THE SEMANTICS OF RUSSIAN INDEFINITE PRONOUNS: SCOPE, DOMAIN WIDENING, SPECIFICITY, AND PROPORTIONALITY AND THEIR INTERACTION By Olga Eremina A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Linguistics 2012 ABSTRACT THE SEMANTICS OF RUSSIAN INDEFINITE PRONOUNS: SCOPE, DOMAIN WIDENING, SPECIFICITY, AND PROPORTIONALITY AND THEIR INTERACTION By Olga Eremina The main goal of this dissertation is to consider the different types of indefinites in Russian as a system and provide a semantic account for each of them that would be able to naturally explain their distribution. The four sets of so-called ‘indefinite pronouns’ (-to, -nibud’, -libo, and koe- indefinites) differ in their degree of specificity. Indefinites with the prefix koe- are epistemically specific (known to the speaker, but unknown to the addressee). They also induce the effect of proportionality (and refer to a part of a given set). As such, they can be analyzed as unambiguously choice-functional proportional indefinites. Indefinites with the suffix -to are usually analyzed as scopally (but not epistemically) specific. The conclusion that they can only take a wide scope often leads to analyzing them as choice-functional items as well. Such an approach, though, requires some additional stipulations to be made to distinguish -to and koe- indefinites. It also fails to account for the cases when -to indefinites are perceived as having a narrow apparent scope. I propose that -to indefinites are instances of the parameterized choice function with an implicit argument which can be bound from outside; in particular, the function can be parameterized by times yielding the quazi- narrow-scope interpretation. Under such an analysis, -to indefinites are not classic choicefunctional indefinites because they can behave similar to quantifiers when the implicit parameter is bound from outside. Traditional approaches to -nibud’ indefinites are based on their non-specificity or/and peculiar distribution (they are only licensed in questions, conditional sentences, imperatives, future tense sentences, in the scope of modal operators, and in some characterizing/generic sentences, e.g. dictionary definitions). I propose that their syntactic distribution is due to their domain-widening semantics. The indefinites with the suffix -libo are very similar to -nibud’ pronouns in their semantics and functions. The only difference lies in the degree of universality. -Nibud’ indefinites can never receive a universal interpretation, -libo items, however, seem to function as universals in comparisons and under negation. This difference is caused by another, stylistic in nature, property of the system: only a formal-style -libo but not a stylistically neutral -nibud’ is able to function as a universal (free choice) indefinite in some contexts, because the truly free-choice items do not meet the stylistic requirements of these contexts. The system of the different sets of Russian indefinite pronouns, then, covers the whole continuum of ‘indefiniteness’ via non-overlapping in their semantics and functions lexical items: from mere choice-functional koe- items through ‘half-quantificational’ -to indefinites to domainwidening (quantificational) -nibud’ indefinite pronouns. The indefinites demonstrate competition and distribution of labor. Each of the indefinites has its own semantics and functions, but at the same time their semantics and behavior are closely tied to and dependent on the other parts of the system. Copyright by OLGA EREMINA 2012 ACKNOWLEDGEMENTS This work would not be possible without the invaluable contribution of a large group of people. My dissertation advisor, Marcin Morzycki, made formal semantics understandable and enjoyable for me. He taught me to concentrate on specific problems and resolve them one by one instead of just being excited and frustrated about an enormous number of puzzles of different nature. He kept me motivated and excited about what I was working on for all these long years. Every meeting with him was always extremely productive; some of them made me re-consider whole chapters of my work. Alan Munn’s critical eye has always been able to catch every little problem with my analysis – from unclear wording to very sophisticated formalism. Many times, preparing a presentation of some not fully developed chunks of my dissertation I relied on his words: “If you don’t completely understand something complicated, try to explain it to others in simple words”, and so many times it worked! Cristina Schmitt brought a broader perspective into my work and made me see connections between things that seemed completely separate before. She managed to find time and supportive words for me in the most stressful situations. Alan Beretta taught me to be very careful with data, not to trust just my own judgments, and pay attention to the contexts. He never forgot to repeat that it never hurts to look closely into someone else’s approach and try to understand it as deeply as you can even if it seems completely opposite to what you believe in; and that one learns more from trying to understand v than from criticizing. In many cases, it made me go back to the same papers again and again, and after the 4th time find something very important there. I am happy I had a chance to also share my ideas with participants of CLS-18 and FASL45. The input from these conference presentations made me re-consider some of my ideas. I am endlessly grateful to the faculty of the Linguistics Department of MSU for creating and encouraging the atmosphere of true love to Linguistics, where students are excited of what they are doing and are not afraid of sharing both their findings and problems with each other. The criticism, and the insights I received from the participants of our Semantics discussion group were priceless. I would like to thank everyone who asked questions, shared their ideas, helped develop formalism, suggested cross-linguistic comparisons, and listened to my presentations and sometimes said, “it makes no sense” – so that next time I’d make it more sensible. Thank you, Matt Husband, Ian Anderssen, Greg Johnson, Curt Anderson, Adam Gobeski, Ai Matsui and many, many others! All the errors and mistakes in the text are, of course, of my own. vi TABLE OF CONTENTS INTRODUCTION...........................................................................................................................1 CHAPTER 1. RUSSIAN INDEFINITE PRONOUNS: DATA AND THE EXISTING APPROACHES 1.1 The jigsaw…………….................................................................................................5 1.2 Semantic non-interchangeability ..................................................................................9 1.3 Different semantics and different types of sentences .................................................19 1.4 A ‘disjunctive’ take on -nibud’ indefinites .................................................................23 1.5 Formally speaking: a choice functional analysis of Russian indefinites ....................26 1.6 Speaker identifiability and -koe indefinites ................................................................35 CHAPTER 2. –TO INDEFINITES 2.1 Semantics and distributional properties .....................................................................37 2.3 Can -to indefinites be quantificational? .....................................................................55 2.4. Parameterized choice function for Russian -to indefinites .......................................60 2.5 Conclusions ...............................................................................................................70 CHAPTER 3. -NIBUD’ AND -LIBO INDEFINITES 3.1. -NIBUD’ vs. other indefinites: distributional properties...........................................71 3.2. Semantic properties of -NIBUD’ indefinites ............................................................77 3.3. Formal analysis: accounting for the semantics and distribution ...............................83 3.4. -LIBO indefinites – same or different .......................................................................95 3.5 Conclusions ..............................................................................................................103 CHAPTER 4. KOE- INDEFINITES 4.1 Choice functional KOE- and problems with it .........................................................105 4.2 An alternative approach ...........................................................................................113 4.3 Conclusions ..............................................................................................................125 CONCLUDING REMARKS AND QUESTIONS FOR FUTURE RESEARCH .....................126 REFERENCES ...........................................................................................................................135 vii INTRODUCTION In this dissertation, I analyze the semantics and distributional patterns of Russian indefinite pronouns. Russian lacks indefinite and definite articles and the so-called indefinite pronouns become means by which different shades of indefinites are decoded in the language. I look closely at four different sets of indefinites: indefinites with the prefix koe-, and the suffixes -to, -nibud’, and -libo). The four types of indefinite pronouns construct a complex and unique system, covering epistemically specific (“known to the speaker”), scopally specific (taking a wide scope in a sentence), and existential domain-widening indefinites. Despite existing multiple studies of particular elements and aspects of this system both in traditional Russian linguistics and in the framework of formal semantics and pragmatics, the whole system with the complex relationships between its members has never received an adequate description. The richness of the system makes it a very attractive research topic. Russian has separate lexical items for what in English is presented by different interpretations of the same lexeme; for example, quantificational and specific (choice-functional) a-indefinites and some-indefinites (see, Kratzer 1998 for discussion and detailed analysis of these English indefinites) are represented by different lexical items in Russian. Russian also has separate items for what is analyzed as a free-choice (universal) any and as a polarity sensitive (existential) any in English (see Kadmon&Landman, 1993; Krifka 1995, Zwarts 1995 for discussion). The strict division of labor between different types of indefinites is very convenient in tearing apart units with different semantics, which in other languages may be hidden under the same lexeme. 1 The other peculiarity of the system of Russian indefinites is their systematic appearance and non-interchangeability in particular syntactic environments. In this dissertation I show that the distribution of the different types of indefinite pronouns that often seems chaotic is tightly connected with the semantics of the indefinites. The structure of the dissertation is as follows. Chapter 1 demonstrates the differences in the semantic and syntactic behavior of the four types of indefinites. I survey the existing literature on Russian indefinites and demonstrate that none of the approaches offered so far are capable of accounting for this behavior. Chapter 2 provides a detailed analysis of pronouns with suffix -to. I show that in different contexts these indefinites are systematically interpreted as having a wide scope; however, there are contexts where they seem to allow for different scope options. I consider the abilities of choice function approach in accounting for the behavior of the Russian scopally specific indefinites. The “pure” CF interpretation of -to indefinites correctly predicts their widescope readings, but cannot account for the instances of intermediate or quasi-narrow scope. I argue that Russian -to indefinites can never be interpreted quantificationally, which means that an account suggesting an ambiguity of -to indefinites between choice-functional and quantificational readings is not grounded, and thus an alternative explanation of the perceived narrow-scope readings is needed. I argue that the behavior of the -to indefinites can be explained in terms of a parameterized choice function interpretation. I consider different sources of the parameterization, and look at contextual limitations and cues facilitating different interpretations of the indefinites. In Chapter 3 I discuss -nibud’ and -libo indefinites, which are claimed to be semantically equivalent but belong to different registers (the former is stylistically neutral and the latter is 2 formal). These indefinites need to be in the scope of a licensor to be grammatical; they can only appear in certain environments, e.g. in questions, with modal verbs, in conditionals, etc. -Nibud’ and -libo indefinites are close to the English any in inducing a domain widening effect but, unlike any, they allow for exceptions and do not pass tests for universal interpretations. I propose that the domain widening effect is a result of a presupposition that requires access to a possible world(s) different from the evaluation world; I show that if an indefinite has such a presupposition it can only be used in contexts which provide access to other possible worlds; this is exactly the reason for the peculiar distribution of -nibud’ and -libo indefinites. I further investigate the relations between the two indefinite pronouns and demonstrate that -libo indefinites can be used where -nibud’ indefinites are rare or impossible: in comparisons and with grammatical negation. However, this difference can be explained as a result of domain-widening / strengthening and is facilitated by a mismatch between the register of the speech and the available lexical resources. I argue that the two pronouns do, in fact, have the same semantics. In Chapter 4 I focus on the epistemically specific indefinites with the prefix koe-. I show that an attempt to account for their semantics by using a classical version of a choice functional approach accounts for their epistemic specificity but fails to account for the other properties they have. I argue that the major semantic feature of these indefinites is their reference to a subset of a given set. They have a proportional value and can only refer to a part of the set that is smaller than its half. I consider two different functions of these indefinites and explain why usages in non-introductory contexts put some limitations on koe- indefinites and lead to the preferred plural interpretation of these pronouns. In the concluding chapter I focus on the Russian indefinite pronouns as a system; I discuss how the four types of pronouns complement each other in fulfilling different functions 3 and how they interact with other parts of the language domain of indefiniteness (e.g. with freechoice items). I then demonstrate the way in which this system is unique and how it correlates with how other languages choose to cover the concept of indefiniteness by lexical resources. I show some points of connection and discuss how insights from the analysis of Russian indefinite pronouns may be useful in analyzing indefinites in other languages. 4 CHAPTER 1 RUSSIAN INDEFINITE PRONOUNS: DATA AND THE EXISTING APPROACHES 1.1 The jigsaw Russian has a rich system of so-called “indefinite pronouns”. They are made by 1 prefixing koe- or adding suffixes -to, -nibud’ , -libo to “question words” chto (‘what’), kto (‘who’), gde (‘where’), kogda (‘when’), kakoj (‘which’) etc. They all share the meaning of indefiniteness (hence the term) and can be all (roughly) translated into English as something, someone, somewhere, sometime, some (as in some boy) etc. Examples of different usages of indefinite pronouns are given in (1) below. As the data shows, some contexts allow for any of the indefinites (1a); some, for instance, dictionary definitions, only allow for -nibud’ or -libo items (1b-c); the register matters as well, and formal style utterances prefer -libo indefinites (1d); some contexts that are bad with -nibud’ indefinites, as in (1e) and (1h) become acceptable when turn into a question (1f) or contain a modal or a verb of desire (1g), or if the verb is used in the imperfective aspect (1i); using the suffix koe- may cause plural interpretation of otherwise singular noun phrases (1j): 1 Morphologically, the suffix -nibud’ has a complex structure, including a negation part -ni- and the root of the verb to be. It is possible that originally the whole combination [what+not+be] meant something like ‘what not’, functioning as a free choice element (like whatever or just any in English). However, in contemporary Russian the suffix does not have such a meaning. The role of ‘free choice’ is given to a different unit, namely, to ugodno (kto ugodno – ‘whoever’, chto ugodno – ‘whatever’ etc.). For a detailed review of all types of indefinites in Russian, see Haspelmath (1997). 5 (1) a. On prigotovit chto-to/chto-nibud’/koe-chto vkusnoje na uzhin. He cook-FUT something (TO/NIBUD’/KOE) delicious for dinner. ‘He will cook something delicious for dinner’. b. Shveja – rabotnitsa po shit’ju chego-nibud’/chego-libo. Seamstress – worker-FEM in sewing something (NIBUD’/LIBO). ‘Seamstress is a worker (specializing) in sewing something’ (dictionary definition). c. Sharakhnut’ – sil’no udar’it’. Sharakhnut’ kamnem kogo-nibud’ /kogo-libo. Strike-INF–strongly hit-INF. Strike-INF with stone someone (NIBUD’/LIBO). ‘To strike – to hit with force. E.g. To strike someone with a stone’ (dictionary definition). d. Molochnyje producty dolzhny khranit’sa gde-libo v kholodnom pomesh’enii. Milk product-PL must-PL store-INF somewhere (LIBO) in cold place. ‘Dairy products have to be stored somewhere in a cold place’ (instructions for handling food). e. Ivan spel *kakuju-nibud’ / kakuju-to pesniu Ivan sing-PAST some (*NIBUD’/TO) song ‘Ivan sang a song’ f. Ivan spel kakuju-nibud’ pesniu? Ivan sing-PAST some (NIBUD’) song ‘Did Ivan sing a song?’ g. Ivan hochet spet’ kakuju-nibud’ / kakuju-to pesniu Ivan wants to sing some (NIBUD’/TO) song ‘Ivan wants to sing a (some) song. 6 h. Ona kupila *kakuju-nibud’/kakuju-to gazetu i poshla domoj She bought-PERF newspaper and went-PERF home ‘She bought a newspaper (once) and went home’. i. Ona pokupala kakuju-nibud’/kakuju-to gazetu i shla domoj She bought-IMPERF some (NIBUD’/TO) newspaper and went-IMPERF home ‘She (usually) bought a newspaper and went home’ j. Ja nashla koe-chto interesnoje v etoj knige. I found something (KOE) interesting in this book ‘I saw something interesting in this book (I know exactly what it is and I may tell you later but at this point I don’t what to specify anything)’. k. Ja videla koe-kogo iz druzej. I saw some (KOE) of friends ‘I saw some of my friends (more than one person, and I know which ones, but it is not important to list them)’. A closer look into the semantics of the members of the four sets of indefinites shows that they are not synonymous. When used in the same sentence, different indefinite pronouns are interpreted differently. In fact, seemingly “the same” sentences belong to different broader contexts, and describe different speech situation/scenarios: (2) a. Masha prigotovit chto-to vkusnoje na uzhin. Masha cook-FUT something (TO) delicious for dinner. ‘Masha will cook something delicious for dinner’ (and she knows what it is going to be, but the speaker does not). 7 b. Masha prigotovit chto-nibud’ vkusnoje na uzhin. Masha cook-FUT something (NIBUD’)) delicious for dinner. ‘Masha will cook something delicious for dinner’ (we don’t know what it is going to be and we don’t know whether Masha has decided either). c. Masha prigotovit koe-chto vkusnoje na uzhin. Masha cook-FUT something (KOE) delicious for dinner. ‘Masha will cook something delicious for dinner’ (Masha knows what she is planning to cook and I (the speaker) know it too, but I don’t want to tell you). d. (?) Masha prigotovit chto-libo vkusnoje na uzhin. Masha cook-FUT something (LIBO) delicious for dinner. ‘Masha will cook something delicious for dinner’ (means the same as (2b) but is 2 stylistically odd . See (1b), (1d), and (1f) for “stylistically proper” instances of –libo). The examples in (1) and (2) allow for a very rough distinction between some of the indefinite pronouns. It is clear that koe- indicates the speaker’s knowledge about the referent and, in parallel, the speaker’s intention to retract from using the actual name of the thing. Strictly speaking, this is not even an indefinite pronoun, it is more a “hidden definite”; 3 sometimes the term “weak definite” is used for them (Paducheva 1985) . Only koe- indefinites are undoubtedly compatible with the actual demonstration of the “secret” thing: 2 It is unlikely to hear (2d) actually uttered (even though it is not an ungrammatical sentence): libo is commonly used in formal writing (dictionaries, manuals, reference materials etc.), and sounds odd in informal conversations. 3 This work is written in Russian; all ciatations from Paducheva here are given in my translation. 8 (3) I came to visit my friend. I brought something wrapped as a gift. I’m giving this thing to my friend, saying: Ja tebe koe-chto /#chto-to/*chto-nibud’ prinesla. I you-DAT something (KOE/#TO/*NIBUD’) bring-PAST. ‘I brought you something’. This property of koe- indefinites attracts the immediate attention of everyone studying Russian indefinites. However, as we will see later, the characteristics of koe- indefinites are not limited to simply referring to something known to the speaker. 1.2 Semantic non-interchangeability The first most obvious observation about -nibud’ indefinites is that they have distributional limitations. These indefinites are used in questions, with future tense verbs, with verbs of desire, with modals etc. They are not licensed in simple declarative present or past tense sentences describing a one-time activity/event. Cases of “parallel” contexts allowing for both -nibud’ and -to indefinites are not very common. But even if both types of indefinites are grammatical in the same sentence, the sentences do not have the same interpretation. The contrastive semantic properties of -to and nibud’ pronouns can be observed in the data in (2a) and (2b), repeated in (4a-b) below for convenience: (4) a. Masha prigotovit chto-to vkusnoje na uzhin. Masha cook-FUT something (TO) delicious for dinner. ‘Masha will cook something delicious for dinner’ (and she knows what it is going to be, but the speaker does not). 9 b. Masha prigotovit chto-nibud’ vkusnoje na uzhin. Masha cook-FUT something (NIBUD’)) delicious for dinner ‘Masha will cook something delicious for dinner’ (we don’t know what it is going to be and we don’t know whether Masha has decided either). The difference seems to be in the scope that ‘something delicious’ takes in these cases: in (4a) it is a wide scope (there is a particular thing she is cooking), and in (4b) it is a narrow scope (she is cooking something). This difference is important, because it makes -to and -nibud’ indefinites non-interchangeable within the same speech situation: (5) a. Scenario 1. Masha called me to say that she found a fantastic recipe on-line and she wants to try it today. I am calling my friend inviting her to join me and come to Masha’s place tonight. I say: Masha prigotovit chto-to / *chto-nibud’ vkusnoje na uzhin. Masha cook-FUT something (TO / *NIBUD’)) delicious for dinner ‘Masha will cook something delicious for dinner’. b. Scenario 2. Masha is my roommate. We cook in turns. Masha is a good cook. It’s her turn to make dinner tonight. I call my friend and I say: Masha prigotovit chto-nibud’ / *chto-to vkusnoje na uzhin. Masha cook-FUT something (NIBUD’ / *TO)) delicious for dinner ‘Masha will cook something delicious for dinner’. So, the simplest way of differentiating between -to indefinites and -nibud’ indefinites is to define the former as “specific” and the latter as “non-specific”. In fact, this is what many 10 linguists and (especially) teachers of Russian as a foreign language do. A very typical example is provided by the textbook used at MSU for intermediate level Russian (Golosa-II); it defines to indefinites as being “more specific than -nibud’ ones”. And this is often the only way the distinction is drawn in the linguistics literature on the nature of indefinites in Russian. However, some instances of -to indefinites seem non-specific because they receive a narrow scope interpretation, as in (6): (6) On ochen’ obsh’itel’nyj chelovek, on (vsegda) priglashajet kakix-to studentov, oni vmeste chitajut kakije-to knigi. He very sociable person, he (always) invite-PRES some (TO) student-PL, they together read-PRES some (TO) book-PL. ‘He is a very sociable person, he (always) invites some students, and they read some books together’. The students and the books in (6) can, in principle, be (the same) particular students and particular books, but the most natural reading of it is ‘some students/books, possibly different all the time’. The difference in specificity is also unable to account for the contrast in (7): (7) a. Zdes’ kakaja-to / *kakaja-nibud’ oshibka. Here some (TO/*NIBUD’) mistake. ‘There is a/some mistake here’. b. Zdes’, navernoje, kakaja-to / kakaja-nibud’ oshibka. Here probably some (TO/NIBUD’) mistake. ‘There is probably a/some mistake here’. 11 The sentence in (7a) does not allow the use of -nibud’. However, a little change (adding “probably”), makes both types of indefinites possible. Importantly, both (7a) and (7b) present the same (non-specific, narrow scope) reading of the indefinites. So the difference here lies not in the change of the meaning of the indefinite, but in the change of context. Another observation from the data in (1) and (2) is the one about the stylistic differentiation between the members of the four sets. The sentences in (2b) and in (2d) have the same meaning, but the latter sounds more formal than the former (the incompatibility of the overall colloquial tone of the sentence and the formal –libo in it makes (2d) stylistically odd). However, some authors indicate that -libo and -nibud’ indefinites are, possibly, not completely 4 equivalent semantically . Due to the complex relationships between the different types of indefinites (semantic, distributional, stylistic etc.) they are usually studied in relation to each other, i.e. as a system. Constructing a system requires defined parameters of comparison, such as specificity vs. nonspecificity, or familiarity vs. non-familiarity to the speaker etc. Various researchers suggest different lines along which such a system can be constructed; often, several different parameters are combined. Sometimes, a triplet “syntactic environments + familiarity to the speaker and/or addressee + existence/non-existence of the referent” serves as the background for comparison. For instance, 4 E.g. Sheliakin (1986: 74) gives several examples where -nibud’ and -libo are not interchangeable, including the one below. (i) Puma begajet bystreje, chem kakoje-libo / #kakoje-nibud’ drugoje zhivotnoje. Puma runs faster than any other animal. 12 5 Pete (1957) and Vsevolodova & Yudina (1963) describe the contrastive indefinites in the following way: -nibud’ is used in questions and other situations, when the referent is unknown and/or its existence is doubtable; -to indefinites refer to something existing in real life and completely unknown or known to either speaker or hearer. Even when only semantic properties of indefinites are considered, at least three separate semantic characteristics may become relevant: (1) indefiniteness / definiteness for the speaker and listener; (2) relation to the actual world (existence/non-existence), and (3) distributivity 6 (Lavrov 1983) . As a result, a definition of each type of indefinite becomes quite complex and is based on a comparison with all other types made along the lines of this three-dimensional system. For instance, for Lavrov, -to indefinites are defined as not known to the speaker (as opposed to koeindefinites), but possibly known to the listener (as opposed to -nibud’ indefinites), indicating the existence of the object/person (as opposed to -nibud’ indefinites which do not presuppose such an existence), and non-distributed (as opposed to both koe- and -nibud’ indefinites). Besides, the differences between the indefinites are often analyzed in non-semantic terms. Lavrov, for example, after proposing the three-dimensional semantically-based definition of the indefinite pronouns, further describes the contrast in terms of syntactic distribution, simply listing the syntactic environments where different types of indefinites are/are not possible. The environments for -nibud’ and -libo indefinites, for instance, are described in the following way: 5 These works are written in Russian; all ciatations from Pete (1957) and Vsevolodova & Yudina (1963) here are given in my translation. 6 This work is written in Russian; all ciatations from Lavrov here are given in my translation. 13 - Declarative sentences in the present (generic) and past tenses with verbs in imperfective aspect referring to regular, repeated activities (distributive interpretation), - Declarative sentences in the future tense with verbs having semantics of indefiniteness, - With infinitives, - With modal words of possibility, - With modus irrealis, including conditionals and imperatives, - Questions (Lavrov, 1983: 17–18) Some researchers claim that the relationships between all the indefinites are based on a 7 contrastive behavior with respect to a single property. E.g., for Yanko-Trinitskaya (1989) , the indefinite pronouns are different in how they present “indefiniteness”. She claims that there are three different types of indefiniteness: (1) unfamiliarity, (2) inessentiality, and (3) incomplete inclusiveness. According to her, -to pronouns are the ones with “indefiniteness of unfamiliarity” (the addressee does not know what or who the speaker refers to), -nibud’ and -libo items have “indefiniteness of inessentiality” (it is not important to what or whom the speaker refers), and koe- indefinites are characterized by “indefiniteness of incomplete inclusiveness” (they refer to some, but not all objects of the group). However, she still needs to address the difference between -nibud’ and -libo indefinites in terms of style, because these two types of indefinite pronouns cannot be differentiated in terms of “indefiniteness”. 7 This work is written in Russian; all ciatations from Yanko-Trinitskaya here are given in my translation. 14 However, even after including an extra parameter (the stylistic factor) into the system, the analysis still fails to cover all the language data: for instance, Yanko-Trinitskaya is not able to account for the cases of koe- indefinites which do not refer to “some but not all objects of a given group”, but, instead, refer to objects known to the speaker but not known to the addressee as in (2c) repeated as (8) below for convenience: (8) Masha prigotovit koe-chto vkusnoje na uzhin. Masha cook-FUT something (KOE) delicious for dinner ‘Masha will cook something delicious for dinner’ (Masha knows what she is planning to cook and I (the speaker) know it too, but I don’t want to tell you). Neither can this analysis explain the contrast between the sentences in (7a) and (7b) repeated as (9) below: (9) a. Zdes’ kakaja-to / *kakaja-nibud’ oshibka. Here some (TO/*NIBUD’) mistake. ‘There is a/some mistake here’. b. Zdes’, navernoje, kakaja-to / kakaja-nibud’ oshibka. Here probably some (TO/NIBUD’) mistake. ‘There is probably a/some mistake here’. 8 Sheliakin (1986) supports the idea that the four types of indefinites differ in the “flavor of indefiniteness”. All indefinites share the same core function, they “indicate objects, people, places, etc., which are not perceived by the speaker as concrete, individualized definite objects” (which creates a problem for describing koe- indefinites – O.E.). The difference in the 8 This work is written in Russian; all ciatations from Sheliakin here are given in my translation. 15 indefinites is due to the fact that “each affix brings in a particular flavor of indefiniteness”. Depending on this flavor, all indefinite pronouns can be characterized as (a) indefinitenonattached (-nibud’, and -libo), indefinite-attached (-to), and weak definite (koe-). According to Sheliakin, indefinite- nonattached/non-fixed pronouns (-nibud’, and -libo) have the following semantic functions: (i) They indicate a whole set of individuals (larger than one element) in which one element is selected (ii) The set is homogenous (by the properties that the individuals share), but the elements of it are different in some respect (iii) An individual has to be selected (it is impossible to refer to the whole set). Further, he makes the following suggestion for -nibud’indefinites: “-nibud’ pronouns indicate disjunctive relationships (“either … or…”) between indefinite and qualitatively different elements of set. The relationships are based on the logical law of excluded middle: out of two propositions, considered at the same time and in the same respect, only one is true and the other one is false. However, both of them can be true if they refer to different time or different participants”. Sheliakin (1986, pp. 69–70) 16 Citing the common assumption that -libo indefinites are bookish variants of -nibud’ pronouns and their semantics is similar, Sheliakin (1986) expresses some doubt in this generalization giving several examples of non-interchangeability of -nibud’ and -libo indefinites. However, he does not provide any account for these controversial examples (see the analysis of -libo indefinites in Chapter 3 for more data). Sheliakin defines indefinite-attached / fixed pronouns (-to) as indicating individual, fixed elements, which are not concrete for the the speaker, and may belong to a definite or indefinite set. According to him, “there are two ways in which the pronoun indicates non-concrete knowledge: (a) larger or smaller degree of familiarity to the speaker, and (b) irrelevance of the concretization for the speaker”. The first type can be introduced in the following cases: - the speaker can only see/perceive some (visible) features of the element; - the speaker compares a perceived element with some other one based on some (but not total) similarity; - the speaker refers to something intuitively perceived or subjectively interpreted. The second type occurs when none of the features of the element are relevant to the speaker but it is important to indicate the existence of this element (pp. 74–76). Sheliakin (1986) is aware of limitations on the usage of -to indefinites (he mentions that they are not used with imperatives and are almost never used in questions), according to him these limitations are “caused by their reference to a fixed but non-concrete element” (p.77). As a weak definite, a koe- pronoun, according to Sheliakin, adds a quantitative interpretation to the noun phrase: such a pronoun refers to some number (amount) of elements of 17 a given set, or, as a particular case, to one member of the set. The meaning of a koe- pronoun then, is an indication of fixed, limited in quantity and non-concrete for the addressee but known to the speaker element(s) of the set. Non-concreteness with respect to the addressee can be of two types: - The speaker is familiar with the element but in purpose does not name it, - The element is irrelevant for the speaker, but it is important to indicate that the element is a member of a given (known to the speaker) set. Similarly to pronouns with the suffix -to, Sheliakin (1986) gives semantic reasons for the limitations in distribution of -koe pronouns: these indefinites cannot be used in imperative sentences and questions because they do not contain information relevant to the addressee about a unique element and, being known to the speaker, are not compatible with the semantics of questions. Sheliakin’s analysis is based on purely semantic properties of the indefinite pronouns. He argues that the distribution of the indefinites is a result of their semantics. His semantic analysis is more descriptive than explanatory (for instance, his indefinite-nonattached pronouns (-nibud’, and –libo items) are described via different properties of the set they belong to; the definition of indefinite-attached (-to) pronouns is given on a case-by-case basis; etc). As a result, the analysis suffers from the same problems as the previously considered approaches: it is not able to account for the cases of seemingly synonymous usage of different types of indefinites and does not give any explanation of licensing conditions on -nibud’ and -libo indefinites 18 (modals, conditionals, future tense, habitual sentences versus one-time present or past tense sentences, etc.). 1.3 Different semantics and different types of sentences The analysis by Paducheva (1985) is based on both the semantics and the syntactic distribution of the indefinites. She argues that indefinite pronouns can be divided into three classes: pronouns of uncertainty (i.e. unknown to the speaker, -to indefinites), weak definite pronouns (koeindefinites), and non-referential existential pronouns (-nibud’ indefinites). According to her, the semantics of all indefinite pronoun is tightly connected with the concept of a set; thus many properties of indefinites are best described in reference to the set to which they belong. For instance, for all four types of indefinites it is important that the set must consist of similar but not identical objects, so that it is possible to individualize the objects (p. 210). Elements of homogenous sets cannot be referred to by using indefinite pronouns. According to Paducheva (1985, p. 211), pronouns of uncertainty (-to indefinites) are concrete-referring, i.e. they correlate with a specific (fixed) object which is unknown to the speaker (the speaker cannot identify it). Because of this property, -to indefinites are impossible in contexts where the speaker is the subject of the sentence, as in (10b) below: (10) a. Ivan xochet spet’ kakuju-nibud’ /kakuju-to pesniu. Ivan wants to sing some (NIBUD’/TO) song Ivan wants to sing some song. 19 b. Ja xochu spet’ kakuju-nibud’ /*kakuju-to pesniu. I want to sing some (NIBUD’/*TO) song I want to sing some song. In (10a) the -to indefinite is fixed by the desire of Ivan but is unknown to the speaker, in (10b) the singer who chooses the song and the speaker is the same person, which makes a -to pronoun impossible, while a -nibud’ indefinite is perfectly acceptable. Paducheva (1985, p. 212) also provides a list of distributional limitations on -to indefinites (see below), but in most of the cases does not give any explanations as to what causes these limitations: 1. -to indefinites are impossible in imperatives (since the object is not known to the speaker, s/he cannot identify it as needed for the purpose of a request) 2. -to indefinites are not, generally, used in questions (no explanation or considering the cases where these indefinites are acceptable in questions) 3. -to indefinites are not used under the scope of negation 4. -to indefinites in many cases (but not always) cannot serve as antecedents for personal pronouns, it is more natural to use a descriptive expression instead (as illustrated in (11 below): (11) Zdes’ kto-to byl. Kto *ego videl? / Kto videl etogo cheloveka? Here someone (TO) was. Who *he-ACC saw? / Who saw this person? There was someone here. Who saw *him? / Who saw that person?) 20 The next type – weak definite pronouns (koe- indefinites) are characterized as referring to individual (=specific) objects that are known to the speaker but supposedly not known to the addressee (Paducheva, 1985, p. 212-14). These indefinites have two distinct usages (functions): (A) Introductory usage: the object / person introduced by the indefinite is later demonstrated or called by name as in (12) below: (12) Ja tebe koe-chto prines. Smotri, kakaja dynia. I you-DAT something (KOE) brought. Look which melon. I brought you something. Look at this melon. A particular case of an introductory usage is a ‘secret definite’ when the object is known to the speaker and to the listener, but the speaker purposefully does not provide a definite reference in order to indicate that this is some kind of a secret. (B) Non-introductory usage: is only possible with plural objects; if a number is not indicated overtly (by plural ending) the indefinite is still interpreted as plural as in (13) below: (13) Koe-kto ushel iz derevni na zarabotki. Some-who (KOE) left village to money-making ‘Some people left the village to make money’ Impossible interpretation: *’Some person left the village to make money.’ Finally, Paducheva (1985, p. 215) suggests that existential pronouns (-nibud’ indefinites) have a semantic component “the element participating in the situation is not chosen yet”. 21 However, this component is not always present in the semantics of this type of indefinites; it appears in future tense sentences but it does not appear, for instance, in cases when it is not known whether or not the situation itself took place (e.g. with possibly), or in “distributed” contexts when an event correlates with (different) multiple participants and /or different (multiple) times. Another way to describe the semantics of -nibud’ indefinites, suggested by Paducheva (pp. 214-216), is via situations where these indefinites are used. She describes four types of such situations: 1. Situations of alternation (with probably, possibly, maybe, either… … or, if-clauses, questions); in these situations even the existence of the object correlating with the pronoun is unknown; 2. Situations referring to the future (imperative sentences, future tense sentences, sentences with need/want/may etc.); in these situations the object does not exist yet, but it is expected to exist in the future; 3. Situations with an unknown participant (only a few cases when the speaker believes that some unknown object exists); 4. Situations of distribution (a repeating event where times/participants vary). Then, the general property of all contexts allowing for -nibud’ indefinites, according to Paducheva, is that the situation is perceived by a speaker as one of many possible situations, and a -nibud’ indefinite corresponds to a possible participant of such a situation. Therefore, these 22 indefinites are banned in affirmative propositions, except for the situations of distribution (pp. 216-17). Further, Paducheva gives several examples of contexts where -libo indefinites are “clearly not interchangeable” with -nibud’ indefinites (including lexical negation and comparisons); therefore she creates a list of rules predicting where -libo pronouns (but not nibud’ indefinites) are used. However, she leaves an important note after the list, saying that “there are multiple records of wrong usage of -libo pronouns” (p.219), which indicates that these “norms” and “rules” are of a prescriptive nature and do not necessarily correspond with the actual language use. Paducheva mentions examples where -nibud’ and -to indefinites appear in the same context; she argues, however, that indefinites in these cases are not synonymous because the examples with -to (unlike the -nibud’ cases) always have additional semantics of a single / particular / specific participant (p. 220). 1.4 A ‘disjunctive’ take on -nibud’ indefinites 9 Seliverstova (1988) carries on some ideas of Sheliakin about the nature of -nibud’ indefinites but develops them in a different way. First of all, she refuses to use the term “indefiniteness” and “different degrees/flavors of indefiniteness” in respect to the distinction between the four types of indefinites, because the terms by themselves are too unclear and are used differently by different researchers creating additional confusion instead of making things clearer. Some of the authors she cites go as far as characterizing -to indefinites as “definite”. For example, Textbook of Russian for Vietnamese 9 Seliverstova’s work is written in Russian; all ciatations here are given in my translation. 23 students (1960) gives the following definition: “Pronouns with suffix -to indicate a definite object, person or property not known to the speaker or to one of the participants of the 10 conversation” . Seliverstova (1988) argues that it is virtually impossible to “measure” the degree of definiteness; besides, in many cases it is clear that the distinction “indefinite/definite” for these pronouns is simply wrong; compare, for instance, a supposedly “more definite” case of (14a) and “less definite” case of (14b) below: (14) a. Vsiu noch’ revelo more, i v rokote etom slyshalos’ chto-to neulovimoje, vechnoje. All night roared sea, and in roaring this hear-REFL something(TO) elusive, eternal ‘The sea roared all night, and one could hear something elusive and eternal in this roar’. b. Kazhdyj den’ ego navesh’al kto-nibud’ iz nas. Every day he-ACC visited some(NIBUD’) of us ‘Every day one of us visited him’. (examples from Seliverstova 1988) Instead of using the contrast in “indefiniteness”, Seliverstova supports an approach that compares -to and -nibud’ pronouns in terms of their “disjunctive properties”. Under this interpretation, -nibud’ pronouns are understood as “disjunctive” (by using these pronouns a speaker indicates that there is always a choice – “this object / person / situation or a different one”), while -to indefinites are “non-disjunctive” (the choice of the object / person / situation has already been made). 10 Very similar definitions have survived till modern textbooks of Russian as a foreign language: see, for instance, Russian Stage Two: Welcome Back! Third edition (2010). 24 The concept of “disjunctive pronouns” can only hold for non-homogenous sets of elements. A -nibud’ pronoun cannot be used if all elements of a set are identical; Seliverstova 11 (1988) illustrates this by comparing examples given in (15a) and (15b) below : (15) a. Ja xochu vypit’ (*kakogo-nibud’) moloka. I want to drink (*some-NIBUD’) milk ‘I want to drink (*some kind of) milk’. b. Ja xochu vypit’ (kakoj-nibud’) mineral’noj vody. I want to drink (some-NIBUD’) mineral water ‘I want to drink (some kind of) mineral water’. Both cases are good with bare nouns, but only (15b) allows for a -nibud’ indefinite because all milk is the same but there are different kinds of mineral water, and by using -nibud’ the speaker implies that he does not care which kind of mineral water he receives. Under the “disjunctive” interpretation, -nibud’ pronouns are defined as indefinites that can apply to more than one element of a set whose members differ from each other in some quality(-ies). However, it is also important that it is not the case that absolutely any member of the set will do; -nibud’ indefinites do not have an absolute interpretation (other pronouns, e.g. liuboj or kakoj ugodno, are used for this purpose). 11 Some speakers note that (15a) becomes better with a -nibud’ indefinite if the speaker knows that he has different sorts of milk available for him at the moment (a somewhat unusual but possible situation). However, the observation made by Seliverstova still holds for all cases of unambiguously non-homogenous contexts; for instances, a sentece like ‘I need some fresh air’ cannot be rendered in Russian with a -nibud’ indefinite. 25 According to Seliverstova (1988, p. 56), it is the “disjunctive” interpretation of -nibud’ indefinites that causes limitations on their usage; she describes three different types of contexts that can satisfy the requirements of “disjunctivity” (or free access to different elements of a set): - Contexts describing a situation which did not happen yet or may never happen, - Contexts where participants (objects/situations) change, - Contexts where participants (objects/situations) are unknown. Unlike -nibud’ indefinites, -to pronouns do not carry this requirement on available alternatives and thus, are not limited to particular contexts of usage. Seliverstova does not discuss koe- and -libo indefinites at all. Her analysis mainly deals with -nibud’ pronouns and why they (and not -to indefinites) are used in particular contexts. Her generalization about the “disjunctive” nature of -nibud’ indefinites seems quite attractive, however it is not clear how exactly -nibud’ pronouns need to be defined in order to compositionally “match” the contexts where they appear. It is clear that such an analysis, despite all well-developed details, again, more describes the data than provides an explanatory background. Aware of these and many other problems with analyzing Russian indefinites, Seliverstova (1988, p. 60) concludes her chapter on indefinite pronouns with the following: “Although particular cases of using different indefinite pronouns are described quite well, an existence and nature of a general feature underlying their distribution and opposition is still a question to be answered”. The absence of a good explanatory theory causes problems in teaching. Cuto (2004), for instance, reports significant problems with the acquisition of Russian indefinites. 26 It is not surprising then, that Russian indefinites become an object of linguistic investigation again and again. They have recently also received close attention from linguists working within the paradigm of formal semantics and pragmatics. 1.5 Formally speaking: a choice functional analysis of Russian indefinites One of the suggested ways of looking at Russian -to and -nibud’ indefinites is a choice functional analysis that, staring from late 1990s, is often used to account for specific interpretations of indefinites in different languages. Initially, such an analisis was used by Reinhart (1997) and Kratzer (1998) to account for specific readings of English indefinites (for instance, noun phrases with a, and some in their wide scope reading). They suggested that these indefinites are interpreted as choice functions (a function that selects a member from a set). Some indefinites (e.g. English a certain) can only have a choice function interpretation. The interpretation of a choice functional a / some man and the only interpretation of a certain man, then, looks like (16): (16) f(man) where the value of f is supplied by the context. Choice function provides a null pronominal element of type <,e> that applies to a DP as shown in (17) below: (17) f<,e> is a CF if P(f(P)) = 1 27 The choice function approach has been successfully used to account for different types of specific indefinites in different languages [see Reinhart (1997) and Kratzer (1998) for CF indefinites in English; Allonso-Ovalle & Menendez-Benito (2003) for analysis of Spanish algun; Kratzer and Shimoyana (2002) for German irgendein; Mattewson (1999) for indefinites in St’at’imcets (Lillooet Salish)]. Yanovich (2006) applies this approach towards both -to and -nibud’ indefinites. He proposes that -to indefinites are instances of “mere choice functions”. In this, they are opposed to -nibud’ indefinites, analyzed as obligatory Skolemized (parameterized) choice functions. The respective denotations for both types of indefinites are cited below: (18) a. 〚to〛<<τt>,τ> = λp<τt>.f(p), where f is a GCF b. 〚kto-to ‘who-TO’〛e = 〚to〛(〚kto〛) = (λp<τt>.f(p))( λx.human(x)) = f<,e>(λx.human(x) (from Yanovich, 2006, #16) (19) a. 〚nibud’〛<<τt>,<υτ>> = λp<τt>λxυ.[f(x,p)], where f is a GCF b. 〚kto-nibud’〛 <υe> = (λp<τt>λxυ.f<<τt>,<υτ>>(x,p))(λy.human(y)) = = λxυ.f(x,λy.human(y)) (from Yanovich, 2006, # 21) 28 According to Yanovich, it is the skolemization that causes the semantic contrast between -to and -nibud’ indefinites. The denotation of -nibud’ includes the reference to the implicit parameter available for parameterization or skolemization – binding from above. It is not clear whether or not Yanovich implies that his “mere choice functions” for -to indefinites can also involve skolemization and because of that can be perceived as having a narrow-scope, like -nibud’ indefinites. He is aware of different scope options for -to indefinites (for instance, he has examples of conditional sentences where -to indefinites can scope out of ifclause but can be also interpreted lower). However, he never discusses how the narrow scope interpretation is obtained. If the definition of a “mere choice function” implies optional parameterization, we should expect that a denotation for -to should look exactly like the denotation for -nibud’; however, then it is not clear what the difference between the two types of the indefinites is and why they are not, in fact, interchangeable, at least when the narrow-scope interpretation is involved. If the notion of a “mere choice function” does not imply optional parameterization (which is the case, say, for Reinhart (1997), and Winter (1997); see also Chierchia (2001) for comparison of the pro-parameterization and against-parameterization approaches towards choice functions) then it cannot predict any scope other than the widest one unless some extra assumptions are made; say, suggesting an existential closure or possible ambiguity of -to indefinites between choicefunctional and quantificational interpretation; the latter possibility is considered and rejected by Yanovich (2006). In its existing form, though, Yanovich’s analysis of -to indefinites does not account for their narrow-scope interpretation as, for instance, in (6) repeated in (20) below for convenience: 29 (20) On ochen’ obshitel’nyj chelovek, on (vsegda) priglashajet kakix-to studentov, oni vmeste chitajut kakije-to knigi. He very sociable person, he (always) invite-PRES some (TO) student-PL, they together read-PRES some (TO) book-PL. ‘He is a very sociable person, he (always) invites some (possibly different) students, and they read some (possibly different) books together’. The other problem with Yanovich’s analysis is that it overgenerates, predicting impossible scope options for -nibud’ indefinites. He claims that -nibud’ indefinites are instances of obligatory skolemized choice functions. The main claim of why a choice-functional interpretation is needed for -nibud’ indefinites is that it accounts for the cases of intermediate scope, which, according to Yanovich, is possible for -nibud’ pronouns (along with the local-scope). His example of the intermediatescope interpretation of a -nibud’ indefinite is given below: (21) Každyj mal’čik budet rad esli vstretit kogo-nibud’ iz svoix odnoklassnic. ‘Every boy will be glad if [he] will-meet whoNIBUD’ of his girl-classmates.’ * Ǝ >∀ > if (There is a girl that every boy will be happy to meet) OK ∀ > Ǝ > if (For every boy, there is a girl he will be happy to meet) OK ∀ > if > Ǝ (For every boy, if there is a girl he meets, he will be happy (from Yanovich, 2006, #19) 30 I do not share this judgment: for me the only interpretation this sentence gets is the last one (the narrow scope). However, since it is sometimes not easy to elicit semantic judgments, I have asked several native speakers of Russian currently living in Russia about speech situations that included sentences similar to the one cited above; I tried to see whether or not these sentences would be perceived as having an intermediate-scope interpretation; (22) below shows one of such scenarios (with the exact sentence used by Yanovich): (22) Situation: There are three boys: Vasja, Petja and Ivan. They go to school together. Tanja, Anja and Marina girls from their class. The boys went to a party. You hear the following sentence about them: Každyj mal’čik budet rad esli vstretit kogo-nibud’ iz svoix odnoklassnic. ‘Every boy will be glad if he meets any (NIBUD’) of his female classmates’. What situation or situations does this sentence describe? A. Anja is so nice! Every boy will be happy to meet her (but not Tanja or Marina). B. Vasja will be happy to meet Anja (but not Tanja or Marina). C. Vasja will be happy to meet Anja, Tanja or Marina. With several people and different sentences, not a single native speaker one chose A or B as possible scenarios, contrary to Yanovich’s analysis which predicts validity of the option (B) as well. 31 Geist (2008) has the same intuition about the scope properties of -nibud’ indefinites. Discussing koe-, -to and -nibud’ indefinites in Russian she defines their scope in the following way: - koe- indefinites: wide scope only - -to indefinites: wide / narrow scope - -nibud’ indefinites: narrow scope only Geist also uses a choice function approach to indefinites; she applies it to all the three types; koe- is a choice function obligatory parameterized by the speaker, -to is a parameterized choice function (the implicit parameter is free and can be bound by something in the context), and -nibud’ is a parameterized choice function locked in its local domain by an existential closure; she gives the denotations of the indefinites as in (23) below: (23) koe- -to -nibud’ P fx (P) P fx (P) P x fx (P) x=speaker She explaines these denotations in the following way: “The only difference between the lexical entries of specificity markers concerns the binding of the implicit argument x. The implicit argument of koe- must be bound by the speaker, yielding identifiability by the speaker and necessary wide scope. 32 The implicit argument of the non-specificity marker -nibud’ is existentially closed at the lexical level yielding non-identifiability of the referent by the speaker, narrow scope, and the lack of co-variation reading. We interpret the fact that for -nibud’ no specific anchor exists in the following way: indefinites accompanied by -nibud’ introduce a completely random referent. The referent of a -nibud’-indefinite is not identifiable. However, the lexical entry of nibud’ is still a simplification since the fact that -nibud’ needs licensing by some operators in the clause is not yet integrated in it.” Geist (2008, p.159) The first part of the cited passage claims that a specificity marker koe- is an instance of a parameterized choice function obligatorily bound by the speaker – this would account for both the semantics (‘known to the speaker’) and the scope properties (‘wide scope only’). The specificity of -to indefinites (also definted as specificity markers) has to be brought in by the context. The second part of the claim, however, is much less clear. If “indefinites accompanied by -nibud’ introduce a completely random referent” then why is an indefinite choice functional at all? What does a choice function, locked under the existential closure, choose? Also, as Geist herself acknowledges, nothing in this representation explains a peculiar distribution of -nibud’ items. And under the denotation of -nibud’ given above it is not clear what property (-ies) of the meaning of -nibud’ items causes them to appear only in a limited set of contexts. 33 The same question may be posited in a different way: what is the nature of a common property for all the contexts in which -nibud’ indefinites are licensed? Then if such a property is found we may want to ask what in the semantics of -nibud’ makes this indefinite compatible with these (but not with the other) contexts. Pereltsvaig (2008) claims that all the contexts allowing for -nibud’ are contexts allowing for co-variation (the indefinite varies with something else varying in the context). She illustrates her point with data like (24) below: (24) a. *Ego o chem-nibud’ sprosili He-ACC about something(NIBUD’) asked-3PS-PL He was asked about something b. Kazhdogo o chem-nibud’ sprosili Everyone-ACC about something(NIBUD’) asked-3PS-PL Everyone was asked about something. In (24b) what was asked varies from person to person, which makes the sentence grammatical with -nibud’; absence of a co-variation makes a -nibud’ indefinite unacceptable in (24a). According to Pereltsvaig, to be licensed -nibud’ needs to be able to choose from more than one individual. It is not compatible with a situation in which there are no alternatives. In other words, -nibud’ indefinites require an operator with respect to which they can take narrow scope. Pereltsvaig calls -nibud’ ‘a marker of co-variation’. She shows that -nibud’ (contrary to claims made by many other researchers) is systematically licensed in declarative sentences under one condition: the sentence has to have different events involved. Thus, contexts containing 34 expressions like every morning, constantly, every time, when, etc. always allow for acceptable usage of - nibud’ indefinites. Pereltsvaig does not discuss sentences with modals, questions, future tense sentences, conditionals etc. allowing for -nibud’ expressions as well. But the generalization seems to be extendable to them because all these cases allow for co-variation in some way. If this generalization is correct, the semantics of -nibud’ needs to be presented in a way that would account for these limitations on the context. Such an analysis of -nibud’ indefinites will be given in chapter 3. 1.6 Speaker identifiability and -koe indefinites Russian indefinites have been also considered from pragmatic point of view. Kagan (2006) for example, builds her analysis on the contrast of koe- and -to items in respect to speaker identifiability / non-identifiability. According to her, koe- items are inherently speaker identifiable and -to items are inherently non-identifiable for the speaker. This property imposes restrictions on the contexts where these items are used. Kagan provides a detailed analysis of the felicity conditions that the two indefinites impose on the context. The analysis, however, does not cover the other “dimensions” of the indefinites; for example, it leaves unexplained the cases of -to pronouns receiving a narrow-scope interpretation. Speaker identifiability accounts for the interpretation of (25a) but it cannot account for the plural interpretation of the indefinite in (25b): (25) a. Ja nashla koe-chto interesnoje v etoj knige. I found something(KOE) interesting in this book ‘I saw something interesting in this book (I know exactly what it is and I may tell you later but at this point I don’t what to specify anything)’. 35 b. Ja videla koe-kogo iz druzej. I saw some(KOE) of friends ‘I saw SEVERAL of my friends (more than one person, and I know which ones, but it is not important to list them)’. So, despite multiple existing approaches to different types of Russian indefinite pronouns, they are still puzzling in many respects. Some of their properties have been described in detail, some other still have not received a sufficient analysis. After two decades of extensive attention to the problem, the pessimistic conclusion made by Olga Seleverstova in 1988 is still true: “a nature of a general [semantic] feature underlying the distribution [of the indefinites]” is still a question to be answered. The goal of the following chapters is to try to answer this question part by part by looking at each item and connecting its semantic properties with its distribution and vice versa; then I will collect the parts together in order to see how these different elements are combined in a system. I will start by looking at -to indefinites, continue with -nibud’ and -libo pronouns, and finally consider koe- indefinites. 36 CHAPTER 2 –TO INDEFINITES 2.1 Semantics and distributional properties The most general interpretation of Russian -to pronouns is to define them as specific with respect to scope (they are usually perceived as wide scope indefinites). There are, however, cases where these indefinites seem to be interpreted as having a narrow scope. Before looking closer at semantic properties of -to items, let’s see when and where they are used. The distribution of Russian -to indefinites depends on various factors. In simple declarative sentences implying a one-time event, limitations on choice of -to may be caused by the choice of a tense or subject. In past tense sentences -to is a default choice. The only other possibility is koe- but it has a different meaning – it refers to something known to the speaker, it is more like a “secret / hidden definite”. Any other situation of an indefinite referent requires -to (if an indefinite is marked in the sentence at all: Russian does not have definite/indefinite articles so the only way of “marking” is by using demonstratives or indefinite pronouns respectively; otherwise the definiteness/indefiniteness has to be “filled out” based on the situation): (26) a. Kakoj-to student napisal stikhotvorenije. Some(TO) student write-PAST poem. ‘Some student wrote a poem’. b. Etot student napisal kakoje-to stikhotvorenije. 37 This student write-PAST some(TO) poem. ‘This student wrote some poem’. c. Ja napisal kakoje-to stikhotvorenije (no teper’ ne pomniu, kakoje). I write-PAST some(TO) poem (but now not remember-1P-SG which). ‘I wrote some poem (but I don’t remember which one now)’. Importantly, (26c) needs the don’t remember part pronounced or assumed – it is pragmatically odd to classify your own poem as something indefinite unless it somehow lost the definite status for you (the fact that it’s your poem presupposes its definiteness for you unless there are other factors involved). As we see, the choice of the subject does not matter for using -to indefinites in the past tense. This is not the case for the present and future tense usages, though, where the 1 st person subject blocks usage of -to: (27) a. Kakoj-to student pishet/napishet stikhotvorenije. Some (TO) student write-PRES/FUT poem. ‘Some student is writing/will write a poem’. b. Etot student pishet/napishet kakoje-to stikhotvorenije. This student write-PRES/FUT some (TO) poem. ‘This student is writing/will write some poem’. c. #Ja pishu/napishu kakoje-to stikhotvorenije . I write-PRES/FUT some (TO) poem. ‘I am writing/will write some poem’. 38 The sentences in (27) are only good if the indefinites in them are perceived as having a wide scope interpretation: there has to be a particular student for (27a) and a particular poem for (27b). It is usually more difficult to model situations facilitating wide scope readings for the future tense. That sometimes causes doubts in grammaticality of sentences like (27b) in the future tense and leads to the conclusion about unacceptability of -to indefinites in such contexts. However, if the right context is created, sentences like (27b) are judged as acceptable. An example of the relevant context/scenario is presented in (28): (28) Scenario: The Department of Linguistics is planning to have a party in a month from now. Some students want to write poems for the party devoted to different areas of Linguistics. “Student-A” will write about Syntax, “student-B" – about Morphology, “student-C” – about Semantics etc. All the topics for the poems are already chosen. Some other students are not good at poetry, so they will sing/dance/do charades etc. All the roles are already assigned. My friend asks me about what “student-B” is doing for the party. I know for sure, that he is doing some poetry, but I don’t remember whether it is going to be on Syntax, or Morphology, or Semantics, or Phonology. So, I say: “B” napishet kakoje-to stikhotvorenije “B” write-FUT some (TO) poem ‘”B” will write some poem’ In such a situation, a -to indefinite is the natural choice. The situation requires a wide scope indefinite, and -to readily provides the needed reading. 39 Obviously, though, when dealing with the future, we usually don’t have well-planned scenarios. And most of the time the things we are talking about are not “some specific” things (requiring wide scope indefinites). Due to this fact, future tense sentences with -nibud’ are much more common than future tense sentences with -to indefinites. This gives some researchers a reason to conclude that -to indefinites are in principle impossible in future contexts. The right part of this conclusion is that they are, in fact, impossible in all the situations dealing with unknown uncertain referents; which are most of the situations about the future. As for the problem with (27c) with the 1st person subject, it seems clear that it has the same nature as the requirement of the don’t remember context in (26c): it is pragmatically odd to say (27c) because if the speaker and the doer coincide, the object of writing has to be known to the speaker (and not indefinite any longer). Indefinites with -to can be used with (but not under the scope of) grammatical negation: the only possible usage in negative contexts is with the indefinite having a wide scope reading 12 as in (29) below: (29) a. Scenario 1: I checked out 5 books from the library. They are due tomorrow and I need to find them all. I only found four. I know that one is missing, but I don’t remember the title. I say: Ja ne nashla kakuju-to knigu. 12 Farkas (2002), discussing the syntactic behavior of English some, also notes that it can only be interpreted as being wide scope relative to negation. She illustrates the point with the following example: (i) Mary didn’t buy some apartment in San Francisco when she could have afforded it and now it’s too late. 40 I not find-PAST some (TO) book. ‘I didn’t find some book’. b. Scenario 2: I suspect that I might have some library books. I looked everywhere but didn’t find any. This is good, it means I don’t need to go there and return anything. I say: Ja ne nashla nikakkoj knigi. I not find-PAST no-which book-GEN. ‘I didn’t find any book’. Negation of a narrow-scope indefinite has to follow the principle of “double negation” or “negative concord/agreement”. This means, besides negation on the verb, we need to have a special – negative – form of the indefinite after it. So-called “negative pronouns” constitute a separate set of pronouns. They are constructed by combining “question words” (who, what, when, where etc.) and a negative prefix ni-. So, if taken out of the context English I didn’t say anything to anyone is translated into Russian, it most likely will receive the “default” form of (30): (30) Ja nikomu nichego ne skazal. I no-who no-what not say-PAST. Literally: ‘I didn’t say no one nothing’. However, such a translation will be completely wrong if we deal with the situation below: 41 (31) My friend complains that his family thinks he is a liar. I asked him why. And he says they are exaggerating. It only happened once, a long time ago, when he failed to tell something important to someone, and since then they don’t trust him anymore even though no one really remembers who that person was and what wasn’t said. To present the situation above correctly we will need to use wide scope indefinites with -to as in (32): (32) Ja komu-to chto-to ne skazal. I someone (TO) something (TO) not say-PAST. ‘I did not say something to someone’. One more limitation on the distribution of indefinites with -to (that can never be violated) comes from imperative sentences: (33) * Daj mne kakuju-to knigu. Give me-DAT some (TO) book. ‘Give me some book’. There are two “correct” ways to make a request (30): it is either to define the book (this, the blue, Shakespeare etc.) or, if it is not of importance what book you will get, to ask for any book using a -nibud’ indefinite: (34) Daj mne etu/siniuju/kakuju-nibud’ knigu. 42 Give me-DAT this/blue/some (NIBUD’) book. ‘Give me this/blue/any book’. The sentence in (33) is not grammatically wrong; it simply does not make sense: if you want a certain book, tell me which book it is; if you want any book, then you do not say kakujuto. Despite the common agreement that questions are the realm of only -nibud’ expressions, -to indefinites can be also used in questions. However, the meaning of a -to question is different from the “normal” -nibud’ question and requires the “right” context to be captured. Compare contexts in (35) and (36): (35) Ty nashel chto-to interesnoje? You find-PAST something (TO) interesting? Did you find something interesting? (in a bookstore, when your friend is holding something in his hands, and his face looks incredibly happy) (36) Ty nashel chto-nibud’ interesnoje? You find-PAST something (NIBUD’) interesting? Did you find anything interesting? (when you are calling your friend to ask about his visit to a new bookstore) Some questions containing a presupposition of the ‘specific’ event directly in them (not in the broader context) are in fact impossible with -nibud’: (37) Kogda ja sdelal chto-to / *chto-nibud’ ne tak? When I do-PAST something (TO/*NIBUD’) not-so? 43 ‘When did I do something wrong’? The presence of both kogda (‘when’) and the past tense in the sentence creates the presupposition that this wrong thing, whatever it was, actually happened (even though the speaker doesn’t know what this thing was), and that it was a specific (even though unknown) thing. This makes using -nibud’ impossible. However, usages like (35) and (37) are rare. Most of the times people ask questions when they do not have any information about the thing they are asking or do not care about the information they already possess because they are asking about the “other side” of the thing in question. Thus, the “standard” question situation (if it involves indefinite) needs to deal with an unambiguously narrow scope reading of it, and thus does not allow -to in Russian: (38) Ty kupila *chto-to / chto-nibud’ na uzhin? You buy-PAST something (*TO/NIBUD’) for dinner? ‘Did you buy anything for dinner?’ The sentences originally allowing “-to only” (e.g. in the past tense) as in (36a) become flexible with the choice between -to and -nibud’ if embedded under verbs like “want”, “wish” etc. as in (39b). (39) a. Etot student napisal kakoje-to / *kakoje-nibud’ stikhotvorenije. This student write-PAST some (TO/NIBUD’) poem. ‘This student wrote some poem’. b. Ja khotel, chtoby etot student napisal kakoje-to/kakoje-nibud’ stikhotvorenije. 44 I wanted that this student write-PAST some (TO/NIBUD’) poem. ‘I wanted this student to write some poem’. (i) -to: I wanted him to write the poem he was assigned (and I did know what poem it was supposed to be when I asked him about it, but now I don’t remember) (ii) -nibud’: I just wanted him to write a poem – whatever he wants to write about. As (i) and (ii) above show, the new sentence in (39b) has different meanings depending on the choice of the indefinite. Imperatives that are normally acceptable only with -nibud’ indefinites, as in (40a), allow both -to and -nibud’ versions when embedded (indirect speech), as in (40b) and (40c): (40) a. Daj mne kakuju-nibud’ / #kakuju-to knigu. Give me-DAT some (NIBUD’/#TO) book. ‘Give me some book’. b. On poprosil menia dat’ jemu kakuju-nibud’ knigu. He ask-PAST me give-INF him-DAT some (NIBUD’) book. ‘He asked me to give him a book’ (any book, doesn’t matter which). c. On poprosil menia dat’ jemu kakuju-to knigu. He ask-PAST me give-INF him-DAT some (TO) book. ‘He asked me to give him some book’ (and he told me which one, but now I don’t remember). 45 Embedded questions can (very rarely, as well as matrix questions) have -to indefinites if they refer to something specific: (41) Ja sprosil, kogda ja sdelal chto-to / *chto-nibud’ ne tak. I ask-PAST when I do-PAST something (TO/*NIBUD’) not-so. ‘I asked when I did something wrong’. Very often, -to indefinites appear in conditionals. And surprisingly, in many cases they are judged as having roughly the same meaning as -nibud’ ones (and can be exchanged for nibud’ conditionals): (42) On vse jesh’e mozhet byt’ obvinen, jesli kto-to predostavit ser’jeznyje dokazatel’stva. He still may be-INF found guilty if someone (TO) provide serious evidence-PL. ‘He might be still found guilty if someone provides serious evidence’. If the sentence (42) is given out of context, the natural reading kto-to (‘someone’) receives is “just anyone”. And this is exactly the reading that the -nibud’ version of the same sentence has. The wide scope reading is also possible but requires an elaborated scenario (or broader context), something like: I have heard that someone (I don’t remember the name now) has some evidence against the defendant and the rumor is that this evidence is really strong. So, if this person brings up this evidence, it will be bad for the defendant. Evidence received from other people, though, will not matter much. The “default” reading of the indefinite in (43) below is also non-specific: (43) Sotrudniki izdatel’stva poluchat premiju, jesli kakaja-to ikh kniga 46 budet polnost’ju rasprodana v etom mesiatse. Personnel-PL publishing house-GEN get-FUT bonus if some (TO) of their books will completely sold-out-PASSIVE in this month. ‘The publishing house personnel will get a bonus if some / one of their books is completely sold out this month’. Out of context, (43) is likely to get the interpretation that when any book is sold out it will result in the bonus. However, it is also likely to be followed by the question “What book?”, because the sentence is ambiguous. Depending on the answer (“I think it’s the one by N” or “Just any”), the correct interpretation will be found. However, for many if sentences the wide reading of -to indefinites is much harder to obtain than the narrow one. This seems unexpected, because both examples of the future tense usages and questions considered above behaved in exactly the opposite way: when a -to indefinite is used in a future tense sentence or in a question it “signals” the wide scope interpretation (otherwise the sentence is ungrammatical). The cases of narrow scope readings in future tense sentences and questions require -nibud’ indefinites. Conditionals, however, allow for both readings and sometimes it is more difficult to obtain a wide scope reading for them. Two scope options for -to indefinites can be also obtained for so-called “Dean examples” (44) 13 (some speakers disagree on the narrow scope reading, though): Kazhdyj prepodavatel’ slyshal, chto kogo-to iz moix studentov vsegda 13 Sentences like (44) are traditionally used in discussing ambiguous properties of English indefinite NPs (a / some student). First used by Janet Fodor and Ivan Sag (1982), the English sentence ‘Each teacher overheard the rumor that a student of mine had been called before the dean’ became a popular illustration for the proposal that, in English, “indefinite NPs are ambiguous between quantificational and referential readings”. 47 vyzyvajut k dekanu. Every teacher hear-PAST that some (TO) from my students always call-PRES-3rdPL(impers.) to dean. ‘Every teacher heard that some (one) of my students is always called before the dean’. (i) (?) narrow scope (‘may be a different student every time’); not all speakers agree; (ii) wide scope (‘some particular, the same student’). The way to fix the sentence for those who judge it as unacceptable in a “narrow scope” context (when a non-specific reading is clearly supported) is to use the -nibud’ indefinite. Then the sentence is judged unambiguously grammatical (but of course only has a narrow scope interpretation). Important information about the scope options for -to indefinites can be obtained from 14 testing so-called “Professor examples” (, sentences with (potentially possible) three scope options. In English, “Professor examples” with a and some indefinites present the third scope possibility (intermediate scope), when an NP in question is neither quantificational nor referential. In Russian, these sentences with -to indefinites also give rise to an intermediate scope reading (along with the other two options) as illustrated in (45): 14 Kratzer (1998) uses sentences like (45) to illustrate that English some indefinites are three-way ambiguous (wide scope, narrow scope and so-called intermediate scope; where students are particular students varying by professors. Examples like this become an important argument to support her parameterized choice function analysis of English some. 48 (45) Kazhdyj professor poxvalil kazhdogo studenta, kotoryj prochital kakuju-to knigu, kotoruju on porekomendoval Every professor praise-PAST every student which read-PAST some (TO) book which he recommended ‘Every professor praised every student who read some book that he recommended’ (i) (?) narrow scope (‘any book’) (some speakers disagree) (ii) intermediate scope (‘there is some particular book for every particular professor) (iii) wide scope (‘some particular book, but the speaker doesn’t care/know which one’) Again, those speakers who cannot get the narrow scope reading for (42) prefer the version with -nibud’ indefinite instead. Having seen all these cases of distribution and scope options for -to indefinites in Russian we are now in a position to proceed with the analysis that would allow us to account for the semantic and syntactic properties illustrated above. 2.2 Choice Functional Analysis – the first approximation The data above shows that, on one hand, the ‘default’ interpretation of -to indefinites is a ‘specific’ one (in terms of scope). However, there are some cases when they, like -nibud’ indefinites, seem to be perceived as having a narrow scope. So, there are two properties to be accounted for: (a) to indefinites are specific, and (b) sometimes they seem non-specific. Let’s start with the more common property – an ability to obtain a wide scope (specific) interpretation. 49 A choice functional analysis seems to be the most natural choice of a tool to use in investigating specific indefinites. It has been successfully used for wide scope indefinites in various languages. Kratzer (1998), for instance, suggests that English noun phrases with a, and some in their wide scope reading and unambiguously specific noun phrases with a certain are interpreted as choice functions (a function that selects a member from a set), as repeated in (46) below: (46) f(man) where the value of f is supplied by the context. A choice functional analisis was used by Yanovich (2006) to account for Russian sets of indefinites (both -to and -nibud’ pronouns). The core part of his analysis is devoted to -nibud’ indefinites, and he claims that they are special type of CF with an obligatory Skolemized argument (I will return to this claim in the next chapter and provide evidence against such an approach to -nibud’ indefinites). The two pages that he devotes to the brief consideration of -to indefinites have interesting observations. Yanovich considers six examples of if sentences with embedded -to indefinites with different core (pronominal) elements (who, what, which, where, to-where etc.). He claims that they can all get wide and narrow scopes, and thus (a) have to be considered as instances of choice functional indefinites (a conclusion I support based on a range of arguments); and (b) since -to can be attached to different types of pronominal elements having different types (nouns, adjectives, adverbs), the denotation of the CF has to be changed (broadened) to include all the possibilities. Thus, he suggests to use the denotation of the general format of f<,τ> instead of the traditionally accepted f<,e> type. 50 Though, in many cases, I cannot perceive the readings he intends for his examples to prove the necessity of the change in denotation, the point still deserves to be discussed. Here is one of the example sentences he gives: (47) Petia budet schastliv, jesli kakaja-to devushka pridet. Petia will be happy if some (TO) girl come-FUT. ‘Petia will be happy if some girl comes. The two readings I have for the sentence in (47) are the following: (48) a. Petia will be happy if there is a girl that comes (narrow scope). Scenario: He will be happy if any girl comes (Tania, Masha, Vera, etc.). b. There is a girl such that if she comes Petia will be happy (wide scope). Scenario: He will be happy only if Tania comes. For Yanovich there is a different reading there, which is cited in (49): (49) There is a property p such that Petia will be happy if a girl y such that p(y)=1 comes. This reading would require the following potential scenario (I am not sure that this is the scenario Yanovich meant but that is what follows from the reading): Petia mentioned that he likes blond girls (but not tall/French/loud etc.) girls. And he was hoping to see any girl with this property soon. I don’t remember what property he mentioned he liked (blond/tall/French/loud 51 or other) but I remember there was a particular one. So, to report Petia’s situation I utter the sentence given in (47). For me both the scenario and the ‘property’ reading above are not plausible. The same happens with the intended ‘manner’ reading of (50): (50) Petia budet schastliv, jesli kak-to smozhet sdat’ ekzamen. Petia will be happy if somehow(TO) can-FUT pass-INF exam. ‘Petia will be happy if somehow (he) will be able to pass the exam. Intended reading: There is a manner p such that Petia will be happy if he manages to p (pass the exam). Potential scenario: Petia mentioned that he has a special way of passing exams (say, hypnotize the professor and make him believe that Petia deserves a good grade even if he does very poorly) and he was hoping to use this manner successfully during the coming exam. I don’t remember what this manner was (may be hypnotizing the professor, studying hard, preparing a cheat-sheet, or something else), so I report Petia’s situation with the sentence above. I do not have this reading for (50) either. This could be a matter of individual perception, because Yanovich himself admits that he cannot get an analogous interpretation for kogda-to (‘some time’). He explains it by a strong conventional bias to a different reading. I did not have a chance to test these sentences on a sufficient number of Russian speakers to come to any conclusions about the judgments for the aforementioned readings. 52 The possibility of readings like (49) and (50) leads Yanovich to the idea that the choice function in Russian must be able to apply to constituents of different semantic types [e.g., the manner one is of type p<> (Adverb)]. Thus, he needs to adjust the definition of CF for Russian to a more general one. Although I am not very comfortable with the data provided by Yanovich, I do believe the intuition behind the idea that not only NPs (or DPs) can be choice-functional indefinites (in which case we should probably be able to find some evidence of this fact in languages other than Russian). However, from my point of view, such a claim definitely needs some stronger support (at least from Russian, and, possibly, from other languages) to be accepted with confidence. It actually leads to a different theoretical question: if is is not the case that CF can apply crosscategorically, why is it so? What properties of CF would require the format f<,τ>? Due to the lack of reliable data, it doesn’t seem plausible to answer any of the questions above so far. There are two different versions of the choice function approach. Reinhart (1997) proposes that choice function has to be existentially closed at some level, and it is the level where the existential closure applies that explains different scope options for the CF indefinites. The widest scope, for instance, can be obtained by applying existential closure at the level of the matrix clause. Then, the specific reading of some man in (51a) can be represented as (51b): (51) a. Some man walked in b. Ǝf [walk in (f(man))] 53 This version of the CF approach makes the right predictions for some Russian data. For example, the wide-scope reading of (52a) can be accounted for and represented as (52b): (52) a. Kakoj-to student chital gazetu na uroke. Some(TO) student read-PAST newspaper in class. ‘Some (particular) student read a newspaper in class’. b. Ǝf [CH(f) & Ǝx: x is a class [f(student) read newspaper during x] However, it is not clear how this approach can explain the contrast in scope options between (52) and (53): (53) a. Kakoj-to student vsegda chitajet gazetu na uroke. Some (TO) student always read-PRES newspaper in class. ‘Some student always reads a newspaper in class’. (i) wide scope: ‘some unknown but the same student every time’; (ii) narrow scope reading ‘may be a different student every time’ The wide-scope reading is the only reading easily available for (52); meanwhile, (53) allows for both wide and narrow scopes of the indefinites. If existential closure can apply anywhere (locally and at the clause level), then both meanings should be equally available for both (52) and (53). If it has to apply at the clause level, then only wide-scope reading has to be available for both sentences. The problem may be resolved by stipulating that Russian -to indefinites, like English a / some indefinites are ambiguous between a quantificational and a CF interpretation with the CF always giving a wide-scope reading, and quantificational items having a local scope. 54 This is definitely a possibility, but in this case we would expect to have some independent evidence showing that -to indefinites in Russian can receive a quantificational interpretation in principle. I am going to look for such evidence below. 2.3 Can -to indefinites be quantificational? Is it possible that -to indefinites have a genuine non-specific (quantificational) interpretation? A relevant case would be a context that only allows for a narrow scope reading. Let’s go back to the library books scenarios, and briefly go through the whole situation again in (54) below: (54) a. Scenario 1: I’m missing 1 book from the library (I found the other four): Ja ne nashla kakuju-to knigu. I not find-PAST some (TO) book. ‘I didn’t find some book’. b. Scenario 2: I don’t have a single book from the library: Ja ne nashla nikakoj knigi. I not find-PAST no-which book-GEN. ‘I didn’t find any book’. The second scenario clearly requires a narrow scope indefinite. And this scenario does not have a -to indefinite, it gets negative agreement instead. Importantly, it is not the case that -to is impossible with negation at all; it is fine in a wide scope context of the first scenario. Let’s look one more time at the direct and indirect requests repeated in (55) below: 55 (55) a. Daj mne kakuju-nibud’ / #kakuju-to knigu. Give me-DAT some (NIBUD’/#TO) book. ‘Give me some book’. b. On poprosil menia dat’ jemu kakuju-nibud’ knigu. He ask-PAST me give-INF him-DAT some (NIBUD’) book. He asked me to give him a book (any book, doesn’t matter which). c. On poprosil menia dat’ jemu kakuju-to knigu. He ask-PAST me give-INF him-DAT some (TO) book. He asked me to give him some book (and he told me which one, but now I don’t remember). The direct request with -to (44a) is not a valid option whatsoever. You will not get any particular book because you do not say what book you are asking for, and you will not get just any book, because you didn’t ask for just any book. The -to version of (55a) simply doesn’t have this narrow scope reading. Is -to possible in requests at all? Yes, in embedded ones (55c). The only condition on such a request is that there has to be a certain book you were asked for initially, thus it does not allow for a narrow-scope indefinite. One other piece of evidence that -to indefinites cannot be interpreted quantificationally comes from the fact that all explanatory dictionaries of Russian never use -to indefinites in their definitions (different dictionaries prefer either -nibud’ or -libo indefinites, but -to is impossible). The nature of an explanatory dictionary definition is such that it needs to describe “a general case” of what the things are; they cannot refer to particular instances of individuals. Saying ‘A singer is a person who can sing something’, a dictionary does not want to refer to a specific 56 “something”, it wants to refer to any instance of a thing that can be sang. If it were possible to get a truly quantificational interpretation out of -to indefinites they would not be incompatible with dictionary definitions. How about the “sociable person” example? I will repeat it here as (56) for convenience: (56) On ochen’ obshitel’nyj chelovek, on (vsegda) priglashajet kakix-to studentov, oni vmeste chitajut kakije-to knigi. He very sociable person, he (always) invite-PRES some (TO) student-PL, they together read-PRES some (TO) book-PL. ‘He is a very sociable person, he (always) invites some students, and they read some books together’. When I described this example before, I said that the students and the books in it can, in principle, be (the same) specific students and (the same) specific books, but the most natural reading of it is ‘some students/books, possibly different all the time’ which seems to be what we are looking for – the (genuine) narrow scope reading! Well, let’s see what happens with this reading if the sentence is slightly changed. We will cut off the “general information” – he is a very sociable person part, and “move” the sentence into the given time of yesterday: (57) Vchera on priglasil kakix-to studentov, oni vmeste chitali kakije-to knigi. Yesterday he invite-PAST some (TO) student-PL, they together read-PAST some (TO) book-PL. ‘Yesterday, he invited some students, they read some books together’. 57 Now the only possible reading is the wide-scope one: the students and the books have to be specific. The same contrast occurs with a “newspaper lover” in (52) and (53); the “always” situation seems to have a wider range of scope options than the past tense situation. However, some speakers hesitate with the narrow scope reading for (53). For them the present tense by itself does not always facilitate (or even support) a non-specific interpretation; they need a broader context to support this reading. The judgments would shift towards non-specific interpretation if the context of (53) is enlarged as in (58) (58) Studenty liubiat chitat’ gazetu na lektsijakh; kakoj-to student vsegda chitajet gazetu u menia na uroke. Students love read-INF newspapers on lectures; some(TO) student always reads newspaper on my class ‘Students love reading newspapers during lectures; some student always reads a newspaper during my classes’. The data described so far suggests that -to indefinites in Russian can never be used in situations requiring a genuine narrow-scope indefinite. The interpretation that seems ‘nonspecific’ is harder to get and usually requires contextual cues. Absence of the cues leads to the ‘default’ wide-scope interpretation. If -to indefinites cannot have a (genuine) quantificational interpretation, we have to return to the same problem again: how can we explain the contrast in scope options between (52) and (53)? Why the non-specific reading is only available sometimes? 58 A similar problem occurs with availability of all theoretically possible scope options in the “Dean” and “Professor” sentences repeated below: (59) Kazhdyj prepodavatel’ slyshal, chto kogo-to iz moix studentov vsegda vyzyvajut k dekanu. Every teacher hear-PAST that some (TO) from my students always call-PRES-3rdPL(impers.) to dean ‘Every teacher heard that some (one) of my students is always called before the dean’. (i) (?) narrow scope (‘may be a different student every time’) – not all speakers agree; (ii) wide scope (‘some particular, the same student’). Many speakers do not have the narrow scope reading for (59), they judge the sentence as ungrammatical and suggest to fix it by using the -nibud’ indefinite. Narrow scope reading of (60) is also questionable for many speakers. (60) Kazhdyj professor poxvalil kazhdogo studenta, kotoryj prochital kakuju-to knigu, kotoruju on porekomendoval Every professor praise-PAST every student which read-PAST some (TO) book which he recommended ‘Every professor praised every student who read some book that he recommended’ (i) (?) narrow scope (‘any book’) (some speakers disagree) (ii) intermediate scope (‘there is some particular book for every particular professor) (iii) wide scope (‘some particular book, but the speaker doesn’t care/know which one’) 59 It is not clear how Reinhart-style analysis can account for such ‘selectivity’. Moreover, the “Professor” sentences bring up one more problem for this analysis. In the same way, as in English, ‘intermediate’ reading can only be perceived in some (but not all!) cases. The change of a pronoun (from 3rd person to 1st person singular) makes the intermediate scope reading impossible: (61) Kazhdyj professor poxvalil kazhdogo studenta, kotoryj prochital kakuju-to knigu, kotoruju ja porekomendoval Every professor praise-PAST every student which read-PAST some (TO) book which I recommended ‘Every professor praised every student who read some book that I recommended’ (i) (?) narrow scope (‘any book’) (some speakers disagree) (ii) wide scope (‘some particular book, but the speaker doesn’t care/know which one’) Again, under the Reinhart-style approach it is not clear why the intermediate scope reading is not available in (50). On the other hand, approach taken by Kratzer (1998) proves capable to account for this seemingly chaotic distribution in scope. 2.4. Parameterized choice function for Russian -to indefinites For Kratzer (1998), choice function is parameterized (Skolemized CF). It means that it has an implicit argument (parameter) that can be bound by a quantifier. First of all, such a hidden parameter is needed to explain the behavior of a certain indefinites in sentences like (62) below: 60 (62) Every husband had forgotten a certain date – his wife’s birthday. The sentence has a reading as in (63): (63) x (husband x))  had forgotten (x, fx (date))) (Kratzer, 1998) The determiner a certain is interpreted as a function variable f. Its implicit argument appears as a subscripted variable. Possible values for f are functions mapping individuals into CFs [in (62) mapping every husband into a choice set defined for one argument, the set of all dates, and picks that man’s wife’s birthday from that set]. So, in (62), the implicit argument of a certain is anaphorically related to a quantifier phrase (each husband) and is interpreted as a bound variable, creating the impression of a narrow scope reading. However, the only possible reading for the indefinite in (64) is a referential one: (64) Is Richard dating a certain woman? Since (64) does not contain any quantifier to bound the implicit argument of a certain, the value of the argument is assigned by the speaker. The (rudimentary) CF, then, picks out a 61 woman that the speaker has in mind, creating the impression of the widest scope for a certain woman. In the same way, in the ‘Professor examples’, the CF picking out a book can be parameterized by the speaker (‘the book that the speaker chose’) if the sentence does not contain the 3rd person singular pronoun (he/she). If such a pronoun is used, the CF is parameterized to different professors (any professor can be referred to as he or she), and thus a/some book receives “intermediate” scope reading. The notion of a parameterized choice function is crucial for Russian -to indefinites. First of all, it is obviously needed to account for the contrast between (60) and (61) in the same way as it explains the difference between the corresponding examples in English. However, the necessity of parameterized CF is not limited to the “Professor” examples. With this notion, we can straightforwardly explain otherwise chaotic scope possibilities in the ‘newspaper lover’ examples. Now, to represent the meaning of (65a) we will use (65c) instead of (65b) (54) a. Kakoj-to student chital gazetu na uroke. Some (TO) student read-PAST newspaper in class. ‘Some (particular) student read a newspaper in class’. b. Ǝf [CH(f) & Ǝx: x is a class [f(student) read newspaper during x]] (Reinhart-style CF) c. CH(fx) & Ǝy: y is a class [fx(student) read newspaper during y] (Kratzer-style CF) 62 For the sentence above both representations describe the meaning correctly. And since the indefinite in it is never quantificational, the only possible meaning of it – the ‘specific’ one – is accounted for by both (65b) and (65c). The scope options of the “always” instance of the same sentence (66a), however, can only be explained by the Kratzer-style approach: (66) a. Kakoj-to student (vsegda) chitajet gazetu (na uroke) Some student (always) read-PRES a newspaper (in class)’ ‘Some student (always) reads a newspaper (in class)’ b. Ǝf [CH(f) & x: x is a class [f(student) reads newspaper during x]] (Reinhart-style CF) c. CH(fx) & y: y is a class [fx(student) reads newspaper during y] f is a free variable, x is supplied by the context (Kratzer-style CF) The interpretation in (66b), as before, accounts for the wide scope possibility, but not for the narrow one (-to indefinites are not quantificational!). The interpretation in (66c), however, accounts for both readings: what seems to be the narrow scope interpretation is an instance of the specific reading in different choice sets: for every class (always implies that more than one class is available in the discourse), a specific (but not necessarily the same every time) student reads a newspaper. 63 It is very similar to the situation with ‘intermediate’ readings in the “Professor” sentences. The choice of the book can be parameterized by the speaker (specific reading) or by different professors (intermediate reading) depending on the contextual cues (I pronoun or he/she pronoun). In the same way, in the present tense sentences, the choice of the student can depend on the speaker (specific reading) or parameterized by different times (a specific student for time a, time b, time c etc.) if there are contextual cues for it (words like always/usually/often). In this case the perceived reading is very close to a narrow scope one. Now we are able to account for the seemingly narrow-scope interpretation of the indefinite in the ‘sociable person’ example, repeated below: (67) On ochen’ obshitel’nyj chelovek, on (vsegda) priglashajet kakix-to studentov, oni vmeste chitajut kakije-to knigi. He very sociable person, he (always) invite-PRES some (TO) student-PL, they together read-PRES some (TO) book-PL. ‘He is a very sociable person, he (always) invites some students, and they read some books together. The sentence is interpreted as (68), where ‘students’ and ‘books’ are specific students and specific books for “time1”, specific students and specific books for “time2” etc. (68) a. … on priglashajet kakix-to studentov … ‘… he invites some students… CF(f) ^ t [he invites ft(students)] b. … oni chitajut kakie-to knigi 64 ‘… they read some books’ CF(g) ^ t [they read gt(books)] The interpretation of the indefinite remains specific, but multiple occurrences of different specific (for given times) individuals become possible, facilitating the seemingly narrow scope reading. This (and not the ‘default’ wide-scope) interpretation is the most salient for (67) because the other choice (“the same people come and read the same books all the time”) is somewhat odd (how long could one read the same books?). Such an analysis predicts that any limitation on times (yesterday, at 5 pm, tonight, perfective aspect on the verb etc.) suppresses the ‘non-specific’ interpretation. And this is exactly the case: we have already seen that the “yesterday” version of the same sentence (repeated below) can only mean that some specific students were invited and some specific books were read. (69) Vchera on priglasil kakix-to studentov, oni vmeste chitali kakije-to knigi. Yesterday he invite-PAST some (TO) student-PL, they together read-PAST some (TO) book-PL. ‘Yesterday, he invited some students, they read some books together’. Since the implicit time parameter can only be satisfied with one given time t (yesterday) the choice function can only choose ones, and so only one specific set of students and one specific set of books can be selected. 65 Unlike English a /some indefinites, the narrow scope interpretation (or, more precisely, quasi-narrow one) for -to indefinites does not come for free. It has to be derived and requires special (contextual) conditions for that. The “natural” reading of -to indefinites is a specific one. If we take -to indefinites of being of type <, e>, then (70) can be represented in the following way: (70) Kakoj-to student chitajet Some (TO) student read-PRES ‘Some student is reading’ t e <,e> kakoj-to student > chitajet Derivationally, we can represent the sentence in (70) as follows: (71) 〚kakoj-to〛,e>> = λp .fx(p) and f is a CF 〚kakoj-to student〛e = 〚kakoj-to〛(〚student]]) = 〚kakoj-to〛 (λx.student(x)) = = f <,e> (λx.student(x)) = fx(student) 〚chitajet〛> = λx. x reads 66 〚Kakoj-to student chitajet〛 = 〚chitajet〛(〚kakoj-to student〛) = = fx(student) reads The data considered so far suggests that -to pronouns in Russian are choice-functional (or “specific”) by nature. Even when, in some contexts, they can be perceived as having (almost) narrow-scope interpretation, they are still instances of “relatively specific” interpretation. This also explains the speakers’ intuition about the inherent difference between the ‘quasi-narrow-scope interpretation’ of -to and the ‘truly-narrow-scope interpretation’ of -nibud’ indefinites. The speakers who disagree with ‘non-specific’ reading of (55), as well as with nonspecific reading of the “Dean” and “Professor” examples, usually give the following reasoning: “if I want to say that it is not a certain student I will say kakoj-nibud’ ”. The type of (perceived) specificity depends on the context. The “purely specific” reading is perceived when the choice set is limited to one. This reading is the most common one for -to indefinites. It is not just a grammatical category of tense (past, present or future) that puts limitations on the number of available sets. It is not the case that the present tense (as a grammatical category) in the sentence automatically makes several sets available for the choice function to choose from. In fact, it is easy to find instances of the present tense sentences with only one choice set available (72): (72) Uzhe konets uroka, vse ustali; kakoj-to student chitajet gazetu. Already end class-GEN, all tired; some (TO) student read-PRES newspaper. ‘It’s the end of the class already; everyone is tired; some (specific) student is reading a newspaper’ (no narrow scope reading). 67 It is the English translation that has the present progressive form of the verb; the actual Russian sentence has the same tense form as the one in (66). Russian has only one form for the present tense. What facilitated “general” reading in (66) was the word “always”. What facilitates the progressive reading of the tense in (72) is the reference to the time of speech: the end of this (current) class. In actual speech, the words “always” or “it is the end of the class” are not necessarily present. The situation for both “always” and “now” contexts can be described by exactly the same sentence: (73) Kakoj-to student chitajet gazetu na uroke. Some(TO) student read-PRES newspaper in class. ‘Some student reads /is reading a newspaper in class’. However, no ambiguity usually occurs in actual communication: the whole scenario in which the sentence is uttered helps in choosing the “right” interpretation. The perceived “ambiguity” of -to (it seems to be able to mean anything from ‘this particular one’ to (almost) ‘anyone’) is illusive; and the fact that there is a number of contexts disallowing -to indefinites demonstrate that the interpretation of -to is not that flexible, CF can only choose from available sets and always chooses the predetermined “candidate”. The “relatively specific” reading is obtained when there is more than one set from which a choice function can pick the candidate. The context usually contains some cues to indicate that the number of choice sets is greater than one. In the “Professor” examples in both Russian and English, the personal pronoun he/she refers to different professors and makes the ‘relatively specific’ reading available. 68 Since always (usually, often etc.) implies multiple times (e.g. of a class: yesterday’s class, today’s class, tomorrow’s class etc.), it might be one specific student for yesterday, and then another student for today, and then a different student for tomorrow – all specific students but all different. Thus, we may perceive the (actual) intermediate scope as being very close to a narrow scope reading, but it is still not the same meaning as for the genuine narrow scope of -nibud’ (therefore, the comments of the speakers: “if I want to say that it is not a certain student I will say kakoj-nibud’ ”). The more cues we receive, the more comfortable we feel with the “quasinarrow scope” reading. For some contexts it might be the first reading that comes to mind. In this respect, we may make some stipulations about the nature of if sentences. In conditionals, indefinites with -to usually readily get the reading close to the narrow scope; it is sometimes the wide scope reading that needs a broader context to be supported. This seems to mean that conditional modality somehow points to the fact that an indefinite number of possible sets can be available. Thus, in Russian, -to pronouns are not ambiguous (like some in English) between specific and non-specific (or between choice-functional and quantificational) readings. They are specific regardless of the context but they are perceived as having different degree/level/type of specificity depending on the context in which they are used (i.e. on what the choice function is choosing from, or how many choice sets are available). If the context does not limit the number of choice sets (e.g. “always” contexts, conditionals etc.), -to pronoun, in many cases, can be understood as having a quasi-narrow interpretation (because it is the only possible reading, then). 69 2.5 Conclusions In this chapter, Russian -to indefinites are analyzed as being context determined choicefunctional items. Unlike English some and a indefinites, Russian -to indefinites are not ambiguous between choice-functional and quantificational interpretations. In Russian, -to indefinites do not have a genuine non-specific (quantificational) interpretation, they can only be interpreted as a choice function. In speech, they might be perceived as having a wide, intermediate, or (close to) narrow scope. The Kratzer-style parameterized choice function approach appears to be able to account for all these scope options. If the CF is parameterized by the speaker, the “purely specific” reading is obtained. If the CF can be parameterized by different people /times etc. (in other words when there is more than one set for the CF to choose from), -to indefinites are perceived as having ‘intermediate’, or “relatively specific” scope. The more sets available for the CF to evaluate, the “narrower” the perceived interpretation. If the situation does not have an obvious limitation for the number of the sets, the quasi-narrow interpretation is obtained. Despite its seeming similarity with a genuine narrow scope reading of Russian nibud’ items, the former is still an instance of a choice function interpretation. The indefinites with -to do not appear where the true quantifiers can appear. Speakers’ intuition also confirms the difference in the interpretation of the genuine narrow scope indefinites with the suffix -nibud’ and quasi-narrow scope -to indefinites. 70 CHAPTER 3 -NIBUD’ AND -LIBO INDEFINITES 3.1. -NIBUD’ vs. other indefinites: distributional properties The general consensus on the relationship between -nibud’ and -libo pronouns is that these two types of indefinites have the same meaning but differ from each other stylistically: nibud’ indefinites are neutral with respect to the style of speech and can be used in wide range of contexts including colloquial conversations and formal speech situations; -libo indefinites are only used in bookish contexts. As I have mentioned before, there is evidence that the relations between the two indefinites is not that simple; later in this chapter, I will do a direct comparison of the two types. For now, however, let us consider the two indefinites as semantically identical. The majority of the examples I am going to look at in next sections are colloquial and thus would prefer -nibud’ indefinites over -libo indefinites, if either of these two types are possible in principle. Therefore, I will keep the data simple and use only -nibud’ examples for now. The reader needs to keep in mind, however, that, unless specifically indicated otherwise, I assume the same conclusions would hold for -libo indefinites as well in appropriate for them formal contexts. The distributional properties of -nibud’ (and, for now, -libo) indefinites are demonstrated below. First, these indefinites are not used in simple declarative sentences with several exceptional environments discussed below: (74) a. Ivan spel *kakuju-nibud’ / kakuju-to pesniu. Ivan sing-PAST some (*NIBUD’/TO) song 71 ‘Ivan sang a song’. b. On kupil *kakoj-nibud’ / kakoj-to tort He buy-PAST some (*NIBUD’/TO) cake ‘He bought a cake’ c. Za stenoj * kto-nibud’ / kto-to smejalsia. Behind wall someone(*NIBUD’/TO) laugh-PAST Someone was laughing behind the wall. The exceptional environments where -nibud’ indefinites are appropriate include future tense sentences, contexts with modals (must, need etc.), sentences with verbs of desire (wish/want etc.); sentences implying multiple events happening in different times (usually, always etc.) or with different individuals (e.g. every boy): (75) a. Ivan hotel spet’ kakuju-nibud’ / kakuju-to pesniu. Ivan want-PAST sing-INF some(NIBUD’/TO) song ‘Ivan wanted to sing a song’. b. On kupit kakoj-nibud’ / kakoj-to tort He buy-FUT some(NIBUD’/TO) cake ‘He will buy a cake’ c. On dolzhen byl kupit’ kakoj-nibud’ tort He must-PAST buy-INF some(NIBUD’/TO) cake ‘He had to buy a cake’ 72 d. Navernoje, on kupil kakoj-nibud’ tort Probably he buy-PAST some(NIBUD’/TO) cake ‘Probably he bought a cake’ e. Za stenoj vsegda kto-nibud’ / kto-to smejalsia. Behind wall always someone(NIBUD’/TO) laugh-PAST Someone was always laughing behind the wall. -Nibud’ indefinites are good in imperative sentences, where -to pronouns are not licensed under any conditions (76a); and in indirect sentences implying requests (both infinitival and that-clauses) with the same meaning as in the corresponding direct requests (76b). Only the latter allow for -to indefinites to be used as well but with a different, ‘specific’, interpretation, as in (76c) below: (76) a. Voz’mi kakuju-nibud’ /* kakuju-to knigu! Take-IMPER some(NIBUD’/*TO) book ‘Take some book!’ b. On skazal mne vziat’ / chtoby ja vzial kakuju-nibud’ knigu. He tell-PAST me take-INF / that I take-PAST some(NIBUD’) book ‘He told me to take a book’ (any book, doesn’t matter which). c. On skazal mne vziat’ / chtoby ja vzial kakuju-to knigu. He tell-PAST me take-INF / that I take-PAST some(TO) book ‘He told me to take a book’ (and he told me which one, but now I don’t remember). 73 One of the most natural environments for -nibud’ indefinites is provided by questions (both direct and indirect). (77) a. Vy vstretily kogo-nibud’ v parke? You-PL meet-PAST-PRF someone(NIBUD’) in the park Did you meet anyone in the park? b. On sprosil nas, vstretili li my kogo-nibud’ v parke. He ask-PAST us meet-PAST whether we someone(NIBUD’) in the park He asked us whether we met anyone in the park. In fact, for most of the questions, a -nibud’ pronoun is the only option for a question if an indefinite must be used. Naturally, the purpose of a question is to request information about something one doesn’t know. Specific individuals usually do not become an object of an inquiry. A special speech situation is needed to ask a question with a -to indefinite (see Chapter 2 for more details). Conditionals provide a natural environment for -nibud’ indefinites, as well. Antecedents of conditionals allow for both -nibud’ and -to indefinites, and the choice of either does not change the meaning of the sentence (78). This is one of the rare cases when the two indefinites seem to be completely interchangeable. (78) a. Esli kto-nibud’/ kto-to opozdajet, nam pridetsa exat’ na taxi. If someone(NIBUD’/TO) be-late-FUT we have-to-FUT go-INF by taxi If anyone comes late we’ll have to go by taxi. b. Kogda kto-nibud’ / kto-to opazdyval, nam prixodilos’ exat’ na taxi. 74 When someone(NIBUD’/TO) be-late-PAST we have-to-PAST go-INF by taxi When someone (anyone) was late we had to go by taxi. Consequents of conditionals also allow for both -to and -nibud’ indefinites, but they differ in their interpretation, and a -to indefinite would require a ‘specific’ scenario to be valid. For example, (79a) below is completely context independent; (79b), on the other hand, sounds very weird as an out-of-the-blue utterance: (79) a. Esli budet xolodno, my ostanemsia doma i posmortim kakoj-nibud’ fil’m. If be-FUT cold we stay-FUT home and watch-FUT some(NIBUD) film If it is cold we’ll stay home and watch a movie (doesn’t matter which one). b. #Esli budet xolodno, my ostanemsia doma i posmortim kakoj-to fil’m. If be-FUT cold we stay-FUT home and watch-FUT someone(TO) film If it is cold we’ll stay home and watch a movie (a particular one). However, in some scenarios, for instance in the one given in (80), the sentence is quite natural with the -to indefinite: (80) Scenario: In a summer camp, children are discussing the schedule for tomorrow. A person who is familiar with what the instructor said about it, shares with the others: Esli budet xolodno, my ostanemsia doma i posmortim kakoj-to’ fil’m. If be-FUT cold we stay-FUT home and watch-FUT someone(NIBUD) film If it is cold we’ll stay home and watch a movie (the name of the movie was given, but at this point the speaker doesn’t remember or doesn’t care which one it was). 75 This is an expected contrast in the meaning: a specific interpretation for -to vs. a nonspecific interpretation of -nibud’; this is exactly what was observed above for questions and embedded requests. For if-clauses this contrast is interesting, because it only occurs in consequents but not in antecedents of conditionals. The other “standard” case where -nibud’ indefinites are extremely common is definitions given in explanatory dictionaries of the Russian language. There is no consensus as some dictionaries prefer using -nibud’ indefinites and some choose to use -libo indefinites in their definitions. No dictionary, however, uses -to indefinites. One other potential candidate – ugodno indefinites, which have a meaning of universal “any” – also appears to be impossible in explanatory dictionaries: (81) Shveja – rabotnitsa po shit’ju chego-nibud’/chego-libo/*chego-to/*chego ugodno. Seamstress – worker-FEM in sewing something(NIBUD’/LIBO/*TO/*UGODNO). ‘A seamstress is a worker (specializing) in sewing (something)’. The purpose of using a -nibud’ indefinite in (70) is to indicate that regardless of what exactly the seamstress sews, she is called a seamstress if she sews anything at all. Using a -to indefinite in this case would require a seamstress to sew some specific thing(s). The other option, ugodno, is also impossible, because it is a universal quantifier and as such it does not allow for exceptions. A universal “any” in English, for instance, is contrasted to “a” indefinites in this intolerance to exceptions. E.g. A cat can catch mice is true even if there is a cat that cannot, however, Any cat can catch mice means that absolutely every cat has to be able to do it in order for the sentence to be true. The sentence in (70) does not need a universal. It says that a 76 seamstress is a person who can sew different things, but it doesn’t say that she can sew everything. She can still be a seamstress even if she can’t sew a parachute. The rich system of indefinite pronouns in Russian creates competition between different lexical items in terms of both distribution and semantics; each set of pronouns has its own unique function in the system due to its own unique properties. What are the semantic properties of -nibud’ pronouns that distinguish these indefinites from the members of the other sets? 3.2. Semantic properties of -NIBUD’ indefinites As we have already seen the most obvious differentiating feature between -to and -nibud’ indefinites is specificity. The declaratives, questions, indirect requests etc. where both types of indefinites can be used, the sentences differ in their interpretation; -to indefinites are always specific and -nibud’ indefinites are non-specific. 15 The other potential competitors of -nibud’ pronouns, ugodno indefinites have their own functions and are not interchangeable with -nibud’ indefinites either. The example in (70) bans an ugodno indefinite because choosing this item would imply a universal interpretation (no exceptions can be made). -Nibud’ indefinites are not genuinely universal even though in some contexts their interpretation is very close to the one of free choice items. Any book will satisfy the request in (82), however, a follow-up question “Which exactly” is not an impossible continuation of a dialogue: 15 Other universal (free choice) indefinites in Russian include liuboj/liubaja/liuboe and kakoj by to ni bylo. They differ from each other stylistically. Free choice items are outside of the scope of this dissertation. When I need to look at them for comparison with -nibud’, I will only include ugodno items in the comparison. 77 (82) – Daj mne kakuju-nibud’ knigu! Give me some(NIBUD’) book ‘Give me a book!’ – Kakuju imenno? ‘Which exactly?’ Such a question does not sound wrong, because out of all the books available in the context of speech there might be one or two that wouldn’t work for the speaker for some reason. A -nibud’ indefinite is not specific, the addressee has a choice of different books, but -nibud’ pronouns are not universal, they do allow for exceptions as well. Using ugodno in a request, on the other hand, means that absolutely any choice would be satisfactory, and, expectedly, the same follow up question is not valid anymore: (83) – Daj mne kakuju ugodno knigu! Give me any(UGODNO) book ‘Give me any book!’ – # Kakuju imenno? ‘Which exactly?’ Only ugodno indefinites can be used in generic sentences; -nibud’ items do not bring the generic interpretation: (84) a. Kakoj ugodno jurist smozhet otvetit’ na vashi voprosy, no ja ne smogu. Which UGODNO lawyer can answer on your questions, but I not can 78 ‘Any lawyer can answer your questions, but I cannot.’ b. Kakoj-nibud’ jurist smozhet otvetit’ na vashi voprosy, no ja ne smogu. Which(NIBUD’) lawyer can answer on your questions, but I not can Possible interpetation: ‘A lawyer can answer your questions but I cannot.’ Impossible interpetation: *‘Any lawyer can answer your questions but I cannot.’ As a result, -nibud’ indefinites do not pass standard tests for universal quantifiers (e.g. almost test): (85) a. Pochti kakoj ugodno jurist smozhet otvetit’ na vashi voprosy. Almost Which UGODNO lawyer can answer on your questions Almost any lawyer can answer your questions. b. * Pochti kakoj-nibud’ jurist smozhet otvetit’ na vashi voprosy. Almost which(NIBUD’) lawyer can answer on your questions Almost (NIBUD’) lawyer can answer your questions. c. Pochti ni odin jurist ne smozhet otvetit’ na vashi voprosy Almost not one lawyer can answer on your questions, but I not can Almost no lawyer can answer your questions. As the comparison above shows, Russian -nibud’ in “almost” sentences patterns with English some, but not with “any” as we could have expected from non-specific nature of -nibud’ indefinites: (86) Almost any lawyer… 79 Almost no lawyer… *Almost some layer… So, -nibud’ indefinites are very different from universal, free choice, any in English. However, they are semantically very similar to polarity sensitive (existential) any. This similarity does not always extend to syntax. For example, -nibud’ pronouns cannot be used in the scope of negation: Russian follows the rule of Negative Concord. But it is easy to see that, for instance, the interpretation of -nibud’ indefinites in questions is parallel to that of any in the same cases. According to Kadmon and Landman (1993), English Do you have potatoes? and Do you have any potatoes? differ from each other; the former is interpreted as an inquiry about potatoes of cooking quality, the latter extends to all kinds of potatoes: small/rotten/etc. This difference comes from two features that presence of any brings into the sentence: (i) widening: any widens the interpretation of the common noun phrase along a contextual dimension; and (ii) strengthening: any is licensed only if the widening that it induces creates a stronger statement. Russian -nibud’ indefinites demonstrate the same effect in comparison with common noun phrases: (87) a. U tebia jest’ gazety? At you be newspapers ‘Do you have newspapers?’ (I want to read something) b. U tebia jest’ kakije-nibud’ gazety? At you be some(NIBUD’) newspapers ‘Do you have any newspapers?’ (I am collecting stuff to recycle; I’ll take anything newspaper-like) 80 c. U tebia est’ kot? At you be cat ‘Do you have a cat?’ (animal) d. U tebia est’ kakoj-nibud’ kot? At you be some(NIBUD’) cat ‘Do you have any cat?’ (a toy cat or a picture of cat will do, too; there is a child here who doesn’t know what ‘cat’ means) Russian -nibud’ indefinites, like English any, provide domain widening / strengthening of the expression. An additional confirmation for this comes from the data where the domain cannot in principle be widened and where – as expected, -nibud’ indefinites are not possible; therefore, a contrast in acceptability between (88a) and (88b) below: (88) a. U tebia jest’ kakije-nibud’ gazety? At you be some(NIBUD’) newspapers ‘Do you have any newspapers?’ b. *U tebia jest’ kakije-nibud’ deti? At you be some(NIBUD’) children ‘Do you have any children (of your own)?’ In a particular speech situation, the sentences in (88b) can become acceptable, but ‘any children’ cannot refer to the children of the addressee of the question only. A plausible scenario for an appropriate usage of (88b) is, for instance, something like the situation in (89): 81 (89) My friend, who is a professional photographer, is working on a web-page to advertize his photo services. He has posted his best pictures and is very proud of how the page looks like. When I see the design, I notice that all the pictures he provides are ones of adults. I know that many customers would like to take professional pictures of their children; it would be nice to show some examples of those as well. I ask my friend the following question: U tebia jest’ kakije-nibud’ deti? At you be some(NIBUD’) children ‘Do you have any children (in your picture collection)?’ The requirement on maximizing the domain could explain the impossibility of -nibud’ in simple episodic declarative sentences in the present and past tense for the same reasons why English does not allow any in such contexts. Kadmon and Landman (1993) give two reasons for why English *I saw anyone/anything is ungrammatical: it violates informativeness (it is too weak a statement and it should not be used under Grice’s maxim of Quantity), and it violates strengthening, which, according to Kadmon and Landman is a lexical feature of any and cannot be ignored. Russian equivalent with -nibud’ is also bad (90a), but it immediately becomes good in a corresponding question (90b) where neither informativeness nor strengthening is violated: (90) a. *Ja videl kogo-nibud’/chto-nibud’ I saw who(NIBUD’) /what(NIBUD’) ‘*I saw anyone/anything’ 82 b. Ty videl kogo-nibud’/chto-nibud’? You-SG saw who(NIBUD’) /what(NIBUD’) 16 ‘Did you see anyone/anything? ’ So, the initial observations on the semantics and syntax of -nibud’ indefinites can be, for the moment, summarized as follows: - -Nibud’ indefinites are never specific; - -Nibud’ indefinites are existential and not universal; - -Nibud’ indefinites induce domain widening and strengthening; - -Nibud’ indefinites have to be licensed by particular operators (including questions, modals, conditionals etc.) An analysis of -nibud’ indefinites needs to account for all these features. Let us see how this can be accomplished. 3.3. Formal analysis: accounting for the semantics and distribution Russian -nibud’ indefinites have attracted the attention of formal semanticists. Since all the peculiarities of this set of pronouns arise only in comparison with other sets, -nibud’ items are usually studied in comparison with -to and koe- indefinites. Because it is easier to present these works in the same way, I gave a detailed analyses of previous approaches to -nibud’ indefinites in the general review of the research on Russian indefinite pronouns (see chapter 1 for 16 For Kadmon and Landman (1993) this is a free choice (universal) ‘any’. Russian nibud’ is never universal. The effect of seemingly universal interpretation of -nibud’ items in some contexts comes from strengthening. 83 details). Here, I will again address two major works on this topic and highlight the problems with these two approaches. Then, I will suggest my own analysis of -nibud’ indefinites. Geist (2008) describes -nibud’ indefinites as always having narrow scope and not allowing for any other interpretations; she defines them as ‘markers of non-specificity’ with the formal representation in (91): (91) 〚-nibud’〛 = P x fx(P) She argues for this representation in the following way: “The implicit argument of the non-specificity marker nibud’ is existentially closed at the lexical level yielding non-identifiability of the referent by the speaker, narrow scope, and the lack of co-variation reading. We interpret the fact that for nibud’ no specific anchor exists in the following way: indefinites accompanied by nibud’ introduce a completely random referent. The referent of a nibud’-indefinite is not identifiable. However, the lexical entry of nibud’ is still a simplification since the fact that nibud’ needs licensing by some operators in the clause is not yet integrated in it.” (Geist, 2008, p. 159) So, she first makes the denotation of a -nibud’ item specific (via implicit argument of a choice function which, normally, needs to have an outside binder in order to be tied to a referent), but then closes the access to the implicit argument by an existential closure to account 84 for the fact that indefinites of this type are never specific. First, it is not clear how this mechanics works. Second, even if it yields the desired result – a non-specific item – it is achieved through an extra complication (making it specific first) which is not validated by any independent need; but even such a complicated approach is still “a simplification”, because, as Geist herself acknowledges, it does not account for the distribution of -nibud’ items (licensing conditions). Yanovich (2006) also looks at -nibud’ indefinites as at choice functions, but in his account they are “obligatory skolemized choice functions”. Contrary to Geist (2008) and many other researchers of Russian indefinites he claims that -nibud’ indefinites, besides narrow scope, can also get an intermediate scope interpretation. This becomes the main argument for him to interpret -nibud’ items choice-functionally; otherwise, the possibility of the claimed intermediate scope reading is hard to explain. His example of the intermediate-scope interpretation of a -nibud’ indefinite is given below along with his interpretation of the sentence: (92) Každyj mal’čik budet rad esli vstretit kogo-nibud’ iz svoix odnoklassnic. Every boy will be glad if [he] will-meet whoNIBUD’ of his girl-classmates. * Ǝ >∀ > if (There is a girl that every boy will be happy to meet) OK ∀ > Ǝ > if (For every boy, there is a girl he will be happy to meet) OK ∀ > if > Ǝ (For every boy, if there is a girl he meets, he will be happy (from Yanovich 2006, #19) 85 As I native speaker of Russian, I do not share the judgement on the possibility of the second (intermediate scope) interpretation. All native spekers I informally consulted with confirmed my opinion. If the particular sentences provided above had the intermediate scope interpretation we would expect that this sentence will be accepted in the following situation: There are three boys: Vasja, Petja and Ivan. They go to school together. Tanja, Anja and Marina are their classmates. The boys went to a party. Vasia is very fond of Anja and will be very happy to see her at the party. He does not, however, care about seeing Tanja or Marina. As I already discussed in Capter 1, this is not an acceptable scenario for (91). The denotation, offered by Yanovich (2006), predicts an intermediate-scope reading in the cases where it doesn’t occur, and thus, overgenerates. The other problem with Yanovich’s approach is that there is no real distinction between the denotations for -nibud’ and -to indefinites, if we accept that the latter also have to be parameterized choice functions. This seems a valid assumption given that they may acquire a pseudo-narrow scope interpretation with an operator that can bind the implicit argument (see Chapter 2 for details). Besides, there is evidence that -nibud’ expressions (unlike -to indefinites) demonstrate quantificational behavior, similar to the one of English polarity sensitive any. They induce domain widening and strengthening effect comparing to the bare nouns; they can never be specific, they are only used in the contexts which all together can be characterized as non-veridical (questions, conditionals, subjunctives, generic sentences, direct requests etc.). -Nibud’ indefinites are also licensed in before-clauses (and not licensed in after-clauses) in the same way as English any as in They left the country before / *after anything happened (see Zwarts (1995) for details). The intuition about the semantics of -nibud’ indefinites is very close to what is 86 observed with respect to any in English; i.e. “the intuition that any statements support counterfactual inferences” so that their “domain of quantification is not a set of particular individuals but the set of possible individuals of the relevant kind” (Dayal 1998). The quantifier approach to for -nibud’ indefinites is also better compatible with observations by Pereltsvaig (2008), who demonstrates that -nibud’ is not compatible with a single instance of an individual being available; it requires an operator/quantifier with respect to which it can take narrow scope. Pereltsvaig calls -nibud’ ‘a marker of co-variation’ and demonstrates that -nibud’ can be licensed in past tense declarative sentences if they provide an environment for co-variation: (93) a. *Ego o chem-nibud’ sprosili. He was asked about something. b. Kazhdogo o chem-nibud’ sprosili. Everyone was asked about something (what was asked varies from person to person). With all these facts, choice functional approach to -nibud’ pronouns doesn’t seem to be grounded. Looking at these indefinites as at existential quantifiers, on the other hand, may not only more successfully account for their properties but will also allow for making some parallels across languages which might be beneficial by itself. In the following analysis, I will not make any preliminary stipulations about the inherently choice-functional nature of -nibud’ indefinites. I will start with the assumption that they are “like common noun phrases”, and then look at how the semantics and syntax of a complex structure “-nibud’ + NP” is different from just an NP in the same context. 87 Let us have a look at simple declarative sentences and corresponding questions first. As the table in (83) shows, grammatical simple declaratives with noun phrases in A(i) and A(ii) become ungrammatical with adding -nibud’ suffix to the noun phrases in B(i) and B(ii). (94) (A) Bare nouns (B) NIBUD’ indefinites Declaratives - (i) On prochital knigu. - ‘ He read a/the book’ - (i) *On prochital kakuju-nibud’ knigu. ‘He read a book’ (ii) U menia est’ kot - ‘I have a cat’ (ii) *U menia est’ kakoj-nibud’ kot ‘I have a cat’ Questions - (iii) On prochital knigu? - ‘Did he read a/the book?’ - (iii) On prochital kakuju-nibud’ knigu? (iv) U tebia est’ kot? ‘Did he read a book?’ ‘Do you have a cat?’ - (iv) U tebia est’ kakoj-nibud’ kot? ‘Do you have a cat?’ The corresponding questions in A (iii-iv) and B(iii-iv) are equally good. The questions in A part and B part mean exactly the same (they are inquiries about having books/a cat), but they are not interchangeable. Asking B(iv) instead of A(iv), the speaker wants to include more books/cats in the range of available/suitable books/cats. With the questions in A, we refer to the books and cats we can plausibly consider under given circumstances in the given context (what counts as a book? what counts as a cat?). With the question in B, we extend what the noun 88 phrases denote to other (normally not considered in the given context) individuals. The domain widening explains the contrast between (95a) and (95b) (95) a. U tebia est’ kot? ‘Do you have a cat?’ (animal) b. U tebia est’ kakoj-nibud’ kot? ‘Do you have any cat?’ (a toy cat or a picture of cat will do, too; there is a child here who doesn’t know what ‘cat’ means). If we look at questions in the Hamblin (1973) – Karttunen (1977) way (when a question denotes a set of answers to it), the denotations for A(iv) and B(iv) look very similar (it doesn’t matter what language to use in denotations, so I will use English to avoid extended glosses): (96) a. Did he read a book? . . ′ b. Did he read a NIBUD’ book? . . ′ ′ ′ So, in both cases, the interpretation of the question is a set of two propositions, the first one is . ( ′ ) , and the second one is . ′ ( ′ ) . The only difference between the two cases is that for A(iv) each of the propositions can be used independently as a separate sentence like in A(i-ii). For B(iv), however, the only possible 89 form of a sentence is the resulting question, in which neither positive nor negative propositions by themselves are good, as demonstrated by B(i-ii). So, whatever the denotation of a -nibud’ item is, the item does not get validated at the stage of the embedded proposition, but only gets validated when the propositions are combined in a set to make a question. The difference between the three sentences in (97) is the absence / presence of a modal; the sentence with the -nibud’ indefinite is only grammatical with the modal but is ungrammatical without it: (97) a. On kupil *kakoj-nibud’ / kakoj-to tort He buy-PAST some (*NIBUD’/TO) cake ‘He bought a cake’ b. On mog kupit’ kakoj-nibud’ tort He can-PAST buy-INF some(NIBUD’/TO) cake ‘He could buy a cake’ c. On dolzhen byl kupit’ kakoj-nibud’ tort He must-PAST buy-INF some(NIBUD’/TO) cake ‘He had to buy a cake’ If we take the approach to modals by Kratzer & Shimoyama (2002), we’ll have the following denotations for can and must: (98) 90 a. A possibility modal says that some proposition in the alternative set it operates over is true in some accessible world: For 〚α〛 w,g D: 〚can〛 w,g = {λw'. w'' [w'' is accessible from w' and p [p 〚α〛 w,g and p(w'') =1]]} b. A necessity modal requires that for every accessible world there be some proposition in the alternative set that is true in it: For 〚α〛 w,g D: 〚must〛 w,g = {λw'. w'' [w'' is accessible from w'  p [p 〚α〛 w,g and p(w'') =1]]} Again, as with questions above, the sentence in (86a) does not allow for a -nibud’ indefinite, but including the same sentence under the scope of a modal licenses the indefinite. Both questions and modals operate above the sentence level and bring in alternatives. As soon as something introducing alternatives appears above -nibud’ indefinites, they become grammatical even though they were ungrammatical otherwise. A logical next step is to accept a working hypothesis: -nibud’ indefinites can be only used in the presence of alternatives and try to develop semantics for -nibud’ pronouns compatible with that. In doing so, we need to keep in mind, that -nibud’ indefinites mean the same thing as bare indefinites, but induce domain widening. Then a denotation for NIBUD’-book is going to be similar to a denotation of 〚book〛 but allow access to more instances of books as in (99) below: 91 (99) 〚-nibud’w' book〛 w,c = . ′ , ℎ ′ ℎ Truth-conditionally, the sentences having indefinites with or without the suffix –nibud’ are the same. The suffix indicates that there is a need of an extended domain for the NP, and the requirement on the domain widening is presuppositional. How does the presupposition work? We will need to assume that the salient context in the actual world may differ from the contexts available in other accessible worlds. Then the context that a common noun phrase has an access to is as given in (100) below: (100) Cw {x: x C and x is in w} Then the domain widening presupposition carried by -nibud’ indefinites is the following relationship between accessible individuals denoted by the noun phrase in the actual world (w) and the world that is accessible from the actual world (w’): (101) {x: x ∈ Cw and x is a book in w} {x: x ∈ Cw' and x is a book in w'} With this presupposition, the resulting denotation for a -nibud’ indefinite (in this case, for NIBUD’-book) looks like (102) below: (102) 〚-nibud’w' book〛 w,c = : {y: y ∈ Cw and y is a book in w} 92 {y: y ∈ Cw’ and y is a book in w'}. x is a book in w' For this denotation, we need to assume that individuals can exist across the worlds. This is not the only possible understanding of how individuals can be distributed in different accessible worlds. A different view is that a different world would have counterparts of the individuals but not the individuals themselves. In that case, the denotation above will need to have an extra piece saying “all books salient in w have counterparts in w’ but not vice versa”. For simplicity of the denotation I will not take this position here. Now it is clear why -nibud’ indefinites are compatible with a question but are not compatible with a single answer to it (Did he read NIBUD’ book? but *He read a book.). In Hamblin-Karttunen style (see Hamblin 1973; Karttunen 1977 for details), questions are made by creating alternatives when the Q-morpheme applies to the set of propositions (answers to the question): (103) Q-morpheme: [[Q]] = < , > . ( ) .¬ ( ) The alternatives are introduced into the structure after the sets of possible answers are combined. Neither of the propositions contains anything to bind a w' index on the -nibud’ indefinite, the binder ( ′) appears only at the stage of question formation as demonstrated below: 93 (104) Question: On prochital kakuju-nibud’ knigu? (Did he read a (NIBUD’) book?) CP t Q > ′ He read NIBUD’w' book Within a simple declarative sentence (TP level), the assignment function assigns a default value to w' for the -nibud’ indefinite, and it’s the same value as w, because nothing else is available at that point. However, w’=w is not an allowed option for the presupposition of nibud’. A possible binder which allows access to w', and thus satisfies the indefinite, comes with formation of a question. An indefinite only containing a common noun phrase does not carry a presupposition of domain widening. It may but doesn’t have to have an access to worlds other than the evaluation world w, and a separate grammatical sentence can be produced at any level, as a TP or as a CP: Did he read NIBUD’ book? and He read a book. The same holds for other contexts where -nibud’ indefinites are used: modals, conditionals, imperatives (which act as bearers of implicit modality ‘you must’) are good contexts for -nibud’ pronouns because they satisfy the requirement of the indefinite providing access to worlds other than the evaluation world. 94 The licensing conditions are explained without any extra stipulations simply by taking into consideration the domain widening semantics of -nibud’ indefinites, which needs to be accounted for independently. This analysis supports the intuition about the special types of contexts (non-veridical ones) that accompany domain-widening indefinites cross-linguistically (see Zwarts 1995, Giannakidou 2001, and Dayal 1998 for data from English, Modern Greek, Dutch and Spanish). With this satisfying result, let’s now look at what is considered to be a semantic equivalent of -nibud’ indefinites and see whether this equivalency actually holds. 3.4. -LIBO indefinites – same or different? The Russian indefinite suffix -libo originates from a disjunction libo which is used either by itself (meaning ‘or’) or as a two-place disjunction libo… libo… (meaning ‘either … or’). Indefinite pronouns with the suffix -libo behave similarly to -nibud’ indefinites. However, a direct comparison is hard to make, because most of the contexts where these indefinites are used are of different stylistic registers as -libo is more formal/bookish, and substituting one indefinite instead of the other produces stylistically inappropriate sentences. However, it is possible to find sentences of the same type where these indefinites demonstrate similar semantics and syntactic limitations. The sentence in (105a), taken from an on-line discussion of jobs in computer science, contains a present tense verb “possess” referring to multiple occurrence of an event (usually) and allows for a -libo indefinite. The same sentence with a one-time past tense event in (105b) is ungrammatical: (105) a. Takije spetsialisty obychno obladajut ochen’ detal’nymi znanijami v kakix-libo voprosax. 95 Such specialists usually possess-PRES very detailed knowledge in some-LIBO questions. ‘Such specialists usually have very detailed knowledge in some things’. b. *On obladal ochen’ detal’nymi znanijami v kakix-libo voprosax He possess-PAST very detailed knowledge in some(LIBO) questions ‘He had very detailed knowledge in some things’ An on-line questionnaire about atmosphere in business offices contains a question given in (106a), but it’s impossible to answer this question using the same indefinite (106b), a -to pronoun has to be used instead: (106) a. Kto-libo ili chto-libo dopekajet vas na rabote? Who-LIBO or what(LIBO) bother-PRES you at work ‘Does someone or something bother you at work?’ b. Chto-to / *Chto-libo dopekajet ego na rabote. What-TO /*What(LIBO) bother-PRES him at work ‘Something bothers him at work’. Imperatives with -libo are harder to find than imperatives with -nibud’, however they are not impossible. An internet forum, where people help each other find information about different things, has the following title: (107) Pomogite najti chto-libo o… Help-IMP find-INF something(LIBO) about… Help (me) find something about…. (insert your request here). 96 Conditionals of appropriate (formal or bookish) register are very common with –libo: (108) Kogda kto-libo delajet chto-libo xorosheje, my dolzhny vygravirrovat’ eto na kamne. When who(LIBO) does what(LIBO) good we must engrave it in stone ‘When someone does something good we must engrave it in stone’. Definitions of words in explanatory dictionaries have an extremely high rate of -libo items. This is probably the only area where stylistic differentiation is completely absent; different dictionaries choose to use one or the other form of the indefinites and use it consistently throughout the volume (or several volumes). The definitions with -nibud’ and -libo indefinites are absolutely identical otherwise. So for most of the environments, the parallel between -nibud’ and -libo indefinites holds. There are two cases, however, where the latter seem to behave differently. One of the problematic cases comes from Sheliakin (1986) who gives examples of comparisons where -libo items sound natural and -nibud’ items are ungrammatical, as in his example cited below: (109) Puma begajet bystreje, chem kakoje-libo /*kakoje-nibud’ drugoje zhivotnoje. Puma runs faster than any(LIBO)/*any(NIBUD’) other animal ‘A puma runs faster than any other animal.’ A possible explanation for this contrast might come from register issues. The sentence in (109), indeed, seems degraded with a -nibud’ indefinite. The same is true for his other examples. However, a closer look at the comparisons Sheliakin uses in his analysis shows that 97 they are all instances of sentences that are stylistically marked as “bookish” (the puma example is taken from a book on animal life etc.). And then it might be the case that it is the stylistic unmarkedness (and not the semantics) of -nibud’ that creates a conflict in such comparisons. Indeed, “informal” comparisons seem to be better with -nibud’ expressions: (110) Petia mozhet s’jest’ bol’she konfet, chem (?) kakoj-nibud’ / kakoj-libo drugoj rebenok. Petia can eat more candies than any(LIBO)/*any(NIBUD’) other child ‘Petia can eat more candies than (?) some other child / any child’. But (110) with -nibud’ is still not absolutely great. A reason for inappropriateness of -nibud’ indefinites in cases like (109) is that the sentence clearly implies an absolute comparison (there is no animal faster than puma) and -nibud’ can never be interpreted as having a universal interpretation. And it’s the -libo indefinite that seems to bring about this “absolutive” flavor. The most natural interpretation of (110) with -libo indefinite is also “absolutive” – ‘there is no one who can eat that many candies as Petia can’. However, the ‘forced’ interpretation for the nibud’ version of the same sentence has a different interpretation of the indefinite: with -nibud’ the sentence means that Petia can eat more candies than some or other child. It seems pragmatically weird to compare a particular person with someone else who is unknown (nonspecific) if the comparison is not absolute, and a -nibud’ indefinites that bans a universal interpretation is not the right choice. The question, then, is whether -libo is actually different in this respect and can be interpreted universally. If it is, then we will have to give up on the assumption that the two types of indefinites are semantically the same but differ in register. 98 Before testing -libo indefinites for universality let us look at the other significant difference in the distribution coming from sentential negation. In Russian, narrow-scope indefinites are generally not allowed under sentential negation (negative concord), but can be used with implicit (lexical) negation: (111) a. On ne videl nikogo znakomogo He not see-PAST no one-ACC familiar-ACC ‘He didn’t see anyone familiar’ (lit.: no one familiar) b. On otkazalsia priglasit’ kogo-libo / kogo-nibud’ iz znakomyh He refuse-PAST invite-INF some(LIBO)/(NIBUD’) of acquaintances ‘He refused to invite any of his acquaintances’. However, -libo indefinites seem to be acceptable not only under lexical negation, but also under grammatical negation: (112) a. Yush’enko ne sobirajetsa k komu-libo prisoediniat’sa. (on-line news) Yush’enko not going-to to someone(LIBO) join-INF ‘Yush’enko is not going to join anyone’. b. My ne pozvolim komu-libo govorit’ s Armenijej na jazyke ul’timatumov. (on-line news) We not let-FUT someone(LIBO) speak-INF with Armenia on language (of) ultimatum-PL ‘We will not let anyone speak with Armenia in the language of ultimatums’. 99 The sentences in (112) are clearly “no-exception” ones (not a single person can speak with us like that!). Interestingly, (112a) and (112b) are perceived as stronger claims than their equivalents with “correct” negation concord would: (113) a. Yush’enko ne sobirajetsa ni k komu prisoediniat’sa Yush’enko not going-to to no one join-INF Yush’enko is not going to join anyone. b. My ne pozvolim nikomu govorit’ s Armenijej na jazyke ul’timatumov. We not let-FUT no one speak-INF with Armenia on language (of) ultimatum-PL We will not let anyone speak with Armenia in the language of ultimatums. The speakers uttering (101) neglected “correctness” to make a stronger claim. This is an expected outcome under the domain-widening/strengthening approach: since negative pronouns are neutral with respect to what domain is taken under consideration when they are used, the sentences in (113) do not have the same pragmatic effect as the ones in (112). The sentences containing -libo indefinites, on the other hand, are not neutral. If we believe that -libo indefinites are semantically equal to -nibud’ items, they would induce domain-widening effect thus making stronger statements. If these sentences are at the same time prescriptively wrong – this is a price to pay for making a stronger claim and the speakers choose to violate the grammar rule but achieve the pragmatic effect. The question to be answered, however, is whether -libo indefinites actually are semantically the same as -nibud’ items. If the universality effect observed with comparisons and under grammatical negation is caused by a genuine semantic property of -libo indefinites, they cannot be treated in the same way as -nibud’ pronouns. If -libo pronouns are not universal, 100 then there is no need to develop an independent denotation for -libo; but then we somehow need to account for the universality effect in (109), (110), and (112). As we had seen, -libo items seem to be less tolerant to exceptions, and in this respect resemble free choice items. On the other hand, they behave differently from truly free choice items (ugodno, liuboj, or kakoj by to ni bylo); instead they completely copy -nibud’ pronouns: (114) a. Kakoj ugodno / liuboj jurist smozhet otvetit’ na vashi voprosy, no ja ne smogu. Which UGODNO / LIUBOJ lawyer can answer on your questions, but I not can ‘Any lawyer can answer your questions, but I cannot.’ b. Kakoj-nibud’ jurist smozhet otvetit’ na vashi voprosy, no ja ne smogu. Which(NIBUD’) lawyer can answer on your questions, but I not can Possible interpetation: ‘A lawyer can answer your questions but I cannot.’ Impossible interpetation: *‘Any lawyer can answer your questions but I cannot.’ c. Kakoj-libo’ jurist smozhet otvetit’ na vashi voprosy, no ja ne smogu. Which(LIBO) lawyer can answer on your questions, but I not can Possible interpetation: ‘A lawyer can answer your questions but I cannot.’ Impossible interpetation: *‘Any lawyer can answer your questions but I cannot.’ As (114b) and (114c) demonstrate, neither of the indefinite pronouns can be interpreted universally in generic sentences, both indefinites are existential. The same is true for standard tests for free choice items (e.g. almost test); both -nibud’ and -libo indefinites fail to appear with almost where the actual free choice ugodno indefinites appear. Initially, both sentences in (115) are grammatical; the first one contains a -libo indefinite, the second one – a free choice universal: 101 (115) a. O kazhdom iz nih mozhno skazat’ chto-libo xorosheje… About each of them possible tell-INF something(LIBO) good ‘About each of them I can tell you some good things…’ (from a newspaper) b. O kazhdom iz nih mozhno skazat’ chto ugodno About each of them possible tell-INF anything (UGODNO) ‘About each of them I can tell you just anything’ The same sentence as in (115a) becomes absolutely unacceptable with adding almost as in (116a). The corresponding sentence with ugodno item, though, is completely grammatical as (116b) below demonstrates: (116) a. *O kazhdom iz nih mozhno skazat’ pochti chto-libo xorosheje. About each of them possible tell-INF almost something(LIBO) good ‘About each of them I can tell you almost some good things.’ b. A kazhdom iz nih mozhno skazat’ pochti chto ugodno. About each of them possible tell-INF almost anything (UGODNO) ‘About each of them I can tell you almost anything.’ So -libo items may in some contexts function as a universal, but are not truly universal. The universality effect may occur in particular contexts as a result of their domain-widening semantics. In certain cases, -nibud’ indefinites can be perceived as universal as well, but it happens more often with -libo indefinites. A reason for this mismatch may come from lexical / stylistic gaps in the system of indefinite pronouns. To make a comparison, for instance, one can 102 use a free choice item, and that is what happens in most of the cases. But all free choice items in Russian have some stylistic limitations: ugodno, for example, is very colloquial; kakoj by to ni bylo is extremely bookish and slightly obsolete. If a comparison is made in a more or less formal text, the speaker may choose a -libo indefinite instead of a true free choice item as rescue strategy to fix the mismatch between what a speaker wants to say and what lexemes are available. -Libo indefinites, then, have the same semantics as -nibud’ pronouns. They induce domain widening effect and therefore have to be licensed by an operator that is able to satisfy the presuppositional requirement of the indefinite – to provide access to worlds other than the evaluation world. Rare cases of non-interchangeability of -nibud’ and -libo pronouns are not caused by a difference in their semantic properties but result from lexical and stylistic gaps in the system of indefinite pronouns in Russian. 3.5 Conclusions In this chapter, I have shown that both -nibud’ and -libo indefinites are limited in their distribution. They are ungrammatical in simple declarative past and present time sentences referring to one-time events. They are licensed in questions, with modal verbs, with verbs of desire, in conditionals, etc. They do not change truth conditions of the utterances comparing to the corresponding sentences with NP expressions; what -nibud’ and -libo indefinites bring with them is a domain widening / strengthening effect. I propose that the domain widening effect is a result of the presupposition that requires an access to a possible world(s) different from the evaluation world. This explains the peculiar distribution of these items – in order to be licensed they need to be in a scope of an operator that provides access to other possible worlds; otherwise 103 -nibud’ and -libo indefinites are not valid. The distributional patterns of the two sets of indefinites are not completely the same, the two cases where -libo indefinites are much more common than -nibud’ items include comparisons and grammatical negation. Despite this distributional difference, I argue that the two sets of indefinites are semantically equivalent. If we assume that -libo items induce domain widening / strengthening, choosing these indefinites under grammatical negation instead of regular negative pronouns in violation of the negative concord receives a natural explanation – negative pronouns do not induce domain widening and therefore regular (and prescriptively correct) cases of double negation are weaker than the corresponding prescriptively “wrong” utterances with -libo indefinites. The ability to have an access to a wider range of individuals allows for using -libo indefinites (along with universal ugodno items) in comparisons. Preferring -libo forms to truly free-choice items in comparisons is facilitated by a mismatch between the register of the speech and the available lexical resources; -libo meets the stylistic requirements of textbook/encyclopedia type of texts better than any of the free-choice lexemes. 104 CHAPTER 4 KOE- INDEFINITES 4.1 Choice functional KOE- and problems with it As we have seen before, Russian indefinite pronouns differ in their semantics. -To indefinites are scopally (but not epistemically) specific. They may be perceived as non-specific (quazi-narrow scope interpretation) if the choice function is parameterized (the parameter is bound by a quantifier). The peruliar disctribution requirements of -nibud’ indefinites are based their domain-widening semantics. The indefinites with the suffix -libo are very similar to nibud’ pronouns; the only difference is their perceived universality in comparisons and under negation. This difference is caused by another, stylistic in nature, property of the system: only a formal-style -libo but not a stylistically neutral -nibud’ is able to function as a free choice indefinite in the contexts, where the truly free-choice items are not appropriate stylistically. We have seen that very different dimensions (semantics, distribution, style of speech, etc.) come into play in the systematic relations between the different types of indefinites. Koepronouns bring in even more diversity. The most ‘visible’ feature that distinguishes koe- indefinites from both -nibud’ and -to items is that the individuals to whom they refer have to be known to the speaker. An illustration of the three-way contrast in specificity is given below: (117) a. Masha prigotovit chto-nibud’ vkusnoje na uzhin. ‘Masha will cook something delicious for dinner’ (Any dish could be cooked). 105 b. Masha prigotovit chto-to vkusnoje na uzhin. ‘Masha will cook something delicious for dinner’ (I, the speaker, don’t know what exactly will be cooked, but I know that the menu has been decided on already) c. Masha prigotovit koe-chto vkusnoje na uzhin. ‘Masha will cook something delicious for dinner’ (I, the speaker, know what she is cooking but I don’t want to tell you) As these examples and their interpretations demonstrate, -to indefinites and koeindefinites are both specific but they have different types of specificity: -to items are scopally specific; they have wide-scope interpretation, the item under discussion is known to exist, but it is not familiar to the speaker; koe- indefinites are epistemically specific, the referent is known to the speaker but for some reason the speaker does not want to mention its name explicitly. In many works on indefinites (e.g. Groenendijk, J. & M. Stokhof 1980, Farkas 2002, Kagan 2006, Ionin 2008), the term ‘specificity’ is used to refer only to epistemic specificity; in these cases, specificity is defined through uniqueness and familiarity to the speaker or speaker identifiability. For the authors above, -to indefinites are non-specific, as well as -nibud’ pronouns. In this paper, I use the term ‘specificity’ in a broader sense and subdivide it further into ‘scopal specificity’ (when referring to -to indefinites) and ‘epistemic specificity’ (when describing koe- indefinites). Kagan (2006) argues that koe- indefinites are obligatory epistemically specific; she provides multiple examples of the contexts banning koe- indefinites because they contain information indicating non-familiarity of the referent for the speaker: 106 (118) a. #Koe-kto pozvonil, no ja ne znaju, kto eto byl. someone called but I NEG know who this was ‘Someone called, but I don’t know who it was.’ b. #Dima razrabatyvaet koe-kakoj proekt, o kotorom ja ne imeju ni malejshegko predstavlenija. Dima works-out some project about which I NEG have not slightest idea ‘Dima is working on some project and I have no idea what this project is about.’ c. Ivan koe na kom ženilsja; #ponjatija ne imeju, na kom imenno. Ivan on-someone married idea NEG have(1st sg) on who exactly ‘Ivan has married someone; I have no idea who exactly it is.’ (examples from Kagan, 2008) This obligatory epistemic specificity can be straightforwardly accounted for by a choicefunctional analysis of koe- items when an indefinite is interpreted as shown below: (119) Masha prigotovit koe-chto vkusnoje na uzhin. ‘Masha will cook something delicious for dinner’ CH(f) ʌ Masha will cook f(a tasty thing/things) for dinner. Standard (non-parameterized) choice function interpretation provides a desirable result; the only way of parameterization would be parameterization by the speaker anyway. Geist 107 (2008) uses this standard choice-function approach and gives the following denotation for koeindefinites: (120) 〚koe-〛= P fx(P), x=speaker According to her, “the only difference between the lexical entries of specificity markers (e.g. -to and koe- indefinites) concerns the binding of the implicit argument x. The implicit argument of koe must be bound by the speaker, yielding identifiability by the speaker and necessary wide scope”. However, the denotation in (120) does not predict the interpretation that the indefinite receives in (121): (121) Ona rasskazala ob etom koe-komu iz svoih druzej. She tell-PAST-PERF about it to some of her friends ‘She told some of her friends about it’. (i) There were SEVERAL friends she talked to (ii) ? There was some (known to me, the speaker) SINGLE friend of her that she talked to The predicted reading is the one in (ii); but this reading is hard to get (it would have to refer to the ‘secret person who we both know but don’t want to call by name’). The preferred way of interpreting the sentence above is as in (i) when there were several people Masha talked 108 to. Interestingly, this ‘multiple referents’ interpretation does not occur with -to indefinites (unless the context somehow calls for the plural interpretation): (122) Ona rasskazala ob etom komu-to iz svoih druzej She told-PERF about it to some of her friends ‘She told someone among her friends about it’ / ?‘She told some of her friends about it’. (i) There was ONE specific friend of her who she talked to, but I don’t know or care who it was (ii) There were SEVERAL specific friends she talked to The interpretation in (ii) is possible but requires a special context in order to be supported – for instance, a continuation like “I don’t know who it was, but I know that several people are familiar with her story now”. The sentence in (121) with the koe- indefinite receives this plural interpretation “for free”. Possibility of (preferred) plural interpretation of koe- indefinites is discussed by Paducheva (1985) and Sheliakin (1986). According to them, koe- is a “weak definite”. It is different from a definite expression, because for a real definite, the speaker wants the addressee to recognize the object/person (s)he is referring to. For a “weak definite”, the speaker also refers to some particular individual(s) but does not want it to be recognizable for the addressee (Paducheva 1985). For Paducheva, koe- indefinites always have the same semantics but show variation in their functions. For her, a koe- indefinite always refers to an individual object that is known to the speaker but supposedly not known to the listener. The ways of realization of this semantics, however, depend on when the indefinite is used. She distinguishes two possible usages 109 (functions) for koe- indefinites: a so-called introductory usage, which in principle allows for both singular and plural interpretation of the indefinite; and so-called non-introductory usage where only plural interpretation is possible. Paducheva does not define these contexts, but provides examples of them. She uses the following sentence to illustrate an introductory usage: (123) Ja tebe koe-chto prines. Smotri, kakaja dynia. I you-DAT something(KOE) bring-PAST. Look what melon I brought you something. Look what a (nice) melon it is. Paducheva (1985) does not discuss the issue of how the singularity of the indefinite is perceived from the context like (123). The morphology doesn’t help – the indefinite is not marked with either singular or plural morphology. It is clear though that at the stage of the indefinite itself the addressee cannot make a decision about the singularity / plurality of its referent. The example in (123) stays ambiguous between singular and plural interpretation of “something” until the second sentence is uttered. Nothing prevents a plural interpretation; if the second sentence was Look at these melons the koe- indefinite in the first sentence would remain possible and grammatical. So, the indefinite by itself is unmarked for number both grammatically and semantically. She considers a ‘secret definite’ (a usage when the object that the koe- indefinite refers to is known to both the speaker and the listener, but the speaker does not call its name to indicate that this is some kind of a secret) a particular case of an introductory usage of the indefinite. Paducheva does not specify what types of contexts facilitate non-introductory usages of koe- indefinites, but they seem to be the cases when the speaker does not present a direct presentation of the object / does not provide a name of the object immediately after it was first 110 introduced by the indefinite. She specifically notes, however, that in their non-introductory usage koe- indefinites can only receive a plural interpretation, as illustrated by the following example: (124) Koe-kto ushel iz derevni na zarabotki. someone(KOE) go-PAST-SG from village to making-money ‘Someone (=some people) left the village to look for better money’. The verb in (124) is marked by a singular ending; the subject of the sentence is not grammatically marked for number: it’s a combination of koe+who, and, like English, Russian has a default singular agreement for wh-words (e.g. I see that at least 10 students are not in the class and I ask, “Who is absent today?”). So, nothing in the grammar of the sentence can be responsible for the plural interpretation of (124). However, the plural interpretation is strong enough to conflict with obvious marks of singularity: an attempt to include an indication to a singular person leads to ungrammaticality of the sentence (125a), even though the same change does not make a corresponding -to sentence ungrammatical (125b) below. Both (125a) and (125b) are grammatical with a different interpretation, when odin (one) means alone, by himself, separately. With this usage of one (125a) is a good sentence, but the koe- indefinite is interpreted as plural: (125) a. Koe-kto odin ushel iz derevni na zarabotki. Someone(KOE) one go-PAST-SG from village to making-money Intended: *‘Someone, one person, left the village to look for better money’. ‘Some people left the village by themselves to look for better money’. 111 b. Kto-to odin ushel iz derevni na zarabotki. Someone(TO) go-PAST-SG from village to making-money ‘Someone, one person, left the village to look for better money’. ‘Someone (or some people) left the village by themselves to look for better money’. Importantly, the second interpretation of (125b) allows for both singular and plural reading of the indefinite, since it is grammatically unmarked for number. This is not a problem for a choice-functional interpretation of -to indefinites: a choice function can choose a single individual or a ‘plural individual’ (a group of individuals). Most of the sentences with -to indefinites are unambiguous for singularity / plurality, because they grammatically encode the information about number: (126) Vchera on vstretil kakuju-to devushku / kakix-to devushek. Yesterday he meet-PAST-PERF some(TO)-SG girl-SG / some(TO)-PL girl-PL ‘Yesterday he met some girl / some girls’ The only cases of ambiguity for -to sentences are who-to (someone) and what-to (something) cases where either number is possible in principle; and only the context can disambiguate between a singular and a plural interpretation of the indefinite. If koe- indefinites are choice-functional in Geist’s way, the same ambiguity is expected for koe-who and koe-what indefinites (in her denotation, it’s the speaker who identifies the indefinite, but there is nothing in the identification process that requires that only a plural or only 112 a singular individual has to be chosen). However, the singular reading is banned and attempts to facilitate it by contextual means fail. At the same time, a singular interpretation of koe- indefinites is not impossible in principle, we obtain it in cases of ‘introductive’ usage of the indefinite. Then the question is: are different “usages” in fact instantiations of the same lexical item or is it a ‘different koe-’? For Paducheva (1985), there is one lexical item with one meaning and two functions. However, it is not clear what meaning it is why it is so context-dependent. Paduchava’s definition of koe- indefinites as referring to individual objects that are known to the speaker but supposedly not known to the listener does not account for the different “usages”. A way to formalize the notion of a “speaker-defined” indefinite is to present it as a choice-function. But the choice-functional denotation suggested by Geist can only explain ‘introductive’ koe-, it cannot account for the non-introductive one. 4.2 An alternative approach A different, set-based, approach towards koe- pronouns was initially proposed by Sheliakin (1986). For him, a koe- indefinite indicates a part / limited amount (minimally – a single representative if the elements of the set can be counted in principle) of (elements of) a contextually provided set, and these individuals are not known to the listener but known to the speaker. The crucial property of koe- pronouns, according to Sheliakin, is to show that out of all elements of the given set of individuals only a subset of them is relevant for the speaker and the subset is always smaller than the available set. The property to indicate a (limited) subset of a set, according to Sheliakin, can be realized by these indefinites in two forms: 113 A. Adjectival forms koe-kakoj (‘some-which-Masc.Sg’), koe-kakaja (‘some-whichFem.Sg’), koe-kakije (‘some-which-Pl’), etc. , which are able to mark number in their forms, are normally used with plural morphology and semantics (127a); singular koe-kakoj sounds very odd (127b); a pronoun one has to be used instead (127c): (127) a. Mne jesh’e nado zabrat’ koe-kakije vesh’i. I-DAT still need to pick-up some(KOE)-PL thing-PL ‘I still need to pick up some things.’ b. #Mne jesh’e nado zabrat’ koe-kakuju vesh’. I-DAT still need to pick-up some(KOE)-SG thing-SG I still need to pick up some thing. c. Mne jesh’e nado zabrat’ odnu vesh’. I-DAT still need to pick-up one-SG thing-SG ‘I still need to pick up one thing.’ B. Nominal forms koe-kto/chto etc. which are not grammatically marked with number, can refer to both plural and singular items. Sheliakin (1986) illustrates the possibilities of singular and plural interpretations by examples which would, according to Paducheva (1985), fall into different categories (introductive and non-introductive contexts), but he never discusses the important differences between the examples (i.e. that the former ones allow for both singular and plural interpretations and the latter can never be interpreted as singular). For him, a single individual is just a particular (minimal) case of a subset, and then a singular interpretation of the 114 indefinite is possible as long as the context and the grammar (e.g. adjectival morphology) do not call for the plural interpretation. As we have seen, this is not always the case. There are some other limitations on the singular interpretation of koe- indefinites, as demonstrated by Paducheva’s example, repeated below for convenience: (128) Koe-kto ushel iz derevni na zarabotki. Someone(KOE) go-PAST-SG from village to making-money ‘Someone (=some people) left the village to look for better money’. Contrary to what Sheliakin claims to be the nature of koe- items (they indicate a limited amount, minimally – a single representative – of elements of a set), the example in (128) demonstrates that “a single representative” cannot always satisfy a koe- item, i.e., (128) is not felicitous if only one person left the village. Thus, a koe- indefinite (at least in its nonintroductory usage) has to mean “more than one”. On the other hand, the sentence is not felicitous if the majority of people left the village either: (129) #Koe-kto ushel iz derevni na zarabotki. Na samom dele pochti vse ushli na zarabotki. Someone(KOE) go-PAST-SG from village to making-money. In fact almost all go-PAST-PL to making-money ‘Someone (=some people) left the village to look for better money. In fact, almost all the people left’. The data in (129) and (130) shows that koe- does not simply indicate some limited amount, it has a proportional value; it can refer to some number of the individuals in the given set that is larger than one member but it cannot be as big as the majority of the set. 115 Since koe- items are indefinites, they are used when the speaker does not intend to give the exact number (e.g. “three out of ten”). The presence of explicit numerical information in a sentence with a koe- indefinite renders it ungrammatical (130b); -to indefinites, however, can be explicitly counted in similar contexts (130c): (130) a. Ja kupil koe-kakije gazety. I buy-PAST some(KOE) newspapers ‘I bought some newspapers (and I know which ones but I don’t want to tell you yet)’. b. Ja kupil (*dve) koe-kakije (*dve) gazety. I buy-PAST (two) some(KOE) (two) newspapers Intended interpretation: ‘I bought two newspapers and I know which ones but I don’t want to tell you yet’. c. Ja kupil dve kakix-to gazety / kakije-to dve gazety. I buy-PAST two some(TO) newspapers / some(TO) two newspapers ‘I bought two newspapers and for each one I don’t care or know what they were / I bought newspapers the names of which I don’t know or care about and there were two of them’. This indefiniteness in respect to exact numbers makes calculation of the proportional value of koe- items problematic. However, it is possible to test speakers’ intuitions about “what seems plausible” in respect to the proportions of the set with koe- indefinites. Many native speakers have very similar judgments about what part of the set koe- refers too. A small part of the test materials with the “standard” responses is shown below: 116 (131) Scenario: There are 10 students in the class. The teacher utters the following sentence. Indicate whether this sentence is felicitous in the cases listed below and explain why: Koe-kto sdelal eto zadanije pravil’no. KOE-who did this assignment correctly ‘Some people did this assignment correctly’. (a) One student did it correctly. NO! The teacher, then, would say odin student (one student)* (b) Two (three, four) students did it correctly. YES! (c) Five students did it correctly. Probably NOT. This is too many… This would be polovina (half) or mnogije (many) (d) Six, seven, eight, nine … NO! It would be mnogije studenty (many students) or bol’shinstvo studentov (most of the students) (e) Ten students did it correctly. NO! It cannot mean “all”. NB: There are speakers who choose the answer in (a) as a possible option as well as the answer in (b). It seems that an adequate proportional value for koe- items is “more than one but less than a half of representatives of the given set of individuals”. Since these individuals are “known to the speaker”, i.e. epistemically specific, the denotation of koe- is based on choice function and includes the proportional value in such a way that the value of the function is in between one and a half of the set in the following way: 117 (132) 〚koec-〛= λP: 1 < |fspeaker (P)| < 1/2 |P ∩ C|. fspeaker P Obviously, this denotation will not hold for “single individual” cases; this is good, because we need to make sure that the case of (131a) is excluded from valid usages of koeindefinites (we will come back to why some native speakers find (131a) plausible). But it also excludes all “introductory” cases of koe-, which, in fact, can refer to both singular and plural individuals depending on what or who exactly the speaker “introduces”. In order to figure out what makes the “introductory” cases different from the rest of koescenarios, let’s see whether the same sentence can be uttered with different functions and compare the usages. It is not impossible for a non-introductory sentence, say, for (131), to become an introductory one as long as the speaker actually introduces the referent in some way, as illustrated below: (133) A teacher made a mistake – he assigned a homework which was supposed to be given to the students a month later. Since the material had never being covered in class, the students had no clue how to do the assignment. Only one student somehow figured it out. Apologizing for the wrong assignment, the amazed teacher says: Koe-kto sdelal eto zadanije pravil’no. Dmitry, mne bylo ochen’ prijatno prochitat’ vashu rabotu! KOE-who did this assignment correctly. Dmitry, me-DAT was very pleased read-PAST-PERF your work! 118 Someone did this assignment correctly. Dmitry, I was very pleased to read your paper! This is a classic case of an introductory usage, when the indefinite is immediately followed by presenting the actual object of its reference. The name does not have to be pronounced at all, even in the continuation of the discourse. Usually it is enough to give some hints to the audience: in the case of (133) it is enough to point, wave or smile at the smart student instead of making direct reference to him. In the case of pointing, the actual utterance is exactly the same as in (131) – at least in writing; the intonation in the “introductory” case will differ – koe-kto (‘someone’) is more likely to be stressed or have a pause after it. So, in the actual speech situation the two usages are easy to distinguish. In written and out-of-the-blue context, however, (131) can be ambiguous between the two usages. The speakers who interpret it in an introductory way report plausibility of (131a). These speakers may also interpret the situation in (131b) in the same way; it is not impossible that there are two/three smart students who did the assignment, and the teacher can point/smile/wave at these students when uttering the sentence. At this point it is not too late to hypothesize that koe- in Russian is ambiguous, or there are two different instantiations of koe-: one (introductory koe-) is simply choice functional as suggested by Geist (2008), and the other one (non-introductory koe-) is proportional and has a denotation given in (132) above. Then the introductory examples of koe-could be described in the following way: (134) Koe-kto sdelal eto zadanije pravil’no. ‘Someone (namely Dmitry / namely, Ivan and Petr) did this assignment correctly’ 119 CH(f) ʌ f(student(s)) did the assignment correctly. However, the ambiguity hypothesis has several problems. First of all, as the judgments of the native speakers prove, referring to a single individual is the only exceptional feature in the behavior of koe- indefinites in their introductory usage; they still cannot refer to half or more than a half members of the set. The people who found (131a) a valid situation for the utterance still protested against the situations (c), (d) and (e). Nothing in the simple choice-functional denotation prevents introducing six out of ten students that “the speaker has in mind”. It means that a choice-functional denotation for koe- by itself cannot account even for introductory cases; we still need to explain the proportional value introduced by the indefinite. Secondly, even in the uncountable introductory contexts, as in (135 below), koeexpressions do not simply indicate “the thing that the speaker has in mind”. Even these contexts have an extra piece of information “a little bit, not everything, less than you may expect” etc. Not only are koe- indefinites epistemically specific, they are also “modest”. In saying (123) the speaker means that he doesn’t want the addressee to overestimate how much he found out or collected: (135) Ja uznal / sobral koe-chto. Slushaj… I learn-PAST collect-PAST something(KOE). Listen-IMP ‘I found out / collected something. Listen…’ The information that may follow may or may not prove that the amount and the importance of the information / the amount and the importance of the material collected is 120 actually not that insignificant; but the speaker doesn’t want to reveal its importance yet, it’s a part of the secret; so, he utters (135) pretending what he found out / collected is very insignificant, it’s probably not as much as was expected from him and he could had done better. This is not necessarily true, but that’s what he wants his addressee(s) to believe for the moment. It is not clear how this “modesty” is can be predicted by assuming the choice-functional nature of koe- indefinites by itself. Besides, proposing two different instantiations of koe- goes against the intuitions behind the previous studies: Paducheva claims that introductory koe- and non-introductory koe- are “different usages” of the same item and depend on the context; Sheliakin analyzes “singular” koe- as a particular and minimal case of his “limited amount” interpretation. With this in mind, providing one denotation which can be somehow further limited by the context seems to be the most reasonable solution. Such a denotation needs to allow for both usages of koe- i.e., it needs to correctly describe an epistemically specific indefinite which refers to a subset of a (contextually) given set, and the subset is greater or equal to “1” but smaller than “1/2 of the set”. There are two ways of approaching this problem. On one hand, we can modify the initial denotation in (132) into a weaker one as in (136) below and give the indefinite the flexibility we need to account for its behavior in different usages: (136) 〚koec-〛= λP: 1 ≤ |fspeaker (P)| < 1/2 |P ∩ C|. fspeaker P Then we will have to explain why non-introductory usages ban the “greater or equal to 1” relation, and limit it to “greater than 1”: 121 The other approach would be to stick to the “stricter” / stronger version of the denotation given in (132) but allow for it to be weakened when the addressee has to accommodate the meaning of the sentence and the rest of the context – namely, the following sentence with a definite, pointing / smiling at a single individual, emphasis on the koe- item, intonation etc. In principle, both approaches would work. However, the first suggestion seems more grounded to me; I will show some evidence in its favor. So, I claim that the denotation of koe- is a weak one, as in (136) above, it is satisfied if a single member of the set is chosen; the requirement “more than one” comes from other language factors. First of all, Sheliakin is not quite right claiming that adjectival forms of koe- indefinites do cannot be used with singular morphology. In some cases koe- items sounds very natural with singular morphology. Recall the example he uses to demonstrate this limitation: (137) a. Mne jeshe nado zabrat’ koe-kakije vesh’i. I-DAT still need to pick-up some(KOE)-PL thing-PL I still need to pick up some things. b. #Mne jeshe nado zabrat’ koe-kakuju vesh’. I-DAT still need to pick-up some(KOE)-SG thing-SG I still need to pick up some thing. c. Mne jeshe nado zabrat’ odnu vesh’. I-DAT still need to pick-up one-SG thing-SG I still need to pick up one thing. 122 An important fact about this example is that it is possible to use the noun “thing” in both singular and plural forms; when the speaker wants to make a statement involving an epistemically specific indefinite and s(he) refers to plural things, s(he) will use a koe- indefinite in the plural form and a plural noun; if s(he) refers to a singular thing, s(he) will use a one indefinite with a singular form of the noun. Mixing the two things, as in (116b) is not acceptable. However, nouns like “furniture”, “literature” etc. are grammatically singular, but they cannot be identified with a single individual and thus “one furniture” is ungrammatical. If a speaker wants to make a statement involving an epistemically specific indefinite with nouns like these, the only grammatical way to do so is to use a singular morphology with a koe-indefinite: (138) a. Ja nashla koe-kakuju prilichnuju mebel’ / literaturu po etoj teme. I found some(KOE)-SG decent furniture / literature on this topic I found some decent furniture / some literature on this topic. However, if a specific piece of furniture or literature is mentioned, koe- is ungrammatical with singular morphology and one+NP has to be used instead: (139) a. #Ja nashla koe-kakoj prilichnyj divan / zhurnal po etoj teme. I found some(KOE)-SG decent couch / journal on this topic I found a decent couch / journal on this topic. b. Ja nashla odin prilichnyj divan / zhurnal po etoj teme. I found one decent couch-SG / journal on this topic I found a decent couch / journal on this topic. 123 c. Ja nashla koe-kakije prilichnyje divany / zhurnaly po etoj teme. I found some(KOE)-PL decent couch-PL / journal-PL on this topic I found some decent couches / journals on this topic. So, adjectival koe- indefinites are not, in principle, incompatible with singular morphology; (138) demonstrates very natural cases of morphologically singular koe- pronouns. Despite the singular ending, however, (138) is still mostly naturally perceived as referring to several pieces of furniture or literature: (138) can be paraphrased as (139c) but not as (139b); because if the speaker meant one item, s(he) would have used “one couch/ journal” instead of using mass nouns “furniture” and “literature”. The Russian system of indefinite pronouns contains two competing epistemically specific indefinites: an unambiguously singular one+NP and an ambiguous koe- which can, in principle, choose one or more individuals from the given set (if the set contains countable items). In this situation of a competition, when a speaker means a singular person, in principle, s(he) has a choice between (140a) and (140b): (140) a. Koe-kto ushel iz derevni na zarabotki. Someone(KOE) go-PAST-SG from village to making-money b. Odin chelovek ushel iz derevni na zarabotki. One person go-PAST-SG from village to making-money ‘One person left the village to look for better money’. The former choice could mean “one person” or “several people” because the semantics of koe- indefinites allows for any number of individuals as soon as it is smaller than half of the 124 contextually available set. The latter sentence unambiguously refers to one person. In this situation of competition, if the speaker still chooses to utter (140a) it means s(he) doesn’t have enough information to make a stronger statement of (140b). When the sentence in (140a) is interpreted, the addressee relies on the choice made by the speaker, s(he) trusts that if the speaker had one person in mind, s(he) would say (140b). This competition between the two indefinite pronouns forms a strong bias towards a plural interpretation of koe- indefinites. The singular interpretation is possible but it needs to have a strong support from the context – the addressee needs to make sure that despite the fact that the speaker did not use the available one+NP form, s(he) still has a singular individual in mind. This is exactly what happens in the cases of so-called ‘introductory usage’ when the addressee observes a singular object. 4.3 Conclusions Thus, koe- indefinites are not ambiguous; they are always proportional and always epistemically specific. They can refer to any part of a given set that is less than a half of it, minimally, to one individual. In the cases where the context does not provide support for a singular interpretation of the indefinite, they are always understood as plural, because there is a competing lexical item that has an unambiguously singular interpretation. Some indirect support in favor of the uni-semantics analysis of Russian koe- pronouns comes from cross-linguistic evidence on proportional indefinites (e.g. Matthewson’s (2009) analysis of a proportional quantifier nukw in Lillooet Salish which has several distinct functions / usages). I will address some cross-linguistic issues in respect to Russian indefinites in the next chapter. 125 CONCLUDING REMARKS AND QUESTIONS FOR FUTURE RESEARCH One of the goals of this dissertation was to study particular properties of the four different sets of Russian indefinite pronouns and provide a semantic account for each of them that would be able to naturally explain their distribution. In the chapters above I have studied the four types of indefinites in detail, described their semantics, and showed that they are all very different in nature: -to indefinites are always choice-functional, nibud’ and -libo indefinites induce domain widening, and koe- indefinite pronouns are epistemically specific and proportional. I also demonstrated that the differences in the distribution of the four types of indefinite pronouns are due to their semantics. However, the goal of accounting for the semantics and distribution of these indefinites was not the only thing I cared about in this dissertation. I think it is also very important to find out how different types of Russian indefinites relate to each other combining into a system. The analysis done in the four previous chapters demonstrates that the semantics and behavior of each set of the indefinites depends on the other parts of the system and on complex relations between its elements. -To and -nibud’/libo indefinites are often found in complementary syntactic distribution because they are inherently different semantically. -To indefinites are instantiations of a parameterized choice function. When the function chooses an individual out of the salient individuals available in the context, its implicit parameter is not bound by anything (or bound only by the immediate context). If the choice function is in the scope of an operator that can bind the implicit parameter (for instance, by always), the choice-functional indefinite may be 126 interpreted as seemingly non-specific, acquiring an apparent narrow (pseudo-)scope. This does not make -to indefinites ambiguous; they are still inherently specific and never become truly quantificational. The quantificational function is left to their competitors in the realm of indefiniteness: to bare noun phrases in their indefinite interpretation (in Russian, lacking definite/indefinite articles, a bare NP is ambiguous between definite and indefinite; this ambiguity is normally resolved by the context) and to “specialized” domain widening indefinites with suffixes -nibud’ and -libo. The latter indefinites are always non-specific, and that is what makes them different from -to indefinites; but they also induce the domain-widening effect that distinguishes them from bare NP indefinites and explains their peculiar syntactic behavior (they need a licenser that satisfies the requirement of their presupposition and provides them with available possible worlds other than the evaluation world). Despite the domain widening effect, -nibud’ and -libo indefinites allow for exceptions and never become truly universal. The place of a universal quantifier is occupied by different lexemes: by free-choice ugodno, liuboj and kto by to ni bylo items. The ability of -libo items to develop a semantics of no-exceptions in some contexts shows that these indefinites can, in principle get very close to universal quantifiers, but having in the system “specialized” free-choice items blocks a possibility of developing true universality for -nibud’ and -libo pronouns. The differences between the behavior of -nibud’ and -libo pronouns with respect to the degree of universality are due to another, stylistic in nature, property of the system: only a formal-style -libo but not a stylistically neutral -nibud’ is able to function as a universal in comparisons and under negation, because the available truly free-choice items do not meet the requirements ‘formal enough but not bookish and obsolete’ that -libo indefinites satisfy. 127 The indefinites with prefix koe- differ from both -to and -nibud’ pronouns in being epistemically specific – i.e. known to the speaker. Since koe- indefinites and their nonproportional epistemically specific indefinite odin/odna/odno (‘one’) refer to someone / something known to the speaker, this interpretation is blocked for -to indefinites (although this interpretation is not, in principle, impossible for choice-function items). The competition between the two types of epistemically specific indefinites – an ambiguous for number koe- and an unambiguously singular epistemically specific indefinite odin/odna/odno (‘one’) leads to strengthening the proportional requirement of the denotation of koe- indefinites from less than a half to more than one but less than a half members of the given set. Thus, it is the internal connections of elements of the whole system that make it work in the way it does and explain the unique features of each of its element. In combination, the four sets of indefinite pronouns and other lexemes (free-choice items etc., bare NPs etc.) cover the semantic space of indefiniteness in a unique way. One more important point brought by the Russian indefinites is that they present a graphic demonstration of non-homogeneity of semantic indefiniteness; the different lexemes represent (and thus make distinct) different sides / aspects of indefiniteness. Russian has separate lexical items to represent three subparts of the space of semantic indefiniteness – complete indefiniteness with the effect of domain widening is realized by -nibud’ and -libo pronouns; wide-scope indefinites are presented by -to pronouns; and epistemically specific proportional indefinites are realized by koe- pronouns. It does not mean that these three areas are the only subparts of the space of semantic indefiniteness, but they are the ones that are lexically encoded in Russian. Different languages lexically / morphologically 128 encode various aspects of indefiniteness and leave other aspects, not represented by any separate lexical or morphological unit, to be figured out based on the context. This causes well-known cases of ambiguity. English indefinites with an indefinite article a, for example, are often analyzed as ambiguous between quantificational and referential / quantificational and choicefunctional interpretations. English any can be existential (polarity sensitive any), and universal (free-choice any). English has proportional “some of” as in the sentence Some of my friends are coming to the party tonight. A proportional meaning of English partitives is often connected with their obligatory definiteness. Matthewson (2009) notes: “Some of the NP requires prior mention of a contextually salient set of NPs” (p. 42). It seems logical – how can people discuss a part of something if no one presented to the discourse what this something is? Russian, however, does not necessarily require definiteness of a proportional expression. In Russian, a referent of a proportional koe- indefinite is not mentioned in the previous discourse, it is known to the speaker, but is truly indefinite for the addressee (except for the ‘secret definite’ cases). This seems to be a random clustering of semantic features – having epistemically specific indefinites being necessarily proportional (or vice versa). That might be the reason why linguists describing koe- indefinites often see only one part of their semantics – their epistemic specificity or their proportionality. Indeed, why does it have to be the case that proportionality is linked with epistemic specificity? However, there might be evidence that Russian is not unique in this respect. Matthewson (2009) analyses a proportional quantifier nukw in Lillooet Salish. She demonstrates that it has several distinct function (usages), which include (a) contexts where English uses other; these are contexts of familiarity, some previous discourse referent(s) must have been introduced or accommodated; (b) contexts referring to a part of definite (mentioned in 129 the previous discourse) individuals; (c) novel, out-of-the-blue discourse contexts. Matthewson argues that what unifies all these seemingly very different usages is that nukw always conveys the notion of ‘some but not all’ of some larger set. Matthewson never discusses whether or not nukw is epistemically specific – it is impossible to figure it out without judgments of native speakers and tests like the ones Kagan (2008) uses to prove epistemic specificity of koeindefinites (repeated below for convenience): (141) a. #Koe-kto pozvonil, no ja ne znaju, kto eto byl. someone called but I NEG know who this was Someone called, but I don’t know who it was. b. #Dima razrabatyvaet koe-kakoj proekt, o kotorom ja ne imeju ni malejshegko predstavlenija. Dima works-out some project about which I NEG have not slightest idea Dima is working on some project which I do not have a slightest idea about. c. Ivan koe na kom ženilsja; #ponjatija ne imeju, na kom imenno. Ivan on-someone married idea NEG have(1st sg) on who exactly Ivan has married someone; I have no idea who exactly it is. (examples from Kagan, 2008) If a nukw-phrase must denote a proper subset of the set of contextually salient individuals satisfying the NP denotation, it is not clear how its appearance in novel discourse contexts can be explained. Matthewson accounts for the cases of out-of-the-blue nukw claiming that in these cases the larger set is the context C that can be vacuously large for discourse initial contexts. A reference to a member (or members) of this unlimited set is perceived as an out-of-the-blue 130 usage of nukw. She notices that this claim “obviously has wider-ranging implications which deserve attention in the future work”, but the idea does not seem implausible (p. 41). The analysis of Russian koe- indefinites, however, may highlight a different possibility. If the tests for epistemic specificity work for the Lillooet Salish proportional quantifier, it might turn out that it is not definite, but, instead, epistemically specific (speaker identifiable). Then brining such an indefinite in a discourse is not completely out-of-the-blue. It is out-of-the-blue for the addressee, but not for the speaker. It is impossible to tell without testing for epistemic specificity, but there are pieces of data in Matthewson (2009) that hint in this direction. She mentions that discourse-initial uses of nukw are common with plural DPs, and are less common with singular DPs. However, when consultants are presented with sentences with a singular DP, they do not seem to have problems understanding them. In the pre-print version of her paper 17 Matthewson gives the following example. The sentence in (142), was elicited with a plural nukw-phrase (the examples below are cited with the numbers as in Matthewson (2009-pre-print)) and then presented to a third speaker with a substituted singular form of the same phrase as in (143) to ask for his judgment about the singular form there. He accepted the sentence with a comment “Yes, that was Leonard Sampson that told me that.” (142) wa7 tsut [i núkw-a úcwalmicw] k-wa-s tu7 nilh i IMPF say [DET.PL nukw-DET person] DET-IMPF-3POSS then FOC DET.PL cácl’ep-mec-a k’al’em-mín-itas kw s-wa7 i wa7 Fountain-people-DET wait-APPL-3PL.ERG DET NOM-be DET.PL IMPF 17 Available at http://faculty.arts.ubc.ca/lmatthewson/pdf/nukw.pdf 131 sáysez’-wit lh-us pipántsek play-3PL HYP-(IMPF)3CONJ summer ‘Some people say that it was the Fountain people who they waited for, to play [sports] during summer.’ [okay as beginning of story] (from Matthewson, 2009-pre-print, example #19) (143) wa7 tsut [ta núkw-a úcwalmicw] k-wa-s tu7 nilh i IMPF say [DET nukw-DET person] DET-IMPF-3POSS then FOC DET.PL cácl’ep-mec-a k’al’em-mín-itas kw s-wa7 i wa7 Fountain-people-DET wait-APPL-3PL.ERG DET NOM-be DET.PL IMPF sáysez’-wit lh-us pipántsek play-3PL HYP-(IMPF)3CONJ summer ‘Someone said that it was the Fountain people who they waited for, to play [sports] during summer.’ [okay as beginning of story] (from Matthewson, 2009-pre-print, example #39) This is the comment of the speaker that interests me the most in this example. According to the speaker, the sentence in (143) is good in a novel context, but the nukw-phrase has to refer to a person that the speaker knows (Leonard Sampson) in order for him to accept the sentence. In the print-version Matthewson (2009) does not cite the consultant’s explanation any more. But she preserves one more example of the speakers (but not necessarily the audience) familiarity with the referent in a novel discourse context: 132 (144) íkena7! wá7 [ta núkw-a xzum sk’ák’y’et] l-ta sácwemten-a! eek be [DET nukw-DET big spider] in-DET bathtub-DET ‘Eek! There’s a big spider in the bathtub!’ Consultant’s comment: “That sounds ok if you saw one.” Do you have to have seen one before? “No, because sometimes you see them so it’s not presuming.” (from Matthewson, 2009-pre-print, example #40; same data in Matthewson, 2009, example #42) The comments of the speakers seem to make the contexts correlate with what Paducheva (1985) would describe as an introductory context (function), when using an indefinite somehow refers back to something/someone you as a speaker know or are presenting at the moment of the utterance. This seems to be a close parallel between two languages that are so different from each other. Even if a further similarity is not to be found, this comparison still seems interesting. The behavior of Russian koe- indefinites in comparison to the proportional quantifier in Lillooet Salish and proportional indefinites in other languages brings up an important issue of comparative semantic research (is it a coincidence or is there any reason for clustering different semantic features together?) and may potentially become an attractive line of the future research. The results obtained in the analysis of the other Russian indefinites may be useful from a cross-linguistic perspective as well. I have shown, for example, that -to indefinites are never truly quantificational; they are specific even in the contexts where they seem to demonstrate an apparent intermediate and narrow scope. It proves that an ability of an indefinite to be perceived as having different scope option is not always a result of ambiguity of the indefinite; sometimes 133 it can be accounted for by the parameterized choice-function approach. This might be an important observation to consider when looking at indefinites in other languages (for instance, at English some indefinites). Analyzing -nibud’ and -libo indefinites as involving a presuppositional requirement to access possible worlds other than the evaluation world may be useful in considering domainwidening indefinites in other languages, especially if they demonstrate different behavior. What are the reasons for the differences (say, between, English any and Russian domain-widening indefinites)? Are they caused by the semantics of the domain-widening item by itself or by other language-specific factors like competition and division of functions with other lexemes, stylistic requirements etc.? All these questions, left beyond the scope of this dissertation, are very intriguing and deserve a special research in the future. 134 REFERENCES 135 REFERENCES Abusch, D. 1994. The scope of indefinites. Natural Language Semantics, 2, 83-135. Alonso-Ovalle, Luis and Menéndez-Benito, Paula. 2003. Some Epistemic Indefinites. Proceedings of the North East Linguistic Society 33, ed. by Makoto Kadowaki and Shigeto Kawahara, University of Massachusetts, Amherst: GLSA, pp. 1-12. Anderssen, Jan. 2006. Generalized Domain Widening überhaupt. In Proceedings of the 25th West Coast Conference on Formal Linguistics, ed. Donald Baumer, David Montero, and Michael Scanlon, 58-66. Somerville, MA: Cascadilla Proceedings Project Chierchia, G. 2001. A puzzle about indefinites. In Semantic Interfaces: Reference, Anaphora, and Aspect. Stanford: CSLI Chierchia, Gennaro. 2006. Broaden your views: Implicatures of domain widening and the logicality of language, Linguistic Inquiry, 37: 535–590. Contemporary Standard Russian Dictionary. Volumes 1-17. Moscow. 1948-1965. Cuto. Eugenija. 2004. Ruske neopredene zamjenice s cesticama. Strani jezici 33 (2004), 3-4: 323-329 Dahl, Osten. 1970. Some notes on indefinites. Language. 46: 33–41. Dayal, V. 1998. ANY as inherently modal. Linguistics and Philosophy. 21: 433-476. Explanatory Russian Dictionary. Edited by Dmitry Ushakov. Volumes 1-4. Moscow. 1935-1940. Geist, Ljudmila. 2008. Specificity as referential anchoring: evidence from Russian. In: Grønn, Atle (ed.): Proceedings of SuB12, Oslo: ILOS 2008, 151-164 Geurts, Bart 2000. Indefinites and Choice Functions. Linguistic Inquiry, 31:4, 731-738 Giannakidou, Anastasia. 2001. The Meaning of Free Choice. Linguistics and Philosophy 24: 659–735. Golosa. A Basic Course in Russian. Book 2. Third edition. New Jersey. 2003 Farkas, D. 2002. Varieties of Indefinites. Proceedings of SALT 12 136 Fodor, J. & I. Sag. 1982. Referential and quantificational indefinites. Linguistics and Philosophy, 5, 355-398 Hackl Martin 2000. Comparative quantifiers. PhD thesis, MIT Hagstrom, Paul. 2003. What questions mean. Glot International Vol. 7, No. 7/8, September/October 2003 (188–201) Haim, I. 1988. The semantics of definite and indefinite noun phrases. Outstanding dissertations in Linguistics. Garland Publishing Inc. New York & London. Hamblin, C.L. 1973. Questions in Montague English. Foundations of Language, 10, 41-53 Haspelmath, M. 1997. Indefinite pronouns. Oxford University Press. von Heusinger, K. 2002. Specificity and Definiteness in Sentence and Discourse Structure. Journal of Semantics, 19: 245-274 Ionin, T. 2008. An experimental investigation of the semantics and pragmatics of specificity. Proceedings of the 27th West Coast Conference on Formal Linguistics. Kadmon, Nirit, and Fred Landman. 1993. Any. Linguistics and Philosophy 16:353-422. Kagan, Olga. 2006. Specificity as speaker identifiability. Logic and Language 9. August 24-26, 2006. Handout Karttunen, Lauri. 1977. Syntax and semantics of questions 1(1), 3–44. Reprinted in Paul Portner and Barbara Partee, eds. 2003. Formal Semantics: The Essential Readings. Blackwell. Kratzer, Angelika. 1991. Modality. Chapter 23 in Arnim v. Stechow and Dieter Wunderlich (eds.): Handbuch Semantik/Handbook Semantics. Berlin and New York (de Gruyter), 639-650. Kratzer, Angelika. 1998. Scope or Pseudoscope? Are there Wide Scope Indefinites? In S. Rothstein (ed.) Events and Grammar. Kluwer. Dodrecht Kratzer, Angelika, and Shimoyama, Junko. 2002. Indeterminate Phrases: the View from Japanese. In The Proceedings of the Third Tokyo Conference on Psycholinguistics, ed. Yokio Otsu, 1-25. Tokyo: Hituzi Syobo. [Also available at http://semanticsarchive.net/Archive/WewNjc4Z/] Krifka, Manfred 1995. The Semantics and Pragmatics of Polarity Items. Linguistic Analysis 25(3-4), pp. 209–257. 137 Martí, L. 2007. Restoring indefinites to normalcy: an experimental study on the scope of Spanish algunos. Journal of Semantics, 24, 1-25. Martí, L. 2009. Contextual restrictions on indefinites: Spanish algunos vs. unos. In Quantification, Definiteness, and Nominalization ed. Anastasia Giannakidou and Monika Rathert. Oxford: Oxford University Press, 108-132. Matthewson, Lisa. 1999. On the interpretation of wide-scope indefinites. Natural Language Semantics 7(1):79. Matthewson, Lisa 2009. (pre-print version). An unfamiliar proportional quantifier. Volume edited by Anastasia Giannakidou and Monika Rathert. http://faculty.arts.ubc.ca/lmatthewson/pdf/nukw.pdf Matthewson, Lisa 2009. An Unfamiliar Proportional Quantifier. In Quantification, Definiteness, and Nominalization ed. Anastasia Giannakidou and Monika Rathert. Oxford: Oxford University Press, 23-52. Paducheva, E. B. 1985. Proposition and its relation to reality: referential aspects of pronouns semantics. Moscow. Science. Pete, I. 1975. Usage of Indefinite Pronouns in Contemporary Russian. Russian at School. V.2. Pereltsvaig, Asia. 2004. Negative polarity items in Russian and the “Bagel Problem”. Negation in Slavic. Bloomington, IN: Slavika Publishers. Pereltsvaig, Asia. 2008. “Russian nibud’-Series as Markers of Co-variation”. In West Coast Conference in Formal Linguistics Proceedings. Pp. 370-378. Reinhart, T.: 1997, Quantifier Scope: How Labor is Divided Between QR and Choice Functions, Linguistics and Philosophy 20, 335–397. Russian for foreign students. Moscow. 1960 Schwarzchild, R. 2002. Singleton Indefinites. Journal of Semantics, 19, 289-314. Shimoyama, Junko. 2006. Indeterminate Phrase Quantification in Japanese. Natural Language Semantics, 14:139–173. Seliverstova, O.N. 1988. Pronoun in Language and Speech. Moscow. Science. Sheliakin, M.A. 1986. Russian pronouns: meaning, grammatical forms and distribution. Tartu. The generic book. 1995. Ed. by Greg N. Carlson, Francis Jeffry Pelletier. The University of Chicago Press 138 Vsevolodova, M.V. and Yudina, L.P. 1963. Russian for Polish speakers. Moscow. Winter, Y. 1997. Choice functions and the scopal semantics of indefinites. Linguistics and Philosophy, 20, 399-467. Yanovich, Igor. 2006. Choice-Function Series of Indefinites and Hamblin Semantics. In Proceedings of Semantics and Linguistic Theory (SALT) 15, eds. Effi Georgala and Jonathan Howell, 309-326. Ithaca, NY: Cornell Linguistics Publications. Zwarts, F. 1995. Nonveridical contexts. Linguistic Analysis. 25: 286-312 139