* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download COMPLEMENT VERB VARIATION IN PRESENT
French grammar wikipedia , lookup
Macedonian grammar wikipedia , lookup
Ukrainian grammar wikipedia , lookup
Kannada grammar wikipedia , lookup
Georgian grammar wikipedia , lookup
Modern Hebrew grammar wikipedia , lookup
Japanese grammar wikipedia , lookup
Chinese grammar wikipedia , lookup
Old English grammar wikipedia , lookup
Swedish grammar wikipedia , lookup
English clause syntax wikipedia , lookup
Polish grammar wikipedia , lookup
Yiddish grammar wikipedia , lookup
Scottish Gaelic grammar wikipedia , lookup
Ancient Greek verbs wikipedia , lookup
Russian grammar wikipedia , lookup
Spanish grammar wikipedia , lookup
Preposition and postposition wikipedia , lookup
Portuguese grammar wikipedia , lookup
Icelandic grammar wikipedia , lookup
Serbo-Croatian grammar wikipedia , lookup
Pipil grammar wikipedia , lookup
Ancient Greek grammar wikipedia , lookup
Latin syntax wikipedia , lookup
Antisymmetry wikipedia , lookup
Lexical semantics wikipedia , lookup
COMPLEMENT VERB VARIATION IN PRESENT-DAY SERBIAN DISSERTATION Presented in Partial Fulfillment of the Requirements for The Degree Doctor of Philosophy in the Graduate School of The Ohio State University By Bojan Belic, M.A. ***** The Ohio State University 2005 Dissertation Committee: Professor Brian D. Joseph, Adviser Professor Charles E. Gribble Professor Daniel E. Collins Approved by ____________________________________ Adviser Graduate Program in Slavic and East European Languages and Literatures ABSTRACT Strictly synchronically speaking, with verbs, nouns, and adjectives as heads of matrix clauses in a sentence, standard Serbian syntax allows for variation between a non-finite complement – that is, a complement headed by a verb not inflected for tense, grammatical person and number – and a finite complement – that is, a complement headed by a verb inflected for tense, grammatical person and number. The non-finite complement is exclusively headed by an infinitive, a non-finite verb form in Serbian, whereas the finite complement is headed by a present tense form, a finite verb form in Serbian, invariably introduced by a complementizer da ‘that’ and, at the same time, in full grammatical agreement in person and number with the matrix. The variation of the two complements, referred to here as complement variation in Serbian (CVS), is a wellknown and a long-documented syntactic phenomenon, though never fully explained, at least not in syntactic terms. In this study I offer a critical view of the previous scholarship about the phenomenon, after which I provide a novel account of CVS. I view the phenomenon from the position of the latest views of control, more specifically unique control as a general linguistic phenomenon. ii I propose that the syntax of CVS is best understood if the role of all other factors responsible for CVS, such as dialectal, regional, socioloectal, idiolectal, semantic, and pragmatic factors, is minimized. I do exactly that in research that I conducted on a sample of native speakers from the territory of the city of Belgrade, the capital of Serbia. The research decisively proved that there is indeed one syntactic factor that crucially determines which complement, infinitival or da+present, is chosen in CVS. The syntactic factor was the presence or absence in the syntax of the matrix of the controller of the complement. This was the basis for a formal theoretical account of CVS. I demonstrate that CVS, as an instantiation of unique control, operates according to the following formula: X((α)) MATRIX (Yβ) [((α))/(β) COMPLEMENT]. iii Slobodanu, Gordani i Branku: jer verujete u mene! iv ACKNOWLEDGMENTS This dissertation, the most voluminous piece I have written in my life so far, I dedicate to my family: my father, my mother, and my brother. It feels only right that with this work, in which I invested as much energy as I possibly could at this moment, I try to begin to thank them for everything they have given me over the years. Here, however, I would like to acknowledge the people in part responsible for the way this dissertation looks. It has been a unique privilege to call Professor Brian D. Joseph my mentor, and to be his student for five years at The Ohio State University. To learn from Professor Joseph and to communicate with him is an experience that I can hardly, if ever, express in words other than “A-haaa!” I would leave every single meeting with him with this feeling of revelation, and with an ever empowering sensation that I have again gained something. As the number of our meetings grew, the sensation kept growing. His discreet guidance in all of what I did as a graduate student has made me discover what it is that I actually like doing in linguistics, and for that I will remain eternally indebted. Professors Charles E. Gribble and Daniel E. Collins have offered constructive support on numerous occasions – working on this dissertation was but the latest one. They have always accepted my ideas with scholarly appreciation for which I am grateful. v At a very important stage in the creation of this dissertation, Professor Peter W. Culicover generously offered his time, his linguistic insights, and then, ultimately, his own theoretical views, on which I rely in this dissertation. I thank him for all of the inspiring conversations both in and out of class. I would like to thank Mary Allen Johnson in particular for reading my whole dissertation before anybody else and offering invaluable comments. Her readiness to help is a rare quality and I therefore cherish it with special care. Pašo, hvala! I would also like to thank Andrea Sims who dedicated a substantial amount of her time providing help with statistical testing of the data presented in this dissertation. She also read and commented on the first two chapters of this dissertation. Finally, the theory developed in this dissertation is based on the contemporary Serbian data obtained from a representative sample of 204 native speakers who participated in my research. Their input provides the initial point in all of the considerations presented here. I thank them all! While all of the people mentioned above did their best in making this dissertation better, I take full responsibility for all of its shortcomings. Бојан Белић 22 August 2005 vi VITA 6 November 1972…………………………………. Born – Belgrade, Serbia 1998……………………………………………….. profesor srpskog jezika i književnosti (B.A.), University of Belgrade 1998-2000………………………………………… Graduate Teaching Assistant, University of Illinois at Chicago 2000……………………………………………….. M.A. Slavic Studies, University of Illinois at Chicago 2000-present………………………………………. Graduate Teaching Associate, The Ohio State University PUBLICATIONS 1. Bojan Belić, “Singular, Plural and Paucal: On Grammatical Number in Serbian,” OSU Working Papers in Slavic Studies, (June 2003). FIELDS OF STUDY Major Field: Slavic and East European Languages and Literatures Slavic Linguistics vii TABLE OF CONTENTS Page Abstract…………………………………………………………………………... ii Dedication……………………………………………………………………........ iv Acknowledgments……………………………………………………………….. v Vita……………………………………………………………………………….. vii List of Tables…………………………………………………………………….. xi List of Figures…………………………………………………………………….. xvi List of Abbreviations…………………………………………………………….. xv Chapters: 1 Introduction………………………………………………………………. 1 2 Language Issues……………………………………………………………9 2.1. The Serbian Language……………………………………………….. 10 2.2. Sources………………………………………………………………. 13 2.3. Conclusion…………………………………………………………… 14 3 Configuration of CVS…………………………………………………….. 16 3.1. Configuration………………………………………………………… 3.1.1. Essentials…………………………………………………… 3.1.2. Motivation………………………………………………….. 3.1.3. Place of Infinitive………………………………………….. 3.1.4. Summary…………………………………………………… 3.2. Matrix………………………………………………………………… 3.2.1. Nouns………………………………………………………. viii 17 18 20 23 32 32 34 3.2.2. Adjectives………………………………………………….. 3.2.3. Verbs……………………………………………………….. 3.2.4. Future Tense……………………………………………….. 3.2.5. Summary…………………………………………………… 3.3. Complement………………………………………………………….. 3.3.1. Infinitive……………………………………………………. 3.3.2. Da+Present…………………………………………………. 3.3.3. Summary…………………………………………………… 3.4. Conclusion…………………………………………………………… 4 Previous Accounts of CVS……………………………………………….. 62 4.1. Dialectology and Sociolinguistics of CVS………………………….. 4.2. Descriptive Accounts………………………………………………… 4.2.1. True Descriptive Accounts………………………………… 4.2.2. Descriptive-Explanatory Accounts………………………… 4.2.3. Summary…………………………………………………… 4.3. Explanatory Accounts……………………………………………….. 4.3.1. Generative Explanatory Accounts…………………………. 4.3.2. Non-Generative Explanatory Accounts……………………. 4.3.3. Summary…………………………………………………… 4.4. Conclusion…………………………………………………………… 5 37 40 48 51 52 53 53 60 61 64 69 69 76 81 82 82 92 98 98 A Study of CVS………………………………………………………….. 102 5.1. Research……………………………………………………………… 104 5.2. Overall CVS………………………………………………………….. 109 5.3. Syntactic Factors in CVS…………………………………………….. 113 5.3.1. Additional Linguistic Material……………………………... 113 5.3.2. Adverbials in CVS…………………………………………. 115 5.3.3. Direct and Indirect Objects in CVS………………………... 120 5.3.4. Verb Aspect and Tense in CVS……………………………. 127 5.3.5. Summary…………………………………………………… 131 5.4. Controller of the Complement in CVS………………………………. 133 5.4.1. Adjectives………………………………………………….. 134 5.4.2. Nouns………………………………………………………. 137 5.4.3. Verbs……………………………………………………….. 142 5.4.4. Summary…………………………………………………….146 5.5. Sociolinguistic factors in CVS……………………………………….. 148 5.5.1. Sex………………………………………………………….. 149 5.5.2. Education…………………………………………………... 150 5.5.3. Age…………………………………………………………. 152 5.5.4. Summary…………………………………………………… 154 5.6. Conclusion…………………………………………………………… 154 ix 6 Towards a Formalization of CVS………………………………………… 158 6.1. Control and CVS…………………………………………………….. 6.2. A Theory of Control………………………………………………….. 6.2.1. Semantic Treatment of Control…………………………….. 6.2.2. Treatment of Matrices……………………………………… 6.2.3. Treatment of Complements………………………………… 6.2.4. Formalizing Control………………………………………... 6.2.5. Summary…………………………………………………… 6.3. CVS as Control………………………………………………………. 6.3.1. CVS Matrices………………………………………………. 6.3.2. CVS Complements…………………………………………. 6.3.3. Summary…………………………………………………… 6.4. Formalizing CVS…………………………………………………….. 6.4.1. Relevant Factors……………………………………………. 6.4.2. Controller with Adjectives…………………………………. 6.4.3. Controller with Nouns……………………………………… 6.4.4. Controller with Verbs……………………………………… 6.4.4.1. Depersonalized Verbs……………………………. 6.4.4.2. Personalized Verbs……………………………….. 6.4.4.3. Additional Controllers with Verbs……………….. 6.4.4.4. Concluding Thought……………………………… 6.4.5. Summary…………………………………………………… 6.5. Conclusion…………………………………………………………… 7 160 164 165 167 168 169 170 170 171 177 179 180 181 182 184 185 186 187 195 197 197 199 Conclusion……………………………………………………………….. 201 7.1. Syntax of CVS……………………………………………………….. 7.2. Account of CVS…………………………………………………….... 7.3. Implications of CVS…………………………………………………. 7.4. Summary……………………………………………………………... 202 204 206 208 Appendices: A Information for Participants…......................……………………………... 209 B Consent Forms …….........………………………………………………... 212 C Questionnaire…….. …………………………………………………….... 215 Bibliography.........………………………………………………………………... 226 x LIST OF TABLES Table Page 3.1. Elements of the formula vis-à-vis sentences (2)-(4)…………………………. 19 3.2. CVS Statistics with the Binder……………………………………………… 22 3.3. CVS Statistics without the Binder…………………………………………… 23 5.1. Overall CVS…………………………………………………………………. 111 5.2. CVS with the MATRIX and COMPLEMENT physically separated………... 114 5.3. CVS with adverbials…………………………………………………………. 116 5.4. CVS with MATRIX and COMPLEMENT adverbials………………………. 117 5.5. CVS with COMPLEMENT adverbials……………………………………… 118 5.6. CVS with manner COMPLEMENT adverbials……………………………... 118 5.7. CVS with place COMPLEMENT adverbials………………………………... 119 5.8. CVS with object arguments in the MATRIX and COMPLEMENT………… 121 5.9. CVS with object arguments in the MATRIX……………………………….. 122 5.10. CVS with object arguments in both the MATRIX and COMPLEMENT….. 122 5.11. CVS with object arguments in the COMPLEMENT………………………. 123 5.12. CVS with direct object arguments in the COMPLEMENT……………….. 124 5.13. CVS with indirect object arguments in the COMPLEMENT……………… 125 5.14. CVS with direct and indirect object arguments in the COMPLEMENT….. 125 5.15. CVS with imperfective MATRIX verbs……………………………………. 128 xi 5.16. CVS with perfective MATRIX verbs………………………………………. 128 5.17. CVS with VOLETI…………………………………………………………. 129 5.18. CVS with ZAVOLETI………………………………………………………. 129 5.19. CVS with past tense MATRIX…………………………………………….. 130 5.20. CVS with present tense MATRIX…………………………………………. 130 5.21. CVS for adjectives…………………………………………………………. 134 5.22. CVS for adjectives without entity…………………………………………. 136 5.23. CVS for adjectives with entity……………………………………………… 136 5.24. CVS for nouns……………………………………………………………… 137 5.25. CVS for nouns without a specific entity……………………………………. 138 5.26. CVS for nouns with a specific entity……………………………………….. 138 5.27. CVS for bare nouns………………………………………………………… 140 5.28. CVS for nouns in expressions………………………………………………. 140 5.29. CVS for bare nouns without a specific entity………………………………. 140 5.30. CVS for bare nouns with a specific entity………………………………….. 141 5.31. CVS for expressions without a specific entity……………………………… 141 5.32. CVS for expressions with a specific entity…………………………………. 141 5.33. CVS for verbs………………………………………………………………. 143 5.34. CVS for TREBATI………………………………………………………….. 144 5.35. CVS for depersonalized verbs……………………………………………… 145 5.36. CVS for personalized verbs………………………………………………… 146 5.37. Overall CVS………………………………………………………………… 148 5.38. CVS for women…………………………………………………………….. 149 xii 5.39. CVS for men……………………………………………………………….. 149 5.40. CVS for high school………………………………………………………... 151 5.41. CVS for college…………………………………………………………….. 151 5.42. CVS for university………………………………………………………….. 151 5.43. CVS for 21-25………………………………………………………………. 152 5.44. CVS for 26-40……………………………………………………………… 152 5.45. CVS for 41-66……………………………………………………………… 153 6.1. Semantic Nuances…………………………………………………………….191 6.2. CVS for the Yβ controller……………………………………………………. 195 xiii LIST OF FIGURES Figure Page 1. CVS trend with object arguments……………………………………………… 126 2. CVS trend with adverbials, object arguments and additional linguistic material 132 3. CVS trend for adjectives, nouns and verbs……………………………………. 147 xiv LIST OF ABBREVIATIONS pparticiple – past participle present – present tense future – future tense infinitive – infinitive verb form – masculine gender M F – feminine gender N – neuter gender SG – singular PL – plural 1ST – first person 2ND – second person 3RD – third person N – nominative case G – genitive case D – dative case A – accusative case I – instrumental case NEG – negation REFLEXIVE – reflexive particle xv CHAPTER 1 INTRODUCTION The goal of the present study is to offer new insights into the phenomenon referred to here – in a very broad sense – as complement variation in Serbian (hereafter CVS) and consequently to provide as thorough an account as possible of CVS. Significant instantiations of CVS are presented in (1)-(9) below. It should be noted that, in this introduction, instantiations of CVS serve purely as an illustration of CVS, which, as this study demonstrates, proves to be much more complex than the given examples might suggest at first. There are various criteria according to which the examples presented below could be grouped together. I group them based on the type of the matrix predicate head, that is whether the head of the matrix predicate is a verb, an adjective, or a noun. Examples (1)-(5) all have a different verb as the head of the matrix predicate: 1 (1) a. Mogao je can.pparticiple.M.SG be.present.3RDSG b. Mogao je can.pparticiple.M.SG be.present.3RDSG uraditi i do.infinitive also da that više. more uradi do.present.3RDSG i also više. more ‘He could have done even more.’1 (2) a. Prestala je stop.pparticiple.F.SG be.present.3RDSG b. Prestala je da stop.pparticiple.F.SG be.present.3RDSG ‘She stopped crying.’ (3) a. Pokušao try.pparticiple.M.SG b. Pokušao try.pparticiple.M.SG ‘He tried to solve it.’ it.A plakati. cry.infinitive plače. that cry.present.3RDSG ga je be.present.3RDSG je be.present.3RDSG da that rešiti. solve.infinitive ga it.A (4) a. Ranije nisu dozvoljavali earlier be.present.3RDPL.NEG allow.pparticiple.M.PL gledati ovaj watch.infinitive reši. solve.present. 3RDSG omladini youth.D film. this.A movie.A b. Ranije nisu dozvoljavali RD earlier be.present.3 PL.NEG allow.pparticiple.M.PL omladini youth.D da that gleda ovaj film. watch.present.3RDPL this.A movie.A ‘They did not let the youth watch this movie before.’ 1 In this study, all translations from languages other than English are mine unless specified otherwise. 2 (5) a. Naučiće ih deliti složenu rečenicu. RD teach.future.3 SG/PL they.A parse.infinitive complex.A sentence.A b. Naučiće ih da teach.future.3RDSG/PL they.A that dele parse.present.3RDPL složenu complex.A rečenicu. sentence.A ‘He/She/They will teach them to parse the complex sentence.’ Examples (6) and (7) present instantiations of CVS with adjectival heads of the matrix predicate: (6) a. Lako easy je be.present.3RDSG b. Lako je easy be.present.3RDSG ‘It is easy to command.’ (7) a. Teško difficult mi I.D zapovedati. command.infinitive da that se REFLEXIVE je be.present.3RDSG b. Teško mi je difficult I.D be.present.3RDSG crime.A ‘It is difficult for me to admit to a crime.’ zapoveda. command.present.3RDSG priznati admit.infinitive da that zločin. crime.A priznam zločin. admit.present.1STSG Finally, in (8) and (9) the heads of the matrix predicate are nouns: (8) a. Sramota je shame be.present.3RDSG govoriti speak.infinitive b. Sramota je da RD that shame be.present.3 SG way ‘It is shameful to speak that way.’ se REFLEXIVE 3 tako. that way govori tako. RD speak.present.3 SG that (9) a. Prilika opportunity mi I.D je be.present.3RDSG b. Prilika mi je be.present.3RDSG opportunity I.D ‘It is an opportunity for me to see Paris.’ videti Pariz. see.infinitive Paris.A da that vidim see.present.1STSG Pariz. Paris.A While an analysis of the nature of the matrix predicates turns out to be an important element in the overall consideration of CVS, it is the complements of those matrix predicates that demonstrate the actual phenomenon. Generalizing over all nine pairs of sentences presented above, and using the most common account of verb forms in Serbian, it is clear that a matrix predicate allows either a non-finite complement (in all examples in a.), that is, a complement headed by a verb not inflected for tense, grammatical person and number, or a finite complement (in all examples in b.), that is, a complement headed by a verb inflected for tense, grammatical person and number. The non-finite complement is exclusively headed by an infinitive, a non-finite verb form in Serbian, whereas the finite complement is headed by a present tense form, a finite verb form in Serbian, invariably introduced by a complementizer da ‘that’ and, at the same time, in full grammatical agreement in person and number with the matrix predicate. There are other differences among the complements in examples given in (1)-(9) as well. In (3a), enclitic ga ‘it.A,’ an argument of rešiti ‘to solve,’ is positioned right after pokušao ‘try.pparticiple.M.SG’ due to its enclitic nature: in Serbian, enclitics cluster in the second position. In (3b), the same enclitic is in the second position within the complement, which indicates that da ‘that’ serves as a barrier for so-called clitic-climbing. Also, in (6b) and (8b), the reflexive particle se ‘REFLEXIVE’ appears in the finite 4 complement, while it does not exist in the non-finite complement. This is required by the generic nature of the matrix, which, in turn, requires the same generic nature in the complement, achieved in a finite complement by the insertion of se ‘REFLEXIVE.’ Neither of those two structural differences between the complements in a. and complements in b. in any way causes a change in meaning. In each pair of sentences from examples (1)-(9), the meaning arguably remains unchanged despite the apparent change in the structure of the complement in general and the head of the complement in particular. It is this particular feature of Serbian syntax, occurring under the circumstances outlined above, that is here referred to as complement variation in Serbian, or CVS. And it is this particular complement variation in Serbian that is the focus of the present study.2 2 The change of the head of the complement does not seem to cause other structural changes in the complement either. In Serbian, predicate adjectives with verb biti ‘to be’ appear in the nominative case: i. a. On je dobar. he be.present.3RDSG good.N.M.PL ‘He is good.’ b. Oni su pristojni. they be.present.3RDPL polite.N.M.PL ‘They are polite.’ If the two are embedded as complements of CVS matrix predicates, the predicate adjective remains in the nominative case despite the change of the complement head: ii. a. Pokušao try.pparticiple je da be.present.3RDSG that bude dobar. be.present.3RDSG good.N.M.PL b. ?Pokušao je biti try.pparticiple be.present.3RDSG be.infinitive ‘He tried to be good.’ iii. a. Naučiću ih teach.future.1STSG they.A da that dobar. good.N.M.PL budu pristojni. be.present.3RDPL polite.N.M.PL b. ?/*Naučiću ih biti teach.future.1STSG they.A be.infinitive ‘I will teach them to be polite.’ pristojni. polite.N.M.PL With činiti ‘to make,’ a predicate adjective is in the instrumental case, 5 The phenomenon, as described above, represents an instance of control, a linguistic phenomenon said by Culicover and Jackendoff (2005:415), who outline one of the latest possible accounts of it, to have been “absolutely central to mainstream theory for forty years.”3 Despite this fact, I have been unable to find any kind of such reference to CVS in the works that I had encountered so far. Culicover and Jackendoff (2005:415) claim, when they speak about examples from English, that “[t]he problem of control concerns how to determine the understood subject of infinitival or gerundive VPs that lack an overt local subject.” Indeed, in examples (1)-(9) there are no overt subjects of either the infinitives or the present tense forms, yet all of the complements are interpreted as if they have one – it is always a matrix argument that is interpreted as the subject of the complement verb infinitive or present tense form. The instance of control represented in examples (1)-(9) above is more commonly referred to as obligatory control (in Landau 2000 and 2004, for example) or, less commonly so, as unique control (in Jackendoff and Culicover 2003 and Culicover and Jackedoff 2005). However, both terms, each in their own way, imply that there is one and iv. Ona me čini she I.A make.present.3RDSG ‘She makes me happy.’ srećnim. happy.I and it remains the same if iv. is embedded as a complement of a CVS matrix predicate: v. a. Nastavila continue.pparticiple.F.SG je da be.present.3RDSG that me I.A čini make.present.3RDSG srećnim. happy.I b. Nastavila je činiti continue.pparticiple.F.SG be.present.3RDSG make.infinitive ‘She continued to make me happy.’ 3 me I.A srećnim. happy.I For some of the latest discussions of the history of linguistic accounts of control, see Culicover and Jackendoff (2005) and Landau (2000). 6 only one possible interpretation of the subject of the complement: the complement is obligatorily, uniquely controlled by a matrix argument. Just as there are two ways of naming the phenomenon, there are at least two in which the phenomenon is indicated in a sentential configuration, as I demonstrate below using my own example sentences in (9), repeated here as (10) and (11). (10) Obligatory Control (based on Landau 2000 and 2004) a. Prilika mii je [Proi/*j videti Pariz.]4 b. Prilika mii je [Proi//*j da vidim Pariz.] ‘It is an opportunity for me to see Paris.’ ‘*It is an opportunity for me that you see Paris.’ (11) Unique Control (based on Jackendoff and Culicover 2003 and Culicover and Jackendoff, 2005) a. Prilika mii je [i/*jvideti Pariz.] b. Prilika mii je [da ividim/*jvidiš Pariz.] ‘It is an opportunity for me to see Paris.’ ‘*It is an opportunity for me that you see Paris.’ It should be noted that, as much as one’s linguistic nomenclature is arbitrary, it often reflects one’s position with respect to the analyzed phenomenon, or at least it should be to whatever extent possible. This is certainly true in the case of obligatory and unique control, as illustrated in (10) and (11). Landau’s (2000:129) position is that “infinitives in O[bligatory] C[ontrol] denote closed propositions, just like finite clauses do. PRO is 4 Landau (2004:n.13) uses Pro “[t]o be neutral on the nature of the controlled null subject” in the Balkan languages. This is his way, albeit temporary, of solving the problem of the controlled subject, whether that subject is PRO or pro. Conventionally, PRO is used to designate a missing and at the same time controlled subject, whereas pro is used to designate a missing but not controlled subject. 7 projected in the syntax and saturates the subject position of the infinitive.” Culicover and Jackendoff (2005) proclaim their neutrality by saying that they use their notation in an example such as (11) above “rather than the conventional null pronoun PRO so as not to prejudice whether the infinitive has a genuine syntactic subject.” Yet, the title of their Chapter 12 is anything but neutral, it reads: The Semantic Basis of Control in English. I provide here a novel explanation for the variation of complements in Serbian, an otherwise very well-known and long-documented phenomenon. That is to say, I view the phenomenon, a well-established fact in Serbian grammar, from the position of the latest views of control, more specifically obligatory, unique control, as a general linguistic phenomenon. My own account is based, in part, on some accounts made several decades ago, during a period when CVS was a focus of the Serbo-Croatian linguistic attention, and, yet in another part, on the latest views of control explicated in Culicover and Jackendoff (2005). After these introductory remarks, I comment, in Chapter 2, on what I refer to as language issues. In Chapter 3, I deal with preliminary considerations that are essential for the study of CVS. I analyze the long-standing tradition of scholarship regarding CVS and present my own view of that tradition in Chapter 4. In Chapter 5, I offer results of my original research which I use as the basis for all of my conclusions. Chapter 6 provides an explanation of the data collected during the research, thus accounting for CVS. I summarize all of my findings in Chapter 7. 8 CHAPTER 2 LANGUAGE ISSUES Before any attempt to explain CVS, several preliminary considerations are in order. All of them I gather under the title language issues. Namely, it is necessary to lay out the position taken in this study with respect to the language in question. This issue is – in a way – highly sensitive, mainly for non-linguistic reasons that arose in the last decade of the 20th century. By outlining my own position here, I hope to establish the parameters that serve as a starting point in the analysis of the focus of the study – the variation itself. Reasons other than linguistic ones demand that the question of the language in the study be addressed here. Paradoxically enough, by taking a stand on the language, I actually suppress all the potential non-linguistic questions that might arise with respect to the study, thus in fact providing more honest and precise linguistic reasoning about CVS. The two sections that follow address two major aspects regarding the language in question: section 2.1 deals with the use of the name of the language, and section 2.2 explicates my stand with respect to previous accounts of CVS as well as various reference sources for the language. Section 2.3 contains concluding remarks. 9 2.1. The Serbian Language The language on which I focus in the study is referred to as the Serbian language, where Serbian is taken as the official language of Serbia and Montenegro (country code SCG), as stated in LK: U SCG u službenoj upotrebi je srpski jezik ekavskog i ijekavskog izgovora. ‘Serbian, of the ekavian and ijekavian dialects (literally, pronunciations), is in official use in SCG.’ This statement implies that varying dialectal forms, namely ekavian and ijekavian, such as the ones presented in (1) below, are equally acceptable for official use. For clarity, I selected lexical items from the sentences given in examples (1)-(9) in Chapter 1: (1) Ekavian vs. (I)jekavian Serbian Forms5 a. rešiti vs. riješiti ‘to solve/resolve’ b. deliti vs. dijeliti ‘to parse/split/divide’ c. zapovedati vs. zapovijedati ‘to command/order’ d. videti vs. vidjeti ‘to see’ In my examples (1)-(9) in Chapter 1, only the first cited forms, the ekavian forms, are found, but not the ijekavian. Consequently, the question whether I, in fact, use Serbian in its totality in my analysis with all of its variants must be posed. The most honest answer 5 This particular dialectal division of the Serbian language is based on the reflexes of Common Slavic *ě, which in the Ekavian dialects gave e and in the (I)jekavian dialects gave either ije or je. 10 to this question seems to be a negative one, for the form of the Serbian language that I use in this study is predominantly ekavian. There are two types of examples used here. On the one hand, I use sentences from various reference works dealing with CVS or the language in question. On the other hand, and predominantly so, I use my own examples, which are often based in part on what I have found in reference works, but also on my own native command: the ekavian Serbian dialect of Belgrade, the capital of Serbia and the administrative center of Serbia and Montenegro, which is at the same time my native dialect and the focus of my linguistic interest. I had easy access to that particular dialect and made it the center of my attention. It is, thus, always of this dialect that I speak here, and my findings reflect what is detected in this dialect only. An argument could be made that, besides the personal reason, there is also a strong sociolinguistic reason for my choice of dialect. The ekavian Serbian dialect of Belgrade is a vernacular very close, if not the closest one, to what might be considered the standard ekavian Serbian language, and – in turn – it is the dominant variant in the majority of the territory of Serbia among the Serbian population (and often among other ethnic groups, though this varies depending on what ethnic group and what part of the territory of Serbia are under analysis). For this reason, although my findings are based on one specific dialect only, the results of my study could easily be applied to a territory larger than that of the original dialect, although admittedly some minor modifications may be necessary. All of my own examples in the study thus contain forms of the ekavian Serbian dialect of Belgrade. However, I occasionally use examples with ijekavian Serbian forms 11 and from various reference works because of a tradition associated with the Serbian language. A careful reader should quickly notice that the only definition of the Serbian language provided thus far has little, if anything, to do with linguistics. Serbian, as defined here, is spoken in Serbia and Montenegro, a country located on the Balkan Peninsula. Serbian is one of the South Slavic languages. Other modern languages belonging to the South Slavic language group include: Slovenian, Croatian, Bosnian (or, both Bosniac and Bosniak)6, Macedonian, and Bulgarian. For the major part of the 20th century, however, the language of Serbia and Montenegro, then part of the former Yugoslavia, was referred to in English as Serbo-Croatian (also Serbo-Croat, Serbocroatian, Croato-Serbian), a term which used to denote an official language of the Socialist Federative Republic of Yugoslavia. Its place is now taken by Croatian, Bosnian (or Bosniac/Bosniak), and Serbian, three new languages in place of a single old one. There is an ongoing discussion regarding whether there are any other languages, besides the six named above, which also belong to the group of the living South Slavic languages (Montenegrin is the first one that comes to mind). Curiously, Bosnian (or Bosniac/Bosniak), Croatian, and Serbian currently are often referred to as BCS in American linguistic usage (but not only American), the motivation for which seems to be as much practical as it is linguistic. Be that as it may, all of what I just briefly sketched has various implications for the target language in the present study: the way the language is treated, as well as the range of the conclusions the study will eventually allow. 6 For various and varying arguments see Neweklowsky (2003). 12 In this study I adopt the following view towards the relevant South Slavic languages: Croatian is the official language of Croatia, Bosnian is the official language of Bosnia and Herzegovina, and Serbian is the official language of Serbia and Montenegro, as they are, indeed, considered to be in their respective countries. At the same time, I cannot avoid the fact that there has been a relatively long convergent linguistic tradition among these currently separate official languages, which resulted in numerous works on what used to be called Serbo-Croatian. And just as I cannot avoid the fact about the converging past, I cannot and do not avoid the works published during such past. It is in them that one finds, besides ekavian examples (of what in the times of Serbo-Croatian used to be called – more loosely – the eastern variant of Serbo-Croatian, or – less often – the Belgrade standard), ijekavian examples (of what in the times of Serbo-Croatian used to be called – more loosely – the western variant of Serbo-Croatian, or – less often – the Zagreb standard) as well. Such examples are used here with caution so as not to imply anything inaccurate or even offensive. 2.2. Sources The relatively long Serbo-Croatian linguistic tradition resulted in significant research dedicated to various aspects of the grammar of the language. One part of this research was also devoted to what is here referred to as CVS. Works that dealt with CVS must, therefore, play an important role in the present study. Before I include them in my discussion, however, I need to address an issue that might arise from the fact that those works view the language question differently from the present study. 13 In the vast majority of the works, the language in question is referred to as SerboCroatian. I, therefore, have to approach them with great care. I find any work on CVS to be invaluable, yet I extract and use only those ideas that explain CVS in what used to be referred to as the eastern variant, or Belgrade standard of Serbo-Croatian. In so doing, I should stay closer to my own position towards the language in question. This is not to say that ideas that make reference to various other standards of Serbo-Croatian are of no use whatsoever; on the contrary, I value all the explanatory suggestions that might be found in all the works dedicated to CVS. It becomes clear later on in this study that some of these suggestions served as a partial inspiration for what I propose here. When I refer to works treating the Croatian standard language or certain Croatian dialects, it is not to imply that I believe that they also represent parts of the Serbian standard language – I have no intention to make claims of that kind whatsoever. My choice of ideas pertaining to CVS in Serbo-Croatian is not determined by the location of the publisher of the work that I consult or the nationality of a researcher. It is determined by linguistic criteria that closely coincide with my own linguistic views of CVS. 2.3. Conclusion By way of concluding this chapter on the language issues, I offer a summarized version of what the interpretation of the term Serbian language means in this study. I deal with Serbian as the official language of Serbia and Montenegro; more precisely, I analyze a dialect of Serbian spoken in the Belgrade area – a vernacular closest to the standard ekavian Serbian language. Since the Serbian language until recently belonged to what 14 used to be called the Serbo-Croatian language, I have to refer to various sources that might make claims about CVS in dialects outside of my linguistic interest at this particular moment. Still, I use them only to the extent that they provide helpful insights with respect to CVS. 15 CHAPTER 3 CONFIGURATION OF CVS It was indicated in the very beginning of this study that the perception of CVS may vary in certain respects. I, for example, choose to view it as a phenomenon occurring in complements of configurations with verbal, adjectival, or nominal matrix predicates, as is generically formalized in (1). (1) a. [MATRIX (Xα, [COMPLEMENT(α)])] b. Xα MATRIX [α COMPLEMENT] This formalization is based on Culicover and Jackendoff’s (2005) view of control, where (1a) represents what they refer to as a standard notation, whereas (1b) stands for what they refer to as a simplified notation. In either representation, however, at this point in the study I use MATRIX and COMPLEMENT in bold in order to indicate the generic character of the two essential elements of the configuration. This is quite different from Culicover and Jackendoff’s (2005) account in which they simply use the actual predicates 16 in their formalizations. Everything else, however, is the same: first, argument positions and their selectional restrictions are in italics, and, second, the apparent connection between the binder, occurring with the matrix predicate, and the bound position, an argument position of the complement, is notated by a Greek variable. From this point on, I use Culicover and Jackendoff’s simplified notation, notably (1b), in all of my formalizations. In this chapter I first argue for the configuration of CVS such as the one presented in (1), more precisely (1b), above. Then, in section 3.2, I deal with the element of the configuration referred to here as MATRIX, both in details and in general. In section 3.3, I address questions pertaining to the element of the configuration referred to here as COMPLEMENT. Finally, section 3.4 summarizes my ideas presented in this chapter. 3.1. Configuration The formal representation of the configuration in which CVS is observed is one possible choice made among many candidates for a formal account of the phenomenon. The choice, however, is neither random nor accidental, nor – unfounded; on the contrary, it is based on several different considerations of the phenomenon. 17 3.1.1. Essentials The formalization in (1b) closely matches what is generally observed regarding CVS, as is demonstrated in (2), (3), and (4) below, where, for convenience, I repeat (1a,b), (7a,b), and (9a,b) from chapter 1, respectively. (2) a. Mogao je can.pparticiple.M.SG be.present.3RDSG b. Mogao je can.pparticiple.M.SG be.present.3RDSG uraditi i do.infinitive also da that više. more uradi do.present.3RDSG i also više. more ‘He could have done even more.’ (3) a. Teško mi difficult I.D je be.present.3RDSG priznati admit.infinitive b. Teško mi je da RD difficult I.D be.present.3 SG that ‘It is difficult for me to admit to a crime.’ (4) a. Prilika opportunity mi I.D je be.present.3RDSG b. Prilika mi je opportunity I.D be.present.3RDSG ‘It is an opportunity for me to see Paris.’ zločin. crime.A priznam zločin. ST admit.present.1 SG crime.A videti Pariz. see.infinitive Paris.A da that vidim see.present.1STSG Pariz. Paris.A To all three pairs of sentences, the following formalization applies; or rather, the formalization is a result of what is observed in this particular sample of sentences: 18 (5) Xα MATRIX [α COMPLEMENT] This formalization is assumed to be the initial formalization for the configuration in which CVS takes place. During the course of this study the formalization itself will undergo several modifications in order to closely account for various aspects of CVS that are yet to be presented. However, the whole process of modifying the formalization along the lines of what is observed in the language data leads to a formula that easily accounts for CVS. The following chart explicates what each element of the formula entails with respect to examples (2)-(4): (2) a. X α b. (4) // // a. /mi/ b. /mi/ a. /mi/ b. /mi/ 3 RDM.SG 3 RDM.SG 1STSG 1STSG 1STSG 1STSG /mogao je/ 3RDM.SG /da uradi i više/ /teško mi je/ 1STSG /priznati zločin/ /teško mi je/ 1STSG /da priznam zločin/ /prilika mi je/ 1STSG /videti pariz/ /prilika mi je/ 1STSG /da vidim pariz/ /mogao je/ 3RDM.SG α COMPLEMENT /uraditi i više/ MATRIX (3) Table 3.1. Elements of the formula vis-à-vis sentences (2)-(4) In the chart, slashes are used in the phonological sense – the material between slashes is phonemic material found in the sentences. For matrix arguments of sentences (2a,b), there is no phonemic material, yet there is still enough information contained in the matrix for the formula to apply – matching of the information associated with the Greek 19 variable in the matrix and the same variable in the complement is what accounts for the obligatory or unique control in (2a,b). The same type of information matching is found for sentences (3a,b) and (4a,b); hence the formula applies as well. It should be noted that there is nothing in infinitival complements in sentences (2a), (3a), and (4a) that provides the actual information said in the chart to be associated with α. The same is true of the da+present complement in (2b) – while α is clearly the 3rd person singular, it is not clear that the grammatical gender of the argument in the complement is indeed masculine. Rather, the grammatical information associated with α in the complement is gathered based on inference – it is the interpretation of the whole configuration that invariably detects features associated with α. 3.1.2. Motivation One other reason to employ the formula in (5) comes from an important discovery in the language data obtained from two different sources: research that I conducted with a representative group of native speakers and an electronic library of Serbian culture – the “Rastko” project. Although this research is laid out in detail in chapter 5, I present here a preview of it, so to speak, as an argument in favor of (5). In my research, all the participants were faced with pairs of sentences, much like the ones in (2)-(4) above, with the task to choose one member of the pair that they would prefer to use in their everyday communication. They were also allowed to choose both sentences if they felt that they would use them with equal probability. When native 20 speakers were making a choice between the sentences such as (3a) and (3b) above, repeated here as (6a) and (6b), respectively, (6) a. Teško mi difficult I.D je be.present.3RDSG priznati admit.infinitive b. Teško mi je da difficult I.D be.present.3RDSG that ‘It is difficult for me to admit to a crime.’ zločin. crime.A priznam zločin. admit.present.1STSG crime.A they would overwhelmingly select (6b), the one with a da+present complement. However, when faced with the following pair of sentences: (7) a. Teško je difficult be.present.3RDSG b. Teško je difficult be.present.3RDSG crime.N priznati admit.infinitive zločin. crime.A da that prizna admit.present.3RDSG se REFLEXIVE zločin. crime.N ‘It is difficult to admit to a crime.’ the participants would, again overwhelmingly, choose (7a), the one with an infinitival complement. When an argument of the matrix, being at the same time the binder of the complement, was clearly specified, as it is in (6a,b) above with a dative noun phrase, the choice of the complement was da+present. However, when an argument of the matrix was not clearly specified, as in (7a,b) above, the choice was an infinitive. This proved to be a pattern, rather than an anomaly, in the language data I gathered during the research. The presence or absence of an argument in the matrix appeared to force the choice of a 21 complement structure in CVS, suggesting that the variation between da+present and the infinitive is governed by the syntactic structure of whatever the matrix may be. It is, thus, necessary to provide a formalization which allows for an accurate account of the observation discussed above; such formalization is precisely the one given in (5). Another support for this line of reasoning comes from a different source. The “Rastko” project, an electronic library of Serbian culture, contains texts on history, archeology, anthropology and ethnology, linguistics and philology, religion, philosophy, literature, visual arts, performing arts, film and television, photography, comic strips, and music. While it was primarily envisioned as a source of various ideas pertinent to all interested parties from various walks of life and expressed in Serbian, it is also a useful source of information about the Serbian language itself. A search of the corpus gave similar results to the ones outlined above with respect to the research with native speakers. Here, I provide statistics for sentences from the “Rastko” project which actually contain the same matrix predicate as sentences in (6) and (7) above, namely teško ‘difficult.’ In CVS configurations with the binder of the complement present in the matrix, I found many more sentences with a da+present complement (P in the table) rather than an infinitival complement (I in the table): TOKEN # % I 19 41.30% P 27 58.70% Table 3.2. CVS Statistics with the Binder 22 TOTAL 46 100% On the other hand, when the binder of the complement was absent from the matrix, I recorded an overwhelming percentage of sentences with an infinitival complement: TOKEN # % I 332 99.40% P 2 0.60% TOTAL 334 100% Table 3.3. CVS Statistics without the Binder In both sources of the language data I discovered essentially one crucial factor which was responsible for the variation of the two complements: the presence or absence of the binder in the matrix. Moreover, depending on the presence or absence of the binder, the da+present or infinitival complement was selected respectively, the two being in complementary distribution. This discovery requires a formula that could indicate such relationship between the matrix and the complement and such formula is the one given in (5) above. 3.1.3. Place of Infinitive One last consideration in this section on the CVS configuration concerns a property of one of the two possible CVS complements – the infinitive. It is well known that the Serbian infinitive can appear in the subject position in a sentence. If it turns out that this infinitive also alternates with da+present in the same way it does in sentences given in 23 (2)-(4) above, this would present a serious challenge for the formula offered in (5) above. I demonstrate here that this in fact is not a problem for the CVS configuration. In Serbian, the infinitive, which is one of the two alternating elements in CVS, can be found in the position of the subject of a sentence, as in (8) below from Arsenijević (1997). (8) Gledati televiziju je neizbežno. unavoidable watch.infinitive television.A be.present.3RDSG ‘To watch TV is unavoidable/It is unavoidable to watch TV.’ Similar claims are also made by Maretić (1963), Stanojčić and Popović (1992), Simić (2002), Simić and Jovanović (2002), as well as by Ivić (1972), Bibović (1976), and Joseph (1983:137), who claims that the infinitive, among other positions, also occupies the position of a “clausal subject of a sentence.” Arguably at the opposite extreme are Novaković’s (1902) and Đukanović’s (1986) accounts of the infinitive. Novaković (1902:319) claims that, “when an infinitive is an added sentence [complement] to a name [noun] or adverb, then the infinitive explicates the action or being for which the MAIN CLAUSE states what the action or being is like or what they are” (emphasis added, BB). Đukanović (1986) makes a claim about examples configurationally similar to the one in (8) above saying that their structure is as follows: (9) Đukanović’s (1986) account Adv/N + Cop + Inf 24 In his formalization, Adv stands for adverb or adverbial phrase, N for noun, Cop for copula, copulative verb, and Inf for infinitive. It is crucial at this point to determine whether an infinitive that occupies the position of the subject in a sentence represents an element of CVS or not; that is to say, whether it alternates with a da+present structure in the same way it does in sentences given in (2)-(4) above. Recall that Arsenijević (1997) claims that an infinitive indeed occupies the subject position, as in (8) above, repeated in (10a) below. Moreover, Arsenijević asserts that such infinitive has its da+present alternant as in (10b). (10) a. Gledati watch.infinitive televiziju television.A b. Neizbežno je unavoidable be.present.3RDSG je be.present.3RDSG neizbežno. unavoidable da that gleda watch.present.3RDSG se REFLEXIVE televizija. television.N ‘To watch TV is unavoidable/It is unavoidable to watch TV.’ Arsenijević provides no evidence that the da+present structure in (10b) is indeed in the position of the subject as the infinitive in (10a) is. He refers to the variation between gledati televiziju and da se gleda televizija ‘to watch TV’ as a transformation, which, in his account, implies “a change in some … elements of the syntactic model” (in this particular example, televiziju ‘TV.A’ changes to televizija ‘TV.N’). In addition, he provides an example in which the same position is occupied by a verbal noun, apparently a second possible alternant of an infinitive in the subject position: 25 (11) Gledanje televizije je watching television.G be.present.3RDSG ‘Watching TV is unavoidable.’ neizbežno. unavoidable What I believe is crucial to notice is that gledati televiziju (in 10a) and da se gleda televizija ‘to watch TV’ (in 10b) in fact do not occupy the same position: the former is in the initial position of the sentence, the latter in the final. Moreover, positioning the latter in sentence-initial position renders a rather unusual sentence in Serbian, a language generally considered to have free word order: (12) ??? Da se that REFLEXIVE gleda televizija RD watch.present.3 SG television.N je be.present.3RDSG neizbežno. unavoidable ‘To watch TV is unavoidable/It is unavoidable to watch TV.’ Now, it appears that the actual order of gledati televiziju (in 10a) and da se gleda televizija ‘to watch TV’ (in 10b) crucially differentiates the two configurations. Also, it is not quite clear that the implications of the meaning of the sentence in (12) are the same as the implications of the meaning of the sentence in (10a): (10a), with gledati televiziju ‘to watch TV,’ clearly utters one’s overall attitude about the event of watching TV, while (12) does not accomplish the same – da se gleda televizija ‘to watch TV,’ does not depict the event of watching TV in the same way, at least not in the configuration as in (12). The sentence in (12) is closer in meaning and event depiction to (10a) in the following configuration: 26 (13) Da that se REFLEXIVE gleda televizija, RD watch.present.3 SG television.N to that je neizbežno. unavoidable be.present.3RDSG ‘To watch TV, that is unavoidable.’ In (13), however, da se gleda televizija ‘to watch TV’ stands in apposition to the demonstrative to ‘that,’ which is here an anaphoric pronoun to, in the sense of Ivić (1972), or else the event pronominal to, in the sense of Progovac (1998). Ivić (1972:134) also accounts for the difference between (10a) as opposed to (12), using her own examples, by saying that what she refers to as the subjectival infinitive “could not be moved [from its original final position] to the initial position without a drastic transformation of the sentence;” placing a subordinate (or dependent, in Ivić’s terms) clause in the initial position does not obey the rules. I, however, take this to imply that the configurations of (10a) and (12) are clearly different and it does not seem to be the case that, when the infinitive is in the subject position, as in (10a), that it easily alternates with da+present, as in (12) or (13). This position apparently does not allow for CVS as it has been defined so far in this study. Furthermore, it seems that a much better equivalent, or alternant – if the goal of one’s analysis is to find as many possible alternants of a structure, as Arsenijević’s (1997) goal seems to be – for an infinitive in (10a) is a verbal noun, as in (11) above. Thus, word order crucially determines exactly which infinitive is in the position in which CVS occurs and which one is not. The one that is in the subject position does not participate in CVS or else CVS takes place in the configuration as given in (5) above. 27 Bibović (1976) makes similar claims to Arsenijević (1997). I indicated that what is crucial to observe in (10a,b) above is word order. In (10a), the infinitive is in the initial position, the one normally occupied by a subject of a sentence in Serbian. The said equivalent of the infinitive, da+present structure, in (10b), nevertheless, occupies the final position of the sentence. Just as the word order in (10a) appears to be easily acceptable, so is the order in (10b). Arsenijević (1997) does not comment on that, but Bibović (1976:7) does, naturally for her own examples. When explaining the varying word order, she states that “[t]he difference between the two word orders may be explained by the different distribution of communicative dynamism (=CD).7 In, let us say, (14) Samovati je tužno8 RD sad lead a lonely life.infinitive be.present.3 SG ‘To lead a lonely life is sad/It is sad to lead a lonely life.’ both the infinitive as subject and the predicate carry the same degree of CD. But in (15) Tužno je samovati9 sad be.present.3RDSG lead a lonely life.infinitive ‘It is sad to lead a lonely life.’ the infinitival subject, samovati, carries a lower degree of CD than the preposed predicate tužno je, the subject being the topic (theme), the predicate being the comment.” 7 Bibović (1976:n9) explains communicative dynamism as follows: “By the degree of CD carried by a linguistic element is meant the extent to which the element contributes to the development of communication, to which, as it were, it ‘pushes the communication forward.’” 8 Bibović’s (1976) example (12SC). 9 Bibović’s (1976) example (13SC). 28 Ivić (1972) also employs the tool of communicative dynamism in her account of the infinitive occupying either the initial or the final position in a sentence. She, however, claims, unfortunately with no statistics to support her claim, that the infinitive is found in the final position more frequently than in the initial position. Ivić states (1972:133): “One should probably look for the reason for this distribution in the fact that the infinitive here is of the secondary origin, it is a transform of the dependent (subordinate) clause, and a subordinate clause converges towards the final position, while the initial position is used as a solution in the case of inversion.” Ivić also mentions that some syntacticians, in particular Stevanović (1970), give a special importance to the distribution of the infinitive as in (14) and (15). One portion of Stevanović’s (1970) description of the infinitive is dedicated to its free use, as he terms it, when the infinitive is functioning as a subject of a sentence. Stevanović claims that he is merely extending a line of reasoning, previously taken by Belić and Maretić (with no particular references mentioned), as well as authors of most recent grammar books (again, no particular references are given). One of his examples is given in (16) below. (16) Pušiti je zabranjeno. forbidden smoke.infinitive be.present.3RDSG ‘To smoke is forbidden/It is forbidden to smoke.’ In (16), the infinitive pušiti ‘to smoke’ functions as the subject. Stevanović (1970:769) claims that this example, as well as the other examples that he provides allow for one to see that the infinitive “becomes the subject of a sentence when the action signified by an 29 infinitive is to be positioned in the center of the utterance.” He continues by saying that all of his examples, (16) included, can be uttered with a different order, as demonstrated in (17). (17) Zabranjeno je forbidden be.present.3RDSG ‘It is forbidden to smoke.’ pušiti. smoke.infinitive He, however, concludes that “this time, the sentences are uttered in a different tone, their components are given in a different order, and most importantly, with different syntactic functions. In them, the infinitive functions as the COMPLEMENT of impersonal expressions” (emphasis added, BB), one of which is zabranjeno je ‘it is forbidden’ from (16) and (17) above. Implicitly, Stevanović makes the exact same claim regarding the importance of word order as I also advocate in this section. The position of the infinitive in a sentence crucially determines whether it is eligible for CVS or not. Besides being important for the present study, Stevanović’s claim is important for considerations of the Serbian language in general, in that it implies that word order in Serbian is not all that free as it is usually considered to be. Consideration of Serbian word order, however, including a proposal that it is not all that free, albeit extremely interesting in its own right, extends beyond the scope of the present study. An argument of a similar, if not the same kind – and, unfortunately, the same approach with little or no syntactic evidence – is presented in Stanojčić and Popović (1992:396). There, they simply claim that in (18), 30 (18) Dosadno boring mi I.D je be.present.3RDSG o svemu about everything tome that govoriti. speak.infinitive ‘It is boring to me to speak about all that.’ the infinitive is used as a complement of an “expression with an adjective with incomplete meaning,” whereas in (19), (19) Razgledati look at.infinitive izloge je department store windows.A be.present.3RDSG zanimljivo. interesting ‘It is interesting to look at department store windows.’ it appears as the subject of the sentence. Stanojčić and Popović provide no further explanation of their examples. My understanding, however, of “an adjective with incomplete meaning” stems from the fact that dosadno has two potential meanings: ‘boring,’ as an adverb and ‘bored/boring,’ as a neuter adjective. In the latter case, dosadno is not associated with any specific masculine or feminine real world entity that might be depicted as ‘bored/boring,’ hence it is ‘an adjective with incomplete meaning’ in Stanojčić and Popović’s interpretation. 31 3.1.4. Summary I conclude this section on the configuration in which CVS occurs by summarizing my claims presented above. I propose that CVS is exclusively observed in the complement component of a configuration that, in addition to the complement, contains matrix component as well, as it is easily observed in example sentences that I provide. CVS occurs only when the complement is obligatorily or uniquely controlled by a matrix argument, which clearly influences the choice of the complement structure. It appears to be the case that the infinitive in Serbian may occupy positions other than of the complement. It is suggested in this study that CVS does not take place in these other positions. 3.2. Matrix It has been indicated previously in this study that the character of the matrix predicates participating in CVS may be one of three types: they can be verbal, adjectival, or nominal. As it turns out, the arguably long-lasting tradition of accounting for CVS did not actually provide a thorough and exhaustive enough account of what exactly can serve as a matrix predicate in CVS. On the one hand, it may be that there are simply no grounds for such an account, for the possible matrix predicates might turn out to be characterized by quite different linguistic parameters. On the other hand, however, it may be that not all the necessary factors have been taken into consideration, which has resulted in varying degrees of agreement about matrix predicates in CVS. In this section I provide an 32 overview of possible CVS matrix predicates. I examine each of the three possible categories of matrix predicates in CVS separately, namely nouns, adjectives, and verbs, and, at the same time, compare several previous accounts in which the same or similar issues have been addressed. First, however, I would like to outline the basic approach to defining the matrix predicates assumed in the present study. There are numerous ways in which matrix predicates participating in CVS can be accounted for. During the long tradition of analyzing CVS, however, essentially two methods clearly emerged, the two that I here loosely term a syntactic approach and a semantic approach, the latter admittedly the more prevalent one. The syntactic approach is one in which the predicate’s attachment to the group of CVS matrix predicates is established based on its subcategorizational properties or else its selectional restrictions. That is to say, the matrix predicates are defined according to the arguments that they (might) take. The semantic approach is one in which semantic characterization of CVS matrix predicates is sought, without necessarily reaching, or even searching for, a point of generalization over all of the established semantic categories. A rare example of the syntactic approach towards defining CVS matrix predicates is found in Đukanović’s (1986) analysis. The author predominantly deals with verbs as matrix predicates and begins by classifying such verbs into essentially two categories: “1. Verbs that exclusively must have an infinitive/da+present as the complement … 2. Verbs that need not have an infinitive/da+present as the complement, but may …” For the sake of accuracy, I should mention that Đukanović actually recognizes three categories, of which his second and third I present here under 2., because the difference between his 33 last two categories lies in the complement and not the matrix. It is clear, however, that his initial position is syntactic – it is the type of argument the verb takes that, first, places the verb in the group of CVS matrix predicates, and, second, determines its category within the group of CVS matrix predicates. The major problem with this approach is that the theory makes little or no prediction regarding possible matrix predicates. The approach towards defining CVS matrix predicates most often used is the semantic approach. In all of the works regarding the semantic approach, the authors provide various classifications of lexical units independently determined to be the possible matrix predicates. These classifications, however, are often supported with vague explanations or, in extreme cases, with no explanations whatsoever, as I will demonstrate below. And none of them makes an attempt to determine a general semantic parameter, if such a parameter exists, that might unify several different semantic parameters said to be relevant for CVS matrix predicates. Still, theories based on the semantic approach provide a certain level of prediction otherwise nonexistent in theories based on the syntactic approach, albeit this may not have been a primary goal of the former theories. In this study, I adopt a semantic approach towards defining all three types of CVS matrix predicates, which I discuss in Chapter 6. 3.2.1. Nouns Although there are differences among scholars with respect to the question of just what can be a matrix predicate in CVS, and rightly so, there is almost a unanimous agreement 34 that one type of such predicate contains nouns. Setting aside this issue, the major problem is determining what kinds of nouns can serve as CVS matrix predicates. Those scholars who address nouns in their accounts of CVS usually approach them from the syntactic, rather than the semantic point of view. To exemplify this, I again quote Đukanović (1986), who simply states that “minimal syntactic constructions … contain two elements that circle around the central (syntactic) particle (or, nucleus) – the copula,” and then he provides two possible formulae: (20) a. N + Cop + To ↔ Inf b. N + Cop + (To ↔) Inf In the formulae, N represents nouns, Cop – a copula or a copulative verb, and Inf – the infinitive; To is the same lexeme earlier referred to as an anaphoric pronoun or an event pronominal. It is quite interesting to note that the formulae make false predictions by implying that all nouns can assume the position of N, which is not quite so. A very common noun such as knjiga ‘book’ cannot, for example, be found in a configuration formed based on either of the formulae: (21) *Knjiga je (to) čitati. be.present.3RDSG that read.infinitive book.N ‘The book is to read./It is the book to read.’ Such is the case with many other nouns. Other accounts of CVS matrix predicates, otherwise predominantly semantic in nature, nonetheless resort to the syntactic approach 35 when it comes to nouns. Compare, for example Brabec et al. (1968) and Ivić (1970), which are similar to the example in Đukanović’s (1986) analysis. Ivić (1970:51), however, claims that she makes an attempt to explain the nouns in CVS matrix predicates from a somewhat semantic standpoint. She claims that there are at least two major groups of nouns that participate in CVS as matrix predicates: 1. nouns most frequently derived from verbs (želja ‘desire,’ zahtev ‘request,’ pokušaj ‘attempt,’ molba ‘appeal,’ moć ‘power,’ predlog ‘suggestion,’ savet ‘advice,’ naredba ‘command’), and 2. native nouns and borrowed nouns for which the presence of an abstract verb is presupposed (volja ‘willingness,’ napor ‘effort,’ svest ‘consciousness’; impresija ‘impression,’ intencija ‘intention,’ ideja ‘idea’). Clearly, whatever the criteria, her account diverges from a strictly semantic explanation; even more so after the application of the theoretical tool that she labels ekspektativnost Exp ‘expectation.’ The implication of this term, however, preserves some semantic element: it implies that “the action [denoted by the verb] has not yet been realized” Ivić (1970:48). Other, generally semantic, accounts of CVS discuss nouns as matrix predicates simply by providing lists of nouns together with example sentences. To the partial list already provided by Ivić (1970) I add several more from Stevanović (1935 and 1970) and Maretić (1963), mainly those often quoted by other authors: dužnost ‘duty,’ grehota ‘wrongdoing,’ namera ‘intention,’ potreba ‘necessity,’ smer ‘direction,’ čast ‘honor,’ način ‘way/fashion,’ običaj ‘custom,’ prilika ‘opportunity’. A close examination of the nouns listed in this section, particularly when they occur in the CVS configuration as proposed in this study, reveals that it is actually not an easy task to find one particular semantic parameter under which they can all be subsumed. 36 In Chapter 6 of this study I offer a semantic account of all CVS matrix predicates, which is in line with the adopted theoretical framework. Finally, I include here what some other scholars treat as a separate group of matrix predicates – expressions, as in (22), for example. (22) a. Nemam not have.present.1STSG vremena time b. Nemam vremena ST not have.present.1 SG time ‘I do not have time to read.’ čitati. read.infinitive da that čitam. read.present.1STSG In (22), both the infinitival and da+present complements are complements to the matrix nemam vremena ‘I do not have time,’ an expression (here, I use the term in the sense of some previous accounts of CVS) which contains both a verb and a noun. The complements, however, depend on the noun and, consequently, complement the noun, and not the verb, in the same sense they complement verbs as CVS matrix predicates. Examples such as the one in (22) are therefore treated in this study as instantiations of CVS with nominal matrix predicates. 3.2.2. Adjectives Much of what has just been said about nouns as CVS matrix predicates equally applies to adjectives that participate in CVS. Once having established that adjectives play an important role in CVS as matrix predicates, the major task is then to determine exactly what kinds of adjectives can be CVS matrix predicates. 37 Again, syntactic accounts such as Đukanović’s fail to recognize that not all Serbian adjectives can in fact be CVS matrix predicates: (23) a. *Plavo Blue je be.present.3RDSG b. *Plavo je blue be.present.3RDSG ‘It is blue to read.’ čitati. read.infinitive da that se REFLEXIVE čita. read.present.3RDSG Again Ivić (1970:52) claims she is making an attempt to provide a semantic account, though just as it was with nouns her account is essentially syntactic. She establishes two different groups, of adjectives this time, united by her syntactic theoretical tool of ekspektativnost Exp ‘expectation’: 1. adjectives denoting characteristics of animate beings whose presence provides conditions for an event to be (gotov ‘done.M,’ spreman ‘ready.M,’ dostojan ‘deserving.M,’ kompetentan ‘competent.M’), and 2. adjectives derived from transitive verbs and used in the passive (određen ‘definite.M,’ izabran ‘chosen.M,’ pozvan ‘invited.M’). While the first of Ivić’s group of adjectives is defined in semantic terms, the second most certainly is not: there is nothing semantic in the fact that određen ‘definite.M,’ for example, is derived from the verb odrediti ‘to determine.’ Finally, again other accounts usually provide simple lists of adjectives and the contexts in which they occur in the CVS configuration. I list some of them from Stevanović (1935 and 1970) and Maretić (1963): pošteno ‘fair.N,’ časno ‘honest.N,’ prijatno ‘pleasant.N,’ dužan ‘obliged.M,’ rad ‘willing.M,’ željan ‘desirous.M,’ neophodno ‘necessary.N,’ potrebno ‘necessary.N,’ važno ‘important.N.’ 38 There is one minor limitation that must be imposed on adjectives, however. The limitation stems from the very meaning of a particular adjective: they may either characterize animate entities, and animate entities only (željan ‘desirous.M,’ rad ‘willing.M’) or they may characterize animate entities as well as whole situations (prijatna ‘pleasant.F’ vs. prijatno ‘pleasant.N,’ važna ‘important.F’ vs. važno ‘important.N’). This limitation accounts for the fact that not all the adjectives that are CVS matrix predicates indeed occur in the CVS configuration in all three possible gender forms, masculine, feminine, and neuter. This last comment brings to attention the fact that the two sentences already cited, but repeated here in (24), can be considered to have adjectival matrix predicates. (24) a. Teško difficult mi I.D je be.present.3RDSG priznati admit.infinitive b. Teško difficult mi I.D je be.present.3RDSG da that zločin. crime.A priznam admit.present.1STSG zločin. crime.A ‘It is difficult for me to admit to a crime.’ Examples similar to this one are sometimes said to contain an adverb as a matrix predicate rather than an adjective. In Serbian, the majority of adverbs indeed resembles in form adjectives when they are in their neuter gender form: teško ‘difficultly’ vs. teško ‘difficult.N,’ but težak ‘difficult.M,’ and lako ‘easily’ vs. lako ‘easy.N,’ but lak ‘easy.M.’ Thus Đukanović (1986) cites the following as instances of CVS with an adverb as the matrix predicate: 39 (25) a. Teško difficultly mu he.D je be.present.3RDSG b. Teško mu je difficultly he.D be.present.3RDSG ‘It is difficult for him to wait.’ čekati. wait.infinitive da that čeka. wait.present.3RDSG There is no difference between (24) and (25), except for minimal differences in certain lexical items in the complement. Basically, the matrix predicate is the same in all four sentences, yet in (24) I claim it is an adjective, and in (25) Đukanović implies that it is an adverb. It is not fully clear in what way the so-called adverb in (25) indeed fulfills its essential adverbial duty of specifying the meaning of the verb. The fact that the form of teško ‘difficult’ coincides with the form of teško ‘difficultly’ is simply a result of the configuration of a Serbian sentence. 3.2.3. Verbs All of the previous analyses dedicated to CVS focus predominantly on verbs as matrix predicates whose complements can vary between infinitival structures and da+present structures. And although all of the analyses treat the verbs essentially the same way, they are not entirely consistent and they provide no unifying account of exactly which verbs can indeed be CVS matrix predicates. Again, a detailed analysis of the semantic nature of verbs as CVS matrix predicates is offered in Chapter 6 of this study. I demonstrated above how Đukanović’s (1986) analysis accounts for all CVS matrix predicates, verbs included, using syntactic parameters only. He is consistent in his 40 approach, but achieves few if any predictive possibilities with his theory. Other analyses are either completely, or almost completely, semantic in nature. The fact that some are not consistently semantic explains problems that still impede a clear understanding of verbal matrix predicates in CVS. A number of studies simply indicate that indeed the given verbs can appear in the position of the matrix predicate in CVS, without providing any linguistic insight whatsoever. In another group of analyses (Stevanović 1970, Stanojčić and Popović 1992, Moskovljević 1936, Stanojčić 1967, Arsenijević 1997) a supposedly semantic parameter is employed: certain verbs are defined as nepotpuni glagoli ‘incomplete verbs’, or else glagoli nepotpunog značenja ‘verbs with incomplete meaning’. Arsenijević (1997) goes so far as to divide his incomplete verbs into two groups: 1. genuine incomplete verbs (morati ‘must’), and 2. false incomplete verbs (želeti ‘to desire’), though he provides no explanation for such a division. It turns out that verbs in the analyses mentioned above referred to as verbs with incomplete meaning, are, in yet other analyses, treated as modal verbs (Simić 2002, Simić and Jovanović 2002, Ivić 1970 and 1973, Craig 1975). It is not quite clear why for some scholars morati ‘must’ and želeti ‘to desire’ are verbs with incomplete meaning and for others they are modal. On the one hand, it could be that the former compared these verbs (in certain usages, it should be emphasized) with other Serbian verbs such as, for example, pevati ‘to sing’ or spavati ‘to sleep,’ which allow for some visualization of what they denote, whereas morati ‘must’ and želeti ‘to desire’ do not. On the other hand, it may be that defining modal verbs, and more generally the concept of modality, is not an easy task, just as Kalogjera (1982) notes: “The term modals, also modal verbs and modal 41 auxiliaries (each term implying a somewhat different point of view) derives from the broad and variously defined linguistic concept of modality.” The varying treatments of modal verbs in Serbian only support this claim. Ivić (1970) and Craig (1975), nevertheless, provide the two most thorough semantic accounts of verbal CVS matrix predicates. In her analysis, which is devoted predominantly to the use of the present in subordinate clauses with a subordinator da ‘that’, Ivić (1970:43) begins by saying that “a dependent sentence (clause) 10 with da ‘that’ is the most frequent complement of a verb, which otherwise functions as a predicate of another sentence (clause).” When what follows in her analysis is translated into the terms used in the present study, Ivić implies that whether the dependent clause with da ‘that’ actually participates in CVS or not “is caused by the meaning of the verb whose complement is this clause.” This is the motivation for her semantic treatment of verbs, among which she also discusses verbs that occur in CVS as matrix predicates. The first class of verbs that Ivić (1970:44) discusses is a group of the prototypical verbal CVS matrix predicates – phase (or aspectual) verbs: početi/počinjati11 ‘to begin,’ prestati/prestajati ‘to end,’ nastaviti/nastavljati ‘to continue,’ produžiti/produžavati ‘to continue.’ Craig (1975:151) speaks of the same verbs as “verbs expressing the different stages of a process.” There is nothing to which one can object in this view expressed by the two scholars, for these are the verbs upon which virtually every study dedicated to CVS agrees. 10 In Serbian, the term rečenica is often used to indicate both a sentence and a clause, and its actual meaning is determined based on the context in which it is found. Here, I translate it as sentence, which is the most common English equivalent, and put clause in parentheses to indicate the actual implication made by Ivić. 11 Verbs are cited in pairs because this is the way they occur in the original analysis. The first member of the pair is always a perfective verb, the second is an imperfective verb. 42 Next, Ivić (1970:48) takes notice of one other class of prototypical verbal CVS matrix predicates – modal verbs. She does that, however, in an interesting way by including in this group a desiderative verb – zaželeti/želeti ‘to desire,’ a voluntative verb – hteti ‘to want,’ and an intentional verb – nameravati ‘to intend.’ She also adds that other modal verbs are moći ‘can,’ smeti ‘to dare,’ morati ‘must,’ and trebati ‘be necessary,’ as well. Craig’s (1975:149-150) semantic account attempts at providing some sort of a generalization. She speaks of “verbs which express wishing or desiring (želeti12 ‘to desire,’ voleti ‘to like,’ hteti ‘to want’) … modal-like verbs (moći ‘can,’ morati ‘must’) … [and verbs] partly comparable to these modal-like verbs (trebati ‘be necessary,’ valjati ‘to be worthwhile’).” It is immediately striking that the two scholars do not duplicate each other’s treatment of this particular class of verbs. I believe that, if anywhere, it is here where the problem of defining modality and, consequently, modal verbs plays a crucial role. Namely, even if the two classifications presented here differ in some details, what seems to be more important is that their use of the term modal verb does not concur with an otherwise purely semantic account. The same applies to, again, virtually every study dedicated to CVS: the term modal verb is used without any precise definition. Thus, a working definition of modality seems to be necessary, for which I turn to Kalogjera’s (1982:1) definition of modality: “This [modality, BB], generally speaking, signifies the attitude of the speaker towards what he is saying, or, to be more specific, the attitude of the speaker towards the meaning expressed by the main verb in a clause.” The specification in his definition applies to cases such as (26a,b), where (26a) would be what 12 Craig’s (1975) examples are all from an ijekavian dialect, and what I present here are their ekavian counterparts. 43 he terms as the main verb in a clause, and (26b) attitude of the speaker towards the main verb. (26) a. He works hard. b. He seems to work hard. Ivić (1970:48) also brings to focus the class of cunctator-like moods: ustezati se ‘to restrain,’ oklevati ‘to hesitate,’ truditi se ‘to put effort,’ nastojati ‘to strive,’ običavati ‘to tend,’ umeti ‘to know how,’ pokušavati ‘to try,’ žuriti ‘to rush,’ očekivati ‘to expect.’ Craig (1975), however, does not discuss the same verbs in her study. In Chapter 6 I present a detailed analysis of CVS matrices and demonstrate that the use of the term modal verb, as well as cunctator-like verb is virtually unnecessary. What is crucial for a classification of CVS matrices is the concept that they denote by their semantics directed towards the complement, as I propose in this study. It is important to indicate here that certain verb classes claimed in some analyses to occur in CVS are not taken into consideration here. This is done because of the way CVS is configurationally set. A slightly more complex explanation is based on the fact that many of the analyses consulted during the work presented in this study start from different starting points: they sometimes simply provide accounts of verb forms that exist in Serbian (such as in the case of the grammar books of Brabec et al. 1968, Stevanović 1970, Stanojčić and Popović 1992, Simić 2002), at other times they are interested mainly in the infinitive (Ivić 1972, Joseph 1983, Đukanović 1986, Arsenijević 1997), yet at other times they deal only with the da-subordinate clause (Ivić 1970), and sometimes with the 44 actual variation of complements (Stevanović 1935 and 1954a,b, Moskovljević 1936, Brozović 1953a,b, Kravar 1953a,b). Different starting points, naturally, may result in inadequate representations of what actually happens in the language. That is to say, the fact that an infinitive may indeed occur as a complement of certain verbs does not automatically imply that it also participates in CVS as it is defined in the present study. One such class of verbs are the verbs of “motion over a certain place,” in Ivić’s (1970:50) terms: poći ‘to set off,’ doći ‘to arrive,’ otići ‘to leave,’ ići ‘to go’ (and many others that also “refer to an action calculated to achieve a certain goal”). Ivić indicates that the infinitive is indeed rare, but provides just enough evidence that configurations with these verbs as matrix predicates are different from CVS configurations: (27) a. Došla arrive.pparticiple.F.SG sam be.present.1STSG da that te you.A sam be.present.1STSG da that bih would.1STSG vidim. see.present.1STSG b. Došla arrive.pparticiple.F.SG te videla. you.A see.pparticiple.F.SG ‘I came (in order) to see you.’ The rare but theoretically, and certainly prescriptively, possible use of the infinitive would result in a sentence as given in (28) below, with the same meaning as the ones in (27a,b). (28) Došla arrive.pparticiple.F.SG sam be.present.1STSG 45 videti te. see.infinitive you.A Even though there is a clear variation of non-finite and finite structures in (27a,b) and (28), this is not an instantiation of CVS. Rather, the structures that constitute the variation in (27a,b) and (28) express the ultimate goal. As such, they can easily be fronted: (29) a. Da te vidim sam došla. b. Da bih te videla sam došla. The fact there is a level of similarity between the varying structures in (27a) and (28) (a da+present structure and an infinitival structure, respectively), on the one hand, and the only two complements in CVS, on the other hand, is not a sufficient argument in favor of (27) and (28) representing an instantiation of CVS. Moreover, the very common syntactic possibility of the structure as in (27b), traditionally labeled as the potential, proves that this cannot be treated as CVS. CVS matrices never allow complements such as the one in (27b). (30) a. Mogu da can.present.1STSG that te vidim. you.A see.present.1STSG b. *Mogu da can.present.1STSG that ‘I can see you.’ bih would.1STSG te videla. you.A see.pparticiple.F.SG Both complements in (30) are finite, with the one in (30a) the only possible complement with CVS matrix verbs. As is clear from (30b), CVS matrices prohibit the use of finite 46 complements formed around a verb in the potential (bih videla ‘would see. 1STSG.F’); the only finite complements allowed in CVS are those formed around the present tense. The structures that express the ultimate goal of the matrix in (27) and (28) are adjuncts, on both semantic and syntactic grounds, rather than complements, which is yet another reason for treating examples such as (27) and (28) with extra caution. While the matrix from (27) and (28) is capable of conveying a substantial amount of information on its own and outside of any context, the matrix from (31) is not: (31) a. Došla arrive.pparticiple.F.SG ‘I arrived/came.’ sam. be.present.1STSG b. #?Mogu. can.present.1STSG ‘I can.’ The case of nemoj ‘do not.2NDSG, nemojmo ‘do not.1STPL, and nemojte ‘do not.2NDPL’, which are ways of introducing a negative imperative, is similar to the one just discussed. Again, there are three possible configurations: (32) a. Nemojte do not.2NDPL pevati. sing.infinitive b. Nemojte do not.2NDPL da that pevate. sing.present.2NDPL c. Nemojte do not.2NDPL ‘Do not sing.’ da that biste would.2NDPL 47 pevali. sing.pparticiple.PL.M At this point I have nothing else to say about it, but that it seems to me that this does not represent what is said to be the CVS configuration for the same or similar reasons to the ones discussed regarding examples in (27) and (28) above. Besides a non-finite complement (as in 32a), there are two possible finite complements (as in 32b,c). The finite complement in (32c) is a structure formed around the potential, which is a verb form not permissible in CVS configurations, as demonstrated in (30b). However, I propose that it is necessary to take a closer look at this particular phenomenon in order to get an ultimate answer to the question whether it is CVS or not. 3.2.4. Future Tense There is one aspect of CVS, namely when variation of the two complements occurs in the future tense, which stands apart from the rest of instantiations of CVS: (33) a. Ja I ću will.1STSG b. Ja ću I will.1STSG ‘I will do more.’ uraditi više. do.infinitive more da that uradim do.present.1STSG više. more The Serbian future tense is a complex verb form in that it contains an auxiliary verb and the main verb. The auxiliary verb used in the future tense formation is: 48 (34) Future Tense Auxiliary ću ‘will.1STSG’ ćemo ‘will.1STPL’ ćeš ‘will.2NDSG’ ćete ‘will.2NDPL’ će ‘will.3RDSG’ će ‘will.3RDPL’ The main verb appears in the infinitive which may alternate with da+present. Diachronically speaking, the future tense auxiliary is derived from the verb hteti ‘to want’; the future tense auxiliary forms are unaccented forms of the present tense of that verb: (35) Present Tense of hteti ‘to want’: Accented/Unaccented Forms hoću/ću ‘want.1STSG’ hoćemo/ćemo ‘want.1STPL’ hoćeš/ćeš ‘want.2NDSG’ hoćete/ćete ‘want.2NDPL’ hoće/će ‘want.3RDSG’ hoće/će ‘want.3RDPL’ The verb hteti ‘to want’ is one of the verbs that appear as CVS matrix predicates, as indicated in the immediately preceding subsection, and is thus semantically described as one belonging to the group of verbal CVS matrices. Strictly synchronically speaking, the unaccented forms of the verb hteti ‘to want,’ used in the future tense formation, carry no semantic characterization otherwise present in the source verb (hteti ‘to want,’ that is) and that is why I refer to this particular case as a pure syntactic instantiation of CVS. In it, the variation between infinitival and da+present complements occurs in the future tense formation and there is no semantic parameter present in the future tense auxiliary that can 49 be held accountable for CVS in the future tense. Even though a probable implication of the future tense is similar to certain CVS matrices, such as nameravati ‘to intend,’ it is the fact that, what in the future tense correlates with the CVS matrix in the CVS configuration (and that is the future tense auxiliary), carries no independent semantic concept even remotely similar to nameravati ‘to intend’ for instance. The issue becomes slightly more complex in the negative future tense. There, the negative auxiliary (as in 36a,b) and the negative present tense of the verb hteti ‘to want’ (as in 37a,b) are exactly the same: (36) a. Ja I neću will.1STSG uraditi više. do.infinitive more b. Ja neću da I will.1STSG that ‘I will not do more.’ (37) a. Ja I neću want.1STSG uradim do.present.1STSG više. more uraditi više. do.infinitive more b. Ja neću da ST that I want.1 SG ‘I do not want to do more.’ uradim do.present.1STSG više. more Stevanović (1954a,b) insists that only (36a) and (37a) are negative future, while (36b) and (37b) are negated willingness to perform an action. Something similar to Stevanović’s claim is found in the results of the research that I conducted with native speakers. According to the results, 6 out of 8 participants claimed that in both the affirmative sentences in the future tense (as in 33a,b), and their interrogative counterparts, 50 there existed semantic similarity between sentences with infinitival complements and da+present complements. However, only 3 out of 8 participants recognized the same level of similarity when the auxiliary was negated. According to the results, 5 out of 8 participants claimed that the negative auxiliary with the da+present complement (as in 36b and 37b) expressed one’s desire, or lack thereof, to perform the complement action, while the negative auxiliary with the infinitival complement (as in 35a and 36a) was negation of a future action. The research results are an additional support for the claim that CVS in the future tense formation is purely a syntactic phenomenon. Furthermore, they are in accord with the claim made earlier in this chapter regarding the presence or absence of the binder of the complement in the matrix of the CVS configuration. With hteti ‘to want’ the importance of the binder is greater than with the future tense auxiliary where the auxiliary verb itself carries little or no semantic content whatsoever. 3.2.5. Summary It is important to reiterate here the most important claim made in this section dedicated to a close examination of nominal, adjectival, and verbal CVS matrix predicates. While it was generally clear that not all Serbian verbs participate in CVS, and only marginally implied that, naturally, not all Serbian nouns and adjectives participate in CVS, I propose that all three types of matrix predicates can be accounted for in a unified fashion. The account is semantically based and its details are presented in Chapter 6 of this study. The semantic account of matrix predicates developed there allows for any lexical unit defined 51 in terms proposed in Chapter to participate in CVS as the matrix predicate. That is to say, even if a primary meaning of a lexical unit does not carry implications required by the proposed account, if they are used in a secondary meaning, they can well participate in CVS. Such is the Serbian verb imati ‘to have.’ In (38) below it is not used in its primary meaning (from Stevanović 1970:764) and thus felicitously participate in CVS. (38) a. Prvog narednog časa first next class svi all imaju doneti should.3RDPL bring.infinitive svi all imaju da should.3RDPL that svoje radove. one’s own papers b. Prvog narednog časa first next class donesu bring.present.3RDPL svoje radove. one’s own papers ‘For the next class, everybody should bring their papers.’ 3.3. Complement I now turn to one remaining part of the CVS formula, namely the complement. Descriptively speaking, in the position of the complement in a CVS configuration either a non-finite complement, i.e. an infinitival structure, or a finite complement, i.e. a da+present structure can occur. While there seems to be no problem with the non-finite complement, it is the finite complement that requires a thorough explanation. 52 3.3.1. Infinitive That the infinitive as a verbal category in its own right exists in Serbian is confirmed by various grammar books of the language. Still, as Joseph (1983:131) notes, “one area of prescriptivist concern centers on the use of the infinitive, so that one must exercise caution when interpreting statements on infinitival usage in Serbo-Croatian.” In what has here been named the Serbian language the actual use of the infinitive had been confirmed by the results of my research, as presented in Chapter 5 of this study. Joseph (1983:147) discusses several aspects of the Serbo-Croatian infinitive only to conclude that the way “Serbo-Croatian contributes to the study and understanding of the Balkan infinitive-loss is through the fact that an infinitive-replacement process is still in progress in literary Serbo-Croatian and in (parts of) the non-Torlak Štokavian dialect area. This situation provides an opportunity to see first hand the variety of factors, social as well as purely linguistic, that can interact in the manifestation of this process.” The present study is an attempt to account for one aspect of infinitive usage, namely its occurrence in CVS. 3.3.2. Da+Present The other possible complement in a CVS configuration is the da+present structure. It has been long noticed that this particular structure presents us with curious linguistic behavior. Browne (1968:27) established that the complementizer da ‘that’ found in what is here referred to as the da+present structure is different from the complementizer da ‘that’ 53 found elsewhere in the language. He termed them da2 and da1 respectively. An example of a da2 is given in (39) below. (39) Zatražio ask for.pparticiple.M.SG sam be.present.1STSG da that Jovan Jovan čita (pročita) novine. read (finish reading).present.3RDSG newspaper ‘I aksed that Jovan reads (finishes reading) the newspaper.’ What is most significant is that, “after da2 there is no choice of tense allowed, but in Serbo-Croatian it is possible to use the present tense of any aspect.” On the other hand, “after da1 any tense that occurs in main sentences (clauses) may occur: hence no SerboCroatian perfective present”: (40) *Čujem hear.present.1STSG da that on he pročita finish reading.present.3RDSG novine. newspaper ‘I hear that he finishes reading the newspaper.’ Unlike Browne, who approached the problem from the point of view of complementizers introducing the complements, Ivić (1970:43) approaches the same phenomenon from the point of view of tense in the complements. She distinguishes two types of the present tense, namely mobile present and immobile present. The criterion for such a division is whether a present tense form can be replaced by other conjugated forms otherwise used for determining past and future tenses. When the substitution is possible, 54 the present tense form under question is mobile present, and when it is not – immobile present. Furthermore, Ivić is absolutely explicit that “immobility refers to the substitution regarding verbal tenses exclusively (... since it will be clear that the immobile present can often be replaced by an infinitive or potential).” The two analyses of the finite complement in CVS suggest that indeed there are some special properties that allow for this complement to demonstrate linguistic features otherwise not encountered in the language: first, any aspect of the present tense is allowed, and, second, the present cannot be replaced by any other tense. Interestingly enough, there exist two seemingly diametrically opposed views of the finite complement, or more precisely, of the verb form occurring in it. One is Kuljbakin’s (1921) view of the phenomenon according to which what is found in the complement is da sa indikativom ‘da with indicative.’ The other is Progovac’s (1993) view according to which the complement in CVS is subjunctive-like. Unfortunately, Kuljbakin provides no support for his claim, because his interest was directed towards a different phenomenon. Progovac (1993:116), however, demonstrates how Serbo-Croatian verbs can be classified as I-verbs, Indicative-selecting verbs, and S-verbs, those selecting Subjunctive-like complements. The latter verbs “are mainly verbs of wishing and requesting, such as želeti (‘wish’), hteti (‘want’), moći (‘be able to’), tražiti (‘ask for’), etc.” Clearly, the verbs that Progovac lists as S-verbs are those that can occur as CVS matrix predicates. Progovac then presents several different syntactic tests based on which she establishes her claim that, indeed, there are subjunctive-like complements in Serbo-Croatian: 55 (41) (42) Negative Polarity13 a. *Ne tvrdim [da vidim ST not claim.present.1 SG that see.present.1STSG ‘I do not claim that I can see anyone.’ nikoga]. no one b. Ne želim [da ST not wish.present.1 SG that ‘I do not wish to see anyone.’ nikoga.] no one vidim see.present.1STSG Topicalization a. *To ne this not tvrdim say.present.1STSG [da that sam be.present.1STSG potpisao t]. sign.pparticiple.M.SG ?? ‘This, I don’t say that I have signed.’ b. To ne želim this not wish.present.1STSG ‘This, I don’t want to sign.’ (43) [da that potpišem t.] sign.present.1STSG Clitic Placement14 a. *Milan gai kaže Milan he.A say.present.3RDSG ‘Milan says that he can see him.’ [da that vidi ti]. see.present.3RDSG b. ?Milan gai želi Milan he.A want.present.3RDSG ‘Milan wants to see him.’ [da that vidi ti]. see.present.3RDSG 13 (41a,b) are Progovac’s (1993) (29) and (30) respectively, (42a,b) her (52) and (11) respectively, and (43a,b) her (13) and (15) respectively. The names of the tests are from the original study. 14 Notice that, in this particular pair of sentences, the distinction between the sentences in a. with I-verbs as matrices, in Progovac’s terms, and sentences in b. with S-verbs as matrices is not as sharp as it is in (41) and (42). On the one hand, (43b) allows clitic-climbing, a complement argument denoted by the enclitic ga ‘he.A’ has climbed from its original second position within the complement to the second position within the sentence. On the other hand, however, this process is, more often than not, impermissible, as the questionable well-formedness of (43b) indicates. This particular example indicates the dual syntactic nature of the finite, da+present complement in CVS, which is explained in the remainder of this section. 56 Rakić (1987:90), however, presents various instances of the verb hteti ‘want’ in order to examine the nature of its complement. He starts from a purely descriptive observation that “[s]ince the particle da is the most common complementizer in SC15, it is normal to assume that da+V (pres) complements are sentence complements.” Nevertheless, the three tests that he applies show that “da+V (pres) complements go together with infinitive complements, not with sentence complements”: (44) Two time adverbs16 a. ?Ja hoću I want.present.1STSG danas da today that sutra tomorrow idem go.present.1STSG u bioskop. in movie theater ‘I want today to go to the cinema tomorrow.’ b. Ja I hoću want.present.1STSG danas da today that dođu. come.present.3RDPL ‘Today I want them to come tomorrow.’ 15 SC – Serbo-Croatian, V – verb, pres – present tense. 16 The names of the tests are from the original study. 57 oni they sutra tomorrow (45) Word order a. Ja I hoću want.present.1STSG danas da today that hoću want.present.1STSG da that idem go.present.1STSG u in bioskop. movie theater b. Ja I idem go.present.1STSG u in bioskop danas. movie theater today ‘I want to go to the cinema today.’ c. On je he be.present.3RDSG oni they juče yesterday hteo want.pparticiple.M.SG da that dođu. come. present.3RDPL ‘Yesterday, he wanted them to come.’ d. On je he be.present.3RDSG hteo want.pparticiple.M.SG da that oni they dođu juče. RD come.present.3 PL yesterday ‘He wanted them to come yesterday.’ (46) Wh-questions a. Koga hoće Milan da want.present.3RDSG Milan that who.A ‘Who does Milan want to deceive?’ prevari? deceive.present.3RDSG b. ??Koga who.A Milena Milena Milan kaže Milan say.present.3RDSG da that voli? love.present.3RDSG ‘Who does Milan say that Milena loves?’ In Rakić’s account, his da+V (pres) complements (44a, 45a,b, and 46a) are contrasted with what he refers to as true sentence complements (44b, 45c,d, and 46b). 58 Rakić then continues with three other tests which demonstrate “that the complements da+V (pres) are sentence complements after all”: (47) Placement of enclitics17 a. Rastko Rastko hoće want.present.3RDSG da that b. ?Rastko ga hoće Rastko he.A want.present.3RDSG ‘Rastko wants to visit him.’ (48) da that poseti. visit.present.3RDSG hoće want.present.3RDPL svi all da that idu go.present.3RDPL hoće want.present.3RDPL da that svi all idu go.present.3RDPL bioskop. movie theater b. Dečaci boys u in bioskop. movie theater ‘The boys want all to go to the cinema.’ (49) Distribution of perfective present Rastko hoće da Rastko want.present.3RDSG that ‘Rastko wants to arrive on time.’ 17 poseti. visit.present.3RDSG Position of quantifiers18 a. Dečaci boys u in ga he.A stigne na vreme. arrive.present.3RDSG on time The names of the tests are from the original study. 18 Rakić claims that (48b) is ambiguous, whereas (48a) is not. He does not explain the ambiguity of (48b), however. In my understanding, (48b) can also mean ‘The boys want everybody to go to the cinema.’ 59 In Rakić’s view, the distribution of clitics, quantifiers, and the perfective present in (47)(49) parallels their distribution in what he refers to as embedded sentences, which are, in terms of the present study, non-CVS complements. Rakić’s analysis, by presenting more relevant evidence, proves that the finite complement in CVS cannot easily be regarded as a subjunctive-like complement, just as it cannot be easily compared to just any other subordinate clause for that matter. I, therefore, continue to use the term da+present complement following the long tradition of the same or similar praxis and treat it as a structure syntactically distinct from the non-finite CVS complement as well as finite subordinate clauses. 3.3.3. Summary I conclude this section on the CVS complements by emphasizing that the non-finite complement, represented by an infinitival structure, and the finite complement, represented by a da+present structure, constitute the only two possible complements in CVS. They are defined as separate syntactic categories and as such they participate in CVS. I continue to use the term da+present structure, for there is not enough conclusive evidence that this structure is either clearly subjunctive or indicative in its character, though it presents linguistic peculiarities of both linguistic moods. 60 3.4. Conclusion In this chapter I provided a thorough linguistic characterization of the three major elements of CVS. First, I have explained what its configuration is and formulated it as in (50) below. (50) Xα MATRIX [α COMPLEMENT] I have, then, discussed the importance of a semantic treatment of CVS matrix predicates whose detailed analysis I present in Chapter 6. Finally, I have illustrated the major syntactic peculiarities of CVS complement predicates. With the three crucial elements of CVS set as they are, I now turn to explaining the very essence of CVS, namely an examination of factors that determine the variation of the complements. 61 CHAPTER 4 PREVIOUS ACCOUNTS OF CVS In her analysis of the Serbo-Croatian infinitive Ivić (1972:121) observes that the “infinitive as a complement has been debated in our local literature [linguistic tradition, BB] a lot, but no one moved beyond simply pointing to two problems: (a) the uneven frequency of the infinitive over a broad territory of the standard Serbo-Croatian, and (b) alleged semantic regulators of the distribution of the infinitive.” The then-existing accounts of the first problem Ivić considers to be only partially satisfying, though she states that all of them generally agree in that “the infinitive occurs more in the western [Serbo-Croatian] territory, while dakanje ‘more frequent use of da+present’ is characteristic of the eastern [Serbo-Croatian] territory.” As for the second problem, however, Ivić explains that the “viewpoints were controversial without one accepted conclusion in the end.” Several years after Ivić published these observations, while speaking of one particular instantiation of CVS, namely CVS with modal verbs, Milošević (1978:110) points out that the “[f]unctional and semantic relationship between the imperfective and 62 perfective present and the infinitive, when they function as complements of modal verbs (in the complex predicate), has been considered in various publications, but that, at the same time, it has been judged differently.” She continues by saying that “the viewpoints have ultimately merged and it has been concluded that only broader investigations, which are being postponed regarding the analysis of verbal forms in Serbo-Croatian studies, will provide a conclusive account.” As it turns out, there had been no major analysis of the phenomenon since Milošević has published her remarks, though there have been some analyses dedicated to either the infinitive or the use of the present with da ‘that.’ Just why this might have been the case is perhaps best summarized by Tanasić (1996:19): “At one time there was a lively ongoing debate regarding the parallel use of the present with the conjunction da and the infinitive, when they function as the complement of a verb. The discussion ended without a consensus on the question. However, this question is on the periphery of the overall problem of the use of the present – the present form here has no independent use, it is rather used for naming the action only.” Clearly, there is no one thorough account of CVS simply because the phenomenon, as it is defined in the present study and as it has been described for more than a century-long linguistic tradition, has almost never become the focal point of linguistic investigations. For those interested in the history of linguistics it is noteworthy that towards the middle of the 20th century two major debates about CVS took place in various linguistic journals (for the first debate see Stevanović 1935, Moskovljević 1936, for the second see Brozović 1953a,b, Kravar 1953a,b, Stevanović 1954a,b). The two debates are often quoted in all the major subsequent analyses dealing with the same or similar issues and 63 they both presented clearly diverging opinions on the character of the factors governing variation in CVS, i.e., whether they are semantic, syntactic, or purely stylistic in nature. In one line of thought that emerged from those debates, that advocated to some extent by Moskovljević (1936) and Brozović (1953a,b), I find some support for what I present as a major governing linguistic, more precisely syntactic, factor in CVS. In this chapter I provide a history of previous accounts of CVS by emphasizing their essential nature and major achievements and, where necessary, explaining their deficiencies. I focus on the weight of their explanations when it comes to the phenomenon of complement variation itself. In section 4.1., however, I first address the problem upon which Ivić (1972) claims some general agreement, namely the importance of understanding and taking into consideration what used to be called the Serbo-Croatian language territory with regard to CVS. I present what I refer to as descriptive accounts in section 4.2. Then in section 4.3., I discuss what I here refer to as explanatory accounts followed by a conclusion in section 4.4. 4.1. Dialectology and Sociolinguistics of CVS Pavlović (1960) provides a very succinct overview of the fate of the infinitive in the Balkans. As for the Serbo-Croatian territory, he determines that Belgrade, the capital city of Serbia, represents the ultimate or easternmost location in which the trait of the loss of the infinitive as a Balkan process can be detected. Still, by providing various examples from a novel by August Cesarec, a Croatian author, Pavlović demonstrates that the loss of 64 the infinitive has an apparent tendency to spread towards the West, more precisely, towards Zagreb, the capital city of Croatia. Kravar (1953:44) focuses his attention on the dakanje phenomenon – “the use of the combination da+present instead of the infinitive” – in the Štokavian dialect of SerboCroatian. The Štokavian dialect, named after the word što ‘what,’ is the dialect upon whose two sub-dialects the standard Serbo-Croatian language was founded. Although Kravar (1953:44) asserts that the actual borders between the use of the infinitive and the use of the da+present structure are yet to be determined, he explains that “the dakanje phenomenon is not equally represented on the Štokavian territory.” Furthermore, he provides a threefold division of the territory under consideration recognizing “three zones: the Eastern zone with the strong dakanje phenomenon and a very weak, almost nonexistent use of the infinitive, then the central zone with the moderate dakanje phenomenon, in which da+present and infinitive are in competition, and the Western zone with the weak dakanje phenomenon and a very developed use of the infinitive. The three zones parallel, grosso modo, the three Štokavian dialects: ekavian, ijekavian, and ikavian.”19 In the literary Serbo-Croatian language, Kravar detects a certain polarization with respect to the dakanje phenomenon along the Belgrade-Zagreb line, though he explains that the actual usage “in the field cannot be marked as Serbian vs. Croatian.” A consequence of such a dialectal picture is the “difference in usage of the infinitive and the 19 This dialectal division is based on reflexes of Common Slavic *ě. Browne (1993:308) explains that the “[r]eflex of ě, often called jat’ vary geographically, a fact on which one well-known dialect classification is made.” Browne (1993:382) also indicates that “Serbo-Croat speakers, conscious of dialect divisions, identify themselves … according to their reflex of jat’ as ekavci/екавци , (i)jekavci/(и)јекавци or ikavci/икавци.” 65 combination da+present in the literary language on two sides [Belgrade and Zagreb, BB].” Interestingly, Kravar (1953:44) also takes into consideration various sociolinguistic factors as well. He claims that “[e]very person, who speaks or writes Štokavian, is faced with an unconscious choice between the two expressive possibilities: the infinitive or the combination da+present. However, only the literate people are perplexed by this choice, and apparently even more so – linguistically educated people. In the field, as always is the case, fixed speaking habits govern the choice.” Kravar also claims that “lately, the dakanje phenomenon has been moving in a different direction [different than the dialectal Belgrade-Zagreb line, BB]: top to bottom, thanks to the administration, military, school, press, radio, etc. It seems that the phenomenon has generally been spreading from the city to the country (village), and not vice versa.” Kravar’s independent observations, essentially sociolinguistic in nature, are extremely refreshing considering the time when they were made. Unfortunately, he did not support them with any scientific evidence, so it is fair to say that his conclusions were based on his own linguistic intuition. Also, and to the best of my knowledge, there has been little or no published research that tests Kravar’s sociolinguistic claims, so his suggestions are yet to be closely examined. Joseph (1983:143) also incorporates sociolinguistic factors into his view of CVS: “The conditions governing this fluctuation [of da-clauses and infinitives, to use his terminology, BB] are complex and seem to involve a mixture of regional, lexical, syntactic, sociolectal, stylistic, and idiolectal factors.” While Joseph (1983:142) claims, however, that “[t]he sociolectal and stylistic factors are harder to get [than the other 66 factors, which he addresses in some detail, BB],” he does maintain “that stylistic differences and prescriptive norms, such as would be socially interpretable, do play a role in the variability of infinitival usage,” and finds that “one additional facet of the infinitive/da-clause fluctuation, namely variation within an idiolect, can safely be assumed to be socially conditioned.” These sociolinguistic factors only add to the list of factors already made by Kravar (1953) and presented above. Since Joseph’s (1983:137) study is dedicated to the Balkan infinitive, SerboCroatian included, he provides a fairly detailed picture of the distribution of the infinitive in Serbo-Croatian dialects, only to conclude that “the general West to East distribution of the infinitive in Serbo-Croatian indicates that the West is more conservative regarding retention of the infinitive, with the Kajkavian 20 maintenance of the infinitive-supine distinction being the extreme on this end, while the East is less so, with the Torlak dialects21 as the opposite extreme.” And although he provides a fairly extensive list of various factors that may have an impact on CVS, Joseph (1983:145,146) still ends by clearly emphasizing the regional factors, saying that “the real value of Serbo-Croatian with regard to the question of the loss of the infinitive in the Balkan languages lies in the distribution of the infinitive in the modern Serbo-Croatian dialects. … Štokavian in general shows the loss of the infinitive more than Čakavian 22 or Kajkavian, and the 20 Kajkavian dialect is a dialect of Serbo-Croatian named after the word kaj ‘what.’ Browne (1993:382) explains that “Kajkavian, Čakavian and Štokavian are named after their words for ‘what’: kaj, … ča, and što ... or šta.” 21 According to Joseph (1983:133), the Torlak dialects of the Southeast have a “basic ‘Balkan’ character and links with Bulgarian and Macedonian.” 22 Čakavian dialect is a dialect of Serbo-Croatian named after the word ča ‘what.’ See footnote 20. 67 Eastern literary standard (Belgrade) shows this feature more than the Western literary standard (Zagreb).” This brief excursus regarding the importance of the Serbo-Croatian linguistic territory for the phenomenon, which is named differently because of different analytical standpoints (the variation of the complements, the Balkan infinitive, to mention only two), and in this study referred to as CVS, indeed demonstrated a sort of agreement about which Ivić (1972) reported. To this I add that, in the present time, as discussed in Chapter 2, the term Serbo-Croatian linguistic territory seems to be outdated as such. One should rather speak of Bosnian/Croatian/Serbian linguistic territory, though many other terms come to mind as well (such as a term indicating the presence of the Montenegrins on that territory, for example, or else to clarify whether one should use Bosniac/Bosniak instead of Bosnian). In any case, one must also remember that there were several and various movements of whole groups of population over the territory in question in the first half of the last decade of the 20th century, which are, as of yet, unaccounted for when it comes to mixing of speakers of various dialects. Even though it would be fascinating to compare the present dialectal situation with the one of before the last decade of the 20th century and analyze them with respect to CVS, such task would be beyond the scope of the present study. By electing to analyze instantiations of CVS from one dialect only, the one spoken around Belgrade, I consciously set aside the dialectal, or regional factors, as well as sociolinguistic factors discussed in this section, and leave them for a future research. In Chapter 3 I focused on semantic factors (lexical, in Joseph’s terms) and in the following sections and chapters to come I focus on syntactic, as well as semantic or conceptual factors that control CVS. 68 4.2. Descriptive Accounts In this section I present one group of selected accounts of CVS, which I arbitrarily term descriptive. Clearly, description is a necessary first step to any analysis. Therefore, all of the already existing accounts begin with some description of the facts relevant for CVS. However, the difference among various accounts of CVS lies in the extent to which they develop an explanatory theory of the data they initially described. I, therefore, choose the term descriptive based on the essential characteristic of the accounts to be presented in this particular section: they simply describe the facts found in the language, but even that not precisely enough sometimes. No two linguistic analyses are the same, so even accounts that start out as purely descriptive in nature often end up providing a certain number of relevant linguistic claims; thus I speak of two kinds of the descriptive accounts: the true descriptive accounts and descriptive-explanatory accounts. 4.2.1. True Descriptive Accounts Most often the true descriptive accounts are found in Serbian grammar books or books dedicated to the Serbian syntax. Accounts vary from simple descriptions such as Lalević’s (1951:122) statement that, “when the infinitive is found in the predicate, it can be replaced with the present with the conjunction da,” 69 (1) a. On he ume know how.3RDSG b. On ume he know how.3RDSG ‘He knows how to sing.’ pevati. sing.infinitive da that peva. sing.present.3RDSG to complex and exhaustingly thorough descriptions offered by Simić and Jovanović (2002,III.-IV.:1100-1117). While it is clear that Lalević randomly selects examples for the purpose of illustrating his simple description, Simić and Jovanović’s examples, by contrast, are a selection of sentences from literary works, without any explanation of the exact criteria for including some and rejecting others, a methodology and praxis traditionally followed by many a, though not every, Serbian grammarian. Similar to these two extremes, but somewhere in between in thoroughness, are other descriptive accounts. Brabec et al. (1968:257) provide their own set of examples and simply state that “instead of the infinitive, the present with the conjunction da may serve as a complement to the verbs.” Stanojčić and Popović (1992:248,249) account for the phenomenon with slightly more details by saying that in certain syntactic contexts with complements, “the complement part consists of a verb with a full meaning used either in the form of the i n f i n i t i v e or (in the form of the) p r e s e n t (which agrees with a subject) with the conjunction da.” Their account then becomes deceptively simple when they claim that the complex predicate (which is their term for what in the present study is the MATRIX COMPLEMENT complex) “may appear in all types of sentences”: 23 23 Examples (2a,b,c,d,e,f) are Stanojčić and Popović’s (1992) (45), (46), (47), (48), and again (48), respectively. 70 (2) a. Ivan hoće da RD Ivan want.present.3 SG that ‘Ivan wants to be the best student.’ bude be.present.3RDSG b. Ivan u pet sati mora Ivan in five hours must.present.3RDSG ‘Ivan has to be here at five o’clock.’ najbolji best đak. student biti ovde. be.infinitive here c. Počelo je da sviće. RD began be.present.3 SG that dawn.present.3RDSG ‘The sun started to come up. (literally: It began to dawn.)’ d. O tome ne sme about that NEG dare.present.3RDSG priča. talk.present.3RDSG ‘It is not allowed to talk about that.’ e. Posao job mora must.present.3RDSG f. Posao se job REFLEXIVE ‘The job must be done.’ da that se REFLEXIVE biti završen. be.infinitive finished mora must.present.3RDSG završiti. finish.infinitive Stanojčić and Popović (1992) provide these six sentences in which, first, by a simple count four out of six sentences have a da+present complement, whereas only two have an infinitival complement, and, second, there is no indication that indeed both possible complements can be used in one and the same syntactic context, as implied in their claim cited above, for they provide no one syntactic context where both complements are possible. Clearly, the claim as they make it does not account for what they present in their own set of examples, or else, the examples are not sufficient to actually prove the claim true. To be completely fair, at a later point in their book Stanojčić and Popović (1992:327) do discuss examples such as (3) below, 71 (3) Marko Marko je be.present.3RDSG želeo want.pparticiple.M.SG krenuti set off.infinitive kući. home ‘Marko wanted to leave for home.’ in which, according to their account, an infinitival complement is an object complement, but they say that “in the Eastern variant at least, more often than the infinitive, a thatclause with the verb in the present tense and the conjunction da is used” 24. Simić (2002:164) briefly states that “there are very few examples in which the infinitive is not replaceable with the present with the conjunction da.” The author employs another methodology often practiced by many a Serbian grammarian – that of replacement. He simply states that in (4) Najbolje je biti lijepo sa best be.present.3RDSG be.infinitive nice with ‘It is the best to be nice with people.’ ljudima. people “it is possible instead of one to take the other form”: (5) Najbolje best je be.present.3RDSG bude be.presen.3RDSG 24 da that se reflexive s with ljudima people lijepo. nice Here, I believe, the term refers to what used to be called the Serbo-Croatian linguistic territory. 72 There is no syntactic evidence, or any other kind, for that matter, that this statement is in fact true unless one accepts this simple replacement as a piece of evidence. Simić does not even indicate a different word order in the complement in (5) when compared to the complement in (4). This becomes an important problematic issue, for there are accounts according to which word order is indeed a factor controlling CVS. Not surprisingly, Simić and Jovanović (2002,I.-II.:329) adopt the same approach and make the following descriptive claim: “In the position of the predicative [in the present study this is the COMPLEMENT, BB], with a certain number of verbs or predicational expressions, an infinitive or a present with the conjunction da can be found.” To prove their claim that supposedly the two types of complements occur with equal frequency, Simić and Jovanović (2002,I.-II.:330) present a set of seven sentences with an infinitival complement of the verb moći ‘can, to be able’ and a set of eight sentences with a da+present complement and then conclude that “not only that they are of the same frequency [the two possible complements, BB], but the infinitive and da+present could freely replace each other in one and the same example without a change in meaning.” Their proof is essentially the same in nature offered for (4) and (5) above. Although it presents various interesting results with respect to CVS, Đukanović’s (1986) analysis is strictly descriptive in nature: he examines various usages of the infinitive in the language of four different authors, two from Serbia and two from Croatia. Đukanović (1986:62) simply lists his findings – the use of the infinitive with various matrix predicates. In the conclusion, however, his otherwise descriptive account turns into a rigorous prescriptive one: “The infinitive is one of those important syntactic categories whose usage, or lack thereof, substantially influences the stratification of our 73 standard language, and even substantially influences the stratification of the Eastern variant of that language.” 25 Đukanović (1986:63) then presents four different pieces of evidence against CVS, as if it does not exist or is just a sporadically occurring phenomenon: “There are at least two strong reasons that negate the legitimacy of this process [CVS, BB]. First: the infinitive is an original Slavic and Serbo-Croatian category. Second: the infinitive is very frequent in the East-Herzegovina dialects, which are still considered to be the basis of the Serbo-Croatian standard language. The fact that within our language-diasystem there are dialects in which there is no infinitive, that is to say the memory of the speakers of those (sub)systems does not register the infinitive at all (except, of course, passively), should not influence the position of the infinitive in the standard language. Be that as it may, the representative speakers of the (sub)systems in which there is no infinitive are ‘obligated’ to recognize the infinitive in the standard language.” Đukanović employs terms such as original, legitimacy, and obligated in order to explain that the variation of the complements should not be recognized as a feature of the standard language, although the results of the research presented in his analysis clearly indicate, and he himself so observes, that there is a fluctuation in the use of the complements both along the dialectal and the temporal line (the four authors are from two different regions and their works – from different time periods). Đukanović (1986:63) does not stop here. He presents two more arguments “in support of the infinitive over the da+present construction:” 26 first, the argument of 25 Again, the term refers to what used to be called the Serbo-Croatian linguistic territory. 26 Interestingly enough, Joseph (1983:73) discusses a similar sentiment towards the infinitive in the Pontic dialects of Greek – a sentiment that lacks any linguistic argumentation. 74 discourse-related economy, and, second, the argument of informational economy. The former argument is illustrated by example (6) below. (6) a. Možemo can.present.1STPL li QUESTION b. Možemo li can.present.1STPL QUESTION ‘Can we lift this weight?’ dići ovaj lift.infinitive this da that teret? weight dignemo lift.present.1STPL ovaj this teret? weight According to Đukanović, in (6a) dići ‘to lift’ has 4 letters (and, I take it, phonemes) in Serbian, whereas da dignemo ‘that we lift’ in (6b) has 9. Indeed, the part of the complement in (6a) that Đukanović takes into consideration is shorter with respect to the number of letters (and phonemes) than the considered part of the complement in (6b), though Đukanović also has examples in which the number of letters is the same. It is not quite clear, though, that this is indeed a factor that governs CVS, for – as already indicated – Đukanović’s account of exactly which complements are found in the literary works that he examines demonstrates that the infinitive is not a dominant choice for the complement. The latter argument, based on informational economy, states that “by using the infinitive one avoids repeating information about the grammatical person, number, and tense, which is already contained in the main verb or in another way already marked.” Here again, Đukanović is logically correct. By looking at (6b), for example, one easily realizes that grammatical information in the matrix predicate (present.1STPL) and grammatical information in the complement predicate (again, present.1STPL) are indeed 75 the same.27 However, Đukanović does not offer any evidence in support of his essentially prescriptive advice. It is absolutely clear that none of Đukanović’s four arguments provide a useful linguistic account of CVS. On the contrary, Đukanović (1986:64) actually insists on a linguistically unnatural process, a forced use of the infinitive, so he states that the “‘restoration’ of the use of the infinitive has a pure linguistic and sociolinguistic justification.” It is just not clear in what way this could be true. 4.2.2. Descriptive-Explanatory Accounts Somewhat similar to Đukanović’s descriptive nature is Kalogjera’s (1971:67) account of the variation. He does, however, provide minimal insight into CVS in that he claims that “[i]n some cases the infinitive and da+present seem to be in free variation (the latter, of course, being more frequent with the Eastern variant), but in some cases (as e.g. nudim da sakupim) the infinitive is only marginally acceptable and the construction da+present seems to be required in both variants.” Kalogjera’s account is semantic in nature – it is directed towards the examination of particular lexical units, though it is not fully developed, rather merely offers insight into the phenomenon of variation. Ćirković (1985:261) also adopts the view that the two complements are in free variation or, as he puts it, “that the function of the present and the particle da and the function of the infinitive are the same in a sentence. But their meaning does not have to 27 Joseph (1983:185) indicates that the idea of redundancy played an important role in explaining the infinitive-loss phenomenon in Greek, though maybe in the opposite sense than the one advocated by Đukanović. By explaining that there is no compelling force in proposals that rely on the idea of redundancy, Joseph criticizes Jannaris’ proposal that a language with “a wealth of finite forms,” like Greek, lost the infinitive since “the infinitive in some way did not fit into the ‘genius’ of the Greek language.” 