Dva stakana moloka: Substances and Containers in Genitive of Measure Сonstructions in Russian * Barbara H. Partee and Vladimir Borschev - PDF

Description
To appear in Russkij jazyk v naučnom osveščenii accepted July 2012 Dva stakana moloka: Substances and Containers in Genitive of Measure Сonstructions in Russian * Barbara H. Partee and Vladimir Borschev

Please download to get full document.

View again

of 16
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Information
Category:

Entertainment & Humor

Publish on:

Views: 17 | Pages: 16

Extension: PDF | Download: 0

Share
Transcript
To appear in Russkij jazyk v naučnom osveščenii accepted July 2012 Dva stakana moloka: Substances and Containers in Genitive of Measure Сonstructions in Russian * Barbara H. Partee and Vladimir Borschev 1. Container nouns and the Russian genitive of measure : Introduction and examples Our topic is the Russian genitive of measure construction(s), studied in [Апресян ; Борщев, Парти 1999, 2011; Borschev, Partee 2001, 2004; Partee, Borschev 2012], and [Rakhilina 2004]. These are a family of closely related constructions all of which can be instantiated by a simple phrase like stakan moloka glass of milk, with a container noun like stakan glass followed by a genitive NP like moloka milk. The basic meaning of a container noun like stakan glass is sortal, but in these constructions it is sometimes sortal, sometimes relational, and sometimes functional. And when we compare canonical genitive constructions with relational nouns, like (1), the genitive with obligatory third term construction in (2), and the constructions studied here, in (3), we find three compositionally different organizations of the semantic parts. (1) брат Пети (2) человек среднего роста In (1), the relational noun is the head and the genitive phrase is its argument. In (2), the obligatory third term construction, the complex genitive NP forms a modifier. And in the constructions with container nouns, we will see that a container noun may be a sortal head, it may be a relational head, and may shift to a functional reading which forms the core of a measure-phrase modifier of the NP in the genitive phrase. These readings arise as the container noun goes through a progression of semantic shifts from a concrete sortal noun through several steps until it becomes an ad hoc unit of measure, and for some nouns, like Russian stakan glass and English cup, it may undergo a further shift to become a lexicalized standard unit of measure. The focus of our previous work has been these shifts in the container noun and in the semantics of the construction; in this article we pay additional attention to the bare noun complement moloka milk-gen (section 5). * We are grateful for discussions of parts of this work with Elena Paducheva, Ekaterina Rakhilina, Louise McNally, Susan Rothstein, Per Anker Jensen and Carl Vikner, and to audiences and participants in the Mathesius Lectures in Prague in 2002, the Possessives Workshop at UMass Amherst in 2002, an Ontology Workshop in Copenhagen, 2006, the Conference on Concept Types and Frames in Language, Cognition, and Science at the Heinrich Heine Universität of Düsseldorf in 2007, and in 2010 at the Ècole Normale Superieure in Paris, at the Conference on Semantics and Formal Modelling JSM10 in Nancy, and in the Workshop on Bare Noun Phrases at Bar Ilan University. Thanks to Georgy Moroz for finding some of the Russian examples in Section 5, and to two anonymous referees for valuable comments and suggestions. We especially thank one of the referees, whom the editors now have kindly identified to us as Maria Kholodilova, for many detailed suggestions and comments, and we have responded to and/or incorporated as many as we could in the time available. All remaining inadequacies are our own. This work was supported by the NSF under grants to Partee and Borschev: BCS Integration of Lexical & Compositional Semantics: Genitives in English and Russian, , and BCS The Russian Genitive of Negation: Integration of Lexical and Compositional Semantics, In his article [Апресян 1999], which appeared at the same time as our article [Борщев, Парти 1999], Apresjan discusses similar examples, but with a slightly different point of view. More information about the differences in our approaches can be found in [Борщев, Парти 2011]. 1 Prototypical examples of the construction we are concerned with are given in (3). The construction resembles the English pseudopartitive (Selkirk [1977] and many others since). (3) a. два стакана молока b. ящик яблок c. машина дров d. полкорзины грибов The genitive of measure examples in (3) are a distinguishable subclass of a broader class of genitive constructions which also includes examples such as those in (4). (4) два литра молока полкилограмма яблок моток проволоки охапка дров стадо овец The interpretation common to the larger class is, roughly, a quantity of some substance or a collection of objects, where the substance or the objects are named by the Genitive NP. The English analogue of the broader class is studied in [Dodge, Wright 2002)]. Formal semantic analyses of measure constructions like two liters of milk in English and other languages can be found in [Krifka 1989a, 1989b; Landman 2004; Rothstein 2009a, 2009b; Schwarzschild 2002, 2006]. We will draw particularly on Landman s and Rothstein s work in what follows. 2. Four readings: From Container to Measure. Rothstein [2009a, 2009b] concentrates on a two-way ambiguity in English three glasses of water, which she characterizes as an individuating reading, in which the DP denotes plural objects consisting of three individual glasses of water, and a measure reading, in which the DP denotes quantities of water which equal the quantity contained in three glasses [Rothstein 2009a: , italics added]. She shows that the two Hebrew constructions for expressing three glasses of water, one of them unambiguously individuating and the other ambiguous, support the analysis of the English ambiguity offered by Landman [2004]. The distinctions argued for by Rothstein are close to the semantic distinctions argued for in the case of Russian container-noun constructions in some of our earlier work [Борщев, Парти 1999, 2011; Borschev, Partee 2001, 2004]. In [Борщев, Парти 1999; Borschev, Partee 2001, 2004] we distinguished a Concrete Portion reading from a measure reading, describing the Concrete Portion reading as close to Pustejovsky s dotted type reading, referring simultaneously to the container and the substance in it (see Section 2.1), although we had no way to formalize that notion, and instead represented it as a predicate true of a portion of matter (quantity of substance) filling a container of the given sort. Later, in an unpublished handout for a conference in Düsseldorf in 2007, we explicitly distinguished and gave a partial formalization of three readings: QUANT+CONTAINER, QUANT, and MEASURE, where the QUANT+CONTAINER reading is a predicate of the container together with the substance it contains; the QUANT reading applies to a particular quantity of substance that fills a container, and the MEASURE reading applies to a standard quantity of the substance, corresponding to a conventional standard size of containers of the given kind. 2 Our QUANT and MEASURE may be considered subtypes of Rothstein s measure readings. Her individuating reading is close to our QUANT+CONTAINER reading. Most recently in [Борщев, Парти 2011; Partee, Borschev 2012] we have further split our QUANT reading, making a three-way distinction within Rothstein s measure readings, giving four readings in all. For example (3a), we distinguish four readings of the construction, which we refer to as follows: the Container + Contents reading: Rothstein s individuating, Pustejovsky s dotted type, a predicate applying simultaneously to the glass(es) and the milk (Sec. 2.1); the Concrete Portion reading: a predicate applying to a concrete quantity of milk that fills some given glass(es) (Sec. 2.2); the Ad Hoc Measure reading: a predicate applying to a quantity of milk that would fill some given glass(es) (Sec. 2.3); the Standard Measure reading: (Rothstein s measure ), where stakan glass has developed a lexicalized sense as a standard unit of measure like litr (Sec. 2.4). For the four readings of the construction, we propose four related but distinct meanings for the container-noun. The first is closest to the basic meaning of the containernoun, and the last is closest to measure-nouns like liter. We start by describing the four distinct readings informally in this section. Their differences are summarized in Table 1 below 2. Table 1. Properties of the four readings of the construction dva stakana moloka CONTAINER + CONTENTS CONCRETE PORTION AD HOC MEASURE STANDARD MEASURE admissibility in contexts that are usually characteristic for words such as stakan, but not + for words such as moloko free cooccurence with fractional numerals * * OK OK if more than one container of substance, can the containers + + be of different sizes? standard size for the container + lexically limited + In Section 3 we discuss some of the main puzzles and problems we have encountered in drawing these distinctions. In Section 4 we formalize and compare the liter-of construction and all four of the genitive of measure constructions, improving on our earlier formalizations by drawing heavily on the Landman-Rothstein analysis. In Section 5 we discuss issues concerning the category, type, and interpretation of the genitive moloka milk and of non-bare-noun genitives in these constructions. 2 We are grateful to Maria Kholodilova for summarizing our diagnostics in this table, which we have copied from her referee s report. 3 2.1. The Container + Contents reading The Container + Contents reading of dva stakana moloka combines reference to the container(s) and reference to the concrete amount of the substance in the container(s) as in (6). (5) CONTAINER + CONTENTS reading: a predicate true of some container(s) together with a substance that fills it/them. (6) a. Он принес бутылку водки. b. Поставь этот ящик яблок в угол. This construction is similar to the English one discussed by Selkirk [1977] under the label pseudopartitive construction. Pustejovsky (1993) introduced the notion dotted type (reminiscent of the Cartesian dot product ) to represent the sort of an expression that simultaneously incorporates two distinct sorts, where sorts are ontological/linguistic semantic classes that play a role in co-occurrence restrictions (sočetaemost ), such as animate, inanimate, human, liquid, institution, location, building,, and in our case container and substance for the container and its contents. So whereas in Rothstein s analysis, one reading of glass of milk has the sort container and the other reading has the sort substance or liquid, on the Container + Contents reading, if one wants to argue that we can refer simultaneously to the container and the substance contained in it, it could be said to have the Pustejovskian dotted type container substance. Dodge and Wright [2002] discuss factors that favor reference to the container vs. reference to the contents without suggesting that one can sometimes refer to both simultaneously. Accompanying verbs may select for one or the other, as in (7). But as Pustejovsky (1993) emphasized, copredication is possible for dotted-type objects, showing that both sorts are simultaneously accessible. The examples in (8) are from [Asher, Pustejovsky 2005: 7]; (9) is a comparable example for our construction. (7) a. Мы выпили бутылку (полбутылки) шампанского. b. Мы разбили бутылку (*полбутылки) шампанского. (8) a. Mary picked up and mastered three books on mathematics. b. Lunch was delicious but took forever. (9) Он выпил стакан молока, который стоял на столе. When Rothstein [2009a] discusses the reading which refers to the container together with its contents as the individuating reading, she does not attempt to give it a dotted type, but takes it to refer to the complex entity consisting of the glass together with its contents. When she formalizes it (for English and for Hebrew), it turns out to actually refer to the glass, which is characterized as containing contents of a given kind. In her referee report, Maria Kholodilova offered a useful argument in favor of something like a dotted type analysis of such a Container + Contents reading. She notes that when one looks at which verbs freely co-occur with various container nouns used in isolation, not all of those co-occurrences are possible when the container noun is used with a Genitive of Measure complement. So while (7b) is normal, (10) is not. (10)??? сломать ящик яблок And she makes the reasonable suggestion that the relevant difference between (7b) and (10) is that in the case of (7b), when the bottle is broken, the champagne is also affected it will all spill out whereas in the case of (10), it s quite possible that the apples are not affected when the box is broken. As she suggests, if this explanation for the difference holds 4 up under further research 3, it could serve as additional indirect evidence that the described object in (7b) is really the bottle plus the champagne, and not just a bottle that happens to be filled with champagne. See also the contrasting example (13) below. For Russian, it would be slightly too weak to just say in an example like (9) that the glass contains milk; we must rather say that the glass is filled with milk 4. There is a different construction, illustrated in (11), which clearly denotes the container and describes it as containing (and not necessarily filled by) contents of a certain sort. Likewise there is a separate construction, illustrated in (12), that picks out just the contents, described as being in a container of a certain sort. (11) стакан с молоком (12) молоко в стакане The two constructions in (11) and (12) unambiguously denote the container and the contents respectively, and neither is a measure phrase. The construction in (9), on the other hand, denotes both the container and the contents together. Kholodilova, who noted the contrast between example (10) and example (7b), notes that the further contrast between the nearly impossible (10) and the only slightly degraded (13) adds additional support to the claim that the construction in (11) and (13) denotes just the container, while that in (9) and (10) denotes both the container and the contents. (13)?сломать ящик с яблоками But adding dotted types to a formal semantic theory is not easy (see brief discussion in Section 3), and since we do not have a formal theory of dotted types, our formalization in Section 4 approximately follows Rothstein s analysis, amending it to specify that the container is filled by, and does not merely contain, contents of a given sort. This will leave some of the puzzles concerning dotted types open 5. In our prose discussion, we maintain our conclusion that the construction refers both to the container and contents. The biggest difference between the Container + Contents reading and the three other readings described below is that on the Container + Contents reading, a glass of milk is a glass, whereas on the other three readings, a glass of milk is milk. The same distinction holds between Rothstein s two readings; it is in that sense that the three readings described below are all variants of Rothstein s measure reading. As a corollary of this basic distinction, the use of numbers like half or two and a half is more restricted for the Container + Contents reading than for the measure readings. It is not impossible to use fractional numbers with the Container + Contents reading, but it is not usual, and it may entail implausible real-world states of affairs. Normally cardinalities are expressed by whole numbers, and many formalisms make that obligatory; but there are 3 Our intuitions are that this explanation will hold up. We can add one supporting example: while (10) seems quite degraded, (i) below strikes the second author as fine, imagining a context where some very heavy object falls and crushes the box together the apples in it. (i) раздавить ящик яблок 4 Susan Rothstein (p.c.) counters that when we speak, for instance, of a glass of wine on this reading, the glass is rarely full; it is only when using a glass or cup as a measure that we normally require it to be full. But at the same time Russian speakers we have consulted have clear intuitions of a difference between kastrjulja vody pot of water and kastrjulja s vodoj pot with water : the first must be relatively full, while the second just has to have some water in it. We believe, as Rothstein herself suggests (p.c.), that the answer lies in the vagueness and context-dependence of the notions of fill and full ; a glass of wine on the Container + Contents reading is relevantly full when it contains a normal serving of wine, while on a measure reading it should be full to the brim. 5 Recent work by Nicholas Asher on dotted types is beyond the scope of this paper but might in principle solve our problem; we discuss it briefly in [Partee and Borschev 2012]. 5 approaches such as those of Hackl [2001] and Krifka [1989a], discussed by Kennedy and Stanley [2009], which treat the counting of (at least some) count nouns as analogous to the measuring of mass nouns, so as to account for the possibility of fractional numbers used with count nouns of suitable divisible sorts, as in (14). (14) a. John ate two and a half sandwiches for lunch. b. We burned two and a half logs in the fireplace last night. If one uses fractional numbers with the Container + Contents reading, it would seem that there must be fractional containers involved; there is no such restriction when fractional numbers are used with measure readings. So while (15) is normal in either of the measure readings discussed in Sections 2.3 and 2.4 below, (16) would require the existence of some salient container that counts as a half glass perhaps a cut-off glass, perhaps a half-filled glass being referred to metaphorically as a half glass 6. (15) Он выпил два с половиной стакана молока. (16)??Он уронил с подноса два с половиной стакана молока 2.2. The Concrete Portion reading The Concrete Portion reading and the Ad Hoc Measure reading of dva stakana moloka were not distinguished in our earlier work; either could be a case of QUANT in [Borschev, Partee 2001, 2004]. In this section we discuss the Concrete Portion reading, which describes a portion of matter, and characterizes its quantity in terms of some concrete containers it actually fills. In Section 2.3 we describe the Ad Hoc Measure reading in which the substance need not be in any container, but its quantity is described using some actual or potential container as a measure, a container that it would fill. In Section 2.4 we describe the Standard Measure reading, which involves a lexical shift, not a formally derivable one like the others discussed here. All three of these readings have a similar syntactic and semantic structure, which we make explicit in Section 4; in all of them the substance phrase in the genitive is the head 7, perhaps surprisingly, and the measure expression phrase (in whatever case the position of the whole NP in the sentence requires) is a modifier. While it may seem surprising that the semantic head of the phrase is in the genitive, that is not so surprising for Russian, where many numerals demand genitive case on the noun, and where genitive may be used to express partitivity even without any overt quantifier or measure expression 8. 6 But a sentence like On prines polkorziny gribov He brought half a basket of mushrooms, is judged intuitively to refer to a half-full basket together with the mushrooms in it. Note that neither of the constructions in (11) and (12) allow half a glass in Russian unless it really is a cut-off glass. The second co-author finds sentences like (16) ungrammatical; the first co-auth
Related Search
Similar documents
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks