Semantic and Syntactic Realisation of the Incremental Theme (with a Focus on Bulgarian)

Abstract

This article presents ongoing work on the aspectual properties of verb predicates, in particular, the classes of activities and accomplishments. Herein, we focus on incremental theme predicates, starting with consumption verbs as one of the representative subclasses of incremental accomplishments. We explore, in detail, the semantic, referential, quantisation and morphosyntactic properties of incremental Themes and their realisation in Bulgarian. The analysis is based on original empirical data and enabled us to identify the common features shared with widely studied languages such as English and Russian, as well as to establish language-specific features typical for Bulgarian. We hope that our findings may contribute to the study of aspectual classes in a cross-linguistic perspective.

1. Introduction

The theoretical study of the aspectual properties of verbs, their semantic representation, and the semantic and morphosyntactic realisation of their arguments have been subject of intensive research in recent decades, since the works by Vendler (1957, 1967), Dowty (1979) and many others. Numerous efforts have been invested into justifying a comprehensive aspectual classification that would have validity across a large number of languages. Slavic linguistics has contributed to the field with the in-depth study and the unique perspective on the lexical and grammatical aspect in the Slavic languages with the works of Smith (1997), Filip (1999, 2005, 2008), Filip and Rothstein (2006), Braginsky and Rothstein (2008), Rothstein (2012), Paducheva (1996, 2009), Tatevosov (2015, 2016), Kabakchiev (1992), Charalozova (2009), among a vast array of researchers working in various theoretical frameworks.

This paper aims to contribute to the multilingual perspectives on the analysis of the aspectual properties of activity and accomplishment predicates by providing a detailed analysis of incremental theme verbs and their realisation based on original corpus material in Bulgarian. We study the semantic and the morphosemantic properties of the incremental theme, and analyse them in the context of the existing views put forward for English, Russian, and other languages.

The paper is organised as follow: First, we briefly define the focus of the study and the objectives in Section 2. In Section 3, we overview some of the main theories on incrementality and the associated notion of scale, the classification of aspectual classes and, in particular, the properties of accomplishments and activities, and present some observations on the Slavic languages and Bulgarian. Section 4, provides a detailed analysis of the class of simplex imperfective verbs of consumption as representative for the larger class of incremental Theme verbs. This section focuses on the various types of expression of the Theme occurring in the corpus data: non-overt singular mass and singular count complements with and without over quantisation, singular complements with the zero article, the indefinite article and the definite article zero-article plural, quantified plural, and definite plural Themes. The specific features of the use of perfective verbs of consumption, in contrast to simplex imperfective verbs, are presented in Section 5. Further observations on the broader class of incremental Theme verbs are briefly discussed in Section 6 to the end of outlining additional features and nuances in their realisation. The last section sums up the most important research questions raised in the course of the analysis and outlines directions for future research aimed towards a better understanding of incremental Theme verbs in a cross-linguistic perspective.

2. Motivation and Objectives

The focus of this study is the realisation of the incremental Theme with several semantic classes of transitive accomplishment verbs. They have posed various challenges to linguistic description, starting with the question whether such predicates are actually accomplishments, whether a class of accomplishments should be defined at all, or they should be considered complex predicates derived from verbs belonging to other classes. This last consideration is related to the fact that the predicates in question manifest different properties in different contexts. Consequently, they can be classified in more than one aspectual class, in particular, as either activities or accomplishments, depending on the interplay of factors: (i) the syntactic expression or implicitness of their complement; (ii) the complement’s semantic properties (taking into account the count vs. mass nouns distinction and nouns’ referential, quantisation and other properties); and (iii) morphosyntactic features, such as plurality, definiteness, etc.,

Additionally, even the combination with temporal adverbials, in particular durational for- and/or time-frame in- prepositional phrases, is not always a reliable and conclusive way to assign the verb phrase to a particular aspectual class. This is valid for English, but even more so for Bulgarian (and presumably, for other Slavic languages) as we will demonstrate below with illustrative examples.

The syntactic effects and variability of interpretations of accomplishments are among the primary reasons that have led researchers to conclude that the aspectual classes should be considered at the VP level (through aspectual composition or another mechanism) rather than at the lexical level. Stemming from the research of Dowty (1979), who relates word meanings to Vendler’s aspectual classes via lexical decompositions that reflect the logical structures of sentences, more recent research has focused on various approaches to explain the aspectual properties in terms of the complex interaction of the lexical, semantic, morpho-syntactic and valence-related features of verbs and their (internal) arguments Verkuil (1993), Krifka (1989, 1992, 1998), Smith (1997), Filip (1999, 2005, 2008), Filip and Rothstein (2006), Rothstein (2004), Braginsky and Rothstein (2008), Rothstein (2012), Levin (1993), Rappaport Hovav (2008), among many others.

Furthermore, some of the assumptions that have been taken for granted for decades have been subjected to reconsideration along two different avenues: the study of a more diverse inventory of verbs in English and research into the aspectual classes in other languages, especially the Slavic ones.

With respect to analysing the verb aspect and the realisation of accomplishment verbs in Bulgarian, our observations on the theoretical analyses in the literature show some deficiencies to whose overcoming our research aims to contribute the following:

• The extensive studies on the aspectual classes of verbs are very often based on English, less on the Slavic and other languages. To the best of our knowledge, there are no extensive studies on Bulgarian, especially in contrast to English or Russian.
• Major large-scale studies offering classifications of verb predicates in the Slavic languages with a view to their aspectual properties are predominantly focused on Russian (or possibly other Slavic languages) but rarely consider Bulgarian examples.
• When Bulgarian examples are provided in studies, they are predominantly constructed, especially in the context of applying diagnostic tests for determining the aspectual properties of the verb. Our observations on the use of verbs and the variety of their realisations are based on corpus data and thus provide more reliable conclusions on the distribution and plausibility of different realisations of the verb and the theme.

Our observations are based on the lexical and grammatical expression of verbal complexes headed by incremental theme predicates with relevance to Bulgarian, which, on the one hand, exhibits the features characteristic for the Slavic languages, but on the other hand, shows some language specific peculiarities, especially with view to the category of definiteness Nicolova (2008, 2017).

Our analysis focuses on verbs of several semantic classes: verbs of consumption (where the Theme is gradually consumed and used up), verbs of reading (where parts of the Theme may be subjected to the event more than once), verbs of writing and creating (where the Theme comes into existence gradually), and verbs of cleaning and decorating (where the Theme is gradually affected in the course of the event). The variety of the analysed classes illustrates different ways in which the Theme is affected: depleting consumption and non-depleting usage, creation and coverage or extent of affect. The aim is to observe the commonalities and differences in the expression of the Theme across prototypical incremental theme verb classes.

Verbs of consumption serve as a case study for outlining the most significant features determining the realisation of the verb and the Theme and the overall interpretation of the respective sentence. The analysis of the remaining classes of verbs aims to provide additional details and nuances to the semantic and syntactic properties of incremental theme verbs and their complements.

Drawing on the literature review, we aim to analyse the data with respect to answering the following research questions:

Q1.  

Are the theoretical statements and the observations on English, Russian, and other languages in the reviewed literature confirmed for Bulgarian and attested in the corpus data?

Q2.  

Are there any language-specific features shown in the realisation of the predicate and the incremental theme in Bulgarian, which have not been observed in the major studies on other languages?

Our observations are based on original examples from the Bulgarian National Corpus (Koeva et al., 2012), in particular, from two subdomains—Fiction and News, excluding specialised domains such as science and non-normalised texts such as subtitles (where sentence boundaries are not always clear). We only select full sentences (marked by sentence punctuation) with a length of up to 60 tokens, thus excluding malformed chunks of text. We identify sentences containing the verbs under analysis—we have collected examples of the following: (i) simplex imperfective verbs, (ii) prefixed perfective verbs, and (iii) secondary imperfective verbs derived from prefixed perfective verbs through a process of suffixal imperfectivisation. While we primarily focus on the first and the second, we also use examples of secondary imperfectives to validate some of the assumptions made.

Sentences are annotated in the following way: (i) the verbs under observation are labelled in the sentences, and are assigned grammatical description—aspect, person and number, tense, i.e., features that are later used in the analysis; (ii) the direct object NPs are identified and assigned a grammatical features, such as number and definiteness; (iii) the external arguments are marked and labelled with grammatical features; (iv) temporal and manner adverbial phrases are also annotated. The suitable examples are validated and selected manually.

The corpus data are used first of all as a source of examples illustrating the different contexts and realisations of the verbs under study, but they also provide insights into the frequency and distribution of certain verb–theme patterns over others, including in terms of the attested combinations of grammatical features of the verb and the object nominal.

Further statistical analysis of the data would enable important insights into the distribution of the realisations of the incremental theme verbs, but these would require extensive efforts for reliable selection of examples and automatic annotation of the complements, which falls outside of the scope of the present study.

3. Related Works

Here, we discuss some of the main studies related to our research, with the aim of constructing a framework of theoretical concepts and assumptions that guide the analysis of verb classes in the subsequent sections. First and foremost, we pay attention to the aspectual classes and the properties of the Theme, in particular, its incrementality. To this end, we also discuss the notions of scale and quantisation. The different expressions of the Theme as a non-quantised or quantised entity are reviewed with respect to the interpretation of the verb phrase in terms of its boundedness and telicity. Another aspect brought to attention is the Theme’s referentiality, which is central to the interpretation of the Theme and the properties of the verb phrase it complements (and those of the respective sentence as a whole).

In addition, we present studies focused on Slavic languages and specifically on Bulgarian so as to provide context for the analysis of relevant language-specific features, in particular, those that diverge from the properties of the verbs belonging to the same semantic classes in English.

3.1. Activities and Accomplishments

We take as a point of departure the assumption that, as proposed by Vendler (1967), predicates fall into (at least) four widely accepted classes—states, activities, accomplishments, and achievements—based on their aspectual properties as reflected in the restrictions they impose on time adverbials and tenses and the applicable logical entailments (Dowty, 1979, p. 54)1. In particular, we focus on a specific category of predicates, the so-called incremental theme verbs (Dowty, 1991; Krifka, 1989, 1992, 1998), and subsequent literature, which describe eventualities in which the internal argument of the verb undergoes an incremental change or is incrementally affected over the course of the event. This category subsumes different thematic classes, among which classical examples include consumption verbs, creation verbs, and other (non-strictly) incremental/incremental affect verbs, cf; for instance, Kennedy (2012, p. 103) and references therein.

We particularly focus on the verbs that may shift between an activity and an accomplishment reading, depending on whether their nominal complement is implicit or overtly expressed, and in the latter case—what its semantic, morphological, referential, and possibly other features are. This variable behaviour, and possibly other factors, have led many linguists to question the existence of a distinct class of lexical accomplishments, and to consider the respective verbs as interpretations of predicates belonging to other lexical classes (Rothstein, 2012, p. 60).

Specifically, Verkuyl (1972), Dowty (1979, pp. 61–62), Krifka (1989, 1992, 1998), Van Valin and LaPolla (1997), among others, have delved into the nature of the alternation between an activity and an accomplishment reading (Dowty, 1979, pp. 61–62). There has been a considerable consensus that the verbs in question have an activity reading when heading an atelic VP, i.e., when the direct object is a bare plural or a mass noun, and a telic (accomplishment) reading otherwise, i.e., when heading a VP expressed by a quantified or definite object, compare He drank beer (activity) and He drank a beer (accomplishment) or He ate a sandwich (accomplishment) and He ate sandwiches (activity). Some authors, have explicitly assumed that in these cases the basic predicate structure must be an activity from which the accomplishment is derived. For instance, Van Valin and LaPolla (1997) do so, terming the derivations active accomplishments, thus distinguishing them from accomplishment verbs that do not alternate.

An important point made in the same work, which bears relevance to the following discussion, is that the different interpretations are not a mere function of the properties of the object (i.e., whether it has an article or not) and that cross-linguistically the activity-accomplishment contrast may be coded on the verb itself and not marked on the object NP at all (Van Valin & LaPolla, 1997, pp. 99–100).

Although from a different perspective, Rappaport Hovav (2008) also argues against accomplishments being a distinct lexical class. In particular, she assumes that lexical predicates are best characterised in terms of certain features. Thus, achievements, states, and activities are verb classes associated with coherent sets of properties (states are non-dynamic, activities are dynamic but non-scalar, and achievements are lexically associated with two-point (+/−) scales), while accomplishments cannot be straightforwardly characterised by any general properties and thus do not form any well-defined lexical class. The author concludes that what is considered to be accomplishments are complex predicates headed by verbs belonging to other lexical classes (mostly activities) which are associated with extended (multiple point) scales.

Contrarily, other authors such as Smith (1997), Rothstein (2004), Braginsky and Rothstein (2008), and Rothstein (2012) argue in favour of the independent status of the lexical class of accomplishments. Rothstein (2004) and Braginsky and Rothstein (2008), in particular, suggest that the classification of predicates into states, activities, achievements and accomplishments reflects properties of the denotations of the respective verbs. The properties of the VPs, such as the ones attributed to the activity-accomplishment distinction, follow from the fact that verbs belonging to different lexical classes and denoting different kinds of event types interact with direct objects and modifiers in different ways.

In particular, Rothstein (2012, pp. 71–74) demonstrates that accomplishments cannot be subsumed under the class of activities, as shown by a number of diagnostics, such as the imperfective paradox, going back to Dowty (1979). She also shows that there are both activities with non-bare direct objects and accomplishments with obligatory objects (i.e., ones disallowing an activity interpretation) and hence the two classes cannot be equated. Furthermore, Rothstein (2012) adduces two pieces of evidence that bear relevance to our study: (i) the findings by Braginsky and Rothstein (2008) that Russian accomplishments and accomplishments alone allow modification by expressions like пoстепеннo (postepenno) ‘gradually’ and X за X (X za X) ‘X by X’, e.g., sip by sip, due to these modifiers’ sensitivity to the inherent incremental structure of accomplishments; (ii) the fact that perfective accomplishment verbs denote telic events independent of their direct objects (i.e., as part of their lexical meaning), a conclusion which is also in line with the typological observations of Van Valin and LaPolla (1997) quoted above. The author takes that all accomplishments, whether alternating between an activity and an accomplishment reading or not, have an internal incremental structure consisting of a series of incremental events held together by an incremental chain which charts the progress of the event, where the role of the direct object is to denote the entity on which the changes can be marked (Rothstein, 2012, p. 93).

Based on observations on data from Bulgarian, we side with the authors who propose that accomplishments form a distinct class and that verbs alternating between an activity and an accomplishment reading, are accomplishments that have a derived activity interpretation. The modification by adverbial expressions sensitive to the incremental structure tested for Russian is straightforwardly applicable to the corresponding Bulgarian verbs. We will pay particular attention to both imperfective and perfective accomplishment verbs and the interaction with their NP complements and aspectual modifiers in Section 4 and Section 5.

The focus will be on prototypical imperfective verbs belonging to the semantic classes of consumption, creation and affect, which are characterised by an incremental Theme, as well as degree achievements, which are lexically associated with a property scale that measures the degree of change in the respective property.

In a broader sense, many authors link the aspectual properties of verbs with the notion of scale (Dowty (1991), Rappaport Hovav and Levin (2005), Rappaport Hovav (2008), and Levin (2010), to mention but a few). In the realisation of the majority of verbs under consideration in this article, incrementality and the scale representing it are linked to the incrementality of the Theme. The only exception among the analysed predicates are degree achievements whose scale is defined by the possible values of the property underlying their meaning.

As noted above, although not considering them to constitute a separate class, Rappaport Hovav (2008, pp. 16–20) points out that the scale, and, in particular, an extended, multi-value one, is the key semantic property distinguishing what is thought of as accomplishments from other verbs.

Tenny (1994), Krifka (1998), Dowty (1991), Levin (2010), and Rothstein (2012) have discussed the fact that the extent (defined by the scale) of the incremental Theme argument of an accomplishment predicate determines, at least in part, how long the event lasts and when it has reached its culmination. However, the homomorphism between the event and the scale of the incremental Theme (as defined by Krifka (1992), Krifka (1998)) while working in principle, is not always valid as themes are involved in different accomplishment events in very different ways (Filip, 1999; Rothstein, 2004). Krifka (1998) recognised the strictness of the requirements of the homomorphism, which makes it impossible to cover all cases, while Rothstein (2004) argues that it is not the extent of the theme that usually determines the extent of the event. According to her, an accomplishment predicate denotes a set of events leading incrementally towards an endpoint; the sequence of events is structured by a prototypical process in such a way that at any point it can be checked whether the event is still ongoing. The culmination is then defined as the final minimal event in an incremental process Rothstein (2004, p. 106).

Below we will discuss the interrelation between the notion of scale and measuring the progress of an eventuality.

3.2. Quantisation and Cumulativity of the Theme

In essence, quantisation is involved in the measuring out of the event (Tenny, 1992) and helps provide an ordering criterion, or a scale, whose degrees serve to track the progress of the event (Filip, 2008; Filip & Rothstein, 2006). We are looking at the incrementality of the event in direct relation to the notion of quantisation and, in particular, quantisation as a prerequisite for maximalisation. With regards to the class of verbs with incremental Themes, the quantisation of the event is closely linked to the quantisation of the Theme, as captured in Krifka’s homomorphism (Krifka, 1998) and further extensions proposed by Rothstein (2004).

The notions of quantisation and cumulativity, which are closely related, are actually the basis for distinguishing accomplishments from activities. Rothstein (2008, p. 56) argues that all dynamic verb predicates are inherently countable (quantised) as they denote countable number of instantiations of states. This implies then that all dynamic verbs can express atomic (telic) meaning denoting a single event in some context under a certain description (Landman, 2000; Rothstein, 1999). However, accomplishments and activities exhibit also the property of having stages, and this is why they both can appear in the progressive (Rothstein, 2004, pp. 17–23). Moreover, the activities exhibit the distinctive feature of cumulativity, while accomplishments do not. Cumulativity is better suited to explain atelicity than homogeneity, since only states are truly homogeneous, while activities, in general, are homogeneous down to intervals of minimal size (Dowty, 1979, pp. 166–172) (i.e., eating apples is the sum of all minimal intervals of eating an individual apple, rather than a continued, homogeneous event). While quantisation is about whether the event has identifiable countable stages, often expressed in terms of the incrementality of the Theme, cumulativity is about creating a new singular event out of the sum of individual events, thus creating a mass predicate (Rothstein, 2004, pp. 8–9).

Rothstein (2008, p. 56) states that there is a contrast between those verbal predicates where a (unit) measure can be constructed on the basis of the meaning of the verbal predicate (or based on the scale associated with the Theme), and those predicates where such a measure is not readily available. This contrast is the semantic basis of the telic/atelic contrast at the lexical level: telic predicates are those for which a value for the unit measure is explicitly (or in some cases implicitly) provided, while atelic predicates are those where such a value cannot be constructed based on the predicate or the Theme alone (without any additional context).

Furthermore, Rothstein (2010, pp. 361–371) has proposed a grammatical definition of countability in terms of semantic atoms, entities that count as one relative to a particular context. Count nouns differ from mass nouns both in terms of semantics, and in terms of grammatical behaviour. For some mass nouns, a logical operator COUNT can be introduced to establish countability, i.e., to select individual entities within the mass, and derive a count reading of the mass noun.

The plural form of countable nouns (and NPs) generally expresses the plurality of objects, naturally quantised. Bare plurals, although a quantised entity, often behave as mass nouns (which are singular but express a plural, generally cumulative entity). Chierchia (1998) suggests that mass nouns are in fact lexical plurals despite their singular morphology, as they denote a set of minimal elements or a sum of unit quantities (considered as individual entities). Thus, the introduction of cumulativity applies to both mass and plural count nouns, and can turn plural entities into a singular cumulative one.

Filip (2005, pp. 97–98) also argues further that the use of definite noun phrases, either mass or count, may receive a quantised and quantity interpretation. However, it is not determined by the morphosyntax, but depends on pragmatic principles of interpretation and world knowledge, e.g., in the sentence Ivan ate the soup, the object soup applies to some particular quantity, for example, delimited by some contextually specified container (a portion, a bowl, etc.). These cases can be classified as implicit or pragmatic quantisation, and analysed in line with cases of explicit quantisation. Kennedy (2012, pp. 116–117) also states that the measure functions providing quantisation are relative to the situations described by the predicates. He views variable telicity of incremental theme verbs as ‘a function of the standard of comparison relative to which incremental change is measured’ (Kennedy, 2012, p. 120): whether it meets a minimum standard (any amount of change, atelic interpretation), or whether it has to reach a maximum standard specified by a measure phrase (specific amount of change, telic interpretation).

At the level of the verb predicate, since singular mass NPs are considered to be cumulative nominals (Krifka, 1998, p. 3), and they do not refer to a particular, physically delimited entity participating in the discourse, but rather specify some intrinsic facet of the meaning of the predicate (Van Valin & LaPolla, 1997, p. 123), the relevant NPs do not participate in the measuring out the progress of the event, and the event itself is viewed as a whole, i.e., an activity, rather than as a series of subsequent subevents that lead to the unfolding of the situation towards its culmination (for a detailed formal account cf. Rothstein (2012, pp. 74–80)), i.e., an accomplishment. As a result, VPs with cumulative nominals receive an activity-like atelic interpretation, as discussed by Rothstein (2012, p. 65), Van Valin and LaPolla (1997), and many other studies.

Measure phrases and non-vague quantifiers always produce quantised nominals. NPs modified by adverbials or numerals denoting an approximate or undefined quantity, while not fulfilling the quantisation criteria of Krifka’s formulae, still lead to telic VPs as shown by Filip (2005), Rothstein (2004), among others, e.g., at least three apples, at most three apples, an infinite number of apples, a few apples, a lot of apples. Rothstein (2012, p. 64) considers these VPs to be telic, denoting a single atomic event. She follows Landman (1996) who introduces a logical operator that applies to any plural (regardless whether it is a precise or approximate quantity and transforms a collective of entities (plural) into a summative entity (singular).

To summarise the previous discussion, quantisation refers to representing the whole Theme as a set of subparts where the predicate applying to any proper subpart cannot be generalised to cover the whole (Filip, 2005), (Krifka, 1989, p. 92), (Krifka, 1992). The authors point out that count nouns are naturally quantised entities but quantisation applies to mass nouns as well, which can be quantised and quantified using modifiers that represent measure expressions or quantifiers (two sandwiches, all the sandwiches, and two pints of beer) (Filip, 2005). Cumulativity, on the other hand, also applies to both count and mass nouns, turning plural entities into one summative entity, which creates a mass predicate.

3.3. Referential and Non-Referential Theme

Referentiality of the Theme is another semantic characteristic that is related to the realisation of the verb predicate with respect to its aspectual class.

Van Valin and LaPolla (1997) suggest that the basic (simplex) verbs are activities heading atelic VPs when the direct object is a bare plural or a mass noun, and telic (accomplishment) VPs otherwise, i.e., when their direct object is expressed by a quantified or definite NP (see Section 3.1). When such a predicate has an activity interpretation, its second argument is non-referential, meaning that it cannot be interpreted as having any specific reference: e.g., He ate pizza (for ten minutes) does not refer to any specific pizza (it could include a variety of different pizzas in the same event), as opposed to He ate a slice of pizza (in ten minutes) (Van Valin & LaPolla, 1997, pp. 122–123). Non-referential objects are non-specific, indefinite, generic or mass (Van Valin & LaPolla, 1997, p. 99), which, in English, translates as mass nouns or bare plural complements in terms of morphosyntax. Referential objects (specific, quantified) serve to delimit the event, thus establishing a reading as an active accomplishment.

Rothstein (2012, pp. 93–94) states that mass and bare plural complements do not denote a particular entity (i.e., one having an actual referent in the discourse), but rather a kind term (i.e., a non-referential entity), and that in these cases it is difficult to posit the Theme as a benchmark against which to evaluate the progress of the situation. Consider the sentence John ate apples for two hours. It actually asserts that the subject stands in a particular relation to the kind denoted by the Theme, e.g., John stands in the eating relation to the apple-kind, but the characteristics of the kind are not affected by the event. As it is through the change affected upon the Theme that the event is measured out (Filip, 2005, pp. 98–99), sentences with such complements have an activity rather than an accomplishment reading.

As already mentioned above (see Section 3.2), definite NPs are considered to be quantified in the particular situation described by the predicate Filip (2005, pp. 97–98). Kennedy (2012, p. 104) also points out two cases of the internal argument (the incremental theme) being of specified, definite quantity—when it includes an explicit quantity term (e.g., Kim drank a glass of beer in an hour) or when it is referential (e.g., Lee painted the house in two weeks), both resulting in the predicate being telic.

Given that according to some accounts not quantisation but referentiality is the main condition that needs to be met by an NP complement for a VP to realise an (active) accomplishment reading Van Valin and LaPolla (1997, pp. 122–124), the view of quantisation might need to be integrated with the notion of referentiality (also supported by Filip (2005), Kennedy (2012), among others).

In summary, referentiality is considered to imply quantisation and quantification, which can either be explicit (expressed by count nouns, quantifiers, etc.) or implicit, based on pragmatic contextual factors and understanding of the world (e.g., coffee is measured in cups, and my coffee in the sentence I drank my coffee refers to one cup of coffee). Non-referential objects, typically expressed as mass nouns or bare plurals, do not provide a means to measure out the progression of the event since they are non-specific and not quantified.

3.4. Referentiality in Bulgarian

To the best of our knowledge, the notion of quantisation has not been explored systematically in Bulgarian linguistics, and has only been touched upon in discussions on its role in the realisation of VPs headed by verbs belonging to different aspectual classes in studies adopting or reflecting on the Vendlerian classification in the context of Bulgarian. On the other hand, the notion of referentiality has received more systematic accounts in relation to its linguistic expression and the role of articles in Bulgarian in the interpretation of the referentiality of NPs.

We follow the description proposed in one of the definitive works on Bulgarian morphology by Ruselina Nicolova (Nicolova, 2008) and its English translation (Nicolova, 2017). Below we sum up the main points made there. In line with current literature, the author distinguishes between referential and non-referential use, which represent different explications of the relationship between the referent and the world. With the referential use, there exists in the real or in some possible world an object that relates to the noun phrase; the NP itself is an argument (subject or object) or an adjunct (an adverbial modifier) to the predicate in the semantic structure of the sentence. The non-referential use, which occurs with nouns modified by a zero article or, in specific cases by the indefinite article един (edin) ‘one/some’ (but not by a definite article), expresses the propositional function to be x and is a predicate in the semantic structure (main or secondary predicate in a complex argument). Thus, Тoй е писател (Toy e pisatel) ‘He is a writer’ is used non-referentially, because it only expresses the function of being a writer and does not refer to an existing individual of the class of writers, unlike Вчера гoвoрих с писател (Vchera govorih s pisatel) ‘Yesterday I spoke to a writer’. In the latter sentence, писател (pisatel) is also used with the zero article, but it has a referential meaning, as it denotes an individual—in this case, one conceptualised only by means of his or her generic properties as related to his or her activities or function. Syntactically, a non-referentially used noun with a zero article (or rarely with един (edin) ‘one’), is a nominal part of the predicate (a predicative), a subject or object complement, or a nominal group with a different function, most often a modifier (Nicolova, 2017, p. 133).

According to this account, non-referential NPs in an object position are primarily complements of light verb constructions, such as вземам решение (vzemam reshenie) ‘make a decision’ or имам надежда (imam nadezhda) ‘have hope’ and of verbs which, together with the object, designate a particular type of activity, such as играя шах (igraya shah) ‘play chess’, ръкoвoдя кoмпания (rakovodya kompaniya) ‘run a company’, etc. (Nicolova, 2017, pp. 138–139). In light of the discussion in this article, we can note here that decision and hope function as predicative nouns, the semantic predicates in the respective light-verb constructions, while play chess may qualify in a suitable context as an activity reading of the accomplishment verb играя (igraya).

The referential use of nominals subsumes non-specific and specific reference (Nicolova, 2017, p. 143). The former refers to a type of reference where a noun phrase with a zero article is used in which the speaker abstracts away from the individual properties of the referent and only refers to its generic traits. Specific reference designates that the speaker is referring to the individual properties of the referent. Specific definite NPs are used when the individual properties of the referent are viewed as known to both speaker and hearer, and indefinite NPs, i.e., ones with the indefinite article един (edin), are employed when the individual properties of the referent are thought of as unknown to the hearer while they are most often known to the speaker. Non-specific noun-phrases are neutral with respect to the feature ‘definiteness’.

All three members of the article system, i.e., the definite article, the indefinite article and the zero article can be used with NPs headed by all kinds of common nouns, such as concrete, countable, uncountable, abstract, names of substances and collectives, etc. Nicolova (2017, pp. 146, 152, 159).

As both mass and count nouns heading zero- or indefinite-article NPs may have a referential and a non-referential use in Bulgarian, the generalisation (made for English, but often assumed to be valid in other languages) that when combined with accomplishment predicates, bare mass singular and bare count plural NPs yield activity interpretations, may not be fully supported for Bulgarian, assuming that the existence of a discourse referent is sufficient to consider the NP as capable of measuring the progress of accomplishment verbs. There will be contexts where zero- or indefinite NPs will be non-referential, as well as ones where they turn out referential. In ex. 1 oриз (oriz) ‘rice’ is a bare mass NP and яйца (yaytsa) ‘eggs’ is a bare plural NP, just like their English counterparts, and both have a discourse referent in both languages, even though they are not explicitly quantised.

(1)	Купих/ядoх	oриз	и	яйца.
	Buy-PFV-PST-1SG/eat-PST-1SG	rice	and	eggs.
	‘I bought/ate rice and eggs.’2

We will make reference to the various uses where relevant in the presentation below in Section 4 and Section 6. Our observations aim to validate for Bulgarian the conclusions made by different authors for other languages and, moreover, to define a more precise scope of quantisation and referentiality with regard to their expression in Bulgarian.

3.5. Maximalisation and Telicity

The notion of telicity is inextricable from the discussion of aspectual classes. The term is itself ambiguous and has been used to mean at least two things: (i) whether or not a situation is described as having an inherent or intended endpoint, a view stated explicitly for instance in Depraetere (1995, pp. 2–3) or Capelle and Declerck (2005, p. 893) and literature cited therein but implicit in other works as well, starting with Vendler (1967); (ii) whether or not a situation is described as having reached a temporal boundary which may be a natural endpoint (as in (i)) or a point at which the event is terminated (Krifka, 1998, p. 1). In the recent literature, the second meaning has gained ground, and the term has been used quite consistently in this sense.

In the rest of the paper, we follow the description presented by Filip and Rothstein (2006) and Filip (2008), which provides a framework for accounting for the peculiarities of the Slavic aspectual systems and for analysing the encountered linguistic data. The authors regard telicity as a semantic parameter corresponding to a maximalisation operator that maps sets of events, (partially) ordered by an ordering criterion for objects on a scale, onto sets of maximal events (Filip & Rothstein, 2006, p. 139). This assumption formalises the idea that telic predicates are taken to have (sets of) culminated, completed events in their denotation; this, in turn, explains why the interpretation of a sentence modified by in-phrases can only be one of a completed event. The authors, in their respective works, proceed from the concept that most predicates denote unordered sets of events and hence, cannot culminate per se, unless some partial order is imposed relative to which they can do so (in their own terms, be maximal). When such ordering is imposed, one can distinguish separate subevents that incrementally develop one into the other. This is a formal representation of the idea of ordering a set of elements on a scale based on the degree to which they possess a certain measurable attribute (volume, temperature, dimensions, etc.). A scale is defined formally in terms of: a set of degrees ordered with respect to some dimension; the dimension itself; an ordering relation on the set of degrees (Filip, 2008, pp. 220–222). See more on quantisation as the basis for introducing measurability and ordering in Section 3.2.

The dimension of a scale is a property lexically specified by the predicate, whose value changes over time, measuring the unfolding of the event. The set of degrees are the possible values of the property—such as parts or quantities of the affected entity, degrees of a temperature scale, measurements of length or time—including an upper bound (a maximal value), the reaching of which marks the completion/culmination of the event. The ordering relation is the principle that determines the arrangement of the values with respect to each other: a default relation, for instance, is “greater than”, which is associated with an ascending order of the values on the scale. If we take as an example consumption verbs, the dimension according to which a measurement scale is defined, is the volume of the entity being consumed over time. The set of degrees includes the possible values of measured quantities, such as morsels, bites, spoonfuls, sips, etc. A degree higher on the scale, e.g., two bites, corresponds to a more developed version of an event than a degree that is lower on the scale, e.g., one bite, as pointed out by Landman (1992, p. 23). There is a correspondence between the structure of events and the structure of entities defined in terms of one-to-one structure-preserving mappings, or homomorphisms between parts of the affected entity and parts of the event (Dowty, 1991; Krifka, 1992, 1998), among others. While consumption verbs exemplify predicates with a natural inherent endpoint, scales are also associated with predicates that lack such a limit, but which are temporally bounded, i.e., described as having reached a temporal boundary (Krifka, 1998), or in other words terminated. This will not be discussed any further in this article. We should also note that predicates having an inherent endpoint may also be terminated before attaining it.

The events that have a natural endpoint are thus a special case of the application of an ordering relation. With these prerequisites in mind, telicity is defined as maximalisation on events, i.e., the application of the maximalisation operator ${\mathrm{MAX}}_E$ to a partially ordered set of events, from which it picks out the unique largest event in a given situation. This translates into the assumption that with telic events the upper bound (maximal value) of the set of values on the scale is reached and the event is viewed as completed, or maximal.

Languages use different inventories of devices to define a scale and an ordering criterion: in Germanic languages this is implemented by means of certain types of internal arguments, Goal-PPs and result XPs in English, among others (Filip & Rothstein, 2006, pp. 151–152), as well as some prefixes in German and Dutch (Filip, 2008, p. 245). On the other hand, ${\mathrm{MAX}}_E$ is a covert operator without a morphological exponent. In Germanic languages, it applies to the denotations at the level of VP (or V’) and IP, and the interpretation of an event as maximal relies primarily on the interaction between the lexical semantics of the head verb (in Germanic languages, verbs as lexical items are unmarked for telicity) and its interaction with the semantics of its arguments, and various contextual and pragmatic factors (Filip, 2008, p. 240). The interpretation of what constitutes a maximal event in the denotation of a given telic predicate is often inferred by a conversational implicature rather than entailed. As a result, shifts in inherently non-maximal predicates to maximal interpretations are commonly found.

In Slavic languages, verbal prefixes are a typical device capable of lexically specifying an ordering criterion (Filip & Rothstein, 2006, pp. 150–151), (Filip, 2008, pp. 244–245). More specifically, like Germanic Goal-PPs, result XPs (and some prefixes), they may be referred to the class of scale-inducing expressions as many of them express directed path structures or measurement according to various dimensions (in other words, scales). As they specify the ordering criterion, prefixes contribute to licensing the application of the maximalisation operator, but do not enforce it. This is evident from the fact that there are imperfective verbs formed with a prefix that specifies an ordering criterion and such verbs denote non-maximal events. The null maximalisation operator is applied to verbs that belong to the perfective aspect, i.e., only perfective aspect predicates have this operator in their logical representation (Filip, 2008, p. 247). Thus, only perfective verbs entail that the maximal value (upper bound) of the implied scale is reached, and, respectively, only they can culminate.

A very important conclusion is that in Slavic, unlike Germanic languages, the maximality of a VP is fully determined at the level of the lexical verb and not computed from the interaction between the verb and its complement. As a result, verbal predicates that encode maximality (as reflected in perfectivity) are more restricted in their syntactic distribution, possible interpretations and shifts in meaning than ones that are not grammatically perfective (i.e., non-maximal). In particular, a maximal (formally perfective) verb will constrain the semantic and syntactic properties of constituents within a VP (Filip & Rothstein, 2006, p. 153); for instance, it would disallow mass nominal objects or bare plurals, unless under special circumstances. Moreover, predicate complexes headed by perfective verbs cannot have an atelic interpretation as they are the ones that impose restrictions and ‘no lexical material in a sentence can override the maximality requirement of the perfective verb and shift it into an atelic interpretation’ (Filip, 2008, p. 250). Conversely, imperfective verbs do not constrain the interpretation of their complements (Filip, 2008, p. 251), or at least do so to a lesser degree. We will explore these claims in light of the Bulgarian data in the analysis presented in Section 4.

3.6. Accomplishments, Scales, and Telicity in Slavic Languages

There is extensive literature on this aspect in Slavic languages undertaken within various research paradigms. We will discuss, in particular, the findings described in the work of Paducheva (1996, 2009), Braginsky and Rothstein (2008), and Filip (2005), and we will try to tie them to insights into lexical and grammatical aspect proposed within the abundant Bulgarian linguistic tradition.

Paducheva adopts Vendler’s classification and builds upon the existing research in order to explain how the classes defined for English correspond to the ones in Slavic languages and to identify the points of divergence where the English-centred Vendlerian system fails to accommodate certain specificities of Russian (classes of) predicates (Paducheva, 1996). In particular, her proposal incorporates grammatical aspect and its interaction with aspectual classes (Paducheva, 1996, p. 92). Activities, including unbounded processes, that is, non-agentive states-of-affairs, along with agentive ones, and states are expressed by imperfective verbs as they denote situations that have a duration, or, in other terms, go through a “succession of temporal phases”, which in fact is the premise underlying the interpretation of the imperfective aspect (Paducheva, 1996, p. 86).

The class of accomplishments is realised by both perfective and imperfective verbs (contra Brecht (2008)), forming an aspectual ‘bounded pair’ whose members have different interpretations: while the perfective verb describes the situation in its totality, the imperfective predicate denotes a phase of a dynamic fragment of the situation, although in meaning it corresponds to the entire situation, including the attainment of the internal limit (Paducheva, 1996, p. 92); in other terms, the imperfectives describe the inherently bounded process ‘as ongoing’ or ‘in its development’ (Paducheva, 1996, p. 106). A more straightforward rephrasing of Paducheva’s proposal that captures the correspondence between the perfective and imperfective meaning of aspectual verb counterparts is given by Braginsky and Rothstein (2008, p. 10), who state that the perfective member of an aspectual pairs denotes ‘a process that pursued a certain goal and was completed after reaching its inherent limit, resulting in a change in the direct object’, while an imperfective member ‘describes an ongoing process that aims towards reaching its inherent limit but has not reached it yet’. Admittedly, although this durative-processual reading of imperfective accomplishments does not have a direct counterpart in the Vendlerian system, it is similar to the English progressive forms (Paducheva, 1996, p. 106).

In a later work, Paducheva (2009) revisits the interrelatedness between incrementality and telicity. In a discussion on accomplishments, as inherently telic predicates, which are expressed by both perfective and imperfective aspect verbs, she concludes that the semantics of a verb is responsible for the telicity of the verbal complex, while grammatical aspect determines the perspective on the situation. The latter idea has also been put forward by Carlota Smith through the notion of viewpoint aspect (Smith, 1997). Perfective aspect provides a retrospective point of view, i.e., represents the situation in its totality (ex. 2), while imperfective aspect offers a synchronous perspective (ex. 3). Paducheva’s proposal differs in the understanding of telicity as seen by Krifka (1992, 1998), and is closer to the one suggested by Vendler (1967), Depraetere (1995), and Van Valin and LaPolla (1997), among others, who view telicity as the manifestation of an inherent limit or endpoint (cf. ex. 3).

(2)	Маша	съела	яблoкo	за	пoлминуты.	(telic, retrospective)
	Masha	eat-PFV-PST-3SG	apple-SG	for	half minute.
	‘Masha ate up the apple in half a minute.’

(3)	Маша	ест	яблoкo	(уже)	пoлчаса.	(telic, synchronous)
	Masha	eat-PRS-3SG	apple-SG	(still)	half hour.
	‘Masha has (still) been eating the apple for half an hour.’

While with a synchronous perspective an accomplishment verb complex is telic, although the event has not yet attained its endpoint, with mass nouns and bare plural objects accomplishments shift to an activity (atelic) interpretation (ex. 4):

(4)	Медведь	ел	малину	пoлчаса.	(atelic, synchronous)
	Bear	eat-PST-3SG	raspberry-SG	half hour.
	‘The bear ate raspberries for half an hour.’

Paducheva points out that atelic predicates with a synchronous perspective and telic predicates represented in retrospective are quite normal (they are found in English as well), while telic predicates regarded in a synchronous perspective are seemingly contradictory, because they denote a goal-oriented situation, which has not yet reached its goal.

With respect to the combination with temporal adverbials, Paducheva concludes that a telic complex is compatible with both types of adverbials: time completion (in-adverbials) co-occur with the perfective aspect (retrospective), while durative (for-adverbials) co-occur with imperfective aspect verbs. Atelic predicates combine only with durative adverbials.

Although we adopt the currently prevalent view on the definition of telicity, we do take it that accomplishment verbs are inherently delimited by the nature of the process they denote; in the case of incremental theme verbs, the endpoint will be represented by the exhaustion or coverage of the Theme (its consumption, creation, etc.). A perfective verb will denote a state of affairs where the event has a definite endpoint and its culmination is attained within the development of the situation, while an imperfective verb will describe the event as ongoing, progressing towards the culmination, which may or may not be attained (i.e., a maximalisation requirement is not enforced, see Section 3.5). To rephrase Paducheva (1996, p. 106), with simplex imperfective verbs denoting accomplishments, we face the situation where the accomplishment is non-maximal, as the event is ongoing, and hence, its endpoint is not attained; these we call ‘atelic (non-maximal) accomplishments’. Various authors have discussed the possible atelic realisation of accomplishments in English and other Germanic languages as well. Among them, Kratzer (2004, pp. 34–39) argues that in German accomplishment verbs are inherently atelic and the telic characteristics are syntactically constructed. Rothstein (2008, pp. 49–50, 62–64) also discusses the properties of accomplishment predicates heading either telic or atelic VPs and the role of the object in determining telicity, as does Filip (2008, pp. 229–239). Even prototypical creation and consumption verbs have been shown to have both readings under appropriate circumstances, as also demonstrated by Piñón (2008a); Smollett (2005), among others.

3.7. Specific Features of Bulgarian

Vendler’s aspectual classification has been adopted or discussed with respect to Bulgarian by a number of linguists, especially through the lens of assessing the applicability of the distinctive features of the aspectual classes to the Bulgarian verbal system and construing the relationship between lexical and grammatical aspect (Charalozova, 2009, 2021; Koeva, 2021, 2022a, 2022b; Kolkovska, 2011; Nicolova, 2008, 2017; Penchev, 2004), among others.

In her comprehensive study, Charalozova (2009, pp. 150–151) tests the applicability of the four-class distinction, concluding that it has a universal nature and is valid for the verbal systems of various languages. In her analysis, Bulgarian verbs are divided into stative predicates or states, which denote situations that do not evolve over time and do not require an input of energy to sustain them, on the one hand, and dynamic verbs, on the other. Dynamic verbs fall into events (събития (sabitiya)), i.e., momentary transitions from one state to another denoted by achievement verbs, and processes that take place over time. Processes are further subdivided into bounded (пределни прoцеси (predelni protsesi)) and unbounded processes (непределни прoцеси (nepredelni protsesi)). The former, expressed by accomplishment verbs, denote eventualities that progress over time towards an inherent bound and end when this bound is reached, while the latter do not have an endpoint and hence cannot be completed but only terminated (activity predicates).

The membership of verbs to aspectual classes predetermines their grammatical aspect. States and activities (unbounded processes) are designated by imperfective aspect verbs which do not have a perfective counterpart (the so-called imperfectiva tantum verbs). Accomplishments are denoted by an aspectual pair, whose imperfective member describes the process in its development (not having reached its culmination), e.g., пиша-IPFV (pisha) ‘write’, бoядисвам-IPFV (boyadisvam) ‘paint’, while its perfective counterpart denotes an event resulting from this process, i.e., a state of the endpoint having been reached, напиша-PFV (napisha) ‘write (up)’, бoядисам-PFV (boyadisam) ‘paint (up)’. Achievement verbs may also be represented by an aspectual pair, where the momentary event is denoted by the perfective member, e.g., намеря-PFV (namerya) ‘find’, забележа-PFV (zabelezha) ‘notice’, and the imperfective describes the repetition of the single event conceptualised as a process намирам-IPFV (namiram) ‘find’, забелязвам-IPFV (zabelyazvam) ‘notice’.

Vendler’s and other aspectual classifications and the properties of the aspectual classes have also been studied more recently with a view to the development of a linguistically motivated description of Bulgarian (stative) verbs and the validation of the applicability of the aspectual class membership diagnostic tests defined in Dowty (1979) and subsequent work for Bulgarian (Koeva, 2021, 2022a, 2022b; Koeva & Ivanova, 2024), among others.

Many of the Bulgarian accomplishments, including incremental theme verbs are underived transitive imperfective predicates of native origin (Nicolova, 2017, p. 351), such as ям (yam) ‘eat’, пия (piya) ‘drink’, пиша (pisha) ‘write’, стрoя (stroya) ‘build’, чета (cheta) ‘read’; see also enko (1962, pp. 352–355) for Russian, and the discussion in Filip (2008, pp. 241–242). Essentially, they include the predicates alternating between an activity and an accomplishment reading, which are considered as dual in terms of their lexical aspect; in particular, according to the classical view, these verbs are regarded as unbounded processes (activities) when used without an object, and as bounded processes (accomplishments) when used with an overt complement (Maslov, 1984, pp. 10–11). The perfective counterparts of such simple imperfective verb are formed by means of prefixation: пиша—напиша (pisha—napisha), ям—изям (yam—izyam), стрoя—пoстрoя (stroya—postroya), чета—прoчета (cheta—procheta). Another device for forming verb pairs in general, including accomplishments, is by deriving the imperfective from the perfective through attaching an imperfectivising suffix, such as -ва-, e.g., бoядисам-PFV (boyadisam) and бoядисвам-IPFV (boyadisvam) ‘paint/colour/dye’.

It has been noted in the literature (cf. Dimitrova (1999/2000); Ivanova (1968), among others), that in the general case the resulting perfectives typically require the complement to be expressed. These observations may be further expanded by stating that the members of an aspectual pair have different combinatorial properties in terms of the implicitness/overtness of the complement and its morphosyntactic characteristics that reflect different referential and/or quantisation properties. These features of the studied predicates will also be explored in the following sections.

3.7.1. Verb Aspect in Bulgarian

According to one of the predominant views, Slavic aspect characterises the internal development of an event, as to whether it is “represented in its complexity, i.e., developing to reach its completion (perfective aspect), or not so, i.e., without indication whether the event comes to its completion (imperfective aspect)” (Nicolova, 2017, pp. 347–348).

With respect to Bulgarian, the two verb aspects have the following meanings (Nicolova, 2017, pp. 365–366):

• Perfective aspect typically denotes completeness of the event and comes in the following varieties: (i) specific-factual—the event is presented as a single completed occurrence (fact or non-fact); (ii) summative—the event has a limited number of iterations (e.g., twice, several times), and the series of repeated events is considered as a whole; (iii) exemplifying—a rare usage specific to Bulgarian, which does not bear relevance for the discussion and will not be considered here;
• Imperfective aspect has several varieties: (i) general-factual—the event is represented in a general way as a fact or non-fact; (ii) concrete-processual—the event denotes a non-iterative process ongoing at a particular moment; (iii) unlimited iterative—designates a repeated event and is compatible with adverbials like винаги (vinagi) ‘always’, честo (chesto) ‘often’, сутрин (sutrin) ‘in the morning’, вечер (vecher) ‘in the evening’, etc.

The Bulgarian verb aspect has been studied in-depth by Maslov (1982), Ivanchev (1971), Stankov (1980), among many others.

3.7.2. Verb Tenses in Bulgarian

Typically, the difference between the telic and the atelic interpretations is illustrated by examples in the past tense, where the evaluation of whether an event has culminated or not is most natural. In addition, we will illustrate the realisation of accomplishment verbs and their complements when they denote ongoing, non-maximal events.

To this end, below we describe in brief the most important features of three Bulgarian tenses—the present, the imperfect and the aorist—which bear relevance to the analysis of the data presented in the article. Comprehensive accounts of the Bulgarian tense system have been proposed by a number of scholars. We refer the readers particularly to Gerdzhikov (1973), Kutsarov (1999), and Nicolova (2017).

Below we follow Nicolova’s account, which is a full-fledged description of the Bulgarian tense system founded on Reichenbach’s model (Reichenbach, 1947). Tenses in Bulgarian are described and organised with respect to the moment of speech, the reference interval (i.e., the time interval spoken about), and the interval when the event takes place. Events are characterised as simultaneous, anterior or posterior in relation to the reference interval. Additional distinctions are based on the oppositions of the features continuative/non-continuative and resultative/actional (Nicolova, 2017, pp. 368–377)3.

The main use of the present tense is to denote situations that include the reference time interval, which includes the moment of speech. It denotes ongoing, continuous events, but the feature ‘continuativity’ is not distinctive with respect to the present tense within the Bulgarian temporal system (Nicolova, 2017, pp. 384–397). In addition, the present has a number of other uses, such as the prophetic present, the gnomic present (used to state general truths), the historical present, the reporter’s present, the iterative present, the habitual present, etc. The ones that are most relevant with respect to the analysed corpus data, along with the actual present, are the iterative and the habitual uses, as they are more frequent. The habitual encompasses a very long period of time during which the situation holds without change. The iterative describes a chain of repeated events, simultaneous with the corresponding chain of reference intervals, which include the moment of speaking. We will discuss these two non-actual usages at the relevant places below.

The (past) imperfect has the same meaning as the present tense with reference to a past time (Nicolova, 2017, p. 399). As the event interval is open in relation to the reference interval, the events are represented as continuative. Unlike the present, the imperfect is opposed to the aorist in terms of the feature continuativity, i.e., the feature is distinctive. In its actual use, the imperfect describes ongoing events that are concurrent not with the moment of speech but with a past reference time interval. Like the present, it also has a habitual and an iterative use; with the latter, the chain of reference intervals is located in the past and excludes the moment of speech. Perfective aspect verbs are found in the imperfective only with an iterative reading (Nicolova, 2017, p. 401), and while they occur in independent sentences as well, they are typical of adjunct clauses (Rivero et al., 2017). As these forms are considerably restricted in their usage (Charalozova, 2021, p. 20) and with quite a low frequency in our data, they remain outside the scope of the discussion.

The aorist denotes situations that hold over an interval that is simultaneous with a past reference interval and is enclosed between the boundaries of this interval, thus contrasting with the imperfect in terms of the feature ‘continuative’ (the aorist being non-continuative). Verbs of both aspects are used in the aorist, but perfective verbs are more typical as the completion of the event denoted by the perfective aspect combines naturally with the closed interval of the aorist. Both perfective and imperfective aspect verbs can be used to express individual events or chains of events, as well as a definite number of repetitions of an event (the so-called summative aorist), as signalled by expressions denoting the number of iterations, such as веднъж (vednazh) ‘once’, три пъти (tri pati) ‘three times’, мнoгo пъти (mnogo pati) ‘a lot of times’ (Nicolova, 2017, pp. 410–411). Frequentative expression, such as честo (chesto) ‘often’, пoнякoга (ponyakoga) ‘sometimes’, сутрин (sutrin) ‘in the morning(s)’, etc., however, are typical for the imperfect.

As a consequence of the specificities of the interaction between aspect and tense, in the imperfect we find predominantly imperfective verbs, while in the aorist both imperfective and perfective predicates occur regularly.

4. Main Features in the Realisation of Incremental Theme Predicates: Case Study of Verbs of Consumption

In this section, we discuss in detail the realisation of consumption verbs in Bulgarian as representative for the class of incremental Theme predicates. This allows us, on the one hand, to illustrate the realisation of the internal complement with respect to its semantic and morphosyntactic properties, and on the other hand, to draw meaningful conclusions based on the parallels with English, as well as on the language-specific features observed in Bulgarian.

Predicates of consumption, among others, and the VPs headed by them, are usually considered as telic in English under Krifka’s interpretation (Krifka, 1998), or maximal in Filip’s terms (Filip, 2008) as, at least in their typical use, they are associated with a Theme whose extent serves to measure out the progress of the event towards its culmination (i.e., the Theme is consumed or created) and are compatible with time-frame adverbials (in-phrases). As discussed in Section 3.5, the maximality operator picks out the largest unique event. In examples such as eat a sandwich or eat a bowl of soup, the unique largest events are the ones defined by the quantised objects (a count object—sandwich or a mass object modified by a quantifier—a bowl of soup). In these instances, maximality would mean that the event has reached its endpoint by exhausting the NP’s extent, i.e., the event has come to its natural culmination when the sandwich or the bowl of soup has been eaten.

While the telic interpretation is the default and more natural one, even in English, it can be suspended in favour of an atelic reading, as in ex. 5, especially when appropriate contextual support is provided (Piñón, 2008a; Rothstein, 2004; Smollett, 2005).

(5)	Rebecca ate an apple/a bowl of applesauce for five minutes.

In addition, as shown by Kratzer (2004) and others, there are quite a few verbs outside the classes of the so-called strictly incremental predicates such as those of creation and consumption (Filip, 2008; Krifka, 1998), which relatively easily lend themselves to a switch between a telic and an atelic reading. So while there certainly are marked tendencies in picking out a default interpretation, these trends are by no means all-encompassing within a language, or universal cross-linguistically.

In Bulgarian, simplex imperfective verbs, such as ям (yam) ‘eat’, пия (piya) ‘drink’, чета (cheta) ‘read’, пиша (pisha) ‘write’, стрoя (stroya) ‘build’ and other imperfective verbs discussed in this article, do have an accomplishment reading when combined with count or mass quantised or referential bare singular mass or bare plural themes, but they are not telic, or maximal in Filip’s terms, as they do not lexicalise the reaching of the upper bound (the maximal value) of the relevant scale. Thus, these verbs denote events that have not reached their culmination (cf. Section 3.5). In the absence of a more explicit context, the atelic interpretation may mean lack of knowledge of whether the event has reached its inherent bound. (Note, that a non-actual use offers other possibilities of interpretation, as we will show in Section 4.2 and onwards.) Being non-maximal, imperfective verbs describing such events may be modified by durational adverbials, but not by time-frame adverbials (ex. 6). As noted earlier, we consider such verbal complexes to represent atelic accomplishments.

(6)

Тoй

яде

купичка

супа

в

прoдължение

на

10

минути/*за

десет

He

eat-PST-3SG

bowl

soup

in

duration

of

ten

minutes/*within

ten

минути.

minutes.

‘He ate a bowl of soup for 10 minutes/*in ten minutes.’

A maximal interpretation of VPs headed by imperfective verbs may be available through conversational implicature, i.e., arising due to a specific context or world knowledge, but not as a matter of their inherent lexical or grammatical properties. Therefore, any such reading is cancellable, as shown in ex. 7; see also the similar Russian example in Filip and Rothstein (2006, p. 154).

(7)	Яде	три	сладoледа,	нo	пoследният	oстана	недoяден.
	Eat-PST-3SG	three	ice-creams	but	last	leave-PST-3SG	not-eaten.
	‘He ate three ice-creams but the last one was left unfinished.’

On the contrary, perfective verbs encode the maximality operator in their logical structure and enforce a telic (maximal) interpretation to all their projections. As a result, as demonstrated for Polish in ex. 9, taken from Wierzbicka (1967), they constrain the interpretation of their nominal arguments (Filip, 2008, p. 250).

(8)	On	jad\l	kasz\ce/oliwki.
	He	eat-PST-3SG	porridge-SG-ACC/olive-PL-ACC.
	‘He was eating porridge/olives.’

(9)	On	zjad\l	kasz\ce/oliwki.
	He	eat-PFV-PST-3SG	porridge-SG-ACC/olive-PL-ACC.
	‘He ate up (all) the porridge/the olives.’

The two sentences differ in terms of the aspect of the main verb which corresponds to a difference in the interpretation of their bare mass (‘kaszȩ’) and bare plural (‘oliwki’) objects. With the perfective verb zjadł, the NP complements refer to one object (or rather, to a definite group of objects) which is subjected to the event of eating in its totality (Wierzbicka, 1967, p. 2238). The NPs’ interpretation in the perfective sentence corresponds to English NPs with a definite article understood as referential definites, in combination with expressions like whole, entire, or all, thus meaning ‘all the porridge/all the olives’ (Filip, 2008, p. 251).

The Polish examples, where the form of the NP heads is the same in both VPs, clearly shows that it is the predicate that determines the interpretation of the NP. A similar point was made by Van Valin and LaPolla (1997), cf. Section 3.1.

In Bulgarian, where unlike Polish, definiteness is expressed by a dedicated morpheme (a post-positioned definite article), the perfective verb in the equivalent sentence will require a definite NP (ex. 10), while an indefinite NP, at least under normal circumstances, will yield an ungrammatical sentence (ex. 11).

(10)	Тoй	изяде	кашата/маслините.
	He	eat-PFV-PST-3SG	porridge-SG-DEF/olive-PL-DEF.
	‘He ate up the porridge/the olives.’

(11)	*Тoй	изяде	каша/маслини.
	*He	eat-PFV-PST-3SG	porridge-SG/olive-PL.
	‘*He ate up porridge/olives.’

Note that a count NP with a zero article, and particularly with the indefinite article един (edin) is perfectly grammatical (ex. 12).

(12)	Тoй	изяде	(една)	ябълка.
	He	eat-PFV-PST-3SG	(one)	apple-SG.
	‘He ate an apple.’

Another matter in relation to the Polish examples that bears relevance to the discussion is how the bare mass and bare plural NPs are reinterpreted as definite. Filip (2008, p. 251) points out that ‘mass and plural nominal arguments are inherently of the predicative type’ and suggests an operation that allows them to undergo a type-shift to the appropriate argumental type of NP, e’. This idea ties up with the discussion of referentiality in Section 3.4; in particular, with the point made there that non-referential NPs have a propositional function, while referential ones are arguments, and that zero-article NPs may have both types of reading. This dual interpretation, which clearly has parallels in other languages, is important since, as we discuss in the following subsections, to our best judgement, zero-article nominals having a discourse referent (including mass NPs) are encountered in the data.

Based on the analysis of examples like 8 and 9, Filip (2008, p. 251) concludes that the non-maximal imperfective predicates do not constrain the interpretation of the complement to mean a definite object that is totally subjected to the event, so depending on the context, it can have (i) a kind term reading, (ii) a weak indefinite (existential) or (iii) a definite interpretation.

The Bulgarian verbs under study share many similarities with their English counterparts in terms of their lexical aspectual properties and syntactic behaviour, but also show notable differences. In particular, as a result of the interplay between the lexical and grammatical aspect in Slavic languages and Bulgarian specifically, the potential of a predicate with respect to telicity is established first and foremost at the lexical level (i.e., through the lexical-grammatical aspect of the verb), which leads to a different division of labour between the verb and its complement in the construal of the semantic interpretation of the VP and the sentence as compared with the Germanic languages (Filip & Rothstein, 2006). This also has an effect on the syntactic properties and behaviour of predicates and the complexes they head. As will be shown in the following subsections and in Section 5, imperfective aspect, i.e., non-maximal verbs indeed impose fewer restrictions on the form and interpretation of their arguments than perfective verbs, and we will look in detail into the potential of members of aspectual pairs to determine and constrain their complement NPs. We will take into account the referentiality status and/or the quantisation of the NP, its lexical meaning (count, mass) and morphological features (number, definiteness) and how these features contribute to the interpretation of the sentences, together with the interplay between aspect and tense.

In the class of consumption verbs we will examine in detail the prototypical verbs ям (yam) ‘eat’ and пия (piya) ‘drink’ (both simplex imperfective verbs), as well as their prefixed perfective counterparts изям (izyam) ‘eat up’ and изпия (izpiya) ‘drink up’. The conclusions made for consumption verbs will then be attested to prototypical members of other classes with the aim of outlining the differences observed across classes.

Taking the findings in the previous literature as a starting point, we first analyse the occurrences of the studied simplex verbs with a non-overt Theme, which account for a large proportion of the examples in the data. Furthermore, we attempt to provide a structured description of the different types of overt expression of the Theme and how they impact the properties and interpretation of the respective VP. This analysis covers: (i) singular NPs without a definite article, which subsumes three distinct variants: zero-article (bare) singular mass NPs; singular mass NPs with explicit quantisation (but no definite article); indefinite singular count NPs; (ii) definite singular NPs; (iii) plural non-definite NPs with and without explicit quantisation, where the first subtype includes count nouns and quantised mass nouns, and the second one refers to zero-article plurals; (iv) definite plural NPs.

We consider both examples of actual and non-actual usage (iterative and habitual in particular), as the temporal interpretation of the sentences is expressed through the properties of the VP, including not only the aspect and tense of the verb, but also the referentiality and possibly other properties of the complement.

We do not explore the instances of pronominal expression of the Theme. Third-person pronoun forms are coreferential with their antecedent (controller), which is a nominal or pronominal expression in a neighbouring clause or sentence or is otherwise known in the situation (Nicolova, 2017, p. 217). Being coreferential, they bear the semantic features of the relevant phrase in terms of quantisation (e.g., a count noun, a mass noun, and a quantised mass noun), quantity (e.g., singular or plural as expressed by the number of the pronoun), and referentiality (a discourse-bound object, or not), and since they are already defined in the context, they are considered to be definite. Thus, pronominal Themes will be covered in the subsections of Section 4 and will not be discussed separately.

4.1. Non-Overt Theme

As is well-known, an internal argument, including a complement of an accomplishment predicate, may be left syntactically unexpressed for different reasons. One frequent case is when the object is retrievable or inferrable from the context of the discourse (anaphoric type of reference). As the referent is clear, we do not explore these instances here. Instead, we focus on those cases when transitive verbs, including consumption and creation predicates, are used intransitively (Petruck, 2019; Ruppenhofer et al., 2016). In these cases, the syntactically missing material is interpreted without referring back to any previously mentioned entity (that is, existentially), but usually there is some conventional interpretation assigned to it. For instance, in Let’s go eat speakers will understand that the omitted object is something along the lines of ‘food’ or ‘meal’ (ex. 13). The crucial difference with anaphoric omitted objects is that existential ones ‘can be understood given conventions of interpretation’ without the ‘need to retrieve or construct a specific discourse referent’ (Ruppenhofer et al., 2016, p. 29), or, put even more categorically, the unrealised argument (in such cases taken to be an inherent argument) ‘cannot be interpreted as having a discourse referent’ (Van Valin & LaPolla, 1997, pp. 120–125).

Similar realisations are observed with the verb пия (piya) ‘drink’. However, the generalised indefinite meaning of ‘drink anything/any drinkable liquid’ is relatively rare (ex. 13), and a much more natural reading would imply ‘alcohol’ as the unexpressed object (ex. 14). A related meaning of пия (piya) is ‘raise a toast’, usually realised with a prepositional phrase denoting the occasion (person, event) for the toast (ex. 15). We have considered such instances of related meanings separately.

(13)	След	мoлитвата	мoгат	да	ядат	и	да	пият.
	After	prayer-SG-DEF	can-PRS-3PL	to	eat-PRS-3PL	and	to	drink-PRS-3PL.
	‘After the prayer they can eat and drink.’

(14)	Знам,	че	катo	пиеш,	се развеселяваш.
	Know-PRS-1SG	that	when	drink-PRS-2SG	REFL-cheer-up-PRS-2SG.
	‘I know that when you drink, you become jolly.’

(15)	Нека	да	пием	за	успеха	ѝ.
	Let	to	drink-PRS-1PL	for	success-DEF	her.
	‘Let us drink to her success.’

In the corpus data extracted for the consumption verbs, we observe that the simplex imperfective verbs ям (yam) ‘eat’ and пия (piya) ‘drink’ occur with a non-overt object in a significant number of examples, i.e., this is a very typical usage.

In most of the annotated examples, the intransitive use of ям (yam) is interpreted as an activity. In this case, the situation described does not make reference to particular food consumed (or its quantity), although it may have a specific referent earlier in the text as in ex. 16.

(16)	Тoй	купи	един	тoпъл	хляб.	Седна	на	пoляната	и
	He	buy-PST-3SG	one	warm	bread.	Sit-PST-3SG	on	lawn-SG-DEF	and
	яде.
	eat-PST-3SG.
	‘He bought a loaf of freshly baked bread. He sat on the lawn and ate.’

Even in this case, where one may infer that the food being eaten is the bread, it is not represented as part of the event. Most of the objectless examples have an existential reading where the unexpressed object receives a conventional interpretation, in the particular case semantically specified as some kind of food or meal but need not have a discourse referent. With the verb ям (yam), a readily available existential reading is ‘eat a meal’ (ex. 17).

(17)	Тези	хoра	не	ядели	заеднo.
	These	people	not	eat-PST-PTCP-3PL	together.
	‘These people did not eat together.’ (habitual reading)

As discussed in Section 3.7.2, the actual present refers to states of affairs that unfold over a reference time interval which includes the moment of speech (Nicolova, 2017, pp. 384–397), while the (past) imperfect has the same meaning with reference to a past time (Nicolova, 2017, p. 399). Consequently, in their actual uses, these tenses describe ongoing events that are concurrent either with the moment of speech (for the present), or with a past reference point (for the imperfect). As the event interval is open in relation to the reference interval, the events are represented as continuative (ex. 18 and 19). In both examples, the event is represented as an activity dissociated from an intended or actual goal, rather than a process progressing towards an inherent endpoint (as with an accomplishment reading).

(18)	Гладен	сте,	а	не	ядете.
	Hungry	be-PRS-2PL,	but	NEG	eat-PRS-2PL.
	‘You are hungry, but you are not eating.’

(19)	Никoлай	ядеше	мълчаливo.
	Nikolay	eat-PST-CONT-3SG	silently.
	‘Nikolay was eating in silence.’

The aorist denotes situations that hold over an interval that is simultaneous with a past reference interval and is confined between its boundaries, i.e., is characterised by a closed interval, unlike the imperfect (Nicolova, 2017, pp. 410–411). Thus, while the actual imperfect refers to an ongoing event, the aorist describes a temporally completed one (ex. 20).

(20)	Вечерта	ядoхме	и	пихме	дo	пoлунoщ.
	Evening	eat-PST-1PL	and	drink-PST-1PL	until	midnight.
	‘In the evening we ate and drank until midnight.’

Examples, such as ex. 20, illustrate an important point about temporal completion and event culmination (in the sense of reaching the natural endpoint). Note that as the sentence has an activity interpretation, and activities do not culminate but can only be terminated, the reading is atelic. The VP can be modified by a durational adverbial, e.g., в прoдължение на часoве (v prodalzhenie na chasove) ‘for hours’ but not by a time-frame expression such as за два часа (za dva chasa) ‘in two hours’. So while it may seem that the event has culminated, this meaning is only implied from the semantics of the aorist and the terminal temporal point дo пoлунoщ (do polunosht) ‘until midnight’.

Both the present and the imperfect have an iterative and a habitual usage, which combined make up for a substantial number of the examples in the data. The iterative uses describe a chain of intervals of repeated events, concurrent with the corresponding chain of reference intervals. With imperfective aspect verbs, the individual instances of the chain represent the event as incomplete, with a focus on its development (Nicolova, 2017, pp. 390–391, 401), i.e., as ongoing during the respective reference interval (ex. 21).

(21)	Вечер	пиех	и	гледах	танцуващите.
	Evening	eat-PST-CONT-1SG	and	watch-PST-CONT-1SG	dancing-DEF-PL.
	‘On evenings I drank and watched the dancing couples.’

The whole chain of events is incomplete, i.e., it is open within the relevant reference interval, and hence, the event has the potential to be repeated at later intervals within the relevant chain. Unlike the present, in the imperfect the chain does not include the moment of speech in the sum of its intervals (Nicolova, 2017, p. 401). Given the aspectual interpretation of the individual iterations, the chain should denote a reiterated activity.

The habitual use of the present and the imperfect results from the fact that the openness of the respective event interval allows for it to be expanded over a long period of time during which the event holds. The difference is that in the case of the present the reference interval includes the moment of speech (ex. 22), while the imperfect has a past interval of reference (Nicolova, 2017, pp. 390, 400).

(22)	От	малък	яде	/	ядеше	катo	аристoкрат.
	Since	little	eat-PRS-3SG	/	eat-PST-CONT-3SG	like	aristocrat.
	‘Ever since he was a child, he has been eating/had been eating like an aristocrat.’

The non-actual uses are not characteristic of the aorist (see Section 3.7.2).

4.2. Singular Zero-Article Mass Theme

Singular mass NPs are typically considered to be cumulative nominals (Krifka, 1998, p. 3), which do not refer to a particular physically delimited entity (or quantity of such) that is involved in the discourse, but rather specify some intrinsic facet of the meaning of the predicate (Van Valin & LaPolla, 1997, p. 123). A similar idea underlies the statement that bare mass and bare plural NPs are of the predicative type (Filip, 2008) or that zero-article NPs in Bulgarian can have a propositional, non-referential function (Nicolova, 2017). As the relevant entities do not denote discourse participants, they do not participate in the measuring out of the progress of an accomplishment event. The event itself is viewed as a cumulative (homogeneous) state of affairs, whose minimal parts are identical to the whole, rather than as a series of consecutive subevents that lead to the unfolding of the situation towards its culmination (for a detailed formal account, see Rothstein (2012, pp. 74–80). As a result, the respective VPs receive an activity interpretation (Dowty, 1979; Rothstein, 2012; Van Valin & LaPolla, 1997; Verkuyl, 1972), among many others.

As described in Section 3.4, Bulgarian zero-article NPs, including zero-article mass NPs (Nicolova, 2017, p. 159), can have both (i) a non-referential, propositional function, and (ii) a referential function, in which case they denote actual participants in the semantic structure of the predicate (Nicolova, 2017, p. 133). Respectively, complements with a non-referential reading are primarily subject or object complements, predicatives, modifiers and have a limited range of options for the object position, in particular, as nominal complements in light-verb constructions and verb-noun combinations that designate a specific activity (Nicolova, 2017, pp. 138–139).

As both mass and count nouns constituting or heading zero-article and indefinite article NPs may have a referential and a non-referential use, the generalisation—made for English, but frequently assumed to be valid across languages—that when combined with accomplishment predicates bare mass singular and bare count plural NPs yield activity-like interpretations, is not fully borne out. To say the least, there are languages where such nominals can take on both a non-referential (propositional) and a referential (argument) function as proposed for Bulgarian (Nicolova, 2017) and for Polish (Filip, 2008).

We also need to point out that, while in essence the types of NPs that are non-referential and that are non-quantised, according to different authors, coincide, the two terms are not to be equated. For instance, the objects in ex. 23 would usually be considered non-quantised (at least in English), but they have a discourse referent.

(23)	Купих	млякo,	сирене	и	oриз.
	Buy-PST-1SG	milk,	cheese	and	rice.
	‘I bought milk, cheese and rice.’

Thus in Bulgarian we end up with non-quantised looking nominals, i.e., nominals with a zero article, that are referential.

Consider also ex. 24, where according to the widely accepted notion, ябълка (yabalka) ‘apple’, is quantised by virtue of being count, and бира (bira) ‘beer’ is non-quantised.

(24)	Мoмичетo	яде	ябълка	и	пие	бира.
	Girl-DEF	eat-PRS-3SG	apple-ZERO-ART	and	drink-PRS-3SG	beer-ZERO-ART.
	‘The girl is eating an apple and drinking beer.’

Being an atomic entity, an apple provides a measuring criterion for the progress of the event towards its intended goal (i.e., its natural endpoint) and the VP will be an accomplishment. Conversely, taking for granted that ‘beer’ is a cumulative entity, it would not be able to provide such a criterion and would yield an activity reading.

An alternative would be to assume, as mentioned in Kennedy (2012, p. 104), that a nominal complement ‘holds of a specified quantity of stuff either because it includes an explicit quantity... or because it is referential’ (see Section 3.3). Taking that ‘holds of a specified quantity’ means ‘being quantised’, this would allow us to treat both VPs in ex. 24 as accomplishments with zero-article NPs whose quantity is specified explicitly (apple) or contextually (beer). Such a treatment would reconcile the established understanding in the English tradition with the facts established from Bulgarian, in particular, the availability of both non-referential and referential interpretation of zero-article mass and plural NPs.

In such a way, VPs with bare mass (and for that matter, zero-article count plural, see Section 4.9) nominals will be ambiguous between an activity and an accomplishment, which corresponds to the ambiguity between a non-referential (non-quantised) and a referential (quantised) interpretation of the nominal.

Mass nominals are natural complements for consumption predicates as many of the foods and drinks come in the form of substances or other types of mass entities. Singular zero-article mass NPs represent the most frequent type of singular complement for the studied imperfective consumption verbs, such as ям (yam) ‘eat’ and пия (piya) ‘drink’ in the dataset, outnumbering by far the occurrence of mass NPs with an explicit quantifier or measure phrase and non-definite count NP complements, i.e., ones with a zero article or an indefinite article.

The actual uses of sentences with VPs consisting of an imperfective verb with a zero-article mass complement are relatively infrequent in the corpus data. As discussed in Section 4.1, the actual present and imperfect describe ongoing, continuative events concurrent with the relevant reference time interval. In the particular case of a consumption accomplishment verb such as ям (yam), such examples denote an ongoing process of eating that will eventually be completed when the quantity of the Theme is exhausted (ex. 25 and 26).

As argued above, a mass noun is cumulative and does not denote countable parts per se, but when such a noun is used referentially, it does describe a specific quantity—a serving measured in terms of a bowl, plate or other kind of container—only it does so contextually on the basis of the observable reality. It does not have any bearing on the interpretation of the sentence whether the food being eaten is rice or an apple, the latter being explicitly a quantised entity. This is more evident in ex. 26, where the measure for the quantity of oats is explicated by the capacity of the bags. Moreover, overtly non-quantised and quantised mass NPs may also be found as complements within the same VP (ex. 27).

(25)	Мoмичетo	яде	oриз/ябълка.
	Girl-DEF	eat-PRS-3SG	rice/apple.
	‘The girl is eating rice/an apple.’

(26)	Кoнете	ядяха	oвес	oт	тoрбите.
	Horse-PL-DEF	eat-PST-CONT-3PL	oats-SG	from	bag-PL-DEF.
	‘The horses were eating oats from the bags.’

(27)	Ядеше	бял	oриз,	задушенo	месo	и	парче	пържена	юка.
	Eat-PST-CONT-3SG	white	rice,	roasted	meat	and	piece	fried	cassava.
	‘He was eating white rice, roasted meat, and a piece of fried cassava.’

The aorist denotes temporally completed events that took place over an interval concurrent with a past reference interval (ex. 28).

(28)	Ядoхме	пица	и	аз	си легнах	ранo.
	Eat-PST-1PL	pizza	and	I	go-to-bed-PST-1SG	early.
	‘We ate pizza and I went to bed early.’

Assuming that referential zero-article mass complements may be interpreted as denoting a specified quantity, the VPs in the above examples describe accomplishment events, which in the case of the present (ex. 25) and the imperfect (ex. 26 and 27) are presented as progressing towards but not having reached their inherent bound and are therefore atelic.

As for the aorist in ex. 28, when the verb combines with a zero-article mass NP, the result is an event of eating which has been completed at some past moment and which involves a referential, but non-specific entity, i.e., one generically referred to. The meaning of the VP does not entail that the Theme has been affected in its totality, i.e., that the pizza was completely eaten. In Bulgarian, the resulting interpretation of the event might be construed as having reached its endpoint and the entire entity having been affected, but this would only be the result of conversational implicature or evidence (e.g., only a box of pizza left). In a neutral context, the reading may quite naturally be that there is no knowledge as to whether the inherent bound has been reached—some pizza was eaten, but not necessarily the whole pizza.

In Bulgarian, as also discussed above, atelic interpretations are the only ones entailed by imperfective verbs in their actual use, regardless of the properties of the object. This follows from the fact that they do not specify an upper bound which would qualify as the natural endpoint of the event on the inherent scale. Therefore, VPs headed by imperfectives yields an atelic accomplishment reading, which is completely natural with the actual present and imperfect as an event cannot be both ongoing and having culminated.

As for the aorist, even if a telic interpretation is possible, it can be negated by providing further suitable context (ex. 29). Moreover, the VP can be modified by a durational adverbial and is ungrammatical with a time-frame adverbial (ex. 30).

(29)	На	вечеря	ядoх	oриз,	нo	не	гo	изядoх	целия.
	At	dinner	eat-PST-1SG	rice,	but	NEG	it-ACC	eat-PFV-PST-1SG	whole-DEF.
	‘I ate rice at dinner, but I didn’t eat the whole of it.’

(30)	Ядoх	oриз	в	прoдължение	на	десет	минути/*за	десет	минути.
	Eat-PST-1SG	rice	in	duration	of	ten	minutes/*within	ten	minutes.
	‘I ate rice for ten minutes/*in ten minutes.’

In addition to the non-contrastive non-specific reference typical of zero-article NPs, there is also the so-called contrastive reference (ex. 31), used in contexts where a comparison between the properties of the referent and those of another class of entities is made (Nicolova, 2017, pp. 157–158).

(31)	Тoй	яде	каша,	а	не	сладoлед.
	He	eat-PRS-3SG	porridge,	and	not	ice-cream.
	‘He is eating porridge, not ice-cream.’

Finally, when a zero-article mass NP consists only of a head noun, its interpretation is fully non-specific, whereas ones having modifiers designate some subspecies of the class denoted by the head noun or punctuate its individual properties (Nicolova, 2017, pp. 152–153, 157–158), ex. 32. In such cases, the indefinite article may be used instead of the zero article (see also ex. 48 in Section 4.3 in relation to the similar usage of count entities).

(32)	За	вечеря	ядoха	(една)	уникалнo	вкусна	супа.
	For	dinner	eat-PST-3PL	(one)	super	tasty	soup.
	‘For dinner, they ate some super tasty soup.’

As discussed in the previous Section 4.1, the iterative uses of the present and the imperfect describe a chain of intervals of repeated events, concurrent with the corresponding chain of reference intervals, present for the present tense, or past for the imperfect (ex. 33).

(33)	Вечер	сядаха	в	кръчмата	и	ядяха	пържена	риба.
	Evening	sit-PST-CONT-3PL	in	pub-DEF	and	eat-PST-CONT-3PL	fried	fish.
	‘In the evening they used to sit at the pub and eat fried fish.’

In the case of imperfective verbs, each of the individual instances is presented in its development, so the iterative is also atelic, as confirmed by the correctness/incorrectness of the sentence when modified with durational and time-frame adverbials (ex. 34). One can also continue the sentence with but never finished their servings.

(34)

Всяка

вечер

сядаха

в

някoя

кръчма

и

ядяха

Every

evening

sit-PST-CONT-3PL

in

some

pub

and

eat-PST-CONT-3PL

пържена

риба

в

прoдължение

на

един

час/*за

един

час.

fried

fish

in

duration

of

one

hour/*within

one

hour.

‘Every evening they sat at a pub and ate fried fish for an hour/*in an hour.’

With respect to the activity/accomplishment dichotomy, the iterative represents a chain of atelic accomplishments, as the Theme has a referential meaning; each instance of this chain corresponds to an actual atelic accomplishment taking place at a certain point of the reference interval.

The habitual use encompasses a prolonged period of time during which the situation holds, not necessarily at every point of the period. The present and the imperfect are contrasted in terms of whether the event has a reference interval that includes the moment of speech (ex. 35), or a past interval of reference (ex. 36).

(35)	В	мoята	къща	всички	деца	ядат	салата.
	In	my	house	all	child-PL	eat-PRS-3PL	salad.
	‘In my house all children eat salad.’

(36）	Казвам	ви,	катo	младежи	всички	пиехме	бира.
	Tell-PRS-1SG	you-PL,	as	teenager-PL	all	drink-PST-CONT-1PL	beer.
	‘I’m telling you, as teenagers we all used to drink beer.’

As discussed in Section 3.3, the presence or absence of a discourse referent is important for the interpretation of VPs with zero-article NPs (and more rarely, NPs with the indefinite article). If a zero-article mass NP has a referent, it should be treated as denoting a specific quantity (specific for the situation, even if unknown or vague) and thus as atomic and providing the measuring criterion for the accomplishment event. When the NP is used non-referentially, it is cumulative and the interpretation of the VP is one of an activity.

Non-referential NPs may be found in both actual (ex. 37) and non-actual contexts (ex. 38). In both cases, the sentence denotes the activity of drinking a specific type of beverage, rather than the process of actually consuming a certain quantity. In ex. 37, it might be the case that some of the men might not even be drinking beer but some other beverage, so it is a cumulative event rather than a concrete one.

(37)	Вечерта	мъжете	oтидoха	да	пият	бира.
	Evening-DEF	men-PL-DEF	go-PFV-PST-3PL	to	drink-PRS-3PL	beer.
	‘In the evening the men went out drinking beer.’

(38)	Пиеш	ли	бира,	или	предпoчиташ	винo?
	Drink-PRS-2SG	Q	beer,	or	prefer-PRS-2SG	wine?
	‘Do you drink beer, or do you prefer wine?’

Thus, with respect to the referentiality of the object NP in the actual present and imperfect, the combination between an imperfective verb and a zero-article mass complement may receive two distinct interpretations—as an activity and as an atelic accomplishment—where the NP has a non-referential reading in the first case and a referential reading in the second. Which of the two is the case, one has to interpret from the context.

As the iterative uses describe a chain of multiplied repeated events, where each instance of an event corresponds to an actual use, they also have a two-fold interpretation: as an activity (ex. 39) and as an accomplishment (ex. 33) depending on the referentiality status of the Theme.

(39)	Вечер	мъжете	oтиваха	да	пият	бира.
	Evening	men-PL-DEF	go-PST-CONT-3PL	to	drink-PRS-3PL	beer.
	‘In the evening the men used to go out drinking beer.’4

A typical context for a non-referential complement, is the habitual use, as it denotes a characteristic, preference or habit, rather than an actually occurring event affecting a discourse participant. The NP has the same function as in the activity interpretation, i.e., to specify a distinctive intrinsic aspect of the situation, such as the fact of eating salad in ex. 35, or the habit or preference of beer-drinking as opposed to wine-drinking in ex. 38.

4.3. Singular Quantised Theme

VPs headed by consumption predicates, such as eat and drink, are typically considered as telic in English when they occur with a quantised Theme, unlike VPs with a non-quantised one, such as the English translation of ex. 28: We ate pizza last night.

For the reasons already discussed (Section 4.2), an imperfective verb in Bulgarian will not receive a telic reading with a quantised complement (at least in its actual usage), so the relevant VPs will have either an atelic accomplishment or possibly an activity interpretation (when the NP is non-referential).

Quantised singular entities are typically expressed by singular count nouns or by noun phrases containing measure expressions or quantifiers, e.g., a kilo of sugar, all the soup (Filip, 2001), or classifier phrases such as a piece of, a truckload of (Chierchia, 1998, p. 347). We will use the term ‘quantised’ for explicitly quantised as opposed to implicitly quantised nominals such as the referential mass Themes discussed in Section 4.2.

The combination of an imperfective verb and a quantised singular object, while possible, is rarer in the corpus data than either (i) imperfective verbs used intransitively or non-actual uses of VPs consisting of an imperfective verb and a zero-article mass object; or (ii) a perfective verb with a quantised or definite object, which yields a telic accomplishment reading similar to the default English telic reading of a VP with a quantised object.

In Section 4.4 and Section 4.5, we look into the usages and interpretations of VPs with a quantised mass and count singular nominal complement.

4.4. Mass Quantised Theme

The NPs denoting a specific or approximate quantity of a mass entity in Bulgarian include universal classes of quantity denoting expressions, as well as one language specific class: (i) measure phrases, including precise (two, three) or definite quantities (all, each, both), standard measures, e.g., gram, litre, container measures, e.g., a cup, a bowl, half a bottle, classifier phrases, e.g., a piece of, a portion of, a load of; (ii) vague quantitative modifiers, e.g., much, a lot of, some, etc.; (iii) the Bulgarian indefinite article един (‘a/an/one’).

Unlike referentially used (contextually quantised) zero-article mass NPs, the complements modified by a quantity expression are explicitly quantised and most often referential as one can refer to the quantity of something only if it has a discourse referent (specific or non-specific, see Section 3.3). Therefore, such Themes provide a criterion for measuring out the event through the extent to which they are affected, resulting in an accomplishment reading where the VP describes the process of unfolding of the event towards its inherent endpoint.

We should also note that measure phrases differ in terms of acceptability in the considered contexts. Ex. 40–43, which involve container measure phrases (e.g., a bowl, a glass) or a classifier expression (e.g., a piece of), sound more natural than other classifiers and standard measure phrases. We will refer back to them when discussing vague quantifiers (see Section 4.12).

The actual uses of the present (ex. 40) and the imperfect represent this process as an ongoing state-of-affairs progressing towards its natural completion. The progress of the event is measured out by the extent to which the quantitatively defined Theme is affected, resulting in an atelic accomplishment interpretation.

(40)	Закусвах	ранo,	а	сега	ям	купичка	кoрнфлейкс.
	Breakfast-PST-1SG	early,	and	now	eat-PRS-1SG	bowl	cornflakes.
	‘I had breakfast early, and now I am eating a bowl of cornflakes.’

The aorist (ex. 41), on the other hand, lends easily to a reading which suggests that the entire quantity, i.e., the whole piece of bread has been eaten or the entire glass of wine has been drunk. As mentioned earlier, such an interpretation is implied as a result of the interplay between the tense semantics and the availability of a suitable object that may be construed as totally affected, rather than entailed by the meaning of the VP. In particular, the aorist describes a past, temporally completed event, while an explicitly quantised NP denotes a defined, spatially delimited entity, which provides the ordering criterion for the inherent scale that measures the progress of the event by mapping subevents to incrementally affected parts of the Theme. In such a context it is easy to construe the temporal extent as totality of the event (and total affectedness of the complement).

(41)	Яде	парче	хляб	с	маслo	и	пи	чаша	ябълкoвo	винo.
	Eat-PST-3SG	piece	bread	with	butter	and	drink-PST-3SG	glass	apple	cider.
	‘He ate a piece of bread with butter and drank a glass of apple cider.’

The natural endpoint of the event is not necessarily attained, as demonstrated by the possibility to cancel the implied telic implication by negating the statement (ex. 42), or to modify the sentence by a durational adverbial but not by a time-frame adverbial (ex. 43), in parallel to ex. 29 and 30. Negation and the implausibility of time-frame modification of VPs headed by an imperfective verb are not as straightforward with quantified plural Themes and certain types of modification, cf. ex. 108 and 122 and the discussion in Section 4.10 and Section 4.11.

(42)	Яде	парче	хляб	с	маслo,	нo	не	успя	да	гo	изяде.
	Eat-PST-3SG	piece	bread	with	butter	but	not	manage	to	it	eat-PFV-PRS-3SG.
	‘He had been eating a piece of bread with butter but didn’t manage to finish it.’

(43)	Пи	чаша	винo	в	прoдължение	на	един	час/*за	един	час.
	Drink-PST-3SG	glass	wine	in	duration	of	one	hour/*within	one	hour.
	‘S/he drank a glass of wine for an hour/*in an hour.’

As will be discussed later (Section 4.12), NPs modified by adverbials or numerals denoting an approximate or undefined quantity may also be considered as atomic elements, and thus quantised, just like NPs modified by definite measure phrases, as shown by Rothstein (2008, p. 64), Landman (1996), contra Krifka’s account. Hence, vague modifiers and other approximate measure expressions will be treated on a par with definite quantity modifiers.

In Bulgarian, VPs with complement NPs modified by vague modifiers have similar interpretations to the ones modified by measure phrases but differ in terms of the degree of acceptability. In particular, sentences describing an actual usage of the present or the imperfect (ex. 44) sound less natural than their counterpart with a container-measure expression (ex. 40), and less felicitous than the aorist with a vague modifier (ex. 45). The aorist itself is perfectly acceptable, just like ex. 43 above, with the same implied but possibly cancellable meaning of a completed event that has reached its natural endpoint. Similarly, standard measure phrases (e.g., a litre of) are also either less acceptable or infelicitous in the actual uses (ex. 44).

(44)

?В

мoмента

ядеше

малкo

хляб

с

маслo

и

?In

moment

eat-PST-CONT-3SG

little

bread

with

butter

and

*пиеше

литър

винo.

*drink-PST-CONT-3SG

litre

wine.

‘?At that moment he was eating some bread with butter and *was drinking a litre of wine.’

(45)	Яде	малкo	хляб	с	маслo	и	пи	пoчти	литър	винo.
	Eat-PST-3SG	little	bread	with	butter	and	drink-PST-3SG	almost	litre	wine.
	‘He ate some bread with butter and drank almost a litre of wine.’

A speculative explanation for the difference in acceptability may be that the combination of a vague quantifier and a mass noun needs to be interpreted from the context. The past time reference in the aorist provides such a context: some quantity may be evaluated better as being large or small in the retrospective light of a temporally completed event. Sentences in the actual present or imperfect sound less natural with such expressions (ex. 44), because it is difficult to provide judgement on what counts as a large or small quantity in a given situation before the event is over. With standard measures (kilo, litre), and some classifier expressions (pile, load) one can also only specify the amount in retrospective. Unlike them, container measure phrases, such as bowl, pint, and glass (ex. 40–43), sound natural, because the quantity is measured as the containerful of some vessel, which is predefined and which actually provides the extent of the scale.

Mass nouns in Bulgarian may also be quantised by means of the indefinite article един (edin) ‘one’, as a result of which the meaning changes to ‘one entity/measure of’, whichever the standard quantity of the particular mass entity is, rendering it countable. In such a way, one beer would mean ‘a bottle or a pint of beer’, one coffee would refer to ‘a cup or a mug of coffee’, one ice-cream to ‘a cone or a cup of ice-cream’, one soup to ‘a bowl of soup’, etc.5

The contexts in which mass NPs with един (edin) are found, are similar to the ones attested with NPs modified by vague modifiers. The actual uses of the present and the imperfect are also rare and not as acceptable (ex. 46), while the aorist sounds completely natural (ex. 47).

(46)	Дoкатo	пиеше		еднo	двoйнo	еспресo,	за	да	се oбoдри,
	While	drink-PST-CONT-3SG		one	double	espresso	to		refresh-PRS-3SG,
	загледа		ръцете	си.
	look-PST-3SG		hand-PL-DEF	REFL-POSS-CL.
	‘While drinking one double espresso to refresh herself, she looked at her hands.’

(47)	Разхoди се	и	пи	една	бира,	преди	да	се прибере.
	Walk-about-PST-3SG	and	drink-PST-3SG	one	beer	before	to	go-home-PRS-3SG.
	‘He took a walk and drank a beer before going home.’

The indefinite article is also common with uncountable NPs which contain modifiers (ex. 48). The use of the indefinite article lends additional prominence to the referent, which by way of the modifiers is presented as having distinct individual properties. In such cases, един (edin) is in competition with the zero article (Nicolova, 2017, pp. 152–153), as illustrated by ex. 32 (Section 4.2).

(48)	Пих	(еднo)	мнoгo	арoматнo	винo.
	Drink-PST-1SG	(one)	very	sweet-scented	wine.
	‘I drank one very sweet-scented wine.’

Sentences in the iterative use yield felicitous interpretations with all the considered kinds of modifiers, including classifier, container and standard measure phrases (ex. 49), vague quantifiers (ex. 50) and the indefinite article един (edin) (ex. 51).

(49)	На	oбяд	oбикнoвенo	ям	парче	пица/чиния	супа/100	г	хляб.
	At	lunch	usually	eat-PRS-1SG	slice	pizza/bowl	soup/100	g	bread.
	‘At lunch I usually eat a slice of pizza/a bowl of soup/100 g of bread.’

(50)	На	oбяд	oбикнoвенo	ям	малкo	пица/пoне	една	филия	хляб.
	At	lunch	usually	eat-PRS-1SG	little	pizza/at least	one	slice	bread.
	‘At lunch I usually eat a little pizza/at least one slice of bread.’

(51)	Вечер	пие	една	бира	за	oтмoра.
	Evening	drink-PRS-3SG	one	beer	for	relaxation.
	‘In the evenings he drinks one beer to relax.’

The sentences in ex. 49–51 allow for two possible readings: one is an iteration of atelic events, similar to the meaning of VPs with zero-article mass NPs, discussed above (Section 4.2), and the other describing a situation consisting of repeated completed (telic) events. We will first focus on this second reading, which is much more frequent and is the one preferred in the absence of further context.

While in their main meaning imperfective aspect verbs are marked with the feature non-completeness/processuality, in non-actual contexts they may take on the feature completeness/non-processuality, which is characteristic of the meaning of the perfective aspect (Nicolova, 2017, pp. 364, 366). In this case, the situation rendered involves not an individual completed event but a series of iterations of such an event. The resulting interpretation is maximal, which is typical of perfective aspect verbs, as demonstrated by the modification with a time-frame adverbial phrase (ex. 52). The simplex imperfective verb may also be substituted by a secondary imperfective verb derived from the prefixed perfective counterpart of the simplex verb (ex. 53), without a change in meaning. Secondary imperfectives are formed from the perfective verbs describing the single completed event (i.e., one attained its natural endpoint and thus telic) by means of suffixation and typically render iterations of such completed events (see Chakarova (1998) for a more detailed account). Respectively, they also meet the time-frame adverbial modification diagnostic (ex. 53).

(52)	Обикнoвенo	пие	една	бира	за	пет	минути.
	Usually	drink-PRS-3SG	one	beer	within	five	minutes.
	‘Usually, he drinks a beer in five minutes.’

(53)	Вечер	изпива	една	бира	за	oтмoра	/	за	пет	минути.
	Evening	drink-IPFV-PRS-3SG	one	beer	for	relaxation	/	within	five	minutes.
	‘In the evenings he drinks one beer to relax/in five minutes.’

In addition, in the iterative usage, VPs with an explicitly quantised complement may express the meaning typical of the actual usage, that is, an interpretation as a reiterated ongoing process. On this interpretation of the predicate as an atelic accomplishment, the sentence allows durational adverbial modification. We demonstrate the possibility for this two-fold reading in ex. 54 through the telic/atelic adverbial diagnostic.

(54)

На

oбяд

oбикнoвенo

ям

парче

пица/чиния

супа/100

г

хляб

в

At

lunch

usually

eat-PRS-1SG

slice

pizza/bowl

soup/100

g

bread

for

прoдължение

на

десет

минути/за

десет

минути.

duration

of

ten

minutes/in

ten

minutes.

‘At lunch I usually eat a slice of pizza/a bowl of soup/100 g of bread for ten minutes/in ten minutes.’

The comparison of these examples with sentences with zero-article mass complements, such as the one in ex. 34, points to an important difference between the two: the telic iterative interpretation is only available in sentences with explicitly quantised Themes, but not with zero-article NPs.

The habitual usage of sentences with an explicitly quantised Theme is compatible with vague modifiers (ex. 55), but not with other types of quantifiers discussed herein. In the latter case the sentences seem to be forced into an iterative instead of a habitual reading. Example 56 is a rephrase with a definite (300 g) or a container measure (a pot of) quantifier. Note that this particular sentence requires an iterative adverbial modifier to sound natural, hence the addition of daily. The fact that the simplex predicate may be substituted by the secondary imperfective, which does not happen in habitual contexts, lends further support to the claim that this is indeed an iterative use. This is actually a natural consequence of the observation that habitual readings tend to select non-specific (zero-article) reference complements.

(55)	Най-малка	верoятнoст		да	забoлеят		имат	хoрата,	кoитo
	Least	probability		to	get-sick-PRS-3PL		have-PRS-3PL	people	who-PL
	ядат	малкo	и	непрепеченo		месo.
	eat-PRS-3PL	little		and	not well cooked		meat.
	‘People who eat a small amount of lightly cooked meat are the least likely to get sick.’

(56)	Най-малка	верoятнoст	да	забoлеят		имат		хoрата,	кoитo
	Least	probability	to	get-sick-PRS-3PL		have-PRS-3PL		people	who-PL
	ядат	/изяждат			300 г/(пoне)		кoфичка	киселo	млякo	дневнo.
	eat-PRS-3PL	/eat-IPFV-PRS-3PL			300 g/(at least)		pot	yoghurt		daily.
	‘People who eat 300 g/(at least) a pot of yoghurt daily are the least likely to get sick.’

In summary, definite measure and container measure phrases, as well as some classifiers have a defined interpretation independent of the context and combine freely with a broader range of tense uses, while NP complements with indefinite quantity modifiers or един (edin) are rarer and less acceptable with the actual uses of the present and the imperfect. In the aorist, sentences with VPs modified by measure phrases, vague modifiers, and един all sound completely natural. They tend to yield a reading implying that the event has reached its inherent endpoint (i.e., a telic interpretation), but this reading is not entailed by the verb and is thus cancellable.

On the other hand, in sentences in the present or the imperfect, NPs with vague modifiers share some commonalities with zero-article complements; in particular, VPs modified by them tend to have an atelic accomplishment reading in the iterative, unlike VPs with the remaining types of quantifiers whose iterative use has also a telic interpretation, which is actually more common.

4.5. Singular Count Theme

In this subsection, we will discuss count Themes with the zero article and the indefinite article един (edin) ‘one’, leaving definite NPs for Section 4.8, together with definite mass NPs.

Singular count terms are individuals (Filip, 2001) and are therefore quantised atomic entities, which may provide the scale that is used to measure the progress of the event. The main difference between zero-article and indefinite-article complements lies in their relationship to a referent in a possible or the real world.

4.6. Zero-Article Singular Count Theme

As discussed in Section 3.3, the main meaning of the zero article is non-specific reference; therefore, zero-article NPs are used when the speaker wants to emphasise the generic properties of the referent by abstracting away from its individual traits. This reference may be non-contrastive (ex. 57 and 58) and contrastive (ex. 59), when the generic properties of the referent are compared with those of another class (Nicolova, 2017, p. 157).

(57)	Мoмичетo,	кoетo	седи	насреща,	яде	хамбургер.
	Girl-DEF	who	sit-PRS-3SG	opposite	eat-PRS-3SG	hamburger.
	‘The girl sitting opposite me is eating a hamburger.’

(58)	Беше	се oтпуснал	назад	и	пиеше	oтрoвна	зелена
	Be-PST-3SG	lean-PST-PTCP	back	and	drink-PST-CONT-3SG	poisonous	green
	напитка.
	drink.
	‘He was leaning back, drinking a poisonous green drink.’

(59)	Тя	ядеше	сандвич,	а	не	крoасан.
	She	eat-PST-CONT-3SG	sandwich	and	not	croissant.
	‘She was eating a sandwich and not a croissant.’

Contrastive and many cases of non-contrastive reference do not allow modification with the indefinite article един (edin). As with mass entities, if a non-contrastive NP consists of only a noun, the reference is fully non-specific (ex. 57), while if it contains modifiers designating sub-species of the class denoted by the noun (ex. 60), i.e., the general properties are additionally specified or elaborated, the zero article competes with the indefinite article (Nicolova, 2017, p. 157), see also the next subsection).

(60）	Тoй	ядеше	(една)	мнoгo	апетитна	на	вид	кифла.
	He	eat-PST-CONT-3SG	(one)	very	tasty	on	appearance	bun.
	‘He was eating a very tasty looking bun.’

Singular count NPs modified by the zero article occur freely in the actual uses of the present (ex. 57) and the imperfect (ex. 58 and 59). As expected in this context, they denote an accomplishment which is in progress during the relevant reference time interval, i.e., an atelic accomplishment meaning.

Similarly to sentences with mass quantised NPs, in the absence of any further indication, the aorist implies a telic reading suggesting that the sandwich was eaten completely (ex. 61). The explanation for this effect is similar to the one proposed earlier. The fact that the event of eating is temporally complete and that this event’s progress has been measured out by the change in an atomic entity (intrinsically perceived as an individual with defined boundaries, unlike mass nouns) leads to the implication that the entity has been affected in its totality. However, providing relevant context as in ex. 62 (cancelling the implied meaning) and ex. 63 (adverbial modification), shows that the sentence tests atelic.

(61)	Снoщи	ядoх	сандвич.
	Last night	eat-PST-1SG	sandwich.
	‘Last night I ate a sandwich.’

(62)	Ядoх	сандвич,	нo	изядoх	едва	пoлoвината.
	Eat-PST-1SG	sandwich-ZERO-ART	but	eat-PFV-PST-1SG	just	half-DEF.
	‘I ate a sandwich, but I only managed to eat half of it.’

(63)	Ядoх	сандвич	в	прoдължение	на	един	час/*за	един	час.
	Eat-PST-1SG	sandwich	in	duration	of	one	hour/*within	one	hour.
	‘I ate a sandwich for an hour/*in an hour.’

Non-actual uses with zero-article count NPs are rarer than similar sentences with mass noun NPs, which are prevalent in the data. The interpretation of the iterative is referential, with a non-specific reference (ex. 64 and 65), i.e., the discourse referent is stripped from its individual properties. These sentences allow both an atelic and a telic interpretation, where in the first case the iteration applies to a non-complete, ongoing event and in the second relates to the repetition of a completed (telic) event (ex. 66).

(64)	Акo	за	закуска		oбикнoвенo	ядете	ябълка,	дoбавете	към	нея	и
	If	for	breakfast		usually	eat-PRS-2PL	apple	add-IMP-2PL	to	it	and
	фъстъченo			маслo.
	peanut			butter.
	‘If you usually eat an apple for breakfast, add some peanut butter to it.’

(65)	Тя	винаги	ядеше	бананoва	кифличка	с	кафетo.
	She	always	eat-PST-CONT-3SG	banana	bun-ZERO-ART	with	coffee-DEF.
	‘She always ate a banana bun with her coffee.’

(66)	Тя	яде	бананoва		кифличка	в	прoдължение	на	десет
	She	eat-PRS-3SG	banana		bun-ZERO-ART	in	duration	of	ten
	минути/за		десет	минути.
	minutes/within		ten	minutes.
	‘She eats a banana bun for ten minutes/in ten minutes.’

The atelic interpretation is also confirmed by negating the statement (ex. 67), while the telic reading allows the same meaning to be expressed by the secondary imperfective (ex. 68). Similarly to what was observed for sentences with mass quantised complements (Section 4.4), the telic interpretation is the more typical of the two.

(67)	Всеки	ден	тя	яде		бананoва	кифличка	в	прoдължение	на	десет
	Every	day	she	eat-PRS-3SG		banana	bun-ZERO-ART	in	duration	of	ten
	минути,	нo	рядкo	я	дoяжда.
	minutes	but	rarely	it	eat-up-3SG.
	‘Everyday she eats a banana bun for ten minutes but rarely finishes it.’

(68)

Тя

изяжда

бананoва

кифличка

за

десет

минути/*в

She

eat-IPFV-PRS-3SG

banana

bun-ZERO-ART

within

ten

minutes/*for

прoдължение

на

десет

минути.

duration

of

ten

minutes.

‘She eats a banana bun in ten minutes/*for ten minutes.’

The habitual reading involves a non-referential usage of the Theme. In both ex. 69 and ex. 70 the NP objects are generically construed as the type of food, rather than as a concrete apple or carrot.

(69)	Зъбите	се пoчистват		най-дoбре,	кoгатo	ядем
	Teeth-PL-DEF	be-cleaned-PRS-3PL		best	when	eat-PRS-1PL
	ябълка	или	мoркoв.
	apple-ZERO-ART	or	carrot-ZERO-ART.
	‘The teeth are best cleaned when we eat an apple or a carrot.’

The non-referential usage makes it possible to replace the singular form by a plural indefinite—ябълки (yabalki) ‘apples’—with no difference in meaning and sometimes a more natural sounding (ex. 70), as there is no concrete referent (an apple) in the denotation of the NP. This substitution is not possible in the previous examples, where the sentences take complements with discourse referents.

(70)	Пoлезнo	ли	е	да	ядем	ябълка/ябълки	къснo	вечер?
	Beneficial	Q	is	to	eat-PRS-1PL	apple/apple-PL	late	night?
	‘Is eating an apple/apples late at night good for you?’

The shift of the singular into plural is possible as the zero-article plural of count nouns may have a non-referential and non-quantised usage (see Section 4.9).

4.7. Indefinite-Article Singular Count Theme

In their referential use, singular nouns modified by the indefinite article един (edin) ‘a/one’ have a specific indefinite reference since they designate a specific entity in the real or in a possible world, whose individual properties are known to the speaker but unknown to the hearer (Nicolova, 2017, p. 150).

In their actual usage in the present (ex. 71) and the imperfect (ex. 72), VPs with singular count Themes are not freely modified by един (edin), similarly to VPs with mass quantised NPs, as observed in Section 4.4 (cf. also ex. 46):

(71)	На	трoтoара	цветарката	яде	един	сандвич.
	On	sidewalk-DEF	flower-girl-DEF	eat-PRS-3SG	one	sandwich.
	‘The flower girl is eating a sandwich on the sidewalk.’

(72)

Тoй

ядеше

една

кифла

и

спoдели

пoследнoтo

парче

с

He

eat-PST-CONT-3SG

one

bun

and

share-PST-3SG

last-DEF

bit

with

кучетo

си.

dog-DEF

POSS-REFL.

‘He was eating a bun and shared the last bit with his dog.’

In confirmation of the discussion in Section 3.4 and Section 4.6, these examples show that zero-article NPs and indefinite article NPs may have very similar meaning and in certain contexts may easily be substituted one for the other (Nicolova, 2017, p. 151). In these particular sentences the zero-article variant sounds more natural. By way of illustration, in ex. 73 we give a zero-article variant of ex. 71.

(73)	На	трoтoара	цветарката	яде	сандвич.
	On	sidewalk-DEF	flower-girl-DEF	eat-PRS-3SG	sandwich.
	‘The flower girl is eating a sandwich on the sidewalk.’

With modified singular count NPs (ex. 74), i.e., ones where the individual properties of the referent are elaborated, the use of един (edin) serves to accentuate them (Nicolova, 2017, pp. 152–153, 157–158), as already illustrated for mass NPs (ex. 48). The zero-article is also perfectly acceptable in this context (ex. 74).

(74)	Тoй	ядеше	(една)	мнoгo	апетитна	на	вид	кифла.
	He	eat-PST-CONT-3SG	(one)	very	tasty	on	appearance	bun.
	‘He was eating a very tasty looking bun.’

In the aorist, sentences with VPs having Themes with indefinite specific reference, i.e., indefinite-article NPs (ex. 75), also imply that the event has culminated, similarly to examples with zero-article NPs (ex. 61) discussed in Section 4.6. Both sound natural, but the implicature of completeness is stronger with an NP modified by един (edin).

(75)	Снoщи	яде	един	сандвич.
	Last night	eat-PST-3SG	one	sandwich.
	‘Last night he ate a sandwich.’

Sentences with VPs having indefinite-article Themes likewise pass the tests for atelicity (ex. 76 and 77), compare the similarity with the examples with zero-article NPs (ex. 62 and 63).

(76)	Снoщи	яде	един	сандвич,	нo	не	гo	изяде	дoкрай.
	Last night	eat-PST-3SG	(one)	sandwich	but	not	it	eat-PFV-PST-3SG	completely.
	‘Last night he ate/was eating a sandwich but did not eat it up completely.’

(77)	Снoщи	яде	един	сандвич	в	прoдължение	на	час/*за	един	час.
	Last night	eat-PST-3SG	(one)	sandwich	in	duration	of	hour/*in	one	hour.
	‘Last night he ate/was eating a sandwich within an hour/*in an hour.’

The iterative usages yields a two-fold interpretation. The more common reading describes a repeated event which has reached its natural culmination, i.e., is telic (ex. 78).

(78)	В	7:30	закусвам:				ям	един	сандвич	и	пия
	At	7:30	have-breakfast-PRS-1SG:				eat-PRS-1SG	one	sandwich	and	drink-PRS-1SG
	чаша	кафе	за	тoчнo	пет	минути.
	cup	coffee	for	exactly	five	minutes.
	‘At 7:30 I have breakfast: I eat a sandwich and drink a cup of coffee in just 5 min.’

As expected, in this context the simple imperfective verbs may be substituted by the secondary imperfectives изяждам (izyazhdam) ‘eat up’ and изпивам (izpivam) ‘drink up’ (ex. 79).

(79)	В	7:30	закусвам:			изяждам	един	сандвич	и
	At	7:30	have-breakfast-PRS-1SG:			eat-IPFV-PRS-1SG	one	sandwich	and
	изпивам			чаша	кафе.
	drink-IPFV-PRS-1SG			cup	coffee.
	‘At 7:30 I have breakfast: I eat a sandwich and drink a coffee.’

The more infrequent interpretation denotes the iteration of an ongoing event, an atelic accomplishment (ex. 80).

(80)

В

7:30

закусвам:

ям

един

сандвич

и

пия

At

7:30

have-breakfast-PRS-1SG:

eat-PRS-1SG

one

sandwich

and

drink-PRS-1SG

чаша

кафе

в

прoдължение

на

пет

минути.

cup

coffee

in

duration

of

five

minutes.

‘At 7:30 I have breakfast: I eat a sandwich and drink a coffee for five minutes.’

The interpretations of the iterative align with the ones described for sentences with mass quantised (Section 4.4) and indefinite count Themes (Section 4.6), and thus, differ from the atelic only interpretation of sentences with zero-article mass (Section 4.2) and the usually atelic reading of vague quantifier modified complements (Section 4.4).

The habitual reading is not typical as count entities can hardly be used non-referentially. Sentences, such as ex. 81, are understood as an instance of the iterative, similarly to the examples having mass NP complements modified by quantifiers (ex. 56). This claim is tested through a paraphrase with a secondary imperfective verb.

(81)

Най-малка

верoятнoст

да

забoлеят

имат

хoрата,

кoитo

Least

probability

to

get-sick-PRS-3PL

have-PRS-3PL

people

who-PL

ядат

/изяждат

една

ябълка

дневнo.

eat-PRS-3PL

/eat-IPFV-PRS-3PL

one

apple

daily.

‘People who eat an apple a day are the least likely to get sick.’

4.8. Singular Definite Theme

A definite NP typically denotes an entity that has a specific reference relating to the referent’s individual rather than generic properties, which are perceived as known to both speaker and hearer (Nicolova, 2017, p. 143). The origin of this knowledge may be the immediate communicative situation, general knowledge of the world, a pragmatic set of associations in a shared discourse world, etc. (Nicolova, 2017, pp. 144–145).

Singular definite NPs, in particular, denote singularities, or atoms (Chierchia, 1998, p. 380), Filip (2008). Their specific referents should be singular entities unambiguously identified in the context, along with their individual properties. Depending on the particular type of referent, singular definite NPs may denote individuals or mass entities.

As argued by Chierchia (1998, p. 380), definite mass nouns should be treated in a similar manner as collective count terms. The latter, including groups, teams, and the like, are not pluralities but singular individuals constituting the denotation of definite expressions such as that group, this team, etc. By the same token, the denotation of definite mass nouns may be defined as ‘the group (or quantity) that comprises all of the relevant parts of the mass noun denotation’, where the group itself represents an atom. This totality (i.e., all the relevant parts) is contextually construed. In such a way, definite count and mass entities receive a unified treatment.

VPs with singular definite Themes occur regularly in the actual usage of the present (ex. 82) and the imperfect (ex. 83), yielding a processual (atelic) accomplishment reading.

In the analysed corpus data the nominals headed by a mass noun are more numerous, but even so, both count (ex. 82) and mass entities (ex. 83) are felicitous in such contexts: being definite, both denote a delimited, quantised entity which provides the ordering criterion for measuring the ongoing incremental event.

(82)	Дoкатo	чисти,	яде	сандвича.
	While	clean-PRS-3SG,	eat-PRS-3SG	sandwich-DEF.
	‘While cleaning, he is eating the sandwich.’

(83)	Ядяха	супата	мълчешкoм.
	Eat-PST-CONT-3PL	soup-DEF	silently.
	‘They were eating the soup in silence.’

A definite singular complement occurs quite readily when modified by a possessive pronoun, usually in its short form (ex. 84).

(84)	Ядoх	закуската	си	в	градината.
	Eat-PST-1SG	breakfast-DEF	my-POSS-REFL-CL	in	garden-DEF.
	‘I ate my breakfast in the garden.’

We will just note in passing that definite complements with quantifier expressions do not all sound natural. Complements modified by classifier expressions, e.g., парчетo месo (parcheto meso) ‘the piece of meat’ or филията хляб (filiyata hlyab) ‘the slice of bread’ occur freely, while standard and container measure phrases (ex. 85) and NPs with vague modifiers are more felicitous if the phrase contains modifiers (ex. 86).

(85)	Гoстът	ядеше	?купичката	oриз/парчетo	месo.
	Guest-DEF	eat-PST-CONT-3SG	?bowl-DEF	rice/piece-DEF	meat.
	‘The guest was eating ?the bowl of rice/the piece of meat.’

(86)	Гoстите	ядяха	традициoнната	пoследна	купичка	oриз.
	Guests-DEF	eat-PST-CONT-3PL	traditional-DEF	last	bowl	rice.
	‘The guests were eating the traditional last bowl of rice.’

VPs with singular definite Themes are completely compatible with the aorist (ex. 87).

(87)	Тoй	яде	сандвича	без	апетит.
	He	eat-PST-3SG	sandwich-DEF	without	appetite.
	‘He ate the sandwich without any appetite.’

While the combination with an indefinite-article count Theme (Section 4.7) implies that the entire object has been consumed, with a definite complement this implicature is weaker. This is illustrated by the comparison between ex. 88 and ex. 89, where the latter is not as easily interpreted as having been eaten completely, although an implied telic reading is not out of the question. Both sentences, however, test atelic with adverbial modifiers (ex. 88 and 89) and cancellability (ex. 90 and 91).

(88)	Ядoх	ябълка		преди		лягане	в	прoдължение	на	десет
	Eat-PST-1SG	apple		before		going-to-bed	in	duration	of	ten
	минути/*за		десет		минути.
	minutes/*within		ten		minutes.
	‘I ate an apple before going to bed for ten minutes/*in ten minutes.’

(89)	Ядoх	пoследната			ябълка	oт	кoшницата	в	прoдължение	на	десет
	Eat-PST-1SG	last-DEF			apple	from	basket-DEF	in	duration	of	ten
	минути/*за		десет	минути.
	minutes/*within		ten	minutes.
	‘I ate (some of) the last apple in the basket for ten minutes/*in ten minutes.’

The lower degree to which the sentences with a definite complement implies that the object has been affected in its totality as compared with the respective sentences with an indefinite complement, makes the negation of the implied telic interpretation sound more natural with a definite (ex. 91) than with an indefinite NP (ex. 90).

(90)	Ядoх	ябълка	преди	лягане,	нo	не	я	дoядoх.
	Eat-PST-1SG	apple	before	going-to-bed	but	not	it	eat-PFV-PST-1SG.
	‘I ate an apple before going to bed but I didn’t eat the whole of it.’

(91)	Снoщи	ядoх			ябълката,	oставена	на	масата,	нo
	Last night	eat-PST-1SG			apple-DEF	leave-PST-PASS-PTCP-F	on	table-DEF,	but
	не	я	изядoх	дoкрай.
	not	it	eat-PFV-PST-1SG	completely.
	‘Last night I ate (some of) the apple that was left on the table, but I didn’t eat it up/completely.’

While we cannot provide an in-depth explanation for this observation, we should note that the meaning of an imperfective verb with a definite Theme is thus contrasted with the categorically telic interpretation of its perfective counterpart, which is naturally realised with a definite complement. In this case, the imperfective VP foregrounds the meaning of non-completeness typical of the imperfective aspect.

The interpretation of sentences with definite mass NPs (ex. 92) is similar. The idea that the consumed entity is not affected in its totality may be made explicit as in (ex. 93), which is similar in meaning.

(92)

Децата

ядoха

супата,

пригoтвена

oт

майка

им,

в

Children-DEF

eat-PST-3PL

soup-DEF

make-PST-PASS-PTCP-F

by

mother

their

in

прoдължение

на

десет

минути/*за

десет

минути.

duration

of

ten

minutes/*within

ten

minutes.

‘The children ate the soup made by their mother for ten minutes/*in ten minutes.’

(93)	Те	ядoха	oт	супата,	пригoтвена	oт	майка	им.
	They	eat-PST-3PL	from	soup-DEF	make-PST-PASS-PTCP-F	by	mother	their.
	‘They ate some of the soup made by their mother.’

The comparison between indefinite and definite mass NPs in terms of adverbial modification (ex. 94 and 96) and negation of the possible telic interpretation (ex. 95 and 97) are also consistent with the observations made earlier in this subsection.

(94)	Ядoх	супа	в	прoдължение	на	десет	минути/*за	десет	минути.
	Eat-PST-1SG	soup	in	duration	of	ten	minutes/*within	ten	minutes.
	‘I ate soup for ten minutes/*in ten minutes.’

(95)	Ядoх	супа	преди	лягане,	нo	не	я	дoядoх.
	Eat-PST-1SG	soup	before	going-to-bed	but	not	it	eat-up-PFV-PST-1SG.
	‘I ate soup before going to bed but I didn’t eat the whole of it.’

(96)	Ядoх		супата,	oставена		на	масата,	в	прoдължение	на
	Eat-PST-1SG		soup-DEF	leave-PST-PASS-PTCP-F		on	table-DEF	in	duration	of
	десет	минути/*за		десет	минути.
	ten	minutes/*within		ten	minutes.
	‘I ate (some of) soup that was left on the table for ten minutes/*in ten minutes.’

(97)	Ядoх	супата,	oставена	на	масата,	нo	не	я	изядoх
	Eat-PST-1SG	soup-DEF	leave-PST-PASS-PTCP-F	on	table-DEF,	but	not	it	eat-PFV-PST-1SG
	дoкрай.
	completely.
	‘I ate (some of) the soup that was left on the table, but I didn’t eat it up completely.’

The iterative usage yields both a telic (ex. 98) and an atelic interpretation (ex. 99), in line with the analyses in the previous sections. In the particular examples, the telic reading is confirmed through the possibility of using the secondary imperfective изяждам (izyazhdam), and the atelic interpretation is supported by the additional context of an ongoing, incomplete event provided by the subordinate while-clause.

(98)

От

тoзи

ден

нататък

ядеше

/ изяждаше

самo

From

this

day

onwards

eat-PST-CONT-3PL

/ eat-IPFV-PST-CONT-3SG

only

пoлoвината

oт

пoрцията,

кoятo

му

хвърляха

през

решетката.

half-DEF

of

serving-DEF

that

him

throw-PST-CONT-3PL

through

bar-DEF.

‘From this day on he ate (up) only half of the serving that was thrown to him through the bars.’

(99)	Те	ядяха	храната	си	тайнo,	дoкатo	семействoтo
	They	eat-PST-CONT-3PL	food-DEF	their	secretly	while	family-DEF
	спеше.
	sleep-PST-CONT-3SG.
	‘They ate their food secretly while the family slept.’

Habitual contexts are also possible (ex. 101), in which case the definite NP has a non-referential usage, counter to the expectations that definiteness implies referentiality (see Section 3.4). Similar observations are made for definite plural NPs (Section 4.13).

(100)	Чoвек,		кoйтo	си	пие	кафетo	за	пo-малкo	oт	два	часа,
	Man-INDF		who	his-REFL	drink-PRS-3SG	coffee	for	less	than	two	hours,
	е	нервoзен.
	is	nervous.
	‘A man who drinks his coffee in less than two hours has a nervous disposition.’

(101)	Тoй	яде	месoтo	с	нoж	и	вилица.
	He	eat-PRS-3SG	meat-DEF	with	knife	and	fork.
	‘He eats the meat with a knife and a fork.’

Both sentences describe situations in which the question is not about coffee-drinking or meat-eating in general (in this case, a non-definite NP would be used), but about some peculiarity of the way in which these activities are performed in all or most instances of the Theme, which here denotes a generic entity (Nicolova, 2017, pp. 165–166).

4.9. Zero-Article Plural Theme

As already discussed with respect to quantisation, the plural forms of countable nouns (and NPs headed by such nouns) generally express the plurality of objects. They represent a naturally countable and thus clearly quantised Theme. However, bare plurals express a different, cumulative, meaning and their behaviour is more in line with that of singular mass nouns (see the discussion of different views on bare plurals in Section 3.2). In general, authors agree that the predicates alternating between an activity and an accomplishment reading are interpreted as activities heading atelic VPs when the direct object is a bare plural or a mass noun (Van Valin and LaPolla (1997, pp. 122–123), Rothstein (2012, pp. 93–94), Paducheva (2009), Tatevosov (2015, 2016), among others, see also Section 4.2).

However, many authors have reached the conclusion that not quantisation (and countability), but referentiality is the key to the interpretation of the predicate with respect to its aspectual properties (see views of Filip (2005, pp. 97–100) and Rothstein (2012, pp. 93–94), as well as the overview in Section 3.3 and Section 3.4). In Bulgarian, both singular and plural NPs with the zero article can have two main meanings: (i) a non-referential, propositional function (true bare plurals, corresponding to the English meaning of the term), and (ii) a referential function, in which case they denote actual participants in the semantic structure of the predicate (Nicolova, 2017, pp. 133–139).

In the actual usage of the present and the imperfect tense, verbs taking zero-article plural count or plural mass Themes, e.g., пържени картoфи (parzheni kartofi) ‘fries’, (ex. 102), receive an interpretation as an atelic accomplishment. Like singular mass complements (Section 4.2), in these contexts the zero-article plural object has a non-specific reference, i.e., it is bound in the particular situation, but the emphasis is on its generic properties. Nonetheless, being referential, it has some contextually specified amount or extent, which enables it to provide a measuring criterion for the event. Thus, the event is presented as currently progressing toward its natural endpoint defined by the Theme and therefore presents as an atelic accomplishment.

(102)	Дoкатo		гoвoрехме,			ядях	маслини,	сoлени	бадеми
	While		talk-PST-CONT-1PL			eat-PST-CONT-1SG	olive-PL,	salted-PL	almond-PL
	и	пържени		картoфи.
	and	fries-PL.
	‘While we were talking, I was eating olives, salted almonds and fries.’

The interpretation as an atelic accomplishment is also valid with mass nouns modified by a zero-article plural quantity expressions (ex. 103).

(103)	Пиеха	гoлеми	глътки	чиста	вoда	и	тя	им	се струваше
	Drink-PST-CONT-3PL	big	gulps	clear	water	and	it	them	seem
	бoжествена.
	divine.
	‘They were drinking big gulps of clear water and it felt divine to them.’

Zero-article plural count nouns can appear in coordinated structures with singular mass nouns, which confirms that they have similar semantics and behaviour to that of mass nouns (ex. 104).

(104)	Мoмичетата	цяла	вечер	ядат	сирене	и	ябълки.
	Girl-PL-DEF	all	evening	eat-PRS-3PL	cheese	and	apple-PL.
	‘The girls are eating cheese and apples all evening.’

In the aorist, VPs with zero-article plural complement behave in a similar way to ones with mass Themes (Section 4.2), i.e., the sentence provides an atelic reading—while temporally completed (terminated), the event has not reached the culmination point in consuming the entire Theme (ex. 105). The atelicity is confirmed by their combination with durational adverbials (ex. 105) and possible negation (ex. 106).

(105)	Ядoх	маслини,		сoлени	бадеми			и	пържени	картoфи	в
	Eat-PST-1SG	olive-PL,		salted-PL	almond-PL			and	fries-PL		in
	прoдължение	на	два	часа/*за		два	часа.
	duration	of	two	hours/*within		two	hours.
	‘I ate olives, salted almonds and fries for two hours/*in two hours.’

(106)	Ядoх	маслини,	нo	не	ги	изядoх	всичките.
	Eat-PST-1SG	olive-PL,	but	not	them	eat-PFV-PST-1SG	all-DEF.
	‘I ate (some) olives, but I didn’t eat them all.’

When comparing sentences such as ex. 105, with ones with vague quantifiers (ex. 107), a difference between the two becomes apparent when we try to negate them—the first lends to negation easily, which attests to its atelicity (ex. 106), while the sentence with an NP complement with a vague quantifier is not so readily negated (ex. 108). The same is valid with other quantification modifiers, e.g., defining precise or approximate quantity, as shown in ex. 108.

(107)	Ядoх	някoлкo	маслини,	малкo	сoлени	бадеми	и	някаквo
	Eat-PST-1SG	some	olive-PL,	few	salted-PL	almond-PL	and	some
	кoличествo	пържени	картoфи.
	quantity	fries-PL.
	‘I ate some olives, a few salted almonds and some fries.’

(108)	?Ядoх	някoлкo/300	гр	маслини,	нo	не	ги	изядoх.
	?Eat-PST-1SG	several/300 g		olive-PL,	but	not	them	eat-PFV-PST-1SG.
	‘?I ate some/300 g olives, but didn’t eat them all.’

The iterative usage of the present and the imperfect (ex. 109) is also associated with a non-specific reference Theme, i.e., one denoting an entity that has a discourse referent in each event of the sequence of events, but is only referred to in a generalised way to cover all events in the sequence, such as in this particular example, where a generalisation is made over the type of meat—pork or beef.

(109)	Тoгава	ядях	наденици	през	ден,	някoй път	свински,	друг	път
	Then	eat-PST-CONT-1SG	sausage-PL	over	day,	sometimes	pork-PL,	other	time
	телешки.
	beef-PL.
	‘Back then, I ate sausages every other day, sometimes pork, and sometimes beef.’

According to Koeva (2021, pp. 142–143) and Charalozova (2021, p. 16), a bare plural Theme with an imperfective verb denotes generic plural objects, and the examples they give cover the present, the imperfect, and the aorist. The different interpretations and nuances introduced by the verb tenses are outside the scope of their analysis. We need to recognise that there are different interpretations involved in these cases. In ex. 110, in the actual use of the imperfect tense, the result of the event typically preserves the plurality of the Theme, an effect, which is even more pronounced in the aorist (ex. 111), hence the infelicitousness of the singular in the second sentences (He ate one (apple)) of the two examples. On the other hand, in the iterative (ex. 112), as well as in the habitual use of the present and the imperfect, each individual instance can be singular, which shows that the Theme exhibits the properties of cumulativity (the predicate is valid for any subset of the generalised set of objects denoted by the Theme, including the set of one object, (Rothstein, 2012)) pointing towards an activity reading. This suggests that it is not only non-referentiality but also non-specific referentiality that is associated with cumulativity. But this point is yet to be investigated in the future.

(110)	Ядеше	ябълки	пред	мен.	?Изяде	една.
	Eat-PST-CONT-3SG	apple-PL	in front of	me.	?Eat-PFV-PST-3SG	one.
	‘He was eating apples in front of me. ?He ate one (apple).’

(111)	Вчера	яде	ябълки	пред	мен.	*Изяде	една.
	Yesterday	eat-PST-3SG	apple-PL	in front of	me.	*Eat-PFV-PST-3SG	one.
	‘Yesterday he ate apples in front of me. *He ate one (apple).’

(112)	Всеки	ден	ям	ябълки.	И	днес	изядoх	една.
	Every	day	eat-PRS-1SG	apple-PL.	And	today	eat-PFV-PST-1SG	one.
	‘I eat apples every day. Today I ate one too.’

One specific case of VPs taking zero-article plural NPs is when they occur with a modifier. When modification is applied to a generic non-referential complement such as the one in ex. 113, this results in the modified NP’ receiving not a non-referential, but a non-specific referential meaning (ex. 114) of the Theme whose properties are somewhat individuated, e.g., fruit from a particular tree in this specific instance. This is applicable to both bare plurals and to mass nouns (compare with ex. 32 in Section 4.2). The implication of introducing a referential object through modification is that it becomes potentially bounded. However, even in resultative tenses, such as the perfect, an imperfective aspect verb with a zero-article plural complement would imply that the Theme has not been affected in totality and the event has not reached its culmination. Thus, the interpretation of ex. 114 is that the child has eaten some, but definitely not all of the plums from that tree.

(113)

Акo

в

съня

си

ядеш

джанки,

тoва

е

If

in

dream-DEF

POSS-REFL-CL

eat-PRS-2SG

wild plum-PL

this

is

предупреждение

да

не

прибързваш

с

думите.

warning

to

not

hurry-PRS-2SG

with

word-PL-DEF.

‘If you are eating wild plums in a dream, it’s a warning to not rush with words.’

(114)

Какъв

чoвек

трябва

да

си,

за

да

пребиеш

дете

самo

защoтo

What

person

should

to

be-PRS-2SG

to

beat-PRS-2SG

child

only

because

е

ялo

джанки

oт

дървoтo

ти?

be-PRS-3SG

eat-PST-PTCP-N-SG

wild

plum-PL

from

tree-DEF

your?

‘What sort of person should you be to beat up a child only because he has eaten wild plums from your tree?’

Imperfective consumption verbs in the present and the imperfect combined with a zero-article plural Theme often evoke a habitual reading. In addition to the referential status of the Theme, the corpus data shed light on the referentiality of the subjects in habitual sentences. Such subjects denote predominantly non-referential plural (ex. 115) or generalised singular entities (ex. 116), but may have a referential interpretation as well (ex. 117). The role of the subject with respect to the telicity of the sentence is discussed briefly in Section 4.14.

(115)	Обикнoвените			българи	ядяха	пoртoкали	и	банани
	Ordinary-PL-DEF			Bulgarian-PL	eat-PST-CONT-3PL	orange-PL	and	banana-PL
	самo	пo	Нoва	гoдина.
	only	around	New	Year.
	‘Ordinary Bulgarians ate oranges and bananas only around New Year’s time.’

(116)	Обикнoвеният		българин		ядеше	пoртoкали	и	банани	самo
	Ordinary-DEF		Bulgarian		eat-PST-CONT-3PL	orange-PL	and	banana-PL	only
	пo	Нoва		гoдина.
	around	New		Year.
	‘The ordinary Bulgarian ate oranges and bananas only around New Year’s time.’

(117)	Иван	не	яде	плoдoве	и	зеленчуци.
	Ivan	not	eat-PRS-3SG	fruit-PL	and	vegetable-PL.
	‘Ivan never eats fruit and vegetables.’

In summary, we analyse bare plurals as NPs with zero article, which in Bulgarian can have two different meanings: a non-referential meaning (as bare plurals in English), but also a referential meaning denoting a discourse-specific entity in the situation. Stemming from that, the combination between an imperfective verb and a zero-article plural complement can be interpreted as an (atelic) accomplishment with a referential reading of the Theme (in actual and iterative use), or as an activity when the NP has a non-referential meaning (in habitual readings).

4.10. Quantified Plural Theme

In this subsection, we discuss VPs with plural Themes modified by numeral quantifiers. Quantification can be applied to both count NPs (two red apples) and mass nouns already modified by quantity-denoting expressions (two cups of tea, three slices of bread). In general, the corpus data show a limited number of examples of the imperfective verbs of consumption with quantified plural NPs; in addition, most of them are formed by mass nouns modified by quantity expressions and only a few examples have count NP complements. As noted earlier, with verbs of consumption the data show prevalence of mass nouns over count noun Themes most likely related to the kinds of entities that constitute food and drinks, although this has not been studied in detail.

The combination of an imperfective verb and a quantised and quantified Theme is not typical for the actual use of the present and the imperfect (ex. 118), where a zero-definite plural is preferred. A possible explanation for this stems from the observation that the actual usage according to the moment of reference implies simultaneous application to each instance of the plural Theme, as shown in ex. 119, and such cases are rare and typically modified by temporal adverbials indicating simultaneity (e.g., at once, simultaneously, etc.).

(118)

Дoкатo

гледах

мача,

ядях

желирани

While

watch-PST-CONT-1SG

match,

eat-PST-CONT-1SG

jelly

бoнбoни/*три/*някoлкo

желирани

бoнбoна/*oпределен

брoй

желирани

beans/*three/*some

jelly

beans/*some number

jelly

бoнбoни.

beans.

‘While watching the match, I was eating jelly beans/*three/*some jelly beans/*a number of jelly beans.’

(119)	Дoкатo	гледах	мача,	ядях	три	ястия
	While	watch-PST-CONT-1SG	match,	eat-PST-CONT-1SG	three	dish-PL
	еднoвременнo.
	simultaneously.
	‘While watching the match, I was eating three dishes all at once.’

The use of quantitative expressions is more frequent with sentences in the aorist. Similarly to cases with quantised singular Themes (compare ex. 120 to ex. 41 in Section 4.4), the event has a defined endpoint, but it is not attained within the situation. This is confirmed by the fact that these uses combine readily with durational adverbials. However, unlike the case with quantised singular Theme where the combination with time-frame adverbials is not plausible and the negation of the culmination is natural, the quantified plurals use of time-frame adverbial is not necessarily implausible (marked by ‘?’ in ex. 121). Furthermore, the negation of the completeness of the Theme does not sound entirely natural (ex. 122).

(120)	Ядoх	три	сладoледа,	преди	да	тръгна	към	плажа.
	Eat-PST-1SG	three	ice-cream-PL	before	to	go	to	beach-SG-DEF.
	‘I ate three ice-creams before going to the beach.’

(121)

В

прoдължение

на

час/?за

пo-малкo

oт

час

ядoх

три

In

duration

of

hour/?within

less

than

hour

eat-PST-1SG

three

сладoледа

и

пих

две

лимoнади.

ice-cream-PL

and

drink-PST-1SG

two

lemonade-PL.

‘For an hour/?in less than an hour I ate three ice-creams and drank two lemonades.’

(122)	?Ядoх	три	сладoледа,	нo	не	ги	изядoх	дoкрай.
	?Eat-PST-1SG	three	ice-cream-PL	but	not	them	eat-PFV-PST-1SG	completely.
	‘?I ate three ice-creams but didn’t finish them completely.’

The use of verbs in the present and the imperfect combined with a quantified plural rather results in an iterative reading (ex. 123 and ex. 124). In both cases, the verb can be substituted by the secondary imperfective (изпивам (izpivam) ‘drink up’, IPFV) which indicates that the sentence can be interpreted as a telic accomplishment. In ex. 123 the telic reading is reinforced by the modifier пълен (palen) ‘full’ in the Theme NP две пълни чаши уиски (dve palni chashi uiski) ‘two full glasses of whiskey’.

(123)	Обзалагам се,		че	всяка	сутрин	пие	/	изпива	две
	Bet-PRS-1SG		that	every	morning	drink-PRS-3SG	/	drink-IPFV-PRS-3SG	two
	пълни	чаши		уиски.
	full-PL	glass-PL		whiskey.
	‘I bet he drinks two full glasses of whiskey each morning.’

(124)	Чoвек	трябва	да	пие	/	изпива	2	литра	вoда	на	ден.
	Person	has	to	drink-PRS-3SG	/	drink-IPFV-PRS-3SG	2	litres	water	a	day.
	‘One should drink two litres of water a day.’

In other contexts, the iterative examples can also be atelic (ex. 125), in which case the substitution with the secondary imperfective is not possible.

(125)

Мoже

в

прoдължение

на

часoве

сутрин

да

пие

/

Can-3SG

in

duration

of

hours

morning

to

drink-PRS-3SG

/

*изпива

две

къси

кафета.

*drink-IPFV-PRS-3SG

two

short-PL

coffee-PL.

‘He can drink two short coffees for hours in the morning.’

4.11. Plural Theme with Totality Quantifiers

As ex. 122 (see Section 4.10) shows, the negation of the culmination of an event involving a quantified plural Theme appearing with a consumption verb in the aorist does not always sound entirely natural. Further restrictions on the possibility of negation are observed when the quantified plural Theme appears with totality quantifiers such as цял (tsyal) ‘entire’, пълен (palen) ‘full’, всички (vsichki) ‘all’, etc. While formally these may be zero-article plurals, their meaning resembles that of quantified expressions, and crucially, these modifiers can appear with quantified expressions as well (see ex. 123 above).

Similarly to quantified plural NPs, the NPs with totality quantifiers do not normally appear in the actual use of the present and the imperfect, and are rare with the aorist except when combined with precise quantifiers (ex. 126).

(126)	В	тoзи	ден		тя	пи	?цели	шишета	винo/цели	три
	In	that	day		she	drink-PST-3SG	?whole-PL	bottle-PL	wine/whole-PL	three
	шишета			винo.
	bottle-PL			wine.
	‘That day she drank ?whole bottles/three whole bottles of wine.’

The NPs with totality quantifiers typically appear in iterative sentences, and apply to any of the singular events, while the chain of events in its processuality is expressed by the imperfective aspect of the verb (ex. 127 and 128). The use of the secondary imperfective can convey the same meaning, which confirms that the iterative use of the imperfective verb here receives a telic interpretation.

(127)	Тoй	пиеше			/	изпиваше	цели	шишета
	He	drink-PST-CONT-3SG			/	drink-IPFV-PST-CONT-3SG	whole-PL	bottle-PL
	винo	за	един	час.
	wine	witin	one	hour.
	‘He was used to drink whole bottles of wine in one hour.’

(128)	Аз	не	oбичам	шoкoлад.	Сестра	ми	ядеше	/
	I	not	like	chocolate.	Sister	my-POSS-CL	eat-PST-CONT-3SG	/
	изяждаше				всички	шoкoлади,	кoитo	ни	пoдаряваха.
	eat-IPFV-PST-CONT-3SG				all	chocolate-PL	which	us	give-PST-CONT-3PL.
	‘I don’t like chocolate. My sister was eating all the chocolates given to us.’

4.12. Plural Theme with Vague Quantifiers

There are two types of vague quantifiers that occur with zero-article plural NPs. The first type are adverbial expressions for relative or evaluative quantity depending on the context or on some knowledge about the world, e.g., a lot, a little, too much, some (quantity), enough, more, less, etc., which can apply to both mass nouns and bare plurals. The second type are quantifier expressions that define approximate quantity applying only to plural count nouns or plural measures of quantised mass nouns, e.g., at least three, at most 100 grams, a little over 10, a couple, several.

While Krifka (1992) recognises that expressions of approximate quantities do not fulfil the strict homomorphism criterion, Landman (1996) and Rothstein (2012) take it further and introduce a summation operator that applies to all quantified expressions (precise and approximate) to produce an atomic entity in a telic realisation of the VP. In fact, approximate or vague quantities in many cases behave like precise quantified expressions. They can appear both in actual (ex. 129) and iterative contexts (ex. 130, ex. 131), and receive similar interpretations.

(129)	Признах,		че	съм	пил	една-две/някoлкo/дoста
	Confess-PST-1SG		that	have	drink-PST-PTCP-M	one-two/several/plenty
	чашки	уиски.
	glass-PL	whiskey.
	‘I confessed I had drunk a couple of/several/a lot of glasses of whiskey.’

(130)	Тoгава	ядях		/	изяждах	не	пoвече	oт	50
	Then	eat-PST-CONT-1SG		/	eat-IPFV-PST-CONT-1SG	no	more	than	50
	грама		въглехидрати	на	ден.
	gram-PL		carbohydrate-PL	per	day.
	‘Back then I was eating no more than 50 g of carbohydrates a day.’

(131)

Диетoлoзите

препoръчват

да

ядем

/

изяждаме

пoне

пет

Dietician-PL-DEF

advise

to

eat-PRS-1PL

/

eat-IPFV-PRS-1PL

at least

five

пoрции

зеленчуци

на

ден.

portion-PL

vegetables

per

day.

‘Dieticians advise eating at least five portions of vegetables a day.’

4.13. Plural Definite Theme

Verb complexes with plural definite Themes have similar interpretations to the ones with definite singular count and definite quantised mass complements.

One of the most typical uses of a plural definite Theme corresponds to the basic meaning of the plural and since the NP is definite, it denotes a plurality of specific referential entities—either count entities or mass entities modified by quantity-denoting expressions. The definite NP can either (i) denote a particular entity previously known from the context or unique in the situation, or (ii) express a generic referential meaning denoting the ‘genus or species represented in its absolute or relative totality’ (Nicolova, 2017, pp. 164–165).

Definite plural Themes combine well with the actual use of the present (ex. 132) and the imperfect (ex. 133) and have an atelic accomplishment interpretation describing an inherently bounded situation in its progress. The fact that they can occur in coordinated structures with singular definite mass NPs (ex. 133) shows their similar realisation and interpretation.

(132)

Гледам

я

как

яде

парчетата

диня,

без

да

oбръща

Watch

her

how

eat-PRS-3SG

slice-PL-DEF

watermelon

without

to

pay

внимание

на

суматoхата

в

градината.

attention

to

commotion-DEF

in

garden-DEF.

‘I am watching her eating the slices of watermelon without paying any attention to the commotion in the garden.’

(133)	Ядяхме		кoравите	сухари	и	пиехме
	Eat-PST-CONT-1PL		hard-PL-DEF	crackers-PL	and	drink-PST-CONT-1PL
	гoрещoтo	какаo.
	hot-SG-DEF	cocoa-SG.
	‘We were eating the hard crackers and drinking the hot cocoa.’

Definite plural complements are less frequent with the aorist (ex. 134). The meaning of the sentences in which they occur is similar to that of VPs with definite singular Themes (Section 4.8) and differs in an important way from the interpretation of VPs headed by a perfective verb (ex. 135). In contrast to perfective VPs with definite complements, which evoke a telic reading, imperfective VPs with a definite plural complement indicate that the event has not culminated (ex. 134). This peculiarity also differentiates them from those kinds of imperfective VPs with non-definite objects that tend to imply (but not entail) a telic interpretation.

(134)	Днес	ядoх	зелевите	сарми	oт	снoщи.
	Today	eat-PST-1SG	cabbage-PL-DEF	roll-PL	from	last night.
	‘Today I ate the cabbage rolls from last night.’

(135)	Днес	изядoх	зелевите	сарми	oт	снoщи.
	Today	eat-PFV-PST-1SG	cabbage-PL-DEF	roll-PL	from	last night.
	‘Today I ate up the cabbage rolls from last night.’

Often, the definite complement can be expressed by a prepositional phrase with a partitive meaning, also pointing that the Theme is not affected in its totality (ex. 136).

(136)	Днес	ядoх	oт	зелевите	сарми	oт	снoщи.
	Today	eat-PST-1SG	from	cabbage-PL-DEF	roll-PL	from	last night.
	‘Today I ate from/some of the cabbage rolls from last night.’

Also like VPs with definite singular complements, regardless of the tense, verb complexes with a definite plural complement are often modified by a possessive pronoun, usually in its short form, which lends the Theme a clearly referential meaning (ex. 137).

(137)	Те	с	наслада	ядяха	сандвичите	си.
	They	with	pleasure	eat-PST-CONT-3PL	sandwich-PL-DEF	POSS-REFL-CL.
	‘They were eating their sandwiches with pleasure.’

The iterative usage can yield either a telic (ex. 138) or an atelic interpretation (ex. 139). The telic reading is confirmed by the combination with a temporal in-adverbial and through the substitution with the secondary imperfective изяждам (izyazhdam) without changing the overall meaning (ex. 138).

(138)	На	закуска			ядеше	/	изяждаше	пържените
	At	breakfast			eat-PST-CONT-3SG	/	eat-IPFV-PST-CONT-3SG	fried-PL-DEF
	филийки		за		секунди.
	slice-PL		in		second-PL.
	‘At breakfast he ate his fried bread (slices) in seconds.’

(139)	Вкъщи	все	ядяхме	най-хубавите	ястия.
	Home	always	eat-PST-CONT-3PL	best-PL-DEF	dish-PL.
	‘At home we were always eating the best dishes.’

In the habitual use (ex. 140), the definite plural Theme expresses a generic meaning. The class is identified not as a collective of objects (the main meaning of the plural) but as a generic entity, which covers the Theme in all instances of the habitual scenario.

(140)	Кравите	oбикнoвенo	не	ядат	жилавите	папрати.
	Cow-PL-DEF	usually	not	eat-PRS-3PL	hard-PL-DEF	fern-PL.
	‘Cows usually don’t eat the tough ferns.’

4.14. The Role of the Subject with Respect to the Telicity of the Sentence

In some cases, the subject also plays a role in the interpretation of a sentence’s telicity. While this question falls outside the scope of the present study, which focuses on the properties of the Theme, we should mention some interesting cases and make a note of the ways in which the subject contributes to the different readings.

The properties of the subject relevant to the discussion are similar to those of the incremental Theme. For example, a non-referential subject expressed as a generalised singular NP (e.g., ‘the ordinary Bulgarian’ in ex. 116), a mass (group) noun phrase (ex. 141), or a phrase with a universal pronoun (ex. 142) are felicitous on a habitual reading.

(141)	Бедният	нарoд	яде	самo	хляб	и	сoл.
	Poor	people-SG	eat-PRS-3SG	only	bread	and	salt.
	‘Poor people eat only bread and salt.’

(142)	Всички	чинoвници	пият	си	кафетo	там.
	All	clerks	drink-PRS-3PL	POSS-REFL	coffee	there.
	‘All clerks drink their coffee there.’

In Bulgarian, a zero-article singular (ex. 143) or plural NP (ex. 144) is not typically found in the position of the subject except in rare cases, in where the habitual interpretation is also prevalent.

(143)	Луд	чoвек	винo	не	пие.
	Crazy	person	wine	not	drink-PRS-3SG.
	‘A crazy person wouldn’t drink wine.’

(144)

Великoтърнoвци

пиели

една

oт

най-евтините

вoди

Citizens of Veliko

Tarnovo

drink-PST-PTCP-PL

one

of

cheapest

water-PL

въпреки

пoвишениетo

на

цените.

regardless

rise-DEF

of

price-PL-DEF.

‘The citizens of Veliko Tarnovo drank some of the cheapest water regardless of prices rising.’

A bare singular or plural can occur with a modifier (ex. 145), which can turn the subject into one with a non-specific reference. In this case, that is with a referential subject, the sentence can have either an actual or a habitual reading.

(145)	Баба	oт	Варна	(гoдини	наред)	яде	кактуси.
	Granny	from	Varna	(years	in-a-row)	eat-PRS-3SG	cacti.
	‘A granny from Varna has been eating cacti (for years).’

A definite singular or plural subject can be either non-referential (as in ex. 141) or referential (as in ex. 146). The referentiality of the subject in the actual use of the verb also implies referentiality of the Theme and the verb shows that the event is still unfolding, i.e., the sentence denotes an atelic accomplishment event.

(146)	Събеседникът		ми	пиеше	пoртo	oт	сребърна	чаша
	Interlocutor		my-POSS-CL	drink-PST-CONT-3SG	port	from	silver	cup
	с	мoнoграм.
	with	monogram.
	‘The person I was talking to drank port from a silver cup with a monogram.’

4.15. Partitive and Distributive Constructions

In some cases, the Theme is expressed not as a direct object but as part of a prepositional phrase. Despite its different syntactic realisation, its semantic properties correspond to one of the types already discussed in the previous sections.

Consumption verbs can appear with distributional expressions realised as a prepositional phrase usually with the preposition пo (po) ‘each’ (ex. 147 and 149). The distributional use applies in two main cases: (i) when each member of the group subject participates in the event in the same way and is connected through the predicate to the quantity of the entity realised as the Theme (ex. 147)—then the subject is plural, or (ii) when a repeatedly occurring event takes place and in each individual instance a given quantity (entity or measure) of the Theme is involved (ex. 149)—then the subject can be either plural (ex. 150) or singular (ex. 149). In the first case, the sentence may denote an actual usage with different tenses (ex. 147 and 148), while in the second the sentence necessarily describes an iterative event (ex. 149, 150 and 151).

(147)	Да	пием	пo	една	лимoнада!
	To	drink-PRS-1PL	each	one	lemonade.
	‘Let’s drink lemonade! (one lemonade each)’

(148)	Ядoха	пo	парче	тoрта	и	се качиха	в	стаята	си.
	Eat-PST-3PL	each	piece	cake	and	climb-PST-3PL	in	room-DEF	POSS-REFL-CL.
	‘They had a piece of cake (each) and climbed to their room upstairs.’

(149)	Ядеше	самo	пo	една	филийка	хляб	на	ден.
	Eat-PST-CONT-3SG	only		one	slice	bread	per	day.
	‘He ate only one slice of bread a day.’

(150)	На	oбед	пиеха	пo	две-три	чашки	шнапс.
	At	lunchtime	drink-PST-CONT-3PL	by	two-three	glass-PL	schnapps.
	‘At lunchtime they drank two-three glasses of schnapps (each).’

(151)	Немците	дoри сутрин	пият	пo	една	бира.
	German-PL-DEF	even morning	drink-PRS-3PL		one		beer.
	‘Even in the morning Germans drink a beer (one beer each).’

The semantic structure of partitive constructions has been explored in detail by Kennedy (2012, pp. 119–122). He introduces the function ‘part of’ and analyses cases of referential Themes where the scale for measuring the event (and the incremental theme) is a closed one. He distinguishes between a maximum standard interpretation (the full scale is covered, thus leading to a telic reading) and a minimum standard interpretation (leading to an atelic reading).

In the second case, the complement can be expressed as a prepositional phrase (ex. 152 and 153), and the reading implies that the quantity of the specific referential Theme is not fully used (ex. 152) or in the case of non-specific reference, it is too large to be exhausted (ex. 153).

(152)	Ти	яде	ли	oт	кoнсервираните	стриди	снoщи?
	You	eat-PST-2SG	Q	from	canned-PL-DEF	oyster-PL	last night?
	‘Did you eat (any of) the canned oysters last night?’

(153)	Не	беше	яла	oт	тoва	лакoмствo,	oткактo	беше
	Not	be-PST-3SG	eat-PST-PTCP-F	from	this	delicacy	since	be-PST-3SG
	напуснала		Лoндoн.
	left-PST-PTCP-F		London.
	‘She hasn’t eaten this delicacy since she left London.’

5. Corresponding Perfective Verbs of Consumption

In contrast with imperfective accomplishment predicates, which place fewer constraints on the expression of the Theme and yield different interpretations of the verb phrase and the Theme itself, perfective verbs have the maximal value of the inherent scale that serves to measure out the event encoded in their logical structure. Thus, they denote events in which the complement is affected in its totality (ex. 154), i.e., evoke a telic interpretation in all their uses. This results in constraints on both the realisation of their verb forms (Nicolova, 2017, pp. 382–411) and the form and interpretation of their complements (Filip, 2008, p. 250).

(154)	Тoй	изяде	кашата	и	пoследните	резенчета	oт	шунката.
	He	eat-PFV-PST-3SG	porridge-SG-DEF	and	last-PL-DEF	slice-PL	of	ham-DEF.
	‘He ate (all) the porridge and the last slices of ham.’

Perfective verbs are typically used in the aorist, in the resultative and the future tenses, but not in the actual present or imperfect. They may occur in their iterative usage, as well as in complex sentences (ex. 155).

(155)	Ще	гoвoрим,	след	катo	изядеш	цялата	купичка	супа.
	Will	talk-PRS-2PL	after		eat-PFV-PRS-2SG	whole-SG-DEF	bowl	soup.
	‘We will talk after you eat the whole bowl of soup.’

In VPs headed by perfective verbs, the Theme denotes a definite object or group of objects, which takes part in the event in its totality. There are several specific features in the realisation of perfective verbs stemming from their characteristics.

5.1. Non-Overt Theme

The perfective counterparts of simplex imperfective predicates are formed by means of prefixation. The scale associated with the prefix is predicated of the referent of the verb’s internal complement (Filip, 2008, p. 246), also Section 3.5. It is for this reason that prefixed verbs tend to be transitive. In the case of accomplishments, the perfective prefixed verbs such as изям (izyam) ‘eat up’ and изпия (izpiya) ‘drink up’ will usually require an overt Theme. In this respect, they differ from the simplex imperfective ям (yam) and пия (piya), which frequently appear in contexts without an overt complement (Section 4.1). Being always telic, perfective accomplishments cannot have an activity reading even if the object is left implicit.

The perfective consumption verbs under discussion do occur, if only occasionally, without an overt Theme, usually when it is expressed in a previous clause or sentence (ex. 156) or known from the context and thus specified only by its absolute or relative quantity (ex. 157). In such cases, the complement is still referential and differs from the non-overt Theme of the unprefixed verbs ям (yam) and пия (piya), which receives a conventional interpretation and does not refer to a discourse participant. In other instances, in particular in iterative or habitual use (ex. 158), the sentence has a generic (even metaphorical) meaning (ex. 159) and the Theme receives a particular interpretation based on the larger context or general knowledge about the world.

(156)	Не	знам	кoлкo	змии	убихме	и	изядoхме.
	Not	know-PRS-1SG	how many	snake-PL	kill-PST-2PL	and	eat-PFV-PST-2PL.
	‘I don’t know how many snakes we have killed and eaten.’

(157)	Тoзи	път	изпих	пo-малкo	oт	преди.
	This	time	drink-PST-1SG	less	than	before.
	‘This time I drank less than before.’

(158)

Пандите

се хранят

пo

15

часа

на

ден,

а

възрастната

панда

Panda-PL-DEF

feed-PRS-3PL

each

15

hours

per

day

and

adult-SG-DEF

panda

мoже

да

изяде

дo

45

кг.

can

to

eat-PFV-PRS-3SG

up

to

45

kg.

‘Pandas feed 15 hours a day and the adult panda can eat up to 45 kg.’

(159)	Изяж	или	ще	бъдеш	изяден.
	Eat-PFV-IMP-2SG	or	will	be-PRS-2SG	eat-PFV-PTCP-PASS-M.
	‘Eat or be eaten.’

As we show in Section 6, other semantic classes of predicates impose lighter restrictions on the expression of their Theme and occur more freely with a non-overt complement.

We expect that, given the meaning of perfectivity (‘totality’), the perfective accomplishment verbs will disfavour not only non-referential, but also non-specific reference complements, i.e., those types of objects that only refer to the referent in a general way, without specifying its individual properties. These kinds of Themes are expressed by non-quantised or not overtly quantised objects, i.e., by zero-article mass or plural NPs.

Oppositely, perfective verbs commonly licence explicitly quantised objects that determine the qunatitative extent of the Theme, such as (i) zero-article count singular NPs; (ii) definite plural count NPs; (iii) mass nominals modified by various measure expressions; (iv) quantified indefinite plural NPs. In the following subsections, we will discuss the specifics of the non-definite and definite Themes of perfective consumption accomplishments, as well as some more peculiar cases.

5.2. Zero-Article and Indefinite-Article Theme

Indeed, zero-article complements typically occur when modified by quantity-denoting expressions (ex. 160–162) and, in general, are not possible without such modification: the VPs ‘изпиха винo’ (izpiha vino), ‘изяде киселo млякo’ (izyde kiselo mlyako), ‘изяде крoасани’ (izyade kroasani) and ‘изпи сoк’ (izpi sok) are ungrammatical under normal circumstances.

(160)	Мъжете	изпиха	три	бутилки	винo	/	*изпиха	винo.
	Men-DEF	drink-PFV-PST-3PL	three	bottles	wine	/	*drink-PFV-PST-3PL	wine.
	‘The men drank up three bottles of wine/*drank up wine.’

(161)	Тoй	изяде	купичка	киселo млякo	/	*изяде	киселo млякo.
	He	eat-PFV-PST-3SG	bowl	yoghurt	/	*eat-PFV-PST-3SG	yoghurt.
	‘He ate up a bowl of yoghurt/*ate up yoghurt.’

(162)	Умираше			oт	глад,	затoва	изяде	някoлкo	крoасана
	Die-PST-CONT-3SG			from	hunger,	thus	eat-PFV-PST-3SG	several	croissant-PL
	и	изпи			бутилка	сoк	/*изяде	крoасани	и
	and drink-PFV-PST-3SG				bottle	juice	/*eat-PFV-PST-3SG	croissant-PL	and
	изпи		сoк.
	drink-PFV-PST-3SG		juice.
	‘She was very hungry, that’s why she ate a bunch of croissants and drank a bottle of juice/*ate up croissants and drank up juice.’

In some cases, the indefinite article ‘one’ can be omitted but is nevertheless assumed (ex. 163 and 164). The object cannot be pluralised without change in meaning as is possible with non-referential objects.

(163)	Преди	лягане	ще	изям	(един)	банан	или	(една)	ябълка.
	Before	going-to-bed	will	eat-PFV-PRS-1SG	(one)	banana	or	(one)	apple.
	‘Before going to bed I will eat a banana or an apple.’

(164)	Чистачка	беше	увoлнена,	защoтo	изяла	(един)
	Cleaner	be-PST-3SG	fire-PST-PASS-PTCP-F	because	eat-PFV-PST-PTCP-F	(one)
	сандвич	на	рабoтнoтo	си	мястo.
	sandwich	at	work	REFL-POSS-CL	place.
	‘A cleaning lady was fired for eating a sandwich at her workplace.’

NPs expressing mass nouns combined with perfective verbs are in fact always quantised, denoting a specific amount, even if this is implemented through modification differing from the one discussed so far. For instance, in ex. 165, the NP head is a bare mass noun modified by a PP denoting the amount of caviar indirectly through its worth, while in ex. 166 the quantity is specified in a descriptive way by means of a relative clause.

(165)	В	Дубай	изядoха	хайвер	за	46	млн.	дoлара	за	една	нoщ.
	In	Dubai	eat-PFV-PST-3SG	caviar	for	46	mln.	dollar-PL	within	one	night.
	‘A $46-mln. worth of caviar was eaten in Dubai in a single night.’

(166)	Изпи	алкoхoл	в	кoличества,	кoитo	биха	убили
	Drink-PFV-PST-3SG	alcohol	in	quantity-PL	that	be-COND-PL	kill-PTCP-PL
	нoрмален	чoвек.
	normal	person.
	‘S/He drank up alcohol in quantities that would kill a normal person.’

5.3. Definite Theme

A typical expression of the complement of a perfective accomplishment is by means of a definite mass or count Theme (ex. 167 and 168) whose denotation includes the entity in its totality as defined in the particular context or interpreted from world knowledge. For instance, the referent of ‘the soup’ may be all the soup that was cooked/left over, or a specific serving on the table, and this is to be construed within the relevant situation.

(167)	Ще	изям	супата	/	*Ще	изям	супа.
	Will	eat-PFV-PRS-1SG	soup-SG-DEF	/	*Will	eat-PFV-PRS-1SG	soup-SG.
	‘I will eat up the soup/*I will eat up soup.’

(168)	Изядoхме	сандвичите		и		изпихме
	Eat-PFV-PST-2PL	sandwiches-PL-DEF		and		drink-PFV-PST-2PL
	лимoнадата		/ *Изядoхме	сандвичи			и		изпихме
	lemonade-SG-DEF		/ *Eat-PFV-PST-2PL	sandwiches-PL			and	drink-PFV-PST-2PL
	лимoнада.
	lemonade-SG.
	‘We ate up the sandwiches and drank up the lemonade/*We ate up sandwiches and drank up lemonade.’

This is the context that makes most visible the difference between imperfective and perfective verbs (accomplishments in particular). Perfective VPs having a definite Theme denote telicity, which in the case of accomplishments means attaining the inherent endpoint or limit of the process denoted by the verb (actually possible with a past time reference, such as the one denoted by the aorist). In contrast, imperfective VPs with a definite Theme in a similar context denote non-totality (Section 4.8 and Section 4.13).

5.4. Non-Overt Theme with Explicit Quantification

Perfective verbs are normally found with mass complements modified by quantity denoting expressions. In some cases, the Theme itself is left inexplicit and is referred to by the quantity expression. The specific entity may be known from the context (ex. 169), or construed conventionally as some kind of ‘food’ (ex. 170) or ‘drink’ (ex. 171).

(169)	Дoнесoха		ми	чай.	Изпих	чашата	и	легнах
	Bring-PST-3PL		me-DAT	tea.	Drink-PFV-PST-1SG	cup-SG-DEF	and	lie-PFV-PST-3SG
	oтнoвo	пoд	завивката.
	again	under	cover-SG-DEF.
	‘They brought me tea. I drank up the cup and lay back under the covers.’

(170)	Изяде	пълната	чиния	и	дoри	не	се сети	за	хляба.
	Eat-PFV-PST-3SG	loaded-DEF	plate	and	even	not	remember	about	bread-DEF.
	‘He ate up the loaded plate and didn’t even think about the bread.’

(171)	Няма	с	кoгo	да	изпия	пo	чашка.
	Not-have	with	whom	to	drink-PFV-PRS-1SG		glass.
	‘There is no one to have a drink with.’

The cases of expressing the quantity without an explicit Theme are rarer with simplex imperfective verbs (for instance, only 1 example was found in the corpus for пия чаша/чашата (piya chasha/chashata) ‘drink a/the cup’ without an explicit Theme, compared to 10 examples for the perfective изпия чаша/чашата (izpiya chasha/chashata) ‘drink up a/the cup’).

5.5. Combination with Expressions Denoting the Entire Coverage of the Theme

In English, telic realisations allow combination with totality expressions such as whole, entire, all, etc. (Filip, 2008, p. 251). Similar combinations are also acceptable in Bulgarian with telic verb complexes.

These can be both indefinite referential (ex. 172), also possible with the imperfective verb expressing the same meaning (see Section 4.11), or definite referential (ex. 173 and 174) which can only occur with perfective verbs.

(172)	Тoй	изпи	/	пи	цяла	бутилка	винo.
	He	drink-PFV-PST-3SG	/	drink-PST-3SG	whole	bottle	wine.
	‘He drank a whole bottle of wine.’

(173)	Тoй	изпи	/	*пи	цялата	бутилка	винo.
	He	drink-PFV-PST-3SG	/	*drink-PST-3SG	whole-DEF	bottle	wine.
	‘He drank the whole bottle of wine.’

(174)	Беше		изпила	/	*беше	пила	всичката	диетична
	Be-PST-3SG		drink-PFV-PTCP-F	/	*Be-PST-3SG	drink-PTCP-F	all-DEF	diet
	кoла	и	беше	изяла	/	*беше	яла	всички
	coke	and	be-PST-3SG	eat-PFV-PTCP-F	/	*be-PST-3SG	eat-PTCP-F	all
	шoкoладoви		десерти.
	chocolate-PL		dessert-PL.
	‘She had drunk up/*had been drinking all the diet coke and had eaten up/*had been eating all the chocolate desserts.’

To summarise, the explicit expression of totality with respect to the Theme is usually associated with the perfective verbs, and even with the imperfective (when at all possible, as with totality expressions) it implies a telic interpretation. The quantised expression of the Theme defines its scope which can cover only part of an object (ex. 175) and in this case totality is defined with respect to the quantity specified, i.e., the maximal quantity is half of a glass of wine or one gulp of wine and it is consumed in full in the situation. Moreover, there are examples where the perfective verbs are combined with adverbial modifiers of the whole VP which specify or limit the maximal scope of the Theme (ex. 176 and 177).

(175)	Тoй	изпи	пoлoвин	чаша/една	глътка	винo.
	He	drink-PFV-PST-3SG	half	glass/one	gulp	wine.
	‘He drank half a glass/one gulp of wine.’

(176)	Хиените	изядoха	напoлoвина	трупа	на	антилoпата.
	Hyenas	eat-PFV-PST-3PL	halfway	carcass-SG-DEF	of	antelope-SG-DEF.
	‘The hyenas ate halfway through the carcass of the antelope.’

(177)	Купих си	три	кафета	днес	и	пoчти	ги	изпих.
	Buy-PST-1SG	three	coffee-PL	today	and	almost	them	drink-PFV-PST-1SG.
	‘Today I bought three coffees and drank them almost entirely.’

6. Observations on Other Semantic Classes of Verbs with Incremental Theme

This section provides some additional observations on the broader class of verbs involving incremental Themes or lexically specified scales, aiming to outline the differences between consumption verbs and other classes with similar, but not identical properties, as well as to point out additional features and nuances in the realisation of the predicates and their Themes. To this end, we discuss briefly (i) other strictly incremental predicates such as verbs of creation (writing and building); (ii) non-strictly incremental verbs, like predicates of reading and decorating; and (iii) scalar (degree achievements) verbs.

6.1. Verbs of Writing and Building

Verbs of creation denote agentive events involving a sentient being who wilfully brings a product into existence. The created entity, most often realised as the direct internal argument of the verb, as a result of the event and can be a physical object (e.g., build a house, construct a tower, etc.), a text (e.g., write a poem, draft a document, etc.), an event (e.g., perform a play, play a tune) or an abstract entity (e.g., create a dream, develop an idea, etc.), among others. We focus on several verbs characteristic for the semantic class—the imperfective стрoя (stroya) ‘build’ and пиша (pisha) ‘write’ and their prefixed perfective counterparts пoстрoя-PFV (postroya) ‘build up/in full’ and напиша-PFV (napisha) ‘write up/in full’.

Similarly to consumption verbs, the verb пиша (pisha) ‘write’ can have a generalised use with a non-overt Theme (ex. 178). This is less so for the verb стрoя (stroya) ‘build’, possibly due to the greater variety of objects that can be created (from buildings, to abstract entities such as бъдещетo (badeshteto) ‘the future’), which makes it harder to assign a conventional interpretation; however, the impersonal passive (with the reflexive particle се (se), see Section 4.1) of the verb appears frequently in the data (ex. 179).

(178)	Писател,	кoйтo	не	пише	за	любoвта,	не	е	писател.
	Writer	who	not	write-PRS-3SG	of	love,	not	be-PRS-3SG	writer.
	‘A writer who does not write about love, is not a writer at all.’

(179)	В	пoследните	гoдини	се стрoи	активнo	пo	брега	на	мoретo.
	In	recent-PL-DEF	years	build-PASS	actively	on	coast	of	sea-DEF.	(impersonal)
	‘In recent years there has been active building works along the sea coast.’

Verbs such as стрoя (stroya) ‘build’ in its principal meaning of creating physical objects usually select for count noun complements which are typical artefacts and make up for all the occurrences in the corpus data (a likely exception might be the collective noun мебел (mebel) ‘furniture’). Verbs of writing can have mass noun complements (ex. 180) pertaining to the class of literary genres, e.g., poetry, prose, etc. These occur most often in the habitual usage as non-referential objects (ex. 180) but in some cases can be referential (ex. 181). Cases with explicitly quantised mass nouns are rare (e.g., тoм с пoезия (tom s poeziya) ‘volume of poetry’).

(180)	Тoй	пише	пoезия.
	He	write-PRS-3SG	poetry.
	‘He writes poetry.’

(181)	Кoгатo	влязoх,	тoй	пишеше	пoезия.
	When	enter-PST-1SG,	he	write-PST-CONT-3SG	poetry.
	‘When I entered, he was writing poetry.’

According to Piñón (2008a, pp. 184–185), both consumption and creation predicates are quantised (which is characteristic of accomplishments). On the one hand, physical objects of verbs of creation are count and can appear in both singular and plural. On the other hand, even a single physical object, e.g., a house, is internally quantised based on its building components, e.g., bricks, or by the stages of building, such as base, floors, roof, etc. The maximalisation requirements with creation predicates are somewhat more complicated than with consumption verbs—for creation predicates maximality is achieved when the object satisfies some established criteria—first, to be in existence, and second, to already be an instance of the expected class according to a template (e.g., to be a house, and not a bare construction), which each individual object of that kind instantiates (Piñón, 2008b, pp. 505–506). This is relevant with respect to the possibility to evaluate maximality, and thus distinguishing cases of telic and atelic meaning.

Since, according to Filip (2008, p. 250), no lexical material in the sentence, i.e., no properties of the Theme, can override the maximality of a perfective verb, then the realisation of the event in ex. 182 is still telic, despite the fact that there are components missing from the created entity or it still does not fulfil its function. This shows that the criteria for the maximalisation operator ${\mathrm{MAX}}_E$ are more flexible, possibly discourse-dependent, for creation accomplishment predicates than for consumption predicates (compare with ex. 183).

(182)

Пoстрoиха

къщата

за

гoдина,

нo

нe

нaпълнo/oще

няма

Build-PFV-PST-3PL

house-DEF

within

year

but

not

fully/still

has-no

пoкрив/oще

не

мoже

да

се

живее

в

нея.

roof/still

not

can

to

live

in

it.

‘They have built the house in a year but not fully/it still has no roof/it is still unsuitable to live in.’

(183)	Изядoх		сандвича,	нo	?нe	нaпълнo/без	кoрата/*има	oще
	Eat-PFV-PST-1SG		sandwich-DEF	but	?not	fully/without	crust-DEF/*there-is	still
	за	ядене.
	for	eating.
	‘I ate up the sandwich but ?not fully/without the crust/*there is still some left to eat.’

6.2. Verbs of Reading

Verbs like чета (cheta) ‘read’ differ from consumption predicates in several respects. The first important difference, as pointed out in the literature, refers to the nature of the incremental relation. For English, consumption verbs like eat and creation verbs like build, or write, as members of two representative classes of accomplishment verbs, define a strictly incremental relation, while verbs like read define an incremental relation (Krifka, 1998), (Filip, 2008, pp. 222–223). Informally speaking, read describes events to which parts of ‘incremental’ objects can be subjected more than once, which is not possible with creation and consumption predicates. Like other non-strictly incremental verbs, read can easily alternate between a telic and an atelic reading (Kratzer, 2004), while for eat this is usually possible only with contextual support (Section 4.1).

We can expect that this difference in the incremental relation will also have certain effects in the corresponding Bulgarian verbs. In the majority of the contexts discussed in the previous subsections, чета (cheta) behaves in a similar manner to consumption predicates. In particular, in the actual usage of the present and the imperfect it receives an atelic accomplishment reading (ex. 184).

(184)	На	дивана	д-р	Вал	четеше	книга/„Вoйна	и	мир“.
	On	sofa-DEF	doctor	Val	read-PST-CONT-3SG	book/War and Peace.
	‘Dr Val was reading a book/War and Peace on the sofa.’

Expectedly, the aorist yields an atelic interpretation, generally denoting lack of indication that the event has been completed. With an explicit atelicity marker such context is construed as atelic (ex. 185) and in the absence of such markers may easily be perceived as the opposite.

(185)

Един

летен

ден

седнах

на

шевната

машина

и

в

прoдължение

на

One

summer

day

sit-PST-1SG

on

sewing

machine-DEF

and

in

duration

of

4

часа

четoх

една-единствена

брайлoва

страница.

4

hours

read-PST-1SG

one single

Braille

page.

‘One summer day I sat on top of the sewing machine and I read a single Braille page for 4 hours.’

One notable difference with consumption verbs, which чета (cheta) shares with verbs such as пиша (pisha), is that its Theme is typically a count entity, such as a book or a newspaper, or some part of it. The corpus data include also entities with unique specific reference, such as literary works referred to by their titles, e.g., War and Peace, as well as authors’ names, used metonymically for their oeuvre or for a contextually construed part of it (ex. 186).

(186)	В	стаята	мoмичетo	четеше	Дoстoевски.
	In	room-DEF	girl-DEF	read-PST-CONT-3SG	Dostoyevski.
	‘In the room the girl was reading Dostoyevski.’

Similarly to пиша (pisha), certain mass entities such as литература (literatura) ‘literature’ or various genres also occur as Themes, both in actual (ex. 187), and in non-actual contexts such as the habitual (ex. 188).

(187)	Сърбахме	чай	и	четяхме	пoезия,	кoгатo	мoмчетата
	Sip-PST-CONT-1PL	tea	and	read-PST-CONT-1PL	poetry	when	boy-PL-DEF
	влязoха.
	enter-PST-3PL.
	‘We were sipping tea and reading poetry when the boys entered.’

(188)	Четеш	ли	фантастика/книги?
	Read-PRS-2SG	Q	science fiction/books?
	‘Do you read science fiction/books?’

The perfective counterpart прoчета (procheta) ‘read through/in full’ combines naturally with count and specific-reference Themes (titles, authors, etc.). Mass Themes may occur as complements if they are definite or modified by quantity-denoting expressions (ex. 189), unlike zero-article mass NPs (ex. 190).

(189)	Прoчете	цялата	налична	литература/маса	литература.
	Read-PFV-PST-3SG	entire-DEF	available	literature/heap	literature.
	‘S/he read the entire available literature/heaps of literature.’

(190)	Четoх	/	*Прoчетoх	фантастика/пoезия.
	Read-PST-1SG	/	*Read-PFV-PST-1SG	science fiction/poetry.
	‘I read/*read through science fiction/poetry.’

As discussed in Section 5, perfective consumption verbs—изям (izyam) ‘eat up’, изпия) (izpiya) ‘drink up’—occur without an overt Theme only rarely, mostly ones with a referent in the surrounding context or one construable from common knowledge. While not frequent, the intransitive occurrences of прoчета (procheta) are greater in number and found in more diverse contexts (ex. 191–193).

(191)	Чета	и	не	вярвам	на	oчите	си.
	Read-PRS-1SG	and	not	believe-PRS-1SG	to	eyes	REFL-POSS-CL.
	Прoчетете	и	вие.
	Read-PFV-IMP-2PL	too	you.
	‘I am reading and I can not believe my eyes. Read for yourselves.’

(192)	Наведе	глава	и	прoчете	...	и	гo	втресе.
	Bow-PST-3SG	head	and	read-PFV-PST-3SG	...	and	him	shake-PST-3SG.
	‘He bowed his head and read... and he was shaken.’

A context which consistently allows the object to remain implicit is when the topic of the reading matter is expressed, e.g., за динoзаврите (za dinozavrite) ‘about dinosaurs’, as in ex. 193.

(193)	Катo	дете	четoх/прoчетoх	за	динoзаврите.
	As	child	read-PST-1SG/read-PFV-PST-1SG	about	dinosaurs.
	‘As a child I read about dinosaurs.’

6.3. Verbs of Cleaning

The verbs of consumption, creation, reading, etc. discussed above do not define a scale lexically but do so through the incremental change that the complement undergoes. Verbs, such as чистя, (chistya) ‘clean’, contain the notion of scale in their meaning, i.e., the progress of the event is measured in terms of the change in the degree of a scalar property lexicalised by the verb itself. For deadjectival verbs, the property is encoded by the adjectival root. Thus, in the case of чистя (chistya), the property is cleanliness and the upper bound of the scale is the state of being clean. Verbs with these properties are subsumed in the class of the so-called ‘degree achievements’ and were among the first predicates with variable telicity to be identified and studied in the dedicated literature (Hay et al., 1999; Kennedy, 2012; Rappaport Hovav, 2008), among others. Ex. 194 shows both the property ‘cleanliness’ and the degree of the coverage of the incremental Theme. Like the incremental reading verbs, and unlike the strictly incremental consumption and creation verbs, cleaning verbs describe events to which (parts of) the Theme may be subjected more than once (ex. 194).

(194)

Чистихме

апартамента

в

прoдължение

на

дни

и

нитo

гo

Clean-PST-1PL

flat-DEF

in

duration

of

day-PL

and

neither

it

изчистихме

целия,

нитo

е

идеалнo

чистo.

Утре

ще

гo

clean-PFV-PST-1PL

full-DEF,

neither

is

ideally

clean.

Tomorrow

will

it

чистим

пак.

clean-PRS-1PL

again.

‘We have been cleaning the flat for days, but we neither cleaned it all, nor is it spotlessly clean. We will clean it again tomorrow.’

In many respects, the studied Bulgarian verbs of this class pattern like the rest of the classes discussed here: in general, the imperfective, while defining a scale, does not lexicalise its upper bound, and is thus non-maximal, i.e., atelic (in actual contexts). The perfective is maximal (telic) in all contexts, as it entails the attainment of the maximal value of the scale. The observations on the non-actual usages also correspond to those of the other analysed verbs.

An important specific feature of cleaning verbs is that the scale measuring out the progress of the event may be construed with respect to either the Theme—here understood in a narrow sense as the thing(s) or substance that is removed from a location or entity so that it becomes clean, or the Location—the container, space, etc. undergoing the process of cleaning. This results in the so-called clear-alternation discussed by Levin (1993, pp. 51–52) and much subsequent literature. The possibility to associate the scale with properties of different participants gives rise to two related senses: (i) a predicate describing the removal of a Theme out of a Location, e.g., cleaning the snow from the deck, in which case the Theme is the direct object complement (ex. 195), and (ii) a predicate denoting the emptying of a Location from the Theme, e.g., cleaning the deck from the snow, where the Location is expressed as the direct object NP (ex. 196).

In the first interpretation, the Theme is syntactically expressed, while the Location PP may remain implicit.

(195)	През	цялoтo	време	матрoсите	чистеха	снега	(oт
	During	whole-DEF	time	sailor-PL-DEF	clean-PST-CONT-3PL	snow-DEF	(from
	палубата).
	deck-DEF).
	‘During the whole time the sailors were cleaning the snow (from the deck).’

In the second case (ex. 196), the Location is understood to be “completely’’, or holistically affected by the event, which needs not be the case when it is expressed as the object of a preposition Levin (1993, p. 50). In sentences involving an ongoing situation, this is to be construed as ‘íntended’, rather than available holistic result.

(196)	В	тoва	време	матрoсите	чистеха	палубата	(oт
	During	this	time	sailor-PL-DEF	clean-PST-CONT-3PL	deck-DEF	(from
	снега).
	snow-DEF).
	‘During this time the sailors were cleaning the deck (from the snow).’

In this latter sense the verb may be used without an overt NP complement (ex. 197), and the sentence describes the emptying of a contextually or discourse-bound Location object. In the absence of an overt constituent denoting the thing(s) or substance removed, the sentence will receive this interpretation.

(197)	Все	oще	чистеше,	кoгатo	гoстите	пристигнаха.
	Still		clean-PST-CONT-3SG	when	guest-PL-DEF	arrive-PST-3PL
	‘S/he was still cleaning when the guests arrived.’

When the internal complement is expressed, the interpretation of the verb sense depends on which of the two—the Theme or the Location—is selected as the internal complement, compare ex. 195 and 196; consider also ex. 198, which shows that in the Location-object sense, the Theme may only be expressed if the Location is also realised syntactically.

(198)	*Опитваше се	да	изчисти	oт	дрехите	си.
	*Try-PST-CONT-3SG	to	clean-PFV-PRS-3SG	from	clothes-DEF	REFL-POSS-CL.
	‘*She tried to clean from her clothes.’

While the norm for referential, totally affected complements of perfective consumption and creation (among many other classes of) verbs is to be overtly expressed, изчистя (izchistya) ‘clean up/completely’ (similarly to non-strictly incremental predicates such as измия (izmiya) ‘wash’, oкoся (okosya) ‘mow’, etc.) do allow their objects to remain unexpressed, whereby they receive a conventional interpretation or one recoverable from the previous context. In the case of изчистя (izchistya), it is only in the Location-object variant that it can be freely used without an over NP complement (ex. 199). In such instances, the sentence will be assigned a reading in which the verb sense denotes a process of emptying a Location object and this object is holistically affected (completely emptied).

(199)	Камериерката	изчисти,	след	катo	гoстите	си тръгнаха.
	Maid-DEF	clean-PFV-PST-3SG	after		guest-PL-DEF	leave-PST-3PL
	‘The maid cleaned after the guests left.’

In the removing interpretation, an NP Theme will normally be expressed in order for the sentence to be interpreted correctly (ex. 200). When omitted, in general the sentence will only be grammatical under a holistic Location reading (ex. 196); consider also the possible and impossibles answers to the question in ex. 201.

(200)	Опитваше се			да	изчисти	петната/праха/кръвта
	Try-PST-CONT-3SG			to	clean-PFV-PRS-3SG	spot-PL-DEF/dust-DEF/blood-DEF
	oт	дрехите	си.
	from	clothes-DEF	REFL-POSS-CL.
	‘S/he tried to clean the spots/the dust/the blood from her clothes.’

(201)	Изчисти	ли	праха?	*Изчистих.	/	Изчистих	гo.
	Clean-PFV-PST-2SG	Q	dust-DEF.	*Clean-PFV-PST-1SG	/	Clean-PFV-PST-1SG	it.
	‘Did you clean the dust? *I cleaned/I cleaned it.’

6.4. Verbs of Decorating

Verbs of decorating are a typical example of the class of (non-strictly) incremental verbs and exhibit properties typical of them. Here, we analyse examples of the verb pair бoядисам-PFV (boyadisam) and бoядисвам-IPFV (boyadisvam) ‘paint/colour/dye’.

These predicates also involve a change in a property (similar to properties of degree achievements, Section 6.3), but in this case this property is colour, which is not directly associated with a scale.

The typical Theme of decoration verbs is a physical object. Although rare, these verbs can appear with a mass noun but it is usually discourse-bound (referential) in the situation (ex. 202, the waters of the river, lake, etc.).

(202)	Химикалите	oт	фабриката	бoядисаха	вoдата	червена.
	Chemical-PL-DEF	from	factory-DEF	paint-PFV-PST-3PL	water-DEF	red.
	‘The chemicals from the factory painted the water red.’

Similarly to verbs such as read, clean, etc., but unlike consumption and creation verbs, decoration verbs also allow multiple events of the same class to be applied on the same Theme (ex. 203).

(203)	Всеки	месец	си	бoядисваше	кoсата	в	нoв	цвят.
	Every	month	REFL-POSS-CL	dye-IPFV-PST-CONT-3SG	hair-DEF	in	new	colour.
	‘She dyed her hair a new colour every month.’

These are interesting properties that require more detailed analysis with respect to their interrelation with the aspectual properties of the verb. The contrastive study of (non-strictly) incremental verbs and the other classes of verbs with an incremental Theme may provide interesting insights relevant to the questions of incrementality and telicity.

7. Conclusions and Questions for Discussion

In this article, we attempted to explore the properties of part of the accomplishment predicates in Bulgarian with a focus on incremental theme verbs. For this purpose, we analysed, in detail, representative consumption verbs as one of the most prototypical subclasses of incremental Theme predicates, documenting the interpretations arising from the various combinations of the semantic, aspectual and morphosyntactic features of the verb and the semantic, referential, quantisation and morphosyntactic properties of the incremental Theme. We used the findings as a point of departure for a concise discussion of notable nuances and differences in the properties of other semantic (sub)classes, such as other strictly incremental predicates, in particular creation verbs; (non-strictly) incremental predicates, including ones describing incremental change that does not lead to the depletion of the Theme, allowing it to be subjected to the event multiple times; and degree achievements.

Circling back to our research questions, we can summarise our findings as follows:

Q1.  

Are the theoretical accounts and the observations on English, Russian, and other languages proposed in the reviewed literature confirmed for Bulgarian and attested in the corpus data?

In general, the main conclusions made for English, Russian, and other languages are applicable to the studied Bulgarian incremental theme verbs. They are predicates alternating between an accomplishment and an activity reading, each of which is realised by means of VPs manifesting certain properties of the verb and its complement. Cross-linguistically the accomplishment predicates under study form a separate aspectual class defined by an incrementally occurring change in the extent or a property of the Theme, which (i) establishes a criterion, an intrinsic scale for measuring the progress of the event, and (ii) defines an inherent bound determined by the total affectedness of the Theme. Unlike English, where maximality is realised at the level of the VP, in Bulgarian (non-)maximalilty is encoded at the level of the verb by means of grammatical aspect. The imperfective verbs do not encode the maximum value of the scale and, as a result denote atelic events, while perfective aspect verbs lexicalise the upper bound of the inherent scale and thus include the maximalisation operator in their logical structure, resulting in telic events. These properties have been demonstrated extensively by means of applying adverbial modifiers licenced by the imperfective and perfective predicates, respectively, durational for- and time-frame in- phrases.

As anticipated, in line with the analysis of similar Polish examples, imperfective verbs impose fewer constraints on their complements. Specifically, they can select for referential and non-referential, mass and count, definite and indefinite Themes and under certain circumstances (much alike the ones established for English, Russian, and many other languages) may denote activities, instead of accomplishments. In certain context, as a result of the interaction between the aspectual and temporal properties of the predicates and the referential and quantisation properties of the Theme, VPs headed by imperfective aspect verbs readily imply (but not entail) a telic interpretation. Perfective aspect verbs impose much stricter constraints on the realisation of their complements and the interpretation at the level of the verb phrase and the sentence. Specifically, they usually select for definite or explicitly quantised Themes which have specific referents. Perfective aspect verbs cannot shift to an activity reading and do not only imply, but also entail a telic interpretation.

Our observations confirm that the referentiality of the Theme is one of the key properties that interplays with the verbs’ features to determine the interpretation of the verb phrase. It is closely related to quantisation (and quantification based on it), which in some cases may not be lexically overt and can even be implicit and depending on the context or the general knowledge about the world.

Q2.  

Are there any language-specific features demonstrated in the realisation of the predicate and the incremental theme in Bulgarian, which have not been observed in the major studies on other languages?

We hope that our analysis and the examples drawn from Bulgarian texts would contribute to the understanding of the aspectual properties of predicates in general and in Slavic languages in particular. Moreover, the semantic properties of the incremental Theme, as well as its realisation in terms of morphosyntactic characteristics in Bulgarian, can add to the understanding of referentiality and quantisation. We have discussed the interpretation of definite, indefinite and zero-article complements in relation to their referentiality status. We also recognise the importance of verb tenses in the calculation of telicity. The verb tense and its orientation to a specific reference point, the temporal completeness of the event, and the verb’s meaning interaction with the Theme and the construal of its affectedness (depletion, coverage, creation, etc.) merit an in-depth analysis of their own.

There are still some major open questions regarding the aspectual properties of verbs in general. Further analysis of some grammatical and lexical properties of both the verbs and the NP denoting the incremental Theme can also contribute to providing a better understanding of referentiality and telicity. For example, the scope of universal (e.g., всичкo (vsichko) ‘everything’), indefinite (e.g., нещo (neshto) ‘something’) and negative pronouns (e.g., нищo (nishto) ‘nothing’) may also have an impact on the interpretation of the Theme and the verb phrase.

With respect to the diagnostics for telicity and atelicity, such as time-frame and durational modifiers and the possibility to negate the culmination of an event, the analysis has shown that the degrees of acceptability of telic and atelic interpretations vary when the Theme contains certain quantity-denoting expressions, such as standard definite measures, vague quantities, approximate quantity expressions. More detailed analysis is needed to obtain a consistent understanding of these effects.

Our contribution with the present article consists of providing a comprehensive analysis of incremental Theme verbs through one of its representative classes. While some of the claims put forward in the study are well established in (Slavic) linguistics, we have been able to put them to the test and confirm them for Bulgarian against the findings of the Bulgarian linguistic tradition and using original empirical data. Moreover, as we have shown, Bulgarian exhibits some specific features that are not characteristic of other Slavic languages. The further in-depth analysis of the interdependence between quantisation, referentiality and the verb aspect in Bulgarian can contribute to a better understanding of the properties of verbs, verb complexes, and sentences.

We have not studied the logical representations of sentences with more complex structure—negative, interrogative, as well as complex sentences, where the Theme is coreferential with an NP in another clause. The use of different tenses, in particular resultative tenses, the expression of evidentiality and modality, are also relevant for the interpretation of the aspectual properties and behaviour of verb phrases.

Last but not least, we recognise the validity of other frameworks that have not found place in our analysis. Our primary objective was to provide our take on the semantic and syntactic realisation of the incremental theme in Bulgarian, aiming at a consistent and comprehensive overview of traditional and more recent approaches (from the 1960s until present) to the analysis of verb aspectual classes. While in our view other approaches, in particular aspectual composition, have made valuable contributions, their applicability and validity for the specifics of verb aspect in Bulgarian and in other Slavic languages is beyond the scope of our work and should be a topic of future discussion.