. Before turning to the sluicing data, I need to introduce a few facts of Bulgarian grammar as background. First, modern Bulgarian has lost almost all case distinctions that were present historically. There is no distinction between subject and object case, except in pronouns, including indefinite and interrogative pronouns and in the definite article. Even for pronouns, the distinction is only optionally expressed in the sense that there is a general form (glossed as g) which can appear in the position of subject, object, and object of prepositions. The other form (historically an accusative and glossed as non-S) is restricted to the positions of object and object of prepositions, (31). The object form of the pronoun is somewhat formal and archaic.8 |njakogo}. (31) a. Ivan sreshtna {njakoi Ivan met.3SG someone.G someone.NON-S ‘Ivan met someone.’ b. Ivan tancuva {s njakoi |s njakogo}. Ivan danced with someone.G with someone.NON-S ‘Ivan danced with someone.’ c. { Njakoi |*Njakogo} tancuva. someone.G someone.NON-S danced ‘Someone danced.’ The other fact about Bulgarian that we will rely on is that preposition stranding is disallowed independently of the form of the interrogative pronoun used: {koi |kogo} tancuva Ivan _______? (32) a. S with who.G who.NON-S danced Ivan ‘With who did Ivan dance?’ b. *Koi tancuva {s ______ Ivan |Ivan s ______ }? who.G danced with Ivan Ivan with c. *Kogo tancuva {s ______ Ivan |Ivan s ______ }? who.NON-S danced with Ivan Ivan with
|| 8 For further discussion see Scatton (1993).
On the interaction of P-stranding and Sluicing in Bulgarian | 19
Object sluices in Bulgarian allow arbitrary combinations of general form and object form of the pronouns in correlate and remnant, (33), though there is a stylistic preference for matching forms in antecedent and remnant. Subject sluices, are, of course, limited to the general form of the pronoun, (34). (33)
(34)
Ivan sreshtna {njakoi |njakogo}, no Ivan met.3SG someone.G someone.NON-S but znam {koi |kogo}. know.1SG who.G who.NON-S ‘Ivan met someone, but I don’t know who.’ { Njakoi |*Njakogo}, tancuva, no ne someone.G someone.NON-S danced but NEG { koi |*kogo}. who.G who.NON-S ‘Someone danced, but I don’t know who.’
ne NEG
znam know.1SG
The same range of possibilities exists in full questions. (35)
(36)
Ivan sreshtna {njakoi |njakogo}, no ne znam Ivan met.3SG someone.G someone.NON-S but NEG know.1SG { koi |kogo} sreshtna Ivan. who.G who.NON-S met Ivan ‘Ivan met someone, but I don’t know who Ivan met.’ { Njakoi |*Njakogo}, tancuva, no ne znam someone.G someone.NON-S danced but NEG know.1SG { koi |*kogo} tancuva. who.G who.NON-S danced ‘Someone danced, but I don’t know who danced.’
This state of affairs is, of course, as expected under the current account. Both forms of the object pronoun give rise to well-formed pre-sluices. For (33) both forms fit into the antecedent. In (34) only the general form fits. This accounts for the restriction seen in (34) even if a suitable pre-sluice could be found that licenses the object form of the pronoun. Syntactic identity theories have an obvious account of these facts, too. Finally, no-syntax accounts require a somewhat abstract view of case, whereby the pronoun in positions that are compatible with both forms share a property (say acc) not present in environments where only the general form is possible. To account for the facts above, the anaphoric relation licensing case on the remnant must be sensitive to this abstract feature rather
20 | Klaus Abels (London)
than to the morphological form. If these assumptions are adopted, the data above can be accounted for. When we turn to questions with prepositions, things become more interesting. If the preposition is part of the remnant, both case forms of the pronoun are allowed – as in full questions – and correlate remnant mismatches are possible in both directions, (37). (37)
Ivan tancuva {s njakoj |s njakogo}, no ne Ivan danced with someone.G with someone.NON-S but NEG znam {s koi |s kogo}. know with who.G with who.NON-S ‘Ivan danced with someone but I don’t know with who.’
Given what we said in the previous paragraph, this is unsurprising under any account. However, some speakers also allow apparent preposition stranding under sluicing. Example (38) illustrates the situation for remnants with the general form of the pronoun. We will discuss examples with the object form in the next paragraph. (38)
Ivan tancuva {s njakoj |s njakogo}, no ne Ivan danced with someone.G with someone.NON-S but NEG znam koi. know who.G ‘Ivan danced with someone but I don’t know who.’
The situation in (38) is not particularly surprising under no-syntax accounts: pied-piping of the preposition is a purely syntactic phenomenon accompanying movement. Since there is no movement in the derivation of sluicing, pied-piping need not happen, hence, the well-formedness of (38) is expected. Example (38) is problematic under syntactic identity accounts, because the sluice would have to be derived from the ill-formed pre-sluice (32b) and should therefore be ungrammatical. The current account allows a mismatch between the syntax of the antecedent and the syntactic structure at the ellipsis site. Under the current approach, (32b) is not the only pre-sluice. We also need to consider (39). (39) is well-formed and semantically recoverable.
On the interaction of P-stranding and Sluicing in Bulgarian | 21
(39)
Ivan tancuva {s njakoj |s njakogo}, no ne Ivan danced with someone.G with someone.NON-S but NEG znam koi beshe tova. know who.G was that ‘Ivan danced with someone but I don’t know who that was.’
The general form of the pronoun we find in (39) fits into the antecedent. Hence, (38) has (39) as recoverable, well-formed source and the remnant fits into the antecedent. This makes the acceptability of (38) the expected state of affairs. Things change when we consider (40), which differs from (38) only in the form of the pronoun in the remnant. The example is unacceptable. (40)
Ivan tancuva {s njakoj |s njakogo}, no ne Ivan danced with someone.G with someone.NON-S but NEG znam kogo. know.1SG who.NON-S ‘Ivan danced with someone but I don’t know who.’
As far as I can see, this is mysterious under no-syntax accounts. Syntactic identity accounts explain (40) straightforwardly. The pre-sluice involves preposition stranding, (32c), and is, therefore, ungrammatical. The copulative sentence that we invoked as the pre-sluice for (38) does not allow the object form of the pronoun: (41)
*Ivan tancuva {s Ivan danced with znam kogo know.1SG who.NON-S
njakoj |s njakogo}, no ne someone.G with someone.NON-S but NEG beshe tova. was that
Lacking a plausible, well-formed, recoverable pre-sluice, (40) is correctly predicted to be ungrammatical. The discussion of (38) and (40) can be summarized in the table below, which shows that, amongst the three accounts considered here, only the semantic identity account with structure at the ellipsis site and the fit condition explains the
22 | Klaus Abels (London)
data. Neither of the other account can deal with the reversal of the judgment depending on the form of the remnant.9 Tab. 2: Acceptability of stranding by remnant
Judgement agrees with prediction
wh.G
wh.non-S *
syntactic identity
no
yes
no-syntax
yes
no
here
yes
yes
The reversal of the judgments depending on the form of the remnant precludes a language-wide solution that could be bolted onto syntactic identity or no syntax accounts. The obvious way to go would be to add conditions regulating how the form of the pronoun interacts with stranding under sluicing directly. The current account has a markedly different logic, since the availability of stranding depends on the availability of a suitable, well-formed, recoverable pre-sluice. Neither of the other theories leave room for maneuver in this direction. No syntax accounts cannot invoke the structure present at the ellipsis site, since they deny it is there. Syntactic identity accounts cannot invoke a structure different from the antecedent without giving up the central assumption of the account (and the explanation of case connectivity). The rest of this section will present two additional arguments suggesting that what matters is the availability of a well-formed, recoverable pre-sluice rather than the form of the pronoun per se. The arguments are straightforward: Stranding even of the general form of the pronoun becomes impossible when the copulative structure in (39) argued to provide the pre-sluice for (38) is unavailable for reasons other than the form of the pronoun. The first argument comes from contrast sluicing. So far, we have concentrated on Chung, Ladusaw, and McCloskey’s merger type of sluicing. However, Bulgarian – like English – also allows contrast sluices, where the remnant queries the identity of an alternative to the correlate:
|| 9 The Bulgarian default pronouns behave like Spanish and Brazilian Portuguese, as described in Rodrigues, Nevins, and Vicente (2009). The object pronouns behave like German, as described in Merchant (2001). This is not an accident under the current theory.
On the interaction of P-stranding and Sluicing in Bulgarian | 23
(42)
Ivan tragna, no ne znam koi oshte. Ivan left but NEG know.1SG who else ‘Ivan left, but I don’t know who else.’
Again like in English, the pre-sluice for such examples cannot be a copulative structure like (39) above, (43a), while a regular full question does work as a presulice, (43b): (43)
a. *Ivan tragna, no ne znam koi oshte beshe tova. Ivan left but NEG know.1SG who else was that *‘Ivan left, but I don’t know who else it is.’ b. Ivan tragna, no ne znam koi oshte tragna. Ivan left but NEG know.1SG who else left ‘Ivan left, but I don’t know who else left.’
The important observation is that contrast sluicing in Bulgarian requires the presence of the preposition independently of the form of the pronoun. (44)
Ivan tancuva s Maria, no ne znam *(s) {koi |kogo} Ivan danced with Maria but NEG know with who.G who.NON-S oshte. else ‘Ivan danced with Maria, but I don’t know with who else.’
This is as expected under the current approach but it compounds the difficulties for the no-syntax and syntactic identity approaches, since, under both accounts, the status of an example cannot depend on the syntactic status of a paraphrase that is not isomorphic to the antecedent. The second argument comes from multiple sluicing. Prepositions must be retained in multiple sluicing, independently of the form of the pronoun: (45)
Ivan zapozna njakoi s {njakogo |njakoi}, Ivan introduced someone.G with someone.NON-S someone.G no ne znam koi *(s) {kogo |koi}. but NEG know.1SG who with who.NON-S who.G ‘Ivan introduced someone to someone, but I don’t know who to who.’
This is as expected under the current account, because the identificational beshe tova–‘was that’ sentence that was suggested as the pre-sluice for the stranding
24 | Klaus Abels (London)
example above only allows a single predicate. The no-syntax and syntactic identity accounts, again, struggle to come to grips with these examples. In this section I hope to have shown that the complex interaction of preposition stranding and sluicing with pronominal morphology in Bulgarian cannot be explained without making reference to the grammatical status of a range of different potential pre-sluices. Since such reference is incompatible both with syntactic identity and with no-syntax accounts, this conclusion serves as an argument against all such accounts. I take the ease with which the pattern flows from the current account to be a strong argument in its favor.
4 Conclusion This paper has proposed the fit condition as an amendment to Merchant’s theory of sluicing. The fit condition demands that the remnant must fit into the antecedent, replacing the correlate if there is one. After illustrating a few of the immediate benefits of fit for the analysis of sluicing in English, I presented a novel set of data from Bulgarian, partially at variance with judgments from the literature, that strongly suggest the current approach. In these final paragraphs I would like to discuss some of the most obvious open issues. These are the issue of idiolectal and dialectal variation, the exact content of the fit condition, and a possible approach to the question of why fit might be a default of universal grammar, as it has to be if the reflections from the introduction on ellipsis as a particularly acute poverty of the stimulus situation are correct. I will take these up in the order given. Merchant (1999, 2001) reports that preposition stranding under sluicing is impossible in Bulgarian. His data involve pronouns in the default form, this is the type of examples reported here to allow preposition stranding under sluicing. The judgment reported by Merchant is not entirely surprising. My own informants prefer the version with the preposition over the version without it in all conditions. Yet, there is a sharp contrast in the prepositionless version between the general and the object form of the wh-pronoun. Thus, there is not necessarily a contradiction in the facts as I report them and as Merchant does. A second reason not to be too surprised comes from the fact that the copulative pre-sluice underlying the prepositionless version appears not to be immediately available for some speakers but emerges only on reflection, altering the judgment on prepositionless sluice. Work with more informants designed to distinguish mere preferences from more categorical effects should be able to determine whether there is
On the interaction of P-stranding and Sluicing in Bulgarian | 25
a split in the population or not. I have run into one informant who prefers stranding with the object form of the pronoun and rejects it with the general form. Again, further work is necessary to determine whether this data point represents a genuine dialect or idiolect or is just noise. A second issue that requires attention has to do with the exact formulation and content of the fit condition. We have seen for Bulgarian that the superficial morphology of the wh-word matters. For the fit condition to work properly in example (38), the wh-pronoun, which is a subject in the pre-sluice, must be able to fit into the antecedent’s complement position within the prepositional phrase. This is possible if fit references the superficial shape of the wh-pronoun. Otherwise, the syncretism couldn’t be made to matter. On the other hand, it is not true that superficial syncretism is always the overriding factor for preposition stranding under sluicing. Consider the following German examples: (46) a. Hans hat auf etwas gesessen, aber ich sage dir nicht Hans has on something sat but I say you not * (auf) was. on what.NOM/DAT ‘Hans has sat on something, but I won’t tell you (on) what.’ b. Hans hat auf jemandem gesessen, aber ich Hans has on someone.DAT sat but I sage dir nicht {*(auf) wem |*wer}. say you not on who.NOM who.NOM ‘Hans has sat on someone, but I won’t tell you (on) who.’ If the overriding factor were simply surface syncretism, we would expect (46a) to allow stranding under sluicing, because there is a copulative source and the remnant wh-pronoun was is syncretic between nominative and dative. And there should, crucially, be a contrast between (46a) and (46b), because in the latter the remnant wh-phrase is not syncretic in the same way. However, at least in my own judgment, this effect does not show up.10 Similarly for Polish. Polish has nominative-accusative syncretism in inanimate masculine and neuter forms but not in animate masculine and feminine ones. As far as I have been able to determine, the judgments on preposition stranding under sluicing seem unaffected by this.
|| 10 For me, which-phrases allow preposition stranding under sluicing but remain degraded unless the part of the noun phrase following, which is also elided. Again, case syncretism does not seem to affect the judgment. Even for which-phrases, preposition stranding becomes much worse in contrast and multiple sluicing.
26 | Klaus Abels (London)
Clearly, too naive a formulation of fit in terms of superficial morphological forms would give rise to the expectation that syncretism should always matter and one in terms of abstract case alone would give rise to the expectation that it never does. It is, at present, unclear how to find the proper balance or whether some kind of parameterization is required. The final question to be discussed here arises in the context of the considerations about the poverty of the stimulus problem from the introduction. How is a system as complicated as the Bulgarian one acquired? In particular, what allows the fit condition to be a default of the language acquisition device? The answer to this question might be tied to the following considerations. Wh-phrases are standardly thought of as foci. If they are foci, then they generate a set of alternatives. In merger type and contrast sluicing, the correlate is usually focused and thus gives rise to a set of alternatives as well. Plausibly, the remnant and correlate are focus alternatives to each other. Under a structural (rather than a semantic) theory of focus alternatives along the lines of Katzir (2008), Fox, and Katzir (2011) it seems natural to impose the condition that the generated alternatives must fit in the position of the generator. Thus, if structurally generated alternatives are used to create coherence between antecedent and sluice, then the fit condition might follow from the imposition that the remnant and correlate be structural focus alternatives to each other. Working out the details of this poses a number of challenges and the issue must be left open here.
References Abels, Klaus. 2011. Don’t repair that island! It ain’t broke. Talk presented at Islands in Contemporary Linguistic Theory. Abels, Klaus. to appear. Movement and islands. In Jeroen van Craenenbroeck, and Tanja Temmerman (eds.), The Oxford handbook of ellipsis, chap. 15. Oxford: Oxford University Press. Aelbrecht, Lobke. 2010. The syntactic licensing of ellipsis. Werner Abraham, and Elly van Gelderen (eds.). Vol. 149 (LA). Amsterdam & Philadelphia: John Benjamins Publishing Company. Baker, C. L., and Michael Brame. 1972. ‘Global rules’: a rejoinder. Language 48(1). 51–75. Barker, Chris. 2013. Scopability and sluicing. Linguistics and Philosophy 36. 187–223. Barros, Matthew. 2014. Sluicing and identity in ellipsis. New Brunswick: Rutgers University PhD thesis. Barros, Matthew, Patrick D. Elliott, and Gary Thoms. 2014. There is no island repair. Ms. Rutgers, UCL, University of Edinburgh. Berwick, Robert C. et al. 2011. Poverty of the stimulus revisited. Cognitive Science. 1207–1242. Chomsky, Noam. 1972. Some empirical issues in the theory of transformational grammar. In Stanley Peters (ed.), The goals of linguistic theory, 63–130. Englewood Cliffs, New Jersey: Prentice-Hall.
On the interaction of P-stranding and Sluicing in Bulgarian | 27
Chomsky, Noam. 2013. Poverty of the stimulus: willingness to be puzzled. In Massimo PiatelliPalmarini, and Robert C. Berwick (eds.), Rich languages from poor inputs, 61–67. Oxford: Oxford University Press. Chung, Sandra. 2013. Syntactic identity in sluicing: how much and why. Linguistic Inquiry 44(1). 1–44. Chung, Sandra, William A. Ladusaw, and James McCloskey. 1995. Sluicing and logical form. Natural Language Semantics 3. 239–282. Chung, Sandra, William A. Ladusaw, and James McCloskey. 2011. Sluicing(:) Between Structure and Inference. Ms. UCSC. van Craenenbroeck, Jeroen. 2010a. Invisible last resort: a note on clefts as the underlying source for sluicing. Lingua 120(7). 1714–1726. van Craenenbroeck, Jeroen. 2010b. The syntax of ellipsis (Oxford Studies in Comparative Syntax). Oxford: Oxford University Press. van Craenenbroeck, Jeroen, and Tanja Temmerman (eds.). to appear. The Oxford handbook of ellipsis. Oxford: Oxford University Press. Culicover, Peter W, and Ray S Jackendoff. 2005. Simpler syntax. Oxford: Oxford University Press. Dalrymple, Mary, Stuart M. Sheiber, and Fernando C. N. Pereira. 1991. Ellipsis and higher-order unification. Linguistics and Philosophy 14(4). 399–452. Dayal, Veneeta, and Roger Schwarzschild. 2010. Definite inner antecedents and wh-correlates in sluicing. In Peter Staroverov et al. (eds.), Rutgers working papers in linguistics, vol. 3, 92–114. LGSA. Fiengo, Robert, and Robert May. 1994. Indices and identity. Cambridge, MA.: MIT Press. Fox, Danny, and Roni Katzir. 2011. On the characterization of alternatives. Natural Language Semantics 19. 87–107. Fukaya, Teruhiko. 2007. Sluicing and stripping in Japanese and some implications. Los Angeles, CA: University of Southern California PhD thesis. Fukaya, Teruhiko. 2012. Island-sensitivity in Japanese sluicing and some implications. In Jason Merchant, and Andrew Simpson (eds.), Sluicing: cross-linguistic perspectives, vol. 38 (Oxford Studies in Theoretical Linguistics), 123–163. Oxford: Oxford University Press. Ginzburg, Jonathan, and Ivan A. Sag. 2000. Interrogative investigations: the form, meaning, and use of English interrogatives. Stanford, CA: CSLI Publications. Griffiths, James, and Anikó Lipták. 2011. Contrast and island-sensitivity in clausal ellipsis. Ms. University of Leiden. Johnson, Kyle. 1996. When verb phrases go missing. GLOT International 2(5). 2–9. Kam, Xuân-Nga Cao et al. 2008. Bigrams and the richness of the stimulus. Cognitive Science 32. 771–787. Katzir, Roni. 2008. Structural competition in grammar. Cambridge, Mass.: MIT PhD thesis. Kim, Christina S. et al. 2011. The acceptability cline in VP ellipsis. Syntax (14). 318–354. Lakoff, George. 1970. Global rules. Language 46. 627–639. Lasnik, Howard. 2001. When can you save a structure by destroying it? In Minjoo Kim, and Uri Strauss (eds.), Proceedings of the north east linguistic society 31, 301–320. Georgetown University: GLSA. Lasnik, Howard. 2005. The syntax of silence. by Jason Merchant. Language 81(1). 259–265. Lasnik, Howard. 2014. Multiple sluicing in English? Syntax 17(1). 1–20. Legate, JA, and CD Yang. 2002. Empirical reassessment of poverty of stimulus arguments. Reply to Pullum and Scholz. Linguistic Rev 19. 151–162.
28 | Klaus Abels (London)
Lidz, Jeffrey, and S Waxman. 2004. Reaffirming the poverty of stimulus argument: a reply to the replies. Cognition 93. (1)57–165. Lobeck, Anne. 1987. Syntactic constraints on ellipsis. Seattle: University of Washington PhD thesis. Marušič, Franc, and Rok Žaucer. 2013. A note on sluicing and island repair. In Steven Franks (ed.), Annual workshop on formal approaches to slavic linguistics, the third indiana meeting 2012, vol. 59 (Michigan Slavic Materials), 176–189. Ann Arbor: Michigan Slavic Publications. Merchant, Jason. 1999. The syntax of silence - sluicing, islands, and identity of ellipsis. UCSC Ph.D. dissertation. Merchant, Jason. 2001. The syntax of silence: sluicing, islands, and the theory of ellipsis. Oxford: Oxford University Press. Merchant, Jason. 2008. An asymmetry in voice mismatches in vp-ellipsis and pseudogapping. Linguistic Inquiry 39(1). 169–179. Merchant, Jason. to appear. Ellipsis: a survey of analytical approaches. In Jeroen van Craenenbroeck, and Tanja Temmerman (eds.), The Oxford handbook of ellipsis. Oxford: Oxford University Press. Merchant, Jason, and Andrew Simpson (eds.). 2012. Sluicing: cross-linguistic perspectives. Vol. 38 (Oxford Studies in Theoretical Linguistics). Oxford: Oxford University Press. Nevins, Andrew Ira. 2010. Two case studies in phonological universals: a view from artificial grammars. Biolinguistics 4(4). 217–232. Nishigauchi, Taisuke. 1998. ‘Multiple sluicing’ in Japanese and the functional nature of whphrases. Journal of East Asian Linguistics 7(2). 121–152. Perfors, Amy, J Tenenbaum, and T Regier. 2011. Poverty of the stimulus: a rational approach. Cognition 118. 306–338. Pullum, G.K., and B.C. Scholz. 2002. Empirical assessment of stimulus poverty arguments. The Linguistic Review 18(1-2). 9–50. Reali, Florencia, and Morten H. Christiansen. 2005. Uncovering the richness of the stimulus: structure dependence and indirect statistical evidence. Cognitive Science 29. 1007–1028. Rodrigues, Cilene, Andrew Ira Nevins, and Luis Vicente. 2009. Cleaving the interactions between sluicing and preposition stranding. In L. Wetzels, and Jeroen van der Weijer (eds.), Romance languages and linguistic theory 2006: selected papers from “Going romance”, Amsterdam, 7–9 December 2006, 175–198. John Benjamins. Ross, John Robert. 1969. Guess who? In Robert I. Binnick et al. (eds.), Papers from the fifth regional meeting of the chicago linguistic society, 252–286. Chicago: Chicago Linguistic Society, University of Chicago. Sag, Ivan A. 1976. Deletion and logical form. MIT Doctoral dissertation. Scatton, Ernest A. 1993. Bulgarian. In Bernard Comrie, and Greville G. Corbett (eds.), The Slavonic languages (Routledge Language Family Descriptions), 188–248. London & New York: Routledge. Takahashi, Daiko. 1994. Sluicing in japanese. Journal of East Asian Linguistics 3(3). 265–300. Winkler, Susanne. 2013. Syntactic diagnostics for extraction of focus from ellipsis site. In Lisa Lai-Shen Cheng, and Norbert Corver (eds.), Diagnosing syntax (Oxford Studies in Theoretical Linguistics), chap. 23, 463–484. Oxford: Oxford University Press.
Loren A. Billings (Bishkek)
A Pilot Corpus Study of Non-verb Experiencer Predicates in Russian 1 Introduction This paper explores the distribution of several Experiencer-taking predicates in Russian. Whereas Billings (2002) reports about verbs of this kind, the current study examines predicates that are etymologically other parts of speech (namely: nouns and adjectives) that also take Experiencers. The models in Billings (1997) are applied to a twentieth-century novel (M. Bulgakov’s Master i Margarita ‘The master and Margarita’, completed in the 1940s but published much later in the twentieth century). Each predicate’s number of occurrences in various constituent orders is reported, with commentary on whether (based on informant judgments) any of these orders requires marked intonation. I assume that only those word orders with neutral intonation and wide focus indicate a sentence’s neutral constituent order. For each of the predicates under investigation, tokens with marked intonation or narrow foci are eliminated systematically. Any remaining tokens are then used to determine the neutral word order. With some predicates there is not even a single discourse-neutral token. Still, it is possible to draw conclusions about the argument structure from these data. This study’s findings specify the information- and argument-structure properties of such predicates. There is overall agreement in the literature that in a canonical Russian clause with a monotransitive verb the discourse-neutral order is subject-verb-object. (Here, I am referring to a clause consisting of the verb and two non-pronominal
|| Note: This work benefited originally from comments by Gerhild Zybatow, Maaike Schoorlemmer, and especially Uwe Junghanns. A draft of it arose during my work on a research project, during the late 1990s, entitled Argumentstruktur und Wortstellung als Mittel der Informationsstrukturierung im Russischen ‘Argument structure and word order as a means of determining the information structure in Russian’, funded by the Deutsche Forschungsgemeinschaft. In order to reflect the work I contributed to the project, I’ve kept the scope of the study the same. More recently, anonymous comments by referees also improved the paper. I also thank Viktor Sundgren for spotting mistakes and Cynthia Vakareliyska for various constructive comments. Any shortcomings remaining in this study are, of course, solely my own responsibility. The paper is dedicated to the memory of Frank J. Miller, whose death occurred during this paper’s revision in January 2016.
DOI 10.1515/9783110517873-003
30 | Loren A. Billings (Bishkek)
arguments. If at least one of the arguments is a pronoun, the clause is not discourse-neutral.) The main tests rest on intonation and the referentiality of the nominal expressions. A truly neutral order will answer a wide-focus question (such as ‘What happened?’), with both nominal expressions being interpreted as indefinites, and the pitch falling slightly during the stressed syllable of the final prosodic word in the clause. For details, see Junghanns and Zybatow (1997). The facts vary for canonical clauses with ditransitive or intransitive verbs. The picture is more complicated with clauses that contain so-called psych verbs, where one of the arguments encodes the Experiencer role. For instance, Bailyn (2012: 16 fn. 20) writes that psych predicates have received plenty of attention in generative linguistics “but have been little discussed with regard to Russian, except in King (199[3]).” Indeed, in my view, very few of the advances with regard to the phrase structure of Russian can be applied directly to psych predicates, especially nonverbal ones. Thus, it has become necessary to establish the facts descriptively. An experimental study of several classes of psych verbs and the neutral word order of their clauses is reported in Billings (2002). In most verb classes, the neutral order is nominative-verb-other (where other refers to a nominal expression in some non-nominative case or even to a prepositional phrase). Only one class of four verbs, each of which selects a dative-case Experiencer and a nominative-case Theme, does not fit this neutral word order. The facts are even more complicated with psych predicates of non-verbal origin. Billings (1997), a working paper, presents many of the facts. The current study, also originally conducted and written up in 1997 (but not yet published), attempts to clarify the picture using a twentieth-century novel (Bulgakov’s Master i Margarita) as its corpus. Only a handful of non-verb predicates have been studied. The findings are very much preliminary but, I hope, useful in the long run. The three main sections of this paper each deal with one type of non-verb predicate, whereby the order of complexity increases. The first predicate is an etymological noun: pora ‘time’. As a psych predicate, pora occurs in Russian expressions such as ‘it is time for us to go’ (where ‘for us’ renders the Experiencer). Since there is relatively little variation in the word order of the sentences, this predicate is used to lay out the general framework of the study. Section 3 then examines the pair of predicates žal' ‘pity’ and žalko ‘pitiful’. As their glosses suggest, these are etymologically nominal and adjectival, respectively. The greater variation in their structures puts this pair of predicates next in the exposition. Finally, three more Experiencer-taking adjectives that behave alike are discussed
A Pilot Corpus Study of Non-verb Experiencer Predicates in Russian | 31
in section 4. They each encode perceivability: /vjid#n-/ ‘visible’, /sliš#n-/ ‘audible’, and /zamjet#n-/ ‘noticeable’.1 Their discussion is complicated primarily because, as predicates (unlike the predicates treated so far), either these three predicates can agree as adjectives (with a nominative-case Theme) or they do not.
2 A sample predicate: pora This section deals with pora ‘(it is) time’, which occurs twenty-three times in the corpus.2 Of these, the majority of the tokens are of the following, minimal types. To begin, eight occurrences are of pora by itself (the present-tense copula being null): ‘It is time.’ Not exemplified here, these tell us nothing about the ordering of pora relative to other elements. Nine additional tokens are of the type (complementizer +) dative-case personal pronoun + this predicate, as in (1a–c), where the present-tense copula likewise remains null.3 (1)
pora. a. Mne 1SG:DAT time ‘It is time for me.’ (two additional, identically punctuated tokens; two others: [sic] and )
|| 1 The use of the /#/ symbol in these stems (and one other in section 3) represents a position where a vowel appears only in certain environments. 2 Two additional occurrences of pora in the corpus are as real nouns: ‘time’. Both instances occur as the subject of nastavat'/nastat' ‘come to pass (IPFV/PFV)’. Certain fixed expressions involving this noun also take the form {s/do} {six/tex} por (literally ‘{since/until} {these/those}:GEN.PL time(F):GEN.PL’) ‘{since/until} {now/then}’; the corpus is replete with such data (though s six por ‘from now on’ is not found). There are also fixed expressions, such as do pory do vremeni (literally ‘until time(F):GEN.SG until time(N):GEN.SG’) ‘for the time being’ (using both the archaic and the current ‘time’ stems), etymologically formed from this noun but now stored as idiom chunks. This use as a real noun is not discussed further in this study. There is also another noun with the same segments, but different stress: póra ‘pore’ (not found in this corpus), as opposed to this predicate: porá ‘(it is) time’. 3 The transliteration from the Cyrillic Russian used in this paper is a hybrid of those used in North America and Europe. As also codified in Schoorlemmer (1995: xiv), the voiceless velar fricative is x (rather than ch), whereas the mid, unrounded vowel indicating no preceding palatal or palatalized consonant is shown here as ė (rather than as è). It would also only be fitting for this paper to conform to the renowned Leipzig Glossing Rules.
32 | Loren A. Billings (Bishkek)
b. Nam pora. 1PL:DAT time ‘It is time for us.’ (two more tokens differing only slightly in their punctuation) c. čto vam pora. that 2PL:DAT time ‘that it is time for you.’ The data in (1a–c) tell us only that the dative-case Experiencer precedes pora. This leaves only six more tokens of pora with any complexity at all: pora vyxodit' (2) a. čto ej that F.3SG:DAT time go_out(IPFV):INF ‘that it is time for her to go out’ b. čto emu pora zagovorit' na that M.3SG:DAT time begin_speaking(PFV):INF on čelovečeskom jazyke human:M.LOC.SG tongue(M):LOC.SG ‘that it is time for him to start speaking in a human language’ c. pora sobirat'sja time gather(IPFV):INF.REFL ‘it is time to get ready’ d. pora brosit' vse k čertu time throw(PFV):INF all:N.ACC.SG toward devil(M):DAT.SG ‘it is time to send everything to hell’ e. Pora i moroženoe podavat'! time even ice_cream(N):ACC.SG serve(IPFV):INF ‘It is time that the ice cream was served!’ f. v tresnuvšem pencne, kotoroe davno in cracked:N.LOC.SG pince_nez(N):LOC.SG which:N.ACC.SG long_ago pora bylo by vybrosit' na pomojku time be:PST.N.SG SBJV discard(PFV):INF onto trash(F):ACC.SG ‘in the cracked pince-nez, which should have been thrown away long ago’ Note that whereas pora is etymologically the nominative-singular form of a feminine noun, the lack of such feminine-singular agreement in the overt copula in (2f) shows conclusively that pora is not the sentential subject of its clause; bylo is the past-tense copula form (used here as default agreement) that would agree only with a neuter-singular subject.
A Pilot Corpus Study of Non-verb Experiencer Predicates in Russian | 33
The differences in (2) are refined schematically in (2´), omitting items irrelevant to the discussion (such as complementizers and adverbials): (2´) a.
DPDAT
b.
DPDAT
c. d. e. f.
DPACC
.. .. ..
VINF VINF PP VINF VINF DPACC PP DPACC VINF COP VINF PP
The predicate pora itself is indicated here by large dots, vertically aligned for ease of comparison, with any other material either preceding or following it (Luhn 1960). As in the nine tokens of dative-case pronoun + pora exemplified above in (1a–c), any dative-case Experiencers, as shown only in (2´a–b), likewise precede pora. Each of the prepositional phrases – namely, in (2´b, d, f) – is an argument of the infinitival verb, not of pora. The accusative-case nominals in (2´d–f) are likewise the (transitive) infinitival verb’s internal argument. Indeed, pora does not subcategorize for any arguments as such. Its dative-case Experiencer is licensed for case purposes by pora, but this nominal expression is selected semantically as an argument of the infinitival verb. I assume that those instances of pora that are not followed by an infinitival verb overtly, including (1a–c), still do so abstractly.4 As the notation in (2´a–f) indicates, there is very little variation in the orders. An overt infinitival verb follows the predicate, as shown in all six examples. Any overt dative-case Experiencer is initial, as in (2´a–b). Prepositional phrases are final, in (2´b, d, f). The only real variation is the positioning of the accusative-case argument (namely, the direct object of the infinitival verb), bold-faced in each of (2´d–f). Beginning with (2´f), this argument is initial because of relativization, fronting for purely syntactic reasons. In (2´d–e) the accusative-case argument is post-/pre-verbal, respectively; indeed, this is the only variation relevant to information structure in all of the pora data. Informant elicitations have confirmed that the order in (2´e) is possible only with marked prosody: sharply falling intonation on the stressed syllable of the pre-final, accusative-case argument. The
|| 4 Indeed, my translations in (1a–c) are incomplete somehow (or at least overly elliptical). In English at least, they require overt infinitival verb phases to be fully acceptable.
34 | Loren A. Billings (Bishkek)
presence of i ‘even’ in (2e) is linked to this marked intonation; it marks the beginning of a focused constituent. See also (8a–b) and (9m–n) below. In (2´d) there is neutral intonation: slightly falling pitch during the last prosodic word’s stressed syllable. Thus, (2´d) is a candidate for neutral order. Because neither ‘everything’ nor ‘to hell’ can be indefinite, in (2e) an additional referentiality test as suggested in Junghanns and Zybatow (1997: 295) is unavailable. Before concluding the current section, it is worth exploring the syntactic structure of sentences with pora. Certain evidence suggests that pora combines with an infinitival verb in the same clause. Namely, (2f) shows that the relative pronominal kotoroe ‘which’ moves to a position before pora, suggesting that pora is part of the extended projection of the verb, in the same clause. However, other evidence suggests that the infinitival verb is in a clause separate from pora. Once again, (2f) provides the key evidence. Structures with pora do not allow the verb to show tense. Rather, in the past tense bylo ‘be:PST.N.SG’ is inserted between pora and the infinitival verb. (To say ‘it will be time’, a different copula, budet ‘be:FUT.3SG’ is used, again between pora and any infinitival verb. Alas, no such examples were found of budet with pora in this corpus. Once more, there is an overt copula in its default-agreeing form and a decidedly tenseless infinitival verb.) Here I follow Junghanns (1997) in considering the ‘be’ forms that co-occur with pora to be copulas rather than auxiliary verbs. As such, bylo ‘was’ in (2f) occupies the lexical-verb position of its clause, further entailing that the infinitival verb is the lexical head of a separate extended verbal projection, perhaps as the copula’s complement. On balance, I am inclined to consider the latter analysis more tenable, with amalgamation to the extent that the lexical verb’s otherwise-nominative argument is case-marked as dative by pora. Certainly much more research needs to be done to pin down the exact syntactic structure, especially how pora itself is situated in this structure.5 This section has demonstrated how a single form’s possibilities are explored in this pilot corpus study. In the remaining sections successively more complicated non-verbal, Experiencer-taking predicates are examined.
|| 5 Just three of the syntactic analyses that Uwe Junghanns (personal communication) has suggested to me as avenues to explore are a small-clause structure (for the infinitival complement of pora), raising, and control.
A Pilot Corpus Study of Non-verb Experiencer Predicates in Russian | 35
3 An additional pair of related predicates: žal' and žalko The predicates žal' ‘pity’ and žalko ‘pitiful’ occur only eleven times in the corpus. The two forms, žal' and žalko, are, roughly speaking, formal- and colloquial-register synonyms, and can be glossed in a number of ways. All eleven tokens are listed in (3a–k). (3) a. Žal' tol'ko, čto … pity only
that
‘It’s just a pity that …’ b. i
vse že
na Stepu
bylo
žalko
vzgljanut'
and nonetheless at Stepa(M):ACC.SG be:PST.N.SG pitiful look(PRF):INF ‘and nonetheless it was pitiful to look at Stepa’ c. A
žal'!
but pity ‘But what a pity!’ d. I
mne
žal'!
and 1SG:DAT pity ‘And it’s a pity for me!’ e. emu
žal'
bylo
M.3SG:DAT
be:PST.N.SG pity
pogibšego
v
plemjannika
ženy,
nephew(M):ACC.SG wife(F):GEN.SG
rascvete
let.
perished:M.ACC.SG in bloom(M):LOC.SG year(N):GEN.PL ‘he felt sorry for his wife’s nephew, who was cut down in the flower of his youth.’ f. No
mne
žalko
tebja,
but 1SG:DAT pitiful 2SG:ACC
Margo Margo
‘But I feel sorry for you, Margo’ g. i
mne
and 1SG:DAT
ego
očen' žal'!
M.3SG:ACC
very
‘and I feel very sorry for him!’
pity
36 | Loren A. Billings (Bishkek)
h. Mne,
vpročem,
ee
1SG:DAT however
ne očen'
F.3SG:ACC NEG
very
žal'! pity
‘I, however, don’t feel that sorry for her!’ i. a
žalko
kotenočka
but kitten(M):DIM.ACC.SG pitiful ‘but (I) feel sorry for the kitty’ j. O, oh
kak ėto
žalko
how this:N.NOM pitiful:N.SG
‘Oh what a pity this is’ k. No
šljapy
bylo
žalko!
but hat(F):GEN.SG be:PST.N.SG pitiful ‘But (he) felt sorry about losing the hat!’ Using the notation in (2´), the data in (3) can be synthesized as follows. (3´) a. b.
PP COP
c. d.
DPDAT
e.
DPDAT COP
f.
DPDAT
g.
DPDAT DPACC
h.
DPDAT DPACC
i.
DPACC
j.
DPNOM
k.
DPGEN COP
.. .. .. .. .. .
CP VINF
DPACC DPACC
A number of similarities with and differences from (2´a–f), and indeed examples (1a–c), are displayed in (3´a–k). I start with the constituent orders that are alike. First, only in (3´b) the infinitival verb follows the large dot, as in each of (2´a–f) above. Next, whenever there is an overt copula, it precedes the dot: (3´b, e, k), unlike (2´f). I consider the ‘be’ forms with žal'/žalko to be copulas rather than auxiliaries for the same reasons (primarily the inaudible present-tense form). In addition, in (3´d–h), as in (1a–c) and (2´a–b), the dative-case Experiencer precedes
A Pilot Corpus Study of Non-verb Experiencer Predicates in Russian | 37
the dot. A final similarity is that an accusative-case nominal can either precede the dot – in (3´g–i), as in (2´f) – or follow it – in (3´e–f), as in (2´d–e). There are also several differences in the corpus between pora and žal'/žalko. To begin, the accusative-case nominal after the dot in (3´e–f) does not co-occur with any infinitival verb. As a consequence, the distinction between (2´d) and (2´e) above is unavailable here. As for the accusative-case nominals preceding the dot, when there is a dative Experiencer, as in (3´g–h), the dative Experiencer precedes the accusative-case nominal. Next, whereas the only prepositional phrases are final in each of (2´b, d, f), in (3´b) the PP precedes the copula. Finally, there are three elements that co-occur only here and not with pora: nominative and genitive phrases, both before the dot, in (3´j–k), as well as a complementizer phrase, after the dot, in (3´a). Looking within just the žal'/žalko data, there is only one bit of variation in constituent order: the accusative-case nominal following the dot in (3´e–f) but preceding it in (3´g–i). Regarding (3e), the accusative-case phrase is not new to the discourse. (The preceding paragraphs mention this nephew.) Still, the sheer size of this nominal expression – namely: plemjannika ženy, pogibšego v rascvete let ‘(his) wife’s nephew, (who was) cut down in the flower of his youth’ – suggests that something akin to heavy-NP shift in English is going on. Similarly, in (3f) there is a clause-final vocative, which might be influencing the positioning of the co-indexed pronominal argument. In (3h) there is a topic pronoun, while in (3i) there is a noun that is a pre-existing topic. Thus, in none of the tokens with this predicate is there neutral intonation with wide focus. Still, certain conclusions about argument structure and syntax can be drawn from these data. As is well known about Russian, only in one declension do nouns distinguish between nominative, accusative, and genitive cases. Only one such noun is represented in the non-dative nominals; šljapy ‘hat(F):GEN.SG’ in (3k) is unmistakably genitive (in the singular). The remaining non-dative nominals are not of this declension and therefore display syncretism. In (3e–i) the nominals have syncretic accusative and genitive forms, distinct from nominative; in (3j), syncretic nominative and accusative forms, distinct from genitive. From the context, it is clear that the speaker pities the kitten if left to fend for itself. (That is, the speaker is not lamenting the loss or death of the animal.) Thus, ‘kitten’ is in the accusative here. Billings (1997: 8) concludes that uses such as (3b), with an infinitival complement, and (3e–j), with an accusative-case nominal complement, belong to one lexical entry, žal'/žalko ‘feel sorry (for/about)’, whereas (3k) with a genitive-case argument, is a separate entry, žal'/žalko ‘feel sad about losing’. For the current
38 | Loren A. Billings (Bishkek)
study’s purposes, both lexical entries also select a dative-case Experiencer.6 Schoorlemmer (1995: 67 fn. 66) reports an example of the former type with a noun that shows unmistakable accusative case. The discourse-neutral order of the non-dative argument of both žal'/žalko predicates is after this predicate. The non-dative phrases in each of (3f–h, j) are pronominal and thus easily topicalized. From the context in the novel, it is clear that the noun ‘kitten’ in (3i) is also a pre-existing topic. The corpus offers no clear examples but it seems that the non-dative expression’s neutral ordering is after žal'/žalko. Recall that žal'/žalko differs from pora in the position of the copula. Though there is not much evidence, in (3´b, e, k), with the copulas before the dot and (2´f) with the opposite order, I am inclined to consider these examples as representative of the predicates’ respective neutral word orders. As such, with žal'/žalko there is a decidedly different argument structure. Though žal'/žalko can take an infinitival-clause complement, as in (3´b), there is no sense in those other examples that take non-dative nominal expressions – namely, (3´e–k) – that there is an understood, inaudible infinitival verb. Mainly for this reason, I propose that žal'/žalko occupies the head of its extended lexical projection. There are two such lexical predicates. One, žal'/žalko ‘feel sorry (for/about)’, is like most any monotransitive verb. It takes either an accusative-case or infinitival-clause complement. The other, usually just žalko (i.e., not žal') ‘feel sorry about losing’, assigns (probably quirky) genitive case (Billings 1997: 8, including footnotes). Only one more example requires discussion in this section: (3j). Here the only nominal expression bears nominative case (though, alas, syncretic with accusative). Note also the absence of any dative-case Experiencer here. My preliminary assessment of (3j) is that this use of žalko is as a garden-variety predicative adjective (i.e., in its short form and not marked for case). In the corpus there are three instances of attributive /žal#k-/ ‘pitiful’ (i.e., in its long form and case-marked). The instance in (3j) appears to be the only one of the agreeing short form of the same adjectival stem used predicatively.7 To summarize the paper thus far, I have argued that whereas žal'/žalko occupies a position that is the head of its extended lexical projection, pora is a func-
|| 6 The genitive-assigning predicate in (3k) is more common with de-adjectival žalko than with de-nominal žal' (Billings 1997: 6 fn. 9). 7 None of the examples of /žal#k-/ in this study show /#/ as being realized phonetically (i.e., only in the masculine-singular short form). Several examples in section 4 show the /#/ as overtly realized: (9c–f, h–j, o, v).
A Pilot Corpus Study of Non-verb Experiencer Predicates in Russian | 39
tional extension of the extended (lexical, verbal) projection. As Maaike Schoorlemmer has pointed out to me personally, this distribution points up the necessity for a syntactic structure autonomous from the morphological properties of these predicate words. While žal' is de-nominal ‘pity’, žalko is de-adjectival (cf. /žal#k-/ ‘pitiful’), the two function alike syntactically, as a lexical head. Also, denominal pora ‘(it is) time’ functions differently from žal'/žalko in the syntax. The former heads a functional category in the copula’s extended lexical projection; the latter is its clause’s lexical head.
4 Predicates of perceivability: /vjid#n-/, /sliš#n-/, and /zamjet#n-/ The remainder of this paper is devoted to the small class of morphologically adjectival predicates that express perceivability: /vjid#n-/ ‘visible’, /sliš#n-/ ‘audible’, and /zamjet#n-/ ‘noticeable’. Because these three predicates’ syntactic properties are identical, and their meanings are similar, I discuss them together. Like (3j) above, these three can show agreement with a nominative-case subject; this allows for considerably more complexity in the data. In the corpus these three predicates occur as clausal predicates sixty-six times (= 43 + 21 + 2, respectively). I do not consider further the additional attributive, long-form occurrences of these adjectives here. As in the preceding sections, six tokens (all of them of non-agreeing vidno) are from very short sentences that reveal nothing of relevance to the current study. Two tokens (not exemplified here) are with no other word in the clause. Another two tokens are as shown in (4a), where there is an adjunct prepositional phrase. The final two tokens take a final complementizer phrase, as in (4b). (4) a. vidno
po
tjažesti
visible according_to
weight(F):DAT.SG
‘it is visible from its weight’ b. Ved'
vidno,
čto …
after_all visible that ‘Well, it’s plain to see that …’
40 | Loren A. Billings (Bishkek)
In each of (4a–b) there is no nominative-case subject with which the predicate can agree. This is also true of all the data in (5) through (8) below. This entails that the default-agreeing form of the predicate vidno (homophonous with the short form that agrees with a neuter-singular subject) is used. Another eighteen tokens consist of just a non-agreeing predicate and a nonagreeing overt copula (all these copulas being in the past tense). Of these, sixteen (= 5 + 10 + 1) take the pattern (initial coordinator/complementizer +) predicate + copula (+ complement clause), as in (5a–b). All remaining fourteen tokens, not listed, show bylo ‘be:PST.N.SG’ after either vidno or slyšno. (5) a. pričem zametno
stalo,
čto …
besides noticeable become:PST.N.SG that ‘what’s more, it became noticeable that …’ b. i
slyšno
and audible
bylo,
kak …
be:PST.N.SG as
‘and … became audible’ By contrast, only two nearly identical tokens take the order (initial coordinator/complementizer +) copula + predicate (+ complement clause), both of which are listed in (6a–b). Both happen to take stalo vidno ‘became visible’. Note also that the non-‘be’ copula in (5a) and (6a–b), stalo ‘become:PST.N.SG’, corroborates the analysis in this study that the ‘be’ forms are copulas (rather than auxiliaries). (6) a. tak čto since
stalo
vidno,
become:PST.N.SG visible
čto … that
‘since it became clear that …’ b. stalo
vidno,
čto …
become:PST.N.SG visible that … ‘it became apparent that’ Based on informant pronunciations of these tokens, I have determined that the pitch accent in both (5) and (6) is on the predicate, not on the copula; cf. my discussion of (2´d–e) above. Thus, despite there being only two tokens so far of one order and eight times as many tokens of the other order, I conclude that the copula is generated as in (6): before /vjid#n-/ ‘visible’, /sliš#n-/ ‘audible’, or /zamjet#n-/ ‘noticeable’. Some of the remaining data below also contain an overt
A Pilot Corpus Study of Non-verb Experiencer Predicates in Russian | 41
copula. The same characterization applies to those examples. That is, the neutral order is with the copula before these three predicates. Only forty-two tokens remain to be discussed. In twenty of these, the predicate of perceivability (and copula, if it is overt) does not display agreement with a nominative-case subject. Of these twenty, in nine there is clausal negation: (7) a. čtoby so_that
ne
bylo
slyšno.
NEG
be:PST.N.SG audible
‘so that it can’t be heard.’ b. neuželi snaruži ne bylo is_it_so outside
NEG
slyšno
muzyki
be:PST.N.SG audible
voobšče groxota
ėtogo
i
music(F):GEN.SG and
bala?
overall rumble(M):GEN.SG this:M.GEN ball(M):GEN.SG ‘is it really so that outside one couldn’t hear the ball’s music and general commotion?’ c. Konečno ne of_course
bylo
NEG
slyšno, koroleva
be:PST.N.SG audible queen(F):NOM.SG
‘Of course one couldn’t, your majesty’ d. ničego nothing:GEN
ne
bylo
vidno
NEG
be:PST.N.SG
visible
‘nothing was visible’ e. i
počemu obo
and why
mne
ničego
about 1SG:LOC nothing:GEN
ne
bylo
slyšno
NEG
be:PST.N.SG audible
ran'še previously ‘and why wasn’t anything heard about me previously’ f. da
čto-to
ničego
v
and somehow nothing:GEN in
volnax
ne
vidno
wave(F):LOC.PL
NEG
visible
‘and somehow nothing can be seen in the waves’ g. no
procesii
but procession(F):GEN.SG
uže
ne
bylo
already
NEG
be:PST.N.SG visible
‘but the procession was no longer visible.’
vidno.
42 | Loren A. Billings (Bishkek)
h. konca
ėtoj
end(M):GEN.SG this:F.DAT
reke
ne
bylo
vidno
river(F):DAT.SG
NEG
be:PST.N.SG visible
‘this river had no end in sight’ i. i
ėtix
ne bylo
rojalej
and these:GEN grand_piano(M):GEN.PL
NEG
be:PST.N.SG
v
grome …
in
thunder(M):LOC.SG saxophone(M):GEN.PL
slyšno audible
saksofonov …
‘and these grand pianos couldn’t be heard over the thunder … of saxophones …’ Incidentally, (7b–c) constitute a question-answer pair in the corpus (and are translated as such here). In (7), negation immediately precedes the copula (if overt); otherwise, the predicate, only in (7f). All overt copulas here precede the adjectival predicate, as in (6a–b) above. In most of these sentences, there is a nominal expression in the genitive case. The case-bearing elements in each such phrase are shown in bold type. (As is well known, in Russian the internal argument of a negated verb often appears in the genitive.) The genitive phrase precedes the (copula and) predicate in (7d–i); only in (7b) does it follow. Still, from informant pronunciations, (7b) is the order that allows for unmarked intonation (though this example is so long that at least multiple intonation phrases are generally pronounced). Additional adverbial expressions, prepositional phrases, and the like precede or follow the negation-through-predicate string; these are not discussed further. In the remaining eleven tokens of adjectival psych predicates of perceivability that show neither agreement nor negation, eight have a preceding overt copula and either a prepositional phrase or adverbial expression, as in (8a). Incidentally, (8a) is the only token – of any of the predicates discussed in this paper, in fact – containing a future-tense overt copula.8 In two other tokens the overt copula follows the predicate, as in (8b). budet vidno iz samoj (8) a. čto i that even be:FUT.3.SG visible from most:F.GEN.SG
|| 8 One could argue that the copula and predicate in (8a) agree with čto ‘what’ in the nominative case, similarly to (3j) above. If so, then (8a) would belong in (9) below, alongside (9r, u), rather than in (8) as currently shown.
A Pilot Corpus Study of Non-verb Experiencer Predicates in Russian | 43
interesnoj časti interesting:F.GEN.SG part(F):GEN.SG ‘which will even be apparent from its most interesting aspect’ b. Daže v nastupavšix grozovyx sumerkax even in advancing:LOC.PL stormy:LOC.PL dusk(F):LOC.PL vidno bylo, kak … visible be:PST.N.SG as ‘Even in the advancing storms of twilight … was visible’ Note also i and daže in these two examples, respectively. Both are glossed and translated as ‘even’ and signify the beginning of a focused expression; cf. my discussion of (2e), with i ‘even’, above. This concludes the discussion of non-agreeing predicates. (Incidentally, none of the data exemplified or even alluded to so far in this section includes a dative-case Experiencer. This appears to be a coincidence. See also my discussion below.) The remainder of this section (indeed, of this entire paper) is devoted to the same three predicate stems with forms that agree with the clausal subject in gender (only if singular) and number. All twenty-two such examples are listed as follows: (9) a. oni ne byli vidny srazu. 3PL:NOM NEG be:PST.PL visible:PL immediately ‘they were not visible right away.’ b. i nikakie dal'nejšie slova slyšny and no_such:NOM.PL subsequent:NOM.PL word(N):NOM.PL audible.PL byt' ne mogut be:INF NEG can:3PL ‘and no further words can be heard’ c. no prosto ne vsem on byl viden but simply NEG all:DAT.PL M.3SG:NOM be:PST.M.SG visible:M.SG ‘but he simply wasn’t visible to everyone’ d. tak, čto ej byl viden Manež. such that F.3SG:DAT be:PST.M.SG visible:M.SG riding_academy(M):NOM.SG ‘so that she could see the Riding Academy.’ e. no im gorod byl viden počti do but 3PL:DAT city(M):NOM.SG be:PST.M.SG visible:M.SG almost until samyx kraev. most:GEN.PL edge(M):GEN.PL ‘but they could see the city as far as its very edges.’
44 | Loren A. Billings (Bishkek)
f. v pustom zdanii nepremenno byl by in empty:N.LOC.SG building(N):LOC.SG surely be:PST.M.SG SBJV slyšen rezkij signal. audible:M.SG sharp:M.NOM.SG signal(M):NOM.SG ‘throughout the empty building such a harsh sound surely would have been heard.’ g. čto vidny grjaznye belye noski that visible:PL dirty:NOM.PL white:NOM.PL sock(M):NOM.PL ‘that dirty white socks are visible’ h. za nimi stal viden koridor behind 3PL:INST become:PST.M.SG visible:M.SG hallway(M):NOM.SG ‘a corridor became visible beyond them’ i. Teper' on byl jasno viden now M.3SG:NOM be:PST.M.SG clearly visible:M.SG ‘Now he was clearly visible.’ j. no on otčetlivo viden. but M.3SG:NOM distinctly visible:M.SG ‘but he can be seen clearly.’ k. Nataša uže byla vidna vperedi Nataša(F):NOM.SG already be:PST.F.SG visible:F.SG in_front ‘Nataša was already visible up ahead’ l. i v nem vidny byli serebristye and in M.3SG:LOC visible:PL be:PST.PL silvery:NOM.PL plavajuščie tela. swimming:NOM.PL body(N):NOM.PL ‘and through it the swimmers’ silvery bodies were visible’ m. No vse-taki i s nee stolby vidny. but nonetheless even off_of F.3SG:GEN pillar(M):NOM.PL visible:PL ‘But even from there the gibbets are visible.’ n. vidny byli za cep'ju i dva visible:PL be:PST.PL behind chain(F):INST.SG even two:NOM sverkajuščie pjatna na grudi glistening:NOM.PL spot(N):GEN.SG on chest(F):LOC.SG kenturiona centurion(M):GEN.SG ‘behind the chain even two glistening spots on the centurion’s chest were visible’ o. tak čto otčetlivo byl viden každyj such that clearly be:PST.M.SG visible:M.SG each:M.NOM.SG
A Pilot Corpus Study of Non-verb Experiencer Predicates in Russian | 45
p.
q.
r.
s.
t.
u.
v.
remešok sandalii. strap(M):NOM.SG sandal(F):GEN.SG ‘such that each strap of the sandal was clearly visible.’ iz kamorki prjamo byla vidna vyxodnaja out_of closet(F):GEN.SG directly be:PST.F.SG visible:F.SG exit:F.NOM.SG dver' … paradnogo. door(F):NOM.SG main_entrace(N):GEN.SG ‘from the hideout (he) could see the main entrance …’ i soveršenno otčetlivo byla vidna v and completely clearly be:PST.F.SG visible:F.SG in vysote polnaja luna height(F):LOC.SG full:F.NOM.SG moon(F):NOM.SG ‘and a full moon shone clearly from above’ na drugom kraju neba bylo on other:M.LOC.SG edge(M):LOC.SG sky(N):GEN.SG be:PST.N.SG vidno belovatoe pjatnyško utrennej visible:N.SG pale:N.NOM.SG speck(N):NOM.SG morning:F.GEN.SG zvezdy. star(F):GEN.SG ‘at the far edge of the sky the morning star’s whitish speck could be seen.’ vidny za sosednim stolikom visible:PL behind neighboring:M.INST.SG table(M):DIM.INST.SG nalitye krov'ju č'i-to … glaza poured:NOM.PL blood(F):INST.SG someone’s:NOM.PL eye(M):NOM.PL ‘someone’s bloodshot … eyes are visible at the next table over’ Otčetlivo byla vidna na polu distinctly be:PST.F.SG visible:F.SG on floor(M):LOC.SG tenevaja spinka kresla … shadowy:F.NOM.SG back(F):DIM.NOM.SG armchair(N):GEN.SG ‘The shadow cast by the back of the armchair … was distinctly visible on the floor’ V osobennosti ėto bylo zametno v in feature(F):LOC.SG this:N.NOM be:PST.N.SG noticeable:N.SG in otnošenii graždanki, stojavšej … relation(N):LOC.SG citizen(F):GEN.SG standing:F.GEN.SG ‘This was particularly apparent in the woman standing …’ kak-to otčetlivo viden na nej kakoj-to somehow distinctly visible:M.SG on F.3SG:LOC some:M.NOM.SG
46 | Loren A. Billings (Bishkek)
zagadočnyj, temnyj – ne to drakon, не … cryptic:M.NOM.SG dark:M.NOM.SG NEG either dragon(M):NOM.SG NEG ‘some sort of enigmatic, dark thing – neither dragon nor … – is somehow distinctly visible on it’ A number of properties vary in these examples. I discuss each in turn. To begin, overt copulas appear in most of these: (9a–f, h–i, k–l, n–r, t–u). Of these, only in (9b, l, n) does the copula follow the adjectival predicate and in each the prosody is marked, with a pitch accent on the adjectival predicate. In (9b) there is an infinitival copula (forced by the modal) and the word order and intonation are both marked. See also discussion immediately after (6) above. Thus, the neutral order is for the copula to precede these adjectival predicates. In this respect, these three adjectival predicates of perceivability are like žal' ‘pity’/žalko ‘pitiful’ and unlike pora ‘(it is) time’. Note also that in (9f) the subjunctive clitic by is between the copula and the predicate adjective; cf. (2f) above. This clitic can be hosted either by the verb or by the domain’s initial constituent (Wackernagel position). Similarly, in (9i) an adverb appears between the copula and an adjectival predicate. The only remaining word-order variation to be discussed in these data is the position relative to the predicate of the phrase that is glossed as nominative. (And here I ignore the relative order of the copula and the adjectival predicate already addressed in the preceding paragraph.) Given that there is an agreeing predicate in this set of examples, it is not surprising that there is an overt nominative expression in each. The nominative expression precedes the predicate in each of (9a–c, e, i–k, m, u) and follows it in (9d, f–h, l, n–t, v). The majority of the prepredicative nominatives are pronominal expressions of some sort, including ‘no_such’ in (9b) and ‘this’ in (9u). There is also one personal name, in (9k). The two common nouns that precede the predicate, in (9e, m), are already mentioned in the preceding discourse (and translated as definite: ‘the city’ and ‘the gibbets’). By contrast, there are no pronominal nominatives at all among the thirteen subjects that appear after the predicate. All are common nouns except for two, in (9d, l, n): Manež ‘the Riding Academy’, a proper noun (i.e., capitalized in the corpus) but mentioned for the first time in the novel; serebristye plavajuščie tela ‘silvery swimming bodies’, where the swimmers are previously mentioned but with two new modifiers added in this sentence; and a numerically quantified expression that is mentioned for the first time. Thus, the order of the nominative subject relative to the predicate has to do with referentiality. Next, dative-case Experiencers, preceding any other (pro)nominal expression, are found in only (9c–e). See also (3´g–h) above. Thus, in all three sections
A Pilot Corpus Study of Non-verb Experiencer Predicates in Russian | 47
we see that the dative-case Experiencer precedes any other (pro)nominal expression. Since Experiencers are almost exclusively human and thus highly topical, this is no surprise.9 In this entire paper, dative Experiencers precede any other (pro)nominal expression as well as the predicate. Just two more tokens deserve mention in this section: (9r, u). In these two sentences the adjectival predicate appears to agree with a neuter-singular nominal expression. In a footnote, I considered this a possibility in (8a) as well. See also (3j) above, with /žal#k-/ ‘pitiful’ appearing to agree in the same way, with a neuter-singular subject. Alas, there is no way to conclude that there is agreement, because of nominative-accusative syncretism with all inanimate neuter (pro)nouns. Thus, the neuter-singular nominal expressions in (3j) and (9r, u) could also be glossed as “ACC” rather than “NOM” in each of those examples. In this connection, it is interesting that not a single token displays unmistakable accusative case on a nominal expression with a non-agreeing adjectival predicate. This gap is consistent with observations, for example, in Bulaxovskij (1954), also cited in Billings and Maling (1995: 187), regarding the increasing inability in standard Russian of non-verb predicates in general to assign unambiguously accusative case to their complements. To summarize this section, three adjectival predicates of perceivability have been investigated. Their structure appears to match that of the predicate pair in section 2. Any overt copula precedes these predicates, which appear to take a complement that optionally can become the subject.
5 Conclusions This paper has investigated several non-verb psych predicates. The first, pora ‘(it is) time’ is unique in this study in three respects. It is the only predicate that is only a noun etymologically. In addition, it also appears to precede its copula if there is one (though the evidence on this particular point is sketchy). Finally, there appears to be an understood infinitival clause with pora (even if there is no overt infinitival verb after it). The second predicate investigated in this study is žal' ‘pity’/žalko ‘pitiful’, etymologically a noun and an adjective, respectively. I argue that there is a single predicate (the noun in the formal register, the adjective in the colloquial one) that is transitive. Its complement can be a complementizer
|| 9 Incidentally, (9m–n) are sequential sentences in the corpus, where vidny ‘visible:PL’ serves as the topic in (9n), hence the copula following it.
48 | Loren A. Billings (Bishkek)
phrase or an accusative-case nominal. There is also a separate predicate (probably just žalko) that takes a genitive-case, probably inanimate complement ‘feel sorry about losing (something)’. Both predicates can take a dative-case Experiencer (though the only example in the corpus that assigns genitive case does not happen to show such an overt dative Experiencer). The final predicates discussed are adjectives of perceivability. Many of these can take agreeing forms (as can the adjectival variant žalko). Since this corpus study was conducted, in 1997, a number of other works have appeared. I have decided against updating the literature. However, two other studies that grew out of the same research project are worth mentioning here: Junghanns and Zybatow (2009) and Zybatow and Junghanns (1998). As a pilot study, this paper has by necessity left unexplored a number of issues. For instance, what triggers whether an agreeing form (of the four adjectival predicates) is selected? Non-agreeing forms are apparently avoided if unambiguous accusative case would be required. Aside from that factor, this question remains unanswered. In addition, a number of other non-verb psych predicates have not been discussed in this paper. Billings (1997) – citing in part work by Chvany (1974), Kondrashova (1994), and Schoorlemmer (1995) – discusses several of these: ne k licu (literally ‘NEG towards face(N):DAT.SG’) ‘not suitable’, len' (literally ‘laziness(F):NOM.SG’) ‘feel too lazy to …’, neoxota (literally ‘NEG + hunt(F):NOM.SG’) ‘not feel like’, nado and /nadob#n-/ (both) ‘need’, možno ‘can’, nel'zja ‘forbidden’, /dolž#n-/ ‘obligated’, /nuž#n-/ ‘must’, and bol'no ‘painful’ – where len', neoxota, and (in the desired function, of selecting an Experiencer) bol'no occur not at all in the current corpus; ne k licu, just once; /nadob#n-/, twice; and the rest, numerous times. Clearly, much more remains to be investigated in this tiny corner of the Russian language. The current study’s findings have merely scratched the surface.
References Bailyn, John Frederick. 2012. The syntax of Russian. Cambridge, England: Cambridge University Press. Billings, Loren A. 1997. Experiencer non-verb predicates in Russian. ZAS Papers in Linguistics 9. 1–21. Billings, Loren A. 2002. Word order and argument structure of Russian psych predicates. In Peter Kosta & Jens Frasek (eds.), Current approaches to formal Slavic linguistics: Contributions of the second European Conference on Formal Description of Slavic Languages FDSL II, held at Potsdam University, November 20–22, 1997 (Linguistik International 9), 415– 424. Frankfurt am Main: Lang.
A Pilot Corpus Study of Non-verb Experiencer Predicates in Russian | 49
Billings, Loren & Joan Maling. 1995. Accusative-assigning participial (–no/–to) constructions in Ukrainian, Polish, and neighboring languages: An annotated bibliography. Part 1 (A–M) Journal of Slavic Linguistics 3(1). 177–217. Part 2 (N–Z) Journal of Slavic Linguistics 3(2). 396–430. Bulaxovskij, L. A. 1954. Vinitel'nyj padež, zavisimyj ot skazuemyx slov i form na -no, -to. §25 in his Russkij literaturnyj jazyk pervoj poloviny XIX veka, 372. Kiev: Gosudarstvennoe učebnopedagogičeskoe izdatel'stvo. Chvany, Catherine V. 1974. The grammar of dolžen: Lexical entries as a function of theory. In Richard D. Brecht & Catherine V. Chvany (eds.), Slavic transformational syntax (Michigan Slavic Materials 10), 78–122. Ann Arbor: Department of Slavic Languages and Literatures, University of Michigan. Junghanns, Uwe. 1997. On byt' (and byti). In Uwe Junghanns & Gerhild Zybatow (eds.), Formale Slavistik (Leipziger Schriften zur Kultur-, Literatur-, Sprach- und Übersetzungswissenschaft 7), 251–265. Frankfurt am Main: Vervuert. Junghanns, Uwe & Gerhild Zybatow. 1997. Syntax and information structure in Russian clauses. In Wayles Browne, Ewa Dornisch, Natasha Kondrashova & Draga Zec (eds.), Annual Workshop on Formal Approaches to Slavic Linguistics [vol. 4]: The Cornell meeting, 1995 (Michigan Slavic Materials 39), 289–319. Ann Arbor: Michigan Slavic Publications. Junghanns, Uwe & Gerhild Zybatow. 2009. Grammatik und Informationsstruktur. In Sebastian Kempgen, Peter Kosta, Tilman Berger & Karl Gutschmidt (eds.): The Slavic languages: An international handbook of their structure, their history and their investigation (Handbooks of Linguistics and Communication Science 32.1), 684–707. Berlin: Mouton de Gruyter. King, Tracy Holloway. 1993. Russian psych verbs and refining the UTAH. In Joel Ashmore Nevis & Vida Samiian (eds.), Proceedings of the Western Conference on Linguistics, volume 5, WECOL 92, 157–170. Fresno: Department of Linguistics, California State University. Kondrashova, Natalia. 1994. Agreement and dative subjects in Russian. In Sergey Avrutin, Steven Franks & Ljiljana Progovac (eds.), Annual Workshop on Formal Approaches to Slavic Linguistics [vol. 2]: The MIT meeting, 1993 (Michigan Slavic Materials 36), 255–285. Ann Arbor: Michigan Slavic Publications. Luhn, H. P. 1960. Keyword-in-context index for technical literature (KWIC index). American Documentation 11(4). 288–295. Schoorlemmer, Maaike. 1995. Participial passive and aspect in Russian. Utrecht: LEd. Zybatow, Gerhild & Uwe Junghanns. 1998. Topiks im Russischen (Sprache und Pragmatik 47). Lund: Germanistisches Institut der Universität Lund.
Petr Biskup (Leipzig)
Decomposing Prepositional Cases in Russian and Polish 1 Introduction This paper is concerned with prepositional cases in Russian and Polish. It treats prepositional cases on a par with structural cases as a reflection of the operation Agree between -features and Tense-features. The type of the assigned prepositional case is determined by semantic properties of particular heads of the decomposed preposition. There is a correspondence between semantic properties of particular heads and their syntactic features. Syntactic features of heads incorporated into the case assigning head are copied on the prepositional complement by Agree. At the level of PF, these features are spelled out as a case by means of a specific vocabulary insertion rule. This approach derives case properties of simple and complex prepositions as well as adverbial prepositions.
2 Prepositions and their case properties It is known that prepositions can assign different cases and that often one case is used for the stative meaning and another one for the dynamic meaning. This pattern can also be observed in Slavic languages, as shown below for Russian and Polish. Švedova et al. (1980) lists twenty four simple primary prepositions for Russian. Most of them (fifteen) assign one case: bez ‘without’, dlja ‘for’, do ‘to’, iz ‘out’, k ‘towards’, krome ‘except’, nad ‘above’, ot ‘away’, pered ‘in front of’, pred ‘in front of’, pri ‘at’, pro ‘for’, radi ‘for’, u ‘at’, čerez ‘over’; as an illustration consider (1).
|| Note: I would like to thank an anonymous reviewer for his/her very helpful comments. I also wish to thank Danuta Rytel-Schwarz for data judgments.
Decomposing Prepositional Cases in Russian and Polish | 51
(1)
a. do avtomobil-ja to car-GEN.SG ‘to the car’ c. k avtomobil-ju toward car-DAT.SG ‘toward the car’ e. u avtomobil-ja at car-GEN.SG ‘at the car’
b. iz avtomobil-ja out car-GEN.SG ‘out of the car’ d. ot avtomobil-ja away car-GEN.SG ‘away from the car’ f. čerez avtomobiľ over car.ACC.SG ‘over the car’
Seven prepositions assign two cases (v ‘in’, za ‘behind’, mež ‘between’, meždu ‘between’, na ‘on’, o ‘about’, pod ‘under’) and two prepositions assign three cases (po ‘along’, s ‘from’). Except the prepositions mež and meždu, all these prepositions express the difference between the locative/stative and directional/dynamic meaning; consider the stative meaning of the instrumental and locative prepositional phrases in (2a) and (3a) and the dynamic meaning of the accusative prepositional phrases in (2b) and (3b). This difference is confirmed by the (in)compatibility of particular prepositional phrases with stative verbs like ležať ‘lie’, as shown in (2c,d) and (3c,d) for prepositions assigning two cases.1 / za jaščik-ami (2) a. pod under / behind box-INST.PL b. pod / za jaščik-i under / behind box-ACC.PL c. ležať pod / za jaščik-ami lie under / behind box-INST.PL d. *ležať pod / za jaščik-i lie under / behind box-ACC.PL (3) a. v / na / o stol-e in / on / about table-LOC.SG b. v / na / o stol in / on / about table.ACC.SG c. ležať v / na / o stol-e lie in / on / about table-LOC.SG
|| 1 O is incompatible with positional verbs like ležať (which also holds for Polish, as shown below) but it can combine with other stative verbs like pomnit’ ‘remember’: pomnit’ o stole / *pomnit’ o stol.
52 | Petr Biskup (Leipzig)
d. *ležať lie
v / na / o stol in / on / about table.ACC.SG
As to prepositions assigning three cases, consider, for instance, the difference between the stative meaning of the instrumental and locative prepositional phrases in (4a) and (5a) and the dynamic meaning of the genitive and accusative prepositional phrases in (4b) and (5b).2 (4) a. stakan s vod-oj glass with water-INST.SG ‘a glass with water’ b. veter s jug-a wind from south-GEN.SG ‘wind from the south’ (5) a. po priezd-e after arrival-LOC.SG ‘after the arrival’ b. po pojas to waist.ACC.SG ‘up to waist’ A similar pattern can be observed in Polish. Bartnicka et al. (2004) lists seventeen primary prepositions (some authors present a more extensive list of prepositions but they also include compound prepositions, e.g. Skibicki 2007, or they list primary and secondary prepositions together, e.g. Kaleta 1995 and contrastive grammar books like Engel et al. 1999, Rytel-Schwarz et al. 2012).3 Eight of them assign one case: bez ‘without’, dla ‘for’, do ‘to’, ku ‘towards’, od ‘away’, przez ‘through’, przy ‘at’, u ‘at’, six assign two cases: na ‘on’, nad ‘above’, o ‘about’, pod ‘under’, przed ‘in front of’, w ‘in’ and three assign three cases: po ‘on’, z ‘from’, za ‘behind’. As to prepositions assigning one case, consider (6), analogous to the Russian examples in (1). There are three differences: since Polish does not have iz, z is
|| 2 As to the third cases, the preposition s also assigns the approximative accusative and po also assigns dative, which can convey various meanings, e.g., the distributive meaning, the reason meaning, the motion along a surface. 3 Prepositions do not receive much attention in Polish grammars. For instance, grammars like Szober (1957), Grzegorczykowa et al. (1984), Wróbel (2001) and Bąnk (2010) discuss prepositions only in connection with other phenomena like parts of speech, nominal cases, the verbal selection and they do not offer a complete overview.
Decomposing Prepositional Cases in Russian and Polish | 53
used in (6b); in the ‘toward’ meaning in (6c), do must be used instead of ku (samochodowi); and in the ‘at’ meaning in (6e), przy (or koło) must be used instead of u since the adessive u in cases like u samochodu is judged as archaic or bookish (Markowski 2000 et al.). (6) a. do samochod-u to car-GEN.SG ‘to the car’ c. do samochod-u to car-GEN.SG ‘toward the car’ e. przy samochodzi-e at car-LOC.SG ‘at the car’
b. z samochod-u from car-GEN.SG ‘out of the car’ d. od samochod-u away car-GEN.SG ‘away from the car’ f. przez samochód over car.ACC.SG ‘over the car’
Concerning prepositions assigning two cases, consider the accusative-instrumental alternation in (7) and the accusative-locative alternation in (8). Similarly to Russian, instrumental and locative prepositional phrases have the stative meaning and accusative prepositional phrases have the dynamic meaning. (7) a. nad / pod / przed skrzyni-ami above / under / in.front.of box-INST.PL b. nad / pod / przed skrzyni-e above / under / in.front.of box-ACC.PL c. leżeć nad / pod / przed skrzyni-ami lie above / under / in.front.of box-INST.PL d. *leżeć nad / pod / przed skrzyni-e lie above / under / in.front.of box-ACC.PL (8) a. w / na / o stol-e in / on / about table-LOC.SG b. w / na / o stół in / on / about table-ACC.SG c. leżeć w / na stol-e lie in / on table-LOC.SG
54 | Petr Biskup (Leipzig)
d. *leżeć w / na stół lie in / on table-ACC.SG In contrast to Russian, Polish stative prepositional phrases like (7a) can also combine with certain dynamic verbs and denote the result location of the figure argument. This, however, does not mean that the prepositional phrases contain a dynamic meaning (cf. Przybylska 2002). Prepositions assigning three cases also express the difference between the stative and dynamic meaning, as shown by the difference between the locative and instrumental prepositional phrases in (9a), (10a) and (11a) and the accusative and genitive prepositional phrases in (9b), (10b) and (11b). The dynamic meaning of the examples in (b) is based on the presence of a scale, path and trajectory. (9) a. po obiedzi-e b. po kolan-a after lunch-LOC.SG to knee-ACC.PL ‘after the lunch’ ‘knee deep’ (10) a. kolega z jabłk-ami b. jabłka z prowincj-i colleague with apples-INST.PL apples from province-GEN.SG ‘a colleague with apples’ ‘apples from the province’ (11) a. strzelać za dom-em b. strzelać za dom shoot behind house-INST.SG shoot behind house.ACC.SG ‘shoot behind the house’ ‘shoot at sth. behind the house’ Both languages also have complex prepositions consisting of two (or three) prepositions, as in (12) and (13), and adverbial prepositions, consisting of two (or three) prepositions or of a preposition and an element belonging to another word class, as shown in (14) and (15). stol-a b. (12) a. iz-za out-behind table-GEN.SG ‘from behind the table’ (13) a. s-przed stoł-u b. from-in.front.of table-GEN.SG ‘from in front of the table’ (14) a. v-pered-i drugich in-in.front.of-LOC.SG others.GEN.PL ‘in front of the others’
iz-pod stol-a out-under table-GEN.SG ‘from under the table’ z-nad stoł-u from-above table-GEN.SG ‘from above the table’
(R)
(P)
(R)
Decomposing Prepositional Cases in Russian and Polish | 55
b. s-zad-i dom-a from-back-GEN.SG house.GEN.SG ‘from behind the house’ (15) a. po-środk-u stoł-u on-middle-LOC.SG table-GEN.SG ‘in the middle of the table’ b. do-koł-a stoł-u to-circle-GEN.SG table-GEN.SG ‘around the table’
(P)
The prepositional case marker can attach to various categories, for instance, to a noun, as in (16a) and (17a), to an adjective, as in (16b) and (17b), to a preposition, as in (16c), and to a deictic pronoun or adverb, as in (17c). (16) a. v Moskv-u in Moscow-ACC.SG ‘to Moscow’ c. po-sred-i on-amidst-DAT.SG ‘in the middle of’ (17) a. do-koł-a to-circle-GEN.SG ‘around’ c. przed-t-em in.front.of-it-INST.SG ‘earlier’
b. s-vysok-a from-high-GEN.SG ‘from above, haughtily’
(R)
b. do syt-a to sated-GEN.SG ‘to one’s fill’
(P)
In the light of these facts, the question arises how the various case assigning properties and their corresponding meanings are derived in the minimalist approach. The remainder of the paper is structured as follows. In section 3, I decompose prepositions into several projections and discuss syntactic and semantic properties of these projections. I am mainly concerned with spatial prepositions. Section 4 shows how prepositional cases are assigned. It establishes the relation between the meaning of particular projections and their syntactic features and proposes vocabulary insertion rules for particular cases.
56 | Petr Biskup (Leipzig)
3 Case and the internal structure of prepositional phrases 3.1
Dynamic phrases are more complex than stative phrases
According to the literature, dynamic prepositional phrases are more complex than stative prepositional phrases (see e.g. Jackendoff 1983, Bierwisch 1988, Wunderlich and Herweg 1991, Kracht 2002, 2008, Van Riemsdijk and Huybregts 2002, Zhang 2002, Den Dikken 2010). This view is supported by the following data. First, there are complex dynamic prepositions containing a stative preposition, as in (12) and (13), but there are no complex stative prepositions containing a dynamic preposition. In the same vein, dynamic wh-adverbs can be derived from stative wh-adverbs, as shown for temporal adverbs below. On the contrary, there are no examples of stative wh-adverbs derived from dynamic wh-adverbs.4 (18) a. kiedy when ‘when’
b. do kiedy to when ‘till when’
c. od kiedy from when ‘since when’
(P)
The same also holds for complex adverbial prepositions. Dynamic prepositions, as in (19b,c) and (20b,c), contain a stative preposition, as in (19a) and (20a), but there is no stative adverbial preposition containing a dynamic preposition.5 b. v-perёd c. s-pered-i (R) (19) a. pered in.front.of in-in.front.of.ACC.SG from-in.front.of-GEN.SG ‘in front of’ ‘to the front of’ ‘from the front of’ (20) a. koł-o b. do-koł-a c. w-o-koł-o (P) circle- NOM/ACC.SG to-circle-GEN.SG in-about-circle-ACC.SG ‘at’ ‘around’ ‘around’
|| 4 It is not decisive for the argument whether or not the preposition forms one word with the adverb (e.g. in Czech and Slovak, they form one word); what is crucial is that the dynamic element is more complex than the stative element and includes it. 5 With respect to (19b), one might object that it is based on the noun perёd ‘front’ but this is problematic for (19c), where the case ending -i does not go together with the masculine gender of perёd.
Decomposing Prepositional Cases in Russian and Polish | 57
Given this, the dynamic phrase, which encodes the dynamic meaning, embeds the stative phrase, which is responsible for stative meanings, as shown below. (21)
[DynamP Dynam [StatP Stat [DP N ]]]
In addition, given the discussed correspondence between cases and the type of the meaning, locative and instrumental prepositional phrases will have the structure in (22) whereas accusative and genitive prepositional phrase will have the structure in (21). (22)
[StatP Stat [DP N ]]
3.2 The case assigning head The data show that case is determined by the highest head in the prepositional structure. Concretely, complex prepositions like (13), repeated here as (23), assign genitive, which is the case assigned by the dynamic s (in (23a)) and z (in (23b)). The prepositions przed and nad do not assign genitive, only accusative and instrumental. (23) a. s-przed stoł-u from-in.front.of table-GEN.SG ‘from in front of the table’
b. z-nad stoł-u from-above table-GEN.SG ‘from above the table’
Similarly, the Russian complex prepositions iz-za and iz-pod in (12)=(24) assign genitive, the case assigned by the dynamic component iz, but not by the stative za and pod. Za and pod assign accusative and instrumental.6 stol-a (24) a. iz-za out-behind table-GEN.SG ‘from behind the table’
b. iz-pod stol-a out-under table-GEN.SG ‘from under the table’
|| 6 Biskup (2009) argues that the quantificational (delimitative) po is a morphological adjunct in Czech, hence it is not visible for syntactic processes and cannot participate in case assignment. The Russian po-nad and po-za, which assign instrumental and have po in the higher position, and the Polish ponad and poza, which both assign accusative and instrumental, could instantiate the same pattern since neither the Russian po nor the Polish po assigns instrumental.
58 | Petr Biskup (Leipzig)
In the case of adverbial prepositions, case is also determined by the highest head. In the Russian (25), accusative, locative and genitive are assigned by the left prepositions v and s, which can be dynamic (as in (25a,c)) in contrast to the stative pered, which assigns only instrumental. In the Polish examples, the accusatives in (26a,c) could theoretically be assigned by both prepositions, na and o and w and o, but genitive in (26b) can be only assigned by the left do. (25) a. v-perёd b. in-in.front.of.ACC.SG ‘to the front of’ (26) a. na-o-koł-o on-about-circle-ACC.SG ‘around’ c. w-o-koł-o in-about-circle-ACC.SG ‘around’
v-pered-i c. s-pered-i in-in.front.of-LOC.SG from-in.front.of-GEN.SG ‘in front of’ ‘from the front of’ b. do-o-koł-a to-about-circle-GEN.SG ‘around’
The heads Dynam and Stat, however, cannot assign case by themselves because Dynam should assign case exactly when Stat does not assign case and Stat in turn should assign case when Dynam does not project. Unfortunately, Stat does not know whether or not Dynam will merge in the derivation, which is the usual lookahead problem. A related problem is that it is not clear why in certain cases Stat could assign case and in others could not. It also does not help to assume optional -features on the head Stat inducing the case assignment process because again, there should be a dependency between the presence/absence of these features on Stat and the presence/absence of the Dynam head in the derivation. For this reason, I follow Biskup (2009) and assume that case is assigned by some higher head, which has all information relevant to case assignment. Analogously to the verbal domain, there is a head T, which c-commands Dynam and Stat and has unvalued -features and the valued T(ense)-feature. This has the advantage that all cases are treated uniformly as Agree between T-features and -features (which is an extension of Pesetsky and Torrego’s (2004) approach to structural cases). Since the case assigned to the prepositional complement is not identical for all prepositions, as we saw above, the head T must be somehow instructed which case it shall assign. This is ensured by incorporation of the lower prepositional heads into T. The whole case assigning process is shown below.
Decomposing Prepositional Cases in Russian and Polish | 59
(27)
Agree [TPT(unval φ-Fs, val T-F)-Dynam-Stat [DynamPDynam-Stat [StatPStat [DP(val φ-Fs, unval T-F)]]]] incorporation incorporation
Although there is no visible agreement morphology on prepositions in Russian and Polish, there are languages with overt prepositional agreement (Baker 2008, Hagège 2010) and with tensed prepositions (Bowern and Aygen-Tosun 2000, Harlow 2007). Agreement morphology on prepositions can be found, for instance, in Abaza, Abkhaz, Hungarian, Irish, Iwaidjan languages, Jacaltec, Tsakhur, Welsh and tensed prepositions in Titan and Māori. The presence of the TP projection in prepositional phrases is also semantically plausible because the prepositional predication holds at a certain time (cf. von Stechow 2006, 2007 for the presence of a time argument in the meaning of prepositions and Kracht’s 2008 function loc’, which takes an object and a time point and returns the region that the object occupies at the time). To allow embedding of prepositional phrases, I assume that the prepositional TP is selected by the little prepositional head p. Since this projection does not play a role in the case assignment process, I will not include it in the following discussion.
3.3 A more detailed decomposition The prepositional structure is more complex; it has been argued that there is a localizing function that situates the external argument (the figure) in space relative to a neighbourhood region of the internal argument (the ground) (e.g. Lang 1991, Wunderlich and Herweg 1991, Kracht 2008). This localizing function has been argued to be morphologically realized in languages like Japanese and Korean; see Wunderlich (1991, 2012), Bierwisch (1996). The neighbourhood region is often expressed by a special relational word, which Svenonius (2006) calls Axial Part. In accordance with the region semantics, I will use the following meaning for the locative head. (28)
Rx[loc(x,R)]
The function loc localizes the referent of the figure argument (x) with respect to neighbourhood region of the ground argument (R), which is specified by the appropriate preposition.
60 | Petr Biskup (Leipzig)
It has been proposed that dynamic prepositional phrases contain the operator become (or change); see, for instance, Dowty (1979), Wunderlich (1991), Stiebels (1996). This operator identifies the transition from one region into the other and takes the final state as its argument. Therefore, it comes in two types, for goal prepositions, consider (29a) and for source prepositions (29b). (29) a. Px[become(P(x))] b. Px[become(P(x))] The become operator embeds the phrase projected by the function loc, hence the prepositional structure of dynamic prepositions looks like (30). For simplicity, I use labels related to semantic properties of particular projections. Thus, BecomeP replaces the former dynamic phrase and the projection Reg(ion)P replaces the stative projection. The localizing function projects LocP above RegP. (30)
[TP T [BecomeP Become [LocP Loc [RegP Reg [DP N]]]]]
As an illustration consider the LF in (31), with the Russian dynamic prepositional phrase iz doma ‘out of the house’. (31)
TP xt[at(become(loc(x,int(y[house(y)]))),t)] 3 Pxt[at(P(x),t)] T BecomeP x[become(loc(x,int(y[house(y)])))] 3 Px[become(P(x))] Become LocP x[loc(x,int(y[house(y)]))] 3 Rx[loc(x,R)] Loc RegP int(y[house(y)]) 3 y[int(y)] Reg DP y[house(y)] iz dom-a
After applying the meaning of Reg to the DP complement, we receive the internal region of the house. The localizing function situates the referent of the figure argument x in that region. The figure, however, merges later in the derivation since it does not intervene between the case assigning T and the ground argument. At the earliest, it merges in Spec,TP; if it is p that is responsible for its introduction, then it merges in Spec,pP (e.g. Svenonius 2003, Biskup and Putnam 2012). Applying the operator Become, the resulting meaning is that it becomes true that the referent of x is not located in the internal region of the house. The head T then
Decomposing Prepositional Cases in Russian and Polish | 61
relates this meaning to a certain time, typically to the reference time introduced by the verbal predicate. Prepositional complements do not have to be overtly realized, as shown by the following examples. Although they do not contain an overt noun, they contain a case ending. (32) a. na lew-o (P) b. po-sred-i (R) c. s-pered-i (R) on left-ACC.SG on-amidst-DAT.SG from-in.front.of-GEN.SG ‘on the left’ ‘in the middle of’ ‘from the front of’ There are several arguments for the presence of a covert noun in examples like these. Firstly, from a theoretical point of view, since case is a reflection of the operation Agree between -features, there must be an element in these examples that has valued -features. Secondly, adverbial prepositional phrases like speredi refer to a certain place; hence we expect a referential element (e.g. a covert noun place) in such phrases. This is supported by the existence of prepositions like vmesto ‘in place of’, in which the place noun (mesto) is present overtly. In na lewo in (32a), which has the neuter accusative ending, the noun is not visible but there are adverbial prepositional phrases in which a neuter noun is expressed overtly; consider the Polish około ‘about’, containing the neuter noun koło ‘wheel’. The next argument is based on case properties of adverbial prepositions. They mostly assign genitive and we know that genitive is the case of complements of nouns.
4 Case in the derivation I assume that there is a correspondence between semantic properties of heads in prepositional phrases and their syntactic features and that syntactic features of heads incorporated into T represent the value(s) of the T-feature on T. These values are copied on the prepositional complement by Agree. At the level of PF, the values are spelled out as a case by means of the vocabulary insertion operation. This proposal has the advantage that the relation between prepositions and their cases is not accidental since case is based on semantic properties of particular heads in the decomposed prepositional phrase (which is in accord with the Jakobsonian approach 1936/1971 to cases, in which cases are bundles of semantic features). To be more concrete, in (31), the internal meaning of Reg corresponds to the syntactic internal-feature and the meaning of the negative Become corresponds
62 | Petr Biskup (Leipzig)
to the syntactic become-feature and source-feature.7 The presence of the general become-feature and the specific source-feature or goal-feature on Become allows us to have just one vocabulary insertion rule for accusatives (among other things), as we will see below. Since source prepositions (see iz, ot in (1b,d) and z, od in (6b,d)) assign genitive, I propose the following vocabulary insertion rule.8 (33)
[become, source] genitive
Now let us look at some stative prepositions. For instance, the meaning of the Russian and Polish za is shown below in the Russian za domom ‘behind the house’. (34)
TP xt[at(loc(x,ext(y[house(y)],-obs)),t)] 3 Pxt[at(P(x),t)] T LocP x[loc(x,ext(y[house(y)],-obs))] 3 Rx[loc(x,R)] Loc RegP ext(y[house(y)],-obs) 3 y[ext(y,-obs)] Reg DP y[house(y)] za domom
The referent of the figure is located in the ext(ernal) region behind (-obs) the house. Obs represents the observer axis and +obs would be used for the ‘in front of’ meaning. The external meaning of Reg corresponds to the syntactic externalfeature and the +/-obs axis (and the vertical axis for the Russian and Polish nad and pod) corresponds to the syntactic projective-feature. Since projective prepositions (nad ‘above’, pod ‘under’, pered ‘in front of’, za ‘behind’/nad, pod, przed, za) assign instrumental case, I assume the following vocabulary insertion rule. (35)
[projective] instrumental
|| 7 To keep the correspondence between semantic properties and syntactic features of the particular heads as simple as possible, I use only privative features. 8 Because of lack of space, I use only simplified vocabulary insertion rules; instead of particular vocabulary items, I put case in the rules. With particular markers, the rules would be more complex because of inner-paradigmatic and trans-paradigmatic syncretism.
Decomposing Prepositional Cases in Russian and Polish | 63
As to the stative na ‘on’, o ‘about’, po ‘along’, v ‘in’/ na, o, po, w, they can be taken to denote a contact between the figure and the ground, in addition to their specific meaning, see (3), (5), (8) and (9); hence the following vocabulary insertion rule can be used.9 (36)
[contact] locative
The dynamic za/za (modulo the Russian pod and Polish nad, pod, przed) has the same LF as the stative za in (34), with the difference that the head Become is present. The positive become meaning corresponds to the syntactic become-feature and goal-feature, which suggests the following vocabulary insertion rule. (37)
[become, goal] accusative
Given the two features, accusative markers are more specific than instrumental markers (see (35)) and fit better in the dynamic feature specification in the instrumental-accusative alternation examples. The rule in (37) can also be used with the same effect for na, o, po, v/na, o, po, w, which alternate between locative and accusative, and also for the accusative preposition čerez ‘over’/przez. This is in line with the fact that goal prepositions mostly assign accusative and with the claim that accusative is characterized as indicating the goal (Van Schooneveld 1986). There is also the goal preposition do ‘to’/do, which assigns genitive. Given the other rules, it is not possible to assume one vocabulary insertion rule for this goal genitive and the source genitive (e.g. [become] genitive). Since with do the referent of the figure argument does not have to end in the ground – it can be located just near the ground – Reg has a proximity-feature. Because of the Specificity Condition, the vocabulary insertion rule in (38) then correctly prevents accusative markers from appearing on the complement of do. (38)
[become, goal, proximity] genitive
As to dative case, it is assigned by the preposition k ‘towards’/ku. With this preposition the figure argument is oriented with respect to the ground argument, which leads to the following vocabulary insertion rule. (39)
[oriented] dative
|| 9 The locative preposition pri ‘at’/przy could also be analyzed in this way.
64 | Petr Biskup (Leipzig)
Complex prepositions like the Russian iz-za ‘from behind’ and the Polish sprzed ‘from in front of’ have a more complex structure; consider (40), with the Polish sprzed domu ‘from in front of the house’. The referent of the figure moves out of the region that is in front of the house. (40)
TP xt[at(become(loc(x,int(ext(y[house(y)],+obs)))),t)] 3 T BecomeP Pxt[at(P(x),t)] x[become(loc(x,int(ext(y[house(y)],+obs))))] 3 Become LocP Px[become(P(x))] x[loc(x,int(ext(y[house(y)],+obs)))] 3 Loc RegP Rx[loc(x,R)] int(ext(y[house(y)],+obs)) 3 s Reg RegP R[int(R)] ext(y[house(y)],+obs) 3 przed Reg DP domu y[ext(y,+obs)] y[house(y)]
The external meaning of the lower Reg corresponds to the syntactic external-feature and -obs corresponds to the projective-feature. The internal meaning of the higher Reg corresponds to the syntactic internal-feature and the meaning of the negative Become corresponds to the syntactic become-feature and source-feature. Since given the Specificity Condition, the vocabulary item with the highest number of matching features is inserted into the terminal, the genitive marker (see rule (33)) wins over the instrumental marker (rule (35)). Analogously, we can analyse adverbial prepositions like the Russian speredi ‘from the front of’, with the difference that, in contrast to (40), the DP is covert and embeds a genitive complement. In this way, we account for the observation that it is the higher preposition that determines case. A comparison of the proposed vocabulary insertion rules shows that dynamic case markers are more specific than stative case markers. For the sake of clarity, I will now present a complete derivation of the Russian v Moskvu ‘to Moscow’. The syntactic derivation of this phrase, with the complex
Decomposing Prepositional Cases in Russian and Polish | 65
T head and appropriate features, is shown in (41a). Given the ordering of prepositions in cases like iz-za ‘from behind’ (R), sprzed ‘from the front of’ (P), znad ‘from above’ (P), speredi ‘from the front of’ (R) and the fact that Become is higher than Reg, incorporation must happen to the right. The T-feature of the DP is valued as [become, goal, internal, contact] via Agree with the complex T. (41) a.
TP 3 T BecomeP 3 3 T Become Become LocP 3 3 Become Loc Loc RegP [become] 3 3 [goal] Loc Reg Reg DP [become, goal, int, cont] v Moskv [int] [cont]
b.
v
Moskv ← -u
The PF of v Moskvu is in (41b). Given the vocabulary insertion rule in (37) [become, goal] → accusative, the accusative marker (for the second declension -u) is suffixed to the noun. The LF of v Moskvu is shown below; the superscript C is a shorthand for the contact relation between the referents of the figure and ground. The positive meaning of Become corresponds to the syntactic become-feature and goal-feature and the internal, contact meaning of Reg corresponds to the syntactic internal-feature and contact-feature. (42)
TP xt[at(become(loc(x,intC(Moskva))),t)] 3 Pxt[at(P(x),t)] T BecomeP x[become(loc(x,intC(Moskva)))] 3 Px[become(P(x))] Become LocP x[loc(x,intC(Moskva))] 3 Rx[loc(x,R)] Loc RegP intC(Moskva) 3 y[intC(y)] Reg DP Moskva v Moskv-u
66 | Petr Biskup (Leipzig)
Generally, case markers are spelled out in accordance with the linearized syntactic structure, that is, on the closest overt element. If the noun is covert, the marker is suffixed to the closest non-nominal element, for instance, to a modifier present in the DP like in the Polish na lewo ‘on the left’. If there is no modifier, the case marker is suffixed to the closest preposition (merged as Reg) like in the Russian posredi ‘in the middle of’.
5 Conclusion We have seen that prepositional cases have a meaning; they are based on semantic properties of particular prepositional heads incorporated into the case assigning head T. Syntactically, prepositional cases result from the operation Agree between -features and Tense-features of the head T and the prepositional complement. The prepositional complement can be overt as well as covert. In either instance, the case marker is spelled out on the closest overt element in the prepositional phrase. I dealt mostly with spatial prepositions but if it is correct that spatial meanings are the underlying meanings of prepositions and other prepositional meanings are based on them, then the proposed analysis can be extended to all prepositions.
References Baker, Mark C. 2008. The syntax of agreement and concord. Cambridge: Cambridge University Press. Bąnk, Piotr. 2010. Gramatyka języka polskiego: Zarys popularny. Warszawa: Wiedza Powszechna. Bartnicka, Barbara, Björn Hansen, Wojtek Klemm, Volkmar Lehmann & Halina Satkiewicz. 2004. Grammatik des Polnischen. München: Otto Sagner. Bierwisch, Manfred. 1988. On the grammar of local prepositions. In Manfred Bierwisch, Wolfgang Motsch & Ilse Zimmermann (eds.), Studia grammatica 29, 1–65. Berlin: AkademieVerlag. Bierwisch, Manfred. 1996. How much space gets into language? In Paul Bloom, Mary A. Peterson, Lynn Nadel & Merrill F. Garrett (eds.), Language and Space, 31–76. Cambridge, MA: The MIT Press. Biskup, Petr. 2009. The syntactic structure of PPs. In Mojmír Dočekal & Markéta Ziková (eds.), Czech in Formal Grammar, 9–29. München: Lincom. Biskup, Petr & Michael Putnam. 2012. One P with two Spell-Outs: the ent-/aus-alternation in German. Linguistic Analysis 38 (1-2). 69–109.
Decomposing Prepositional Cases in Russian and Polish | 67
Bowern, Claire & Gülşat Aygen-Tosun. 2000. Titan's Tensed Prepositions. In Arika Okrent & John P. Boyle (eds.), Proceedings of the 36th Chicago Linguistics Society. Vol. 2, 35–48. Chicago, IL: Chicago Linguistic Society. den Dikken, Marcel. 2010. On the functional structure of locative and directional PPs. In Guglielmo Cinque & Luigi Rizzi (eds.), Cartography of syntactic structures, 74–126. New York: Oxford University Press. Dowty, David R. 1979. Word meaning and Montague grammar: the semantics of verbs and times in generative semantics and in Montague’s PTQ. Dordrecht: Reidel. Engel, Ulrich, Danuta Rytel-Kuc, Lesław Cirko, Antoni Dębski, Alicja Gaca, Alina Jurasz, Andrzej Kątny, Paweł Mecner, Izabela Prokop, Roman Sadziński, Christoph Schatte, Czesława Schatte, Eugeniusz Tomiczek & Daniel Weiss. 1999. Deutsch-polnische kontrastive Grammatik. Band 2. Warszawa: Wydawnictwo Energeia. Grzegorczykowa, Renata, Krystyna Kallas, Krystyna Kowalik, Roman Laskowski, Alicja Orzechowska, Jadwiga Puzynina & Henryk Wróbel. 1984. Gramatyka współczesnego języka polskiego. Morfologia. Warszawa: Państwowe Wydawnictwo Naukowe. Hagège, Claude. 2010. Adpositions. Oxford: Oxford University Press. Harlow, Ray. 2007. Māori: A linguistic introduction. Cambridge: Cambridge University Press. Jackendoff, Ray. 1983. Semantics and cognition. Cambridge, MA: MIT Press. Jakobson, Roman. 1936/1971. Beitrag zur allgemeinen Kasuslehre: Gesamtbedeutungen der russischen Kasus. In Roman Jakobson. Selected Writings II: Words and Language, 23–71. The Hague: Mouton. Kaleta, Zofia. 1995. Gramatyka języka polskiego dla cudzoziemców. Kraków: Uniwersytet Jagielloński. Kracht, Marcus. 2002. On the semantics of locatives. Linguistics and Philosophy 25 (2). 157– 232. Kracht, Marcus. 2008. The fine structure of spatial expressions. In Anna Asbury, Jakub Dotlačil, Berit Gehrke & Rick Nouwen (eds.), Syntax and semantics of spatial Ps, 35–62. Amsterdam & Philadelphia: John Benjamins. Lang, Ewald. 1991. A two-level approach to projective prepositions. In Gisa Rauh (ed.), Approaches to Prepositions, 127–167. Tübingen: Gunter Narr Verlag. Markowski, Andrzej (ed.). 2000. Nowy słownik poprawnej polszczyzny. Warszawa: Państwowe Wydawnictwo Naukowe. Pesetsky, David & Esther Torrego. 2004. Tense, case, and the nature of syntactic categories. In Jacqueline Guéron & Jacqueline Lecarme (eds.), The syntax of Time, 495–539. Cambridge, MA: The MIT Press. Przybylska, Renata. 2002. Polisemia przyimków polskich w świetle semantyki kognitywnej. Kraków: Universitas. van Riemsdijk, Henk & Riny Huybregts 2002. Location and locality. In Marc van Oostendorp & Elena Anagnostopoulou (eds.), Progress in Grammar: Articles at the 20th Anniversary of the Comparison of Grammatical Models Group in Tilburg, 1–23. Amsterdam: Meertens Instituut. Rytel-Schwarz, Danuta, Alina Jurasz, Lesław Cirko & Ulrich Engel. 2012. Deutsch-polnische kontrastive Grammatik. Bd. 4: Die unflektierbaren Wörter. Hildesheim, Zürich & New York: Georg Olms Verlag. van Schooneveld, Cornelis H. 1986. Jakobson’s case system and syntax. In Richard D. Brecht & James S. Levine (eds.), Case in Slavic, 373–385. Columbus: Slavica. Skibicki, Monika. 2007. Polnische Grammatik. Hamburg: Helmut Buske Verlag.
68 | Petr Biskup (Leipzig)
von Stechow, Arnim. 2006. Spatial prepositions in interval semantics. Paper presented at the Semantics Network Workshop, Barcelona, September 2006. von Stechow, Arnim. 2007. Syntactic and lexical causativization: CAUSE and BECOME again. Draft 10.1.2007. http://www.sfs.uni-tuebingen.de/~astechow/Handouts/StechowBruessel.pdf Stiebels, Barbara. 1996. Lexikalische Argumente und Adjunkte: zum semantischen Beitrag von verbalen Präfixen und Partikeln. Berlin: Akademie Verlag. Svenonius, Peter. 2003. Limits on P: Filling in Holes vs. Falling in Holes. Nordlyd 31 (2). 431– 445. Svenonius, Peter. 2006. The emergence of Axial Parts. Nordlyd 33 (1). 49–77. Szober, Stanisław. 1957. Gramatyka języka polskiego. Warszawa: Państwowe Wydawnictwo Naukowe. Švedova, Natalja. Ju. (ed.). 1980. Russkaja grammatika. Moskva: Nauka. Wróbel, Henryk. 2001. Gramatyka języka polskiego: podręcznik akademicki. Kraków: Od nowa. Wunderlich, Dieter. 1991. How do prepositional phrases fit into compositional syntax and semantics? Linguistics 29 (4). 591–621. Wunderlich, Dieter. 2012. Lexical decomposition in grammar. In Markus Werning, Wolfram Hinzen & Edouard Machery (eds.), The Oxford Handbook of Compositionality, 307–327. Oxford: Oxford University Press. Wunderlich, Dieter & Michael Herweg. 1991. Lokale und Direktionale. In Arnim von Stechow & Dieter Wunderlich (eds.), Semantik. Ein internationales Handbuch der zeitgenössischen Forschung (= HSK 6), 758–785. Berlin & New York: de Gruyter. Zhang, Niina Ning. 2002. Movement within a spatial phrase. In Hubert Cuyckens & Günter Radden (eds.), Perspectives on prepositions, 47–63. Tübingen: Max Niemeyer.
Kristina Gehring (Kaliningrad)
Semantic and Morphosyntactic Features of Verbal Prefixes: A Case Study of the Russian Prefix pere- ‘over’ 1 Introduction The article focuses on verbal prefixation in Russian, in particular on the prefix pere- ‘over’. Verbal prefixation in Russian - as well as in other Slavic languages has come into focus of formal linguistic research that regards prefixation as a derivational process in the domain of syntactic structure (see e.g. Babko-Malaya 1999, 2003; Biskup 2007, 2008, 2012; Gehrke 2008; Markova 2011; Ramchand 2004; Romanova 2004, 2006; Svenonius 2004; Tatevosov 2008, 2009; Žaucer 2009, 2012, 2013). It has been widely recognised that verbal prefixes form a heterogeneous class with respect to their semantic and syntactic properties. Proponents draw a line between i) lexical, or internal prefixes and ii) superlexical, or external prefixes. For Russian, the distinction had been already established by Isačenko (1962). He divided Russian verbal prefixes into qualifying and modifying ones, whereby the former correspond to lexical, and the latter to superlexical prefixes. More recent work proposes the existence of a third group, called intermediate prefixes (see Tatevosov 2008, 2009). In a nutshell, these groups of prefixes differ with respect to their position in the syntactic derivational structure. However, in recent literature one cannot find a general consensus on how the lexical-superlexical-distinction can be represented syntactically (for an overview see Tatevosov 2008: 425–427). Nonetheless, all these different proposals agree on that lexical prefixes have to originate inside the verbal phrase (VP), while superlexical prefixes are located outside, higher up in the syntactic derivation. In the following this approach will be called the Prefix-Split-Theory. With this in mind, the current paper shows that pere- ‘over’ supports the idea of a more fine-grained distinction of verbal prefixes (see Tatevosov 2008, 2009). It investigates the prefix according to its status as a lexical, superlexical and/or intermediate prefix. The article is organised as follows: section 2 presents six selected readings of pere- (see table 1), the possible distinguishing criteria for lexical and superlexical
DOI 10.1515/9783110517873-005
70 | Kristina Gehring (Kaliningrad)
prefixes (see table 2) as well as the current state of research with respect to pere‘over’ against the background of the Prefix-Split-Theory (see table 3). The arguments for a tripartite classification of pere- ‘over’ are demonstrated in section 3. First, the reader gets a short introduction to the chosen distinguishing criteria, and then, these criteria are applied to certain occurrences of pere-prefixed verbs. Thus, general assumptions in the relevant literature concerning verbal prefixation are either veri- or falsified. It will be shown that several assumptions based on single readings of pere- ‘over’ (e.g. the excessive reading) are much too superficial and have to be revised. Backed by the findings of Tatevosov (2008, 2009), this article illustrates that instances of pere- ‘over’, some of which are usually considered to belong to the class of superlexicals, carry characteristics more similar either to lexical prefixes (e.g. the duration-reading), or intermediate ones (e.g. the excessive-reading in Romanova 2006). The main results of this paper are summarised in the conclusion in section 4.
2 Status quo According to Flier (1975, 1985), the Russian prefix pere- ‘over’ has twelve different readings, of which six have been chosen for this investigation (see table 1). Looking at grammars and dictionaries of Russian, one can see that the assumed number of meanings ranges from six (Russkaja Grammatika 1960) to fourteen (see Kuznecov 1998). For the present analysis the most prominent ones have been picked (see table 1). Tab. 1: Selected readings of Russian pere- ‘over’ to be investigated.
Reading
Russian example
English translation
translocation-reading
pere-katit’
‘to roll over’
(b)
duration-reading
pere-ždat’
‘to wait out’
(c)
repetitive-reading
pere-izdat’
‘to republish’
(d)
distributive-reading
pere-lomat’
‘to break one by one’
(e)
excessive-reading
pere-varit’
‘to overcook’
(f)
superiority-reading
pere-chitrit’
‘to outwit’
(a)
To investigate the above readings in detail, a list of lexical, morphological and syntactic criteria (see table 2) taken from the recent literature has been compiled.
Semantic and morphosyntactic features of the Russian verbal prefix pere- ‘over’ | 71
According to Gehrke 2008, Romanova (2004, 2006), Svenonius 2004 and Tatevosov (2008, 2009, 2011), these criteria make it possible to distinguish between prefixes located inside or outside of the VP, since these linguistic tests operate on different structural layers. Tab. 2: Criteria to distinguish between lexical and superlexical prefixes.
Distinguishing criteria
lexical
superlexical
semantic (non)-compositionality
either or
strictly compositional
(b)
influence on the argument structure
yes
no
(c)
selection of unselected objects
yes
no
(d)
change in case-marking restrictions
yes
no
(e)
telicity
yes
no
(f)
derivation of past passive participles
yes
no
(g)
multiple prefixation & stacking
stem-adjacent
non stem-adjacent
(h)
selectional restrictions
[+perfective]
[−perfective]
(a)
As one can see from table 3, Russian pere- ‘over’ – and its equivalents in other Slavic languages, indicated by round parentheses – have been the subject of research several times. However, a closer look shows that the given readings of pere- ‘over’ have not all been analysed thoroughly: while the duration- and superiority-readings have not been analysed at all, pere-repetitive as well as pere-excessive have been investigated in several papers, but with different results. In contrast to the aforementioned, pere-translocation is unanimously treated as a lexical and pere-distributive as a superlexical prefix. They do not show any striking features. Thus, for the following analysis the prefixed verbs of the duration-, repetitive-, excessive- and superiority-readings are of primary interest, since they have either not been discussed at all, or analyses have led to contradictory results. Even though the status of pere-distributive as a superlexical prefix is not controversial, it will be discussed, because its behaviour gives information about the structural position of pere-repetitive and excessive (see section 3).
72 | Kristina Gehring (Kaliningrad)
Tab. 3: Current state of research: Classification of pere-prefixed verbs in the recent literature.
Reading
lexical
superlexical
intermediate
translocation
Svenonius 2004; Romanova 2006
/
/
duration
(Žaucer 2009)1
(Borik 2002; Romanova 2004)2
/
repetitive
LeBlanc 2010; Babko-Malaya 1999
Svenonius 2004
Tatevosov 2008
distributive
/
Romanova 2004, 2006
/
Romanova 2004, 2006: vP-external degree modifier
/
excessive
Žaucer 2013: VP internal result modifier superiority
/
/
/
The following presents several criteria that help classify a prefix either as a lexical, superlexical or intermediate prefix. Verbs of the translocation-reading will not be discussed, because pere-translocation undoubtedly belongs to the cluster of lexical prefixes that originate inside of the VP.
3 Pere-prefixed verbs and their distinguishing criteria in detail The following distributional tests are mainly taken from the recent literature on verbal prefixation (see Gehrke 2008; Romanova 2004, 2006; Svenonius 2004; Tatevosov 2008, 2009). This section concentrates on eight selected criteria: (i) semantic (non)-compositionality, (ii) argument structure augmentation, (iii) selection of unselected objects, (iv) changes in case marking restrictions, (v) telicity, (vi) derivation of past passive participles, (vii) multiple prefixation and stacking and (viii) selectional restrictions.
|| 1 Žaucer (2009) analyses the Slovenian equivalent pre- ‘over’ of Russian pere- ‘over’ in its duration reading. 2 Borik (2002) and Romanova (2004) investigate the Russian durative prefix pro- ‘through’ which is partly similar to pere- ‘over’ in its semantics.
Semantic and morphosyntactic features of the Russian verbal prefix pere- ‘over’ | 73
3.1 Semantic (non-)compositionality Superlexical prefixes differ from lexical ones in their semantic properties. Lexical prefixes tend to have spatial or idiosyncratic meanings, superlexicals express temporal, quantifying, adverbial or measure-like meanings. In contrast to lexical prefixes, they combine with the semantics of a verbal stem in a predictable and systematic way; thus, the meaning of a superlexical-prefixed verb can always be deduced compositionally and does not allow idiosyncrasy (see e.g. Romanova 2004; Svenonius 2004; Tatevosov 2008). Pere-verbs of the duration-reading are said to have temporal meanings similar to perdurative pro-prefixed verbs (e.g. pro-sidet’ vsju noč – ‘to pull an allnighter’). However, those verbs cannot be interpreted strictly compositional as is the case for verbs with superlexical prefixes; prefix verbs like pere-žit’ ‘to survive’ in (1) or pere-sidet’ ‘to sit out’ in (2) cannot be interpreted according to the meaning of the base verbs žit’ ‘to live’ or sidet’ ‘to sit’ (see Kagan 2013: 496-500). Thus, these verbs fulfil a criterion which is actually attributed to lexical prefixes; and this is not an exception. In the upcoming tests it will become obvious that pereduration shares a lot of features with lexical prefixes. (1)
pere-žit’
duševn-yj
krizis
over-live
soul-ACC
crisis[ACC]
‘to survive a mental crisis’ (2)
pere-sidet’ ėpidemi-ju over-sit
epidemic-ACC
‘to sit out an epidemic disease’ Recent work in the Prefix-Split-Theory ascribes pere-repetitive a fully transparent meaning, similar to the English adverb again (see Tatevosov 2008: 429), for example pere-čitat’ knigu ‘to read the book again’ in (3). Because of its compositional meaning, the prefix is associated either with the class of superlexical (see Svenonius 2004) or intermediate, precisely positionally restricted prefixes3 (see Tatevosov 2008, 2009). Still, examples like pere-stroit’ dom ‘to rebuild the house’ in (4) show that the meaning of repetitive-verbs cannot only be computed compositionally. To interpret verbal phrases as in (4) properly, one needs additional assumptions: in case of creation-verbs with incremental Theme-arguments like || 3 Tatevosov (2009) divides the group of superlexical prefixes into ‘positionally restricted’ and ‘selectionally restricted’ prefixes.
74 | Kristina Gehring (Kaliningrad)
stroit’ dom – ‘to build a house’ in (4) one will always receive an interpretation of the repetitive-reading including a change of the Theme-argument. It is not possible to interpret such a verbal phrase as a mere repetition of the verbal action. Thus, it is not possible to interpret all instances of repetitive-verbs fully compositionally. In my opinion, this behaviour of pere- ‘over’ may be explained by its intermediate position (see Tatevosov 2008, 2009). (3)
pere-čitat’
knig-u
over-read
book-ACC
‘to read the book again (4)
pere-stroit’ dom over-build house[ACC] ‘to rebuild the house’
In contrast to duration and repetitive verbs, pere-prefixed verbs of the distributive reading as in (5) show a fully compositional pattern. During my research on pereverbs, I could not find a single counterexample. All prefixed verbs of this type can be interpreted distributively, i.e. that a verbal event is spread to all contextually available objects (see Isačenko 1962; Romanova 2006), with the amount of objects determining the number of repetitions of the verbal event. The particular number of objects is not of importance. (5)
pere-probovat’ vs-e over-try
jablok-i
all-ACC.PL apple-ACC.PL
‘to try all apples one after another’ Similar to the distributive-reading, excessive-verbs like pere-gret’ ‘to overheat’ in (6) do not show any striking features with respect to their interpretation. They can be deduced compositionally and never lose their excessive meaning: “This use of pere- makes sure that the event exceeds, along some dimension or other, a contextually determined expectation value or norm” (Kagan 2013: 500). (6)
pere-gret’ raztvor over-heat solution[ACC] ‘to overheat the solution’
(Kagan 2013: 500, (18a))
Semantic and morphosyntactic features of the Russian verbal prefix pere- ‘over’ | 75
Verbs of the superiority-reading like pere-rasti ‘to outgrow’ in (7), pere-pljunut’ ‘to outdo’ in (8) and pere-chrapet’ ‘to snore more loudly’ have never been investigated against the background of the Prefix-Split-Theory. These verbs are similar to the verbs of the excessive-reading in so far as the prefix provides comparative meanings, “[…] as it involves a comparison between two individuals along some dimension or other” (Kagan 2013: 504). Pere-verbs of the superiority-type are often used metaphorically, e.g. pere-rasti ‘to outgrow’ in (7) may refer to the subject’s professional growth, pere-pljunut’ in (8) is usually understood “[…] in the sense of the subject exceeding the object in terms of some contextually specified achievement, possibly in terms of the quality of some performance” (Kagan 2013: 505). In addition to the metaphorical use of superiority-verbs, I will follow Kagan (2013: 504-507) in her assumption that these verbs cannot be interpreted purely compositionally. Similar to the repetition-reading in (4) additional assumptions have to be made. In (7), the degree of change of the measured quality or property undergone by the subject is compared to the object’s degree at the end of the verbal event, i.e. at the same time. In contrast, with pere-chrapet’ ‘to snore more loudly than someone else’ in (9), the subject and the object do not even have to snore at the same time. The object can be exceeded by the subject in loudness at any time. (7)
pere-rasti ko-go–nibud’ over-grow some-ACC–one ‘to outgrow someone’
(8)
pere-pljunut’ ko-go–nibud’ over-spit
some-ACC–one
‘to outdo someone’ (9)
pere-chrapet’ ko-go–nibud’ over-snore
some-ACC–one
‘to snore more loudly than someone else’
76 | Kristina Gehring (Kaliningrad)
3.2 Argument structure augmentation The base verbs of the duration-reading almost all belong to the cluster of intransitive verbs of the state-type (e.g. sidet’ – ‘to sit’; ležat’ – ‘to lay’).4 If one follows the general assumptions of prefixes with temporal meaning that are usually considered to belong to the superlexical-cluster, one should not expect changes in the argument structure of the prefix verbs in contrast to their simplexes. Nevertheless, the opposition of (10a) and (10b) – on the basis of intransitive sidet’ ‘to sit’ – proves argument structure augmentation after prefixation with respect to the temporal prefix pere-duration. Furthermore, in contrast to other temporal prefixes, like for instance the perdurative Russian pro- ‘through’, the example in (10b) shows that pere-duration is able to co-occur with event-oriented determiner phrases like groza ‘storm’, and not only with temporal expressions, resp. adverbials like dva časa ‘two hours’.5 (10) a. *sidet’ sit
groz-u storm-ACC
‘to sit a storm’ b. pere-sidet’ groz-u over-sit
/ dv-a
čas-a6
storm-ACC / two-ACC hour-GEN
‘to wait until the storm is over / to sit for two hours’
|| 4 I will follow Maienborn (2000) who splits the traditional Vendler state-category into i) stateverbs (e.g. sidet’ – ‘to sit’, stojat’ – ‘to stand’, ležat’ – ‘to lay’, ždat’ – ‘to wait’, spat’ – ‘to sleep’ etc.) and ii) stative expressions (e.g. znat’ – ‘to know’, ljubit’ – ‘to love’, verit’ – ‘to believe’ and nenavidet’ – ‘to hate’). 5 If duration-verbs are followed by temporal expressions (e.g. dva časa – ‘two hours’), I will argue that they occupy the object position and, therefore, they are arguments of the prefix verb. See Fowler & Yadroff (1993), Schoorlemmer (1995) and Žaucer (2009, 2012) for several tests of how to proof the argument-status of temporal expressions. 6 To co-occur with temporal expressions like dva časa ‘two hours’, pere-verbs of the durationtype usually require a larger context, like in (i). To verify the co-occurrence of duration-verbs with temporal expressions, I refer the reader to Kagan (2013): 497, (12c).
Semantic and morphosyntactic features of the Russian verbal prefix pere- ‘over’ | 77
Argument structure augmentation of pere-prefixed repetitive-verbs has been either denied (see Tatevosov 2008: 430) or neglected so far. In my opinion, this is due to the fact that pere-repetitive mainly operates on transitive base verbs and, therefore, changes in the argument structure are more difficult to observe. However, a more detailed investigation shows several argument structure effects: transitive simplex verbs like pisat’ ‘to write’ in (11a) that co-occur with non-obligatory arguments (e.g. pis’mo – ‘letter’) have to realise their arguments obligatorily after prefixation with pere-repetitive, see (11b). Unergative base verbs like rabotat’ ‘to work’ in (12a) occur with obligatory direct objects as in (12b). Furthermore, these objects do not satisfy the selectional restrictions of cognate objects the simplex verb is usually able to select. It is a known fact that unergatives have the possibility to appear in cognate object constructions, i.e. ‘to think a wonderful thought’; ‘to live ones’ life’; ‘to smile a happy smile’. Thus, unergatives may appear in transitive constructions. (11) a. Julija pisa-l-a
(pis’mo).
Julia write-PST-3SG.F letter[ACC] ‘Julia wrote (a letter).’ b. Julija pere-pisa-l-a
*(pis’mo).
Julia over-write-PST-3SG.F letter[ACC] ‘Julia rewrote *(the letter).’ (12) a. Vasja rabota-l
(*sv-oj
plan).
Vasja work-PST[3SG.M] his-ACC plan[ACC] ‘Vasja worked (*his plan).’
|| Žena prišla tol'ko večerom. Ožidaja eë, on poguljal po gorod-u, (i) wife came only in.the.evening waiting she[ACC] he walked round city-DAT poobedal, pere-side-l dva čas-a v mašin-e i poexal had.lunch over-sit-PST[3SG.M] two hour-GEN.SG in car-PREP and drove vstrečat’. eë she[ACC] meet ‘His wife didn’t arrive before the evening. While he was waiting for her he went for a walk around the town, had lunch, sat in his car for two hours and finally drove off to pick her up.’
78 | Kristina Gehring (Kaliningrad)
b. Vasja pere-rabota-l
*(sv-oj
plan).
Vasja over-work-PST[3SG.M] his-ACC plan[ACC] ‘Vasja reworked (*his plan).’ Verbs of the distributive-reading like pere-tancevat’ ‘to dance all dances’ in (13a) and pere-šutit’ ‘to retell all the jokes’ in (14a) are able to select cognate objects. In contrast to verbs of the duration- or repetitive-reading this cannot be considered an instance of argument structure augmentation, since the examples in (13b) and (14b) verify that the objects can hardly be subject to passivation (see 13b) and are blocked from deriving relative clause structures (see 14b). (13) a. pere-tancevat’ vs-e over-dance
tanc-y
all-ACC.PL dance-ACC.PL
‘to dance all possible dances one after another’ b. ?vse tancy all
by-l-i
pere-tancova-n-y
dances is-PST-PL over-dance-PST.PASS.PTCP-PL
‘All dances have been danced one after another (by us).’ (14)
a. pere-šutit’ over-joke
vs-e
star-ye
šutk-i
all-ACC.PL old-ACC.PL joke-ACC.PL
‘to retell all the old jokes one after another’ b. *vse šutki, kotor-ye all
my pere-šuti-l-i
…
jokes which-ACC.PL we over-joke-PST-PL
‘All the old jokes that we have made one after another…’ Throughout my investigation of pere-prefixed verbs, pere-verbs of the excessiveinterpretation did not show changes in their argument structure: after prefixation, transitive base verbs like solit’ ‘to salt’ as in (15) and intransitive base verbs like rabotat’ ‘to work’ as in (16) keep their former argument structure, i.e. transitive verbs stay transitive, whereas intransitive ones stay intransitive. Unlike verbs of the repetitive-meaning, simplex verbs with non-obligatory objects, e.g. verbs of consumption (pit’ – ‘to drink’) in (17), continue to realise their arguments optionally even after prefixation.
Semantic and morphosyntactic features of the Russian verbal prefix pere- ‘over’ | 79
(15)
(16)
(17)
Povar pere-soli-l sup. cook over-salt-PST[3SG.M] soup[ACC] ‘The cook put too much salt into the soup.’ Vera pere-rabota-l-a. Vera over-work-PST-3SG.F ‘Vera burned out.’ / ‘Vera is tired from too much work.’ Maksim pere-pi-l (vodk-i). Maxim over-drink-PST[3SG.M] vodka-GEN ‘Maxim had too much Vodka.’
As mentioned before, verbs of the excessive- and the superiority-reading are similar in their semantics in terms of their comparative character. However, verbs of the superiority type as in (18) regularly show the argument structure augmentation of their base verbs. Unlike the objects of distributive-verbs above in (13a) and (14a), the objects of verbs of the superiority-reading will be treated as real arguments of the particular prefix verb. Examples in (19) and (20) show that the objects of superiority-verbs can be subject to relative-clause structures and passivation respectively. The latter is possible in predicative (20a) and attributive (20b) use. The observations about argument structure augmentation in the case of superiority-verbs have never been discussed before in recent literature.7 (18)
mir-a Anglija […] v matč-e čempionat-a in game-PREP champion-GEN world-GEN England pere-igra-l-a Argentin-u […]. over-play-PST-3SG.F Argentina-ACC ‘England won against Argentina in the World Championship game.’8
|| 7 I assume that this is due to the fact of the similar semantics of both the excessive- and the superiority-reading: because of its regular and transparent meaning, pere-excessive is mostly considered to belong to the cluster of superlexicals (see Svenonius 2004; Romanova 2006). Peresuperiority that establishes only a small subpart of all pere-prefixed verbs sometimes seems to be treated identically. My assumption is proven by the following example which is considered as an instance of the excessive-reading in the given source. (ii)
Džejms pere-brosa-l James
Arenasa i
over-threw-PST[3SG.M] Arenas
prinës
pobedu “Klivlendu”.
and brought victory Cleveland
‘James outdid Arenas (in a basketball match) and led “Cleveland” to victory.’ (Romanova 2006: 257, (77b)) 8 Navoša, Dmitrij 2002: Zaščita Anglii. Ėriksson – B''elsa – 1:0. In: „Izvestija“; 07.06.2002.
80 | Kristina Gehring (Kaliningrad)
(19)
Rossijane igrali protiv SŠA, kotor-uju pere-igra-l-i […]. Russians played against USA who-ACC over-play-PST-3PL ‘Russia played against the USA and won.’9 (20) a. SŠA by-l-a pere-igra-n-a Rossi-ej USA is-PST-3SG.F over-play-PST.PASS.PTCP-F Russia-INSTR ‘The USA has been outplayed by Russia.’ b. polnost’ju pere-igra-nn-aja komand-a completely over-play-PST.PASS.PTCP-NOM.F team-NOM.F ‘total losers’ (lit.: ‘a completely outplayed team’)
3.3 Selection of unselected objects Along with argument structure augmentation, selection of unselected objects is one of the strongest criteria to distinguish lexical from superlexical prefixes. In case of the duration-reading, verbs with non-obligatory objects like ždat’ ‘to wait’ in (21) show that pere-duration causes the selection of unselected objects. More precisely, as can be seen in (21b), it prevents the prefixed verb to cooccur with its prior objects. In contrast to the simplex verb ždat’ ‘to wait’ in (21a), the prefixed verb pere-ždat’ ‘to wait for the end of something’ in (21b) can no longer co-occur with objects of the [+anim]-type like podruga ‘friend’. (21) a. Katja žda-l-a podrug-u / avtobus. Katja wait-PST-3SG.F friend-ACC / bus[ACC] ‘Katja waited for her friend / for the/a bus.’ b. #Katja pere-žda-l-a podrug-u avtobus. Katja over-wait-PST-3SG.F friend-ACC bus[ACC] Verbs of the repetitive-reading demonstrate their ability of selecting unselected objects in the case of unergative simplex verbs: prefixation with pere-repetitive extends the selectional restrictions of the base verb as is clearly visible from the opposition between (22a) and (22b) using the unergative simplex verb dumat’ ‘to think’ as an example. (22) a. Vasja duma-l (#svo-ë perv-oe rešen-ie). Vasja think-PST[3SG.M] his-ACC first-ACC decision-ACC ‘Vasja thought (#his first decision).’
|| 9 http://ria7.ru/3116/; 28.08.2014.
Semantic and morphosyntactic features of the Russian verbal prefix pere- ‘over’ | 81
b. Vasja pere-duma-l svo-ë perv-oe rešen-ie.10 Vasja over-think-PST[3SG.M] his-ACC first-ACC decision-ACC ‘Vasja has changed his first decision.’11 Verbs of the superiority-type show the most striking cases of unselected objects. One can see in (23b) that, after prefixation the objects of superiority-verbs like pere-pit’ ‘to drink more than someone else’ carry the selectional feature [+anim], whereas the base verb pit’ ‘to drink’ in (23a) is not compatible with [+anim]-objects. (23) a. pit’ (#ko-go–nibud’) drink some-ACC–one b. pere-pit’ ko-go–nibud’ over-drink some-ACC–one ‘to drink more than someone else’ To my knowledge, verbs of the distributive- as well as of the excessive-reading do not show any instances of unselected objects.
3.4 Changes in case marking restrictions Besides the aforementioned criteria in 3.1, 3.2 and 3.3 that bring pere-duration closer to the class of lexical than superlexical prefixes, verbs of the duration-reading also show changes in case marking restrictions. As can be seen by the opposition between (24a) and (24b), the objects of duration-verbs – in contrast to those of the base verbs – are always marked with the Accusative-case. The simplex ždat’ ‘to wait’ in (24a) allows both Genitive- and Accusative-case marking, while the verb is only compatible with objects in the Accusative-case after prefixation, as shown in (24b). (24) a. ždat’ zim-u / zim-y wait winter-ACC / winter-GEN ‘to wait for the/a winter’ || 10 The majority of native speakers of Russian prefers pere-dumat’ ‘to change ones’ mind’ without a direct object. Nevertheless, research results on yandex.ru show examples like pere-dumat’ (svoj) plan / (svoë) rešenie ‘to change one’s plan / decision’. I am aware of the fact that transitive pere-dumat’ might only be found in few cases. 11 See e.g. http://history-gatchina.ru/shline1.php?id=19; 14.12.2015.
82 | Kristina Gehring (Kaliningrad)
b. pere-ždat’ zim-u *zim-y over-wait winter-ACC winter-GEN ‘to wait for the end of the winter’ In case of base verbs such as stradat’ ‘to suffer’ in (25a) that occur with non-Accusative-case marked objects (boljami – aches-INSTR), prefixation with pere-duration as in (25b) overwrites the oblique Instrumental-case. Thus, the objects of duration verbs in (24b) as well as in (25b) are both marked with the “real” object case. (25) a. stradat’ bol-jami v pozvonočnik-e suffer aches-INSTR in backbone-PREP ‘to suffer from pain in the backbone’ b. pere-stradat’ ėt-u bol’ over-wait that-ACC pain-ACC ‘to suffer through the pain’ In contrast to pere-duration in (24) and (25), one can see in (26) that prefixation with pere-distributive does not change the case marking restrictions of the base verbs. Verbal phrases like bolet’ čem-to ‘to be ill with something’ in (26a) occur with objects in the Instrumental-case. After prefixation in (26b), the objects are still marked with the lexical Instrumental-case. Hence, prefixation with pere-distributive does not cause an overwriting of oblique cases. (26) a. bolet’ razn-ymi zabolevani-jami be.ill different-INSTR illness-INSTR ‘to be ill with different illnesses’ b. pere-bolet’ razn-ymi zabolevani-jami over-be.ill different-INSTR illness-INSTR ‘to have gone through infectious diseases one after another’ In case of verbs of the excessive-type, changes in the case-marking restrictions can be observed only when dealing with verbs of consumption like est’ ‘to eat’ in (27a). The simplex verbs regularly assign their objects the Accusative-case. After prefixation in (27b), objects of these verbs occur only in the Partitive Genitivecase, whereas Accusative-case marking is prohibited.
Semantic and morphosyntactic features of the Russian verbal prefix pere- ‘over’ | 83
(27) a. est’ kaš-u eat porridge-ACC ‘to eat porridge’ b. pere-est’ kaš-i *kaš-u over-eat porridge-GEN porridge-ACC ‘to eat too much porridge (so that you don’t want to eat it ever again)’ Regarding the repetitive- and superiority-verbs, changes in the case-marking restrictions cannot be easily proven. As pointed out before, pere-repetitive usually takes transitive base verbs as its input, i.e. verbs with Accusative-marked objects, see (3) and (4). After prefixation, repetitive-verbs mark their objects with the Accusative-case as well. Thus, the objects of both simplex and prefixed verbs carry the same case-marking. Therefore, changes in case-marking cannot be observed on the surface. If the Accusative-cases of simplex and prefixed verbs differ in any features, this will have to be investigated separately. Base verbs of the superiority-reading are usually intransitive, i.e. they occur without any object. After prefixation with pere-superiority they get transitivised, see for instance (18). It follows that the former and the latter structure cannot be compared to each other and changes in the case-marking restrictions cannot be observed.
3.5 Telicity It will be argued that all pere-prefixed verbs investigated here establish telic predicates. Thus, one may infer that telicity is not a suitable criterion to distinguish between lexical and superlexical prefixes. This assertion is actually right, but only with regards to the classical telicity tests based on the time-frame (in-X-time) and time-span adverbial (for-X-time). At first sight, all investigated pere-verbs – with exception of the more difficult case of duration-verbs, which will be discussed at the end of section 3.5 – are compatible only with the time-frame adverbial. However, single telicity tests show differences with respect to certain readings, as for example verbs of the excessive-type in (33). It will be demonstrated that this is due to inherent characteristics of particular pere-prefixed verbs. Hence, telicity-tests give information about the inherent structure of pere-prefixed verbs resp. predicates. As shown in (28), prefixed verbs of the repetitive-reading establish telic predicates since they can be followed only by the time-frame adverbial za polgoda ‘in half a year’. Furthermore, they show an event structure typical of telic accom-
84 | Kristina Gehring (Kaliningrad)
plishments. A fact that can be demonstrated through their compatibility with incremental modifiers like postepenno ‘gradually’ in (29). As it is known for accomplishments, their event structure is composed of a process- and a change-of-statecomponent. Incremental modifiers operate on the underlying process structure of the predicate (see Braginsky & Rothstein 2008). Additionally, unlike prefixed verbs of the duration-, excessive- or superiority-reading, repetitive-verbs as in (30) react positively, i.e. ambiguously to almost-modification (see Hay, Kennedy & Levin 1999). Both interpretations in (30i) and (30ii) are possible, which is typical of accomplishment structures. (28)
(29)
(30)
My s muž-em pere-stroi-l-i dom (za) polgoda. we with husband-INSTR over-build-PST-1PL house for half.year ‘It took me and my husband half a year to rebuild the house.’ My s mužem postepenno pere-stroi-l-i dom. we with husband-INSTR gradually over-build-PST-1PL house ‘My husband and I have gradually rebuilt the house.’ My s mužem počti pere-stroi-l-i dom. we with husband-INSTR almost over-build-PST-1PL house ‘My husband and I have almost rebuilt the house.’ (i) The described event is claimed to have occurred, but has not quite been completed; thus, there is still some work left to do. (ii) The event is asserted not to have occurred at all, i.e. they did not even start to rebuild the house.
Verbs of the distributive-reading sustain all telicity tests, except for the almostmodification in (31). This behaviour can be explained by the fact that distributiveverbs express a chain of already completed sub-events (see Romanova 2006) and each of these sub-events has its own process-component. Thus, ‘almost’ is not able to have scope over the process-component of each single sub-event. Therefore, it is not possible to interpret počti ‘almost’ in (31ii) so that the process is negated as a whole.
Semantic and morphosyntactic features of the Russian verbal prefix pere- ‘over’ | 85
(31)
My počti pere-kupa-l-i-s’ na vs-ech pljaž-ach we almost over-swim-PST-1PL-REFL on all-GEN beaches-GEN ‘We almost swam at all beaches of Goa.’ (i) The described event is claimed to have occurred, but has not quite been completed; thus, there are some beaches left they did not swim at. (ii) *The event is asserted not to have occurred at all, i.e. they did not swim at all.
As can be seen in (32), verbs of the excessive-type derive telic predicates: they combine only with time-frame adverbials (za polčasa – ‘in half an hour’). The construction in (32) becomes ungrammatical with time-span adverbials (polčasa – ‘for half an hour’). Pere-verbs of the excessive-reading show deviations in the case of almostmodification and modification with incremental modifiers, see (33) and (34). In my opinion, such behaviour can be explained by the fact that pere- ‘over’ in excessive-verbs does not have scope over the process-component, but only over the resultant-state of the verbal phrase (see also McIntyre 2003 for over- and out-predicates in English). (32)
(33)
(34)
Oni pere-gruzi-l-i sudno *(za) polčasa. they over-load-PST-1PL vessel[ACC] for half.hour ‘They overloaded the vessel within half an hour.’ Oni počti pere-gruzi-l-i sudno (sverch norm-y). they almost over-load-PST-1PL vessel[ACC] beyond standard-GEN ‘They have almost overloaded the vessel over its standard.’ (i) The described event is claimed to have occurred, but has not quite been completed; thus, they did not load the vessel beyond its limits. (ii) *The event is asserted not to have occurred at all, i. e. they did not even start to load the vessel. *Oni postepenno pere-gruzi-l-i sudno (sverch normy). they gradually over-load-PST-1PL vessel[ACC] beyond standard ‘They gradually overloaded the vessel.’
The prefix-verbs of the superiority-reading do not show any deviations of the telicity standard test. As the example in (35) shows, superiority-verbs can only be followed by the time-frame adverbial. Thus, prefixation with pere-superiority derives telic predicates. Оn the one hand, similar to the excess-reading, these verbs show that they cannot be interpreted ambiguously with respect to the almost-
86 | Kristina Gehring (Kaliningrad)
modification test, see (36). On the other hand, modification on the basis of incremental-modifiers like postepenno ‘gradually’ in (37) demonstrates the underlying accomplishment-structure of these verbs. (35)
(36)
(37)
Anglija pere-igra-l-a Argentin-u *(za) 90 minut. England over-play-PST-3SG.F Argentina-ACC for 90 minutes[GEN] ‘England outdid Argentina within 90 minutes.’ Anglija počti pere-igra-l-a Argentin-u. England almost over-play-PST-3SG.F Argentina-ACC ‘England almost outdid Argentina.’ (i) The described event is claimed to have occurred, but has not quite been completed; England played against Argentina, but they did not win. (ii) *The event is asserted not to have occurred at all; they did not even start to play Anglija postepenno pere-igra-l-a Argentin-u. England gradually over-play-PST-3SG.F Argentina-ACC ‘England has gradually outplayed Argentina.’
As the example in (38) shows, verbs of the duration-reading can be followed only by the time-span adverbial dva časa ‘for two hours’.12 The prefixed verb cannot co-occur with the time-frame adverbial za dva časa ‘in two hours’. At first glance, this observation suggests that verbs of the duration-type derive only atelic structures. Nevertheless, it will be argued that the incompatibility of duration-verbs with time-frame adverbials as in (38) is due to the fact that the temporal expression here is not a temporal adverbial in adjunct position. It is rather the argument of the prefixed verb, occupying its object position (see e.g. Žaucer 2009). (38)
Petja pere-side-l v mašin-e (*za) dv-a čas-a. Petja over-sit-PST[3SG.M] in car-PREP for two-ACC hour-GEN ‘Petja sat for two (spare) hours in the car.’ (Kagan 2013: 497, (12c))
If dva časa ‘two hours’ in (38) is in fact the object of the prefixed verb, it should be possible to combine the object with a temporal adverbial as in (39); but this
|| 12 In footnote 5 I have already pointed out that duration-verbs usually trigger a larger context to co-occur with temporal expressions. Thus, without a larger context, (38) might not be accepted by all native speakers of Russian. In narrow contexts such as (38) the preferred prefix is pro‘through’.
Semantic and morphosyntactic features of the Russian verbal prefix pere- ‘over’ | 87
obviously leads to a not well-formed structure. Following Žaucer (2009: 165–169), incompatibilities like in (39) are based on pragmatic, not on structural reasons: as soon as a duration-verb with a temporal expression in object position (dva časa – ‘two hours’) co-occurs with a time-frame adverbial that matches the denotation of the obligatory temporal expression (za dva časa – ‘in two hours’), redundancy is generated, since the structure already contains a temporal measure expression. If the information given by the time-frame adverbial (za tri časa – ‘in three hours’) does not match the denotation of the obligatory temporal expression, pragmatic deviance is produced. (39)
Petja pere-side-l v mašin-e dv-a čas-a Petja over-sit-PST[3SG.M] in car-PREP two-ACC hour- GEN (??za tri čas-a / ??za dv-a čas-a). for three hour-GEN / for two-ACC hour-GEN ‘Petja sat in the car for two hours (??in three hours / ??in two hours).’
In my opinion, the assumption that a temporal expression like dva časa ‘two hours’ in (38) and (39) occupies the object position of the duration-verb can be proven by two facts: firstly, instead of temporal expressions, verbs of the duration-type are able to select event oriented determiner phrases (see Kagan 2013) like groza ‘storm’ in (40) that occupy the object position of the verb. (40)
Petja pere-side-l groz-u. Petja over-sit-PST[3SG.M] storm-ACC ‘Petja outsat the storm.’
Secondly, if dva časa ‘two hours’ in (38) and (39) was a temporal expression in adjunct position, its syntactic status should allow the co-occurrence with an object-DP like groza ‘storm’. As the construction in (41) shows, this is prohibited under the given reading; no matter if a time-frame or a time-span adverbial is present in the structure.13
|| 13 As soon as pere-sidet’ ‘to sit out’ in (39) is combined with the time-frame adverbial za dva časa ‘in two hours’, native speakers of Russian tend to interpret the structure in the sense of the superiority-reading, i.e. that the subject has been sitting longer than the given object.
88 | Kristina Gehring (Kaliningrad)
(41)
*Petja pere-side-l groz-u (za) dva čas-a. Petja over-sit-PST[3SG.M] storm-ACC for two hour-GEN ‘Petja has outsat the storm (in) two hours.’
However, further telicity-tests show that duration-verbs like pere-ždat ‘to wait for the end of something’ in (42) cannot be interpreted ambiguously with respect to the modifier počti ‘almost’. Furthermore, as shown by the example in (43), they cannot be modified by incremental modifiers like postepenno ‘gradually’ or X-zaX ‘X-by-X’. It follows that duration-verbs do not provide an accomplishmentstructure.14 Similar behaviour can be detected with prefix verbs of the excessivetype. (42)
(43)
Petja počti pere-žda-l groz-u. Petja almost over-wait-PST[3SG.M] storm-ACC ‘Petja has almost waited for the storm to pass.’ (i) The described event is claimed to have occurred, but has not quite been completed. (ii) *The event is asserted not to have occurred at all. *Petja postepenno / molnija za molni-ej Petja gradually / lightning for lightning-INSTR pere-žda-l groz-u. over-wait-PST[3SG.M] storm-ACC ‘Petja gradually/lightning by lightning waited for the storm to pass.’
Since the average of Russian perfective verbs derive telic predicates as soon as they are followed by direct objects, I suggest that duration-verbs derive telic predicates. However, as soon as an event oriented DP (e.g. groza – ‘storm’) is present in the structure, see (41), duration-verbs cannot be followed by a time expression. In my opinion, this is due to the semantics of the object-DP and its unpredictable duration, i.e. the subject does not have any influence on how long the storm will last. Nevertheless, additional telicity-tests have to be found to fully proof the telicity-status of duration-verbs.
|| 14 For further investigation: there is still no answer to the question what kind of event-structure underlies the duration-verbs.
Semantic and morphosyntactic features of the Russian verbal prefix pere- ‘over’ | 89
3.6 Derivation of Past Passive Participles Verbs of the duration-reading do not show a complete past passive participle paradigm, i.e. it is not possible to form past passive participles of all pere-prefixed duration-verbs: while pere-žit’ ‘to survive’ in (44) offers suitable input to derive the form pere-žityj ‘survived’, the past passive participle pere-ždannyj ‘outwaited’ derived from pere-ždat’ ‘to wait for something to pass/end’ in (45) seems to be less acceptable. The third example in (46) shows that pere-ležat’ ‘to wait for something to pass/end in a lying position’ does not provide proper input to derive a past passive participle. This observation can presumably be reduced to one of two reasons: on the one hand, the argument-structure of the underlying base verb may play a role, since unprefixed žit’ ‘to live’ as well as ždat’ ‘to wait’ can select objects, at least cognate ones, whereas ležat’ ‘to lay’ is not able to select objects at all. On the other hand, the observations in (44) to (46) may be explained by the frequency of the prefixed verbs. Internet research shows that the prefixed verbs building the derivational basis for (44) and (45) (pere-žit’ – ‘survive’; pere-ždat’ – ‘to wait for the end of something’) occur much more frequently than the basis for the last one in (46). Thus, one might suggest that frequency according to the speaker’s “habit” of using prefix-verb-complexes increases the possibility to derive a proper past passive participle. Still, if this assumption is true and, in the case of duration-verbs one has to deal with argument structure augmentation, there is actually no structural reason that prohibits the derivation of past passive participles as in (46). (44)
(45)
(46)
pere-ži-t-yj over-live-PST.PASS.PTCP-NOM.M ‘the survived crisis’ ?? pere-žda-nn-yj over-wait-PST.PASS.PTCP-NOM.M ’the waited-for-moment’ *pere-leža-nn-yj over-lay-PST.PASS.PTCP-NOM.M
krizis crisis[NOM.M] moment moment[NOM.M]
In contrast, verbs of the repetitive-reading show a complete paradigm of past passive participles: as demonstrated by the examples in (47), repetitive-verbs – even on the basis of unergative simplex verbs like rabotat’ ‘to work’ – are able to derive past passive participles, either in attributive, see (47a), or predicative usage, see (47b).
90 | Kristina Gehring (Kaliningrad)
(47) a. pere-rabota-nn-aja n-ami stat’-ja over-work-PST.PASS.PTCP-NOM.F we-INSTR article-NOM.F ‘the revised article (by us)’ b. stat’-ja by-l-a pere-rabota-n-a article-NOM.F is-PST-3SG.F over-work-PST.PASS.PTCP-F ‘the revised article’ According to the standard assumption for lexical and superlexical prefixes, pereverbs of the distributive reading like pere-tancevat’ ‘to dance all dances one after another’ in (48) and pere-lovit’ ‘to catch one after another’ in (49) should not be able to form past passive participles, since pere-distributive is said to be an SLP located high up in the syntactic structure. But, as one can see from the examples in (48b) and (49b), the derivation of past passive participles is possible, even in the case of unergative base verbs like tancevat’ ‘to dance’. The predicative structures in (48c) and (49c) work pretty well and are accepted by the majority of the native speakers surveyed. It should be added that (49b) and (49c) were more easily decoded by native speakers than (48b) and (48c).15 Interestingly, the derivation is blocked in the case of pere-otkryvat’ ‘to open one after another’ in (50b), although the prefixed verb is transitive and perfective. For the derivational structure of the prefixed verb, see (67). (48) a. Ja uže pere-tanceva-l-a vs-e tanc-y I already over-dance-PST-3SG.F all-ACC.PL dance-ACC.PL vs-ech narod-ov mir-a. all-GEN.PL people-GEN.PL world-GEN.SG ‘I have already danced all world‘s folk dances (one after another).’ b. ??vs-e pere-tancova-nn-ye mn-oj all-NOM.PL over-dance-PST.PASS.PTCP-NOM.PL I-INSTR tanc-y vs-ech narod-ov mir-a dance-ACC.PL all-GEN.PL people-GEN.PL world-GEN.SG ‘all one after another danced world’s folk dances (by me)’
|| 15 Two of six native speakers indicated the attributive use in (48b) as grammatically bad, one of six did not accept the predicative structure in (48c). Native speakers who accepted the examples in (48b) and (48c) mostly described them as ‘somewhat odd’ or ‘sounding strange’, but ok. The reason for this has to be further investigated.
Semantic and morphosyntactic features of the Russian verbal prefix pere- ‘over’ | 91
c. ?Vs-e tanc-y vs-ech narod-ov all-NOM.PL dance-NOM.PL all-GEN.PL people-GEN.PL mir-a by-l-i pere-tancova-n-y mn-oj. world-GEN.SG is-PST-PL over-dance-PST.PASS.PTCP-PL I-INSTR ‘All world’s folk dances have been danced one after another (by me).’ (49) a. Otec pere-lovi-l vs-ech myš-ej. father over-catch-PST[3SG.M] all-ACC.PL mouse-ACC.PL ‘The father has caught all mice one after another.’ b. ?vs-e pere-lovle-nn-ye otc-om all-NOM.PL over-catch-PST.PASS.PTCP-NOM.PL father-INSTR myš-i mouse-NOM.PL ‘all one after another caught mice (by the father)’ c. Vs-e myš-i by-l-i pere-lovle-n-y all-NOM.PL mouse-NOM.PL is-PST-PL over-catch-PST.PASS.PTCP-PL otc-om. father-INSTR ‘All mice were caught one after another (by the father).’ (50) a. Otec pere-otkryva-l vs-e bank-i. father over-open-PST[3SG.M](PF) all-ACC.PL can-ACC.PL ‘The father has opened all the cans one after another.’ b. *vs-e pere-otkry-t-ye otc-om all-NOM.PL over-open-PST.PASS.PTCP-NOM.PL father-INSTR bank-i can-NOM.PL ‘all one after another caught mice (by the father)’ Verbs of the excessive-reading as in (51) show that the derivation of past passive participles – either in attributive (51a) or predicative (51b) usage – is blocked if the prefixed verb has been derived on the basis of an intransitive simplex verb like rabotat’ ‘to work’. As soon as excessive predicates are derived from transitive simplex verbs, e.g. žarit’ ‘to fry’ in (52), the derivation of both kinds of past passive participles is perfectly fine. The blocking of past passive participles in (51a) and (51b) can be easily explained by the fact that intransitive simplex verbs even after prefixation with pere-excessive stay intransitive (see section 3.2). (51) a. *pere-rabota-nn-yj Oleg over-work-PST.PASS.PTCP-NOM.SG.M Oleg[NOM.SG.M] ‘the over-worked Oleg’
92 | Kristina Gehring (Kaliningrad)
b. *Oleg by-l pere-rabota-n Oleg[NOM.SG.M] is-PST[3SG.M] over-work-PST.PASS.PTCP-M ‘Oleg was overworked’ (52) a. pere-žare-nn-oe mjas-o over-fry-PST.PASS.PTCP-NOM.SG.N meat-NOM.SG.N ‘the overdone meat’ b. mjas-o by-l-o pere-žare-n-o meat-NOM.SG.N is-PST-3SG.N over-fry-PPP-N ‘the overdone meat’ Verbs of the superiority-reading, as pere-igrat’ ‘to outplay’ in (53) do not show any gaps in deriving past passive participles, even in the case of an intransitive base verb like igrat’ ‘to play’. Furthermore, as the examples in (53a) and (53b) show, the derivation of past passive participles of verbs of the superiority-reading is possible both in attributive and in predicative structures. Explaining the structural difference between attributive and predicative Russian past passive participles is beyond the limits of this paper. For a more detailed discussion of predicative and attributive past passive participles in Russian the reader is referred to Schoorlemmer (1995). (53) a. pere-igra-nn-aja na prošloj nedele komand-a over-play-PST.PASS.PTCP-NOM.F on last week team-NOM.SG.F ‘the team that has been outdone last week’ b. na prošloj nedele komand-a by-l-a pere-igra-n-a on last week team-NOM.SG.F is-PST-F over-play-PST.PASS.PTCP-F ‘the team was outdone last week’
3.7 Multiple prefixation and stacking Following Tatevosov (2008: 431–436), multiple prefixation and stacking describe processes in which at least two prefixes in adjacent order occur above a verbal stem. In the case of multiple prefixation it is already a common fact that superlexical prefixes occur only above lexical ones, not the other way round (see for instance Svenonius 2004: 229; Tatevosov 2008: 431–434; Wiland 2012: 310). According to Tatevosov (2008: 435), stacking - in contrast to multiple prefixation denotes the co-adjacence of two prefixes of the same class. This view will be adopted.
Semantic and morphosyntactic features of the Russian verbal prefix pere- ‘over’ | 93
(Self-)recursiveness of prefixes, see for example (58), will be treated as instances of stacking and, therefore, as features of intermediate prefixes (for a contrary view see Wiland 2012, who argues that stacking of syncretic prefixes in Polish is excluded). The following will focus on recursive stacking to show that pere-duration and pere-excessive share features with pere-repetitive, which has already been classified as an intermediate resp. positionally restricted prefix (Tatevosov 2008, 2009). In contrast to the standard assumption that prefixes with temporal meaning belong to the pattern of superlexicals (see for instance Borik 2002; Romanova 2004, Svenonius 2004), pere-duration in (54) displays behaviour similar to Tatevosov’s (2008) intermediate prefixes. Firstly, it can be merged with prefixes like do-completive (completion of an action) that have already been classified as intermediate (see Tatevosov 2008).16 Secondly, both prefixes in (54a) and (54b) are able to co-occur with each other in recursive order without changing or losing their meaning. (54) a. […] sel za kamnem i pere-do-žda-l17 tam bur-ju. sat behind stone and over-till-wait-PST[M] there storm-ACC ‘[…] (he) hid behind the stone to finish sitting out the storm.’18 b. […] my vmeste do-pere-žda-l-i dožd’. we together till-over-wait-PST-PL rain[ACC] ‘[…] we have finished sitting out the storm together.’ (M. Yastrebova: p. c.) Pere-repetitive co-occurs regularly and productively above lexical prefixes like za- ‘behind’, as for instance in (55). Thus, at first sight, the prefix fulfils the standard assumption made for superlexical prefixes (see Svenonius 2004). Tatevosov (2008: 431) has already shown in detail: if pere-repetitive co-occurs with superlexical prefixes like accumulative na- in (56), the position of the former is always inside, adjacent to the verbal stem like in (56a). The configuration in (56b) shows that pere-repetitive is blocked from occurring outside na-accumulative. As soon as the prefix is part of a configuration with other intermediate prefixes, e.g. do-
|| 16 This observation can be taken as evidence only if one supports Tatevosov’s (2008: 434–435) view that stacking of superlexical and lexical prefixes is prohibited. 17 The prefix complex pere-do-ždat’ ‘to finish sitting out’ and do-pere-ždat’ ‘to finish sitting out’ as in (54a), (54b) and (66b) is mainly found in Internet chats and forums. Therefore, I assume that it can be treated as an instance of colloquial language. 18 http://mallonpokerpg.mybb.ru/viewtopic.php?id=923&p=4; 21.07.2013.
94 | Kristina Gehring (Kaliningrad)
completive in (57), stacking is possible with changed prefixed order without losing or changing the semantics of the given prefixes, see (57a) and (57b). Furthermore, the example in (58) demonstrates that the prefix is able to stack recursively with itself. (55)
(56) a.
b. (57) a.
b.
(58)
pere-za-pisat’ DVD-disk over-behind-write(PF) DVD-disc[ACC] ‘re-write the DVD-disc’ na-pere-pis-yva-t’ on-over-write-IVA-INF(IPF) ‘accumulate a quantity of sth as an outcome of re-writing’ *pere-na-pis-yva-t’ over-on-write-IVA-INF(IPF) (Tatevosov 2008: 431) Vasja do-pere-pisa-l stat’-ju. Vasja till-over-write-PST[3SG.M] paper-ACC ‘Vasja completed re-writing a paper’ Vasja pere-do-pisa-l stat-’ju. Vasja over-till-write-PST[3SG.M] paper-ACC ‘Again, Vasja performed a final stage of writing a paper’ Vasja pere-pere-pisal stat’-ju. Vasja over-over-write-PST[3SG.M] paper-ACC ‘Vasja re-wrote his paper again.’ (Tatevosov 2008: 435, (17d))
As one can see in (59), pere-excessive regularly gets merged above lexical prefixes, e.g. vy- ‘out’. Parallel to pere-duration and pere-repetitive, the prefix is able to co-occur with other prefixes, for example na-saturative, see (60), or do-completive, see (61): as the examples in (60a) and (60b) as well as (61a) and (61b) show, pere-excessive can occur in changeable prefix order without losing its excessive meaning. Parallel to (58), pere-excessive is suitable for recursive selfmerging, see (62).19 (59)
pere-vy-polnit’ plan over-out-fill(PF) plan[ACC] ‘do more than planned’
|| 19 Native speakers of Russian pointed out to me that stylistically it sounds not very good, but nevertheless it is interpretable.
Semantic and morphosyntactic features of the Russian verbal prefix pere- ‘over’ | 95
(60) a. Daže na-pere-el-a-s’ slegka vs-em ėt-im. even on-over-eaten-F-REFL(PF) slightly all-INSTR that-INSTR ‘I have slightly overeaten too much.’20 b. […] na paru dnej appetit poterjala, pere-na-el-a-s’ […]. on pair days appetite lost over-on-eaten-F-REFL(PF) ‘[…] I’ve lost my appetite for some days, […] I have overeaten too much.’21 (61) a. […] uže ruka ne podnimaetsja do-pere-pit’ […]. already hand not raise:3SG:REFL till-over-drink(PF) ‘I want some alcohol, but I have drunk too much and I can’t drink anymore.’22 b. […] kto nedo-pere-pi-l smožet pere-do-pit’ who not.till-over-drink-PST[3SG.M] can(PF) over-till-drink(PF) prjamo na tancpole. straight on dancefloor ‘Those who haven’t had too much to drink will have the possibility to catch up on excessive drinking on the dance floor.’23 (62) pere-pere-gruzit’ lodk-u over-over-load(PF) boat-ACC ‘overload the boat way too much’ (Tatevosov: p. c.) The following section will concentrate on the fact that there is a structural difference between pere-distributive on the one hand, and pere-repetitive as well as pere-excessive on the other. Tatevosov’s (2009) view on selectionally and positionally restricted prefixes will be supported and it will be shown why pere-excessive is considered to belong to the pattern of positionally restricted prefixes.
3.8 Selectional restrictions The standard assumption for lexical and superlexical prefixes is that lexical prefixes select perfective and superlexicals imperfective stems, see examples (63a) and (63b) respectively (cf. Romanova 2004; Svenonius 2004). To demonstrate this pattern, authors usually refer to verbs like brosit’/brosat’ ‘to throw’PF/IPF or kinut’/kidat’ ‘to throw’PF/IPF; here the perfective-imperfective distinction is done
|| 20 http://forum.say7.info/topic28172-10800.html; 21.07.2013. 21 http://dukandiet.ru/forum/viewtopic.php?f=18&t=2828&start=2490; 21.07.2013. 22 http://stfw.ru/forums.php?m=posts&q=5606&d=30; 13.11.2012. 23 http://vk.com/wall-48385424?offset=40; 29.01.2015.
96 | Kristina Gehring (Kaliningrad)
through the theme vowel. Contrary to this assumption, it will be argued that it is not possible to predict a prefix’s class membership based on the aspect-feature of the unprefixed base verb the prefix is combined with. (63) a. pere-kinut’ mjač čerez setk-u over-throw(PF) ball[ACC] over net-ACC ‘to throw the ball over the net’ b. pere-kidat’ over-throw(IPF) ‘throw one by one’
(Svenonius 2004: (71b))
As the examples in (64) and (65) show, pere-excessive can be merged with both perfective brosit’ ‘throw’PF as well as kinut’ ‘throw’PF. (64)
(65)
Rybak pytalsja za-kinut’ udočk-u v simpatičnoe fisherman tried behind-trow(PF) fishing.rod-ACC in nice mesto u kamyš-ej, no čut’-čut’ pere-brosi-l. place at rushes-GEN but a.little.bit over-throw-PST[3SG.M](PF) ’A/the fisherman tried to throw the rod into a nice place at the rushes, but he has thrown it too far.’24 Paren’ brosi-l okurok v urn-u […], no guy throw-PST[3SG.M](PF) stub[ACC] in bin-ACC but pere-kinu-l, i okurok upal na gazon. over-throw-PST[3SG.M](PF) and stub[NOM] fell on lawn[ACC] ‘A/the guy tried to throw the cigarette stub into the bin, but he has thrown it too far and it fell on the lawn.’25
Nevertheless, the aspect-feature of the prefix’s input can be used as a distinctive feature, but only in the case of secondary imperfectivisation. It will be shown that the [−perf]-feature of a prefixed and secondary imperfectivised verb can be used to make a more fine-grained distinction between the prefixes usually considered as superlexicals. If the suffix for secondary imperfectivisation (-iva-/-yva-/-va-) is present in the structure of a pere-prefixed verb, the [+perf]feature of the verbal complex provides information about the position of the prefix relative to the suffix, i.e. one finds out whether the prefix or the suffix was
|| 24 http://onlineslovari.com/tolkovyiy_slovar_dmitrieva/page/perebrasyivat.3402/; 12.03.2012. 25 http://onlineslovari.com/tolkovyiy_slovar_dmitrieva/page/perekidyivat.3440/; 12.03.2012.
Semantic and morphosyntactic features of the Russian verbal prefix pere- ‘over’ | 97
merged first. Further, it becomes obvious that individual readings of pere- ‘over’ differ with respect to their merging position. To begin with, pere-duration in (66), contrary to the standard assumption about superlexicals, can be merged with both an imperfective input, see (66a), or a perfective one, see (66b): the given input can be either a simplex or a prefixed verb. If pere-duration were a typical superlexical prefix, the construction in (66b) should be prohibited. (66) a. pere-ždat’ [over-[wait](IPF)](PF) ‘to wait for something to pass’ b. pere-do-ždat’ [over-[till-wait](PF)](PF) ‘to wait for something to pass’ An even more interesting observation is the fact that the suffix for secondary imperfectivisation provides information about the position of pere-distributive relative to pere-repetitive and pere-excessive. Pere-distributive in (67) as well as pere-repetitive in (68) co-occur productively above lexical prefixes. However, unlike pere-repetitive pere-distributive triggers the presence of the suffix for secondary imperfectivisation. The latter cannot be merged with an already prefixed [+perf]-stem; evidence for this is the grammatically poor structure in (67b). Thus, the position of pere-distributive relative to the position of the suffix for secondary imperfectivisation shows that the prefix has to take a higher position in the structure than the suffix. (67) a. pere-ot-kry-va-t’ vs-e bank-i [over-[[away-open](PF)-VA](IPF)](PF)-INF all-ACC.PL can-ACC.PL ‘open all cans one after another’ (Tatevosov 2011: 235, (40b)) b. *pere-ot-kry-t’ vs-e bank-i over-[away-open](PF)-INF all-ACC.PL can-ACC.PL intented: ‘open all cans one after another’ In contrast to pere-distributive in (67), pere-repetitive in (68a) can be merged with prefixed [+perf]-input. In case of secondary imperfectivisation, the prefix never gets merged above the suffix for secondary imperfectivisation. This fact can be proven by the morphological aspect of the verb, see (68b): as soon as the suffix is present in the structure, the multi-prefixed verb carries the morphological imperfective aspect. In contrast to pere-distributive in (67b), pere-repetitive in (68a) can
98 | Kristina Gehring (Kaliningrad)
be merged with [+perf]-input as well. Evidence for the fact that we are dealing with two different derivational structures in (67a) and (68b) is displayed by the aspect feature of both structures; (67a) carries a [+perf]-feature, whereas (67b) is [−perf], regardless of the presence of the suffix for secondary imperfectivisation. (68) a. pere-do-pisa-t’ over-[till-write](PF)-INF ‘repeat the final stage of writing something’ (Tatevosov 2008: 435, (17b)) b. pere-do-pis-yva-t’ [[over-[till-write](PF)](PF)-YVA](IPF)-INF ‘repeat the final stage of writing something’ (Tatevosov 2009: 142, (123)) The position of prefixes with respect to the location of the suffix for secondary imperfectivisation is one of the reasons why Tatevosov (2009) argues that the group of superlexical prefixes has to be divided into a minimum of three groups: prefixes with selectional restrictions like pere-distributive, prefixes with a positional restriction like pere-repetitive and left-peripheral prefixes like po-distributive. The class of positionally restricted prefixes is consistent with his 2008 intermediate prefixes. His view is supported here and it will be shown in the following that pere-excessive belongs to the group of positionally restricted prefixes, since it displays the same behaviour as pere-repetitive. As one can see in (69a), pere-excessive can be merged on top of already prefixed and perfective verbs. If one derives the secondary imperfective of (69a) like in (69b), the aspectual feature of the verbal complex shows that the prefix has to be merged below the suffix for secondary imperfectivisation, since the pere-prefixed complex carries a [−perf]-feature.26 Thus, the derivational structure in (69b) on the basis of pere-excessive is parallel to the structure in (68b) based on pererepetitive. (69) a. pere-vy-polni-t’ [over-[out-fill](PF)](PF)-INF ‘to do more than planned’
b. pere-vy-polnj-a-t’ [[over-[out-fill](PF)](PF)-A](IPF)-INF ‘to do more than planned’
|| 26 In the case of (69b) it has to be further investigated if the fact that the secondary imperfective is derived by the a-morpheme and not the typical suffix for secondary imperfectivisation has any impact on the derivational structure?
Semantic and morphosyntactic features of the Russian verbal prefix pere- ‘over’ | 99
4 Conclusion According to the distributional tests on pere-prefixed verbs from (1) to (69) it can be concluded that pere- ‘over’ in its investigated readings (duration-, repetitive-, distributive-, excessive- and superiority-reading) does not comply with each criterion of the standard assumptions for verbal prefixes against the background of the Prefix-Split-Theory. To make the results more transparent, they will be summarised in table 4. Tab. 4: Summary of the distributional tests.
Distinguishing criteria
duration
repetitive
distributive
excessive
superiority
(a)
non-compositionality
+
+
−
−
+
(b)
argument structure
+
+
−
−
+
(c)
unselected objects
+
+
−
−
+
(d)
case-marking restrictions
+
/
−
+
/
(e)
telicity time-frame adverbial almost-modification incremental-modification
+ + − −
+ + + +
+ + − /
+ + − −
+ + − +
(f)
past passive participles
+
+
+
+
+
(g)
multiple prefixation & stacking
− +
+ +
+ /
+ +
− −
(h)
selectional restrictions
+ perf
+ perf
− perf
+ perf
/
Starting with the duration-verbs – the most controversial reading of pere-prefixed verbs against the background of the Prefix-Split-Theory in the present investigation – I would like to remind of the fact that prefixes with temporal meaning are usually treated as instances of superlexicals that only modify the base verb. Regarding the criterion of (a) compositionality, duration-verbs like e.g. pere-sidet’ ‘to sit out’ in (2) can be interpreted compositionally, insofar as these verbs describe eventualities – usually expressed by the base verb –, whereby pere- ‘over’ contributes a meaning of exceeding the temporal interval of the given eventuality. But, strictly speaking, the meaning of the prefix and the base verb is not fully predictable. One needs additional information. Thus, pere-sidet’ ‘to sit out’ in (2) cannot be understood according to the meaning of its base verb sidet’ ‘to sit’ (see Kagan 2013: 496-500). Of course, these verbs cannot be treated as cases of full
100 | Kristina Gehring (Kaliningrad)
idiosyncrasy, that is why they have been marked with [+]. It should be highlighted that the partly non-compositionality of duration-verbs has never been discussed before and duration-verbs are less compositional than it is expected against the background of the Prefix-Split-Theory for temporal prefixes. A similar situation can be observed in the case of the criteria (b) changes in the argument structure, see (10), (c) selection of unselected objects, see (21), and (d) changes in the case-marking restrictions, see (24). To sum up, duration verbs react positively to all three of them, which contradicts the standard assumption for temporal prefixes. A more difficult case presents the criterion (e) telicity, because in contrast to all the other investigated readings, duration-verbs cannot be followed by timespan or time-frame adverbials as soon as a direct object, see (41), is present in the structure. Furthermore, duration-verbs cannot be modified by almost or incremental modifiers, see (42), suggesting that they do not belong to the pattern of accomplishment-verbs. A pattern that can be observed in the case of the excessive-reading as well. With respect to criterion (f), the derivation of past passive participles, duration verbs form a very heterogeneous picture; nevertheless, as can be seen in (44), the derivation is not blocked. This serves as additional evidence that duration-verbs derive telic predicates, since the derivation of past passive participles requires the presence of a resultant state. Concerning criterion (g), multiple prefixation and stacking, pere-duration shows similar stackingcharacteristics as pere-repetitive, see (54), whereby the latter has been clustered as an intermediate prefix (see Tatevosov 2008). However, throughout the whole investigation no cases of multiple prefixation where pere-duration has co-occurred above a lexical prefix have been discovered. But, if pere-duration was a typical superlexical prefix, it should be regularly merged above lexical ones. Finally, testing criterion (h) shows pere-duration is able to select perfective as well as imperfective stems, see (66). Summing up all these observations, pere-duration shows much more similarities with lexical than with superlexical prefixes. Continuing with the repetitive-reading and criterion (a), similar to pere-duration, these verbs only partly show compositional interpretation, see (3) and (4). Especially in the case of creation-verbs with incremental Theme-arguments like pere-stroit’ dom ‘to rebuild the house’ in (4), one needs additional assumptions to interpret the structure properly. Looking at criteria (b) and (c), parallel to the duration-reading, repetitive-verbs show changes in the argument structure, see (11) and (12), and they allow the selection of unselected objects, which is treated as the most important criterion to distinguish superlexical from lexical prefixes. Regarding the issue in (d), it has to be separately investigated if the object’s Accusative-case marking – before and after prefixation – differ in any feature. On the surface, no changes in case marking can be detected. Unlike the duration-
Semantic and morphosyntactic features of the Russian verbal prefix pere- ‘over’ | 101
reading, repetitive-verbs show a very clear pattern concerning (e) telicity, i.e. all tests are found to be positive, see (28) to (30). Furthermore, pere-repetitive shows a very systematic pattern in deriving (f) past passive participles, see (47). In contrast to the duration-verbs, pere-repetitive is able to co-occur in configurations of (g) multiple prefixation, see (55), and allows stacking, cf. (57). Finally, pere-repetitive, similar to pere-duration and pere-excessive, is able to select perfective as well as imperfective stems, see (68). Pere-distributive can be summarised quickly, since the prefix shows hardly any non-characteristic features of superlexical prefixes. Distributive-verbs can be (a) interpreted fully compositionally, cf. (5), the verbs do not show (b) any changes in the argument-structure of the base verb, (c) do not select unselected objects, and (d) no changes in case-marking restriction can be discovered. However, contrary to the standard-assumption for superlexical prefixes, the prefixation with pere-distributive brings about (e) telic predicates. Yet they do not allow almost-modfication, which can be explained by the verb’s distributive structure, see (31). Similar to verbs of the duration reading, not all instances of distributiveverbs are able to derive (f) past passive participles, see (48) and (49). But, as has already been pointed out at the beginning, the mere derivation of past passive participles may not be meaningful as a criterion to distinguish between lexical and superlexical prefixes. The prefix’s characteristics in (g) multiple prefixation were not discussed in detail, because it is a known fact that pere-distributive cooccurs only above lexical prefixes. For further research it might be interesting to have a closer look at the prefix’s stacking behaviour, i.e. with delimitative po- ‘for a while’ and/or distributive po- ‘one after another’. Here, the reader is referred to Tatevosov (2009). Finally, (h) selectional restrictions show that pere-distributive differs clearly from all the other investigated instances of pere-, since distributivepere selects only imperfective stems, see (67). As can be seen from table 4, pere-excessive is similar to pere-distributive according to criteria (a), (b) and (c). Thus, excessive-verbs can be interpreted strictly compositionally, do not show any changes in the verb’s argument structure and do not allow unselected objects. In so far, excessive verbs fulfil all criteria for a superlexical prefix. However, when it comes to consumption verbs, see (27), changes in (d), the case marking restrictions, can be observed. Furthermore, the excessive-predicates are telic (e), since they can be followed by time-frame adverbials, cf. (34). However, similar to duration-verbs, they do not allow incrementalmodification, see (34), and cannot be interpreted ambiguously with regards to almost, see (33). Much like the repetitive-reading, excessive-verbs derive (f) past passive participles, see (52), occur in (g) multiple prefixation above lexical prefixes, see (59), and stack, see (60) and (61). Parallel to repetitive-pere, the prefix
102 | Kristina Gehring (Kaliningrad)
is able to be merged recursively with itself, see (62). Finally, similar to pere-duration and pere-repetitive, pere-excessive can be merged with perfective and imperfective stems, cf. (64), (65) and (69). To sum up, on the one hand, as far as the criteria (a) to (c) are concerned, pere-excessive shares a lot of features with peredistributive, but, on the other hand, its behaviour is very similar to pere-repetitive according to the criteria (f) to (h). Finally, superiority-verbs, which have never been discussed in detail, actually fulfil all criteria expected for lexically prefixed verbs: they can be interpreted (a) compositionally and non-compositionally, see (7) to (9), (b) always show changes in the argument structure, cf. (18) and allow (c) selection of unselected objects, see (23). Changes in the case-marking restrictions, criterion (d), cannot be observed, since superiority-verbs are derived on the basis of intransitive simplex verbs. Parallel to pere-repetitive and pere-excessive, pere-superiority causes (e) telic structures that can be attested by the time-frame adverbial, see (35). But, unlike repetitive-verbs and parallel to excessive-verbs, superiority-predicates cannot be modified with almost, see (36). Yet they can co-occur with incremental modifiers that are prohibited in the case of excessive-verbs, see (37). Superiorityverbs derive (f) past passive participles, even if the base verb is intransitive, see (53). This observation is considered as evidence that prefixation with pere-superiority leads to real argument structure augmentation, where the “new” object becomes the direct object of the superiority-verb. Finally, throughout my whole research on pere-prefixed verbs, not a single instance could be found on (g) multiple prefixation or stacking with pere-superiority. Nevertheless, the possibility of co-occurrences, such as po-delimitative (‘for a while’) above pere-superiority, is not excluded; however, this is beyond the limits of this paper. Note that pere-superiority shuns all kinds of co-occurrences with other prefixes; therefore, it does not make sense to investigate its (h) selectional restrictions, because pere-superiority occurs above imperfective simplex verbs, and is, therefore, very similar to pere-translocation (pere-echat’ – ‘to drive over’; to move), which is uncontroversially analysed as a lexical prefix. At this point, research on prefixes suggests a minimum of three structurally different groups of prefixes: lexical, superlexical and intermediate – with selectionally and positionally restricted prefixes (see e.g. Tatevosov 2009): According to my research on pere- ‘over’, I suggest that pere-translocation – which has not been discussed here – and pere-superiority can be treated as lexical prefixes, since the latter matches all criteria, and, therefore, takes a low position inside the verbal phrase. Pere-excessive is very similar to pere-repetitive; both can be included in Tatevosovs (2008, 2009) intermediate prefixes. But, it is unclear on which syntactic level the intermediate position is exactly located. According to
Semantic and morphosyntactic features of the Russian verbal prefix pere- ‘over’ | 103
the test results in table 4, pere-distributive differs from all the other instances of pere- ‘over’. The prefix has to be ranked among the group of superlexical prefixes, and, against the background of its selectional restrictions, it is suggested that it has to be located higher up in the structure than pere-repetitive and pere-excessive (see e.g. Tatevosov 2009). Some authors assume the position of distributive prefixes in the area of the aspect-projection (see e.g. Ramchand 2004; Svenonius 2004). In the case of pere-duration more research will be necessary to determine the exact position of the prefix. Nevertheless, as shown in section 3.1, 3.2 and 3.3, the prefix displays behaviour which is usually contributed to the group of lexical prefixes. In my opinion it is not right, to classify pere-duration as a superlexical prefix solely on the prefix’s temporal meaning.
References Babko-Malaya, Olga. 1999. Zero Morphology: A Study of Aspect, Argument Structure, and Case. New Brunswick: Rutgers University dissertation. Babko-Malaya, Olga. 2003. Perfectivity and Prefixation in Russian. Journal of Slavic Linguistics 11 (1). 5–36. Biskup, Petr. 2007. P(refixe)s and P(reposition)s. MS. (t.a.) In Boštian Dvořák & Elena Gorishneva (eds.), Proceedings of the 2nd Congress of SLS 2007. Biskup, Petr. 2008. The Syntax of Prefixes. In Alena Jaklová (ed.), Člověk - jazyk - text. Sborník z mezinárodní lingvistické konference konané u příležitosti životního jubilea prof. PhDr. Jana Kořenského, České Budějovice, 18.-22. září 2007, 339–344. České Budějovice: Jihočeská univerzita. Biskup, Petr. 2012. Slavic Prefixes and Adjectival Participles. In Markéta Ziková & Mojmír Dočekal (eds.), Slavic Languages in Formal Grammar: Proceedings of Formal Description of Slavic Languages 8.5, Brno 2010, 271–289. Frankfurt am Main et al.: Peter Lang. Borik, Olga. 2002. Aspect and Reference Time. Utrecht: Utrecht University dissertation. Braginsky, Pavel & Rothstein, Susan. 2008. Vendler Classes and the Russian Aspectual System. In Journal of Slavic Linguistics 16 (1). 3–55. Flier, Michael. 1975. Remarks on Russian Verbal Prefixation. The Slavic and East European Journal 19 (2). 218–229. Flier, Michael. 1985. Syntagmatic Constraints on the Russian Prefix pere-. In Michael Flier & Richard Brecht (eds.), Issues in Russian Morphosyntax, 138–155. Columbus, Ohio: Slavic Publishers. Fowler, George & Yadroff, Michael. 1993. The Argument Status of Accusative Measure Nominals in Russian. Journal of Slavic Linguistics 1 (2). 251–279. Gehrke, Berit. 2008. Ps in Motion: On the Semantics and Syntax of P elements and Motion Events. Utrecht: LOT Publications.
104 | Kristina Gehring (Kaliningrad)
Hay, Jennifer, Kennedy Christopher & Levin, Beth. 1999. Scalar Structure underlies Telicity in “Degree Achievements”. In Tanya Mathews & Devon Strolovitch (eds.), Proceedings of Semantics and Linguistic Theory 9, 127–144. Cornell University, Ithaca: Cornell Linguistics Circle Publications. Isačenko, Aleksandr. 1962. Die russische Sprache der Gegenwart. Teil 1: Formenlehre. 2. Ausgabe. [The contemporary Russian language. Part 1: Morphology. 2nd edn.] München: Hueber. Kagan, Olga. 2013. Scalarity in the Domain of Verbal Prefixes. Natural Language and Linguistic Theory 31 (2). 483-516. Kuznecov, Sergei. 1998. Bol’šoj tolkovyj slovar’ russkogo jazyka [Big Russian dictionary with definitions]. Sankt Peterburg: Norint. LeBlanc, Nicholas. 2010. The Polysemy of an “Empty Prefix”: A Corpus-based Cognitive Semantic Analysis of the Russian Verbal Prefix po-. Chapel Hill: University of North Carolina dissertation. Maienborn, Claudia. 2000. Zustände – Stadien – stative Ausdrücke: Zur Semantik und Pragmatik von Kopula-Prädikativ-Konstruktionen [States – stages – stative expressions: On the semantics and pragmatics of copula-predicative-constructions]. In Linguistische Berichte 183, 271–307. Hamburg: Buske. Markova, Angelina. 2011. On the Nature of Bulgarian Prefixes: Ordering and Modification in Multiple Prefixation. Word Structure 4 (2). 244–271. McIntyre, Andrew. 2003. Preverbs, Argumentlinking and Verb Semantics: Germanic Prefixes and Particles. In Geert Booij & Jaap van Marle (eds.), Yearbook of Morphology 2003, 119– 144. Dordrecht, Boston, London: Kluwer. Ramchand, Gillian. 2004. Time and the Event: The Semantics of Russian Prefixes. Nordlyd 32 (2). 323–361. Romanova, Eugenia. 2004. Superlexical vs. Lexical Prefixes. Nordlyd 32 (2). 255–278. Romanova, Eugenia. 2006. Constructing Perfectivity in Russian. Tromsø: University of Tromsø dissertation. Russkaja Grammatika. 1960. Akademija Nauk SSSR, Institut russkogo jazyka. Tom 1: Fonetika i Morfologija. Moskva. Schoorlemmer, Maaike. 1995. Participial Passive and Aspect in Russian. Utrecht: Utrecht University dissertation. Svenonius, Peter. 2004. Slavic Prefixes inside and outside VP. Nordlyd 32 (2). 205–253. Tatevosov, Sergei. 2008. Intermediate Prefixes in Russian. In Andrei Antonenko, Christina Bethin & John Bailyn (eds.), Formal Approaches to Slavic Linguistics 16. The Stony Brook Meeting 2007, 423–445. Ann Arbor: Michigan Slavic Publications. Tatevosov, Sergei. 2009. Množestvennaja Prefiksacija i anatomija russkogo glagola [Multiple prefixation and anatomy of the Russian verb]. In Ksenija Kiseleva, Vladimir Plungjan, Ekaterina Rachilina & Sergei Tatevosov (eds.), Korpusnye issledovanija po russkoj grammatike [Corpus studies on Russian grammar], 92–145. Moskva: Probel. Tatevosov, Sergei. 2011. Severing Perfectivity from the Verb. Scando Slavica 57 (2). 216–244. Wiland, Bartosz. 2012. Prefix Stacking, Syncretism, and the Syntactic Hierarchy. In Markéta Ziková & Mojmír Dočekal (eds.), Slavic Languages in Formal Grammar. Proceedings of Formal Description of Slavic Languages 8.5, Brno 2010, 307–324. Frankfurt am Main et al.: Peter Lang. Žaucer, Rok. 2009. A VP-internal/Resultative Analysis of 4 “VP-External” Uses of Slavic Verbal Prefixes. Ottawa: University of Ottawa dissertation.
Semantic and morphosyntactic features of the Russian verbal prefix pere- ‘over’ | 105
Žaucer, Rok. 2012. The Syntax of Perdurative-prefixed Verbs and the VP-internal/VP-external Prefix Distinction. In Markéta Ziková & Mojmír Dočekal (eds.), Proceedings of Formal Description of Slavic Languages 8.5, Brno 2010, 339–355. Frankfurt am Main et al.: Peter Lang. Žaucer, Rok. 2013. A Case for Result-Modifying Prefixes. In Formal Approaches to Slavic Linguistics 20. Ann Arbor: Michigan Slavic Publications.
Claudia Hurtig (Leipzig)
Zur Darstellung lexikalischer Eigenschaften belarussischer Prädikativa in einsprachigerklärenden Wörterbüchern 1 Einführung Dieser Beitrag widmet sich der Darstellung lexikalischer Eigenschaften ausgewählter belarussischer Lexeme im traditionellen einsprachig-erklärenden Wörterbuch Tlumačal’ny sloŭnik belaruskaj movy (dt. ‚Erklärendes Wörterbuch der belarussischen Sprache‘, im Weiteren TSBM)1. Der Fokus der Untersuchung liegt auf Einträgen für Prädikativa, wobei sich die Begriffserklärung für Prädikativum in slavischen Sprachen von der in der deutschen Grammatik unterscheidet.2
|| 1 Das genannte Wörterbuch kann stellvertretend für die einsprachig-erklärende belarussische Lexikographie herangezogen werden. Die fünfbändige Akademieausgabe des Tlumačal’ny sloŭnik belaruskaj movy (TSBM) in sechs Büchern ist das erste und bisher einzige seiner Art wissenschaftliche einsprachige ‚Großwörterbuch‘, welches alle Stile der belarussischen Sprache berücksichtigt. Dieser Thesaurus, welcher von einem Autorenkollektiv der Belarussischen Akademie der Wissenschaften erarbeitet wurde und im Zeitraum von 1977 bis 1984 in Minsk erschien, hat aufgrund noch fehlender Neuauflagen und ähnlicher Großprojekte bis heute Gültigkeit als normatives Lexikon der belarussischen Sprache, was m. E. die Auswahl dieses Wörterbuches für die vorliegende Untersuchung hinreichend begründet. Eine Neuauflage des TSBM wird derzeit im Minsker Verlag „Belaruskaja Navuka“ in der Belarussischen Akademie der Wissenschaften teilweise zum Druck vorbereitet; die neue Auflage soll insgesamt 15 Bände umfassen (Informationsstand 09.01.2016). 2 Den Terminus ‚Prädikativum‘ verwende ich in Anlehnung an Isačenko (1954: 359; Isačenko 2 2003: 279) als Bezeichnung für eine lexikalische Klasse in den (ost)slavischen Sprachen. Es handelt sich hierbei um eine historisch entstandene, relativ heterogene Gruppe non-verbaler lexikalischer Einheiten mit unterschiedlichen morphologischen Eigenschaften, deren einzige syntaktische Funktion es ist, im Satz (in Verbindung mit einer Kopula) als Prädikat zu fungieren. In der traditionellen Grammatik wird anstelle von ,Prädikativum‘ häufig von ‚Zustandskategorie‘ (russ. kategorija sostojanija, belar. katėhoryja stanu, vgl. Ščerba 1928: 17; Galkina-Fedoruk 1957; Galkina-Fedoruk (Hrsg.) 1964: 202 ff.; Vinogradov 1972: 319 ff.; Šuba 1975: 30 ff.; Šanskij (Hrsg.) 1981: 382 ff.; Šuba 1987: 270 ff.; Butyrina et al. 1998 u.a.) oder ‚prädikativen Adverbien‘ (russ. predikativnye narečija, belar. prėdykatyŭnyja prysloŭi, vgl. Vinogradov et al. 1960: 630 ff, Atrachovič & Bulachaŭ (rėd.) 1962: 441 ff.; Burak 1974: 198 ff.; Biryla (rėd.) 1987; Blinava et al. 1994: 130 ff.; Novikov et al. 2001: 573–574, Sudnik & Kryŭko (rėd.) 1996, 21999, 32002, 42005 u.a.) ge-
DOI 10.1515/9783110517873-006
Zur Darstellung lexikalischer Eigenschaften belarussischer Prädikativa | 107
Die Prädikativa blieben in der lexikalischen Forschung bislang weitgehend unberücksichtigt. Selbst in gängigen ostslavischen Wörterbüchern fristen sie bis heute eine bescheidene Existenz. Traditionelle Wörterbücher und Grammatikbeschreibungen enthalten sehr wenige Informationen zur Syntax und Semantik der Prädikativa; wissenschaftliche Untersuchungen zu deren lexikalischen Eigenschaften existieren bisher kaum. Die Darstellung von Prädikativa in traditionellen Wörterbüchern wird selten hinterfragt. Einen ersten Ansatz einer solchen Untersuchung zu den Prädikativa im Russischen, wenn auch nicht aus semantischsyntaktischer Sicht, liefert z.B. Otkupščikova (1995). Als erste Untersuchungen zur wissenschaftlichen Beschreibung lexikalischer Eigenschaften der Prädikativa können u.a. folgende Aufsätze und Studien gelten: Fehrmann et al. (2007); Fehrmann & Junghanns (in Vorbereitung); Junghanns (2012); Junghanns & Lenertová (in Vorbereitung); Pitsch (2012a, 2012b); Hurtyh (2012); Hurtig (in Vorbereitung). Hier schließt sich die vorliegende Untersuchung an. Ausgehend von den linguistischen Anforderungen an einen wissenschaftlichen Lexikoneintrag, wie sie mit dem Begriff der lexikalischen Eigenschaften3 hinreichend formuliert werden, wird nach einem Überblick über die allgemeine Darstellung sprachlicher Daten im TSBM unter besonderer Berücksichtigung der Prädikativa (Abschnitt 2) der
|| sprochen. In ihrem syntagmatischen Verhalten sind die (ost)slavischen Prädikativa uneinheitlich. Einige können bzw. müssen einen Infinitiv zu sich nehmen, andere nicht, sie können verschiedene Nominalphrasen bei sich haben usw. Eine Besonderheit der Prädikativa ist, dass sie i.d.R. nicht mit einem personalen Subjektargument im Nominativ verknüpft werden können. Die personale Ergänzung bei Prädikativa steht meist im Dativ, z. T. sind aber auch andere personale Komplemente wie z.B. Präpositionalphrasen zulässig. Einige Prädikativa lassen sogar zwei Ergänzungen, Dativ- und Nominativ-Komplement, zu, wie es die Analyse des Prädikativums patrėbna (dt. ‚(man) muss; (man) braucht‘) in 3.4 zeigen wird. 3 Unter lexikalischen Eigenschaften verstehe ich die Gesamtheit der für einen Lexikoneintrag relevanten Merkmale eines Lexems bzw. einer lexikalischen Einheit. Nach Bierwisch (1996: 135) sind lexikalische Einheiten „Datenstrukturen, die drei Arten von Informationen verbinden: (a) die Phonetische Form PF (E) der Einheit E; (b) die Semantische Form SF (E) der Einheit E; (c) die Grammatische Form GF (E) der Einheit E“. Die vorliegende Untersuchung beschäftigt sich nur mit (b) und (c). Für Bierwisch ist SF (E) „eine aus Konstanten und Variablen bestehende FunktorArgument-Struktur, die die begriffliche Interpretation sprachlicher Ausdrücke steuert, und GF (E) determiniert durch syntaktische und morphologische Merkmale die grammatische Struktur komplexer Ausdrücke.“ (Bierwisch 1996: 135). Mit anderen Worten, die lexikalischen Eigenschaften eines Lexems umfassen neben den hier nicht berücksichtigten phonologischen und phonetischen Eigenschaften wie z.B. Lautfolge und Wortakzent die Kategorisierungsmerkmale (morphologischen Eigenschaften) wie Wortart und Flexionsschema, die kombinatorischen Eigenschaften wie syntaktische Valenz und syntaktische Beschränkungen und schließlich die semantischen Eigenschaften wie Polysemie, semantische Rollenstruktur und Argumentstruktur.
108 | Claudia Hurtig (Leipzig)
Aufbau zweier traditioneller Lexikoneinträge analysiert (Abschnitt 3). Wie sich herausstellen wird, sind die Stichwortartikel im TSBM in syntaktischer und semantischer Hinsicht unvollständig. Deshalb gebe ich für die von mir untersuchten Prädikativa trėba und patrėbna (dt. beide ‚(man) muss; (man) braucht‘) zunächst eine Übersicht über die jeweils mögliche Komplementierung und stelle hiernach erste Überlegungen zu deren Argumentstruktur an. Abschließend fasse ich meine Untersuchungsergebnisse zusammen und gebe einen Ausblick auf weitere Forschungen (Abschnitt 4). Ziel meines Beitrags ist es, auf Probleme in der traditionellen belarussischen Lexikographie hinzuweisen und Vorschläge für die Optimierung eines Lexikoneintrags der lexikalischen Klasse Prädikativum einzubringen.
2 Aufbau eines typischen Lexikoneintrags im TSBM Die Stichwörter (Lemmata) sind in alphabetischer Reihenfolge angeordnet. Grammatische, phonetische oder Akzent-Varianten werden dabei durch die koordinierende Konjunktion i (dt. ‚und‘) angegeben, vgl. Beispiel (1). Alle Stichwörter und deren Varianten sind mit einem Akzent versehen, welcher graphisch durch einen Akut (´) über dem betonten Vokal markiert ist.4, 5
|| 4 Bei flektierbaren Wörtern wird der Akzent in den neben der Nennform angegebenen Wortformen lediglich dann angegeben, wenn sich die Akzentverhältnisse bei der Flexion ändern oder besonders sind: (i) (ii)
НЕАБХÓДНЫ, -ая, -ае. […] ‚nötig, notwendig, unentbehrlich‘ ДАРАГÍ, -áя, -óе […] ‚teuer, wertvoll; kostbar; lieb, geliebt‘
(TSBM, Band 3: 347) (TSBM, Band 2: 138)
Für die unflektierbaren Prädikativa (sieht man von der Bildung von Steigerungsstufen bei einigen Klassen von Prädikativa ab, kann man in einem gewissen Sinne von Unflektierbarkeit sprechen) sind solche Akzentregelungen jedoch nicht von Relevanz und können deshalb hier außer Acht gelassen werden. 5 Die deutschen Übersetzungen in den folgenden Beispielen stammen von mir.
Zur Darstellung lexikalischer Eigenschaften belarussischer Prädikativa | 109
(1)
МÓХАВЫ, -ая, -ае і МАХÁВЫ2, -áя, -óе. […] ‚moosig; Moos-‘
(TSBM, Band 3: 175)6, 7
Das TSBM ist nicht nach dem Wortnest-Prinzip aufgebaut. So werden u.a. folgende Lexeme als eigenständige Lemmata angegeben (vgl. TSBM, Band 1: 7–8): – abstrakte Substantive auf -ннe, -aсць und -ствa: z.B. адлюстраванне; слабасць; умельства (dt. ‚Widerspiegelung; Schwäche; Geschicktheit‘) – Substantive mit Augmentativ- bzw. Diminutivsuffixen: z.B. тапарышча; мячык; сучок (dt. ‚große Axt; kleiner Ball; Ästchen‘) – Adjektive mit dem Suffix -aвa-, die die Bedeutung der Unvollständigkeit haben: z.B. велікаваты (dt. ‚ziemlich groß / ein bisschen zu groß‘) – die wichtigsten Possessiv- und Relationsadjektive: z.B. бацькаў; вавёрчын (dt. ‚väterlich / des Vaters; Eichhörnchen-‘) – nicht regulär gebildete Superlativformen von Adjektiven: z.B. малады – маладзейшы (dt. ‚jung – jünger / am jüngsten) und малодшы (Junior- / Unter-‘) – passive Partizipien mit präteritaler Bedeutung sowie andere Partizipien und Adverbialpartizipien, die in andere Wortarten (i.d.R. in Substantive) übergegangen sind oder Besonderheiten in der Bildung aufweisen: z.B. ранены; дзякуючы (dt. ‚Verletzter; dank‘) – Adverbien, darunter auch von Adjektiven gebildete, sofern sie sich in der Bedeutung von den Adjektiven unterscheiden: z.B. важна; добра (dt. ‚wichtig; gut‘) Homonyme werden im TSBM durch hochgestellte arabische Ziffern angegeben, wobei jedes homonyme Lexem ein eigenes Lemma aufweist:
|| 6 In den Beispielen (1)–(17) verwende ich die Original-Notation aus dem TSBM. Es erfolgt hier keine vollständige Übersetzung der kyrillischen Zitate, welche ausschließlich zu Demonstrationszwecken dienen. 7 In den aus dem Wörterbuch zitierten Beispielen werden entsprechend dem TSBM folgende belarussische Abkürzungen verwendet (in alphabetischer Reihenfolge): безас. – безасабовая форма (dt. ‚unpersönliche Form‘), без дап. – без дапаўнення (dt. ‚ohne Ergänzung‘), гл. – глядзі (dt. ‚siehe‘), Д – давальны склон (dt. ‚Dativ‘), ж. – жаночы род (dt. ‚feminin‘), з інф. – з інфінітывам (dt. ‚mit Infinitiv‘), М – месны склон (dt. ‚Lokativ‘), мн. – множны лік (dt. ‚Plural‘), Прысл. / прысл. – прыслоўе (dt. ‚Adverb‘), разм. – размоўнае слова (dt. ‚umgangssprachliches Wort‘), са злучн. – са злучнікам (dt. ‚mit Konjunktion‘), у знач. вык. / ў знач. вык. – у значэнне выказніка (dt. ‚in Prädikatsbedeutung‘), фр. – французскае слова (dt. ‚Wort französischer Herkunft‘).
110 | Claudia Hurtig (Leipzig)
(2)
ВÁРТА1, -ы, ДМ -рце, ж. 1. […] (TSBM, Band 1: 465) 2. […] ‚Wache, Wachposten‘ ВÁРТА2, безас. ў знач. вык. 1. […] Неабхoдна, трэба; мае сэнс; карысна. […] 2. […] ‚(es ist) nötig; (es ist) nützlich‘
Als Lemma dient traditionsgemäß die Grundform eines Wortes. Für Verben ist dies z.B. der Infinitiv, für Substantive der Nominativ Singular, für Adjektive die maskuline Singularform usw. Ein vollständiger Lexikoneintrag enthält laut Vorwort des TSBM folgende Elemente: i) Stichwort, ii) grammatische Charakteristik, iii) stilistische Charakteristik, iv) semantische Charakteristik, v) Illustrationsmaterial, vi) terminologische Wendungen und Phraseologismen, sowie vii) etymologische Information (vgl. TSBM, Band 1: 8). Im Folgenden sollen besonders die Elemente ii) und iv) näher erläutert werden. Alle anderen können aufgrund ihrer minderen Relevanz für die vorliegende Untersuchung sehr kurz nebenbei abgehandelt werden.
2.1 Grammatische Charakteristik des Lemmas Dem Lemma folgt zunächst eine grammatische Charakterisierung. In der Regel beginnt diese mit einem Hinweis auf die Wortart. Als eigenständige Wortarten (im TSBM als „grammatische Klassen“ aufgefasst (TSBM, Band 1: 9)) und somit als eigene Lemmata werden im TSBM folgende aufgeführt: Substantive (belar. назоўнікі), Adjektive (прыметнікі), Numeralia (лічэбнікi), Pronomina (займеннікi), Verben (дзеясловы), Partizipien (дзеепрыметнікi), Adverbialpartizipien (дзеепрыслоўi), Adverbien (прыслоўi), Präpositionen (прыназоўнікi), Konjunktionen (злучнікi), Partikeln (часціцы) und Interjektionen (выклічнікі) (vgl. TSBM, Band 1: 9). Die Kennzeichnung eines Wortes als eine bestimmte Wortart wird überwiegend über die Angabe bestimmter morphologischer Kategorien derselben erzielt. Für Substantive z.B. ist dies die Angabe verschiedener Kasus, mindestens aber des Genitivs Singular und die Angabe des Genus, wenn notwendig, auch ein Vermerk zum Numerus-Gebrauch sowie besondere (v. a. in akzentologischer Hinsicht) Pluralformen. Für Verben erfolgt mindestens die Angabe der grammatischen Endungen der 1., 2. und 3. Person Singular, die Angabe des Aspekts und
Zur Darstellung lexikalischer Eigenschaften belarussischer Prädikativa | 111
mitunter (d. h. nicht generell) die Anführung der Präterital-Formen und des Imperativs der 2. Person Singular. Für Adjektive, Partizipien, Ordinal-Numeralia und einige Pronomina erfolgt obligatorisch die Angabe der Genus-Endungen aller drei Genera (vgl. TSBM, Band 1: 9–10). Sofern die Kurzform eines Adjektivs gebräuchlich ist, wird auch diese sofort mit aufgeführt, ungeachtet dessen, dass es zwischen Lang- und Kurzform mitunter zu Bedeutungsverschiebungen kommen kann. Die entsprechenden grammatikalischen Angaben zur Flexion werden im Normalfall über die reine Flexionsendung angegeben, sofern der Wortstamm bei der Flexion keinen phonetischen, morphologischen oder akzentologischen Änderungen unterworfen ist. Andernfalls wird als grammatische Form ein Wortteil angeführt, welcher „eine korrekte Rekonstruktion des Gesamtwortes einschließlich seiner Akzentverhältnisse sicherstellt“ (TSBM, Band 1: 9).8 Eine explizite Benennung der Wortart, welcher das Lemma zuzuordnen ist, erfolgt lediglich für die „grammatischen Klassen“ (TSBM, Band 1: 9) der Numeralia (Vermerk: ліч. = лічэбнік), Pronomina (Vermerk: займ. = займеннік), Partizipien (Vermerk: дзеепрым. = дзеепрыметнік), Adverbialpartizipien (Vermerk: дзеепрысл. = дзеепрыслоўе), Adverbien (Vermerk: прысл. = прыслоўе), Präpositionen (Vermerk: прыназ. = прыназоўнік), Konjunktionen (Vermerk: злучн. = злучнік), Partikeln (Vermerk: часціца) und Interjektionen (Vermerk: выкл. = выклічнік). Für Substantive, Adjektive und Verben existieren außer den Hinweisen zu den genannten morphologischen Kategorien keine expliziten Angaben zur Wortart. Prädikativa werden, wie aus den obigen Angaben ersichtlich, im TSBM nicht als eigenständige Wortart gekennzeichnet und sind demzufolge mehrheitlich nicht mit einem eigenen Stichwort verortet. In jedem Fall werden sie als sogenannte ‚unpersönliche, prädikativ verwendete Formen‘ mit dem Vermerk безас. ў знач. вык. / безас. у знач. вык. (= безасабовая форма ў значэнні выказніка) (dt. ‚unpersönliche Form in der Bedeutung des Prädikats‘)9 oder у знач. вык. (= у значэнні выказніка) (dt. ‚in Prädikatsbedeutung‘) vermeintlich grammatisch gekennzeichnet, nämlich als Lexeme mit zulässiger oder ausschließlicher prädikativer Verwendung, vgl. die Beispiele (2)–(14). Unter безас. (= безасабовая
|| 8 Auf Einzelheiten der grammatischen Charakterisierung der genannten flektierbaren Wortarten wird in der vorliegenden Untersuchung nicht eingegangen. Auch werden sie nicht im Einzelnen kritisch erörtert. Allgemeine kritische Anmerkungen dazu siehe 2.3. Meine Ausführungen beschränken sich im Weiteren auf die Darstellung der Prädikativa. 9 Der grammatische Vermerk erscheint im TSBM in zwei Varianten: mit und ohne unsilbisches u (belar. у нескладовае, graphematisch ў). Das unsilbische u in der Abkürzung rührt fälschlicherweise von der ausgeschriebenen Bezeichnung, in der u auf Vokal folgt (vgl. belar. безасабовая форма ў значэнні выказніка). Richtig wäre jedoch die Abkürzung ohne unsilbisches u, da dort u nicht auf Vokal folgt: безас. у знач. вык.
112 | Claudia Hurtig (Leipzig)
форма) (dt. ‚unpersönliche Form‘) wird hier gemäß der traditionellen Grammatik der Gebrauch einer Form ohne Bezug auf eine ‚Person‘ im Sinne eines Handlungsurhebers bzw. der Gebrauch ohne Subjekt im Nominativ verstanden (vgl. Z.B. die Definition des „unpersönlichen Satzes“ in Scjacko et al. (1990); Biryla & Šuba (rėd) 1986: 223 / Anmerkung). Die meisten Prädikativa, hier v. a. die quantitativ sehr umfangreiche Gruppe der Prädikativ-Formen auf -a, welche i.d.R. homonym zu Adverbialformen sind, werden, konformgehend mit der traditionellen belarussischen Grammatiktheorie (vgl. hierzu den Begriff der ‚prädikativen Adverbien‘ in Abschnitt 1), als lexikalische Unterformen von Adverbien charakterisiert und stets unter dem entsprechenden Adverb-Eintrag gelistet. Die Zugehörigkeit des Lexems zur lexikalischen Klasse der Adverbien und die vermeintlich syntaktische Verwendungsart derselben als Prädikativum (Adverb vs. prädikative Form) wird dabei in der Art einer lexikalischen Bedeutungsunterscheidung durch Nummerierung mit arabischen Zahlen geregelt, vgl. Beispiel (3). Die lexikalisch übergeordnete Wortart wird über die übliche Adverb-Kennzeichnung mit dem Vermerk прысл. (= прыслоўе) (dt. ‚Adverb‘) markiert, vgl. u.a. die Beispiele (3), (8) und (11)–(14).10 (3)
ЖÓЎТА. 1. Прысл. да жóўты. 2. безас. ў знач. вык. […] ‚gelb‘
(TSBM, Band 2: 258)
In ähnlicher Weise wird mit den aus Substantiven hervorgegangenen Prädikativa verfahren, welche – mit Ausnahme einiger Sonderformen wie Beispiel (2) – als Unterform bzw. Unter-‚Bedeutung‘ der jeweiligen Substantive innerhalb deren Eintrag aufgeführt werden, vgl. Beispiel (4): (4)
ЛЯНÓТА, -ы, ДМ -нóце, ж. 1. Адсутнасць (TSBM, Band 3: 74) жадання працаваць, […] 2. у знач. вык., […] Разм. Не хочацца, няма ахвоты. […] ‚Faulheit, Trägheit‘
Sofern ein prädikativ gebräuchliches (non-verbales, non-adjektivisches) Lexem neben der prädikativen keine andere syntaktische Verwendung zulässt und so-
|| 10 Hier und im Folgenden arbeite ich mit Unterstreichungen zur Hervorhebung der jeweils relevanten Komponenten des Lexikoneintrags.
Zur Darstellung lexikalischer Eigenschaften belarussischer Prädikativa | 113
mit (nach Ansicht der traditionellen Grammatik) keiner Adverb- oder Substantivform zugeordnet werden kann, wird es als eigenständiges Lemma erfasst, dabei jedoch konsequenterweise keiner konkreten Wortart zugeordnet. In diesen Fällen bezieht sich der Hinweis auf die prädikative Verwendung auf das gesamte Lemma, was zumeist nur mit у знач. вык. (dt. ‚in Prädikatsbedeutung‘) markiert wird, vgl. u.a. die Beispiele (5) und (9), nicht selten aber auch mit безас. у знач. вык. (dt. ‚unpersönliche Form in Prädikatsbedeutung‘), vgl. u.a. die Beispiele (7) und (10). Eine Logik in der jeweiligen Merkmalszuweisung scheint es hier nicht zu geben. (5)
ТРЭБА1, у знач. вык. 1. […] 2. […] 3. безас. […] ‚(man) muss; (man) braucht‘
(TSBM, Band 5 / 1: 539)
Mitunter werden Prädikativa im TSBM als zu anderen Wortarten homonyme Lexeme separat als eigenständiges Stichwort gelistet, vorzugsweise im Falle von zu Substantiven klanggleichen Formen. Das Prädikativum steht dabei i.d.R. an zweiter Stelle, vgl. Beispiel (2). Unterscheidet sich ein Prädikativum in seiner Grundbedeutung nicht vom entsprechenden Substantiv, so erhält es, wie bereits in Beispiel (4) gezeigt, kein eigenes Stichwort, vgl. auch Beispiel (6). (6)
ПАРÁ, -ы ´ , мн. пóры […]; ж. 1. […] 2. у знач. вык. ‚1. Zeit; 2. es ist (an der) Zeit‘
(TSBM, Band 4: 35)
Zur grammatischen Charakteristik der Lexeme gehören neben der Zuordnung zu einer Wortart Angaben zur syntaktischen Verwendung des jeweiligen Lexems bzw. seiner Bedeutungen. Speziell für Prädikativa werden als mögliche Ergänzungen Infinitive (Vermerk: з інф. = з інфінітывам (dt. ‚mit Infinitiv‘)), vgl. die Beispiele (7), (9) und (10), Dativ-Determinansphrasen, Akkusativ-Determinansphrasen, Genitiv-Determinansphrasen und konjunktional eingeleitete Nebensätze mit што (dt. ‚dass‘) bzw. калі (dt. ‚wenn; immer dann, wenn‘) (Vermerk: са злучн. “што”, “калі” = са злучнікамі […]) (dt. ‚mit den Konjunktionen što, kali‘), vgl. Beispiel (11), aufgeführt. Die nominalen Ergänzungen werden dabei traditionsgemäß als Rektionskasus angesehen (vgl. dazu auch Biryla & Šuba (Hrsg.) 1986: 144, 220) und durch Fragepronomina im entsprechenden Kasus ausgedrückt: каму (dt. ‚wem‘) für Dativ, vgl. die Beispiele (8), (9) und (12), каго-што (dt. ‚wen-was‘) für Akkusativ, vgl. Beispiel (9), und каго-чаго (dt. ‚wessen‘) für
114 | Claudia Hurtig (Leipzig)
Genitiv, vgl. Beispiel (10). Weist ein Prädikativum in einer Bedeutung keine Ergänzung auf, dann fehlt die Angabe, vgl. die Beispiele (8) und (11); in wenigen Fällen steht auch der Vermerk без дап. (= без дапаўнення) (dt. ‚ohne Ergänzung‘), vgl. Beispiel (12). (7) (8)
(9)
(10)
(11)
(12)
МÓЖНА, безас. у знач. вык. 1. з інф. […] (TSBM, Band 3: 171) ‚(man) kann, (es ist) möglich; (man) darf‘ (TSBM, Band 5 / 2: 209) ХÓЛАДНА. 1. Прысл. […] 2. безас. у знач. вык. […] 3. безас. у знач. вык., каму […] ‚kalt; kühl‘ ТРЭБА1, у знач. вык. 1. з інф. […] (TSBM, Band 5 / 1: 539) 2. каго-што і з інф. […] 3. безас., каму. […] ‚(man) muss; (man) braucht‘ ШКАДÁ. 1. безас. у знач. вык., каго-чаго або з (TSBM, Band 5 / 2: 364) інф. […] 2. безас. у знач. вык., чаго або з інф. […] ‚(es ist) schade, (es ist) bedauerlich, (es) tut (jmdm.) leid‘ (TSBM, Band 2: 185) ДÓБРА. 1. Прысл. […] 2. […] 3. безас. у знач. вык. […] 4. безас. у знач. вык, са злучн. “што”, “калі”. […] ‚gut; schön‘ ГÓРАЧА. 1. Прысл. […] (TSBM, Band 2: 70) 2. безас. у знач. вык. […] 3. безас. у знач. вык., каму і без дап. […] 4. безас. у знач. вык, каму і без дап. […] ‚heiß; schwül‘
Mitunter finden sich im TSBM Prädikativformen ohne explizite syntaktische Angaben. Frequente syntaktische Verwendungen der Formen werden gegebenenfalls implizit über das Belegbeispiel angezeigt, vgl. die Beispiele (13) und (14).
Zur Darstellung lexikalischer Eigenschaften belarussischer Prädikativa | 115
(13)
(14)
(TSBM, Band 1: 452) ВÁЖНА. 1. Прысл. […] 2. у знач. вык. Істотна ў якіх-н. адносінах. Важна было, каб поезд па той ці іншай прычыне прыпыніўся там, дзе трэба. Якімовіч. ДЗÍЎНА. 1. Прысл. […] (TSBM, Band 2: 175) 2. безас. у знач. вык. Незвычайна, выклікае здзіўленне. [Юрыю] самому потым дзіўна было, як гэта шпарка ўсё адбылося. Чорны. Як гэта хораша і дзіўна Вокам абняць далячынь. Звонак. ‚seltsam, merkwürdig, erstaunlich‘
Der grammatischen (d. h. morphologischen und gegebenenfalls syntaktischen) Charakteristik des Lemmas folgt, wenn nötig, eine stilistische Charakterisierung desselben. Gekennzeichnet werden umgangssprachliche (belar. Vermerk: разм. = размоўнае слова), regionale (Vermerk: абл. = абласное слова), buchsprachliche (Vermerk: кніжн. = кніжны стыль), veraltete (Vermerk: уст. = устарэлае слова) bzw. historische (Vermerk: гіст. = гістарычны тэрмін) und poetische (Vermerk: паэт. = паэтычнае слова) Wörter sowie Wörter gehobenen Stils (Vermerk: высок. = высокі стыль) und spezielle fachsprachliche Termini (Vermerk: спец. = спецыяльны тэрмін). Darüber hinaus gibt es auch Angaben für eventuelle emotionale Wertungen, wie z.B. Koseformen (Vermerk: ласк. = ласкальная форма), derbe (Vermerk: груб. = грубае слова), verächtliche (Vermerk: пагард. = пагардлівае) bzw. herabsetzende (Vermerk: зневаж. = зневажальнае) Ausdrücke u.a. (vgl. TSBM, Band 1: 10–11, 25–26).
2.2 Semantische Charakteristik des Lemmas Den grammatischen und gegebenenfalls stilistischen Angaben folgen die semantischen Explikationen. Hierbei werden verschiedene Explikationstypen unterschieden: Am häufigsten sind (i) Explikationen, die aus einer kurzen Formulierung der realen Bedeutung des Lexems bestehen. Daneben gibt es (ii) Explikationen, die „nicht nur die Semantik des […] Stamms, sondern auch die Bedeutung der Affixe“ (TSBM, Band 1: 11) erläutern, (iii) Explikationen, in denen die Bedeutung des formbildenden Affixes im Vordergrund steht, das an einen realen Stamm herantritt, (iv) Explikationen, die aus einer Synonymreihe bestehen, (v) Explikationen, in denen lediglich die semantische Übereinstimmung des Stichwortes mit einem anderen Wort betont wird, wobei die grammatische und stilistische Charakteristik unterschiedlich sein kann und (vi) Explikationen, in denen nicht die Semantik des Wortes geklärt wird, sondern lediglich dessen
116 | Claudia Hurtig (Leipzig)
grammatische Relation zu demjenigen Wort, mit welchem es in der Bedeutung übereinstimmt (vgl. TSBM, Band 1: 11). Letztere betreffen lediglich verbale Wortformen, wie z.B. Aspekt-Formen, Partizipien, reflexive Passivformen oder Aktionsarten, welche im TSBM allesamt als eigenständige Lemmata aufgelistet werden. Werden bei einem Lemma mehrere Bedeutungen unterschieden, sind diese innerhalb des Eintrags fortlaufend mit arabischen Zahlen nummeriert. Die Reihenfolge der Bedeutungsvarianten wird sowohl von der logischen Beziehung, als auch von der Gebräuchlichkeit in der Standardsprache diktiert. An erster Stelle wird dabei stets die direkte Bedeutung angegeben, hernach Bedeutungsvarianten mit den eingrenzenden stilistischen Vermerken разм. (= размоўнае слова), кніжн. (= кніжны стыль), высок. (= высокі стыль) und спец. (= спецыяльны тэрмін) (dt. ‚umgangssprachlich, buchsprachlich, gehoben, fachsprachlich‘), und zum Schluss folgen übertragene Bedeutungen (belar. Vermerk: перан. = пераноснае значэнне). Jede Bedeutung bzw. Bedeutungsnuance verfügt über eine eigene Explikation. Bedeutungsnuancen werden durch das Zeichen „//“ voneinander abgetrennt (vgl. TSBM, Band 1: 12). Jede Bedeutung bekommt ihre eigene syntaktische Charakteristik, sofern sich diese von jenen der anderen Bedeutungen unterscheidet, vgl. dazu die Beispiele (7)–(12). Gilt eine syntaktische Verwendungsweise für alle unter dem Lemma explizierten Bedeutungen, dann erfolgt deren Angabe für das gesamte Lemma noch vor der ersten ausgegliederten Bedeutung:11 (15)
ЗАГНÁЦЬ, […], каго-што. 1. […] 2. […] ‚(hinein)jagen, hetzen‘
(TSBM, Band 2: 301)
Die Bedeutungsexplikation wird in der Regel durch Belegbeispiele für das Lemma bzw. die entsprechende Bedeutungsvariante aus der schöngeistigen Literatur (= Illustrationsmaterial) ergänzt. Die aus „maßgebenden Quellen“ (TSBM, Band 1: 12) stammenden Zitate sollen die Zugehörigkeit des jeweiligen Wortes zur Standardsprache unterstreichen, dessen richtige standardsprachliche Verwendung dokumentieren, die Verwendungssphäre bestätigen und die typischsten syntaktischen Verwendungsweisen (‚Wortumgebungen‘) aufzeigen (vgl. TSBM, Band 1: 12). Mitunter beschränkt sich das Illustrationsmaterial auf gebräuchliche
|| 11 Für Prädikativa wurden keine Beispiele gefunden. In der Regel wird bei diesen jeder angenommenen Bedeutung eine eigenständige syntaktische Charakteristik zugewiesen, vgl. die Beispiele (7)–(12).
Zur Darstellung lexikalischer Eigenschaften belarussischer Prädikativa | 117
Zwei- bis Dreiwortverbindungen. Als illustrierendes Material werden außerdem teilweise Sprichwörter und Redewendungen aufgeführt. Nach der letzten Bedeutungsexplikation eines Eintrags werden terminologische Wendungen und Phraseologismen angegeben und erläutert, welche vom übrigen Artikel durch die Zeichen „○“ bzw. „◊“ abgetrennt sind. Die Explikation des jeweiligen Phraseologismus bzw. der terminologischen Wendung erfolgt unter dem Hauptwort der Wendung. Unter den jeweils anderen Einzelwörtern der Wendung, welche ihrerseits selbständige Lemmata im TSBM bilden, wird lediglich die entsprechende Wendung ohne Explikation aufgeführt (vgl. TSBM, Band 1: 12) und mit dem Vermerk гл. (= глядзі) (dt. ‚siehe‘) auf das Hauptwort verwiesen: (16)
КÉПСКА […] ◊ Кепска кончыць гл. кончыць. ‚schlecht, schlimm; übel‘
(TSBM, Band 2: 680)
Bei entlehnten Wörtern wird der Eintrag schließlich durch eine in eckigen Klammern stehende etymologische Information mit dem Hinweis auf die Ursprungssprache, die ursprüngliche (originale) Schreibweise und gegebenenfalls die ursprüngliche Bedeutung abgerundet (vgl. TSBM, Band 1: 13):12 (17)
БЛАНК […] [Фр. Blanc – белы.] ‚Formular; Vordruck‘
(TSBM, Band 1: 383–384)
2.3 Kritische Anmerkungen und Fazit Im Hinblick auf die Frage, wie die als für einen vollständigen und wissenschaftlichen Anforderungen genügenden Lexikoneintrag relevant herausgestellten lexikalischen Informationen (vgl. Fn 3) im TSBM dargestellt werden, kann Folgendes gesagt werden. Die grammatischen Angaben im TSBM sind sehr unvollständig. Kombinatorische (syntaktische) Eigenschaften der Prädikativa werden nur teilweise berücksichtigt und kaum im Zusammenhang mit den semantischen Eigenschaften gesehen, sodass sich mitunter relativ willkürliche Bedeutungsuntergliederungen in den Einträgen ergeben. Die kategorialen Eigenschaften (d.h. die Zuordnung zu einer lexikalischen Kategorie bzw. Wortart) werden in Bezug auf die Prädikativa m. E. nicht korrekt bzw. verzerrt dargestellt. Als übergeordnete Wortart wird für Prädikativa im TSBM die Kategorie des Adverbs angenommen, was als unhaltbar
|| 12 Für Prädikativa wurden hier keine Beispiele gefunden.
118 | Claudia Hurtig (Leipzig)
zurückgewiesen werden kann, da sich ein Adverb im Belarussischen (bzw. Ostslavischen) per definitionem syntaktisch immer an ein anderes Wort anlehnt und im Satz als Adverbiale fungiert, keinesfalls jedoch als Prädikat, welches sich seinerseits syntaktisch gesehen nicht anderen Wortformen unterordnet. Folglich ist die im TSBM in den Einträgen verwendete Klassifizierung der Prädikativa als ‚unpersönliche Formen‘ von Adverbien „in Prädikatsbedeutung“ (TSBM, Band 1: 10 u.a.) absurd. Es ist allerdings zu betonen, dass die im TSBM vorgefundenen Daten konform gehen mit der Auffassung der belarussischen traditionellen Grammatik und Lexikographie, welche die Prädikativa nach wie vor als Adverbien betrachtet (einschlägige Literatur dazu siehe Begriffsklärung in der Einführung).13 Abgesehen davon birgt die Handhabung der Prädikativa im TSBM noch ein weiteres Problem. Die grammatische Charakterisierung besteht hier genau genommen nicht wie bei anderen Wortarten aus einem direkten Hinweis auf die Zugehörigkeit zu einer (eigenständigen) Wortart (also zu einer grammatisch-lexikalischen Klasse im Sinne der traditionellen Grammatik), sondern aus einem Hinweis auf die syntaktische Funktion des Lexems. Damit werden zwei unterschiedliche Klassifizierungsprinzipien angewandt: das semantische und das syntaktische. Die Prädikativa (v. a. die, welche homonym zu Lexemen anderer Wortarten sind) werden als eine Art lexikalische Bedeutung der Adverbien, seltener der Substantive, oder anderer (z. T. nicht explizierter) Wortklassen präsentiert (vgl. die arabische Nummerierung u.a. in den Beispielen (3), (4), (8) und (11)–(14), welche im TSBM sonst, d. h. bei anderen Lemmata, zur Unterscheidung der lexikalischen Bedeutungen verwendet wird), werden aber gleichzeitig als besondere syntaktische Verwendung dieser Lexeme dargestellt. Die Behandlung der Prädikativa im TSBM – als lexikalische ‚Bedeutung‘ anderer Wortklassen – lässt sich mit der in der russischen Akademiegrammatik vertretenen Auffassung der Prädikativa als „syntaktische Derivate“ (Švedova 1970: 309) in Verbindung bringen. Eine syntaktische Funktion, wie z.B. die des Prädikats, ist jedoch keine semantische Größe und kann deshalb m. E. nicht als ‚Bedeutung‘ (vgl. belar. u značėnni vykaznika – dt. ‚in Prädikatsbedeutung‘) oder ‚Derivat‘ bezeichnet werden. Allenfalls könnte der Vermerk bei den entsprechenden Lexemen lauten: ‚u funkcyi vykaznika‘ (dt. ‚in Prädikatsfunktion‘), allerdings dennoch nicht in Bezug auf ein Adverb. Gesetzt den Fall, ein Adverb könnte im Satz als Prädikat fungieren, brächte jedoch selbst diese Variante der grammatischen Charakteristik meiner Meinung nach
|| 13 Auch die aus Substantiven hervorgegangenen Prädikativa wie час (dt. ‚(es ist) Zeit‘), грэх (dt. ‚(es ist (eine)) Sünde‘), лянота (dt. ‚zu faul sein‘, wörtlich: ‚(es ist) Faulheit‘) u. ä. werden, sofern sie nicht ganz außer Acht gelassen werden, traditionell als Adverbien betrachtet; alternativ verbleiben sie in der Klasse der Substantive.
Zur Darstellung lexikalischer Eigenschaften belarussischer Prädikativa | 119
keine befriedigende Lösung für das Lexikon. Alle Wörter übernehmen im Satz bestimmte syntaktische Funktionen, fast alle Wortarten können in verschiedenen syntaktischen Funktionen auftreten. So kann z.B. ein Substantiv oder ein Adjektiv auch als Prädikat fungieren. Gleichwohl wird diese Funktion nicht explizit in den entsprechenden Wörterbuchartikeln aufgeführt, da die lexikalische Bedeutung des Lexems in der Regel gleich bleibt. Warum also sollte dies bei den vermeintlichen ‚prädikativen Adverbien‘ (einschlägige Literatur zu diesem Begriff siehe Einführung / Begriffsklärung) getan werden? Im Zusammenhang mit der fragwürdigen Handhabung der Prädikativa innerhalb der Lexikoneinträge stellt sich auch die Frage, warum im TSBM z.B. das Partizip (belar. дзеепрыметнік) und das Adverbialpartizip (belar. дзеепрыслоўе) als eigenständige „grammatische Klassen“ (TSBM, Band 1: 9) (hier im Sinne von Wortarten) zusätzlich zu den üblichen zehn Wortarten als selbständige Lemmata aufgeführt werden14, wohingegen den Prädikativa dieses Los nicht zuteil wird. Sowohl Partizipien als auch Adverbialpartizipien sind grammatische Formen des Verbs und müssten konsequenterweise unter den entsprechenden Einträgen für Verben verortet werden. Aus den genannten Gründen ist die Darstellung des Wortschatzes im TSBM inkonsequent. Die konkrete syntaktische Charakterisierung der einzelnen Prädikativa ist unzureichend. Sehr häufig werden mögliche syntaktische Ergänzungen gar nicht angegeben, wenngleich sie durchaus im Illustrationsmaterial der einzelnen Einträge auftreten können. Mitunter fehlen Informationen zur Komplementierung der Lexeme ganz. So sind z.B. bei важна (dt. ‚(es ist) wichtig‘) (vgl. Beispiel (13)) keinerlei syntaktische Ergänzungen angegeben, obgleich das Lexem mit DativDeterminansphrase bzw. Präpositionalphrase (mit der Präposition dlja – dt. ‚für‘), Infinitiv, dass-Satz u.a. oder auch ohne Ergänzung verwendet werden kann, um nur ein Beispiel zu nennen. Für einen korrekten Gebrauch der Lexeme im Satz sind die syntaktischen Informationen allerdings unumgänglich. Der Wörterbuchbenutzer (sicherlich nicht nur der nichtmuttersprachliche, sondern u. U. auch der muttersprachliche, der etwa die Standardsprache nicht gut beherrscht oder aber einfach nur ein Wort zur Kontrolle nachschlagen will) erhält keine Klarheit über die möglichen Verwendungsweisen der Lexeme. Zuweilen ist unersichtlich, ob die entsprechenden syntaktischen Ergänzungen weggelassen wurden,
|| 14 Die Auffassung von Partizip und Adverbialpartizip als eigenständige Wortarten wird auch von einigen wenigen Grammatik- und Lehrbuchautoren des Russischen vertreten (vgl. z.B. Dolbik et al. 42008).
120 | Claudia Hurtig (Leipzig)
weil es sich dabei um hochfrequente Formen handelt, die möglichweise als Wissen vorausgesetzt werden. (Ich denke hier z.B. an die obligatorische Dativ-Determinansphrase bei einigen Gruppen von Prädikativa oder an Infinitive bei ModalPrädikativa u. ä.). In einem Lexikoneintrag, der wissenschaftlichen Anforderungen genügt, sollten alle notwendigen syntaktischen Angaben (Komplemente) erfasst sein. Auf die Angabe einer phonetischen Form der Lexeme wird im TSBM, wie auch in den meisten einsprachigen Wörterbüchern des Belarussischen und Russischen, ganz verzichtet. Es sollte wenigstens eine trivialisierte Notation derselben angegeben werden. Ein weiterer Nachteil des TSBM besteht darin, dass sehr viele Prädikativa gänzlich unberücksichtigt bleiben und selbst als ‚lexikalisch-syntaktische Variante‘ anderer Lexeme nicht genannt werden. Besonders betroffen sind hier von Adjektiven abgeleitete und negierte Formen: So findet man z.B. keinerlei Informationen zu solchen Prädikativformen wie aбразлівa (dt. ‚(es ist) verletzend, kränkend‘), нармальнa (dt. ‚(es ist) normal, üblich‘), неразумнa (dt. ‚(es ist) unklug‘) etc. Alles in allem wird die lexikalische Klasse der Prädikativa im TSBM zwar in großem Umfang berücksichtigt, jedoch trotz allem recht stiefmütterlich behandelt. Ein wesentlicher Grund dafür ist m. E. in der im Wörterbuch vorgenommenen Aufgliederung des Wortschatzes in Wortarten zu suchen. Die traditionelle belarussische Grammatik unterscheidet zehn Wortarten, welche sich auch im TSBM manifestieren: 1. назоўнік (Substantiv), 2. прыметнік (Adjektiv), 3. лічэбнік (Numerale), 4. займеннік (Pronomen), 5. дзеяслоў (Verb), 6. прыслоўе (Adverb), 7. прыназоўнік (Präposition), 8. злучнік (Konjunktion), 9. часціцы (Partikeln) und 10. выклічнік (Interjektion) (vgl. Atrachovič & Bulachaŭ (Hrsg.) 1962: 19; Biryla & Šuba (Hrsg.) 1985: 52). Die Prädikativa, welche besonders durch die Arbeiten von Ščerba (1928), Vinogradov (21972), Vinogradov et al. (red) (1960), Galkina-Fedoruk (1957, 1964) und Isačenko (1954, 1962) als eigenständige Wortart innerhalb des Wortartensystems des Russischen etabliert wurden (zur Geschichte der Wortart vgl. Isačenko 1962: 194–205) und als solche auch im Belarussischen angenommen werden können – wenngleich sie auch heute noch sehr umstritten sind –, bleiben in der oben erwähnten (traditionellen) Wortschatzgliederung gänzlich unberücksichtigt. Im TSBM sind sie dennoch zum großen Teil als lexikalische Einheiten vertreten, allerdings zumeist nicht als eigene Lemmata, sondern, wie bereits erwähnt, in der Regel als lexikalische bzw. grammatische (syntaktische?) ‚Unterklassen‘ bzw. Varianten der Vertreter anderer Wortarten: der Adverbien und z. T. der Substantive.
Zur Darstellung lexikalischer Eigenschaften belarussischer Prädikativa | 121
Allgemein ist zum Wortschatz im TSBM zu sagen, dass aus unersichtlichen Gründen (abgesehen von ökonomischen Gründen) (bei flektierbaren Wortarten) einerseits nicht konsequent alle grammatischen Formen der Lexeme aufgeführt werden: Z.B. müssten bei Verben generell mindestens alle (bildbaren) Personalformen (mit Ausnahme des Konjunktivs), die Partizip- und Adverbialpartizip-Formen sowie der Aspekt-Partner (dazu gehört u.a. auch, dass beide Aspekt-Partner jeweils ein eigenes Lemma bilden) in voller Gestalt aufgeführt sein. Andererseits werden bestimmte grammatische Formen (z.B. viele Partizipien und Adverbialpartizipien, jedoch längst nicht alle) als eigenständige Lemmata ausgegliedert. Dies würde mit der Ausweitung des Lexikoneintrags auf alle Formen des jeweiligen Lexems überflüssig werden und könnte u. U. Platz sparen. Darüber hinaus sollte nach Möglichkeit für jedes Lexem mindestens ein Beispiel angeführt werden, das die Verwendung des Lexems im Satz oder in der Wortgruppe zusätzlich illustriert. Aus den genannten Gründen ist das TSBM aus lexikographischer Sicht und unter Berücksichtigung linguistischer Aspekte unzureichend. Es genügt nicht modernen sprachwissenschaftlichen Beschreibungsprinzipien. Für praktische Zwecke ist es ebenfalls unzureichend. Zugute halten muss man dem TSBM jedoch, dass es die Prädikativa überhaupt in so großer Zahl berücksichtigt, wenn auch nicht in korrekter Art und Weise. Für die belarussische Lexikographie ist das nicht selbstverständlich. In den meisten anderen Wörterbüchern werden die Prädikativa entweder gar nicht oder nur in sehr geringem Umfang berücksichtigt, so u.a. auch im einbändigen einsprachig-erklärenden Tlumačal’ny sloŭnik belaruskaj litaraturnaj movy (TSBLM) (dt. ‚Erklärendes Wörterbuch der belarussischen Standardsprache‘) aus dem Jahr 2005 (Sudnik & Kryŭko (Hrsg.) 1996, 2 1999, 32002, 42005). Die geäußerte Kritik soll jedoch nicht darüber hinweg täuschen, dass das TSBM auch aus heutiger Sicht das mit seiner immensen Datensammlung bisher umfangreichste und beststrukturierte allgemeinsprachliche einsprachige Wörterbuch der belarussischen Sprache ist, dessen enormer Wert für die belarussische Lexikographie keinesfalls geschmälert werden soll.
3 Die Lexeme trėba und patrėbna im TSBM Im Folgenden sollen die Lexikoneinträge für die Lexeme trėba und patrėbna (dt. beide ‚(man) muss; (man) braucht‘) analysiert werden. Ich werde zeigen, dass auch diese beiden Einträge aus wissenschaftlicher Sicht nicht befriedigend sind.
122 | Claudia Hurtig (Leipzig)
Deshalb wird anschließend ein Überblick über die Komplementierung dieser Lexeme gegeben, und es werden Aussagen zur Semantik und Argumentstruktur beider Lexeme getroffen.
3.1 Eintrag für trėba Das TSBM unterscheidet zunächst zwei homonyme Lemmata: ТРЭБА1 und ТРЭБА2. Das für uns relevante ist der erste Eintrag: das Prädikativum. Der zweite Eintrag betrifft das feminine Substantiv trėba, welches 1) eine ‚Opfergabe‘ und 2) einen ‚religiösen Brauch‘ bzw. eine ‚Amtshandlung eines Geistlichen‘ bezeichnet (TSBM, Band 5: 539), vgl. auch Lejn (Hrsg.) (131999: 636). Das prädikativ gebrauchte trėba, welches etymologisch mit dem homonymen Substantiv verwandt ist und wahrscheinlich auf Urslavisch *terb- zurückgeht, ist in dieser Funktion neben dem Belarussischen u.a. auch im Ukrainischen, Bulgarischen, Serbischen, Polnischen, Tschechischen, Slovakischen, Obersorbischen und Altrussischen belegt, nicht aber im modernen Russischen (vgl. u.a. den Eintrag trebovat’ (dt. ‚fordern, verlangen; bedürfen‘) in Vasmer 2 1980 / 3: 133–134 und den Eintrag treba (dt. ‚(man) muss; (man) braucht‘) in Mel’nyčuk et al. 2006 / 5: 625–626). Im Russischen entsprechen dem Lexem die Wörter nado und nužno (dt. beide ‚(man) muss; (man) braucht‘), welche im Wesentlichen dieselben Bedeutungen wie das belarussische trėba (dt. ‚(man) muss; (man) braucht‘) aufweisen und sich auch syntaktisch ähnlich verhalten, auch wenn in den Wörterbüchern unterschiedlich viele Bedeutungen explizit angegeben werden (vgl. u.a. die Einträge für nado (dt. ‚(man) muss; (man) braucht‘) und nužnyj (dt. ‚nötig, notwendig; erfoderlich; unentbehrlich‘) in Ožegov & Švedova 4 1998: 379, 423; die Einträge für nado und nužno (dt. beide ‚(man) muss; (man) braucht‘) in Kuznecov (Hrsg.) 1998: 578, 659). Der TSBM-Eintrag zu trėba stellt sich im Wesentlichen wie folgt dar (TSBM, Band 5 / 1: 539):15
|| 15 Aus Platzgründen lasse ich im Zitat einige Beispielsätze und phraseologische Wendungen aus. Die entsprechenden Stellen sind mit […] gekennzeichnet. Eine Übersetzung des Stichworteintrags kann hier aus Platzgründen nicht erfolgen und ist auch nicht erforderlich. Zur Erläuterung der im Eintrag verwendeten Abkürzungen siehe Fn. 7.
Zur Darstellung lexikalischer Eigenschaften belarussischer Prädikativa | 123
ТРЭБА1, у знач. вык. 1. з інф. Небаходна, абавязкова патрэбна. Трэба вячэраць. □ […] 2. каго-што і з інф. Патрабуецца, патрэбна. Для гэтага трэба яшчэ шмат матэрыялаў. Трэба звярнуцца за дапамогай. Не трэба было так рабіць. □ [Валодзька:] – Не, дзеду, не трэба парома. Ляжы сабе, куры. Брыль. [Урач:] – Трэба хіба насілкі ці што... Мележ. [Жанчына:] – Вам жа кватэра трэба? Даніленка. 3. безас., каму. Пажадана. Што вам тут трэба? – раптам спытала жанчына з падкрэсленай зычнасцю і патрабаваннем. Кулакоўскі. – Што табе трэба? – крычыць лёкай. Якімовіч. 4. Разм. Належаць. Колькі з мяне трэба? □ […] ◊ Гэта ж трэба – вокліч, які выражае здзіўленне. […] Дзе трэба, дзе не трэба – недарэчы, не да месца. Лепш (і) не трэба гл. лепш. […] Што трэба (у знач. вык.) – а) вельмі добры (пра каго-, што-н.) […]; б) вельмі добра, як належыць, як мае быць. […] Як трэба – у адпаведнасці з нормамі, правіламі, існуючым палажэннем.
Wie aus dem Zitat ersichtlich, erfolgt die grammatische Kennzeichnung des Lexems trėba als Prädikativum im TSBM indirekt durch den Verweis „у знач. вык.“ (dt. ‚in Prädikatsbedeutung‘), was bereits in 2.1 erläutert wurde. Hernach folgt sofort die Bedeutungsexplikation. Das TSBM nimmt vier Bedeutungen für trėba an. Die einzelnen Bedeutungen sind jeweils an unterschiedliche syntaktische Merkmale gekoppelt, welche sich wie folgt darstellen: i)
неабходна, абавязкова патрэбна – dt. ‚(es ist) nötig, notwendig; müssen‘: mit Infinitiv ii) патрабуецца, патрэбна – dt. ‚benötigen, brauchen‘: mit Akkusativ oder mit Infinitiv iii) пажадана – dt. ‚wünschen, wollen‘: unpersönlich, mit Dativ iv) належаць – dt. ‚zukommen, zustehen, zu bekommen haben‘: keine syntaktische Spezifikation.
Der Kasus einer nominalen Ergänzung – hier nur Akkusativ und Dativ – muss vom Leser aus den Kasusformen der Fragepronomen für ‚wer?‘ (belar. хтo?) und ‚was?‘ (belar. што?) abgeleitet werden.16 Die syntaktischen Angaben zum Lexem trėba sind sehr lückenhaft. Außer den im Eintrag genannten sind noch weitere syntaktische Ergänzungen möglich.17
|| 16 Die Kasuskennzeichnung erfolgt in der traditionellen (ost)slavischen Grammatik und Lexikographie i.d.R. über die Anführung der Interrogativpronomina ,wer?‘ und ,was?‘ im erforderlichen Kasus. Dabei steht die syntaktische Angabe каго-што (dt. ‚wen-was‘) für eine Ergänzung im Akkusativ und die Angabe каму (dt. ‚wem‘) für eine Ergänzung im Dativ, vgl. die Bedeutungen 2 und 3 im TSBM-Eintrag. 17 Neben dem Eintrag für trėba findet sich im TSBM auch ein Eintrag für die Kurzform trė (ТРЭ), welche gleichbedeutend zu trėba ist (vgl. TSBM, Band 5 / 1: 539). Bei diesem Lemma beschränkt
124 | Claudia Hurtig (Leipzig)
3.2 Zur Komplementierung von trėbna Das Lexem trėba weist folgende Komplemente auf:
1 Infinitiv (18)
pajsci (dachaty). Trėba ADV PRAED INF ‚Wir (ich / …) müssen (muss / …) nach Hause gehen.‘
Ein absoluter Infinitiv ist bei trėba nicht möglich. Trėba braucht immer eine personale Ergänzung, welche durch eine Determinansphrase im Dativ ausgedrückt wird. Fehlt das Dativ-Komplement in der Oberflächenstruktur, kann es i.d.R. aus dem Kontext erschlossen werden. Die Personenangabe wird immer mitverstanden: (19)
Nam (mne / tabe / …) trėba pajsci (dachaty). DPDat. PRAED INF ADV ‚Wir (ich / du / …) müssen (muss / musst / …) nach Hause gehen.‘
2 Dativ-Determinansphrase (DPDat.) (20)
Mne trėba (padumac’). DPDat. PRAED INF ‚Ich muss nachdenken.‘
Die Dativ-Determinansphrase kann bei trėba ebenfalls nicht absolut auftreten. Das Lexem braucht mindestens eine weitere Ergänzung: Infinitiv oder Nominativ-Determinansphrase bzw. Genitiv-Determinansphrase oder konjunktional eingeleiteter Nebensatz oder Katapher + Nebensatz. Eine Ausnahme bilden elliptische Sätze vom Typ (22 B).
|| sich die Bedeutungsexplikation auf den Verweis auf das Lemma ТРЭБА (Vermerk: Тое, што і т р э б а (dt. ‚dasselbe wie trėba‘)) und das Aufführen zweier Zitate aus der Belletristik. Explizite syntaktische Hinweise enthält der Eintrag nicht. In meinen weiteren Ausführungen bleibt das Lexem trė deshalb unberücksichtigt.
Zur Darstellung lexikalischer Eigenschaften belarussischer Prädikativa | 125
(21)
(22)
*Mne trėba. DPDat. PRAED ‘Ich muss / brauche.‘ A: Tabe sloŭnik trėba? DPDat. DPNom./mask. PRAED ‚Brauchst Du das Wörterbuch?‘ B: Trėba (mne). PRAED DPDat. ‚Brauche ich.‘
3 Nominativ-Determinansphrase (DPNom.) (23)
Dlja hėtaha jašče šmat trėba PPGen. (P: dlja) PRAED noch DP matėryjalaŭ. DPGen. ‚Dafür ist noch viel Material nötig.‘
(TSBM, Band 5 / 1: 539)
Im TSBM werden für trėba unter der zweiten Bedeutung die Fragepronomen kaho / što (dt. ‚wen / was‘) angegeben18, was als Akkusativforderung zu deuten ist.19 Dies entspricht jedoch nicht den Sprachdaten, welche für eine Nominativ-Ergänzung sprechen. Aus dem in (23) zitierten Beispiel des TSBM ist zunächst nicht ersichtlich, um welchen konkreten Kasus es sich bei der Form šmat (dt. ‚viel‘) in der Wendung šmat matėryjalaŭ (dt. ‚viel Material‘) handelt: Akkusativ oder Nominativ? Die Form šmat (dt. ‚viel‘), welche ihrerseits den Genitiv verlangt, ist unveränderlich und kann somit keinem eindeutigen Kasus zugewiesen werden. Als weiteres Beispiel finden wir im TSBM außer dem in (23) zitierten noch das Beispiel
|| 18 Dieselbe Notation findet sich u.a. auch im einbändigen einsprachigen Wörterbuch (TSBLM: 666). 19 Aufgrund von Kasussynkretismus im Akkusativ und Genitiv bei belebten maskulinen Nomina sowie belebten Nomina im Plural bzw. im Akkusativ und Nominativ bei unbelebten maskulinen und neutrischen Nomina sowie allen unbelebten Nomina im Plural kann die Kombination der Pronominalformen kaho-što (dt. ,wen-was‘) nur akkusativisch interpretiert werden. Möglich wäre jedoch auch die Deutung des Pronomens kaho (dt. ‚wen‘) als Akkusativ und des Pronomens što (dt. ‚was‘) als Nominativ, was bedeuten würde, dass nur belebte Nomina die Form des Akkusativs annähmen, während unbelebte Nomina im Nominativ stünden. Es ist allerdings recht unwahrscheinlich, dass für belebte und unbelebte Nomina unterschiedliche Kasus verwendet werden.
126 | Claudia Hurtig (Leipzig)
(24), welches jedoch die Frage nach der Form der Ergänzung ebenso offen lässt, da es sich bei der entsprechenden Determinansphrase um ein unbelebtes Substantiv im Plural handelt. Der Akkusativ der unbelebten Substantive fällt im Ostslavischen bekanntlich mit dem Nominativ zusammen. Die nominale Ergänzung in (24) kann deshalb als Nominativ interpretiert werden. (24)
Trėba chiba (TSBM, Band 5 / 1: 539) nasilki, PRAED vielleicht DPNom./pl. ci što abhängiger Nebensatz ‚Da brauchen wir vielleicht eine Trage, oder was …‘
Daneben findet sich im TSBM das in (25) zitierte Beispiel, in welchem – entgegen der syntaktischen Anmerkung – eindeutig ein Nominativ eines femininen Substantivs vorliegt: (25)
kvatėra trėba? Vam ža DPDat. doch DPNom./fem. PRAED ‚Sie brauchen doch sicher eine Wohnung?‘
(TSBM, Band 5 / 1: 539)
Dieses und viele weitere Beispiele (vgl. (26)–(27)) zeigen, dass bei trėba entgegen den Angaben im TSBM nicht der Akkusativ gebraucht wird, sondern der Nominativ. Sätze vom Typ (28) sind im Belarussischen ungrammatisch. (26)
(27)
(28)
mne jak letašni sneh! (Internet) Trėba jana PRAED DPNom./fem. DPDat. wie DPNom./mask. ‚Darauf kann ich gut und gerne verzichten!‘ (Wörtlich: ‚Ich brauche sie wie den Schnee des vergangenen Jahres!‘) Mne trėba vol’naja dėmakratyčnaja Belarus’, a ne mersedėsy. DPDat. PRAED DPNom./fem. und NEG DPNom./pl. ‚Ich brauche ein freies demokratisches Belarus, und keine Mercedes.‘ (Ljavon Vol’ski) *Mne trėba hėtuju knihu. DPDat. PRAED DPAkk./fem. ‚Ich brauche dieses Buch.‘
Evidenz für den Gebrauch des Nominativs in Sätzen wie (23) und (24) liefern auch Beispiele mit Präterital-Formen. Die Subjektkongruenz der Kopula verweist hier
Zur Darstellung lexikalischer Eigenschaften belarussischer Prädikativa | 127
darauf, dass die Determinansphrase im Nominativ steht, vgl. die Beispiele (29)– (31). (29)
(30)
(31)
nechta čužy […] […] jamu trėba byŭ DPDat. PRAED COPmask. DPNom./mask. ‚[…] er brauchte einen Fremden, […]‘ Jamu trėba byli ruki, […] DPDat. PRAED COPpl. DPNom./pl., ‚Er brauchte Helfer (wörtlich: Hände), […] ‘ pomač, […] […] druhim ljudzjam trėba byla PRAED COPfem. DPNom./fem., DPDat. ‘[…] andere Menschen brauchten meine Hilfe, […]‘
(Internet)
(Internet)
(Internet)
4 Genitiv-Determinansphrase (DPGen.) Eine Genitiv-Determinansphrase tritt in zwei Fällen an die Stelle der NominativDeterminansphrase: a) bei Negation, vgl. die Beispiele (32)–(33)20, und b) wenn es sich um partitive Bedeutung handelt (= Genitivus partitivus), vgl. (34): (32)
(33)
(34)
Mne tvajho spačuvannja ne trėba. DPDat. DPGen./neutr. NEG PRAED ‚Ich brauche dein Mitgefühl nicht.‘ Nja trėba asablivych vysilkaŭ, kab […] (Internet) NEG PRAED DPGen./pl., dass-Satz ‚Es bedarf generell keiner besonderen Anstrengungen, um […]‘ Dzicjaci trėba malaka. DPDat. PRAED DPGen./neutr. ‚Das Kind braucht Milch.‘
|| 20 Bei Negation kann aber auch der Nominativ stehen. Dies gilt sowohl für trėba als auch für patrėbna: (i)
(Internet) škadoba da Tvaich pakutaŭ. Tabe ne patrėbna našae spačuvanne i und DPNom./fem. PPGen. [P: da] DPDat. NEG PRAED DPNom./neutr. ‚Du brauchst unser Mitgefühl und Bedauern Deiner Krankheit (wörtlich: Deiner Leiden) nicht.‘
128 | Claudia Hurtig (Leipzig)
5 Konjunktional eingeleiteter Nebensatz mit kab (dt. ‚dass‘) (35)
Trėba kab i ŭ Belarusi takija festyvali byli. PRAED dass-Satz ‚Es müsste auch in Belarus solche Festivals geben.‘
(Internet)
3.3 Eintrag für patrėbna Der Eintrag für das Lexem patrėbna (dt. ‚(man) muss; (man) braucht‘), welches im Belarussischen synonym zu trėba verwendet wird, fällt im TSBM (wie im Übrigen in allen anderen Wörterbüchern auch) extrem spärlich aus. Er beschränkt sich auf wenige Zeilen:
ПАТРЭБНА,
безас. у знач. вык. Неабходна, трэба. – Каб толькі так надвор’е пастаяла які тыдзень – то ўжо і касіць патрэбна. Гартны. Патрэбна дараваць Таму, хто памыліўся, А ворага – бязлітасна караць. Корбан. (TSBM, Band 4: 109)21
Die grammatische Kennzeichnung erfolgt auch hier indirekt über den Verweis ‚unpersönliche Form in Prädikatsbedeutung‘. Anschließend wird eine einzige Bedeutung expliziert. Die Explikation besteht in diesem Falle aus der bloßen Nennung zweier Synonyme – neabchodna (dt. ‚(es ist) notwendig‘) und trėba (dt. ‚(man) muss; (man) braucht‘) –, gefolgt von zwei Belegbeispielen. Hinweise auf syntaktische Ergänzungen fehlen ganz. Einzig der Infinitiv wird implizit durch die zitierten Beispielsätze berücksichtigt.22
|| 21 Da für diese Untersuchung allein die grammatischen und semantischen Merkmale des Stichworts patrėbna von Interesse sind, wird hier auf eine Übersetzung des Zitats verzichtet. 22 Neben patrėbna findet sich im TSBM auch ein Eintrag zur adjektivischen Form patrėbny bzw. patrėben (dt. beide ‚notwendig; wird gebraucht‘), deren Bedeutungsexplikation jedoch ebenso knapp ausfällt wie die im Lemma patrėbna. Auch hier finden sich keinerlei explizite Hinweise auf die syntaktische Fügung des Lexems. Minimale implizite Informationen zur syntaktischen Verwendung (als Attribut und als Prädikat) erfährt der Leser aus den angefügten Belegbeispielen aus der schöngeistigen Literatur. Jedoch sind für diese Erschließungsoperation grundlegende Grammatikkenntnisse des Belarussischen nötig, welche bei einem Nichtmuttersprachler nicht vorausgesetzt werden können. Diese Feststellung trifft auch auf die meisten, wenn nicht sogar auf alle Einträge für Prädikativa im TSBM zu, welche zwar zum Teil mehr oder minder syntaktische Abgaben beinhalten, aber, wie bereits mehrfach gezeigt, dennoch nicht zufrieden stellen können.
Zur Darstellung lexikalischer Eigenschaften belarussischer Prädikativa | 129
3.4 Zur Komplementierung von patrėbna Das Lexem patrėbna tritt mit folgenden Komplementen auf:
1 Infinitiv (36)
(37)
Patrėbna kupic’ (chleb). PRAED INF DPAkk./mask ‚Wir (ich / du / …) müssen (muss / musst / …) Brot kaufen.‘ [C]i patrėbna ŭ Belarusi budavac’ Fragepartikel PRAED PPLok. [P: u] INF (atamnuju ėlektrastancyju)? ‚Sollte in Belarus ein Kernkraftwerk gebaut werden?‘
(Internet)
Auch patrėbna braucht eine personale Ergänzung. Diese kann in der Oberflächenstruktur fehlen (vgl. u.a. Beispiel (37)), wird aber immer mitverstanden und muss aus dem Kontext erschlossen werden. Die personale Ergänzung nimmt die Form des Dativs an. Im Satz (37) könnte dieses Komplement z.B. folgendermaßen lauten: nam [patrėbna budavac’ …] (dt. ‚wir sollten bauen‘), uradu [patrėbna budavac’ …] (dt. ‚die Regierung sollte bauen‘) u. ä. Im Deutschen kann ‚fehlendes‘, d. h. non-overtes Dativ-Komplement durch eine Übersetzung als man-Konstruktion sichtbar gemacht werden (‚Sollte man in Belarus ein Kernkraftwerk bauen?‘).
2 Dativ-Determinansphrase (DPDat.) (38)
patrėbna (čytac’). Tabe PRAED INF DPDat ‚Du musst lesen.‘
Es gilt dasselbe wie bei trėba mit Dativ-Komplement. Die Dativ-Determinansphrase kann nicht absolut gebraucht werden: (39)
*Nam patrėbna. DPDat. PRAED ‚Wir müssen / brauchen.‘
130 | Claudia Hurtig (Leipzig)
3 Nominativ-Determinansphrase (DPNom.) Bezüglich einer vermeintlichen Akkusativ-Ergänzung gilt für patrėbna dasselbe wie für trėba (vgl. die Diskussion in 3.2). Die sprachlichen Daten belegen den nominativischen Status der betreffenden Determinansphrase: (40)
(41)
(42)
(43)
(44)
(45)
*Joj patrėbna knihu. DPDat. PRAED DPAkk./fem ‚Sie braucht das Buch.‘ *Kraine patrėbna herojaŭ. DPDat PRAED DPAkk./mask./pl ‚Helden braucht das Land.‘ Šmat moladzi ličac’ (Internet) što naša mova ne patrėbna, ale ja z imi ne zhodzen! ... dass DPNom./fem NEG PRAED […] ‚Viele junge Menschen sind der Meinung, dass unsere Sprache nicht gebraucht wird, aber ich bin mit ihnen nicht einverstanden! …‘ Mne ne patrėbna šoŭ, mne patrėbna „surovaja iscina“. DPDat NEG PRAED DPNom./neutr., DPDat PRAED DPNom./fem. ‚Ich brauche keine Show, ich brauche die harte Wahrheit.‘ (Internet) […] usё-tki, jany mne patrėbna … (Internet) […] dennoch DPNom./pl. DPDat. PRAED ‚[…], dennoch brauche ich sie …‘ patrėbna, pamrė, a nekatoryja buduc' žyc' večna. Chto V3.Ps.Sg./pf., […] DPNom./mask. PRAED, ‚Wer gebraucht wird, stirbt, einige aber werden ewig leben.‘ (L. Vol’ski)
4 Genitiv-Determinansphrase (DPGen.) (nach Negation) (46)
Mne ne patrėbna hrošaj. DPDat. NEG PRAED DPGen./pl. ‚Ich brauche kein Geld.‘
Zur Darstellung lexikalischer Eigenschaften belarussischer Prädikativa | 131
5 Konjunktional eingeleiteter Nebensatz mit kab (dt. ‚dass‘) (47)
Mne patrėbna, kab ljudzi byli pobač, […] (Internet) DPDat. PRAED, dass-Satz, ‚Ich brauche Menschen in meiner Nähe (wörtlich: dass die Menschen in der Nähe sind), […] .‘
3.5 Lexikalische Bedeutung und Argumentstruktur Nachdem die Möglichkeiten der Komplementierung der Lexeme trėba und patrėbna dargestellt wurden, soll nun für beide Lexeme jeweils das Verhältnis zwischen syntaktischer Realisierung und Bedeutung untersucht werden. Abgesehen davon, dass die syntaktischen Angaben in den untersuchten Lexikoneinträgen des TSBM sehr unvollständig sind, wird dort auch nicht hinreichend klar, wie sich diese konkret auf die Semantik der Lexeme auswirken bzw. wie sich die Semantik auf die Komplementierung auswirkt. Die Varianz der syntaktischen Ergänzungen ist ein generelles Problem solcher Lexikoneinträge. Ich glaube allerdings nicht, dass dieses Problem dadurch gelöst werden kann, dass man jeder syntaktischen Ergänzung eine neue Bedeutung zuordnet, wie dies im TSBM u.a. für trėba versucht wird. Es stellt sich die Frage, wie viele Bedeutungen überhaupt sinnvoll sind. Schaut man sich die oben angeführten Belegbeispiele an, dann fällt zunächst auf, dass die verschiedenen Verwendungsweisen von trėba und patrėbna keine einheitliche Bedeutung aufweisen. Aufgrund der analysierten Daten gehe ich davon aus, dass wir es sowohl bei trėba, als auch bei patrėbna jeweils mit zwei Lexemen (bzw. zwei Lesarten jeweils eines Lexems) zu tun haben, deren Semantik und Argumentstruktur sich wie folgt darstellen23:
|| 23 Aufgrund der prädikativen Funktion der Prädikativa kann davon ausgegangen werden, dass diese ähnlich wie Verben Argumente an sich binden und folglich über eine Argumentstruktur verfügen.
132 | Claudia Hurtig (Leipzig)
Tab. 1: Semantik und Argumentstruktur von trėba und patrėbna
trėba 1:
mit der Grundbedeutung: x müssen y Das Prädikativum trėba 1 stellt eine Relation zwischen einer Variable x und einer Variable y dar (vgl. Bierwisch 1983: 74). X bezeichnet dabei im weitesten Sinne die semantische Repräsentation eines ‚Subjekts‘ und y die semantische Repräsentation eines ‚Objekts‘.24 X wäre in unserem Falle die Person, die etwas tun muss bzw. sollte, und y wäre das, worauf sich der Musszustand richtet, das, was ‚gemusst wird‘, was ‚notwendig ist‘. X nimmt die Form des Dativs an, und y ist propositional. Die Grundbedeutung könnte also so paraphrasiert werden: x muss y (tun) / x sieht eine Notwendigkeit in y.
trėba 2:
mit der Grundbedeutung: x brauchen / nötig haben y X hat wiederum Dativform, und y hat i.d.R. Nominativform. Die Grundbedeutung könnte so paraphrasiert werden: x braucht y / x hat y nötig.
patrėbna 1:
mit der Grundbedeutung: x müssen y x muss y (tun) / x sieht eine Notwendigkeit in y
patrėbna 2:
mit der Grundbedeutung: x brauchen / nötig haben y x braucht y / x hat y nötig
Die Annahme von zwei (bzw. vier) Lexemen (Lesarten jeweils eines Lexems) hätte den Vorteil, dass sich das Problem der Komplementierungsvarianz, wie es in den TSBM-Einträgen (v.a. bei trėba) zu beobachten ist, aufheben würde. Unter trėba 1 (= ‚müssen, notwendig sein‘) würde nur die syntaktische Realisierung mit Infinitiv und mit dass-Satz fallen. Im entsprechenden Eintrag im TSBM (TSBM, Band 5 / 1: 539; siehe 3.1) trifft das auf die erste angenommene Bedeutung und teilweise auf die zweite Bedeutung zu. Auch die dort aufgeführte
|| 24 Die Begriffe Subjekt und Objekt sind in der traditionellen Grammatik festgelegt auf vom Verb geforderte (abhängige) Nomina im Nominativ bzw. Akkusativ. Abgesehen davon, dass die bei Prädikativa auftretenden Komplemente diesen Termini insofern nicht entsprechen, dass sie eine andere syntaktische Form aufweisen (vgl. v. a. die bei Prädikativa auftretenden Nomina im Dativ, die traditionell häufig als ‚semantisches Subjekt‘ bezeichnet werden), verwende ich hier der Einfachheit halber ebenfalls die Variablen x und y, zumal bislang auch noch nicht geklärt werden konnte, welchen konkreten syntaktischen Status die jeweiligen Argumente der Prädikativa haben. Dies ist eine spezielle Diskussion, welche für diese Arbeit nicht relevant ist. Ausführlicher dazu siehe z.B. Hrabě (1964); Švedova (1964); Voinova (1965); Guiraud-Weber (1984); Zybatow (1994); Hurtyh (2007), Hurtig (in Vorbereitung) u.a.
Zur Darstellung lexikalischer Eigenschaften belarussischer Prädikativa | 133
vierte Bedeutung (= naležyc’ – dt. ‚(es) muss / soll sein‘) kann zu trėba 1 gezählt werden. Alle anderen syntaktischen Realisierungen (mit Nominativ und Genitiv) fallen dann unter trėba 2 (= ‚brauchen, nötig haben‘). Im TSBM entspricht dies z.T. der zweiten Bedeutung. Die dritte im TSBM für trėba angenommene Bedeutung (= pažadac’ – dt. ‚(etw.) wollen / wünschen‘) würde ich unter trėba 2 (= ‚brauchen, nötig haben‘) verorten. Einzig die Komplementierung mit dass-Satz kann vermutlich bei beiden Lexemen (trėba 1 und 2) auftreten. Dies müsste durch gesonderte Analysen überprüft werden. Eine Dativ-Ergänzung ist für beide Lesarten notwendig. Diese wird jedoch häufig nicht overt realisiert. In Abhängigkeit davon, ob eine Dativ-Determinansphrase in Kombination mit einer Nominativ-Determinansphrase overt oder nonovert realisiert wird, kann allerdings die Bedeutung divergieren. Bei non-overter Realisierung des Dativ-Komplements und gleichzeitig realisierter Nominativ-Determinansphrase gilt vermutlich immer die Bedeutung von trėba 2 (= ‚brauchen, nötig haben‘), während bei overt realisierter Dativ-Determinansphrase und fehlender Nominativ-Determinansphrase offenbar die Bedeutung von trėba 1 (= ‚müssen, notwendig sein‘) Gültigkeit hat. Dasselbe gilt für die zwei Lexeme (Lesarten) von patrėbna. An die overte Realisierung einer Dativ-Determinansphrase ist darüber hinaus auch die Realisierung eines Infinitivs oder einer anderen Ergänzung gekoppelt, vgl. dazu die Beispiele (19)–(22). Auch hier gilt selbiges für patrėbna, vgl. die Beispiele (38)–(39). Evidenz dafür könnte es auch in der unterschiedlichen Übersetzung der Aussagen in (45) und (48) geben: In Abhängigkeit davon, ob das Personalpronomen chto (dt. ‚wer‘) im Nominativ oder im Dativ steht, variiert zum einen die Bedeutung des Prädikativums patrėbna als ‚brauchen‘ vs. ‚müssen‘, zum anderen wird das Infinitiv-Komplement blockiert vs. besetzt. So erfordert die Dativ-Determinansphrase in (48) die gleichzeitige Anwesenheit einer Infinitiv-Ergänzung, welche zwar in der Oberflächenstruktur fehlen kann, aber dennoch immer mitverstanden wird. Die Nominativ-Determinansphrase in (45) lässt hingegen keinen Infinitiv beim Prädikativum patrėbna zu.25 (48)
Kamu patrėbna (pamerci), pamrė, a nekatoryja buduc' žyc' večna. DPDat. PRAED, INF V3.Ps.Sg./pf., […] (Linguistische Variation des Zitats von Ljavon Vol’ski: Ja naradziŭsja tut, Songtext, Booklet „Belarusian Red Book“) ‚Wer (sterben) muss, (der) stirbt, einige aber werden ewig leben.‘
|| 25 Für die Diskussion dieses Beispiels danke ich Uwe Junghanns und Taccjana Ramza.
134 | Claudia Hurtig (Leipzig)
Für patrėbna 1 und patrėbna 2 sieht die syntaktische Verteilung insgesamt ähnlich aus wie für trėba 1 und trėba 2. Unter patrėbna 1 (= ‚müssen, notwendig sein‘) würde die Komplementierung mit Infinitiv und eventuell die Realisierung mit dass-Satz zählen. Letztere kann aber auch unter patrėbna 2 (= ‚brauchen‘) fallen. Die syntaktischen Realisierungen mit Nominativ bzw. Genitiv gehören zu patrėbna 2 (= ‚brauchen, nötig haben‘). Grundlegend kann festgestellt werden, dass trėba und patrėbna Synonyme sind, wie auch im TSBM, im Wörterbuchartikel zu patrėbna (TSBM, Band 4: 109), vermerkt wird. Offenbar unterscheiden sie sich aber teilweise in ihrer Komplementierung. Bei der Komplementierung mit Determinansphrase im Nominativ lassen trėba 2 und patrėbna 2 offenbar nicht exakt dieselben Argumente zu. Trėba 2 kann, wie es scheint, Nomen aller Genera und Numeri als Nominativ-Ergänzung zu sich nehmen (vgl. die Beispiele (29)–(31)), während patrėbna 2 regulär wohl nur Nominativ-Ergänzungen mit femininem und neutralem Nomen (vgl. die Beispiele (42)–(43)) bzw. selten Nomen im Plural (vgl. Beispiel (44)) zu sich nehmen kann.26
|| 26 Dagegen sprechen allerdings vereinzelt auftretende Beispiele von patrėbna 2 mit eindeutiger Nominativ-Determinansphrase im Maskulinum bzw. Plural, vgl. die Beispiele (44)–(45). Eine derartige Verwendung ist möglicherweise nicht-normativ, was hier jedoch nicht geklärt werden kann. Weitere Belegbeispiele für maskuline Nominativ-Determinansphrasen bei patrėbna 2 wurden bisher nicht gefunden. Vor allem Belege mit Substantiven vom Typ ??bratDPmask. mne patrėbna / ??bratyDPpl. mne patrėbna (dt. ‚ich brauche den Bruder / ich brauche die Brüder‘) sind sehr schwer zu finden und müssen in Zukunft noch verifiziert oder falsifiziert werden. Indessen bieten die Sprachdaten folgendes Bild: Soll als Ergänzung für den Ausdruck von ,brauchen / nötig haben‘ eine maskuline Nominativ-Determinansphrase oder eine Determinansphrase im Plural verwendet werden, steht statt des Prädikativums patrėbna i.d.R. die adjektivische Kurz- oder Langform patrėben / patrėbny (= Kurz- / Langformmask.) (dt. ‚nötig‘) bzw. patrėbny / patrėbnyja (= Kurz- / Langformpl.) (dt. ‚nötig‘), vgl. (i)–(ii): (i)
(ii)
Ci patrėbny ŭ Mensku pomnik Dzjaržynskamu? Fragepartikel ADJmask.LF PPLok. [P: u] DPNom. DPDat. ‚Brauchen wir in Minsk ein Dzjaržynski-Denkmal?‘ Specyjalisty patrėbny zaŭsjody. DPPl. ADJpl.KF ADV ‚Spezialisten werden immer gebraucht.‘
(Internet)
(Internet)
Es stellt sich nun die Frage, ob patrėbna 2 (dt. ,brauchen, nötig haben‘) sich nicht generell im Genus an das Nomen anpasst und demzufolge auch in den Beispielen mit femininen bzw. neutralen Nomen (vgl. u.a. (42)–(43)) vom Typ Adjektiv ist und nicht mehr der lexikalischen Kategorie der Prädikativa angehört. Dann wäre der Lexikoneintrag patrėbna 2 hinfällig, man brauchte dann nur die Lexeme patrėbna 1 (dt. ‚müssen, notwendig sein‘) und patrėbny, patrėbnaja,
Zur Darstellung lexikalischer Eigenschaften belarussischer Prädikativa | 135
Die sprachlichen Fakten zeigen, dass die lexikalische Bedeutung eines Lexems nur im Zusammenhang mit dessen Argumentstruktur analysiert werden kann. Dadurch lässt sich die Anzahl der Bedeutungen pro Lexem zwar einschränken, gleichzeitig erhöht sich aber die Anzahl der Lemmata, was im Hinblick auf Platzökonomie auch nicht die optimale Lösung für ein Wörterbuch ist. Denkbar wäre, dass die von mir ausgegliederten Lexeme (bzw. Lesarten) trėba 1 (dt. ‚müssen‘) und trėba 2 (dt. ‚brauchen‘) (sowie analog dazu die Lexeme patrėbna 1 und patrėbna 2) beide unter einem Stichwort verortet, jedoch streng in ihrer Bedeutung getrennt werden. Unabhängig davon sollten für jede der beiden Lesarten alle zulässigen Varianten der Komplementierung aufgeführt werden.
4 Zusammenfassung und Ausblick Ziel meiner Untersuchung war es, die Darstellung lexikalischer Eigenschaften einer ausgewählten Klasse von Lexemen in einsprachig-erklärenden Wörterbüchern des Belarussischen zu untersuchen. Da das TSBM als erstes und bisher einziges belarussisches Großwörterbuch sozusagen das Fundament der modernen belarussischen einsprachigen Lexikographie bildet, ist es ein logischer Schluss, in einer kleinen Untersuchung wie der vorliegenden dieses Wörterbuch heranzuziehen. Im Mittelpunkt des Beitrags stand die Analyse von Lexikoneinträgen für Prädikativa. In Abschnitt 2 wurde gezeigt, wie ein solcher Eintrag im TSBM aufgebaut ist bzw. laut Wörterbuch-Vorwort aufgebaut sein sollte. Anschließend wurden in Abschnitt 3 zwei ausgewählte Lexikoneinträge speziell auf Existenz und Explizierung wichtiger lexikalischer Informationen überprüft. Insgesamt konnte gezeigt werden, dass die Prädikativa allgemein nicht befriedigend beschrieben werden und die Einträge zu einem großen Teil essenzieller Bestandteile und lexikalischer Angaben entbehren. Als wesentliche Probleme der traditionellen Lexikographie konnten u.a. folgende festgestellt werden: 1) Die mangelnden lexikalischen Informationen zu den Prädikativa sind in erster Linie auf die mehr oder minder traditionelle Klassifizierung des Wortschatzes in zehn Wortarten (bzw. zwölf – mit Partizip und Adverbialpartizip) zurückzuführen, in der Prädikativa nicht berücksichtigt werden. 2) Die Behandlung von Homonymen ist, nicht zuletzt auch aufgrund der genannten Wortschatzeinteilung, unzulänglich. Prädikativa, die homonym zu
|| patrėbnae, patrėbnyja (= Adjektiv in der Langform, dt. ‚nötig‘) vs. patrėben, patrėbna, patrėbna, patrėbny (= Adjektiv in der Kurzform, dt. ‚nötig‘) zu unterscheiden.
136 | Claudia Hurtig (Leipzig)
Lexemen anderer Wortarten sind, werden nicht als eigenständige Lemmata aufgeführt, sondern als eine Art „syntaktischer Derivate“ (Švedova (Hrsg.) 1970: 309) innerhalb der Bedeutungen der vermeintlich übergeordneten Lexeme unterschieden. 3) Die syntaktische Struktur der Prädikativa bleibt zu einem erheblichen Teil sehr vage. Die syntaktischen Angaben sind hochgradig unvollständig und teilweise fehlerhaft. Der syntaktische Status von Ergänzungen wird nicht geklärt: Es ist nicht ersichtlich, welche Ergänzungen notwendig und welche frei sind. Als Komplemente sind alle notwendigen Ergänzungen zu erfassen. 4) Es wird nicht hinreichend berücksichtigt, dass eine Lexembedeutung verschiedene Komplemente aufweisen kann. Häufig wird jede syntaktische Ergänzung mit einer neuen lexikalischen Bedeutung verbunden. Dies ist jedoch nicht immer notwendig. Die lexikalische Bedeutung sollte immer im Zusammenhang mit der Argumentstruktur des Lexems erfasst werden. Natürlich stellt dieser Beitrag nur eine minimale Auswahl an potenziellem lexikographischem Material zur Verfügung. Jedoch konnten mit dem Aufzeigen einiger wesentlicher Probleme der traditionellen belarussischen Lexikographie anhand ausgewählter Einträge für Prädikativa bereits erste Vorschläge zur Optimierung eines solchen Lexikoneintrags für Prädikativa gemacht werden, die für die Erstellung zukünftiger Wörterbücher von Relevanz sein könnten. Um wissenschaftlich repräsentative Ergebnisse in Bezug auf die Handhabung der Prädikativa in belarussischen einsprachigen Wörterbüchern zu erzielen, müssten in weiterführenden Arbeiten die Lexikoneinträge für alle Prädikativa analysiert werden und dabei auch andere Wörterbücher (wie z.B. das TSBLM) in die Untersuchungen einbezogen werden. Darüber hinaus sind auch die Lexikoneinträge für alle anderen Wortarten auf ähnliche Weise zu analysieren. Es bleibt zu wünschen, dass die Prädikativa in Zukunft von der (ost)slavischen Lexikographie und Grammatiktheorie, v. a. auch in der zu erwartenden überarbeiteten Neuauflage des TSBM, stärker berücksichtigt und wissenschaftlich konsequenter behandelt werden.
Literatur TSBM. Tlumačal’ny sloŭnik belaruskaj movy. Siehe Atrachovič, K. K. (Krapiva, K.) (Hrsg.) (1977– 1984). TSBLM. Tlumačal’ny sloŭnik belaruskaj litaraturnaj movy. Siehe Sudnik, M. R. & M. N. Kryŭko (Hrsg.) (1996, 21999, 32002, 42005). Atrachovič, K. K. (Krapiva, K.) (Hrsg.) (1977–1984): Tlumačal’ny sloŭnik belaruskaj movy: U 5 tamach. Minsk: „Belaruskaja Saveckaja Ėncyklaperdyja“ imja Petrusja Broŭki.
Zur Darstellung lexikalischer Eigenschaften belarussischer Prädikativa | 137
Atrachovič, K. K. &. M. H. Bulachaŭ (Hrsg.) (1962): Hramatyka belaruskaj movy. Tom 1. Marfalohija. Minsk: Vydavectva Akadėmii navuk BSSR. Bierwisch, M. (1983): Semantische und konzeptuelle Repräsentation lexikalischer Einheiten. In: Růžička, R. & W. Motsch (Hrsg.): Untersuchungen zur Semantik (Studia Grammatica XXII). Berlin: Akademieverlag, 61–101. Bierwisch, M. (1996): Lexikon und Universalgrammatik. In: Weber, N. (Hrsg.): Semantik, Lexikographie und Computeranwendungen (Sprache und Information; 33). Tübingen: Niemeyer, 129–165. Biryla, M. V. (Hrsg.) (1987): Sloŭnik belaruskaj movy. Arfahrafija. Arfaėpija. Akcėntuacyja. Slovazmjanenne. Minsk: „Belaruskaja saveckaja ėncyklapedyja“ imja Petrusja Broŭki. Biryla, M. V. & P. P. Šuba (Hrsg.) (1985): Belaruskaja hramatyka. Častka 1. Fanalohija, arfaėpija, marfalohija, slovaŭtvarėnne, nacisk. Minsk: Navuka i tėchnika. Biryla, M. V. & P. P. & Šuba (Hrsg.) (1986): Belaruskaja hramatyka. Častka 2. Sintaksis. Minsk: Navuka i tėchnika. Biryla, M. V. & P. P. Šuba (Hrsg.) (1985): Belaruskaja hramatyka. Častka 1. Fanalohija, arfaėpija, marfalohija, slovaŭtvarėnne, nacisk. Minsk: Navuka i tėchnika. Blinava, Ė. D., N. V. Haŭroš, M. Kavalëva et al. (21994): Belaruskaja mova. 2-e vydanne, dapracavanae i dapoŭnenae. Minsk: Vysėjšaja škola. Burak, L. I. (1974): Sučasnaja belaruskaja mova. Minsk: Vyšėjšaja škola. Butyrina, L. N., A. V. Lykov & L. I. Šuljak (1998): Slova kategorii sostojanija v sovremennom russkom jazyke. Taganrog: Izdatel’stvo Taganrogskogo gosudarstvennogo pedagogičeskogo instituta. Chomsky, N. (1995): A Minimalist Programm for Linguistic Theory. In: Chomsky, N.: The minimalist programm (Current studies in linguistics; 28). Cambridge, Mass. / London, England: MIT Press, 167–217. Dolbik, E. E., V. L. Leonovič & V. A. Sanikovič (42008): Russkij jazyk: Tablicy, schemy, upražneniaj. Dlja postupajuščich v vuzy. 4-e izdanie, ispravlennoe. Minsk: Vyšėjšaja škola. Fehrmann, D. & U. Junghanns (in Vorbereitung): Infinitive-like Perception Predicates in Slavic. https://www.uni-goettingen.de/de/gemeinsame-aufs%C3%A4tze/153182.html (09.01.2016) Fehrmann, D., Junghanns, U. & D. Lenertová (2007): Was zu sehen und zu hören ist: Infinitivartige Perzeptionsprädikate. Handout zur Projektpräsentation. 16. JungslavistInnen-Treffen, Dresden, 22.09.07. http://www.jungslavisten.de/treffen/16/#fehrmannjunghannslenertova (09.01.2016) Galkina-Fedoruk, E. M. (1957): O kategorii sostojanija v russkom jazyke. In: Russkij jazyk v škole 4 (1957), 6–17. Galkina-Fedoruk, E. M. (Hrsg.) (1964): Sovremennyj russkij jazyk. Čast’ II: Morfologija. Sintaksis. Moskva: Izdatel’stvo Moskovskogo universiteta. Guiraud-Weber, M. (1984): Les propositions sans nominatif en russe moderne. Paris. Hrabě, V. (1964): K otazce nepřimého subjektu v ruské skladbé. In: Bulletin ústavu ruského jazyka a literatury, Jg. 1964, H. VIII, 45–49. Hurtig, C. (in Vorbereitung): Die Prädikativa im Belarussischen (Arbeitstitel). Manuskript. Universität Leipzig, Institut für Slavistik.
138 | Claudia Hurtig (Leipzig)
Hurtyh, K. (2007): Ab adnym sintaksičnym kampanence ŭ prostych skazach z prėdykatyvami [Zu einer syntaktischen Komponente in einfachen Sätzen mit Prädikativa]. In: Hramatyčny lad belaruskaj movy. Šljachi histaryčnaha razviccja i sučasnyja tėndėncyi. Matėryjaly mižnarodnaj navukovaj kanferėncyi, Minsk, 29–30 kastryčnika 2007 h. Minsk: Prava i ėkanomika, 154–159. Hurtyh, K. (2012): Leksičnyja prykmety belaruskich prėdykatyvaŭ (na prykladze slova „trėba“). [Lexikalische Eigenschaften belarussischer Prädikativa (am Beispiel des Lexems „trėba“)]. In: Zaprudski, S. M. & Cychun, H. A. (Hrsg.): Novae slova ŭ belarusistycy. Movaznaŭstva. Matėryjaly V Mižnarodnaha kanhrėsa belarusistaŭ, Minsk, 20–21 maja 2010 hoda. (Belarusika / Albaruthenica; 32). Minsk: Četyre četverti, 204–212. Isačenko, A. V. (1954): Grammatičeskij stroj russkogo jazyka v sopostavlenii s slovackim. Morfologija. Čast’ I. Bratislava: Izdatel’stvo Slovackoj Akademii nauk. Isačenko, А. V. (1962): Die russische Sprache der Gegenwart. Teil I: Formenlehre. Halle (Saale): Max Niemeyer. Isačenko, A. V. (22003): Grammatičeskij stroj russkogo jazyka v sopostavlenii s slovackim. Morfologija. (Jazyki sljavjanskoj kul’tury: Venskij slavističeskij al’manach). (Klassiki otečestvennoj filologii; čast’ 1). Moskva, Wien. Junghanns, U. (2012): Nekatoryja zaŭvahi da ŭzaemaadnosinaŭ pamiž leksikonam i marfasintaksisam: prėdykatyvy ŭsprymannja [Betrachtungen zum Verhältnis von Lexikon und Morphosyntax: Perzeptionsprädikativa]. In: Zaprudski, S. M. & Cychun, H. A. (Hrsg.): Novae slova ŭ belarusistycy. Movaznaŭstva. Matėryjaly V Mižnarodnaha kanhrėsa belarusistaŭ, Minsk, 20-21 maja 2010 hoda. (Belarusika / Albaruthenica ; 32). Minsk: Četyre četverti, 181–193. Junghanns, U. & D. Lenertová (in Vorbereitung): On the syntax and semantics of Slavic perception predicates. https://www.uni-goettingen.de/de/gemeinsameaufs%C3%A4tze/153182.html (09.01.2016) Kuznecov, S. A. (Hrsg.) (1998): Bol’šoj tolkovyj slovar’ russkogo jazyka. Sankt-Peterburg: Norint. Lejn, K. (131999): Bol’šoj russko-nemeckij slovar’. 13-oe izdanie, stereotipnoe. Moskva: Russkij jazyk. Mel’nyčuk, O. S., І. К. Bilodid, V. T. Kolomiec’ et al. (Hrsg.) (2006): Ėtymolohičnyj slovnik ukrains’koi movy. Tom 5: R–T. Kyiv: Naukovo-vydavnyčyj viddil filolohii, chudožn’oi litėratury ta slovnykiv. Novikov, L. A., L. G. Zubkova, V. V. Ivanov et al. (2001): Sovremennyj russkij jazyk. Fonetyka. Leksikologija. Slovoobrazovanie. Morfologija. Sintaksis. 3-e izdanie. Sankt-Peterburg: Lan’. Otkupščikova, M. I. (1995): Bezlično-predikativnye slova (kategorija sostojanija) v novom akademičeskom slovare. In: Očerednye zadači russkoj akademičeskoj leksikografii. Sbornik statej. Sankt-Peterburg: Izd. ILI RAN, 140–144. Ožegov, S. I. & N. Ju. Švedova (41998): Tolkovyj slovar’ russkogo jazyka. 4-e izdanie, dopolnennoe. Moskva: Azbukovnik. Pitsch, H. (2012a): Ličnye modaľnye glagoly i modaľnye predikativy v belorusskom jazyke. [Modalverben und Modalprädikativa im Belarussischen]. In: Zaprudski, S. M. & Cychun, H. A. (Hrsg.): Novae slova ŭ movaznaŭstve: matėryjaly V Mižnarodnaha kanhrėsa belarusistaŭ, Minsk, 20–21 maja 2010 h. (Belarusika / Albaruthenica; 32). Minsk: Četyre četverti, 193– 201.
Zur Darstellung lexikalischer Eigenschaften belarussischer Prädikativa | 139
Pitsch, H. (2012b): Zur Syntax und Semantik von Modalprädikativa. In: Szucsich, Luka, Natalia Gagarina, Elena Gorishneva & Joanna Leszkowicz (Hrsg.): Linguistische Beiträge zur Slavistik. XIX. JungslavistInnen-Treffen in Berlin 16.–18. Dezember 2010. (Specimina Philologiae Slavicae; 171). München: Otto Sagner, 181–194. Scjacko, P. U., M. F. Hulicki & L. A. Antanjuk (1990): Sloŭnik linhvistyčnych tėrminaŭ. Minsk: Vyšėjšaja škola. Sudnik, M. R. & M. N. Kryŭko (Hrsg.) (1996, 21999, 32002, 42005): Tlumačal’ny sloŭnik belaruskaj litaraturnaj movy. 2-e vydanne, dapracavanae i dapoŭnenae, 3-e vydanne. Minsk: „Belaruskaja Ėncyklapedyja“ imja Petrusja Broŭki. Šanskij, N. M. (Hrsg.) (1981): Sovremennyj russkij literaturnyj jazyk. Leningrad: Prosveščenie. Ščerba, L. V. (1928): O častjach reči v russkom jazyke. In: Ščerba, L. V. (1957): Izbrannye raboty po russkomu jazyku. Moskva: Gos. uč.-ped. izd-vo Ministerstva Prosveščenija RSFSR. Šuba, P. P. (1975): Lekcyi pa belaruskaj marfalohii. Minsk: Vydavectva Belaruskaha dzjaržaŭnaha universitėta imja U. I. Lenina. Šuba, P. P. (1987): Sučasnaja belaruskaja mova: Marfanalohija. Minsk: Universitėckae. Švedova, N. Ju. (1964): Determinirujuščij ob’jekt i determinirujuščee obstojatel’stvo kak samostojatel’nyje rasprostraniteli predloženija. In: Voprosy jazykoznanija 1964 (6), 77–93. Švedova, N. Ju. (Hrsg.) (1970): Grammatika sovremennogo russkogo literaturnogo jazyka. Moskva: Nauka. Vasmer, M. (21980): Russisches Etymologisches Wörterbuch. 2., unveränderte Auflage. Band 3: Sta–Ÿ. Heidelberg: Carl Winter, Universitätsverlag. Vinogradov, V. V. (21972): Russkij jazyk. Grammatičeskoe učenie o slove. Moskva: Vyjsšaja škola. Vinogradov, V. V., E. S. Istrina & S. G. Barchudarov (Hrsg.) (1960): Grammatika russkogo jazyka. Tom 1: Fonetika i morfologija. Moskva: Izdatel’stvo Akademii nauk SSSR. Voinova, E. I. (1965): Predikativnye slova na -o, sočetajuščiesja s infinitivom, v sovremennom russkom literaturnom jazyke. Avtoreferat dissertacii na soiskanie učenoj stepeni kandidata filologičeskich nauk. Leningrad: Leningradskij ordena Lenina gosudarstvennyj universitet imeni A. A. Ždanova. Zybatow, G. (1994): Infinitive, kleine und große Pros und (in)kohärente Konstruktionen im Russischen. In: Steube, A. & G. Zybatow (Hrsg.): Zur Satzwertigkeit von Infinitiven und Small Clauses (Linguistische Arbeiten; Bd. 315). Tübingen: Max Niemeyer, 141–154.
Uwe Junghanns (Göttingen), Denisa Lenertová (Berlin) and Dorothee Fehrmann (Göttingen)
Parametric Variation of Slavic Accusative Impersonals 1 Introduction Accusative Impersonals (AIs) such as morphologically marked AIs and Adversity AIs (A-AIs) have received much attention in the literature due to their unexpected case licensing on the internal argument of a transitive verb. It is marked accusative although the external argument does not show up on the surface. The question arises how the external argument is, if at all, realized. The examples in (1)– (2) illustrate morphologically marked AIs.1 Here, the reflexive marker (refl) and no/to prevent the canonical realization of the external argument. With so-called Adversity-AIs2 (A-AIs), as in (3), the canonical external argument realization is || Note: The reported work is part of the project “Microtypological variation in argument structure and morphosyntax” of the DFG research group “Grammar and Processing of Verbal Arguments” (FOR 742). Earlier versions of the paper were presented at the 8th European Conference on Formal Description of Slavic Languages at the University of Potsdam (December 2009), the meeting of the research group “Grammar and Processing of Verbal Arguments” at the University of Leipzig (January 2012), the Slavic linguistic colloquia at the University of Göttingen (June 2012, May 2013) and the 10th European Conference on Formal Description of Slavic Languages at the University of Leipzig (December 2013). We would like to thank the audiences for valuable comments and especially Ilse Zimmermann for inspiring discussion, as well as Bożena Cetnarowska, Petr Biskup, Franc Marušič and Hana Gruet-Škrabalová for helpful comments. Thanks are also due to Hagen Pitsch and Roland Meyer for constructive discussion at various stages of the paper. All errors are our own. || 1 The following abbreviations are used in the glosses: 1/2/3 = 1st/2nd/3rd person; acc = accusative case; aux = auxiliary; dat = dative case; f = feminine; gen = genitive case; ger = gerund; inf = infinitive; instr = instrumental case; ipf = imperfective aspect; loc = locative case; m = masculine; n = neuter; neg = negation; nom = nominative case; ns = null subject; past = past tense; pf = perfective aspect; pl = plural; prep = preposition; pres = present tense; prt = particle; ptcp = participle; refl = reflexive; sg = singular. 2 We use the term “adversity impersonal” introduced by Babby (1994). The type of structure is also known as “accidental construction” (e.g., Markman 2004) or “stixijnaja konstrukcija”, where Russian stixija means ‘element(ary force)’ (e.g., Mustajoki and Kopotev 2005, adopting the term from Šeljakin 2001 and Zolotova 1988). Lavine and Freidin (2002) refer to the structure as “finite accusative unaccusative”.
DOI 10.1515/9783110517873-007
Parametric Variation of Slavic Accusative Impersonals | 141
prevented without additional morphological marking. The forms of the transitive verbs found in A-AIs look like ordinary active predicates, carrying the 3rd singular neuter (3sg.n) ending. Morphologically marked AIs – refl and no/to AIs (1)
(2)
Niszczy się starą zabudowę (Po)3, 4 destroy.3sg refl old.acc architecture.acc (*przez władze / * powodzią). by authorities.acc flood.instr ‘They are destroying the old architecture.’ (oblique agent/causer impossible) (Nym) bulo vidkryto tuberkulin. (Ukr) he.instr was.sg.n discover.TO tuberculin.acc ‘Tuberculin was discovered (by him).’ (Šyrobokov, V. P. (red.). 2011. Medyčna mikrobiolohija, virusolohija ta imunolohija. Vydannja 2-e. Vinnycja: Nova Kniha. p. 43)
Adversity AIs (3)
Lodku uneslo (vetrom). boat.acc carry-away.past.sg.n wind.instr ‘The boat was carried away (by the wind).’ (Mustajoki and Kopotev 2005: 12)
(Ru)
At least since Lavine and Freidin (2002) the data have been investigated as a challenge for Burzio’s Generalization (BG).
|| 3 Languages are abbreviated as follows: spk/dlctl spoken/dialectal BCS, stdrd standard BCS Bosnian/Croatian/Serbian; Bg Bulgarian; BRu Belarusian; Cz Czech; Po Polish; Ru Russian; Slk Slovak; Slvn Slovenian; Ukr Ukrainian; Upper Sorb(ian). 4 For data elicitation, native speakers have been consulted for all examples in each language. Special thanks are due to: Taccjana Ramza and Valiantsin Solakhau (BRu), Petr Biskup and Kristina Krchňavá (Cz), Alicja Butkiewicz, Ola Gogłoza, Małgorzata Małolepsza, Wojtek Siółkowski and Edyta Zander (Po), Genia Böhnisch, Nadja Dückmann, Rita Graf, Elena Grimmig, Nadja Herdt, Ol’ga Karpova, Shanna Koppmeier, Marianna Leonova, Olga Liebich, Natalya Maischeva, and Inga Pagel (Ru), Jana Orieščiková (Slk), Boštjan Dvořák (Slvn), Svitlana Adamenko and Iryna Parkhomenko (Ukr), and Božena Braumanowa, Marko Malink and Hync Rychtaŕ (Upper Sorb).
142 | U. Junghanns (Göttingen), D. Lenertová (Berlin) and D. Fehrmann (Göttingen)
(4)
“A verb Case-marks its object if and only if it θ-marks its subject.” (Burzio 1986: 178)
Thus many accounts of Slavic AIs question BG in its classical formulation and propose that (i) BG can be violated in the presence of defective, φ-incomplete, T, which licenses accusative assignment (Lavine and Freidin 2002, Harves 2006, Tsedryk 2004), or (ii) BG is rather a correlation between structural acc(usative) assignment and the presence of external causation in the event-structure (cf. Pylkkänen’s 2008 v-caus as acc licenser); languages with AIs project the syntactic heads introducing the causing event and the external argument separately (Markman 2004, Lavine 2010, 2013). However, (i) and (ii) do not account for the cross-Slavic variation with respect to the availability of particular AI-types and their properties. In Ru and BRu, for example, A-AIs as in (3) are quite frequent, whereas morphologically marked AIs are not available at all. Importantly, the conditions proposed by the accounts in (i) and (ii) should equally apply to morphologically marked AIs and A-AIs. For languages allowing only one type of AIs, (i) and (ii) would have to stipulate additional special restrictions. Table 1 summarizes the varying distribution and properties of morphologically marked AIs across Slavic.5 Slavic languages differ wrt whether they allow oblique agents (by-phrases)6 with refl(exive) pass(ive) – as shown in the first row of the table. A subgroup of the languages disallowing by-phrases – Po, Slvn, and spoken/dialectal BCS – show evidence (e. g., anaphor binding) for the presence of a null syntactic subject in refl(exive) impers(onals) – given in the second row. In this group, morphologically marked AIs are attested – as shown in row three and row four. They do not allow by-phrases – compare the Po example in (1) – and show evidence of null syntactic subjects (for the latter see Kupść 2000, Rivero || 5 Note that the order of languages in the table does not reflect the traditional partition into East, West and South Slavic but is arranged for expository purposes such as to group languages that share relevant properties. 6 There is variation wrt how oblique agents, if possible at all, are realized in Slavic – as instrumental NP (e. g., in East Slavic or Cz) or grammaticalized PP (headed by, e. g., przez ‘by’ in Po or od/ot ‘from’ in South Slavic). Importantly, they are to be clearly distinguished from elements such as free adverbials that may be expressed by the same or similar morphological/lexical means (e. g., instrumental case, or various PPs such as ze/zo strone + NP in Slvn).
Parametric Variation of Slavic Accusative Impersonals | 143
2000, Rivero and Milojević Sheppard 2003, Lavine 2005, Kibort 2006, Szucsich 2008, 2009). On the other hand, Ukr has morphologically marked AIs too. However, they allow by-phrases as does the reflexive passive and reflexive impersonal – compare the example in (2). None of the three show any evidence of a null syntactic subject in Ukr.7 Tab. 1: Typology of morphologically marked AIs (and reflexive passive/impersonal).
Upper Bg
Ru
BRu
Ukr
Po
refl pass
BCS
BCS
spk/dlctl
stdrd
Slvn
Sorb
Cz
Slk
oA
oA
oA
oA
oA
*oA
*oA
*oA
*oA
*oA
*oA
-nS
-nS
-nS
-nS
-nS
+nS
+nS
+nS
-nS
-nS
-nS
refl AI
oA, -nS
*oA, +nS
*oA, +nS
*oA, +nS
no/to AI
oA, -nS
*oA, +nS
refl impers
‘oA’ = oblique Agent (by-phrase) ok, ‘*oA’ = oA impossible, ‘+nS’ = null syntactic subject, ‘-nS’ = no null subject, filled (grey) cells means type not available
Tab. 2: Distribution of Adversity AIs across Slavic, partly based an Mrázek (1990).
Bg
Upper
Ru
BRu
Ukr
Po
Slvn
Sorb
BCS
BCS
spk/
stdrd
Cz
Slk
ok
ok
dlctl A-AIs
ok
ok
ok
ok
ok
ok
Filled (grey) cells means type not available
Table 2 shows the distribution of A-AIs in Slavic. They are attested in all Slavic languages except Bg and BCS. In addition to their varying availability, there are also differences wrt their productivity and their compatibility with FORCEphrases (compare the optional instrumental NP in example (3)), which will be discussed in detail below.
|| 7 We abstract away from the fact that the speaker community is not homogeneous in Ukraine and base our argumentation on facts taken from the standard literature. We concede that speakers from Eastern Ukraine, especially Ukrainian/Russian bilinguals, may have different judgements.
144 | U. Junghanns (Göttingen), D. Lenertová (Berlin) and D. Fehrmann (Göttingen)
We see that, on the one hand, morphologically marked AIs cannot receive a unique analysis, and, on the other hand, the distribution of morphologically marked vs. A-AIs is different. Thus a simple feature or parameter can account neither for the availability of AIs in Slavic, nor for the cross-Slavic typology of AIs. In JFL (2009) and FJL (2010), we discussed the variation wrt morphologically marked AIs in Slavic and attributed the difference between Ukr vs. Po, Slvn, and spoken/dialectal BCS to two types of refl/no/to and to a special property [-agr] T in Ukr (and Po). The goal of this paper is to find explanations for the parametric variation concerning A-AIs, ultimately allowing to draw the overall picture of Slavic AIs (morphologically marked and unmarked): their typology and parametrization. The following variation with respect to A-AIs has to be accounted for: (i) their availability across Slavic, (ii) their productivity/status as core vs. periphery in each language, (iii) the availability of an optional FORCE-phrase. The main issues to be addressed are the status of the external argument with A-AIs, the licensing of accusative case, whether there is a null subject or not and, if so, what kind of null subject, as well as capturing the nature of FORCE-phrases. In our proposal we will attribute the properties of A-AIs solely to the presence of a null indefinite pronoun specified as [+FORCE], which realizes the external argument. It refers by definition to non-animate, hence non-volitional, non-agentive entities. The availability of A-AIs depends on the presence of the null pronoun in the lexicon of a given language. The varying productivity is related to the status of the null pronoun – free/separately available in the lexicon vs. fixed in the lexical entries of the relevant verbs. The variation wrt the availability of FORCE-phrases will be attributed to two lexical types of the null pronoun – one that is not yet existentially quantified and can therefore be further specified, viz. by an optional FORCE-phrase (adjunct), and one that is already existentially quantified excluding further specification by a FORCE-phrase. The structure of the paper is as follows. In section 2 we discuss in detail the properties and distribution of A-AIs in Slavic, including a special subtype of AAIs – with the overt pronoun to (subsection 2.5). In section 3 we develop our analysis of A-AIs. The main claim is that the external argument of the transitive causative verb is realized as a null element – a pronoun with FORCE semantics. The problems of occurrence of A-AIs as well as restrictions on use in the languages are tackled. Finally, in section 4 we give an outlook on a possible overall parametrization of Slavic AIs, including morphologically marked AIs and A-AIs as well as Accusative Impersonalia tantum.
Parametric Variation of Slavic Accusative Impersonals | 145
2 Properties and distribution of A-AIs Due to the productivity of A-AIs in East Slavic, the discussion so far concentrated mostly on Ru (and Ukr), cf., e.g., Babby (1994), Mel’čuk (1995), Lavine and Freidin (2002), Markman (2004), Mustajoki and Kopotev (2005), Szucsich (2007, 2008), Lavine and Franks (2008), Lavine (2010, 2013). For Po, see Włodarczyk (1993) and Kibort (2006). The cross-Slavic availability of A-AIs is discussed in Mrázek (1990).
2.1 Verb morphology As shown in (3) above, A-AIs involve transitive verbs assigning accusative case to their internal argument despite the lack of an overt external argument. The verbs appear in the active voice and exhibit 3rd person singular neuter morphology8. The examples in (5) illustrate a typical pattern with A-AIs, namely perfective past tense verbs, for the whole range of Slavic. However, Mustajoki and Kopotev (2005) show for Ru that aspect, tense, and mood of the transitive verbs in A-AIs are variable and there are no special restrictions on word order. See (6a, b) for examples with imperfective verbs in the present resp. past tense. As illustrated for Slk in (7a, b) (imperfective aspect, conditional mood), A-AIs in the other Slavic languages are not restricted with respect to the verbal categories either. (5) a. Žyto vybylo (hradom). rye.acc beat.past.sg.n hail.instr ‘The rye was beaten down (by hail).’ (adapted from Mrázek 1990: 104) b. Žyta pavybivala (hradam). rye.acc beat.past.sg.n hail.instr ‘The rye was beaten down (by hail).’ c. Zabity. Miną go rąbnęło. killed.ptcp.sg.m mine.instr him.acc blow-up.past.sg.n ‘Dead. He was blown up by a mine.’ (Żukrowski, ex. from Mrázek 1990: 104)
(Ukr)
(BRu)
(Po)
|| 8 Note that 3sg.n morphology on a verb in Slavic either signals agreement with a singular neuter subject or the default realization in case of missing matching -features (for the latter see, e. g., Szucsich 2007: 426).
146 | U. Junghanns (Göttingen), D. Lenertová (Berlin) and D. Fehrmann (Göttingen)
d. (Prívalom vody) odnieslo (Slk) torrent.instr water.gen carry-away.past.sg.n tamten most. that bridge.acc ‘The bridge was carried away (by the torrent of water).’ (adapted from Mrázek 1990: 104) e. Ladjico je odneslo. (Slvn) boat.acc aux.3sg carry-away.past.sg.n ‘The boat was carried away.’ f. Tam je nas tak prawje trjechiło. (Upper Sorb) there aux.3sg us.acc so really hit.past.sg.n ‘We were really hit hard there.’ g. Obilí potlouklo. (Cz) corn.acc beat-down.past.sg.n ‘The corn was beaten down.’ (Mrázek 1990: 96) (6) a. U nego pod burkoj po krajnosti (Ru) prep him.gen under felt-cloak.instr at least doždëm ne probiraet.9 rain.instr neg soak.ipf.pres.3sg ‘Under his felt cloak she at least does not get soaked by the rain.’ (Leskov: Ledi Makbet) b. Kogda menja unosilo tečeniem, […] as me.acc carry-away.ipf.past.sg.n current.instr ‘As I was being carried away by the current, …’ (Mustajoki and Kopotev 2005: 13) (7) a. Vo vzduchu lietali strechy, stromy, konáre … (Slk) ‘Roofs, trees, branches were flying through the air …’ a odnášalo ich desiatky až and carry-away.ipf.past.sg.n them.acc tens till stovky metrov. hundreds meters.gen ‘… and they were carried away tens to hundreds of meters.’
|| 9 Due to ellipsis, the accusative object is deleted.
Parametric Variation of Slavic Accusative Impersonals | 147
b. No v rýchlom prúde by ich but in quick current would.3sg them.acc odnášalo a tak to ani neskúšali. carry-away.ipf.past.sg.n ‘so they didn’t even try it’ ‘But they would be carried away in the quick current, so they didn’t even try it.’
2.2 Verb semantics A-AIs are based on transitive causative verbs and are causative. pulej. (Ru) (8) a. Soldata ubilo soldier.acc kill.past.sg.n bullet.instr ‘The soldier got killed by a bullet.’ b. λy λx λs [[s INST [P x]] : [[s CAUSE s’] : [s’ INST [BECOME NOT ALIVE y]]]] The form ubilo in (8a) is a form of the transitive verb ubit’ (pf) ‘to kill’. The meaning representation for this verb, part of its lexical entry, is given in (8b).10 The binding of the variables y and x by the operators λy and λx respectively indicates that the verb has two structural arguments, normally realized with accusative and nominative case. In the meaning representation, s and s’ (the causing and the caused situation or cause and effect, cf. Bierwisch 2005) are related via the semantic primitive CAUSE. This makes the verb causative.11 CAUSE remains in the semantic representation of the verb form as used in (8a), following standard assumptions excluding elimination of semantic components. However, transitive causative verbs as used in A-AIs obligatorily receive a non-agentive interpretation. Thus they are lexically restricted to predicates that are compatible with non-agentive subjects (see, e. g., Babby 1994). The verb in (9) is obligatorily agentive (cf. Padučeva 2003). As expected, such a verb cannot be used in an A-AI. Moreover, A-AIs are incompatible with agent-oriented adverbs – (10), controlled instruments – (11), controlled purpose clauses – (12), and oblique agents – (13).
|| 10 Unbound variables remain semantic parameters and undergo binding by default at the level of CS (for the term see below, sec. 3.1), cf. FJL (2014, sec. 3.2). On binding by default see also Chierchia (2004 [1989]: 37). 11 For more details of the analysis of causative verbs see FJL (2014), section 3.2.
148 | U. Junghanns (Göttingen), D. Lenertová (Berlin) and D. Fehrmann (Göttingen)
(9) a. Ivan zaper kalitku. (Ru) Ivan.nom bolt.past.sg.m gate.acc ‘Ivan bolted the gate.’ (adapted from Padučeva 2001: 25) b. *Veter zaper kalitku. wind.nom bolt.past.sg.m gate.acc c. *Kalitku zaperlo. gate.acc bolt.past.sg.n Intended interpretation: ‘The gate got bolted.’ (10) Dom sožglo (*special’no). (Ru) house.acc burn.past.sg.n purposefully ‘The house got burned down.’ (Markman 2004: 426) (11) Ego rezko udarilo (balkoj /kirpičom / (Ru) him.acc hard hit.past.sg.n beam.instr brick.instr *palkoj / *nogoj / *nožom). 12 stick.instr foot.instr knife.instr ‘He got struck by a beam/brick.’ (Babby 1994: 29) (12) Kryšu sneslo, (*čtoby ubit’ ved’mu). (Ru) roof.acc carry-away.past.sg.n in-order-to kill.inf witch.acc ‘The roof got blown away.’ (13) Vetram / *Lesnikom zvalila (BRu) wind.instr forest-warden.instr uproot.past.sg.n al’xu. 13 alder.acc ‘The alder got uprooted by wind.’ (Plotnikaw and Antanjuk 2003: 110)
2.3 Variation wrt productivity A-AIs seem to be unattested only in BCS and Bg. They are available in East and West Slavic and Slovenian. However, their status differs in the languages. According to the literature (e. g., Mrázek 1990: 96 and 103-4) and to our informants,
|| 12 Note that only controlled instruments are excluded. However, FORCE-phrases are possible. 13 Observe the contrast between the instrumental FORCE-phrase, which is fine, and the oblique agent, which is not.
Parametric Variation of Slavic Accusative Impersonals | 149
the pattern is productive in East Slavic, and frequent in Slvn. For these languages, the expressions are more or less central. In West Slavic, on the other hand, they belong to the periphery – Po, Slk – or are rather marginal – Upper Sorb, Cz (but see below).14
2.4 Variation wrt the availability of FORCE-phrases A-AIs may allow instrumental NPs or PPs with FORCE interpretation, see (14)– (18). The Ru example in (14b) illustrates a PP. The ‘FORCE’-semantics involves processes occurring in the world – nature, physical world, see (14)–(16) (and also (3), (5a-d), and (6a-b) above), but also ‘weapons’ or ‘vehicles’ as a force as in (17) and (18), causing an “uncontrolled human physical or emotional experience” (Babby 1994: 27). See Mustajoki and Kopotev (2005) for discussion of the various types of “kauzatory” (causing forces). The languages differ with respect to whether A-AIs allow a FORCE-phrase. Importantly, in languages allowing FORCE-phrases in A-AIs, they are optional. We indicate this by the brackets in the examples. (14) a. Lodku uneslo (vetrom). (Ru) boat.acc carry-away.past.sg.n wind.instr ‘The boat was carried away (by the wind).’ (Mustajoki and Kopotev 2005: 12) b. Zontik sognulo (ot vetra).15 umbrella.acc bend.past.sg.n from wind.gen ‘The umbrella was bent (by the wind).’ (15) Poraziło go (prądem). (Po) strike.past.sg.n him.acc current.instr ‘He was struck down by electricity.’ (Mrázek 1990: 104) (16) Kusy pobrežného ľadu odnieslo (Slk) pieces.acc shore.gen ice.gen carry-away.past.sg.n (vetrom) do zálivu. wind.instr in bay.gen ‘The wind carried away the ice floes into the bay.’ || 14 Note that in languages where A-AIs are peripheral speaker judgements on the acceptability of A-AIs and, in particular, FORCE-phrases contained therein may vary. 15 We would like to point out that not all native speakers of Russian like A-AIs with a PP. Some examples can be found in internet texts produced by native speakers. The example shown in the text was judged acceptable by our informants.
150 | U. Junghanns (Göttingen), D. Lenertová (Berlin) and D. Fehrmann (Göttingen)
(17)
(18)
Ego pereexalo (avtomobilem). him.acc run-over.past.sg.n car.instr ‘He was run over by a car.’ (adapted from Babby 1994: 28) Ivanovi pokaličylo ruku (mašynoju). Ivan.dat cripple.past.sg.n hand.acc machine.instr ‘Ivan’s hand was severed (by a machine).’ (adapted from Mrázek 1990: 104)
(Ru)
(Ukr)
While FORCE-phrases are allowed in East Slavic (Ru, BRu, Ukr), Po, and Slk, they are ungrammatical in Cz – (19) (based on (5g)), Slvn – (20) (based on (5e)), and Upper Sorb – (21). (19)
(20)
(21)
Obilí potlouklo (*krupobitím). (Cz) corn.acc beat.past.sg.n hail.instr ‘The corn was beaten down.’ (‘by hail’ impossible) Ladjico je odneslo (*od vetra). (Slvn) boat.acc aux.3sg carry-away.past.sg.n from wind.gen ‘The boat was carried away.’ (‘by the wind’ impossible) Při tym njewjedrje je dweju mužow (Upper Sorb) in this storm aux.3sg two.acc men.gen zabiło. kill.past.sg.n ‘Two men were killed during the storm.’ (FORCE-phrase impossible)
An A-AI, thus, minimally consists of the accusative nominal phrase and the verb. The FORCE-phrase, if allowed, is optional.16 The two variables wrt the status (productive vs. marginal) and the availability of FORCE-phrase are summarized in Table 3. All combinations are attested, in other words, the availability of FORCE-phrases does not correlate with the central/peripheral parameter. There also is no connection between the possibility of a by-phrase with refl Passive and the availability of FORCE-phrases with A-AIs. Ru allows by-phrases in refl Passive, but Po does not. Both languages allow FORCE-phrases in A-AIs. Upper Sorb has by-phrases in refl Passive, like Ru, but no FORCE-phrases in A-AIs. (For by-phrase variation with refl Passive, see Table 1).
|| 16 Mrázek (1990: 96) and AG80 II, 353 use the term rasprostranitel’ (modifying element).
Parametric Variation of Slavic Accusative Impersonals | 151
Table 3: Cross-Slavic variation with respect to A-AIs.
central
peripheral
FORCE-phrases
Ru, BRu, Ukr
Po, Slk
= Group A
no FORCE-phrases
Slvn
Cz, Upper Sorb
= Group B
An account must be found that allows FORCE-phrases in some languages and excludes them in others. Also, we have to bear in mind that a FORCE-phrase, where allowed, is optional. See section 3.2 for the proposed solution.
2.5 A-AIs with overt pronoun A final remark is due on Cz. A-AIs are limited to a few verbs. On the other hand, there is a version of A-AIs with an overt pronoun to as shown in the examples in (22a, b). To in Cz is a multifunctional element: demonstrative pronoun, 3sg.n pronoun, (focus, intensifying) particle, eventive topic pronoun (see Junghanns 1997 and Progovac 1998).17 Importantly, to in (22a, b) cannot be classified according to these categories. (22a) is a headline without an immediately preceding context, which to could possibly refer to. In (22b), one cannot understand to as referring to the event of falling into the river. Contextually inferrable natural elements like proud, příval (m.) ‘current’, voda (f.) ‘water’ do not agree with to in gender. Thus to rather is a pronoun interpreted as non-volitional causer of pulling down and hurling out (or burying somebody with stones as in (22a)). It is the same interpretation as with A-AIs. Note the infelicitous English paraphrase with it (‘A girl fell into a river. #It
|| 17 If to is unstressed, it occupies the second position like other second position clitics (it is placed in the final position of the clitic cluster, cf. (22b)). Note that to in cases like (22a, b) is not equivalent to the expletive use of it in English or es in German. Cz is a pro-drop language and sentences like (i) have regularly no (expletive) subject, cf. (ia) vs. (ib). The rise of to in Cz was discussed already in Zubatý (1909), see Meyer (2011). (i)
a. Ø prší. rains.3sg.n ‘It rains.’ b. * Prší
to.
rains.3sg.n it.nom
152 | U. Junghanns (Göttingen), D. Lenertová (Berlin) and D. Fehrmann (Göttingen)
pulled her down and then hurled her out …’). It would obligatorily refer to the river, which is less plausible/felicitous as a causer of pulling down/hurling out than an unspecified element inferred by a passive-like sentence). (22) a. Bagrista, kterého to zavalilo v (Cz) excavator-op.nom whom.acc to.nom bury.past.sg.n in tunelu Blanka, skončil na JIP. tunnel.loc Blanka end-up.past.sg.m in intensive-care-unit ‘An excavator operator who was buried in the Blanka tunnel ended up in the intensive care unit.’ b. … spadla do divoké řeky. ‘… a girl [she] fell into a wild river.’ Stáhlo ji to úplně dolů, poté ji to pull.past.sg.n her.acc to.nom fully down then her.acc to.nom vymrštilo a chytila catapult.past.sg.n and catch-hold-of.past.sg.f se kamene. refl stone.gen. ‘She got pulled under, then she was hurled out and caught hold of a rock.’ In some cases of A-AIs in Cz, like in (19), the null is possible, but in most cases – as in (22a, b) – to is obligatory.18 There is further cross-linguistic evidence. Szucsich (2007) points out to AIs in Bavarian and other German dialects with es ‘it’. In these examples, es is obligatory. Note that in (23), es occupies a middlefield position, which is not available for the expletive es, cf. (24b). Consequently, es in (23) cannot be analysed as an expletive element. (23)
Mi z’reißt * (’s) voa Loch’n. me.acc burst.3sg -it from laughter ‘I’m going to burst from laughter.’ (Szucsich 2007: 428) (24) a. Es wurde viel gelacht. it was.3sg much laughed ‘One laughed a lot.’
(Bavarian)
(German)
|| 18 Slk does not have the option of overt to, which is interesting, since it is a neighbouring language and it is closely related with Cz.
Parametric Variation of Slavic Accusative Impersonals | 153
b. Hier wurde ( * es) here was.3sg it ‘One laughed a lot here.’
viel gelacht. much laughed
Czinglar (2002) analyses obligatory es in German and Alemannic accusative impersonal existentials – cf. the examples in (25a) and (25b) – as a semantically empty quasi-argument19 in the position of the external argument licensing accusative case. * (es) (25) a. In meinem Garten gibt in my garden give.3sg it viele Gänseblümchen. many daisies.acc ‘There are many daisies in my garden.’ b. In minam Garta hot * (’s) in my garden have.3sg -it an Hufa Gänseblüamle. a pile.acc daisies ‘There are many daisies in my garden.’ (Czinglar 2002: 87)
(German)
(Alemannic)
We take overtly realized to as in (22a, b) as supporting evidence for postulating a null pronoun in Slavic A-AIs. Note that the parallel between the two structures in Slavic has not been drawn in the literature, yet.
3 Analysis 3.1
Realization of the external argument
With the exception of Cz to, in Slavic A-AIs the external argument is not overtly realized. However, an operation on argument structure cannot possibly be invoked, since there is no morphological exponent in A-AIs that would signal it. Apart from the parallel between structures without an overt pronominal and with
|| 19 A quasi-argument is a syntactically obligatory expression that does not receive a thematic role. See, e. g., Růžička (1999: 170) on the subject of German regnen ‘to rain’, a quasiargument.
154 | U. Junghanns (Göttingen), D. Lenertová (Berlin) and D. Fehrmann (Göttingen)
overt to in Cz, there is further evidence for a null realization of the external argument in Slavic A-AIs. Szucsich (2007) following Testelec (2001) takes control into gerunds as argument for the presence of a null subject (ns) in East Slavic A-AIs. The argumentation is based on the assumption that an embedded gerundive expression contains a PRO subject that needs to be controlled and that in Ru, this PRO must be controlled by the matrix-clause subject, cf. Testelec (2001).20 That the gerundive expression is possible with the A-AI can be taken as evidence for a syntactically realized subject in the matrix, i. e. in the A-AI, cf. (26). (26)
Mašinu zaneslo na povorote, (Ru) car.acc ns make-skid.past.3sg.n at bend [PRO razvernuv poperëk šosse]. turn.ger across highway ‘The car skidded at the bend, turning 90 degrees across the highway.’ (Testelec 2001: 312)
Apart from gerunds, control is also possible from an A-AI into an infinitival expression – an “ironical” purpose clause21 as in the following example. (27)
Možet potomu Vas v bol’nicu i zaneslo, (Ru) maybe for-that you.acc in hospital.acc prt bring.past.sg.n čtoby PRO spasti ot takogo poganogo vrača in-order-to save.inf from such.gen terrible.gen doctor.gen buduščix pacientov ètoj bol’nicy. future.acc patients.acc this.gen hospital.gen ‘Maybe you have been brought to the hospital, so that future patients of this hospital are saved from such a terrible doctor.’
There have been several proposals concerning null subjects with A-AIs. In Mel’čuk’s (1995) account a ‘force’ null element – ØFORCE – occupies the subject position in Ru A-AIs. Szucsich (2007) proposes a semantically bleached nominal expression D[-φ] without φ-features and with low referentiality which, however, does not lack any referentiality; its semantic role need not necessarily be ‘natural force’ (following Babby 1994). For Po, Włodarczyk (1993) proposes a zero subject with the features [-hum, sg, n]. Kibort (2006) argues that Po A-AIs are cases of
|| 20 For more details on control into Ru gerunds see, e. g., Růžička (1982, 1990a, 1990b, 1994). 21 The term refers to expressions that have been associated with fate/fortune – “fügung des schicksals” (cf. Bech 1957: 123-124, see also Leys 1971: 50 and Junghanns 1994: 63-64).
Parametric Variation of Slavic Accusative Impersonals | 155
pro-drop: what is dropped is the indefinite (singular neuter) pronoun coś ‘something’ referring to non-humans. We also assume a null subject with an indefinite interpretation. It refers to non-animate entities belonging to the domain of FORCE. The null pronoun realizes the external argument. Since the external argument is syntactically realized, the accusative is licensed as usual. Consequently, sentences with A-AIs have a normal agreeing Tense head – T[+agr]. The predicate is finite and agrees with the null subject pronoun, the -features of which are specified as 3sg.n. Note that this is different from Szucsich (2007) where the semantically bleached pronoun D has no -features and finite T’s -features have to be valued as [default]. Moreover, whereas Szucsich (2007) assumes one type of pronoun for both morphologically marked AIs and A-AIs, in our account the pronoun in A-AIs differs from its counterpart in morphologically marked AIs in its semantic specification: [+arbhum] for the pronoun in morphologically marked AIs vs. [FORCE] for the pronoun in AAIs, which excludes reference to humans in the latter case. Babby (1994: 32) argues against the presence of a null subject: a null subject “denoting an unknown or unidentifiable natural force that employs the INST [instrumental] case NP as its instrument” is semantically implausible. However in our account, the instrumental NP, where realized, is not to be understood as an instrument used by the natural force subject, but as a free adjunct semantically specifying the highly unspecified null pronoun, see below, section 3.2 for details. FORCE-phrases – NPs or PPs – are optional. In some languages they are excluded at all, see section 2.4. In order to account for the variation, we propose two types of the null indefinite FORCE pronoun. The semantic representations, part of the respective lexical entries, are given in (28) and (29). (28) (29)
Null pronoun 1 x [[FORCE x] : [Q x]] Null pronoun 2 x x [+FORCE]
Null pronoun 1 has the usual semantics of an indefinite expression with the conjunction of two properties, cf., e. g., Partee (1987).22 Null pronoun 2 is just the variable x restricted to the domain of FORCE. Importantly, with null pronoun 2 the
|| 22 On entering the derivation, the pronoun is assigned a referential index (cf. Bierwisch 1988: 8-9).
156 | U. Junghanns (Göttingen), D. Lenertová (Berlin) and D. Fehrmann (Göttingen)
variable x is not yet existentially quantified at the level of S(emantic) F(orm)23. Here, FORCE hast to be interpreted as a member of an arbitrary set. Null pronoun 1 can be referred to as existentially quantified FORCE, and Null pronoun 2 as not existentially quantified FORCE. In the process of meaning composition (semantic amalgamation), the representation of the null pronoun available in the language – either null pronoun 1 or null pronoun 2 – replaces variable x in the meaning representation of the verb.24 (30)
Semantic amalgamation involving null pronoun 1
(31)
λx λs [[s INST [P x]] : [[s CAUSE s’] : [s’ INST [BECOME … ]]]] (xi [[FORCE xi] : [Q xi]]) λs [[s INST [P [xi [[FORCE xi] : [Q xi]]]]] : [[s CAUSE s’] : [s’ INST [BECOME … ]]]] Semantic amalgamation involving null pronoun 2 λx λs [[s INST [P x]] : [[s CAUSE s’] : [s’ INST [BECOME … ]]]] (x) λs [[s INST [P x]] : [[s CAUSE s’] : [s’ INST [BECOME … ]]]]
As a result, we get a complex meaning representation containing existentially quantified FORCE in the case of some languages (those using null pronoun 1), and not existentially quantified FORCE in the case of some other languages (those using null pronoun 2). In either case, there is a predicate P applying to the respective FORCE item. Since P remains a semantic parameter, FORCE is conceptually interpreted as being involved in some situation or other. Importantly, further semantic specification of an expression is precluded if this expression is existentially quantified at the level of SF.25 Hence, semantic specification through an optional FORCE-phrase is excluded in languages using null pronoun 1, cf. (30). Such specification is, however, possible in languages using null pronoun 2, cf. (31). Examples as, e. g., Ru (3) vs. Cz (19) or Slvn (20) can serve as illustration of the two cases. See also below, section 3.2. An alternative analysis of A-AIs could employ blocking of the external argument without overt morphological marking (e. g., via a semantic template). The
|| 23 For the characterization of SF see Bierwisch (1986, 2007) and Lang and Maienborn (2011), a. o. 24 In a preceding step, the meaning representation of the object expression replaces variable y in the meaning representation of the verb. This is not shown here. 25 See FJL (2010) for discussion.
Parametric Variation of Slavic Accusative Impersonals | 157
external argument is blocked, but not yet bound, hence optional oblique realization, e. g., as an argument-adjunct (cf. Grimshaw 1990), would be possible. Crucially, such an analysis still leaves us with the problem of exceptional acc licensing, as the external argument is not projected canonically, in the subject position. Haider (2000: 45) argues that the inflection system of Slavic provides an impersonal paradigm not admitting nom licensing, which leaves acc licensing as the only option. However, the distribution of such a [-agr] T across Slavic languages would have to be different for morphologically marked AIs (available only in Ukr, Po, Slvn, spoken/dialectal BCS) vs. A-AIs (available in East Slavic generally and West Slavic peripherally), cf. Tables 1 and 2. As A-AIs involve verbs in the active voice not marked with special morphology that would signal a change of argument structure, all languages allowing A-AIs should also allow refl AIs.
3.2 Variation wrt the availability of FORCE-phrases In personally constructed sentences corresponding to A-AIs, the nominative FORCE-phrase is an external argument. Such sentences can be passivized, as argued in Szucsich (2007: 425) contra Babby (1994). See (32). (32) a. Molnija oslepila Ivana. lightning.nom.f blind.past.sg.f Ivan.acc ‘A flash of lightning blinded Ivan.’ b. Ivan byl osleplën molniej. Ivan.nom aux.past.sg.m blind.ptcp.sg.m lightning.instr ‘Ivan was blinded by a flash of lightning.’ (Szucsich 2007: 425)
(Ru)
Accounts of A-AIs assuming blocking of the external argument analyse FORCEphrases as oblique realization of the external argument, e. g., as an argumentadjunct. See Kwon (2010) for a proposal to analyse the FORCE-NP/PPs along these lines. However, such an account is not possible on our assumptions if the external argument is analysed as a syntactic null.
158 | U. Junghanns (Göttingen), D. Lenertová (Berlin) and D. Fehrmann (Göttingen)
Moreover, in Po there is a difference in realization: a genuine by-phrase in participial passive is realized as przez-PP (33b), a FORCE-phrase in A-AIs is realized as bare instrumental (33c).26 zabił konia. (Po) (33) a. Piorun lightning.nom.m kill.past.sg.m horse.acc ‘A flash of lightning killed a horse.’ b. Koń został zabity horse.nom.m aux.past.sg.m kill.ptcp.sg.m przez żołnierzy / przez piorun. by soldiers.acc by lightning.acc ‘The horse was killed by soldiers / by a flash of lightning.’ c. Konia zabiło piorunem / *przez piorun. horse.acc kill.past.sg.n lightning.instr by lightning.acc ‘The horse got killed by a flash of lightning.’ So it is quite clear that the FORCE-phrase in Po A-AIs is not an oblique agent/ non-volitional causer. Thus it is not unmotivated to assume that it is an adverbial adjunct. This is compatible with its optionality (cf. Szucsich 2007 for a similar argumentation on Ru). Recall the empirical observation that there are two groups of languages with A-AIs. Group A allows optional FORCE-phrases (Ru, BRu, Ukr, Po, Slk), Group B does not (Slvn, Cz, Upper Sorb), see Table 3. In our account, the variation is related to the two types of null pronoun presented in (28) and (29) above. Group A languages employ null pronoun 2, a null pronoun that semantically corresponds to a variable x restricted to the domain of FORCE and is not yet existentially quantified at SF so that it gets an interpretation as an indefinite belonging to an arbitrary set, see (29). Since the variable remains unbound, it can still be semantically specified at SF. Semantic specification is the result of overtly realizing a FORCE-phrase (an NP with instrumental case or an NP selected by a P). In a second step of interpretation – at the level of CS27 – x will be identified with || 26 Note that not all native speakers of Po readily accept instrumental expressions in A-AIs (see also fn 14). However, such FORCE-phrases are attested as (5c) and (15) from Mrázek (1990) – see above– show. The sources are authentic printed texts. The variation may be attributed to the coexistence of both types of the null pronoun in Po. This may be a matter of register or diachronic change. 27 C(onceptual) S(tructure) is the second level of interpretation in a two-level semantics. See Bierwisch (1986, 2007) and Lang and Maienborn (2011) for the details of such a framework. For SF, the first level, see fn 23.
Parametric Variation of Slavic Accusative Impersonals | 159
the FORCE-phrase’s NP referent. If the FORCE-phrase is not realized, then the variable will be existentially quantified at CS by default. Group B languages have null pronoun 1. Its meaning representation conjoins two properties predicated of x, the first one characterizing the entity as FORCE. Existential quantification makes the pronoun an indefinite. Since x is bound by the existential quantifier at SF, further semantic specification of the pronoun is excluded. Realization of a FORCE-phrase would not yield a licit interpretation. The effect would be characterized as caused by an indefinite FORCE (with a referential index) and yet another FORCE, which necessarily would have to be referentially distinct from the indefinite FORCE.28 Note that the exclusion of FORCE-phrases cannot be tied with the possibility of an overt FORCE pronoun. Slvn, e. g., has no overt pronoun and does not allow a FORCE-phrase. Cz, on the other hand, has overt to and does not allow a FORCEphrase either. There have been other proposals analysing the FORCE-phrase as a second internal argument with lexical case (see, e. g., Babby 1994, Lavine 2005, Lavine and Freidin 2002). At the same time these authors postulate a derivational relation between sentences with a nominative FORCE-NP in subject position and their A-AI-counterparts with the oblique FORCE-phrase. Arguments against such an approach have been brought up, e. g., by Szucsich (2007), including, among others, problems such as overriding of lexical case by structural nominative (under standard assumptions lexical case cannot be overridden), examples of passivization of sentences with a nominative FORCE-NP – cf. our (32b) – (passivization should be excluded in an analysis assuming that the sentence lacks an external argument).
3.3 Variation wrt availability and productivity Whether a language allows A-AIs depends on the availability of an appropriate null element in the lexicon of the language. Attributing the varying distribution of A-AIs across Slavic to the availability of an indefinite null FORCE pronoun is compatible with the option of overt realization of the pronoun, as in Cz equivalents of A-AIs with to. Thus to in (22a, b) has the same status and semantics as null pronoun 1, cf. (28). With some verbs in Cz it seems to be lexically required for the FORCE pronoun to be overt, cf. (22a, b). We may think of restrictions of the FORCE pronoun – overt
|| 28 According to our intuition, A-AIs do not render scenarios with multiple causing factors.
160 | U. Junghanns (Göttingen), D. Lenertová (Berlin) and D. Fehrmann (Göttingen)
or covert – in Cz (laid down in the lexical entries of the relevant verbs), which would be one possible explanation of the peripheral status of A-AIs in this language as compared with, e. g., East Slavic where the (null) pronoun quite productively combines with transitive verbs.
4 Outlook Note that the minimal structure of A-AIs without adjunct FORCE-phrases (see above, section 2.4) resembles Accusative Impersonalia tantum, cf. (34)–(36). The latter are attested in all Slavic languages. (34)
(35)
(36)
Mdli mnie (od tego zapachu / z bólu). (Po) nauseate.3sg me.acc from this smell.gen because-of pain.gen ‘Something makes me nauseous.’ (‘This smell/The pain makes me nauseous.’) (Kibort 2006) Menja tošnit. (Ru) me.acc nauseate.3sg ‘I feel nauseous.’ (Perlmutter and Moore 2002: 628) Zamrazilo mě (z toho). (Cz) chill.past.sg.n me.acc from this.gen ‘It made me shiver.’
It seems to be possible to control the non-overt subject of a gerund from within Accusative Impersonalia tantum, which may be taken as evidence for a structural subject in the matrix. (37)
Menja tošnit, ne davaja mne spat’. me.acc nauseate.3sg neg give.ger me.dat sleep.inf ‘I feel nauseous, which prevents me from sleeping.’
(Ru)
Attested Cz cases of the overt pronoun to with, e. g., dávit ‘choke, retch’, cf. (38b), as opposed to null pronoun use, cf. (38a), probably point in the same direction.
Parametric Variation of Slavic Accusative Impersonals | 161
(38) a. […] stále ji natahovalo a dávilo, až (Cz) constantly her.acc choke.past.sg.n and retch.past.sg.n till to vypadalo, že se snad udusí. it.nom look-like.past.sg.n that refl probably choke-to-death.3sg ‘She constantly choked and retched, till it looked like that she would choke to death.’ b. Kocoura občas natahuje, ale je pravda, že už cat.acc time-to-time choke.3sg but is.3sg truth.nom that already ho to dávilo včera […] míň a míň. him.acc to.nom retch.past.sg.n yesterday […] less and less ‘The cat chokes from time to time, but it’s true, that yesterday, he retched less and less.’ Therefore, it seems not to be implausible to suggest a unified analysis for A-AIs and Accusative Impersonalia tantum as a syntactic transitive structure.29 However, different null elements seem to be involved. Accusative Impersonalia tantum have a causer that appears to be an internal (physical, mental) entity, whereas A-AIs have a causer that is an external entity. The null internal causer is available also in Bg and BCS, i. e. in all Slavic languages (cf. Mrázek 1990: 95-96). To sum up, we have been concerned with various cases of AIs: (i) morphologically marked AIs, and (ii) A-AIs. In this paper, we have argued that A-AIs should receive a different analysis than morphologically marked AIs. Morphologically marked AIs comprise (a) refl AIs and (b) no/to AIs. In these cases, morphological items (refl, no/to) signal an operation on the argument structure of the verb they combine with so that the canonical realization of the external argument is prevented. We have brought out two crucial properties concerning subject realization and agreement, yielding the following cases: ±null subject, [-agr] T vs. no [-agr] T. The following picture can be drawn for the languages investigated, cf. JFL (2009) and see Table 1: (39) a. Po: +null subject, [-agr] T b. Ukr: -null subject, [-agr] T c. Slvn, spoken/dialektal BCS: +null subject, no [-agr] T A-AIs contain a transitive causative verb. By contrast to morphologically marked AIs (see above), with A-AIs there is no morphological marker to indicate a
|| 29 On this point see also Lavine and Franks (2008) and Lavine (2010: 122) who conclude for predications as nauseating and cramping in Ru that they must be causative.
162 | U. Junghanns (Göttingen), D. Lenertová (Berlin) and D. Fehrmann (Göttingen)
changed argument structure. The 3sg.n ending potentially corresponds to either default agreement (non-agreement) or normal agreement with a sg.n subject. We have proposed that the external argument of verbs used in A-AIs is realized as null pronoun. Thus, assignment of accusative case to the internal argument proceeds in the usual way. The T head agrees with the null in the relevant features. So we assume that there is an external argument and it is not demoted. This leaves us with the task to account for FORCE-phrases (NPs with instrumental case or PPs). Group A languages (Ru, BRu, Ukr, Po, Slk), which allow a FORCE-phrase in an A-AI, have null pronoun 2 (= not existentially quantified FORCE) as a lexical item. Group B languages (Slvn, Cz, Upper Sorb) employ null pronoun 1 (= existentially quantified FORCE). Since we invoke a two-level semantics framework distinguishing between SF and CS, we can explain realization of a FORCE-phrase by the relevant variable not being bound at SF – Group A languages. In contrast, in Group B languages using null pronoun 1, such realization is excluded, since variable x is existentially quantified, the pronoun has a referential index, and FORCE cannot be semantically specified. So our account of the possibility vs. exclusion of a FORCE-phrase relies on the assumption of two lexical types of the indefinite null pronoun. We analyse FORCE-phrases as free adverbial adjuncts. Importantly, such FORCE-phrases are optional. So the minimal structure of A-AIs comprises the verb form and the accusative nominal expression (internal argument). It can be shown that Accusative Impersonalia tantum structurally resemble A-AIs. This has been claimed before. It remains to be seen whether there is more evidence beyond control facts and an overt subject pronoun. This is left for future research.
References AG80. 1980. Russkaja grammatika [Russian Grammar]. Vol. II. Sintaksis [Syntax]. Moskva: Nauka. Babby, Leonard H. 1994. A theta-theoretic analysis of adversity impersonal sentences in Russian. In Sergey Avrutin, Steven Franks, and Ljiljana Progovac (eds.), Annual Workshop on Formal Approaches to Slavic Linguistics: The MIT Meeting 1993, 25–67. Ann Arbor: Michigan Slavic Publications. Bech, Gunnar. 1957. Studien über das deutsche verbum infinitum. Band 2 (= Historisk-filologiske meddelelser; 36.6). København: Munksgaard. Bierwisch, Manfred. 1986. On the nature of Semantic Form in natural language. In Friedhart Klix and Herbert Hagendorf (eds.), Human memory and cognitive capabilities – mechanisms and performances, 765–784. Amsterdam & New York: North-Holland Publishers.
Parametric Variation of Slavic Accusative Impersonals | 163
Bierwisch, Manfred. 1988. On the Grammar of Local Prepositions. In Manfred Bierwisch, Wolfgang Motsch, and Ilse Zimmermann (eds.), Syntax, Semantik und Lexikon, 1–65. Berlin: Akademie-Verlag. Bierwisch, Manfred. 2005. The event structure of CAUSE and BECOME. In Claudia Maienborn and Angelika Wöllstein (eds.), Event arguments: Foundations and applications, 11–44. Tübingen: Niemeyer. Bierwisch, Manfred. 2007. Semantic Form as interface. In Andreas Späth (ed.), Interfaces and interface conditions, 1–32. Berlin & New York: de Gruyter. Burzio, Luigi. 1986. Italian Syntax. Dordrecht: Reidel. Chierchia, Gennaro. 2004 [1989]. A semantics for unaccusatives and its syntactic consequences. In Artemis Alexiadou, Elena Anagnostopoulou, and Martin Everaert (eds.), The unaccusativity puzzle: Explorations of the Syntax–Lexicon interface, 22–59. Oxford & New York: OUP. Czinglar, Christine. 2002. Decomposing Existence: Evidence from Germanic. In Werner Abraham and Jan-Wouter Zwart (eds.), Issues in Formal German(ic) Typology, 85–126. Amsterdam: Benjamins. FJL 2010 = Fehrmann, Dorothee, Uwe Junghanns, and Denisa Lenertová. 2010. Two reflexive markers in Slavic. Russian Linguistics 34.3, 203–238. FJL 2014 = Fehrmann, Dorothee, Uwe Junghanns, and Denisa Lenertová. 2014. Slavic Reflexive Decausative. Russian Linguistics 38.3, 287–313. Grimshaw, Jane. 1990. Argument Structure (= Linguistic Inquiry Monographs; 18). Cambridge, Mass. & London, England: MIT Press. Haider, Hubert. 2000. The license to license: Licensing of structural case plus economy yields Burzio’s Generalization. In Eric Reuland (ed.), Arguments and case: Explaining Burzio’s Generalization, 31–55. Amsterdam & Philadelphia: Benjamins. Harves, Stephanie. 2006. Non-agreement, unaccusativity, and the external argument constraint. In James E. Lavine, Steven Franks, Mila Tasseva-Kurktchieva and Hana Filip (eds.), Annual Workshop on Formal Approaches to Slavic Linguistics: The Princeton Meeting 2005, 172–188. Ann Arbor: Michigan Slavic Publications. JFL 2009 = Junghanns, Uwe, Dorothee Fehrmann, and Denisa Lenertová. 2009. Accusative Impersonals and the issue of external selection. Paper presented at the 8th European Conference on Formal Description of Slavic Languages, University of Potsdam, 2–5 December. Junghanns, Uwe. 1994. Syntaktische und semantische Eigenschaften russischer finaler Infinitiveinbettungen (= Slavistische Beiträge; 315). München: Sagner. Junghanns, Uwe. 1997. On the so-called èto-cleft construction. In Martina Lindseth and Steven Franks (eds.), Annual Workshop on Formal Approaches to Slavic Linguistics: The Indiana Meeting 1996, 166–190. Ann Arbor: Michigan Slavic Publications. Kibort, Anna. 2006. On three different types of subjectlessness and how to model them in LFG. In Miriam Butt and Tracy H. King (eds.), LFG-06 Conference, Universität Konstanz, 2006. CSLI Publications Online. Kupść, Anna. 2000. Lexical Analysis of Polish Multifunctional Reflexive Marker. In Tracy H. King and Irina A. Sekerina (eds.), Annual Workshop on Formal Approaches to Slavic Linguistics: The Philadelphia Meeting 1999, 214–237. Ann Arbor: Michigan Slavic Publications. Kwon, Kyongjoon. 2010. The Adversity Impersonal Construction in Diachrony. In Wayles Browne, Adam Cooper, Alison Fisher, Esra Kesici, Nikola Predolac and Draga Zec (eds.), Annual Workshop on Formal Approaches to Slavic Linguistics: The Second Cornell Meeting 2009, 255–269. Ann Arbor: Michigan Slavic Publications.
164 | U. Junghanns (Göttingen), D. Lenertová (Berlin) and D. Fehrmann (Göttingen)
Lang, Ewald and Claudia Maienborn. 2011. Two-level semantics: Semantic Form and Conceptual Structure. In Claudia Maienborn, Klaus von Heusinger, and Paul Portner (eds.), Semantics: An international handbook of natural language meaning. Volume 1, 709–740. Berlin & New York: Mouton de Gruyter. Lavine, James. 2005. The Morphosyntax of Polish and Ukrainian -no/-to. Journal of Slavic Linguistics 13.1, 75–117. Lavine, James. 2010. Case and Events in Transitive Impersonals. Journal of Slavic Linguistics 18.1, 101–130. Lavine, James. 2013. Passives and near-passives in Balto-Slavic: On the survival of accusative. In Artemis Alexiadou and Florian Schäfer (eds.), Non-canonical Passives, 185–212. Amsterdam & Philadelphia: Benjamins. Lavine, James and Steven Franks. 2008. On Accusative First. In Andrei Antonenko, John F. Bailyn, and Christina Y. Bethin (eds.), Annual Workshop on Formal Approaches to Slavic Linguistics: The Stony Brook Meeting 2007, 231–247. Ann Arbor: Michigan Slavic Publications. Lavine, James and Robert Freidin. 2002. The subject of defective T(ense) in Slavic. Journal of Slavic Linguistics 10.1–2, 253–289. Leys, Odo. 1971. Die Präpositionalinfinitive im Deutschen. Einige Beobachtungen. Leuvense Bijdragen 60, 1–56. Markman, Vera G. 2004. Causatives without Causers and Burzio’s Generalization. In Keir Moulton and Matthew Wolf (eds.), NELS 34 (2003), 425–440. Amherst, MA: Graduate Linguistic Students Asociation. Mel’čuk, Igor’ A. 1995. Russkij jazyk v modeli “smysl – tekst” [The Russian language in the “smysl – tekst” model]. Moskva & Wien: Škola “Jazyki russkoj kul’tury”. Meyer, Roland. 2011. The history of null subjects in North Slavonic. Regensburg: University of Regensburg Habilitation thesis. Mrázek, Roman. 1990. Sravnitel’nyj sintaksis slavjanskix jazykov. Isxodnye struktury prostogo predloženija [Comparative Syntax of Slavic Languages. Underlying Structures of Simple Clauses]. Brno: Univerzita J.E.P. Mustajoki, Arto and Mikhail V. Kopotev. 2005. Lodku uneslo vetrom: Uslovija i konteksty upotreblenija russkoj “stixijnoj” konstrukcii [Lodku uneslo vetrom: Conditions and contexts of use of the Russian “elementary” construction]. Russian Linguistics 29.1, 1–38. Padučeva, Elena V. 2001. Russkij dekauzativ i ego formal’nye modeli [The Russian Decausative and its Formal Models]. Naučno-texničeskaja informacija. Serija 2. Informacionnye processy i sistemy, 23–34. Moskva: RAN VINITI, 1 (2001). Padučeva, Elena V. 2003. Is there an “ANTICAUSATIVE” component in the semantics of decausatives? Journal of Slavic Linguistics 11.1, 173–198. Partee, Barbara. 1987. Noun Phrase Interpretation and Type-Shifting Principles. In Jeroen Groenendijk, Dik de Jongh, and Martin Stokhof (eds.), Studies in Discourse Representation Theory and the Theory of Generalized Quantifiers, 115–143. Dordrecht: Foris. Perlmutter, David and John Moore. 2002. Language-internal explanation: The distribution of Russian impersonals. Language 78, 619-650. Plotnikaw, Branislaw A. and Ljubow A. Antanjuk. 2003. Belaruskaja mova. Linhvistyčny kampendyum [Belarusian. Linguistic Compendium]. Minsk: Intèrprèssèrvis/Knižny Dom. Progovac, Ljiljana. 1998. Event Pronominal To. Journal of Slavic Linguistics 6.1, 3–39. Pylkkänen, Liina. 2008. Introducing arguments. Cambridge, MA: MIT Press.
Parametric Variation of Slavic Accusative Impersonals | 165
Rivero, María Luisa. 2000. Impersonal się in Polish: A Simplex Expression Anaphor. Journal of Slavic Linguistics 8.1–2, 199–237. Rivero, María Luisa and Milena Milojević Sheppard. 2003. Indefinite Reflexive Clitics in Slavic: Polish and Slovenian. Natural Language and Linguistic Theory 21, 89–155. Růžička, Rudolf. 1982. Kontrollprinzipien infinitiver Satzformen: Infinitiv und Gerundium (deepričastie) im Russischen und in anderen slavischen Sprachen. Zeitschrift für Slawistik 27.3, 373–411. Růžička, Rudolf. 1990a. Über Syntax und Semantik russischer Gerundien (Adverbialpartizipien, Deepričastija). Wiener Slawistischer Almanach 25/26, 361–378. Růžička, Rudolf. 1990b. Leere Kategorien und Verstehensprinzipien partizipieller Adjunkte (Deepričastija) des Russischen. In Walter Breu (ed.), Slavistische Linguistik 1989. Referate des XV. Konstanzer Slavistischen Arbeitstreffens, Bayreuth 18.–22.9.1989, 269–284. München: Sagner. Růžička, Rudolf. 1994. Russische partizipielle Adjunkte (Deepričastija) als Instanz der Ökonomie sprachlicher Struktur und theoretischer Derivation. In Hans Robert Mehlig (ed.), Slavistische Linguistik 1993. Referate des XIX. Konstanzer Slavistischen Arbeitstreffens, Kiel 21.–23.9.1993, 293–306. München: Sagner. Růžička, Rudolf. 1999. Control in Grammar and Pragmatics. Amsterdam & Philadelphia: Benjamins. Šeljakin, Mixail A. 2001. Funkcional’naja grammatika russkogo jazyka [Functional Grammar of Russian]. Moskva: Russkij jazyk. Szucsich, Luka. 2007. Nothing Wrong with Finite T: Non-Agreeing Accusative Impersonal Sentences. In Richard Compton, Magdalena Goledzinowska, and Ulyana Savchenko (eds.), Annual Workshop on Formal Approaches to Slavic Linguistics: The Toronto Meeting 2006, 417–435. Ann Arbor: Michigan Slavic Publications. Szucsich, Luka. 2008. Evidenz für syntaktische Nullen im Burgenland-Kroatischen, Polnischen, Russischen und Slovenischen. Merkmalsausstattung, Merkmalshierarchien und morphologische Defaults. Zeitschrift für Slawistik 53.2, 160–177. Szucsich, Luka. 2009. Animacy and Multiple Feature Sharing: Restrictions with Accusative Reflexive Impersonals in Burgenland-Croatian and Varieties of BCS. In Steven Franks, Vrinda Chidambaram, and Brian Joseph (eds.), A Linguist’s Linguist: Studies in South Slavic Linguistics in Honor of E. Wayles Browne, 423–444. Bloomington, Indiana: Slavica. Testelec, Jakov. 2001. Vvedenie v obščij sintaksis [Introduction to General Syntax]. Moskva: RGGU. Tsedryk, Egor. 2004. Case and agreement in Russian adversity impersonal constructions. In Olga Arnaudova, Wayles Brown, María Luisa Rivero, and Danijela Stojanović (eds.), Annual Workshop on Formal Approaches to Slavic Linguistics: The Ottawa Meeting 2003, 419–438. Ann Arbor: Michigan Slavic Publications. Włodarczyk, Hélène. 1993. Sentences without a Nominative NP in Polish. In Gerd Hentschel and Roman Laskowski (eds.), Studies in Polish Morphology and Syntax (= Specimina Philologiae Slavicae; 99), 209–227. München: Sagner. Zolotova, Galina A. 1988. Sintaksičeskij slovar’: Repertuar èlementarnyx edinic russkogo sintaksisa [Syntactic Dictionary: Repertoire of elementary units of Russian syntax]. Moskva: Nauka. Zubatý, Josef. 1909. O jistém způsobu užívání zájmen an a on a jeho původu [On a certain kind of use of the pronouns an and on and its origin]. Listy filologické 36, 25–34, 106–117, 230–240, 334–355.
Irina M. Kobozeva (Moskva)
Russian namek and English hint as Ordinary Language Hyponyms of Grice’s Term Implicature 1 Introduction The topic of semantic relations between the term implicature and ordinary words associated with it was touched upon by Н.P Grice himself in his famous (1975) paper “Logic and conversation”. In the first paragraph of the section Implicature he gives the first example of conversational implicature, reproduced in (1) below in reduced form: (1)
A asks B how their mutual friend C is getting on in his job. B: …he has not been to prison yet.
and comments on it: “…whatever B implied, suggested, meant, etc., in this eхample, is distinct from what he said… [emphasis mine – I. K.] I want to introduce, as terms of the art, the verb implicate and the related nouns implicature (cf. implying) and implicatum (what is implied)” (Grice 1975: 43–44). Grice did not say exactly whether he thinks that English verbs are synonyms and adds implicate as a professional synonym to this row, or whether he places implicate as their hypernym. Не characterized the semantic relations of his term implicate with its ordinary language correlates in an informal and rather a vague manner: the verb implicate “is to do general duty” for “this or that member of this family of verbs” (Grice 1975: 44). In the Russian translation of this passage the interpreter (V.V. Turovskij) used the expression родовое понятие ‘generic notion’ (Grice 1985: 220) thus treating implicate as the hypernym of a certain number of ordinary language co-hyponyms. Such an interpretation seems plausible and may be taken as a starting point for further discussion. Then two issues for lexical semantics arise: 1) the membership of what Grice called the implicate family of verbs; 2) the distinctive features of each member of the family. As for issue 1, Grice explicitly mentioned only three members of the semantic class of English verbs related to the terminological verb implicate – imply, suggest, mean, indicating with etc. that there are others. And he added: “I shall …
DOI 10.1515/9783110517873-008
Russian namek and English hint as Hyponyms of Grice’s Term mplicature | 167
have to assume … an ability to recognize particular verbs as members of the family, with which implicate is associated” (Grice 1975: 44) thus relying on introspective evidence. The second issue was not relevant for Grice. As far as I know the semantics of ordinary language hyponyms of implicate has not been exhaustively studied neither for English nor for Russian1. In what follows I shall try to concentrate on one particular member of the implicate family – the verb hint and its Russian translational equivalent namekat’ ‘to hint’ (with their derivative nouns hint and namek). The reason for this choice is that hint was not explicitly mentioned by Grice among the co-hyponyms of implicate / implicature being at the same time stylistically neutral and a commonly used name of the significant part of the phenomena that he analyzed. The aim of this study is to outline the borders of the ordinary language concept HINT underlying both the English verb and its Russian equivalent in the realm of the scientific concept IMPLICATURE as defined by Grice. Before I proceed, two remarks are in order. First, we need to recall that Grice also uses the verb say as a term of the art. For him to say P is to express / convey the conventional meaning of P minus the conventional implicatures, i.e. those aspects of conventional meaning that are not truth conditions of P. Utterance meaning 2 Conventional 2 SAID
Non-conventional
IMPLICATED 2
conventional implicature
conversational implicature
Fig. 1: Types of utterance meaning in Grice’s theory of implicature.
|| 1 Indirect speech acts of different types were studied in the perspective of various linguistic pragmatic theories, cf. six types identified in (Cheng and Warren 2003): 1) conversational implicatures (Grice 1975); 2) indirect speech acts (Searle 1975); 3) indeterminate illocutionary acts (Sew 1997); 4) nonliteral illocutionary acts (Bach and Harnish 1979); 5) collateral acts (Bach and Harnish 1979); 6) pre-sequences (Levinson 1983). But the question about their relation to the meaning and denotation of natural language names of the corresponding speech acts was not at issue in this literature.
168 | Irina M. Kobozeva (Moscow)
The ordinary verb say is used in a way that does not suit the definition of the Gricean term say. It is clear from the following exchange, containing quite a natural use of the verb say: (2)
Jill: Where’s my favorite dress? Jack: Your ugly dress is under the table. Jill: How can you say (?imply, *mean, *hint) that my dress is ugly?
In her second turn Jill reports the conventional implicature (here also the presupposition) of Jack’s answer as something that was said by Jack, whereas the term say as defined by Grice would be inappropriate here and the term implicate should be used. Note that the verbs of the implicate family are also inappropriate in such a context. Here I leave open whether there are conventional implicatures that can be reported as something implied or hinted at. In this paper I check the applicability of the verbs hint and namekat’ only for the cases of conversational implicature and so below the term implicature stands for conversational implicature only. The second remark concerns derivational relations of the terminological nouns implicature and implicatum to implicate. Grice likened the first noun to implying. So he may have considered implicature at that point as an event nominalization. The noun implicatum stood for what is implied. But later he himself and all his linguistic followers used the term implicature as a synonym of implicatum, i.e. as denoting the information conveyed as a result of implicating. So we can speak not only of ordinary language verbs associated with implicate but also of a correlative set of ordinary language nouns associated with implicature (see below).
2 Implicate and implicature families in English and Russian Now let us have a closer look at the ordinary words of the implicate and implicature families in English and Russian. According to various dictionaries of synonyms, the corresponding sets include at least the following verbs and verbal collocations:
Russian namek and English hint as Hyponyms of Grice’s Term mplicature | 169
(3) Implicate family: a. English: imply, hint, suggest, mean, intimate, insinuate, allude at the fact, let it be known, be driving at b. Russian: намекать, подразумевать, иметь в виду, хотеть сказать, давать понять, клонить к (4) Implicature family: a. English: hint, clue, mention, suggestion, implication, indication, reminder, tip-off, pointer, allusion, innuendo, inkling, intimation, insinuation, word to the wise; b. Russian: намек, инсинуация, аллюзия, недомолвка, укол, обиняки, экивоки, околичности, закавыка, камешек в чей-либо огород The members of the family are by no means semantically identical. The main dividing line is the one that separates the true speech act verbs in (5) from those in (6) that only presuppose speech acts in the context of speech communication: (5) a. b. (6) a. b.
English: Russian: English: Russian:
suggest, hint, intimate, insinuate, allude намекать, давать понять imply, mean подразумевать, иметь в виду, хотеть сказать
Depending on the context, the verbs in (5) may denote acts of speech or non-verbal communicative acts as such. Bach and Harnish (1979, 96-97) classified speech acts of this type as collateral speech acts, i.e. speech acts that are performed together with illocutionary acts or instead of these. It is not the case with the expressions in (6). If a speaker S is said to imply or to mean some information Q, then S must have said (or communicated) something and he must have used some utterance (or some other communicative device) U. But the given verbs assert only the mental relation between S and the content Q of U (S had Q in his mind, while saying U). This semantic difference is brought to light in (7): (7) а. Mary told John that he might read more books and in his turn he hinted / *implied / *meant that she was too fat. b. Маша сказала Ивану, что ему следовало бы больше читать, a тот в ответ намекнул ей /*подразумевал / *имел в виду, что она слишком толстая. As far as «conversation» is a key word in the title of Grice’s famous paper we have chosen the propotypical speech act verb of the IMPLICATE family for the purpose
170 | Irina M. Kobozeva (Moscow)
of comparing the notion IMPLICATURE to a naïve pragmatic category dubbed by the ordinary language word. It is the Russian verb namekat’ nominalized as namek and its closest translational equivalent in English - hint (verb and noun). Judging by lexical entries of hint and namek in dictionaries and by data of the parallel Russian-English sub-corpus of the Russian National Corpus (RNC) we can treat both words as having the same lexical meaning in the sense of speech act verbs. This meaning is represented in dictionaries as ‘to say indirectly or covertly’. In other words, they express the same concept that I capitalize as HINT. In fact those two verbs have yet another speech act meaning: ‘to say something directly but in such a way that the addressee would not suspect that the topic is really important to the speaker’ disregarded in dictionaries. This secondary meaning is illustrated by (8): (8)
When she'd mildly complained over breakfast one day of her loneliness and boredom, Gerard had suggested charity organisations, flower-arranging classes and cordon bleu cookery courses. When she'd hinted at a baby instead, he'd vetoed that for another year at least. (Miranda Lee. Fugitive Bride (1998)).
As expected in RNC hinted in (8) is translated by nameknula. It is clear from the context that the woman expressed her wish to have a baby overtly and directly (the anaphoric that refers to the idea of having a baby, expressed by the woman). If her message was ‘indirect or covert’, the hearer could well pretend he did not get it, and thus escape the need to enter into direct conflict with his wife’s wishes. The use of the verb hint here instead of say marks some special quality of the speech act: the wish was expressed directly, but as mildely as a previous complaint, without insistence, cf. the parameter of “strength” with which the speaker introduces the illocutionary goal in Searle’s classification of illocutionary acts (Searle 1976). Such “weak” speech acts, sometimes denoted by the verbs hint and namekat’ are not subsumed under the notion of HINT represented in dictionaries as ‘to say indirectly or covertly’ and we shall not discuss them here.
Russian namek and English hint as Hyponyms of Grice’s Term mplicature | 171
3 Namekat’ and hint: differences in argument structure. Нint (v) is always given as a translational equivalent to namekat’ and vice versa in bilingual dictionaries, but their properties are not fully identical. They differ in their argument structure. Both verbs have 4 semantic arguments: 1 – Agent (the one who hints) 2 – Theme (the implied meaning) 3 – Addressee 4 – Instrument (the utterance used or its parts) The verb hint normally realizes only arguments 1 and 2, as is shown in (9): (9) a. b. c.
[The Minister ]1 hinted at [a possible change of heart ]2 (OED) [Не]1 hinted [that the sale might be delayed]2 (OED) ? Нint [to him]3 that you are ready for a date. (Internet) = OKDrop / Give him a hint that you are ready for a date.
(9c) shows that when we want to mention the Addressee, the synonymous lightverb collocation is preferable. The verb namekat’ may realize all the 4 arguments, as in (10): (10)
[Упоминанием о занятиях в Москве]4 [Сергей Иванович]1 намекал [Левину]3 на [славянский вопрос]2. (abridged from RNC) [Lit.: By mentioning his affairs in Moscow Sergej Ivanovič hinted to Levin at the Slavic question].
4 The previous study of hinting using data of Russian The phenomenon of hinting was studied by I.M. Kobozeva and N.I. Laufer (1988) on the basis of linguistic data of Russian lexis and discourse. The authors pursued two goals:
172 | Irina M. Kobozeva (Moscow)
a) to give a definition of this ordinary language concept by analyzing the meaning of the verb namekat’ in proper contexts; b) to extract the linguistic techniques of hinting from the data of literary dialogues (about 100 contexts with words namek, namekat’ and their contextual synonyms). I shall summarize the results of the study beginning from the modes of hinting in order to show that the phenomena denoted by the Russian verb and its contextual synonyms are closely connected to those studied by H.P. Grice. Then I shall proceed to the definition of the concept HINT based on the meaning of namekat’, the Russian equivalent of the verb to hint.
4.1 Modes of hinting The contexts we have analyzed contained the indirect utterance of some speaker and either the explication of its implicatum or some other material that enabled the addressee (and the analyzer) to infer the implicatum. In the majority of cases the speech act of the utterer was described in the text with the help of namek, namekat’ or one of their synonyms. If the name of some indirect utterances was not mentioned in a text we checked whether the corresponding speech act could be reported with the help of namek, namekat’. We came to distinguishing five types, or modes of hinting. I shall enumerate them giving one typical example of each type. At the same time I shall connect this material with Grice’s theory of implicature. The latter was not done in (Kobozeva and Laufer 1988).
4.1.1 Hinting via indeterminacy This mode is represented by the remark of the clerk in (11): (11)
Context: A man (Čičikov) comes to the Council to register his purchase and says to a clerk (Ivan Antonovič) he wants to finish with the matter this very day. The clerk denies the possibility. Then the following exchange takes place: – Впрочем, что до того, чтоб ускорить дело, так Иван Григорьевич, председатель, мне большой друг... – Да ведь Иван Григорьевич не один; бывают и другие, – сказал сурово Иван Антонович. (И.В. Гоголь. Мертвые души)
Russian namek and English hint as Hyponyms of Grice’s Term mplicature | 173
[“Then, to expedite matters, let me say that Ivan Grigorievich, the President of the Council, is a very intimate friend of mine.” “Possibly” said Ivan Antonovitch without enthusiasm, “but the President alone will not do – it is customary to have others as well.”] (Gogol’ “Dead Souls”. Translated by Hogarth) Saying it is customary to have others as well the clerk in the given context hints that in order to expedite matters Čičikov has to be friends not only with the President, but with him as well (and applying common knowledge Čičikov correctly interprets it as a coercion to bribe). The main linguistic feature of such a mode of hinting is either the use of universal, generic or some other plural indefinite NP for a definite specific referent or the omission (or zero expression) of some part or aspect of the propositional content to be conveyed by the utterance: of some arguments or adjunct, of the predicate itself, of one of the elementary propositions of the complex proposition. Now if we look at such utterances through the prism of the Gricean theory of implicature we may say that all these tricks make the utterance less specific and so less informative than is expected in a given situational and dialogical context.
4.1.2 Нinting via premise This mode may be illustrated by the second remark in a dialogical fragment (12): (12)
– …«Левиафан» – настоящий плавучий дворец. Комфортабельно, удобно и, главное, совершенно безопасно. Ведь так? Что может случиться с таким великаном? – Я слышал, «Левиафан» застрахован на такую сумму, что на одни лишь годовые страховые взносы можно было бы построить неплохой каботажный пароход. – Что вы хотите этим сказать? – Вы коммерсант, и вам нетрудно сделать вывод: если бы не было риска, пароходная компания не выбрасывала бы денег на страхование. (А. Беляев. Чудесное око)
174 | Irina M. Kobozeva (Moscow)
[“– «Leviathan» is a real floating palace. Comfortable, convenient and, what is most important, safe. Isn’t it? What can happen with such a giant?” “– I’ve heard «Leviathan» was insured for such a huge sum of money that one could build a good coastwise steamer for annual premiums alone.” “– What are you trying to say?” “– You are a businessman and you can easily come to the conclusion: if there was no risk, the shipping company would not throw money about for insurance.”] (A. Beljaev. The magic eye. [Translation mine – I. K.]). In this case the second speaker hints that the risk exists by supplying a small premise for the addressee who is supposed to evoke the relevant big premise and infer the proposition that the speaker indirectly expressed in such a manner. Another kind of this type of hinting is based on the overt expression of a big premise (e.g. in the form of a proverb) that must be interpreted with the help of a small premise that the addressee is expected to evoke from the situational or dialogical context and come to the intended meaning. From the point of view of Gricean theory in this type of hints the maxim of Relation is exploited.
4.1.3 Hinting via complementarity This mode could be represented by the first remark in (13) this time constructed by us for the sake of brevity: (13)
– Из тебя мог бы выйти хороший повар. – Ты намекаешь, что из меня вышел плохой детектив? [“– You could have made a good cook. – Are you hinting that I have made a bad detective?”] (From a movie). [Translation mine – I. K.]
This type of hinting exploits the complementary relation between members of some set (here the set of alternative professions the addressee could have chosen and the one he had chosen in reality). Explicitly ascribing some property to a certain member of this set (here it is the positive evaluation of professionalism addressee could have if he became a cook) the speaker implies that the other members do not possess the given property (thereby implicitly gives negative evaluation of addressee’s professionalism in other spheres). Again here as in type 1 Gricean maxim of Quantity is at work together with the Economy Principle in
Russian namek and English hint as Hyponyms of Grice’s Term mplicature | 175
communication, cf. the analysis of the similar example in (Padučeva 1985: 44, example 2).
4.1.4 Hinting via circumlocution It is the most elaborate of all the modes of hinting. It presupposes the play of words, the intended ambiguity on the speaker’s part and the ability of the addressee to uncover the second meaning under the disguise of the first one. An example of such a hint is (14). (14)
[Answering the joke of his chief that now the addressee has two Annas on the neck (one Anna is the order of St Anna, the other is addressee’s wife) Modest Alekseevič, the addressee, says:] – Теперь остается ожидать появления на свет маленького Владимира. (А.П. Чехов. Анна на шее.) [“– Now I have only to look forward to the arrival of a little Vladimir”]. (A.P. Čechov «Anna on the Neck». [Translated by C. Garnett.]
The remark in (14) is to be understood as ‘Now I have only to look forward to be awarded with the order of St Vladimir of the lowest (fourth) grade’. The use of an ambiguous name in ambiguous context flouts one of the Gricean maxims of Manner (“Avoid ambiguity”) and serves as a signal to seek for a non-literal interpretation.
4.1.5 Hinting via equivoque This mode is exemplified by (15) addressed by a boy to a girl in case the boy expects her to understand that he is speaking about his feelings towards her: (15)
– Один мой приятель полюбил девушку, но не знает, как сказать ей об этом. [“A friend of mine fell in love with a girl but he doesn’t know how to tell her about it.”]
This seems to be the only type of hints where the Gricean maxims do not help, because none of them is flouted by speakers using so-called Aesop language without announcing it. In fact introducing a topic about one’s friend need not
176 | Irina M. Kobozeva (Moscow)
violate the maxim of Relation or any other conversational maxims. In order to model the functioning of this type of hinting some other explanatory principles are in order. In cases like (15) it is a principle of interpersonal relations captured by I. Goffman’s notion of face (Goffman 1955). The boy might lose his face if he spoke of his feelings overtly and the girl rejected him. Knowing that his interlocutor shares the same principle he can expect that his hidden message will be worked out on the basis of some analogy between the situation spoken of and the actual communicative situation. In cases of literary hints (allusions) the only explanatory principle often is the extralinguistic ban on expressing some ideas. It seems that capturing the analogy between the situation overtly described in the utterance and the actual state of affairs (whether interpersonal or political) is the crucial part of the mechanism of interpreting this type of hinting.
4.2 The definition of the concept HINT as a result of the semantic analysis of the Russian verb namek(at’) On the basis of the analysis of the use of the verb namekat’, Kobozeva and Laufer proposed the definition of the concept, expressed by this verb (1988, 465). I argue that this definition in a slightly revised form (16) can be treated as a definition of the concept HINT, expressed by hint in English, namekat’ in Russian and their translational equivalents in other languages. (16)
The Speaker (S) HINTS at the information (‘I’) to the Hearer (H) using the text (T) = 1) S intends to convey ‘I’ to H (intentionality condition); 2) for that purpose S uses T with literal meaning ‘T’ in the given context (linguistic means condition); 3) ‘T’ ≠ ‘I’ (indirectness condition); 4) there is a special reason for S not to express ‘I’ as I, i.e. conventionally (condition of justifiability of the indirectness); 5) S believes that H can infer ‘I’ from ‘T’ using the context and common knowledge (calculability condition); 6) inference of ‘I’ from ‘T’ is not trivial (non-triviality condition).
One has to bear in mind that this concept is expressed by the core uses of the corresponding verbs which may have peripheral uses that belong to some other speech act concept (see the peripheral use of hint in (8) that obviously does not fit the definition).
Russian namek and English hint as Hyponyms of Grice’s Term mplicature | 177
This definition of the concept HINT is in some respects similar to Grice’s definition of conversational implicature (Grice 1975: 49-50), although it does not mention Cooperative principle explicitly, subsuming it under the notion of common knowledge in condition 5. But it comprises two conditions, namely 4 and 6 that are not part of Grice’s definition. These conditions need elaboration.
4.3 Comments to the definition of HINT The condition 4 that the speaker must have special reasons for indirectness was introduced on empirical grounds. In all the analyzed examples the hinting person had one of the following reasons to use non-conventional means for the expression of some information ‘I’: – to conceal the information ‘I’ from some other people present (over-hearers); – to avoid responsibility for breaking some rules (ethical or legal norms) that could arise from the conventional expression of ‘I’ as I; – to introduce an element of play into communication. The non-triviality of inferring ‘I’ from ‘T’ in the given context (condition 6) is a gradable parameter. The different degree of non-triviality is expressed by attributes, mostly metaphoric that combine with the verbs and deverbative nouns verbalizing the concept HINT, e.g. in Russian prozrachno namekat’ ‘to hint transparently’, tumannyj namek ‘hint that it is not easy to understand, lit. ‘a foggy hint’ or in English to hint clearly, opaque hint and the like. Our data show that non-triviality of inference in hinting has the following sources: – the difficulty in identifying the reference of some term or proposition on the basis of the literal meaning ‘T’; – the difficulty in recognizing the aspect of similarity (in metaphors); – the need to appeal to some background knowledge that is not common. By common knowledge I mean here knowledge that is accessible to an average adult person. It comprises the following kinds of information: – most trivial common knowledge that is fixed in lexical meanings of everyday vocabulary; – the minimal extralinguistic knowledge acquired at school or through traditional educational practices; – cultural stereotypes (some of them are fixed linguistically as connotations of words);
178 | Irina M. Kobozeva (Moscow)
–
–
communicative knowledge, comprising unspoken rules scientifically represented in linguistic pragmatics as felicity conditions of illocutionary acts, conversational maхims, rules of politeness etc.; axioms / postulates of natural logic.
Now after comparing the explicit definition of the concept HINT (16) and Grice’s definition of implicature and stating that the former concept is more restricted than the latter, we can expect that the word hint and its equivalents in other languages should behave as hyponyms of the term (conversational) implicature. In what follows we shall substantiate this claim by testing the cases of implicature from (Grice 1975) for the applicability of the word hint to them. If hint is a hyponym of implicature, then it should be the case that at least some of the examples of implicature discussed by Grice cannot be correctly named by hint, verb or noun.
5 Hint as a kind of (conversational) implicature Does the verb hint (and Russian namekat’) really denote a subclass of speech acts that form the denotation of the term implicate and does the noun hint serve as an ordinary language name of a special kind of implicit information that Grice called (conversational) implicatures? If the answer is yes, are the distinctive features correctly captured by conditions 4 (justifiability of indirectness) and 6 (non-triviality)? In section 4 commenting on different types of hinting we have shown that the common knowledge used to infer the implied information includes the maxims of Cooperative principle in each type of hinting with the exception of type 5 (hinting via circumlocution). But if we include other general principles of interpersonal communication, e.g. the Politeness Principle or the Principle of Saving Face as a part of the mechanism of implication (and Grice admitted the possibility of non-conventional implicatures generated in a dialogue by social and moral maxims), then the hints of this type may also be regarded as representatives of nonconventional implicature. So we can suggest that ultimately all cases of hinting are analyzable as non-conventional implicatures. This point is hard to prove or falsify. The empirical fact is that all НINTS in the data discussed in (4) are not calculable without resorting either to the Cooperative Principle with some of its maxims or to some other general pragmatic principle. The reverse is not true. Not everything that is a non-conventional implicature is a hint. Among the cases discussed in Grice (1975) we have found several kinds
Russian namek and English hint as Hyponyms of Grice’s Term mplicature | 179
of those that do not belong to the category HINT (that is they cannot be reported as cases of hinting). Let us have a closer look at them in order to see whether the reason for this is that they violate some of the conditions of the definition of the concept HINT (16). We shall reproduce Grice’s examples using the index of a group to which they belong and their number given in (Grice 1975).
5.1 Grice’s Group A example (1): А: I am out of petrol. B: There is a garage round the corner. В conversationally implicates ‘The garage is open, and has petrol to sell’. We cannot report B’ s remark as in (17а), while (17b) is all right: ??
B said that there is a garage round the corner thereby hinting that the garage is open… b. OK...implying…
(17) a.
The inapplicability of hint is explained by the fact that two of the conditions of definition (16) are not satisfied in this case. The condition of justifiability is violated because B has no special reason for indirectness, (s)he just choses the optimal (most economical) way to be cooperative. As for the condition of non-triviality, we can cite Grice’s own comment: “the unstated connection of В’s remark and A’s remark is so obvious…” [Emphasis mine – I. K.] (Grice 1975: 51). In a situational context of a man standing next to an immobilized car the lexical meanings of petrol and garage are related within the semantic field of a CAR in a way that guarantees the unequivocal calculation of the connection between having no petrol and there being a garage nearby. Compare it to Grice’s Group A example (2): А: Smith doesn’t seem to have a girlfriend these days. B: He has been paying a lot of visits to New York lately. B’s remark successfully passes our test for the applicability of hint, see (18): (18)
OK
B said Smith has been paying a lot of visits to New York lately thereby hinting that he has or might have a girl there.
180 | Irina M. Kobozeva (Moscow)
As predicted by definition (16) both conditions violated in the previous example are satisfied in this one: justifiability – there is a special reason for indirectness: not to violate privacy of Smith overtly; non-triviality – there is no steadfast connection between going to some place and having a girl there although it can be the case. Cf. Grice’s comment about this case: “The next example is perhaps less clear in this respect [in respect of unstated connection of B’s remark to A’s, comment and emphasis mine – I. K.]” (Grice 1975: 51). In Grice’s interpretation B’s remark contradicts A’s conjecture. B could begin his turn with “I doubt it, because …” But the same remark can be reinterpreted as an argument in favor of A’s conjecture. B could as well begin with “You may be right, because…” assuming that if A had a girl he wouldn’t go to New York so often. Hence non-triviality of calculation: A and B need to resort to very particular facts about Smith, e. g. that he has no close relatives or business goals in NY.
5.2 Grice’s Group B, example (3): А: Where does C live? B: Somewhere in the South of France. B implicates that he does not know in which town C lives. Here again B’s turn cannot be reported as (19a), while (19b) and (19c) are all right: ??
B answered C lives somewhere in the south of France, thereby hinting that he does not know in which town C lives. b. OK…thereby demonstrating / showing… c. OK…as if (s)he didn’t know exactly.
(19) a.
Explanation is the same: the only reason for not expressing the implicatum is to satisfy A’s communicative needs in the optimal way and the calculation of the implicatum is based only on conversational principles, thus trivial.
5.3 Grice’s Group C (1a). A flouting of the first maxim of Quantity In the case of patent tautologies the use of hint in reported speech is anomalous, see (20): (20)
??
War is war, hinted Mr. Х.
Russian namek and English hint as Hyponyms of Grice’s Term mplicature | 181
Again here we have no special reason for indirectness and in any actual situational context the calculation of the implicatum ‘I’ is trivial: it is based on the stereotype beliefs about war and consists in actualization of the one which is the most relevant for the given situation. By the way, patent contradictions, which Grice did not discuss in (1975), produce different results. At least some of them may be reported as namekanie / hinting, see an abridged example from RNC (21а), and its translation (21b): (21) а. Похороны, они не всегда похороны, – намекнул дед. b. Funerals are not always funerals, hinted the old man. The implicatum ‘I’ here: either the funeral under discussion was false or it lacked some property of a prototype. It is more difficult to determine what was meant by the speaker.
5.4 Grice’s Group C (2a) – examples where the first maxim of Quality is flouted In cases of patent irony, when e. g. A says Х is a fine friend (T) meaning the opposite (‘I’ = ‘X is a bad friend’), verbs namekat’ and hint are inapplicable, as (22a) and (22 b) show: (22) a. b.
??
Saying “Х is a fine friend” A hinted that Х is a bad friend Говоря «Х- хороший друг», А намекал, что Х плохой друг
??
The correct way to describe the ironic utterance would be something like (23a) and (23b): (23) a. b.
OK
Saying “Х is a fine friend” A spoke ironically / A meant the opposite. Говоря «Х- хороший друг», А иронизировал / A имел в виду противоположное.
OK
Grice wrote about the implicature of the irony type: “thе most obviously related proposition is the contrary one … [emphasis mine – I. K.]”(1975: 53). So the reason of the anomalousness of hint here is the violation of non-triviality condition. Meiosis and Hyperbole are not HINTS for the same reason: their calculation is trivial. Metaphoric utterances may or may not be HINTS depending on the degree to which the metaphor is «fanciful».
182 | Irina M. Kobozeva (Moscow)
All the other cases discussed in (Grice 1975) may successfully be reported with the help of the verb hint and its Russian equivalent. We have shown that several kinds of implicature discussed by H.P. Grice are not conceptualized as instances of a HINT in ordinary languages that have such a concept because this concept includes two additional conditions on the indirect utterance: justifiability of indirectness and non-triviality of inference. At the same time all the cases in our corpus of indirect speech acts conceptualized as HINTS satisfy the definition of conversational implicature or some other kind of non-conventional implicature. Cooperative Principle with its conversational maxims (as well as some social and ethical rules governing human communication) constitutes the basis for the step by step reconstruction of the process that results in taking a hint (or suspecting hints where there is none).
6 Conclusion In the paper we studied the semantic relations between the notion of conversational implicature introduced by P. Grice and the ordinary language concept HINT expressed as hint in English and namek / namekat’ in Russian. We introduced classification of hints and the definition of this concept which is a slightly revised version of the definition that was given in (Kobozeva and Laufer 1988) on the basis of the analysis of cases of hinting taken from literary dialogs. The definition comprises six conditions a communicative act must satisfy in order to be subsumed under the concept НINT: 1) intentionality; 2) use of linguistic means; 3) indirectness; 4) justifiability of indirectness; 5) calculability and 6) non-triviality. All kinds of hints can be treated as non-conventional implicatures, because their transmission and interpretation can be explained with the help of Cooperative Principle and its maxims and/ or some other general principle of communication (e.g. Politeness Principle). Оn the other hand, by testing the cases of conversational implicature from (Grice 1975) for the possibility to use the verb hint in reported speech constructions in such cases we demonstrated that some types of conversational implicatures are definitely not conceptualized as hints. It turned out that all these types do not satisfy the condition of non-triviality of hints. Consequently, HINT is a special kind of non-conventional IMPLICATURE and the translational equivalents hint and namek have all the rights to be included into the implicature family as ordinary language hyponyms of this pragmalinguistic term.
Russian namek and English hint as Hyponyms of Grice’s Term mplicature | 183
References Bach, Kent & Robert M. Harnish. 1979. Linguistic communication and speech acts. Cambridge: M.I.T. Press. Cheng, Winnie & Martin Warren. 2003. Indirectness, inexplicitness and vagueness made clearer. Pragmatics 13 (3), 381-400. Goffman, Erving. 1955. On Face-work: An Analysis of Ritual Elements of Social Interaction. Psychiatry: Journal for the Study of Interpersonal Processes 18 (3). 213–231. Grice, H. Paul. 1975. Logic and conversation. In Peter Cole and Jerry Morgan (eds.) Syntax and Semantics, vol.3, 41–58. New York: Academic Press. Grice, H. Paul. 1985. Logika i rečevoje obščenije [Logic and conversation]. In Elena Paducheva (ed.) Novoje v zarubežnoj lingvistike [New trends in foreign linguistics]. Moskva: Progress. Kobozeva Irina M. & Natalia I. Laufer. 1988. Ob odnom sposobe kosvennogo informirovanija [On one way of indirect informing]. Izvestija Akademii nauk SSSR. Serija literatury i jazyka. 47 (5). 462–471. Levinson, Stephen C. 1983. Pragmatics. Cambridge, England: Cambridge University. Padučeva, Elena V. 1985. Vyskazyvanie i ego sootnesennost’ s deistvitelnost’ju: Referencialnye aspecty semantiki mestoimenij [Utterance and its relation to reality: Referential aspects of semantics of pronouns]. Moskva: Nauka. Searle, John R. 1975. Indirect speech acts. In Peter Cole & Jerry L. Morgan (eds.) Syntax and Semantics, vol.3, 59–82. New York: Academic Press. Searle, John R.1976. A classification of illocutionary acts. Language in Society 5 (1). 1–23. Sew, Jih Wee. 1997. Power pragmatics in Asian languages. Language Sciences 19 (4). 357–367.
Kersten Krüger (Leipzig)
Ukrainische nominale Flexion – zur automatischen Generierung der Substantivformen in einem ukrainisch-deutschen Wörterbuch 1 Einführung Das elektronische ukrainisch-deutsche Wörterbuch, um das es hier geht1 und das u.a. zu allen flektierenden Worteinträgen das jeweilige Formenparadigma liefern soll, nutzt für die Generierung der Formen die algorithmische Grundlage des Russischen Universalwörterbuchs von B. Bendixen (2014), mit dem es auch anderweitig verknüpft ist. Ausgangspunkt für das auf das Ukrainische anzuwendende System der Wortformengenerierung ist hier der jeweilige Eintrag im Lexikon des Wörterbuchs in seiner graphematischen Form, aus der zunächst – soweit möglich – mittels autodetect-Algorithmen die Wortart und für Substantive der Stamm und die Endung der Nennform sowie das Genus ermittelt werden. In den Fällen, in denen sich die genannten Ausgangsgrößen nicht aus der Form gewinnen lassen, sind diese explizit anzugeben. Im Weiteren erfolgt auf der Grundlage von Genus, Stammauslaut und Endung bzw. Graphemfolgen am Wortende automatisch die Zuordnung zu einer definierten Flexionsklasse, welche die morphologischen Kasusendungen liefert. Die Notwendigkeit zur expliziten Angabe der Flexionsklasse ist eher die Ausnahme. In einem nächsten Schritt sind die morphologischen Kasusendungen in ihre jeweilige graphematische Form zu transformieren, wobei gegebenenfalls morphonologische Alternationen nach entsprechenden Regeln zu realisieren sind. Als letztes erfolgt die Betonung der Flexionsformen, die wiederum von einem zugeordneten Akzentschema vorgegeben wird, das nur bei Substantiven mit absolut unbeweglichem Akzent auf einund derselben Silbe des Stamms oder der Endung nicht explizit angegeben werden muss. Aus den beschriebenen Schritten zur Formengenerierung ergeben sich für die Anwendung auf ukrainische Substantive im Wesentlichen vier Erfordernisse
|| 1 Das Ukrainisch-deutsche Wörterbuch, herausgegeben von K. Krüger et al, wird voraussichtlich zunächst in einer Basisversion mit ca. 30.000 Lemmata im ersten Quartal 2018 im Harrassowitz Verlag erscheinen. Im Weiteren wird mittels Updates der Umfang auf ca. 250.000 Lemmata erweitert.
DOI 10.1515/9783110517873-009
Automatische Generierung der Flexionsformen der Substantive im Ukrainischen | 185
und Zielstellungen für den vorliegenden Beitrag. Im Einzelnen geht es – auch kontrastiv zum Russischen – um die Bestimmung und Beschreibung 1. der Entsprechungen zwischen graphematischer und morphologischer Ausdrucksebene, speziell im Bereich der substantivischen Flexion; 2. der morphologischen Formen der Substantivstämme und der morphonologischen Alternationen; 3. der Flexionsklassen und 4. der Akzentschemata ukrainischer Substantive. Die Entsprechungen zwischen graphematischer und morphologischer Ausdrucksebene sind sowohl für die Gewinnung der morphologischen Stammformen aus der graphischen Präsentation als auch für die Transformation der morphologischen Endungen in ihre jeweilige graphematische Notation relevant. Der folgende Abschnitt 2 ist dem zweiten Aspekt gewidmet. Der erste Aspekt wird im Zusammenhang mit den morphologischen Formen der Substantivstämme in Abschnitt 3 beleuchtet. Ausgehend von den in diesen beiden Abschnitten festgestellten Beziehungen, liefern die Abschnitte 4 und 5 eine vollständige, formenorientierte Beschreibung der Flexion ukrainischer Substantive. In Abschnitt 6 wird die Arbeitsweise der vorgeschlagenen Algorithmen anhand eines Beispiels demonstriert.
2 Morphologische Flexionsendungen und ihr graphematischer Ausdruck Die weiter unten vorgestellten Flexionsklassen ukrainischer Substantive geben die Endungen morphologisch-phonematisch vor, sodass diese zur Abbildung in graphematischer Notation unter Berücksichtigung des jeweiligen Stammauslauts zu transformieren sind, wozu es entsprechender Regeln bedarf. Hierbei lässt sich unterscheiden zwischen Regeln, die für die Sprache insgesamt gelten, und solchen, die nur im Rahmen der Substantivflexion bzw. für bestimmte Paradigmen Gültigkeit besitzen. Allgemeingültigkeit fürs Ukrainische haben in erster Linie die komplementären Regeln zum graphematischen Ausdruck von Härte und Weichheit paarig auftretender Konsonanten sowie zur Wiedergabe von /j/ durch die sogenannten jotierten Vokalbuchstaben. Ihre Anwendung auf die Flexion der Substantive, also auf die zu generierenden graphematischen Endungen in den Kasus obliqui nach konsonantisch hartem oder weichem Stammauslaut oder nach Vokal ergibt:
186 | Kersten Krüger (Leipzig)
(1a) -/a/, -/am/, -/amy/, -/ax/ nach hartem Konsonant → , , , 2 (1b) -/a/, -/am/, -/amy/, -/ax/ nach weichem Konsonant oder Vokal → , , , (2а) -/u/ nach hartem Konsonant → (2b) -/u/ nach weichem Konsonant oder Vokal → (3a) -/i/, -/iv/ nach hartem oder weichem Konsonant außer /j/ → , (3b) -/i/, -/іv/ nach /j/ oder Vokal → , Zu (3a) ist zu sagen, dass im Ukrainischen vor /i/ stets weiche Konsonantenlaute bzw. ggf. weiche Varianten gesprochen werden. Vor den Endungen und unterliegen stammauslautende harte Konsonanten somit immer der phonetischen Alternation. Für die Formenbildung auf morphologischer bzw. graphematischer Ausdrucksebene sind derartige Alternationen allerdings ohne Relevanz und können hier also vernachlässigt werden. Zum stammauslautenden weichen Konsonant /j/ ist anzumerken, dass dieser genau wie im Russischen im Falle einer leeren Endung nach Vokal durch wiedergegeben und ansonsten durch den jeweiligen jotierten Vokalbuchstaben mit ausgedrückt wird. Steht links von /j/ bzw. vom entsprechenden jotierten Vokalbuchstaben ein harter Konsonant, so wird dieser durch Apostroph vom folgenden /j/ bzw. jotierten Vokalbuchstaben getrennt, wie u.a. bei сім’я́ 'Familie'. Nach einem weichen Konsonanten erfolgt die Trennung von /j/ durch das Weichheitszeichen, so z.B. bei гери́лья 'Guerilla'. Neben den genannten allgemeingültigen Phonem-Graphem-Entsprechungen gelten für die graphematische Wiedergabe der substantivischen Flexionsendungen in den Kaus obliqui hinsichtlich des Stammauslauts außerdem folgende komplementäre Regeln: (4а) -/ɛ/, -/ɛvi/, -/ɛm/, -/ɛju/, -/ɛj/ nach hartem oder weichem Konsonant außer /j/ → , , , , (4b) -/ɛ/, -/ɛvi/, -/ɛm/, -/ɛju/ nach /j/ oder Vokal → , , ,
|| 2 Da die Entsprechungen zwischen Morphemen bzw. Phonemen und Graphemen ein zentrales Thema des Beitrags sind und die Transliteration der kyrillischen Grapheme das Auseinanderhalten der beiden Darstellungsebenen nur unnötig erschweren würde, werden hier und im gesamten Beitrag die kyrillischen Zeichen beibehalten und erscheinen außerhalb von Tabellen – sofern es sich nicht um ganze Wörter handelt – stets in spitzen Klammern.
Automatische Generierung der Flexionsformen der Substantive im Ukrainischen | 187
(5a) -/y/ nach hartem Konsonanten außer /ȝ/, /ʃ/, /ʧ/, [ʃt͡ʃ] bzw. , , oder → (5b) -/y/ nach /ȝ/, /ʃ/, /ʧ/ oder [ʃt͡ʃ] bzw. , , oder → (6а) -/o/, -/ovi/, -/om/, -/oju/ nach hartem Konsonant außer /ȝ/, /ʃ/, /ʧ/ oder [ʃt͡ʃ] bzw. , , oder → , , , (6b) -/o/, -/ovi/, -/om/, -/oju/ nach /ȝ/, /ʃ/, /ʧ/, [ʃt͡ʃ] bzw. , , , → , , , Aus den Regeln (4а) und (4b) geht hervor, dass graphematische Endungen mit im Bereich der Substantivflexion nach Konsonanten außer /j/ eigentlich überhaupt nicht auftreten, lediglich in adjektivisch flektierenden Substantiven wie z.B. in майбу́тнє 'Zukunft' kommen diese hier vor. Graphematisches wiederum steht für Härte unmittelbar vorangehender Konsonanten, sodass also weiche konsonantische Stammauslaute in dieser Position alternieren, was im Folgenden analog zu (3a) jedoch nicht weiter berücksichtigt wird. Zur Regel (4b) ist außerdem zu bemerken, dass analog zu (4a) diese auch in Bezug auf -/ɛj/ zu erwarten wäre, doch tritt die Endung -/ɛj/ niemals nach /j/ auf und im Übrigen kommt die Zeichenfolge auch sonst nicht im Wortinneren vor, sondern lediglich im Anlaut und auch das nur äußerst selten. In Bezug auf die Regeln sollte klar sein, dass sie nur die Zeichendistribution in den substantivischen Flexionsendungen in Abhängigkeit von bestimmten stammauslautenden Konsonanten vorgeben und nicht etwa allgemein für das Ukrainischen gelten, was z.B. жир 'Fett', ді́вчина 'Mädchen', ко́шик 'Korb' oder я́щик 'Kasten' genauso wie жолобо́к 'Rinne', чоло́ 'Stirn', шов 'Naht' oder щока́ 'Wange' belegen. In Entlehnungen, die wie z.B. ба́нджо 'Banjo' oder ле́чо 'Letscho' auf , , oder auslauten und die in der Regel indeklinabel sind, ist das auslautende nicht als Endung zu betrachten. Lediglich bei леда́що 'Faulpelz' tritt in der Nennform und folglich auch im Akkusativ Singular nach einem prädorsalem Konsonanten die graphematische Endung auf, die ansonsten in den Kasus obliqui nur als absolute Ausnahme vorkommt (s. Anmerkung zur folgenden Tabelle). Die Entsprechungen zwischen phonematischen Endungen und ihrem graphematischen Ausdruck lassen sich tabellarisch wie folgt zusammenfassen, wobei allerdings nur die produktiven Endungen aufgenommen wurden. Die Tabelle zeigt, dass die Endungen -/i/ und -/iv/ in ihren graphematischen Formen bzw. an harten wie an weichen Stammauslauten auftreten. Alle stammauslautenden harten Konsonanten unterliegen hierbei einer positionellen Palatalisierung. Im Weiteren ist nach hartem Stammauslaut sowohl auf phone-
188 | Kersten Krüger (Leipzig)
matischer als auch auf graphematischer Ebene zwischen -/i/ und -/y/ zu unterscheiden. Das gilt auch für die Endungen -/ɛ/ und -/o/, wobei allerdings vor -/ɛ/ bzw. stammauslautende weiche Konsonanten außer /j/ auf der phonetischen Ebene ihre Weichheit aufgeben und hart ausgesprochen werden. Tab. 1: Phonem-Graphem-Entsprechungen.
phonematische graphematische Endungen nach Stammauslaut auf: Endungen harten Konsonant ʃt͡ʃ oder ein- weichen Konsoaußer ȝ, ʃ, ʧ oder faches ȝ, ʃ oder ʧ nant außer /j/ ʃt͡ʃ bzw. ж, ш, ч, щ bzw. ж, ш oder ч3
Vokal oder /j/
a, am, amy, ax
а, ам, ами, ах
а, ам, ами, ах
я, ям, ями, ях
я, ям, ями, ях
і
і
і
i
ї
y
и
і
-
-
iv
ів
ів
ів
їв
ɛ
e
e4
e
є
5
o
o
-
-
-
ɛj
ей
ей
ей
-
ɛvi, ɛm, ɛju
-
еві, ем, eю
еві, ем, eю
єві, єм, єю
ovі, om, oju
ові, ом6, ою
-
-
-
ю
ю
u
у
у
7
|| 3 Die gedoppelten Konsonantenbuchstaben , , stehen für lange weiche Konsonanten, nach denen wie auch sonst nach weichen Konsonanten Vokale durch jotierte Vokalbuchstaben wiedergegeben werden, so z.B. in бездорі́жжя 'unbenutzbarer Straßenzustand', обли́ччя 'Antlitz' oder ту́шшю (Instrumental Singular zu туш 'Tusche'). Das gilt ohne Einschränkung allerdings nur für die Endungen der substantivischen Flexion und sagt also nichts über die generellen Distributionsverhältnisse im Ukrainischen aus, wenngleich und nach oder absolute Ausnahmeerscheinungen im ukrainischen Schriftbild sind. 4 Eine Ausnahme macht hier eigentlich nur der Vokativ Singular бри́жо (zu бри́жа 'Falte'). 5 Ausnahmen zeigen nur die Vokative auf zu den Substantiven, vgl. Fn. 6. 6 Lediglich bei einigen wenigen Substantiven mit hartem /r/ bzw. im Stammauslaut treten analog zu , , und aufgrund früherer Weichheit die graphematischen Endungen bzw. auf, so z.B. bei весля́р ' Ruderer', гусля́р 'Guslaspieler', пла́стир 'Pflaster' und школя́р 'Schüler', womit im Vokativ Sg. anstelle die graphematische Endung einhergeht. 7 Davon ausgenommen sind nur ganz wenige Fälle mit stammauslautendem im Instrumental Singular (Feminina der Klasse fw2), wo -/u/ als erscheint: z.B. bei не́хворощю (zu не́хворощ 'Feldbeifuß'), па́хощю (zu па́хощ 'Duft') oder при́горщю (zu при́гірщ 'Handvoll').
Automatische Generierung der Flexionsformen der Substantive im Ukrainischen | 189
Harte Konsonantenvarianten werden auch stets vor -/ɛj/ bzw. sowie vor /ɛvi/, -/ɛm/ und -/ɛju/ bzw. , und realisiert. Bei den Endungen -/ɛvi/ und -/ɛm/ sowie -/ovi/ und -/om/ ist im Übrigen auf graphematischer Ebene mit wenigen Ausnahmen eine komplementäre Verteilung gegeben, was deren phonematische Zusammenfassung ermöglichen und zur Vereinfachung der Flexionsklassen und Formengenerierung der m-Klassen-Substantive genutzt werden könnte.
3 Morphologische Stammformen und flexionsklassenübergreifende Alternationen Der für die Formenbildung primär anzusetzende Stamm ist bei einem Großteil der Substantive mit der so genannten Nennform, also mit der Form des Nominativs Singular identisch oder er lässt sich aus dieser Form gewinnen, indem die in einer entsprechenden Flexionsklasse für den Nominativ Singular ausgewiesene Endung abgetrennt wird. Das ist z.B. bei абажу́р 'Lampenschirm' mit Nullendung bzw. bei ка́ша 'Buchweizengrütze' oder го́рло 'Kehle' der Fall, bei denen sich von den Stämmen абажу́р-, ка́ш- und го́рл- durch Anhängen der entsprechenden Endungen problemlos alle Flexionsformen bilden lassen. Allerdings lässt sich aus der Nennform – vor allem aufgrund der fürs Ukrainische charakteristischen historischen Auslautverhärtung bei Labialen und bei -/rᶨ/ – oft nicht zuverlässig auf paradigmatische Härte oder Weichheit des Stammauslautes schließen, wie z.B. верф 'Werft', кров 'Blut', лі́кар 'Arzt' oder мо́ре 'Meer' belegen, die allesamt aufgrund der Endungen bzw. Formen in den Kasus obliqui den Flexionsklassen mit weichem Stammauslaut zuzuordnen sind. Insofern wird zur Feststellung von Härte oder Weichheit des Stammauslautes – der übrigens genau wie im Russischen immer konsonantisch ist – eine Kontroll- oder Referenzform benötigt. Hierzu eignet sich die Dativ- oder Lokativform des Plurals genauso wie die Form des Genitivs Singular und mit gewissen Einschränkungen auch die des Instrumentals Plural. In anderen Fällen wiederum werden Stämme, die auf bestimmte Konsonantencluster enden, in den Formen mit Nullendung jeweils durch einen so genannten flüchtigen oder fluktuierenden Vokal modifiziert bzw. erweitert. Hier ist für die Generierung der Flexionsformen nicht von den tatsächlichen Stämmen auszugehen, die mit der jeweiligen Nennform identisch sind bzw. sich aus deren graphematischer Form ableiten lassen, sondern von entsprechend modifizierten Stämmen, die im Weiteren „zugrundeliegende Stämme“ genannt werden sollen.
190 | Kersten Krüger (Leipzig)
Zaliznjak (2002: 246–252) spricht in diesem Zusammenhang von bedingten Formen, was sich aber ohne weitere Erklärungen, für die hier nicht der Platz ist, nicht so einfach erschließt. Fluktuierende Vokale bzw. die Positionen fluktuierender Vokale in den zugrundeliegenden Stämmen werden im Folgenden durch das Rautenzeichen repräsentiert. Somit steht in einer zugrundeliegenden Stammform die Raute für einen Vokal, der – als Reflex eines geschwundenen Jers – nur vor leerer Endung erscheint. Im Falle von голубо́к 'Täubchen' mit den Formen голубка́, голубко́ві/ голубку́ usw. ist also von einem Stamm golub#k- auszugehen. Analog dazu erfolgt die Bildung aller Formen zu ба́бка 'Großmutter' und zu зло 'Übel' von den zugrundeliegenden Stämmen bab#k- und z#ł-, wobei hier nur im Genitiv Plural anstelle # der Vokal /o/ bzw. erscheint. In anderen Fällen ist die Rauten-Position nicht durch /o/, sondern durch /ɛ/ bzw. graphematisch durch oder zu füllen, wie z.B. bei весло́ 'Ruder', день 'Tag' oder кита́єць 'Chinese' mit den zugrundeliegenden Stämmen ves#ł-, d#nᶨ- und kitaj#cᶨ- und der tatsächlichen Form des Genitiv Plurals ве́сeл bzw. den Nennformen день und кита́єць. Ob in einem zugrundliegenden Stamm die Vokalisierung einer RautenPosition bei Nullendung durch /o/ oder durch /ɛ/ – andere Vokale kommen bis auf die im Folgenden unter (8) angeführte Ausnahme nicht vor – vorzunehmen ist, muss für den Nominativ Singular nicht gesondert angegeben werden, da diese Form ja im Lexikon gegeben ist. Hier ist es ausreichend zu vermerken, dass für die Bildung aller übrigen Formen der letzte Stammvokal der Nennform entfällt. Dagegen ist bei Stämmen mit Raute für die Formen des endungslosen Genitivs Plural stets der fluktuierende Vokal vorzugeben. Das kann in den meisten Fällen durch mehr oder weniger konsequent bzw. unbeschränkt oder mit Beschränkung auf einzelne Flexionsklassen geltende Regeln geschehen, nach denen der einzusetzende Vokal in Abhängigkeit von der linken lautlichen Umgebung und ggf. auch unter Berücksichtigung der Betonung bestimmt wird. In einigen wenigen verbleibenden Fällen muss der Vokal allerdings konkret angegeben werden. Ohne Einschränkungen oder nur mit wenigen Ausnahmen gelten für den endungslosen Genitiv Plural folgende Vokalisierungsregeln: (1) -/#k/- → (2) -/#m/ → ( tritt nur in zwei Fällen nach auf: тюрма́ 'Gefängnis' und ярмо́ 'Joch') (3) -/#lʲ/- → (4) -/#ł/- → – mit Ausnahme von зол (zu зло 'Übel') und сте́кол (zu скло 'Glas') (5) -/#r/- → – mit Ausnahme von і́гор (zu гра 'Spiel') und і́скор (zu і́скра 'Funke')
Automatische Generierung der Flexionsformen der Substantive im Ukrainischen | 191
(6) -/#͡tsᶨ/- → , wobei ggf. ein vor entfällt und vor in aufgeht: z.B. bei кі́лець (zu кільце́ 'Ring') und яє́ць (zu яйце́ 'Ei') (7) In Stämmen mit fluktuierendem Vokal auf -/#n/- oder -/#nᶨ/- schließt der endungslose Genitiv Plural in der Regel auf bzw. , es sei denn, vor dem fluktuierendem Vokal befindet sich ein Velar. In diesem Falle erscheint vor hartem /n/ oder weichem /nʲ/ der Vokal , so z.B. in ба́гон (багно́ 'Sumpf'), ві́кон (вікно́ 'Fenster'), воло́кон (волокно́ 'Faser'), па́хон (пахно́ 'Geruch'), сто́гон (сто́гна 'Boulevard'), су́кон (сукно́ 'Tuch') bzw. ку́хонь (ку́хня 'Küche'), пове́рхонь (пове́рхня 'Oberfläche') und су́конь (су́кня 'Jackenkleid'). Schwankungen zwischen und zeigen sich z.B. bei стегно́ 'Lende' und сосна́ 'Kiefer'. (8) In einer überschaubaren Anzahl von ca. 15 Substantiven erfolgt die Vokalisierung einer Rauten-Position nach /j/ bzw. oder Apostroph durch , z.B. in гери́лей (гери́лья 'Guerilla') oder сіме́й (сім’я́ 'Familie'), oder ausnahmsweise auch durch , und zwar nur dann, wenn vor /j/ ein /nʲ/ steht: до́ній (до́нья 'Töchterchen') oder піра́ній (піра́нья 'Piranha'). Zu dieser Art von Stammmodifikationen durch fluktuierende Vokale, die auch der russischen Nominalflexion eigen sind, kommen im Ukrainischen zwei weitere, dem Russischen fremde Vokalalternationen hinzu, die ebenfalls den letzten Stammvokal betreffen. Gemeint sind hier die sich im Schriftbild zeigenden Alternationen // (mit dem Sonderfall //) sowie //, die insgesamt bei zahlreichen Substantiven in bestimmten Kasus auftreten, wobei die Alternation // deutlich dominiert. Welche Substantive eine solche Alternation aufweisen, lässt sich nur zum Teil an deren Form festmachen und in entsprechenden Regeln erfassen. So zeigen z.B. alle Substantive mit dem Suffix in ihren Paradigmen die Alternation //. In den übrigen und meisten Fällen oder auch generell ist in den betreffenden Ausgangsstämmen nicht von einem Phonem /o/ oder /ɛ/, sondern von einem Hyperphonem /oi/ bzw. /ɛi/ auszugehen. Für die Hyperphoneme /oi/ bzw. /ɛi/ gilt im Weiteren, dass sie in endungslosen Kasus, also in der letzten Silbe einer Wortform – bedingt könnte man hier auch von geschlossenen Silben sprechen – als /i/ und ansonsten als /o/ bzw. oder oder als /ɛ/ bzw. realisiert werden. Vokalisch leere Endungen treten in der Flexion ukrainischer Substantive nur im Nominativ Singular von Maskulina und Feminina sowie im Genitiv Plural von Feminina und Neutra auf: z.B. in баті́г 'Peitsche' – батога́, батогу́ oder батого́ві¸ баті́г, батого́м usw.; папі́р 'Papier' – папе́ру, папе́ру oder папе́рові usw. Nach weichem /lʲ/, also wie z.B. bei výlʲoitoder kо́lʲoir-, erscheint das Hyperphonem in geschlossener Silbe als /i/ bzw. – ви́літ 'Abflug' oder ко́лір 'Farbe' – und ansonsten als /o/ bzw. als Zeichenfolge
192 | Kersten Krüger (Leipzig)
: ви́льоту und ко́льору. Beispiele für die genannten Vokalalternationen im endungslosen Genitiv Plural sind: вдів (zu вдова́ 'Witwe'), слів (zu сло́во 'Wort'), сліз (zu сльоза́ 'Träne'), бері́з (zu бере́за 'Birke') oder сіл (zu село́ 'Dorf'). Doch reiht sich auch der Instrumental Singular einiger Feminina, die im Nominativ Singular eine Nullendung haben, in die Formen mit /i/ anstelle /o/ oder /ɛ/ ein: ва́ртість 'Preis'– ва́ртості, …, ва́ртістю, ва́ртості usw.; піч 'Ofen'– пе́чі, …, пі́ччю, пе́чі usw. Die Vokalalternation im Instrumental Singular wie auch die Konsonantendoppelung bei піч u.a. ist ein Charakteristikum der Flexionsklasse fw2, worauf weiter unten noch näher einzugehen sein wird. Bekanntlich ist die ukrainische Substantivflexion generell auch durch Alternationen der Velare geprägt. Diese Lautwechsel sind mit der so genannten zweiten regressiven Palatalisation der Velare vor Vokalen der vorderen Reihe entstanden und haben sich hier – anders als im Russischen – erhalten. So alternieren vor der Endung /i/, die im Dativ und Lokativ Singular auftritt, die stammauslautenden Velare /g/, /k/ und /x/ sowie der Pharyngal /ɦ/, der im Schriftbild durch wiedergegeben wird, mit den weichen Dentalen bzw. Präpalatalen /zᶨ/, /t͡sᶨ/ und /sᶨ/, was sich auf der graphematischen Ebene so darstellt: , //; //; //. Schließlich ist an dieser Stelle noch auf die regelmäßigen phonetischen Alternationen des Stammauslautes von Härte zu Weichheit vor den Endungen -/i/ und -/iv/ zu verweisen und von Weichheit zu Härte vor -/ɛ/, -/ɛj/, -/ɛvi/, -/ɛm/ und -/ɛju/, die bereits im vorangehendem Kapitel angesprochen wurden.
4 Flexionsklassen und klassenspezifische Stammalternationen Das Flexionssystem ukrainischer Substantive ist historisch bedingt dem der russischen Substantive sehr ähnlich. Im Wesentlichen geht es nur in zwei Punkten über dieses hinaus. Zum einen gibt es im Ukrainischen einen Vokativ, der allerdings im Plural stets und im Singular in etlichen Flexionsklassen mit der entsprechenden Nominativform zusammenfällt. Zum anderen ist die Anzahl der Substantivendungen umfangreicher als im Russischen und die Endungen selbst erfordern bereits auf der morphonologischen Ebene eine konsequente Unterscheidung zwischen hartem und weichem Stammauslaut. Die Flexionsklassen sind als Mengen von Substantiven zu verstehen, die lediglich im Hinblick auf die phonematischen Formen der Kasus-Endungen homogen sind. Substantive einer Klasse können also durchaus unterschiedliches Genus haben, doch überwiegt in
Automatische Generierung der Flexionsformen der Substantive im Ukrainischen | 193
jeder Klasse jeweils eines der drei Genera deutlich, worauf der erste Buchstabe der hier verwendeten Abkürzungen zur Bezeichnung der Klassen verweist. So stehen also fh1 und fh2 für zwei Klassen überwiegend femininer Substantive mit hartem Stammauslaut, wohingegen die beiden Kürzel fw1 und fw2 Klassen überwiegend femininer Substantive mit weichem Stammauslaut und den angegebenen phonematischen Endungen bezeichnen. Analog dazu gibt es m-Klassen und n-Klassen, die primär maskuline bzw. neutrale Substantive umfassen. Wie im Russischen haben auch alle ukrainischen Substantive produktiver Flexionsklassen im Dativ, Instrumental und Lokativ Plural identische Endungen. Daher werden im Folgenden diese lediglich für die Substantive der f-Klassen aufgeführt. Bei den übrigen Klassen werden die Pluralformen nur bis zum Dativ, der Kontrollform für harten oder weichen Stammauslaut angegeben. Im Akkusativ Plural wird wie im Russischen zwischen grammatisch belebten und unbelebten Substantiven unterschieden. Grammatisch unbelebte Substantive, zu denen auch Bezeichnungen für Pflanzen gehören, fallen hier formal mit dem Nominativ zusammen, Personenbezeichnungen – mit dem Genitiv. Allerdings können Bezeichnungen für Tiere im Akkusativ Plural die Form des Genitivs oder des Nominativs Plural aufweisen. Panzer (1999, S. 55) merkt hierzu an, dass es im Plural nur eine Personalkategorie und keine Beseeltheitskategorie gibt, insofern “nur Bezeichnungen von Personen die A=G-Endung haben, während Bezeichnungen von Tieren und anderen Lebewesen im Gebrauch schwanken“.
4.1 Die f-Klassen Bei den f-Klassen kann über das Genus, den Stammauslaut und die Endung in der Nennform problemlos eine automatische Zuordnung der Substantive vorgenommen werden. Von den vier Klassen umfasst die Klasse fh2 die mit Abstand wenigsten Substantive, die im Übrigen allesamt auf auslauten und zwischen femininem und neutralem Genus und entsprechenden Flexionsendungen schwanken. Bei dieser Klasse sowie bei der Klasse fw1 treten keinerlei besondere Stammalternationen auf, in den anderen beiden Klassen dagegen regelmäßig, und zwar in Klasse fh1 die bereits erwähnte Alternation der Velare einschließlich /ɦ/ bzw. vor /i/ und in Klasse fw2: (a) der Wechsel о//і bzw. ɛ//і auch im Instrumental Singular bei eigentlich offener Silbe sowie
194 | Kersten Krüger (Leipzig)
(b) Konsonantenlängung bzw. -doppelung und damit einhergehender Wechsel von hartem zu weichem Stammauslaut im Instrumental Singular bei Stammende auf Vokal + , , , , , oder , sowie . Tab. 2: Singular.
Klassen
fh1
fh2
fw1
fw2
Kasus
Endg.
'Birke'
Endg.
'Oma'
Endg. 'Banja'
Endg.
'Hafen'
Nom.
a
бере́за
ɛ
баби́ще
a
ба́ня
–
га́вань
Gen.
y
бере́зи
Dat.
i
бере́зі
y
баби́щі
i
ба́ні
i
га́вані
i
баби́щі
i
ба́ні
i
га́вані
Akk.
u
бере́зу
ɛ
баби́ще
Instr.
oju
бере́зою
ɛju
баби́щею
u
ба́ню
–
га́вань
ɛju
ба́нею
(j)8u
га́ванню
Lok.
i
бере́зі
i
баби́щі
i
ба́ні
i
га́вані
Vok.
o
бере́зо
ɛ
баби́ще
ɛ
ба́не
ɛ
га́ване
Tab. 3: Plural.
Nom.
y
бере́зи
y
баби́щі
Gen.
–
Dat.
am
бері́з
–
баби́щ
бере́зам
am
баби́щам
i
ба́ні
i
га́вані
–
бань
ɛj
га́ваней
am
ба́ням
am
га́ваням га́вані
Akk.
y/–
бере́зи
i/–
баби́щ
i / – ба́ні
i / ɛj
Instr.
amy
бере́зами
amy
баби́щами
amy
ба́нями
amy
га́ванями
Lok.
ax
бере́зах
ax
баби́щах
ax
ба́нях
ax
га́ванях
Vok.
y
бере́зи
y
баби́щі
i
ба́ні
i
га́вані
Die Vokalalternation о//і geht, wie oben bereits festgestellt wurde, eigentlich mit der Position des Vokals in offener und geschlossener Silbe einher, was aber mit der Form des Instrumentals Singular von Substantiven der Klasse fw2 mit dem Hyperphonem /oi/ bzw. /ɛi/ nicht völlig vereinbar ist. Bei diesen erscheint wie im endungslosen Nominativ Singular auch im Instrumental Singular regelmäßig /i/, obwohl die Silbe eigentlich als offene zu betrachten ist: vgl. z.B. бі́дність 'Armut'
|| 8 Die Endungsvariante mit Phonem /j/ tritt nur nach den harten Labialen /b/, /v/, /m/, /p/, /r/ und auf und wird im Schriftbild durch ausgedrückt.
Automatische Generierung der Flexionsformen der Substantive im Ukrainischen | 195
– бі́дності, бі́дності, бі́дність, бі́дністю, бі́дності, бі́дностe; міць 'Stärke' – мо́ці, …, мі́ццю; те́мінь 'Dunkelheit' – те́мені, …, те́мінню; сіль 'Salz' – со́лі, …, сі́ллю; ніч 'Nacht' – но́чі, …, ні́ччю. Das letzte Beispiel zeigt zugleich die Alternation zwischen hartem, weichem und gelängtem weichen konsonantischen Stammauslaut. Nur wenige Substantive der Klasse fh1 wie z.B. ба́ба 'Oma' bzw. праба́ба 'Uroma', губа́ 'Lippe' und ма́ма 'Mama' zeigen Schwankungen im Genitiv Plural zwischen Nullendung und Endung , welche hier die Maskulina гайдама́ка ' Haidamake' und ста́роста 'Ältester' wie auch Maskulina der Klasse fw1 (z.B. суддя́ 'Richter' und те́сля 'Zimmermann') ausschließlich haben. Einige weitere Substantive der Klassen fh1 und fw1 weisen einen unregelmäßigen Genitiv Plural auf auf, darunter z.B. дити́на bzw. ді́ти 'Kinder', люди́на bzw. лю́ди 'Leute' und ми́ша 'Maus' sowie баддя 'Kübel', бро́ня 'Reservierung', куде́ля 'Werg', парасо́ля 'Schirm', пеня́ 'Hindernis', свиня́ 'Schwein', смаже́ня 'Rührei', стаття́ 'Artikel', тля 'Blattlaus', wobei Geminaten vor der Endung vereinfacht werden, was z.B. баде́й und стате́й zeigen. Außerdem zeigen alle Substantive der Klasse fw1 mit stammauslautendem weichen /rᶨ/ im endungslosen Genitiv Plural Auslautverhärtung: бу́ря 'Sturm' – бур, вече́ря 'Abendbrot' – вече́р u.a. Schließlich hat ро́зкіш 'Vergnügen' als Ausnahme in der Klasse fw2 im Genitiv Plural nicht , sondern , was auch auf das völlig isolierte ма́ти 'Mutter' zutrifft, das mit der Stammerweiterung in allen Kasus obliqui bzw. im Akkusativ Singular nach Danylenko & Vakulenko (1995: 22) quasi seine eigene Flexionsklasse bildet.
4.2 Die m-Klassen Bei den m-Klassen ist wiederum zwischen hartem und weichem Stammauslaut zu trennen, doch unterscheiden sich die Flexionsklassen der überwiegend maskulinen Substantive nicht so sehr durch divergierende Endungen bei hartem und weichem Stammauslaut als vielmehr durch die Formen im Genitiv, Lokativ und Vokativ Singular voneinander. In den genannten Kasus treten im Prinzip jeweils zwei Endungen auf, und zwar -/a/ und -/u/ im Genitiv9, -/i/ und -/u/ im Lokativ sowie -/ɛ/ und -/u/ im Vokativ. Die Endungen kommen bei Substantiven mit har-
|| 9 Danylenko & Vakulenko (1995: 18–20) gehen hier von zwei Kasus aus, und zwar bei -/a/ vom Genitiv und bei -/u/ vom Partitiv, wobei aber in Abhängigkeit von der Bedeutung jeweils nur einer von beiden gebildet wird.
196 | Kersten Krüger (Leipzig)
tem Stammauslaut in allen möglichen Kombinationen vor, woraus sich acht Flexionsklassen ergeben, wohingegen es bei denen mit weichem Stammauslaut nur drei sind. Die Klassen werden in den folgenden Tabellen jeweils durch Nennung der Endungen in den signifikanten Kasus bezeichnet. Die Bezeichnung mhaie steht somit für maskulin – harter Stammauslaut: -/a/ (im Genitiv), -/i/ (im Lokativ) und /ɛ/ bzw. im Vokativ; die Klasse mwuuu umfasst hingegen überwiegend maskuline Substantive mit weichem Stammauslaut und mit der Endung /u/ in den drei signifikanten Kasus. Die genannten Endungen sowie die Endungen -/ovi/ bzw. -/ɛvi/ konkurrieren z. T. auch innerhalb der Flexionsklassen miteinander, und zwar im Dativ und Lokativ Singular. Allerdings kommt -/ovi/ bzw. -/ɛvi/ immer nur als zusätzliche Variante neben einer der anderen oder beiden anderen Endungen vor, sodass diese Endungen für die Unterscheidung der Flexionsklassen sekundär sind und hierbei vernachlässigt werden können. Dennoch ist die Endungsvariante -/ovi/ bzw. -/ɛvi/ für den Dativ und ggf. auch für den Lokativ Singular konsequent zu berücksichtigen, da diese hier, wie auch Danylenko & Vakulenko (1995: 16) für den Dativ Singular von Maskulina feststellen, bei belebten Substantiven in der Regel vorrangig ist. Daneben gilt für den Ausdruck von Belebtheit bei Maskulina dasselbe wie im Russischen, also der Zusammenfall des Akkusativs mit dem Genitiv Singular, wohingegen bei unbelebten Substantiven der Akkusativ dieselbe Endung wie der Nominativ hat. Wenn man zunächst von den stammauslautenden Konsonanten und der Endung im Vokativ ausgeht, gestaltet sich die Aufteilung der Maskulina mit hartem Stammauslaut auf Flexionsklassen wie folgt: Tab. 4: Harter Stammauslaut 1, Singular.
Klasse Kasus
mhaie
'Pilz'
mhaue
'Posten'
mhuie
'Beweis'
mhuue
'Ordnung'
Nom.
–
гриб
–
пост
–
до́каз
–
лад
Gen.
a
гриба́
a
поста́
u
до́казу
u
ла́ду
Dat.
ovi/u10
грибу́/ -о́ві
ovi/u
посту́/-о́ві
ovi/u
до́казу/ -ові
ovi/u
ла́ду/ -ові
|| 10 Die Reihenfolge der angegebenen phonematischen Varianten ist rein alphabetisch. Bei den Beispielen steht dagegen zuerst die häufigere Variante.
Automatische Generierung der Flexionsformen der Substantive im Ukrainischen | 197
Akk.
N./G.
гриб
N./G.
пост
Instr. Lok. Vok.
N./G.
до́каз
N./G.
лад
om
грибо́м
om
посто́м
om
i
грибі́
u
посту́
i
до́казом
om/
ла́дом
до́казі
u
ладу́
ɛ
гри́бе
ɛ
по́сте
ɛ
до́казе
ɛ
ла́де
Tab. 5: Harter Stammauslaut 1, Plural.
Nom.
y
гриби́
y
пости́
y
до́кази
y
лади́
Gen.
iv
грибі́в
iv
пості́в
iv
до́казів
iv
ладі́в
Dat.
am
гриба́м
am
поста́м
am
до́казам
am
лада́м
Tab. 6: Harter Stammauslaut 2, Singular.
Klasse Kasus
mhaiu
'Ufer'
mhauu
'Sünde', 'Vater'
mhuiu
'Bewegung'
mhuuu
'Schrei'
Nom.
–
бе́рег
–/o
гріх, ба́тько
–
рух
–
крик
Gen.
a
бе́рега
a
гріха́, ба́тькa u
ру́ху
u
кри́ку
Dat.
ovi/u
бе́регу/ -ові
ovi/u
гріху́/-о́ві ба́тькові /-у
ovi/u
ру́ху/ -ові
ovi/u
кри́ку/ -ові
Akk.
N./G.
бе́рег
N./G.
гріх, ба́тькa
N./G.
рух
N./G.
крик
Instr.
om
бе́регoм
om
гріхо́м, ба́тьком
om
ру́хом
om
кри́кoм
Lok.
i
бе́резі
u
гріху́, ба́тьку/-кові
i
ру́сі
u
кри́ку
Vok.
u
бе́регу
u
грі́ху, ба́тьку u
ру́ху
u
кри́ку
Tab. 7: Harter Stammauslaut 2, Plural.
Nom.
y
береги́
Gen.
iv
берегі́в
Dat.
am
берега́м
y
гріхи́, батьки́
y
iv
гріхі́в, батькі́в
iv
am
гріха́м, батька́м am
ру́хи
y
кри́ки
ру́хів
iv
кри́ків
ру́хам
am
кри́кам
Wie sich an den Beispielen bereits erkennen lässt, sind den vier Flexionsklassen mit -/u/ im Vokativ vorwiegend Maskulina zugeordnet, deren Stamm auf einen Velar bzw. den Pharyngal -/ɦ/ ausgeht. Daneben folgen diesen Mustern
198 | Kersten Krüger (Leipzig) hauptsächlich noch Stämme auf die Präpalatale -/ȝ/, -/ʃ/, -/ʧ/ oder -[ʃt͡ʃ] bzw. , , oder sowie in Klasse mhauu einige Maskulina auf -/ko/. Anders auslautende Stämme sind in diesen Klassen die Ausnahme. Demgegenüber sind die Flexionsklassen mit Vokativ auf -/ɛ/ in Bezug auf den Stammauslaut recht heterogen, doch Stämme auf Velar bzw. Pharyngal oder einen der Präpalatale kommen bis auf wenige Ausnahmen überhaupt nur in Klasse mhaue vor, und zwar auch nur solche auf -/k/ und -/ȝ/ und vereinzelt auf -/ɦ/ und -/х/. Somit erfolgt die Aufteilung vorwiegend maskuliner Substantive mit hartem Stammauslaut auf die angegebenen Klassen mit einigen Ausnahmen bei Klasse mhaue danach, ob ihr Stamm auf einen der durch , , , , , , oder bezeichneten Konsonanten endet oder nicht. Auslautendes tritt übrigens nur bei зигза́г 'Zickack' auf. Bei den Stämmen auf Velar bzw. Pharyngal der Flexionsklasse mhaue mit Vokativ auf -/ɛ/ kommt es zu den Konsonantenalternationen //, // sowie //. Belege hierfür sind z.B. во́роже (zu во́рог 'Feind'), риба́че (zu риба́к 'Fischer'), проро́че (zu проро́к 'Prophet') und пасту́ше (zu пасту́х 'Hirte'). Bezüglich der Endungsverteilung im Genitiv und Lokativ Singular lässt sich keine konsequente Korrelation mit bestimmten Stammauslauten bzw. Konsonantengruppen feststellen, obgleich sich Stämme auf einen Velar bzw. Pharyngal und insbesondere auf -/k/ im Lokativ wesentlich häufiger mit der Endung -/u/ als mit -/i/ verbinden. Bei Stämmen, die auf -/ȝ/, -/ʃ/, -/ʧ/ oder -[ʃt͡ʃ] bzw. , , oder enden, ist wiederum mehr oder weniger regelmäßig die eine wie die andere Endung gebräuchlich und zulässig, sodass in diesen Fällen eine Zuordnung zu zwei Flexionsklassen erfolgt, was z.B. auf кря́ж 'Gebirgskette' – Lok. Sg.: кря́жі (mhaie)/ кря́жу (mhaue); м’яч 'Ball' – Lok. Sg.: м’ячі́ (mhaiu)/ м’ячу́ (mhauu) oder ви́граш 'Gewinn' – Lok. Sg.: ви́граші (mhuiu)/ ви́грашу (mhuuu) zutrifft. Bei allen belebten, aber auch bei etlichen unbelebten Substantiven kommt in der Regel als dritte Variante die Endung -/ɛvi/ hinzu, die zugleich eine regelmäßige Variante im Dativ Singular ist: чита́ч 'Leser' – Lok. Sg.: читачі́/ читачу́/ читаче́ві; ніж 'Messer' – Lok. Sg.: ножі́/ ножу́/ ноже́ві. Im Falle von ніж kommt eine weitere Variante im Akkusativ Singular (ніж/ ножа́) hinzu, die auch andere unbelebte Substantiven zeigen, obgleich die mit dem Genitiv zusammenfallende Form eigentlich für Belebtheit steht. Eine weitere, eigene Flexionsklasse erfordern die maskulinen Substantive auf harten Stammauslaut und mit der Singularstammerweiterung wie z.B. датча́нин 'Däne' oder росія́нин 'Russe', für die sich die Bezeichnung mhin anbietet. Die Substantive der mhin-Klasse folgen im Singular der Klasse mhaie und im Plural der Klasse fh1, das heißt, sie unterscheiden sich von den Formen der
Automatische Generierung der Flexionsformen der Substantive im Ukrainischen | 199
Klasse mhaie lediglich durch die Nullendung im Genitiv Plural. Zudem ist der Akkusativ Singular hier stets mit der Form des Genitivs identisch, da alle Substantive dieser Klasse Personen bezeichnen. Tab. 8: Weicher Stammauslaut, Singular.
Klasse Kasus
mwaie/ mwaue
'Schneider'
mwaiu/ mwauu
'Tag'
mwuiu/ mwuuu
'Leid'
Nom.
–
краве́ць
–
день
–
жаль
Gen.
a
кравця́
a
дня
u
жа́лю
Dat.
ɛvi/u
кравце́ві/ кравцю́
ɛvi/u
дню/ дне́ві
ɛvi/u
жа́лю/ жа́леві
Akk.
N./G.
кравця́
N./G.
день
N./G.
жаль
Instr.
ɛm
кравце́м
ɛm
днем
ɛm
жа́лем
Lok.
i/ u
кравце́ві/ кравці́/ кравцю́
i/ u
дні / дню
i/ u
жа́лі/ жа́лю
Vok.
ɛ
кра́вче
u
дню
u
жа́лю
y
жалі́
Tab. 9: Weicher Stammauslaut, Plural.
Nom.
y
кравці́
y
дні
Gen.
iv
кравці́в
iv
днів
iv
жалі́в
Dat.
am
кравця́м
am
дням
am
жаля́м
Eine Besonderheit der weichen Flexionsklassen sind die durchgängig gegebenen Varianten nicht nur im Dativ, sondern auch im Lokativ Singular. Von diesen kommt -/ɛvi/, wie schon das Pendant -/ovi/ in den Klassen mit hartem Stammauslaut, in der Regel nur bei belebten Substantiven vor. Doch in Bezug auf die Endungen -/i/ und -/u/ gibt es innerhalb der hier unterschiedenen sechs Klassen keine oder nur wenige Substantive, wie z.B. бій 'Kampf' (mhuuu) – бою́, die ausnahmsweise nur die eine oder die andere Endung zulassen. Insofern ist es begründet, alle Substantive mit Endungsvarianten im Lokativ jeweils zwei Klassen zuzuordnen oder aber die Klassen paarweise zusammenzufassen und im Lokativ zwei Endungen vorzugeben. Im zweiten Fall müssen dann natürlich Lemmata wie бій von der automatischen Formengenerierung ausgenommen werden, was am einfachsten durch manuelle Vorgabe der einen normativen Endung ge-
200 | Kersten Krüger (Leipzig)
schieht. Ein solches Vorgehen ist freilich auch noch in anderen Fällen erforderlich, so z.B. bei друг 'Freund', wo für den Singular die Endungen der Klasse mhaue zu übernehmen sind, im Plural jedoch aufgrund des Konsonantenwechsels im Stammauslaut von /g/ zu /zᶨ/ – die Endungen zum weichen Stammauslaut: дру́зі, дру́зів, дру́зям. Für die Klassen mwaie bzw. mwaue ist schließlich noch bemerkenswert, dass im Vokativ Singular vor der Endung -/ɛ/ weiches /t͡sᶨ/ und hartes /ʧ/ bzw. // alternieren
4.3 Die n-Klassen Tab. 10: Harter Stammauslaut, Singular.
Klasse Kasus
nh1
'Eimer'
nh2
'Armee'
nh3
'Erscheinung'
nh4
'Gänse-küken'
Nom.
o
відро́
o
ві́йсько
ɛ
я́вище
–
гусеня́
Gen.
a
відра́
a
ві́йська
a
я́вища
t-y
гусеня́ти
Dat.
u
відру́
ovi/u
ві́йську/-ові u
я́вищу
t-i
гусеня́ті
Akk.
o
відро́
o
ві́йсько
ɛ
я́вище
–
гусеня́
Instr.
om
відро́м
om
ві́йськом
ɛm
я́вищем
m
гусеня́м
Lok.
i
відрі́
u
ві́йську
i/ u
я́вищі / -у
t-i
гусеня́ті
Vok.
o
відро́
o
ві́йсько
ɛ
я́вище
–
гусеня́
Tab. 11: Harter Stammauslaut, Plural.
Nom.
a
ві́дра
a
війська́
a
я́вища
t-a
гусеня́та
Gen.
-
ві́дер
–
військ
–
я́вищ
t- –
гусеня́т
Dat.
am
ві́драм
am
війська́м
am
я́вищам
t-am
гусеня́там
Die Aufteilung überwiegend neutraler Substantive mit hartem Stammauslaut auf die ersten drei nh-Klassen bedarf sicher nur weniger Erklärungen. Die Klasse nh2 umfasst im Prinzip alle Neutra auf und , die Klasse nh1 – eigentlich alle restlichen mit der Endung -/o/ bzw. nach beliebigem harten Stammauslaut. Nur die Handvoll Neutra mit Stamm auf -/x/ verhalten sich unterschiedlich und schwanken zwischen diesen beiden Klassen. Die Endungen für den Dativ kommen hier von der Klasse nh2 und diese werden entweder auch für den Lokativ
Automatische Generierung der Flexionsformen der Substantive im Ukrainischen | 201
übernommen oder der Lokativ endet wie z.B. bei ву́хо 'Ohr' - ву́сі auf -/i/ und es kommt zur bekannten Alternation von /x/ und /s/ bzw. //. Ansonsten sind Konsonantenalternationen bei diesen Klassen die absolute Ausnahme, so wie bspw. im Falle von о́ко 'Auge' mit dem Lokativ Singular о́ці und den Pluralformen о́чі, оче́й, о́чам, о́чі bzw. ві́чі im Akkusativ und очи́ма im Instrumental, die sich insgesamt keinem produktiven Paradigma zuordnen lassen. Zur Klasse nh3 gehören fast ausschließlich Substantive auf , von denen einige alternativ feminines Genus haben und die Klasse fh2 repräsentieren. Deklinable Substantive auf , oder kommen im Ukrainischen so gut wie nicht vor oder sind wie плече́ 'Schulter' als Sonderfälle zu betrachten. Die wenigen Substantive der Klasse nh4, die so ähnlich auch das Russische kennt, sind Bezeichnungen für kleine, junge Lebewesen. Für diese Klasse gilt, dass in allen Kasus, die formal nicht mit dem Nominativ Singular identisch sind, vor der Endung ein stammerweiterndes Phonem /t/ einzufügen ist. Neben den Substantiven der Klasse fh4 und der unter 3.2. beschriebenen Klasse mhin kennt das Ukrainische genau wie das Russische natürlich noch weitere Substantive mit Stammerweiterungen. In den meisten Fällen handelt es sich jedoch um isolierte Einzelfälle. Wenn überhaupt, dann repräsentieren lediglich noch die Substantive auf mit der Stammerweiterung je eine entsprechende Flexionsklasse auf harten und weichen Stammauslaut. Allerdings weisen hier im Singular jeweils alle Formen mit Stammerweiterung konkurrierende Formen ohne Stammerweiterung auf, die der Klasse nw3 zugeordnet werden können, so z.B. bei пле́м’я 'Volksstamm' – Gen.: пле́мені/ пле́м’я, Dat.: пле́мені/ пле́м’ю, Instr.: пле́менем/ пле́м’ям, Lok.: пле́мені/ пле́м'ї/ пле́м'ю. Die Pluralformen werden hingegen konsequent durch Anfügen der Stammerweiterung und die Endungen gebildet, die auch für alle übrigen Neutra mit hartem Stammauslaut gelten: племена́, племе́н, племена́м. Tab. 12: Weicher Stammauslaut, Singular.
Klasse Kasus
nw1
'Ring'
nw2
'Meer'
nw3
'Rang'
nw4
'Gipfel'
Nom.
ɛ
кільце́
ɛ
мо́ре
a
звання́
a/ ɛ
верхі́в’я
Gen.
a
кільця́
a
мо́ря
a
звання́
a
верхі́в’я
Dat.
u
кільцю́
u
мо́рю
u
званню́
u
верхі́в’ю
Akk.
ɛ
кільце́
ɛ
мо́ре
a
звання́
a/ ɛ
верхі́в’я
202 | Kersten Krüger (Leipzig)
Instr.
ɛm
кільце́м
ɛm
мо́рем
am
звання́м
am
верхі́в’ям
Lok.
i/ u
кільці́/ -ю
i
мо́рі
i/ u
званні́/ -ю
i/ u
верхі́в’ї/ -ю
Vok.
ɛ
кільце́
ɛ
мо́ре
a
звання́
a/ ɛ
верхі́в’я
nw3
Beispiel
nw4
Beispiel
Tab. 13: Weicher Stammauslaut, Plural.
Klasse Kasus
nw1
Beispiel
nw2
Beispiel
Nom.
a
кі́льця
a
моря́
a
звання́
a
верхі́в’я
Gen.
–
кі́лець
iv
морі́в
–
звань
iv/ –
верхі́в’їв
Dat.
am
кі́льцям
am
моря́м
am
звання́м
am
верхі́в’ям
Das mit Abstand häufigste Flexionsmuster bei Neutra mit weichem Stammauslaut repräsentiert die Klasse nw3, zu der alle Substantive mit den auslautenden Graphemfolgen , , , , , , , und sowie gehören. Zu Abweichungen von den angegebenen Endungen kommt es dabei ausgesprochen selten und nur im Lokativ Singular, wo einige wenige Substantive – vor allem solche auf – allein die Endung -/i/ zulassen, sowie im Genitiv Plural, wo in der Regel dieselben Substantive anstelle der Nullendung die Endung -/iv / bzw. haben und somit quasi eine eigene, besondere Flexionsklasse bilden. In allen Fällen mit Nullendung im Genitiv Plural werden auslautende Doppelkonsonanten zu einfachen, was wiederum bei denen mit der Nennform auf und mit Verhärtung und Schreibung ohne Weichheitszeichen einhergeht. Die vom Umfang her ebenfalls nicht unbeträchtliche Klasse nw1 umfasst ausschließlich Neutra auf und ist auch in Bezug auf die Endungen äußerst homogen. Gelegentlich ist im Lokativ Singular nur die Form auf -/i/ belegt und im Genitiv Plural unterbleibt bei einigen Substantiven, wie z.B. bei місць (zu мі́сце 'Ort') und сонць (zu со́нце 'Sonne') der Vokaleinschub, der sonst außer nach Vokal regelmäßig stattfindet, sodass also annähernd alle zugrundeliegenden Stämme dieser Klasse eine Raute beinhalten. Die Klasse nw2 repräsentieren im Prinzip nur го́ре 'Leid', мо́ре 'Meer' und по́ле 'Feld', von denen го́ре Singularetantum ist und по́ле im Genitiv Plural als Variante die endungslose Form піль aufweist, und die Flexionsklasse eigentlich nur durch zahlreiche Komposita mit diesen Wörtern begründet ist. Ähnlich verhält es sich mit ви́м’я 'Euter', сі́м’я 'Samen' und ті́м’я 'Scheitel', die im Genitiv, Dativ, Instrumental und Lokativ Singular zwar dieselben Varianten wie das bei
Automatische Generierung der Flexionsformen der Substantive im Ukrainischen | 203
den Neutra mit hartem Stammauslaut beschriebene пле́м’я 'Volksstamm' haben, ansonsten aber die Endungen der Flexionsklasse nw4 zeigen und eben dieser zugeschlagen werden können. Diese letzte Klasse umfasst wiederum eine größere Anzahl von Neutra, deren Stämme allesamt auf -/j/ enden und die im Nominativ Singular die Endung -/a/ bzw. Konsonant + oder die Endung -/ɛ/ bzw. haben, das hier stets auf i folgt und nach Vokal das stammauslautende /j/ mit bezeichnet. Streng genommen haben wir es hier mit zwei Flexionsklassen zu tun, denn die Substantive auf sind im Genitiv Plural endungslos bzw. gehen auf aus, doch sind es weniger als zwei Dutzend und die Hälfte davon sind Singulariatantum.
5 Akzentuierung Die Betonung der Flexionsformen, die nach der Zuweisung der Endungen erfolgt, basiert auf entsprechenden Akzentschemata. Damit kommt die morphologische Akzentologiekonzeption ins Spiel, wie sie vor allem durch Zaliznjak (2002/1967, 149-276; 2003, 25-142) sowie auch durch Lehfeldt (2003) in Bezug auf das Russische herausgearbeitet worden ist. Diese unterscheidet prinzipiell zwischen abstrakten und konkreten Formen, was mit unterschiedlichen Ebenen der linguistischen Beschreibung und mit den Begriffen Akzent und Betonung korreliert. Somit lässt sich der Akzent als ein Merkmal abstrakter Wortformen auffassen, das diesen im Sinne jeweils einer akzentuierten Silbe auf der phonologischen bzw. morphonologischen Ebene zukommt, wohingegen die Betonung als Eigenschaft konkreter Wortformen auf der phonetischen Ebene realisiert wird. Im Folgenden wird für die Erfassung der Betonungsverhältnisse in abstrakten Akzentschemata von den im dritten Abschnitt beschriebenen zugrundeliegenden morphonologischen Stammformen ausgegangen, in denen also die Positionen fluktuierender Vokale in allen Formen durch Rauten ausgewiesen sind. Des Weiteren werden in den Akzentschemata auch vokalisch leere Endungen durch Rauten angegeben, die genau wie die für fluktuierende Vokale als akzentogene Positionen behandelt werden. Folglich können in den Akzentschemata alle Formen den Akzent auf dem Stamm oder auf der Endung haben und das sind auch die beiden Positionen, die durchgängig unterschieden werden. Das heißt, jede beliebige Form eines beliebigen Akzentschemas ist entweder stamm- oder endungsakzentuiert. Auf dieser Grundlage werden die Akzentverhältnisse in allen primären Formen des Paradigmas erfasst. Die Vorgaben hierfür liefert die Betonung der analogen konkreten Wortformen, wobei sekundäre bzw. sogenannte individuelle Wortformen, wie Lokative mit der Präposition по, z.B. по боло́ту (zu
204 | Kersten Krüger (Leipzig)
боло́то 'Sumpf'), oder Numerative wie z.B. два бра́ти (zu брат 'Bruder'), zu vernachlässigen sind. Die Akzentverhältnisse der beiden Teilparadigmen des Singulars und Plurals werden einzeln erfasst so wie auch die Angabe zur Akzentuierung des Vokativs Singular gesondert erfolgt. Für den Vokativ Plural erübrigt sich eine Akzentangabe, da dieser stets mit dem Nominativ Plural zusammenfällt. Die besondere Behandlung des Vokativs Singular ergibt sich aus der Übernahme der Grundlagen für die Akzentbeschreibung der russischen Substantive, die ja keinen systematischen Vokativ kennen, aber ansonsten – also ohne diesen – alle Akzenttypen bzw. -schemata mit dem Ukrainischen teilen. Die Zuordnung eines Substantivs zu je einem Akzenttyp im Singular und Plural und dann weiter zu einem Akzentschema erfolgt zunächst über die Betonung in jeweils einer der tatsächlichen Formen, die als Referenz- oder Kontrollform dienen. Hierfür kommen alle Formen in Betracht, in denen keiner der weiter unten beschriebenen Akzentwechsel stattfindet und die keine vokalisch leere Endung aufweisen, also z.B. der Dativ Singular und der Dativ Plural. So ist z.B. bei бобе́р 'Biber' bzw. dem zugrundeliegenden Stamm bob#r- mit den entsprechenden Kontrollformen бобру́/ бобро́ві und бобра́м sowohl im Singular als auch im Plural von paradigmatischem Endungsakzent auszugehen, der im Weiteren in keinem Kasus durchbrochen wird und daher als durchgehend zu betrachten ist. Zwar erscheint die tatsächliche Form des Nominativ Singulars бобе́р an der Oberfläche stammbetont, doch das nur deshalb, weil der Nominativ Singular mit der zugrundeliegenden Form bob#r-#´11 eine vokalisch leere Endung hat und sich die Betonung quasi automatisch auf den Stamm verlagert. Bei automatischer Betonung, der alle Substantive mit endungsakzentuierten Kontrollformen in endungslosen Kasus unterliegen, wird die Betonung in der Regel auf die letzte Stammsilbe verlagert. Diese Regel wird nur in sehr wenigen Fällen durchbrochen, zu denen bspw. der Nominativ Singular чо́вен 'Kahn' zu den Singularformen човна́, човну́ / човно́ві usw. oder die endungslosen Pluralformen ма́сел (zu ма́сло 'Butter') und ли́жень (zu лижня́ 'Spur') im Vergleich zu den übrigen Pluralformen масла́ und масла́м bzw. лижні́ und лижня́м zählen. Im Unterschied zur automatischen Betonung bei бобе́р liegt z.B. im Falle des Akkusativs Singular го́ру (zu гора́ 'Berg') echter Akzent- bzw. Betonungswechsel vor. Auch hier gibt die Kontrollform des Dativs mit горі́ paradigmatischen Endungsakzent vor, jedoch wird dieser im Akkusativ Singular durchbrochen, obwohl Endungsakzent möglich wäre. Ein Beispiel für Akzentwechsel und bedingten, automatischen Betonungswechsel im Plural liefert брова́ 'Braue' mit den Pluralformen бро́ви und
|| 11 Zugunsten der besseren Lesbarkeit steht das Akzentzeichen nicht über, sondern unmittelbar nach der Raute.
Automatische Generierung der Flexionsformen der Substantive im Ukrainischen | 205
брів im Verhältnis zur Referenzform брова́м, der wiederum die Formen des Instrumentals und Lokativs Plural folgen: брова́ми, брова́х. Welche Akzentschemata die ukrainischen Substantive insgesamt aufweisen und wie sich die tatsächlichen Betonungsverhältnisse jeweils gestalten, zeigen die Tabellen am Ende dieses Abschnitts. Zur Bezeichnung der Akzenttypen bzw. Akzentschemata gelten dabei folgende Konventionen: – alle Akzentschemata werden durch zwei der Kleinbuchstaben a, b, c, d und e für die Akzentverhältnisse vom Nominativ bis zum Lokativ und ggf. zusätzlich durch < für den Vokativ Singular bezeichnet – der erste Buchstabe steht für das Teilparadigma des Singulars, der zweite – für den Plural. Beide gemeinsam bezeichnen jeweils einen grundlegenden Akzenttyp – a steht für durchgängigen Stammakzent auf ein und derselben morphologischen Silbe im betreffenden Teilparadigma – b steht für durchgängigen Endungsakzent im betreffenden Teilparadigma, wobei zweisilbige Endungen stets auf der ersten Silbe akzentuiert sind – c, d oder e stehen für unterschiedliche Typen von Wechselakzent innerhalb eines Teilparadigmas – bei c für das Singularparadigma hat allein der Akkusativ eine andere Akzentstelle als die übrigen Formen, die hier auf der ersten Stammsilbe und bei allen übrigen Singularformen auf der Endung liegt – bei c für das Pluralparadigma ist allein der Nominativ der akzentuelle Ausreißer, der allerdings bei unbelebten Substantiven im Akkusativ und bei Bezeichnungen für Tiere als Akkusativvariante übernommen wird; der Akzent liegt hier auf der ersten Stammsilbe und sonst im Pluralparadigma auf der Endung – d kommt nur in Bezug auf den Singular vor und bezeichnet hier Wechselakzent vom Stamm auf die Endung im Lokativ – e kommt nur in Bezug auf den Plural vor und bezeichnet hier Wechselakzent vom Stamm auf die Endung im Genitiv – < steht für Verlagerung des Akzents im Singular Vokativ um eine Silbe nach links Das Akzentschemata aa steht für Akzent auf ein- und derselben Stammsilbe in allen Singular- und Pluralformen einschließlich des Vokativs, bb bezeichnet Endungsakzent in allen Singular- und Pluralformen, ab meint Akzent auf ein- und derselben Stammsilbe in allen Singularformen einschließlich des Vokativs und Endungsakzent in allen Pluralformen, ba steht für den umgekehrten Fall, wobei die Akzentstelle im Plural im Vergleich zum Singular um eine Silbe nach links verlagert ist. Alle übrigen Akzentschemata, also z.B. ac, ae, ba