76 be the same.” Still, based on lexical grounds, he lists examples in which “the present with the particle da cannot be replaced with the infinitive: moli da uđe ‘he/she is asking to come in’– moli ući ‘he/she is asking to come in’; zahtijeva da krene ‘he/she is asking/requesting to set off’ – zahtijeva krenuti ‘he/she is asking/requesting to set off’ and the like.”28 Ćirković unfortunately does not explain exactly how the cited sentences differ in meaning. His major claim, however, is made about the infinitive preceded by negation, as in (7) below. (7) Zabrinut sam concerned be.present.1STSG jesti s eat.infinitive with za for Džudit, Judith neće want.present.3RDSG nama. us According to Ćirković (1985:261,262), the sentence in (7) “can carry two meanings:” (8) a. Zabrinut sam za Džudit zato što neće da jede sa nama. ‘I am concerned for Judith because she won’t eat with us.’ b. Zabrinut sam za Džudit i zbog toga joj zabranjujem da jede sa nama. ‘I am concerned for Judith and therefore I forbid her to eat with us (she WILL not eat with us). ’ Ćirković’s (1985:262) explanation is somewhat convoluted in that he states that “the two sentences are missing something. And that something is the use of negation + infinitive instead of present + da.” What Ćirković is saying is that sentence (7) can have two meanings as in (8a,b): (8a) denotes one’s concern over the state in which Judith is (she has no desire to eat with others), whereas (8b) denotes one’s concern over Judith for 28 All translations in this quote are mine, BB. 77 which reason she is not allowed to eat with others (probably because of something she did). Sentence (7) is, thus, clearly ambiguous, so one should never use (7) in order to convey meanings as in (8a,b); rather only (8a) should be used when the particular meaning of (8a) is intended, and (8b) when the speaker’s intention is to say just what (8b) is saying. While this may be the case, it is not apparent to me that (7) is an instantiation of CVS at all, as Ćirković indicates. This is the first out of many examples of how cautious one must be in linguistic analysis in general: Ćirković presents his analysis as if it explains an instantiation of CVS, and his title (Negacija uz infinitiv i rječcu da + present ‘Negation With Infinitive and Particle da + Present’) implies that he speaks about the negation of the infinitive and the particle da + present. While in (7) the infinitive is indeed preceded by negation and only in (8a) a structure introduced by da ‘that’ is, it is quite clear that, as presented, (7) and (8a,b) are not an instantiation of CVS. Gallis’s (1970) analysis examines CVS in the future tense structure only as it occurs in one novel by a Serbian author who is originally from central Serbia. As such, it is descriptive in nature, and Gallis explicitly agrees with the essence of the descriptive accounts discussed so far in that there is no difference in meaning between the two variants, although he eventually provides an interesting semantic explanation of just what might determine the variation of the complements. His analysis shows that verbs such as kazati ‘to say’ and misliti ‘to think’ are more likely to appear in the da+present structure in the future tense than in the infinitive, whereas the opposite is true for many other verbs (biti ‘to be’, čekati ‘to wait’, dati ‘to give’, dolaziti/doći ‘to arrive’, imati ‘to have’, isterati ‘to chase out’, izgoreti ‘to burn’, obesiti (se) ‘to hang (oneself)’, odlaziti/otići ‘to 78 leave’, pojesti ‘to eat up, reći ‘to say’, roditi ‘to give birth’, ubiti ‘to kill’, uzeti ‘to take’, zapamtiti ‘to memorize’, znati ‘to know’). On various occasions Stevanović (1935, 1954a,b, 1970) has stated his view of CVS and each time concluded that the two complements are synonymous in nature. Still, there are some curious explanations in his claims. Stevanović (1935:287) emphasizes the stylistic criterion as dominant by saying that “where there are more of the words with incomplete meaning, the beauty of the style poses a request for having a feel for using their complements.” Stevanović (1954a:97) instructs that “it is incorrect to say that the present with the conjunction da is used instead of the infinitive or vice versa – the infinitive instead of the clause da+present, since both forms are used in their true function.” He does pose the right question, however: “Just what sustains the use of one or the other?” Unfortunately, he provides no answer to his own question, for his accounts, as I said, always insist on the synonymy of the complements. Stanojčić (1967:176) closely follows Stevanović’s views of CVS and concludes that “one must add – to the basically correct scientific claims that ‘here, the different meanings of the main verb are the only relevant factor’ and that to, ‘on the one hand, the function of the complement, and, on the other hand, the nature of the basic contemporary meaning of the infinitive and aspectual meaning of the present’, which allow the parallel use of the two forms as complements – one must add the element of rhythm, of course always for a defined context.” Stanojčić also speaks of examples in which Stevanović’s instruction about a beautiful style cited above is apparently rejected by the actual language, which is, in both of their cases, always the language of literature. 79 A very interesting debate about CVS took place between Brozović (1953a,b) and Kravar (1953a,b). The latter basically claimed that there was no difference in sense between the two complements. The former, however, insisted on stylistic and rhythmic differences and essentially accepted Moskovljević’s (1936) account of CVS. Moskovljević (1936:108) presents various examples in support of his crucial point that the difference lies in the fact that there are incomplete verbs of two types: those that can be used with a verbal complement whose subject is the same as the subject of the main verb, and those whose verbal complement expresses the intention of the main verb so that the two do not have to have the same subject. Moskovljević notes that it is the subject of the matrix (in terms of the present study) that determines what might happen with the complement, though he does not develop this idea further, for his primary concern was to use this argument in order to explain the diachrony of CVS. A descriptive-explanatory account of CVS is also offered by Katičić (1986). He refers to CVS as infinitivizacija ‘infinitivization’ and conceives of it as one type of joining two clauses during which the clause that is added to another clause loses its clausal features. Infinitivization in particular Katičić (1986:465) regards as the process in which the added clause’s “predicate verb is transformed into the infinitive. … Infinitivization is a transformation by which, an already inserted dependent clause, declarative or final/purpose clause, establishes an even stronger connection with the main clause, so much that in the resulting transformed configuration it loses its own predicational categories and the predicate of an original dependent clause is completely joined with the predicate of an original main clause.” 80 In Chapter 3 of this study I already explained that final/purpose clauses do not belong to what is defined as CVS. As for the reminder of Katičić’s claims, they appear to be interesting for at least two reasons. First, Katičić’s view of CVS is interesting in that he implies that a clause-like structure is a source of the resulting infinitival structure, as is clear from his definition of infinitivization. Second, Katičić presents a substantial amount of examples and uses all of them to indicate in exactly what configurations infinitivization is more probable than not. His ultimate conclusions are both stylistic and prescriptive in nature, more than anything else, for he employs terms such as starinski stil ‘old-fashioned style’ and svečani stil ‘special-occasion style.’ Ultimately, Katičić does not offer any particularly telling synthesis of his examples, but leaves them as a list of individual descriptions. 4.2.3. Summary What is characteristic of these descriptive accounts is that they all treat CVS in a very simple way, first, by acknowledging it, and, second, by demonstrating how both possible complements are interchangeable. Also characteristic of the descriptive accounts is that they usually start from example sentences found in the language of literature that illustrate the authors’ claim of the equal distribution of the complements. Unfortunately, except for Gallis (1970) and Đukanović (1986), there are no statistical data offered in support of claims about even distribution. And even these two linguists fail to use their findings to the fullest extent. 81 In describing CVS, the descriptive accounts sometimes attempt to provide possible explanations for CVS. Their explanations are focused on the semantics of the matrix or complement, and the style and rhythm of the utterance. Moskovljević (1936) and Brozović (1953), who expresses some support for Moskovljević’s claims, may be considered exceptions to this generalization, since they notice that there is an important syntactic factor, i.e., the involvement of the matrix subject, that accounts for the diachrony of CVS. 4.3. Explanatory Accounts In this section I present accounts arbitrarily referred to as explanatory accounts. They differ from the descriptive-explanatory accounts discussed in section 4.2.2. above in that their main goal is to actually provide an explanation of CVS and not simply to describe the language. I further divide the explanatory accounts into two kinds: generative and non-generative. 4.3.1. Generative Explanatory Accounts In two different analyses by Ivić (1970, 1972) of various linguistic phenomena in SerboCroatian, she develops an idea that is essentially generative in nature and supposedly explains the variation in complements. Ivić (1972:119) treats the infinitive as a complement to certain verbs that “in a given syntactic context, receives the character of a piece of information. Such an infinitive is, actually, a grammatical tool that serves to 82 condense the sentence. … It is, therefore, always a variant of a d e p e n d e n t clause.” Then, in her footnote 19, Ivić notes that she “speaks of this phenomenon as well as of the conditions which account for the ‘immobile’ character of the predicate during the generating of a sentence,” in one of her previous works, Ivić (1970). 29 Indeed, Ivić (1970:48) provides a generative-like explanation of what exactly happens with the verbal matrix predicates in CVS. She states that “in fact, the meaning of those verbs automatically implies the establishment of a syntactically relevant piece of information which I would name here ekspektativnost ‘expectancy’ and mark with a symbol Exp. The Exp information shows up in front of the dependent clause in the deep structure as a sentential adverbial of a kind, signaling with its presence that the realization of the action entailed by the predicate is being expected”: (9) S = S1 + Exp + S230 Ivić goes so far as to provide a very precise semantic (or, lexical) equivalent of Exp, which is, in her view, the expression da bude ‘may it be.’ According to her account, the derivation of a sentence would then be as in (10a) or (10b) below. 29 The terminology characteristic for Ivić (1970, 1972) has already been discussed in Chapter 3. 30 S – sentence, S1 – main clause, S2 – dependent clause. 83 (10) a. želeo sam I wanted + da bude may it be + ja zaspati → I to fall asleep želeo sam zaspati // da zaspim I wanted to fall asleep.infinitive // da+present ‘I wanted to fall asleep.’ b. nameravam + I intend da bude + may it be ja se ženiti → I to get married nameravam se ženiti // da se ženim I intend to get married.infinitive // da+present ‘I intend to get married.’ There are two problems with this particular account. First, it is clear from (10a,b) above that, while the account may well explain why the two types of complements occur with certain verbal matrix predicates, it does not explain why they vary the way they do – the account actually does not say much about CVS itself. Second, although Exp is treated as a syntactic unit, it is still said to be “a relevant piece of information” and then also the one implied by the “lexical meaning of certain verbs.” Both characterizations made by Ivić are more semantic in nature than they are syntactic, so the presence of Exp in the syntax of sentences such as (10a,b) above is not quite justified. It may be better if Exp is said to be present in the semantics of CVS matrix predicates, as it was suggested in Chapter 3, for different semantic factors, however. Three different authors in four different accounts, notably Browne (1968), Craig (1985), and Rakić (1986,1987), analyze CVS in the classical fashion of the generative linguistics and argue for two different and theory-internal explanations. Browne (1968) proposes that sentence (11) has the surface structure illustrated in sentence (12) and the deep structure of (13). 84 (11) Jovan želi pročitati RD Jovan want.present.3 SG finish reading.infinitive ‘Jovan wants to finish reading the newspaper.’ (12) [SJovan Jovan novine.31 newspapers [VPželi want.present.3RDSG [VPpročitati finish reading.infinitive [VPželi want.present.3RDSG [SJovan Jovan novine]]]32 newspapers (13) [SJovan Jovan [VPpročita novine]]]] RD finish reading.present.3 SG newspapers Browne’s generative account is essentially transformational. He suggests that there are at least two possible transformations: one is applied in the case of the addition of a complementizer infinitive,33 the other if the complementizer da is added. In the case of the former, “a transformation that removes the redundant subject Jovan will be applied and the transformation will integrate Jovan’s predicate into the upper S,” while in the case of the latter, “the lower node ‘S’ is still present [though] the repeated subject (Jovan) is usually left out, although in cases of emphasis it may be present in the form of a pronoun … [I]n any case the person and number specifications that the subject requires are given to the verb pročita ‘he finishes reading.’” 31 Examples (11), (12), and (13) are Browne’s (1986) examples (1), (2) and (4) respectively. 32 All the square bracket representations in this section stand in the place of tree representations found in the original studies, unless otherwise indicated. I believe that this in no way takes away from any of the original studies’ explanatory power. 33 Browne’s treatment of the infinitive. 85 Craig (1985:153) discusses the same or similar type of sentences as Browne (1968). Her generative account, however, is a theoretically updated version in that she does not speak of transformations, but of the Equi-NP deletion rule being applied to various structures. For example, (14) Ja želim I want.present.3RDSG ‘I want to go.’ ići. go.infinitive derives from (15) [Sja I [VPželim want.present.3RDSG [Sja I idem]]] go.present.3RDSG “as the result of the application of an Equi-NP deletion rule applying only in case of coreferentiality of the subjects.” Interestingly, there is another rule, namely that of complementizer insertion, in case there is no co-referential relationship between the matrix and embedded subject. Thus (16) [Sja I [VPželim want.present.3RDSG [Sti you ideš]]] go.present.2NDSG cannot be used for derivation of either (17a) or (17b). 86 (17) a. *Želim want.present.3RDSG ‘I wish you to go.’ ti ići. you.D go.present.2NDSG b. *Želim want.present.3RDSG ‘I want you to go.’ te ići. you.A go.present.2NDSG Both (17a,b) are ungrammatical for reasons of theoretical stipulations – the rule of EquiNP deletion does not apply unless the matrix and embedded subjects are co-referential, which is not the case in (16) and (17a,b). Therefore in (16) only the complementizer insertion rule can apply and when it does, the result is in (18) below. (18) Želim da want.present.3RDSG that ‘I want/wish that you go.’ (ti) ideš. you.N go.present.2NDSG The examples discussed by Browne and Craig are referred to now as instantiations of the phenomenon of control, as CVS has already been characterized in Chapter 1. The particular sentences analyzed by Browne and Craig are said to represent subject control, i.e., the matrix subject controls the subject of the complement. Craig (1985:154) also makes note of what is known as object control – when the object of the matrix controls the subject of the complement. In Craig’s view, (19) Pomagao help.pparticiple.M.SG sam be.present.1STSG raditi. work.infinitive ‘I helped Jasna work.’ 87 Jasni Jasna.D derives from (20) [SJasna Jasna.N [Sja I [VP[Vpomogao34 help pparticiple.M.SG sam] be.present.1STSG [NPJasni] Jasna.D raditi]]] work.infinitive again due to the application of the rule of Equi-NP deletion. There is an apparent inconsistency between the representation in (16) and the one in (20) in that the embedded constituent in (16) is S ti ideš ‘you go,’ a finite constituent, whereas in (20) the embedded constituent is non-finite: S Jasna raditi ‘Jasna to work.’ Although this seems to be an important difference between the two representations, because it may be that the subject control phenomenon originates from one, and the object control phenomenon from a different configuration, Craig unfortunately does not explain exactly what, if anything, is implied by this discrepancy. Curiously, Craig also claims that, (21) Čujem hear.present.1STSG ‘I hear Ivan sing.’ Ivana da Ivan.A that pjeva. sing.present.3RDSG 34 The discrepancy between this verb and the one that is in its place in (19) is due to Craig (1985). I believe it is a simply typographical error. In any case, in (19) the matrix verb is pomagati ‘to help.imperfective aspect,’ whereas the matrix verb in (20) is pomoći ‘to help.perfective aspect.’ 88 with a perception verb as the matrix predicate, derives from the same structure as (20) above, “but the embedded clause shows up as a da construction then.” Not only is there no explanation for why one and the same structure, as in (20), serves to illustrate the derivation of two clearly different phenomena (19) and (21), but Craig also decides to name da pjeva ‘that he sings’ in (21) a da construction, which it clearly is not as it does not participate in CVS. Rather, (21) contains a prototypical that-clause as can be seen from the fact that it allows a tense-inflected predicate: (22) Čuo sam hear.pparticiple.M.SG be.present.1STSG Ivana da Ivan.A that je pjevao. RD be.present.3 SG sing.pparticiple.M.SG ‘I heard Ivan sing.’ as well as a different complementizer, unlike any configuration that represents CVS: (23) a. Čujem hear.present.1STSG ‘I hear Ivan sing.’ Ivana kako Ivan.A that b. Čuo hear.pparticiple.M.SG pjeva. sing.present.3RDSG sam be.present.1STSG Ivana kako Ivan.A that je pjevao. be.present.3RDSG sing.pparticiple.M.SG ‘I heard Ivan sing.’ c. *Želim want.present.3RDSG ‘I want/wish that you go.’ kako that (ti) ideš. you.N go.present.2NDSG 89 On two different occasions Rakić (1986,1987) outlines his basically transformational generative account of CVS. More importantly, in Rakić (1987) the two possible generative accounts presented above, transformational and rule oriented, are compared and a conclusion is reached in favor of the transformational account. Rakić (1987:94) proposes the Subjugation of Infinitive transformation, as in (24) below. (24) X hteti V 1 2 3 → 1 2 [da Present α num β per + 3] α num β per This particular transformation accounts, for example, for (25) [Sja [VPhoću I want.present.1STSG ‘I want to go to school.’ [VPići [PPu školu]]]] go.infinitive in school [VPhoću [Sja I want.present.1STSG ‘I want to go to school.’ [Sda that giving (26) [VPidem go.present.1STSG [PPu školu]]]]] in school As I demonstrated in Chapter 3, Rakić (1987:94) insists that, while finite complements in CVS do indeed behave like that-clauses, they also behave as non-finite 90 CVS complements as well. He, therefore, maintains that (26) is derived from (25) after the application of Subjugation of Infinitive transformation, which “introduces … an S node which hasn’t existed previously [, so this transformation] is a kind of lowering transformation.” Due to such a view, Rakić finds three different theory-internal arguments against the Equi NP Deletion rule and ultimately rejects it in favor of the Subjugation of Infinitive transformation. The Equi NP Deletion rule provides ad hoc explanations of two time adverbs, adverbial placement, and Wh-Fronting tests that Rakić (1987) utilized in his analysis, which I presented in Chapter 3. What, among other things, seems to be an important feature of both the convergence and divergence among all the generative accounts of CVS that I have discussed above is that all of them insist on the derivation of a CVS configuration with one possible complement (either non-finite or finite) from a CVS configuration with the other possible complement (consequently, either finite or non-finite). But at the same time, they differ in exactly which configuration might be the starting point one. Ivić (1970,1972), Browne (1968), and Craig (1985) maintain that a clause-like structure gives rise to an infinitival structure, whereas Rakić (1987) maintains that an infinitival structure transforms into a clause-like structure. This is the essence of all the generative accounts discussed here and this is where they end. What is lacking in all of them is an explanation of what exactly might control the apparent variation of the complement. 91 4.3.2. Non-Generative Explanatory Accounts Besides the generative explanatory account of CVS discussed above, there are some accounts that are explanatory in nature, but do not rely entirely on the achievements of generative linguistics. Arsenijević’s (1997:47) account, though interesting in its approach, does not provide many real explanations of CVS. The author is concerned with the character of the infinitive in Serbian and “its position among the parts of speech,” so he concludes that the nature of the infinitive can be viewed on a scale ranging from the nominal extreme to the adverbial extreme. Arsenijević, thus, concludes that CVS is due to the adverbial character of the infinitive, which can then be replaced with a da+present structure. Here is Arsenijević’s line of reasoning (1997:49). (27) Moram must.present.1STSG ‘I have to study.’ učiti. study.infinitive is a syntactic context in which the infinitive may appear. In particular, the infinitive is here “a semantic fulfillment of a true incomplete verb.” Possible substitutions of the infinitive in this syntactic context are: 92 (28) a. Moram must.present.1STSG ‘I have to study.’ da that b. Moram ovako ST must.present.1 SG this way ‘I have /to work/ this way.’ učim. study.present.1STSG /raditi/35. work.infinitive According to Arsenijević’s example sentences (28a,b), a da+present structure and an adverb can replace an infinitive here, thus the infinitive here performs the “complementary/adverbial” function. However, without the assumed infinitive in between slashes on the one hand, (28b) does not mean much without a proper context. On the other hand, with the infinitive in its place, it is clear that the infinitive is not replaced with an adverb after all – both an infinitive and an adverb are present. Arsenijević himself admits that there are two major problems with this account, though he develops his whole theory based on it. First, he notices the problem of (28b) as I have just indicated it, and, second, he suggests that different linguistic schools have different perspectives on the function that he calls “complementary/adverbial.” Clearly, his explanation does not seem to account for CVS sufficiently. I have already mentioned that Joseph (1983:141) draws the crucial conclusion that several different factors actually account for the variation of the complements. Besides the factors discussed in section 4.1. above, he cites two possible syntactic factors as well. One, originally due to Bibović (1976), in fact involves two different factors which happen to operate jointly in a CVS configuration: the presence of a specified subject and word order. In Chapter 3 I discussed Bibović’s views on CVS, or at least what according 35 Slashes are here used as they are in the original – to indicate an assumed infinitive. 93 to her analysis might seem to be CVS. Here, I simply say that it might be the point of view of her analysis (she aims at accounting for the infinitive as subject in English and Serbo-Croatian, as the title of her analysis suggests) that influenced some of her claims. The other syntactic factor is due to Pavlović (1960) and, just like Bibović’s, it also involves word order. According to Joseph (1983:141), Pavlović notes “that the infinitive persists when it follows the governing verb directly, but (generally) not otherwise:” (29) a. zaboravio forget.pparticiple.M.SG ‘I forgot to say.’ sam be.present.1STSG reći36 say.infinitive b. zaboravio forget.pparticiple.M.SG sam be.present.1STSG da that vam you.D kažem say.present.1STSG ‘I forgot to tell you.’ Pavlović’s claim, cited by Joseph, implies that the verbal complement in (29a) directly follows the verb, whereas the one in (29b) does not. While this may be generally true, it is also clear that the verbal complement in (29a) actually follows the whole matrix verbal complex, not the verb itself. Also, it is worth noting that the complement verbs are in fact not the same in (29a,b), though they are semantically very close (if not indeed the same as Joseph’s glosses suggest). This is particularly interesting if one knows that the verb reći ‘to say, to tell’ is rarely, if at all used in the present tense, except maybe in certain dialects, in either of its two possible present tense paradigms: 36 Joseph’s (1983) example (46a,b). Pavlović, however, states that all of his examples presented here are from journalistic language, precisely from the Belgrade newspaper Politika for June 16 1960, p. 570. 94 (30) a. rečem/rečeš/reče/rečemo/rečete/reku b. reknem/rekneš/rekne/reknemo/reknete/reknu say.present.1STSG/2NDSG/3RDSG/1STPL/2NDPL/3RDPL In its place in the present tense, or else in the place of the concept ‘to say.present,’ kazati ‘to say, to tell’ is almost exclusively used. Although this may turn out to be an unimportant observation, it is still a valid one, and one to keep in mind when judging examples such as (29a,b), for it may be that the variation of the complements there is due to frequency of use of the actual verbs involved. Ultimately, Pavlović’s example given in (29) above turns out not to be an instantiation of CVS after all: (31) zaboravio forget.pparticiple.M.SG sam be.present.1STSG da that sam be.present.1STSG vam kazao you.D say.pparticiple.M.SG ‘I forgot that I told you.’ While an infinitive may complement the verb zaboraviti ‘to forget,’ this verb is not among the verbs that actually participate in CVS as matrix predicates. As explained in Chapter 3, not all infinitives are necessarily instantiations of CVS. Pavlović (1960:44) cites one other syntactic factor that might account for CVS. He notes that “with the final proposition, a construction with the infinitive is preferred:” 95 (32) Ako hoćete da se vidimo, If you want us to meet, moram must.present.1STSG vam reći you.D say.infinitive da sad idem u grad. that I am going to the city right now ‘If you want us to meet, I have to tell you that I am going to the city right now.’ though Pavlović says that “in the same text, I note also the fact of a generalization of the use of da ‘that’:” (33) Izgleda da se odlučila da se ne udaljuje, da It seems that she decided not to move away that bi be.aorist.3RDSG mogla da mi bude svedok. can.pparticiple.F.SG that I.D be.present.3RDSG witness ‘It seems that she decided not to move away in order for her to be my witness.’ While (33) is indeed a final proposition, it is not clear if (32) is as well. It is, in fact, a conditional proposition, rather than final one. Therefore Pavlović’s argument does not achieve much in this particular instance. In addition, Pavlović himself claims that he finds (32) and (33) in one and the same source, but provides no explanation whatsoever as to what accounts for either the use of the infinitive or the use of da ‘that.’ In his account of CVS, Gudkov (1958:106) insists on what might be called semantic, or perhaps more precisely and accurately, conceptual factors. He demonstrates that “the contemporary Serbian authors invariably use the da construction under the condition of getting closer to a result:” 96 (34) Kada je ovaj odrastao, this one grew up, no way nikako sa njime with him NEG nije When mogao da se slaže. can.pparticiple.M.SG that agree.present.3RDSG ‘When this one grew up, he could not get along with him in any way.’ It is rather difficult to clearly understand in exactly what way the concept of ‘getting closer to a result’ is present in (34) and Gudkov does not provide any explanation in that regard nor does he give any real statistical support for his claim that what is observed in (34) indeed happens invariably. Gudkov then discusses the verb moći ‘can, to be able’ in particular and explains that “the da construction appeared with this verb in positions that better than anything else fitted the construction’s meaning … ‘getting closer to a realization of an action,’ that is to say forward direction, into the nearest futurity. The infinitive was replaced where an indication of the future had been realized.” Finally, Gudkov extends his findings to the contemporary language, based on the language of literature. There, he claims that the use of what he refers to as the da construction has a complex character based precisely on what he has found in the language of literature – directionality into futurity. He then discusses yet another factor that might cause the presence of a da+present complement in CVS According to Gudkov a finite complement is chosen “if the subject can and is ready to realize an action, can and will do something, could and did something. The infinitive is used in case an action is possible, no matter whether or not it is going to be realized.” Here Gudkov correctly notes the importance of the semantics of the two possible complements. I, however, present a similar, albeit somewhat different, view of the same in Chapter 6. 97 4.3.3. Summary The explanatory accounts of CVS discussed in this section all attempt to provide an explanation for CVS, though almost all of them are based on different assumptions. And even when they originate in the same theoretical framework, as the generative explanatory accounts do, they often differ in such essential points as the precedence of either the finite or non-finite complement in the process of derivation. For those particular accounts this issue is an important one, yet there is no agreement among the different proposals regarding this issue. The generative explanatory accounts dismissed all but syntactic factors, and those were only theory-internal, in accounting for CVS. It seems that the non-generative explanatory accounts at least provide more insights into CVS, if not a full explanation of it. And even when those accounts make claims that otherwise do not withstand more detailed syntactic tests, they are still useful in that their insights might serve as a point of departure for further investigations. 4.4. Conclusion In this chapter I surveyed a substantial number of previous accounts of CVS. Not all of the studies were necessarily concerned with CVS explicitly, though all of them discussed it at one point or another. Both Ivić’s (1972:121) and Milošević’s (1978:110) observations, quoted in the very beginning of the chapter, still hold – according to the former, probably the only point of agreement among the various studies is the problem of 98 the “uneven frequency of the infinitive over a broad territory,” while according to the latter, there is still a need for “broader investigations … [which] should provide a final account.” Indeed, the section on the regional aspect of CVS proved that various scholars generally recognize that the question of dialects plays an important, if not the crucial role in the use of either one or the other possible complement. A substantial level of agreement is also reached with respect to the sociolinguistic aspect of CVS, if these issues were discussed at all (as they were in Kravar 1953, and Joseph 1983). Still, when those two groups of factors were set aside, in what remains there is essentially little agreement as to what exactly should be, or – actually – are factors that control the variation of the complements in CVS. I classified all the accounts into four different groups, though this is not to say that there are clear divisions separating the groups. The true descriptive accounts insisted on pure descriptions of CVS, and those descriptions were, naturally, never complete, as it is impossible to list every single instantiation of CVS. This is precisely why the true descriptive accounts engage in a methodology that ultimately achieves little – they demonstrate how it is both simple and possible to place one complement where the other one is found as if nothing changes, and, according to them, it does not. The most important problem is that often one encounters sentences that are actually never and nowhere found in the language, which in turn lessens the value of the accounts themselves. When this does not happen, however, then prescriptivism takes over and dominates the actual findings with little or no true linguistic justification. 99 The descriptive-explanatory accounts resemble the true descriptive accounts in that they also maintain that the difference among the two possible complements is nonexistent, and in that they heavily rely on what is found in the language and how it can be manipulated. Still, under the pressure of the actual data, the descriptive-explanatory accounts provide certain linguistic insights into the stylistics and rhythm of the configurations, for, if everything else is equal as they claim, those might be the only differences between the complements. Moskovljević’s (1936) view of the diachrony of CVS provides a very useful and, as it turns out, a real linguistic, more precisely syntactic insight into the importance of the matrix subject for the choice of the complement. The generative accounts of CVS were exclusively concerned with the derivation of one or the other possible configuration. Thus, the question of CVS does not exist for them at all. They start from an assumption about an initial configuration and direct their focus on explaining in what way the other possible configuration comes into being. After that particular mechanism may (or may not) be explained, there are no questions left for the generative accounts. The non-generative accounts appear to be concerned with the actual explanation of CVS. In them, one finds real attempts at explaining the factors that influence one’s choice of either the finite or non-finite complement in CVS. By looking at them and by closely analyzing them, it becomes clear that an explanation might (or even must) involve at least two major groups of factors – syntactic and semantic. Unfortunately, not even non-generative explanatory accounts provide a satisfying explanation of CVS. It is clear from this sketch of the various CVS accounts’ achievements that CVS needs, and indeed deserves still more analyses that will take into consideration all the 100 previous accounts, if only to find a possible novel view of CVS. This novel view must certainly emphasize syntactic and semantic factors as controlling factors in CVS and give them appropriate treatment, which they have always lacked before. All this should be done in an attempt to provide an answer (that has yet to be given) to the question that Stevanović (1954:97) so sharply posed when he himself pondered the two different complements in CVS: “Just what sustains the use of one or the other?” 101 CHAPTER 5 A STUDY OF CVS The following facts about CVS should be well established by now: not every occurrence of the infinitive, let alone the da+present structure, is an instantiation of CVS; CVS occurs only in the configuration illustrated in (1) below. (1) Xα MATRIX [α COMPLEMENT] There are numerous factors that control the variation of the CVS complements.37 With this in mind, I now present results of my own study of CVS. I attempted to control for a set of the factors that are generally recognized to influence the choice of the complements in CVS (mainly, dialectal/regional, stylistic, rhythmic, and even certain sociolinguistic factors), and concentrated predominantly on syntactic factors: both those factors previously claimed to be important and critically discussed in Chapter 4, and those that I myself have discovered to be crucial for CVS. 37 See Chapter 4. 102 The organization of the present chapter is as follows. In section 5.1., I provide a description of the research itself. Then, in section 5.2., I present an overall picture of CVS based on the sample sentences used in the research including an analysis of the matrices and the complement variation in the language data. I test one syntactic factor claimed to be important for CVS, namely the physical distance between the matrix verb and the complement verb, as suggested by Pavlović (1960) and cited by Joseph (1983). In section 5.3., I explore various aspects of the proposal made by Pavlović and expand on it syntactically, semantically, and conceptually. In section 5.4., I demonstrate what I believe is the crucial syntactic factor responsible for CVS – the actual syntactic presence of a matrix entity (affected by the meaning of the matrix head), which invariably figures as a matrix argument. As I suggested in Chapter 4, this theory was somewhat implied, though never fully developed, by Moskovljević (1936). I here present data which prove conclusively that the choice of complement depends on whether a matrix argument figures in the CVS configuration. I use the opportunity of having conducted the research on a relatively large sample of native speakers to examine the data from a sociolinguistic perspective in section 5.5., albeit not as fully and closely as would be necessary for a complete understanding of the sociolinguistic aspect of CVS. Finally, I summarize my findings in section 5.6. 103 5.1. Research My study38 was conducted during the summer of 2004, specifically between 18 August and 19 September. The research was conducted on a representative sample of speakers only. In order to control for the dialectal or regional factors, the importance of which in CVS is well known and documented, I decided to conduct my research in one particular part of the Serbian-language-speaking area, the one that can generally be described as the territory of the city of Belgrade, the capital of Serbia. In this area, the dominant dialect is the Štokavian ekavian dialect of the Šumadija-Vojvodina type.39 However, one should keep in mind that language policy pertaining to the Serbian language originates almost exclusively in this particular area, where major linguistic, philological, and cultural institutions are located, yet this, being the capital, is an area that attracts a multitude of speakers of other dialects. Still, I believe that the speakers who participated in the research offer a fair representation of the local dialect with all its peculiarities: they were either born in the area of my interest or else spent a major portion of their life in the area. Their dialect is also very close to what might be considered the standard Serbian language, the one taught in schools and used in the media. What the participants all have in common is that they are all residents of the territory of the city of Belgrade, though not all of them were necessarily born in this area. 38 The research on which I base my theories and conclusions in this chapter has been named Complement Variation in Serbian, and it has been filed with the Ohio State University’s Behavioral and Social Sciences Institutional Review Board under the research protocol number 2004B0159. 39 See Browne (1993:382-386) for a map and description of the dialects. 104 There were 204 participants in the research, 159 women and 45 men. The participants were chosen in as arbitrary a manner as possible, but all of them participated in the research voluntarily and, in return, received no compensation for their participation. There is no intentional reason behind the discrepancy between the number of women and men, nor have I controlled for that. The age of the participants, at the time when the research took place, ranged from 21 to 66. They also differed with respect to the level of education: some participants had completed high school, others a two-year college, still others the university at an undergraduate level, or university at a graduate level (people with B.A., M.A., and Ph.D. degrees). They represented a range of professional fields: education, administration, journalism, industry, art, economy. Thus, they represent a well-educated group of people more often than not exposed to what might be understood as the standard Serbian language on a daily basis, being themselves speakers of that particular register. While all of the participants worked on the same questionnaire, they did so in different ways and at their own convenience. They were all instructed to work individually. However, this individual work varied in form: from a true individual work, in which an individual would be the only person working on the questionnaire at a given moment, to individual work in groups of varying sizes, in which the participants still worked individually, albeit at the same time surrounded by other participants. All 204 participants completed the research questionnaire, and 8 of them were asked to participate in a post-questionnaire interview as well. The choice of interviewees was as arbitrary as possible. Of the 8 interviewees, 6 were women and 2 men. Their ages ranged from 27 to 46 at the time of the research. The educational levels of those who 105 participated in the interviews ranged from high school to Ph.D. degrees. Both their vocations and the positions that they held at the time of the research differed as well. The control for stylistic factors in the language was achieved by providing everyday sentences, those that generally do not ask for much stylization in the sense of fiction authors or, for that matter, professional writers of any kind. Also, the sentences were not given in a continuum, which additionally decreased, if it did not completely nullify the issue of style in them. I controlled for the rhythmic factors as well, which are also claimed to play a role in CVS; I provided sentences whose configuration essentially resembled the formula in (1) above in the vast majority of cases. Thus the rhythm of the sentences presented to the participants was not an issue, or certainly was considerably minimized. There was only one questionnaire that all the participants were asked to complete following a set of precise instructions. 40 The participants were first informed of the purpose of the research. Then, they were given basic information about the phenomenon of complement variation in general and CVS in particular. Finally, they were told that they should complete all the tasks that the questionnaire poses based on their understanding of the given instructions as well as their own native-speaker intuition. After the participants signed the consent form 41 , they were given the questionnaire 40 See Appendix A for the actual INFORMATION FOR THE PARTICIPANTS both in English and in Serbian. 41 See Appendix B for the actual CONSENT FORM both in English and in Serbian. 106 without any time limit for completing it. 42 Still, an average time for completing the questionnaire was approximately 25 minutes. The questionnaire itself contained three major sections: the first one asked the participants for basic personal information and assigned each participant an identification code by which the participants would later be identified; the middle section was the major part of the questionnaire – it contained the actual sentences which the participants were supposed to judge; the sentences were preceded by a set of detailed instructions on how to judge the sentences; the third section consisted of three debriefing questions. The first section of the questionnaire was used for gathering data about the participants that were eventually used for a sociolinguistic analysis of the language data, which is presented in section 5.5. of this chapter. The last section of the questionnaire allowed the participants to express their own opinions about the sentences and explain how they judged the sentences. Besides being the largest part of the questionnaire, the middle section was also the most important part of the questionnaire, for it tested native-speaker intuition with regard to CVS. The set of instructions explained to the participants how to focus their attention on the sentences: they were told that all of the sentences were organized in pairs, they were next asked to judge the sentences based on their native-speaker intuition, and then they were instructed to pick the sentence within each pair which they believed (or thought or were completely sure) they would use more often than the other sentence. If they believed, or thought, or were completely sure that they would use both sentences in the given pair with the same frequency, they were allowed to circle the numbers in front of 42 See Appendix C for the actual QUESTIONNAIRE both in English and in Serbian. 107 both of them. The participants were forced to deal with only two sentences at a time and to make their judgment within that forced choice. There were 80 pairs of sentences in the questionnaire. Each pair presented an instantiation of CVS, that is, the two sentences differed only in the form of the complement, whether it was syntactically an infinitival complement or a da+present complement. All other elements of the sentences were exactly the same. The sentences were positioned parallel to each other so that the participants could look at both of them at the same time. Also, before each sentence there was a number, by which the sentences were later identified. The participants were asked to circle the number in front of the sentence or sentences of their choice. Finally, the participants were allowed to comment on the sentences; a box was placed parallel to each sentence where they could write their comments. During the course of designing the questionnaire, one of the most important decisions to make was to choose exactly what CVS matrices to include in the questionnaire. Since there is no one study or reference book that offers a list of CVS matrices, they were, first, compiled from all the previous studies to which I had access, and, second, the list of the matrices to ultimately be included in the questionnaire was made based on the frequency of its occurance in the studies that I consulted. A smaller part of the research consisted of interviews with selected participants. The interviews were held after the selected participants completed the questionnaire. They were either given a blank questionnaire or else they would keep their own questionnaire while answering particular questions regarding the sample sentences. The interviewees were asked to comment on the sentences and their overall meaning (that is 108 to say whether the sentences had the same meaning or not), to comment on their choice, to explain why they judged the sentences as they did, to determine whether there were any reasons they could cite that dictated their choice. The interviewees were not allowed to change their original choices. There was one basic question all interviewees were asked and that was to comment on the semantic sameness (or contrast for that matter) between the two sentences in each pair. I asked for no sophistication, but for the most common explanation of what the interviewees felt was important to discuss. Often, I did not even have to pose the question, but only to direct the interviewees’ attention towards the next pair of sentences. The responses made by interviewees were written down and later used in the study. After the survey portion of the research was over, all the consent forms and questionnaires were collected and stored, and the data were subsequently analyzed in various ways. As for the validity of the language data obtained through the research, a chi-squared test indicated that speakers’ choice was significantly influenced by the presence of a da+present complement vs. an infinitival complement. The chi-squared test for each individual table presented in this chapter has p<0.0001 (the actual value of p is considerably lower than 0.0001). 5.2. Overall CVS As I indicated earlier, there were 80 pairs of sentences in the questionnaire, of which only 65 fulfilled the criteria required by CVS both semantically and syntactically. Hence, only 109 those 65 pairs of sentences were analyzed for this particular study and all the counts are based on the 65 pairs of sentences only. I group the matrix predicates from the questionnaire sentences according to their part of speech. The adjectives used in the questionnaire were: lako ‘easy’ and teško ‘difficult.’ The nouns were: mogućnost ‘possibility,’ običaj ‘custom,’ pravo ‘right,’ prilika ‘opportunity,’ sramota ‘shame,’ and vreme ‘time.’ Of those six nouns, three of them, namely običaj ‘custom,’ prilika ‘opportunity,’ sramota ‘shame,’ appear as sole matrix predicates, whereas the other three, mogućnost ‘possibility,’ pravo ‘right,’ vreme ‘time,’ appear in what is traditionally referred to as izrazi ‘expressions,’ for the lack of a more precise term, but I here include them together with other nouns. Finally, the verbs used in the questionnaire were: dati ‘to give,’ dati se ‘to give in,’ dozvoljavati/dozvoliti43 ‘to allow,’ hteti ‘to want,’ imati ‘to have,’ izvoleti ‘to deign,’ mrzeti ‘to hate,’ moći ‘can, to be able,’ morati ‘must,’ nameravati ‘to intend,’ nastaviti ‘to continue,’ odbijati/odbiti ‘to refuse,’ početi ‘to begin,’ pokušati ‘to try,’ prekinuti ‘to quit,’ prestati ‘to stop,’ pristajati ‘to adhere,’ produžavati ‘to continue,’ smeti ‘to dare,’ trebati ‘to be necessary,’ učiti/naučiti ‘to study, to learn, to teach,’ umeti ‘to know how,’ uspevati/uspeti ‘to succeed,’ usuđivati se/usuditi se ‘to venture,’ uzeti ‘to take,’ voleti/zavoleti ‘to like, to love,’ znati ‘to know,’ želeti/poželeti ‘to desire.’ I now look at the actual ratio of the two possible complements with all the lexical units used in the questionnaire as CVS matrix predicates. In the present study, I refer to this overall CVS ratio simply as overall CVS. This constitutes the point of comparison for 43 When for one meaning or concept two different aspectual possibilities were used, the imperfective and the perfective variant, the two are given in that particular order, but the pair is alphabetically ordered based on the imperfective variant only. All other lexical units are given in alphabetical order also. 110 all other considerations of CVS in this research in particular and in the present study in general. The results found for the matrix predicates used in the research are taken to be the general trend in CVS in the standard Serbian language as spoken by the native speakers in the research sample. The overall statistics of the research results are given in Table 5.1. below. TOKEN # % I 2031 15.35% I/P 789 5.96% P 10413 78.69% TOTAL 13233 100% Table 5.1. Overall CVS44 What Table 5.1. roughly suggests is that, in the analyzed sample, and consequently in the Serbian language in general, when it comes to CVS more than five times as many da+present complements are used than infinitival complements. In addition to this, in only slightly less than 6% of cases, native speakers actually chose both possible complements; this suggests that there is not much vacillation in the intuition of native speakers when it comes to the phenomenon of complement variation. The percentages presented in Table 5.1., however, provide results taken across all the participants as well as all the sample sentences. Thus, it should be noted that, while the percentages appear to be clear-cut, the actual statistics with individual CVS matrices may, and in fact do, as the present chapter demonstrates, vary in various directions. 44 In all the charts in this chapter I stands for infinitival complement, I/P for infinitival and da+present complement, P for da+present complement; TOKEN # is the actual number of sentences circled by the participants and % is the percentage. 111 One other interesting piece of information obtained from the interviews and relevant for the overall understanding of CVS concerns the native speakers’ judgment of the semantic sameness of the paired sentences, which was usually taken for granted in many of the previous accounts of CVS. The eight interviewees judged all 65 CVS sentences in the questionnaire and in 71.54% (372 out of 520 pairs of sentences) of instances they determined that there exists an absolute semantic sameness, while in 28.46% (148 out of 520 pairs of sentences) of cases they had various as well as varying comments regarding the semantic nuances that differentiated the pairs of given sentences. In the latter instance, however, the interviewees never judged the two sentences to be radically different semantically. The overall statistics confirm what has already been said about CVS, namely that, on the one hand, the two possible sentential realizations of the CVS configuration present two semantically very similar if not identical sentences, and, on the other hand, that CVS is clearly dialectally or regionally conditioned, with the area under consideration in the research falling within the region in which da+present complements are a more common choice, but infinitives still robustly attested. Indeed, that is the case according to Table 5.1. above, and that was rarely, if ever, a disputed part of previous claims regarding CVS. It is other, often unmentioned or only occasionally mentioned factors, syntactic and semantic (and/or conceptual) factors, which are more closely examined in the following two sections. 112 5.3. Syntactic Factors in CVS As I indicated in Chapter 4, a claim has been made by Pavlović (1960), and cited by Joseph (1983), that one possible syntactic factor which controls the variation of the complements in CVS is the actual position of the complement with respect to the matrix in “that the infinitive persists when it follows the governing verb directly. (Joseph 1983:141)” In this section I test this particular claim that word order plays a role in CVS. I examine first whether the actual additional linguistic material between the matrix and the complement causes a change in the choice of the complement. Second, I analyze whether adverbial additional linguistic material in the complement exclusively, or in both the matrix and the complement at the same time, provokes a different choice of the complement. Third, I examine whether the presence of arguments, such as direct and indirect objects, in the matrix, in the complement, or in both the matrix and the complement results in a different ratio of the choice between the complements in CVS. Finally, I analyze whether the change of the matrix verb aspect and tense also causes a different choice of the complement in CVS. 5.3.1. Additional Linguistic Material in CVS While in many questionnaire sentences it was the case that the complement verb immediately followed the matrix, with no additional linguistic material intervening between the two, there were also sentences in which additional linguistic material physically separated the matrix from its complement verb. The additional linguistic 113 material is of various sorts: there are auxiliary verbs, personal pronouns, as well as nouns and adverbs intervening. One example of a physical separation of the matrix from the complement in the linear order of sentential elements is given in (2) below. (2) a. Morao sam must.pparticiple.M.SG be.present.1STSG mu he.D REFLEXIVE se da that mu he.D izviniti. apologize.infinitive b. Morao sam must.pparticiple.M.SG be.present.1STSG se REFLEXIVE izvinim. apologize.present.1STSG ‘I had to apologize to him.’ In (2a,b) auxiliary sam ‘be.present.1STSG’ and personal pronoun mu ‘he.D’ separate ‘must.pparticiple.M.SG’ morao from se izviniti/se izvinim ‘apologize.infinitive/present.1STSG.’ In addition, in (2b) da ‘that’ intervenes between the two as well, which is a given intervener once a finite complement appears in CVS. The CVS statistics for sentences that resemble (2a,b) in the syntactic organization, where the matrix and the complement are physically separated, are given in Table 5.2. below. TOKEN # % I 108 3.53% I/P 77 2.52% P 2871 93.95% TOTAL 3056 100% Table 5.2. CVS with the MATRIX and COMPLEMENT physically separated 114 According to the chart, it seems that Pavlović’s (1960) claim is true, in a way. Namely, it is clear from comparing Table 5.2. with Table 5.1. above that the percentage for da+present complements has increased more than 15%, while it has decreased almost 12% for infinitival complements. Another apparent fact, I believe, can be detected in the participants’ tendency to choose both complements: in the case when the matrix and the complement are separated by additional linguistic material, the percentage is less than half of what it is in overall CVS. All this indicates that, in this particular context and with this particular syntactic factor operating, the participants are extremely likely to choose da+present complements over infinitival complements as their ultimate CVS complement. 5.3.2. Adverbials in CVS I now take Pavlović’s (1960) insight regarding additional linguistic material intervening between the matrix and the complement and examine it in more detail. Here I analyze whether adverbial linguistic material in the CVS configuration has any influence on the choice of the complement. When adverbials (and by adverbials I mean adverbs and adverbial phrases) are present in a CVS configuration, besides the matrix and the complement, a CVS configuration appears as in (3) below, for example. 115 (3) a. Ume know how.present.3RDSG govoriti izokola. speak.infinitive indirectly b. Ume da govori izokola. know how.present.3RDSG that speak.present.3RDSG indirectly ‘He/She knows how to speak indirectly.’ CVS statistics for when this syntactic factor is operating are given in Table 5.3. below. TOKEN # % I 561 22.96% I/P 244 9.98% P 1639 67.06% TOTAL 2444 100% Table 5.3. CVS with adverbials While adverbials, on the one hand, are indeed additional linguistic material, on the other hand, they do not often appear to physically separate the matrix from the complement; they are not necessarily positioned between the two, as in (3) above. However, not only do the percentages differ from Table 5.2. above now that nothing intervenes between the matrix and the complement, but they also differ from the percentages in Table 5.1. as well. It is either that the presence of adverbials results in a higher percentage for infinitival complements and the choice of both complements, or else it could simply be that the particular matrices in the CVS configurations with adverbials allow for more infinitival complements. In either case, the apparently confirmed claim about additional linguistic material lowering the overall percentage of infinitival complements clearly needs further specification: although adverbials are additional linguistic material in the 116 CVS configuration, they do not provoke the same or even similar pattern in the selection of the ultimate CVS complement as general additional linguistic material does. The adverbials in the questionnaire sentences appeared in both the matrix and the complement at the same time, or else in the complement alone. An example of the former case is given in (4), and the CVS statistics for that particular CVS example are in Table 5.4. below. (4) a. Srpski narod je Serbian people be.present.3RDSG tada then ulaziti enter.infinitive istoriju. history u in modernu modern b. Srpski narod je Serbian people be.present.3RDSG tada then počeo begin.pparticiple.M.SG počeo begin.pparticiple.M.SG da ulazi u modernu istoriju. RD that enter.present.3 SG in modern history ‘The Serbian people had then begun to enter modern history.’ TOKEN # % I 35 17.16% I/P 6 2.94% P 163 79.90% TOTAL 204 100% Table 5.4. CVS with MATRIX and COMPLEMENT adverbials In the majority of cases, an adverbial appeared in the complement only – as in (5), with the CVS statistics in Table 5.5. 117 (5) a. Nastaviće raditi RD continue.future.3 PL work.infinitive po according b. Nastaviće da rade continue.future.3RDPL that work.present.3RDPL ‘They will continue to work as they have.’ TOKEN # % I 526 23.48% I/P 238 10.63% starom. old po according P 1476 65.89% starom. old TOTAL 2240 100% Table 5.5. CVS with COMPLEMENT adverbials While there is a difference in the percentages between Table 5.4. and Table 5.5., it may be unreasonable to draw any conclusions or even implications from that, since Table 5.4. illustrates only one example sentence – sentence (4) above. What I find to be much more important is the clear similarity in the percentages presented in Table 5.3. and Table 5.5. It seems that there is no particular difference in CVS with adverbials no matter whether the adverbials are found in the matrix or in the complement. Finally, I closely analyzed the complement adverbials based on their semantics. All of the complement adverbials fell into two categories: place and manner adverbials. The sentence in (5) is an example of a CVS configuration with a manner adverbial. For sentences such as this, the CVS statistics are given in Table 5.6. below. TOKEN # % I 186 15.23% I/P 116 9.50% P 919 75.27% Table 5.6. CVS with manner COMPLEMENT adverbials 118 TOTAL 1221 100% In (6) below is a CVS configuration with a place adverbial and in Table 5.7. below the respective CVS statistics. (6) a. Jovan Jovan će b. Jovan Jovan će RD FUTURE-ENCLITIC.3 SG RD FUTURE-ENCLITIC.3 SG ići na go.infinitive on da that koncert. concert ide go.present.3RDSG na on koncert. concert ‘Jovan will go to the concert.’ TOKEN # % I 375 30.66% I/P 128 10.47% P 720 58.87% TOTAL 1223 100% Table 5.7. CVS with place COMPLEMENT adverbials The difference in percentages given in Table 5.7. and Table 5.6. are interesting. While the percentage for choosing both complements is only slightly different (less than 1%), the percentages for the other two categories differ considerably. The percentage of choosing da+present complements is more than 16% higher with complement adverbials of manner than of place. Also, with place complement adverbials, infinitival complements are twice as likely to occur as with manner complement adverbials. This fact deserves a more detailed analysis, and what I discover in my data may be used just as an initial working hypothesis. Still, a priori, it would seem that there is no apparent connection between manner adverbials and the present tense, on the one hand, and 119 between place adverbials and the infinitive, on the other hand; just what the connection might be, if truly motivated, remains a question for further research. 5.3.3. Direct and Indirect Objects in CVS I further expand on the presence of additional linguistic material in the matrix and the complement in CVS. Among the questionnaire sentences there were those with direct and indirect objects as arguments, which appear in addition to subjects, of course. The direct and indirect object arguments appeared in both the matrices and the complements, as well as in both at the same time. Here, I examine whether those arguments influence the choice of the complements in CVS, and if that indeed turns out to be the case, in exactly what manner. In (7), sentences are given with object arguments in both the matrix and the complement. (7) a. Ranije nisu earlier NEGATION.3RDPL gledati ovaj watch.infinitive film. this omladini youth.D movie.A b. Ranije nisu earlier NEGATION.3RDPL da that dozvoljavali allow.pparticiple.M.PL dozvoljavali allow.pparticiple.M.PL gleda ovaj film. RD watch.present.3 SG this movie.A ‘Earlier, they did not allow youth to watch this movie.’ 120 omladini youth.D There is an indirect object in the matrix, omladini ‘youth.D,’ and a direct object in the complement, ovaj film ‘this movie.A.’ For instances of CVS such as that depicted in (7), the CVS statistics are given in Table 5.8. below. TOKEN # % I 393 8.72% I/P 225 4.99% P 3891 86.29% TOTAL 4509 100% Table 5.8. CVS with object arguments in the MATRIX and COMPLEMENT The percentages in the CVS statistics in Table 5.8. are different from the overall CVS statistics given in Table 5.1., the CVS statistics with additional linguistic material given in Table 5.2., and the CVS statistics with adverbials given in Table 5.3. above. Although the differences are not dramatic, they are still noticeable and they exist and, therefore, they must be mentioned. I now examine the presence of object arguments in either part of the CVS configuration, first in the matrix only, then in both the matrix and the complement, and finally in the complement alone. Example sentences with a matrix object argument are given in (8) below. (8) a. Uči teach.present.3RDSG ga he.A pevati. sing.infinitive b. Uči ga da teach.present.3RDSG he.A that ‘He/She teaches him how to sing.’ peva. sing.present.3RDSG 121 For sentences such as the one in (8), the CVS statistics are in Table 5.9. TOKEN # % I 17 2.79% I/P 13 2.13% P 579 95.07% TOTAL 609 100% Table 5.9. CVS with object arguments in the MATRIX Sentences from (7) above, repeated here in (9) for convenience, are example sentences for the CVS configuration in which an object argument appears in both the matrix and the complement. For such sentences, the CVS statistics is in Table 5.10. below. (9) a. Ranije nisu earlier NEGATION.3RDPL gledati ovaj watch.infinitive film. this dozvoljavali allow.pparticiple.M.PL omladini youth.D movie.A b. Ranije nisu earlier NEGATION.3RDPL dozvoljavali allow.pparticiple.M.PL omladini youth.D da gleda ovaj film. that watch.present.3RDSG this movie.A ‘Earlier, they did not allow the youth to watch this movie.’ TOKEN # % I 7 1.72% I/P 2 0.49% P 398 97.79% TOTAL 407 100% Table 5.10. CVS with object arguments in both the MATRIX and COMPLEMENT 122 Finally, in (10) below, I give example sentences with an object argument in the complement only. The respective CVS statistics for such sentences are given in Table 5.11. (10) a. Uspeo je succeed.pparticiple.M.SG be.present.3RDSG sve all teškoće. difficulties.A b. Uspeo je succeed.pparticiple.M.SG be.present.3RDSG savladati overcome.infinitive da that savlada sve teškoće. overcome.present.3RDSG all difficulties.A ‘He succeeded in overcoming all the difficulties.’ TOKEN # % I 369 10.57% I/P 210 6.01% P 2914 83.42% TOTAL 3493 100% Table 5.11. CVS with object arguments in the COMPLEMENT The percentages in Table 5.9., Table 5.10., and Table 5.11. show that, at least with respect to CVS, sentences with object arguments only in the matrix and sentences with object arguments in both the matrix and the complement at the same time pair together. Sentences with object arguments only in the complement, however, display quite different results for CVS. They allow for more infinitival complements and they also allow a higher degree of choice of both complements, something that is almost nonexistent in the case of sentences with object arguments in the matrix and the complement. 123 Finally, I analyze CVS in sentences with different types of object arguments in the complement, that is, with only the direct objects, only the indirect objects, and both types of objects at the same time. In (11), sentences with only a complement direct object are given. (11) a. Pokušao try.pparticiple.M.SG ga je he/it.A be.present.3RDSG rešiti. solve.infinitive b. Pokušao try.pparticiple.M.SG je be.present.3RDSG ga he/it.A da that reši. solve.present.3RDSG ‘He tried to solve it.’ For all the questionnaire sentences such as the ones in (22) above, the CVS statistics are in Table 5.12. below. TOKEN # % I 335 13.54% I/P 183 7.39% P 1957 79.07% TOTAL 2475 100% Table 5.12. CVS with direct object arguments in the COMPLEMENT In (12), example sentences with an indirect object in the complement are given and in Table 5.13. their respective CVS statistics are presented. 124 (12) a. Odbio refuse.pparticiple.M.SG je be.present.3RDSG pomoći mu. help.infinitive he.D b. Odbio refuse.pparticiple.M.SG je be.present.3RDSG da that mu he.D pomogne. help.infinitive ‘He refused to help him.’ TOKEN # I 20 4.90% % I/P 20 4.90% P 368 90.20% TOTAL 408 100% Table 5.13. CVS with indirect object arguments in the COMPLEMENT Finally, in (13) and Table 5.14. below, CVS configurations with both direct and indirect objects in the complement and their CVS statistics are given, respectively. (13) a. Namerava intend.present.3RDSG kazati mu tell.infinitive he.D b. Namerava da mu intend.present.3RDSG that he.D ‘He intends to tell him everything.’ TOKEN # % I 14 2.30% sve. everything.A sve kaže. everything.A tell.present.3RDSG I/P 7 1.15% P 589 96.55% TOTAL 610 100% Table 5.14. CVS with direct and indirect object arguments in the COMPLEMENT In the percentages in Table 5.12., Table 5.13., and Table 5.14 a certain kind of trend can be observed. I present it in Figure 1. below. 125 100.00% 90.00% 80.00% 70.00% 60.00% 50.00% 40.00% 30.00% 20.00% 10.00% 0.00% DO IO I DO&IO I/P P 45 Figure 1. CVS trend with object arguments 45 In Figure 1, DO stands for the CVS statistics with direct object arguments in the complement, IO for indirect object arguments, and DO&IO for both direct and indirect object arguments. 126 It is clear now that the likelihood of da+present complements appearing in CVS increases from CVS configurations with only direct object arguments in the complement, over CVS configurations with only indirect object arguments in the complement, to CVS configurations with both object arguments in the complement. Consequently, both the likelihood of infinitival complements and the choice of both possible complements decrease in that same order. This detected pattern is perhaps a trend only, for there seems to be no major or abrupt change in the choice of the complement (as the case is with parameters yet to be discussed) besides the observed apparent fluctuation in the choice. 5.3.4. Verb Aspect and Tense in CVS Finally, I analyze the possible influence of the matrix verb aspect and tense on the choice of the complements in CVS. Serbian verbs can be either of the imperfective or of the perfective aspect, as voleti ‘to like, to love’ and zavoleti ‘to begin to like, to begin to love,’ respectively. The two verbs appear in the questionnaire sentences given in (14) and (15) below. (14) a. Neka let dođe kad voli come.present.3RDSG when like.present.3RDSG plivati. swim.infinitive b. Neka let dođe kad voli come.present.3RDSG when like.present.3RDSG pliva. swim.present.3RDSG ‘Let him/her come when he/she likes to swim.’ 127 da that (15) a. Neka let dođe kad zavoli RD come.present.3 SG when begin to like.present.3RDSG plivati. swim.infinitive b. Neka let da that dođe kad zavoli RD come.present.3 SG when begin to like.present.3RDSG pliva. swim.present.3RDSG ‘Let him/her come when he/she begins to like to swim.’ For all the questionnaire sentences with the same matrix verb variation regarding verbal aspect, as in (14) and (15) above, the CVS statistics are given in Table 5.15. and Table 5.16. below, respectively. TOKEN # % I 88 3.93% I/P 82 3.66% P 2069 92.41% TOTAL 2239 100% Table 5.15. CVS with imperfective MATRIX verbs TOKEN # % I 104 5.11% I/P 56 2.75% P 1877 92.14% TOTAL 2037 100% Table 5.16. CVS with perfective MATRIX verbs Based on Table 5.15. and Table 5.16., it does not seem that a change in the matrix verb aspect significantly influences the choice of the complement. The chi-squared test yields 128 p<0.1. What happens with the percentages here does not even seem to be close to the variation observed earlier with some other syntactic factors and referred to in the present study as trends. One more set of CVS statistics are given in Table 5.17. and Table 5.18. below, corresponding, respectively, to the sentences in (14) and (15) above, with their difference in aspect between voleti ‘to like, to love’ and zavoleti ‘to begin to like, to begin to love.’ TOKEN # % I 11 2.70% I/P 13 3.18% P 384 94.12% TOTAL P 371 91.60% TOTAL 408 100% Table 5.17. CVS with VOLETI TOKEN # % I 26 6.42% I/P 8 1.98% 405 100% Table 5.18. CVS with ZAVOLETI While the percentage discrepancies between Table 5.17. and Table 5.18. are slightly higher than those between Table 5.15. and Table 5.16., I believe that they still do not possess enough force to be fully considered an influential syntactic factor when it comes to CVS, just as it has already been said about the matrix verb aspect variation. The chisquare test yields p<0.1. One last factor to be examined is the matrix verbal tense variation. 129 (16) a. Htela want.pparticiple.F.SG ga he.A je be.present.3RDSG b. Htela want.pparticiple.F.SG je be.present.3RDSG da that zvati. call.infinitive ga he.A zove. call.present.3RDSG ‘She wanted to call him.’ (17) a. Hoće want.present.3RDSG raditi. work.infinitive b. Hoće want.present.3RDSG ‘He/She wants to work.’ da that radi. work.present.3RDSG Sentences (16a,b) are examples of CVS with the matrix verb in the past tense, whereas sentences in (17a,b) present CVS with the same matrix verb, but in the present tense. The CVS statistics for the former case are given in Table 5.19. and for the latter in 5.20. below. TOKEN # % I 49 6.01% I/P 45 5.51% P 722 88.48% TOTAL 816 100% Table 5.19. CVS with past tense MATRIX TOKEN # % I 132 12.99% I/P 34 3.35% P 850 83.66% Table 5.20. CVS with present tense MATRIX 130 TOTAL 1016 100% While there are some differences in the percentages for infinitival complements and the choice of both complements, only a small difference exists in the percentages for the da+present complements. I take this to be only a small variation in CVS. It is necessary to emphasize at the end of this subsection that, while indeed both those syntactic (and/or conceptual) factors operate in the examples presented above, various other factors operate as well and at the same time: additional linguistic material, object arguments, as well as the factor of the actual individual matrices, which also influences CVS. For those reasons, results presented in this section, while initially significant and certainly true, must be taken with the necessary scientific caution. 5.3.5. Summary In this section I examine several syntactic and semantic (and/or conceptual) factors that might operate in CVS, specifically, the actual physical separation of the matrix and the complement, the presence of adverbials, and the presence of object arguments in CVS configurations, as well as matrix verbal aspect and tense. While it appeared that aspect and tense changes in the matrix do not cause significant changes in the choice of the CVS complement, the presence of various kinds of additional linguistic material did provoke some fluctuation in the choice of the CVS complement. I compare the three factors with the overall CVS count and present the results in Figure 2. below. 131 100.00% 90.00% 80.00% 70.00% 60.00% 50.00% 40.00% 30.00% 20.00% 10.00% 0.00% ADVERBIALS OVERALL OBJECT ARGUMENTS I I/P EXTRA LINGUISTIC MATERIAL P Figure 2. CVS trend with adverbials, object arguments and additional linguistic material 132 The clear trend represented in the figure is that the least number of da+present complements and the most infinitival complements are chosen with adverbials in the CVS configuration, whereas the opposite holds for the CVS configurations with any other additional linguistic material. Also, the choice of both complements decreases much like the choice of infinitival complements, though not as sharply. The three syntactic factors represented in Figure 2., with the exception of the overall CVS statistics, indeed constitute a set of syntactic factors that influence the choice of the complement in CVS and consequently influence the CVS counts presented in this section. One can notice that there is a correlation between adverbial presence and choice of infinitival complements and also between object presence and choice of da+present complements. However, it does not seem to me that either adverbial or object presence actually causes the choice of a CVS complement, for it does not seem that there is an inherent connection of any sort among the categories involved in CVS. While the categories under consideration are indeed involved in CVS, still, the variation, as the trend lines in the figure imply, is not a drastic one and it does not decisively give precedence to one syntactic factor over the others. 5.4. Controller of the Complement in CVS There seem to be one syntactic factor that clearly makes a difference in the choice of the complement in CVS: the actual syntactic presence of an entity in the matrix, which is expressed as a matrix argument, actually appears to control the choice of the complement. As I have already indicated, in Moskovljević’s (1936) view of the diachrony of CVS 133 there is only a germ of an idea that in some way coincides with what I here claim is the most important factor in CVS. In this section I demonstrate that this is indeed the syntactic factor crucially responsible for what structure is found in the complement in that that there is a correlation between this factor and the choice of either infinitival and da+present complements. I examine how this factor operates in CVS with adjectival, nominal, and verbal matrix predicates separately. 5.4.1. Adjectives In the questionnaire sentences one group of matrix predicates contained adjectives as CVS matrix predicates. The CVS statistics for adjectives as CVS matrix predicates are given in (38) below. TOKEN # % I 331 41.22% I/P 77 9.59% P 395 49.19% TOTAL 803 100% Table 5.21. CVS for adjectives The most important fact about CVS indicated by the percentages from Table 5.21. is that, unlike certain claims made by Brabec et al. (1968:258) that “today, with the majority of [CVS] adjectives, the present with da is more common,” with adjectives, in fact, the infinitive complement percentage is almost as high as the da+present complement percentage. The percentage of the choice of both complements is similar to the values shown above for the overall CVS. 134 Adjectives as CVS matrix predicates appear in two possible configurations, as illustrated in (18) and (19) below. (18) a. Teško difficult je be.present.3RDSG priznati admit.infiitive zločin. crime.A b. Teško difficult je be.present.3RDSG da that prizna admit.present.3RDSG se REFLEXIVE zločin. crime.N ‘It is difficult to admit to a crime.’ (19) a. Teško difficult mi I.D je be.present.3RDSG priznati admit.infiitive b. Teško difficult mi I.D je be.present.3RDSG da that zločin. crime.A priznam admit.present.3RDSG zločin. crime.A ‘It is difficult for me to admit to a crime.’ Adjectives either appear as what might be referred to as plain matrix predicates, as in (18), that is, their meaning is not associated with any particular entity (no entity is affected by the difficulty of the situation denoted in 18). However, there is the possibility that adjectives may indicate meaning that affects a very specific entity, as in (19) above. It is quite interesting to observe the CVS statistics for these two different sentential realizations of the CVS configuration. 135 TOKEN # % I 282 69.63% I/P 47 11.60% P 76 18.77% TOTAL 405 100% Table 5.22. CVS for adjectives without entity TOKEN # % I 49 12.31% I/P 30 7.54% P 319 80.15% TOTAL 398 100% Table 5.23. CVS for adjectives with entity The differences that exist between the percentages in Table 5.22. and Table 5.23. are greater than in the previous section. It seems that the syntactic presence of a matrix entity, plays a crucial role in exactly which CVS complement will be chosen. If the entity is syntactically present in the matrix and appears as an argument, as in (19) above, the choice of the complement is as in Table 5.23. above. The percentages in Table 5.23. are very similar to those of the overall CVS statistics. However, if the entity is not syntactically present in the matrix, and therefore there is no argument affected by the meaning of an adjective, as in (18) above, the choice of the complement is as in Table 5.22. above: the percentage values are almost reversed; the choice of infinitival complements is almost as high as the choice of da+present complements usually is. 136 5.4.2. Nouns What has just been said regarding adjectives as CVS matrix predicates seems to hold true for nouns as matrix predicates as well. The CVS statistics for nouns is given in Table 5.24. below. TOKEN # % I 317 17.28% I/P 124 6.75% P 1394 75.97% TOTAL 1835 100% Table 5.24. CVS for nouns Clearly, the overall CVS statistics for nouns in Table 5.24. do not differ much from what is already well established for CVS – da+present complements indeed dominate. Just as adjectives, nouns also appear in the questionnaire sentences in two possible sentential realizations of the CVS configuration. (20) a. Sramota shame je be.present.3RDSG govoriti speak.infinitive b. Sramota shame je be.present.3RDSG da that govori tako. speak.present.3RDSG that way ‘It is shameful to speak that way.’ 137 se REFLEXIVE tako. that way (21) a. Prilika mi opportunity I.D je be.present.3RDSG videti Pariz. see.infinitive Paris.A b. Prilika mi opportunity I.D je be.present.3RDSG da that vidim see.present.1STSG Pariz. Paris.A ‘It is an opportunity for me to see Paris.’ In (20), the matrix noun has no entity associated with its meaning; example (20) expresses a generalized attitude, more an observation of some kind. In (21), however, the meaning indicated by the noun is at the same time associated with an entity syntactically present in the matrix, where it figures as an argument. The CVS statistics for case sentences (20) and (21) above are given in Table 5.25. and Table 5.26. below, respectively. TOKEN # % I 291 35.71% I/P 106 13.00% P 418 51.29% TOTAL 815 100% Table 5.25. CVS for nouns without a specific entity TOKEN # % I 26 2.55% I/P 18 1.76% P 976 95.69% TOTAL 1020 100% Table 5.26. CVS for nouns with a specific entity Even though the differences are not as sharp as with adjectives as CVS matrix predicates (as in Table 5.22. and Table 5.23. above), they are obvious and further support the claim 138 that whether or not the entity, whose attitude (and/or initiative) towards the complement is indicated by the matrix, is syntactically present in the matrix determines the choice of the CVS complement. As already mentioned above, along with “bare” nouns as CVS matrix predicates as in (20) and (21) above, I also include what are traditionally referred to as izrazi ‘expressions,’ for they contain nouns that carry the essence of the overall meaning directed towards the complement. Example sentences are given in (22) and (23) below. (22) se a. Nema not have.present.3RDSG se b. Nema not have.present.3RDSG se REFLEXIVE REFLEXIVE vremena time čitati. read.infinitive vremena time da that čita. read.present.3RDSG ‘One does not have TIME to read.’ REFLEXIVE (23) a. Nemamo Americi. not have.present.1STPL America mogućnost studirati u opportunity study.infinitive in b. Nemamo mogućnost da studiramo not have.present.1STPL opportunity that study.present.1STPL Americi. America ‘We do not have an OPPORTUNITY to study in America.’ u in I use capital letters to indicate what the essence of the overall meaning of the matrix is (in 22 it is TIME, which one lacks, and in 23 it is OPPORTUNITY, which one misses). As is clear from (22), there is a possibility that the matrix can be depersonalized, although with both a verb and a noun, in which case no particular entity is associated with an attitude 139 indicated by the matrix. Still, as in (23), there is also a possibility that the verb is fully personalized, thus clearly indicating what the entity associated with the matrix attitude is. I analyze in detail CVS statistics for bare nouns, on one hand, and nouns in ‘expressions,’ on the other hand, as CVS matrix predicates. TOKEN # % I 204 20.00% I/P 104 10.20% P 712 69.80% TOTAL P 682 83.68% TOTAL 1020 100% Table 5.27. CVS for bare nouns TOKEN # % I 113 13.87% I/P 20 2.45% 815 100% Table 5.28. CVS for nouns in expressions There is no surprise in the actual percentages: they are not much different than those in Table 5.24. above. However, I refine the data even more and analyze whether or not the presence of a matrix entity with bare nouns (as in 20 and 21 above), on the one hand, as opposed to the presence of a matrix entity with expressions (as in 22 and 23 above), on the other hand, show any clear difference in CVS statistics. TOKEN # % I 190 31.05% I/P 102 16.66% P 320 52.29% Table 5.29. CVS for bare nouns without a specific entity 140 TOTAL 612 100% TOKEN # % I 14 3.43% I/P 2 0.49% P 392 96.08% TOTAL 408 100% Table 5.30. CVS for bare nouns with a specific entity The CVS statistics in Table 5.29. and Table 5.30. correspond to configurations such as the ones in (20) and (21) above, respectively. Again, if there is no entity in the matrix, with which an attitude towards the complement is associated, the percentage of the infinitival complements is higher than if such entity is present. TOKEN # % I 101 49.75% I/P 4 1.97% P 98 48.28% TOTAL 203 100% Table 5.31. CVS for expressions without a specific entity TOKEN # % I 12 1.96% I/P 16 2.62% P 584 95.42% TOTAL 612 100% Table 5.32. CVS for expressions with a specific entity The CVS statistics in Table 5.31. and Table 5.32. correspond to configurations such as the ones in (22) and (23) above, respectively. Once again, the syntactic presence of a matrix entity determines a significantly higher percentage of da+present complements as 141 in Table 5.32., whereas its absence determines a significantly higher percentage of infinitival complements as in Table 5.31. 5.4.3. Verbs As is well known, Serbian verbs, more often than not, inflect for grammatical person, among other grammatical categories, thus indicating the exact entity associated with the concept denoted by the verb. An example is given in (24) below. (24) Bezbojne zelene ideje spavaju colorless green ideas.N.PL sleep.present.3RDPL ‘Colorless green ideas sleep furiously.’ besno. furiously In (24), the concept denoted by the verb spavaju ‘sleep.present.3RDSG’ is inflected for 3rd person plural, just as it should be, for its subject in the sentence is bezbojne zelene ideje ‘colorless green ideas’ a plural entity, of neither 1st nor 2nd, but rather 3rd person, in the nominative case. Consequently, the same is true of Serbian verbs as CVS matrix predicates, which constitute the largest group of possible CVS matrix predicates. It is, therefore, extremely difficult to analyze them as CVS matrix adjectives and nouns have been analyzed above, for a matrix entity associated with the matrix verb more often than not will be present in the configuration. The overall CVS statistics for verbs are given in Table 5.33. below. 142 TOKEN # % I 1383 13.05% I/P 588 5.55% P 8624 81.40% TOTAL 10595 100% Table 5.33. CVS for verbs The percentages in Table 5.33. are very similar to those presented for overall CVS in Table 5.1. above, only with a slightly higher percentage for da+present complements, as can be expected based on what has just been said about Serbian verbs. One way to detect whether the syntactic factor currently under investigation also operates with verbs as CVS matrix predicates is to take a closer look at verbs that, in a way, allow for a matrix entity not to be syntactically present. In Serbian, such verbs would still be inflected for grammatical person, only this time the grammatical person would be the neutral one – 3rd person singular, neuter gender (if the form indeed inflects for grammatical gender) – and in addition, there would be a reflexive particle se ‘self,’ but in these cases without that actual reflexive meaning, added to such verbs. A verb whose grammatical person is neutral in the above sense is traditionally referred to as a depersonalized verb. Curiously enough, in the standard Serbian language there is a single verb, namely trebati ‘to be necessary,’ which is not, and cannot be, depersonalized, but is rather impersonal. This means that the verb always appears in its neutral form with respect to grammatical person, as for example in (25) below. 143 (25) a. Treba be necessary.present.3RDSG ‘We need to go to the concert.’ da that idemo go.present.1STPL b. Trebalo be necessary.pparticiple.N.SG je be.present.3RDSG na on koncert. concert da that ideš na koncert. go.present.2NDSG on concert ‘You should have gone to the concert.’ It is thus interesting to see how this particular verb behaves with respect to CVS. The questionnaire sentences are given in (26) below and the respective CVS statistics in Table 5.34. below. (26) a. Treba be necessary.present.3RDSG raditi. work.infinitive b. Treba be necessary.present.3RDSG ‘One needs to work.’ da that TOKEN # % I 64 31.37% se REFLEXIVE I/P 64 31.37% radi. work.present.3RDSG P 76 37.26% TOTAL 204 100% Table 5.34. CVS for TREBATI The difference in the percentages in Table 5.33. and Table 5.34. proves that, even with verbs as CVS matrix predicates, the factor of a matrix entity being syntactically present is an important one. The following set of examples further supports this claim. 144 Among the questionnaire sentences, there were those with depersonalized verbal matrix predicates, as in (27) below. (27) a. Može se can.present.3RDSG REFLEXIVE pogrešiti. make a mistake.infinitive b. ?Može se can.present.3RDSG REFLEXIVE da that se REFLEXIVE pogreši. make a mistake.infinitive ‘It is possible to make a mistake.’ For them, the CVS statistics are given in Table 5.35. TOKEN # % I 266 65.36% I/P 37 9.09% P 104 25.55% TOTAL 407 100% Table 5.35. CVS for depersonalized verbs Clearly, with depersonalized verbal CVS matrix predicates infinitival complements are the preferred choice, just as is the case with adjectival and nominal CVS matrices with no entity syntactically present in the matrix. The same verbs used for the statistics in Table 5.35. also appear in their respective personalized forms, as in (28) below. 145 (28) i also a. Mogao can.pparticiple.M.SG je be.present.3RDSG uraditi i do.infinitive also b. Mogao can.pparticiple.M.SG je be.present.3RDSG da that više. more uradi do.present.3RDSG više. more ‘He could have done even more.’ For them, the CVS statistics are given in Table 5.36. TOKEN # % I 51 12.53% I/P 58 14.25% P 298 73.22% TOTAL 407 100% Table 5.36. CVS for personalized verbs Again, as expected, the percentages are crucially different than those in Table 5.35., thus proving again that the syntactic presence of a matrix entity associated with matrix head is the most prominent syntactic factor in CVS. 5.4.4. Summary In this section I present what turns out to be the crucial syntactic factor in CVS – the syntactic presence in the matrix of an entity affected by the matrix semantic or conceptual implications towards the complement. The true impact of this factor on CVS is even more obvious when the data from this section are presented as in Figure 3. below. 146 100.00% 90.00% 80.00% 70.00% 60.00% 50.00% 40.00% 30.00% 20.00% 10.00% 0.00% ADJECTIVES+ ADJECTIVES- NOUNS+ I NOUNS- VERBS+ I/P VERBSP Figure 3. CVS trend for adjectives, nouns and verbs46 46 In the figure, ADJECTIVES, NOUNS, and VERBS indicate the CVS matrix type, and + and – respectively indicate the presence or absence of the matrix entity. 147 While the changes of the percentages for the choice of both complements are not all that dramatic, though the way they change is different than the way the same variable changes in Figure 1. and Figure 2. above, the changes in the percentages for infinitival complements and da+present complements are extremely abrupt and even drastic, both if considered by themselves, as in Figure 3., and particularly if considered in comparison with the same variable changes in Figure 1. and Figure 2.. When this particular syntactic factor operates in the CVS configuration, the choice of infinitival complements becomes much more important to the extent that, in the absence of a matrix entity, its percentage exceeds the percentile for da+present complements, something never witnessed before with any other syntactic factor. This is exactly why I claim that whether the CVS matrix contains an argument affected by the matrix head uniquely controls the syntax of the complement. 5.5. Sociolinguistic Factors in CVS In this section I present a sociolinguistic view of the data that I obtained during the research. The sociolinguistic factors taken into consideration are sex, level of education and age. All the comparisons are based on what I referred to and presented in Table 5.1. above as overall CVS but for convenience, repeated here in Table 5.37. TOKEN # % I 2031 15.35% I/P 789 5.96% Table 5.37. Overall CVS 148 P 10413 78.69% TOTAL 13233 100% In this section, the major goal is to observe whether the sociolinguistic factors under consideration provoke changes in the choice of the CVS complement, and if so, exactly what kind of changes. 5.5.1. Sex I indicated earlier that there were 204 participants in the research of whom 159 were women and 45 men. Their respective overall CVS statistics are given in Table 5.38. and Table 5.39. below. TOKEN # % I 1746 17.06% I/P 566 5.52% P 7925 77.42% TOTAL P 2253 77.10% TOTAL 10237 100% Table 5.38. CVS for women TOKEN # % I 446 15.26% I/P 223 7.63% 2922 100% Table 5.39. CVS for men The percentages in Table 5.38. and Table 5.39. allow for at least two important observations. First, there seems to be no apparent difference in CVS between women and men who participated in the research. Second, neither the women’s nor the men’s 149 percentages differ from the overall CVS percentages. Still, one interesting observation could be made about the fact that, apparently, women are slightly more likely (less than 2%) to use infinitival complements than men are. On the other hand, that is compensated by the fact that men allow for a higher percentage (slightly over 2%) of the choice of both complements. 5.5.2. Education The level of education of the participants varied from those who had completed high school to those with a Ph.D. degree. I now examine whether the level of education, at the moment when the research took place, causes any changes in the choice of the complement. I divide all the participants into three groups. In the first group are those who, at the time when they worked on the questionnaire, have completed high school; it did not matter whether, at the given moment, they were attending university or not. The total number of the participants in the first group is 38. The second group is composed of those who have completed what is in the Serbian educational system known as viša škola ‘higher school,’ the approximate equivalent of a two-year level college in the U.S.A. educational system. The number of the participants in this group is 60. Finally, all the participants who have earned their B.A., M.A., and Ph.D. degrees are in the third and largest educational group. There are 105 participants in this group. One person did not declare his or her level of education, so the total number of the participants under consideration in this count is 203. 150 The CVS statistics for the three respective educational groups determined based on the level of education of the participants are given in Table 5.40., Table 5.41., and Table 5.42. below. TOKEN # % I 407 16.48% I/P 130 5.26% P 1933 78.26% TOTAL P 3018 78.49% TOTAL P 5375 78.79% TOTAL 2470 100% Table 5.40. CVS for high school TOKEN # % I 669 17.40% I/P 158 4.11% 3845 100% Table 5.41. CVS for college TOKEN # % I 946 13.87% I/P 501 7.34% 6822 100% Table 5.42. CVS for university There seems to be no one clear trend in the change of the CVS percentages in Table 5.40., Table 5.41., and Table 5.42. That is to say, the percentages for da+present complements are very similar, though there is a slight increase from Table 5.40. over Table 5.41. to Table 5.42.; still the differences in the percentages are less than one half of one percent. On the other hand, there is no decrease in the percentages for infinitival complements that would conversely parallel the increasing trend of the percentages for da+present 151 complements. The highest percentage for infinitival complements is found in Table 5.41., a slightly smaller percentage in Table 5.40., and the lowest percentage is in Table 5.42. This last observation would appear to indicate that higher educational level does not necessarily imply that more infinitives will be used. Also, it seems very clear that different levels of education do not play an important role in CVS. 5.5.3. Age The age of the participants ranged from 21 to 66. Here, I analyze whether the age of the participants influences their choice of the complement in CVS. I decided to group all the participants into three age groups. There are 11 participants in the first group, those ages 21 to 25. In the second group there are 107 participants whose age ranges from 26 to 40. Finally, those ages 41 to 66 are in the third group, 86 of them. Their respective CVS statistics are presented in Table 5.43., Table 5.44., and Table 5.45. below. TOKEN # % I 148 20.73% I/P 22 3.08% P 544 76.19% TOTAL P 5270 75.96% TOTAL 714 100% Table 5.43. CVS for 21-25 TOKEN # % I 1165 16.79% I/P 503 7.25% Table 5.44. CVS for 26-40 152 6938 100% TOKEN # % I 877 15.72% I/P 264 4.73% P 4437 79.54% TOTAL 5578 100% Table 5.45. CVS for 41-66 Interestingly, the youngest age group allows for the most infinitival complements, whereas the oldest age group allows for the most da+present complements. Đukanović’s (1986) analysis indicated that, over time, the percentage of infinitival complements decreases and, consequently, the percentage of da+present complements increases. While the time span in the research that I conducted is not as long as the time span of the novels analyzed by Đukanović, it is still important to emphasize that, at least based on the percentages in Table 5.43., Table 5.44., and Table 5.45., a possible loss of the infinitive is not as clear a process as sometimes may seem or is implied. The percentage of infinitival complements is lowest in the oldest age group, it rises slightly over 1% in the middle age group, and then it rises an additional almost 4% in the youngest group. Also, the sum of the percentages for infinitival complements and the choice of both complements in the middle and youngest age groups is very similar, around 24%, while only less than 21% in the oldest age group. Admittedly, my research is but a snapshot, a short-term view of what is a constant long-term trend. 153 5.5.4. Summary This brief analysis of three sociolinguistic factors, namely sex, level of education, and age, and their impact on CVS, demonstrates that, when controlled for dialectal or regional factors, as in the case of this particular research, they do not cause major variation in the choice of complements in CVS. The percentages presented in this section did not indicate that the analyzed factors had demonstrable effect in causing dramatic departures from the overall CVS percentiages given in Table 5.1. and repeated in Table 5.37. above. There were only minor fluctuations in the percentages in this section. Still, two interesting observations with respect to those fluctuations can be made. First, it appears that female participants utilized more infinitival complements than male participants in the research; the latter, however, allowed for more of the choice of both complements. Second, the youngest participants used more infinitival complements than the other two age groups in the research, and, together with the middle age group, they used more infinitival and both complements than the rest of the participants. 5.6. Conclusion The main goal in this chapter was to present findings based on the research that I conducted in order to test as many different factors that might influence CVS as possible. The vast majority of those factors were syntactic, although I also discussed some semantic and sociolinguistic factors as well. The investigation of the obtained results indicated, first, that while there were some significant fluctuations in the percentages for 154 semantically different adverbials, i.e., manner and place adverbials, a further investigation into the question of possible correlations between different adverbials and the two possible CVS complements is necessary for any kind of a firmer claim to be made in this regard. Second, the sociolinguistic factors did not play a significant role in CVS when compared to both the semantic and syntactic factors. Third, the syntactic factors did not influence CVS in a uniform way: there was only one syntactic factor that actually caused significant changes in the choice of the complement, while others were responsible only for various trends in choosing the complement. The sociolinguistic factors that I tested were sex, educational level, and age of the participants. The first factor naturally dealt with two variables, female and male, whereas the other two dealt with three variables each: high school, college, university, and 21-25 years of age, 26-40 years of age, 41-66 years of age, respectively. Generally speaking, none of the variables induced significant changes in the choice of the CVS complement. It seems that, when full control of dialectal or regional, as well as stylistic and rhythmic factors is achieved, as in the case of this particular research, sociolinguistic factors such as sex, educational level, and age do not influence CVS at all. Still, the most interesting observation was that the infinitive is not quickly disappearing from the language of the youngest group of people who participated in the research. On the contrary, they allowed for the most infinitival complements of all three age groups. Exactly why this is so cannot be said based solely on the research. Still, it is possible that the level of education played a role: just 4 out of 11 members of the youngest research age group completed only high school. However, it is possible that other sociolinguistic factors, such as exposure to foreign languages, English before all, particularly nowadays with the Internet culture as a 155 source of exposure to English, were at play as well, though they were not tested in this research, nor did I control for them. Be that as it may, the sociolinguistic take on the research results proved that, although Serbian is in the area characterized by the loss or retraction of the infinitive as a verbal category, it appears that this category is still very much alive in the language. The syntactic factors that I tested in the research were much more significant. As a point of departure I used Pavlović’s (1960) insight, also cited by Joseph (1983), according to which “the infinitive persists when it follows the governing verb directly, but (generally) not otherwise.” I analyzed whether the presence of any additional linguistic material, as I referred to it, which would physically separate the matrix from the complement, influenced CVS in any way. It turned out that there was a minor change in the choice of the complement in that, when there was any additional linguistic material, the participants would choose more da+present complements. However, I utilized Pavlović’s insight and expanded its implications to very specific additional linguistic material, namely adverbials and object complements. This resulted in observations summarized in Figure 2 in this chapter: with adverbials, the choice of infinitival complements was the highest, and the choice of da+present complements was the lowest. The decrease of the former and the increase of the latter surpassed the overall CVS statistics, as well as the percentages with sentences with object arguments, only to reach the opposite extremes with sentences with any other kinds of extra-linguistic material. This is a somewhat telling observation, which – in a way – does support Pavlović’s claim, though it specifies and clarifies it in greater detail, still, I maintain that the changes in the choice of the complement induced by these particular syntactic factors should be treated 156 as trends. It did not seem that any of the analyzed factors caused changes that would actually be critical for CVS. This proposal becomes even more significant when a syntactic factor that induces such crucial changes is discovered. My analysis of the research data demonstrated that the most crucial factor in CVS is the actual syntactic presence (or absence, for that matter) in the CVS matrix of an entity affected by the concepts employed by that same matrix, which is invariably a matrix argument. In section 5.3. and in Figure 3. above I demonstrated that, on one hand, the presence of such entity indeed yields CVS counts similar to the other CVS counts. On the other hand, however, conceptual and syntactic absence of the entity yields CVS counts that are unparalleled in that the choice of infinitival complements either became several times greater than with the entity present or simply overall dominant in the sense that the choice of da+present complements in CVS normally is. It is in this factor that, all other things being equal, I find the true syntactic impetus for CVS and it is this factor that will, before all other factors, be in the focus of an attempt to formalize the syntactic mechanism of the CVS configuration. 157 CHAPTER 6 TOWARDS A FORMALIZATION OF CVS In Chapter 5 I presented various descriptive generalizations with respect to possible syntactic factors that control CVS, of which only one proved to be crucially responsible for the actual variation of the complements. In this chapter I provide one possible way to formalize the most important syntactic factor in CVS, the actual syntactic presence (or absence, for that matter) in the CVS matrix of an entity invariably expressed as a matrix argument. In addition to this, I also provide a more thorough account of the two parts of the CVS configuration, namely the matrix and the complement; this account will be in line with the theoretical framework that I adopt in the present study. A configuration claimed to be the only one in which CVS takes place is repeated in (1) below, for convenience. (1) Xα MATRIX [α COMPLEMENT] 158 The formula has been patterned after Culicover and Jackendoff’s (2005) 47 views of English phenomena in certain respects similar to CVS. Also, until this point, the formula has been kept in as generic a form as possible in order to include all the different instantiations of CVS discussed thus far. Most importantly, however, the formula – in most general terms – represents the linguistic phenomenon known as control, of which CVS is an interesting example. It is not my goal in this study to engage in the debate about the precise nature of control; CJ (2005:415-427) provide an overview of this debate. Although an examination of control in CVS may have an impact on our understanding of control as a general linguistic phenomenon, the focus of this study is CVS alone. By focusing on CVS I hope to provide an indicator of exactly how complex the phenomenon of complement variation in Serbian actually is, but, at the same time, how complement variation phenomena in languages other than Serbian could possibly be analyzed. In order to address all these issues, I begin by explaining the essence of the phenomenon of control and the place CVS occupies within this phenomenon in section 6.1. In section 6.2., I present the theoretical framework whose basic assumptions best suit what has been said about CVS in the present study up to this point; the framework is that of CJ (2005) which they refer to as simpler syntax. I then explain how the CVS matrices and complements fit the essential assumptions of the adopted theoretical framework in section 6.3. In section 6.4., I develop a formula based on the theoretical framework as well as my findings presented in Chapter 5; the formula accounts for the crucial syntactic 47 Hereafter CJ. 159 factor that controls CVS. Finally, in section 6.5. I summarize my claims presented in the chapter. 6.1. Control and CVS To develop the claim that CVS is indeed an instantiation of control, some basics about control are needed. CJ (2005:415) indicate that control “has been absolutely central to mainstream theory for forty years.” Linguistically the most fascinating aspect of control is probably best understood from a descriptive observation that there exist certain complements without one of their arguments overtly present in the syntax, but interpreted (i.e., understood) nevertheless; furthermore, the syntactically absent argument appears to be the same as an argument of the matrix. To illustrate this descriptive observation, CJ give the following (CJ 2005:415 example 1a). (2) Johni likes [to idance with Sarah] In (2), John likes is the matrix and to dance with Sarah is its complement. The matrix verb, likes, is, in Culicover and Jackendoff’s terms, a two-place function or else the function of two arguments; that is to say, likes has two arguments. In (2) above, one argument of likes is John, and the other argument is to dance with Sarah. The complement verb to dance is a one-place function; it has one argument, namely the dancer, which is expressed as its subject. However, in the configuration in (2) above, the dancer is not syntactically present in the complement itself. Intuitively, however, the 160 dancer is understood as John, already the matrix verb argument. Thus, John is the controller of the complement, and John is said to control the complement. It is this kind of correlation between a matrix argument and a syntactically non-existent complement argument, which is invariably the complement’s local subject, which is referred to as control. In order to formally indicate this kind of correlation between the two arguments, conventionally a set of indices, or else subscripted numerals, is used. Unconventionally, however, CJ (2005:416,n1) decide to place the indices as in (2) above “so as not to prejudice whether the infinitive has a genuine syntactic subject.” Index i is used merely to indicate the descriptive generalization about the sentence in (2) above. It does not substitute for any argument, particularly not in the complement. In the present study, I follow CJ’s manner of indexing control. According to CJ, the phenomenon of control has been widely debated, but all of the individual debates can generally be characterized as either those insisting on syntactic factors as the basis for the analysis of control or those insisting on semantic factors as the basis. Be that as it may, ultimately “the problem of control concerns how to determine the understood subject (CJ 2005:415)” in configurations such as the one in (2) above. CJ’s solution for control is semantic in essence, though they cannot avoid some syntactic factors that play an important role in control. The major argument for an essentially semantic approach to control is found in the fact that numerous instantiations of control, arguably more complicated than the one in (2) above, simply cannot be explained in purely syntactic terms without ad hoc theoretical internal stipulations, as CJ (2005:41-420) demonstrate. 161 It has already been explained in various places in this study, most clearly in Chapter 1, that CVS is an instantiation of control. I use three prototypical pairs of sentences to illustrate this: (3) a. {Mogao je}48i can.pparticiple.M.SG be.present.3RDSG b. {Mogao can.pparticiple.M.SG je}i be.present.3RDSG iuraditi i do.infinitive also that više. more da iuradi do.present.3RDSG i više. also more ‘He could have done even more.’ (4) a. Prilika mii opportunity je I.D Pariz. ivideti be.present.3RDSG see.infinitive Paris.A b. Prilika mii je da RD opportunity I.D be.present.3 SG that ‘It is an opportunity for me to see Paris.’ (5) a. Teško mii difficult I.D je be.present.3RDSG ividim see.present.1 ipriznati admit.infinitive b. Teško mii je da difficult I.D be.present.3RDSG that ‘It is difficult for me to admit to a crime.’ ST SG Pariz. Paris.A zločin. crime.A ipriznam zločin. admit.present.1STSG crime.A The indices in (3)-(5) demonstrate the relationship of control that exists between matrix arguments and the complements. 48 Curly brackets are used to fully encompass periphrastic verb forms in Serbian, of which past tense and future tense will figure in the present study. The periphrastic verb forms generally contain an auxiliary verb and a main verb; only together do the two parts provide complete information on the argument expressed as subject of the verb. Thus, in (3a,b), only mogao je ‘he could’ indicates that the subject is an entity grammatically characterized as 3RD.SG.M. 162 One note is in order here. Since Serbian verbs can be inflected for grammatical person, number and gender, the explicit presence of the local subject is not required, rather, the absence of the local subject is preferred. Such is the case in (3a,b) above, where the local subject on ‘he’ is only implicitly present in the sentence but is recoverable from the morphology of the verb form mogao je ‘can.3RD.SG.M.’ Still, this does not mean that in (3a,b) above, while clearly there is no local subject in the complement, there is also no local subject in the matrix. On the contrary, the local matrix subject is understood but recoverable from the verbal morphology of the matrix. Moreover, it is the controller of the complement, more precisely the syntactically nonexistent local complement subject, which is not recoverable from the verbal morphology of the complement, but only from the fact that the matrix subject controls the complement subject. CVS, as just analyzed above, thus presents an instantiation of control. What is more interesting about examples in (3)-(5) above is that they actually avoid the most important question of the problem of control, namely how the understood local subject of the complement is determined. CVS could not take place unless there is an absolute equation, a co-reference in the sense of (3)-(5), between the matrix argument and the complement argument expressed as the subject. This is the critical condition under which CVS occurs. Only under this condition are all the variations of the complements, observed in (3)-(5) above, possible. If this condition is not met, then there is no CVS. 163 (6) a. Teško difficult I.D mii je be.present.3RDSG that da zločin. jpriznate ND admit.present.2 PL crime.A b. *Teško mii je jpriznati difficult I.D be.present.3RDSG admit.infinitive ‘It is difficult for me that you admit to a crime.’ zločin. crime.A In (6a), there is no correlation between the matrix argument and the complement local subject argument, as indicated both by the glosses and indices. Therefore, (6b) is ungrammatical under the assumption that it should denote the same meaning as the one in (6a). Crucially here, while (6b) is a grammatical Serbian sentence in (5a) above, and in many other contexts as well, it is ungrammatical in (6) under the proposed indexing. Consequently, there is no CVS. It seems then that once particular instantiations of CVS are clearly determined, the most important question of the problem of control according to Culicover and Jackendoff becomes irrelevant here, because only a matrix argument can be the controller of the complement, more precisely its local subject argument. For CVS itself, however, the most important question becomes why there are two possible complements if there is only one unique controller and only one uniquely controlled argument. I provide a formal mechanism in answer to this question in section 6.4. 6.2. A Theory of Control In Chapter 3 I demonstrated how the majority of previous CVS accounts analyzed the matrix CVS predicates from a semantic point of view, which I myself have found to be the most accurate approach. Thus even a novel theoretical account of CVS should take 164 into consideration the semantic aspect of the phenomenon itself. Consequently, a view of control which recognizes semantic (and/or conceptual) factors as crucial is therefore a suitable approach to apply to CVS. I, therefore, adopt CJ’s view of control and propose that it best formally defines the descriptive truths that hold in CVS. Before I demonstrate exactly how CVS is formalized with regards to the theory of control, in this section I present only those aspects of CJ’s theory that are relevant for CVS. 6.2.1. Semantic Treatment of Control CJ base their essentially semantic treatment of control, in part, on examples such as (7) below. (7) a. Johni persuaded Sarahj to j/*idance. b. Johni promised Sarahj to i/*jdance.49 Syntactically, (7a) and (7b) are the same, CJ claim, although the control properties of persuaded and promised are different: with persuaded, Sarah controls dance, while with promised, John does, as indicated by indices. CJ (2005:419) assert that “[s]hould one wish to find a relevant syntactic difference between [(7a) and (7b)], it has to be motivated by the dogma that control is syntactic; there is no independent motivation.” However, 49 Brian Joseph (personal communication) indicates that there are some English speakers that allow the following reading of (7b): i. Johni promised Sarahj to i/jdance. where both John and Sarah appear as possible controllers of dance, though not at the same time. Here, however, I follow CJ’s judgments. 165 there is a certain level of constancy in (7a,b) as they claim, “a constancy at the level of thematic role.” In their view, this gives the level of conceptual structure (CS) precedence over the level of syntactic structure; according to them, control is “a relation stated over the level of conceptual structure rather than over syntactic structure.” Theoretically, CJ present three reasons for why control should be accounted for like this: “[first,] [a]t the level of CS, syntactically implicit arguments are explicit; [second,] [a]t the level of CS, the meanings of verbs are explicitly represented, in such a way that they can directly bear on control relations without special added machinery; [and third,] the association of control with constant thematic roles is most natural at CS, the level at which thematic roles are structurally represented.” Even though semantic treatment of control is not a novel approach to this linguistic phenomenon – in fact CJ list an extensive list of previous semantic, as well as syntactic, accounts of control stemming from as early as Jackendoff (1969) and Rosenbaum (1967) – this approach has an advantage over the previous semantic accounts in that it does not require that control be “handled in terms of item-by-item lexical marking”; rather it is handled “as an organic part of meaning. CJ (2005:420)” What this means “is that the control behavior of persuade and promise is an essential part of their meanings; there could not be a verb that meant the same thing as persuade but that had the control behavior of promise. This requires a level of representation where the requisite aspects of meaning are structurally explicit: conceptual structure.” When analyzing some previous accounts of CVS in Chapter 3, I implied that semantic (and/or conceptual) character of the matrix predicates indeed plays an important 166 role in CVS. This is why CJ’s view of control is the one that will account both for what has already been said and for what is yet to be said about CVS. 6.2.2. Treatment of Matrices CJ’s (2005:444-445) semantic view of control is put to practical use in their semantic treatment of both matrices and complements. The matrices, more precisely lexical items that control their respective complements, “fall into a delimited number of semantic classes”: INTEND, OBLIGATED, ABLE, SHOULD, CS50, REQUEST. According to the authors, “each class determines a particular thematic role that serves as controller; each of these can serve as a component of the meaning of verbs, nouns, and/or adjectives,” and, crucially, “each basic predicate establishes a control relation” between its complement and one other argument. CJ base their treatment of matrices on the fact that there is “a limited number of basic predicates in CS that select actions a arguments.” Only when all of the conditions listed above are met does unique control obtain. CJ (2005:432-444) present a set of examples in order to demonstarte that the difference in controller choice as in (7a,b) above, with persuade and promise, has “to do with the meanings of these predicates” and consequently develop their theory of control based on this. The complements are treated semantically as well. In the complement, “any sort of state or event can appear” and for those CJ (2005:427) use the term situation. One subclass of situations is actions, which, furthermore, can be divided into voluntary and 50 In CJ (2005) CS stands for force-dynamic predicates. 167 non-voluntary actions, the latter thus being non-actions. It is in this particular semantic characterization of possible complements, as well as matrices, that CJ find the basis for their typology of control. Here too I find the particular theoretical framework appropriate for CVS. 6.2.3. Treatment of Complements Being concerned with the question of how to determine the understood subject in the controlled complements, CJ (2005:421-424), recognize one important dimension “of variation in control … the choice among ‘free,’ ‘nearly free,’ and ‘unique’ control.”51 According to CJ, “free control is a configuration in which the range of possible controllers includes (a) any NP in the sentence or surrounding discourse plus the speaker and hearer, (b) the possibility of split antecedents, and (c) the possibility of a generic controller”; nearly free control occurs if “the controller may be either of two NPs in the sentence; [and] split antecedents and generic controllers are also possible” as well as discourse control in certain circumstances; finally, unique control is “[t]he most restricted form of control … [which, in standard examples, occurs when] there are two possible targets of control in the matrix clause, but only one of them can serve as controller.” In addition to this, and within unique control, there is one other possible choice of control, that of ‘unique+generic’ control. Unique+generic control occurs when both unique and generic control are possible, the latter being control by an entity conceived of in the abstract. 51 CJ discuss two additional dimensions of variation in control: ‘exhaustive’ vs. ‘partial’ control and ‘obviative’ vs. non-obviative’ control; however, they are not relevant for the present study. 168 As I indicated earlier in this section, in CVS the choice of the controller is invariably clear, thus CVS is an instantiation of unique control. According to CJ (2005:427), a very specific semantic kind of complement participates in unique control, namely actional complements. The authors formulate their theory of unique control by proposing the Unique Control of Actional Complements (UCAC) Hypothesis: “Infinitival and gerundive complements that are selected by their head to be of the semantic type Voluntary Action have unique control. The unique controller is the character to which the head assigns the role of Actor for that Action – whatever its syntactic position.” 6.2.4. Formalizing Control CJ (2005:444-445) formalize the descriptive notion of a voluntary action as x ACT, and the notion of an actional complement, which appears as the semantic argument as [x ACT]. Basic semantic predicates that select for actional complements are formalized as in 6.2.2. above. Using one of them, INTEND, as an example, the structure of a basic predicate is formalized as in (8) below (CJ’s 2005:445 example 72): (8) a. standard notation: [INTEND (Xα, [ACT(α)])] b. simplified notation: Xα INTEND [α ACT] Both formalizations indicate the same thing: there exists a type of semantic binding between an actional complement of INTEND and one other (the only one possible, that is, for INTEND is a two-place function) argument of INTEND. The semantic binding is 169 notated by a Greek variable in the controlled actional complement that corresponds to the other argument’s superscript. Also, both argument positions and semantic (selectional) restrictions on INTEND are notated in italics. The formulae in (8a,b) do not hold for the basic predicate INTEND only; they hold for “any verb that contains this predicate as part of its meaning.” In the present study, the simplified notation is adopted for the formalization of CVS. 6.2.5. Summary CJ’s view insists on the idea that control is semantically based. It requires a precise semantic account of both matrices and complements, because the semantics of the two establishes a certain connection between their arguments on the level of conceptual structure. This also provides the necessary conditions for a typology of control that eliminates the syntactic position of an argument; it becomes irrelevant whether the controller is a subject or an object. What is important, however, is that an argument’s thematic role is based on the basic semantic predicate with which the argument occurs. 6.3. CVS as Control The claim that CVS is an instantiation of unique control must be tested according to CJ’s UCAC Hypothesis. The hypothesis itself accounts for unique control in English, about which the authors themselves note: “Our study suffers from the limitation that it is 170 restricted to English. Our impression from the literature … is that control behaves crosslinguistically in much the same fashion” CJ (2005:417). In this section I demonstrate that CVS generally follows the basic assumption of the UCAC Hypothesis, though it requires that the hypothesis be slightly modified to fully accommodate language-specific facts. The first necessary modification concerns the type of complements in a unique control configuration. According to the hypothesis, infinitival and gerundive complements have unique control. As it is clear by now, infinitival and da+present complements are the two CVS complements and they have unique control. The second necessary modification is discussed in sub-section 6.3.2. 6.3.1. CVS Matrices Here I analyze CVS matrix predicates from the standpoint of the adopted theoretical framework. CVS matrices indeed fall into the proposed set of basic semantic predicates, though not as elegantly as it might be desired. Namely, not all the proposed basic semantic predicates participate in CVS as it is defined in the present study, and not all CVS matrices are found among the proposed basic semantic predicates. In the discussion concerning the semantics of the MATRIX, I use the generic term COMPLEMENT, in bold, to represent all the possible complements. As is the case in English unique control, in CVS the predicate NAMERAVATI ‘INTEND’ is one of the predicates that require unique control. This basic predicate is a two-place function, thus having two arguments, as it is formalized in (9) below. 171 (9) Xα NAMERAVATI [α COMPLEMENT] One argument of NAMERAVATI ‘INTEND’ is the intender, while the other is the COMPLEMENT. The COMPLEMENT itself has at least one argument that is semantically bound to the intender, that is to say, the intender’s intention is about the COMPLEMENT. The actual CVS verbs used in the questionnaire which contain this basic semantic predicate as part of their meaning are nameravati ‘to intend,’ pokušati ‘to try,’ pristajati ‘to adhere,’ usuđivati se/usuditi se ‘to venture,’ and uzeti ‘to take.’ Another CVS basic semantic predicate that requires unique control is UMETI ‘ABLE.’ It is again a predicate that has two arguments, so its CVS configuration is formalized in (10). (10) Xα UMETI [α COMPLEMENT] The semantic binding between the entity that possesses an ability and an argument of the complement is the same kind as for NAMERAVATI ‘INTEND.’ There were several verbs in the questionnaire whose meaning contains this basic semantic predicate as its component, primarily moći ‘can, to be able,’ umeti ‘to know how,’ roughly ‘to possess ability to do something,’ uspevati/uspeti ‘to succeed,’ roughly ‘to reach a goal by being able to do something,’ and znati ‘to know,’ roughly ‘to possess ability to do something.’ CJ explain, however, that this same basic predicate figures as a component of meaning of such verbs as učiti/naučiti ‘to study, to learn,’ roughly ‘to come to be able to do something’ and učiti/naučiti ‘to teach,’ roughly ‘to cause to come to able to do 172 something.’ The former verb requires unique control much the same as NAMERAVATI ‘INTEND’ and UMETI ‘ABLE;’ the formal configurations in which unique control occurs are exactly the same as in (9) and (10) above, as I demonstrate in (11). (11) Xα UČITI/NAUČITI [α COMPLEMENT] The latter verb, however, is different in that the argument of the COMPLEMENT is semantically bound to the argument of UČITI/NAUČITI ‘TEACH’ that receives instruction. This is formalized as in (12) below. (12) Xα UČITI/NAUČITI Yβ [β COMPLEMENT] The notion of normativity represents another basic semantic predicate for which CJ (2005:447) assert ranges “over the various senses of the root modal should.” Hence, the formalization of the corresponding CVS matrix verb appears as given in (13). (13) Xα TREBATI [α COMPLEMENT] Just as in the case of CJ’s study, various verbs contain, or may contain, the basic semantic predicate in its meaning. Such CVS verbs from the questionnaire are imati ‘to have,’ roughly ‘to be required to do something,’ morati ‘must,’ smeti ‘to dare,’ roughly ‘to possess what is necessary to do something,’ and trebati ‘to be necessary.’ 173 The so-called force-dynamic predicates are, in CJ (2005:447) words, “a slightly more complicated case.” No doubt because they involve different nuances of meaning, which are “causing, preventing, enabling, and helping; … variants in which the outcome is uncertain, such as pressuring and hindering; … predicates both in the physical domain such as pushing and in the social domain such as encouraging.” A prototypical CVS verb from the questionnaire with the force-dynamic basic semantic predicate as a component of its meaning is DOZVOLJAVATI/DOZVOLITI ‘ALLOW,’ which appears in the configuration formalized in (14) below. (14) Xα DOZVOLJAVATI/DOZVOLITI Yβ [β COMPLEMENT] DOZVOLJAVATI/DOZVOLITI ‘ALLOW’ is a three-place function, a predicate with three arguments, one of which is the COMPLEMENT, and the other two are the antagonist or agent (Xα) and the agonist (Yβ). As the formula in (14) indicates, the semantic binding relation extends between one complement’s argument and the agonist. Besides dozvoljavati/dozvoliti ‘to allow,’ dati ‘to give,’ roughly ‘let,’ dati se ‘to give in,’ izvoleti ‘to deign,’ and odbijati/odbiti ‘to refuse’ also belong to this group. The former two indeed operate according to the formula in (14). The latter three, however, operate according to the formula in (15). (15) Xα ODBIJATI/ODBITI [α COMPLEMENT] 174 One basic semantic predicate that CJ do not list among those that require unique control, but which indeed occurs in CVS, is what I term FAZA ‘PHASE.’ This is a twoplace function, thus the configuration in which all the verbs that contain this as part of their meaning is formalized as it is given in (16) below. (16) Xα FAZA [α COMPLEMENT] The CVS verbs from the questionnaire that contain this basic semantic predicate in their meaning are nastaviti ‘to continue,’ početi ‘to begin,’ prekinuti ‘to quit,’ prestati ‘to stop,’ and produžavati ‘to continue.’ The CVS verbs from the questionnaire that were not classified according to any of the basic semantic predicates discussed above are hteti ‘to want,’ mrzeti ‘to hate,’ voleti/zavoleti ‘to like, to love,’ and želeti/poželeti ‘to desire.’ All four of them belong to the group of verbs named by CJ (2005:464) experiencer verbs. According to CJ, the semantic nature of verbs such as hteti ‘to want,’ mrzeti ‘to hate,’ voleti/zavoleti ‘to like, to love,’ and želeti/poželeti ‘to desire,’ accounts for the fact that they too require unique control. Finally, nouns and adjectives also occur as CVS matrix predicates. They all contain some or all of the basic semantic predicates in their meaning and appear in configurations that are formalized as it is given in (17) below. (17) Xα MATRIXnoun, adjective [α COMPLEMENT] 175 This is true of nouns mogućnost ‘possibility,’ which contains ABLE, običaj ‘custom,’ which contains SHOULD, pravo ‘right,’ which contains SHOULD, prilika ‘opportunity,’ which contains INTEND and ABLE, sramota ‘shame’ contain SHOULD (NOT), and vreme ‘time’ contains ABLE and at the same time express, in a way, some sort of experience similar to experiencer verbs. Adjectives such as lako ‘easy’ and teško ‘difficult’ are also similar to experiencer verbs as well, though they might be thought of as containing ABLE. What is important for them, just as it is important for CVS verbs as matrix predicates, is that their meaning includes the basic semantic predicates discussed above. A generic CVS formula, which unifies findings presented in this sub-section, is proposed in (18). (18) Xα MATRIX(BSP:verb, noun, adjective) Yβ [α/β COMPLEMENT] This formula accounts for both possibilities in CVS: the COMPLEMENT can be uniquely controlled either by the Xα MATRIX argument or by the Yβ MATRIX argument. Just which MATRIX argument actually controls the COMPLEMENT is determined by the basic semantic predicate type. In the formula, the MATRIX is characterized by containing a basic semantic predicate (BSP) in its meaning. In the sections that follow, I use a somewhat simplified formula, as given in (19) below. (19) Xα MATRIX Yβ [α/β COMPLEMENT] 176 6.3.2. CVS Complements According to CJ’s (2005:427) UCAC Hypothesis, the semantics of both the MATRIX and the COMPLEMENT plays the crucial role in unique control. The former needs to be of a certain basic semantic type in order to select the latter; the latter must be a voluntary action. CVS matrices indeed belong to the set of unique control basic semantic predicates. I now examine CVS complements in order to determine whether they necessarily belong to the semantic type of voluntary actions. Based on all the examples of CVS cited thus far in the present study it is clear that voluntary actions are the primary type of complement in CVS. In this particular chapter, in (3)-(5) above, the following three voluntary actions appear: uraditi i više ‘to do even more,’ videti Pariz ‘to see Paris,’ and priznati zločin ‘to admit to a crime.’ That the three are actions in the first place, is confirmed by a standard test for actions, What X did was CJ (2005:427). (20) Ono što je Petar uradio bilo je da je... what Petar did was that he... … uradio i više/video Pariz/priznao zločin. did even more/saw Paris/admitted to a crime ‘What Petar did was do even more/see Paris/admit to a crime.’ The defining characteristic of voluntary actions, according to CJ, is that the actor of an action is animate, in which case the default interpretation is that the action is performed voluntarily. Indeed, the actions of doing more, seeing Paris, and admitting to a crime must have an animate actor; hence, they are performed voluntarily. Two tests that 177 are used to separate voluntary actions from other actions include using the adverbial voluntarily or on purpose in the sentence and putting the utterance in the imperative. They give satisfying results with the three actions. (21) a. Uradi i više! ‘Do even more!’ Petar je dobrovoljno/namerno uradio i više. ‘Petar voluntarily/on purpose did even more.’ b. Vidi Pariz! ‘See Paris!’ Petar je dobrovoljno/namerno video Pariz. ‘Petar voluntarily/on purpose saw Paris.’ c. Priznaj zločin! ‘Admit to a crime!’ Petar je dobrovoljno/namerno priznao zločin. ‘Petar voluntarily/on purpose admitted to a crime.’ According to the standard tests, all three actions analyzed here can appear in the imperative and also with adverbials voluntarily and on purpose, which makes them voluntary actions. The most important problem with respect to voluntary actions, as Culicover and Jackendoff confirm, comes from pragmatics. Namely, there exists a possibility for an utterance, which is normally and logically a non-voluntary action, such as grow taller, to be interpreted as a voluntary action, as illustrated in (22) below. (22) Petar pije vitamine da bi porastao. ‘Petar drinks vitamins in order to grow taller.’ Here, it is the whole pragmatic context that allows for the complement to be a voluntary action of a certain kind, even though grow taller is not otherwise a voluntary action. Still, 178 such examples should only serve as a reminder of what can be a voluntary action at all times – and what only under certain pragmatic circumstances. One particular exception to the UCAC Hypothesis is that not only voluntary actions can be selected as the COMPLEMENT in the configuration of unique control. This is characteristic of the above mentioned experiencer verbs. CJ (2005:464) simply observe that “there are sources of unique control other than being a selected actional complement,” which is the case of experiencer verbs as the MATRIX. They explain this as a factor of the semantic nature of such verbs. An example from English is given in (23) below (CJ 2005:464 example 113a): (23) Judyj thinks that Henryi hopes/wishes to i/j/*gen redeem himself/*herself/*oneself/ *myself. As in unique control, generic, long-distance, and speaker-hearer control are impossible, yet the complement is not a voluntary action, rather it is a situational infinitival complement.52 This, however, does not seem to be a problem for the present analysis. 6.3.3. Summary In this section I examine both the MATRIX and the COMPLEMENT parts of the proposed CVS configuration from the standpoint of the UCAC Hypothesis of CJ’s view of control. According to this view, the basis of unique control (as well as control in 52 For some speakers of English at least Judy thinks that Henry hopes/wishes to redeem herself. Is an acceptable sentence. Here, I closely follow CJ’s judgements as well as CJ’s treatment of (23). CJ do not offer any particularly elaborate discussion of this example. 179 general, for that matter) is found in semantics, more precisely, certain semantic characteristics of both the MATRIX and the COMPLEMENT. Just as the adopted theoretical framework proposes, all of the CVS matrix predicates belong to a small group of basic semantic predicates, notably NAMERAVATI ‘INTEND,’ UMETI ‘ABLE,’ TREBATI ‘SHOULD,’ DOZVOLJAVATI/DOZVOLITI ‘ALLOW,’ and FAZA ‘PHASE.’ Experiencer verbs also require unique control. What is characteristic of all these predicates is that they require unique control in that they select for a semantically very specific type of complement – voluntary actions. It was confirmed that, generally speaking, the complements that occur in CVS are indeed of the semantic type voluntary actions. 6.4. Formalizing CVS The fact that CVS is an instantiation of the phenomenon of control is only one part of the problem. The other part is that it is also a phenomenon of variation of complements, arguably without any variation in meaning. CVS was formalized in (19) above, repeated in (24) for convenience. (24) Xα MATRIX Yβ [α/β COMPLEMENT] In this section I further specify the formula in order to account for exactly how the variation of the two possible complements in CVS operates. For this, the facts about CVS, presented in Chapter 5, are relevant. It is important to indicate at this point that the Greek 180 variables in the formula have a twofold function. First, they indicate the kind of semantic binding that exists in CVS between a complement argument and its controller, a matrix argument. Second, in Serbian the Greek variables necessarily represent a set of grammatical features shared between the controller and the COMPLEMENT, which in turn indicates control. I first explain how I use all of the different factors discussed in Chapter 5. Then, I propose specification of the formula in (19) and (24) based on the CVS statistics for adjectives, nouns and verbs. As conclusive support for what I propose to be the ultimate CVS formula I use findings based on interviews that I conducted during the research process. 6.4.1. Relevant Factors In Chapter 5, I analyzed sociolinguistic, semantic and/or conceptual, and syntactic factors and the impact they have on CVS. I demonstrated that the former two had little or no significant impact on CVS. The participants’ sex, age, and level of education produced no important difference in the CVS statistics. The same was true of the factors such as the matrix verb aspect and the matrix verb tense. The impact of syntactic factors, however, was noticeable. Among them, however, only one syntactic factor caused changes in the choice of the complement that none other factor did. The crucial syntactic factor was the actual presence or absence of the controller in the CVS configuration. In comparison with this factor, all other syntactic factors produced mere trends of changes in the choice of the complement. 181 Whether or not the controller is present in the CVS configuration is a vital factor in whether native speakers choose an infinitival complement or a da+present complement. It is for this reason that this factor is given precedence over all other syntactic factors in the mechanism of CVS and consequently in the CVS formula that accounts for that mechanism. Other syntactic factors must necessarily be indicated, and semantic and/or conceptual, as well as sociolinguistic factors must necessarily be mentioned in order to provide a full understanding of CVS. However, because of its importance, only the first factor will figure in the CVS formula. 6.4.2. Controller with Adjectives Adjectives appear as one type of CVS matrix predicates. A pair of prototypical sentences is given in (5) above and repeated here in (25) for convenience. (25) a. Teško difficult mii I.D je be.present.3RDSG ipriznati b. Teško difficult mii I.D je be.present.3RDSG da that admit.infinitive zločin. crime.A ipriznam admit.present.1STSG zločin. crime.A ‘It is difficult for me to admit to a crime.’ All CVS matrix adjectives are two-place functions. That is to say, they have one complement and one other argument. In a CVS configuration, the complement’s argument is semantically bound by the only other matrix argument. The CVS formula for 182 matrix adjectives must therefore be adjusted in order to represent those descriptive observations, as given in (26). (26) Xα MATRIX Yβ [α/β COMPLEMENT] In the formula, the shaded parts are virtually non-existent in this particular instantiation of CVS and they require no theoretical consideration whatsoever. The other parts, however, do. The factor indicating the actual presence of the matrix argument in the configuration formalized in (26) produces substantially different results when it comes to the choice of the complement in CVS. When the matrix argument Xα, the controller, was present in the syntax of the matrix, the choice of da+present complements, α COMPLEMENT, was the dominant one. When the controller was absent, the choice of infinitival complements, COMPLEMENT, was the dominant one. One way of formally accounting for this fact is given in (27) below. (27) (Xα) MATRIX Yβ [(α)/β COMPLEMENT] The formula in (27) implies exactly that what is found to be true of CVS with adjectival matrix predicates. The use of parentheses accounts for both possible sentential realizations of the CVS configuration: the one with the controller in the syntax of the matrix, without parentheses, and the one without the controller in the syntax of the matrix, with the parentheses. Crucially, the COMPLEMENT without a Greek variable before it 183 represents an uninflected verb, non-finite, infinitival complement, whereas the COMPLEMENT with a Greek variable before it represents an inflected verb, finite, da+present complement. 6.4.3. Controller with Nouns Nominal CVS matrix predicates are represented with the following prototypical pair of sentences from (4) above: (28) a. Prilika mii opportunity I.D je be.present.3RDSG ivideti b. Prilika mii opportunity I.D je be.present.3RDSG da that Pariz. see.infinitive Paris.A ividim see.present.1STSG Pariz. Paris.A ‘It is an opportunity for me to see Paris.’ The case of CVS with nominal matrix predicates mirrors that of adjectival matrix predicates. All CVS nominal matrix predicates are two-place functions; hence, the CVS formula is as the one given in (29). (29) Xα MATRIX Yβ [α/β COMPLEMENT] Furthermore, the syntactic presence of the controller Xα in the matrix causes the preferred choice of a da+present complement, α COMPLEMENT, while its absence significantly 184 raises the probability of the choice of an infinitival complement, COMPLEMENT. All of this is best accounted for using the formula in (30) below. (30) (Xα) MATRIX Yβ [(α)/β COMPLEMENT] The implications of the formula in (30) are exactly the same as the implications of the formula in (27) above. 6.4.4. Controller with Verbs The most frequent kind of CVS matrix predicates are verbs. They also present the most problematic group of CVS matrix predicates for the CVS formula. A sample prototypical pair of sentences is given in (3) above and repeated here in (31) for convenience. (31) a. {Mogao can.pparticiple.M.SG je}i be.present.3RDSG iuraditi i do.infinitive also b. {Mogao can.pparticiple.M.SG je}i be.present.3RDSG da that više. more i also više. more ‘He could have done even more.’ 185 iuradi do.present.3RDSG 6.4.4.1. Depersonalized Verbs In Chapter 5, I discussed one characteristic of Serbian verbs, namely that, more often than not, they are inflected for grammatical person; their morphology indicates which exact entity is associated with the concept denoted by the verb. However, I demonstrated in section 5.4.3. of Chapter 5 that it is possible to depersonalize Serbian verbs by providing their neutral forms – 3rd person singular forms, neuter gender. In this case, the controller is not associated with any specific entity, it is generalized, more precisely, it is generic.53 One verb that appeared in the questionnaire sentences as both personalized, as in (31) above, and depersonalized, is moći ‘can, to be able.’ This is a two-place function, just like adjectival and nominal CVS matrices, as demonstrated in (32) below, (32) a. Može se can.present.3RDSG REFLEXIVE pogrešiti. make a mistake.infinitive b. ?Može se can.present.3RDSG REFLEXIVE da that se REFLEXIVE pogreši. make a mistake.infinitive ‘It is possible to make a mistake.’ and the corresponding CVS formula given in (33) below holds as well. 53 I am aware of the fact that this argument may depend on one’s view of what a controller is and how it relates to a depersonalized structure with the 3rd person singular form, neuter gender. While clearly there is no controller that can, in any way, be present in the configuration such as (32a,b), still the only two possible complements are precisely controlled by the generic nature of the matrix. 186 (33) Xα MATRIX Yβ [α/β COMPLEMENT] The CVS statistics for verbs that were tested in both personalized and depersonalized forms indicate that, in the former case, the finite complement, α COMPLEMENT, is the dominant choice, and in the latter case, the non-finite complement, COMPLEMENT, is the dominant choice. Consequently, the same formula that accounted for adjectival and nominal CVS matrix predicates, also accounts for verbal CVS matrix predicates. The formula in (34) also has the same implication as the formulae in (27) and (30) above. (34) (Xα) MATRIX Yβ [(α)/β COMPLEMENT] 6.4.4.2. Personalized Verbs One potential problem with the formula in (34) is that, although the matrix verb is depersonalized, it is not true that the controller is not present in the syntax of the matrix. If the controller Xα stands for a generic entity, the value of α is the 3rd person singular, neuter gender, as explained above. This generic grammatical information then controls the complement: generic grammatical information requires a non-finite verb form in the complement, the infinitive. Unlike CVS with adjectival and nominal matrix predicates, where the controller may be absolutely non-existent, here the controller exists only in its generic form. Another potential problem is that, while Serbian verbs can be and indeed are easily depersonalized, as, for example, moći ‘can, to be able’ is in (32) above, they more 187 often appear in the language as personalized, with a very specific personal entity associated with the concept denoted by the verb; this is how they appear in CVS as well. Still, an infinitival complement is a possible choice in such CVS configurations also. Exactly how this is possible and what the mechanics are is best understood not from the research questionnaires themselves, but from the interviews that were also a part of the research process. As I have explained in Chapter 5, in the majority of cases the interviewees judged CVS as a variation of two possible complements with no change in meaning (71.54%). In the remainder (28.46%), the interviewees’ judgments never indicated anything but varying semantic, conceptual, or contextual nuances between two sentences in a pair. One such pair is given in (35) below, with an adjective as a CVS matrix predicate. (35) a. Teško difficult mii I.D je be.present.3RDSG ipriznati b. Teško difficult mii I.D je be.present.3RDSG da that 54 admit.infinitive zločin. crime.A ipriznam admit.present.1STSG zločin. crime.A ‘It is difficult for me to admit to a crime.’ One way in which the interviewees explained the difference in meaning between (35a) and (35b) was based on the difference between a generic act of admitting to a crime (35a) as opposed to a very specific, concrete act of admitting to a crime by the person who committed the crime (35b). 54 Note that *Teško mii je da jpriznaš zločin. and *Teško mi je da se prizna zločin. are not acceptable. 188 A similar claim was made for sentences with nominal CVS matrix predicates, given in (36). (36) a. Prilika mii opportunity I.D je be.present.3RDSG b. Prilika mii je opportunity I.D be.present.3RDSG ‘It is an opportunity for me to see Paris.’ ivideti Pariz. see.infinitive Paris.A da that ividim see.present.1STSG Pariz. Paris.A According to the interviewees, (36a) implies a somewhat generic act of seeing Paris, whereas (36b) speaks of the matrix entity’s actual, concrete opportunity to see Paris. Interestingly, even when sentences parallel to (36a,b) in everything but the presence of the controller in the matrix are compared, the interviewees detect semantic nuances that differentiate the two, presented in (37) below. (37) a. Prilika je opportunity be.present.3RDSG videti Pariz. see.infinitive Paris.A b. Prilika je opportunity be.present.3RDSG da that se REFLEXIVE vidi see.present.3RDSG Pariz. Paris.N ‘It is an opportunity to see Paris.’ The sentence in (37a) is said to speak of a more general action of seeing Paris, indicated by English sentence ‘To see Paris/Seeing Paris is (quite) an opportunity,’ whereas (37b) 189 is said to speak of a somewhat more personal action, indicated by English sentence ‘It is (quite) an opportunity for one to see Paris.’55 Finally, the interviewees had similar comments for certain instantiations of CVS with verbal matrix predicates. (38) a. Prestala stop.pparticiple.F.SG je be.present.3RDSG plakati. cry.infinitive b. Prestala stop.pparticiple.F.SG ‘She stopped crying.’ je be.present.3RDSG da that plače. cry.present.3RDSG The sentence in (38a) implies that a female person lost the overall ability to cry, while the one in (38b) implies that the person stopped crying just now, at this very moment. (39) a. Naučila learn.pparticiple.F.SG je be.present.3RDSG b. Naučila je be.present.3RDSG learn.pparticiple.F.SG ‘She learned (how) to read.’ čitati. read.infinitive da that čita. read.present.3RDSG The sentence in (39a) speaks of an action that is quite normal and was probably completed a long time ago. On the other hand, the sentence in (39b) speaks of a very recent, if not present achievement – the person finally learned (how) to read. 55 I thank Brian Joseph for providing the two English sentences. 190 (40) a. Treba be necessary.present.3RDSG raditi. work.infinitive b. Treba be necessary.present.3RDSG ‘One needs to work.’ da that se REFLEXIVE radi. work.present.3RDSG Finally, the interviewees agreed that (40a) presents a generic recommendation, a life truth of some kind, whereas (40b) was perceived as a recommendation directed to a specific person or else uttered in a very specific, concrete situation. All of the semantic nuances discussed above indicate that there are fine differences between the sentences in each pair. Furthermore, the fine semantic nuances are easily grouped as indicated in Table 6.1. below. infinitival complement generic act of admitting to a crime da+present complement very specific, concrete act of admitting to a crime generic act of seeing Paris; more general actual, concrete opportunity to see Paris; action of seeing Paris somewhat more personal action lost the overall ability to cry stopped crying just now, at this very moment action that is quite normal and probably very recent, if not present achievement – completed a long time ago the person finally learned (how) to read generic recommendation, a life truth of recommendation directed to a specific some kind person or else uttered in a very specific, concrete situation Table 6.1. Semantic Nuances Clearly, when the semantic differences are detected between the two possible choices for CVS complements, infinitival complements are chosen when a more generic action is indicated, while da+present complements are chosen to indicate a more concrete action. 191 This has also been noticed by Moskovljević (1936) and then later supported by Brozović (1953a,b). Moskovljević (1936:111-112) provides a set of examples, one of which is given in (41) below. (41) u in a. Danas (sutra) ne today (tomorrow) NEG mogu can.present.1STSG ići u go.infinitive in b. Danas (sutra) ne today (tomorrow) NEG mogu can.present.1STSG da that školu. school idem go.present.1STSG školu. school ‘Today (tomorrow) I cannot go to school.’ According to Moskovljević, (41a) implies that the person “will not go to school due to a general setback of sorts, a setback coming from the outside with respect to that person.” In this case, and with this choice of the complement, the sentence speaks of a situation in which “the true stimulus for the fulfillment of the action in the complement is an outer stimulus with respect to the subject.” On the other hand, (41b) indicates that the person “has no energy to go … and speaks of going at one specific point in time,” which is because “the completion of the action in the complement depends on the subject of the main verb, i.e., the grammatical and the logical subject are the same, or the action in the complement is performed in the present or in an otherwise specified moment.” Brozović (1953a:15) essentially agrees with Moskovljević’s examples. He explains them using two different scenarios for one and the same context, as presented in (42) below. 192 (42) Guest at the door… a. one is paralyzed and in the bed, thus cannot get out of bed; one says: Ne NEG mogu can.present.1STSG nikako ustati in no way get up.infinitive.1STSG s kreveta kao da sam prikovan. from bed as if I am nailed-up ‘There is no way I can get out of bed as if I am nailed-up.’ b. one is naked in the bed, thus cannot get out of bed; one says: Ne NEG mogu can.present.1STSG da that ustanem, izađi get up.present.1STSG leave časak, da se obučem. for a moment in order for me to get dressed ‘I cannot get up; leave the room for a moment so I can get dressed.’ Brozović consequently concludes that “if the complementary verb indicates something that lasts or something general, it is better to use the infinitive, and if it indicates something momentary or concrete, it is better to use the present with the conjunction da ‘that.’” Moskovljević’s and Brozović’s findings are essentially the same as the claims made in the present study and summarized in Table 6.1. above. More importantly, the claims regarding the fine semantic differentiation between the choice of an infinitival complement, on the one hand, and a da+present complement, on the other, must find its place in the CVS formula, the latest version of which is given in (34) above and is repeated here in (43) for convenience. (43) (Xα) MATRIX Yβ [(α)/β COMPLEMENT] 193 It is clear now that, although the controller Xα is present in the syntax of the MATRIX, it does not mean that the choice of the infinitive in the COMPLEMENT is absolutely impossible, though it is largely minimized. Still, the infinitive may appear in the COMPLEMENT where the implication is generic or simply less concrete. The controller is syntactically present, therefore the formalization (Xα) does not hold as firmly anymore, for it implies that the controller either exists, Xα, or it does not, (Xα). Also, even though the controller is syntactically present, the control relationship that it now imposes on the COMPLEMENT does not require that the complement be inflected for the grammatical information of the controller. The Greek variable now indicates a sort of bare semantic binding between the controller and the COMPLEMENT, without any implications regarding the inflection of the latter. In other words, the controller is syntactically present in the matrix, but it is semantically transparent for CVS, thus allowing a possibility of an infinitival complement under the semantic circumstances discussed above. I, therefore, propose the following improvement to the formula in (44): (44) X(α) MATRIX Yβ [(α)/β COMPLEMENT] The use of parentheses in the new formula has the same implications as it did in the previous version of the formula. Now, however, the formula allows for the actual controller to be obvious in the syntax, but somehow transparent in the semantics of the matrix. 194 6.4.4.3. Additional Controllers with Verbs The formula in (44) does not account for all the basic semantic predicates in CVS, notably DOZVOLJAVATI/DOZVOLITI ‘ALLOW’ and one type of UMETI ‘ABLE’ represented by UČITI/NAUČITI ‘TEACH.’ The generic CVS formula for these basic semantic predicates is as given in (45) below. (45) X(α) MATRIX Yβ [(α)/β COMPLEMENT] In the formula, the shaded parts are not pertinent for these particular basic semantic predicates. The controller is Yβ and, consequently, Greek variable β is placed with the COMPLEMENT. What needs to be analyzed is whether control by Yβ operates in the same way as control by Xα, which will, in turn, lead to a potential improvement of the formula in (45). The CVS statistics for the basic semantic predicates of the type DOZVOLJAVATI/DOZVOLITI ‘ALLOW’ and UČITI/NAUČITI ‘TEACH’ are given in Table 6.2. below. TOKEN # % I 21 2.06% I/P 7 0.69% P 990 97.25% Table 6.2. CVS for the Yβ controller 195 TOTAL 1018 100% The percentage for the choice of da+present complements is decisively the highest here: it seems that, in this particular configuration, the choice of the complement is virtually absolutely clear. Moreover, the interviews provide a very significant piece of information about the possibility of an infinitival complement in CVS configuration such as the one in (44) above, whose actual realization is given in (46) below. (46) a. #*Naučiće teach.3RDSG/PL ih deliti they.A parse.infinitive složenu rečenicu. complex sentence.A b. Naučiće ih da dele RD teach.3 SG/PL they.A that parse.present.3RDPL ‘He/She/They will teach them how to parse a sentence.’ složenu rečenicu. complex sentence.A Sentential realizations of this particular CVS configuration, with the infinitive as the COMPLEMENT, as in (46a), were described by the interviewees as “hard to judge, impossible to understand, requiring slower processing, never used, and confusing.” Such descriptive generalizations imply that, when the Yβ MATRIX argument is the controller in CVS, there is almost no room whatsoever for variation of the complements. Thus, the two possible controllers, Xα and Yβ, do not operate in the same way with respect to CVS and the formula in (45) is improved as given in (47) below. (47) X(α) MATRIX (Yβ) [(α)/(β) COMPLEMENT] 196 For the basic semantic predicates such as DOZVOLJAVATI/DOZVOLITI ‘ALLOW’ and UČITI/NAUČITI ‘TEACH’ the actual presence of the controller in the syntax of the MATRIX is the only factor that dictates the choice of the complement in CVS. 6.4.4.4. Concluding Thought In this subsection I formally account for all possible kinds of CVS, the one characteristic of adjectival and nominal CVS matrix predicates, the one with depersonalized verbs, and then the two kinds of CVS with personalized verbs (depending on what the controller is). Ultimately, two formulae have emerged, (42) and (44). They account in a formal way for the established descriptive generalizations about CVS. 6.4.5. Summary In concluding this section on a formal account of CVS, I attempt to provide a unifying formula that accounts for CVS in general. The two formulae proposed for CVS given in (44) and (47) above are repeated in (48a,b) below, respectively. (48) a. X(α) MATRIX Yβ [(α)/β COMPLEMENT] b. X(α) MATRIX (Yβ) [(α)/(β) COMPLEMENT] 197 In the formulae, parentheses imply the syntactic presence or absence of the controllers, while the shaded parts of the formulae represent elements of CVS configurations not pertinent for certain basic semantic predicates. Instead of the shaded parts, I propose the use of another set of parentheses, which then allows that the two formulae be united into one, ultimate CVS formula, given in (49) below. (49) X((α)) MATRIX (Yβ) [((α))/(β) COMPLEMENT] The use of the slash in the formula implies the option of one or the other controller controlling the COMPLEMENT. The single and double parentheses in the formula are ordered in that the single parentheses have precedence over the double parentheses. If controller Yβ is present in the syntax of the MATRIX, it controls the COMPLEMENT, precisely in the way discussed above. If, however, controller Yβ is not present in the syntax of the MATRIX, then controller Xα controls the COMPLEMENT. It may seem that the CVS formula, as given in (49) above, does not present a desired generalization, for the use of parenthesis in the formula, as well as the line of reasoning that lead towards it, clearly suggest that the formula collapses at least two (as in 48a,b) or possibly more than two formulae. However, the goal of the present study has always been to provide a better understanding of CVS. The fact that the study produced a formula to account for CVS is a consequence of the results presented in the study as well as the adopted theoretical framework. In addition to this, during the course of this study, CVS has never been treated as a simple phenomenon that can easily be accounted for by 198 an easy-to-use formula. Rather, it has always been said that CVS is deceptively simple, but essentially an extremely complex phenomenon. The variation of complements is controlled by numerous factors at the same time, as I explain in Chapter 4. Thus, it should come as no surprise that the CVS formula in (49) allows for several different interpretations. It is still both the first and an important step in a better understanding of CVS. 6.5. Conclusion In this chapter I ultimately provide a formal theoretical account of all the crucial descriptive generalizations of CVS. I demonstrate that CVS is an instantiation of control, more precisely unique control, in terms of the adopted theoretical framework proposed by Culicover and Jackendoff (2005). Following their view of control as a linguistic phenomenon based in semantics, I explain how CVS is indeed semantically based in that it exemplifies control between a set of basic semantic predicates, which invariably select for voluntary actions as complements, and those voluntary actions. In addition to this, experiencer verbs, because of their own semantic properties, also require uniquely controlled complements. I then proceed with accounting for the actual possible variation of the complements, which is the essence of CVS. I propose that CVS be explained with one formula only, the one that takes into consideration two different mechanisms of control by two different possible controllers. 199 Finally, observations made regarding CVS in particular have certain implications for the theory of control in general. While it is true that control, as in CVS, is semantically based, the role of syntax is not, nor can it be neglected. The actual choice of complements in CVS, once the essentially semantic basis of control has been detected, is largely determined by the presence or absence of the controller in the syntax of the MATRIX. In the particular case of the Xα controller, syntactic information denoted by the Greek variable α is of crucial importance. I, therefore, maintain that, although control appears to be semantic in nature, it is, probably syntactic in nature as well. Only a combination of the two levels of linguistic analysis, as presented in this chapter, provides both a thorough account of CVS in particular and of the phenomenon of control in it in general. 200 CHAPTER 7 CONCLUSION The major goal of the present study was to provide new insights into CVS and, eventually, to account for them in a formal way. I consider the goal – reached. Just as I emphasized in the beginning of the study, CVS, as it is conceived of in the study, appears to be a deceptively simple linguistic phenomenon – two possible complements appear in one and the same syntactic configuration, though the meaning remains the same. The study, however, demonstrates how complex CVS actually is – in reality, factors of different kinds control the variation of the possible complements. Ultimately, when syntactic factors have been taken into consideration, I propose a formula that accounts for the syntax of CVS, as it is found in Serbian. In this chapter I, first, summarize major claims regarding CVS presented in various places in the study. Then, I explain the position of the present study with respect to the previous accounts of CVS. Finally, I discuss consequences of the account of CVS given here. 201 7.1. Syntax of CVS Since I was concerned with the syntax of CVS, the most important task to accomplish was to define CVS. Defining CVS involved providing answers to two major questions: first, what the configuration of CVS is and, second, what complements vary in CVS. The answer to the first question was given in a proposal that CVS invariably takes place in a very specific configuration whose ultimate formal shape is given in (1) below. (1) X((α)) MATRIX (Yβ) [((α))/(β) COMPLEMENT] The mechanism according to which the formula operates has been discussed in the previous chapter. Crucially for CVS, the variation of complements occurs in this particular configuration and in it alone. This proposal immediately excludes any other configuration in which either of the two complements that figure in CVS may appear as well, such as in the subject position, for example. The answer to the second question indicated that infinitival and da+present complements are the only two possible complements in CVS. This excludes the possibility that other structures can appear in place of the infinitive in Serbian, as illustrated in (2) below. 202 (2) a. Gledati televiziju je watch.infinitive television.A be.present.3RDSG ‘To watch TV is unavoidable/It is unavoidable to watch TV.’ b. Gledanje televizije watching television.G ‘Watching TV is unavoidable.’ je be.present.3RDSG neizbežno. unavoidable neizbežno. unavoidable Examples of this kind, or any variation between an infinitival structure and a structure other than da+present, are not considered to belong to CVS. Once CVS has been defined syntactically in those terms, I address the issue of the nature of the two possible complements, more precisely the da+present complement. It is clear that, on the one hand, this complement exhibits characteristics of regular Serbian subordinate clauses, and, on the other hand, properties more characteristic of infinitival complements and – in languages other than Serbian – subjunctive complements. However, no conclusive evidence exists at the moment to demonstrate that the da+present complement is clearly a subordinate clause or a subjunctive. Hence, I continue to refer to it as the da+present complement. This term both captures the uncontested descriptive generalization about the complement and emphasizes its unique character in the syntax of CVS. After the basics have been defined, I begin a detailed analysis of the syntax of CVS. It turned out, first, that the long linguistic interest in CVS produced little or no true explanation of the actual variation of the complements, with the exclusion of dialectal and regional as well as some sociolectal factors responsible for the variation. Second, the question of syntactic factors in CVS was almost non-existent. Those two facts account for an insufficient understanding of CVS, the long tradition of study notwithstanding. 203 I propose that the syntax of CVS is best understood if the role of all other factors responsible for CVS is minimized. I do exactly that in research that I conducted on a sample of native speakers from the territory of the city of Belgrade, the capital of Serbia. The research decisively proved that there is indeed one syntactic factor that crucially determines which complement, infinitival or da+present, is chosen in CVS. The syntactic factor was the presence or absence in the syntax of the matrix of the controller of the complement. This was the basis for a formal theoretical account of CVS. I adopt Culicover and Jackendoff’s (2005) theoretical account of control and demonstrate that CVS, as an instantiation of unique control, operates according to the formula given in (1) above. 7.2. Account of CVS My account of CVS relies on both a theory of control and actual language data. This fact alone makes the particular account of CVS presented in this study different than any other previous account of CVS. To the best of my knowledge, there is no indication in previous accounts of CVS that the phenomenon of complement variation is an instantiation of control. Recognizing this allowed me to employ theoretical views said to hold for control as a linguistic phenomenon. Relying on the actual language data, however, as well as on the existing tradition of accounting for CVS, enabled me to adopt just the right theoretical framework, one that recognizes the importance of semantics. The role of semantics has usually been emphasized in CVS in the past and the present research results demonstrated the same. 204 Consequently, I proposed an account that groups all CVS matrix predicates into a small set of basic semantic predicates that select voluntary actions as their complements. The result of this semantic relationship between the matrix and the complement is the phenomenon of unique control. The role of syntax, however, proved to be just as important in CVS. Prototypical sentences are given in (3) below. (3) a. Teško difficult I.D mii je be.present.3RDSG ipriznati admit.infinitive b. Teško mii je difficult I.D be.present.3RDSG that ‘It is difficult for me to admit to a crime.’ da zločin. ipriznam ST admit.present.1 SG crime.A c. Teško je difficult be.present.3RDSG priznati admit.infiitive d. Teško je difficult be.present.3RDSG that da zločin. crime.A se REFLEXIVE zločin. crime.A prizna admit.present.3RDSG zločin. crime.N ‘It is difficult to admit to a crime.’ The presence of the controller is what differentiates (3a,b) from (3c,d), where the controller is absent. Furthermore, the choice of the complement in (3a,b) is overwhelmingly in favor of the da+present complement, and in (3c,d) the infinitival complement is favored. Thus, I find the role of syntax in CVS to be equally important as that of semantics. The explanation of CVS presented in this study, although it relies on past achievements, as well as the latest theoretical advances in linguistic theory, provides a 205 novel treatment with respect to both. On the one hand, it dispenses with heavy descriptive machinery that was employed in accounting for CVS in the past simply because the importance of the syntax of CVS and proper involvement of the semantics in CVS were never fully understood. On the other hand, the present account utilizes crucial theoretical achievements of Culicover and Jackendoff’s semantically-based theory of control only to expand on it in terms necessary to capture the essence of CVS, which is the actual variation of the two complements. 7.3. Implications of CVS I have already emphasized earlier in this chapter, as well as in the previous chapter, what the importance of CVS is for the overall phenomenon of control. My findings contribute to an ongoing debate concerned with whether control can be explained in purely syntactic terms, or if it must be analyzed in terms of semantics only. It turns out that CVS, and consequently control in it, cannot be fully explained if only one or the other view is taken. Rather, CVS is based in semantics, but fully dependent on syntax, which is crucial for determining exactly what complement is chosen, and, when syntax does its part, semantic or conceptual nuances come into play, allowing complements, otherwise not predicted by syntax, to appear. It is impossible to pinpoint the line that divides semantics from syntax in CVS and CVS itself does not allow for such a defined separation. The theoretical account proposed in the present study may have implications for at least two other linguistic phenomena. 206 One phenomenon is known as the loss of the infinitive in the Balkan languages. Joseph (1983:145-148) insists on the value of Serbo-Croatian in particular with regard to the infinitive-loss. While his concern is mainly the dialectal distribution of the infinitive in Serbo-Croatian, he still recognizes the importance of a variety of factors in the infinitive-loss: “An additional way in which Serbo-Croatian contributes to the study and understanding of the Balkan infinitive-loss is through the fact that an infinitivereplacement process is still in progress… This situation provides an opportunity to see first hand the variety of factors, social as well as purely linguistic, that can interact in the manifestation of this process. (Joseph 1983:147)” The present study contributes to the question of how the infinitive is retracting in Serbian, or else how it might have retracted in the Balkan languages that lost the category of the infinitive, by emphasizing one syntactic factor over all others. That is to say, the study indicates what takes place both in configurations with one potential controller and in configurations in which there are two possible controllers in the matrix, as is formalized in (1) above and explained in the previous chapter. Another phenomenon is complementation, as CVS is an instantiation of that as well. As such, CVS confirms how complementation that involves two (or more, for that matter) complements appearing in one and the same syntactic configuration seems simple, but in fact is a complex problem, for the variation of the complements is often, as in the case of CVS, determined by a combination of different factors (listed in random order): dialectal, regional, socioloectal, idiolectal, syntactic, semantic, and pragmatic. All of those factors operate together often, if not at all times, as I indicate in Chapter 5, for CVS. 207 The analysis presented here is only one way in which complementation, which at the same time involves variation of the complements, may be treated. 7.4. Summary The goal of the present study has been reached and a better understanding of CVS provided, not, however, as an ultimate answer to the general question of CVS, but as a set of particular descriptive observations eventually formalized in what has been named the CVS configuration. The formula must be tested, understandably, on even more data from the language, as well as, preferably, on cross-linguistic data in order to sustain possible criticism. 208 APPENDIX A INFORMATION FOR PARTICIPANTS 209 INFORMATION FOR THE PARTICIPANTS This research is an important part of an ongoing work on the phenomenon of certain verbs’ and structures’ complement variation in Serbian. I analyze the phenomenon in my dissertation on which I am working at the Ohio State University in Columbus, Ohio, USA, under the mentorship of Professor Brian D. Joseph. The research itself is relevant for linguistics in general as well. A similar phenomenon exists, or used to exist in other South Slavic languages (Bulgarian and Macedonian) as well as in non-Slavic languages neighboring Serbian (Romanian, Albanian, and Greek). The research will, therefore, demonstrate what the similar process might have looked like in those languages – through history (of which we know very little, or nothing, due to the lack of relevant historical records). It will be demonstrated what the situation with the phenomenon in the contemporary Serbian language is, and it will also help in understanding the phenomenon in languages other than Serbian, thus helping general and theoretical linguistics understand how language in general operates and changes. You are expected to contribute to the research by completing all the tasks that the questionnaire poses before you. You should complete based on your understanding of the given instructions as well as your own native-speaker intuition. I hope I will thus receive very precise judgments of the phenomenon from native speakers of the contemporary Serbian language. This will serve as an initial point in the writing of my dissertation. Your participation in the research is voluntary. You may withdraw at any moment without penalty. Thank you for participating in this research! 210 INFORMACIJA ZA UČESNIKE U ISTRAŽIVANJU Ovo istraživanje je važan deo mog rada na razumevanju fenomena varijacije dopuna određenih glagola i konstrukcija u srpskom jeziku. Fenomen varijacije dopuna je predmet moje doktorske disertacije na Univerzitetu države Ohajo, u Kolumbusu u Ohaju (Sjedinjene Američke Države), gde je moj mentor profesor Brian D. Joseph. Ovo istraživanje je značajno i za druge lingvističke discipline takođe. Naime, sličan fenomen postoji, ili je postojao, u drugim južnoslovenskim jezicima (bugarski i makedonski), ali i neslovenskim jezicima koji se nalaze u susedstvu srpskog jezika (rumunski, albanski, grčki). Tako će ovo istraživanje pokazati kako je izgledao sličan proces u navedenim jezicima – kroz istoriju (o kojoj se zna vrlo malo ili čak ništa zbog toga što ima vrlo malo informacija o istorijskom razvoju tih jezika). Istraživanje će zato pre svega pokazati kakva je situacija sa ovim fenomenom u savremenom srpskom jeziku, ali će pomoći u razumevanju fenomena i u drugim jezicima i omogućiti opštoj i teorijskoj lingvistici da razumeju na koji način jezik uopšte funkcioniše i kako se menja. Od Vas se očekuje da doprinesete istraživanju na taj način što ćete obaviti zadatke koje pred Vas postavlja upitnik, na način na koji razumete instrukcije koje su tamo date i na osnovu svog jezičkog osećanja. Na osnovu toga, ja se nadam da ću moći da dobijem vrlo precizne procene fenomena od strane govornika savremenog srpskog jezika, što će poslužiti kao polazište za pisanje disertacije. Vaše učestvovanje u ovom istraživanje je potpuno slobodno, pa se iz istraživanja možete i povući u bilo kom trenutku, bez ikakvih negativnih posledica po sebe. Hvala Vam na pristanku da učestvujete u istraživanju! 211 APPENDIX B CONSENT FORMS 212 CONSENT FOR PARTICIPATION IN SOCIAL AND BEHAVIORAL RESEARCH Protocol title: Complement Variation in Serbian Protocol number: 2004B0159 Principal Investigator: Professor Brian D. Joseph I consent to my participation in research being conducted by Professor Brian D. Joseph of The Ohio State University and his assistants and associates. The investigator(s) has explained the purpose of the study, the procedures that will be followed, and the amount of time it will take. I understand the possible benefits, if any, of my participation. I know that I can choose not to participate without penalty to me. If I agree to participate, I can withdraw from the study at any time, and there will be no penalty. I have had a chance to ask questions and to obtain answers to my questions. I can contact the investigators at the following address: 206 Oxley Hall, 1712 Neil Avenue, Columbus, OH 43210, or at (614) 292-4981, or at: [email protected]. If I have questions about my rights as a research participant, I can call the Office of Research Risks Protection at (614) 688-4792. I have read this form or I have had it read to me. I sign it freely and voluntarily. A copy has been given to me. Print the name of the participant: ______________________________________________________ Date: Signed: _________________________________ ___________________________________ (Participant) Signed: ________________________________ Signed: ___________________________________ (Principal Investigator or his/her authorized representative) (Person authorized to consent for participant, if required) Witness: _______________________________ (When required) HS-027 (Rev. 05/01) (To be used only in connection with social and behavioral research.) 213 PRISTANAK ZA UČESTVOVANJE U ISTRAŽIVANJU Naziv protokola: Complement Variation in Serbian Broj protokola: 2004B0159 Rukovodilac istraživanja: Profesor Brian D. Joseph Izjavljujem da pristajem da učestvujem u istraživanju koje sprovode profesor Brian D. Joseph i njegov saradnik sa Univerziteta države Ohajo iz Kolumbusa u Ohaju (Sjedinjene Američke Države). Istraživač mi je objasnio svrhu ovog istraživanja, procedure prema kojima će se istraživanje odvijati, kao i vremenski period neophodan za samo istraživanje. Takođe razumem i kakva je moja moguća korist, ukoliko korist uopšte postoji, od učestvovanja u ovom istraživanju. Znam da mogu odlučiti da ne učestvujem bez ikakvih negativnih posledica po sebe. Ukoliko pristanem da učestvujem, mogu se povući iz istraživanja u svakom trenutku bez ikakvih negativnih posledica. Imao/Imala sam priliku da pitam i dobijem odgovore na svoja pitanja. Istraživaču se mogu obratiti na adresu: 206 Oxley Hall, 1712 Neil Avenue, Columbus, OH 43210, ili na telefon: (614) 292-4981, ili na adresu: [email protected]. Ukoliko imam pitanja u vezi sa svojim pravima, kao osoba koja je učestvovala u istraživanju, mogu se obratiti Kancelariji za zaštitu od rizika istraživanja na telefon: (614) 688-4792. Pročitao/Pročitala sam ovaj formular ili mi je formular pročitan. Potpisujem ovaj formular uz potpunu slobodu. Kopija ovog formulara mi je data. Ime učesnika (štampanim slovima): ____________________________________________________ Datum: ________________________________ Svojeručni potpis:________________________ (učesnik) Svojeručni potpis:_________________________ Svojeručni potpis:________________________ (rukovodilac ili predstavnik rukovodioca istraživanja) Svedok:________________________________ (kada je neophodno) HS-027 (Rev. 05/01) 214 (osoba koja daje pristanak u ime učesnika istraživanja ako je neophodno) APPENDIX C QUESTIONNAIRE 215 QUESTIONNAIRE IDENTIFICATION CODE:__________ Name:__________________________________________________________________ Contact Information (e-mail or phone):________________________________________ Date of birth:____________________ Gender: ■ male ■ female Place of birth:__________________ Place where growing up:__________________ Place of residence: _____________________________ Education: ■ elementary school ■ high school ■ college ■ university Profession: _____________________ Current Employment:_______________________ INSTRUCTIONS INSTRUCTIONS INSTRUCTIONS There are 80 pairs of sentences in this questionnaire. It is your task to judge each pair of sentences, based on your native speaker intuition, in the following way: 1. Read both sentences in the given pair of sentences; 2. Based on your native speaker intuition, circle the number in front of the sentence for which you believe (or think or are completely sure) that you would use it more often than the other sentence (if you believe, or think, or are completely sure that you would use both sentences in the given pair with the same frequency, circle the numbers in front both of them); 3. Any additional comment that you might have for the given pair of sentences you may write in the provided space; 4. Proceed with the next pair in the same fashion. INSTRUCTIONS INSTRUCTIONS INSTRUCTIONS SENTENCES 011. To take three times a day after meals. 021. Will Jovan go to the concert? 031. They are continuing to talk. 041. You have to study. COMENTARY 012. To take three times a day after meals. 022. Will Jovan go to the concert? 032. They are continuing to talk. 042. You have to study. 216 051. (S)he is afraid to approach. 061. At that moment, Serbian people began to enter the modern history. 071. This is an opportunity to see Paris. 081. It is easy for you to command. 091. (S)he wants to work. 101. (S)he is ashamed to look him in the eyes. 111. She stopped crying. 121. He was refusing to take the book. 131. She intended to visit her girlfriend. 141. Here you go, eat dinner. 151. He has gotten used to getting up early. 161. To disclose the real causes of the boys’ death. 171. (S)he is studying how to read and write. 181. She is afraid to travel alone. 191. She was not supposed to run a printing job without permission. 201. Take this to read. 211. I am ashamed to sing. 221. It is difficult to admit a crime. 231. But the Turks in Serbia did not allow themselves to be confused. 241. I have the right to say everything. 251. He did not dare to call. 261. He decided to travel away. 271. (S)he is afraid to remember everything. 281. He refused to accept them. 291. She learned how to read. 052. (S)he is afraid to approach. 062. At that moment, Serbian people began to enter the modern history. 072. This is an opportunity to see Paris. 082. It is easy for you to command. 092. (S)he wants to work. 102. (S)he is ashamed to look him in the eyes. 112. She stopped crying. 122. He was refusing to take the book. 132. She intended to visit her girlfriend. 142. Here you go, eat dinner. 152. He has gotten used to getting up early. 162. To disclose the real causes of the boys’ death. 172. (S)he is studying how to read and write. 182. She is afraid to travel alone. 192. She was not supposed to run a printing job without permission. 202. Take this to read. 212. I am ashamed to sing. 222. It is difficult to admit a crime. 232. But the Turks in Serbia did not allow themselves to be confused 242. I have the right to say everything. 252. He did not dare to call. 262. He decided to travel away. 272. (S)he is afraid to remember everything. 282. He refused to accept them. 292. She learned how to read. 217 301. I do not feel like going shopping. 311. I had to admit everything to him. 321. They want to buy us a computer. 331. Let him/her come since he likes to swim. 341. (S)he is afraid to call. 351. Jovan will not go to the concert. 361. Slowly, she was forgetting how to rejoice. 371. One can make a mistake. 381. (S)he/They will teach them how to parse a complex sentence. 391. The school board decided to award all of the nominated students. 401. He quit studying. 411. One needs to work. 421. He succeeded in overcoming all the difficulties. 431. This is an opportunity for me to see Paris. 441. It is easy to command. 451. I do not have time to read. 461. To mix until it blends. 471. She wanted to call him. 481. (S)he knows how to talk around things. 491. (S)he tries to make everybody happy. 501. She wished to get a bike. 511. He did not let him buy us bread. 521. She liked to draw. 531. One must work. 541. He refused to help him. 551. It does not suit her to yell. 561. It is a shame for him to talk 302. I do not feel like going shopping. 312. I had to admit everything to him. 322. They want to buy us a computer. 332. Let him/her come since he likes to swim. 342. (S)he is afraid to call. 352. Jovan will not go to the concert. 362. Slowly, she was forgetting how to rejoice. 372. One can make a mistake. 382. (S)he/They will teach them how to parse a complex sentence. 392. The school board decided to award all of the nominated students. 402. He quit studying. 412. One needs to work. 422. He succeeded in overcoming all the difficulties. 432. This is an opportunity for me to see Paris. 442. It is easy to command. 452. I do not have time to read. 462. To mix until it blends. 472. She wanted to call him. 482. (S)he knows how to talk around things. 492. (S)he tries to make everybody happy. 502. She wished to get a bike. 512. He did not let him buy us bread. 522. She liked to draw. 532. One must work. 542. He refused to help him. 552. It does not suit her to yell. 562. It is a shame for him to talk 218 like that. 571. It is hard for me to admit a crime. 581. There is no time for one to read. 591. (S)he wants to take a stenography class. 601. So back then they did not let the youth to watch this film. 611. Let him/her come when (s)he falls in love with swimming. 621. We have no chance for studying in America. 631. It is a custom to drink coffee. 641. He did not dare to object. 651. He was successful in overcoming all of the difficulties. 661. He tried to solve it. 671. He refused to accept the gift. 681. Jovan will go to the concert. 691. (S)he is teaching him how to swim. 701. (S)he/They will continue to work as always. 711. (S)he intends to tell him everything. 721. He has gotten used to greet. 731. (S)he hates to sing. 741. He could have done more. 751. (S)he knows how to listen. 761. She forgot how to rejoice. 771. I wish you to get well fast. 781. Earlier they did not allow the youth to watch this film. 791. She did not let him talk. 801. He fell in love with drawing. like that. 572. It is hard for me to admit a crime. 582. There is no time for one to read. 592. (S)he wants to take a stenography class. 602. So back then they did not let the youth to watch this film. 612. Let him/her come when (s)he falls in love with swimming. 622. We have no chance for studying in America. 632. It is a custom to drink coffee. 642. He did not dare to object. 652. He was successful in overcoming all of the difficulties. 662. He tried to solve it. 672. He refused to accept the gift. 682. Jovan will go to the concert. 692. (S)he is teaching him how to swim. 702. (S)he/They will continue to work as always. 712. (S)he intends to tell him everything. 722. He has gotten used to greet. 732. (S)he hates to sing. 742. He could have done more. 752. (S)he knows how to listen. 762. She forgot how to rejoice. 772. I wish you to get well fast. 782. Earlier they did not allow the youth to watch this film. 792. She did not let him talk. 802. He fell in love with drawing. 219 Finally, please answer the following questions: 1. Do you know (or you think you know) what the essence of the questionnaire is? ________________________________________________________________________ 2. How did you determine what sentence you will circle? ________________________________________________________________________ 3. What is your overall attitude toward this questionnaire? ________________________________________________________________________ THANK YOU!!! THANK YOU!!! 220 THANK YOU!!! UPITNIK IDENTIFIKACIONI BROJ:__________ Ime i prezime:___________________________________________________________ Kontakt informacija (e-mail adresa ili telefon):________________________________ Datum rođenja:____________________ Pol: ■ muški ■ ženski Mesto rođenja:_____________________ Mesto odrastanja:______________________ Mesto stanovanja:______________________________ Školska sprema: ■ osnovna ■ srednja ■ viša ■ visoka Zanimanje:_______________________ Trenutno zaposlenje:____________________ INSTRUKCIJE INSTRUKCIJE INSTRUKCIJE INSTRUKCIJE U ovom upitniku postoji 80 parova rečenica. Vaš zadatak je da, koristeći svoje jezičko osećanje, procenite svaki par rečenica na sledeći način: 1. Pročitajte obe rečenice u datom paru rečenica; 2. Na osnovu svog jezičkog osećanja, zaokružite redni broj ispred one rečenice za koju verujete (ili mislite ili ste sasvim sigurni) da je upotrebljavate češće od one druge (ukoliko verujete, ili mislite, ili ste sasvim sigurni, da biste jednako često koristili obe rečenice u datom paru, zaokružite redne brojeve ispred obe); 3. Svaki svoj komentar na dati par rečenica možete upisati u predviđeni prostor. 4. Nastavite sa sledećim parom rečenica na isti način. INSTRUKCIJE INSTRUKCIJE INSTRUKCIJE 221 INSTRUKCIJE REČENICE KOMENTAR 011. Da se uzima tri puta na dan 012. Uzimati tri puta na dan posle posle jela. jela. 021. Da li će Jovan da ide na 022. Da li će Jovan ići na koncert? koncert? 031. Produžavaju pričati. 032. Produžavaju da pričaju. 041. Imaš da učiš. 042. Imaš učiti. 051. Boji se prići. 052. Boji se da priđe. 061. Srpski narod je tada počeo 062. Srpski narod je tada počeo da ulaziti u modernu istoriju. ulazi u modernu istoriju. 071. Prilika je da se vidi Pariz. 072. Prilika je videti Pariz. 081. Lako vam je zapovedati. 082. Lako vam je da zapovedate. 091. Hoće raditi. 092. Hoće da radi. 101. Stidi se da ga pogleda u oči. 102. Stidi se pogledati ga u oči. 111. Prestala je da plače. 112. Prestala je plakati. 121. Odbijao je uzeti knjigu. 122. Odbijao je da uzme knjigu. 131. Nameravala je da poseti 132. Nameravala je posetiti prijateljicu. prijateljicu. 141. Izvolite večerati. 142. Izvolite da večerate. 151. Navikao se da ustaje rano. 152. Navikao se ustajati rano. 161. Objaviti prave uzročnike 162. Da objave prave uzročnike smrti dečaka. smrti dečaka. 171. Uči čitati i pisati. 172. Uči da čita i piše. 181. Plaši se da putuje sama. 182. Plaši se putovati sama. 191. Nije smela da drži 192. Nije smela držati štampariju štampariju bez dozvole. bez dozvole. 201. Uzmi čitati. 202. Uzmi da čitaš. 211. Sramota me je pevati. 212. Sramota me je da pevam. 222 221. Teško je priznati zločin. 222. Teško je da se prizna zločin. 231. Ali se Turci u Srbiji nisu 232. Ali se Turci u Srbiji nisu dali dali da se zbune. zbuniti. 241. Imam pravo sve kazati. 242. Imam pravo da sve kažem. 251. Nije se usudio da se javi. 252. Nije se usudio javiti se. 261. Rešio je otputovati. 262. Rešio je da otputuje. 271. Plaši se da se seti svega. 272. Plaši se setiti se svega. 281. Odbio je primiti ih. 282. Odbio je da ih primi. 291. Naučila je da čita. 292. Naučila je čitati. 301. Mrzi me ići u kupovinu. 302. Mrzi me da idem u kupovinu. 311. Morao sam mu se izviniti. 312. Morao sam da mu se izvinim. 321. Žele da nam kupe 322. Žele nam kupiti kompjuter. kompjuter. 331. Neka dođe kad voli plivati. 332. Neka dođe kad voli da pliva. 341. Boji se javiti se. 342. Boji se da se javi. 351. Jovan neće ići na koncert. 352. Jovan neće da ide na koncert. 361. Polako je zaboravljala 362. Polako je zaboravljala da se radovati se. raduje. 371. Može se da se pogreši. 372. Može se pogrešiti. 381. Naučiće ih deliti složenu 382. Naučiće ih da dele složenu rečenicu. rečenicu. 391. Savet škole je odlučio da 392. Savet škole je odlučio nagradi sve predložene učenike. nagraditi sve predložene učenike. 401. Prekinuo je studirati. 402. Prekinuo je da studira. 411. Treba raditi. 412. Treba da se radi. 421. Uspeo je da savlada sve 422. Uspeo je savladati sve teškoće. teškoće. 431. Prilika mi je videti Pariz. 432. Prilika mi je da vidim Pariz. 223 441. Lako je zapovedati. 442. Lako je da se zapoveda. 451. Nemam vremena čitati. 452. Nemam vremena da čitam. 461. Mešati dok se ne izjednači. 462. Da se meša dok se ne izjednači. 471. Htela ga je zvati. 472. Htela je da ga zove. 481. Ume govoriti izokola. 482. Ume da govori izokola. 491. Trudi se da ugodi svakome. 492. Trudi se ugoditi svakome. 501. Poželela je dobiti bicikl. 502. Poželela je da dobije bicikl. 511. Nije mu dao da nam kupi 512. Nije mu dao kupiti nam hleb. hleb. 521. Volela je crtati. 522. Volela je da crta. 531. Mora se raditi. 532. Mora se da se radi. 541. Odbio je pomoći mu. 542. Odbio je da mu pomogne. 551. Ne pristaje joj vikati. 552. Ne pristaje joj da viče. 561. Sramota je da se govori 562. Sramota je govoriti tako. tako. 571. Teško mi je priznati zločin. 572. Teško mi je da priznam zločin. 581. Nema se vremena da se čita. 582. Nema se vremena čitati. 591. Želi završiti kurs za 592. Želi da završi kurs za stenografiju. stenografiju. 601. Ni tada nisu dozvolili 602. Ni tada nisu dozvolili omladini omladini da gleda ovaj film. gledati ovaj film. 611. Neka dođe kad zavoli da 612. Neka dođe kad zavoli plivati. pliva. 621. Nemamo mogućnost 622. Nemamo mogućnost da studirati u Americi. studiramo u Americi. 631. Običaj je da se popije kafa. 632. Običaj je popiti kafu. 641. Nije se usuđivao 642. Nije se usuđivao da protivreči. protivrečiti. 224 651. Uspevao je savladati sve 652. Uspevao je da savlada sve teškoće. teškoće. 661. Pokušao je da ga reši. 662. Pokušao ga je rešiti. 671. Odbio je primiti poklon. 672. Odbio je da primi poklon. 681. Jovan će ići na koncert. 682. Jovan će da ide na koncert. 691. Uči ga da peva. 692. Uči ga pevati. 701. Nastaviće raditi po starom. 702. Nastaviće da rade po starom. 711. Namerava kazati mu sve. 712. Namerava da mu sve kaže. 721. Navikao se da se pozdravlja. 722. Navikao se pozdravljati se. 731. Mrzi pevati. 732. Mrzi da peva. 741. Mogao je uraditi i više. 742. Mogao je da uradi i više. 751. Zna slušati. 752. Zna da sluša. 761. Zaboravila je da se raduje. 762. Zaboravila je radovati se. 771. Želim ti brzo ozdraviti. 772. Želim ti da brzo ozdraviš. 781. Ranije nisu dozvoljavali 782. Ranije nisu dozvoljavali omladini da gleda ovaj film. omladini gledati ovaj film. 791. Nije mu dala govoriti. 792. Nije mu dala da govori. 801. Zavoleo je crtati. 802. Zavoleo je da crta. Na kraju, odgovorite na sledeća pitanja: 1. Da li znate (ili mislite da znate) šta je suština ovog upitnika? ________________________________________________________________________ 2. Na koji način ste određivali koju rečenicu ćete izabrati? ________________________________________________________________________ 3. Kakvo je Vaše opšte raspoloženje prema ovom upitniku? ________________________________________________________________________ HVALA!!! HVALA!!! HVALA!!! HVALA!!! HVALA!!! 225 BIBLIOGRAPHY: ARSENIJEVIĆ, BOBAN. 1997. Sintaksičko-morfološki status infinitiva. Književnost i jezik. 45/2- 3:47-55. BELIĆ, ALEKSANDAR. 1955. O značaju posmatranja vlastitog jezika za poznavanje jezika uopšte. Naš jezik (nova serija). 6/7-10:197-208. BELIĆ, ALEKSANDAR. 1969. Istorija srpskohrvatskog jezika. Belgrade. BIBOVIĆ, LJILJANA. 1976. The Infinitive as Subject in English and Serbo-Croatian. Yugoslav Serbo-Croatian-English Contrastive Project : A; Reports. 10:3-19. BOŠKOVIĆ, RADOSAV. 1968. Osnovi uporedne gramatike slovenskih jezika. Belgrade. BRABEC, IVAN, MATE HRASTE, and hrvatskosrpskoga jezika. Zagreb. SRETEN ŽIVKOVIĆ. 1968. Gramatika BROWNE, WAYLES E. 1968. Srpskohrvatske enklitike i teorija transformacione gramatike. Zbornik za filologiju i lingvistiku. 11:25-29. BROWNE, WAYLES E. 1974. On the Problem of Enclitic Placement in Serbo-Croatian. Slavic Transformational Syntax. Ann Arbor. 36-52. BROWNE, WAYLES E. 1993. Serbo-Croat. Bernard Comrie and Greville G. Corbett, eds. The Slavonic languages. London and New York. 306-387. BROZOVIĆ, DALIBOR. 1953a. O vrijednosti infinitiva i prezenta s veznikom da. Jezik; časopis za kulturu hrvatskoga književnog jezika. 2/1:13-18. BROZOVIĆ, DALIBOR. 1953b. O razlikama između infinitiva i prezenta s veznikom da. Jezik; časopis za kulturu hrvatskoga književnog jezika. 2/5:153-154. 226 BROZOVIĆ, DALIBOR and PAVLE IVIĆ. 1988. Jezik, srpskohrvatski/hrvatskosrpski, hrvatski ili srpski. Zagreb. CRAIG, COLETTE. 1975. On Serbo-Croatian Complement Sentences. Yugoslav Serbo-Croatian-English Contrastive Project : B; Studies. 6:148-164. CULICOVER, PETER W. and RAY JACKENDOFF. 2005. Simpler Syntax. Oxford. ČASULE, ILIJA. 1988. The Development of the Macedonian Verbal Noun in the Context of the Loss of the Infinitive. International Journal of Slavic Linguistics and Poetics. 37:61- 71. ĆIRKOVIĆ, MILORAD. 1985. Negacija uz infinitiv i rječcu da + prezent. Naš jezik. 26/4-5:261-263. ĐUKANOVIĆ, VLADO. 1986. Upotreba infinitiva u jeziku A. Šenoe, M. Glišića, Z. Majdaka i D. Nenadića. Naš jezik. 27/1-2:38-66. GALLIS, ARNE. 1970. Zum Gebrauch von DA+Präsens versus Infinitiv beim Futurum I in Serbokroatischen: Untersuchungen über die ‘Koreni’ von Dobrica Ćosić. Zbornik Matice srpske za filologiju i lingvistiku. 13/2:217-223. GRICKAT, IRENA. 1952. O jednom slučaju mešanja sveza da i što. Naš jezik (nova serija). 3/5-6:196-207. GOLQB, Z. 1954. Funkcja syntaktyczna partykuły da w językach pd.-słowianskich (bułgarskim, macedońskim i serbo-chorwatskim). Biuletyn polskiego towarzystwa językoznawczego. 13:67-92. GUDKOV, V.P. 1958. Употребление инфинитива и конструкции с да в сербском языке (сочетания с глаголами). Вопросы лингвистики и методики преподавания иностранных языков. 6:98-112. HOCK, HANS HENRICH. 1991. Principles of historical linguistics. Berlin and New York. HOCK, HANS HENRICH and BRIAN D. JOSEPH. 1996. Language history, language change, and language relationship: an introduction to historical and comparative linguistic. Berlin; New York. IVIĆ, PAVLE. 1955. O vidu glagolskog oblika b u d e m. Naš jezik (nova serija). 6/7-10: 237-245. IVIĆ, PAVLE. 1971. Srpski narod i njegov jezik. Beograd. 227 IVIĆ, PAVLE, IVAN KLAJN, MITAR PEŠIKAN and BRANISLAV BRBORIĆ. 1991. Језички приручник. Belgrade. IVIĆ, MILKA. 1970. O upotrebi glagolskih vremena u zavisnoj rečenici: prezent u rečenici s veznikom da. Zbornik Matice srpske za filologiju i lingvistiku. 13/1:43-55. IVIĆ, MILKA. 1972. Problematika srpskohrvatskog infinitiva. Zbornik Matice srpske za filologiju i lingvistiku. 15/2:115-138. JACKENDOFF, RAY. 1969. Some Rules of Semantic Interpretation for English. Unpublished doctoral dissertation, MIT. JACKENDOFF, RAY, and CULICOVER, PETER W. 2003. The Semantic Basis of Control. Language. 79:517-556. JOSEPH, BRIAN D. 1983. The synchrony and diachrony of the Balkan infinitive : a study in areal, general, and historical linguistics. Cambridge and New York. KALOGJERA, DAMIR. 1971. The Expression of Future Tense in Eglish and in SerboCroatian. Yugoslav Serbo-Croatian-English Contrastive Project : A; Reports. 4:50-72. KALOGJERA, DAMIR. 1982. The English Modals and Their Equivalents in Serbo-Croatian. Zagreb. KATIČIĆ, RADOSLAV. 1986. Sintaksa hrvatskoga književnog jezika. Zagreb. KORDIĆ, SNJEŽANA. 1997. Serbo-Croatian. Munich and New Castle. KRAVAR, MIROSLAV. 1953a. O “razlici” između infinitiva i veze da+prezenta. Jezik: Časopis za kulturu hrvatskoga književnog jezika. 2/2:43-47. KRAVAR, MIROSLAV. 1953b. O “razlici” između infinitiva i veze da+prezenta. Jezik: Časopis za kulturu hrvatskoga književnog jezika. 2/3:70-74. KULJBAKIN, STJEPAN. 1921. Nekoliko reči o konstrukciji “da sa indikativom” mesto infinitiva u staroslovenskom prevodu Jevanđelja. Zbornik filoloških i lingvističkih studija A. Beliću povodom 25-godišnjice njegova naučnog rada posvećuju njegovi prijatelji i učenici. 229-232. LALEVIĆ, M. S. 1951. Sintaksa srpskog jezika. Belgrade. LANDAU, IDAN. 2000. ELEMENTS OF CONTROL Structure and Meaning in Infinitival Constructions. Dordrecht, Boston, and London. 228 LANDAU, IDAN. 2004. The Scale of Finiteness and The Calculus of Control. Natural Language & Linguistic Theory. 22:811–877. LK. Internet ID of Serbia and Montenegro. http://www.gov.yu/start.php?je=s&id=6 MARETIĆ, TOMO. 1963. Gramatika hrvatskoga ili srpskoga književnog jezika. Zagreb. MELVINGER, JASNA. 1982. Supstandardni prijedložni infinitiv i odgovarajuća sintaksička sredstva u hrvatskom književnom jeziku. Jezik: Časopis za kulturu hrvatskoga književnog jezika. 29/3:74-76. MILOŠEVIĆ, KSENIJA. 1978. O proučavanju vremenskih glagolskih oblika u serbokroatistici. Zbornik za filologiju i lingvistiku. 21/2:93-121. MOSKOVLJEVIĆ, MILOŠ S. 1936. Još o upotrebi svezice “da” s prezentom mesto infinitiva. Naš jezik. 4:106-113. MUSIĆ, A. 1913. Značenje perfektivnog prezenta u negativnim pitanjima u hrvatskom ili srpskom jeziku. Južnoslovenski filolog. 1/1-2:27-34. NEWEKLOWSKY, GERHARD (ed.). 2003. BOSANSKI-HRVATSKI-SRPSKI BOSNISCHKROATISCH-SERBISCH Aktuelna pitanja jezika Bošnjaka, Hrvata, Srba i Crnogoraca. Wien NOVAKOVIC, STOJAN. 1902. Srpska gramatika. Belgrade. PAVLOVIC, MILIVOJ. 1960. La perte de l’infinitif dans les langues bakaniques. Zbornik za filologiju i lingvistiku. 3:35-45. PESIKAN, MITAR. 1985. Napomene uz dva prethodna priloga. Naš jezik. 26/4-5:269-272. PROGOVAC, LJILJANA. 1992. Subjunctive: Transparency of Functional Categories. The Penn Review of Linguistics. 16:122-134. PROGOVAC, LJILJANA. 1993. Locality and Subjunctive-like Complements in Serbo-Croatian. Journal of Slavic Linguistics. 1/1:116-144. PROGOVAC, LJILJANA. 1998. Event Pronominal to. Journal of Slavic Linguistics. 6/1:3-40. RADIĆ, PRVOSLAV. 2002. Balkan Features in Standard and Colloquial Serbian. Presentation at the 13th Biennial Conference on Balkan and South Slavic Linguistics, Literature, and Folklore, Chapel Hill, North Carolina. RAKIĆ, STANIMIR. 1984. Glagol hteti i struktura pomoćnih i modalnih glagola u srpskohrvatskom jeziku. Linguistica. 24:355-368 229 RAKIĆ, STANIMIR. 1986. Definicija nekih osnovnih transformacija srpskohrvatskog jezika. Južnoslovenski filolog. 42:95-106. RAKIĆ, STANIMIR. 1989. The Evidence for Inverse Cycle in Serbo-Croatian. Lingua Posnaniensis: Czasopismo poswiecone jezykoznanstwu porownowczemu i ogolnemu. 30: 89-95. Rastko. http://www.rastko.org.yu ROSENBAUM, PETER. 1967. The Grammar of English Predicate Complement Constructions. Cambridge, Mass.: MIT Press. SEDLÁČEK, JAN. 1970. Srpskohrvatske potvrde o razvitku rečenica sa da u južnim slovenskim jezicima. Zbornik Matice srpske za filologiju i lingvistiku. 13/2:5971. SEDLÁČEK, JAN. 1989. Serbocharvatske protejsky latinskoromanske konstrukce s habere (debere) + infinitive a romanskeho habeo-futura. Časopis pro Slovanskou Filologii. 58/1-2:106-114. SIMIĆ, RADOJE. 2002. Srpska gramatika II Sintaksa. Belgrade. SIMIĆ, RADOJE and JELENA JOVANOVIĆ. 2002. Srpska sintaksa I.-II./III.-IV. Belgrade. STANOJČIĆ, ŽIVOJIN. 1967. Jezik i stil Iva Andrića. Belgrade. STANOJČIĆ, ŽIVOJIN and LJUBOMIR POPOVIĆ. 1992. Gramatika srpskoga jezika. Belgrade and Novi Sad. STEVANOVIĆ, MIHAILO. 1935. Infinitiv i svezica “da” sa prezentom. Naš jezik. 3:282-288. STEVANOVIĆ, MIHAILO. 1954a. Naporedna upotreba infinitiva i prezenta sa svezicom da. Naš jezik (nova serija). 5/3-4:85-102. STEVANOVIĆ, MIHAILO. 1954b. Naporedna upotreba infinitiva i prezenta sa svezicom da. Naš jezik (nova serija). 5/5-6:165-185. STEVANOVIĆ, MIHAILO. 1970. Savremeni srpskohrvatski jezik II. Belgrade. STEVANOVIĆ, VIDA. 1986. Oko alternativne upotrebe dvaju oblika u drugom delu složenog futura. Naš jezik. 27/1-2:118-124. SUBIRATS-RÜGERBERG, CARLOS. 1990. The relation between infinitival and sentential complements in traditional grammar. Linguisticæ Investigationes. 14/1:81-93. 230 TANASIĆ, SRETO.1996. Prezent u savremenom srpskom jeziku. Belgrade. TERZI, ARHONTO. 1992. PRO in Finite Clauses A Study of the Inflectional Heads of the Balkan Languages. New York: The City University of New York Ph. D. Dissertation. VOJVODIĆ, DOJČIL. 1996. Status futura I u književnoumjetničkom i publicističkom stilu. Južnoslovenski filolog. 52:107-121. VRZIĆ, ZVJEZDANA. 1996. Categorial Status of the Serbo-Croatian “Modal” da. Annual Workshop on Formal Approaches to Slavic Linguistics The College Park Meeting 1994. 291-312. ZEC, DRAGA. 1987. On Obligatory Control in Clausal Complements. In: Masayo Iida, Stephen Wechsler, and Draga Zec (eds.). Working Papers in Grammatical Theory and Discourse Structure. 139-168. 231