University of Southern California Dissertations and Theses

Decomposing Slavic aspect: the role of aspectual morphology in Polish and other Slavic languages

(USC Thesis Other)

Decomposing Slavic aspect: the role of aspectual morphology in Polish and other Slavic languages

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content DECOMPOSING SLAVIC ASPECT: THE ROLE OF ASPECTUAL MORPHOLOGY IN POLISH AND OTHER SLAVIC LANGUAGES by Agnieszka Agata Łazorczyk A Dissertation Presented to the FACULTY OF THE USC GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (LINGUISTICS) August 2010 ii ACKNOWLEDGEMENTS This has been a long, but wonderful journey, and I feel both happiness and sadness it has come to an end. I am so grateful for the many amazing people I have met along the way, and especially for and to those who have made this experience possible in the first place. The most important person in this regard is Prof. Hagit Borer. It is safe to say that I might have never pursued my interest in linguistics to the extent that I did without her encouragement. She supported me through my challenging family situation, has been a wonderful mentor, and a role model as a person and a scholar. I am so thankful for having had the opportunity to have met her and worked with her. I feel the same gratitude and admiration towards Prof. Roumi Pancheva, who has been extremely generous with her time and always very supportive. I have benefited enormously from her expertise and learned a lot from her in terms of practicalities of doing research and being a scholar. If there is any value or insight to the work presented here, I owe it in a large part to the direction of these two wonderful women and exceptional linguists. I also want to appreciate my other dissertation committee members, who have dedicated their time and provided very useful comments: Prof. Elena Guerzoni, Prof. Thomas Seifrid, as well as Prof. Tania Ionin, who worked with me in the past. I have also benefitted a lot in the course of the program from classes and discussions with other professors at UCS Linguistics as well as from fellow graduate students. The company of the latter, especially Jelena Krivokapic, Isabelle Roy, Eunjeong Oh, Basia Tomaszewicz, Nihan Ketrez, Cristian Iscrulescu, Asier Alcazar, and Simona Montanari, is what made the time spent at USC such a great experience. Reflecting on my development as a linguist, I would also like to acknowledge the influence of my undergraduate and master‘s program iii professors at the Department of English Philology at the University of Wrocław: Prof. Michał Post, Prof. Andrzej Skrzypiec, Prof. Przemysław Pawelec, but most importantly Prof. Bożena Rozwadowska, who introduced me to generative linguistics, gave direction as an advisor, and encouraged me to pursue doctoral studies. Last but not least, I would like to thank Mr. Władysław Panaś, my primary school Polish language teacher, who first introduced me to syntactic trees, albeit not of the generative kind. The constant exercise in breaking up sentences into constituents and arranging them hierarchically was hated by most students, but to me it was fascinating and it planted a seed that kept on growing my entire life. I hope he is smiling from up there as I am writing this. This journey, while certainly enjoyable, has also been challenging in many different respects, and I would not been able to continue on it had it not been for the love, patience, and support of my family and friends. My family has had to make a lot of sacrifices, in particular my husband and our three little kids – Hannah, Misali and Saraya. I love them with all my heart and I hope this work of mine will inspire them to follow their dreams when they grow up. I am also incredibly grateful for the steady support, both moral and financial, of my father, Edward. Ogromnie Ci dziękuje, Tato! Lastly, I want to express my appreciation to my incredible friends Magdalena Mączyńska (Iwanik) and Jelena Krivokapic, who have shown me so much love and never stopped to encourage me, and also to many others at the First Mennonite Church of Upland family (a special ‗thank you‘ goes to Henny van der Zwaag for staying on my case and helping with the kids), to those near and far in the United States, and to countless others in Poland. You are all a part of this accomplishment. Los Angeles/Wrocław, June 2010 iv TABLE OF CONTENTS ACKNOWLEDGEMENTS ............................................................................................ ii LIST OF TABLES ....................................................................................................... vii LIST OF FIGURES ..................................................................................................... viii ABSTRACT .................................................................................................................. ix CHAPTER 1: INTRODUCTION ....................................................................................1 1. The goal ...................................................................................................................1 2. Two-tiered theory of aspect ......................................................................................2 3. The peculiarity of Slavic aspect ................................................................................4 4. Recent approaches to Slavic aspect ...........................................................................6 5. The overlooked problem of secondary imperfectivization and a novel analysis .........8 6. Theoretical assumptions ...........................................................................................9 7. The outline of the dissertation ................................................................................. 10 CHAPTER 2: ASPECTUAL MORPHOLOGY IN SLAVIC AND THE PERFECTIVE- IMPERFECTIVE DISTINCTION .................................................................................. 14 1. The Verbal Template in Slavic ............................................................................... 14 2. Aspectual Morphology in Slavic and the perfective-imperfective distinction ........... 15 3. Verbal prefixes ....................................................................................................... 21 3.1. General characteristics and prefix list .............................................................. 21 3.2. Meaning-modifying/changing prefixes ............................................................ 24 3.3. Purely aspectual prefixes ................................................................................. 29 3.4. Quantificational prefixes.................................................................................. 30 4. The secondary imperfective morphology ................................................................ 31 5. The Slavic perfective-imperfective distinction: the diagnostics ................................ 33 Appendix 1 .................................................................................................................... 41 CHAPTER 3: THE ASPECTUAL ROLE OF VERBAL PREFIXES ............................. 45 1. Two Types of Aspect .............................................................................................. 45 1.1. Aktionsart and Telicity: definitions and tests ................................................... 45 1.1.1. Telicity as non-homogeneity ........................................................................ 45 1.1.2. Commonly used tests for telicity .................................................................. 48 1.1.3. Basic syntax of telicity ................................................................................. 50 1.2. Viewpoint Aspects: PERFECTIVE and IMPERFECTIVE ........................................ 51 1.1.4. Semantics of Viewpoint Aspects .................................................................. 51 1.1.5. Tests for PERFECTIVITY and IMPERFECTIVITY ................................................ 53 1.1.6. Syntax of Viewpoint Aspect ......................................................................... 56 2. The nature of the Slavic aspectual contrast: B-imperfectives vs perfectives ............. 56 3. Telicity and the perfective-imperfective distinction ................................................. 62 3.1. Telicity diagnostics .......................................................................................... 62 3.2. Prefixed = telic ................................................................................................ 67 v 3.3. Considering counterevidence ........................................................................... 68 3.3.1. Filip‘s (1999, 2000, 2004) Objections .......................................................... 68 3.3.2. Borik‘s (2002) Objections ............................................................................ 73 3.3.2.1. Perdurative pro- ....................................................................................... 73 3.3.2.2. ‗Beginning verbs‘ ..................................................................................... 76 3.4. Prefixes as telicity-markers: concluding thoughts ............................................ 78 4. Viewpoint aspect and the perfective-imperfective distinction .................................. 79 4.1. Slavic B-imperfective = IMPERFECTIVE? ........................................................... 79 4.2. Slavic perfective = PERFECTIVE? ...................................................................... 81 4.3. Verbal prefixes and viewpoint ......................................................................... 84 4.4. Slavic aspectual contrasts: summary ................................................................ 89 5. The Syntax of Slavic aspect .................................................................................... 92 5.1. The morphosyntax of prefixation ..................................................................... 92 5.2. Projection and interpretation of arguments ....................................................... 95 5.3. Composing with outer aspect ......................................................................... 101 6. Summary .............................................................................................................. 102 CHAPTER 4: THE SECONDARY IMPERFECTIVE ................................................... 104 1. Secondary imperfective: the problem .................................................................... 104 1.1. Secondary Imperfective morphology .............................................................. 104 1.2. Secondary imperfectivization – the semantic effect ........................................ 105 1.3. SI – marker of viewpoint or …?..................................................................... 106 2. Old Church Slavonic ............................................................................................ 108 2.1. Viewpoint aspect markers in OCS ................................................................. 109 2.2. Aspectual paradigm in OCS and the role of the SI ......................................... 110 2.2.1. Compositional possibilities ........................................................................ 110 2.2.2. The distribution of the SI vis-à-vis the aorist/imperfect distinction ............. 110 2.2.3. The *[perfective IMPERFECTIVE] rule in OCS and S-imperfectivization ....... 113 3. Bulgarian .............................................................................................................. 118 3.1. The Bulgarian paradigm ................................................................................ 118 3.2. *perfective IMPERFECTIVE rule in Bulgarian ................................................... 121 4. The Secondary Imperfective in Polish ................................................................... 124 4.1. S-imperfectives and atelicity .......................................................................... 124 4.2. The S-imperfective: viewpoint aspect marker vs. Aktionsart marker .............. 126 4.2.1. Distribution of the S-imperfective morphology ........................................... 126 4.2.2. Viewpoint Interpretation of S-imperfectives in Polish ................................. 128 4.3. Conclusions ................................................................................................... 132 5. Analysis ............................................................................................................... 132 5.1. Setting the stage ............................................................................................ 132 5.2. SI as a partitive homogenizer ......................................................................... 135 5.3. Why do we have S-imperfectivization? .......................................................... 139 5.4. Imperfective = S-imperfective = atelic ............................................................ 142 6. Aspectual triplets a.k.a. verbal triangles ................................................................ 145 6.1. The problem .................................................................................................. 145 6.2. What‘s behind the restriction in Polish? ......................................................... 149 6.3. Verbal triangles in Bulgarian (and Upper Sorbian)......................................... 153 vi 6.4. Non-lexical prefixes: summary ...................................................................... 156 7. The interpretation of arguments under the S-imperfective ..................................... 156 8. Conclusions .......................................................................................................... 159 CHAPTER 5: QUANTIFICATIONAL PREFIXES: UNIQUE CHARACTERISTICS 162 1. Q-prefixes: unique properties................................................................................ 162 2. Quantificational-like function ............................................................................... 163 2.1. Quantificational role of na Q - .......................................................................... 165 2.1.1. na Q - and the interpretation of the direct object ............................................ 165 2.1.2. na Q - as a quantifier over the event .............................................................. 180 2.2. The quantificational uses of po- ..................................................................... 189 2.2.1. The distributive po D - .................................................................................. 189 2.2.2. po Q - ‗some, a little‘ .................................................................................... 192 2.3. Summary ....................................................................................................... 197 3. Stackability .......................................................................................................... 198 4. Summary .............................................................................................................. 205 CHAPTER 6: QUANTIFICATIONAL PREFIXES AND ASPECT ............................ 207 1. Q-prefixed verbs and the perfective-imperfective distinction ................................. 208 2. Quantificational prefixes and telicity .................................................................... 211 2.1. Po D - and time adverbial test ........................................................................... 211 2.2. Na Q - and time adverbial test ........................................................................... 216 2.3. Po Q - and time adverbials ................................................................................ 221 2.4. Other telicity tests and na Q - and po Q - with scope over the object .................... 230 2.4.1. Other telicity tests and na Q -prefixed verbs .................................................. 231 2.4.2. Other telicity tests and po Q -prefixed verbs .................................................. 233 2.5. Summary ....................................................................................................... 234 3. Q-prefixes and viewpoint aspect ........................................................................... 235 4. Quantificational prefixes: previous classifications and accounts............................ 238 4.1. Q-prefixes: not special ................................................................................... 238 4.2. Lexical versus superlexical prefixes ............................................................... 239 4.2.1. Babko-Malaya (1999) ................................................................................ 239 4.2.2. Svenonius (2004) ....................................................................................... 242 4.2.3. Internal versus external prefixes (DiSciullo & Slabakova) .......................... 247 4.3. Summary ....................................................................................................... 250 5. Syntactic projection of Q-prefixes ........................................................................ 250 5.1. po Q - with non-degree achievement verbs ....................................................... 251 5.2. The syntax of other Q-prefixes (na Q -, po D - and po Q - with DAs) ...................... 252 6. Conclusions .......................................................................................................... 260 CHAPTER 7: CONCLUSIONS AND OUTSTANDING ISSUES ............................... 261 BIBLIOGRAPHY ....................................................................................................... 266 vii LIST OF TABLES Table 1: Aktionsart properties and event types ............................................................... 46 Table 2: The relationship between Slavic aspectual categories, telicity and viewpoint aspect .............................................................................................................. 90 Table 3: The OCS Imperfect paradigm (cf. Lunt 2001:100) ......................................... 109 Table 4: The OCS Aorist paradigm (cf. Lunt 2001: 102, 104) ...................................... 109 Table 5: The interaction of Aktionsart with viewpoint aspect in the OCS corpus (Dostál 1954:600) …………………………………………………………. ... 113 Table 6: The aspectual paradigm for the verb 'write' in the past tense in Bulgarian ....... 119 Table 7: The aspectual paradigm for the verb 'drink' in the past tense in Bulgarian ...... 119 Table 8: Quantification-related properties of the stackable prefixes na Q -, po Q -, and po D - .............................................................................................................. 198 Table 9: Superlexical prefixes (Svenonius 2004) ......................................................... 243 Table 10: Internal vs. external prefixes (DiSciullo & Slabakova 2005) ........................ 248 viii LIST OF FIGURES Figure 1: Preliminary basic syntax and atelicity and telicity ........................................... 51 Figure 2: Theoretical combinatorial possibilities of telicity and viewpoint aspect........... 56 Figure 3: Prefixed Root Nominals .................................................................................. 86 Figure 4: Morphosyntax of telicity in Slavic ................................................................... 93 Figure 5: Syntax of aspect marking via prefix in Slabakova (2004) ................................ 98 Figure 6: Atelic and telic structures and the direct object ............................................. 101 Figure 7: Combiniatorial possibilities of (a)telicity and viewpoint aspect in Slavic and their syntax .................................................................................................................. 102 Figure 8: The syntax of the secondary imperfective ..................................................... 139 Figure 9: Types of Aktionsart structures in Slavic ........................................................ 139 Figure 10: Lexical vs. non-lexical prefixes................................................................... 146 Figure 11: Secondary imperfectivization of perfective forms ........................................ 146 Figure 12: *Secondary imperfective with pure perfectivizers ........................................ 153 Figure 13: Pereltsveig's 2006 analysis of the cumulative na- ........................................ 175 Figure 14: Reflexivization in the context of na Q - ......................................................... 184 Figure 15: Po Q - combining with B-imperfectives ......................................................... 252 Figure 16: Po Q - combining with secondary imperfectives ............................................ 252 Figure 17: The syntactic projection of na Q -, po Q -DegA and po D - ................................. 255 Figure 18: The syntax of na Q - and po D - attaching to S-imperfectivized forms ............... 255 Figure 19: The syntax of iterated po Q -'s ....................................................................... 258 ix ABSTRACT This dissertation considers the problem of the semantic function of verbal aspectual morphology in Polish and other Slavic languages in the framework of generative syntax and semantics. Three kinds of such morphology are examined: (i) prefixes attaching directly to the root, (ii) secondary imperfective suffixes, and (iii) three prefixes that have a quantificational function and are able to stack on top of already prefixed verbs: na- ‗a lot,‘ po- ‗some, a little‘ and the distributive po-. Assuming a two-tiered theory of aspect (Smith 1991), where two distinct types of grammatically encoded aspect are distinguished – Aktionsart (telicity, durativity, dynamicity) and viewpoint (PERFECTIVITY vs. IMPERFECTIVITY) – the aspectual contribution of the above listed markers is analyzed using various syntactic and semantic tests, leading to a number of new conclusions and novel accounts of various problems. It is determined that Slavic verbal prefixes are uniformly markers of telicity, and then proposed that they project as heads of the telicity- encoding projection InnerAsp. Prefixed forms are also viewpoint-PERFECTIVE (= temporally bound), but for a number of reasons this fact is attributed to a compositional restriction whereby telic predicates compose only with the PERFECTIVE viewpoint, rather than to the prefixes fulfilling that role. With regard to secondary imperfective suffixation, careful examination of data from Old Church Slavonic, Bulgarian and Polish reveals, contrary to what is commonly assumed, that secondary imperfective forms need not be interpreted IMPERFECTIVELY. Their interpretation is, on the other hand, always atelic. This leads to the novel proposal that the secondary imperfective morphemes correspond to a partitive-homogenizing operator, which applies to telic predicates and returns atelic ones. It is further proposed that this operator heads the InnerAsp SI projection, merging right x above InnerAsp. Lastly, with regard to quantificational prefixes, they are subjected to a two-part analysis, with the first one focusing on their unique properties, i.e., their role as intensifiers and variable quantifiers as well as their stacking ability, and the second one addressing their aspectual role. Aspectually, quantificational prefixes are shown to behave largely like other, ―regular‖ prefixes, with only po- ‗some, a little‘ outside of degree achievement predicates being Aktionsart neutral. This observation puts into question a number of previous analyses which claim that such quantificational prefixes do not participate in telicity-marking. Their newly reported Aktionsart role necessitates a recursive telicity-encoding structure for all cases where a quantificational prefix attaches to a secondary imperfective form. This is proposed to be accomplished through an additional telicity-encoding projection InnerAsp Q , which merges with InnerAsp SI . 1 CHAPTER 1: INTRODUCTION 1. The goal The goal of this dissertation is to identify the semantic role and the syntax associated with morphological markers of aspect in Polish and other Slavic languages. The task consists largely in disambiguating the kind of aspectual meaning(s) that the relevant morphology encodes, which continues to elude researchers and is still a matter of controversy, despite quite a substantial body of research produced within diverse linguistic traditions 1 . The morphology in question includes: (a) preposition-like verbal prefixes or preverbs, whose number is close to 20, and which attach to simplex forms and derive so-called perfective verbs, (b) 2 suffixes with additional allomorphs which derive so-called secondary imperfective verbs from perfective forms, and (c) prefixes with quantificational functions which can attach to simplex verbs or to secondary imperfective forms and derive perfective verbs. The application of the three types of morphology are illustrated in (1) below. (1) a. bare form: pisać ‗to write‘ b. prefixed form: prze-pisać ‗to copy, re-write, prescribe‘ c. prefixed-suffixed form: prze-pis-yw-ać ‗to copy, rewrite, prescribe‘ d. form with a Q uant-prefix: po-pisać ‗to write some‘ 1 An extensive bibliography of works on the topic of aspect in Slavic languages can be found at http://www.utsc.utoronto.ca/~binnick/TENSE/OnSlavic.htm. Special mention for foundational work in non-generative traditions has to be been given to Comrie (1976), Forsyth, (1970), Isačenko (1960) and Jacobson (1936). 2 po-prze-pi-yw-ać ‗to copy, rewrite, prescribe some; to copy, rewrite, prescribe distributively‘ The remaining aspectual markers, namely the semelfactive suffix and the habitual suffix are not considered extensively in this work. 2. Two-tiered theory of aspect It has become customary since Smith (1991), who builds on much previous work on situation types and grammatical aspect 2 , to recognize two distinct layers of aspectual meaning: situation type and viewpoint aspect. Situation type, also referred to as Aktionsart, 3 lexical aspect or inner aspect, concerns the internal constituency of events or situations, and involves properties such as telicity (i.e., whether or not an event or situation is culminating/non-homogenous or homogenous), dynamicity (i.e., whether or not a situation or event is dynamic or stative), and durativity (i.e, whether a situation is punctual or has duration). Viewpoint aspect, on the other hand, also referred to as grammatical or outer aspect, concerns the perspective that the speaker takes on an event or situation, and typically distinguishes PERFECTIVE viewpoint (i.e., bounded, viewed from the outside, inclusive of the initial and final endpoints of an event) and IMPERFECTIVE viewpoint (i.e., unbounded, viewed from the inside, exclusive of the initial 2 Previous seminal works on verbal aspect include Vendler (1957/1967), Comrie (1976), Dowty (1979), and Verkuyl (1972, 1989). There was also an earlier rich descriptive and structuralist tradition of trying to tease apart the aspectual contribution of various morphological markers in the verb forms of Indo-European languages, which relied on the Aktionsart-grammatical aspect differentiation. The opinion that grammatical aspect is what is expressed by inflectional means comes from these traditions (Filip, to appear). 3 Originally, as recounted in Filip (forthcoming), ―[T]he term AKTIONSART(EN) was coined by Agrell (1908) to cover the lexicalization of various ‗manners of action‘ (e.g., terminative, resultative, delimitative, perdurative, iterative, semelfactive, attenuative, augmentative) by means of overt derivational word- formation devices, and set apart from grammatical aspect.‖ The current usage of the term Aktionsart is more narrow, as outlined in the text. 3 and final endpoints of an event) 4,5 . There is also another, third layer of aspectual meaning that can be further distinguished, namely one that has to do with pluractionality and singularactionality, and involves notions such as habituality and iterativity, among others van Geenhoven (2005), but these will not be considered at length in this work. I assume here following Smith (1991) that the different kinds of aspect are independent of each other and should in principle combine freely. Taking just telicity and viewpoint as an example, this would give us the following possibilities: (a) atelic IMPERFECTIVE, (b) atelic PERFECTIVE, (c) telic IMPERFECTIVE, and (d) telic PERFECTIVE. While some languages may realize the full paradigm, e.g. Greek and Bulgarian, other languages sometimes align a particular viewpoint with a particular Aktionsart (Bohnemeyer & Swift 2004, Bertinetto 2001). In Georgian and Upper Sorbian, for example, telic predicates are reported to be realized only in the PERFECTIVE viewpoint, while atelic ones in the IMPERFECTIVE (Georgian: Dahl 1985, Upper Sorbian: Stone, 1993). In Greek, which otherwise allows all combinatorial possibilities, for predicates which are not unambiguously telic or atelic, the application of the PERFECTIVE viewpoint correlates with the telic reading, and of the IMPERFECTIVE viewpoint with the atelic one (Mavromanolaki 2001, Horrocks & Stavrou 2003). Also, in French, the PERFECTIVE aspect (passé simple) applied to a stative predicate gives it a change of state reading (Cox 4 In this work, I will use the words PERFECTIVE and IMPERFECTIVE in small caps to refer to the universal semantic notions of bounded and unbounded readings respectively, to be defined in detail in the subsequent chapter, while the Slavic categories of perfective and imperfective (simplex and secondary) will be given in italics and should be understood to correspond to sets of verbs with a particular distribution with a number of syntactic-semantic tests, which also correlated with particular morphological complexity, both the be discussed in Chapter 2. The use of the terms perfective and imperfective is necessitated by considerations of space (as compared to possible, longer-winded descriptive alternatives), as well as by the established Slavic tradition, and should not be taken to indicate any particular aspectual interpretation of such forms, as will be shown shortly. 5 Smith (1991) distinguishes also another kind of viewpoint aspect, namely the neutral viewpoint, which is defined as including the initial endpoint and at least one internal stage of an event. 4 1982, Smith 1991, Vikner, 1994). Lastly, it is well know that the application of the IMPERFECTIVE viewpoint to telic predicates invariably returns an atelic interpretation (cf. Dowty‘s 1977 ‗imperfective paradox‘), revealing interaction that goes beyond the additive semantics of composing semantically and structurally distinct categories. All these examples show that while the two kinds of aspect are in principle independent, they can interact and converge, making it difficult sometimes to separate them in analysis. 3. The peculiarity of Slavic aspect Such meshing or alignment of the two kinds of aspect, which underlies the historic use of the same term to refer to both Aktionsart and viewpoint, obtains to a degree also in Slavic, and is one reason why analyzing the semantic contribution of individual markers of aspect in Slavic is so challenging. The other issue posing a problem for an easy classification of Slavic aspectual oppositions is the unusual combination of morphology used to mark them. Smith (1991) observes that typically languages mark viewpoint aspect through dedicated, inflection-like verbal morphology, as illustrated for Spanish in (2) and (3) 6 , while Aktionsart can be either part of the verb‘s meaning, or come from its arguments and modifiers. (2) a. Juan leyó el libro. (Bhatt & Pancheva, 2005) Juan read.past.PERFECTIVE the book ‗Juan read the book.‘ b. Juan leía el libro. Juan read.past.IMPERFECTIVE the book ‗Juan was reading the book.‘/‘Juan used to read the book.‘ 6 This often takes the form of so-called aspectual tenses, i.e., inflectional forms that convey both viewpoint aspect and tense, with the contrast typically restricted to past tense only. Such is also the case in Spanish. 5 (3) a. E tisa ena spiti. (Greek) build.past.PERFECTIVE a house ‗I built a house.‘ b. E tiza ena spiti. build.past.IMPERFECTIVE a house ‗I was building a house.‘/‗I used to build a house.‘ This assertion is not entirely correct with regard to Aktionsart, as there are languages that mark Aktionsart distinctions morphologically in a systematic way. Examples of that include telicity marking particles ra and xong/hêt in Vietnamese, as shown in (4) and (5) (Fukuda, 2007), and the telicity-marking prefix ha- in Malagasy (Travis, 2005). (4) a. Lan tìm hai quyến sách. (Fukuda 2007) Lan search two CL book ―Lan looked for two books.‖ b. Lan tìm ra hai quyến sách. Lan search telic-PART two CL book ―Lan found two books.‖ (5) a. Tân đọc hai quyến sách. Tân read two CL book ―Tân read two books.‖ b. Tân đọc xong/hết hai quyen sách Tân read telic-PART/telic-PART two CL book ―Tân has read two books.‖ 6 Nevertheless, the observation made by Smith (1991) does seem to reflect a general tendency. In Slavic, aspectual oppositions are encoded through morphology, which would seem to suggest viewpoint aspect, but while part of that morphology is regular and inflection-like (i.e., the secondary imperfective suffixes), another and a very prominent part is word-formational (i.e., the prefixes which typically change the lexical meaning of the verbs to which they attach), and still another part is quantificational, corresponding to free standing quantifiers in other languages. The overall picture that emerges is quite complicated and unlike what is observed in any other language or language family where aspect has been studied. This much has been concluded by Dahl (1985) in his extensive survey of aspectual systems in diverse languages. This uniqueness of Slavic aspect is a challenge to an account that aims at an analysis in terms of universal categories, but it is also an opportunity to add to the understanding of what kind of grammatical devices can be used in natural languages to mark various aspectual categories. 4. Recent approaches to Slavic aspect The more recent attempts at analyzing Slavic aspect fall into a few different camps. The first one are those who argue that Slavic perfective-imperfective opposition corresponds to differences in viewpoint, and therefore perfective=PERFECTIVE and imperfective=IMPERFECTIVE (e.g., Smith 1991, Nagórko 1998, Młynarczyk 2004, Borik 2002, and Pereltsvaig 2005, to appear, and Mezhevich, 2008). The second camp, focusing on the contribution of prefixes, sees the contrast between simplex and prefixed verbs as an Aktionsart contrast and consequently viewed prefixes as markers of telicity (e.g., Brecht 1984, Krifka 1992, Piñón 1995, Schmitt 1966, Verkuyl 1999, Slabakova 2005, 7 Borer 2005, and Travis 2005). In this group we can place also those whose analysis of verbal prefixation is more fine-grained, and who attribute telicity-marking role to only the so-called internal prefixes (as opposed to the so-called external prefixes, which are adverbial-like and always semantically transparent, and which are argued to be telicity neutral) (e.g., Svenonius 2004, Ramchand 2004, Di Sciullo & Slabakova 2005), and also the proposal advanced in Babko-Malaya (1999, 2003), where telicity-marking role of prefixes is indirect in that prefixes are said to have the role of deriving result-state predicates, which then compose (obligatorily, it appears) with a morphologically null BECOME affix (in the sense of Dowty 1979) and yield telic, change of state predicates. The proponents of the Aktionsart approach to prefixation are typically uncommitted with respect to the viewpoint aspect role of prefixes. 7 Their focus is simply on their telicizing natures, and no argument is ever made that prefixes could not simultaneously also be viewpoint markers. That possibility is acknowledged by a yet another group, who take the function of verbal prefixes to be that of telicity markers (sometimes also limiting it to internal/lexical prefixes) as well as PERFECTIVITY markers (e.g., Bertinetto 2001, Paslawska & von Stechow 2003, Ramchand 2004, Svenonius 2004). Lastly, there is the approach taken by Filip (2000, 2003, 2005) who strives to derive telicity in Slavic by compositional means, and denies any systematic grammatical role that would characterize all Slavic prefixes. As I will show in Chapter 3, the viewpoint-only position cannot be ultimately defended. Rather, prefixes are clearly involved in telicity marking, with the only exception to that being one of the quantificational prefixes. As for the viewpoint marking 7 In the case of Slabakova (2004) and DiSciullo & Slabakova (2005), this is unproblematic, as these analyses concern Bulgarian, and Bulgarian has a separate system of viewpoint aspect marking. 8 function of prefixation, I will show that prefixed forms are always viewpoint PERFECTIVE, but will argue that the prefixes themselves do not have a viewpoint-marking role and that the correlation is due to compositionality restrictions in that the telic predicates can compose only with the PERFECTIVE viewpoint. 5. The overlooked problem of secondary imperfectivization and a novel analysis With the exception of scholars who are squarely in the viewpoint aspect camp, the work of others focuses mostly, if not exclusively, on the role of aspectual prefixation. In the process, the contribution of the secondary imperfective as an aspectual marker is either overlooked or taken for granted. Likely due to the regular nature of the involved morphology, as well as to the frequent comparisons with the English progressive, the secondary imperfective is almost universally assumed to be a marker of the IMPERFECTIVE viewpoint (e.g., Schmitt 1996, Hewson & Bubenik 1997, Babko-Malaya 1999, Svenonius 2004, Jabłońska 2004, Borer 2005, inter alia). As I will show in Chapter 4, however, the data does not support this view. Furthermore, when the totality of facts from various Slavic languages is considered, in particular from those languages that have a separate system of marking PERFECTIVE and IMPERFECTIVE readings, such as Bulgarian and the historic Old Church Slavonic, an analysis of secondary imperfectivization as a marker of viewpoint aspect simply stops making sense. In this work I propose an alternative account where the secondary imperfective is analyzed as a partitive atelicizer and therefore a marker of Aktionsart, on a par with aspectual prefixes. This account not only squares better with the actual aspectual interpretation of secondary imperfective forms, but also substantiates the native intuition that the contrast between a prefixed perfective and the corresponding secondary imperfective verb is exactly of the same nature as the 9 contrast between a bare imperfective and the corresponding perfective, with both kinds of aspectual pairs differing principally in telicity. The new analysis also helps to understand why we find secondary imperfective morphology on deverbal nominals such as names of agents, professions, or instruments. Such nominals do not have any temporal properties, and therefore the presence of viewpoint aspect markers inside such forms would be completely unmotivated. The affixes deriving such nominals could, however, be understood to select for forms that denote processes or activities, which are atelic (in contrast to the telic achievements and accomplishments), and the secondary imperfective marker delivers just that. 6. Theoretical assumptions In terms of the theoretical background, this work assumes a neo-constructionist approach to aspect, and to the syntax/lexicon interface in general, where a sentence meaning is constructed through structure independently of the meaning of lexical terminals. The lexical items themselves function as modifiers rather than determinants of that structure, and the final meaning is a result of combining the two (Borer 2003, 2005). With respect to aspect, I take both telicity and viewpoint aspect to be anchored in syntactic structure, and more concretely in dedicated functional projections of Inner Aspect (InnerAsp) and Viewpoint Aspect (ViewAsp) respectively 8 . With regard to the syntactic model, I will assume the Minimalist bare phrase structure approach, as well as the operations of Merge and Move (alternatively, Copy and Merge) (Chomsky 1993). I will not, however, assume a Checking Theory approach to morphology, where fully 8 Outer Aspect could also be used as a term denoting the viewpoint aspect projection, but I will refrain from using it in this work as it may wrongly suggest Viewpoint Aspect represents the outermost layer of aspectual meaning. As I indicated earlier, another layer and consequently a higher projection is possible, namely one that concerns the pluractionality or singularactionality of an event and notions such as habituality and iterativity (Van Geenhoven 2005), which can be also overtly marked in Slavic. 10 derived and inflected vocabulary items are inserted into the structure, and those vocabulary items are pre-specified for features that are later checked against functional structure. Rather, instead of viewing fully derived and inflected words as something that is produced in the lexicon and later matched against structure, I assume them to be the product of the structure in the sense of Distributed Morphology (Halle & Marantz 1993) and Borer (2005). With respect to the Slavic aspectual morphology, it will mean that individual pieces of aspectual morphology must be related to syntactic structure, and that barring independent phonological principles, the morphological hierarchical structure (i.e., the order of attachment) reflects the syntactic hierarchy. 7. The outline of the dissertation This work is organized as follows. In the next chapter, Chapter 2, I provide a description of the morphology involved in aspect marking in Slavic, and review the various grammatical tests that are used to classify verb forms as imperfective or perfective. In Chapter 3, I discuss in more detail the two types of aspect, provide definitions of the categories of telicity, IMPERFECTIVITY and PERFECTIVITY, and list the test that can be employed to diagnose aspectual properties of predicates. Then I apply these tests to bare imperfectives (i.e., verbs with no aspectual morphology) and perfectives derived by prefixation of the non-quantificational kind, in order to uncover the aspectual contribution of verbal prefixes. I conclude that prefixes are telicity markers. The tests also reveal that the interpretation of perfectives is always PERFECTIVE, while bare imperfectives can have either IMPERFECTIVE or PERFECTIVE interpretation. Based on that I propose that PERFECTIVE viewpoint is marked by zero morphology, and I attribute the fact that prefixed verbs must be PERFECTIVE to the compositional restriction on 11 composing the IMPERFECTIVE with telic predicates, which may be a universal (recall the previously mentioned atelicizing function of the IMPERFECTIVE viewpoint). In Chapter 4, I consider the problem of secondary imperfectivization in Old Church Slavonic (OCS), the oldest Slavic language recorded and our best approximation at what Common Slavic (the parent of all Slavic languages) looked like, Bulgarian and Polish. The first two languages have a separate system of viewpoint aspect marking through the aorist and imperfect morphology for the IMPERFECTIVE and PERFECTIVE viewpoint in the past, respectively, and as such they provide a valuable case study in identifying the contributions of other, pan-Slavic aspectual markers. As it turns out, the use of the secondary imperfective morphology in those languages is independent of viewpoint distinctions. Secondly, the secondary imperfective is applied to prefixed forms in an apparent effort to ―undo‖ the grammatical contribution of the prefix. Given that in Modern Bulgarian and OCS, the role of verbal prefixes is uncontroversially purely Aktionsart in nature (i.e., a separate marker is used to signify PERFECTIVE readings), I conclude that the data justifies an analysis of the secondary imperfective purely in the Aktionsart terms as an atelicizer. This is a novel conclusion that goes against the general claim that the secondary imperfective as a marker of the IMPERFECTIVE viewpoint. I further consider distributional and interpretive data from Polish, which confirms the conclusions drawn on the basis of OCS and Bulgarian. I show that the interpretation of the secondary imperfective forms is atelic and viewpoint-aspect neutral (in the sense of allowing both IMPERFECTIVE and PERFECTIVE interpretation), again in contrast to what is commonly assumed. Consequently, they are exactly like simplex imperfectives, a fact which provides further evidence for viewing the secondary imperfective marker as one 12 that undoes the grammatical contribution of the prefix, i.e., telicity. I formalize this role by interpreting the secondary imperfective as a partitive-homogenizing operator, which applies to a telic event and return an atelic one. This new analysis has not only the advantage of accounting for the common distribution of simplex and secondary imperfectives, but also allows us to understand the presence of secondary imperfective morphology in nominal forms that do not have temporal interpretation but are built on verbal stems that denote activities/process. Finally, this newly proposed function of the secondary imperfective is necessary in the context of Slavic, where the grammatical role of prefixes makes all the verbs derived through prefixation, and therefore the overwhelming majority of Slavic verbs, telic (in contrast to, for example, English particles or verbal prefixes in Romance, which do not uniformly determine the verb‘s Aktionsart in this way). The secondary imperfective marker, as here conceptualized, makes it possible to derive their atelic counterparts. In Chapters 5 and 6, I look at the group of aspectual prefixes in Polish that have a quantificational function. Quantificational verbal prefixes are a pan-Slavic phenomenon, and efforts are typically made to distinguish them from ―regular‖ prefixes. It is normally observed that such prefixes have stable, predictable meanings, can stack on top of other prefixes, do not affect verb valency, and are not involved in marking Aktionsart distinctions. While the first two, possibly three properties do characterize quantificational prefixes in Polish, it is not the case that they are all neutral with respect to telicity. Rather, the distributive quantifier po- and the cumulative quantifier na- do in fact make predicates telic. Only one of the quantificational prefixes, the attenuative po- is telicity neutral. Given that, when stacked, such prefixes must attach to secondarily 13 imperfectivized forms, the Slavic system provides for a semi-recursive (a)telicity marking system, where a predicate can be made telic, then atelic, and then telic again 9 . While this may seem somewhat exotic as compared to aspectual composition in other well studied languages, so is the Slavic aspectual system itself (hence the term ‗Slavic aspect‘), and the proposed structural account is what is needed to account for the Slavic data. At the same time, the Slavic system operates within the constraints imposed by Universal Grammar, and the fact that telicity and atelicity marking, including recursive marking, is available in Slavic, indicates that it is also available universally and that other languages, some of them yet unstudied, can be expected to use systematic morphological marking of telicity and atelicity. Finally, in Chapter 7, the conclusions from the previous chapters are summarized together with a short discussion of issues for further research. 9 Žaucer (2006), based on varying interpretations of the secondary imperfective with stacked prefixes in Slovenian argues that secondary imperfectivization can apply more than once, and the fact that we only see one secondary imperfective marker is due to haplology avoidance. 14 CHAPTER 2: ASPECTUAL MORPHOLOGY IN SLAVIC AND THE PERFECTIVE-IMPERFECTIVE DISTINCTION 1. The Verbal Template in Slavic The Slavic verb form is a complex morphological construct, encoding distinctions in tense, finiteness, agreement, as well as aspect. The generalized verbal morphological template is shown in (1) below, and (2) gives some examples of verb forms in Polish. (1) Verbal Template in Slavic (PREFIX(ES)) + ROOT + (SECONDARY IMPERFECTIVE/ HABITUAL/ SEMELFACTIVE) + THEME VOWEL + TENSE (past vs. non-past)/ INFINITIVE + AGR 1 (2) a. kop-a-ć ‗to kick, to dig‘ (ROOT-THEMEVOWEL-INFINITIVE) b. wy-kop-a-l-i-śmy ‗we kicked out, we dug out - completive‘ (PREFIX-ROOT-THEMEVOWEL-PAST-PLURAL-1PERSON) c. wy-kop-yw-a-ł-em ‗I kicked out, I dug out‘ (PREFIX-ROOT-SECONDARYIMPF-THEMEVOWEL-PAST-1PERSON.SING) d. wy-kop-n-i-Ø-esz ‗you will kick out once‘ (PREFIX-ROOT-SEMELFACTIVE-THEMEVOWEL-PRESENT 2 -2PERSON.SING) 1 The person-number-gender agreement morphemes in Polish are mobile in that they may cliticize to other words in a clause. To keep things simple, in all the Polish examples in this work, the agreement markers will be placed on the verb. 15 There are issues concerning the template in (1), which may be of general interest and helpful in decomposing verb forms, but are not implicated in aspect marking. Those are discussed in Appendix 1 at the end of this chapter. 2. Aspectual Morphology in Slavic and the perfective-imperfective distinction The aspectual contrast in Slavic, commonly referred to as the perfective- imperfective distinction, is encoded in Slavic verbs through the following elements: prefixes, secondary imperfective (henceforth S-imperfective) suffixes, the habitual suffixes (homophonous with the S-imperfective suffixes, but with a more limited distribution), and the semelfactive suffix. The perfective-imperfective distinction is based on the distribution of the relevant forms with respect to a number of syntactic-semantic test that will be discussed later, but it also correlates with morphology in the following ways (non-aspectual morphology is collapsed into V): (3) a. V → imperfective b. PREFIX + V → perfective c. V + SEMELFACTIVE → perfective d. PREFIX + V + S-IMPERFECTIVE/HAB → imperfective The generalizations in (3) have some exceptions. Those concern the statements in (3)a and (3)b. With regard to (3)a, there do exist some bare forms that have the distribution of perfectives. These facts are discussed in detail in Schuyt (1990), who also cites sources giving exhaustive lists of bare perfectives for individual Slavic languages. The fact that such exhaustive lists are possible to make and have been put together, indicates that bare perfectives are felt to be exceptional and in need of enumeration (in 2 The present tense forms of verbs with a prefix and no secondary imperfective suffix, or with a semelfactive suffix, have future tense interpretation. Please see the discussion of the perfective-imperfective distinction in Slavic later in this chapter. 16 contrast to bare imperfectives, which are never listed in this way), just as it is done, for example, for irregular past tense forms in English. The exceptionality of bare perfectives is further confirmed by the fact that they are typically high frequency verbs, and therefore more prone to preserving synchronically idiosyncratic features, e.g., jęti ‗to take,‘ sĕsti ‗to sit down,‘ lešti ‗to lie down,‘ pasti ‗to fall down,‘ rešti ‗to say,‘ dati ‗to give‘ (forms are from Old Church Slavonic, but the same verbs function as bare perfectives in many modern Slavic languages), kaža ‗say,‘ kupja ‗buy,‘ blagoslobja ‗bless,‘ (Bulgarian 3 ), or séči, séže- ‗to reach,‘ séči, séče- ‗to cut,‘ vréči ‗to throw‘ (Slovene). The largest number of bare perfectives is reported to have existed in Old Church Slavonic, with Bulgarian and Macedonian still preserving a relatively large group. In other modern Slavic languages, on the other hand, the number of such forms is very limited, e.g., 9 in Slovene, 5 in Czech, 4 and virtually none in Sorbian (Schuyt 1990) 5 . The exhaustive list of bare perfectives in Polish is given in (4) (cf. Schuyt 1990:188, Czochralski 1972:107 6 ). (4) i. chwycić ‗to seize, to grab‘ ii. chybić (archic) ‗to miss‘ iii. czepić się ‗to grab hold of sth., to cling to‘ iv. dać ‗to give‘ v. kupić ‗to buy‘ 3 Bulgarian does not have an infinitive form. 4 Many formerly bare perfectives in Czech have been reanalyzed as containing the perfective-marking suffix –n. 5 Drawing again on the comparison with irregular past tense forms in English, these numbers are quite small. The count for English irregular verbs stands at about 200. 6 These sources also list a few other forms, but these could arguably be analyzed as diachronically involving prefixation, e.g., obrócić (o- ‗around + brać ‗to take‘?), or starczyć ‗to be sufficient‘ (s- ‗from‘ + tarcza ‗a shield‘?), as also indicated in Schuyt (1990:189). 17 vi. lec (archaic) ‗to lie down‘ vii. paść ‗to fall down‘ viii. puścić ‗to release, to let go‘ ix. rzec (archaic) ‗to say‘ x. ruszyć ‗to set in motion‘ xi. rzucić ‗to throw‘ xii. skoczyć ‗to jump‘ xiii. stawić ‗to place, to put‘ xiv. strzelić ‗to shoot, to fire‘ xv. trafić ‗to reach; to happen upon, to hit‘ The fact that Old Church Slavonic possessed a relatively larger number of bare perfectives, while most modern Slavic languages have over time limited them to just a handful of forms, points to an ongoing grammaticalization process whereby bare forms are uniquely assigned to one aspectual category (i.e., the imperfective), while prefixed forms are assigned to another (i.e., the perfective). The other counterexample to the generalization in (3)a concerns the existence of bi-aspectual bare forms, i.e., bare forms that can alternate between perfective and imperfective uses. These fall into three categories. The first one is common native verbs, and here Bulgarian appears to stand out from other Slavic languages, including its immediate relative, Macedonian, in that it once again allows a much bigger number of such forms as compared to a handful of forms, if any, attested in other languages. The Bulgarian examples include the verbs moga ‗be able to,‘ čuja ‗hear,‘ vidja ‗see,‘ menja ‗change,‘ sipja ‗dissipate,‘ imam ‗have,‘ kašljam ‗cough,‘ bija ‗beat,‘ kalja ‗harden‘ 18 (Schuyt 1990). The second category, and one that is attested more prominently across Slavic, are loan verbs. Those are again very common in Bulgarian, but also in Russian (Gladney 1982, Janda 2007). One can find them in Czech and Polish as well, but there the languages seem to make a greater effort to nativize the vocabulary aspectually and absorb it into the perfective-imperfective system. This is done by creating ―prefixed, explicitly perfective partners to bi-aspectual loan-verbs, [with] the simplex becoming more or less exclusively used as an imperfective‖ (Schuyt 1990:111). Thus, while forms such as demilitarizirovat‟ ‗to demilitarize,‘ evropeizirovat‟ ‗to Europeanize,‘ or gratulirovat‟ ‗to congratulate‘ are bi-aspectual in Russian, in Polish the corresponding forms are imperfective, with the following prefixed perfective counterparts: z- demilitaryzować ‗to demilitarize,‘ z-europenizować ‗to Europeanize,‘ and po-gratulować ‗to congratulate.‘ Finally, the last category of bi-aspectual verbs are performatives, e.g., kanonizować ‗to canonize‘ or koronować ‗to crown/to make a king or queen,‘ or mianować ‗to name (to an office)‘ (all examples from Polish, cf. the Czech data in Adler, 1994). The existence of simplex perfectives and bi-aspectuals does not, in my opinion, invalidate the generalization in (3)a. Rather, both could be analyzed as involving phonologically null ―pure perfectivizers.‖ As I will explain later, ―pure perfectivizers‖ are prefixes that derive perfective forms from bare imperfectives without any lexical meaning change. The form of that prefix for any given verb is not predictable, and has to be memorized by the language learner. In the case of apparent bare perfectives, the learner will conclude that the prefix is phonologically null. The exceptionality of such forms will then lie not in the fact that they are not prefixed, but rather than the prefix in the relevant 19 forms is restricted to a small number of verbs. The situation will be somewhat different in Bulgarian and Russian, where the null prefix is synchronically productive with respect to loan words. This is more expected in Bulgarian, which has a bigger number of apparent bare perfectives, and the null prefix could be viewed as generally productive. For Russian, one would need to assume a separation between native and foreign vocabulary with respect to morphological rules, which happens to be a common cross-linguistic phenomenon (cf. the Germanic / Latinate split in English, McCawley 1968; or Yamato (native) / Sino-Japanese/foreign / onomatopoetic vocabulary split in Japanese, Itȏ & Mester 2002). Another set of counterexamples to (3) concern (3)b. The unexpected pattern this time is the occurrence of forms which are persistently imperfective when prefixed. Two groups of verbs produce this pattern. The first group are some statives, and here the derived prefixed imperfectives could be viewed as fossilization of an older stage in the grammaticalization of Slavic aspect, when, we could speculate, states could not be perfectivized at all. To see that we are dealing with listed exceptions rather than a general rule, consider how unstable this phenomenon was with respect to some verbs already in Old Church Slavonic: (i) always imperfective when prefixed: ležati ‗to lie,‘ sĕdĕti ‗to sit,‘ stojati ‗to stand,‘ (ii) imperfective with some prefixes, perfective with others: drъžati ‗to hold‘ (imperfective except for formations with the prefix u- and vъz- 7 ), imati ‗to have‘ (imperfective only with regard to the form ne NEG -do-imati ‗not to have enough‘), slušati ‗to listen, to hear‘ (formations with prefixes o- and pro- are bi-aspectual), umĕti ‗to be able‘ (imperfective with do- and iz- only), vĕdĕti ‗to know‘ (imperfective except for u- vĕdĕti ‗to get to know, to learn‘), vidĕti ‗to see‘ (imperfective except for obidĕti), and 7 See Section 2 for the list of prefixes and their core meanings. 20 vlasti ‗to dominate, to reign‘ (imperfective with the prefix sъ- only). In Modern Polish, only one verb from this group continues to allow imperfective readings when prefixed, namely leżeć ‗to lie,‘ which has the distribution of an imperfective in semantically opaque formations (i.e., when the meaning could not be computed from the meaning of the verb and the prefix), i.e., za-leżeć ‗to depend,‘ and ‗na-leżeć ‗to belong,‘ but becomes perfective in semantically more transparent combinations, e.g., do-leżeć ‗to lie until,‘ prze-leżeć ‗to lie through (X time)‘ and wy-leżeć się ‗to lie sufficiently long (with reflexive).‘ A more regular and pan-Slavic kind of apparent persistent imperfectivity involves indeterminate verbs of motions, i.e., verbs of motion indicating undirected or habitual movement, in contrast to determinate verbs, which denote a one-time, directed movement. Examples from Polish include chodzić ‗to walk, to go - undirected‘ (cf. iść ‗to go - directed‘), biegać ‗to run - undirected‘ (cf. biec ‗to run - directed), pływać ‗to swim, to move through/on water - undirected‘ (cf. płynąć ‗to swim, to move through/on water - directed‘), and nosić ‗to carry - undirected‘ (cf. nieść ‗to carry – directed‘). These verbs, however, have to be analyzed as involving S-imperfective morphology, which has habitual interpretation when used with unprefixed verbs. This is evidenced in the fact that the undirected motion forms are used to derive S-imperfectives from prefixed verbs of directed motion, e.g., do-płynąć (perfective) → do-pływać (S-imperfective) ‗to get to by swimming,‘ przy-nieść (perfective) → przy-nosić (S-imperfective) ‗to bring,‘ or wyjść (perfective) → wy-chodzić (S-imperfective) ‗to leave, to go out, to exit.‘ Consequently, we have to conclude that it is not that prefixes attach to undirected motion verbs and are unable to make them perfective, but rather that they attach to directed motion verbs, 21 resulting in perfective forms, and can then be secondarily imperfectivized, which is where we get the supposed persistent imperfectives. With all the irregularities described above, the correlation between aspectual morphology and aspectual contrast is very strong in Slavic, indicating a definite grammatical role for prefixes, the semelfactive suffix, and the S-imperfective morphology. Of the three, in this work I consider only the role of verbal prefixes and of S-imperfectivization. The distribution and the semantic function of the semelfactive suffix is set aside here as a topic for further research. In the sections below, I discuss the general characteristics of prefixes and the S-imperfective morphology, mostly on the example of Polish, setting a stage for the analysis of their aspectual function in the subsequent chapters. 3. Verbal prefixes 3.1. General characteristics and prefix list Slavic aspectual prefixes are preposition-like in form and to a degree also in meaning, and are comparable as such to English prepositions in prepositional verbs. They attach to the left of the root, and derive perfective verbs from bare imperfectives (henceforth B-imperfectives), i.e., from stems without any aspectual morphology. The prepositional origin of the aspectual prefixes is evident in the fact that many of them are formally identical to free standing prepositions and may share their meaning. At the same time, verbal prefixes show independence from prepositions in that not all prepositions synchronically have a prefix counterpart, and that they have either developed secondary meanings or have lost their semantic connection to the corresponding preposition altogether. The list in (5) gives all the verbal prefixes used in the aspectual 22 role in Polish, as well as their core meanings (unavailable meanings characterizing the corresponding preposition are indicated with an asterisk; while new meanings not (fully) present in the prepositions are given in italics) (partially based on Polish Dictionary PWN). (5) Aspectual Prefixes in Polish a. Verbal prefixes synchronically corresponding to prepositions: 1. do- ‗to;‘ expresses reaching a goal, adding something 2. na- ‗on;‘ ‗a lot, to satisfaction‟ 3. nad- ‗over, above;‘ affects the quantity the object, e.g., nad-gryźć „to take a bite of sth‟ cf. gryźć „to bite, to chew‘ 4. o- ‗about, around;‘ directs the activity downward or backward 5. od- ‗from, away, away from;‘ back, indicates undoing sth, taking sth away, or re-doing sth, gaining sth‘ 6. po- ‗over;‘ ‗some, a little,‟ DISTRIBUTIVE marker, inceptive wrt verbs of motions and states (*after) 7. pod- ‗under, below;‘ up to, indicates causing sth, increasing intensity or partial fulfillment of a goal 8. przed- ‗before‘ 9. przy- ‗at, near,‘ indicates reaching a spatial goal, affixing sth 10. u- ‗away;‘ manage to; used also to derive denominal and deadjectival verbs indicating making sth have the property of noun/adjective (*at) 11. w(e)- ‗in‘ 23 12. z(e)-/s- ‗from, with;‘ indicates completion (default pure perfectivizer in Polish – see section 3.3 below) 13. za- ‗behind, for;‘ inceptive marker b. Verbal prefixes that cannot synchronically function as prepositions: 14. ob- around (historically an allomorph of o-) 15. prze- through, over (cf. preposition przez ‗through, by,‘ not used as a prefix) 16. roz- indicates spreading, separating, distributing 17. wy- out 18. wz- upward, indicates increase or intensification, used with some verbs as inceptive The exact set of aspectual prefixes as well as their phonological form can vary from one Slavic language to another, but the differences are minimal 8 . All Slavic languages have by and large preserved the set of prefixes that were used in aspectual function at the time the earliest Old Church Slavonic texts were recorded (10-11 th century AD). Those were (glosses are in Latinized script with ъ representing an ultrashort vowel): do-, iz-, na-, nadъ-, niz-, o(b)-, ot(ъ)-, po-, podъ-, prě-, prědъ-, pri-, pro-, raz-, sъ(n)-, u-, vy-, vъ(n)-, vъz- and za- (Brajerski 1993). The core meanings have also remained the same. It should be stressed at this point, that the core meanings listed above, whether of the prepositional or non-prepositional kind, cannot account for all uses of verbal prefixes. More specifically, prefixes can sometimes be used simply to mark perfectivity, in which case the non-aspectual meaning of the prefix is bleached. Further, over the years, some prefix- 8 For a comparative list of prefixes in different Slavic languages, please see Svenonius (2004), in particular tables (26) on p. 194, and (28) on p. 195. 24 V combinations have acquired meanings that are completely opaque, and so could not be computed from the meaning of the root and the prefix as listed in (5). Some examples of that will be provided in the subsequent section. As a consequence, cognate formations in various Slavic languages do not always mean the same thing, and so the reader should not assume that the same exact prefix would be used with a particular root to derive a particular idiomatic meaning in every Slavic language. Before we move on to the discussion of prefixes with various lexical import, it needs to be clarified that not all verbal prefixes have an aspectual function in Slavic. In Polish, borrowed prefixes, such as kontr- ‗contra-,‘ re- ‗re-,‘ and dez- ‗dis-‗ are aspectually inert. Also, the Polish native prefix wspόł- ‗together,‘ translatable into English as co-, as in coexist, does not change the aspectual value of the verb to which it attaches. In that last case, one could argue that wspόł- is different from other prefixes in that it is more a bound content morpheme than a prefix, as it functions as the root in the adjective wspól-ny ‗common,‘ which in turn gives rise to wspól-n-ota ‗community‘. If this is the correct classification, we would preserve the generalization that all native verbal prefixes in Slavic have an aspectual function. 3.2. Meaning-modifying/changing prefixes The list in (6) provides combinatorial possibilities of the Polish verb pisać ‗to write‘ with different aspectual prefixes and the resultant meanings, while (7) and (8) illustrate the varying lexical contributions of the prefixes do- ‗to‘ and prze- ‗through‘ with respect to different verbs. It should be made clear that the fact that a given combination is not attested in Polish does not mean that it is also impossible in other Slavic languages. To the contrary, such ‗paradigm gaps‘ are typically idiosyncratic to 25 individual Slavic languages, and corresponding cognate forms can be found. This in turn indicates that such gaps are accidental and not due to some semantic incompatibility between the root and the prefix. (6) PREFIX + pisać ‗to write‘ a. do-pisać ‗to write sth additional; to be favorable‘ b. na-pisać ‗to write up‘ c. (nad-pisać) not attested d. o-pisać ‗to describe‘ e. od-pisać ‗to write back; to write off‘ f. po-pisać ‗to write over sth; to write a little; to write-distributive,‘ ‗to show off (with the reflexive)‘ g. pod-pisać ‗to sign‘ h. prze-pisać ‗to copy; to prescribe; to rewrite; to sign sth over to sb‘ i. (przed-pisać) not attested j. przy-pisać ‗to ascribe‘ k. roz-pisać ‗to make a schedule; to make a dry pen write; to write extensively (with the reflexive się) l. (u-pisać) not attested m. w-pisać ‗to write in; to sign up‘ n. wy-pisać ‗to write out‘ o. (wz-pisać) not attested p. s-pisać ‗to copy down,‘ ‗to come up/through (with the reflexive)‘ q. za-pisać ‗to write down; to prescribe‘ 26 (7) do- ‗to‘ a. dać ‗to give‘ → dodać ‗to add‘ b. pisać ‗to write‘ → dopisać ‗to add by writing‘ ‗to be favorable (of weather, health) c. biec ‗to run‘ → dobiec ‗to reach by running or by sound (of news, noise)‘ d. kończyć ‗to finish‘ → dokończyć ‗to finish up‘ e. chować ‗to hide‘ → dochować ‗to keep (secret, faith)‘ f. chować się ‗to hide self‘ → dochować się ‗to have offspring‘ g. stać ‗to stand‘ → dostać ‗to stand to the end; to get‘ h. stać się ‗to become‘ → dostać się ‗to get into‘ i. znać ‗to know‘ → doznać ‗to experience‘ (8) prze- ‗through, over‘ a. ciąć ‗to cut‘ → przeciąć ‗to cut through‘ b. pisać ‗to write‘ → przepisać ‗to copy; prescribe; to rewrite; to sign sth over to sb‘ c. biec ‗to run‘ → przebiec ‗to run through, to run across‘ d. mówić ‗to speak‘ → przemówić ‗to speak/utter words, to give a speech‘ e. chować ‗to hide‘ → przechować ‗to keep, to store‘ f. malować ‗to paint‘ → przemalować ‗to paint over (to a different color)‘ g. stać ‗to stand‘ → przestać ‗to stop, to cease; to stand through‘ h. żyć ‗to live → przeżyć ‗to experience; to live through; to survive‘ The examples above illustrate that the lexical contribution of the prefix can be compositional (i.e., the meaning is computed from the meaning of the root and the available meanings of a given prefix), but it can also be idiosyncratic, as previously 27 indicated. One would expect that meaning would always be compositional for any new formations, e.g., those involving new verbs such as klikać ‗to click‘ or blogować ‗to blog,‘ as such forms are too recent to be idiomized. This is, indeed the case, as illustrated in (9) and (10), which show possible combinations and their meanings, as based on native speaker intuitions and internet usage via Google search. (9) klikać ‗to click‘ (computerese) a. do-klikać ‗to get to by clicking, to add by clicking‘ b. na-klikać ‗to click a lot,‘ ‗to click on‘ c. (nad-klikać) not attested d. o-kilkać ‗to click all around‘ e. od-kilkać ‗to click back, to undo by clicking‘ f. po-klikać ‗to click for a little while‘ g. pod-klikać ? ‗to click under,‘ (very few examples) i. (przed-klikać) not attested j. przy-klikać ‗to get to by clicking‘ k. u-klikać ‗to (manage to) get by clicking‘ l. w-klikać ‗to click in, to put in by clicking‘ m. z-klikać ‗to bring together by clicking,‘ ‗to copy by clicking‘ n. za-klikać ‗to click – inceptive,‘ ‗to cover by clicking‘ o. prze-klikać ‗to spend time clicking; to transfer by clicking‘ p. roz-kilkać ‗to go on for too long clicking (with the REFL się),‘ ‗to open the contents of a directory,‘ ‗to distribute by clicking‘ q. wy-klikać ‗to produce sth.with clicking, to get sth. out with clicking‘ 28 r. (wz-klikać) not attested (10) blogować ‗to blog‘ (computerese) a. do-blogować ‗to hit sb. (metaphorically) by blogging‘ (no examples found) b. na-blogować ‗to blog a lot‟ c. (nad-blogować) not attested d. o-blogować ‗to blog about‘ e. od-blogować ‗to blog back,‘ ‗to sign off from a blog,‘ ‗to take a break in blogging‘ f. po-blogować ‗to blog for a little while‘ g. pod-blogować ? ‗to send by blogging‘ (1 example found) i. (przed-blogować) not attested j. przy-blogować ‗to add to a blog‘ k. u-blogować ‗to manage to blog sth (e.g., a story, 4 pages, etc.)‘ l. w-blogować ‗to blog in‘ m. z-blogować ‗to blog (as pure perfectivizer), ‘put into a blog‘ n. za-blogować ‗to blog – inceptive‘ o. prze-blogować ‗to spend time blogging‘ q. roz-blogować ‗to go on for too long blogging (with the REFL się),‘ ‗to spread the habit of blogging‘ r. wy-blogować ‗to produce sth.with blogging, to blog out‘ s. (wz-blogować) not attested 29 As the forms above illustrate, most prefixes can combine productively with new verbs. Those that could not attach to the verbs in (9) and (10), might be unproductive or not fully productive in Modern Polish, though more verbs would need to be considered to make this conclusion definite. 3.3. Purely aspectual prefixes Just as the meaning of already established prefixed forms is often idiosyncratic, so it is often not predictable which prefix will serve as the so-called ―pure perfectivizer.‖ I will use the term ―pure perfectivizer‖ to refer to prefixes that do not contribute any lexical meaning to the verb and have only aspectual function (see example (6)-b above), something that can be compared to the use of the English particle up in drink up, wash up or wake up. In glosses, I will represent this prefix use as PREF. Typically Polish verbs have one such prefix each 9 , and it is not an easy guess what prefix it would be for a given verb. There are some sub-regularities, for example the frequent use of the prefix po- with psychological verbs, which gives them the inceptive reading (e.g., po-kochać ‗to start to love,‟ po-lubić ‗to start to like,‘ po-czuć ‗to get a feeling‘), as well as uses that can be attributed to the core meaning of a given prefix, as in the use of the prefix na- ‗on‘ with verbs denoting an activity involving putting something on a surface with a writing/painting instrument, such as na-pisać ‗to write up,‘ na-malować ‗to paint up,‘ na- szkicować ‗to sketch up,‘ or na-bazgrać ‗to scribble up.‘ It is sometimes suggested that all prefixes have a semantic content, and in the case of ―pure perfectivizers‖ it is simply not easily perceived because of the overlap with verb meaning (e.g., Isačenko 1962, 9 Occasionally, there may be two prefixes that appear purely aspectual, e.g., płowieć → s-płowieć/wy- płowieć ‗to lose color,‘ brudzić → po-brudzić/u-brudzić ( (Laskowski 1998; the source has za-brudzić as an alternative for po-brudzić, but this form can be shown not to involve a pure perfectivizer through the secondary imperfectivization test, to be discussed later). Other verbs may lack a prefixed form with identical lexical meaning altogether, e.g., spać ‗to sleep,‘ czołgać się ‗to crawl.‘ 30 Czochralski 1975). However, there is plenty of evidence against this claim. First, the choice of the prefix is often simply not explainable in this way, e.g., kończyć ‗to finish‘ has s-kończyć as its perfective counterpart, and not, for example, do-kończyć ‗TO-end,‘ which would seem semantically motivated; kochać się – the reflexivized version of ‗to love‘ with the meaning of ‗to be in love,‘ has za-kochać się ‗to fall in love‘ and not po- kochać się, which would correspond to the use of po- in the unreflexivized po-kochać ‗to start to love.‘ Secondly, different meaning nuances of the same verb may involve the use of different ―purely perfectivizing‖ prefixes, e.g., krzyżować → s-krzyżować ‗to cross,‘ krzyżować → u-krzyżować ‗to crucify,‘ and krzyżować → po-krzyżować ‗to thwart, to hinder.‘ Thirdly, cognate verbs in different Slavic languages can take different prefixes to form their perfective counterparts, for example: (11) a. to call (on the phone) dzwonić → za-dzwonić Polish zvonit‘ → po-zvonit‘ Russian b. to drink pić → wy-pić Polish piti → po-piti Serbian All these facts underscore the idiosyncratic nature of ―pure perfectivizers‖ and indicate that their form for a given root must be listed, just like any prefix-V combination with opaque meaning. 3.4. Quantificational prefixes Beside lexical meaning-changing prefixes and ―pure perfectivizers,‖ another group of prefixes need to be distinguished, namely the quantificational ones. In Polish, 31 these are na- with the meaning of ‗a lot, satisfactorily,‘ po- with the meaning of ‗a little, some,‘ and po- as a distributive marker. These prefixes are also perfectivizing, but differ from others in that they are always semantically transparent, have quantificational properties, and can stack on top of already prefixed verbs. I will postpone a more detailed description of these prefixes until Chapter 5 and 6, where their properties will be considered at length. 4. The secondary imperfective morphology As its name indicates, secondary imperfective morphology creates derived imperfective forms. It attaches to prefixed verbs, which are perfective, and returns forms that pattern with B-imperfectives. The morphology used in S-imperfectivization is quite different from prefixation. Semantically, it never changes the verb‘s lexical meaning, and consequently, it is a more pure grammatical marker. Morphologically, the elements deriving S-imperfective forms are suffixes, and their number is substantially smaller, which goes together with their strictly functional role. Slavic languages may differ in the exact phonological form of the formatives used to derive S-imperfectives and with respect to their relative productivity. In Old Church Slavonic and Old Russian, the most frequent type was the suffix –a, which was often applied to verbs taking the theme vowel –i (the presence of the thematic i, though not overt, is felt in the consonantal alternations, e.g., vъz-glasiti/vъz-glašati ‗to proclaim‘), and produced vowel alternations in the root (o→a, e→ĕ, ъ→y, and ь→i). The suffix –a had also a variant in –va, which has its origins ―in roots in *-u- (e.g., OCS pokryti/pokryvat [‗to cover‘]), and was extended to roots and stems in other vowels (e.g., OCS uspĕti/uspĕvati [‗to benefit‘], otъvĕštati/otъvĕštavati [‗to reply‘], ubiti/ubivati [‗to 32 kill‘].‖ (Schuyt 1990:7). Much less common in these two historical languages was the derivation with the suffix –ova with the allomorph –uje (previously mentioned in reference to the Slavic verbal template), which is arguably –a augmented with –ov, and which served as an alternative for certain forms where the difference between the perfective and the S-imperfective derived with –a suffixation, might have been felt to be not sufficiently pronounced (e.g., because the root vowel was originally a or ĕ, and so the suffix –a did not effect any mutations, e.g., izvĕstiti/izvĕštati, izvĕštovati ‗to assure, convince‘). This –ova/-uje suffix has subsequently developed into the most productive of the S-imperfective morphemes in modern Slavic languages (Schuyt 1990), but the derivation with the –a suffix has also been preserved. With respect to phonological changes, the most pronounced ones concern the form of the suffix –ova in its S-imperfective uses. In Bulgarian and Macedonian, its form is now -(u)va, in Serbo-Croat and Russian it is –iva, in Slovene, Czech and Slovak it continues to be –ova, but Slovak has also a variant in –(j)eva, while in Polish it has the form –ywa (pronounced as [ɨva]). This contrasts with the previously mentioned use of the –ova suffix as a verbalizer, which has not been subject to such changes. The examples below from Polish illustrated the general two types of morphology employed in deriving S-imperfectives in Slavic, as well as yet unmentioned type, namely suppletion, which is very limited, but needs to be acknowledged: (12) a. –ywa suffix (historically –ova): prze-pisać → prze-pis-yw-ać ‗to copy‘ od-czytać → od-czyt-yw-ać ‗to read out‘ 33 b. -a suffix with vowel mutation wy-robić → wy-rabi-a-ć ‗to manufacture‘ wy-jeść → wy-jad-a-ć c. suppletion przy-jść → przy-chodzić ‗to come‘ za-łożyć → za-kładać ‗to put on‘ 5. The Slavic perfective-imperfective distinction: the diagnostics As I mentioned before, the perfective-imperfective distinction, while it has clear morphological correlates, is based on the distribution of verbs with respect to a number of syntactic-semantic tests. Some of these tests are listed in (13)-(16) below. The illustrative examples provided for each test will be, for the reasons of space, limited to the following forms: a B-imperfective, a perfective with a ―pure perfectivizer,‖ a perfective with lexical content (a different one each time to show that the tests work for all prefixes), and a S- imperfective form. (13) Present versus future denotation for present tense forms GENERALIZATIONS: 1. Imperfective verbs in the present tense forms have present tense interpretation. 2. Perfective verbs in the present tense forms have future tense interpretation. EXAMPLES: a. Hania pisze listy. Hannah.NOM writes.B-IMPF letters.ACC ―Hannah is writing/writes letters.‖ b. Hania na-pisze listy. 34 Hannah.NOM PREF-writes.PERF letters.ACC ―Hannah will write (the) letters.‖ c. Hania od-pisze na listy. Hannah.NOM from-writes.PERF on letters ―Hannah will reply to (the) letters.‖ d. Hania od-pis-uj-e na listy. Hannah.NOM from-writes.SI on letters ―Hannah replies/is replying to (the) letters.‖ (14) Analytic future forms with the future form of the verb BE.FUT + V in the infinitive or past tense form GENERALIZATIONS: 1. Imperfective verbs can be part of the analytic future constructions. 2. Perfective verbs cannot be part of the analytic future constructions. EXAMPLES: a. Hania będzie pisać / pisała listy. Hannah.NOM be.FUT write.B-IMPF.INF / wrote.B-IMPF letters.ACC ―Hannah will be writing letters.‖ b. *Hania będzie na-pisać / na-pisała listy. Hannah be.FUT PREF-write.PERF.INF / PREF-wrote.PERF letters.ACC ―Hannah will be writing letters.‖ c. *Hania będzie prze-pisać / prze-pisała listy. Hannah.NOM be.FUT through-write.PERF.INF / through-wrote.PERF letters.ACC ―Hannah will be copying letters.‖ 35 d. Hania będzie prze-pis-yw-ać / prze-pis-yw-ała listy. Hannah.NOM be.FUT through-write.SI.INF / through-wrote.SI letters.ACC ―Hannah will be copying letters.‖ (15) Compatibility with phase verbs such as begin or end GENERALIZATIONS: 1. Imperfective verbs can be complements to phase verbs. 2. Perfective verbs cannot be complements to phase verbs. EXAMPLES: a. Hania zaczęła pisać listy. Hannah.NOM began write.B-IMPF.INF letters.ACC ―Hannah began to write (the) letters.‖ b. *Hania zaczęła na-pisać listy. Hannah.NOM began PREF-write.PERF.INF letters.ACC ―Hannah began to write (the) letters.‖ c. *Hania zaczęła wy-pis-ać listy. Hannah.NOM began out-write.PERF.INF letters.ACC ―Hannah began to write out (the) letters.‖ d. Hania zaczęła wy-pis-yw-ać listy. Hannah.NOM began out-write.SI.INF letters.ACC ―Hannah began to write out (the) letters.‖ (16) Compatibility with the V 1 and V 1 ‗on and on‘ construction indicating prolonged length and continuity or repetition (Adler 1994) GENERALIZATIONS: 36 1. Imperfective verbs can be part of the V 1 and V 1 construction. 2. Perfective verbs cannot be part of the V 1 and V 1 construction. EXAMPLES: a. Hania pisała i pisała numery. 10 Hannah.NOM wrote.B-IMPF and wrote.B-IMPF numbers.ACC ―Hannah wrote and wrote (the) numbers.‖ b. *Hania na-pisała i na-pisała numery. Hannah.NOM PREF-wrote.PERF and PREF-wrote.PERF numbers.ACC ―Hannah wrote and wrote the numbers.‖ c. *Hania za-pisała i za-pisała numery. Hannah.NOM INCPT-wrote.PERF and INCPT-wrote.PERF numbers.ACC ―Hannah wrote down and wrote down the numbers.‖ d. Hania za-pis-yw-ała i za-pis-yw-ała numery. Hannah.NOM INCPT-wrote.SI and INCPT-wrote.SI numbers.ACC ―Hannah wrote down and wrote down (the) numbers.‖ There are also other tests which are more language specific, showing differences between Slavic languages in whether or not imperfective and perfective verbs can give rise to certain participial formations or deverbal nouns. In Polish, for example, adjectival active participle and the present/concurrent adverbial participle (i.e., the participle used in the adverbial function, corresponding to the English –ing participle, as in (While) reading this book, I realized…) are derived only from imperfectives, while the adverbial past/prior participle (corresponding to the English ‗have + past participial‘ construction, as in 10 In this construction, only one overt argument is possible on the intended reading. Here it is placed after the second V, but it could also be located after the first V. 37 Having read this book, she decided…) is derived only from perfectives. The passive adjectival participle can be built on both perfectives as well as imperfectives. The derivational possibilities for different participles in Polish are shown in (17). (17) a. adjectival active participle imperfectives: piszący ‗writing‘ grający ‗playing‘ perfectives: *na-piszący ‗writing-completive‘ *prze-piszący ‗copying, re-writing‘ *za-piszący ‗writing down, enrolling‘ *za-grający ‗playing-completive‘ *wy-grający ‗winning‘ *prze-grający ‗losing‘ b. passive adjectival participle imperfectives: pisany ‗being written‘ grany ‗being played‘ perfectives: na-pisany ‗written‘ prze-pisany ‗copied, re-written‘ za-pisany ‗written down, enrolled‘ za-grany ‗played‘ wy-grany ‗won‘ prze-grany ‗lost‘ c. adverbial present/concurrent participle imperfectives: pisząc ‗writing‘ 38 grając ‗playing‘ perfectives: *na-pisząc ‗writing-completive‘ *prze-pisząc ‗copying, re-writing‘ *za-pisząc ‗writing down, enrolling‘ *za-grając ‗playing-completive‘ *wy-grając ‗winning‘ *prze-grając ‗losing‘ d. adverbial past/prior participle imperfectives: *pisawszy ‗having been writing‘ *grawszy ‗having been playing‘ perfectives: na-pisawszy ‗having written‘ prze-pisawszy ‗having copied, having re-written‘ za-pisawszy ‗having written down, having enrolled‘ za-grawszy ‗having played‘ wy-grawszy ‗having won‘ prze-grawszy ‗having lost‘ In Russian, the pattern is the same, only this time it extends to bigger number of participles, as Russian has a full paradigm of adjectival participles: present active, present passive, past active and past passive. In that paradigm, the imperfective-perfective distinction is aligned perfectly with the present-past distinction, i.e., present participles are uniformly derived from imperfectives while past participles are derived from perfectives. 39 As for the derivation of deverbal nouns, West Slavic languages such as Polish or Czech allow both perfective and imperfective verbs to give rise to –ing type nominals, while in East and South Slavic, e.g., in Russian and Bulgarian, the input to such formations is typically limited to imperfective verbs. This contrast is illustrated for Polish and Bulgarian in (18). (18) Polish Bulgarian a. B-imperfective: pisanie pisane ‗writing‘ b. perfective: na-pisanie *na-pisane ‗writing-completive‘ c. perfctive: od-pisanie *od-pisane ‗writing back-compl‘ d. S-imperfective: od-pis-yw-anie od-pis-va-ne ‗writing back‘ The varying nature of the tests for the Slavic perfectivity and imperfectivity does not give a clear picture as to what kind of aspectual property governs the different distribution of perfective and imperfective verb, i.e., whether it is viewpoint aspect, or Aktionsart. Certainly, many of the diagnostics seem to be related to the temporal properties of perfective and imperfective forms, and more specifically to whether something can or cannot be viewed concurrently with the time of speaking or with the temporal reference of the main event, as is the case with the present vs. future time denotation test, and with the participle formation, respectively. As such, those tests would seem to point in the direction of viewpoint aspect, which is a temporal category. This is also the interpretation of the relevant tests as proposed in e.g., Borik (2002), Pereltsvaig (2005) or Mezhevich (2008). Other tests, however, (might) point in a different direction. The analytic future – be.FUT + V – is limited to imperfective forms, which could be attributed to the fact that perfectives already get the future interpretation in their basic 40 present tense form, but it could also be that the stativizing copula requires a homogenous, and therefore atelic, complement on the active (as opposed to passive) reading. The V 1 and V 1 ‗on and on‘ construction also appears to require that the co-ordinate predicates be atelic, as indicated, for example, by the impossibility of applying the ‗on and on‘ adverbial to telic predicates in English (e.g., *He fell on and on vs He sang on and on). The difference in telicity likely underlies also the restriction in the formation of deverbal –ing-like nominals in East and South Slavic, as they do not appear to have viewpoint aspect interpretation, but could be construed as naming activities only (as opposed to accomplishments). The mixed nature of the tests that distinguish perfective verbs from imperfective ones indicates that analyzing the Slavic aspectual contrast in terms of a single feature may not be possible. Instead, both kinds of aspect – Aktionsart and viewpoint – will likely need to be engaged to provide an explanatorily adequate account for the distribution of perfective and imperfective forms. In the next two chapters, I consider how the two types of aspect interact in B-imperfective and prefixed perfective forms, and in S- imperfectives. 41 Appendix 1 1. Morphology missing from the template in (1) There are two pieces of morphology not represented in the template in (1) because they are neither obligatory for all verbs nor optional in the sense of being able to attach freely to any root. 1.1. -ow The first one is the suffix –ow in Polish (and its variants in other Slavic languages) 11 , which follows the root and precedes the S-imperfective/habitual marker, and which is found obligatorily on some verbs. There is no simple way to characterize the role of this morpheme, which occurs in verbal as well as adjectival forms and their derivates. It is required to derive verbs from an overwhelming majority of roots of foreign origin, e.g., koszt-ow-a-ć ‗to cost‘ (cf. *koszt-a/i/etc-ć), or kalkul-ow-a-ć ‗to calculate‘ (cf. kalkul- a/i/etc-ć), amerykaniz-ow-a-ć ‗to Americanize‘, or identyfik-ow-a-ć ‗to identify‘ (the latter two roots are morphologically complex in the source language, English, and even include verbalizing suffixes, but that information is clearly not accessible to Polish morphology), but is also found with some native roots, e.g., mal-ow-a-ć ‗to paint,‘ chor- ow-a-ć ‗to be ill,‘ bud-ow-a-ć ‗to build.‘ In addition, -ow is used to derive verbs (and adjectives) from nouns, e.g., głów-k-ow-a-ć ‗to rack one‘s brain‘ from głów-k-a little head‘ (k=dimunitive suffix), dar N -ow-a-ć ‗to gift‘ (cf. da-ć ‗to give‘), bed-stv-ova-t‟ ‗to live in poverty‘ (Russian), or u-pań-stw-ow-i-ć ‗to make into a state/of a state‘ via the adjectival derivate pań-stw-ow- ‗of the state.‘ This latter use has prompted the occasional 11 Uniquely in Bulgarian and Macedonian, in addition to this suffix, which takes the form of –yv, -ov and – v, the suffix –ir is used in the same function with loan verbs of Germanic origin (cf. Ger. –ieren).as well as with a handful of native roots (e.g., obiskiram ‗search,‘ skladiram ‗store,‘ dogovoriram ‗agree‘). Historically, these verbs originated as involving the –ov suffix added onto the loan stem containing –ir, but towards the end of the 19 th century, -irov was reduced to –ir ( Maslov 1963, Schuyt 1990:44). 42 characterization of –ow as a denominalizer, while its addition to roots have been attributed to its role as a verbalizer on a par with theme vowels. However, -ow does not take the place of a theme vowel in a verb, but rather co-occurs with it: typically with the theme vowel –a, and in transitive deadjectival verbs with –i (see fn. 2 below). Its role, therefore, seems to be different and more general, likely predicative. It could also be speculated that it marks duration, as it is incompatible with the semelfactive suffix –n. Further, in Old Church Slavonic (the oldest recorded Slavic language) -ow was one of the suffixes used to derive the S-imperfective forms, and, as stated in Schuyt (1990), ―has developed into the most productive secondary imperfective type of the modern Slavic languages‖ (Schuyt 1990:7). In Polish, from which most of the data considered in this work will be derived, -ow incidentally does not have this role, save for a couple counterexamples, and so I will set the problem of its multi-faceted function and complicated distribution aside, with the understanding that in a different Slavic language it may play a more prominent role in aspectual marking. 1.2. The second morpheme not represented in (1) is the suffix –n, homophonous with the semelfactive suffix but clearly with a different function. Jabłońska (2004) calls this suffix inchoative (she analyzes it as a unit together with the vowel that always follows it). However, what it has in common with inchoatives is only the lack of an external causer. Otherwise it basically derives degree achievement verbs with atelic interpretation, e.g., marz-n-ą-ć ‗to get cold,‘ sch-n-ą-ć ‗to get dry, chud-n-ą-ć ‗to lose weight,‘ kwit-n-ą-ć ‗to bloom,‘ więd-n-ą-ć ‗to wither.‘ (The same holds for Russian, but in Czech, -ną equivalent nou can also be used synchronically in transitive contexts, (Newman & Towsend 1972). In terms of its position, it would belong in the secondary imperfective/ 43 habitual/semelfactive column, as it does not co-occur with any of these markers. This suffix does not have an aspectual function. 2. Theme Vowels Another issue in need of some explanation is the theme vowels, which are not considered aspect markers per se, but which do sometimes participate in aspectual alternations. The main role of theme vowels is to decide the conjugational class of the verb. In Polish, there are five such vowels: -a, -e, -a(j)), -i, -ej/ew, plus, for a very limited number of verbs, a phonologically null element is assumed. The semelfactive marker –ną and the homophonous degree achievements marker described above are also typically listed in this category, alongside –owa. –Owa is clearly bi-morphemic and the –ow portion does not belong in this group, as explained earlier. As for the semelfactive and the degree achievement suffixes, the matter is more complicated, because they are not compatible with the secondary imperfective and habitual suffixes, but participate in alternations with other theme vowels. My inclination would be to split them into –n, marking semelfactivity and gradual change, and the theme vowel, but the matter clearly calls for more consideration. The selection of a particular theme vowel for a given verb stem can be motivated morphologically (when they attach to already suffixed stems), be idiosyncratic (when they attach to roots), but can also reflect structural properties. For example, in Polish and Russian -ej typically derives unaccusative verbs (and was historically related to stative readings), while -i is found on transitive or unergative verbs (Svenonius 2004a, Jabłońska 2004). Also, sometimes the same root can take two different theme vowels, with the alternative vowels marking the causative-inchoative alternation (Svenonius 2004, 44 Milićević 2004, Jabłońska 2004). Attributing a clear semantic role to theme vowels may be less applicable to Czech and Slovak, for example, where theme vowels have lost many of older functions and serve mostly as a syllabic element (Isačenko 1966, Newman & Towsend 1972). Theme Vowels can be thought of as verbalizers (for examples of the implementation of this idea in Polish, see (Czaykowska-Higgins 1997 and Jabłońska 2004) and/or verbal classifiers (Isačenko 1966). It needs to be noted, however, they are not all verbalizers in the derivational sense (i.e., being able to change a category X into a V). Only some of the theme vowel (-i and –e, specifically) appear to have this role, otherwise the addition of the –ow suffix is required prior to the addition of the theme vowel. I would suggest that the proper characterization of theme vowels is that they are overt realizations of a functional head in the extended verbal projection. In other words, theme vowels are overt reflexes of verbalization through structure, rather than verbalization by a categorizing morpheme (cf. Borer 2005). This conclusion is supported by a number of facts. First, theme vowels are external to all aspectual morphology, if distinct aspectual morphology is present, and precede only tense and agreement morphemes, so they are clearly related to a (higher) functional head, and not to V. Second, theme vowels show up only on forms that function in a clause as verbs, and do not show up on words that may have a verb in their derivational history, but are ultimately of a different category. Lastly, theme vowels can co-exist with the overt verbalizing /predicative suffix –ow. All these facts indicate that the role of the theme vowels is not to verbalize its input, but rather to indicate the presence of certain functional structure. I will return to the matter of theme vowels briefly when discussing the morphology of S-imperfectivization. 45 CHAPTER 3: THE ASPECTUAL ROLE OF VERBAL PREFIXES 1. Two Types of Aspect In this chapter, I consider the aspectual contrast between B-imperfectives and prefixed perfectives in Polish, in an effort to determine the exact grammatical role of Slavic verbal prefixation. As I stated in the introductory chapter, I assume a two-tiered theory of aspect, where aspectual composition involves two kinds of aspectual information: Aktionsart (or inner aspect), and for the purposes of this work more narrowly telicity, and viewpoint aspect (or outer aspect) (Smith 1991). These are understood to be independent of each other both semantically (as defined below) and structurally, the latter meaning essentially that they are syntactically encoded through separate, dedicated projections. 1.1. Aktionsart and Telicity: definitions and tests 1.1.1. Telicity as non-homogeneity In most general terms, Aktionsart provides information about the internal structure of an event. Semantically, it is a predicate of events specifying the event type (Paslawska & von Stechow 2002, Kratzer 2004, Pancheva 2008): [[ AKTIONSART ]] = λe (v) .P(e). Following Vendler (1967), four such event types are normally distinguished: states, activities, accomplishments and achievements. These types are, however, not primitives, but rather represent combinations of Aktionsart features such as telicity, durativity and dynamicity. Table 1 below defines individual event types in terms of these three properties. 46 Table 1: Aktionsart properties and event types Telic dynamic durative States - - + Activities - + + Accomplishment + + + Achievement + + - Of the three properties, telicity is the one most frequently discussed, and has been over the years defined in different ways, from the presence of an inherent end point (telos) (Garey 1957, Comrie 1976), through event-argument homomorphism (Dowty 1991, Krifka 1998, 1989, 1992, Ramchand 1997, Tenny 1992, 1994), to atomicity (Kiparsky 1998, Rothstein 2004), to non-homogeneity/quantity (Borer 2005). Here I will assume the non-homogeneity definition proposed in Borer (2005) 1 , which in turn is based in part on earlier proposals by Krifka (1992) and Kiparsky (1996, 1998), and keeps with the long tradition of comparing the telic-atelic distinction in the verbal domain to the count-mass distinction in the nominal domain. 2 On that view, telic predicates (Quantity predicates in Borer‘s terminology) are those predicates which are not homogenous, where homogeneity is defined as a property of being simultaneously cumulative and divisive, as given in (1). 1 This choice of the homogeneity view is also a choice against the often assumed quantization approach of Krifka (1998). The problems with the latter are outlined in detail in Borer (2005: 143-9). 2 The parallel between the mass-count/quantity distinction in the nominal domain and the atelic/process- telic/event distinction in the verbal domain has been discussed by many authors, beginning with Bach (1986), who based on the proportion events:processes = things:stuff, proposed that events are analogous to singular and plural individuals, while temporally delimited processes (i.e., PERFECTIVE processes, on the terminology used here) are analogous to portions of matter. 47 (1) a. A predicate P is homogenous iff P is cumulative and divisive. (Borer 2005: 147) b. P is cumulative iff ∀x [P (x) ∧ P (y) → P (x ∪ y)]. “A predicate is cumulative iff whenever it holds of two arguments, it holds of their union as well.” c. P is divisive iff ∀x [P (x) → ∃y (P (y) ∧ y < x)] ∧ ∀x,y [P (x) ∧ P (y) ∧ y < x → P (x−y)]. “A predicate if divisive iff the following holds: For any argument that P is true of there is a part of that argument that P is also true of and in addition for any argument and part of it P holds true of, P is true also of the difference between the argument and its part.” If a predicate meets both the cumulativity and the divisity requirements, that predicate is atelic. Conversely, a predicate which is not cumulative or not divisive is telic. To see how these diagnostics work in practice, consider the following examples: (2) a. John read. i. cumulativity ‗John read‘ + ‗John read‘ = ‗John read‘ – cumulative ii. divisiveness A part of ‗John read‘ = ‗John read‘ – divisive > It follows that ‗read‘ is atelic. b. John read books. i. cumulativity ‗John read books‘ + ‗John read books‘ = ‗John read books‘ – cumulative 48 ii. divisiveness A part of ‗John read books‘ = ‗John read books‘ – divisive > It follows that ‗read books‘ is atelic. c. John read 3 books. i. cumulativity ‗John read 3 books‘ + ‗John read 3 books‘ ≠ ‗John read 3 books‘ (instead, = ‗John read 6 books‘) – not cumulative i. divisiveness A part of ‗John read 3 books‘ ≠ ‗John read 3 books‘ (instead, = ‗John read 1 book, 1.5 books, 2 books, etc.‘) – not divisive > It follows that ‗read a book‘ is telic. d. John read many books. i. cumulativity ‗John read many books‘ + ‗John read many books‘ = ‗John read many books‘ – cumulative ii. divisiveness A part of ‗John read many books‘ ≠ ‗John read books‘ – not divisive > It follows that ‗read many books‘ is telic. 1.1.2. Commonly used tests for telicity The non-homogeneity of telic predicates translates into the existence of a culmination point (or goal), i.e., a point at which the event could naturally terminate 49 because the intended change of state/result has been achieved. 3 In other words, telic predicates signify a property of the event such that the event can be completed. Standard diagnostics used to distinguish telic and atelic predicates are precisely tests for completivity. These are exemplified in (3)-(5) below. (3) Adverbial modification  Duration adverbials like for an hour are incompatible with telic predicates  ‗Frame‘ adverbials like in an hour are incompatible with atelic predicates atelic a. John read books for an hour/*in an hour. telic b. John read 3 books in an hour/*for an hour. (4) Conjunction with temporal modification (an implementation of the cumulativity test)  An atelic predicate allows for single event or two event interpretations with relevant temporal modification.  A telic predicate allows only for a two event interpretation with relevant temporal modification. a. John read books on Monday and Tuesday. (one or two events) b. John read 3 books on Monday and Tuesday. (two events) (5) Progressive test  An atelic predicate licenses inference from past progressive to past simple sentences.  A telic predicate does not license such inference. 3 Borer (2005) argues that ―culmination point is a side effect of the existence of an event structured by quantifiable division, rather than an independent property that holds of quantity events‖ (Borer 2005:221). Here I will use the notion of event culmination as a descriptor of an event that is not homogenous, such that it can be evaluated in terms of completion rather than just temporal termination. 50 a. John was reading books => John read books. b. John was reading 3 books. ≠> John read 3 books. In the discussion of Slavic aspect, I will use the adverbial modification, the conjunction and the homogeneity tests exemplified in the preceding section. The progressive entailment test is not directly adaptable to Slavic, but I will discuss the aspectual contribution of the English progressive briefly when considering the role of the secondary imperfective in chapter 4. 1.1.3. Basic syntax of telicity Telicity as a predicate of events composes with another predicate of events, namely the Verb Phrase. Structurally, for those who assume that telicity is explicitly represented in syntax, the functional projection encoding telicity (Asp(ect)P or of some other related name), projects in close proximity to VP, usually directly over the (lower) VP (e.g., Borer 2005, 1994; Travis 2005, 1994, 1991, Kratzer 2004, Tenny & Pustejovsky 2000, van Hout 1996, 1992, Schmitt 1996, Schoorlemmer 1995, or Tenny 1994, 1987). 4 There is also agreement as to the fact that only telicity, and not atelicity, should motivate the projection of additional/dedicated functional structure, or in other words, that atelicity is structurally more basic than telicity. 5 On this approach, atelicity is essentially the absence of telicity, which translates into syntax as the absence of a telicity- encoding projection or the absence of the ingredients of telicity (i.e., as in the models that utilize independently needed projections to derive telicity, as with the Agreement 4 A different model is offered in Ramchand (2004), where the telicity-encoding Result projection is a complement to the Process projection. 5 The only exception to that I am aware of is the proposed in Slabakova (2004), who postulates both a [+telic] and a [-telic] feature in the representation of situation types in Slavic. I will return to her proposal in section 6.2. 51 projections of the 1990‘s). I take this to be generally the case, but in anticipation of the discussion of the role of the secondary imperfective, I will also allow that atelicity can be derived. Until then, however, only the two structures in (6) will be assumed: (6) Figure 1: Preliminary basic syntax and atelicity and telicity a. atelic predicates b. telic predicates VP InnerAsp                                  [+telic] VP                            1.2. Viewpoint Aspects: PERFECTIVE and IMPERFECTIVE In the classic characterization of Comrie (1976), viewpoint aspect represents ―different ways of viewing the internal temporal constituency of a situation (…) perfectivity indicates the view of a situation as a single whole (…), while the imperfective pays essential attention to the internal structure of the situation‖ (Comrie 1976:16). In Smith (1991) a further, third kind of viewpoint aspect is distinguished, namely neutral viewpoint, but it remains controversial and will not be considered here. 1.1.4. Semantics of Viewpoint Aspects The semantics of the PERFECTIVE and IMPERFECTIVE 6 viewpoints has been formalized by Klein (1994), who used the theory of tense developed by Reichenbach‘s (1947), and cast the difference between the two viewpoint aspects as the difference in relationship between Topic Time (also referred to as Reference Time), and Event Time. On this approach, the role of viewpoint aspect is to locate an event temporally with respect to reference time: PERFECTIVE viewpoint conveys a reading where the Event Time 6 As a reminder, throughout this work, I use the terms PERFECTIVE and IMPERFECTIVE in small caps to refer to the universal viewpoint aspect categories, and the italicized perfective and imperfective to refer to the Slavic categories with particular morphology and distribution. 52 interval is contained within the Reference Time interval (and so when we look at the event from the outside), while the IMPERFECTIVE viewpoint refers to a situation where the Reference Time interval is properly contained within the Event Time (that is when we look at the event from the inside). It is uncontroversially assumed that viewpoint aspect composes with other predicates of events (i.e., VP, InnerAsp) before tense (TP). Therefore, viewpoint aspect is the first predicate that introduces Event Time into the equation, and turns predicates of events into predicates of intervals. As it asserts that a given event holds true of a time interval, it consequently existentially binds the event variable, which itself is associated with V, VP, vP/VoiceP, or EP depending on the model. Lastly, it situates the event time in relation to an evaluation interval, i.e., the reference time (Pancheva 2008, Bhatt & Pancheva, 2005, Kratzer 1994, 1996). Viewpoint aspect construed in this way is assumed to apply for all predicates. This makes Viewpoint Aspect different from Inner Aspect, which does not need to project for all events. The complete semantics for the PERFECTIVE and IMPERFECTIVE 7 viewpoints inclusive of their existential quantification are provided in (7) below. (7) from Bhatt & Pancheva (2005) a. [[ PERFECTIVE]] = λP (vt) .λt (i) . ∃e (v) [τ(e) ⊆ t & P(e)] ―The reference time is included or equal to the event time.‖ b. [[ IMPERFECTIVE]] = λP (vt) .λt (i) . ∃e (v) [t ⊂ τ(e) & P(e)] ―The reference time is properly included in the event time.‖ 7 I leave aside here the modal analysis of the IMPERFECTIVE viewpoint, as proposed for the English progressive by e.g., Dowty 1979, Landman 1992, Portner 1998, Zucchi 1999, or for the Spanish imperfecto in Cipria & Roberts (2001). I will come back to this issue briefly in discussing secondary imperfectivization in Chapter 4. 53 where ‗v‘ = the type of events, and ‗i‘ = the type of intervals, and where τ is a function from events to the time intervals they span 1.1.5. Tests for PERFECTIVITY and IMPERFECTIVITY It is rather hard to find tests that could diagnose viewpoint aspect without reference to (a)telicity. The tests that are typically employed distinguish between telic PERFECTIVES and everything else, as the IMPERFECTIVE viewpoint has an atelicizing effect when applied to telic predicates. Those are exemplified in (8)-(10) below (all tests and examples are based on Bhatt & Pancheva 2005). Note that English simple past is taken to express the PERFECTIVE viewpoint (Klein 1994, Giorgi & Pianesi 1998), while the progressive is a sub-type of the IMPERFECTIVE viewpoint. (8) Entailment patterns (the imperfective paradox of Dowty 1979) A goal-oriented (=telic) predicate in the PERFECTIVE entails the goal has been reached, while the same predicate in the IMPERFECTIVE does not have such entailment. a. PERFECTIVE: John read the book.  John finished reading the book. b. IMPERFECTIVE: John was reading the book. -/  John finished reading the book. a‘. #John read the book but didn‘t actually finish reading it. b‘. John was reading the book but didn‘t actually finish reading it. (9) Interaction with temporal clauses telic PERFECTIVE a. John read the book when the earthquake hit. (sequencing) telic IMPERFECTIVE 54 b. John was reading the book when the earthquake hit. (simultaneity) atelic PERFECTIVE c. John read when the earthquake hit. (simultaneity) atelic IMPERFECTIVE d. John was reading when the earthquake hit. (simultaneity) (10) Interaction with completive adverbials telic PERFECTIVE a. John read the book in an hour. telic IMPERFECTIVE b. John was reading the book (*in an hour). atelic PERFECTIVE c. John read (*in an hour). atelic IMPERFECTIVE d. John was reading (*in an hour). The atelicizing effect of the IMPERFECTIVE viewpoint, to which I will come back briefly later in this chapter as well as in Chapter 4, obscures the validity of the above tests as diagnostics for viewpoint aspect. One test that does not suffer from such limitations is the time interval adverbial test as in (11). (11) telic PERFECTIVE a. Between 10 and 11, Maya walked to the park. (durative or inclusive) atelic PERFECTIVE b. Between 10 and 11, Maya walked in the park. (durative or inclusive) 55 telic IMPERFECTIVE c. Between 10 and 11, Maya was walking to the park. (durative) atelic IMPERFECTIVE d. Between 10 and 11, Maya was walking in the park. (durative) This tests shows the difference between the IMPERFECTIVE and PERFECTIVE viewpoint in that only events viewed PERFECTIVELY can have an inclusive reading with respect to a time interval, i.e., they do not need to fill the whole reference time provided by the adverbial. Another version of this test uses coordination with a predicate that, because of our world knowledge, does not allow a concurrent activity, e.g., sleep. This is illustrated in (12). (12) telic PERFECTIVE a. Between 1 and 5, Maya read a book and slept. (inclusive) atelic PERFECTIVE b. Between 1 and 5, Maya read and slept. (inclusive) telic IMPERFECTIVE c. #Between 1 and 5, Maya was reading a book and sleeping. atelic IMPERFECTIVE d. #Between 1 and 5, Maya was reading and sleeping. In the sections to follow I will use only the time interval tests to diagnose viewpoint properties of Slavic predicates. 56 1.1.6. Syntax of Viewpoint Aspect I will assume here, uncontroversially, that viewpoint aspect is encoded syntactically through a viewpoint aspect projection ViewAsp, which can be headed by two features: PERFECTIVE or IMPERFECTIVE. This projection dominates any inner aspect (i.e., Aktionsart) projections, giving us the following four theoretical possibilities: (13) Figure 2: Theoretical combinatorial possibilities of telicity and viewpoint aspect a. atelic IMPERFECTIVE b. atelic PERFECTIVE ViewAsp ViewAsp          [IMPERF] VP [PERF] VP                           c. telic IMPERFECTIVE d. telic PERFECTIVE ViewAsp ViewAsp          [IMPERF] InnerAsp [PERF] InnerAsp            [telic] VP [telic] VP                            2. The nature of the Slavic aspectual contrast: B-imperfectives vs perfectives In the previous chapter, I discussed the morphological characteristics of aspect marking in Slavic, and shows that Slavic verbs fall into two categories of contrasting distribution: imperfectives and perfectives. Bare verbs (i.e., verbs with no aspectual morphology) and almost always imperfective, and the addition of a prefix to such bare forms creates perfective verbs. In the remainder of this chapter I consider the actual aspectual contribution of verbal prefixation in Slavic, and consequently the nature of the 57 contrast between B-imperfectives and prefixed perfectives 8 . I will limit the discussion in this chapter to non-stacking prefixes (i.e., those attaching directly to the bare stem), and will consider the distribution and function of stacking ones in Chapters 5 and 6. The addition of a prefix to a bare verb creates a pair with the following contrast: (14) a. Hania pisała list. Hannah.NOM wrote.3SG letters.ACC ‗Hannah wrote/was writing a/the letter.‘ (-/  Hannah wrote a/the letter.) b. Hania na-pisała list. Hannah.NOM PREF-wrote.3SG letter.ACC ‗Hannah wrote a/the letter.‘ The contrast consists essentially in (14)a describing an activity of writing a/the letter, and (14)b describing the event of having completed a/the letter. The accomplishment or achievement interpretation characterizes also prefixed forms that modify the verb‘s lexical meaning, including idiomatic cases. This is illustrated in (15) 9 : (15) a. Hania prze-pisała notatki. Hannah.NOM through-wrote.3SG notes.ACC ‗Hannah copied the notes.‘ b. Hania od-pisała na listy. Hannah.NOM from-wrote.3SG on letters.ACC ‗Hannah replied to the letters.‘ 8 As opposed to semelfactive perfectives, which are set aside in this work. 9 For the prefix list as well as their core meanings, please refer to Chapter 2, Section 3.1. 58 c. Hania za-pisała się na gimnastykę. Hannah.NOM INCPT-wrote.3SG REFL on gymnastics.ACC ‗Hannah enrolled in gymnastics.‘ d. Hania do-pisała do rachunku $10. Hannah.NOM to-wrote.3SG to bill.GEN big sum.ACC ‗Hannah wrote an additional $10 on the bill.‘ e. Hania o-pisała swoją koleżankę. Hannah.NOM about-wrote.3SG self‘s friend.ACC ‗Hannah described her friend.‘ f. Hania po-pisała się. Hannah.NOM po-wrote.3SG REFL ‗Hannah showed off.‘ (completive interpretation) g. Hania pod-pisała dokumenty. Hannah.NOM under-wrote.3SG documents.ACC ‗Hannah signed the documents.‘ h. Hania przy-pisała ten pomysł bratu. Hannah.NOM at-wrote.3SG this idea.ACC brother.DAT ‗Hannah ascribed this idea to her brother.‘ i. Hania roz-pisała długopis. Hannah.NOM out-wrote.3SG pen.ACC ‗Hannah got the (dry) pen to write.‘ j. Hania w-pisała adres do komputera. Hannah.NOM in-wrote.3SG address.ACC to computer 59 ‗Hannah typed in the address into the computer.‘ k. Hania wy-pisała recepty lalkom. Hannah.NOM out-wrote.3SG prescriptions.ACC dolls.DAT ‗Hannah wrote out prescriptions for her dolls.‘ l. Hania s-pisała się dobrze. Hannah.NOM s-wrote.3SG REFL well ‗Hannah came through.‘ By contrast, B-imperfective forms can never have achievement or accomplishment interpretation. 10 This is best illustrated with verbs that would correspond to achievements in English, but alternate between activity and accomplishment, or activity and achievement reading in Polish, e.g., (16) a. Hania kończyła pisać list. Hannah.NOM finished.3SG write.INF letter.ACC ‗Hannah was finishing writing a/the letter.‘ b. Hania s-kończyła pisać list. Hannah.NOM PREF-finished.3SG write.INF letter.ACC ‗Hannah finished writing a/the letter.‘ (17) a. Hania słyszała śpiew ptaka. Hannah.NOM heard.3SG sining.ACC bird.GEN ‗Hannah was hearing a/the bird‘s singing.‘ b. Hania u-słyszała śpiew ptaka. Hannah.NOM PREF-heard.3SG sining.ACC bird.GEN ‗Hannah heard a/the bird‘s singing.‘ 10 Except for a very small group of bare perfectives – see Chapter 2, Section 2, 60 (18) a. Hania traciła nadzieję. Hannah.NOM lose.3SG hope.ACC ‗Hannah was losing hope.‘ b. Hania s-traciła nadzieję. Hannah.NOM PREF-lose.3SG hope.ACC ‗Hannah lost hope.‘ Other textbook examples of achievement verbs in English, such as die, win, come, arrive, find, see or say, are all prefixed verbs in Polish and do not have an unprefixed counterpart, as shown in (19). (19) a. u-mrzeć ‗to die‘ (from *mrzeć in Modern Polish) b. wy-grać ‗to win‘ (from grać ‗to play‘) c. przy-jść ‗to come‘ (from iść ‗to go‘) d. przy-jechać ‗to arrive‘ (from jechać ‗to go (in a motorized way)‘) e. z-naleźć ‗to find‘ (from na-leźć, itself a prefixed form not part of Standard Polish, but found in some dialects, e.g., Kashubian Polish) f. zo-baczyć ‗to see‘ (from baczyć ‗to watch, pay attention (archaic)‘; the verb widzieć ‗to see‘ is used for the activity reading) g. po-wiedzieć ‗to say‘ (from wiedzieć ‗to know‘) Based on the above examples, it appears that the difference between B- imperfectives and the prefixed forms is their Aktionsart interpretation, with B- imperfectives being atelic (i.e., denoting activities or states), and prefixed forms being telic (i.e., denoting accomplishments and achievements). The same conclusion has been drawn, for example, to quote just more recent literature, in Brecht (1984), Krifka (1992), 61 Piñón (1995), Slabakova (2004), Schmitt (1996), Verkuyl (1999), Borer (2005), Travis (2005), and indirectly Babko-Malaya (1999) 11 , but there are also dissenting voices, most notably Filip (1999, 2000, 2004), but also the proponents of viewpoint aspect approach, as noted below. The assumed telicizing function of prefixes does not, naturally, preclude a viewpoint aspect role for verbal prefixes. In fact, in the light of previously noted atelicizing effect of the IMPERFECTIVE viewpoint, it would have to be concluded that the interpretation of prefixed forms, to the extent that it is always telic, is also PERFECTIVE. And since here I assume that the two kinds of viewpoint are in principle independent, this would in turn suggest that prefixes might have a double function, marking telicity as well as the PERFECTIVE viewpoint. Some observations to this effect have been made in the literature (e.g., Bertinetto 2001, Ramchand 2004, Svenonius 2004 12 , Nossalik 2009). However, there are also those believe that the imperfective-perfective distinction is essentially a distinction in viewpoint aspect, and consequently that prefixes are markers of the PERFECTIVE viewpoint, e.g., Smith (1991), Nagórko (1998), Młynarczyk (2001), Borik (2002), and Pereltsvaig (2005, forthcoming), of which Borik (2002) argues specifically against a telicity-marking function of prefixes in general. To see which is the above analysis is on the on the right track, and consequently to disprove alternative ones, I consider below how simplex imperfectives and perfective verbs fare with regard to tests for telicity and, more importantly, viewpoint aspect. 11 As described in Chapter 1, in Babko-Malaya‘s (1999) analsys, prefixes are said to have the role of deriving result-state predicates, which then compose (obligatorily, it appears) with a morphologically null BECOME affix (in the sense of Dowty 1976) and yield telic, change of state predicates. 12 Ramchand (2004) and Svenonius (2004) attribute telicity-marking property to only some prefixes. I will return to this issue in Chapter 6. 62 3. Telicity and the perfective-imperfective distinction 3.1. Telicity diagnostics If verbal prefixes are telicity markers and B-imperfective verbs are atelic, we expect prefixed verbs to be compatible with completive adverbials and incompatible with durative ones, while converse should be the case for B-imperfectives. This is, indeed, the case, as shown in (20). (20) a. Hania czytała książkę 2 godziny / *w 2 godziny. Hannah.NOM read.3SG book.ACC (for) 2 hours / in 2 hours ‗Hannah read / was reading a / the book for 2 hours.‘ b. Hania prze-czytała książkę *2 godziny / w 2 godziny. Hannah.NOM PREF-read.3SG book.ACC (for) 2 hours / in 2 hours ‗Hannah read the / a book in 2 hours.‘ c. Hania od-czytała referat *pół godziny / w pół godziny. Hannah.NOM from-read.3SG paper.ACC (for) half hour / in half hour ‗Hannah presented the/a paper in half an hour.‘ d. Hania w-czytała dane do komputera *5 minut / w 5 minut. Hannah.NOM in-read.3SG data.ACC to computer (for) 5 minutes/in 5 minutes ‗Hannah loaded (the) data into the computer in 5 minutes.‘ e. Hania roz-czytała dane *5 minut / w 5 minut. Hannah.NOM apart-read.3SG data.ACC (for) 5 minutes / in 5 minutes ‗Hannah decoded (the) data in 5 minutes.‘ (21) a. Hania biegła 10 minut / *w 10 minut. Hannah.NOM ran.3SG (for) 10 minutes / in 10 minutes 63 ‗Hannah ran for 10 minutes.‘ b. Hania po-biegła *10 minut / w 10 minut. Hannah.NOM PREF-ran.3SG (for) 10 minutes / in 10 minutes ‗Hannah ran the run in 10 minutes.‘ c. Hania przy-biegła *10 minut / w 10 minut. Hannah.NOM at-ran.3SG (for) 10 minutes / in 10 minutes ‗Hannah came by running in 10 minutes.‘ d. Hania do-biegła *10 minut / w 10 minut. Hannah.NOM to-ran.3SG (for) 10 minutes / in 10 minutes ‗Hannah got [to some place] by running in 10 minutes.‘ e. Hania pod-biegła do sklepu *10 minut / w 10 minut. Hannah.NOM under-ran.3SG to store (for) 10 minutes / in 10 minutes ‗Hannah made a quick run to a/the store in 10 minutes.‘ Another test that confirms the atelic nature of B-imperfectives and the telic nature of prefixed forms is the cumulativity test through conjunction with temporal modification. This test has the advantage of being applicable also to predicates which (grammatically) lack duration, i.e., the achievement class, and which for that reason can take neither completive nor durative adverbials, as exemplified in (22). (22) a. Hania wy-biegła *10 minut / *w 10 minut. Hannah.NOM out-ran.3SG (for) 10 minutes / in 10 minutes ‗Hannah ran out.‘ b. Hania od-biegła od tematu *10 minut / *w 10 minut. Hannah.NOM from-ran.3SG from subject (for) 10 minutes /in 10 minutes 64 ‗Hannah digressed from the subject matter.‘ c. Hania za-biegła drogę *10 minut / *w 10 minut. Hannah.NOM over-ran.3SG way (for) 10 minutes / in 10 minutes ‗Hannah run and blocked the way.‘ The results of the cumulativity test for the predicates in (20)-(22) are given in (23) and (24). As a reminder, a telic event can receive only a two-event reading in this test, while an atelic event can be interpreted as a single event or as two separate events. (23) a. Hania czytała książkę w środę i czwartek. Hannah.NOM read.3SG book.ACC in Wednesday and Thursday ‗Hannah read a book on Wednesday and Thursday.‘ (= 1 or 2 events) b. Hania prze-czytała książkę w środę i czwartek. Hannah.NOM PREF-read.3SG book.ACC in Wednesday and Thursday ‗Hannah read a book on Wednesday and Thursday.‘ (= 2 events) c. Hania od-czytała referat w środę i czwartek. Hannah.NOM from-read.3SG paper.ACC in Wednesday and Thursday ‗Hannah presented a paper on Wednesday and Thursday.‘ (2 events) d. Hania w-czytała dane do komputera w środę i czwartek. Hannah.NOM in-read.3SG data.ACC to computer in Wed and Thur ‗Hannah loaded the data into the computer on Wed & Thur.‘ (= 2 events) e. Hania roz-czytała dane w środę i czwartek. Hannah.NOM apart-read.3SG data.ACC in Wednesday and Thursday ‗Hannah decoded the data on Wednesday and Thursday.‘ (= 2 events) 65 (24) a. Hania biegła we wtorek i środę. Hannah.NOM ran.3SG in Tuesday and Wednesday ‗Hannah ran on Tuesday and Wednesday.‘ (= 2 events or 1 event, if unusual given world knowledge) b. Hania po-biegła we wtorek i środę. Hannah.NOM PREF-ran.3SG in Tuesday and Wednesday ‗Hannah ran the run on Tuesday and Wednesday.‘ (= 2 events) c. Hania przy-biegła we wtorek i środę. Hannah.NOM at-ran.3SG in Tuesday and Wednesday ‗Hannah came by running on Tuesday and Wednesday.‘ (= 2 events) d. Hania do-biegła we wtorek i środę. Hannah.NOM to-ran.3SG in Tuesday and Wednesday ‗Hannah got [to some place] by running in 10 minutes.‘ e. Hania pod-biegła do sklepu we wtorek i środę. Hannah.NOM under-ran.3SG to store in Tuesday and Wednesday ‗Hannah made a quick run to the store on Tues and Wed.‘ (= 2 events) f. Hania wy-biegła we wtorek i środę. Hannah.NOM out-ran.3SG in Tuesday and Wednesday ‗Hannah ran out on Tuesday and Wednesday.‘ (= 2 events) g. Hania od-biegła od tematu we wtorek i środę. Hannah.NOM from-ran.3SG from subject in Tuesday and Wednesday ‗Hannah digressed from the subject matter on Tues & Wed.‘ (= 2 events) 66 h. Hania za-biegła drogę we wtorek i środę. Hannah.NOM over-ran.3SG way in Tuesday and Wednesday ‗Hannah run and blocked the way on Tues and Wed.‘ (= 2 events).‘ Yet another set of supporting data for the telicizing function of prefixes comes from the domain of stative predicates. There, the addition of a prefix invariably turns a state verb it into an inchoative (when a ―pure perfectivizer‖ is added), or a general change-of-state verb, and thus transforms a homogenous predicate into a non- homogenous one (Młynarczyk 2001). Consequently, prefixes in such cases must be viewed as telicity markers, as confirmed by the examples below. (25) a. Hania pamiętała o swoich urodzinach 2 miesiące/*w 2 miesiące. Hannah remembered.3SG about self‘s birthday (for) 2 months/in 2 months ‗Hannah remembered about her birthday for 2 months.‘ b. Hania za-pamiętała datę swoich urodzin *2 dni/w 2 dni. Hannah INCPT-remembered.3SG date self‘s birthday (for) 2 days/in 2 days ‗Hannah memorized the date of her birthday in 2 days.‘ (26) a. Hania wierzyła jej godzinę/*w godzinę. Hania believed her (for) hour/in hour ‗Hannah believed her for an hour.‘ b. Hania u-wierzyła jej *godzinę/w godzinę. Hania PREF-believed her (for) hour/in hour ‗Hannah got to believe her in an hour.‘ (27) a. Hania pamiętała o tym we wtorek i środę (1 or 2 eventualities). Hannah remembered.3SG about this on Tuesday and Wednesday 67 ‗Hannah remembered about this on Tuesday and Wednesday.‘ b. Hania za-pamiętała to we wtorek i środę. (2 events) Hannah INCPT-remembered.3SG this in Tuesday and Wednesday ‗Hannah memorized this On Tuesday and Wednesday.‘ (e.g. Hannah suffers from short memory loss, and had to memorize the same thing twice) (28) a. Hania wierzyła jej w poniedziałek i wtorek (1 or 2 eventualities) Hannah believed.3SG her in Monday and Tuesday ‗Hannah believed her on Monday and Tuesday.‘ b. Hania u-wierzyła jej w poniedziałek i wtorek (2 events) Hannah PREF-believed her in Monday and Tuesday ‗Hannah got to believe her on Monday and Tuesday.‘ 3.2. Prefixed = telic The diagnostics above show conclusively that (i) B-imperfectives are atelic, and (ii) aspectual prefixes, both when they are purely perfectivizing as well as when they modify the verb‘s lexical meaning, are markers of Aktionsart, and derive telic predicates from atelic ones. As indicated earlier, this is by no means a novel conclusion. In fact, essentially the same sentiment was expressed as early as in Streitberg (1889) (after de Swart & Verkuyl 1999), as well as in later, pre-generative seminal works on Slavic aspect such as Vinogradov (1947) 13 , Isačenko (1960), Maslov (1961), and in a more reserved way in Towsend (1980). It could be argued that the use of the perfective and imperfective category labels in the traditional Slavic terminology really misleads and misrepresents 13 Vinogradov (1947) introduced the concept of ‗inner endpoint,‘ the goal of an action (Vinogradov 1947:497) 68 much of what has been uncovered in terms of their semantic correlates. 14 More modern and formalized implementations of this idea include the aforementioned Brecht (1984), Krifka (1992), Piñón (1995), Schmitt (1996), Babko-Malaya (1999, 2003), Verkuyl (1999), Slabakova (2004), Borer (2005), Travis (2005), to name just a few. Filip (2004) refers to the analysis of Slavic prefixes as telicity markers as the common view of Slavic prefixes. In the light of all this, it may even be surprising that there are any critics of this view, and yet there are. Their objections are considered below. 3.3. Considering counterevidence 3.3.1. Filip’s (1999, 2000, 2004) Objections Filip acknowledges that prefixed verbs typically have telic readings, but the correlation is not perfect enough for her, and, consequently, she considers telicity of prefixed verbs to be only epiphenomenal (stemming from the lexical semantics of the prefix-verb combinations) and not a product of a grammatical process. The specific irregularities cited by Filip are (i) the presence of bare perfective verbs and the possibility of attaching prefixes to them, (ii) the possibility of prefix stacking, (iii) the presence of prefixes with quantificational or modal functions, and (iv) the lack of a single purely telicizing prefix. While her objections have factual basis, they either concern a small number of fossilized forms or are easily explainable, and therefore are not sufficient to reject the possibility of a grammatical role for Slavic prefixation, especially in the light of 14 This is not to suggest that viewing Slavic aspectual oppositions as related to differences in Aktionsart was in that period the only or even the standard analysis. To the contrary, some of the more prominent works on the development of aspect in Slavic have claimed that the Slavic categories of perfective and imperfective have replaced the older viewpoint categories of the aorist and the imperfect, and consequently mark the PERFECTIVE and IMPERFECTIVE viewpoints respectively (e.g., Borodič 1953, Kuznecow 1953, Kölln 1958). 69 overwhelming evidence to the contrary (see Slabakova 2004, who reaches an identical conclusion). With regard to the existence of bare perfectives, their number in Slavic languages is rather small, which indicates their exceptional and fossilized nature. For a detailed discussion of bare perfectives the reader is referred to Chapter 2, Section 3; here I repeat only the list of such verbs in Polish: (29) i. chwycić ‗to seize, to grab‘ ii. chybić (archaic) ‗to miss‘ iii. czepić się ‗to grab hold of sth., to cling to‘ iv. dać ‗to give‘ v. kupić ‗to buy‘ vi. lec (archaic) ‗to lie down‘ vii. paść ‗to fall down‘ viii. puścić ‗to release, to let go‘ ix. rzec (archaic) ‗to say‘ x. ruszyć ‗to set in motion‘ xi. rzucić ‗to throw‘ xii. skoczyć ‗to jump‘ xiii. stawić ‗to place, to put‘ xiv. strzelić ‗to shoot, to fire‘ xv. trafić ‗to reach; to happen upon, to hit‘ Verbs in (29) could easily be analyzed as involving a phonologically null prefix, as indicated in (30), which would be restricted to this particular set of verbs. 70 (30) Bare perfectives: [ InnerAsp Ø- [ VP V]], e.g., a. [ InnerAsp Ø- [ VP kup]] ‗buy‘ b. [ InnerAsp Ø- [ VP pad]] ‗fall‘ c. [ InnerAsp Ø- [ VP rzut]] ‗throw‘ d. [ InnerAsp Ø- [ VP skok]] ‗jump‘ e. [ InnerAsp Ø- [ VP da]] ‗give‘ Such forms would be comparable to, for example, English irregular plurals with Ø- marking such as sheep, fish, or offspring, or irregular past tense forms such as put, let, or cost, where Ø-marking is also restricted to a small set of forms. The fact that verbs in (29) can combine also with other, overt prefixes, in turn brings to mind the parallel with pluralia tantum nouns, which function only as bound forms: they must either be affixed with a plural marker, or compounded with another stem, e.g., *trouser – trousers – trouser press, trouser socks, trouser sale; *scissor – scissors – Edward Scissorhands. The verbs which appear to be bare perfectives could similarly be analyzed as involving a bound root that must combine with a prefix, null or otherwise. For a root like √kup ‗buy,‘ this would give rise to the following set of forms: (31) a. [ InnerAsp Ø- [ VP kup]] ‗buy‘ b. [ InnerAsp prze- [ VP kup]] ‗bribe‘ c. [ InnerAsp pod- [ VP kup]] ‗steal (an idea)‘ d. [ InnerAsp o- [ VP kup]] ‗ransom, pay a price‘ e. [ InnerAsp s- [ VP kup]] ‗concentrate (attention)‘ f. [ InnerAsp s- [ VP kup]] ‗accumulate by buying‘ g. [ InnerAsp wy- [ VP kup]] ‗buy out‘ 71 h. [ InnerAsp od- [ VP kup]] ‗buy from‘ i. [ InnerAsp do- [ VP kup]] ‗buy something additional‘ j. [ InnerAsp roz- [ VP kup]] ‗buy exhaustively‘ Such an analysis, which appeals only to the rather uncontroversial processes of null- affixation and morphological boundedness, eliminates the problem of prefixation of an already telic predicate and consequently of double delimitation per predicate, to which Filip objects citing Tenny‘s (1987, 1994) Single Delimiting Constraint: ―The event described by a verb may only have one measuring-out and be delimited only once (Tenny 1994:79). The second objection raised by Filip concerns the phenomenon of prefix stacking, where a prefix can attach to an already prefixed verb. This situation replicates the problem of prefixed (seemingly) bare perfectives and the consequent apparent double telicity-marking. I will consider the issue of stacking prefixes in detail in Chapters 5 and 6. Here, I will only note that this objection relies on viewing all prefixes, including all the stacking ones, as telicity markers – something that is not fully accurate. Secondly, in most Slavic languages, 15 additional prefixes attach not to perfective stem, but rather to secondarily imperfectivized forms, which I will argue are atelic. The third concern of Filip has to do with special function prefixes, such as the measure prefixes na- ‗a lot‘ and po- ‗a little, some‘ (both to be discussed in Chapters 5 and 6), and u- with the meaning ‗manage, be able to.‘ It seems that Filip finds it hard to reconcile the diverse, sometimes quantificational or possibly modal, meanings and uses of Slavic prefixes with a single grammatical function. The only problem is that except for 15 Bulgarian behaves differently in this respect, which must be seen as correlating with the many other differences it exhibits, some of which will be discussed in the following chapter. 72 one case, again to be addressed in Chapters 5 and 6, all Slavic prefixes do in fact mark telicity. Further, there is really no alternative in Slavic. If one believes that telicity is a universal grammatical property, then it needs to be expressible in every natural language (cf. Olsen 1994). In some languages, this is done through the properties of arguments and modifiers (e.g., English, Dutch), in others through case marking (e.g., Finnish, Estonian), in yet others through a dedicated morpheme (e.g., Vietnamese, Malagasy). In Slavic, none of these strategies is available globally to derive different Aktionsart readings. 16 Something else then, must be the mechanism Slavic languages employ, and given the correlation between imperfectivity and atelicity on the one hand, and perfectivity and telicity on the other, it is hard to argue it is anything other than aspectual morphology. Lastly, with regard to the problem of the lack of a prefix that would have a purely grammatical function (as opposed to meaning modifying prefixes), a claim that was previously advanced in Isačenko 1962, Czochralski 1975, it needs to be clarified that there are, in fact, purely perfectivizing prefixes (see Chapter 2, Section 3.3) – it is the prefixes we find in the so-called aspectual pairs (e.g., pisać–na-pisać ‗to write,‘ czytać– prze-czytać ‗to read,‘ budzić–o-budzić ‗to wake‘, nieść–za-nieść ‗to carry,‘ jeść–z-jeść ‗eat,‘ pić–wy-pić ‗to drink,‘ etc.). The fact they are different from ―content-full‖ prefixes is best evidenced by the fact that forms with pure perfectivizers typically cannot be secondarily imperfectivized. If, as I will argue here, the role of the secondary imperfective is to atelicize a verb, then one can interpret this restriction in functional terms as a 16 Some of these strategies may be used in a more limited way. For example, it is reported that the properties of the object can influence the Aktionsart interpretation of biaspectual verbs in Bulgarian (Slabakova 1997). Also, some perfective verbs allow accusative-partitive alternation with their objects (Franks 2005), but it is not clear if the result of applying the partitive case is atelicity or rather degraded telicity, where the role of the partitive is simply to block the definitive interpretation of the direct object. Lastly, Slavic languages may be gradually converging on a default perfectivizer, e.g, po- in Russian, z-/s- in Czech (Dickey 2000), and also z-/s- in Polish. 73 resistance to create a morphologically complex form for a bundle of meaning that already exists, namely the unprefixed counterpart of the verb with a ―pure‖ perfectivizer. In addition to that, in response to Filip, Slavic languages seem to be settling on a default perfectivizer – i.e., one that is most frequently applied to new verbs. In Russian it is po-, and in Czech and Polish z- (Dickey 2000). It might be that what Filip suggests characterized an older stage in the development of Slavic aspect, but the functional role of prefixes has since been grammaticalized to a degree that can no longer be denied. 3.3.2. Borik’s (2002) Objections Borik (2002) discusses two classes of predicates which appear problematic for generalizing the role of prefixes as telicity markers. These three groups are: (i) delimitative verbs prefixed with po-, (ii) perdurative verbs prefixed with pro-, and (iii) ‗beginning‘ verbs‘. She takes these groups to contradict the generalization that prefixed verbs are always telic, based on their performance is different telicity/ homogeneity tests. As I briefly mentioned before, the prefix po- is indeed different in that it belongs to the group of quantificational prefixes, which will be considered in Chapter 5. As for the other two problematic classes, let‘s look at them below. 3.3.2.1. Perdurative pro- Borik (2002) argues that the perdurative Russian prefix pro- (Polish prze-), with the meaning ‗through, for (of time)‘ is not telicizing because pro-prefixed verbs perform as atelics in the adverbial modification tests. This is illustrated in (32)-(34) (all Russian examples below are from Borik 2002, with some modifications suggested by Tania Ionin – p.c.). 74 (32) Petja pro-sidel v tjur‘me pjat‘ let/*za pjat‘ let Peter.nom pro-sat.3SG in prison (for) five years/*in five years ‗Peter was in prison for five years.‘ (33) Petja pro-sidel v tjur‘me maj i ijun (one or two events) Peter.NOM pro-sat.3SG in prison May and June ‗Peter was in prison in May and in June.‘ (34) Petja pro-sidel v tjur‘me pjat‘ let ę → Petja pro-sidel v tjur‘me dva goda (divisive) Peter pro-sat in prison five years → Peter pro-sat in prison two years ‗Peter was in jail for 5 years‖ → ―Peter was in jail for 2 years.‘ The examples above can be translated into Polish (with the corresponding prefix prze-) with identical results. So does that mean that pro-/prze- is not a telicizing prefix? As it turns out, it is Borik herself that provides a hint that ultimately leads to the conclusion that the answer to this question is actually a ‗no‘. She notes that the prefix pro- ―requires the presence of an explicit duration temporal expression‖ (Borik 2002:55). These temporal expressions, I would argue, are actually arguments of the pro-prefixed verbs, and not adverbial modifiers. Note that these temporal expressions could not be PPs, e.g., (35) Piotr prze-siedział w więzieniu (*przez) 5 lat Peter.NOM prze-sat.3SG in prison (*for, through) 5 years ‗Peter was in prison for five years.‘ (36) Piotr prze-mieszkał we Wrocławiu (*przez) całe życie Peter.NOM prze-lived.3SG in Wrocław (*for, through) whole life ‗Peter lived in Wrocław his whole life.‘ 75 Consequently, they cannot count as adverbial modifiers for the purposes of telicity tests. With regards to the test in (32), the reason why the completive adverbial ‗in 5 years‘ is impossible, is not because it is semantically incompatible with the predicate, but because it is rather odd to add a modifier to a predicate that has its equivalent as an argument. It would be as natural as to say in English, ‗I spent 5 years in jail in 5 years.‘ That it is a problem of redundancy rather than grammaticality can be seen if we construct a context where the temporal argument and temporal modifier can be different. Consider two such examples in (37) and (38), where completive adverbials turn out to be perfectly fine, while durative ones are out. (37) Motyl prze-żył swoje całe życie w kilka dni/*kilka dni. Butterfly.NOM prze-lived.3SG self‘s whole life in a few days/(for) a few days ‗The butterfly lived its entire life in a few days.‘ (38) Lecąc statkiem kosmicznym prze-żyliśmy 10 lat w 2 tygodnie/*2 tygonie. riding spaceship.INSTR prze-lived.1PL 10 years in 2 weeks/*(for) 2 weeks ‗While riding the spaceship, we lived 10 years in 2 weeks.‘ We can contrast that with unprefixed forms of the same verbs, where completive adverbials become impossible, and durative ones become fine: (39) Motyl żył swoje całe życie *w kilka dni/kilka dni. Butterfly.NOM lived.3SG self‘s whole life in a few days/(for) a few days ‗The butterfly lived its entire life for a few days.‘ (40) Lecąc statkim kosmicznym żyliśmy 10 lat *w 2 tygodnie/2 tygonie. riding spaceship.INSTR lived.1PL 10 years *in 2 weeks/(for) 2 weeks ‗While riding the spaceship, we lived 10 years for 2 weeks.‘ 76 It is worth noting that the atelic-like appearance due to the presence of a temporal argument is not a unique property of pro-/prze-, but rather characterizes all verbs that take temporal arguments. A few examples are given below. (41) Piotr od-siedział w więzieniu 5 lat/* 5 lat. Peter.NOM from/away-sat.3SG in prison (for) 5 years/*in 5 years ‗Peter did a sentence in prison for five years.‘ (42) Piotr za-czekał 10 minut/*10 minut. Peter.NOM PREF-wait.3SG (for) 10 minutes/*in 10 minutes ‗Peter waited 10 minutes.‘ The prefix in (41) modifies the verb meaning, while the one in (42) is a pure perfectivizer. These examples show that the perdurative pro- is not special, and that it does not provide evidence against the telicity-marking role of prefixes. To the contrary, when the right context is construed, pro-prefixed verbs test as telics, showing that pro- is a telicity marker, just like any other prefix. 3.3.2.2. ‘Beginning verbs’ The second problematic set of cases according to Borik (2002) are inceptives (‗beginning verbs‘ in Borik‘s terminology). Such verbs are formed in Russian and Polish (and generally in Slavic) with the prefix za-, and they focus the attention on the beginning of the event, e.g., (43) Marta zapaliła papierosa i usiadła na ławce. Marta.NOM INCPT-burn.3SG cigarette and sat.3SG on bench ‗Marta lighted a/the cigarette and sat on the bench.‘ 77 Borik reports that verbs with the inceptive za- prefix show a mixed behavior with respect to different telicity tests. Specifically, they come out as telic on the adverbial modification and conjunction test (i.e., they receive 2 event interpretation), but atelic on the progressive test. The application of the progressive test is exemplified below with Polish translations of Borik‘s examples (Borik 2002:60): (44) Jak przy-szedł Piotr, komputer działał → Komputer za-działał. when came.3SG Peter, computer worked.3SG → computer INCPT-worked.3SG ‗When Peter came, computer was working‘ → ‗The computer started worked.‘ (45) Jak przy-szedł list, Piotr śpiewał. → Piotr zaśpiewał. when came.3SG letter, Peter sang.3SG → Peter INCPT-sang.3SG ‗When the letter arrived, Peter was singing‘ → ‗Peter sang something.‘ According to Borik, sentences such as those in (44)-(45) indicate that the ‗beginning‘ verbs are not telic because they pattern with atelic predicates on the progressive entailment test. However, one has question the validity of applying the progressive entailment test to the examples above because the B-imperfective forms are not the progressive, or even, as I will show shortly, necessarily viewpoint IMPERFECTIVE, versions of the inceptive forms. The inceptive prefix adds a meaning component that is not present in the B-imperfective form, and so the difference between the two is not just aspectual. Secondly, given that za- is equivalent ‗begin to‘ in English, it is trivial that the entailments in (44) and (45) hold, cf. The computer worked → The computer started to work; Peter sang → Peter started to sing. Therefore, it has to be concluded that the progressive entailment test as applied in (44) and (45) is not a valid diagnostics, and as in 78 all other tests za-prefixed verbs come out as telic, the inceptive za- must in fact be understood to be a marker of telicity. 3.4. Prefixes as telicity-markers: concluding thoughts The conclusion from the previous subsections should be quite clear: Slavic B- imperfectives are atelic, while prefixed perfectives 17 are telic. Consequently, verbal prefixes must be viewed as markers of telicity in Slavic, without which telic readings do not arise. While this conclusion is based on telicity tests applied to a small sample of predicates, they are also supported by the results of a quantitively much broader investigation presented in Młynarczyk (2004), who herself assumes the viewpoint aspect approach. The corpus of data studied by Młynarczyk revealed that (i) adding a prefix to a state verb invariably turns it into a change-of-state verb, and (ii) verbs with modifying prefixes (i.e., prefixes that change the verb meaning) are all transition verbs (i.e., verbs involving a change of state). If prefixes are not first and foremost telicity markers, it becomes quite a coincident that adding a prefix uniformly has this effect with respect to these two groups. Thus, Młynarczyk‘s work provides quantitative backing to the here- presented conclusions. In the following section, I look at the viewpoint aspect contrast between B- imperfectives and prefixed perfectives in order to establish whether the traditional labels correspond to the viewpoint aspect categories of IMPERFECTIVITY and PERFECTIVITY, respectively, and whether Slavic verbal prefixes should also be understood as markers of viewpoint aspect. 17 As I mentioned in the introductory chapter, I am restricting the discussion of perfectives to those involving prefixation. The other class of perfectives, i.e., semelfactives, are set aside as a topic for further research. 79 4. Viewpoint aspect and the perfective-imperfective distinction 4.1. Slavic B-imperfective = IMPERFECTIVE? It is often assumed that Slavic imperfectives have a temporally unbounded, i.e., IMPERFECTIVE, interpretation. That may be generally true. With no contextual information and no adverbial modifiers, the IMPERFECTIVE interpretation seems to be the natural reading for B-imperfectives, e.g., (46) Hania czytała książki. Hania.NOM read.3SG books.ACC ‗Hannah was reading books/read books habitually.‘ However, in the presence of a time frame adverbial, a B-imperfective form can receive a temporally bounded, i.e., PERFECTIVE, interpretation, e.g, (47) Między 10 a 11, Hania pisała list. (durative OR INCLUSIVE) between 10 and 11, Hannah.NOM wrote.3SG letter.ACC ‗Between 10 and 11, Hannah was writing a letter/did some writing of a letter.‘ Other examples showing bounded readings of B-imperfective forms have already been shown earlier, though were not introduced as such. Those are repeated below. (48) Hania czytała książkę w środę i czwartek. Hannah.NOM read.3SG book.ACC in Wednesday and Thursday ‗Hannah read a book on Wednesday and Thursday.‘ (= 1 or 2 events) (49) Hania biegła we wtorek i środę. Hannah.NOM ran.3SG in Tuesday and Wednesday ‗Hannah ran on Tuesday and Wednesday.‘ (= 2 events or 1 event, if unusual given world knowledge) 80 (50) Hania pamiętała o tym we wtorek i środę (1 or 2 eventualities). Hannah remembered.3SG about this on Tuesday and Wednesday ‗Hannah remembered about this on Tuesday and Wednesday.‘ (51) Hania wierzyła jej w poniedziałek i wtorek (1 or 2 eventualities) Hannah believed.3SG her in Monday and Tuesday ‗Hannah believed her on Monday and Tuesday.‘ In all the examples above, B-imperfective verbs receive a bounded interpretation, and hence a 2 event reading with a conjoined modifier. Perhaps an even clear indication of that is a test with a specified time interval and a conjoined predicate that precludes a concurrent activity, such as sleep, as well as with a modifier clause ‗but not the whole time.‘ These are applied below: (52) Między 10 a 11, Hania czytała książkę i spała. between 10 and 11, Hannah.NOM read.3SG book.ACC and slept.3SG ‗Between 10 and 11, Hannah did some reading of a book and slept.‘ (53) Między 10 a 11, Hania czytała książkę, ale nie przez cały czas. between 10 and 11, Hannah.NOM read.3SG book.ACC, but not for the entire time ‗Between 10 and 11, Hannah did some reading of a book, but not for the entire time.‘ The examples above all illustrate the point that the interpretation of B- imperfectives does not need to be IMPERFECTIVE, and therefore that the imperfective= IMPERFECTIVE equivalence does not hold. This fact has been noted previously in Forsyth (1970), Comrie (1976), and Gawrońska (1993), but remains mostly ignored in viewpoint- 81 based analysis of Slavic aspect (e.g., Mezhevich 2008, Pereltsvaig 2005, Borik 2002, Młynarczyk 2001). 4.2. Slavic perfective = PERFECTIVE? This equivalence, in contrast to the previous one, does in fact turn out to be true. Perfective verbs always denote temporally bounded events, or more informally, events which are viewed ―from the outside‖. The time interval test in (54)-(55) shows that perfective forms can have an inclusive interpretation, while the interaction with temporal clauses test in (56)-(57) shows that it cannot have simultaneous interpretation: (54) Między 10 and 11, Hania na-pisała list. (inclusive) between 10 and 11, Hannah.NOM PREF-wrote.3SG letter.ACC ‗Between 10 and 11, Hannah wrote a letter.‘ (55) Między 10 and 11, Hania prze-czytała książkę. (inclusive) between 10 and 11, Hania.NOM PREF-read.3SG book.ACC ‗Between 10 and 11, Hannah read a book.‘ (56) Hania na-pisała (właśnie) list jak za-dzwonił telefon (sequencing) Hannah.NOM PREF-wrote.3SG (just) letter.ACC when INCPT-rang.3SG phone 1. with just: ‗Hannah has just written a letter when the phone rang.‘ 2. without just: ‗Hannah wrote a letter when (= right after) the phone rang.‘ (57) Hania prze-czytała (właśnie) książkę jak za-dzwonił telefon (sequencing) Hannah.NOM PREF-read.3SG (just) book.ACC when INCPT-rang.3SG phone 1. with just: ‗Hannah has just read a book when the phone rang.‘ 2. without just: ‗Hannah read a book when (= right after) the phone rang.‘ 82 If the viewpoint aspect interpretation of Slavic (simplex) imperfectives is variable, but the interpretation of prefixed forms is always PERFECTIVE, we can ask why. With the framework assumed here, where the two kinds of aspect are in principle independent of each other and should be able to compose with each other freely, this restriction may seem unexpected. Judging by the available literature, to most people to answer to this problem would be obvious: prefixed verbs can in fact compose with the IMPERFECTIVE viewpoint, which is realized through the secondary imperfective marking. However, as I will argue in the next chapter, the secondary imperfective morphology does not in fact mark IMPERFECTIVITY, but instead has an atelicizing function which is independent of viewpoint aspect. With that, we are back to the same question, namely why cannot perfective forms compose with the IMPERFECTIVE viewpoint? I would suggest that the answer to this question has to do with the basic, i.e., non- modal, semantics of IMPERFECTIVITY, as defined in (7) in Section 1.2.1., and repeated in (58) below. (58) [[ IMPERFECTIVE]] = λP (vt) .λt (i) . ∃e (v) [t ⊂ τ(e) & P(e)] ―The reference time is properly included in the event time.‖ This semantics will return an atelic reading for any telic predicate where the event culmination coincides with the event‘s left or right temporal boundary, as proper inclusion of the reference time interval within the event time will exclude both boundaries. It has been noted, however, that there are events where the culmination does not need to coincide with initial or final temporal boundaries, e.g., to float under a bridge, to fill a room with a smoke, to run past the barn, Borer 2005). In such cases, the IMPERFECTIVE viewpoint (and more narrowly the progressive in English) selects not just 83 for any properly included subinterval, but crucially for a homogenous subinterval preceding the culmination. 18 But if the IMPERFECTIVE/progressive viewpoint has to make reference to the culmination, then is it really independent of Aktionsart? It may be that it is not, at least in the sense that it cannot apply indiscriminately to telic predicates (i.e., it cannot select for just any subinterval). Instead, we might speculate that it can apply only to homogenous (i.e., atelic) or homogenized (i.e., once telic) predicates. This could be formulated as the following constraint: (59) The IMPERFECTIVE viewpoint can compose only with homogenous or homogenized predicates. Alternatively, using the terminology of Borer (2005), we could say that the IMPERFECTIVE viewpoint cannot by itself remove quantifiable divisions and thereby make the event atelic. The consequence of (59) is that the application of the progressive in English or, e.g., imperfecto in Spanish, must be seen as entailing more than just adding a viewpoint aspect predicate, which is incidentally independently suggested by the modal analyses of the relevant categories proposed specifically in relation to telic predicates (e.g., Dowty 1979, Landman 1992, Portner 1998, or Zucchi 1999 for English, Cipria & Roberts 2001 for Spanish). Coming back to the Slavic case, we could postulate that the IMPERFECTIVE aspect in Slavic is transparently just a viewpoint aspect, i.e., it does not involve a phonologically covert atelicizing operation. That operation, as I will argue in the next chapter, is in fact overt in Slavic, with its phonological reflex being the secondary imperfective morphology. The prefixed verbs, which are telic, cannot consequently compose with the IMPERFECTIVE viewpoint, and so they compose with the PERFECTIVE one (the reader might recall that the addition of viewpoint aspect is assumed to be 18 Schein (2002, fn 41) makes an observation to the same effect. 84 necessary for the composition of event semantics, and more specifically for the existential binding of the event variable). 4.3. Verbal prefixes and viewpoint The question that needs to be asked given the discussion above is whether the alignment between prefixed forms and the PERFECTIVE viewpoint has been grammaticalized in the sense of assigning a viewpoint-marking function to the verbal prefixes in addition to their telicity-marking role, or whether it has a form of a compositional restriction. I suggest it is the latter one based on three considerations. The first one is the fact that the relevant correlation is present in Bulgarian and was present already in Old Church Slavonic. Both of these languages have a separate system of viewpoint aspect marking through the aorist-imperfect distinction (i.e., the PERFECTIVE- IMPERFECTIVE distinction in the past tense), and perfective verbs can be only expressed in the aorist. Of course, in other Slavic languages that have since lost the aorist-imperfect distinction, prefixes could have picked up an additional, viewpoint marking function. However, what the Bulgarian and Old Church Slavonic rule shows at the minimum, is that the prefixes need not be implicated in viewpoint marking and it still holds that prefixed forms cannot compose with the IMPERFECTIVE viewpoint. We can extend this to other Slavic languages by limiting prefixes to their Aktionsart role, and assuming the rule in (59) would force the composition of prefixed verbs with the PERFECTIVE viewpoint, which in languages such as Polish or Russian would on that approach be marked with zero morphology. The other two reasons for not charging prefixes with viewpoint aspect marking role have to do with the fact that prefixes can be found in forms that (i) do not have to 85 have PERFECTIVE interpretation, and (ii) do not have viewpoint aspect interpretation at all. The first set of forms are secondary imperfective verbs. They are forms derived from prefixed verbs by means of suffixation, and their aspectual interpretation, as I will show in the subsequent chapters, is aspectually identical to B-imperfectives, i.e., atelic(ized) and viewpoint variable. Clearly, in such forms the prefix does not mark viewpoint, and making the prefix an obligatory viewpoint marker would be a problem there. One could try to salvage the viewpoint marking role of the prefix in such cases by claiming that the secondary imperfective blocks the movement of the prefix to ViewAsp and consequently prevents it from marking the PERFECTIVE aspect. In other words, the prefix would mark the perfective viewpoint only when it could, i.e., when its movement to a viewpoint- encoding projection is unobstructed, or when such a projection actually exists (see the discussion of nominals below). I do not discount this possibility, but also do not find overwhelming evidence that it is superior to the compositional restriction. The second set of forms of consequence are nominals with verbal prefixes which decidedly lack any temporal, and therefore viewpoint, properties. Nouns in Polish are generally resistant to prefixation with the exception of some foreign prefixes such as, e.g., vice-, ex-, or super-. Prefixes known from the verbal domain do not attach to nouns. Nominals can, however, be derived from prefixed verbs as well as from what looks like prefixed roots. A lot of such nominals are not of the complex event type and clearly do not involve any temporal, and therefore viewpoint aspect, interpretation. The fact that prefixes are found abundantly in such forms indicates that they are not portmanteau morphemes necessarily realizing Aktionsart and viewpoint features, but rather that they are limited to Aktionsart. There is no doubt that the main role of prefixes in such 86 nominals is meaning changing, but there is also evidence for InnerAsp projections in them, as when we get a result reading for PREFIX+ROOT combinations, as in (60), or when prefixed stems require secondary imperfectivization in order to compose with suffixes that require an activity-denoting (as opposed to an accomplishment- or achievement- denoting) stem, as in (61). In the latter case, secondary imperfective suffixes act as atelicizers which are able to convert accomplishments and achievements into activities, but they clearly need to do it within the bounds of the syntactic structure, and therefore provide evidence for the existence of Aktionsart-encoding projections in the relevant nominals in the absence of ViewAsp. (60) Figure 3: Prefixed Root Nominals 19 N   -Ø InnerAsp    PREF- V e.g., √pis: *pis N , pis-a-ć V ‗to write‘ a. na-pis ‗an inscription, caption‘ (cf. na-pisać ‗to write-telic‘) b. przy-pis ‗footnote‘ (cf. przy-pisać ‗to attribute‘) c. za-pis ‗a note, record;‘ za-pisy pl ‗sign up, enrollment‘ (cf. za-pisać ‗to write down;‘ za-pisać się ‗to sign up, to enroll‘) d. wy-pis ‗a copy of an official document‘s text, hospital discharge‘ (cf. wy-pisać ‗to write out, to discharge from hospital‘) 19 Despite embedding verbal structure up to InnerAsp projection, prefix root nominals cannot take arguments. We could attribute this to the restriction on –Ø N : it does not attach to structures with filled specifiers (assuming all arguments merge in specifiers). 87 e. prze-pis ‗recipe‘ (cf. prze-pisać ‗copy, prescribe‘) (61) Names of Instruments and Agents a. s-prysk-iw-acz ‗sprinkler‘ (S-imperfective stem) cf. *s-prysk-acz (perfective stem) ~ s-prysk-iw-a-ć ‗to sprinkle‘ (S-imperfective) b. od-gruz-ow-yw-a-cz ‗remover of debris‘ (S-imperfective stem) cf. *od-gruz-ow-a-cz (perfective stem) ~ od-gruz-ow-yw-a-ć ‗to remove debris‘ (S-imperfective) c. pod-żeg-iw-a-cz ‗an instigator‘ (S-imperfective stem) *pod-żeg-acz (perfective) ~ pod-żeg-iw-a-ć ‗to instigate‘ (S-imperfective) As a final comment, note that the fact that Slavic perfectives are always viewpoint PERFECTIVES allows us to still analyze the distribution of imperfectives and perfectives with respect to tests concerned with temporal characteristics, such as present versus future time denotation or the compatibility with delimiting phase verbs (see Chapter 2, Section 5 for details), in terms of viewpoint aspect distinctions, as has been done, for example, in Borik (2002), Pereltsvaig (2005) and Mezhevich (2008). Such tests, however, must now be seen as sensitive not to PERFECTIVE interpretation, which is available for both perfective as well B-imperfective forms, as previously shown in examples (47)-(53) in Section 4.1., but to the availability of IMPERFECTIVE readings, which is limited to imperfectives. Consequently, such tests should show restricted distribution for perfectives, and unrestricted one for imperfectives. Take, for example, the present versus 88 future interpretation test as an example, for which the following generalizations were stated in Chapter 2: (62) Present versus future denotation for present tense forms: 3. Imperfective verbs in the present tense forms have present tense interpretation. 4. Perfective verbs in the present tense forms have future tense interpretation. We now expect that while the perfective forms are restricted to future tense interpretation, as temporally bounded readings appear to preclude a coincidence relation with the speech time (cf. Mezhevich 2008), imperfectives should be compatible with either present tense interpretation (on their IMPERFECTIVE reading) or future tense interpretation (on their PERFECTIVE reading). This is indeed the case, as illustrated in (63) and (64). (63) a. Idę do kina. (IMPERFECTIVE reading) go.1SG to theater ‗I am going to the movies.‘ (present interpretation) b. Idę jutro do kina. (PERFECTIVE reading) go.1SG tomorrow to theater ‗I go to the movies tomorrow.‘ (64) a. Biorę urlop. (IMPERFECTIVE reading) take.1SG vacation.ACC ‗I am taking vacation.‘ (present interpretation) b. Biorę za tydzień urlop. take.1SG for week vacation.ACC ‗I take vacation in a week.‘ 89 The fact that B-imperfectives have PERFECTIVE interpretation when used to refer to future, is confirmed by the impossibility of obtaining progressive reading for B-imperfectives in the relevant contexts. This is shown in (65), where the ‗while‘ clause forces a progressive interpretation for the matrix verb. (65) *Jutro śpię gdy będziesz w pracy. tomorrow sleep.1SG while be.FUT.2SG at work ‗I will be sleeping tomorrow while you are at work.‘ To get the progressive reading, the analytic future form is required instead: (66) Jutro bedę spać gdy będziesz w pracy. tomorrow be.FUT.1SG sleep.INF while be.FUT.2SG at work ‗I will be sleeping tomorrow while you are at work.‘ These examples confirm once again that B-imperfective forms are actually ambiguous between IMPERFECTIVE and PERFECTIVE interpretation, and also indicate that the temporal tests for (im)perfectivity should be understood to reveal only the restrictions imposed on the distribution of the perfectives due to their necessarily PERFECTIVE interpretation, rather than any definite properties of imperfectives. 4.4. Slavic aspectual contrasts: summary The discussion of the Aktionsart and viewpoint aspect interpretation of the Slavic perfective and B-imperfective categories in the last two sections revealed the following pattern: 90 Table 2: The relationship between Slavic aspectual categories, telicity and viewpoint aspect Slavic aspectual category Aktionsart Viewpoint Aspect Simplex Imperfective Atelic IMPERFECTIVE or PERFECTIVE Perfective Telic PERFECTIVE I have concluded that Slavic prefixes are responsible for marking telicity and discussed the objections to such an analysis that have been raised in the literature. The arguments I presented strengthen the consensus that has been developing for quite some time that verbal prefixes are the grammatical exponents of Aktionsart, and that verbal prefixation is the primary way of deriving telic readings in Slavic. I also argued that while Slavic perfectives are always viewpoint PERFECTIVE, it may be best to not charge prefixes with marking viewpoint as well. I drew in this proposal on evidence from the presence of prefixes in nominal forms without any temporal or viewpoint aspect meaning, on the lack of viewpoint aspect marking by prefixes in secondary imperfective forms (to be discussed in more detail in the following chapter), as well as on the independence of prefixation from viewpoint aspect marking in Old Church Slavonic and Bulgarian despite the presence of the exact same correlation between telicity and PERFECTIVITY, or, as stated in (59), the impossibility of combining the IMPERFECTIVE viewpoint with telic predicates. I have suggested instead that viewpoint aspect in languages such as Polish and Russian is marked by zero morphology. Admittedly, one could question postulating across-the-board zero morphological marking for a given grammatical category. This sentiment essentially motivates the idea expressed in Smith‘s (1991) that while all languages encode Aktionsart distinctions, not 91 all languages have to make grammatical viewpoint aspect distinctions. Such an approach seems particularly attractive when viewpoint aspect interpretation is contextually determined, as is the case with B-imperfective forms in Slavic (other than languages with the aorist-imperfect distinction). After all, contextually motivated interpretation is not available once the grammar ―fixes‖ a particular interpretation through projection of a particular functional structure headed by (elements with) features of a particular value (cf. Borer 2005), and so when it is available, it signals the lack of such functional structure. With regard to Slavic aspect, this approach would suggest that viewpoint aspect is projected only for perfective forms, as there the viewpoint interpretation is fixed – it is always PERFECTIVE. For the imperfective forms, on the other hand, there would be no ViewAsp projection. The reason why this analysis is not advanced here has to do with the way that the compositional event semantics is assumed to proceed. As explained in section 1.2.1, I follow the approach where the role of ViewAsp is to introduce the Event Time and to existentially bind the event variable. Consequently, ViewAsp must project in order for the composition of the event semantics to proceed. If one assumed that ViewAsp projects only for some events, but not others, then the existential binding as well as the role of viewpoint aspect as turning the predicate into a predicate of intervals would have to be delegated to a different functional projection, e.g., TenseP(hrase). That, however, would mean that that higher projection would not have a unified semantics – it would have a more narrow function if ViewAsp is projected, and a bigger one when there is no ViewAsp. Assuming universal projection of ViewAsp in Slavic, for which projection language learners have positive evidence in the form of necessarily 92 PERFECTIVE readings for perfective forms, helps to prevent such complications, and will therefore be upheld in the here presented analysis. 5. The Syntax of Slavic aspect 5.1. The morphosyntax of prefixation The syntactic representation of Slavic aspect has to account for a number of facts. The first and obvious one is the telicity-marking role of prefixes. I will interpret Slavic aspectual prefixes to be functional heads carrying the [telic] feature, and more specifically to be the heads of InnerAspP, the first projection dominating VP. These prefixes are inserted into the structure with their phonological specification. For cases where the prefix is of the ―pure perfectivizing‖ kind, I will assume that InnerAspP is headed by just the feature [telic], and its overt realization is post-syntactic. Given that the form of the ―pure perfectivizer‖ is frequently not predictable, the speller (i.e., spelling mechanism) will in such cases look for an entry in the encyclopedia of a form that corresponds to the syntactic chunk being spelled out and includes a given root (inserted at the terminal under VP). When such entry does not exist, as, for example, with many new verbs, the language-particular default prefix will be used. Lastly, for a very small group of verbs in Polish (and a larger set in South Slavic), the realization of the [telic] feature will be phonologically null, e.g., Ø-kupić ‗to buy‘, Ø-dać ‗to give.‘ 20 One could possibly conceive of a different approach to ―pure perfectivizers,‖ namely that they, too, are inserted into the structure as functional head together with their phonological specification, and it just happens that such prefixed forms correspond to 20 For kupić in Polish, there is also a stylistic alternative with the inceptive za-: za-kupić, where the prefix also functions as a ―pure perfectivizer‖ (i.e., the secondary imperfective form cannot be derived), and which sounds a bit more formal. A choice also exists for a few other verbs, where more than one overt prefix can function as a pure telicity-marker. 93 encyclopedic entries that are synonymous, sans the additional element of telicity, with their unprefixed counterparts. Such a take would be in line with the arguments made in Isačenko (1962) and Czochralski (1975), where it is argued that there is no significant difference between meaning-modifying and ―pure perfectivizers,‖ with the latter only appearing to be lexically empty due to the overlap with verb meaning. However, as I have pointed out earlier, the grammar of most Slavic languages 21 treats ―pure perfectivizers‖ differently from other prefixes in that forms that include them cannot be secondarily imperfectivized. The interpretation of ―pure perfectivizers‖ as realization of a feature rather than a prefixes subject to early insertion is motivated by this very fact, and I will discussed it in more detail when the phenomenon of secondary imperfectivization is considered in Chapter 4. Based on the above discussion, we can represent different types of prefixation in the following way (the theme vowel and the infinitive hosting projections are omitted): (67) Figure 4: Morphosyntax of telicity in Slavic a. meaning modifying prefixes, e.g, za-pis(-a-ć) ‗to write down‘ InnerAsp           PREF- [telic] VP   (za-, etc)        | pis b. ―pure perfectivizers,‖ e.g, na-pis(-a-ć) ‗to write‘ InnerAsp          [telic] VP     | pis 21 The languages with the aorist-imperfect distinction are exceptions to this rule. 94 c. exceptional apparent B-perfectives. e.g., kup(-i-ć) ‗to buy‘ InnerAsp          Ø- [telic] VP       | kup A note is due here with respect to the problem of compositionality. We have seen in Chapter 2 that while aspectual prefixation is generally a synchronically productive process (in Polish, only three prefixes do not seem to be fully productive; see Ch. 2, exx (9) and (10)), there is also a number of prefixed forms whose meaning is idiosyncratic, i.e., it is not derived from the meaning of the parts. The question is whether radical decomposition of all prefixed forms into the root/verb stem and the prefix, with the latter being merged independently of V as the head of InnerAspP, does not violate lexical integrity of idiomized forms. I would offer here a suggestion due to Borer (2007, 2008) where lexical interpretation is assigned at the first functional node dominating the projection where a given vocabulary item is merged. For verbs, that first functional node would be InnerAspP, and so the projection where verbal prefixes merge as heads. 22 At that junction, a given PREF+V combination would be compared with the list of encyclopedic entries of idiosyncratic sound-meaning, and in this case also –structure combinations. If a corresponding entry is found, the idiosyncratic meaning is made available. Crucially, however, the compositional meaning as constructed syntactically 22 This idea may require some reworking of how internal arguments are merged. On Borer‘s (2005) theory, they are not merged in VP, as that would make VP a functional projection, and interpretation would have to be assigned there. Instead, internal arguments are merged in Spec, AspQ (i.e., telicity-encoding projection) for telic predicates, and in a semantically vacuous functional shell projection dominating VP for atelics. Alternatively, we could assume that VP can host arguments, but that VP does not count as a functional projection due to being headed by a lexical (as opposed to functional) head. In the discussion below I will allow this possibility for a more economical presentation of argument projection. 95 should also be available. Whether and to what extent this is true for all idiomatic cases remains a topic for further empirical study. 5.2. Projection and interpretation of arguments A second issue in the syntax of Slavic aspect is the projection of arguments of imperfective and perfective verbs. While this problem is not the focus of this work and has not been discussed here, there are interpretational differences between arguments of perfectives and imperfectives. More specifically, in Slavic languages without articles (i.e., all languages except Bulgarian and Macedonian), bare plural and mass arguments of perfectives are said to receive definite/specific interpretation (Piñón 2001), while the interpretation of arguments of imperfectives is not restricted in this way, but is rather contextually determined. This could be seen in the translations given to some prior examples, but is also illustrated in (68) below. (68) a. Hania czytała książkę / książki / prasę. Hannah.NOM read.3SG book.ACC / books.ACC / press.ACC ‗Hannah read/was reading a/the book / (the) books / (the) press.‘ b. Hania prze- czytała książkę / książki / prasę. Hannah.NOM PREF-read.3SG book.ACC / books.ACC / press.ACC ‗Hannah read a/the book//the books/the press.‘ What happens on the side of the perfective verb is essentially equivalent to what we see in English, only that in Slavic it is not the properties of the object that determine the interpretation of the predicate, but rather the other way around. One can see why telic predicates would be incompatible with mass and bare plural objects – while telicity is marked on the verb in Slavic, when a predicate has a direct object, the culmination is 96 computed with respect to the object (cf. argument-event homomorphism of Dowty 1991, Krifka 1998). Hannah read 5 books means that Hannah completed reading all 5 books, and Hannah read the books means that Hannah completed reading all the books in a contextually defined set. Culmination could not be computed in such cases if the object were unbounded. This is also essentially the difference in the interpretation in the objects of perfective verbs. The property they impose on their objects is not specificity or definiteness, as is typically stated, but rather boundedness. That is why the interpretation of singulars is not affected – they are bounded, and therefore compatible with telic interpretation. The mass and bare plural nouns, on the other hand, are unbounded, and receive a bounded interpretation in the context of telic predicates. 23 The notion of object boundedness is naturally relevant for computing culmination, but one has to wonder if it could not be related to the viewpoint aspect as well. After all, temporal boundedness surely imposes limits on event participants. One cannot read an 23 Slavic languages have a way of encoding non-specific reading for the objects of some perfective verbs, which is accomplished through the use of genitive marking. Verbs participating in such accusative/genitive alternation include e.g., drink, eat, buy, bring, and take.Some illustrative examples are given below. (i) a. Hania gotowała zupę/*zupy. Hannah cooked.impf soup.ACC/soup.GEN ‗Hannah cooked soup.‘ b. Hania u-gotowała zupę/zupy. Hannah PREF-cooked.perf soup.ACC/soup.GEN ‗Hannah cooked the soup/soup.‘ (ii) a. Bartek o-brał ziemniaki/ziemniaków. Bartek.NOM peel.perf potatoes.ACC/potatoes.GEN ‗Bart peeled the potatoes/some potatoes.‘ b. Bartek o-bierał ziemniaki/*ziemniaków. Bartek.NOM peel.2ndImpf potatoes.ACC/potatoes.GEN ‗Bart peeled potatoes.‘ BUT (iii) a. Edward budował domy/*domów. Edward.NOM built.imperf houses.ACC/houses.GEN ‗Edward built houses.‘ b. Edward z-budował domy/*domów. Edward.NOM PREF-built.perf houses.ACC/houses.GEN ‗Edward built houses.‘ I attribute the differences in the interpretation to non-existential reading associated with otherwise unmotivated genitive marking. For more details, see Chapter 5, Section 2.1.2. 97 infinite number of books in a finite time or drink an infinite amount of water in a delimited time interval. Consequently, one might speculate that the PERFECTIVE viewpoint could impose bounded interpretation on the mass and bare plural objects. Such an approach would have an advantage of explaining why bounded interpretation, while obligatory with perfective verbs, is also possible with imperfective verbs. Imperfective verbs, as the reader might recall, can have either a viewpoint IMPERFECTIVE or PERFECTIVE interpretation, and should consequently allow either unbounded or bounded readings for their object, which is in fact the case. This approach would also eliminate the rationale behind an analysis like the one proposed in Slabakova (2004) and shown in (69) below, where a (Slavic-specific?) PerfAsp projection is assumed, which merges over AspP but is not in any obvious way a viewpoint aspect phrase, 24 all to ensure that the prefix has the scope over the object (raised to Spec, AspP) and assigns it an appropriate interpretation. 25 24 Slabakova (2004) argues for the telicity-marking role of Slavic prefixes and builds her main argument on the fact that in Bulgarian prefixes cannot mark viewpoint aspect, as there are separate markers dedicated to that function (i.e., the aorist form vs. the imperfect form). Consequently, looking from the Bulgarian perspective, the prefix-hosting PerfP cannot possibly be a viewpoint aspect phrase. 25 Slabakova (2004) proposes the following structures for situation types in Slavic: (i) STATE: (ii) ACHIEVEMENT: AspP AspP                          DP obj Asp‘ DP obj Asp‘                   Asp VP Asp VP [−telic]                [+telic]   t obj V‘ t obj V‘ | | V V [−telic] [+telic] 98 (69) Figure 5: Syntax of aspect marking via prefix in Slabakova (2004) vP                         t subj v'         v PerfP CAUSE  Perf AspP Prefix=[+telic]   DP obj Asp‘         Asp VP  t obj V‘ | V (iii) ACCOMPLISHMENT: (iv) ACTIVITY: PerfP PerfP         Perf‘ Perf‘                                Perf AspP Perf AspP [+telic]      [−telic]   DP obj Asp‘ DP obj Asp‘                       Asp VP Asp VP            t obj V‘ tobj V‘ | | V V [α telic] [α telic] In her analysis, both telicity and atelicity are marked syntactically, which is a departure from the usual syntax-based accounts of telicity. A more curious fact, however, is that (a)telicity is marked in different projections for states and achievements on the one hand, and for accomplishments and activities on the other (in AspP and PerfP respectively), and that in the case of the first two, the (a)telic specification seems to come from the verb (i.e., it is lexical), while in the case of the latter two, it is imposed on the telicity- neutral verb by the feature heading the PerfP. There are a number of problems with this approach. Assuming lexically specified (a)telicity in a language with morphological, transparent and fully productive marking of telicity, seems rather unnecessary. Secondly, prefixes can productively derive inchoatives (=achievements) from states as well as achievements from activities, which the representations above do not allow. Lastly, it is not clear what the function of AspP is in (iii) and (iv) beyond hosting an object. If the goal of postulating PerfP was to ensure that the prefix takes scope over the object, this is accomplished also if the object stays in Spec, VP. The author clearly felt that AspP needs to part of the representation and that the object must move to its specifier, an assumption that I share with many others with respect to telic predicates, but ended up dissociating from the Aspect Phrase the very property that it is designed to encode. 99 Of course, there are ways to derive the appropriate interpretation for the object of a perfective verb without involving viewpoint aspect as outlined above, or without assuming a vacuous AspP projection as in (69). In Krifka (1989, 1992) and Verkuyl (1999), prefixes are viewed as aspectual operators taking scope over the verb and inducing telic readings by requiring the VP to be quantized (Krifka), or to be [+T] or terminative (Verkuyl 1999), which in turn affects the interpretation of the object (presumably in Spec, VP). Alternatively, in Borer (2005), this is accomplished through specifier-head agreement within InnerAsp (her Asp Q P). This solution of course requires that (a copy of) the object be merged in Spec, InnerAsp. In Borer‘s (2005) account, Quantity objects (i.e., delimited objects deriving telic readings) are merged directly in Spec, Asp Q P. In Slavic, as I mentioned earlier, the properties of the object do not impact Aktionsart interpretation of the predicate. We could therefore assume that the direct object is merged in Spec, VP, as traditionally assumed, and then moves up to Spec, InnerAsp for case, following the movement of the verb from under VP to InnerAsp 0 , which is independently necessary to ensure adjacency between the verb and the prefix (Slavic verbal prefixes cannot be separated from the verb, as is possible, for example, with some preverbs in German or Hungarian) (e.g., Babko-Malaya, 2003, Svenonius, 2004). A different situation would arguably arise in cases where the prefix appears to introduce an argument. This is illustrated in (70) and (71). (70) a. spać ‗to sleep‘ (unergative) 26 b. za-spać ‗to oversleep‘ (unergative) BUT 26 The unaccusativity and unergativity diagnostics for Polish have been discussed in Biały (1998) and Centarowska (2000, 2002). The designations above are made based on the tests described therein. 100 c. prze-spać ‗to sleep through sth‘ (obligatory direct object) d. wyspać *(sie) ‗to get enough sleep‘ (obligatory reflexive clitic się) e. roze-spać *(się) ‗to fall into a deep sleep, to sleep to long, to feel sleepy after sleeping too short‘ (obligatory reflexive clitic się) f. ode-spać ‗to make up for lost sleep (during named time)‘ (optional dir. object) (71) a. pisać ‗to write‘ (optional direct object) b. od-pisać ‗to reply‘ (unergative, optional PP argument) c. do-pisać ‗to be favorable‘ (unaccusative) d. o-pisać ‗to describe‘ (optional direct object) e. pod-pisać ‗to sign‘ (optional direct object) BUT f. wy-pisać ‗to write out, to prescribe‘ (obligatory direct object), also possible with the reflexive clitic: wy-pisać się ‗to disenroll‘ g. przy-pisać ‗to ascribe‘ (obligatory direct object) h. s-pisać ‗to copy, to lift the content (by writing)‘ (obligatory direct object) i. roz-pisać ‗to make sth work/write‘ (obligatory object), also possible with the reflexive clitic: roz-pisać się ‗to write extensively‘ The argument-introducing/requiring role of Slavic prefixes was discussed in Spencer & Zaretskaya (1998) and Slabakova (2004). In those cases, we could assume that the argument is merged directly in Spec, InnerAsp. As a result, we would have the following possibilities: 101 (72) Figure 6: Atelic and telic structures and the direct object a. atelic structures w/direct object VP                                      DP V | V b. telic structures w/direct object of V InnerAsp           DP InnerAsp‘               PREFIX [telic] VP              t DP V‘ | V c. telic structures w/direct object of PREFIX InnerAsp           DP InnerAsp‘              PREFIX [telic] VP              |   V 5.3. Composing with outer aspect The final step in aspectual composition is the addition of viewpoint aspect. As I explained earlier, I assume ViewAsp to project universally in Slavic, with the phonological realization of ViewAsp heads being null in most Slavic languages (e.g., Polish, Czech, Russian, but not Bulgarian or Macedonian). This gives us the following representations: 102 (90) Figure 7: Combiniatorial possibilities of (a)telicity and viewpoint aspect in Slavic and their syntax a. atelic, IMPERFECTIVE viewpoint ViewAsp         -Ø [IMPERF] VP                                  b. atelic, PERFECTIVE viewpoint ViewAsp         -Ø [PERF] VP                      c. *telic, IMPERFECTIVE viewpoint d. telic, PERFECTIVE viewpoint ViewAsp           -Ø [PERF] InnerAsp                        PREF [telic] /[telic] VP                        6. Summary In this chapter I considered the aspectual contribution of (non-quantificational) Slavic verbal prefixes, and the contrast in aspectual interpretation between unprefixed and prefixed forms with respect to Aktionsart and viewpoint aspect. Based on the clear and almost universal correlation between unprefixed forms and atelicity on the one hand, and prefixed forms and telicity on the other, I concluded, in agreement with many previous accounts, that prefixes are markers of telicity, and argued that the objections raised by others are either invalid or insufficient to deny this generalization. An analysis 103 to the contrary would also essentially claim that there is no grammatical telicity-marking strategy in Slavic, as neither of the strategies known from other languages, such marking through the properties of an argument or a modifier PP, case-marking, or marking through a dedicated particle, is used in Slavic. I have further proposed that prefixes merge as functional heads in the telicity-encoding InnerAsp projection, except for the ―pure perfectivizing‖ prefixes, which are overt realization of a [telic] feature in InnerAsp 0 . With regard to viewpoint aspect, I showed that despite the traditional label, Slavic B- imperfectives can have either IMPERFECTIVE or PERFECTIVE interpretation, while perfectives are always PERFECTIVE. I have attributed this asymmetry to the restriction that the IMPERFECTIVE viewpoint may not compose with telic predicates, which can be grounded in the difficulty in applying interval semantics to culminating events. I proposed that viewpoint aspect in the case of both viewpoints is marked in Slavic through phonologically null morphology, except for languages such as Bulgarian and Macedonian, where the aorist-imperfect distinction has been preserved and is functional (where it has been only partially preserved, e.g, Serbo-Croat, or where it has been aligned with the perfective-imperfective distinction, it has likely lost its viewpoint-marking function). In the next chapter, I consider the interpretation of secondary imperfectives – forms derived from prefixed verbs through regular and non-lexical in nature suffixation. It is frequently noted that the role of the secondary imperfective is to undo the contribution of the prefix. If our analysis so far is correct, it predicts that the secondary imperfective undoes telicity, rather than marks the IMPERFECTIVE viewpoint. As I will show in Chapters 4, this prediction is borne out. 104 CHAPTER 4: THE SECONDARY IMPERFECTIVE 1. Secondary imperfective: the problem In the previous chapter I argued that verbal prefixes in Slavic are markers of telicity, and that prefixed forms are interpreted PERFECTIVELY (i.e., as temporally bounded), while atelic, unprefixed forms can have either IMPERFECTIVE (i.e., temporally unbounded) or PERFECTIVE readings. 1 Now we turn our attention to another prominent kind of aspectual morphology in Slavic, namely the so-called secondary imperfective (henceforth S-imperfective) suffixation, which derives imperfective verbs from prefixed ones. While a lot of attention has been given recently to the role of aspectual prefixation, the process of secondary imperfectivization has remained largely understudied and its semantic role is typically assumed rather than investigated. 1.1. Secondary Imperfective morphology In comparison to aspectual prefixation, S-imperfectivization is much more regular – both in terms of the range of the involved morphology as well as its semantic effect. The morphological details of S-imperfective derivation were discussed in Chapter 2, Section 4. Here I repeat only illustrative examples of ways of S-imperfective formation from Polish, noting that the type in (a) is the most productive, while the type in (c) limited to a handful of forms. 1 The definitions of the IMPERFECTIVE and the PERFECTIVE viewpoints are repeated below for easy reference: i. [[ PERFECTIVE]] = λP (vt) .λt (i) . ∃e (v) [τ(e) ⊆ t & P(e)] (Event time is included in Reference time) ii. [[ IMPERFECTIVE]] = λP (vt) .λt (i) . ∃e (v) [t ⊂ τ(e) & P(e)] (Reference time is properly included in Event time) 105 (1) a. –ywa suffix (historically –ova): prze-pisać → prze-pis-yw-ać ‗to copy‘ od-czytać → od-czyt-yw-ać ‗to read out‘ b. -a suffix with vowel mutation wy-robić → wy-rabi-a-ć ‗to manufacture‘ wy-jeść → wy-jad-a-ć c. suppletion przy-jść → przy-chodzić ‗to come‘ za-łożyć → za-kładać ‗to put on‘ 1.2. Secondary imperfectivization – the semantic effect As for the semantic effect, S-imperfectivization never changes the verb‘s lexical meaning, in contrast to much prefixation. Instead, its function appears to be to undo/cancel the aspectual contribution of the prefix (lexical meaning change introduced by the prefix is retained), which is evidenced in the fact that S-imperfectives have exactly the same distribution as bare imperfectives (i.e., unprefixed imperfectives, henceforth B- imperfectives). This, again, was shown previously in Chapter 2, exx. (13)-(18), with respect to a number of Slavic tests for imperfectivity, including the availability of present tense denotation for present tense forms, participation in the formation of analytic future, compatibility with phase verbs such as ‗begin‘ or ‗end,‘ compatibility with the V 1 and V 1 ‗on and on‘ construction, the availability of deverbal nominal forms (this contrast obtains only in South and East Slavic), and the pattern in the formation of participles. It is worth noting that there are no grammatical test where S-imperfectives would pattern together with perfectives to the exclusion of B-imperfectives. Thus, clearly, S-imperfectives must 106 share some semantic feature(s) with B-imperfectives that are not present in perfective verbs. 1.3. SI – marker of viewpoint or …? We have seen previously that prefixed perfectives are all telic 2 and viewpoint- PERFECTIVE. This would seem to give us two possible candidates for the semantic feature that would distinguish between perfective and imperfectives verbs: telicity and PERFECTIVITY. Recall, however, that B-imperfectives are in fact viewpoint aspect ambiguous, and can have either IMPERFECTIVE or PERFECTIVE interpretation. This has been pointed out in Forsyth 1970, Comrie 1976, Gawrońska 1993, and Paslawska & von Stechow 2003, and was also shown in the previous chapter through the application of various viewpoint-sensitive tests. If we add to this fact the strong native intuition that the relationship between corresponding perfective and S-imperfective verbs, like the ones in (2)b, is of exactly the same kind as the relationship between corresponding B- imperfective and perfective verbs, as the ones in (2)a – the intuition that is commonly referred to as aspectual pairing (see, e.g., Młynarczyk 2004) – it would seem natural to hypothesize that the role of S-imperfectivization could be to undo the telicity-marking contribution of the prefix, and therefore render the predicate atelic. (2) a. B-imperfective: pisać ‗to write atelic ‘ → perfective: na-pisać ‗to write telic ‘ b. perfective: prze-pisać ‗to copy telic ‘ → S-imperfective: prze-pis-yw-ać ‗to copy‘ 2 I set aside here semelfactive forms in Slavic, which are said to be atelic (Smith 1991), but which pattern with perfectives in tests for (im)perfectivity, and clearly are compatible only with the PERFECTIVE viewpoint – as they are punctual, they could not have an unbounded interpretation. It is not clear to me why semelfactives must in fact be categorized as atelic. Smith‘s (1991) argument seems to be that they do not produce a change of state (e.g., knock, blink). However, while in English iterative interpretation could be assigned to such predicates, in Slavic in the absence of a multiplying adverbial (e.g., many times, 5 times, etc.), they have a single, punctual event interpretation, and would not pass any tests for homogeneity. With the multiplying adverbial, by the way, the predicate would be clearly telic if the compatibility with a completive adverbial were to be any indication. Without it, semelfactives can take neither completive nor durative adverbials, just like achievements. 107 Yet, despite the clear logic of this hypothesis, it is never entertained. In fact, even those who otherwise depart from the traditional viewpoint aspect characterization of the Slavic aspectual contrast and identify the role of prefixes as telicity markers, continue to assume a viewpoint aspect role for the S-imperfective as a marker of the IMPERFECTIVE viewpoint (e.g., Hewson & Bubenik 1997; Jabłońska 2004, Slabakova 2004, Svenonius 2004, among many others). This so-far unchallenged assumption seems to be rooted in at least four different considerations. First, despite the frequent recognition of the Aktionsart role of verbal prefixes, there is the historical bias toward viewing Slavic aspectual contrast in terms of viewpoint aspect, which is evident in the employed terminology. Secondly, the S- imperfective morphology is regular, with a limited number of exponents, and non-lexical in nature, and therefore quite different from the notorious prefixes. As such, it is more likely to be viewed as a marker of viewpoint aspect (cf. Smith 1991). Thirdly, there is the issue of the S-imperfective morphology being added onto prefixed verbs, and so arguably representing a second layer of aspectual meaning, and therefore likely outer (=viewpoint) aspect. Lastly, there are the similarities with the English progressive, which has been considered a species of viewpoint aspect, and which, when applied to telic predicates, gives rise to atelic readings (cf. John built a house and John is building a house). The last two reasons, i.e., morphosyntactic and comparative, would seem to encourage a viewpoint aspect analysis of the S-imperfectives. And yet, there are facts that clearly suggest a different approach. First, there is the question of why the S-imperfective morphology, assuming it is a marker of the IMPERFECTIVE viewpoint, does not attach to 108 simplex verbs to unambiguously mark IMPERFECTIVE readings. 3 Note that the English progressive applies to atelic and telic predicates alike. Secondly, there is the issue of the actual viewpoint aspect interpretation of S-imperfectives. As it turns out, and as I will show later, S-imperfectives do not need to be interpreted IMPERFECTIVELY. Instead, they are like B-imperfectives in that they allow both IMPERFECTIVE as well as PERFECTIVE readings. Lastly, there is the problem of the role of S-imperfectivization in Old Church Slavonic and Modern Bulgarian, languages with a separate system of marking PERFECTIVE and IMPERFECTIVE readings in the past tense, which is the topic of the next two sections, and which provides strong evidence against viewing the S-imperfective morphology as a marker of viewpoint aspect. 2. Old Church Slavonic Old Church Slavonic (henceforth OCS) is the language of the oldest Slavic manuscripts, dating back to the 10 th or 11 th century A.D, and ―the earliest form of Slavic known, a form very close to the language called Proto-Slavic or Common Slavic which was presumably spoken by all Slavs before they became differentiated into separate nations‖ (Lunt 2001:1). OCS is special within the Slavic language family in that it possessed, alongside the previously described aspectual morphology (i.e., prefixation and S-imperfectivization), a system of overt marking of PERFECTIVE and IMPERFECTIVE viewpoint aspects in the past tense. As such, this language can provide a unique opportunity of examining how the S-imperfective morphology interacts with the 3 S-imperfective morphology can in fact attach to simplex imperfectives to a limited degree in some Slavic languages (e.g., Polish and Russian, but not Bulgarian), but the relevant forms have unambiguosly habitual interpretation rather than the more general IMPERFECTIVE one. Historically, the S-imperfective morphemes might have originated as markers of habituality, and then assumed a different function. The residual use of the S-imperfective morphemes with simplex imperfectives is the evidence for that historically earlier and now parallel use of the relevant morphology. 109 viewpoint categories of PERFECTIVE and IMPERFECTIVE, and can help decide its actual aspectual function. 2.1. Viewpoint aspect markers in OCS In OCS, the markers of viewpoint aspect were the aorist (for the PERFECTIVE viewpoint in the past), and the imperfect (for the IMPERFECTIVE viewpoint in the past). The figure in (3) shows the morphological template for past tense forms, while the Tables 3 and 4 show generalized paradigms for the imperfect and the aorist. (3) V-stem + IMPERFECTIVE/ + past tense + AGR PERFECTIVE marker marker Table 3: The OCS Imperfect paradigm (cf. Lunt 2001:100) Singular dual plural 1 st person V-ĕa-h-ъ V-ĕa-h-ovĕ V-ĕa-h-omъ 2 nd person V-ĕa-š-e V-ĕa-š-eta V-ĕa-š-ete 3 rd person V-ĕa-š-ete V-ĕa-h-õ Table 4: The OCS Aorist paradigm (cf. Lunt 2001: 102, 104) singular dual plural 1 st person V-(o)- h/Ø/s -ъ V-(o)- h/Ø/s- ovĕ V-(o)- h/Ø/s -omъ 2 nd person V-(e)-Ø- Ø e (e)s V-(o)- s-ta V-Ø- eta V-(o)- s-te ete 3 rd person V-(o)- s-te V- Ø- ete V-(o)- šẽ õ sẽ 110 2.2. Aspectual paradigm in OCS and the role of the SI 2.2.1. Compositional possibilities The OCS aorist-imperfect system co-existed with the already well-established (though of later origin) system of aspectual prefixation as well as the SI derivation (Schuyt 1990, Lunt 2001). The two aspectual systems combined together to produce the following set of forms: (4) (i) simplex imperfective imperfect (ii) simplex imperfective aorist (iii) perfective imperfect – semantically restricted, see the discussion below (iv) perfective aorist (v) secondary imperfective imperfect (vi) secondary imperfective aorist 2.2.2. The distribution of the SI vis-à-vis the aorist/imperfect distinction As the list in (4) shows, the S-imperfective morphology in OCS was clearly independent of the IMPERFECTIVE-marking morphology in that it (i) did not appear in all imperfect forms, and (ii) could co-occur with the aorist. The number of attested examples for the pattern in (4)-vi is somewhat limited, but they do exist, as illustrated in (5) and (6). (5) a. po-klanjasta sę jemu (Codex Suprasliensis 103a,22) bowed-down.SI.AOR REFL him.DAT ―They bowed down before him.‖ Cf. b. the perfective aorist form: po-kloni c. the S-imperfective, imperfect form: po-klanjaaše (Suprasliensis 103a,3) 111 (6) a. tvoimь imenemь vĕsy iz-gonixomъ your.INSTR name.INSTR evil-spirits.ACC drove-out.SI.AOR ―In your name we drove out the evil spirits‖ (Codex Marianus, Matt 7:22) Cf. b. the perfective aorist form: vьsę iz-gъna i crkve (Codex Marianus, John 2:15) all.ACC drove-out.AOR out-of church ―he drove all out of the temple‖ More examples if this kind will be given when discussing Bulgarian. The fact that they can be found indicates that the role of the S-imperfective was not to mark the IMPERFECTIVE viewpoint. This is further confirmed by the possibility of adding the imperfect morphology to perfectives in the absence of the S-imperfective marker, which makes it impossible to argue, for example, that the S-imperfective morphology is a spell- out of the IMPERFECTIVE feature in the presence of a telic predicate. It should also be pointed out that the S-imperfective in OCS was not just a marker of habituality, as it is sometimes suggested (Van Wijk 1927, Amse de-Jong 1974), because while S- imperfectives could certainly have habitual interpretation, as they do in modern Slavic languages, this was by no means a necessity. This is illustrated in the following examples (relevant forms are shown in bold): (7) Codex Marianus, Matt 27:48 J abie tekъ edinъ otъ nixъ . J priemъ gõbõ . Jsplьnь ocьta . J vьznezъ na trь{ь}stь napaěše i . 112 ‗Immediately one of them ran and got a sponge filled with vinegar, put it on a stick, and offered it to Jesus to drink.‘ (8) Codex Marianus, John 2:23-24 egda že bě vъ iměx . vъ pasxa vъ prazdьnikъ . mъnodzi věrovašę vъ imę ego . vidęšče znameniě ego eže tvorěaše . samъ že isъ ne vъdaaše sebe vъ vro, ixъ ‗While he was in Jerusalem at the Passover Feast, many people believed in his name having seen the miracles he was doing. But Jesus did not entrust himself to them.‘ (9) Codex Marianus, John 5:18 ěko ne tъkmo razarěaše soboto, . nъ i oca svoego glaaše ba . ‗Because not only was he breaking the Sabbath, but he was even calling God his own Father.‘ The facts above show clearly that the analysis of the S-imperfective as a marker of viewpoint aspect has to be rejected, at least with respect to OCS, and that an alternative analysis is needed. What the alternative should be becomes clear when one considers what kind of forms the S-imperfective attaches to, and it attaches to perfectives, which in the OCS system, where viewpoint aspects are marked by a distinct set of morphemes, are characterized by just one property: telicity. Consequently, the role of the S-imperfective must be related to Aktionsart marking, and more specifically consist in reversing what the prefix does, and making predicates atelic. If this is correct, we have to wonder why the S- imperfective marker does not always attach to perfectives before the IMPERFECTIVE viewpoint applies, as indicated in (4)-iii, given the previously identified Slavic (and likely 113 universal) principle that the IMPERFECTIVE viewpoint may not combine directly with a telic predicate. This issue is discussed in the section below. 2.2.3. The *[perfective IMPERFECTIVE] rule in OCS and S-imperfectivization As I have proposed in the previous chapter, Slavic is subject to the following restriction: (10) IMPERFECTIVE aspect cannot compose with telic predicates. If correct, this would predict that the imperfect in OCS, the exponent of the IMPERFECTIVE viewpoint, should not combine with perfective forms. This is in fact generally the case. While perfective imperfect forms were possible in OCS, they were rare (Amse-deJong 1974, Schuyt 1990, Lunt 2001). This is reflected in the following table summarizing the use of such forms in the corpus of OCS (Dostál 1954:600). Table 5: The interaction of Aktionsart with viewpoint aspect in the OCS corpus (Dostál 1954:600) Imperfective Perfective Imperfect 99% 1% Aorist 40% 60% Further, the perfective imperfect forms were restricted to non-referential contexts, e.g., subjunctive or habitual, i.e., contexts which express possibility or probability rather than the assertion that a particular event has taken place. Examples of such uses of the perfective imperfect are provided in (11)-(14) below (the relevant forms are shown in bold) (Amse-deJong 1974). 114 (11) Suprasliensis 207,15 ašte prošaaše kto otъ ništiixъ nĕčьso dati sebĕ to ašte sę sъlojčaaše ne imĕti jemoj ničьsože dati jemoj to kotygo, sьvlĕkъ sъ sebe dadĕaše ništojojmoj tako že tvoraaše vьsa lĕta žitija svojego ‗If one of the beggars asked him to give something, then – if it so happened that he had nothing to give him – he would take off his shirt and give that to the beggar. He acted thus all the years of his life.‘ (12) Suprasliensis 346,12 bъšъjo, vo nijednoћe bĕ otь nego cъtvoreno dĕlo ježe nĕstъ dĕlo ojspĕšъn ni jedno ježe ne vьpijaše istino, ni jedno ježe ne kъ nebecьnojojmoj priętiju podvigiĕaše ojmъ. ‗For surely there was absolutely not a single deed done by him which is not a deed of good, not one which did not proclaim the truth, not one which would not lift up the soul to the heavenly reception.‘ (13) Suprasliensis 476,24 proklinaaše smokьvnico, i isъxnĕjaše ‗[If] he cursed the fig tree, then it would wither.‘ (14) Suprasliensis 73,9 ne vьsegda li jegda načьnĕĕxomъ sę brati glagolaaxomъ ĕalmosa sego ‗Did we not always say this psalm whenever we were to begin the fight?‘ How could we account for such use of the perfective imperfect in the light of the general prohibition on such forms? We can start with the observation that, as the English translations indicate, the interpretation of the relevant forms is actually bounded, 115 PERFECTIVE. We could therefore hypothesize that the role of the imperfect (=IMPERFECTIVE) morphology in such contexts is not actually to mark a viewpoint aspect, and that the imperfect morphology is not interpreted in its usual way there. Similar facts have been reported by Iatridou (2000) for aspect (and tense) marking in future less vivid counterfactuals in Greek, with extension to other languages. She notes that in such contexts, exemplified in (15)-(17) below, the IMPERFECTIVE and past tense morphology is ‗fake,‘ i.e., the IMPERFECTIVE viewpoint and the past tense are not actually interpreted. Instead, the aspectual interpretation is actually PERFECTIVE, 4 and neither the antecedent nor the consequent of the conditional are interpreted as referring to the past. (15) An eperne afto to siropi θa γ 1 inotan kala. (Iatridou 2000:236) if take.PST.IMP this syrup FUT become.PST.IMP well ‗If he took this syrup, he would get better.‘ (16) An peθene o arχ 1 iγos θa ton θavame stin korifi tu vunu. if die.PST.IMP the chief FUT him bury.PST.IMP on-the top the mountain ‗If the chief died, we would bury him on the top of the mountain.‘ (17) An pandrevotan mia prigipisa, θa esoze tin eteria tu. if marry.PST.IMP a princess FUT save.PST.IMP the company his ‗If he married a princess, he would save his company.‘ As Iatridou explains, (16) does not mean that if the chief were in the process of dying, we would be in the process of burying him. Similarly, (17) does not mean that during the wedding ceremony the groom would be saving his company (Iatridou 2000:236). The fact that IMPERFECTIVE forms above actually get PERFECTIVE interpretation is also 4 A different situation arises for Present Counterfactuals (i.e., statements contrary to the present situation), where IMPERFECTIVE interpretation is possible (see Iatridou 2000:255). 116 confirmed by their compatibility with completive adverbials, something which is not normally possible. This is shown in (18). (18) An eχtizes to spiti (mesa) se ena mina θa prolavenes if build.IMP the house in one month FUT ―have time enough‖.IMP na to pulisis prin to kalokeri. to it-sell before the summer ‗If you built this house in a month, you would be able to sell it before the summer.‘ In contrast to the use of the IMPERFECTIVE, PERFECTIVE marking cannot be ―fake‖ in this way. Instead, it must be properly interpreted, alongside with tense, leading to the future neutral vivid reading, in the terminology of Iatridou, where future-oriented interpretation is not possible (cf. If he had married a princess, he would have saved his company). Exactly the same facts obtain with respect to the IMPERFECTIVE and PERFECTIVE marking in wish-expressing constructions in Greek. Iatridou explains the ―fake‖ uses of past tense marker by reanalyzing its function in a more general and non-time specific way. She proposes that it corresponds to an exclusion relation such that Topic(x) (―the x that we are talking about‖) excludes C(x) (―the x that for all we know is the x of the speaker‖), where the variable x can range over times, and in the modal environments, over worlds. When x ranges over times, the exclusion feature, realized overtly through past tense morphology, conveys that the topic time excludes the utterance time. When x ranges over worlds, it conveys that the topic worlds exclude the actual world. With respect to the IMPERFECTIVE morphology, she does not have a similar reanalysis, but rather proposes that the IMPERFECTIVE marking in the 117 relevant environments is a default, and that aspect is not specified but rather interpreted contextually (it will be interpreted as PERFECTIVE as in the examples above, but as IMPERFECTIVE in Present Counterfactuals 5 ). The same, it is argued, extends to the unexpected use of the IMPERFECTIVE in habitual and generic contexts, and in the subjunctive. The conclusion that we might draw from the discussion above is that IMPERFECTIVE (and tense) morphology found in modal contexts does not have its regular semantic function. 6 Returning now to OCS, this can explain why we find perfective imperfects there in the first place. If the IMPERFECTIVE morphology is default/not actually interpreted in the relevant contexts, whatever semantic incompatibility exists between telic forms and the IMPERFECTIVE viewpoint is now not relevant. The scarcity of such forms proves now not that there is some morphological incompatibility between the two, but rather that there is a semantic incompatibility, and that the contexts where that compatibility is not at play (because the relevant morphology is dissociated from its usual 5 Iatridou attributes this difference to the fact that ―when the discourse is about the present/now, (…) eventualities are discussed as ongoing; that is, they have an imperfective/progressive interpretation. But when the discourse is about the future, an eventuality can be discussed as either ongoing or completive.‖ (Iatridou 2000:256). 6 I am not entirely sure whether the default analysis is in the end the correct one, and there are two reasons for that. One is that in some languages, it is the PERFECTIVE marker that might be considered the primary/default one, either because it is morphologically simpler, or because PERFECTIVE forms are used for basic statement of facts. Secondly, habitual contexts (which pattern with CFs, generics and subjunctives in licensing the ―default‖ IMPERFECTIVE) as well as non-referential present (e.g., historic present), seem to also neutralize the grammatical contribution of other markers. For example, in Polish, SI forms, which I will argue to be structurally and morphologically marked as atelic/homogenous, allow telic interpretation in habitual contexts or in historic present, e.g., (i) Codziennie/Wtedy przepisuję notatki godzinę/w godzinę. every-day/then I-copy.2NDIMPF notes hour/in hour ‗Everyday/Then I copy (the) notes for an hour/in an hour.‘ Another example of special contexts neutralizing aspectual marking would be Upper Sorbian, where it is reported that present tense forms of perfectives, which can normally have only future time denotation, can be used as performatives and for non-referential (e.g., historical) present (Toops 1998, following Faβke 1981). Nevertheless, I certainly agree with Iatridou that modal and other non-referential contexts are special and might license unexpected morphology. 118 semantics) just happen to not be common in the recorded narratives that supply us with language data. Once the examples in (11)-(14) are explained away, the generalization in (10) is proven to extend to OCS: the IMPERFECTIVE viewpoint cannot combine with telic predicates. For prefixed forms to combine with the IMPERFECTIVE viewpoint, they must first undergo S-imperfectivization. Thus, S-imperfectivization can be viewed as a semantic process with a morphological reflex which allows for IMPERFECTIVE interpretation to apply to perfective (=telic) verbs, which in turn means that the S- imperfective renders telic predicates atelic. This function, while it facilitates the composition of the IMPERFECTIVE viewpoint with prefixed verbs, is nevertheless independent of viewpoint aspect, which is evidenced by the fact that atelicized forms can combine with either viewpoint aspect. Why such an operation would be needed in Slavic can be easily understood when one considers the fact that Slavic prefixes in the overwhelming number of cases modify the verb‘s lexical meaning. Therefore S- imperfectivization provides a mechanism to obtain atelic counterparts of telic predicates that do not have a corresponding simplex atelic form. 3. Bulgarian 3.1. The Bulgarian paradigm Just like with OCS, the peculiarity of Modern Bulgarian within the Slavic context is that it has a productive system of overt marking of viewpoint aspects as well as aspectual prefixation and S-imperfectivization. Bulgarian aspectual system is a continuation of the system inherited from OCS, with some inevitable 119 morphophonological changes. Examples of the inflected verb forms are given in 0 and 0 below. Table 6: The aspectual paradigm for the verb 'write' in the past tense in Bulgarian imperfect aorist simplex imperfective pish-e-še pis-a prefixed perfective (na-pish-e-še) na-pis-a secondary imperfective na-pis-va-še (na-pis-va) Table 7: The aspectual paradigm for the verb 'drink' in the past tense in Bulgarian Imperfect aorist simplex imperfective pi-e-še pi prefixed perfective (iz-pi-e-še) iz-pi secondary imperfective iz-pi-va-še (iz-pi-va) The examples in (19) and (20) show the use of the forms that are not in brackets and the range of allowed interpretations 7 . (19) write a. Ivan pisheshe knigata si Ivan wrote the-book REFL ‗Ivan was writing his book.‘ (as in e.g., answer to the question "What was Ivan doing when you called him?"); ‗Ivan worked on his book.‘ (as in e.g., answer to the question "What did Ivan do every weekend last year?") b. Ivan pisa knigata si Ivan wrote the-book REFL ‗Ivan worked on his book.‘ (as in e.g., answer to the question "What did Ivan do 7 I would like to thank Roumi Pancheva for help with all the Bulgarian examples. 120 yesterday?", or as in answer to the question "What did Ivan do every weekend last year?" – the habitual reading) c. Ivan napisvashe knigata si Ivan wrote the-book REFL ‗Ivan was about to finish his book.‘ (as in e.g., answer to the question "What was Ivan doing when you called him/when his wife left him?"); also habitual: ‗Every day last year Ivan was about to finish his book.‘ d. Ivan napisa knigata si Ivan wrote the-book REFL ‗Ivan finished his book.‘ (as in e.g., answer to the question "What did Ivan accomplish last year?") (20) drink a. Ivan pieshe kafe/kafeto Ivan drank coffee/the-coffee ‗Ivan was drinking coffee/drank from the coffee.‘ (when Mary called him); also habitual, as in 'Every day last week Ivan drank coffee/drank from the (packet of) coffee.‘ b. Ivan pi kafe/kafeto Ivan drank coffee/the-coffee ‗Ivan drank coffee/drank from the coffee.‘; also habitual, as in ‗Every day last week Ivan drank coffee/drank from the (packet of) coffee.‘ c. Ivan izpivashe (*kafe)/edno kafe/kafeto Ivan drank (coffee)/one cup of coffee/the coffee 121 ‗Ivan was about to finish a cup of coffee/the coffee.‘; also habitual as in ‗Every day he drank a cup of coffee.‘ d. Ivan izpi (*kafe)/edno kafe/kafeto Ivan drank (coffee)/one cup of coffee/the coffee ‗Ivan finished a cup of coffee/the coffee.‘ 3.2. *perfective IMPERFECTIVE rule in Bulgarian The noteworthy fact about the Tables 6 and 7 is that two of the paradigm forms are either restricted or not frequently used. The restricted forms are, once again, the perfective imperfects. Such forms are not normally listed as part of the verbal paradigm in Bulgarian, as they would be ungrammatical in regular declarative contexts such as (19)c and (20)c. Instead, the perfective verb requires the addition of the S-imperfective suffix (expressed in the forms above with –va) in order to combine with the IMPERFECTIVE viewpoint, just as we saw earlier for OCS. Just like in OCS as well, perfective imperfects are marginally possible, and the contexts which license such forms are, again, modal and non-episodic: subjunctive or habitual. This restriction is shown in the examples (21)-(23), with sentences in (a) representing referential contexts, and sentences in (b) and (c) representing non-referential contexts. (21) a. *Ivan napisheshe knigata si Ivan wrote the-book self ‗Ivan was finishing his book./ Ivan was about to finish his book.‘ b. Ako Ivan napisheshe knigata si, shteshe da poluchi kola za nagrada if Ivan wrote the-book self would SUBJ receive car for award 122 ‗If Ivan were to finish his dissertation, he would get a car as an award.‘ c. Kogato Ivan napisheshe pismo, poluchavashe nagrada when Ivan wrote letter received award ‗Whenever Ivan wrote a letter, he received an award.‘ (22) a. *Ivan izpieshe kafe/kafeto Ivan drank coffee/the coffee ‗Ivan was drinking up the coffee/Ivan was about to drink up the coffee.‘ b. Ako izpieshe (*kafe)/edno kafe/kafeto, shteshe da mu mine glavata if drank (coffee)/one cup of coffee/the coffee, would SUBJ him pass the-head 'If he were to drink a cup of (the) coffee, his headache would go away.‘ c. Kogato izpieshe (*kafe)/edno kafe/kafeto, mu minavashe glavata. when drank (coffee)/one cup of coffee/the coffee, him passed the-head ‗Whenever he drank up a cup of (the) coffee, his headache would go away.‘ (23) a. * Ivan si zapisheshe adresa Ivan REFL wrote-down the-address ‗Ivan was writing down the address.‘ b. Ako Ivan si zapisheshe adresa, ne bichme da se zagubim if Ivan REFL write down the-address, not would SUBJ REFL get-lost ‗If Ivan were to write down the address we would not get lost.‘ c. Kogato si zapisheshe adresa ne se zagubvashe when REFL wrote-down the-address not REFL get lost 123 ‗Whenever he wrote down the address he did not get lost‘. The Bulgarian data above allows us to empirically confirm the generalization based on the statistical data from OCS, namely that the IMPERFECTIVE viewpoint cannot and could not combine with perfective verbs in contexts where viewpoint aspect morphology is interpreted in the expected way. The observed incompatibility is resolved, just as in OCS, through the S-imperfective derivation. The other paradigm form that is not used regularly is the S-imperfective aorist form. The reasons for that, however, are likely not semantic but contrast-related. As it turns out, such forms are homophonous in Bulgarian with present tense forms of S- imperfectives (Roumyana Pancheva, p.c.), and are likely disfavored for pragmatic reasons. It should be stressed, however, that they are in principle possible and semantically unproblematic, as evidenced by the following examples taken from Gramatika Na Sĕvremennja Bâlgarski Knizhoven Ezik (The Grammar of the Contemporary Literary Bulgarian, 1993, pp. 314-316) (the S-imperfective morphology in bold) (24) Jurtalana govori dъlgo, vika, ryza go, za-plash-v-a go, Jurtalana spoke.AOR long, cried.AOR, swore.AOR him, threatened.AOR him, moli my se begged.AOR him REFL (25) Toj se v-slush-v-a edna-dve mi-nuti he REFL in-listened.AOR one-two minutes The facts presented above reaffirm the conclusions drawn on the basis of OCS, and support the alternative approach to S-imperfectivization pursued here. In the next section, 124 I consider how the Aktionsart approach fairs with respect to Slavic languages without the aorist-imperfect distinctuion. 4. The Secondary Imperfective in Polish In the previous two sections I showed that in OCS and Bulgarian, languages with a system of marking viewpoint in the past distinct from verbal prefixation and S- imperfectivization, the role of the S-imperfective was not related to viewpoint, but instead belonged in the domain of Aktionsart. In this section, I will consider, on the example of Polish, whether this novel analysis of S-imperfectivization can be extended to other Slavic languages, or whether Slavic languages with no imperfect/aorist contrast may warrant a different analysis. The key questions that will need to be answered are: (A) Do S-imperfectives forms denote telic or atelic situations? (B) What kind of viewpoint interpretation can a S-imperfective form receive? 4.1. S-imperfectives and atelicity As I showed in Chapter 2, and as is uncontroversially accepted, B-imperfectives in Slavic have atelic interpretation. Applying the durative/completive adverbial test to such predicates always yields the same result: durative adverbials are allowed, completive ones are not. S-imperfectives, as it turns out, pattern exactly the same way, providing a two-way contrast between B-imperfectives and S-imperfectives on the one hand, and perfectives on the other. This is exemplified in (26)-(27): (26) a. Jan pisał list 15 minut/*w 15 minut. John wrote.IMPF letter 15 minutes/in 15 minutes ‗John was writing a letter for 15 minutes/*in 15 minutes.‘ 125 b. Jan za-pis-yw-ał adres 3 minuty/*w 3 minuty. John wrote-down.SI address 3 minutes/in 3 minutes ‗John was writing down a/the address for 3 minutes/*in 3 minutes.‘ Cf. c. Jan za-pisał adres *3 minuty/w 3 minuty. John wrote-down.PERF address 3 minutes/in 3 minutes ‗John wrote down a/the address *for 3 minutes/in 3 minutes.‘ (27) a. Rząd prowadził rozmowy w sprawie reform 5 lat/*w 5 lat. government conducted.IMPF talks in matter reforms 5 years/in 5 years ‗The government conducted reform talks for 5 years/*in 5 years.‘ b. Rząd prze-prowadz-a-ł reformy 5 lat/*w 5 lat. government implemented.SI reforms 5 years/in 5 years ‗The government was implementing reforms for 5 years/*in 5 years.‘ Cf. c. Rząd prze-prowadził reformy *5 lat/w 5 lat. government implemented.PERF reforms 5 years/in 5 years ‗The government implemented reforms *for 5 years/in 5 years.‘ The telic entailment test similarly shows that S-imperfectives do not have telic interpretation: (28) Jan zapisywał.SI adres ≠> Jan zapisał.PERF adres ‗John was writing down a phone number‖ ≠> ―John wrote down a phone number‘ (29) Rząd przeprowadzał.SI reformy ≠> Rząd przeprowadził.PERF reformy ‗Government was implementing reforms‖ ≠>―Government implemented reforms‘ Naturally, one could argue, the failure of telic entailment and the incompatibility of the S- imperfective forms with completive adverbials could stem either from the atelic nature of 126 the involved predicates or from the application of the IMPERFECTIVE viewpoint to telic predicates, as it happens in the case of the English progressive 8 . While in OCS and Bulgarian the existence of dedicated IMPERFECTIVE markers pointed to an alternative role for the S-imperfective morphology, in most modern Slavic languages, the aorist/imperfect distinction does not exist, and so the S-imperfective could have potentially assumed the role of the IMPERFECTIVE viewpoint marker. This in fact is the universal understanding of the role of the S-imperfective offered in the literature, as I mentioned earlier. Fortunately, the two competing analyses – the viewpoint analysis advocated in the literature, and the Aktionsart analysis proposed here – make two specific predictions that allow us to test how each analysis squares with the facts. One prediction is distributional and the other semantic, and they are discussed in the following section. 4.2. The S-imperfective: viewpoint aspect marker vs. Aktionsart marker 4.2.1. Distribution of the S-imperfective morphology The first prediction concerns the distribution of the S-imperfective morphology. On the Aktionsart analysis, where the S-imperfective has an atelicizing function, it is expected to attach only to telic verbs. On the viewpoint aspect analysis, on the other hand, we would expect the S-imperfective to be able to mark the IMPERFECTIVE viewpoint on a wider range of predicates, i.e., including B-imperfectives, which are perfectly compatible with IMPERFECTIVE interpretation. In fact, since these verbs are viewpoint aspect neutral (Forsyth 1970, Comrie 1976, Gawrońska 1993, Paslawska & von Stechow 2003), allowing either viewpoint interpretation, and since in Polish there is no 8 I suggested in the previous chapters, that the role of the English progressive is likely more complex than just relating temporal intervals. Best evidence for that are the many modality-based analysis of the progressive that have been proposed (e.g., Dowty 1979, Landman 1992, Portner 1998, Zucchi 1999). For the sake of argument, however, I will ignore this complication, especially that the viewpoint aspect analysis of secondary imperfectivization will ultimately be concluded to not hold. 127 aorist/imperfect contrast, the S-imperfective morphology, on its assumed viewpoint aspect role, could be particularly useful as a way to disambiguate the intended viewpoint. This, however, does not happen. The S-imperfective suffix does not attach productively to simplex imperfectives. Further, in the limited number of cases when such forms are licensed, the role of the SI morphology is not to mark the IMPERFECTIVE viewpoint, but rather to mark habituality. This is illustrated below, where (30)d shows limited productivity of S-imperfectivization with respect to B-imperfectives. 9 (30) a. pisywać: pis ROOT -yw SI -a THEMEVO -ć INFINITIVE ‗to write-habitual‘ b. czytywać: czyt ROOT -yw SI -a THEMEVO -ć INFINITIVE ‗to read-habitual‘ c. bywać: by ROOT -(y)w SI -a THEMEVO -ć INFINITIVE ‗to be-habitual‘ d. *malowywać ‗to paint;‘ *płakiwać ‗to cry;‘ *liczać ‗to count;‘ *budowywać ‗to build,‘ etc. The allowed forms above are unambiguously habitual and cannot have general IMPERFECTIVE interpretation. This is demonstrated by the examples in (31), which show that progressive interpretation in particular is excluded: (31) a. Maria pisywała *list / listy. Mary wrote letter / letters ‗Mary wrote letters.‘ / ‗*Mary wrote a letter.‘ b. Maria pisywała listy (*przez godzinę). Mary wrote letters (for an hour) ‗Mary used to/would write letters.‘ # ‗Mary was writing a letter/letters.‘ 9 The Russian speakers I consulted rejected such forms in habitual contexts in favor of simplex imperfectives. Borik (2002) also reports that the relevant forms in Russian have a memoir flavor, suggesting they are focilized, archaic formations, and not part of the modern grammar of Russian. 128 The habitual-marking role of the S-imperfective morphology can be understood as a remnant of the older, possibly original function of the relevant suffixes, which has been preserved to a limited degree till the modern times (Schuyt 1990). Its remnant status is confirmed by the fact that habitual derivation has disappeared completely from some Slavic languages (e.g., Bulgarian), and that it is limited to a subset of the vocabulary in languages which still allow it (e.g., Polish). In Polish in particular, this habitual morphology is (a) most readily found on verbs that form S-imperfectives by means of the less productive suffixes and/or vowel alternation (e.g., jeść ‗to eat‘ – jadać ‗to eat- habitual; pić ‗to drink‘ – pijać ‗to drink-habitual‘), (b) found idiosyncratically on forms which take the more productive S-imperfective –yw suffix (e.g., być ‗to be‘ – bywać ‗to be-habitual,‘ but szyć ‗to sow‘ - *szywać cf. √przy-szywać ‗to sow onto‘), and (c) never found on forms which contain the currently most productive verbalizing suffix –owa (e.g., malować ‗to paint‘ – *malowywać cf. √za-malowywać ‗to paint over‘; pracować ‗to work‘ – *pracowywać cf. √od-pracowywać ‗to make up work‘). The limited productivity and a distinct role of the S-imperfective morphology in the context of simplex imperfectives argue for the separation of the two kinds of the S- imperfective derivation, and they support the generalization that the currently productive kind of S-imperfectivization applies only to perfective, and therefore telic, verbs. This, in turn, supports the Aktionsart analysis of S-imperfectivization. 4.2.2. Viewpoint Interpretation of S-imperfectives in Polish The other prediction that can help in selecting the right analysis concerns the viewpoint aspect interpretation of S-imperfectives. On the viewpoint aspect approach, they are expected to always have IMPERFECTIVE interpretation. If they can be interpreted 129 PERFECTIVELY, the analysis is incorrect. On the Aktionsart approach, the prediction is different, namely that the S-imperfectives are not overtly marked for viewpoint aspect, and because the viewpoint interpretation of atelic predicates is not restricted in Slavic (in contrast to telics, which can compose only with the PERFECTIVE viewpoint), they should be able to receive either viewpoint interpretation. In other words, on the Aktionsart approach, S-imperfective forms should behave just like B-imperfectives and allow either IMPERFECTIVE or PERFECTIVE reading. To see what the actual interpretational facts are, I will employ a test with two predicates and a time frame adverbial. The time frame adverbial will specify the reference time for the speaker‘s viewpoint, and the two predicates will allow us to see if the events described by the S-imperfective forms have to be understood as filling the whole reference time (= IMPERFECTIVE viewpoint) and therefore overlapping with the coordinated event, or if they can be interpreted as occurring in a sequence, and therefore coextensive only with a sub-interval of the reference time (= PERFECTIVE viewpoint). In the first example, (32), the S-imperfective form is coordinated with the predicate ‗sleep,‘ which, due to world knowledge, should preclude a concurrent reading for another agentive activity. If the resulting sentence is felicitous, it indicates that PERFECTIVE interpretation for the S-imperfective form is possible. In the second example, (33), two S- imperfective forms are coordinated, which, if the prediction made here is correct, should allow either an IMPERFECTIVE or a PERFECTIVE reading. (32) √Między 5:00 a 8:00, Hania spała i odrabiała lekcje. between 5:00 and 8:00, Hania slept.IMPF and did.SI homework ‗Between 10 and 11, Hania slept and did her homework.‘ 130 (33) Wczoraj rano wystawiałam stopnie i wypisywałam świadectwa. yesterday morning I-give.SI grades and wrote-out.SI report cards ―Yesterday morning I gave grades [to students] and wrote out report cards.‖ Reading 1: I gave grades and wrote a report card one student at a time throughout the morning. (IMPERFECTIVE interpretation) Reading 2: I first have grades to all the students and then wrote the report cards. (PERFECTIVE interpretation) The examples above indicate that PERFECTIVE reading is certainly available for S- imperfectives. In fact, in (32), this is the only possible reading, as further confirmed by the fact that the same sentence would not be possible with a modifier such as ‗from X time to X time,‘ which appears to require the IMPERFECTIVE interpretation for the involved predicates: (34) a. #Od 5:00 do 8:00, Hania spała i odrabiała lekcje. from 5:00 till 8:00, Hania slept.IMPF and did.SI homework ‗From 5:00 till 8:00, Hania slept and did her homework.‘ Cf. b. Od 5:00 do 8:00, Hania spała.IMPF. ‗From 5:00 till 8:00, Hania was sleeping.‘ c. Od 5:00 do 8:00, Hania odrabiała.SI lekcje. ‗From 5:00 till 8:00, Hania was doing her homework.‘ d. Od 5:00 do 8:00, Hania spała.IMPF i chrapała.IMPF. ‗From 5:00 till 8:00, Hania was sleeping and snoring.‘ e. #Od 5:00 do 8:00, Hania odrobiła.PERF lekcje. ‗From 5:00 till 8:00, Hania did homework.‘ 131 Example in (34)e shows that a perfective verb, which has only viewpoint PERFECTIVE interpretation, cannot co-occur with the ―from X-time till X-time‖ adverbial. Since our test example in (34)a is also incompatible with this adverbial, it shows once again that the two events must be interpreted PERFECTIVELY, allowing for a sequential reading. Consequently, it must be concluded that the S-imperfective form can receive a PERFECTIVE reading. Finally, we can also use the test which relies on the interaction between tense and aspect. As I reported in Chapter 2, present tense forms of PERFECTIVE verbs can have only future time interpretation, which can be attributed to their temporally bounded interpretation, which disallows a temporal coincidence relation with the utterance time. Simplex imperfectives, on the other hand, do allow temporal coincidence, which shows they can be interpreted IMPERFECTIVELY, but can also have future denotation, which in turn indicates they can also have bounded, PERFECTIVE readings. As it turns out, the exact same interpretational possibilities characterize S-imperfectives, as illustrated below. (35) O-trzym-uj-ę jutro nagrodę. I-receive.SI tomorrow award ‗I am receiving an award tomorrow.‘ (36) Od dziś, za-prasz-a-m do nas tylko rodzinę. from today I-invite.SI to us only family ‗From now on, I will invite only family.‘ If these future uses of S-imperfective present tense forms are truly anchored in their PERFECTIVE viewpoint interpretation, as I take to be the case here, we can expect that they could not have progressive reading. That, too, turns out to be true, as shown in (37). 132 (37) *Jutro od-pocz-yw-am gdy będziesz w pracy. tomorow I-rest.SI while you-will-be at work ‗I will be sleeping tomorrow while you are at work.‘ To get the progressive reading, an analytic future construction has to be used instead: (38) *Jutro bedę od-pocz-yw-ał/od-pocz-yw-ać gdy będziesz w pracy. tomorow I-will rest.SI.PAST/rest.SI.INF while you-will-be at work ‗I will be resting tomorrow while you are at work.‘ These examples indicate that, when used to refer to the future, S-imperfectives have bounded (=PERFECTIVE) reading. This, in turn, means that S-imperfective suffixes are not markers of IMPERFECTIVITY, contrary to what is normally assumed, and therefore provides support for the Aktionsart analysis of S-imperfectivization as proposed here. 4.3. Conclusions The distributional and interpretational facts from Polish in conjunction with the OCS and Bulgarian facts show unequivocally that the universally adopted analysis of the S-imperfective as a marker of the IMPERFECTIVE viewpoint is incorrect. The Aktionsart approach taken here, allows us to account for the interpretational patterns, and explains why S-imperfectives pattern together with the viewpoint-neutral B-imperfectives – they are all atelics, allowing either viewpoint interpretation. How this new analysis should be formalized and related to syntax is considered in the next section. 5. Analysis 5.1. Setting the stage The data considered above shows clearly that the role of S-imperfectivization could not be viewpoint aspect marking. Instead, its function must be to undo the 133 grammatical contribution of the prefix, i.e., telicity. Consequently, the secondary imperfective must be interpreted as a marker of Aktionsart, on a par with verbal prefixes, which is able to maps telic predicates onto atelic ones. This role of the S-imperfective is crucially independent of any viewpoint aspect consideration (i.e., it is not systematically motivated by viewpoint). At the same time, thanks to its semantics, it helps to resolve the incompatibility between the IMPERFECTIVE viewpoint and telic predicates which characterizes all Slavic languages and is likely a cross-linguistic phenomenon. As I argued earlier, that incompatibility or difficulty with computing IMPERFECTIVE interpretation on a telic predicate is not unexpected. To reiterate the point made in the previous chapter, notice that when the English progressive applies to a telic predicate, the reference time representing the speaker‘s viewpoint must exclude the culmination point. In other words, only the process part of the event, which is not necessarily co-extensive with a contiguous time interval, can be included in the reference time. In consequence, the English progressive consistently gives rise to atelic readings, cf., (39) simple past (PERFECTIVE viewpoint) a. John built a house. => he finished it (event culminated) b. John built a house in one year/for one year. (40) progressive (IMPERFECTIVE viewpoint) a. John was building a house. ≠> he finished it (event did not culminate) b. John was building a house *in one year/for one year. While the standard definition of the IMPERFECTIVE viewpoint (i.e., ―The reference time is included in the event time‖), does effectively exclude the culmination point for 134 predicates where the culmination coincides with the left or right temporal boundary of the event, something else is needed for predicates where the culmination point is not so perfectly aligned (e.g., to fill sth, to cross sth, to spill sth). In those cases, the reference time cannot simply pick any subpart of the event, but must instead select only (a part of) the activity/process component of the event. This essentially involves ignoring or ―seeing through‖ the InnerAspP layer in telic predicates. The English progressive appears able to do that, but that does not necessarily mean it is a one-step operation. On the contrary, the many modal analyses of the English progressive that have been advanced (e.g., Dowty 1979, Landman 1992, Portner 1998, Zucchi 1999, among others), motivated largely by the problems posed by telic predicates, clearly indicate that something more than a viewpoint operator is at play. We might speculate that when it comes to telic predicates, a two-part process is needed for the application of the English progressive: an atelicizing/homogenizing part (essentially neutralizing the InnerAsp [telic] layer) and the addition of the viewpoint predicate. We could further hypothesize that the –ing morpheme is the exponent of the first operation (which could be related to its commonly recognized stativizing role, e.g., Frajzyngier 1985), and the copula indicates temporal coincidence between the time of the event and the reference time. How many languages make such a two-step process available remains an empirical issue. Bohnemeyer & Swift (2001) report that cross-linguistically one observes a preference to apply IMPERFECTIVE viewpoint to atelic predicates only, and PERFECTIVE viewpoint to telic ones. Language acquisition data also indicates that children prefer to use atelic predicates in the IMPERFECTIVE and progressive aspect, and telic ones in the PERFECTIVE aspect only (English: Bloom, Lifter & Hafitz 1980, Sachs 1983, Shirai and 135 Anderson 1995, Clark 1996; Hebrew: Berman 1985; Italian: Antinucci and Miller 1976; Mandarin: Li 1995; Turkish: Aksu-Koc 1998; Walpiri: Bavin 1992; Inuktitut: Swift 2000; all cited in Bohnemeyer & Swift 2001). It is possible that some languages do not have a mechanism to derive IMPERFECTIVE readings for telic events. At the same time, there are clearly languages that do. English would be one example, as outlined above. Slavic languages would be another, with the S-imperfective morphology fulfilling the ―telicity-neutralizing‖ role, taking as its input a telic predicate, and returning at atelic one. 10 5.2. SI as a partitive homogenizer When one considers the apparent two-step process in the application of the progressive described above, one parallel that comes to mind is the use of the partitive of in the quantification of definite phrases. 11 Consider the following examples, with formal representations in (41)b and (41)c: (41) a. some of the girls, some of the soup b. [ QP Q-DET [ PP of [ DP D [ NP N]]]] c.  Q-det <e,t>           of <e>      D <e,t> a. some of the girls b. some of the soup 10 A modal analysis of the progressive/IMPERFECTIVE could attribute to the SI suffix the introduction of modality (i.e., access to alternative worlds). That would not, however, explain the use of the S-imperfective marker with the PERFECTIVE viewpoint, as illustrated earlier, as well as its occurrence in nominalized forms with no temporal or viewpoint interpretation, as will be illustrated later. 11 I am grateful to Roumi Pancheva for bringing this parallel to my attention. 136 In the structure in (41)c above, of represents a function which takes a group-level individual and returns the set of atoms corresponding to the generator set (Ladusaw 1982, Comorovski 1995). Consequently, of the girls is the same type of object as girls, and so in effect the contribution of the definite article ‗the‘ as creating an individual is neutralized by of. Notice that here the logical type of the argument is changing, but the meaning of the girls (unique set) does not disappear. Such is also the case with S- imperfectivization: the Aktionsart of the predicate is reversed back to the original, atelic value, and for the purpose of composition with viewpoint aspect as well as with respect to tests for (a)telicity it behaves as an atelic predicate, but the semantic contribution of the prefix does not go away entirely. Rather, the residual effect of the prefix in the form of the existence of some culmination point, which is not part of the asserted event, but which exists in the possible extension of the timeline on which the event time is located is maintained (cf. Klein 1994; Maslov 1958 talks, in reference to Bulgarian, about maintaining the nuance of terminativity introduced by the prefix). Another parallel between S-imperfectivization and the partitive construction in (41) is that it applies to definite descriptions. It is a frequently expressed sentiment that perfective verbs in Slavic are comparable to definites in the nominal domain (most recently the idea of the definiteness as a feature of perfectives was expressed in Ramchand 2004). More narrowly, they seem to have properties of definite mass descriptions, e.g., the soup, in that the addition of the prefix turns a mass-like predicate (i.e., activity or state) into an indivisible whole. 12 That definite mass descriptions are 12 Definite descriptions have also referential interpretation, which seems to be the property of perfectives as well, evidenced in their strictly referential use (as opposed to the habitual, modal, and other non-episodic uses of the imperfective, with the exception of Czech, where perfectives can have habitual interpretation, and Upper Sorbian, where they are reported to have various non-referential uses, Toops 1998). However, it 137 indivisible is evidenced by the fact that they cannot be quantified, e.g., *some/*much the soup. The same property in Slavic perfectives translates into their exclusively episodic interpretation. 13 Now, if the prefix applies to a mass object (activity, state) and by virtue of supplying culmination, returns an individualized, indivisible event, then in order to access parts of that event, a partitive operator has to be used. In the nominal domain, the semantics of the partitive operator has been proposed to be the following: (42) [[of PART ]] = λxλPλy[P(y)  y  x] (Ladusaw 1982, Hoeksma 1984) 14 Notice that the seal-off/indivisible nature of perfectives could incidentally provide an alternative explanation for why the IMPERFECTIVE viewpoint cannot apply to such predicates – the event they denote simply cannot be analyzed in terms of parts. In order to make it analyzable and return it to a quantifiable state, we need a partitive operator. While the parallel with partitivity in the nominal domain is certainly compelling, note that the role of the S-imperfective could not be reduced to just partitivity. More specifically, it is not just the access to the internal structure of the event that the S- imperfective is concerned with, but also that the part(s) it returns is/are homogenous (=atelic). In fact, intuitively, this is its primary role – to apply to a telic predicate and return an atelic one. This is why S-imperfectivization does not need a grammatical trigger, unlike nominal partitivity, which applies only in the context of quantification is not clear if the referentiality comes from the prefix, or rather from the ViewAsp projection, which I assume here to be responsible for existential closure of the event variable, without which referential reading should not be possible. 13 Again, as indicated in fn. 17, Czech and Upper Sorbian might provide counterexamples to this generalization. 14 Barker (1998) proposes a modification to this scheme, where inclusion is replaced with proper inclusion. As he argues ―[t]his will guarantee that any property denoted by a partitive will have at least two entities in extension, and cannot uniquely identify an individual.‖ (Barker 1998:679). 138 (i.e., I like (*of) the girls, I am trying (*of) the soup). We could call this role homogenizing and add it as a condition to the partitive operator, whereby event 2 (e 2 ), a part of event 1 (e 1 ), has to be homogenous. This can be formulated as follows (the definitions of cumulativity and divisity, the two conditions of homogeneity, are adopted from Borer 2005: 147). (43) [[SI]] = λeλPλe’[P(e)  e’  e  HOM(e’)] “The S-imperfective operator takes an event of which predícate P holds, and returns a homogenous subpart of that event.” [[HOM]] = λe [CUM(e)  DIV(e)] [[CUM]] = e [P(e)  P (e2‘) → P(e ∪ e‘)]. [[DIV]] = e [P(e) → e‘[(P(e‘)  e‘ < e)]]  e,e‘[P(e)  P(e‘)  y < x → P (e−e‘)]. Note that (43), even though it selects a sub-part of a telic event, is different from the IMPERFECTIVE viewpoint aspect, which selects a subinterval. More specifically, the IMPERFECTIVE viewpoint does not make reference to the event‘s internal structure, but only to its temporal duration. The [[SI]], on the other hand, does what is implicit in the semantics of the English progressive, thought not a part of how it is normally defined, i.e., it returns the process portion of the event. Once (43) applies, either viewpoint can then apply, but they will now be evaluated with respect to the homogenous event returned by (43), and not the embedded event described by InnerAsp [telic] . In other words, we can now receive a temporally bounded or unbounded reading of the process part, which I have shown to be the case for Polish S-imperfectives, and which is evident in the use of the imperfect as well as the aorist marking with S-imperfectivized forms in OCS. 139 Syntactically, I will assume that the SI operator merges as the head of InnerAsp [SI] phrase, which itself projects over InnerAsp [telic] . Merging it with a VP would be unmotivated, hence we do not see forms *śpiew-yw-ać ‗to sing,‘ or *szuk-iw-ać ‗to look for.‘ 15 The application of InnerAsp [SI] gives us a third, new kind of an Aktionsart structure: (44) Figure 8: The syntax of the secondary imperfective InnerAsp [SI]     SI InnerAsp [telic]   [telic]/PREFIX [telic] VP   With this in place, the following Aktionsart structures are now possible in Slavic: (45) Figure 9: Types of Aktionsart structures in Slavic (a) atelic predicates (b) telic predicated (c) derived atelic predicates InnerAsp [SI]                InnerAsp [telic] SI InnerAsp [telic]                                         VP [telic]/PREFIX [telic] VP [telic]/PREFIX [telic] VP                         5.3. Why do we have S-imperfectivization? As mentioned above, unlike with partitive of in the nominal domain, the S- imperfective can apply without a trigger – it can happen whether or not a predicate needs to compose with the IMPERFECTIVE viewpoint. This begs the question why we would have of the soup-type expressions in the verbal domain. The answer to that seems to be of the functional kind. Consider the previously highlighted fact that for the overwhelming 15 Again, as explained in section 4.2.1., such forms are possible for some verbs, but they have strictly habitual reading and involve homophonous, non-productive morphology. 140 majority of prefixed verbs, the aspectual prefix also changes the verb‘s lexical meaning. Some illustrative examples are provided in (46) below (for more, please see Chapter 2). (46) PREFIX + pisać ‗to write‘ a. do-pisać ‗to write sth additional; to be favorable‘ b. na-pisać ‗to write up‘ (non-lexical prefix) c. o-pisać ‗to describe‘ d. od-pisać ‗to write back; to deduct (from taxes)‘ e. pod-pisać ‗to sign‘ f. prze-pisać ‗to copy; to prescribe; to rewrite; to sign sth over to sb‘ g. przy-pisać ‗to ascribe‘ h. roz-pisać ‗to make a schedule; to make a dry pen write; to write extensively (with the reflexive się) i. w-pisać ‗to write in; to sign up‘ j. wy-pisać ‗to write out‘ k. s-pisać ‗to copy from‘ l. za-pisać ‗to write down; to prescribe‘ Many of these derived meanings would not necessarily correspond to telic predicates in other languages, and yet in Slavic, because preverbs are charged with a grammatical role, prefixed forms are always telic. Consequently, S-imperfectivization provides a mechanism for deriving atelic variants of telic verbs which do not have simplex atelic counterparts. For the verbs in (46), these would be: (47) PREFIX + pisać + SI ‗to write‘ a. do-pis-yw-ać ‗to write sth additional; to be favorable‘ 141 b. *na-pis-yw-ać ‗to write up‘ (non-lexical prefix) c. o-pis-yw-ać ‗to describe‘ d. od-pis-yw-ać ‗to write back; to deduct (from taxes)‘ e. pod-pis-yw-ać ‗to sign‘ f. prze-pis-yw-ać ‗to copy; to prescribe; to rewrite; to sign sth over to sb‘ g. przy-pis-yw-ać ‗to ascribe‘ h. roz-pis-yw-ać ‗to make a schedule; to make a dry pen write; to write extensively (with the reflexive się) i. w-pis-yw-ać ‗to write in; to sign up‘ j. wy-pis-yw-ać ‗to write out‘ k. s-pis-yw-ać ‗to copy from‘ l. za-pis-yw-ać ‗to write down; to prescribe‘ This process is referred to in Slavic as aspectual pairing, and it is felt by native speakers to be equivalent in function to aspectual pairing by means of prefixation (Młynarczyk 2004). This intuition is accounted for by the analysis of S-imperfectivization presented here, as it identifies one common feature that underlies the pairing process, namely (a)telicity: (48) Aspectual pairing: creating atelic-telic pairs (i) by means of telicity-marking prefix: atelic (pis-) → telic (na-pis-) ‗write‘ (ii) by means of secondary imperfectivization telic (za-pis-) → atelic (za-pis-va-) ‗write down‘ 142 5.4. Imperfective = S-imperfective = atelic Importantly, the nature aspectual pairing goes beyond some vague intuition, as it is evidenced in the common distribution of simplex imperfectives and S-imperfectives in a wide range of syntactic-semantic contexts, which were discussed earlier in the chapter. While some of the tests are sensitive to the IMPERFECTIVE viewpoint aspect interpretation allowed by imperfectives and disallowed by perfectives, other tests clearly refer to Aktionsart rather than viewpoint. One example of such test is the formation of –ne nominals in Bulgarian (comparable to the English –ing nominals). In Bulgarian (and also in Russian and other East Slavic languages), such nominals can only be derived from imperfective stems, be they simplex or secondary. This is shown in (49) and (50): (49) ‗writing‘ a. pisane : pis Root -a ThemeVowel -ne N b. *napisane : na Prefix -pis Root -a ThemeVowel -ne N (50) ‗re-writing/copying‘ a. *prepisane : pre Prefix -pis Root -a ThemeVowel -ne N b. prepisvane : pre Prefix -pis Root -v SI -a ThemeVowel -ne N The forms above do not show any evidence of the presence of viewpoint aspect, as there is no temporal interpretation as well as no aorist or imperfect morphology. The Aktionsart morphology, on the other hand, is notably and uncontroversially present in the form of the prefix in prepisvane. 16 Given that the prefixed perfective forms *napisane and *prepisane are excluded, we can infer that –ne nominals cannot be built on telic stems. The atelic, activity-denoting pisanie, on the other hand, can be nominalized in this 16 Also in Greek, Bulgarian neighbor with the aorist and imperfect distinction, deverbal nouns can include inner aspect, but no viewpoint aspect (Markantonatou 1995, Alexiadou 2001). 143 way. Now, if the role of S-imperfectivization is to derive atelic predicates from telic ones, we would expect to be able to derive –ne nominals from S-imperfectivized forms. As the example in (50)b shows, this is exactly what happens. Consequently, the presence of the S-imperfective morphology in –ne nominals provides further evidence that it is a marker of Aktionsart, and more specifically – as argued here – an partitive-homogenizing operator. Another data proving the same point comes from Polish, where certain nominalizing affixes cannot attach to perfective stems. These are, for example, the suffix –acz, naming agents or instruments of individual-level predicates, and the profession/instrument/tool naming suffix –arka. It can be seen easily why these suffixes would attach to imperfective verbs – they are names of agents of activities or possessors of habits/abilities. As it turns out, these suffixes attach to either B-imperfective or S- imperfective forms, which further confirms the atelicizing role of the secondary imperfective morphology, as activities and states are atelic. This is exemplified in (51) and (52) below. (51) –acz 17 a. pal-acz ‗a stoker‘ b. kręt-acz ‗a dishonest person, lit. barber lit. shaver‘ c. o-twier-acz ‗a (can) opener‘ cf. * o-tworz-acz d. s-prysk-iw-acz ‗a sprinkler‘ cf. *s-prysk-acz ‗a sprinkler‘ (52) –arka a. wiert-arka ‗a drill‘ 17 Similar examples can be found in Bulgarian (aspectual morphology in bold), e.g., pre-pis-v-ač ‗person who copies on a test, homework, etc.,‘ raz-kaz-v-ač ‗person who tells a story.‘ 144 b. śmieci-arka ‗a garbage truck‘ c. wy-ładow-yw-arka ‗a (mechanical) off-loader‘ cf. * wy-ładow-arka d. z-my-w-arka ‗a dish washer‘ cf. *z-myj-arka Lastly, both prefixes as well as S-imperfective suffixes can be part of nominals without any categorizing morphology beside gender marking (which is null for Nominative masculine forms in Polish and Slavic in general). Some examples are given below, with (54)c and (54)d showing forms with S-imperfective morphology. (53) √bieg: bieg N ‗a run; a gear (automobile)‘, biec ‗to run-directed motion,‘ bieg-a-ć V ‗to run-habitual, nondirectional motion (used in SI formation)‘ (a) wy-bieg ‗runway, outlet‘ (cf. wy-biec ‗to run out‘) (b) prze-bieg ‗course (of an event)‘ (cf. prze-biec ‗to run across, run (a distance, a course)‘) (c) za-bieg ‗medical procedure‘ (cf. za-biec ‗to get in front of sb/sth by running;‘ za-biegać ‗to get in front of sb/sth by running; to pursue‘ (d) z-bieg ‗a runaway criminal‘ (cf. z-biec ‗to run down, to run away/escape‘) (54) √rw: rwa N ‗lumbago‗*ryw, rwać ‗to pull, to tear‘ (a) wy-rwa ‗a big dent‘ (cf. wy-rwać ‗to pull, tear out‘) (b) prze-rwa ‗a break‘ (cf. prze-rwać ‗to break‘) 145 (c) pod-ryw ‗a pick-up (of a girl or a boy)‘ (cf. pode-rwać ‗to pick up (a girl, a boy),‘ pod-rywać.SI) (d) po-ryw ‗a gust of wind‘ (cf. po-rwać ‗to snatch, to kidnap,‘ po-rywać.SI) The presence of prefixes and S-imperfective morphemes in forms in (53) and (54), which have no temporal or viewpoint interpretation whatsoever, indicates that both types of morphology are not markers of viewpoint. Though certainly it may be challenging to, if one were to attempt that, accommodate the functional projections hosting these morphemes in the syntactic representations underlying the forms above, Aktionsart morphology is incomparably more likely to be part of encyclopedically stored representations with idiosyncratic meanings than are any viewpoint markers. 6. Aspectual triplets a.k.a. verbal triangles 6.1. The problem I have suggested earlier that the key motivation for having a general atelicizing strategy is the need to derive atelic counterparts for verbs that lack a more basic (i.e., unprefixed, unsuffixed) mates. That leads to a question of what happens when the InnerAsp [telic] is headed not by a meaning-modifying prefix, but by a [telic] feature, which I argued to be the case when the prefix is of the ―pure perfectivizing‖ (i.e., non-lexical) kind. The corresponding structures for both types of situations are repeated in 0 below. 146 (55) Figure 10: Lexical vs. non-lexical prefixes a. meaning modifying prefixes, e.g, za-pis(-a-ć) ‗to write down‘ InnerAsp [telic]      PREFIX [telic] VP | | za- pis b. ―pure perfectivizers,‖ e.g, na-pis(-a-ć) ‗to write down‘ InnerAsp [telic]         [telic] VP | pis → na-pis- Adding the partitive-homogenizing projection to 0a,b would give us the following: (56) Figure 11: Secondary imperfectivization of perfective forms a. meaning modifying prefixes, e.g, za-pis-yw(-a-ć) ‗to write down‘ InnerAsp [PART-HOM]      SI InnerAsp [telic]            PREFIX [telic] VP | | za- pis  za-pis-yw- b. ―pure perfectivizers,‖ e.g, na-pi-yws(-a-ć) ‗to write‘ InnerAsp [PART-HOM]      SI InnerAsp [telic]            [telic] VP | pis  na-pis-yw- As it turns out, Slavic languages different in which of the derivations in (56) are allowed. (56) a is possible and fully productive in all Slavic languages. (56)b, on the 147 other hand, is allowed only in Bulgarian and two other languages that have also retained the aorist-imperfect marking, namely Macedonian, and to a degree Upper Sorbian. All other Slavic languages generally disallow it. S-imperfective forms with a ―pure perfectivizing‖ prefix have been already exemplified for Bulgarian in the tables in 0 and 0, and later in contextualized examples, though I did not draw attention to their exceptional nature. Corresponding forms in Polish, Russian, and most other Slavic languages, are ungrammatical. This is illustrated in (57) and (58). (57) Polish a. pisać  napisać  *napisywać ‗write‘ cf. przepisywać ‗rewrite; copy‘ podpisywać ‗sign‘ b. czytać  przeczytać  *przeczytywać ‗read‘ cf. odczytywać ‗read aloud‘ zaczytywać się ‗read passionately‘ c. robić  zrobić  *zrabiać ‗do/make‘ cf. przerabiać ‗modify, process‘ odrabiać ‗make up for, do homework‘ d. płacić  zapłacić  *zapłacać ‗pay‘ cf. odpłacać ‗pay back‘ spłacać ‗pay off‘ (58) Russian a. pisat‟  napisat‟  *napisywat’ ‗write‘ 148 cf. perepisywat‟ ‗rewrite, copy‘ zapisywat‟ ‗write down‘ b. chitat‟  prochitat‟  *prochitivat’ 18 ‗read‘ cf. otchitivat‟ ‗tell off‘ zachitat‟ ‗read part of‘ c. delat‟  sdelat‟  *sdelyvat’ ‗do/make‘ cf. otdelyvat‟ ‗improve, polish‘ dodelyvat‟ ‗finish, complete‘ d. platit‟  zaplatit‟  *zaplachivat’ ‗pay‘ cf. vyplachivat‟ ‗pay out‘ priplachivat‟ ‗prepay‘ Given the correlation between the presence of the aorist-imperfect marking mechanism and the availability of forms such as na-pis-va-she, one might be tempted to relate the two and possibly attribute the ungrammaticality of *na-pis-yw-ać/*na-pis-yv- at‟ to a double role of ―pure perfectivizers‖ in Polish or Russian as markers of Aktionsart, but also of the PERFECTIVE viewpoint. This would provide a straightforward explanation. Specifically, we could hypothesize that ―pure perfectivizers‖ are exponents of two features: [telic], heading InnerAsp [telic] , and [PERFECTIVE], heading ViewAsp. With the SI operator heading an intermediate projection (e.g., InnerAsp [SI] ) and therefore standing in the way of adjacency between the features [telic] and [PERFECTIVE], the non-lexical prefixes could not surface. In Bulgarian, this problem would not arise, as a separate 18 My informant accepted this form on the habitual reading, but rejected it on the progressive one. 149 morphology marks the PERFECTIVE viewpoint, so the prefix spells out only one feature, and can therefore surface without a problem. 6.2. What’s behind the restriction in Polish? As elegant as this explanation might be, it suffers from a problem that afflicts most elegant, simple explanations, namely a limited, but nevertheless present, number of counterexamples. More specifically, forms like *na-pis-yw-ać or *prze-czyt-yw-ać can surface sometimes in Polish in the presence of a quantificational prefix, such as the cumulative na- and the distributive po-, e.g, (59) a. po-na-pis-yw-ać DISTR-[TELIC]-write-SI-INF ‗to write all over/distributive-completive‘ b. na-prze-czyt-yw-ać się (A LOT/ENOUGH-[TELIC]-read.SI-INF REFL ‗to read a lot of (stuff, books, etc)-completive‘ (the same is reported in Jabłońska 2004). At the same time, other parallel examples sound plain ungrammatical, e.g., (60) a. *po-z-rabi-a-ć DISTR-[TELIC]-do/make-SI-INF ‗to make/do distributively-completive‘ b. *na-u-gotow-yw-ać (A LOT/ENOUGH-[TELIC]-cook.SI-INF ‗to cook a lot of (food, dishes, etc)-completive‘ 150 While limited, the examples like po-na-pis-yw-ać indicate that the non-lexical prefix is not in fact a portmanteau morpheme, because the S-imperfective marking is present and the prefix still surfaces. Consequently, an alternative explanation is needed. How might we then approach examples like the ones in (59) and (60)? As it turns out, one thing that unifies the examples in (59) and contrasts them with those in (60) is the possibility of adding habitual morphology (homophonous with the S-imperfective) to B-imperfective stems. This is shown in (61). (61) a. pis-yw-ać ‗to write habitually‘ b. czyt-yw-ać ‗to read habitually‘ c. *rabi-a-ć ‗to do/make habitually‘ d. *gotow-yw-ać ‗to cook habitually‘ Further, verbs with ―pure perfectivizing‖ prefixes that allow habitual marking, can derive S-imperfective forms even in the absence of quantificational prefixes, as illustrated in (62). (62) a. jeść ‗to eat‘ → jadać ‗to eat-habitual‘ z-jeść ‗to eat-telic‘ → z-jad-a-ć b. pić ‗to drink‘ → pi-ja-ć ‗to drink-habitual‘ wy-pić ‗to drink-telic‘ → wy-pi-ja-ć I left out the interpretation of S-imperfectives in (62) on purpose, as it is not consistent across speakers. While for me they allow some progressive readings (see (63)a-c and (64)a-b below), similar to those seen in Bulgarian in (21) and (22), other speakers I consulted accepted them only as habituals (i.e., as in (63)d and (64)c). 151 (63) a. *Hania zjadała jabłko gdy zadzwonił telefon. Hania ate-up.SI apple while rang phone. ‗Hannah was finishing the apple when the phone rang.‘ b. ?Kiedy weszłam do kuchni, Hania właśnie zjadała moje ciasto. when I-entered to kitchen Hannah just ate-up.SI my cake ‗When I entered the kitchen, Hannah was just finishing up my cake.‘ c. Wchodze do kuchni, a tu Hania zjada mi moje ciasto. I-enter to kitchen, and here Hannah eats-up.SI me my cake ‗I enter the kitchen, and here is Hannah eating up my cake!‘ d. Hania zjadała codziennie jedno jabłko. Hannah eats-up.SI every day one apple ‗Hannah ate one apple a day.‘ (64) a. ??Hania wypijała kawę kiedy zadzwonił telefon. Hannah drank-up.SI coffee when rang phone ‗Hannah was finishing the coffee when the phone rang.‘ b. Hania wypijała ostatni łyk kawy kiedy zadzwonił telefon. Hannah drank-up.SI last swallow of-coffee when rang phone ‗Hannah was drinking the last swallow of coffee when the phone rang.‘ c. Hania wypijała codziennie dwie kawy. Hannah drank-up.SI every day two coffees ‗Hannah drank two cups of coffee a day.‘ In Russian, it appears, the interpretation must be habitual, as shown in (65): 152 (65) a. Ol‘ga prochityvala stranitsu knigi kazhdyi den‘. Olga read.SI page of-book every day ‗Olga read a page of a book every day.‘ b. *Olga prochitivala stronitsu knigi tseloe utro. Olga read.SI page of-book whole morning ‗Olga read a page of a book the whole morning.‘ The marginal status of these examples on the progressive reading (in Polish), and the limited number of verbs that allow such forms in the first place, all point to their fossilized nature. The more readily accepted habitual readings (exclusive in Russian), on the other hand, indicate that these forms actually involve the habitual marking look-alike to S-imperfective morphology, which I discussed briefly earlier in relation to B- imperfectives. Note also that quantificational prefixes: the cumulative na- and the distributive po-, which license otherwise ungrammatical forms *prze-czyt-yw-ać ‗to read‘ and *na-pis-yw-ać ‗to write,‘ promote pluractional readings. With these facts established, we can now account for the availability of forms such as those in (59) as well as for the idiosyncratic nature of their formation (i.e., why some are and why some are not allowed), by analyzing the S-imperfective suffix in such forms as a habitual marker. The application of this marker is, as mentioned earlier, restricted and idiosyncratic, which in turns leads to gaps in the formation of the so-called verbal triangles (Townsend 1968), and so [simplex imperfective – perfective – SI] paradigms with the same encyclopedic meaning. 153 This account leaves us with the generalization that perfective verbs with ―pure perfectivizing‖ non-lexical cannot be S-imperfectivized and/or that the following structure does not arise: (66) Figure 12: *Secondary imperfective with pure perfectivizers * InnerAsp [PART-HOM]   SI InnerAsp [telic]        [telic] VP | pis → na-pis-yw- I would argue that there are in fact no restrictions on generating this structure, and the fact that forms such as *na-pis-yw- do not surface has to do with the fact that the [telic] feature and the S-imperfective operator neutralize their semantic contribution and are not spelled out. For a structure like 0 and the root √pis, the output will be phonologically identical to the one for basic atelic structure, and so VP, i.e., pis-. 6.3. Verbal triangles in Bulgarian (and Upper Sorbian) In contrast to Polish as well as an overwhelming majority of Slavic languages, Bulgarian does not appear to restrict the formation of S-imperfectives from perfectives with pure perfectivizing prefixes. Two such examples were already shown in (19)c and (22)c. It is interesting to note, however, that even in Bulgarian, the status of forms such as na-pis-va-she is not completely clear. My informants allowed episodic, progressive readings for such forms 19 , but elsewhere it is reported that such forms are not used to 19 With thanks for language judgments to Roumi Pancheva and Todor Koev. 154 denote praesens actualis (i.e., are not referential) (e.g., Toops 1998) 20 . The habitual- marking explanation would not, however, be applicable to Bulgarian, as this language does not employ specifically habitual marking (i.e., it does not have equivalents of the Polish B-imperfective habitual forms such as pis-yw-ać ‗to write habitually‘ or jad-a-ć ‗to eat habitually‘). Therefore something else must be responsible for the unexpected surfacing of S-imperfectives with pure perfectivizers. I would suggest that in Bulgarian, in contrast to most other Slavic languages, all prefixes merge with their phonological content at InnerAsp [telic] . Consequently, all prefixed forms can participate in S-imperfectivization. That would naturally mean that in Bulgarian, ―pure perfectivizers‖ are less grammaticalized that in other Slavic languages. But this is, as it turns out, what is reported for the language. Schuyt (1990), for example, writes: ―[a]s concerns the existence of “empty” or aspectual prefixes in Bulgarian and Macedonian, it must be stated that their status is marginal. This opinion is expressed in Maslov, alongside others, who also mentions the possibility of making, without difficulty, such forms as napisvam to pf. napiša, clearly in order to maintain the nuance of terminativity introduced by the prefix‖ 21 (Schuyt 1990). 20 The actual context of use for the relevant forms reported in Toops (1998) are: historical present, modal present, iteration of completed actions in temporal subordinate clauses, and iterative/habitual present (Toops 1998: 285-286). 21 My analysis does not account for the fact that there is an interpretational difference between SIs derived from verbs with non-lexical prefixes versus those with lexical ones, namely that the ones with a non-lexical prefix always emphasize the lead-up to the culmination point (‗was about to finish doing sth.‘), while the other kind simply picks the process part of the event without any orientation with respect to its culmination. This is true of Bulgarian as well as of those secondary imperfectives with a non-lexical prefix that are marginally allowed in other Slavic languages (see comments on the next page). One could offer a functional explanation for this more narrow interpretation and say that secondary imperfectives in aspectual triplets needed to differentiate their meaning from the corresponding simplex atelics, and since what sets the two verb forms apart is the presence of the (negated) culmination point, they foreground it instead of backgrounding it as is other secondary imperfectives. 155 As for Upper Sorbian, which like Bulgarian has preserved the aorist/imperfect distinction (though it now seems secondary to the perfective/imperfective distinction, as imperfectives are expressed in the imperfect and the perfectives in aorist), and where forms like na-pis-ow-a-ć ‗to write‘ (-ow being the S-imperfective marker here) are supposed to be allowed, it is reported by Toops (1998) that while a 1954 dictionary lists a number of such forms, they have all but disappeared from more recent dictionaries (1989, 1991). Also, when such forms are used synchronically, as Toops (1998) describes, the prefix is reinterpreted as lexical (for na-pis-ow-a-ć, that would be ‗write down‘ instead of ‗write,‘ the core meaning of na- being ‗on‘ → write sth. on sth. ≈ write down). These facts might be taken to indicate that we are witnessing progressing grammaticalization of ―pure perfectivizers‖ in Upper Sorbian. Based on the facts in Polish, Bulgarian and Upper Sorbian, we can deduce that, originally, all prefixes started as lexical, merging as heads of InnerAsp [telic] (i.e., once they acquired the telicity-marking function). As the lexical meaning of some was bleached in the context of a given stem, they have become ―pure perfectivizers,‖ and therefore simply phonological exponents of an abstract feature. The process of the grammaticalization of ―pure perfectivizers‖ is advanced in most Slavic languages, a little behind in Upper Sorbian, and in early stages in Bulgarian. It can be expected based on these considerations, that Bulgarian (in contrast to, for example, Russian, Czech or Polish) does not have a default perfectivizer, i.e., a prefix with the sole role of marking telicity (and therefore a non-lexical prefix) which would be employed to derive perfective forms from newly coined or borrowed verbs. Whether that is indeed that case remains an empirical issue and a topic for further study. 156 6.4. Non-lexical prefixes: summary The discussion of Bulgarian and Polish facts above have revealed a difference in the grammar of Aktionsart marking between the two languages. While in Polish the InnerAsp [telic] projection can be headed by either an abstract [telic] feature or a prefix with phonological content, in Bulgarian only the latter is possible. This difference accounts for the availability of S-imperfectives forms that include seemingly ―pure perfectivizing‖ prefixes in Bulgarian, and for the lack of such forms in Polish and most Slavic languages, where the spell out of the complex root plus the abstract features [telic] and [SI] will return an output equivalent to the root, where the two features cancel each other out. 7. The interpretation of arguments under the S-imperfective In this final section, I will consider the interpretation of direct arguments under S- imperfectivization, this time limiting the probe to Polish only. As I indicated in the previous chapter, section 6.2, mass and bare plural arguments of perfective verbs in Slavic must receive bounded/definite/specific interpretation. This contrasts with the interpretation of argument of B-imperfectives, where no such restriction obtains. This is illustrated below with modified examples from Chapter 2. (67) a. Hania czytała książki/prasę. Hannah read.impf.PAST books/press ‗Hannah read/was reading (the) books/(the) press.‘ b. Hania prze-czytała książki/prasę. Hannah PREF-read.perf.PAST books/press ‗Hannah read the books/the press.‘ 157 This interpretational pattern allows us to make another falsifiable prediction with respect to the S-imperfectives: if it is indeed an Aktionsart-changing element and projects as the head of an InnerAsp projection, it should allow the movement of V to the head of the relevant projection (for morphological unity), and consequently allow object movement to the specifier of that projection, for case. Further, since the S-imperfective head is homogenous, non-quantity (in the sense of Borer 2005), it should not assign a bounded interpretation to the raised object. In other words, the interpretation of the object should be free, just as it is the case with B-imperfectives. To see whether this is in fact the case, consider the following examples from Polish: (68) a. Autobus ode-brał dzieci/młodzież ze szkoły. bus.NOM from-took.perf children.ACC/youth.ACC from school ‗A/the bus picked up the children/the youth from school.‘ b. Autobus od-bier-a-ł dzieci/młodzież ze szkoły. bus.NOM from-took.SI children.ACC/youth.ACC from school ‗A/the bus picked/was picking up (the) children/(the) youth from school.‘ (69) a. Magda przy-gotowała śniadanie/kanapki. Magda.NOM at-ready.perf breakfast.ACC/sandwiches.ACC ‗Magda made (lit. got ready) the breakfast/the sandwiches.‘ b. Magda przy-gotow-yw-ała śniadanie/kanapki. Magda.NOM at-ready.SI breakfast.ACC/sandwiches.ACC ‗Magda was making (lit. getting ready) (the) breakfast/(the) sandwiches.‘ As the examples indicate, the prediction is borne out – the direct object does indeed have free interpretation under the S-imperfective, i.e., it can have either a bounded reading or 158 an unbounded (i.e., mass, bare plural) reading. This result is further confirmed by the facts of genitive/accusative alternation. As I discussed briefly in the footnotes in the previous chapter, Polish (as well as other Slavic languages) allows a partitive genitive/accusative alternation in the direct object position in the context of the perfective form of selected verbs as a way to obtain an indefinite/unbounded interpretation for a mass or bare plural object. When such interpretation is available for accusative-marked objects, as is the case with B-imperfectives, the accusative/genitive alternation does not obtain. As it turns out, S-imperfectives disallow the genitive/accusative alternation, as illustrated in (70) and (71). This in turn indicates that the structural relationship between the direct object and the perfective form changes with the addition of the S-imperfective, which change could be understood as movement of the object to Spec, InnerAsp SI . (70) a. Bartek o-brał ziemniaki/ziemniaków. Bartek.NOM peeled.perf potatoes.ACC/potatoes.GEN ‗Bart peeled the potatoes/some potatoes.‘ b. Bartek o-bier-a-ł ziemniaki/*ziemniaków. Bartek.nom peeled.SI potatoes.ACC/potatoes.GEN ‗Bart peeled potatoes.‘ (71) a. Saraya za-pakowała zabawki/zabawek do worka. Saraya.NOM ZA-packed.perf toys.ACC/toys.GEN to bag ‗Saraya packed the toys/(some) toys into the bag.‘ b. Saraya za-pakow-yw-ała zabawki/*zabawek do worka. Saraya.NOM ZA-packed.perf toys.ACC/toys.GEN to bag ‗Saraya packed (the) toys into the bag.‘ 159 As shown above, the interpretation of arguments of S-imperfectives is identical to that of simplex imperfective, further confirming their aspectual equivalency. And as the bound/definite interpretation of the objects of perfectives and unrestricted interpretation of the objects of imperfectives is, by most accounts, related to (a)telicity and Aktionsart- encoding projections (e.g., Krifka 1992, Verkuyl 1999, Slabakova 2004, Borer 2005), we can conclude one more time that S-imperfectives, just like their morphologically and structurally more simple relatives, B-imperfectives, are atelic predicates. 8. Conclusions In this chapter I considered the problem of S-imperfectivization in two types of Slavic languages: those with a distinct system of viewpoint aspect marking in the past tense (i.e., the aorist/imperfect distinction), and those where prefixes and S-imperfective suffixes is all the aspectual morphology there is. Starting with the first type, I looked at the combinational possibilities with regard to the perfective/ imperfective distinction and the imperfect (= IMPERFECTIVE) and aorist (= PERFECTIVE) marking in Old Church Slavonic and Modern Bulgarian. As it turned out, in OCS, the process of S- imperfectivization was clearly independent of viewpoint aspect, as S-imperfective forms could be expressed in either the IMPERFECTIVE or PERFECTIVE viewpoint, and the IMPERFECTIVE morphology could be applied in the absence of the S-imperfective marking. In Bulgarian, the findings were similar, though the S-imperfective aorist form was not used frequently as it was homophonous with the present tense form. Bulgarian also helped to confirm the generalization proposed in the previous chapter whereby the IMPERFECTIVE viewpoint in Slavic is incapable of combining with telic predicates. Both in Bulgarian as well as in OCS, perfective (=telic) imperfect forms were not possible 160 except in modal contexts which are special cross-linguistically in that they allow tense and aspect morphology that is not interpreted in the usual way. Given the distribution of the S-imperfective forms with respect to the aorist/imperfect distinction in OCS and Bulgarian, I concluded that, contrary to what is universally assumed, the S-imperfective is not a marker of viewpoint, but rather of Aktionsart, and I proposed informally that it was an atelicizer. I then considered data from Polish vis à vis the tests for telicity and viewpoint aspect, and concluded that the aspectual interpretation of S-imperfective forms is identical to B-imperfectives, namely it is atelic and neutral with respect to viewpoint aspect (in the sense of being able to combine with either the IMPERFECTIVE or PERFECTIVE viewpoint). This confirmed the earlier working hypothesis that the S-imperfective marker belongs in the domain of Aktionsart and that it has an atelicizing role. I proposed that its exact role is that of homogenizing partitive operator, with the following semantics: [[SI]] = λeλPλe‘[P(e)  e‘  e  HOM(e‘)], where HOM stands for homogenous, as defined in (43). While proposed in the context of Slavic languages, this operator could have a more universal application. For example, in conjunction with the basic semantics of the progressive (i.e., the progressive as limited to relations between time intervals and not of the modal kind), it would give the right interpretation with respect to all kinds of telic predicates, especially those where culmination does not coincide in a simple way with the right or left temporal boundary. The novel analysis of the S-imperfective marker as an Aktionsart modifier allows us to not only account for the atelic readings of the S-imperfective forms, especially when their actual interpretation is PERFECTIVE, and so one could not attribute the atelic reading 161 to the application of the IMPERFECTIVE viewpoint, but also for the presence of the S- imperfective morphology in derived nominal forms that do not have any temporal properties, and whose derivational suffixes appear to select for a verb stem that denotes an activity. The new analysis also allows us to account for the Slavic speakers intuition that the aspectual pairing between B-imperfectives and their telic counterparts (e.g., pisać atelic ~ na-pisać telic ‗to write‘) is of the same kind as the pairing between a perfective verb where the prefix modifies the encyclopedic meaning of the verb and its S- imperfective form (e.g., prze-pisać telic ~ prze-pis-yw-ać atelic ‗to copy‘). Now the pairing can be viewed as involving just one grammatical property, i.e., (a)telicity, so the commonality felt between two kinds of pairings can be better understood. Lastly, with all prefixed verbs being grammatically marked for telicity, a huge portion of the verbal vocabulary, including those prefixed verbs where meaning is not compositional and/or whose equivalents in other languages could be used in telic as well as atelic contexts (e.g., prze-pisać ‗to copy, lit. through-write,‘ pod-robić ‗to counterfeit, lit. under-make,‘ po-móc ‗to help, lit. after/some can‘), are restricted to describing telic events only. On the analysis of the S-imperfective presented here, there is now a mechanism to allow for atelic used of such verbs, irrespective of the speaker‘s viewpoint. 162 CHAPTER 5: QUANTIFICATIONAL PREFIXES: UNIQUE CHARACTERISTICS While discussing the aspectual role(s) of Slavic verbal prefixation in Chapter 2, I concentrated on prefixes whose grammatical function (as opposed to their role in changing the verb‘s encyclopedic meaning) is limited to telicity-marking. In this chapter, I turn to the problem of three prefixes common to all Slavic languages that also have another systematic grammatical role, which could be termed as quantificational (Q). These are: na- with the meaning ‗a lot, a large unspecified quantity, enough‘ (the latter reading available in the presence of the reflexive clitic) (henceforth na Q -), po- with the meaning ‗some, a little,‘ (henceforth po Q -) and the distributive-marking po- (henceforth po D -). In the sections below, I will consider the unique properties of these prefixes on the example of Polish, while their aspect-marking properties, which will turn out to be almost identical to those of other, non-quantificational prefixes, will be discussed in Chapter 6. 1. Q-prefixes: unique properties In addition to the telicity marking prefixes, Slavic languages also have a more limited number of prefixes which differ from the rest in the following two ways: (a) they have systematic quantificational role or adverbial meanings, and (b) they can stack, i.e., in addition to bare imperfectives (B-imperfectives), they can attach also to already prefixed stems. The number of such prefixes differs from language to language. Bulgarian is reported to have as many as eight stackable prefixes 1 (repetitive pre-, 1 South Slavic languages seem to be most generous in the number of stackable prefixes they allow. We could speculate that in having a larger than usual inventory of stackable prefixes, Bulgarian and other South Slavic languages reflect an older stage in the development of verbal prefixation in Slavic, just as they do also with the preservation of the aorist-imperfect system. It is reported that the prominent 16 th century 163 excessive raz-, cumulative na-, completive iz-, inceptive za-, and three variants of po- (distributive, attenuative, and delimitative 2 ) (Istratkova 2006), while other languages, e.g., Polish, may have just two: the cumulative na- and po-, the latter with two distinct uses: po Q - and po D -. Na Q -, po Q - and po D - will be the only stackable prefixes considered here, as they are the only ones that can stack in all Slavic languages, and because their quantificational-like properties make them particularly intriguing. 2. Quantificational-like function Each of the three considered prefixes, na Q -, po Q -, and po D -, are quantificational in nature and have stable, transparent meanings. They differ in this regard from non- stacking prefixes, which can, and often do give rise to synchronically opaque meanings. Semantically, na Q - conveys the meaning of ‗a lot / an unspecified large amount / enough / satisfactorily,‘ while po Q - means ‗some / a little bit / to a certain extent / for a short while,‘ and po D - marks distributive readings, which involve universal quantification. Some illustrative examples are given in (1)-(5) below, with the examples in (b) showing minimal pair readings when the relevant Q-prefix is not present. Polish New Testament translator, Stanisław Murzynowski, used such doubly prefixed forms as z-u-ciekać ‗to run away‘ and prze-na-śladować ‗to follow, copy,‘ where the first/outer prefix was of the prepositional (i.e., spatial, relational) rather than adverbial or quantificational kind (e.g., z in z-u-ciekać appears to have had the meaning ‗from, away‘). These forms are ungrammatical in contemporary Polish, and were also criticized already by his contemporaries. We could infer, however, that they were once possible, just as preverb stacking was possible in, for example, Sanskrit, where ―combinations of two [preverbs] are quite usual; of three, much less common; of more than three, rare‖ (Whitney 1080; quoted in Papke 2005). Over time, in South the stacking ability has been limited to prefixes with adverbial and quantificational meanings, and the West and East Slavic languages went even further and now allow only a handful of such prefixes. In Polish these happen to be all quantificational ones. 2 In this work, I will consider the attenuative and delimitative uses to be not two different prefixes, but rather one prefix applied to different kind of scales: intensity for the attenuative reading, and time for the delimitative reading. Other scales are also possible as po-‗s quantifiees, e.g., the scale of gradual change with degree achievements, as will be discussed later. 164 (1) a. Na-z-bierałem grzybów. na Q -picked.SI.1SG mushroom ‗I‘ve picked a good quantity of mushrooms.‘ b. Z-bierałem grzyby. picked.SI.1SG mushroom ‗I picked/was picking mushrooms.‘ (2) a. Na-płakałam się. na Q -cried.1SG self ‗I cried a lot/enough.‘ b. Płakałam. cried.1SG ‗I cried/was crying.‘ (3) a. Najpierw po-mieszkam u rodziców. first po Q -live.1SG at parents ‗I will live with my parents for a little while first.‘ … before I rent an apartment.‖ b. Mieszkam u rodziców live. 1SG at parents ―I live with my parents‘.‖ (4) a. Muszę o tym po-myśleć. must about that po Q -think.INF ‗I have to think about this some/a little.‘ 165 b. Muszę o tym myśleć. must about that think.INF ‗I have to think/be thinking about this.‘ (5) a. Dzieci po-przynosiły kwiatki nauczycielom. children.NOM po D -brought.3PL flowers.ACC teachers.DAT ‗The children each brought flowers for teachers.‘ or ‗The children brought flowers for each teacher.‘ or ‗The children brought each of the flowers for the teachers.‘ b. Dzieci przynosiły kwiatki nauczycielom. children.NOM brought.SI.3PL flowers.ACC teachers.DAT ‗The children brought/were bringing flowers for teachers.‘ 2.1. Quantificational role of na Q - 2.1.1. na Q - and the interpretation of the direct object A notable property of the prefix na Q - that distinguishes it from other prefixes, including other quantificational-like ones, is that it typically affects the quantity interpretation of the internal argument (i.e., the direct object or the single argument of an unaccusative verbs), and that normally the presence of such argument is required. This characteristic of na Q - has been prominently featured in Filip‘s work on Czech (e.g., Filip 1996, 2000, 2005), and has been also discussed at length in Pereltsveig (2006) and Romanova (2006) with respect to Russian, though Romanova (2006) does not in the end recognize the special relationship between the internal argument and na Q -, to which I will return shortly. The exceptions to this rule are constructions with the reflexive clitic się, to 166 be discussed in the next section, as well as some verbs which are able to combine with na Q - without an object, in contrast to most other predicates 3 , e g., (6) Kto tu tak na-brudził / na-bałaganił / na-mieszał ? who here like-this na Q -dirtied.3SG / na Q -made-mess.3SG / na Q -mixed-up.3SG ‗Who dirtied up the place this much / made such a mess / created this much confusion here?‘ (7) Na-paliłem ci w piecu. na Q -burned.1SG you.DAT in furnace ‗I got the furnace going for you.‖ (lit. ―I burned it in the furnace a lot/enough.‘) (8) Pan Blankfein na-kłamał podczas przesłuchania. Mr. Blankfein na Q -lied.3SG during hearing ‗Mr. Blankfeit did a lot of lying during the hearing.‘ Cf. (9) Ona *na-czytała/*na-pisała/*na-prosiła/*na-myślała/*na-biegała/*na-jadła/etc she na Q -read.3SG / na Q -wrote.3SG / na Q -asked.3SG / na Q -thought.3SG / na Q -ran.3SG / na Q -ate.3SG / etc. The examples in (6)-(8) are exceptional and I set them aside here. I will only note that they all have a clearly resultative reading (the same goes for all the Russian examples in fn. 3), and could be analysed as involving a null argument. In all other situations, the presence of an overt internal argument is required and na Q - restricts the interpretation of 3 For Russian, Romanova (2006:188) lists the following na Q -prefixed forms without an object: (i) na-dyshat‘ ‗to breath a lot, so that it becomes warm in some closed space‖ (ii) na-toptaj‘ ‗to trample a lot, so that the floor becomes dirty.‘ (iii) na-kurit‘ ‗to smoke a lot, so that the air becomes foggy‘ (iv) na-bolet‘ ‗to ache (psychologically) a lot, so that the emotion has to be let out‘ The equivalents of examples in (i)-(iii) would also be possible in Polish. 167 that argument. The direct relationship between na Q - and the internal argument is evident not only in the obtained interpretations, but also in the fact that, in Polish 4 , the internal argument of na Q -prefixed verbs must be assigned the genitive case, which otherewise characterizes quantificational structures, where most quantifiers as well as numerals other than 1-4, assign genitive case to the NPs they modify. The examples below show genitive case assignment due to the presence of a numeral/quantifier and the prefix na Q -, and contrast that with non-quantifcational contexts where the accusative case is assigned. I also indicate the possible and impossible readings for predicates with na Q -. (10) a. Kupiłam dużo / pięć *flamastry/flamastrów. Ø-bought.1SG many / five markers.ACC/markers.GEN ‗I bought many/five markers.‘ b. Na-kupiłam *flamastry/flamastrów. na Q -bought.1SG markers.ACC/markers.GEN ‗I bought a lot of markers.‘ # ‗I did a lot of marker buying.‘ Cf. c. Kupiłam flamastry. Ø-bought.1SG markers.ACC ‗I bought markers.‘ 4 Based on the data included in Filip (2000) and Filip (2005), we can infer that in Russian either genitive or accusative is possible, while in Czech arguments quantified by na Q - are expressed in the accusative. It remains a topic for an empirical study to see if the difference in case assignment correlates with differences in syntactic or semantic properties. 168 (11) a. S-padło dużo / sporo śniegu/*śnieg. PREF-fell.3SG much / quite a bit snow.ACC/snow.GEN ‗A lot/Quite a bit of snow fell down.‘ b. Na-padało śniegu/*śnieg. na Q -fell.SI.3SG snow.ACC/snow.GEN ‗A lot of snow fell down.‘ # ‗It snowed a lot (e.g., for a long time)‘ …but little snow fell Cf. c. S-padł 5 *śniegu/śnieg. PREF-fell.3SG snow.ACC/snow.GEN ‗Snow fell down / It snowed.‘ (12) a. U-gotowałam dużo / trochę *zupę / zupy. PREF-cooked.1SG much / a little soup.ACC / soup.GEN ‗I cooked a lot of / a little of soup.‘ a. Na-gotowałam *zupę / zupy. na Q -cooked.1SG soup.ACC / soup.GEN ‗I cooked a lot of soup.‘ # ‗I did a lot of cooking of soup.‘ Cf. c. Gotowałam zupę / *zupy. cooked.1SG soup.ACC / soup.GEN ‗I cooked/was cooking (the) soup.‘ 5 Śnieg has a masculine gender, but in the cases when it is quantified – either by an over quantifier or na Q -, the verb agrees with the gender of the quantifier - in the above examples, that gender is neuter. 169 (13) a. Student roz-wiesił dużo / pełno *ogłoszenia / ogłoszeń. student out/around-hang.3SG many / a lot notices.ACC / notices.GEN ―A/the student posted a lot of/a number of notices in the hallway.‖ #―A/the student did a lot of posting of notices in the hallway.‖ b. Student na-roz-wieszał *ogłoszenia / ogłoszeń. student na Q -out/around-hang.SI.3SG notices.ACC / notices.GEN ‗A/the student posted a lot of/a number of notices in the hallway.‘ #‘A/the student did a lot of posting of notices in the hallway.‘ Cf. c. Student roz-wiesił ogłoszenia / *ogłoszeń. student out/around-hang.3SG notices.ACC / notices.GEN ‗A/the student posted notices in the hallway.‘ The obligatory genitive case assignment effected by na Q - as well as the allowed interpretations indicate that despite its more adverbial-like syntactic position, na Q - quantifies clearly over the internal argument rather than over the event. Similar observation was made in Filip (1996) with respect to Czech, and in Pereltsveig (2006) with regard to Russia, and appears to be a pan-Slavic property. The conclusion that the quantification is obligatorily with respect to internal argument is further supported by the fact that internal arguments of na Q -prefixed verbs must be quantifiable, and so either mass or plural. Singular arguments are ungrammatical with na Q -, as shown in (14) and (15). 170 (14) Na-łuskałam *orzecha / orzechów. na Q -shelled.1SG nut.GEN / nuts.GEN ‗I have shelled a good quantity of nuts.‘ (15) Na-za-praszałam *ważnego gościa / ważnych gości. na Q -invited.SI.1SG important guest.GEN / important guests.GEN ‗I invited a lot of important guests.‘ (16) Na-padało *liścia / liści. na Q -fell.SI.3SG leaf.GEN / leaves.GEN ‗A lot of leaves fell.‘ Compare this with non-quantificational uses of na-: (17) Na-pisałam książkę. PREF-wrote.1SG book.ACC ‗I wrote a book.‘ (18) Na-łożyłam ci obiad na talerz. on-put.1SG you.DAT dinner.ACC on plate ‗I put dinner on the plate for you.‘ It is interesting to note that while alternative readings where na Q - quantifies over the event and not the internal argument would not be cognitively impossible (e.g., ‗I did a lot of shelling of one particular nut,‘ and ‗I did a lot of inviting of one particular important guest/I invited one particular guest many times‘), and one might have expected them to be forced in the presence of a non-quantifiable object, they are nevertheless not available in the sentences above (cf. Filip 1996). 171 The fact of the dependency between the quantifier na Q - and the internal argument might lead one to think of it as a nominal quantifier, merged in the extended nominal projection of the relevant argument, and then moved to a higher, pre-verbal position. This would be similar to a movement account of Quantification-at-a-Distance phenomenon in French (Pollock 1989, Cyr 1991), where the nominal quantifier can appear in two positions, as shown in (19). 6 (19) a. J'ai mangé beaucoup de pommes I-have eaten a-lot of apples ‗I have eaten a lot of apples.‘ b. J‘ai beaucoup mangé de pommes I-have a-lot eaten of apples ‗I have eaten a lot of apples.‘ However, there are facts that clearly argue against such an analysis. The first one is the aspectual role of na Q -, to be discussed in the next chapter, which does not otherwise characterize nominal quantifiers in Slavic. A second and related fact is the incompatibility of na Q -with perfective stems – it must attach either to B-imperfectives or S-imperfective verbs. This latter selection, in particular, makes it unlikely that na Q - is merged as part of the internal argument (i.e., in Spec, VP or Spec, InnerAsp [telic] ), which would be below the projection hosting the secondary imperfective operator. Further, na Q - 6 Doetjes (1995) argues the Quantification at a Distance is misnomer, as in sentences like (19)b, the quantifier has scope not over the object, but over the event variable (so the actual reading is: there were many events of me eating an apple). Consequently, beaucoup in (19)b should be seen as base-generated in the preverbal position, and quantification is local. Quantification at a Distance has, however, been argued to be real in the Quebeck dialect of French (Bouchard & Burnett 2007). There also comparable phenomenon in Japanese, where numeral quantifiers can be found away from the NP, but these have ben analyzed in terms of movement when the numeral clearly quantifies over the NP (rather than the event) (Nakanishi 2006). Neither the French nor the Japanese case is an actual match to what we find with na Q - in Slavic, as na Q - is not floating, and it does not show scope ambiguities. 172 cannot function as an independent nominal quantifier, so a scenario like the one in (19)a cannot arise. Lastly, there is the crucial fact of na Q - being able to co-occur with overt nominal quantifiers, measure phrases, and sometimes also numerals 7 , as shown in (20). (20) Na-kleiłam dużo / mnóstwo / dość / pełno / masę / kupę / 100 pierogów. na Q -glued.1SG many / lots / enough / loads / mass / pile / 100 dumplings.GEN ―I have made many/lots of/enough/loads of/mass of dumplings.‖ The obtained reading in such cases is not one where na Q - measures over a part of an already quantified object, i.e., a lot OF the many/lots/loads/mass/pile/etc. dumplings, but rather one where the contribution of na Q - and the nominal quantifier or the measure phrase overlap, with na Q - appearing to function more as an intensifier. This can be further seen in the fact that the quantifiers and measure expressions co-occuring with na Q - must be semantically compatible with it, i.e., also indicating a large quantity, or else ungrammaticality results, as shown in (21). (21) Na-kleiłam (*mało / *kilka / *talerzyk / *kupkę / *10) pierogów. na Q -glued.1SG *little / *several / *a small plate / *a little pile / *10 dumplings.GEN ―I have made a lot of dumplings.‖ Secondly, the measure expressions co-occuring with na Q - that are inflectable for case are invariably expressed in the accusative and not the genitive that would be expected if na Q - were to quantify into a quantity expression. This is illustrated in (22). (22) Na-kleiłam masę/*masy / tonę/*tony / miskę/*miski pierogów. na Q -glued.1SG mass.ACC/GEN / ton.ACC/GEN / bowl.ACC/GEN dumplings.GEN ―I have made a mass of/ a ton of / a bowl of dumplings.‖ 7 Numerals seem to be possible with some verbs but ruled out with others. Na Q -prefixed kleić ‗to glue‘ above is fine with a numeral, but zaprosić ‗to invite‘ or kupić ‗to buy‘ sound ungrammatical or degraded. I do not have any suggestions as to why that might be. 173 On the basis of these facts, which have been reported to also hold in Russian (Romanova 2006 8 ), we can conclude that while na Q - appears to quantify over the internal argument, it is not in itself a nominal quantifier. Instead, we can characterize it as an intensifier that comes with the presupposition that the quantity of the internal argument is large or sufficient according to some contextually provided scale. 9 This conclusion is further supported by facts concerning the interaction, or rather lack thereof, of na Q - with negation. In Polish, just as in English, preverbally marked negation can be interpreted narrowly with respect to any constituent in its scope, including nominal quantifiers and measure expressions, e.g., (23) Nie z-robiłam wielu / miski pierogów. NEG PREF-made.1SG many / bowl.GEN 10 dumplings.GEN ‗I did not make many / a bowl of dumplings.‘ A. Widest scope reading: ‗There was no event of me making many/a bowl of dumplings.‘ B. Narrow scope reading with respect to the quantifier/measure phrase: ‗I made dumplings, but not many / a bowl of.‘ However, it is impossible to obtain narrow scope as in (23)-B with na Q -. This is illustrated in (24) and (25) below. 8 They do not, however, lead Romanova (2006) to the same conclusions as the ones presented here. Instead, she analyzes na Q - as a variable quantifier. I will return to her analysis in the discussion below. 9 I am grateful to Elena Guerzoni, Roumi Pancheva and Hagit Borer for extremely useful discussion of pertinent data and suggesting this possibility to me. 10 Genitive marking here is due to Genitive-of-Quantification phenomenon, where in Polish internal arguments invariably receive genitive case under sentential negation.The phenomenon in pan-Slavic, but is subject to some variability wth regard to the obligatory nature of genitive marking. 174 (24) Nie na-kleiłam pierogów. NEG na Q -glued.1SG dumplings.GEN ‗I did not make a lot of dumplings.‘ Widest scope reading only: ‗There was no event of me making a lot of dumplings. (25) Nie na-gotowałam zupy. NEG na Q -cooked.1SG soup.GEN ‗I did not cook a lot of soup.‘ Widest scope reading only: ‗There was no event of me cooking a lot of soup.‘ Once a nominal quantifier or measure phrase is added, narrow scope becomes possible: (26) Nie na-kleiłam wielu / miski pierogów. NEG na Q -glued.1SG many / bowl.GEN dumplings.GEN ‗I did not make many / a bowl of dumplings.‘ A. Widest scope reading: ‗There was no event of me making many / a bowl of dumplings.‘ B. Narrow scope reading with respect to the quantifier/measure phrase: ‗I made dumplings, but not many / a bowl of.‘ (27) Nie na-gotowałam dużo / garnka zupy. NEG na Q -cooked.1SG much / pot.GEN soup.GEN ‗I did not cook much / a pot of soup.‘ A. Widest scope reading: ‗There was no event of me cooking much / a pot of soup.‘ B. Narrow scope reading with respect to the quantifier/measure phrase: ‗I cooked soup, but not much / a pot of.‘ 175 The facts discussed above indicate strongly that na Q - is not a nominal quantifier. But if it is not, how do we account for the genitive marking on the internal argument in the context of na Q -? It seems two approaches are possible here. One is to postulate a null quantifier local to the internal argument, whose meaning (i.e., some large quantity) is derived from the presupposition supplied by na Q -. The local relationship between the null quantifier and the argument would account for the genitive case assignment in a straightforward way as an instance of Genitive of Quantification. A proposal incorporating the null quantifier idea has been previously presented in Pereltsveig (2006), who assumes that the prefix na Q - selects for a QP (Quantity Phrase) as its argument, which can be headed by an overt or a null quantifier. The relevant structure is shown in 0. (28) Figure 13: Pereltsveig's 2006 analysis of the cumulative na- na-P   QP na-P              Q NP na- …               … VP      V 0 On this analysis, na Q - itself is not a quantifier, but simply requires a quantified object. As we have seen earlier, however, na Q - can co-occur not only with quantifiers, but also with measure phrases with a nominal head. Therefore it would be better to state the generalization as na Q - requiring a quantity object, where quantity is obtained either via quantification (and so QP) or measuring (and so Measure Phrase). The other possibility would be to assume that there is no null quantifier, and that the large quantity interpretation comes solely from the presupposition introduced by na Q -. 176 On this approach, the observed genitive case marking on the internal argument of na Q - prefixed verbs would have to be attributed to something else. That something else could be, following the proposal concerning small nominals in Pereltsveig (2005), the syntactic status of the relevant arguments as NPs rather than DPs, with the grammar assigning genitive case to the former, and accusative to the latter. Pereltsveig (2005) has shown that genitive-marked nominals in Slavic (and more narrowly Russian) without an overt genitive case assigner (e.g., a quantifier) have properties that otherwise characterize bare plurals and bare mass nouns in English, or incorporated nominals cross-linguistically (see, e.g., Baker 1988, van Geenhoven 1998, or Dayal 1999), such as lack of specific/definite interpretation, lack of scopal force (e.g., with respect to negation), inability to control PRO and be an antecedent in binding. Pereltsveig proposes that such nominals are NPs rather than DPs, hence the term ‗small nominals,‘ and as such are property-denoting and lack a referential index. Their different status can be illustrated on the example of select perfective verbs which allow accusative/genitive alternation with their objects, as in (29). The following example, (30), shows that a genitive-marked NP does not have referential interpretation, as it cannot be an antecedent to a pronoun. (29) a. U-gotowałam zupy. PREF-cooked.1SG soup.GEN ‗I cooked soup.‘ cf. b. U-gotowałam zupę. PREF-cooked.1SG soup.ACC ‗I cooked the soup.‘ 177 (30) a. U-gotowałam zupy i (*i przelałam ją i do garnka). PREF-cooked.1SG soup.GEN (and through-poured.1SG her.ACC to pot) ‗I completed soup-cooking (and transferred it into the pot).‘ Cf. b. U-gotowałam zupę i (i przelałam ją i do garnka). PREF-cooked.3SG soup.ACC (and through-poured.1SG her.ACC to pot) ‗I cooked the soup and (and transferred it into the pot).‘ We could speculate that the genitive-marked arguments of na Q -prefixed verbs are also NPs. 11 They need to be NPs, because NPs but not DPs are amenable to quantification (cf. *many the boys, *some the cheese). The additional assumption would be required here, namely that quantification can be either direct – through a quantifier, or indirect – through the presupposition, as proposed for na Q - here. With that assumption, the necessity for a direct argument with reduced functional structure follows naturally. To see if this hypothesis is on the right track, we can test the binding possibilities of genitive-marked objects of na Q -prefixed verbs. This is done in (31)-(33) below. (31) Na-gotowałam zupy i (i przelałam ją i do garnka). na Q -cooked.1SG soup.GEN (and through-poured.1SG her.ACC to pot) ‗I cooked a lot of soup (and transferred it into the pot).‘ (32) Na-zbierałam grzybów i (i je i wysuszyłam). na Q -gathered.1SG mushrooms.GEN (and them.ACC dried.1SG) ‗I picked a lot of mushroom (and dried them).‘ 11 A proposal similar in spirit has been previously advanced in Romanova (2006), who analyzes genitive marked NPs as bare plurals. The objects of na Q -prefixed verbs do not actually need to be plural – they can also be mass – and there is no overt indefiniteness/definiteness marking in most Slavic languages that would contrast, e.g., books with the books, but the intent of the analysis is clear, namely to derive the properties of genitive-marked phrases outlined in Pereltsvaig (2006) drawing on the comparison with the properties of bare plurals in English. 178 (33) On na-gadał kłamstw i (a potem za nie i przepraszał). he na Q -spoke.3SG lies.GEN (and then for them.ACC apologized.3SG) ‗He told a lot of lies (and then was apologizing for them).‘ As the examples above illustrate, the genitive-marked objects of na Q -prefixed verbs can be antecedents in binding, and therefore cannot lack the referentiality index. Consequently, they could not be NPs, and so the observed genitive marking could not be attributed to impoverished nominal functional structure. This leads us back to the first considered analysis, namely the null quantifier account, which at this point appears to be most adequate. It would have to be assumed, however, that the null quantifier is not visible to negation, or else we have no way of accounting for the lack of interaction of negation with the internal argument in the presence of na Q - and in the absence of an overt quantifier. The fact that the internal argument of a na Q -prefixed verb must be quantified or measured – if not by anything else than by a null quantifier – underscores once more the special relationship of na Q - and the internal argument. It also argues against the analysis advanced in Romanova (2006), who proposes that na Q - actually measures over the event, but with flexibility as to the scale it specifically measures over in the verbal context. She attributes to na Q - the Quantificational Variability Effect (Nakanishi & Romero 2004), where a verbal quantifier can be non-selective and measure over different entities or scales which are part of the event, and where predicate type will decide on the possible measurable scales (as recounted by Romanova, Schwartzchild (2006) identifies five such scales in verbal contexts: degree (a point on a scale), ranges (a set of degrees), amount of 179 events, amount of stuff, and duration (where the last three are kinds of range)). She states the role of na Q - as follows: (34) “Na-, being a range predicate in the sense of Schwartzchild (2006), measures an amount of event mapped onto quantifiable entities such as time, space or stuff. The measured range includes all the degrees between the two degrees specified by na-, and overt adverbials and measure phrases it combined with …” (Romanova 2006:208) Romanova then illustrates the uses of na Q - as a quantifier over such different scales in Russian, including temporal, spatio-temporal and participant (i.e., internal argument) scale. The first two alleged uses of na Q - are shown in (35) and (36) (exs. adopted from Romanova 2006:205). (35) Na-rabotal sto chasov na Q -worked.3SG 100 hours.GEN ‗He accumulated 100 working hours.‘ (36) Na-jezdil tysjashu kilometrov na Q -drove.3SG 1000 kilometers.GEN ‗He drove cumulatively 1000 km.‘ Note, however, that these sentences still essentially involve quantification over the object, which is evidenced in the case marking as well as in the interpretation. Consequently, they do not show variable quantification possibilities for na Q -. True variable quantification would obtain if for a sentence like Nancy drove to Los Angeles, the addition of na Q - could be interpreted as indicating that she drove for a long time or that 180 she drove a long distance. This is not the case in Slavic. Rather, an explicit argument referring to a distance or time has to be used to obtain the relevant reading. Another piece of evidence proving that na Q - does not have free interpretation are examples with degree achievement predicates. Degree achievements are predicates with a built-in degree of change scale, and should therefore facilitate an alternative interpretation for na Q - that is not object oriented, namely one with respect to the degree scale. This however, does not happen. Instead, na Q - is completely incompatible with degree achievements, as shown in (37) and (38). (37) Mleko *na-drożało / drożało. milk.NOM na Q -got-more-expensive / got-more-expensive ‗Milk was getting expensive (*a lot).‘ (38) Tato *na-siwiał / siwiał. dad.NOM na Q -got-more-gray / got-more-gray ‗Dada was getting gray (*a lot).‘ Lastly, there is the fact that in order to measure over the event and not the object, na Q - requires the addition of the reflexive clitic się (these constructions are discussed in detail in the section below). All these fact point clearly in the direction of a dependency between na Q - and the internal argument, and support the characterization of na Q - as an object-oriented quantity intensifier. 2.1.2. na Q - as a quantifier over the event In addition to its role with respect to the internal argument, the prefix na Q - can be also used productively as a quantifier over the event – with accomplishments as a quantifier over scales of duration or intensity, and with achievements as a quantifier over 181 plurality of events. This reading of na Q -, however, requires the presence of the reflexive accusative clitic się, which otherwise functions as a subject-oriented reflexive with a bound variable reading (Lubowicz 1999). 12 The relevant interpretation obtains both when the verb is inherently reflexive and the presence of the clitic is obligatory, when the verb is syntactically reflexivized independently of na Q -, as well as when the clitic is added specifically to obtain the relevant reading for na Q - (i.e., event level quantification) and reflexivization is otherwise not possible. This is illustrated in the examples in (39)-(40), (41)-(42), and (43)-(44), respectively. In the relevant constructions, na Q - can be interpreted as meaning ‗a lot‘ and/or ‗enough,‘ depending on the context. 13 (39) a. pocić *(się) ‗to sweat‘ b. Na-pociłam się przy tym. na Q -sweated.1SG REFL.ACC at this ‗It took a lot of effort to do this.‘ (lit. ‗I sweated quite a lot at this.‘) (40) a. śmiać *(się) ‗to laugh‘ b. Na-śmiałam się dziś. na Q -laughed.1SG REFL.ACC today ‗I laughed today a lot/enough.‘ 12 As a clitic, się does not have any particular attachment properties other than being an enclitic to some other element in the sentence to the left of its base-generated position (cf. Franks 1998). Only colloquial/dialectal uses allow sentence-initial position of this clitic in impersonal contructions with the meaning ‗one does/did/will V,‘ e.g., Się zrobi ‗It will be/get done‘ or Się przyjdzie ‗One will come.‘ 13 The context may be situational or sentential in the sense of presence vs. absence of additional modifiers. For example, the exclamatory particle ale promotes the ‗a lot‘ reading, while the adverbial już ‗already‘ advances the ‗enough‘ interpretation. 182 (41) a. męczyć (się) ‗to tire, to toll, bother‘ b. Na-męczyłam się przy tym. na Q -tired.1SG REFL.ACC at this ‗It took a lot of effort to do this.‘ (lit. ‗I got tired a lot doing this.‘ (42) a. prze-wrócić (się) ‗to fall‘ b. Na-prze-wracałam się dziś jeżdżąc na łyżwach. na Q -fell.SI.1SG REFL.ACC today riding on skates ‗I fell many times today while skating.‘ (43) a. biegać (*się) ‗to run‘ b. Na-biegałam się za tą książką. na Q -ran.1SG REFL.ACC after this book ‗I ran around a lot to get this book.‘ (44) a. czytać (*się) ‗to read‘ b. Na-czytałam się o tym w internecie. . na Q -read.1SG REFL.ACC about this in internet ‗I read a lot/enough about this on the internet.‘ The reflexive accusative clitic się, crucial to obtaining the event-level reading for na Q -, has been analyzed in Lubowicz (1999) with respect to Polish as a lexical de- transitivizing operator, attaching directly to V o through head adjunction. She argues that się could not be an actual argument in the direct object position, because it does not have any of the typical DP properties such as the ability to be declined for cases, or the ability to be a complement of a preposition. Also, unlike with its DP-like, prosodically independent counterpart siebie ‗self.ACC,‘ quantifiers like only cannot take scope over 183 się, and instead treat the verb and się as a unit. While the non-argument analysis must certainly be taken as valid, the specific ‗lexical detransitivizing operator‘ function clearly does not motivate the addition of się in (43)b and (44)b, where the involved verbs do not otherwise function as transitive predicates, as indicated by the impossibility of reflexivizing them outside of the context supplied by na Q -, illustrated in (43)a and (44)a. Further, an overt object can co-occur with the reflexivized uses of na Q -, as shown in (45) and (46), which is not possible for other reflexivized predicates, as expected. (45) Na-gotowałam się ziemniaków / *ziemniaki. na Q -cooked.1SG REFL.ACC potatoes.GEN / potatoes.ACC ‗I did a lot/enough of cooking potatoes.‘ (could be just a few potatoes) #‘I cooked a lot of/a good number of potatoes.‘ (46) Na-rąbałem się drewna / *drewno. na Q -chopped.1SG REFL.ACC wood.ACC / wood.GEN ‗I did a lot/enough of chopping wood.‘ (could be just one plank of wood that was particularly hard to chop) #‘I chopped a lot/a good amount of wood.‘ In such cases, even though the object is in the genitive, quantification is still necessarily with respect to the event. The interpretation of the object is in fact undefined quantity- wise. The fact that na Q - does not affect the quantity interpretation of the object in the presence of się is further confirmed by the fact that the object co-occurring with the reflexive can be singular, as shown in (47) and (48). 184 (47) Na-kopałem *(się) dziś piłki / *piłkę. na Q -kicked.1SG REFL.ACC today ball.GEN / ball.ACC ‗I played soccer a lot/enough today.‘ (48) Na-pchałem *(się) wczoraj tego auta / *to auto na Q -pushed.1SG REFL.ACC yesterday this car.GEN / this car.ACC ‗I pushed this car a lot/enough yesterday.‘ The examples above indicate that the reflexive clitic as used with na Q -, is not the same as the detransitivizing się found elsewhere. And yet, it is detransitivizing in the sense of (a) operating on a transitive structure, which can be taken to be required with na Q - because of its object-oriented character, and (b) making the internal argument invisible for na Q -, thereby resulting in the event-level quantification. For inherent and structural reflexives, we can assume, following Lubowicz (1995) that the reflexive is head-adjoined to V low in the verbal functional structure. With się dependent on the presence of for na Q - or appearing alongside an overt object, I propose that it head-adjoins to na Q -, or rather the na Q -V complex head following verb movement for morphological composition, within the projection hosting the prefix (to be determined in the next chapter). This is illustrated in (49). (49) Figure 14: Reflexivization in the context of na Q -                …              się    na Q - (NP) V It also needs to be noted that się added in the context of na Q - also remains a subject- oriented reflexive in the sense that it introduces an additional scale along which event- 185 level quantification can be interpreted, namely the scale of satisfaction determined with respect to the subject. This scale manifests itself as the ‗enough‘ reading. In the discussion of na Q - as a quantifier over the object, I concluded that the prefix is not a quantifier over individuals, but rather an intensifier requiring a quantified or measured internal argument, whose ‗large quantity‘ it then emphasizes. The question we need to answer with respect to the reflexivized uses of na Q - is whether na Q - in such cases functions as a quantifier, or also an intensifier of some kind. The first fact to consider in this regard is whether na Q - can co-occur with adverbials of comparable meanings. This is done in (50)-(55). (50) Na-pociłam się przy tym (bardzo / ??dużo / sporo / dość). na Q -sweated.1SG REFL.ACC at this (very (much)/??a lot / quite a bit / enough) ‗I sweated very much / ??a lot / quite a bit / enough doing this.‘ (51) Na-śmiałam się dziś (bardzo / ???dużo / sporo / dość). na Q -laughed.1SG REFL.ACC today (very (much)/ ???a lot / quite a bit / enough) ‗I laughed very much / ???a lot / quite a bit / enough today.‘ (52) Na- męczyłam się przy tym (bardzo / ???dużo / sporo / dość). na Q -tired.1SG REFL.ACC at this (very (much) / ???a lot / quite a bit / enough) ‗I put in very much / ???a lot / quite a bit / enough effort to do this.‘ (53) Na-prze-wracałam się dziś (*bardzo/??dużo / sporo / dość) jeżdżąc na łyżwach. na Q -fell.SI.1SG REFL.ACC today (*very (much)/??a lot / quite a bit / enough ) riding on skates ‗I fell *very much / many times / quite a number of times / enough times today while skating.‘ 186 (54) Na-biegałam się (*bardzo / dużo / sporo / dość) za tą książką. na Q -ran.1SG REFL.ACC (*very (much)/a lot / quite a bit / enough) after this book ‗I ran around *very much / a lot / quite a bit / enough to get this book.‘ (55) Na-czytałam się o tym (*bardzo / dużo / sporo / dość) w internecie. na Q -read.1SG REFL.ACC about this (*very (much) / a lot / quite a bit / enough) in internet ‗I read *very much / a lot / quite a bit / enough about this on the internet.‘ With the exception of the example in (55), the ungrammaticality or the degraded status of a particular adverbial has to be attributed to the presence of na Q -, as without na Q - the same adverbial is unproblematic. This in turn suggests that what is disallowed is the adverbial duplicating the semantic contribution of na Q - in a particular context, where that contribution depends on the scale na Q - is applied to: for the intensity scale it would be bardzo ‗very (much),‘ and the for duration scale – dużo ‗a lot.‘ This hypothesis is further confirmed by the fact that when only the satisfaction scale is accessible, adding the adverbial dość ‗enough‘ is ruled out, as illustrated below. (56) Na-jadłem się (*dość). na Q -ate.1SG REFL.ACC (*enough) ‗I ate enough/had enough to eat.‘ (57) Zuzia na-cieszyła się (*dość) lalką. Zuzia na Q -enjoyed.3SG REFL (*enough) doll.INSTR ‗Susie had her fill of being excited over/playing with the doll.‘ The data above indicates that in the reflexivized uses, na Q - acts as an actual quantifier, rather than an intensifier. If this conclusion is correct, we predict that the 187 reflexivized na Q - should behave differently with respect to negation than the na Q - acting as an intensifier. More specifically, negated na Q -prefixed forms should allow the same interpretation as negation with an overt quantificational adverbial, as illustrated in (58) and (59). (58) Nie z-męczyłam się przy tym bardzo / dużo. NEG PREF-tired.1SG REFL.ACC at this very (much) / much ‗I did not get tired a lot doing this.‘ = ‗I got tired, but not a lot.‘ (59) Nie chodziłam dość po sklepach. NEG went-around enough over shops ‗I did not go around shopping enough.‘ = ‗I went around shopping, but not enough.‘ This is, indeed, the case, as exemplified in (60) and (61). (60) Nie na-męczyłam się przy tym. NEG na Q -tired.1SG REFL.ACC at this. ‗I did not take a lot of effort to do this.‘ = ‗It took effort to do this, but not a lot of effort.‘ (61) Nie na-chodziłam się po sklepach. NEG na Q -went-around REFL.ACC over shops ‗I did not go around shopping a lot / enough.‘ = ‗I went around shopping, but not a lot / enough.‘ These examples confirm the conclusion that when na Q - applies to the event, it functions as a quantifier, even though its role with respect to the internal argument is that of an intensifier. The difference between the two applications of na Q - could be attributed to the 188 syntactic placement of na Q - that is more adverbial like, and not local to the internal argument. Consequently, na Q - can easily function as an event quantifier, but less so as a nominal one. Before we close the discussion of the prefix na Q -, one last issue needs to be considered, namely is the realization of the direct object of na Q -prefixed verbs when it co- occurs with the reflexive. As one could see in the previously given examples, such objects must be assigned the genitive case even though they are not quantified over. We could attribute this fact to the inability of the direct object in the relevant cases to move to a position where it could be assigned the accusative case, which I take to correspond to be the specifier position of the highest inner aspect marking projection. I will argue in the next chapter that na Q -, beside its quantificational function, is also a marker of Aktionsart, and so the journey of the object for accusative case has to go through the specifier of the projection headed by na Q -. The reflexive clitic się, which I assume to be head-adjoined to V at some point in the derivation no later than the na Q -hosting projection, can be uncontroversially taken to block accusative case assignment. The object in such situation is assigned the genitive case instead, which we can assume to be the default case. The formal identity between the genitive direct objects when co-occuring with the reflexive, and the genitive internal arguments when there is no reflexive, is therefore only apparent. This analysis is further confirmed by the fact that in Bulgarian, a language without overt case marking, the objects co-occuring with the reflexive in the context of na Q -, take the form of a PP, in contrast to the NPs quantified by na Q -. This is illustrated in (62). 189 (62) a. Na-gotvih se *(na) kartofi. na Q -cooked.1sg REFL on/to potatoes ‗I did enough (perhaps even too much) of cooking potatoes.‘ (i.e. I am tired of cooking potatoes) b. Na-gotvih kartofi na Q -cooked.1sg potatoes ‗I cooked a lot of potatoes.‘ 2.2. The quantificational uses of po- As mentioned earlier in this chapter, quantificational prefix po Q - has a more diverse set of uses than na Q -. Here I distinguish between two such uses: the distributive po D - and po Q - with the meaning ‗some, a little‘ that can measure over various scales supplied by the predicate, as well as the internal argument. I take po D - and po Q - to be two distinct prefixes, contra Jabłońska (2004), who analyzes all uses of po- as involving one prefix of variable meaning, with its various interpretations being motivated by the kind of predicate to which it attaches. As I will show later, there are facts such as differences in case assignment and the availability of either distributive or ‗some, a little‘ interpretation with some predicates, that support the differential approach taken here. 2.2.1. The distributive po D - The distributive prefix po D - is related to the free standing distributive marker po D . The two are in complementary distribution: po D - distributes the content of the predicate over other elements in the clause, while the free standing po D distributive-marks DPs/NPs (and assigns locative case to it, just like the preposition(s) po ‗after, over‘) and distributes 190 them over another constituent. Some examples of this complementary use are given in (63) and (64) below 14 . (63) po D - (distributive-marks the predicate and distributes over other elements in the clause) distribution wrt OBJECT: a. Nauczyciel po-otwierał okna. teacher.NOM po D -opened.3SG windows.ACC ‗The/a teacher opened all of the windows.‘ distribution wrt OBLIQUE ARGUMENT: b. Nauczyciel po-otwierał okno wszystkim klasom. teacher.NOM po D -opened.3SG window.ACC every class.DAT ‗The/a teacher opened a window for (each) class (of students).‘ distribution wrt SUBJECT & multiple event interpretation: c. Nauczyciele po-otwierali okno (jeden za drugim). 15 teachers.NOM po D -opened.3PL window.ACC (one after another) ‗(The) teachers each opened a/the window (one after another).‘ 14 It is possible to have the two po D ‘s to co-occur, as in (i), although the speakers I consulted prefer a ―simpler‖ version without the prefix po D -, as in (ii). It is not immediately clear what the prefix po D - contributes to the meaning in (i). (i) Mama po-nakładała im po dużej gałce /po trzy gałki lodów. mother po D -put them.DAT DISTR big scoop /DISTR three scoops ice-cream ‗Mother gave every person a big scoop/three scoops of icecream.‘ (ii) Mama nałożyła im po dużej gałce /po trzy gałki lodów. mother put them.DAT DISTR big scoop /DISTR three scoops icecream ‗Mother gave every person a big scoop/three scoops of icecream.‘ 15 This sentence might not sound natural without providing proper context, e.g., The teachers each opened the window - as they were instructed/as part of a fire drill, etc. 191 distribution wrt UNACCUSATIVE SUBJECT: d. Nauczyciele po-przy-chodzili na zebranie parami. teachers.NOM po D -at-came.3PL to meeting in-pairs ‗(The) teachers came to the meeting in pairs.‘ (64) freestanding po D (distributive-marks a DP/NP and distributes over another constituent in the clause) distribution wrt SUBJECT, OBLIQUE ARGUMENT, or EVENT (MODIFIER): a. Dzieci przyniosły po kwiatku dla ulubionych nauczycieli. children.NOM brought.3PL DISTR flower.LOC for favorite teachers 1. ‗(The) children brought one flower for each of their favorite teachers.‘ 2. ‗(The) children each brought one flower for their favorite teachers.‘ b. Na każdą lekcje przyszło po 20 dzieci. to every class came.3SG DISTR 20 children.GEN ‗20 children attended each class.‘ c. Co roku przyznawano po jednej nagrodzie. what year awarded.IMPERS DISTR one prize ‗Each year they awarded one prize.‘ The distributive po D -differs from na Q - in that it does not take genitive marked objects, even when the predicate is distributed over the object. We could possibly attribute it to the fact that distributivity involves universal quantification, and universal quantifiers are not genitive case assigners in Polish. In other respects, po D - is also rather unproblematic, and there is not much to be said about it before we consider its aspectual function, to which I will return in Chapter 6. 192 2.2.2. po Q - ‘some, a little’ The other kind of quantificational po Q - has the meaning of ‗some, a little,‘ but could also connote satisfactory amount and mean ‗quite a bit.‘ In this respect, po Q - is more like na Q - than the distributive po D -. Po Q - can, just like na Q -, can apply to the interal argument, in which case the object is in the genitive case, but it can also as easily quantify over the event, and more concretely over different scales supplied by the predicate, including time, intensity, etc. 16 This is illustrated in (65)-(67) below. (65) Transitive predicate; po Q - interpreted wrt the object: Po-zrywałam kwiatków. po Q -plucked.1SG flowers.GEN ‗I picked some flowers.‘ (66) Transitive predicate; po Q - quantifies over the event (object is non-quantifiable): Po-kopał piłkę/*piłki. po Q -kicked.3SG ball.ACC/*ball.GEN ‗He played soccer for a little while/he did some soccer playing.‘ 16 Given the examples below, one might be led to believe that po Q - actually quantifies over the event time, and so it means ‗for a while, for a short amount of time.‘ This is essentially the interpretation proposed in Jabłońska, who analyzes this use of po- as a temporally delimiting adverbial with a presupposition that the event could continue. However, while this may appear to be the case for many examples, others indicate that temporal delimitation is an inferences and not the actual role of po Q -, e.g., (i) Po-cierpiał 10 lat i w końcu zmienił pracę. po Q -suffered.3SG 10 years and finally changed.3SG job ―He suffered a little for 10 years and finally changed his job.‖ Just like na Q - quantifying over the event does not necessarily entail an extended duration (e.g., one can get a lot/enough reading done in a short time), po Q -, too, may indicate a temporally prolonged event of a, so to say, low intensity. In that case, the quantifier po Q - measures over intensity scale rather than time scale. 193 (67) Intransitive predicate; po Q - quantifies over the event: a. Po-chodziłam po parku. po Q -walked.1SG about park ‗I walked in the park for a little while/I did some walking in the park.‘ b. Po-marzył sobie. po Q -dreamt.1SG self.DAT ‗He did some/a little dreaming.‘ As the example in (66) shows, quantification over the event is possible with po Q - in the presence of a direct object, and the relevant reading obtains without the need for the reflexive accusative clitic się, in contrast to what we observed with na Q -. In fact, adding się to a po Q -prefixed verb that is not inherently reflexive leads to ungrammaticality, as shown in (68) and (69). (68) Po-śpiewałam (*się). po Q -sang.1SG REFL.ACC ‗I sang some / a little.‘ (69) Po-kopał (*się). piłkę/*piłki. po Q -kicked.3SG REFL.ACC ball.ACC/*ball.GEN ‗He played soccer for a little while/he did some soccer playing.‘ In a further departure from the properties of na Q -, po Q - is able to quantify over the event even if the direct object is plural, and therefore quantifiable. Consider the following actual usage examples found online. 194 (70) Poradziłem, aby po-czytał gazety oleśnickie z 1922 r... advised.1SG that po Q -read.3SG.M newspapers.ACC of-Oleśnica from 1922 year… ‗I advised him to do some reading of the Oleśnica newspapers from 1922…‘ (71) Po-czytałam sobie, bardzo dużo informacji. po Q -read-I self.DAT very many information.PL.GEN ‗I did some/a bit of reading of a lot of information.‘ In (70), the interpretation is not one where one advises another to read some of the relevant newspapers (that interpretation would be possible with genitive marked ‗newspapers‘), but rather to do some reading of those newspapers, possibly all of them as a best case scenario. In (71) the fact that po Q - does not quantify over the object is even clearer, with the object being independently quantified by a quantifier with opposite meaning. The above discussed freedom of po Q - to quantify over the event rather than the direct object is reflected in the fact that, unlike with na Q -, the direct object of a po Q - prefixed verb need not always be in the genitive case. Instead, genitive is assigned only when po Q - is interpreted with respect to the direct object, as in (65). In other cases, the object receives the expected accusative case marking, as illustrated in (66) and (70) (in (71), the object is in the genitive because it is independently quantified by dużo ‗many‘). To account for the genitive marking when po Q - is interpreted with respect to the internal argument we can extend the analysis earlier proposed for na Q -, where the quantificational prefix functions as an intensifier with respect to the internal argument, and where that argument must be independently quantified by a nominal quantifier or a measure phrase denoting a large quantity. The nominal quantifier in such situation can be 195 overt or covert, and it is that quantifier or a measure phrase that assigns the genitive case to the NP. If correct, extending this account to po Q - predicts that po Q -prefixed verbs can take as their object either a genitive marked nominal with a null quantifier, as in (65), as well as an object with an overt quantifier or measure phrase. The examples in (72) and (73) show that the prediction is borne out. (72) Po-zrywałam trochę / kilka / garstkę kwiatków. po Q -plucked.1SG some / a few / little handful flowers.GEN ‗I plucked some / a few / a small bunch of flowers.‘ (73) Podczas wakacji po-czytałam trochę / kilka / parę książek. during vacation po Q -read.1SG some / a few / a couple books.GEN ‗During my vacation, I read some / a few / a couple books.‘ While the generalizations above are generally true, it needs to be noted that not all verbs participate in the genitive/accusative alternation and in this way encode two different functions of po Q -. For example, the verb po-jeść ‗to eat quite a bit‘ is compatible only with a genitive-marked object, while po-brudzić ‗to dirty a bit‘ takes only accusative objects. Further, the alternation is limited to verbs where po Q - attaches to an unprefixed verb. In all other cases, i.e., when po- attaches to an already prefixed verb, the case assignment on the quantifiable direct object can be used to disambiguate the role of po-: if the object is gentitive-marked, po Q - is interpreted as ‗some, a little;‘ if the object is accusative-marked, po- is interpreted as a distributive marker. 17 This is illustrated in (74). 17 When po- attaches to an unprefixed verb, its interpretation is often limited to that of ‗some, little.‘ For example, a sentence like Poczytałam książki ‗I po-read books‘ can only mean ‗I did some reading of books‘ – the distributive interpretation ‗I read/was reading each of the books‘ is not available. The reason for that might be that distributive interpretation is computed more easily on clearly individuated events. When an unprefixed verb has more of a delimited event meaning, distributed reading for po- is perfectly possible, e.g., 196 (74) a. Po Q -wy-rzucałam starych mebli. po Q -out-threw.1SG old furniture.PL.GEN ―I threw away some old furniture.‖ #‘I threw away all old furniture.‘ b. Po D -wy-rzucałam stare meble. po D -out-threw old furniture.PL.ACC ―I threw away all old furniture.‖ #‘I threw away some old furniture.‘ Given the above discussion of the uses of po Q -, we can conclude that po Q - would be best identified as an element with two functions: (i) an intensifier highlighting a small or unspecified quantity of the internal argument, which intensifier, like na Q -, has the ability to license a phonologically null quantifier of like meaning, and (ii) a variable quantifier in the sense of Nakanishi and Romero (2004), as previously, and mistakenly, proposed for na Q - in Romanova (2006). As a variable quantifier, po Q - can quantify over any of the scales made available by a predicate, including temporal duration and intensity, as well as other scales that could be used to measure event amount. This conclusion is further confirmed by the fact that po Q - can attach to degree achievement predicates (something that na Q - could not do), and is interpreted in such cases as a quantifier over ‗degree of change‘ scale or range in the sense of Schwartzchild (2006), (i) Ksiądz po-chrzcił dzieci i zaczął kazanie. priest DISTR-baptized children.ACC and started sermon ‗The priest baptized (each of) the children and started the sermon.‘ (ii) Wyborcy po-głosowali na kandydatów do sejmu. voters DISTR-voted on candidates to parliament ‗The voters each voted for parliamentary candidates.‘ or ‗The voters voted for all of the offices that the parliamentary candidates were running for.‘ 197 i.e., ‗some (= a visible amount of) change.‘ I will discuss po Q -prefixed degree achievements in more detailed in the next chapter in the context of Aktionsart interpretation of Q-prefixed verbs, and here I include just two illustrative examples: (75) a. Mleko po-drożało. milk.NOM po Q -got-expensive.3SG ‗Milk got more expensive.‘ cf. b. Mleko drożało. milk.NOM got-expensive.3SG ‗Milk was getting more expensive.‘ (76) a. Prezydent bardzo się po-starzał. president much REFL po Q -got-old-looking.3SG ‗The president got old-looking a lot in a year.‘ cf. b. Prezydent *(bardzo) starzał się. president (much) got-old-looking.3SG REFL ‗The president was getting old-looking for a year.‘ 2.3. Summary In this section I discussed the quantificational properties of the Slavic stackable prefixes na Q -, po D - and po Q -. The main observations are summarized in the table below. In the next section, I consider the other unique feature of Q-prefixes, namely stackability. 198 Table 8: Quantification-related properties of the stackable prefixes na Q -, po Q -, and po D - na Q - po Q - po D - MEANING a lot / in great quantity / to a great extent / enough / satisfactorily some / a little / a few distributive quantifier QUANTIFIES OVER EVENT (MEASURES OVER DIFFERENT SCALES) YES, but only in the presence of the reflexive clitic się YES YES QUANTIFIES OVER DIRECT OBJECT Not directly – it acts as an INTENSIFIER which has the ability to license a null nominal quantifier with the meaning ‗a lot‘ Not directly – it acts as an INTENSIFIER which has the ability to license a null nominal quantifier with the meaning ‗some, a little, a few‘ YES – distributes TAKES GENITIVE OBJECT YES, always YES, but only when applying to the object NO QUANTIFICATION VARIABILITY EFFECT NO YES YES – free choice of distributee 3. Stackability While quantificational properties make na Q -, po D - and po Q - different from all other prefixes, there is another characteristic they possess that makes them a standout among Slavic prefixes in general, namely their stacking ability. What this means is that in addition to attaching to simplex imperfective forms, Q-prefixes can also be added to forms that are already prefixed. This is illustrated in (77)-(79) below. 199 (77) na Q - Na-z-bierałam grzybów. na Q -from-took.1SG mushroom.PL.GEN ‗I picked a lot of mushroom.‘ (78) po Q - Po-wy-szywałam sobie dzisiaj. po Q -out-sewed.1SG self.DAT today ‗I did a little/some embroidering today.‘ (79) po D - Student po-prze-pisywał notatki. student.NOM po Q -through-wrote.3SG notes.ACC ‗The student copied all the notes.‘ Other prefixes cannot be stacked this way, even if interpretation could be easily computed based on the meaning of the coupled prefixes, e.g., (80) a. Student *wy-prze-pisywał notatki. student.NOM out-through-wrote.3SG notes.ACC ‗The student copied the notes completely.‘ b. Student *od-prze-pisywał notatki. student.NOM from-through-copied.3SG notes.ACC ‗The student copied the notes from somebody.‘ 200 Further, stackability of na Q -, po Q - and po D - is not limited to one or another stacked prefix per verb. Instead, they can co-occur, creating multi-stacked structures. 18 In such cases, however, each prefix must be easily interpreted as applying to a different semantic argument. Forms where two Q-prefixes could be taken to apply to the same item, even though an alternative interpretation would in principle be possible (i.e., one prefix applies to the direct object and the other quantifies over the event) are judged as infelicitous. This is illustrated in (81). (81) na Q - & po Q -; direct object in the genitive a. *Po-na-zbieraliśmy grzybów. po Q -na Q -gathered.3PL mushrooms.GEN b. *Na-po-zbieraliśmy grzybów. na Q -po Q -gathered.3PL mushrooms.GEN Could not mean: ‗We did some picking of a lot of mushroom‘ (the alternative interpretation: ‗We did a lot of picking of some mushroom is independently precluded due to the fact that na Q - must quantify over the object).‘ Also, a combination of na Q - & po Q - where both are to quantify over the event (but over different scales), would create conflicting demands with respect to the reflexive: it is required with na Q - and impossible with po Q -. Consequently, such configurations are excluded, as shown in (82). (82) na Q - & po Q - quantifying over the event and creating conflicting a. *Student po-na-męczał (się). student.NOM po Q -na Q -tired.3SG REFL 18 Istratkova (2006) gives examples of as many as seven stacked prefixes in Bulgarian, noting, however, that they may be hard for a native speaker to process. 201 b. *Student na-po-męczał (się). student.NOM na Q -po Q -tired.3SG REFL Potential reading: ―He got tired a lot for some time‖ or ―He got tired a little quite a bit.‖ The combinations where conflicts do not appear to arise are those involving the distributive po D - and either na Q - or po Q -. First I show some examples involving na Q -: (83) a. Wujek po-na-obiecywał dzieciom presentów. Uncle.NOM po D -na Q -promised.3SG children.DAT presents.GEN ‗A/the uncle promised each of them many presents.‘ b. Wujek na-po-obiecywał dzieciom presentów. Uncle.NOM po D -na Q -promised.3SG children.DAT presents.GEN ‗A/the uncle promised each of them many presents.‘ (84) a. Studenci po-na-za-praszali się tego nauczyciela. students.NOM po D -na Q -invited.3PL REFL this teacher.GEN ‗The students each did a lot of inviting of this teacher.‘ b. ???/*Studenci na-po-za-praszali się tego nauczyciela. students.NOM po D -na Q -invited.3PL REFL this teacher.GEN ‗The students each did a lot of inviting of this teacher.‘ Admittedly, good examples involving reflexivized uses of na Q - and the distributive po D - are hard to find, and when one does find them, only the po D -na Q - order seems fully acceptable. The same restriction does not appear to apply to cases where na Q - empasizes 202 the quantity of the the object, and both orders are possible. 19 However, the different orderings do not translate into different interpretations. In the case of na Q -, this can be attributed to the fact that the distributive quantifier excludes a collective reading of the distributee such that na Q - + QP could take scope over it (i.e., for (83), ‗the many presents that the children were promised as a group‘). The very same situation obtains with the combination of po D - and po Q -, where the latter applies to the object. There, the homophony of the two involved prefixed makes is rather hard to decide which is which, e.g., (85) Mikołaj po-po-przy-nosił wszystkim zabawek. Santa po Q -po D -at-carry.3SG all.DAT toys.GEN Santa po D -po Q -at-carry.3SG all.DAT toys.GEN ‗Santa brought some toys to each of them.‘ Given the aspectual properties of po Q - and po D -, I will eventually propose that the actual order is po Q -po D -, but at this point it is not clear whether this claim is falsifiable. The same problem concerns situations where po Q - applies not to the object but to the event. Consider the examples in (86). (86) Dzieci po-po-prze-wracały sobie krzesła. children po-po-knocked-over.3PL self.DAT chairs.ACC 1. ‗(The) children did a little/some of knocking over each of the chairs.‘ 19 While both combinations are possible, the po-na- order appear to be statistically more frequent in unreflexivized contrexts. The number of Google hits for po-na- greatly outnumbers those for na-po-, e.g., 872 tokens for the 3 rd person singular masculine form ponawypisywał vs. 53 for napowypisywał, and 300 tokens for the 1 st person singular form ponawypisywałem vs. 3 napowypisywałem. This may be reflective of a general tendency in Slavic to keep na Q - relatively low, likely due to the fact that it quantifies over the object. Istratkova (2006) argues that in Bulgarian the distributive and the attenuative po- (‗to do something to a certain extent, with low intensity‘) is merged over na-, and that deliminative po- (‗for a while‘) does not allow any other prefixes on top of it. This is in agreement with the suggestions I will make for Polish below, but it needs to be noted Istratkova (2006) argues for a strict ordering of all stackable prefixes in Bulgarian, while in Polish I will assume that only po Q - needs to be marged higher than other Q-prefixes. 203 2. ‗(The) children knocked over each of the chairs a few times.‘ 3. ‗(The) children each did a little/some knocking over the chairs.‘ The first two interpretations indicate a po Q -po D - order, while the last one would seem to call for the po D -po Q - ordering. However, with the interpretation po D - being free in the sense that it is able to select any eligible (i.e., plural) distributee, it is not clear if a po D - over po Q - ordering is necessary to obtain the interpretation in (86)-3. What further indicates that the actual order is po Q -po D - is the fact that the additional po- in (86) sounds more natural with the reflexive sobie ‗to self,‘ which spells out the scale for po Q - ‗a set of degrees to someone‘s liking .‘ How sobie improves the second po- is shown in (87). (87) a. Dzieci po-prze-wracały krzesła. children po D -knocked-over.3PL chairs.ACC 1. ‗The children knocked over all of the chairs.‘ 2. ‗The children each knocked over the chairs.‘ b. ??Dzieci po-po-prze-wracały krzesła. children po-po-knocked-over.3PL chairs.ACC c. Dzieci po-po-prze-wracały sobie krzesła. children po-po-knocked-over.3PL self.DAT chairs.ACC (interpretation as in (86)). This pattern strengthens the case for the po Q -over po D - ordering, which will prove important for accounting for the differences in aspectual properties of different Q- prefixes. Beside the combinations of the distributive po D - with either po Q - or na Q -, it is also possible to stack two po Q -‗s. This provides an additional strong argument for the analysis 204 of po Q - as variable quantifier, i.e., a quantifier free to measure over any of the scales provided by the predicate as well as the object. In the example (88)a below, one po Q - quantifies over the object, and the other one over event (time, frequency/temporal density, or intensity), while in (88)b, both po Q -‗s measure over event-level scales – intensity and frequency. 20 (88) a. Po-po-czytywali sobie 21 gazet. po Q -po Q -read.3PL self.DAT newspapers.GEN ‗They read some newspapers for a short time/from time to time/ unattentively.‘ b. Po-po-czytywali sobie gazety. po Q -po Q -read.3PL self.DAT newspapers.ACC ‗They read newspapers for a short while from time to time.‘ With na Q -, my conclusion was different – I concluded it was not normally (i.e., when not reflexivized) a variable quantifier, but rather had to be interpreted with respect to the internal argument. This predicts that stacked na Q -‘s should not be possible, which is born out: (89) a. *Na-na-budowali domów. na Q -na Q -built.3PL houses.GEN Hypothetical reading: ―They built a lot of houses many times.‖ 20 I would like to thank Barbara Tomaszewicz for verifying judgements as to the interpretation of the examples in (88). 21 Here again the dative reflexive siebie assists in differentiating the scales for the two po-‗s. Without it, the relevant clause is not felicitous. The same is true of the example below: (i) Muzycy po-po-grywali sobie na gitarze. musicians.NOM po D -po Q -played.3PL self.DAT on guitar. ‗The musicians played a little on the guitar from time to time.‘ 205 b. *Na-na-męczył się przy budowie domu. na Q -na Q -tired.3PL REFL at building house Hypothetical reading: ―He got tired a lot many times while building a/the house.‖ The facts presented above show, albeit in a very general way, the unique distributional characteristics of Q-prefixes and give an approximation of their syntactic ordering. The reader may already anticipate that they may require an alternative and higher merge position from non-quantificational prefixes, with which they can co-occur. This problem is considered in the next chapter, where I look at the aspectual role of quantificational prefixes. 4. Summary In this chapter I described the special properties of three (and only) Q-prefixes in Polish: na Q -, po Q -, and po D -. With all their quirks, some of which will have to remain a topic for further research, the quantificational nature of the relevant prefixes is quite clear, and it is this quantificational quality together with their stacking ability that clearly separates these prefixes from all other ones. However, as I will show in the next chapter, they also share surprising similarities with non-quantificational prefixes with respect to aspectual interpretation, which has previously led sometimes to indiscriminatory discussion of aspectual properties of Slavic prefixes in general, with some unfortunate conclusions for the non-stacking, telicity marking prefixes, e.g., Filip (1999, 2004). I will investigate in that chapter the Aktionsart and viewpoint contribution of the relevant prefixes, and show that other than po Q -, quantificational prefixes yield telic reading, 206 which in situations where they attach to secondarily imperfectivized forms, requires recursion of a telicity-encoding projection. 207 CHAPTER 6: QUANTIFICATIONAL PREFIXES AND ASPECT In the previous chapter, I described the distribution and the quantificational function of Q(uantificational)-prefixes, including their scope and the impact on internal argument realization. Both the quantificational function and the ability to attach to already prefixed verbs make Q-prefixes different from the telicity-marking prefixes discussed in Chapter 3. Here I consider the properties that unite Q-prefixes with non- quantificational ones, starting with their classification with respect to the Slavic perfective-imperfective distinction, to their Aktionsart role, and finishing with their viewpoint interpretation As I will show, of the considered Q-prefixes, only verbs prefixed with po Q - have an atelic interpretation. All other uses of Q-prefixes lead to telic readings. Since their quantificational nature could not by itself account for that in the family of languages where free-standing quantifiers of comparable meanings do not give rise to telic readings, prefixes na Q - and po D - must be analyzed as telicity markers on a par with other verbal prefixes. This in turn necessitates a recursive telicity-encoding structure. Q-prefixed verbs, as I will show, are even more uniform in their viewpoint interpretation – it is always PERFECTIVE. For telic forms that follows from the previously identified semantic restriction of viewpoint composition in Slavic, i.e., only PERFECTIVE viewpoint can compose with telic predicates. In the case of po Q -, on the other hand, it must be stipulated that po Q - either marks the PERFECTIVE viewpoint itself or has to agree with a PERFECTIVE feature heading ViewAsp. 208 1. Q-prefixed verbs and the perfective-imperfective distinction A surprising fact about stacking prefixes, given their unique quantificational role and less-constrained distribution (i.e., their stacking ability), is that Q-prefixed verbs pattern exactly the same way as verbs with non-quantificational prefixes with respect to various Slavic aspectual classification tests. For that reason, verbs prefixed with quantificational prefixes are traditionally classified in Slavic as perfective alongside other prefixed verbs. This applies both to forms where na Q -, po Q -, and po D - are the only prefix as well as to those where they are stacked, as illustrated in (1)-(2) below for two common tests for Slavic (im)perfectivity. (1) Obligatory FUTURE TIME denotation for present tense forms of perfective verbs perfectives imperfectives (primary or secondary) i. as a single prefix a. Na-jem się u babci. vs. Jem obiad u babci. na Q -eat.1SG REFL at grandma eat.1SG dinner at grandma ‗I will eat to satisfaction at grandma‘s.‘ ‗I am eating dinner at grandma‘s.‘ #―I eat/am eating to satisfaction at grandma‘s.‖ b. Na-gotuję zupy. vs. Gotuję zupę. na Q -cook.1SG soup.GEN cook.1SG soup.ACC ‗I will cook a lot of soup.‘ ‗I am cooking soup.‘ #‘I cook/am cooking a lot of soup.‘ c. Po-chodzę po parku. vs. Chodzę po parku. po Q -walk.1SG around park walk.1SG around park ‗I will take a little walk in the park.‘ ‗I walk/am walking in the park.‘ 209 # ‗I take/am taking a little walk in the park.‘ d. Po-głosuję na nich vs. Głosuję na nich. po D -vote.1SG on them vote.1SG on them ‗I will vote for each of them.‘ ‗I vote/am voting for them.‘ #‘I vote/am voting for each of them.‘ ii. as a stacked prefix e. Na-przy-mierzam się dziś. vs. Przy-mierzam ubrania. na Q -at-measure.SI.1SG REFL today at-measure clothes.ACC ―I will do a lot of trying things on today.‘ ‗I try on/am trying on clothes.‖ #‘I do/will do a lot of trying things on today.‘ f. Na-prze-stawiam krzeseł. vs. Prze-stawiam krzesła. na Q -through-stand.SI.1SG chairs.GEN through-stand.SI.1SG chairs.ACC ‗I will move a lot of chairs.‘ ‗I move/am moving chairs.‘ #‘I move/am moving a lot of chairs.‘ g. Po-o-palam się. vs. O-palam się. po Q -about-burn.SI.1SG REFL about-burn.1SG REFL ‗I do some tanning.‘ ‗I tan/am tanning.‘ #‘I tan/am tanning.‘ h. Po-od-rzucam wszystkie oferty. vs. Od-rzucam wszystkie oferty. po D -away-throw.SI.1SG all offers away-throw.1SG all offers ‗I will reject all offers.‘ ‗I reject/am rejecting all offers.‘ #‘I reject/am rejecting all offers.‘ 210 (2) Incompatibility with PHASE VERBS perfectives imperfectives (primary or secondary) i. as a single prefix a. *Skończyłam się na-jeść. vs. Skończyłam jeść. finished.1SG REFL na Q -eat.INF finished.1SG eat.INF ‗I finished filling my stomach. ‗I finished eating.‘ b. *Skończyłam na-gotować zupy. vs. Skończyłam gotować zupę. finished.1SG na Q -cook.INF soup.GEN finished.1SG cook.INF soup.ACC ‗I finished cooking a lot of soup.‘ ‗I finished cooking soup.‘ c. *Zaczęłam po-chodzić po parku. vs. Zaczęłam chodzić poparku started.1SG po Q -walk.INF around park started.1SG walk.INF around park ‗I started taking a little walk in the park.‘ ‗I started walking in the park.‘ d. *Przestał po-głosować na nich. vs. Przestał głosować na nich. stopped.3SG po D -vote.INF on them stopped.3SG vote.INF on them ‗He stopped voting for each of them.‘ ‗He stopped voting for them.‘ ii. as a stacked prefix e. *Skończyłam na-przy-klejać znaczków. vs. Skończyłam przy-klejać znaczki. finished.1SG na Q -at.glue.SI.INF stamps finished.1SG at.glue.SI.INF stamps ‗I finished putting a lot of stamp on.‖ ―I finished putting stamps on.‖ f. *Zaczęłam po-przy-jeżdżać do domu. vs. Zaczęłam przy-jeżdżać do domu. started.1SG po Q -at-go.SI.INF to home started.1SG at-go.SI.INF to home ‗I started coming home a bit.‘ ‗I started coming home.‘ 211 g. *Przestał po-roz-wieszać ulotki. vs. Przestał roz-wieszać ulotki. stopped.3SG po D -out-hang.SI.INF fliers stopped.3SG out-hang.SI.INF fliers ‗He stopped posting the fliers.‘ ‗He stopped posting fliers.‘ The fact that na Q -, po D -, and po Q -prefixed verbs test as perfectives would suggest that they must share the semantic features defining Slavic perfectives, i.e., telicity and PERFECTIVITY by the account presented here (the reader will recall from the previous chapters that imperfectives in Slavic languages without overt imperfect-aorist marking system can be interpreted as either IMPERFECTIVE or PERFECTIVE, while Slavic perfective verbs can have only viewpoint PERFECTIVE interpretation). To determine if this is indeed the case, I consider below the behavior of Q-prefixed verbs with respect to various tests for telicity and PERFECTIVITY. I will start with the former, using the standard adverbial modification test as well as the cumulativity and divisity diagnostics when needed, and then use the coordination test for concurrent or sequential interpretation to look at the viewpoint aspect interpretation of na Q -, po D -, and po Q -prefixed verbs. 2. Quantificational prefixes and telicity 2.1. Po D - and time adverbial test Perfectives, as the reader may recall from earlier chapters, test uniformly as telic with respect to time adverbial modification test: they are compatible with completive adverbials and incompatible with durative ones. 1 As it turns out, when this test is applied to Q-prefixed verbs, the results are mixed. Of the three considered Q-prefixes, only verbs prefixed with the distributive po D - behave consistently like perfectives with non- quantificational prefixes. This is illustrated in (3)-(5). 1 This generalization naturally sets aside achievements, which would be incompatible with either adverbial, but are uncontroversially telic. 212 (3) Po-prał firanki *godzinę / w godzinę. po D -washed.3SG curtains.ACC hour/in hour ‗He washed (all) the curtains for an hour/in an hour.‘ cf. Prał Impf firanki godzinę / *w godzinę. washed.3SG curtains.ACC hour/in hour ‗He washed/was washing the curtains for an hour/in an hour.‘ (4) Po-roz-wieszał pranie *10 minut / w 10 minut. po D -out-hang.3SG laundry.ACC 10 minutes / in 10 minutes ‗He hang all the laundry for 10 minutes/in 10 minutes.‘ cf. Roz-wieszał 2ndImpf pranie 10 minut / *w 10 minut. out-hang.3SG laundry.ACC 10 minutes / in 10 minutes ‗He hang/was hanging the laundry for 10 minutes / in 10 minutes.‘ (5) Ksiądz po-chrzcił pięcioro dzieci *godzinę / w godzinę. priest po D -baptized.3SG five children hour / in hour ‗A/the priest baptized all five children in an hour.‘ cf. Ksiądz chrzcił pięcioro dzieci godzinę / *w godzinę. priest baptized.3SG five children hour / in hour ‗A/the priest baptized five children for an hour.‘ The examples above indicate that po D - derives telic readings. The question is whether this is because po D - is a marker of telicity, or due to some other factors that underlie distributive interpretation. The problem that comes to mind concerns pluractionality. The general issue of Aktionsart interpretation in pluractional contexts has been set aside in this work, but needs to be brought in, albeit rudimentarily, in the context 213 of po D -. What is special about such contexts is that they allow for telic readings of otherwise atelic imperfectives, as evidenced by their compatibility with completive adverbials. This is shown in (6) and (7). (6) Codziennie / wiele razy czytałam książkę godzinę / w godzinę. Everyday / many times read.1SG book.ACC hour / in hour ‗Everyday / Many times I read a book for an hour / in an hour.‘ (7) Zawsze / kilka razy prze-pis-yw-ałam notatki godzinę / w godzinę. Always / a few times through-wrote.2IMPF.3SG notes.ACC hour / in hour ‗I always copied (the) notes for an hour / in an hour / I copied (the) notes for an hour / in an hour many times.‘ The examples in (3)b, (4)b, and (5)b would also all allow the completive adverbial on the habitual reading. The availability of telic readings with imperfectives in habitual/iterative contexts can be related to the fact that in Polish (also in East and South Slavic), perfectives have only episodic, single event interpretation, as shown in (8)-(10). (8) (*Codziennie / *często / *w soboty) na-pisałam 5 mejli. (everyday / eoften / *en Saturdays) [telic]-wrote.1SG 5 emails ‗I wrote 5 emails (*everyday / *often / o*n Saturdays).‘ (9) (*Zawsze / *zazwyczaj) użyłam oleju z oliwek. always / usually used.1SG oil from olives ‗I (always / usually) used olive oil.‘ 214 (10) Na śniadanie z-jadłam płatki. on breakfast ate.1SG cereal ―I ate cereal for breakfast‖ (no habitual reading) Consequently, only imperfective forms can express habituality / iterativity, and when they do, their Aktionsart characteristics appear neutralized, so that they can receive either telic or atelic interpretation (note that both completive and durative adverbials are possible in (6) and (7). This is, however, also what makes their situation different from po D -prefixed verbs, which cannot have atelic interpretation (durative adverbials are excluded). Consequently, pluractionality itself cannot account for the necessarily telic readings of po D -prefixed verbs. The other property that might be potentially considered in relation to the telic nature of po D -prefixed verbs is distributivity itself. It could be hypothesized that distributive readings require an iteration of atomic events, and telic events are more naturally conceptualized as atomic. In English, however, distributivity does not correlate in any way with telicity, i.e., one can derive both atelic and telic distributive readings, e.g., (11) Each student read a book for 5 hours / in 5 hours. (12) The student read each book for 5 hours / in 5 hours. (13) I lived in each place for 2 years. (14) I made each of them a gift in 1 day. In Polish, too, distributive readings, which can be marked overtly with the free standing distributive marker po, are compatible with either telic or atelic interpretations, as decided by the verb form. This is shown in (15)-(20). 215 (15) Simplex imperfective (=atelic) Ochotnicy sadzili po 10 drzew 3 miesiące / *w 3 miesiące. volunteers planted DISTR 10 trees 3 months / in 3 months ‗Volunteers planted 10 trees each for 3 months.‘ (the distribution could also with respect to some contextually specified time intervals: days, weeks, months, etc, or with respect to the time adverbial included in the sentences, in which case w 3 miesiące ‗in 3 months‘ becomes fine as it functions as the distributee; for a sentence to sound natural, context is required, e.g., Volunteers planted 10 trees each for 3 months, and 20 trees each for 2 months) (16) Perfective (telic) Ochotnicy za-sadzili po 10 drzew 3 miesiące / *w 3 miesiące. Volunteers INCPT 2 -planted DISTR 10 trees 3 months / in 3 months ‗Volunteers planted 10 trees each in 3 months.‘ (the distribution can be only with respect to the subject) (17) Secondary Imperfective (atelic) Ochotnicy za-sadzali po 10 drzew 3 miesiące / *w 3 miesiące. Volunteers INCPT-planted DISTR 10 trees 3 months / in 3 months ‗Volunteers planted 10 trees each in 3 months.‘ 2 The meaning of za-sadzić, though glossed as involving the inceptive prefix, is not too different from the form po-sadzić, where po- acts as a pure telicizer, e.g., (i) Ochotnik po-sadził drzewo *godzinę / w godzinę. volunteer [TELIC] -planted tree (for) hour / in hour ―A/the volunteer planted a/the tree in an hour.‖ The fact that there is some additional meaning nuance which is hard to capture in translation (focus on the initial stages of planting?) manifests itself in the ability of za-sadzić to derive the secondary imperfective form, which is not possible with po-sadzić. 216 (same quanlifications as under (15) above) (18) Simplex imperfective unaccusative (=atelic) 2 tygodnie / *w 2 tygodnie na każdą lekcję szło po 20 dzieci. 2 weeks / in 2 weeks on each class went DISTR 20 children ‗For 2 weeks, 20 children went to each class.‘ (19) Perfective unaccusative (telic) *2 tygodnie / w 2 tygodnie na każdą lekcję prz-szło po 20 dzieci. 2 weeks / in 2 weeks on each class at-went DISTR 20 children ‗In 2 weeks, 20 children came to each class.‘ (20) Secondary Imperfective unaccusative (atelic) 2 tygodnie / *w 2 tygodnie na każdą lekcję przy-chodziło 3 po 20 dzieci. 2 weeks / in 2 weeks on each class at-went(non-dir) DISTR 20 children ‗For 2 weeks, 20 children were coming to each class.‘ Consequently, the fact that po D -prefixed verbs can have only telic interpretation indicates that in addition to po D - being a marker of distributivity, it is also a marker of telicity, just like non-quantificational prefixes. 2.2. Na Q - and time adverbial test When it comes to na Q -prefixed verbs, the matter is more complicated. Let us consider first the reflexivized uses: (21) Na-śpiewał się *godzinę / ? w godzinę. na Q -sang.3SG REFL (for) hour / in hour ‗He sang a lot/enough (?in an hour).‘ 3 This is a suppletive form. For verbs of motion, the secondary imperfective is derived using the form that when unprefixed denotes a non-directed motion. 217 (22) Na-pływał się *godzinę / ? w godzinę. na Q -swam.3SG REFL (for) hour / in hour ‗He swam a lot/enough (?in an hour).‘ As the examples indicate, na Q -prefixed verbs in constructions with the reflexive are incompatible with durative adverbials, but also degraded with completive ones. We could speculate that this is due to the ‗enough‘ reading, which makes comparable English sentences also sound somewhat odd, e.g., ?He sang enough in an hour, ?He are enough in an hour. Note, however, that the English examples improve noticeably if a modifier is added, ―objectivizing‖ or specifying the meaning of enough, e.g., He sang enough in an hour to make me wanna hire him, or He ate enough in an hour to run the bill to $200. Similarly in Polish, such modification improves the compatibility with completive adverbials dramatically, e.g., (23) Na-śpiewał się dość w godzinę, żeby zapamiętać tekst piosenek. na Q -sang.3SG REFL enough in hour so-that remember.INF lyrics.ACC songs.GEN ‗He did enough singing in an hour to remember the songs‘ lyrics.‘ (24) Jak myślisz, na-pływasz się w godzinę? how think.2SG na Q -swam.2SG REFL hour/in hour ‗What do you think, will you get enough swimming done in an hour.‘ We can therefore conclude that the readings of na Q -prefixed verbs with the reflexive are in fact telic. With regard to the uses of na Q - where it functions as an intensifier with respect to the object, we obtain the following pattern: 218 (25) Hania na-łuskała orzechów *godzinę / *w godzinę. Hannah na Q -shelled.3SG nuts.GEN (for) hour / in hour ‗Hannah shelled a lot/a quantity of nuts *for an hour / *in an hour.‘ (26) Lekarz na-wy-pisywał recept *pół godziny / *w pół godziny. Doctor na Q -prescribed.3SG prescriptions.GEN (for) half hour / in half hour ‗A/the doctor wrote a lot of prescriptions *for half an hour / *in half an hour.‘ Just as with the reflexivized uses, durative adverbials are not possible when na Q - quantifies over the internal argument. In addition, however, completive adverbials are equally bad. This pattern normally characterizes achievements, and so telic events without duration, but the events described in (25) and (26) are clearly durative. What could then be the reason for the observed incompatibility? It appears that the problem has to do with genitive marked objects, and more specifically, with genitive marked objects without a local, nominal quantifier. As I mentioned previously, select perfective verbs allow an alternation between an accusative- marked and a genitive-marked object, with the first one receiving a specific/definite interpretation characteristic of the objects of perfective verbs, while the latter receiving a non-referential reading. This is illustrated in below. (27) a. Hania wy-łuskała orzechy (i zjadła je). Hannah.NOM out-shelled.3SG nuts.ACC (and ate them) ‗Hannah shelled the nuts (and ate them).‘ b. Hania wy-łuskała orzechów (*i zjadła je). Hannah.NOM out-shelled.3SG nuts.GEN (and ate them) ‗Hannah completed shelling of nuts (*and ate them).‘ 219 (28) a. Mama od-grzała zupę (i posoliła ją). Mom.NOM back-heated.3SG soup.ACC (and salted it) ‗Mom heated up the soup (and put salt in it).‘ b. Mama od-grzała zupy (*i posoliła ją). Mom.NOM back-heated.3SG soup.GEN (and salted it) ‗Mom completed heating up of soup (*and put alt in it).‘ The application of the adverbial tests to the examples with genitive-marked objects gives results identical to those in (25)-(26), as shown below. (29) Hania wy-łuskała orzechów *godzinę / *w godzinę. Hannah.nom out-shelled.3sg nuts.GEN (for) hour / in hour ‗Hannah completed shelling of nuts *for an hour / *in an hour.‘ (30) Mama od-grzała zupy *5 minut / *w 5 minut Mom.NOM back-heated.3SG soup.GEN (for) 5 minutes / in 5 minutes ‗Mom completed heating up of soup *for 5 minutes / in 5 minutes.‘ Once a nominal quantifier or measure phrase is added, completive adverbials become fine: (31) Hania wy-łuskała dużo orzechów *godzinę / w godzinę. Hannah.nom out-shelled.3sg many nuts.GEN (for) hour / in hour ‗Hannah shelled a lot of nuts *for an hour / in an hour.‘ (32) Mama od-grzała garnek zupy *5 minut / w 5 minut Mom.NOM back-heated.3SG pot.ACC soup.GEN (for) 5 minutes / in 5 minutes ‗Mom heated up a pot of soup *for 5 minutes / in 5 minutes.‘ 220 The same can be shown to be true of genitive NPs licensed by the exclamatory particle ale. There, too, neither type of adverbial is felicitous, but judgments change altogether when a local nominal quantifier is added. The example in (33) shows how ale licenses a genitive marked object, while (34) shows adverbial modification possibilities. (33) a. *Chomsky na-pisał książek. Chomsky PREF-wrote.3SG books.GEN b. Ale Chomsky na-pisał książek. EXCL Chomsky PREF-wrote.3SG books.GEN ―What a number of books did Chomsky write!‖ (34) a. Ale Chomsky na-pisał książek *3 lata / ??? w 3 lata. EXCL Chomsky PREF-wrote.3SG books.GEN (for) 3 years / in 3 years ―What a number of books did Chomsky write *for 3 years/??? in 3 years!‖ b. Ale Chomsky na-pisał dużo książek *3 lata / w 3 lata. EXCL Chomsky PREF-wrote.3SG many books.GEN (for) 3 years / in 3 years ―What a number of books did Chomsky write *for 3 years/in 3 years!‖ Adding an overt quantifier or measure phrase, as it turns out, also changes the outcome of the adverbial modification test for na Q -prefixed verbs – they become compatible with completive adverbials, as illustrated in (35). (35) Na q -kleiłam dużo / pełno / masę / kupę pierogów *godzinę / w godzinę. na q -glued.1SG many / loads / mass / pile dumplings.GEN *(for) hour / in hour ―I made many/enough/loads of/mass of dumplings *for an hour / in an hour.‖ All these data suggest that there is some problem with computing telic interpretation for predicates with genitive-marked objects without an overt genitive case assigner. It is not 221 that such predicates become atelic, because durative adverbials are decidedly ungrammatical in the relevant contexts, but they are not straightforwardly telic. It might be tempting to speculate at this point that maybe in Slavic the form of the direct object does matter in constructing telic readings. Maybe, for example, telic readings require the objects to be DPs – hence their specific / definite readings, while genitive marked objects are NPs or QPs. As we will see shortly, however, the simultaneous incompatibility of durative and completive adverbials is not limited to telic predicates with genitive objects, and therefore it should not be related specifically to requirements that telic readings may impose on the form of the direct object. The exact source of this incompatibility will have to remain for now a topic for further research. Given the telic-like pattern exhibited by na Q -prefixed verbs with the reflexive, as well as the consistent infelicity of durative adverbials with all uses of na Q -, we can conclude that na Q - does in fact mark telic readings, just like the distributive po D -. That leaves us with one final Q-prefix to examine, namely po Q -. 2.3. Po Q - and time adverbials The situation with the prefix po Q - is complex and different from both po D - and na Q -. Po Q -prefixed verbs where the prefix quantifies over the event are compatible with durational adverbials and incompatible with completive ones. In this regard they come out as atelic and pattern with Slavic imperfectives. This is illustrated in (36)-(38) below 4 . (36) Po-chodził po parku pięć minut / *w pięć minut. po Q -walked.3SG over park five minutes / in five minutes 4 Same results have been reported for Russian in Borik (2002), e.g., (1) Petja (po)iskal knigu polčasa/*za polčasa Peter (PF-)look.for-pst book-acc half-hour/*in half-hour ‗Peter looked for a book for half an hour/*in half an hour.‘ 222 ‗He did some walking in the park for 5 minutes / in five minutes.‘ (37) Po-kopał piłkę godzinę / *godzinę. po Q -kicked.3SG ball.ACC hour / in hour ‗He kicked/was kicking the ball for an hour / in an hour.‘ (38) Po-czytał gazety godzinę / *w godzinę. po Q -read.PST.3SG newspapers.ACC hour / in hour ‗I read newspapers a little for an hour/in an hour.‘ The same generalization extends to cases where po Q - is stacked, i.e., when it attaches to an already prefixed, secondary imperfective form, as in (39) and (40). (39) Po-pod-rzucał piłkę godzinę / *godzinę. po Q -up-threw.3SG ball.ACC hour / in hour ‗He threw the ball up for an hour / in an hour.‘ (40) Po-ob-gadywali ją godzinę / *w godzinę. po Q -around-talked.3PL her.ACC hour / in hour ‗They gossiped about her for an hour / in an hour.‘ However, in the contexts where po Q - applies to the internal argument, and, consequently, that argument is expressed in the genitive case, both durative as well as completive adverbials are impossible. This is exemplified in (41) and (42). (41) Po-czytał gazet *godzinę / *w godzinę. po Q -read.3SG newspapers.GEN hour / in hour ‗I read newspapers a little for an hour / in an hour.‘ 223 (42) Po-szyła sukienek *5 godzin / */???w 5 godzin. po Q -sewed.3SG dresses.GEN 5 hours / in 5 hours ‗I read newspapers a little for an hour/in an hour.‘ These last two examples show the pattern identical to the one just discussed with na Q - prefixed verbs. This, in turn might lead to either of the following conclusions: (i) all po Q - prefixed verbs are all atelic, and genitive-marked objects are simply incompatible in Slavic with either durative or completive adverbials, or (ii) po Q -prefixed verbs are atelic when po Q - quantifies over the event, but telic when it quantifies over the direct object; in the latter case, the genitive-marked object makes the modification by a completive adverbial impossible just as it is the case with na Q -. The first conclusion preserves a common Aktionsart function for po Q -, while the second one is able to link the incompatibility of durative adverbials with telicity and the incompatibility of completive ones with genitive marking on the object. Deciding on either of them as correct at this point would be speculative, and so more tests and data need to be considered, which is done in section 2.4.2 below. Lastly, there is the yet another pattern with respect to the adverbial modification possibilities, obtained when po Q - attaches to degree achievement predicates. As observed in Dowty (1979) and further discussed in Abusch (1986) and Hay, Kennedy & Levin (1999), among others, in English such predicates are often ambiguous between telic and atelic readings. This is illustrated in (43) and (44) with the adverbial modification test 5 (cf. Dowty 1979 and Hay, Kennedy & Levin 1999). 5 Other tests show the Aktionsart interpretation of degree achievements may depend on the real-world knowledge concerning their object, e.g., 224 (43) The soup cooled for an hour / in an hour. (44) The crew lengthened the rope for an hour / in an hour. Hay, Kennedy & Levin (1999) relate the availability of telic and atelic readings with DAs to the scalar structure underlying the relevant predicates, where a scale is a set of points ordered along some dimension. When the degree of change in the property associated with the predicate is seen as unbounded on the scale, the interpretation is atelic; when it is seen as bounded, the interpretation is telic. In the latter case, the bound on the scale can be specified overtly by a modifier (e.g., The soup cooled 5 degrees; The tailor lengthened the pants 2 inches), but can also come from the lexical semantics of the underlying adjective, e.g., to lengthen, to shorten are unbounded (or open-range), but to straighten and to empty are bounded (or closed-range) because their scale has a maximal value. It is interesting to note that when degree achievements are expressed in Polish in the simplex imperfective form, those forms can only have the open-scale interpretation, which goes in hand with the general atelic interpretation of imperfectives. That includes predicates which, by Hay, Kennedy & Levin‘s (1999) classification, based on their unambiguously telic readings in English obtained without any ‗bounding‘ modifiers and their compatibility with adverbials such as completely, are closed-range, e.g., prostować ‗to straighten‘ (interpreted only as ‗to make straighter‘), or ciemnieć ‗to get dark‘ (i) a. The tailor almost lengthened my pants. (telic interpretation, sentence ambiguous between a reading when the event occurred, but was not completed, and one where the event has not occurred) b. The teacher almost lengthened the exam. (atelic interpretation, only one reading possible – the event has not occurred) or may be different for different types of DAs based on the semantics of the underlying adjectives, as shown in (ii) (ii) a. Kim is lengthening the role => Kim has lengthened the rope. b. Kim is straightening the rope ≠> Kim has straightened the rope. (all exs. from Hay, Kennedy & Levin 1999, cf. also Dowty 1979): 225 (interpreted only as ‗to get darker‘). In order to obtain a closed-range, bounded interpretation, one has to add a telicity-marking prefix, e.g., wy-prostować ‗to straighten,‘ z-ciemnieć ‗to get dark.‘ The same contrast between simplex imperfectives and prefixed forms with regard to open-scale and closed-scale readings respectively, obtains for predicates with underlying open-range adjectives, e.g., drożeć ‗to get more expensive‘ vs. z-drożeć ‗to get expensive,‘ starzeć się ‗get older-looking (with refl)‘ vs. ze-starzeć się ‗to get old-looking (with refl),‘ or młodnieć ‗to get younder-looking‘ vs. wy-młodnieć or od-młodnieć ‗to get young-looking.‘ Incidentally, these examples confirm the analysis of the Slavic verbal prefixes as telicity-markers, because if marking temporally bounded readings (i.e., the PERFECTIVE viewpoint) were their role, it could not guarantee a telic interpretation for degree achievements as evidenced by the fact that the following is not a contradiction: (45) The soup cooled for an hour (= temporally bounded reading), but it didn‘t cool completely. Returning to the question of po Q - and degree achievement verbs, as reported in Jabłońska (2004), po Q - can attach to a limited number of such predicates, in which case we get a bounded, telic reading which is mid-scale. In other words, the interpretation is that of some noticeable movement on the scale, but without any indication of the existence, or reaching thereof, of an upper/lower limit on the scale, in contrast to the forms prefixed with a telicizing prefix. This is illustrated in 0 (examples adopted from Jabłońska 2004). 226 (46) Simplex imperfectives po Q -prefixed perfectives perfectives with a telicizing prefix a. ciem-ni-e-ć dark-ADJ-V-INF ―to be getting dark‖ po-ciem-ni-e-ć po Q -dark-ADJ-V-INF ―to get darker‖ z-ciem-ni-e-ć [TELIC]- dark-ADJ-V-INF ―to get dark‖ b. droż-e-ć expensive-V-INF ―to be getting expensive‖ po-droż-e-ć po Q - expensive-V-INF ―to get more expensive‖ z-droż-e-ć [TELIC]-expensive-V-INF ―to get expensive‖ c. starz-e-ć się old-V-INF REFL ―to be getting old-looking‖ po-starz-e-ć się po Q -old-V-INF REFL ―to get older-looking‖ ze-starz-e-ć się [TELIC]-old-V-INF REFL ―to get old-looking‖ Such interpretation of po Q - with respect to degree achievements is consistent with its general semantics as a quantifier denoting ‗some, a little‘: with degree achievements that quantifier measures over the scale supplied by a DA predicate, i.e., the degree of change scale. However, while with other uses of po Q - as a quantifier over event (i.e., as a quantifier measuring over any other kind of scale supplied by the predicate), we get an atelic interpretation, the interpretation of po Q -prefixed degree achievements telic. This is evidenced by their behavior in the adverbial modification test, which is different from the previously illustrated uses of po Q - as an event-level quantifier. This is illustrated below, including parallel examples where the prefix is straightforwardly telicizing and encoding closed-scale readings, as well as where there is no prefix, and, unlike in English, completive adverbials are not possible. 227 (47) a. Mleko po-drożało *3 dni / w 3 dni. milk.NOM po Q -got-expensive.3SG (for) 3 days / in 3 days ‗Milk got more expensive *for 3 days / in 3 days.‘ b. Mleko z-drożało *3 dni / w 3 dni. milk.NOM PREF-got-expensive.3SG (for) 3 days / in 3 days ―Milk got expensive *for 3 days / in 3 days.‖ Cf. c. Mleko drożało 3 dni / *w 3 dni. milk.NOM got-expensive.3SG (for) 3 days / in 3 days ―Milk was getting more expensive for 3 days.‖ (48) a. Prezydent bardzo się po-starzał *rok / w rok. president much REFL po Q -got-old-looking.3SG (for) a year / in a year ―The president got old-looking a lot *for a year / in a year.‖ b. Prezydent ???/*(bardzo) się ze-starzał *rok / w rok. president (much) REFL PREF-got-old-looking.3SG (for) a year/in a year ―The president got old-looking *for a year / in a year.‖ Cf. c. Prezydent *(bardzo) starzał się rok / *w rok. president (much) got-old-looking.3SG REFL a year / in a year ―The president was getting old-looking for a year.‖ 228 Jabłońska (2004) suggests that the meaning of po Q - in the cases above is comparable to the adverbials like significantly or visibly, 6 which, as reported in Hay, Kennedy and Levin (1999), give rise to telic readings in English. However, given the general semantics of po Q -, it would be better translated as ‗somewhat.‘ 7 The question is now whether the observed telic reading could be attributed to the combination of the adverbial/quantificational meaning of po Q - and the semantics of the degree achievement predicate. We can test this using an unprefixed degree achievement and an adverbial./quantificational free standing equivalent of po Q -, as in (49) and (50). (49) Niebo ciemniało znacznie / widocznie / trochę godzinę / *w godznię. sky.NOM got-darker.3SG significantly / visibly / some (for) an hour / in an hour ‗The sky was getting siginificantly / visibly / somewhat darker for an hour.‘ (50) Mleko drożało znacznie / widocznie / trochę 3 dni / *w 3 dni. milk.NOM got-more-expensive.3SG significantly/visibly/somewhat (for) 3 days/in 3 days ‗Milk was getting siginificantly / visibly / somewhat more expensive for 3 days.‘ As the examples above show, the presence of an adverbial/quantifier is not sufficient to derive telic readings of degree achievements in Polish. This fact is consistent with the way telic-atelic readings are marked in Slavic: they are not derived through the properties of objects or modifiers, including nominal and adverbial quantifiers, but are rather morphologically marked on the verb. However, there is another possibility that has to be considered, namely a situation where po Q - quantifies over a scale of times, and gives us a reading ‗move on a scale of change for some time.‘ The telic reading one obtains would 6 This seems to me different from the concurrently made claim that the role of po Q - in the relevant cases is to ―introduce an arbitrary Reference Time that seals off the event,‖ (Jabłońska 2004:5) which is would suggest that po Q - is a viewpoint marker. 7 This suggestion was made to be by Hagit Borer. 229 then be epiphenomenal, and not due to Aktionsart properties of po Q -. At first, this analysis seems supported by the following implications: (51) Niebo ciemniało przez jakiś czas. → Niebo po-ciemniało. sky.NOM got-darker.3SG for some time sky.NOM po Q -got-darker.3SG ‗The sky was getting darker for some time.‘ → ‗The sky got darker.‘ (52) Mleko drożało przez jakiś czas. → Mleko po-drożało milk.NOM got-more-expensive.3 SG for some time milk.NOM poQ-got-more-expensive.3 SG ‗Milk was getting more expensive for some time.‘ → ‗Milk got more expensive.‘ However, it is also possible to get a telic, achievement-like interpretation of po Q -prefixed degree achievements when po Q - clearly does not measure over a time scale, e.g., (53) a. On nagle po-smutniał. he suddenly po Q -got-sadder.3SG ‗He suddenly got sadder.‘ b. Niebo nagle po-ciemniało.. sky.NOM suddenly got-darker ―The sky got suddently darker.‖ Consequently, we are forced to conclude that po Q - as a prefix of degree achievements signifying some perceptible change, is a telicity marker. The telicity-marking role, therefore, now characterizes all Q-prefixes other than the uses of po Q - outside of degree achievements, and even there that was some uncertainty with regard to cases with genitive-marked objects. In the following section, I look at other tests for telicity in an attempt to validate the conclusions drawn so far, as well as to clarify the Aktionsart interpretation of po Q -prefixed verbs when they quantify over the direct object. 230 2.4. Other telicity tests and na Q - and po Q - with scope over the object With the adverbial modification tests being inconclusive with respect to na Q - and po Q - when they apply to the internal argument and that argument is consequently genitive-marked, it is useful to turn to other tests for telicity. Such tests were discussed in Chapter 2, and here I will employ the following (some illustrative examples are provided for easy reference): (54) Conjunction with temporal modification  An atelic predicate allows for single event or two event interpretations with relevant temporal modification.  A telic predicate allows only for a two event interpretation with relevant temporal modification. c. John read books on Monday and Tuesday. (ATELIC: one or two events) d. John read a book on Monday and Tuesday. (TELIC: two events) (55) Cumulativity & Divisiveness (Borer 2005) i. A predicate P is homogenous iff P is cumulative and divisive. ii. P is cumulative iff ∀x [P (x) ∧ P (y) → P (x ∪ y)]. ―A predicate is cumulative iff whenever it holds of two arguments, it holds also of their union.‖ iii. P is divisive iff ∀x [P (x) → ∃y (P (y) ∧ y < x)] ∧∀x,y [P (x) ∧ P (y) ∧ y < x → P (x−y)]. ―A predicate is divisive iff the following holds: For any argument, if P is true of that argument, it is also true of a part of that argument, and in addition, for 231 any argument and a part of it that P holds true of, P is also true of the difference between the argument and its part.‖ = Telic predicates are those which are either not cumulative or not divisive. Atelic predicates need to be both cumulative and divisive. a. cumulativity ‗John read a book‘ + ‗John read a book‘ ≠ ‗John read a book‘ (instead, = ‗John read two books‘). ‗John read books‘ + ‗John read books‘ = ‗John read books‘ b. divisiveness A part of ‗John read a book‘ ≠ ‗John read a book‘ (instead, = ‗John read a book part‘) A part of ‗John read books‘ = ‗John read books‘ = ‗John read a book‘ is TELIC; ‗John read books‘ is ATELIC 2.4.1. Other telicity tests and na Q -prefixed verbs The results of the tests above with respect to na Q -prefixed verbs are shown below. I included both examples with the reflexive, which I earlier concluded were telic, and examples where na Q - is interpreted with respect to the object, which is the ambiguous kind. (56) Conjunction: 1 or 2 events a. Na-śpiewał się w poniedziałek i wtorek. = 2 events (TELIC) na Q -sang.3SG REFL in Monday and Tuesday ‗He sang a lot/enough on Monday and Tuesday.‘ 232 b. Lekarz na-wy-pisywał recept w poniedziałek i wtorek. = 2 events, ???1 event Doctor na Q -prescribed.3SG prescriptions.GEN in Monday and Tuesday ‗A/the doctor wrote a lot of prescriptions for on Monday and Tuesday.‘ (57) Cumulativity a. Na-śpiewał się + Na-śpiewał się = Na-śpiewał się (CUMULATIVE) na Q -sang.3SG REFL ‗He sang a lot/enough.‘ b. Lekarz na-wy-pisywał recept + Lekarz n-awypisywał recept = Lekarz na- wypisywał recept Doctor na Q -prescribed.3SG prescriptions.GEN (CUMULATIVE) ‗A/the doctor wrote a lot of prescriptions.‘ (58) Divisiveness a. A part of ‗Na-śpiewał się‘ ≠ Na-śpiewał się (NOT DIVISIVE) na Q -sang.3SG REFL ‗He sang a lot/enough.‘ b. Lekarz na-wy-pisywał recept ≠ Lekarz na-wy-pisywał recept Doctor na Q -prescribed.3SG prescriptions.GEN (NOT DIVISIVE) ‗A/the doctor wrote a lot of prescriptions.‘ It needs to be noted that with respect to the divisiveness test, the vague meaning of the quantifier na Q -as well as its use to convey a satisfactory amount make it sometimes hard to conceptualize what a part of an event described by a na Q -prefixed predicate would mean. But even with this caveat, na Q -prefixed verbs are clearly non-divisive, just as it would be the case with a quantifier signifying a large number. Consequently, they are 233 non-homogenous, and, therefore, telic. This is also confirmed by their performance in the conjunction test. 2.4.2. Other telicity tests and po Q -prefixed verbs In the examples below, the same tests are applied to po Q -prefixed verb. Here again I include those where po Q - quantifies over the event, which came out as atelic in the adverbial modification test, as well as those where po Q - applies to the direct object, which were problematic. (59) Conjunction: 1 or 2 events a. Po-chodziła po lesie w poniedziałek i wtorek.= 1 or 2 events (ATELIC) po Q -walked.3SG on forest in Monday and Tuesday ‗She walked in the forest a bit on Monday and Tuesday.‘ b. Po-szyła sukienek w poniedziałek i wtorek.= 1 or 2 events (ATELIC) po Q -sewed.3SG dresses.GEN in Monday and Tuesday ‗She sewed some dresses on Monday and Tuesday.‘ (60) Cumulativity a. Po-chodziła po lesie + Po-chodziła po lesie = Po-chodziła po lesie (CUM) po Q -walked.3SG on forest ‗She walked in the forest a bit.‘ b. Po-szyła sukienek + Po-szyła sukienek = Po-szyła sukienek (CUM) po Q -sewed.3SG dresses.GEN ‗She sewed some dresses.‘ 234 (61) Divisiveness a. A part of ‗Pochodziła po lesie‘ = Pochodziła po lesie (DIV) po Q -walked.3SG on forest ‗She walked in the forest a bit.‘ b. A part of ‗Poszyła sukienek‘ = Po-szyła sukienek (DIV) po Q -sewed.3SG dresses.GEN ‗She sewed some dresses.‘ The results above indicate that po Q -prefixed verbs where po Q - functions as an intensifier are atelic, just as po Q -prefixed verbs where po Q - is a variable quantifier. This in turn suggests that the previously noted incompatibility of completive and durative adverbials with na Q -and po Q -prefixed verbs when they function as intensifiers and the object is genitive-marked, is not reflective of the Aktionsart interpretation of the predicate, nor does it show a problem with obtaining unambiguously telic interpretation when the object is genitive-marked. Instead, it must be something about genitive-marked objects without an overt genitive-case assigner that somehow inhibits the application of temporal interval modifiers. What it is exactly will have to remain a topic for further research. 2.5. Summary The application of various telicity tests to Q-prefixed verbs has revealed a mixed picture in terms of inner aspect properties of individual prefixes. What can be concluded based on the obtained results as well as some additional considerations concerning the impact of genitive-marked objects, is that po D - and na Q -prefixed forms are telic, while po Q -prefixed verbs are atelic, with the exception of po Q - combining with degree 235 achievements, where po Q - gives rise to telic readings. 8 The particular and contrasting Aktionsart interpretation for na Q - and po Q -prefixed forms is clearly related to their quantificational meaning, with na Q - ‗a lot, enough‘ being non-divisive, and hence non- homogenous, and with po Q - ‗some‘ (ignoring for the moment the ‗a little‘ reading) being divisive and cumulative, and hence homogenous. However, the quantificational meaning by itself could not account for the telic interpretation of na Q -, po D - and po Q - with DAs, because telicity in Slavic is not derived through the properties of objects or modifiers, including adverbials and quantifiers – free standing adverbial and nominal quantifiers with meanings identical or comparable to Q-prefixes would not yield telic readings. Therefore, the telic nature of po D - and na Q -prefixed verbs, and of po Q -prefixed DAs, has to be attributed to the telicity-marking function of the relevant Q-prefixes, which unifies them with non-quantificational prefixes in Slavic. 3. Q-prefixes and viewpoint aspect In the preceding sections we saw that verbs with quantificational prefixes are not a uniform group with respect to their inner aspect properties. Here I consider their viewpoint interpretation. Based on the previous discussion, we should expect Q-prefixes that give rise to telic readings to allow only PERFECTIVE interpretation, which is due to the proposed restriction allowing telic predicates to compose only with the PERFECTIVE viewpoint, discussed in Chapter 2 and 3. Less certain is the status of po Q -prefixed, non- degree achievement verbs, which are atelic. I examine the viewpoint interpretation of all Q-prefixed verbs below, using the sequential (=bounded, PERFECTIVE) vs. simultaneous (=unbounded, IMPERFECTIVE) interpretation of two coordinated predicates, and the 8 Similar conclusions have also been reached for Bulgarian with respect to po D - and po Q - in non-degree achievements uses in Istratkova (2006). 236 contrast between inclusive (=bounded, PERFECTIVE) and durative (=unbounded, IMPERFECTIVE) readings of a predicate with respect to a time interval. 9 These two kinds of test can be merged into one by coordinating two predicates and evaluating them with respect to a time interval, which is what I do below. The coordinated predicates in all examples are simplex imperfectives, which are – as the reader may recall – ambiguous for viewpoint aspect, and so cannot by themselves force the predicted sequential interpretation of Q-prefixed verbs. (62) Sequential vs. simultaneous interpretation: po Q - Q: What did you do in the afternoon? a. Po-spacerowałam po parku i słuchałam ptaków. (sequential) po Q -walked.1SG over park and listened.1SG.IMPF birds ‗I walked in the park a little and listened to birds.‘ b. Po-przy-mierzał ubrań i oglądał się w lustrze. (sequential) po Q -tried-on.3SG clothes.GEN and watched.3SG.IMPF REFL in mirror ‗He tried on some clothes and looked at himself in the mirror.‘ (63) Sequential vs. simultaneous interpretation: po D - Q: What did you do in the afternoon? 9 In English, atelic verbs in the simple past form, which are understood to be PERFECTIVE, allow both sequential and simultaneous readings for coordinated predicated, as well as both inclusive and durative readings with respect to a time interval. This contrasts with the exclusively sequential readings exhibited by Slavic perfectives, including the atelic po Q -perfectives. The reason for the difference may lay in the fact that the English PERFECTIVE viewpoint is not grammatically marked, but rather a default. It is the IMPERFECTIVE (or rather the more narrow progressive) viewpoint that is marked, and forms morphologically marked as IMPERFECTIVE/progressive can only have simultaneous/durative interpretation. Note that in Slavic it is the IMPERFECTIVE viewpoint that is morphologically unmarked, and so forms unmarked for viewpoint aspect (i.e., Slavic imperfectives) may receive the default IMPERFECTIVE interpretation but do not have to, resulting in the possibility of both sequential and simultaneous, or inclusive and durative readings for the relevant forms. 237 a. Po-chrzcił dzieci i kropił wodą święconą. (sequential) po D -baptized.3SG children and sprinkled.3SG.IMPF water sanctified ‗He baptized (all) the children and sprinkled holy water.‘ b. Po-przy-mierzał ubrania i oglądał się w lustrze. (sequential) po D -tried-on.3SG clothes.ACC and watched.3SG.IMPF REFL in mirror ‗He tried on (all) the clothes and looked at himself in the mirror.‘ (64) Sequential vs. simultaneous interpretation: na Q - Q: What did you do in the afternoon? a. Na-gotowałam pierogów i słuchałam radia. (sequential) na Q -cooked.1SG dumplings and listened.1SG.IMPF radio ‗I cooked a lot of dumplings and listened to the radio.‘ b. Na-przy-mierzał się ubrań i oglądał się w lustrze. (sequential) na Q -tried-on.3SG REFL clothes.GEN and watched.3SG.IMPF REFL in mirror ‗He tried on a lot of the clothes/he tried on clothes till he had enough, and looked at himself in the mirror.‘ In all the examples above, Q-prefixed forms have decidedly sequential interpretation, which means that they are necessarily inclusive with respect to the time interval. The evidence from this test therefore confirms the prediction that telic Q-prefixed verbs are PERFECTIVE. However, it also tells us that po Q -prefixed verbs are PERFECTIVE even though they are atelic. This has consequences for how we view the Slavic tests for (im)perfectivity, which group po Q -prefixed verbs with perfectives which are telic. Such tests, with the exception of the nominalization tests discussed in Chapter 2, where Q- prefixes would not participate, have to be now understood as sensitive to viewpoint 238 aspect categories. The upside of this result is that the careful accounts of the patterning of perfectives with respect to such tests in terms of viewpoint categories and temporal coincidence (e.g., Borik 2002, Pereltsvaig 2005, forthcoming, Mezhevich 2008) can retain their explanatory power. 4. Quantificational prefixes: previous classifications and accounts The generalizations concerning the special nature of Q-prefixed verbs and their aspectual properties uncovered in the sections above do not necessarily agree with claims previously made in the literature. The differences, which concern both the general approach to Q-prefixes as well as their aspectual role, and their ability to mark telicity in particular, are discussed below. 4.1. Q-prefixes: not special One type of approach to Q-prefixes, characterizing, for example, much of Filip‘s work on Czech (e.g., 1996, 1999, 2004), is to use examples of Q-prefixed verbs to question the growing consensus that Slavic prefixes are markers of telicity. Filip highlights the quantificational function of here discussed prefixes, arguing that, as measure expressions, they could not be analyzes as markers of aspect. Filip does not consider in the end whether the interpretation of Q-prefixed verbs is in fact telic or atelic, likely taking that to be irrelevant once the special character of the said prefixes under has been shown, and she does not entertain the possibility that some prefixes can have more than one grammatical function. A slightly different, though parallel, approach is taken in Borik (2002), who uses po Q -prefixed verbs with atelic interpretation in Russian (in addition to the misunderstood perdurative pere-prefixed forms, discussed in Chapter 3) as evidence against telicity-marking role of Russian prefixes. Here, while the Aktionsart- 239 related generalizations are correct, the lack of recognition of the special status of Q- prefixes leads to unfortunate conclusions for the rest of verbal prefixes in Russian. 4.2. Lexical versus superlexical prefixes Another approach to Q-prefixes assumes a two-way classification of Slavic prefixes into lexical and superlexical ones, and groups quantificational prefixes together with other ones that have adverbial-like uses (e.g., Babko-Malaya 1999, Svenonius 2004) 10 . In most general terms, it is argued that lexical prefixes are those that syntactically attach low and have irregular meanings, while superlexical prefixes attach high and have transparent meanings. It needs to be stressed before we get into the discussion of the specifics of two such proposals, that the distinction between lexical and superlexical prefixes does not uniquely assign each and every prefix to one or the other category. Rather, the distinction is made between two kinds of uses for a verbal prefix, with some prefixes being able to functioning as either lexical or superlexical depending on the context (including the context of the verb stem to which they attach). 4.2.1. Babko-Malaya (1999) Babko-Malaya‘s (1999) distinction between lexical and superlexical prefixes is based pre-theoretically on the perceived (ir)regularity of meaning as well as the type of meanings prefixes convey. Lexical prefixes are said not to have stable meanings, while superlexical prefixes are understood to have regular meanings and ―correspond to aspectual words or adverbial phrases in English and other languages‖ (Babko-Malaya 1999:76). The superlexical prefixes listed by Babko-Malaya for Russian as representative examples are: za- ‗to begin,‘ ot-, do- ‗to finish,‘ po- ‗to do for a while,‘ and pro- ‗to do 10 A third type is also distinguished in both Babko-Malaya (1999) and Svenonius (2004), namely the pure perfectivizer type (i.e., what was referred earlier in this paper as ―empty‖ prefixes), but it can be subsumed under the ―lexical prefix‖ category with regard to the distinction discussed in this chapter. 240 for a long time.‘ Based on this semantic criterion, it is proposed that lexical prefixes are adjoined to a lexical head – i.e., prefixation is pre-syntactical, while superlexical prefixes are adjoined to a functional category. Babko-Malaya (1999) lists a number of facts in support of the proposed distinction: lexical prefixes require an internal argument, while superlexical prefixes do not; verbs with superlexical prefixes cannot form participial passive or undergo secondary imperfectivization (Schoorlemmer 1995); verbs that assign instrumental case cannot take lexical prefixes. She goes on to propose that only lexical prefixes can form accomplishments, as they ―characterize the result or a terminal point of the action denoted by a verbal root," while superlexical prefixes ―do not specify a result or a final state of the action, but rather focus attention on a particular stage of the action.‖ (Babko- Malaya 1999: 79-80). As evidence for that claim, two examples are provided where verbs prefixed with the inceptive za- are shown to be incompatible with a completive adverbial that presupposes a durative event (i.e., ‗in X time‘). There are a number of assertions in Babko-Malaya (1999) that could be disputed. First, while the adverbial-like uses of prefixes are regular, semantic regularity can also be found in the prepositional-like/spatial uses of prefixes. It is this regularity that guarantees productivity of Slavic prefixation – both with regard to adverbial as well as prepositional- like uses. For the semantically transparent uses of prepositional-like prefixes, there is no need to assume a different, sub-X 0 point of insertion, especially that they have the same aspectual function as many of the adverbial-like prefixes (including the above listed za-, ot- and do-). 241 Second issue is the achievement versus accomplishment distinction. It appears that using verbs with the inceptive za- prefix as representative examples in this regard can be misleading. Inceptive za-, due to its semantics, will likely yield achievements rather than accomplishments, as starts are typically conceptualized as punctual. However, even here one can find counterexamples when the starting process can be understood to have some duration, e.g., (65) Za-palił papierosa w kilka sekund. INCPT-burn cigarette in few seconds ―He lit a cigarette in a few seconds.‖ The same verb can appear without an overt object, and can undergo secondary imperfectivization, contradicting further the generalizations proposed in Babko-Malaya for superlexical prefixes. Also, the reader may recall that the distributive prefix po-, which by Babko-Malaya‘s criteria is clearly superlexical, is easily compatible with completive adverbials, and therefore yields accomplishments and not achievements. Lastly, with respect to argument structure, the quantificational (and therefore superlexical) prefix na Q - does select for an internal argument, which is due to the fact that it typically needs to quantify over an object. When that requirement is satisfied through the reflexive, that typically involves changing the argument structure of the underlying verb, which is again not something unexpected under Babko-Malaya‘s account. While the observation that Slavic verbal prefixes can have prepositional-like and adverbial-like uses and that prepositional-like uses are more likely to be lexicalized is not unimportant, from the point of view of aspectual composition it is not relevant. Neither (a)telicity nor (IM)PERFECTIVITY distinguish the superlexical prefixes as a class from 242 lexical ones. Instead, with the exception of the quantificational po Q -, superlexical prefixes mark telicity just as the lexical prefixes do, and all prefixed verbs are interpreted as PERFECTIVE. Therefore, the distinction between lexical and superlexical prefixes as presented in Babko-Malaya (1999) must be understood as semantic/lexical and not related to aspect as a functional category. 4.2.2. Svenonius (2004) Svenonius (2004) builds on the distinction made in Babko-Malaya (1999). Lexical prefixes in Slavic are compared there to Germanic verbal particles, and are said to have the following characteristics, among others: (i) have core spatial meanings, (ii) readily produce idiomatic combinations, (iii) attach closest to the V stem (in case there is more than one prefix), and (iv) allow the formation of SIs. Syntactically, it is proposed that lexical prefixes originate either as the head of the Result Phrase (with the result head being part of the lexical entry of resultative verbs, Ramchand 2004), or as the complement of Result head (Svenonius 2004a). The superlexical prefixes, on the other hand, are those which have functional uses. A list of such prefixes for selected Slavic languages in Svenonius (2004a) and reproduced in Table 9 below, with Q-prefixes discussed in this work shown in bold. 243 Table 9: Superlexical prefixes (Svenonius 2004) (R=Russian, P=Polish, Cz=Czech, SC=Serbo-Croat, B=Bulgarian) Function R P Cz SC B INCEPTIVE za za za za TERMINATIVE ot od do COMPLETIVE do do iz iz PERDURATIVE pro prze pro DELIMITATIVE po po po po ATTENUATIVE po pod po po DISTRIBUTIVE po po po po po CUMULATIVE na na na na na SATURATIVE na na na REPETITIVE pere prze pre pre EXCESSIVE pere prze raz Here, again, the distinction seems to be predicated on preposition-like vs. adverbial-like meanings. In terms of independent evidence, what is submitted, following Milićević (2004), is that when verbs show perfective-imperfective alternation without a prefix 11 (e.g., Russian končatj.IMPF - končitj.PERF ‗to end,‘ brosatj.IMPF - brositj.PERF ‗to throw‘ (Townsend 1975); Serbian bacati.IMPF - baciti.PERF ‗to throw,‘ skakati.IMPF - skočiti.PERF ‗to jump,‘ udarati.IMPF - udariti.PERF ‗to hit‘ (Milićević 2004)), ―it is usually the perfective stem that appears with lexical prefixes, and the imperfective stem that 11 These are the few cases discussed in Chapters 2 and 3, where a simplex form is telic and has the distribution of a perfective. 244 appears with superlexical prefixes.‖ (p. 195-196). This is illustrated for Serbian in (66) (from Milićević 2004). (66) a. baciti ~ iz-baciti throw P out.throw P ‗throw‘ — ‗throw out‘ b. bacati ~ iz-bacati throw I CMPL-throw P ‗throw‘ — ‗throw extensively‘ The operative word here appears to be ‗usually,‘ as in Polish, for example, the superlexical prefixes listed in (66) can attach to simplex perfectives. This is shown in (67). (67) kupić.perf : kupować.impf ‗buy‘ a. prze-kupić / prze-kupować through/across.buy ‗to bribe‘ b. za-kupić / ?za-kupować 12 INCPT-buy ‗to buy, to purchase‘ c. na-kupić / na-kupować CUMUL.buy ‗to buy a lot‘ 12 This form returns a substantial number of hits in Google search; I find it highly degraded, possibly due to my reanalysis of za- as a pure telicizer in za-kupić. 245 As I argued in Chapter 3, simplex perfective forms like those above actually involve a null prefix, and so the bare stem is not actually perfective/telic – it is simply that it does not exist in the encyclopedia without a prefix (or alternatively un unprefixed form of such verbs are not morphologically well formed). Consequently, the relevant forms should not impose any special attachment requirements on the prefix pool. Polish data in (67) for the verb buy indicates that is true. As for other verbs mentioned in Milićević (2004), the fact that they are not be compatible with superlexical prefixes may be due to the fact that those forms have non-durative, semelfactive like meanings (e.g., throw, jump, hit) and therefore cannot be modified by adverbial prefixes that presuppose either a durative, non- punctual event, or a multiplicity of events (recall that primary imperfectives with secondary imperfective morphology have repetitive/habitual interpretation, which is exactly the interpretation of the imperfective variants of simplex perfective verbs skakać.impf ‗jump‘, rzucić.impf ‗throw,‘ and uderzyć.impf ‗hit‘). 13 Consequently, the occasional problem with attachment of superlexical prefixes to primary perfectives (which are quite limited as a group anyway) cannot be seen as a characteristic that defines superlexical prefixes and indicates their different grammatical status. The second piece of evidence proposed in support of distinguishing superlexical prefixes from lexical ones is the fact that verb forms with superlexical prefixes generally do not form secondary imperfectives (Svenonius 2004). Here again, however, the 13 The fact that the restriction in question is semantic rather than grammatical/syntactic is also evidenced by the fact that superlexical prefixes such as po- and na- do not attach to directed motion verbs, which denote a single event, and instead require a non-directed motion verb as their stem, e g., biec (directed motion) ~ biegać (non-directed motion)  *na-biec się/na-biegać się ‗to run a lot/enough;‘ *po-biec/po-biegać ‗to run a little.‘ Both directed and non-directed motion verbs are imperfective, but only the latter one allows multiple event interpretation, which is what po- and na- appear to seek in such verbs. 246 generalization is qualified, and it is indicated that exceptions to this rule are quite numerous. Finally, Svenonius (2004) brings up the stacking ability of certain prefixes as evidence for a two-way classification: lexical prefixes are merged lower (below the IMPERFECTIVE position filled by the secondary imperfective suffix), while superlexical prefixes are merged higher (above the IMPERFECTIVE position) and can consequently stack. While I believe the stacking ability is a valid diagnostics for classifying Slavic prefixes and identifying their merger position, that ability does not characterize all the prefixes in Table 9. While in Bulgarian all of them are reported to be stackable, other languages are not as generous. In fact, and as I mentioned earlier, Polish allows only two stackable prefixes: na- and po- (the latter one in its different quantificational instantiations). The same is true of Russian. If all superlexical prefixes are merged higher and in a different position than the lexical ones, it is not clear why all the prefixes listed in Table 9 cannot be stacked in Polish, or Russian, or Czech, etc. The fact that they cannot, suggests instead that they share a merger/target landing site with lexical prefixes. Consequently, it is not clear whether the prefixes in Table 9 truly constitute a class across Slavic languages, or whether a more differential classification is needed 14 . It appears that the classification of prefixes into lexical and superlexical ones as proposed in Svenonius (2004) is again based on the difference between spatial and possibly idiosyncratic meanings on the one hand, and adverbial and necessarily systematic meanings on the other, and not consistently supported by syntactic or morphological evidence. While the differences in meaning could translate into different 14 Žaucer (2005) and Arsenijević (2006) also question the inclusion of some superlexical verbs in that category, e.g., inceptives or completives, though Žaucer (2006) also proposes that the reflexivized used of na- also ought not to be classified as superlexical/external, with which I disagree. 247 merger sites, the classification in question has two other unfortunate outcomes. One is that it fails to recognize a shared telicity-marking role of lexical and almost all superlexical prefixes. Secondly, it obscures the differences that exist within the superlexical prefix group, particularly as they pertain to the unique semantics, syntax and morphology of quantificational prefixes. The solution to these problems is either (i) a three-way classification with three merger positions plus a requirement that the lowest merged, lexical prefixes must move so as to ensure that they do not co-occur with non- stacking adverbial prefixes, or (ii) a re-aligned two-way distinction. A version of the three-way approach has been offered in Ramchand (2004), who proposes that lexical prefixes are merged in the complement of V (in the result Phrase), some superlexical prefixes are hosted by the (viewpoint) Asp projection, and finally the superlexical prefixes quantifying over DP arguments such as na Q - are merged higher than (viewpoint) AspP. This analysis clearly relies on the assumption that the highest-merged superlexical prefixes are not implicated in telicity marking, which, as we have seen earlier in this chapter, is not correct. As for the option in (ii), it can be found in DiSciullo & Slabakova (2005), as outlined below. 4.2.3. Internal versus external prefixes (DiSciullo & Slabakova) DiSciullo & Slabakova (2005) propose a two-way distinction, self described as similar to Babko-Malaya‘s (1999) own, and identify two levels of prefixation: internal and external. This time, the distinction is modeled on the Romance languages, where two types of verbal prefixes are proposed by DiSciullo (1997): ―an external prefix (temporal- sequential, iterative or inverse) [which] modifies the full verbal projection but does not affect its argument structure or internal aspect‖ and ―an internal prefix (spatial) [which] 248 affects the internal aspect of the verbal projection, and in some cases, its argument structure as well.‖ The properties of the two classes of prefixes in the extension of the proposal to Bulgarian are summarized in Table 10 below. Table 10: Internal vs. external prefixes (DiSciullo & Slabakova 2005) INTERNAL PREFIXES EXTERNAL PREFIXES i mark telicity neutral wrt telicity ii attach to the root  cannot be stacked, one such prefix per word attach to the root or to a prefixed stem  can be stacked iii produce specific readings of the object DP (and the subject) do not produce specific reading of the object DP (and the subject) iv can effect argument structure changes do not change argument structure v transparent and idiosyncratic meanings transparent, stable meanings only The discussion in Di Sciullo & Slabakova (2005) is limited to Bulgarian and the authors identify only three prefixes as external: po- (‗briefly‘), pre- (‗repeated action‘) and iz- (‗distributive reading‘). Istratkova (2006), as I mentioned earlier, lists eight stacking prefixes for Bulgarian, not counting the different uses of po-. Coming from Bulgarian, using stackability as a defining feature (hence the name ‗external‘) does not necessarily distinguish this approach from the lexical vs. superlexical distinction, and the authors imply that much. When applied to other Slavic languages, however, it narrows the set of external prefixes – properly, I believe – to those that can be shown syntactically and morphologically to be truly external, i.e., those that can stack. In Polish, specifically, external prefixes would be those that I have so far referred to as quantificational. 249 The other generalizations/assertions made in DiSciullo & Slabakova (2005) with respect to verbal prefixes concern their syntax and inner aspect role. External prefixes are said not to be implicated in inner aspect marking, unable to produce specific/definite readings of the direct object, which is otherwise a property of all perfectives, and unable to effect argument structure changes. The Polish data partially contradicts each of these claims. With respect to telicity, we saw that while Q-prefixes attach to imperfective bases only – which are uncontroversially atelic when in simplex form and here also proposed to be atelic when derived (i.e., as secondary imperfectives) – the resulting forms remain clearly atelic only in the case of the quantitative po Q -. The other external prefixes: na Q - and the distributive po D - yield telic readings. Istratkova (2006) indicates the same is true of Bulgarian, saying that some stacking prefixes imply the existence of an end point and/or result. As for the interpretation of the direct object, the assertion in Table 10-iii is undoubtedly true anytime the object is in the genitive case. When the direct object is accusative-marked, on the other hand, its interpretation is decidedly specific/definite (for one thing, the universal quantification underlying distributivity presupposes the existence of a definite set), contradicting the generalization in Table 10-iii. Lastly, with regard to argument structure, the generalization is largely true, but with one notable exception: that of the prefix na- when it introduces the reflexive. This very fact has led Žaucer (2006) to propose that this instance of na- should be analyzed as an internal/resultative prefix, clearly in an effort to keep the generalization to external prefixes are not argument- structure changing. However, this same na- has no only a semantically transparent, clearly quantificational role, but can also stack on top of other prefixes – both of the internal/lexical kind as well as the distributive po-. Consequently, its classification as 250 internal in the sense of Babko-Malaya (1999), Svenonius (2004) or DiSciullo & Slabakova (2005) faces more cons than pros. 4.3. Summary The recently proposed classifications of Slavic verbal prefixes into lexical vs. superlexical prefixes or internal vs. external ones, with na Q -, po Q -, and po D - falling into the latter categories, rely primarily on the distinction between prepositional-like, and often idiosyncratic uses versus adverbial-like, and typically transparent uses. However, the same analyses propose certain generalization concerning the grammatical role of lexical/internal vs. superlexical/external prefixes that turn out not to be entirely correct. For the purposes of this work, the most significant mischaracterization concerns the supposed Aktionsart inertness of external/superlexical prefixes. As I showed earlier on the example of Polish, within the group of quantificational prefixes, which seem to be most prototypical representatives of the external/superlexical category, only the prefix po Q - used in the context of non-DA predicates returns atelic readings. In all other situations, Q-prefixes yield telic predicates. In the section below I undertake to task of expressing these facts syntactically, while accounting also for the unique stacking ability of Q-prefixes. 5. Syntactic projection of Q-prefixes In the previous sections, I concluded that the following properties characterize Q- prefixed verbs: 251 (68) Q-prefixes and telicity & viewpoint aspect i. na Q -prefixed verbs  telic, PERFECTIVE ii. po D -prefixed verbs  telic, PERFECTIVE iii. po Q -prefixed, non-DA verbs  atelic, PERFECTIVE iv. po Q -prefixed DA verbs  telic, PERFECTIVE The task before us is to account for how Q-prefixes mark these properties through syntactic structure. The most straightforward case, it seems, is that of po Q -prefixed, non- degree achievement verbs, and I consider it first. All other Q-prefix uses will be considered later together, as they have the same aspectual properties. Before I get into that, however, a general comment is due with respect to how much new syntactic structure will be proposed. In some alternative accounts (e.g., Istratkova 2006), quantificational prefixes are awarded their own syntactic projections, and they become heads of such projections. This will not be the approach I will take. Instead, I will strive to accommodate Q-prefixes within independently motivated structures whenever possible, and minimize any structural additions. 5.1. po Q - with non-degree achievement verbs To summarize again facts concerning po Q -, it attaches to atelic bases – either simplex or secondary imperfectives 15 – and yields atelic predicates. In other words, it is Aktionsart neutral. In contrast to other atelic predicates, however, po Q -prefixed verbs have only PERFECTIVE interpretation. All these facts combined suggest that the merge site of po Q - is within ViewAsp, which is higher and external to Inner Aspect projections. The potential problem for such an analysis would be cases of verbs with multi-stacked 15 This generalization is true of Polish and Russian, but not of Bulgarian, where Q-prefixes can attach to perfectives. In Czech, it appears that they mostly attach to imperfectives, but there are some exceptions. 252 prefixes, where po Q - could be seen as merged prior to another, telicity-marking Q-prefix. I argued in the previous chapter that such an ordering could not be firmly established, and that the data involving po Q - as quantifier over the event actually argues against it. I will consequently assume that po Q - merges in Spec, ViewAsp, with po Q - having to agree in the [PERFECTIVE] feature with the null head of ViewAsp, and so essentially selecting for a PERFECTIVE-marked verb. The resulting, simplified syntactic representations of po Q - affixed to a B-imperfective and a secondary imperfective are given in (69) and (70) respectively. (69) Figure 15: Po Q - combining with B-imperfectives ViewAsp  po- [PERF] ViewAsp      [PERF] VP             (70) Figure 16: Po Q - combining with secondary imperfectives ViewAsp  po- [PERF] ViewAsp      [PERF] InnerAsp SI             PART-HOM InnerAsp             PREFIX VP       5.2. The syntax of other Q-prefixes (na Q -, po D - and po Q - with DAs) The other (uses of) quantificational prefixes are different from po Q - in non-degree achievement contexts in that they give rise to telic readings. Consequently, the relevant 253 prefixes need to merge in a projection responsible for marking Aktionsart distinctions – here in some InnerAsp projection. In the instances when Q-prefixes attach directly to simplex imperfective forms, that projection could be the InnerAspP, immediately dominating VP – the same one that otherwise hosts non-quantificational prefixes. This predicts, contrary to the claims made in the literature (e.g., Svenonius 2004), 16 that Q- prefixed verbs of this kind should be able to undergo secondary imperfectivization. This can be shown to be true for na Q -prefixed verbs, as illustrated below. (71) a. Mama na-gotowała nam pierogów. Mom.NOM na Q -cooked.3SG us.DAT dumplings.GEN ―Mom cooked us a lot of dumplings.‖ b. Mama po-na-gotow-yw-ała nam pierogów. Mom.NOM po D -na Q -cooked.2IMPF.3SG us.DAT dumplings.GEN ―Mom cooked each of us a lot of dumplings.‖ (72) a. Wszyscy się na-jedli. all.NOM REFL na Q -ate.3PL ―Everybody had enough to eat.‖ b. Wszyscy się po-na-jadali. all.NOM REFL po D -na Q -ate.3PL ―Each person had enough to eat.‖ Secondarily imperfectivized forms for po Q -prefixed degree achievements and verbs with distributive po D - are much harder to derive. For degree achievements, the problem lies 16 The claims are likely bases on the inability of Q-prefixed secondary imperfective forms (i.e., forms that already contain a different prefix) to be secondarily imperfectivized again, as well as the impossibility of coupling some Q-prefixed, as described in the previous chapter, which can be erroneously interpreted as prohibition of secondary imperfectivization of Q-prefixed forms in general. 254 mostly in the fact that unprefixed DAs are typically inchoatives, and the theme vowel marking inchoatives is not compatible with the secondary imperfective morphology (Jabłońska 2004). Outside of degree achievement inchoatives, po Q -prefixed degree achievements appear to allow secondary imperfectivization: (73) a. Po-starzyli ich na tych zdjeciach. po Q -made-look-older.3PL them.ACC on these pictures ―They made them look older in these pictures.‖ b. Po-po-starzali ich na tych zdjeciach. po D -po Q -made-look-older.2IMPF.3PL them.ACC on these pictures ―They made each of them look older in these pictures.‖ As for non-stacked po D -prefixed verbs, these are rather infrequent, due to the previously mentioned need to have unitisable, atomic events for the purpose of assigning distributive meaning. However, examples can be found, as shown in (74). (74) a. Po-głosowaliśmy na kandydatów. po D -voted.3PL on candidates.ACC 1. ―We each voted for the candidates.‖ 2. ―We voted for each candidate.‖ b. Po-po-głosow-yw-aliśmy sobie na kandydatów. po Q -po D -voted.2IMPF.3PL self.DAT on candidates.ACC 1. ―We each voted a little/did some voting for the candidates.‖ 2. ―We voted a little/did some voting for each candidate.‖ The data above supports the conclusion that telicity-marking Q-prefixes attach below the secondary imperfective deriving InnerAsp [PART-HOM] projection, and consequently within 255 InnerAsp. I will assume that they merge in the head of InnerAsp, just like other prefixes, as shown in (75). (75) Figure 17: The syntactic projection of na Q -, po Q -DA and po D - InnerAsp   na Q - [telic] VP po Q-DA - [telic]  po D - [telic] For all the stacking cases, where Q-prefixes attach to secondarily imperfectivized forms, I will propose a new telicity marking projection, InnerAsp Q . This will result in the following structure: (76) Figure 18: The syntax of na Q - and po D - attaching to S-imperfectivized forms InnerAsp Q    na Q - [telic] InnerAsp SI po D - [telic]   PART-HOM InnerAsp            PREFIX VP       Assuming more than one telicity-marking projection has its precedent in the literature. More specifically, it has been proposed in the aforementioned Istratkova (2006) in the context of her analysis of the syntax of all superlexical prefixes in Bulgarian. She proposes that while telicity can be marked through the Result Phrase, where lexical prefixes are merged, it can also happen in the functional projection she calls Quantization Phrase (QP), which projects over vP and below Viewpoint Aspect (AspP in her model). The proposed structure is shown in (77) below, where projections above AspP represent 256 the stacking possibilities in addition to an earlier merged one superlexical prefix in Q[uantization]P. (77) Syntax of stacked prefixes in Bulgarian (Istratkova 2006:22) [ AttentuativeP po- [ InceptiveP za- [ CompletiveP po- [ DistributiveP po- [ CumulativeP na- [ ExcessiveP raz- [ RepetitiveP pre- [ AspP Ø/-va- [ QP SUPERLEXICAL PREFIX [ Q‘ Q [ vP v [ VP V [ RP INTERN.ARG [ R‘ LEXICAL PREFIX]]]]]]]]]]]]]] Istratkova concedes that ―there is no single uniform structural position identified with inner aspect as a whole‖ (Istratkova 2006:23), and departs from the proposals where only lexical prefixes are implicated in telicity marking. In this model, telicity is marked below the outer aspect projection AspP by one telicity-marking element: either a lexical prefix in RP or a superlexical one in QP. Stacking prefixes are proposed to merge in a functional projection above AspP, and select for a telic predicate as their input ([+quantized] in the terminology used by Istratkova 2006) 17 . Consequently, they are not involved in telicity marking but preserve the semantic type of the predicate which they modify. The analysis presented in Istratkova (2006) attributes different properties to the same prefix depending on its position. When a superlexical prefix is merged lower, in QP, it telicizes the predicate; when it is merged higher, above outer aspect, it does not by itself mark predicates as telic but instead requires them to be telic at merger. This is clearly motivated by the fact that in Bulgarian, stacking prefixes appear to be able to attach to perfective forms, which is not possible in Polish, Russian, Czech, and other Slavic languages. When superlexicals attach to secondarily imperfectivized forms in 17 It is hard to be exactly sure if ‗quantized‘ can be equated with ‗telic‘ on Istratkova‘s (2006) account, as she assumes that non-telicizing, delimitative po Q - also merges in Spec, QP. 257 Bulgarian, Istratkova (2006) keeps the analysis constant by interpreting the secondary imperfective suffix in the typical way as a marker of the IMPERFECTIVE viewpoint, and counting secondary imperfectivized forms as quantized, even though such forms have all the characteristics of atelics. As I argued in Chapter 4, the role of the secondary imperfective marker is different from viewpoint aspect marking, and Bulgarian in particular provides a convincing case for it being analyzed as a partitive-homogenizing operator, applying to telic predicates and returning atelic ones. In consequence, secondary imperfective forms cannot be viewed as telic/quantized, and therefore the superlexicals attaching to such forms must be viewed as markers of telicity rather than selectors for a telic complement. The structure in (76) accounts for that, though admittedly a different one might be needed for the uniquely Bulgarian cases of stacking prefixes attaching to perfectives. With the basic analysis in places, there are a few issues that still need resolving. One concerns the fact that across Slavic languages, when more than one prefix stacks, only the first stacking prefix requires an imperfective base. 18 In other words, secondary imperfective derivation is not recursive. This in turn argues against awarding individual projections to each prefix, as it is done, e.g., in Istratkova (2006). Admittedly, the Bulgarian case is more complex because of the sheer number of stacking prefixes as well as their varying character (only some are quantificational – see (77) above), and I will not attempt an analysis that would cover them all. For the three Q-prefixes analyzed here, it seems motivated to accommodate them within a single projection. The only combinations that we really need to account for are: (i) po D - and na Q - within InnerAsp Q , and (ii) 18 I.e., for languages which have this requirement, such as Polish, Russian, and mostly Czech. 258 multiple po Q -‗s in ViewAsp. For the latter case, I will propose that stacked po Q -‗s are a reflex of a multiple specifier structure, as shown in (78)a and (78)b for previously unprefixed and prefixed verbs, respectively. (78) Figure 19: The syntax of iterated po Q -'s a. ViewAsp          po-[ PERF ] ViewAsp     po-[ PERF ] ViewAsp       [PERF] VP                       b. ViewAsp         po-[ PERF ] ViewAsp                       po-[ PERF ] ViewAsp                       [PERF] InnerAsp SI                PART-HOM InnerAsp                   PREFIX VP               For the telicity marking prefixes po D - and na Q -, the matter is more complicated. Three logical possibilities present themselves here. One is a somewhat unorthodox multi- headed analysis, with po D - and na Q - forming an adjunction structure in the head of InnerAsp Q . The other one would be to locate one of the prefixes in the head of InnerAsp Q , and the other one in its specifier. That however would leave unexplained the impossibility of merging a Q-prefix in Spec, InnerAsp – the lower telicity marking projection. After all, if a stacked Q-prefix can merge with a telicity-marking Q-prefix in 259 the head of InnerAsp Q , it should be able to merge with a telicity-marking ―regular‖ prefix in InnerAsp 0 . The third possibility is to assume that all stacked Q-prefixes merge as iterated specifiers of InnerAsp Q . They could be viewed in that configuration as agreeing in [telic] feature with a phonologically null Q [telic] head. Unlike with ViewAsp, which is by assumption necessary, InnerAspQ would not be able to project with just phonologically null material (i.e., just with Q [telic] in its head), and would instead be legitimate only if overt quantifiers are present. The problem with the specifier analysis presented above is that it precludes the movement of the object to the specifier of the projection hosting na Q - which gives us the configuration like the one proposed in Pereltsveig (2006) and replicated in (90) (see also the discussion on pp. 173-174 in Chapter 5). (90) na-P   QP na-P              Q NP na- …               … VP      V 0 The issue is not just how to account for the relationship between na Q - and the direct object, but also that the direct object needs to move to/through the InnerAsp Q projection in order to receive the interpretation assigned to the objects of perfective verbs, namely the necessarily specific/definite interpretation that is not obtained in the context to imperfectives – bare or secondary. The head analysis makes this possible, while the specifier approach does not. Based on the above consideration, we have to conclude that the multiple / conjoined head analysis, though somewhat unorthodox, must be preferred. 260 6. Conclusions In this chapter I considered the problem of aspectual interpretation of verbs prefixed with pan-Slavic quantificational prefixes: na Q - ‗a lot, enough,‘ po Q - ‗some, a little‘ and the distributive po D -. I concluded that they constitute a mixed group with respect to their Aktionsart characteristics: na Q - and po D - derive telicit readings, while po Q - generally does not. One thing they do share is that Q-prefixed verbs are viewpoint PERFECTIVE. This is also the only feature that unifies all prefixes in Slavic, quantificational and non-quantificational alike. Consequently, (at least some of) the commonly used Slavic tests for (im)perfectivity (e.g., future vs. present tense denotation for present tense forms, compatibility with phase verbs, participle formation, etc.), which group all prefixed verbs together, including atelic po Q -prefixed verbs, must be understood as test for viewpoint aspect, and more specifically as tests for the unavailability of IMPERFECTIVE readings, which characterizes perfectives to the exclusion of imperfectives. In terms of the syntax of Q-prefixes, I proposed that telicity-marking ones are merged in either of the two Aktionsart projections: as heads of InnerAspP for non- stacking cases, or as heads of the new InnerAsp Q phrase – a telicity-encoding projection merging above the atelicizing secondary imperfective projection. The result is a recursive telicity-marking structure, which while unorthodox, is necessary in Slavic to derive the correct interpretation. 261 CHAPTER 7: CONCLUSIONS AND OUTSTANDING ISSUES In this dissertation, I considered the problem of the grammatical function of three kinds of Slavic aspectual morphology: (i) prefixes attaching directly to the root, (ii) secondary imperfective morphemes, and (iii) three quantificational prefixes, which have the ability to stack on top of previously prefixed verbs: na Q - ‗a lot, enough,‘ po Q - ‗some, a little‘ and the distributive po D -. I concluded, lending further support to the growing consensus, that (native) Slavic verbal prefixes are all telicity markers. I also showed that prefixed forms have necessarily viewpoint PERFECTIVE interpretation, in contrast to imperfectives, which can be interpreted IMPERFECTIVELY as well as PERFECTIVELY. I attributed the invariable PERFECTIVE interpretation of prefixed form to the compositional restriction, assumed to be of cross-linguistic validity, that the IMPERFECTIVE viewpoint may not compose directly with a telic predicate. With regard to the phenomenon of secondary imperfectivization, Bulgarian and Old Church Slavonic – two languages with a separate system of viewpoint aspect marking through the aorist-imperfect distinction – were found to provide a strong case against the treatment of the secondary imperfective as a marker of the IMPERFECTIVE viewpoint, contrary to the widely held belief. I proposed a novel analysis of the secondary imperfective as an Aktionsart modifier, and more specifically as a PARTITIVE- HOMOGENIZING operator, taking a telic predicate as its input and returning an atelic one. Syntactically, I assumed it to head the InnerAsp SI projection, itself merging with the telicity-encoding InnerAsp. The accuracy of this new analysis was furher confirmed by 262 the availability of PERFECTIVE interpretation for secondary imperfective forms in Polish, while their Aktionsart interpretation was consistently atelic. Lastly, with regard to the issue of quantificational prefixes, I first discussed their unique distributional and semantic properties as compared with other prefixes. I arrived at a novel conclusion that na Q - ‗a lot‘ and po Q - ‗some, a little,‘ when applied to the internal argument, function not as quantifiers but as intensifiers of quantity descriptions, and that the internal arguments in the relevant contexts must be independently quantified / measured by either an over or covert quantifier or a measure phrase. Conversely, when applied to the event, na Q - and po Q - act as variable quantifiers, measuring directly over any of the scales supplied by the predicate (e.g., duration, intensity, degree of change), with the qualification that in order to do so, na Q - must be blocked from applying to the internal argument through by the addition of the reflexive clitic się ‗self.ACC.‘ I then considered the aspectual role of Q-prefixes and concluded, contrary to what is often proposed, that prefixes na Q -, po D - and po Q - in the context of degree achievements are in fact telicity markers, while only po Q - is not. All Q-prefixed forms, however, share the property of being PERFECTIVE, which unifies them with verbs prefixed with non-quantification prefixes. With regard to Aktionsart-neutral po Q -, I proposed that it merges in the specifier of Viewpoint Aspect projection, and that it has to agree in there with the null head bearing the feature [PERFECTIVE]. For all other Q-prefixes, the necessarily PERFECTIVE interpretation follows from the compositional requirement identified earlier. These prefixes, which are telicity markers, were proposed to merge in InnerAsp (i.e., the first telicity-encoding projection) in non-stacking configurations, and in InnerAsp Q otherwise. Since InnerAsp Q itself merges with the atelicizing InnerAsp [SI] , it provides a mechanism 263 to recursively derive telic readings which is not known to exist in the well-described languages. There are a number of issues related to the morphological marking of Aspect in Slavic that were not considered in this work and remain outstanding. The first one is the distribution and role of semelfactive markings. Semelfactives, which are predicates denoting punctual events (i.e., events conceptualized as lacking duration), are marked in Slavic with the suffix –n. This suffix can sometimes co-occur with a prefix, but prefixation of semelfactives is not completely productive. This raises the question of the role of the prefix in semelfactive forms in the light of the general assumption that semelfactives are PERFECTIVE but not telic (Smith 1991). The larger question is whether an event lacking duration can be at all evaluated in terms of homogeneity, which underlies the notion of (a)telicity. A second significant issue concerns habitual interpretation and habitual marking. Slavic languages differ in the extent to which habitual marking is allowed with respect to bare imperfective forms: some use it with a number of verbs (e.g., Polish), some find it ―bookish‖ (e.g., Russian), and others have lost it altogether (e.g., Bulgarian). However, given that habitual marking is homophonous with secondary imperfective morphology, and assuming that habituality is encoded by a projection distinct from and structurally higher than viewpoint, it should be examined whether cases where secondary imperfective forms are interpreted habitually involve a secondary imperefective marker, or whether they might be analyzed as involving a habitual marker attaching to a perfective-marked and telic predicate. 264 A problem related to this is the general issue of pluractionality and Aktionsart interpretation. While in principle pluractionality is idependent of Aktionsart and vice versa, habitual, iterative and counted readings (i.e., once/1 time, twice, 10 times, etc.), appear to allow telic interpretation for otherwise atelic predicates. This can be demonstrated using the adverbial modification test, as in (1) and (2) for bare imperfective and secondary imperfective forms, respectively. (1) a. Czytałam tą książkę 3 godziny / *w 3 godziny. read.PAST.1SG this book.ACC (for) 3 hours / in 3 hours ―I was reading this book for 3 hours / * in 3 hours. Cf. b. Czytałam tą książkę raz/wiele razy/często 3 godziny / w 3 godziny. read.PAST.1SG this book.ACC one time/many times/often (for) 3 hours / in 3 hours ―I read this book one time / many times / often for 3 hours / in 3 hours.‘ (2) a. Prze-pis-yw-ałam notatki godzinę / *w godzinę. through-wrote.1SG notes.ACC (for) hour / in hour ―I was re-writing (the) notes for an hours / * in an hour. Cf. b. Prze-pis-yw-ałam notatki 3 razy / zazwyczaj / zawsze godzinę / w godzinę. through-wrote.1SG notes.ACC 3 times / usually / always (for) hour / in hour ―I re-wrote (the) notes 3 times / usually / always for an hours / in an hour. The examples above show that the pluractional/counted event reading allows a different set of interpretations that the progressive reading does. This in turn leads to questions about the nature of the interaction between different types of aspect as well as the nature 265 of the commonly used tests for telicity. These and other issues remain a topic for future research. 266 BIBLIOGRAPHY Abusch, D. (1986). Verbs of change, causation and time, Technical Report CSLI-86-50, Center for the Study of Language and Information. Stanford University. Adler, M. (1994). Contributions to the study of aspect, with special reference to Czech. M.A. Thesis. Simon Fraser University. Amse-de Jong, T. H. (1974). The Meaning of the Finite Verb Forms in the Old Church Slavonic Codex Suprasliensis. The Hague; Paris: Mouton. Babko-Malaya, O. (2003). Perfectivity and prefixation in Russian. Journal of Slavic Linguistics 11, 5–36. Babko-Malaya, O. (1999). Zero Morphology: A Study of Aspect , Argument Structure, and Case. Ph.D. Dissertation. Rutgers University. Bach, E. (1986). The algebra of events. Linguistics and Philosophy 9 , 5-16. Baker, M. C. (1988). Incorporation: A theory of grammatical function changing. Chicago, IL: University of Chicago Press. Barker, C. (1998). Partitives, double genitives and anti-uniqueness. Natural Language and Linguistic Theory 16 , 679-717. Bertinetto, P. M. (2001). On a frequent misunderstanding in the temporal-aspectual domain: The ‗Perfective = Telic Confusion‘. In C. Cecchetto, G. Chierchia, & M. T. Guasti, Semantic Interfaces [Reference, Anaphora and Aspect] (pp. 177-210). Stanford, CA: CSLI Publications. Bhatt, R., & Pancheva, R. (2005). LSA 130: The Syntax and Semantics of Aspect. Lecture Notes. Bialy, A. (1998). Unaccusativity in Polish. Proceedings of the Third ESSLLI Student Session, Geert-Jan Kruijff, Ivana Kruijff-Korbayova (eds.). Saarbrücken, Germamy. Bohnemeyer, J., & Swift, M. (2001). Default Aspect: The Semantic Interaction of Aspectual Viewpoint and Telicity. In Proceedings of Perspectives on Aspect. Utrecht. Borer, H. (2003). Exo-skeletal vs. endo-skeletal explanations: syntactic projections and the lexicon. In J. Moore, & M. Polinsky, The Nature of Explanation in Linguistic Theory. Stanford, CA: CSLI. Borer, H. (2005). The Normal Course of Events. Structuring Sense Vol. II. Oxford, UK & New York: Oxford University Press. 267 Borer, H. (1994). The projection of arguments. In E. Benedicto, & J. Runner, University of Massachusetts Occasional Paper in Linguistics 17. Amherst: GLSA, University of Massachusetts. Borik, O. (2002). Aspect and Reference Time. Ph.D. Dissertation. Utrecht: OTS Dissertation Series. Borodič, V. (1953). K voprosu o formirovanii soveršennogo i nesoveršennogo vida v slavjanskih jazákah . Vopr. jazakoznanija, br. 6 , 68-86. Bouchard, D.-E., & Burnett, H. (2007). Quantification at a Distance: A Critical Review of the Literature. in Milica Radisic (ed.). Proceedings of the 2007 Meeting of the Canadian Linguistics Association. Brajerski, T. (1993). Slowianskie przedrostki czasownikowe i aspekt [Slavic verbal prefixes and aspect]. In M. Basaj, & Z. Zygmunt, Munera linguistica Ladislao Kuraszkiewicz dedicata (pp. 63-67). Wroclaw: Zaklad im. Ossolinskich. Brecht, R. D. (1984). The Form and Function of Aspect in Russian. In M. S. Flier, & R. D. Brecht, Issues in Russian Morphosyntax (pp. 9-34). Columbus, OH: Slavica Publishers. Carlson, G. (2006). The meaningful bounds of incorporation. In S. Vogeleer, & L. Tasmowski, Non-Definiteness and Plurality. In the Linguistik Aktuell series (pp. 35-50). Amsterdam: Benjamins. Chomsky, N. (1993). A minimalist program for linguistic theory. In K. Hale, & S. J. Keyser, The view from Building 20 (pp. 1--52). Cambridge, MA: MIT Press. Cipria, A., & Roberts, C. (2000). Spanish Imperfecto and Preterito: Truth Conditions and Aktionsart Effects in a Situation Semantics. Natural Language Semantics 8:4 , 297–347. Comorovski, I. (1995). On Quantifier Strength and Partitive Noun Phrases. In E. e. Bach, Quantification in Natural Languages (pp. 145-177). Dordrecht: Kluwer. Comrie, B. (1976). Aspect. Cambridge, UK.: Cambridge University Press. Cox, T. (1982). The Inchoative Aspect in French . French Review 55.2 , 228-40. Cyr, F. (1991). La quantification distance en français québécois. MA Thesis. Université de Montréal. Czaykowska-Higgins, E. (1997). Verbalizing Suffixes and the Structure of the Polish Verb . In Yearbook of Morphology (pp. 25-58). 268 Czochralski, J. (1972). Verbalaspekt und Tempussystem im Deutschen und Polnischen: eine konfrontative Darstellung (Aspect and tense system in Polish ang German: a contrastive study). Warszawa: Wydawnictwa Uniwersytetu Warszawskiego. Dahl, Ö. (1985). Tense and Aspect Systems. Oxford, UK & New York: Basil Blackwell. Dayal, V. (1999). Bare NP's, Reference to Kinds, and Incorporation. Proceedings of SALT IX. Ithaca, NY: Cornell University, Cornell Linguistics Club. de Swart, H., & Verkuyl, H. (1999). Tense and Aspect in Sentence and Discourse. Utrecht: Utrecht Institute of Linguistics OTS. Di Sciullo, A. M., & Slabakova, R. (2005). Quantification and Aspect. In H. Verkuyl, H. de Swart, & A. van Hout, Perspectives on Aspect. Dordrecht: Kluwer AP. Dickey, S. M. (2000). Parameters of Slavic Aspect. A Cognitive Approach. Stanford, CA. Doetjes, J. (1995). Quantification at a distance and iteration. Proceedings of NELS (North Eastern Linguistic Society) 25, University of Massachusetts, Amherst. Dowty, D. R. (1977). Toward a Semantic Analysis of Verb Aspect and the English 'Imperfective' Progressive. Linguistics and Philosophy, Vol. 1, No. 1 (Jan., 1977) , 45-77. Dowty, D. R. (1979). Word Meaning and Montague Grammar. Dordrecht: Reidel. Dowty, D. R. (1979). Word Meaning and Montague Grammar. Dordrecht: Reidel. Dowty, D. (1991). Thematic Proto-Roles and Argument Selection . Language 67 , 547- 619. Filip, H. (forthcoming). Aspect: Theoretical and Cross-linguistic Perspectives. Oxford: Oxford University Press. Filip, H. (2005). On Accumulating and Having it All: Perfectivity, Prefixes and Bare Arguments. In H. Verkuyl, d. S. Henriette, & v. H. (eds.), Perspectives on Aspect (pp. 125-148). Dordrecht: Kluwer Academic Publishers Group. Filip, H. (2003). Prefixes and the Delimitation of Events. In W. Browne, & B. (. Partee, Journal of Slavic Linguistics, Volume 11.1, a special issue (pp. 55-101). Filip, H. (2000). The Quantization Puzzle. In C. Tenny, & J. (. Pusejovsky, Events as grammatical objects, from the combined perspectives of lexical semantics, logical semantics and syntax (pp. 3-60). Stanford, CA: CSLI Press. Forsyth, J. (1970). A grammar of aspect: usage and meaning in the Russian verb. Studies in the modern Russian language. Cambridge: Cambridge University Press. 269 Franks, S. (1998). Clitics in Slavic. Position paper for Comparative Slavic Morphosyntactic workshop. Franks, S. (1994). Parametric properties of numeral phrases in Slavic. Natural Language and Linguistic Theory 12 , 570-649. Franks, S. (2005). Slavic Languages. In G. Cinque, & R. Kayne, Handbook of Comparative Syntax. Fukuda, S. (2007). The Syntax of Telicity in Vietnamese. Proceedings of WECOL 34, Vol 17. Eed by Erin Bainbridge & Brian Agbayani (pp. 109-120). Fresno: CSU Fresno. Garey, H. B. (1957). Verbal aspects in French. Language 33 , 91–110. Gawrońska, B. (1993). An MT Oriented Model of Aspect and Article Semantics. Lund University Press. Giorgi, A., & Pianesi, F. (1998). Tense and Aspect: From Semantics to Morphosyntax. Oxford: Oxford University Press. Gladney, F. Y. (1982). Bi-aspectual verbs and the syntax of aspect in Russian. Slavic and East European Journal 26: , 202–215. Gramatika Na Sĕvremennja Bâlgarski Knizhoven Ezik (The Grammar of the Contemporary Literary Bulgarian). (1993). Halle, M., & Marantz, A. (1993). Distributed Morphology and the pieces of inflection. In K. Hale, & S. J. Keyser, The View from Building 20 (pp. 111-176). Cambridge, MA: MIT Press. Hay, J., Kennedy, C., & Levin, B. (1999). Scale structure underlies telicity in ‗degree achievements‘. Semantics and Linguistic Theory 9, Tanya Matthews and Devon Strolovitch (eds.) , 127–144. Ithaca, NY: CLC Publications. Hewson, J., & Bubenik, V. (1997). Tense and Aspect in Indo-European Languages. Theory, Typology and Diachrony. Current Issues in Linguistic Theory, Vol. 145. Amsterdam & Philadelphia: John Benjamins. Hoeksema, J. (1984). Partitives, [Unpublished paper] University of Groningen. Horrocks, G., & Stavrou, M. (2003). Actions and their Results in Greek and English: The Complementarity of Morphologically Encoded (Viewpoint) Aspect and Syntactic Resultative Predication. Journal of Semantics 20(3) , 297-327. Iatridou, S. (2000). The grammatical ingredients of counterfactuality. Linguistic Inquiry, Vol. 31, No. 2, Spring 2000 , MIT Press. 270 Isačenko, A. V. (1962). Die Russische Sprache der Gegenwart, volume I. Niemeyer: Halle (Saale). Isačenko, A. V. (1960). Grammaticeskij stroj russkogo jazyka v sopostavleni s slovackim. Morfologija. Bratislava. Isačenko, A. V. (1966). The morphology of the Slovak verb. Travaux linguistiques de Prague 1 , 183-201. Istratkova, V. (2006). When Prefixes Become Adverbs. Proceedings of the Second Graduate Colloquium on Slavic Linguistics, vol. 6, Ivanova, T., Sims, A. & Whiting, M. (eds.) (pp. 1-31). Columbus, OH: Ohio Slavic Papers, The Ohio State University. Itô, J., & Mester, A. (2002). The Phonological Lexicon. In N. Tsujimura, A Handbook of Japanese Linguistics (pp. 62-100). Oxford: Blackwell. Jabłońska, P. (2004). When the prefixes meet the suffixes. In P. Svenonius, Nordlyd 32.2. Special issue on Slavic prefixes (pp. 363-401). Tromsø: CASTL. Jacobson, R. (1936). Beitrag zur allgemeinen Kasuslehre. English translation in Jakobson, Russian and Slavic Grammar Studies, Ed. by Linda R. Waugh and Morris Halle. Berlin: Mouton de Gruyter. 1984. Janda, L. A. (2007). What makes Russian Bi-aspectual verbs Special. In D. Divjak, & A. Kochanska, Cognitive Paths into the Slavic Domain. Cognitive Linguistics Research. (pp. 83-109). Berlin & New York: Mouton de Gruyter. Kiparsky, P. (1998). Partitive Case and Aspect. In M. Butt, & W. Geuder, The Projection of Arguments: Lexical and Compositional Factors (pp. 265-307). Stanford: CSLI Publications. Kiparsky, P. (1996). Partitive Case and Aspect, Ms. Stanford University. Klein, W. (1994). Time in Language. London: Routledge. Kolln, H. (1958). Die Entstehung des slavischen Verbalaspektes. Scando-Slavica IV , 308-313. Kratzer, A. (1996). Severing the External Argument from its Verb. In J. Rooryck, & L. A. Zaring, Phrase Structure and the Lexicon (pp. 109–137). Dordrecht: Kluwer Academic Publishers. Kratzer, A. (2004). Telicity and the Meaning of Objective Case. In J. Guéron, & J. Lecarme, The Syntax of Time (pp. 389–424). Cambridge, MA: MIT Press. Kratzer, A. (2005). The Syntax of Time. Cambridge, MA: MIT Press. 271 Krifka, M. (1989). Nominal reference, temporal constitution, and quantification in event semantics. In R. Bartsch, J. van Bentham, & P. van Emde Boas, Semantics and Contextual Expressions. Dordrecht: Foris. Krifka, M. (1998). The origins of telicity. In S. Rothstein, Events and Grammar (pp. 197–235). Dordrecht: Kluwer Academic Publishers. Krifka, M. (1992). Thematic relations as links between nominal reference and temporal constitution. In I. Sag, & A. Szabolsci, Lexical Matters, CSLI Lecture Notes. Stanford, CA: University of Stanford. Krifka, M. (1992). Thematic relations as links between nominal reference and temporal constitution. In I. Sag, & A. Szabolsci, Lexical Matters. Stanford, CA: CSLI. Kuznecov, P. (1953). K voprosu o genezise vido-vremennáh otnošeniè drevnerusskogo jazáka. In Trudá in-ta jazákoznanija AN SSSR, Moskva, tom 2 (pp. 220-256). Moscow. Ladusaw, W. A. (1982). Semantic constraints on the English partitive construction. In Flickinger et al. (eds.), Proceedings of WCCFL 1. The Stanford Linguistics Association. Landman, F. (1992). The Progressive. Natural Language Semantics 1:1 , 1-32. Laskowski, R. (1998). Czasownik [The Verb]. In R. Grzegorczykowa, R. Laskowski, & H. Wrobel, Morfologia [Morphology]. Vol 1. (pp. 225-269). Warszawa: PWN. Lubowicz, A. (1999). Two Views of Polish Reflexives. Proceedings of the 18th West Coast Conference on Formal Linguistics (pp. 337-350). Cascadilla Press. Lunt, H. G. (2001). Old Church Slavonic Grammar. 7th Edition. Berlin; New York: Mouton de Gruyter. Maslov, J. (1963). Morfologija glagol'nogo vida v sovremennom bolgarskom jazyke [Morphology of Verbal Aspect in Modern Bulgarian]. Moscow and Leningrad. Maslov, J. (1961). Rol' tak nazyvaemoj perfektivizacii i imperfektivacii v procese vozniknovenija slavjanskogo glagol'nogo vida. In S. B. Bernshtejn, Issledovanija po slavjanskomu jazykoznaniju (pp. 165–95). Moscow. Maslov, J. (1958). Rol‘ tak nazvyvaemoj perfektivacii i imperfektivacii v processe vozniknovenija slavjanskogo glagol‘nogo vida. Proceedings of the IV International Congress of Slavists, (pp. 3-39). Moscow. Mavromanolaki, G. (2001). Verbs of change of state and tests for unaccusativity. In Studies in Greek Linguistics, Vol.22 (pp. 429-440). Thessaloniki: Aristotle University. 272 McCawley, J. D. (1968). The Phonological Component of a Grammar of Japanese. The Hague: Mouton. Mezhevich, I. (2008). A feature-theoretic account of tense and aspect in Russian. Natural Language and Linguistic Theory, 26.2 , 359–401. Milićević, N. (2004). The lexical and superlexical verbal prefix iz- and its role in the stacking of prefixes. In P. Svenonius, Nordlyd 32.2: Special issue on Slavic prefixes (pp. 279-300). Tromsø: CASTL. Młynarczyk, A. (2001). A Verb Classification for Polish. Presented at Perspectives on Aspect, December, Utrecht. Młynarczyk, A. (2004). Aspectual Pairing in Polish. Ph.D. Dissertation. Utrecht: OTS Dissertation Series. Nagórko, A. (1998). Zarys gramatyki polskiej (ze słowotwórstwem). (An Outline of Polish Grammar (with wordformation)). Warszawa: PWN. Nakanishi, K. (2008). The syntax and semantics of floating numeral quantifiers. In S. Miyagawa, & M. Saito, The Oxford Handbook of Japanese Linguistics. Oxford: Oxford University Press. Nakanishi, K., & Romero, M. (2004). Two constructions with most and their semantic properties. The Proceedings of the 34th Conference of the North East Linguistic Society (NELS 34). Newman, L., & Townsend, C. (1972). Towards a new classification of Czech Verbs. Slavic and East European Journal, Vol. 16. No. 3 , 324-. Nossalik, L. (2009). Slavic Inceptive Verbs: Accomplishments or Achievements. Proceedings of the 2009 annual conference of the Canadian Linguistic Association. Pancheva, R. (2008). Layers of Aspectual Structure in Bulgarian. Workshop on "Temporality: Typology and Acquisition". CNRS Paris, 24 June 2008. Papke, J. (2005). Order and Meaning in Sanskrit Preverbs. Paper presented at ICHL XVII. Madison, WI. Paslawska, A., & von Stechow, A. (2003). Perfect Readings in Russian. In A. Alexiadou, M. Rathert, & A. (. vin Stechow, Perfect Explorations. De Gruyter. Pereltsvaig, A. (2005). Aspect Lost, Aspect Regained: Restructuring of aspectual marking in American Russian. In P. Kempchinsky, & R. Slabakova, Aspectual Inquiries (pp. 369-395). Dordrecht: Kluwer. 273 Pereltsvaig, A. (to appear). Compositional semantics of Russian tense and aspect. Journal of Slavic Linguistics . Pereltsvaig, A. (2006). Small Nominals. Natural Language and Linguistic Theory 24(2) , 433-500. Piñón, C. (1995). A mereology for aspectuality. Ph.D. Dissertation. Stanford, CA: Stanford University. Pollock, J.-Y. (1989). Verb Movement, Universal Grammar, and the Structure of IP. Linguistic Inquiry 20 , 365-424. Portner, P. (1998). The Progressive in Modal Semantics. Language 74:4 , 760–787. Ramchand, G. (1997). Aspect and Predication:The Semantics of Argument Structure. Oxford: Oxford University Press. Ramchand, G. (2004). Time and the event: The semantics of Russian prefixes. In P. Svenonius, Nordlyd 32 (2) (pp. 323-361). Tomsø: CASTL. Reichenbach, H. (1947). Elements of Symbolic Logic. New York: MacMillan. Romanova, E. (2007). Constructing perfectivity in Russian. Ph.D. Dissertation. University of Tromsø. Rothstein, S. (2004). Structuring Events. Oxford: Blackwell Publishing. Schmitt, C. J. (1966). Aspect and the syntax of noun phrases. Ph.D. Dissertation. University of Maryland. Schoorlemmer, M. (1995). Participial passive and aspect in Russian, Ph.D. dissertation. Utrecht: Utrecht University. Schuyt, R. (1990). The Morphology of Slavic Verbal Aspect. A Descriptive and Historical Study. Studies in Slavic and General Linguistics, Vol. 14. Amsterdam & Atlanta: Rodopi. Schwarzschild, R. (2006). The Role of Dimensions in the Syntax of Noun Phrases. In Syntax 9.1 , 67-110. Slabakova, R. (2005). Perfective Prefixes: What they are, what flavors they come in, and how they are acquired? Formal Approaches to Slavic Linguistics 13: The South Carolina Meeting (pp. 324-341). Ann Arbor, MI: Michigan Slavic Publications. Smith, C. (1991 ). The Parameter of Aspect. Dordrecht: Kluwer Academic Press. Spencer, A., & Zaretskaya, M. (1998). Verb prefixation in Russian as lexical subordination. Linguistics 36 , 1-39. 274 Stone, G. (1993). Sorbian (Upper and Lower. In B. Comrie, & G. G. Corbett, The Slavonic languages. London and New York: Routledge. Streitberg, W. (1889). Perfective und imperfective Actionsart im Germanischen I. Halle. Svenonius, P. (2004). Slavic prefixes inside and outside VP. In P. Svenonius, Nordlyd 32.2, Special issue on Slavic prefixes (pp. 205-253). Tomsø: CASTL. Tenny, C. (1994). Aspectual Roles and the Syntax-Semantics Interface. Dordrecht: Kluwer. Tenny, C. (1987). Grammaticalizing aspect and affectedness, Ph.D. dissertation. MIT. Tenny, C. (1992). The Aspectual Interface Hypothesis. In I. Sag, & A. Szabolcsi, Lexical Matters (pp. 1-27). Stanford, CA: CSLI. Tenny, C., & Pustejovsky, J. (2000). Events as grammatical objects. Stanford, CA: CSLI. The Dictionary of Polish. (www.sjp.pwn.pl). PWN. Toops, G. H. (1998). On the Functional Status of Derived Imperfectives in Contemporary Upper Sorbian. Slavic and East European Journal 42 (1998):2 , 283–297. Townsend, C. E. (1980). Russian Word-Formation. Slavica Publishers. Travis, L. (2005). Articulated vPs and the computation of aktionsart. In P. Kempchinsky, & R. Slabakova, Aspectual Inquiries (pp. 69-94). New York: Springer. Travis, L. D. (2005). Agents and Causes in Malagasy and Tagalog. In T. Rapoport, The Syntax of Aspect: Deriving Thematic and Aspectual Interpretation (pp. 174-89). New York: Oxford University Press. Travis, L. (1994). Event phrase and a theory of functional categories. Priceedings of thr 1994 Annual Conference of the Canadian Linguistic Association, (Toronto Working Papers in Linguistics), ed. by P. Koskinen, (pp. 559-570). Toronto. Travis, L. (1991). Inner Aspect and the Structure of VP. Cahiers de Linguistique de l‟UQAM 1 , 132-146. van Geenhoven, V. (2005). Atelicity, pluractionality and adverbial quantification. In H. Verkuyl, H. de Swart, & A. van Hout, Perspectives on Aspect. Dordrecht: Springer. Van Geenhoven, V. (1998). Semantic incorporation and indefinite descriptions: semantic and syntactic aspects of noun incorporation in West Greenlandic. Stanford, CA:: CSLI Publications. 275 van Hout, A. (1996). Even Semantics of Verb Frame Alternations. Tilburg: TILDIL dissertation series, 1996-1. Tilburg University. van Hout, A. (1992). Linking and projection based on event structure, Ms. Tilburg University. Van Wijk, N. (1927). Die sogenannten Verba iterativa und die Bezeichnung der wiederholten Handlung im Altkirchenslavischen. Vendler, Z. (1967). Linguistics in Philosophy. Ithaca: Cornell. Vendler, Z. (1957). Verbs and times. The Philosophical Review LXVI , 143-160. Reprinted in a revised version in Vendler (1967). Verkuyl, H. J. (1989). Aspectual classes and aspectual composition. Linguistics and Philosophy 12 , 39-94. Verkuyl, H. J. (1972). On the Compositional Nature of the Aspect. Dordrecht: Reidel. Verkuyl, H. (1999). Tense, aspect, and aspectual composition. In M. Dimitrova- Vulchanova, & L. Hellan, Topics in South Slavic Syntax and Semantics (pp. 125- 162). Amsterdam: John Benjamins. Vikner, C. (1994). Change in homogeneity in verbal and nominal reference. In C. Bache, H. Basboll, & C.-E. Lindberg, Tense, Aspect and Action. Empirical and Theoretical Contributions to Language Typology (pp. 139-164). Berlin & New York: Mouton de Gruyter. Vinogradov, V. .. (1947). Russkij Jazyk (Grammatičeskoe učenie o slove). Moscow: Vysšaja škola. Zucchi, S. (1999). Incomplete Events, Intensionality, and Imperfective Aspect. Natural Language Semantics 7:2 , 179–215.

Abstract (if available)

Abstract This dissertation considers the problem of the semantic function of verbal aspectual morphology in Polish and other Slavic languages in the framework of generative syntax and semantics. Three kinds of such morphology are examined: (i) prefixes attaching directly to the root, (ii) secondary imperfective suffixes, and (iii) three prefixes that have a quantificational function and are able to stack on top of already prefixed verbs: na- ‘a lot,’ po- ‘some, a little’ and the distributive po-. Assuming a two-tiered theory of aspect (Smith 1991), where two distinct types of grammatically encoded aspect are distinguished – Aktionsart (telicity, durativity, dynamicity) and viewpoint (PERFECTIVITY vs. IMPERFECTIVITY) – the aspectual contribution of the above listed markers is analyzed using various syntactic and semantic tests, leading to a number of new conclusions and novel accounts of various problems. It is determined that Slavic verbal prefixes are uniformly markers of telicity, and then proposed that they project as heads of the telicity-encoding projection InnerAsp. Prefixed forms are also viewpoint-PERFECTIVE (= temporally bound), but for a number of reasons this fact is attributed to a compositional restriction whereby telic predicates compose only with the PERFECTIVE viewpoint, rather than to the prefixes fulfilling that role. With regard to secondary imperfective suffixation, careful examination of data from Old Church Slavonic, Bulgarian and Polish reveals, contrary to what is commonly assumed, that secondary imperfective forms need not be interpreted IMPERFECTIVELY. Their interpretation is, on the other hand, always atelic. This leads to the novel proposal that the secondary imperfective morphemes correspond to a partitive-homogenizing operator, which applies to telic predicates and returns atelic ones. It is further proposed that this operator heads the InnerAspSI projection, merging right above InnerAsp.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Functional categories: the syntax of DP and DegP

PDF

The morphosyntax of states: deriving aspect and event roles from argument structure

PDF

Narrowing the focus: experimental studies on exhaustivity and contrast

PDF

Silence in answers: a study of ellipsis in Hindi

Asset Metadata

Creator Łazorczyk, Agnieszka Agata (author)

Core Title Decomposing Slavic aspect: the role of aspectual morphology in Polish and other Slavic languages

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Linguistics

Publication Date 08/05/2010

Defense Date 06/04/2010

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag Aktionsart,aspect,Bulgarian,OAI-PMH Harvest,Old Church Slavonic,Polish,Slavic,verbal morphology,viewpoint aspect

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Borer, Hagit (committee chair), Guerzoni, Elena (committee member), Pancheva, Roumyana (committee member), Seifrid, Thomas (committee member)

Creator Email chigoji@verizon.net,lazorczy@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m3291

Unique identifier UC1302773

Identifier etd-Lazorczyk-3856 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-374228 (legacy record id),usctheses-m3291 (legacy record id)

Legacy Identifier etd-Lazorczyk-3856.pdf

Dmrecord 374228

Document Type Dissertation

Rights Łazorczyk, Agnieszka Agata

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu