The expression of definiteness in Russian within Functional Discourse Grammar

(1)

MA thesis

The expression of definiteness

in Russian within

Functional Discourse Grammar

University of Amsterdam

MA: Linguistics of European Languages: Slavonic

Submitted by: Richard Udes 5735203 Supervisors: Dr. A.V. Peeters-Podgaevskaja Dr. H.G. Olbertz

(2)

List of figures and tables

Figures

1. Summary of the various terms concerning identifiability and activation 16

2. The Topic Acceptability Scale 17

3. The realization of (in)definiteness in Russian 19

4. The syntactic interplay of information structure and definiteness 35 5. Hierarchy of (in)definiteness markers in Russian 39

6. Hierarchy of activation states 65

7. The revised Topic Acceptability Scale 65

8. The syntactic interplay of information structure and activation states 69

Tables

1. Most popular constituent orders per focused constituent per sentence type 28 2. Distribution of constituent orders according to definiteness statuses of subjects

per age group 48

3. Distribution of constituent orders according to definiteness statuses of subjects

per age group, after exclusion of each first sentence of each story 49 4. Distribution of direct object placement in transitives with one verb with regard

to definiteness status over age groups 50

5. Distribution of direct object placement in transitives with two verbs with regard

6. Distribution of placement of direct and indirect objects with regard

7. Distribution of the results for intransitives (with and without thetic sentences) 52

8. Distribution of the results for transitives 53

9. Distribution of preverbal placement of DOs containing either a definite bare NP

or a pronoun 53

10. Placement of constituents containing the demonstrative ėtot 'this' 54 11. Placement of constituents containing a possessive pronoun 54 12. Placement of constituents containing a marker of indefiniteness 55 13. Distribution of constituent orders with regard to definiteness statuses of subjects

(5)

14. Distribution of constituent orders with regard to definiteness statuses of subjects

and directional PPs 56

15. Distribution of constituent orders with regard to definiteness statuses of direct

and indirect objects 56

16. Distribution of constituent orders in the adult data with regard to activation states

of subjects and locative/directional PPs 57

17. Activation states of definite referents in SVP orders within the children's data 57 18. Distribution of constituent orders with regard to activation states of direct and

indirect objects in ditransitives 58

19. Distribution of constituent orders with regard to activation states of subjects

(6)

Acknowledgments

I would like to thank some people who have helped me through the completion of this thesis. Firstly, I would like to express my gratitude to my supervisors. My thesis would have not been possible without the continuous input and feedback from dr. Alla Peeters-Podgaevskaja. The same applies to the extensive discussions I have had with dr. Hella Olbertz, which ultimately led to a satisfying conclusion that fits the framework of Functional Discourse Grammar. Secondly, I would like to express my gratitude to the Netherlands Instititute in Saint Peterburg. Their research grant enabled me to do conduct my experiment in Saint Petersburg, and I received considerable help from the staff of this institute. I would also like to thank Bibi Janssen in particular, who gave me a lot of advice on experimental design and further helped me with conducting the experiment. Finally, I would like to thank my friends and family, who have always supported me and gave me advice whenever I needed it.

Additional comments

All the examples, figures and tables in this thesis are mine, unless otherwise stated. For the examples, the Leipzig glossing rules were followed:

(7)

Introduction

According to the The World Atlas of Language Structures Online, there is a total of 198 languages in the world that lack (in)definite articles (Dryer 2013). It is commonplace that most Slavic languages belong to this group, and Russian is no exception. The question of how it is possible to express the notion of definiteness in this type of languages, has been intriguing linguists to date. Indeed, the concept of definiteness (which in my view is centered around a speaker's and hearer's ability to identify a referent) is important to such an extent that it is expected to exist in practically every language. In the case of Russian, constituent order and case are most commonly mentioned as relevant aspects. Indeed, the relatively free constituent order and rich case system in Russian seem to at least partly encode definiteness in noun phrases. This has been supported by a number of studies (Nesset 1999; Brun 2001; Bayer 2005; among others).

Another term that is intimately related to definiteness is information structure. It deals with the distinction between topic and focus (or theme and rheme). Since the topic is usually known information to both speaker and hearer, and focus is about adding new information regarding that topic, they are commonly referred to as 'old' and 'new' information. This, however, is only partly true, which is why topic and focus will not be treated in this definition here. Information structure does indeed influence constituent order to a high degree. This already presents a problem, because it seems to indicate that constituent order is a marker for both information structure and definiteness. Consider the following examples (Dyakonova 2004: 93):

(1a) Priexali gosti

arrive.PST.PL guest.NOM.PL

'There arrived some guests'

(1b) Gosti priexali.

guest.NOM.PL arrive.PST.PL

'The guests have arrived.'

In (1a), the guests are in focus and indefinite, while in (1b) they are topical as well as definite, which is exactly how these categories often overlap. Note, however, that information structure and

definiteness are not the same thing and do not always correspond to each other. Still, it is generally assumed that they are both expressed through constituent order (Dyakonova 2004; Kallestininova 2007; among others). Evidently, both phenomena are inextricably connected, and information structure will continue to play an important role in the discussion.

(8)

Apart from constituent order, case and information structure, other aspects that have been claimed to be relevant are verbal prefixes and grammatical markers (pronouns, determiners, etc.). Despite all these observations, a coherent theory of definiteness in Russian is lacking. This is partly due to the fact that the phenomenon is very complex, which is already illustrated by the multitude of relevant aspects. It can, however, also be due to the absence of focus on one form of speech. For instance, it greatly matters whether an analysis involves written texts or oral speech. In Russian, the type of communication heavily affects intonation contours and constituent order, which in turn affect the expression of definiteness. Therefore, the main goal of this thesis is to formulate a coherent theory of definiteness in Russian. In order to facilitate achieving this goal, I will only focus on one form of oral speech, namely narratives.

In the discussion of definiteness, a psychologically adequate perspective seems essential. The key issue is the status of referents for both the speaker and hearer. The (assumed) knowledge of both interlocutors generally determines the usage of articles in English:

(2a) I have read the book. (2b) I have read a book. (2c) I want to read a book.

The definite article in (2a) presupposes that hearer and speaker are both aware of which book the speaker is talking about. It is, however, always about what the speaker assumes the hearer to know. (2b) and (2c) show two possible readings of the indefinite article. In (2b), the speaker finished the book by himself, so he naturally refers to a specific book, but the hearer is unaware of which book it is. Although in (2c) this reading is still available, it is also possible that the speaker is equally

unaware of what book is being referred to. The speaker then refers not to a specific book, but to a kind of object. In other words, he refers to the type, and not the token. This distinction of speaker knowledge is the general interpretation of specificity (Lambrecht 1994: 81; Hengeveld & Mackenzie 2008: 122).1

Apparently, it is not sufficient to merely consider morphosyntax when dealing with

1 As argued in Lyons (1999: 173) this is somewhat a simplification of the term:

There have been attempts to reduce both kinds of ambiguity to a single specific versus non-specific distinction, though they are generally held to be distinct. One might try to unify them by taking the pragmatic concept of the speaker having a particular referent in mind. On this view the vagueness between specific and non-specific is always present (Lyons 1999: 173).

Lyons distinguishes between opaque and transparent contexts, in which narrow-scope and non-referential readings are deemed to be non-specific, and wide-scope and referential readings specific. According to him, these distinctions and labels are, in fact, a simplification as well. Nevertheless, the present definition (described above) will suffice here.

(9)

definiteness. In the field of Generative Grammar, much research has been done on this topic, but Functional Discourse Grammar (FDG) seems a more suitable theory in this respect and will shed new light on this phenomenon. FDG is the successor of Functional Grammar (FG) and a

typologically based theory of language structure. The major advantage of FDG is its assumption that linguistic utterances are built up on four different levels and in a set order:

 The Interpersonal level, which accounts for pragmatics;  The Representational level, which accounts for semantics;

 The Morphosyntactic level, which accounts for morphology and syntax;  The Phonological level, which accounts for phonology.

Together, these levels form the Grammatical Component, which is connected to three other Components. Every linguistic utterance starts from the Conceptual Component, where the

communicative intention arises. The utterance is then formulated and encoded in the Grammatical Component. Note that Formulation and Encoding in the Grammatical Component always take place in correspondence with the history of discourse and environment that are formulated in the

Contextual Component. The utterance is then realized in the Output Component through writing, speaking or signing. Evidently, the Interpersonal Level and the Contextual Component are essential to determine the status of referents for both interlocutors, while the Morphosyntactic Level is necessary to encode this information in the utterance. FDG allows us to analyze language material while taking into account all relevant aspects (Hengeveld & Mackenzie 2008: 1-24).

In addition to a clear theoretical framework, an experiment will also give further insight into definiteness in Russian. In order to learn to what extent and under which circumstances definiteness is expressed in the actual usage of Russian in narratives, an experiment has been conducted with both children and adults as participants. On the basis of four sets of pictures, participants either had to devise a story by themselves, or retell one which they had heard before. As a result, a database was created using the recorded stories of the participants. It was then analyzed in correspondence with the assumed theory. The decision to also include children is important. Children are considered to not take discourse context into account (Schaeffer & Matthewson 2005: 68-69). This results in a 'egocentric' view of the world, where only the speaker's knowledge matters. The question then remains, to what extent this is of influence on children's speech performance in relation to encoding definiteness. The expected divergence in performance could give more insight into the main aspects that determine the expression of definiteness.

(10)

general. A brief introduction to the relevant aspects of FDG is also provided, alongside which an explanation is given on how definiteness is represented in FDG. Chapter 2 covers definiteness in Russian within the theoretical framework of FDG. Each type of definiteness marking is considered, yielding my own theory. Chapter 3 consists of a description and the results of the experiment in all its facets. On the basis of the results and analyses, I refine my own theory in chapter 4, and give suggestions for further research.

(11)

1. Theoretical framework

In order to formulate a new theory on the expression of definiteness in Russian, all the relevant aspects need to be properly introduced. In section 1.1, a general overview of definiteness and

indefiniteness is given. In section 1.2, the theoretical framework of Functional Discourse Grammar is introduced. Finally, in section 1.3, a discussion follows on related but different categories, namely, specificity, information structure and accessibility/activation.

1.1 Definiteness and indefiniteness

In general, definiteness is considered to be a grammatical category that is encoded in noun phrases. However, the exact semantics it tries to evoke is not immediately clear. Since a wide range of descriptions on definiteness is available in the literature, no consensus has been reached on what definiteness exactly entails. As has been discussed in the introduction, determining the status of referents seems to be its core concept. In Lambrecht (1994), this is the focus as well: "The

grammatical category of definiteness is a formal feature associated with nominal expressions which signals whether or not the referent of a phrase is assumed by the speaker to be identifiable by the addressee" (Lambrecht 1994: 79). Identifiability, whether a referent is identifiable to the addressee by previous mention in discourse, is considered by Lambrecht to be the foundation of definiteness. The usage of indefinite NPs is thus justified when the speaker does not expect the addressee to be able to identify the referent. This is not the only relevant concept. As Lambrecht himself adds: "(...) the correlation between the cognitive category of identifiability and the grammatical category of definiteness is at best an imperfect one" (Lambrecht 1994: 79). Identifiability appears to not cover all possible expressions of definiteness. Other important concepts, such as inclusiveness (Hawkins 1978, Lyons 1999) and accessibility (Lambrecht 1994; Eppstein 2002), have also been introduced in the literature to explain different usage of definite NPs. In what follows, these alternative concepts will be discussed in more detail. Their importance notwithstanding, in this study, I consider identifiability to be the core characteristic of definiteness.2

Rijkhoff (2008) describes four contexts in which definite NPs are most commonly used. The first context involves the most typical usage, the anaphoric use (Rijkhoff 2008: 97):

(3) I just bought a book and a calendar. Surprisingly, the book was much cheaper than

the calendar.

2 The focus of this thesis on narratives justifies this decision. Indeed, context and the status of referents for the addressee are of great importance for narratives, which require proper introduction of referents.

(12)

Secondly, a referent can be directly available in a shared physical context (instead of in a discourse context). Situational or deictic use of definite NPs then occurs (Rijkhoff 2008: 98):

(4) Tell me - what do you see on the monitor?

The third type of context involves a more indirect way of reference. 'Bridging-cross-reference'3_can be at work, which yields associative usage of definite NPs:

(5) I am taking a Chinese class right now. The funny thing is that the teacher is not even Chinese himself.

The addressee is then able to make the connection between 'course' and 'teacher' without previous reference of the latter:

(...) after a previous linguistic mention of a class, the speaker can immediately talk of the professor, the textbook, the final exam. All members of the relevant linguistic community know that the set of things which make up a class typically include these (Hawkins 1991: 409).

Finally, there are referents that do not designate entities within a (non-)linguistic setting, but entities that are unique. On the basis of general knowledge, the addressee is capable of identifying these kinds of referents (Rijkhoff 2008: 98):

(6) The moon was very bright last night.

However, the notion of uniqueness is regularly incorporated in inclusiveness, a term that was introduced in Hawkins (1978). In Lyons (1999) it was assumed that uniqueness is merely a subcategory of inclusiveness, which covers still other contexts. In clauses, such as Move the sand

away from my driveway, the definite article refers to the object or mass in its totality. Consequently,

inclusiveness mainly comprises non-referential usage of plural and mass nouns with a definite article (Rijkhoff 2008: 69; Hawkins 1978: 159).

An alternative viewpoint is that under the various conditions that require the usage of definite NPs, the referent is made accessible in the addressee's mind. This is the notion of accessibility. Rather than identifiability, Eppstein (2002) considers accessibility to be the core of definiteness: "the definite article signals the accessibility of a discourse referent, or more precisely, the availability of an ‘‘access path’’ through a configuration of mental spaces" (Eppstein 2002: 335). Hence, the definite article in front of a noun creates a path towards the mental representation of the referent in

(13)

the addressee's mind. The activation of this representation can be achieved by a variety of means, as has been described above. Even though I consider identifiability to be the most important

characteristic of definiteness, accessibility and activation will also prove to be important for the theory to be proposed.

So far, mainly English examples have been used. In English, definiteness is expressed through contrastive articles, which either denote definiteness or indefiniteness. In Lyons (1999: 48), the explicit marking of definiteness is called an areal feature, which mainly characterizes the

languages of Europe and Middle East. However, within that area different systems also occur. Irish, Arabic and Bulgarian have only definite articles, while Turkish has an indefinite one. Apart from articles, definiteness can also be expressed through other means. In Hungarian, for example,

definiteness is marked on the verb. In Serbian/Croatian and Lithuanian, definite adjectives are used. Note that all these are overt grammatical means to express definiteness. Languages with no explicit markers of definiteness provide the real mystery, such as Russian. It is often assumed that in Russian constituent order and case are the primary markers of definiteness. This will be discussed in detail in chapter 2.

1.2 Definiteness and FDG

It has become clear that the realization of definiteness can differ both within and between languages. Therefore, Functional Discourse Grammar will be used as a theoretical framework, as it has been formulated in Hengeveld & Mackenzie (2008). This way, the explanatory capacity of the upcoming comparisons and conclusions will be enhanced. FDG consists of the Grammatical Component at the basis, which is connected to the Conceptual, Contextual and Output Components. The Conceptual Component marks the starting point of every communicative intention. The intention is then

transferred to the Grammatical Component, where its Formulation and Encoding take place. This is always done in accordance with the Contextual Component that stores all the language-relevant previous discourse information for both interlocutors. The Output Component then converts the grammatical information into an observable form (Hengeveld & Mackenzie 2008: 1-7).

The Grammatical Component is further subdivided into four levels of analysis. These are repeated here for convenience:

 The Interpersonal Level, which accounts for pragmatics;  The Representational Level, which accounts for semantics;

(14)

 The Phonological Level, which accounts for phonology.

The Interpersonal and Representational Levels are responsible for Formulation. The Interpersonal Level covers all aspects that are related to the goal of the speaker's communicative intention and the assumed knowledge of the addressee. The Representational Level provides the semantics. It is used by the speaker to establish a relation between linguistic units with the corresponding extralinguistic entities, which the speaker wishes to describe. Basically, the Representational Level involves designation, whereas the Interpersonal Level involves evocation. Finally, the Morphosyntactic and Phonological Levels are in charge of Encoding. On the basis of the information received from the Interpersonal and Representational Levels, first, a morphosyntactic representation is made, and after that a phonological one. The representation is then ready to be transmitted to the Output Component (See Appendix A for figures; Hengeveld & Mackenzie 2008: 12-14).

In order to explain the expression of definiteness in narratives in Russian, not all levels and components in FDG are equally important. For this study, the Contextual Component and the

Interpersonal, Representational and Morphosyntactic Levels within the Grammatical Component will be most significant. Unsurprisingly, in FDG definiteness is expressed on the Interpersonal Level. However, within the Interpersonal Level not all units are equally important for this study, which is why I will only discuss the Communicated Content (C1). This is all that the speaker tries to convey to the addressee and is organized as follows (See Appendix A for the full organization of the

Interpersonal Level as represented in Hengeveld & Mackenzie 2008: 49):

Organization of Communicated Content at the Interpersonal Level

(π C1: [ Communicated Content

(π T1: [...] (T1))Φ Ascriptive Subact (π R1: [...] (R1))Φ Referential Subact

] (C1))Φ Communicated Content

π = Any operator, expressed by grammatical or phonological means; Φ = Any pragmatic function.

Each Communicated Content consists of one or more Subacts. There are two types: Subacts of Ascription (T1) involve a speaker's attempt to evoke a property, whereas Subacts of Reference (R1) constitute an attempt of a speaker to evoke a referent. Note that between Subacts no hierarchical differences exist, they are all equal (Hengeveld & Mackenzie 2008: 46-68, 87-124).

(15)

Since Referential Subacts revolve around referents, they are strongly connected with

definiteness. In FDG, definiteness is determined by two operators that apply to the R-variable. They are two binary positions in the representation that indicate whether the addressee (identifiability (id)) and the speaker (specificity (s)) are familiar with the referent (+) or not (-). Consider the following representations of (2a) and (2b), which are repeated here as (7a) and (7b) for convenience:4

(7a) I have read the book.

(CI: [(TI) (+id RI: [+S, -A] (RI))(+id +s RJ)] (CI)) (7b) I have read a book.

(CI: [(TI) (+id RI: [+S, -A] (RI))(-id +s RJ)] (CI))

In both (7a) and (7b), RJ is the mental representation of 'book'. In (7a) both speaker and addressee are aware of which book is meant, which means that it is a definite NP. This why in the representation 'id' and 's' are both positive. However, in (7b) only the speaker knows which specific book is meant. For the addressee, this is likely to be new information, so here 'book' is indefinite. This is indicated in the representation with a positive 's', but a negative 'id' (Hengeveld & Mackenzie 2008: 122).

The Representational Level deals with designation, i.e. the semantics of a linguistic unit. Its layers are therefore defined according to the semantic categories they designate. Again, not all units are of equal importance to this study, only those that are organized as follows (See Appendix A for the full organization of the Representational Level and an example clause based on Hengeveld & Mackenzie 2008: 142):

Organization of States-of-affairs at the Representational Level

(π e1: State-of-Affairs [(π f1: [ Configurational Property (π f1: [σ (f2)Φ]) Lexical Property (π x1: [σ (x1)Φ]) Individual ] (f1): [σ (f1)Φ]) Configurational Property ] (e1): [σ (e1)Φ]) State-of-Affairs

σ = A modifier, which can be drawn from the lexicon or be internally complex.

The State-of-Affairs (e1) is characterized by the Configurational Property (f1) which includes Lexical

4 Referential Subacts involving personal pronouns are additionally encoded through [+/-S] 'involving the speaker or not' and [+/-A] 'involving the addressee or not'. Naturally, I only involves the speaker.

(16)

Properties (f2) and/or Individuals (x1).5 Individuals can be further characterized by Lexical Properties. Despite semantic aspects being crucial for any given clause, the Representational Level is not

directly involved in denoting definiteness. Nevertheless, the Representational Level will prove to be important in chapters 2 and 3. Note that the representation includes no conjugation and inflection. These are encoded on the following level (Hengeveld & Mackenzie 2008: 140-142).

In case a language (overtly) expresses definiteness, it somehow has to be encoded in the utterance. Therefore, logically, the Morphosyntactic Level is of importance. The following scheme shows how the Morphosyntactic Level is organized, with a Linguistic Expression consisting of at least one Clause. Each constituent can occur more than once (Hengeveld & Mackenzie 2008: 291):

Organization of the Morphosyntactic Level

(Le1: [(Cl1: [(Xw) (Xp1: [(Xw) (Xp2) (Cl2)] (Xp1)) (Cl3)] (Cl1))] (Le1)) Le = Linguistic Expression

Cl = Clause

Xp = Phrase (of the type x) Xw = Word (of the type x)

As has been stated earlier, the encoding of definiteness varies significantly between languages. In case of English, encoding definiteness takes place on Phrase-level, since it is expressed through articles. The Referential Subact arrives from the Interpersonal and Representational Levels to the Morphosyntactic Level, where it is then converted into a Noun Phrase. An article is then added, which indicates the referent's status in terms of definiteness (See (8) for an example; Hengeveld & Mackenzie 2008: 291, 378-381).

Despite this, it is assumed for Russian that the encoding of definiteness is done primarily through syntactic means, i.e. constituent ordering. Logically, syntactic encoding takes place on the Morphosyntactic level. Three absolute positions are (theoretically) always available for the

placement of units. Cross-linguistically, these are Clause-initial (PI_{), Clause-final (P}F_{), and} Clause-medial (PM_{). The Postinitial (P}I+1_{) and Penultimate (P}F-1_{) relative positions can only be filled when the} absolute positions are no longer available. Of course, with the addition of, for instance, a subordinate clause, a linguistic expression may become complex. In this study, however, I will (mainly) focus on the level of the simple Clause. The number of available positions and how the ordering of

constituents takes place is language-specific. For English, this can be illustrated with (8):

5 In Hengeveld & Mackenzie (2008) more semantic categories are distinguished that may be included in the Configurational Property. For the present study, only the two presented here are required.

(17)

(8) A man has read the book. IL: ML: PI _{(-id +s R} I) (Cli: [ (Npi: [ (Gwi: a (Gwi)) (Nwi: man (Nwi)) (Npi))Subj PM (fin_Vw i: has <SING> (Vwi))6 PM+1 (Vpi: inVwj: read <PSTPTCP> (Vwj)) (Vpi)) PF _{(+id +s R} J) (Npj: [ (Gwj: the (Gwj)) (Nwj: book (Nwj)) (Npj))Obj ] (Cli))

So far, the examples only dealt with the placement of hierarchically related units. However, the expression of definiteness is a pragmatic function, which outweighs hierarchical structure.7 Consequently, the expression of definiteness in Russian will involve the placement of

non-hierarchical units (Hengeveld & Mackenzie 2008: 311-316, 333-334). This will be discussed in the next chapter.

Finally, the Contextual Component will be discussed. In Hengeveld & Mackenzie (2008: 9), FDG is said to make "no effort tot offer anything like a complete description of the overall discourse concept." Therefore, the description of the Contextual Component in this book was fairly limited. Nevertheless, attempts have been made to give more importance to the Contextual Component. In a later article, Hengeveld and Mackenzie (forthcoming) elaborated further on the relation between grammar and context. They provided a more extensive description of the Communicated Content that I will follow in this study. In FDG, the Contextual Component constantly communicates with the whole Grammatical Component. Indeed, contextual factors can affect the grammatical information at every level. The Grammatical and Contextual Components thus always cooperate to generate correct output. Nevertheless, there are several constraints. Not all aspects of context are equally important: only contextual information that affects the grammar of a language is part of the Contextual

Component. Purely stylistic choices are, therefore, irrelevant. Furthermore, this means that the

6 The Grammatical Morpheme have is treated here as a Verbal Word (Vw), rather than a Verbal Phrase (Vp), because it can be placed independently of the main verb, as illustrated in (9):

(9) A man has already read the book.

7 Indeed, in the top-down and centripetal approach of FDG, it can be assumed that for the ordering of constituents, aspects that are relevant on a higher level are taken into consideration before lower classified ones. Identifiability and specificity act as operators on Referential Subacts at the Interpersonal Level, which is relatively high. This would imply that in languages that encode such features through constituent order, the placement of constituents according to identifiability and specificity takes place prior to placement of lower placed units and is, therefore, crucial for the structure of the clause.

(18)

Contextual Component is language-specific, since not every language is affected by context in the same way (Hengeveld & Mackenzie, forthcoming: 1-3).

The Contextual Component consists of four Strata, every Stratum of which corresponds with a Level from the Grammatical Component. Each Stratum covers Discoursal Information, but only the Formulation Strata (Interpersonal and Representational) additionally include Situational

Information. Situational Information comprises language-specific selection of relevant elements that are present in the speech situation. There are three different dimensions: speech participants, space and time. To illustrate the first dimension, in Russian, the gender of participants of the interaction is marked in finite verbs in the past tense. Logically, space and time cover their relative and absolute indications at the time of the speech situation. In addition, the physical world that is not directly included in the interaction can also affect the grammar, such as individuals or events (Hengeveld & Mackenzie, forthcoming: 4-5).

Discoursal Information consists of a series of pushdown stacks, which jointly comprises all the formulated and encoded information from the Grammatical Component. This should be regarded as a stack of elements, to which new elements can only be added on top, while older elements can only be removed from underneath. Consequently, the Discoursal Information on each Stratum contains an analysis at its own Level (for instance, the Interpersonal Stratum and an Interpersonal Level analysis). The most recent elements are placed on top. Lower items can disappear, which corresponds to actual memory decay in interaction. This set-up of the Contextual Component has two major advantages. Firstly, it allows for anaphoric and cataphoric reference within FDG. Secondly, it indicates a clear difference between given/old information (in the Contextual

Component) and new information (in the Conceptual Component). This will prove to be important in chapter 3.

Finally, (10) illustrates what stacking exactly entails. (10a) shows the actual utterances, (10b) the order of entities as they are stored in the Contextual Component at the Representational Stratum for concrete entities:

(10a) Iank met Maryl yesterday. Hek had just visited Annm, whom is looking after Maryl's dogn. (10b) (xk) (xl) (xk) (xm) (xm) (xl) (xn) (xk) (xl) (xk) (xk) (xm) (xl) (xl) (xl) (xk) (xm) (xk) In (10a), Ian is mentioned in the first clause. Despite the interruption of a second entity (Mary),

(19)

which causes Ian to descend one position down the stack, he is then referred back to in the second clause by means of a pronoun ('he'). The presence of a second more recent referent did not cause problems for referring to Ian anaphorically, since he was the only available male entity. However, the second reference to Mary turns out to be slightly different. 'Mary's dog' does not include a personal pronoun, but Mary's full name. Mary landed underneath Ann in the stack, which rendered it

impossible to refer to her anaphorically by means of a pronoun (Hengeveld & Mackenzie, forthcoming: 10-11).

1.3 Remaining relevant concepts: specificity, information structure and activation

The study of definiteness requires the explanation of three other important notions: specificity, information structure and activation. Specificity can be an exceptionally confusing term, since its characteristics are sometimes mistaken for those of definiteness and vice versa (Rijkhoff 2005: 86). The following description of specificity by Lambrecht shows what is mostly understood under specificity:

A 'specific indefinite NP' is one whose referent is identifiable to the speaker but not to the addressee, while a 'non-specific indefinite NP' is one whose referent neither the speaker nor the addressee can identify at the time of utterance (Lambrecht 1994: 81).

Like definiteness, specificity is a contrastive concept. However, it is only affected by the discoursal and situational knowledge of the speaker. Evidently, this is correctly expressed in FDG: in

representations on the Interpersonal Level, 's', which constitutes speaker knowledge, can either be positive or negative. In sum, the distinction between specificity and definiteness can be summarized as follows:

 Known by speaker and addressee (+id +s): definite, specific;  Known by speaker only (-id +s): indefinite, specific;  Known by neither speaker nor addressee (-id -s): indefinite, non-specific. Note that hearer's knowledge is presumed knowledge of the addressee by the speaker. Furthermore, as can be seen in the schema above, a definite NP is necessarily specific.8_{However, an indefinite NP} can be either specific or non-specific.

Rijkhoff & Seibt (2005) and Rijkhoff (2008) mention another important distinction. It concerns how (in)definite and (non-)specific referents are related to reality and how they are placed

8 This only applies to declaratives. In interrogatives, the fourth possibility is completely natural (Hengeveld & Mackenzie 2008: 122).

(20)

('grounded') in the world of discourse. In this respect, definite and indefinite NPs differ in that only definite NPs occupy a temporal-spatial position in the shared world of discourse, the so-called Common Ground. Initially, indefinite NPs do not occupy this position. Once they are introduced to the addressee, they start to occupy such a position and become definite. For specificity, the situation is slightly more complex. The difference between specific and non-specific NPs is that the former occupy a physical place in the world, which makes them groundable in the world of discourse. Hence, non-specific NPs are ungroundable. For example:

(11a) I want to buy a book. It is a thriller.

(11b) I want to buy a book. It has to be a thriller.

In (11a), reference is made to a specific existing book. Clearly, it occupies a position in time and space, and through reference in the world of discourse as well. No reference is made to a particular book in (11b). Notwithstanding the fact that it is possible to refer to 'book' in (11b) anaphorically in subsequent clauses (which explains the use of 'it'), it is not grounded in the world of discourse. As has been explained in the introduction, merely the type is being referred to, but not the token. In short, specific NPs can be grounded in the world of discourse because they occupy a position in time and space in reality, while non-specific NPs do not and thus cannot be grounded (Rijkhoff 2008: 71-74).

Information structure is also often related to definiteness, since they are both pragmatic concepts. In Lambrecht (1994) information structure is described as "the formal expression of the pragmatic structuring of a preposition in discourse".9_{From Lambrecht's perspective, a referent is a} topic when the proposition is construed as being about this referent. In contrast, Focus covers the information that designates the difference between 'presupposition' and 'assertion'. Consequently, topical information comprises known information, while focal information is 'unpredictable' and cannot be inferred from the context. This is why, as has been mentioned in the introduction, topics are usually associated with old information and foci with new information (Lambrecht 1994: 5-6, 118-119, 206-207).

Nevertheless, Lambrecht does not consider topic and focus to be complementary to each other. To illustrate this point: a clause is always required to carry some form of new information. Otherwise it would not be necessary to produce it. This is why each clause needs to have focal

9 Lambrecht distinguishes three main categories: 1) presupposition and assertion, structuring propositions in terms of what information is new to the addressee and what is not; 2) identifiability and activation, the former term being familiar, while the latter will be discussed later in this section; 3) topic and focus, which deal with the speaker's assumptions of the (un)predictability of the relation between propositions and their elements. The latter categories, topic and focus, are usually ascribed to information structure alone, which is what I will continue to do here as well.

(21)

constituents, but not necessarily topics (Lambrecht 1994: 206). Likewise, FDG does not consider topic and focus to be complementary to each other either. One of the following pragmatic functions can be assigned to Ascriptive and Referential Subacts: Topic or Focus.10_{In a Discourse Act where a} Subact fulfills one of these functions, the remainder of the Communicated Content forms the Comment or Background, respectively. In other words, instead of a Topic-Focus contrast, there are Topic-Comment (which Lambrecht mentioned as well) and Focus-Background relations. In line with FDG, I speak only of a topical or a focal referent when this function is realized in linguistic form. Again, Topic and Focus marking is language-specific. Incidentally, their counterparts, Comment and Background, respectively, tend not to be linguistically marked (Hengeveld & Mackenzie 2008: 88-96).

Despite the fact that Topic and Focus are not in a complementary relation to each other, it is certainly possible to combine them in one Discourse Act. Such combinations are called Content Frames (since they are the frames of Communicated Contents). The standard Content Frames that involve Topic and Focus are the following (Hengeveld & Mackenzie 2008: 101):

 Thetic: The whole Communicated Content is in Focus. [(SA)N_]

FOC

 Categorical: A combination of topical and focal Subacts. [(SA)TOP (SA)N (SA)FOC]

 Presentative: A Subact is both Topic and Focus, a new referent is introduced. [(SA)N_(SA)

TOPFOC]

Again, the expression of Information is entirely language-specific. Not only can Topic and Focus be marked differently, the possibilities for using them may also vary. Therefore, different context frames exist for Topic- and Focus-oriented languages (Hengeveld & Mackenzie 2008: 99-102).

Finally, activation will be discussed. In Lambrecht (1994), identifiability is linked to

knowledge, while activation is linked to consciousness. In this thesis, identifiability is considered to be the core of definiteness, for which knowledge seems to be a suitable term. Indeed, whether a referent is definite or indefinite mainly depends on the addressee's familiarity with this referent. Nevertheless, whenever reference is made by a speaker, a mental link needs to be established

between the referent and its mental representation in the mind of the addressee. This process is called activation. A referent's degree of accessibility can greatly differ. Whenever a referent is identifiable

10 Or Contrast, but I will not further elaborate on this pragmatic function. In addition, note that in line with FDG, pragmatic functions will from now on be written with a capital letter.

(22)

to an addressee, it is in one of the three activation states: active, inactive or accessible. Logically, an active referent is mentioned recently and is therefore easy to remember. Inactive (or unused)

referents are located in the mind of the addressee, but are not yet activated. Consequently, it then requires more processing effort to determine the necessary referent. Accessible referents form an intermediate stage between the active and inactive referents. They are not actively used at the speech moment, but they are nevertheless more active in the addressee's memory than inactive referents or easier to infer from the context (Lambrecht 1994: 105-108; Eppstein 2002: 334, 344).

It should be clear that activation is dependent on the information that is stored in the

Contextual Component, and that the various activation states are reminiscent of stacking. Therefore, again consider example (10).11_{Active referents are generally expressed through personal pronouns.} In the second sentence, Ian is an active referent and is referred back to by means of a pronoun. The same applies to 'Ann' and 'who'. Both Ian and Ann were located high enough on the stack to be considered as active. However, when Ian, Mary and Ann were first mentioned, they were all inactive referents. Given the fact that they are referred to by their first name implies that the addressee is familiar with all of them. Despite this, they still needed to be introduced in the discourse, which is why the use of pronouns would be inappropriate. An example of accessibility is the second reference to Mary. She is then located at the bottom of the stack, which makes her less active than Ian and Ann. To avoid confusion by using a pronoun, reference is again made to Mary by the use of her first name. Still, since reference to Mary has been made earlier, at this stage she is textually accessible. She is deactivated but not removed from the stack, which is why it is relatively easy to reactivate her. Referents can be accessible in two more ways: situationally and inferentially. Situationally accessible referents are accessible due to the situational information from the Contextual Component.

Inferentially accessible referents are those referents that can be identified by means of context (such as 'the teacher' in (5)). The following figure shows the relationship between identifiability and activation, as well as the internal possibilities (borrowed from Lambrecht 1994: 109-111):

Unidentifiable

Identifiability Active Textually

Identifiable Activation Accessible Situationally Inactive Inferentially

Figure 1. Summary of the various terms concerning identifiability and activation

11 Stacking, as described in Hengeveld & Mackenzie (2011: 11), was actually explicitly not attributed to activation state but to contextual salience. This point will be taken up again in section 3.4.4. For the time being, the procedure of stacking is simply used to more clearly illustrate the notion of activation.

(23)

Despite their differences, it is clear that these three concepts strongly overlap. Nevertheless, definiteness, activation and information structure are distinguished from one another. As has been discussed above, definiteness (identifiability) concerns an interlocutor's knowledge of a referent, whereas activation concerns an interlocutor's consciousness of a referent. Lambrecht unites both descriptions under the category of expressing the (temporary) cognitive state of discourse referents, as opposed to information structure that expresses the different relations between referents and propositions. Evidently, the three concepts discussed here have closely related but different functions (Lambrecht 1994: 161-162).

To give an example of the interaction between activation and information structure, I will introduce Lambrecht's Topic Acceptability Scale. To some degree, activation and information structure are interdependent. However, the possibility to interpret a referent as a Topic depends to a large degree on the activation state of that referent. Topics are generally composed of known information/referents. Nevertheless, the difficulty of activating them can differ. This gives us the following scale (borrowed from Lambrecht 1994: 165):

Active Most acceptable

Accessible Unused

Brand-new12 _{Least acceptable}

Figure 2. The Topic Acceptability Scale

The most acceptable Topics are the most recently mentioned ones. As the ease of activating the referent diminishes, so does its likelihood of becoming a Topic. However, the advantage of this scale is that it leaves open the theoretical possibility for a lower acceptable referent to function as a Topic. Logically, this is unlikely to occur with brand-new referents, but accessible referents form a gray zone. Incidentally, it may appear as if activation and information structure overlapped, i.e. the most acceptable referents were necessarily Topics, while the least acceptable referents were assigned the Focus function, but this is not the case. Active referents are likely to become Topics, but do not have to. If emphasis is put on an active referent, it can still receive Focus function (Lambrecht: 165-166). This is the case in the following examples (Nesset 1999: 89):

(12a) Whom did Bill hit?

(12b) He hit me.

12 Lambrecht (1994: 86) actually distinguishes two kinds of brand-new referents, anchored and unanchored. However, I do not consider this distinction to be relevant to my theory as well as to my experiment in Chapter 2.

(24)

2. Definiteness in Russian within FDG

Russian is a language with (relatively) free word order and no articles. The absence of articles in Russian (and in most other Slavic languages) has continuously raised questions concerning the overt or covert existence of definiteness. When describing different ways of expressing this category, Lambrecht also mentions Russian. Despite the fact that no grammatical markers exist to denote definiteness, he argues that native speakers of Russian nevertheless have a concept of identifiability and can express it in an indirect way. Due to the dissimilarity between definiteness and identifiability, Lambrecht (1994: 87) considers identifiability to be a language universal rather than definiteness, albeit as a cognitive category instead of a grammatical one. However, Diana Brun in her article (2001) says the following:

The distinction between definite and indefinite nominals is an important element of discourse and, therefore, should be universally present in any natural language. Hence, the apparent difference among languages is not the presence or absence of the definite/indefinite distinction but rather in the ways this distinction is expressed (Brun 2001: 45).

Interestingly, Lambrecht draws his conclusion after a brief discussion of Czech and Russian examples, but Brun seems to be convinced that definiteness exists in Russian as a grammatical category. In the following, both views will be taken into account.

So far, definiteness has been discussed in general within the FDG framework only. The present section will discuss earlier findings on the expression of definiteness in Russian and consider them in accordance with FDG. In what follows, two articles will be discussed: Nesset (1999) and Brun (2001). Both articles provide a profound analysis of the expression of definiteness in Russian. Furthermore, both articles overlap in certain crucial aspects. They both presuppose that definiteness in Russian can be expressed in three ways: lexically, morphologically and syntacticly.13_{Nesset also} adds a number of subcategories that deal with the manner of realization itself (see Figure 3 below from Nesset 1999: 84). According to both publications, definiteness is directly (overtly) realized by means of lexical markers; in contrast, morphological and syntactic means can only express

definiteness indirectly. This is considered to be possible in two ways: either (in)definiteness is implied by a given linguistic marker, or (in)definiteness statistically corresponds with this marker. Brun slightly deviates from Nesset by leaving open the possibility for unmarked NPs, i.e. NPs that

13 In Brun (2001), this category is actually called the 'non-lexical' realization of definiteness. Notwithstanding the fact that referring to this category as morphological is somewhat inconsistent with a manner of definiteness realization defined by Brun, I will continue to do so for convenience.

(25)

are not marked for definiteness. The system will be clarified in the following sections.

Realization

Direct Indirect

Implicational Statistical

Figure 3. The realization of (in)definiteness in Russian

This chapter is organized as follows. The first three sections consider the realization of definiteness in Russian by describing each of the three manners of realization separately. Hence, section 2.1 deals with lexical marking and 2.2 with morphological marking. Section 2.3 contains three subsections, of which the first deals with Russian syntax in general and with regard to

information structure; the second with previous theoretical and empirical findings in the literature on the syntactic realization of definiteness; and the third with a new theory of definiteness marking through syntax within FDG. Section 2.4 outlines the mutual relations between all the manners of definiteness realization discussed. A certain hierarchy seems to exist between them. Note that each manner of definiteness realization will be considered in light of the FDG framework, on the basis of which I will formulate my own theory. The chapter concludes with summarizing remarks.

2.1 Lexical realization of definiteness in Russian

Russian has the following lexical markers of definiteness:  Demonstratives (e.g. ėtot 'this', or tot 'that');  Anaphoric pronouns (e.g. on 'he', or ono 'it');

 Possessive pronouns (e.g. naš 'our', or svoj 'one's' (refers to the subject of the clause));  Definite quantifiers (e.g. oba 'both' and ves' the whole').

Evidently, these lexical markers are not equivalents of definite articles, since they carry additional information. For instance, anaphoric pronouns can be used independently, something that is

impossible for articles. Nevertheless, an entity that is anaphorically referred to is logically definite, because, otherwise, anaphoric reference would be impossible. The following lexical markers denote indefiniteness:

 Pronouns with: -to, -nibud', -libo, ne-, and koe- 'something/someone/anything';  Odin 'one, a';

(26)

 Vsjakij/ljuboj 'any'.

Concerning the suffixes, in addition to their marking of indefiniteness, they can also denote specificity. Whereas -nibud' expresses non-specificity, -to indicates the opposite. Furthermore,

vsjakij also designates indefiniteness, in contrast to its 'synonym' každyj, which is unmarked.

Unstressed odin also expresses indefiniteness, although there are some limitations as to its usage (Nesset 1999: 96-99; Brun 2001: 45-46).14

Incorporating this form of definiteness marking in FDG seems relatively simple, since it is overt realization of definiteness. However, it is again important to note that these markers are not pure markers of (in)definiteness. Consider the following example and possible representations (Brun 2001: 46):

(13a) Ėti cvety vjanut.

this.NOM.PL flower.NOM.PL wither.3PL

'These flowers are withering.' (13b) IL: (+id +s RI) RL: (m prox xi: [(fi: flower (fi)) (xi)]) (13c) IL: (+id +s RI) RL: ≠(m xi: [(fi: flower (fi)) (xi)])

In (13) 'these flowers' is unambiguously definite and this status is inherent to the demonstrative, which, obviously, does not express definiteness only. If a language has articles, their usage is determined at the Interpersonal Level on the basis of identifiability and specificity. The usage of demonstratives, however, is also determined on the Representational Level, since demonstratives mainly designate deictic relations. This is why at the speech moment of (13a), the flowers need to be present in vicinity or the mental space of the speaker, or have to be distinguished from other flowers.

14 Specificity oppositions within these markers actually seem to be more complicated than just differences in speaker's knowledge. Lyons (1999: 174) mentions the distinction between interrogatives with either -to and -nibud'. They can both only be used in specific contexts along which the concept of specificity changes in terms of scope. In Ionin (2013) Russian odin is studied in the following context (Ionin 2013: 2):

(14) Maša pročitala včera odnu interesnuju knigu.

Masha.NOM read.PST.F yesterday one.ACC.F interesting.ACC.F book.ACC.F

'Masha read an interesting book yesterday.'

It appears that such usage of odin has certain semantic prerequisites. It needs the referent to convey a sense of 'identifiability', which in this case means that enough information is being transferred to the addressee to be able to identify the referent. Therefore, odin corresponds closely to a certain in English. Despite all this, I will simply consider -nibud' as an indefinite non-specific marker and -to and odin as indefinite specific markers.

(27)

In (13c) proximity is absent, which makes (13a) incorrect. (13b) contains the representation that makes (13a) correct, since it does include proximity. Logically, this situational information comes from the Contextual Component. In addition, the representations seem to demonstrate that the lexical realization of definiteness is, in fact, far from direct. Indeed, definiteness information is not directly transmitted to the Morphosyntactic Level, it is interrupted on the Representational Level, where additional semantic information is formulated.15_{Naturally, this also applies to other languages with} such markers.

Finally, in spite of the fact that Brun considers the usage of verbal prefixes a non-lexical manner of definiteness marking, this is not entirely the case. Firstly, the addition of a prefix to a verbal stem can be considered to be word formation and is, thus, part of the lexicon. Secondly, the manner of definiteness marking through the use of verbal prefixes strongly resembles that of the realization of definiteness through the markers described above. Consider the following examples (Brun 2001: 47, cited from Apresjan 1995):

(16a) On napisal pis'mo.

he write.PFV.PST.M letter.ACC

'He has written a/?the letter.'

(16b) On dopisal pis'mo.

he to.the.end.write.PFV.PST.M letter.ACC

'He has written the letter to the end.'

(16c) On dopisyval pis'mo.

he to.the.end.write.IPFV.PST.M letter.ACC

'He was finishing writing the letter.'

Certain verbal prefixes can indirectly imply definiteness, especially those that denote an

accomplishment. This is largely independent of Slavic aspect.16_{In (16a), the prefix na- functions}

15 Etymologically, definite articles are generally derived from demonstratives. As illustrated above, Russian demonstratives still carry extra information that distinguishes between proximal and distal referents, but a certain language change seems to be in progress in Russian. An unstressed demonstrative seems to form an intermediate stage for a definite article (Nesset 1999: 98, cited from Birkenmaier 1979: 90):

(15) Dokument ėtot byl neobxodim.

document.NOM.M this.NOM.M was.PST.M necessary.NOM.M

'He needed the document.'

A similar process took place in Bulgarian with demonstrative pronouns (Kalsbeek 2012: 210). Certain Russian dialects even already have something similar to the Bulgarian article, the post-positive participle: kryša-ta 'roof-this/the roof'. Despite this, usage of unstressed demonstratives is not obligatory in standard Russian, which is why it cannot be considered an article at this point (Nesset 1999: 96-97; Mendoza 2011: 249).

16 It has been claimed that in Russian, aspect correlates with (in)definiteness, such as in Bayer (2005: 13-14). Perfective verbs would denote definiteness, whereas imperfective verbs would denote indefiniteness. However, Bunčić managed

(28)

simply as a marker of perfectivity but not of definiteness of the object. On the other hand, in (16b) and (16c) do- in fact designates both accomplishment and definiteness of the object, regardless of the aspectual status of the verb (Brun 2001: 47). Nevertheless, as is the case with the other lexical markers, verbal prefixes are not pure markers of definiteness and contain additional semantic information. For instance, the use of do- is justified not only because of the definite status of the letter, but also due to the specific manner of writing, namely finishing it. In case this meaning component is not applicable to the situation, the use of the prefix is also unavailable, regardless of whether the reference to the letter is definite or not. In sum, I argue that lexical markers do not denote definiteness directly, but they do so indirectly, and that verbal prefixes are also part of the lexical marking of definiteness.

2.2 Morphological realization of definiteness in Russian

The morphological realization of definiteness is also a form of indirect realization. The marker used designates something else, but, nevertheless, implies definiteness or statistically corresponds to it. According to Nesset (1999), case marking is such a manner of indirect definiteness marking. He distinguishes between two instances of case marking, those of quantification and negation. Consider the following examples taken by Nesset from Mathiassen (1996: 209):

(17a) Ona kupila saxar. she bought sugar.ACC 'She bought (the) sugar.'

(17b) Ona kupila saxaru she bought sugar.GEN

'She bought some sugar.'

While in (17a) 'sugar' receives an accusative case marking, in (17b) saxar is in the genitive partitive.17_{Nesset (1999: 92-93) argues that the difference in case in this context is that the}

accusative merely indicates that the NP is an object (but statistically often also designates that it is definite), whereas the genitive indicates the same but, additionally, that there is an indefinite quantity of the relevant entity. Therefore, the use of the genitive partitive implies indefiniteness according to this scholar. However, the difference between (buying) sugar and some sugar seems to be vague to such an extent, that I assume here that usage of the genitive partitive is not motivated mainly by a

to falsify this hypothesis (Bunčić 2012: 8-9).

17 Note that the genitive partitive only occurs with concrete products, materials and other similar substances (Podgaevskaja & Honselaar 2007: 433).

(29)

speaker's wish to express quantity, but to express that the object is indefinite. In fact, in the literature, indefinite articles have sometimes even been directly related to quantification (Bunčić (2012: 4), cited Lyons (1999: 34), and Vater (2005: 106-107)). Considering that the accusative does not necessarily designate definiteness, it is likely to be unmarked. In this sense, the genitive case can be considered to be a pure marker of indefiniteness.

The genitive case expressing negation and definiteness have a slightly more complex relationship. Consider the following examples:

(19a) Griby zdes' ne rastut mushrooms.NOM.PL here NEG grow.3PL

'(The) mushrooms do not grow here.' (19b) Gribov zdes' ne rastet mushrooms.GEN.PL here NEG grow.3SG

'No mushrooms grow here'/'There are no mushrooms here.'

Whereas in (19a) 'mushrooms' receives nominal case marking and is, therefore, the grammatical subject of the clause, in (19b) it receives genitive case marking combined with an impersonal verb form without a subject. The same can occur in transitive clauses with direct objects (DO) that can have either accusative or genitive marking (which in the latter case means that it is technically not a DO anymore). Nesset considers a prototypical genitive NP to be indefinite, and a nominative or accusative one to be definite (Nesset 1999: 94-96).18_{Now consider the representations of (19a) and} (19b) for 'mushrooms':

(20a)

IL: (+id +s RI)19

RL: (neg ei: [(fi: [(m xi: grib (xi))A ] (fi)) (ei)])

ML: (Npi: [(Nwi: [(Nsi: /grib/ (Nsi)) (Affi: /y/ (Affi))] (Nwi))] (Npi)) (20b)

IL: (-id +s RI)

18 However, as Nesset also noted, in Timberlake (1985: 339), case designation in the context of negation was linked to the 'individuation of the referent of the relevant NP'. According to Timberlake, individuation is "the degree to which the participant is characterized as a distinct entity or individual in the narrative event". The nominative and accusative cases indicate a high level of individuation, while the genitive case indicates a low level. The degree of individuation depends on a number of categories, such as abstractness and countability, but also definiteness. Since other categories are present, definiteness and individuation are not exclusively linked.

(30)

RL: (neg ei: [(fi: [(m xi: grib (xi))A ] (fi)) (ei)])

ML: (Npi: [(Nwi: [(Nsi: /grib/ (Nsi)) (Affi: /ov/ (Affi))] (Nwi))] (Npi))

Just as in (17b), the genitive appears to be a marker of indefiniteness here as well, but there is a difference. In Russian, negation determines the genitive case (as a default case) of subjects of intransitive verbs and (direct) objects of transitive verbs, regardless of their semantic role. As is visible in the representations, the genitive case marking, thus, originates from the Representational Level, not the Interpersonal Level. Given that negation as an operator of a State-of-Affairs is located higher in the hierarchy than semantic functions, which belong to dependents of the Configurational Property, it logically takes precedence over them. But whenever a referent is definite, it appears that inflection in the genitive is canceled and the NP takes the regular nominate/accusative case.

Likewise, this can be explained through FDG, since the Interpersonal Level is hierarchically higher than the Representational Level. Considering the fact that genitive case under negation implies indefiniteness, but receives its case marking from negation on the Representational Level and not through its information status on the Interpersonal Level, it can be argued that case marking here functions as a marker of definiteness (rather than of indefiniteness). In conclusion, marking of (in)definiteness through case contrasts heavily with lexical marking.

Finally, let me turn to the last manner of indirect morphological realization of definiteness involves agreeing adjectives and non-agreeing genitive-marked or possessive nouns in this context. Consider the following examples from Brun (2001: 46, cited from Apresjan 1995):

(21a) Za dver'ju slyšalsja ženskij golos.

behind door.INS hear.PST.M.PASS female.NOM.M voice.NOM.M

'There was a woman's voice heard from behind the door.'

(21b) Za dver'ju slyšalsja golos ženščiny.

behind door.INS hear.PST.M.PASS voice.NOM.M woman.GEN.SG

'The voice of a/the woman was heard from behind the door.'

As becomes evident from the translation in (21a), the use of an adjective indicates that the corresponding referent is indefinite. This, however, does not imply that the alternative denotes definiteness. Instead, the relevant NP is then unmarked for definiteness, and other factors determine its information status (Brun 2001: 46). However, when considered within the FDG framework, agreeing adjectives do not seem to be actual markers of definiteness. Consider the representations of (21a) and (21b) for 'a woman's voice/the voice of a woman':

(31)

(22a)

IL: (-id +s RI) (?id ?s R?) RL: (xi: [

(fi: golos (fi): [(fj: ženskij (fj)) (fi)]) ] (xi)U])

ML: (Npi: [

(Awi: [(Asi: /žensk/ (Asi)) (Affi: /ij/ (Affi))] (Awi)) (Nwi: [(Nsi: /golos/ (Nsi)) (Nwi))

] (Npi)) (22b)

IL: (+id +s RI) (?id +s RI) RL: (xi: [ (fi: [ (fj: golos (fj)) (xj: [(fk: ženščina (fk)) (xj)U])Ref ] (fi)) (xi)U]) ML: (Npi: [ (Nwi: [(Nsi: /golos/ (Nsi)) (Nwi))

(Nwj: [(Nsj: /ženščin/ (Nsj)) (Affi: /y/ (Affi))] (Nwj)) ] (Npi))

In (22b), the woman is indeed unmarked for definiteness. However, the representation in (22a) reveals that the adjective does not denote possession, but a property. This could be due to what is referred to in FDG as alienable and inalienable possession. In the case of 'a woman's book', it is evident that this book is not inseparable from the woman. In contrast, this does not apply to 'a

woman's voice'. Therefore, it is doubtful to consider (21a) as the direct counterpart of (21b), since the former seems to focus on the female property of the voice, and the latter on the female (inalienable) possessor of the voice. Furthermore, it is questionable whether this contrast in the expression of possession is even possible in cases of alienable possession or different kinds of adjectives. Apart from this, in the original Russian clause in (21a), 'woman' is not even the head of the NP ('voice' is), which casts further doubt on to what referent the expression of indefiniteness actually applies.

(32)

2.3 Russian syntax

2.3.1 An overview of Russian syntax

In Kallestinova's dissertation (2007), constituent order in Russian is thoroughly discussed.20 Theoretically, a simple transitive clause may occur in every possible constituent order.

(23) Ivan uvidel Borisa.

Ivan.NOM see.PST.M Boris.ACC.ANIM

'Ivan saw Boris.'

So a SVO clause such as (23) can also occur in a SOV, VSO, VOS, OSV and OVS order. This is possible, because, in contrast to English, it is case marking rather than constituent order that expresses grammatical relations in Russian. However, this does not mean that constituent order in Russian is arbitrary. In the literature, it is generally assumed that Russian syntax is mainly influenced by information structure. The following examples illustrate this (derived from Kallestinova (2007: 1-2)):

(24a) Kogo uvidel Ivan?

who.ACC.ANIM see.PST.M Ivan.NOM

'Whom did Ivan see?'

(24b) Ivan uvidel Borisa.

Ivan.NOM see.PST.M Boris.ACC.ANIM

'Ivan saw Boris.'

(25a) Kto uvidel Borisa?

who.NOM see.PST.M Boris.ACC.ANIM

'Who saw Boris?'

(25b) Borisa uvide-l Ivan.

Boris.ACC.ANIM see.PST.M Ivan.NOM

'Ivan saw Boris/Boris was seen by Ivan.'

In each response, the referent given in the question is placed first in the clause, whereas the new referent in the answer to the question is placed last. For many scholars, this is reason to believe that Russian distinguishes between Topic and Focus positions in each clause, respectively Clause-initial and Clause-final, and that 'discourse-neutral information' can be added to this distinction for any remaining information in a given clause that is neither Topic nor Focus. This yields a tripartite

20 I use the term 'constituent order' instead of 'word order', because the focus of this study is not the order of any type of word, but the focus of constituents, i.e. a word or a sequence of words that functions as a single unit.

The expression of definiteness in Russian within Functional Discourse Grammar

MA thesis