Bulgarian edin: The Rise of an Indefinite Article∗ Ljudmila Geist University of Stuttgart Abstract In many languages, the indefinite article emerges from the numeral ONE. In the course of its diachronic development it acquires additional functions such as marking of specificity and genericity. Thus, different functions of indefinite articles can be seen as stages in the process of grammaticalization. This paper examines the use of edin in Bulgarian as a marker of indefiniteness. It shows that it has reached the stage of a specificity marker and is developing the predicative and the generic use. This leads to the conclusion that edin has not yet entirely achieved the stage of indefinite article but merely has the intermediate status, the status of an indefinite determiner. The findings concerning the current functions of edin in this paper lead to a modification of the implicational scale of functions of indefinite articles proposed in the literature. 1 Introduction It has commonly been assumed that Slavic languages have no indefinite articles. However, there are hints that the counterpart of the numeral ONE in Bulgarian can serve as a marker of indefinite reference. Consider example (1). (1) Marija se omâži za edin lingvist. Mary refl married prep one ‘Mary married a linguist.’ linguist Here edin does not highlight the cardinality (the sentence is not understood as being an answer to “How many linguists did she marry last year?”) but indicates indefinite reference, i.e., it is used to contribute an existential claim and to introduce a new referent into the discourse. Besides this function edin has some other functions typical of markers of indefinite reference. This led Friedman (1976) to hypothesize that Bulgarian edin has the status of an indefinite article. It is a well-known fact that in many languages the counterpart of the numeral ONE has given birth to indefinite articles. In many languages indefinite ∗ This research was funded by the German Science Foundation (project C2 Case and Referential Context, SFB 732 Incremental Specification in Context), which I gratefully acknowledge. I would like to thank Klaus von Heusinger, Uwe Junghanns, Elena Karagjosova, and Svetlana Petrova for very helpful discussions. I also thank two anonymous reviewers for critical comments and the editors for including the paper in the volume. Erschienen 2013 in: Junghanns, U. , Fehrmann, D., Lenertová, D. & Pitsch, H. (eds.): Formal Description of Slavic Languages: The Ninth Conference. Proceedings of FDSL 9, Göttingen 2011 Frankfurt am Main: Peter Lang, 125-148. Ljudmila Geist articles have the same root as the numeral in the respective language, cf. German ein, Turkish bir, and Spanish un. Hence, it would not be a surprise if Bulgarian edin also had the function of an indefinite article in addition to its numeral function. The aim of this paper will be to test the hypothesis of Friedman (1976) that Bulgarian edin can be used as an indefinite article. In particular, the questions that I want to answer are: • What functions can edin as a marker of indefinite reference cover and how are they related? • What is the contribution of edin to the meaning of the noun phrase? • What might be the outlines of its further development? To determine the functions of edin and its semantic contribution I will turn to the work on the grammaticalization of indefinite articles in Givón (1981) and Heine (1997). The main idea going back to Givón is that numerals on the way of developing into indefinite articles undergo a process of semantic bleaching, for which a scale like in (2a) can be assumed. I will call the corresponding grammaticalization stages as in (2b). (2) Grammaticalization of indefinite articles a. quantification > referentiality b. numeral Stage I > non-referentiality1 > indefinite determiner > indefinite article Stage II Stage III The function as numeral (Stage I) is the starting point on the scale. The indefinite article (Stage III) represents the terminal point with the highest degree of grammaticalization. But besides the starting stage numeral and the terminal stage indefinite article there is an intermediate stage called indefinite determiner, where the numeral loses its function of indicating cardinality or quantity and acquires the function of indicating indefinite reference. At this stage the indefinite marker is used only with referential indefinite phrases; at the last stage the indefinite marker can be used with non-referential NPs as well. This scale will be more finely differentiated in Section 2. Although I will focus on synchronic functions of edin in modern Bulgarian, the work on grammaticalization of indefinite articles by Givón (1981) and Heine (1997) will be helpful in this discussion since different stages in the diachronic process of grammaticalization of indefiniteness markers can be seen as their synchronically attested functions. 1 Givón uses the term genericity instead of non-referentiality. Bulgarian edin: The Rise of an Indefinite Article The paper is organized as follows. In Section 2 criteria for distinguishing the numeral from the marker of indefinite reference will be discussed. In Section 3 we concentrate on the intermediate status of edin as an indefinite determiner, and in Section 4 we test whether formal and semantic criteria for the terminal stage, the stage of indefinite articles, apply to edin. To illustrate the numeral and indefinite functions of edin I will use examples from the linguistic literature and complement them with constructed examples that I tested with informants.2 2 Distinguishing the numeral from the marker of indefinite reference The original function of edin in Bulgarian is to indicate the cardinality of one, which is the numeral function. In this section I will introduce some tests for singling out the numeral function3 from other functions of edin partly known from the literature (e.g. Nikolaeva 1979, Šmelev 2002 for Russian odin, Weiss 2004 for Macedonian eden): (i) Combination with focus particles and numeral modifiers The numeral edin can be combined with focus particles such as samo ‘only’, or modifiers such as točno ‘exactly’ or naj-malko ‘at least’, which emphasize the cardinality. (3) Samo edin telefon li imate? only one telephone Q.part you-have ‘Do you have only one telephone (or two)?’ 2 3 I would like to express my gratitude to Elena Karagjosova, Milena Kühnast, Svetlana Petrova, Zorica Trpcevska, Boryana Max Bincheva Bittner and Nadia Varley for evaluating and commenting on my Bulgarian examples. Although edin, like its counterparts in other Slavic languages, can semantically serve as a numeral, it differs morphologically from other numerals. It agrees with the noun in gender and resembles adjectives in this respect. Three gender forms have to be distinguished in the singular: the masculine edin ‘one.masc’, the neuter edno ‘one.neut’, and the feminine form edna ‘one.fem’. In Slavic languages with morphological case forms, the counterpart of ONE agrees with nouns not only in gender but also in case; cf. Russian Ja govoril s odnim sosedom ‘I spoke with one.inst neighbor.inst’. Hence, it differs from numerals like TWO, THREE and FOUR, which govern the genitive case on the quantified noun; cf. in Russian the example of genitive government in dva soseda ‘two.nom neighbor.gen’ with the example of the case agreement in odin sosed ‘one.nom neighbor.nom’. Ljudmila Geist The indefiniteness marker edin cannot be combined with focus particles and numeral modifiers; cf. (4), where edin is used as an indefiniteness marker and the use of the focus particle točno is excluded. (4) Marija se omâži za (*točno) edin lingvist. Mary refl married prep exactly one ‘Mary married exactly one linguist.’ linguist (ii) Contrast in cardinality A typical characteristic of numerals in general and of the numeral edin in particular, which follows from characteristic (i), is that they can be contrasted with other numerals. In (5) the numeral function of edin is singled out by placing it in opposition to the numeral for two. (5) Ivan ima edin sin (, а nе Ivan has one son and not ‘Ivan has one son (and not two).’ dvama). two Sentence (5) can be understood as an answer to the question “How many sons does Ivan have?” Here edin unambiguously indicates the number of sons. (iii) Restrictions on the morphological form Besides the meaning difference and the difference in the compatibility with focus particles, there is another formal property of the indefiniteness marker edin which distinguishes it from the corresponding numeral. The numeral edin can occur with count nouns such as amerikanec ‘an American’ only in the singular. However, edin can be combined with count nouns in the plural too (Friedman 1976). Since edin in the plural cannot indicate a cardinality of ONE, it is not a numeral in this use. Edin in the plural also cannot be combined with numeral modifiers. The use of edin as an indefiniteness marker in the plural is demonstrated in (6). (6) Včera vidjax edni amerikanci, koito poznavam. Yesterday I-saw one.pl Americans whom ‘Yesterday I saw some Americans whom I know.’ I-know One remark should be made concerning the plural use of the numeral edin. Morphologically plural forms for the numeral edin do exist, however these forms are semantically singular and can be used only with pluralia tantum (cf. edni očila ‘one.PL glasses = one pair of glasses’). The numeral edin is incompatible with count nouns in the plural, since it semantically indicates a cardinality of one while plural count nouns indicate reference to more than one individual. Hence, in (6) edin clearly has an indefinite rather than a numeral Bulgarian edin: The Rise of an Indefinite Article function. Note that from a cross-linguistic point of view, the availability of a plural form for one for pluralities of objects is not an obligatory but a sufficient property of edin as an indefiniteness marker. For example, in Spanish the indefiniteness marker un, which is related to the numeral, has a plural form unos, which can be used with plural count nouns (Le Bruyn 2010). The indefiniteness marker un has the status of an indefinite article. In English and German, by contrast, the indefiniteness markers a(n)/ein have no morphologically plural form, but still have the status of indefinite articles because of other characteristics that I will discuss in Section 4. 3 The indefinite determiner as an intermediate stage Heine (1997) and Givón (1981) have assumed that the development of indefinite articles from numerals is a continuous process which goes through different stages. As shown in the introduction, at least three stages can be distinguished: besides the stage of the numeral, the stage of the indefinite determiner and the stage of the indefinite article per se. Heine (1997) suggests a more fine-grained view of the grammaticalization process of the numeral to the article: he distinguishes 5 stages, as shown in (7). The relation between Heine’s stages and the more simple scale introduced in (2) from Givón (1981) is indicated by the curly braces: (7) Grammaticalization stages of indefinite articles: Heine (1997) and Givón (1981) 1. the numeral > 2. the presentative marker > 3. the specificity marker I. numeral II. indefinite determiner > 4. the non-specific marker > 5. the generalized article III. indefinite article Scales such as these can be interpreted as synchronic implicational scales. This means, for example, that an indefiniteness marker at a given stage may have properties of all preceding stages, but not vice versa. Besides the synchronic view, which is our main interest here, such scales can provide a diachronic view of the evolution from the first, oldest, stage to the last, current, stage. Heine shows that languages have different relative degrees of grammaticalization that an indefinite article has reached. He gives the following examples: Stage 1: Swahili; Stage 2: Russian (but see Geist in prep.); Stage 3: Ljudmila Geist Hebrew; Stage 4: Germanic, most Romance languages, Chinese; Stage 5: Spanish. In what follows I will analyze edin with respect to particular stages on the scale of Heine. In Section 3.1 I will concentrate on the stage of presentative marker – Stage 2 in (7) – and in Section 3.2 on the stage of specific or specificity marker – Stage 3 in (7). The stages of non-specific marker and generalized article (Stages 4 and 5) will be discussed in Section 4. 3.1 Edin as a presentative marker At Stage II indefinite determiner the counterpart of ONE in different languages is used only with referential NPs, i.e., with NPs which introduce a new referent into the discourse.4 Wright & Givón (1987) show that in addition to indicating the semantic status of referentiality the indefinite determiner in many languages also indicates the pragmatic status of the importance of NP referents in the discourse. The difference in the importance of discourse referents can be made clear via the short discourses in (8a-b): (8) a. I was in the city and bought a paper. It had a very interesting article. I read it … . b. I was in the city and bought a paper. Then I went home, ate something and went to sleep. In both stories, the NP a paper is referential. However, in the first story the referent is taken up in the second statement. This statement contributes new information about the referent, hence the speaker considers the referent to be important. In the second story the paper is mentioned only in the first statement; it plays no role in the next statement. It is not taken up again. To capture the intuittive idea of the importance of the referent for the subsequent discourse, Givón (1983) introduces the term referential continuity or discourse persistence. While the referent of a paper is discourse persistent in (8a), it is not in (8b). Givón states that some languages mark discourse persistence of NP-referents by lexical means, e.g. in Hebrew the indefiniteness marker xad ‘one’ can be used in the 4 I assume that discourse referents introduced by referential NPs can be anaphorically accessed by personal pronouns in non-modal contexts such as in (i). Non-referential phrases either do not introduce any referents, like the predicate NP a teacher in (ii) – in this case they denote properties and do not license pronominal anaphors – or they have discourse referents with a short life span, i.e., are restricted to possible worlds. Such referents cannot be accessed by personal pronouns in non-modal contexts, cf. (iii). (i) Gertrud is looking for a bicyclereferential. It disappeared on Friday. (ii) Gertrud is a teachernon-referential by profession. (iii) Gertrud is looking for a bicyclenon-referential. She couldn’t find *it/one for her size. Bulgarian edin: The Rise of an Indefinite Article story (8a) to indicate high discourse persistence of the indefinite NP, while in (8b) the NP remains unmarked. Another example of a language which uses a special marker of discourse persistence is Romanian with the particle pe (Chiriacescu & von Heusinger 2010). But what about edin, can it indicate discourse persistence? In the literature on edin it has been pointed out that in using it the speaker seems to convey that the referent is important in the discourse and he/she wants to tell more about it. This function of edin is attested, for example, in the beginnings of fairy tales; cf. (9), where the referent introduced with edin is immediately taken up in the next sentence. (9) Imalo edno vreme edin starec. Toj imal trima sina. had one time one old-man he had three ‘Once upon a time there was an old man. He had three sons.’ sons Hence, edin can indicate discourse persistence. It serves as a presentative marker. However, if the function of edin were exclusively that of a presentative marker, its use in contexts such as (8b), where the introduced referent is not discourse persistent since it is not taken up in the subsequent discourse, would be inappropriate. But as (10) shows, edin can be used even if the object is not taken up in the subsequent discourse. (10) Predi da predam statijata ja dadox na edin kolega za korekcii. Sled tova podadox statijata na edno spisanie. ‘Before submitting my paper I gave it to a colleague for proofreading. Then I sent the paper to a journal.’ Thus, the function of edin as an indefiniteness marker does not seem to be confined to the presentative function. 3.2 Edin as a specificity marker At this stage the indefiniteness marker does not just indicate discourse persistence of the NP referent, but specificity as well. Specificity is understood by Heine in its traditional sense: the speaker conveys that he/she has a particular referent in mind, but the referent is presumed to be unknown to the hearer (Heine 1997, 7). I will show that edin can serve as a specificity marker in this sense: it can indicate identifiability of the referent by the speaker (cf. Stankov 1984). I will elaborate on this point in Subsection 3.2.1. Compared with other indefinites, indefinites anchored to the speaker are special in their interaction with other components of the clause: they have wide scope with respect to all quantifiers and operators in the clause and exhibit an Ljudmila Geist exceptional scope-taking behavior. We have to check whether edin shows this exceptional behavior too. In Subsection 3.2.2 we will test the scope-taking behavior of edin NPs in the interaction with quantifiers and then with intensional operators in intensional contexts. Afterwards we will examine their behavior in scope islands. 3.2.1 Identifiability In the literature on edin (e.g. Ivanova & Kovaľ 1994) some hints can be found that edin indicates speaker identifiability of the referent. This is illustrated in (11). (11) a. Čete mi se edno spisanie. read I.dat refl one journal ‘I would like to read a journal.’ b. Continuation compatible with (11a): A imenno, poslednijat broj na and namely, last.def issue of ‘Namely the last issue of Novo Vreme.’ Novo Vreme. Novo Vreme c. Continuation not compatible with (11a): #Kakvoto i da e. which and inf.part be ‘Any journal would do.’ Only the continuation in (11b), which conveys identifying information about the referent, is felicitous. The continuation in (11c), indicating speaker nonidentifiability of the referent, is odd. The absence of edin in (11a) would allow (11c) to be felicitous. Thus, edin seems to serve as a marker of specificity in terms of identifiability by the speaker. Identifiability, however, is a vague concept. I will adopt the definition of identifiability suggested in Ionin (in press). Following Abusch & Rooth (1997), who elaborate on the identifiability contributed by certain-indefinites in English, she proposes that the identifiability of the referent x “requires the speaker to be able to answer the question »Which x is it?«” (Ionin in press). Applied to example (11), felicitous use of edno spisanie ‘one journal’ requires the speaker to be able to give an answer to the question “Which journal is it?” The response to the question must contain an identifying property that singles out a specific journal, distinguishing it from all other journals. Ionin points out that there is a further requirement that the identifying property must be distinct from properties ascribed to the selected individual by the sentence. For example, in (11) the property of the journal that Bulgarian edin: The Rise of an Indefinite Article the speaker wants to read it is not an identifying property in the narrow sense. According to Ionin (in press) “an identifying property must come from outside of the sentence.” In the above example (11), the journal is identified by naming the title and the volume number. Assuming that edin serves as a marker of specificity, i.e., indicates the identifiability of the referent of the NP by the speaker, one remark should be made concerning the specification of the term “the speaker”. Note that this term can be differently specified depending on the type of context. In the examples I have discussed so far “the speaker” is identical to the actual speaker of the utterance. However, in indirect speech, the perspective can be shifted away from the actual speaker to the speaker of embedded speech; consider the following example: (12) Baštata kaza, če e govoril s edin jurist. father.def said that have spoken with one ‘The father said that he has spoken with a lawyer.’ lawyer The NP edin jurist ‘a lawyer’ refers to a person the father can identify. Thus the referent is anchored to the father − the speaker of the embedded speech act − and not to the actual speaker of the whole utterance. 3.2.2 Scope-taking behavior In this subsection I will describe the behavior of edin-indefinites in intensional contexts and in syntactic islands. I will compare their behavior with that of indefinites with a(n) in English. Consider the English example (13). Indefinite singular NPs such as a doctor in English are ambiguous: they can have narrow or wide scope with respect to intensional operators. In (13a) the indefinite NP receives a wide scope interpretation with the continuation (13b) and a narrow scope interpretation with the continuation (13c). (13) a. Ann wants to marry a doctor. b. Continuation compatible with (13a): I know who it is. c. Continuation compatible with (13a): Unfortunately, there are no candidates yet. As opposed to a(n) in English, edin as a specificity marker triggers the wide scope reading indicated by the continuation (14b). The narrow scope reading required by the continuation (14c) is not compatible with (14a). Ljudmila Geist (14) a. An iska da se omâži za Ann wants inf.part refl marry ‘Ann wants to marry a doctor.’ prep edin lekar.5 one doctor b. Continuation compatible with (14a): I know him. c. Continuation not compatible with (14a): #Unfortunately, there are no candidates. The narrow scope reading of the indefinite indicated by the continuation (14c) can only be achieved by the use of a bare NP without edin. Finally, it can be shown that NPs with the specificity marker edin scope out of syntactic islands. This behavior goes along with their wide scope taking behavior in intensional contexts. Syntactic islands are syntactic constructions such as relative clauses, which constitute barriers for quantifiers. The quantifiers cannot scope out of relative clauses and have to be interpreted within them. As Fodor & Sag (1982) have shown, specific indefinites in English exhibit exceptional behavior since they can escape from an island and take maximally wide scope with respect to any quantifier or logical operator outside the island. (15) can have a reading that Mary read every book recommended by one professor or other (narrow scope reading) or Mary read every book recommended by one particular professor ((exceptional) wide scope reading). (15) a. Mary read every book that a professor at this university has recommended. b. narrow scope reading: ∀ > NP c. (exceptional) wide scope reading: NP > ∀ In the counterpart of (15) with the specificity marker edin in (16) the narrow scope reading of the indefinite is excluded. The only available reading is that Mary read every book recommended by one particular professor, i.e., the 5 Note that if in a context with intensional operators a numeral is used instead of the indefiniteness marker edin, the numeral can receive either wide scope as with the continuation in (ib) or narrow scope as with the continuation in (ic). (i) a. Jana iska da pročete dve knigi za nasekomite. Jana wants inf.part read two books about insects.def ‘Jana wants to read two books about insects.’ b. Obače ne moža da gi nameri na neinija raft. ‘But she cannot find them on her shelf.’ c. Njama značenie koi. ‘Any books would do.’ Bulgarian edin: The Rise of an Indefinite Article (exceptional) wide scope reading in which the indefinite escapes the scope island formed by the relative clause. (16) a. Marija pročete vsjaka kniga, kojato edin professor ot tozi Mary read universitet ì every book which one professor in this preporâča. university her recommended ‘Mary read every book that a professor at this university has recommended.’ b. #narrow scope reading: ∀ > NP c. (exceptional) wide scope reading: NP > ∀ As with the examples above, the narrow scope reading can be achieved by using bare indefinite NPs without edin. Edin-NPs can have narrow scope inside the island only if they are modified by focus particles or numeral modifiers such as exactly. As we have seen in Section 2 such modifiers bring out the numeral reading. Numeral NPs unlike specific NPs can take scope either inside or outside the island. To conclude Section 3.2, edin in the contexts investigated here indicates that the indefinite NP is specific, i.e., the speaker intends to refer to a particular individual he/she can identify. The feature of specificity determines the behavior of indefinites with respect to intensional operators and scope islands: edin indefinites always take wide scope. 4 The stage of the indefinite article In the preceding section we have seen that edin has achieved at least Stage 3, specificity marker, on the grammaticalization scale of Heine. In this section I want to determine whether clues are available for the further development of this indefiniteness marker towards the last stage, generalized indefinite article, the extreme point on the grammaticalization scale of Heine (1997) in (7). 4.1 Obligatoriness as a formal criterion Indefinite articles are purely grammatical markers of indefiniteness. As elements of a high degree of grammaticalization, articles are obligatory, i.e., they cannot be left out. To check whether this criterion applies, we consider first the occurrence of NPs in the position of an object as in (17). As Ivanova & Koval’ (1994) show, the argument NP dete ‘child’ in argument position without edin receives an indefinite interpretation. Moreover, the indefinite interpretation is Ljudmila Geist the only available interpretation in this case, since definiteness in Bulgarian has to be marked by the clitic definite article. The argument NP without the definite article and without edin can only be interpreted as indefinite. (17) V stajata vleze dete. (Ivanova & Kovaľ 1994, 59) in room.def came child ‘A child came into the room.’ Thus, bare NPs in Bulgarian need not be marked with edin in order to receive an indefinite interpretation. In this sense, edin is not obligatory. However, this generalization seems to hold only for non-topical NPs. Things are different in topic position: this is sentence initial position, where subjects but also other constituents can occur. Ivančev (1957, 515, cited in Friedman 1976, 338) shows that NPs undergoing topicalization (indicated by clitic doubling) require the occurrence of edin as in (18a) or the definite article as in (18b). Bare NPs are excluded in this position; consider (18c). (18) a. Edna žena ja risuva edin xudožnik. (Ivančev 1957, 515) one woman her painted one painter ‘A woman was painted by a painter.’ b. Ženata ja risuva edin xudožnik. woman.def her painted one painter ‘The woman was painted by a painter.’ c. *Žena ja risuva edin xudožnik. woman her painted one painter Why is edin obligatory with argument NPs only if they are topics? I think that this has to do with the specificity condition on topics: According to Reinhart (1981) indefinite topics (or, to be more precise, indefinite aboutness topics) must be specific in order to provide an address in the common ground under which the information about the topic – the comment – is stored. Since bare NPs in Bulgarian are not marked for specificity (Gorišneva in prep., among others), edin as a specificity marker must be added. In the literature on Bulgarian edin, however, the item that I call aboutness topic has been referred to as “theme” (Dyer 1993) or “sentence initial subject” (Ivanova & Kovaľ 1994, 61). But since theme is often used as a synonym for topic, and sentence initial subjects very often serve as topics, we can assume that what is called theme in the literature on Bulgarian is the same as what is denoted by the term aboutness topic. The correlation between topicality and obligatoriness of edin was also observed by Izvorski (1994). She points out that bare NPs in Bulgarian cannot serve as subjects of predicates such as tall, intelligent, and heavy, which denote Bulgarian edin: The Rise of an Indefinite Article permanent properties (individual-level predicates in the sense of Carlson 1977). According to Jäger (2001), sentences with individual-level predicates in different languages have a fixed information structure with the subject as aboutness topic and the predicate as comment. As shown in (19), bare NPs in Bulgarian cannot serve as indefinite topics in sentences with individual-level predicates: edin cannot be left out. (19) *(Edni) studenti sa inteligentni. (Izvorski 1994, 237) one.pl students are intelligent ‘Some students are intelligent.’ On the basis of these observations edin seems at first glance to fulfil the obligatoriness criterion for articles. However, in topic position edin is obligatory not for purely grammatical reasons but for the semantic reason that indefinite aboutness topics must be specific, i.e., edin is required in this position as a marker of specificity. For this reason I conclude that edin does not fulfil the formal criterion of obligatoriness. This speaks against the article status of this element. 4.2 Semantic criteria for indefinite articles Givón (1981) claims that the transition from indefinite determiner to indefinite article is accompanied by the bleaching out of referentiality. He assumes that this process within the stage of indefinite article is gradual and goes through three substages, which can be represented on an implicational scale. (20) Substages of the stage indefinite article (à la Givón 1981, 49)6 predicative use > generic use > non-referential use in modal and negative scope In these stages markers of indefinite reference lose their referentiality and come to serve some secondary purpose. According to the scale, the use of an indefiniteness marker with generic NPs indicates a more advanced stage than the predicative use, and the use as a non-specificity marker in the scope of modal operators and negation is the most advanced stage in the hierarchy. Although Givón’s general idea of the gradual bleaching out of referentiality is right, I find the order of substages of the referentiality bleaching in (20) questionable. The empirical grounds that led Givón to assume the hierarchy in (20) were a comparison between English and Spanish. The indefinite article in English is an article in the narrow sense since it can be used not only with referential NPs, but 6 Givón considers the use in the scope of a future operator to be a separate group. I lump this use together with the use in modal scope into one group. Ljudmila Geist also with non-referential NPs in the contexts in (20); consider the examples in (21). (21) a. Mary is a nurse. (predicative use) b. A gentleman always opens the door for ladies. (generic use) c. If a woman shows up, let her in, but if a man does, don’t. (modal scope) d. I didn’t read a magazine, I read a book. (negative scope) By comparison, the indefinite article in Spanish can be used with generic NPs in subject positions and also with non-referential NPs in modal scope and in the scope of negation, hence it approximates the terminal stage of the development of English. However, as Givón himself shows, the indefinite article in Spanish cannot be used with non-referential predicate NPs, as shown in (22a). If un is used in the predicate position, the NP is referential, cf. (22b). (22) a. Juán es professor de idiomas. John is professor of languages ‘John is a language professor.’ b. Juán es un professor que encontré el John is one professor that met-I ‘John is a professor I met last year …’ the año passado. year past (Givón 1981, 48, 49) Note that Givón treats both sentences as predicational sentences with predicate NPs despite the difference in their referentiality. In contrast to Givón, I assume that (22a) and (22b) are different types of sentences and only (22a) contains a predicate NP. Sentence (22a) can be understood as an answer to the question “What is John by profession” or “What does John do?”, where the precopular NP is the subject and the postcopular NP describes one of the properties of the referent introduced by the precopular NP, namely his profession. In such sentences, called predicational copular sentences, the predicate noun denotes a property which is predicated of the subject referent. Since predicate nouns denote properties, they are non-referential. Example (22b) is different. It is understood as an answer to the question “Who is John?”. Although this sentence has a copula as a main verb, its complement is not a predicate NP but a referential NP. Such sentences have been called identity or identificational sentences (see Geist 2008 on types of copular sentences). Example (22a) shows that predicate nouns, which are always non-referential, are not accompanied by un. Hence, although un can display a generic function and a non-referential function in modal and negative scope, it has not yet achieved the Bulgarian edin: The Rise of an Indefinite Article stage of predicative use. This situation is not consistent with the implicational scale in (20), which also predicts the predicative use for un. This suggests that in order to reflect the appropriate relation between different non-referential uses the scale must be modified. In what follows I will elaborate on particular points of this scale for edin. 4.2.1 Predicative use As can be seen in examples from Ivanova & Kovaľ (1994) in (23), edin as an indefiniteness marker cannot be combined with predicate nouns: (23) a. Peter e (*edin) učitel. Peter is one ‘Peter is a teacher.’ b. Mladežât e teacher (*edin) vojnik. young-man.def is one ‘The young man is a soldier.’ c. Toj stana (Ivanova & Kovaľ 1994, 59) (*edin) he became one ‘He became a student.’ soldier student. student The sentences in (23) can be understood as answers to the question “What is x by profession?” or “What does x do?”, i.e., these are predicational copula sentences with non-referential predicate nouns in the postcopular position. The example in (24) from Ivanova & Kovaľ (1994) is different. It is understood as an answer to the question “Who is Stojčev?” and states something about his identity. This is an identity sentence. Here the sentence final NP is referential; it refers to an individual. The prepositional phrase po profesija ‘by profession’ as an indicator of the predicate status of the NP cannot be added here. Edin is allowed and indicates, as in other cases we discussed in the previous section, specific indefinite reference. (24) Stojčev e edin žurnalist (kogoto poznavam otdavna). Stojchev is one journalist whom I-know long-ago ‘Stojchev is a journalist whom I have known for a long time.’ (Ivanova & Kovaľ 1994, 60) Consider the minimal pair: (25) a. Toj e (*edin) žurnalist po profesija. he is one journalist by profession ‘He is a journalist by profession.’ (predicational sentence) Ljudmila Geist b. Toj e (edin) žurnalist, kogoto poznavam otdavna. (identity sentence) he is one journalist whom I-know long-ago ‘He is a journalist who I have known for a long time.’ In the predicational sentence (25a) indicated by the modifier po profesija ‘by profession’ edin cannot be used with the predicate NP žurnalist. In the identity sentence (25b) žurnalist is a referential NP. Edin can be used with this NP to indicate specific indefinite reference. Note that in sentences with tova ‘this’ as in (26) the postcopular phrase is the referential subject and tova is the predicate NP. In such sentences, edin as a marker of specific reference can optionally be used with the postcopular subject NP: (26) Tova e (edin) žurnalist, kogoto poznavam otdavna. this is one journalist whom I-know long-ago ‘This is a journalist who I have known for a long time.’ Ivanova (1994) wrongly assumes that sentences with tova ‘this’ like (26) are predicational and the edin NP is a predicate NP. At least one argument can be presented against this assumption. If such sentences were predicational, it would be possible to replace the postcopular phrase with an adjective or a prepositional phrase, since such phrases also qualify as predicates in copular sentences. However, in sentences with tova, APs and PPs are excluded and only NPs can occur. (27) a. Tja e (*edna) žena / umna / v gradinata. (predicational sentence) she is one woman smart in garden.def ‘She is a woman / smart / in the garden.’ b. Tova e (edna) žena / *umna / *v gradinata. (specificational sentence) this is one woman smart in garden.def ‘This is a woman / smart / in the garden.’ The sentences (26) and (27b) are not predicational, they are similar to identity sentences but belong to a separate type, the so-called specificational type. In such sentences the postcopular NP specifies the identity of the referent indicated by the precopular predicate NP (cf. Geist 2008 for this type of copular sentence). The postcopular NP is the subject and is referential. To conclude the discussion so far, predicate NPs denoting a profession or social status cannot be combined with edin. However, the literature on Bulgarian mentions an exception to this generalization. Edin can be used with predicate nouns such as glupak ‘fool’ or genij ‘genius’, which involve some evaluative characterizing component. Consider (28): Bulgarian edin: The Rise of an Indefinite Article (28) Ivan e edin glupak! Ivan is one fool ‘Ivan is such a fool!’ However, according to Ivanova & Koval’ (1994) and Stankov (1984), in this case edin does not indicate specificity but has a so-called ‘intensifying function’. Gorišneva (2009) assumes that characterizing predicate nouns such as glupak ‘fool’, unlike names of professions such as učitel ‘teacher’, are inherently gradable in the sense that they denote properties which can apply to individuals to different degrees: x can be just a fool, a big fool or a complete fool. Combining edin with such scalar nouns yields intensification: edin glupak is, so to speak, more of a fool than just glupak. Elena Karagjosova (p.c.) pointed out to me that the intensificational edin with rising intonation and a special gesture can even be combined with nouns such as učitel ‘teacher’ triggering a meaning shift from a non-scalar property to a scalar property with a pejorative connotation:7 (29) a. Petâr e učitel. (neutral) Peter is teacher ‘Peter is a teacher.’ b. Petâr e edin učitel. (pejorative connotation) Peter is one teacher ‘Peter is a teacher. (intended meaning: Peter is a bad teacher)’ The sentence with edin (with rising intonation and the special gesture) means that Peter cannot teach well. Interestingly, he need not be a teacher by profession. This utterance just evaluates his ability to teach. Note that in other languages such as German, in cases where the indefinite article is optional with predicate nouns, the use of it can trigger a meaning shift of a similar kind: (30) a. Peter ist Athlet. (German) Peter is athlete ‘Peter is an athlete.’ b. Peter ist ein Athlet. Peter is an athlete ‘Peter is an athlete.’ While (30a) means that Peter is an athlete by profession, (30b) states that Peter’s body constitution is athletic. In the latter case Peter need not be a professional athlete. The function of ein is not one of intensification but merely 7 Svetlana Petrova (p.c.) pointed out to me that without special intonation edin still triggers a meaning shift if used with a predicate noun such as učitel. In this case (29b) means that Peter is “not more than just an ordinary teacher”. Ljudmila Geist that of a type shifter which transforms neutral properties into evaluative ones. Interestingly, while with names of professions the article can generally be omitted in German, with evaluative nouns the indefinite article is obligatory: (31) Peter ist *(ein) Idiot. Peter is a ‘Peter is a fool.’ fool Besides these semantically motivated uses there is a fully grammatically motivated use of the indefinite article with non-scalar nouns if the noun is combined with a modifier, such as in ein guter Lehrer ‘a good teacher’. Such only grammatically motivated uses are not available for edin in Bulgarian. To conclude, the analysis of the Bulgarian data shows that the development of edin as a predicative marker is not completed. 4.2.2 Generic use It is known that in languages with indefinite articles such as English, singular indefinites can refer not only to particular objects but can also have generic reference – cf. Krifka (1995), among others. Under “generic use” on the scale in (20) Givón is referring to the uses of indefinite generics in the subject position. (32) a. A gentleman always opens doors for ladies. b. A cat is usually smart. c. A monkey would eat the banana. (32a) is a statement about a typical representative of the kind of gentlemen and (32b) is a statement about a typical representative of cats. Since such generic NPs refer not to particulars but to abstract representatives of a kind, they are commonly assumed to be non-referential. As the Bulgarian counterpart of (32a) in (33) shows, edin must be used with the generic subject. Edin indicates here the identification of a member of a class as a representative of an entire class (Friedman 1976, Ivanova & Koval’ 1994). (33) *(Edin) džentâlmen vinagi otvarja vrata na damite. one gentleman always opens doors prep ladies.def ‘A gentleman always opens doors for ladies.’ Another example of the obligatory use of edin with generic subjects is given in (34). Bulgarian edin: The Rise of an Indefinite Article (34) *(Edin) lekar bi pomognal. one doctor con.part ‘A doctor would help.’ help However, there are other contexts where edin with generic subjects, although its use is preferred, can in principle also be omitted, cf. (35). (35) (Edin) lekar ne bi postâpil one doctor neg con.part act ‘A doctor would never act this way.’ taka. such Such variability in the use of edin with generic subjects can be understood as a signal of the non-stability of this grammaticalization stage for edin. It can be concluded that the use of edin for marking generic subjects has not yet firmly taken root. 4.2.3 Non-referential use in modal and negative scope In this subsection, we consider two other contexts in which NPs can receive a non-referential interpretation: modal contexts and contexts with sentence negation. Modal contexts as used by Givón (1981) are intensional contexts, which involve an intensional operator with respect to which indefinite NPs can take wide scope to yield a referential interpretation or narrow scope to yield a non-referential interpretation. As I have shown in Subsection 3.2.2, indefinites with edin tend to take wide scope with respect to intensional operators and are specific. This is illustrated in (36). Edin obligatorily triggers the referential specific interpretation of the indefinite NP. (36a) can be followed by the continuation (36b), which forces the referential reading of the NP, but it is incompatible with the continuation in (36c), which triggers a non-referential interpretation of the NP. (36) a. Tja iska da se omâži za she wants inf.part refl marry prep ‘She wants to marry a certain Russian man.’ b. Continuation compatible with (36a): edin rusnak. one Russian I know him. c. Continuation not compatible with (36a): #There are no candidates yet. A similar behavior is exhibited by edin-indefinites in sentences with negation that establish an extensional context. Izvorski (1994, 243) points out that indefinites with edin can never be interpreted in the scope of negation. Indeed, sentence (37) with an unstressed edna can only be understood in the Ljudmila Geist sense that there is a particular detail that the person didn’t mention. It cannot be interpreted as “He didn’t mention any details.” (37) Toj ne spomena edna podrobnost. he neg mentioned one detail ‘He didn’t mention some detail.’ In order to be interpreted in the scope of negation, the NP with edin must be accompanied by a negative particle nito: (38) Toj ne spomena nito еdna podrobnost. he neg mentioned neg.part one ‘He didn’t mention any detail.’ detail (38) can be paraphrased as “Not even one detail was mentioned,” highlighting the cardinality. Thus, edin functions here not as a marker of indefinite reference but as a numeral. To conclude, NPs with edin in contexts with modal operators and negation are referential. 4.3 Edin as indefinite article In this section we have concentrated on the issue of whether edin can be used with non-referential NPs – the use typical for indefinite articles. We found two contexts where such a use is possible. First, edin can occur as a marker of generic reference. However, this use has not yet firmly taken root. Second, edin can even be used with predicate nouns in predicational sentences; however, this use is subject to restrictions concerning the type of the noun (primarily nouns denoting scalar properties are involved) and special intonation. A “neutral” use of edin with predicate nouns as we know from English or German is however not possible. Furthermore, edin cannot be used with non-referential NPs in the modal scope and the scope of negation. These findings make it necessary to consider some modifications to the scale of Givón (1981) for the substages of the indefinite article in (20), repeated in (39). (39) Substages of the stage indefinite article (à la Givón 1981, 49) predicative use > generic use > non-referential use in modal and negative scope Based on the evidence from Bulgarian, the following modifications to the implicational subscale for indefinite articles can be assumed, cf. (40). Bulgarian edin: The Rise of an Indefinite Article (40) Substages of the stage indefinite article predicative use > generic use non-referential use in modal and negative scope > the generalized article First, since we found no evidence for the assumption that the generic use is more advanced than the predicative use, we can consider the stages predicative use and generic use to be parallel stages. Edin is beginning to develop both uses, but this development is not completed. From this I conclude that these uses are hierarchically equal. The parallel representation of these uses in (40) also allows languages to skip one of the uses, the generic or the predicative use, on the way to the next stage. This is what seems to be the case in Spanish, where the indefinite article has already achieved the non-referential use in modal and negative scope, and can be used with generic NPs, but cannot occur with nonreferential predicate nouns. Second, for completeness I have added the stage generalized article from Heine (1997) as the terminal stage on the scale. At this stage the indefinite article can be combined with all kinds of nouns, irrespective of whether they are count or mass – with the effect that any number-specific behavior is eradicated. Bulgarian edin has not achieved this stage. It can only be used with count nouns. Mass nouns accompanied by edin obligatorily shift to count nouns; consider (41). (41) Az piex edin čaj. I drank one tea ‘I drank one (cup of) tea.’ In this example, edin čaj refers to a discrete countable portion of tea or a kind of tea (e.g. Ceylon tea), i.e., čaj is shifted from mass to count noun. In the former case edin serves as a numeral, while in the latter case it is ambiguous between a numeral and an indefinite determiner. 5 Conclusion This paper had the aim of determining the status and the semantic contribution of the indefiniteness marker edin in Bulgarian. I tested the hypothesis by Friedman (1976) that Bulgarian edin is an indefinite article. According to Givón (1981) and Heine (1997) numerals pass through different intermediate stages before they become indefinite articles. At intermediate stages the marker of Ljudmila Geist indefinite reference can only be combined with referential NPs. It does not have the status of an indefinite article but the status of an indefinite determiner. At the terminal stage, the stage of the indefinite article, the indefiniteness marker can also occur with non-referential NPs. The question was to what extent edin has developed from the numeral through the stage of the indefinite determiner towards the stage of the indefinite article on the integrated grammaticalization scale of Givón (1981) and Heine (1997). The examination of edin has shown that it has entirely achieved the stage of the indefinite determiner since it can be used as a marker of discourse persistence or specificity with referential NPs. However, it is not the terminal stage for edin but further development on the grammaticalization scale towards the indefinite article was attested. Edin is beginning to spread to non-referential uses, since it can occur with generic and predicative phrases. However, unlike the indefinite articles in English and German, edin cannot be used with non-referential NPs in the modal scope and the scope of negation and cannot be combined with mass nouns without a meaning shift. For this reason edin cannot yet be assessed as a “true” indefinite article. Also formally edin does not behave as a “true” article since it is not obligatory with NPs in argument positions. Bare NPs without edin can receive an indefinite interpretation. If edin is added in argument NPs, it indicates discourse persistence or specificity. We also made some progress in deepening our understanding of relations between various functions that indefiniteness markers may have in different languages. I applied work on the grammaticalization and typology of indefinite articles in which different functions of indefiniteness markers are seen as stages in the process of grammaticalization. In this work the diachronic stages of indefiniteness markers are represented on an implicational scale. However, the diachronic stages can also be seen as synchronically available functions of indefiniteness markers in an implicational hierarchy. This means that an indefiniteness marker at a given stage may synchronically have the functions of all the preceding stages. The findings concerning the current functions of edin in Bulgarian led us to modify the implicational subscale proposed by Givón (1981) for the stage of indefinite article. The resultant integrated scale can be represented as below: Bulgarian edin: The Rise of an Indefinite Article (42) Stages and functions of markers of indefinite reference the nuthe speci> the presentameral tive marker > ficity > marker I. numeral II. indefinite determiner predicative use generic use non-referential use in > modal and negative scope > the generalized article III. indefinite article Bulgarian edin On this scale the different functions which indefinite determiners may have are represented in an implicational hierarchy. The box for Bulgarian edin indicates its development status on the scale. The examination of more languages is necessary to show whether this integrated scale has cross-linguistic validity. References Abusch, D. & Rooth, M. (1997): Epistemic NP modifiers. In: Abusch, D., Rooth, M. & Lawson A. (eds.): Proceedings of SALT VII. Itaca, NY, 1-18. Carlson, G. (1977): Reference to Kinds in English. Ph.D. diss., UMass [published 1980: New York: Garland]. Chiriacescu, S. & von Heusinger, K. (2010): Discourse prominence and pe-marking in Romanian. International Review of Pragmatics 2.2: 298- 332. Dyer, D. L. (1993): Determinedness and the pragmatics of Bulgarian sentence structure. The Slavic and East European Journal 37.3: 273-292. Fodor, J. & Sag, I. (1982): Referential and quantificational indefinites. Linguistics and Philosophy 5: 355-398. Friedman, V. A. (1976): The questions of a Bulgarian indefinite article. In: Butler, T. (ed.): Bulgarian Past and Present. Columbus: American Association for the Advancement of Slavic Studies, 334-339. Geist, L. (2008): Predication and equation in copular sentences. Russian vs. English. In: Comorovski, I. & von Heusinger, K. (eds.): Existence. Semantics and Syntax. Dordrecht: Springer, 79–105. Geist, L. (in prep.): The rise of an indefinite article. The case of Russian odin and Bulgarian edin. Givón, T. (1981): On the development of the numeral ‘one’ as an indefinite marker. Folia Linguistica Historica II.1: 35-53. Ljudmila Geist Givón, T. (1983): Topic continuity in discourse. An introduction. In: Givón, T. (ed.): Topic Continuity in Discourse. A Quantitative Cross-Language Study. Amsterdam, Philadelphia: John Benjamins, 1-41. Gorišneva, E. (2009): ONE. Between numeral, indefinite marker and intensifiers. In: Saxena, A. & Viberg, Å. (eds.): Multilingualism. Proceedings of the 23rd Scandinavian Conference of Linguistics. Uppsala University 1-3 October 2008. Uppsala: Uppsala University, 37-50. Gorišneva, E. (in prep.): Bare vs. non-bare nouns. Two kinds of indefinites in Bulgarian. Ms. Heine, B. (1997): Cognitive Foundations of Grammar. New York, Oxford: Oxford University Press. Ionin, T. (in press): Pragmatic variation among specificity markers. To appear in: Hinterwimmer, S. et al. (eds.): Volume on “Funny Indefinites”. Dordrecht: Springer. Ivančev, S. (1957): Nabljudenija vârxu upotrebata na člena v bâlgarskija ezik. Bâlgarski ezik 7: 499-529. Ivanova, E. Ju. (1994): Neopredelennyj artikľ EDIN v imennyx prisoediniteľnyx konstrukcijax. Vestnik Sankt-Peterburgskogo Universiteta 16: 112-115. Ivanova, E. Ju. & Kovaľ, S. (1994): Bolgarskoe EDIN s točki zrenija referenciaľnogo analiza. Vestnik Sankt-Peterburgskogo Universiteta 23: 58-64. Izvorski, R. (1994): On the semantics of the Bulgarian “indefinite article”. In: Avrutin, S., Franks, S. & Progovac, L. (eds.): Formal Approaches to Slavic Linguistics: The MIT Meeting 1993. Ann Arbor, Mich.: Michigan Slavic Publications, 235-254. Jäger, G. (2001): Topic-comment structure and the contrast between stage level and individual level predicates. Journal of Semantics 18: 83-126. Krifka, M. (1995): Introduction to genericity. In: Carlson, G. N. & Pelletier, F. J. (eds.): The Generic Book. Chicago: Chicago University Press, 1-124. Le Bruyn, B. (2010): Indefinite Articles and Beyond. Utrecht: LOT. Nikolaeva, T. M. (1979): Slovosočetanija s leksemoj “odin”. Forma, značenija i ix kontekstnaja markirovannosť. In: Zolotova, G. A. (ed.): Sintaksis teksta. Moskva: Nauka, 134-152. Reinhart, T. (1981): Pragmatics and linguistics. An analysis of sentence topics. Philosophica 27: 53-94. Šmelev, A. D. (2002): Jazyk i vnejazykovaja dejstviteľnosť. Moskva: Studia Philologica. Stankov, V. (1984): Za kategorijata na neopredelenost na imenata v bâlgarskija ezik. Bâlgarski ezik 3: 195-205. Weiss, D. (2004): The rise of an indefinite article. The case of Macedonian eden. In: Bisang, W., Himmelmann, N. P. & Wiemer, B. (eds.): What Makes Grammaticalization? A Look from Its Fringes and Its Components. Berlin, New York: de Gruyter, 139-165. Wright, S. & Givón, T. (1987): The pragmatics of indefinite reference. Quantified text-based studies. Studies in Language 11: 1-33. [email protected]
© Copyright 2026 Paperzz