Formal and Lexical Semantics and the Genitive in Negated ...

Borschev, Vladimir and Barbara H. Partee (1998). Formal and lexical semantics and the genitive in
negated existential sentences in Russian. In: ðeljko Boškoviƒ, Steven Franks and William Snyder, eds.,
Formal Approaches to Slavic Linguistics 6: The Connecticut Meeting 1997, Ann Arbor: Michigan
Slavic Publications, 75-96.
Formal and Lexical Semantics and the Genitive
in Negated Existential Sentences in Russian*
Vladimir Borschev and Barbara H. Partee
VINITI, Russian Academy of Sciences, Moscow and
University of Massachusetts, Amherst
1
Goals
1.1 Theoretical concerns and general goals
The theoretical concern of this paper is the integration of formal and lexical semantics, more specifically the
traditions of (post-) Montague Grammar and the Moscow semantic school, respectively. We propose to
represent lexical meaning in the form of meaning postulates, and the output of compositional semantic
interpretation in a formula of intensional logic in which lexical items are primitives, and to integrate lexical
and compositional information via entailments from these (and other) sources.
We think of the content of a text as a theory determined by a set of axioms together with their
entailments. The axioms come from various sources: lexicon, compositional semantics, context and
background knowledge. (Broader and narrower notions of semantic or semantico-pragmatic interpretation
correspond to the inclusion or exclusion of various potential sources of axioms.) Such a theory
characterizes the class of all models that are consistent with the content of the given text, or of the text
together with aspects of its context. Some of the most general axioms, which may be taken to form part
of the theory of any text, are those that represent some of the most general constraints on possible models
of a given language, axioms which contribute to what the Moscow School calls naivnaja kartina mira ‘the
naive picture of the world’ (Apresjan 1974), and what formal semanticists, following Bach (1986), call
Natural Language Metaphysics.
We do not pretend to have an articulated view of the nature of all the different sorts of axioms that may
play a role in the “theory” of a text, but here we will illustrate some of the possibilities.
1.2 The Genitive in Negated Existential Sentences
*
W e are grateful for valuable discussions and comments to Leonard Babby, Wayles Browne, Catherine Chvany, Hana
Filip, Natasha Kondrashova, Elena Paducheva, Orin Percus, Asya Pereltsvaig, Ekaterina Rakhilina, and Tanya Yanko,
and to the editors; our gratitude does not imply their agreement with our conclusions, for which we take full
responsibility.
This work was supported in part by a grant to the first author from the Russian Foundation for Basic Research, Project
No. 96-06-80315a, and in part by a grant from the National Academy of Sciences to the second author under the program
for Collaboration in Basic Science and Engineering to host the first author for collaboration on the project “Towards an
Integration of Formal and Lexical Semantics: Meaning Postulates and Fine-grained Sortal Structures.” Neither of the
sponsoring organizations is responsible for the views expressed.

---

The Russian genitive with subjects of negated existential sentences (the NES construction, in Babby’s
terms) provides an interesting empirical domain for examining the interaction of lexical and compositional
semantics and testing theoretical approaches. From the work of Babby (1980), Padu…eva (1992,1997),
and others it seems clear that an account of the NES construction involves at least the syntax and Theme-
Rheme (or topic-focus) structure of negative sentences, the lexical semantics of verbs, and often additional
context-specific presuppositions or implicatures.
We will follow Babby in analyzing the NES construction as implying the negation of existence of “the
NP”; but not as “denying that the NP has a referent,” rather as denying that “NP” exists in a given
“location.” What the relevant “location” is depends in part on the Theme-Rheme structure. The
relativization of “existence” to a “location” makes it possible to subsume byt’ under the class of verbs
covered by Babby’s analysis, and to account for NES sentences with proper names and other definite NPs
as their (genitive) subjects.
We suggest a compositional interpretation of the NES construction which involves an assertion part
negating the literal predication of the given verb to the given subject and location, together with a
presupposition that that verb in that sentence is equivalent to “be” or “exist.” The role of the additional
axioms that we discuss (lexical, encyclopedic, contextual) is seen to consist in providing the support needed
for such a presupposition to hold in a given context.1
2
“Existence” and “existential sentences”
2.1 Babby’s distinction between NES and NDS
Babby contrasts negated existential sentences (NES), with genitive subjects, from “negated declarative
sentences” (NDS), with nominative subjects, as in the following (his (81a-b), from Ickovi… 1974):
(1) Otvet
iz polka ne prišel.
Answer-NOM-m-sg from regiment NEG arrived-m-sg
‘The answer from the regiment has not arrived.’
(2) Otveta
iz polka ne prišlo.
Answer-GEN-m-sg from regiment NEG arrived-n-sg
‘There was no answer from the regiment.’
Chart (3), from Babby, shows a “scope of assertion” difference, argued by Babby to follow from
Theme-Rheme differences.
1
We neglect much existing syntactic work to focus on semantics. Among important issues we do not address are the
potential unification of subject and object genitives, Pesetsky’s (1982) assumption of an empty quantifier governing the
genitive NP, and the suggestion of Perelstvaig (1997) and others of a connection with negative polarity phenomena.

---

(3)
AFFIRMATIVE
NEGATED
EXISTENTIAL
[Scope of A VP NP] YNEG
[ne VP NPgen]
DECLARATIVE
NP [Scope of A VP] YNEG
NPnom [neVP]
Although Babby generally characterizes NES’s as having the entire sentence inside the scope of
negation, he notes that there may be an optional locative outside the scope of negation. Below we will make
“location” obligatory in NES’s but possibly implicit.
2.2 Sentences with byt'
Babby does not claim that his analysis applies to sentences with byt'. He argues that since the main
assertion in an NES is a denial of existence of the referent of the subject NP, NES’s should normally not
permit definite NPs as subjects; but sentences with byt' do commonly allow the genitive of negation with
proper names and other definite NPs. Babby claims that the sentence (4) below cannot be an existential
sentence because of its definite subject and therefore must be a “locative sentence,” a type of NDS, with
“be at the lecture” as the negated part, which goes contrary to the generalization in 2.1 above.
(4) Ivana ne bylo na lekcii
Ivan-GEN-m-sg NEG was-n-sg at lecture
‘Ivan wasn’t at the lecture.’
However, byt' is in a sense a “basic” verb of existence (“being”), and as many have observed, all NES’s
can be approximately paraphrased as NES’s with byt', as illustrated by the following “equivalences,” the
nature of which will be discussed in what follows.
(5) a. Otveta ne prišlo =
Answer-GEN-m-sg NEG arrived-n-sg =
Otveta ne bylo
Answer-GEN-m-sg NEG was-n-sg
‘No answer came.’ = ‘There was no answer.’
b. Moroza ne …uvstvovalos’ (Babby 1980, p.59) =
Frost-GEN-m-sg NEG be-felt-n-sg =
Moroza ne
bylo
Frost-GEN-m-sg NEG was-n-sg
‘No frost was felt.’ = ‘There was no frost.’
c. Posudy na stole ne stojalo
Dishes-GEN-f-sg on table-LOC-m-sg NEG stood-n-sg
= Posudy na stole
ne bylo

---

= Dishes-GEN-f-sg on table-LOC-m-sg NEG were-n-sg
‘No dishes stood on the table.’ = ‘There were no dishes on the
table.’
We believe with more careful attention to the interpretation of “existence” in “existential sentences,”
sentence (4) can indeed be interpreted as an existential sentence, and Babby’s analysis can work for
existential sentences with byt' as well as for existential sentences with lexical verbs.

2.3 “Being” and the roles of “thing” and “location”
We understand existence, or “being,” in the sense relevant to NES’s, as a potentially temporary relation
between some “thing” and some “location.” We may accept Jackendoff’s (1972) metaphorical-structural
extensions of “being in a location” to include “being in some state,” “occurring in some spatiotemporal
region,” “being in someone’s possession,” extending also to “being in the speaker’s (or an observer’s)
perceptual field” (Padu…eva 1992, 1997). We will treat “thing” and “location” as basic roles of verbs of
being, or better, as roles of the situations denoted by existential sentences: BE(THING,LOC).1 Theme-
Rheme differences, the subject of Section 3, distinguish ES’s, in which the “location” is the Theme, from
DS’s, in which the “thing” is the Theme.
In “existential sentences,” then, some “location” is given (Thematic Location) or contextually
presupposed (implicit Thematic Location), and it is asserted that in that location there is (“exists”) some
“thing” of some sort.
(6)
“EXISTENCE IS RELATIVE” PRINCIPLE:
Existence (in the sense relevant to AES’s and NES’s) is always relative to a
“location.”
The principles that determine which “location” is the one relative to which an existence claim is being
made (if any) in a given sentence are related to Theme-Rheme structure. We believe that these principles
make the analysis of sentences with byt’ consistent with Babby’s analysis of sentences with lexical verbs.
We discuss these principles in Section 3 below, and their interaction with existence presuppositions and
their location roles.
3
Theme-Rheme structure, presupposition, and locations
3.1 Conditions for Genitive of Negation: Babby’s basic scheme Babby’s final formulation of his rule
of genitive marking in NES’s (his (160)) is given in (7) below.
1
Our “Thing” role may well be Jackendoff’s (1972) thematic role Theme (not to be confused with Theme vs. Rheme),
which would fit analyses of the relevant NPs as “underlying objects” of Unaccusative verbs (Pesetsky 1982 and others),
and might predict non-obliqueness. Hana Filip (p.c.) suggests that our Thing and Loc roles probably have a status
between conceptual structure and syntax, as argued for by Fillmore and as found in Dowty’s work on lexical meaning.

---

NEG
(7) [Rheme V NP] Y [ ne V NPgen]
Conditions:
(a) NP is indefinite [we disagree – BHP and VB]
(b) V is semantically empty [discussed in Section 4 below]
We agree with Babby, and with Prague school linguists such as Haji…ová and Sgall, that scope of
negation is directly correlated with Theme-Rheme structure. In “existential sentences,” the location is
Thematic, and both the “thing” and the verb are Rhematic and hence fall within the scope of negation; so
negation in NES’s negates existence in the Thematic location. In “Declarative” sentences, the “thing” is
Theme, and “its existence” stays outside the scope of negation; the Verb Phrase is Rheme and is negated.
(When NP subject alone is Rheme, negating gives constituent negation.)
In our terms, the minimal difference between pairs such as Babby’s and Arutjunova’s examples (8a,b)
below would be schema-tized as follows:
Existential S’s: Location = Theme; ‘Thing-being-in-it’ = Rheme.
Declarative S’s, including “Locative S’s”: Thing = Theme; ‘Being-in-location’ [or other predicate] =
Rheme.
(8) a. [Theme Na stole] [Rheme byli knigi i
On table-LOC-m-sg were-m-pl book-NOM-f-pl and
ñurnaly]
magazine-NOM-m-pl
‘On the table there were books and magazines.’
b. [Theme Knigi i ñurnaly]
Book-NOM-f-pl and magazine-NOM-m-pl
[Rheme byli na stole]
were-m-pl on table-LOC-m-sg
‘(The) books and magazines were on the table.’
The negation of (8a) would use genitive, that of (8b) nominative.
We disagree with Babby’s claim that the NP subject in an NES must be indefinite; Babby’s text itself
includes a number of counterexamples, and we have argued that there are many NES’s with byt’ that
provide further counterexamples. Rhematic NPs are typically indefinite, but definite NPs may also be
Rhematic and may show up in genitive of negation in NES’s.
3.2 Theme-Rheme, presuppositions and assertions
We follow Haji…ová (1973,1984) and Peregrin (1995) in the analysis of the connection between Theme-

---

Rheme structure and presuppositions (and assertions) corresponding to this structure.2 For simplicity, we
limit our discussion to presuppositions and assertions of existence. On their analysis, an NP like knigi
‘books/the books’ will carry an existence presupposition when it occurs in the Theme but not when it
occurs in the Rheme; this is related to the function of the Theme in anchoring the sentence to the conversa-
tional background.
But existence, including the existence relevant for existence presuppositions, is always existence in some
location. Let us informally label the different “locations” relevant to NES’s and NDS’s according to their
roles in different “being-situations.”
Thematic location: the “location” of the “being-situation” of the sentence when that “location” is the
Theme of the sentence; this includes both explicit Thematic location as in the NES’s (4) and (5c) and
implicit Thematic location as in the NES’s (2) and (5a,b).
Rhematic location: the “location” of the “being-situation” of the sentence when that “location” is the
Rheme of the sentence: (11a).
Reference location: (or “Anchor location”): the “location”of the “being-situation” of the existence
presupposition associated with the Theme of the sentence. For a sentence expressing a “being-situation,”
if the “thing” (typically the subject) is Theme (as in NDS’s), then Reference location will be the “location”
contextually associated with that “thing” – a part of the conversational background, analogous to
“Reference time.” Existence in the Reference location is what we often informally describe as existence in
the “universe of discourse.” If the “location” is Theme, as in (4), the associated existence presupposition
guarantees the existence of that location (see 3.4). In that presupposition, the lecture plays the role of
“thing,” and its location is the Reference location; in that case Thematic location is identical to or within
Reference location.
Resource location: “location” associated with a presupposition of existence of a thing denoted by a
referential NP like Maša ‘Masha’ in (9); where a Resource location is depends on the knowledge source,
not on the structure of the sentence. This term is modeled on Barwise and Perry’s (1981) resource
situations.
These distinctions are what enable us to subsume byt’ under the verbs covered by Babby’s analysis.
Sentence (4) illustrates the distinction between the situation of existing in the Thematic location and a
backgrounded situation of existence in a Resource situation. Sentence (4) asserts that Ivan did not exist in
the Thematic location “at the lecture,” while presupposing he does exist in a Resource location “in the
world.”
The “Thematic location” is sometimes a speaker’s (or observer’s) “point of view” location (see
Padu…eva 1992, 1997). This “Thematic location” may be implicit, as in (2) and (5a,b). And we believe that
a perceptual verb always has a “location”role, explicit or implicit, which can be Thematic; we offer this as
a possible reason behind Padu…eva’s observation that perception verbs can always be used as “existential
verbs” supporting a genitive of negation, as in (9).
2
A few of our colleagues disagree with our (and Babby’s) claim that the genitive NP is always rhematic regardless of
word order, but agree with our claims about what is presupposed. If these presuppositions do not follow from Theme-
Rheme structure in the way sketched here, we do not know how to derive them.

---

(9) Maši ne vidno
Masha-GEN-f-sg NEG seen-n-sg
‘Masha isn’t to be seen.’
Sentence (9) asserts the nonexistence of Masha within the speaker’s perceptual field (the implicit Thematic
location) without denying her existence “in the world” (the Resource location for the proper name).
3.3 NES’s and NDS’s: their assertions and presuppositions
An informal statement of the assertion made by an NES is given in the NES Principle below; it will be
expanded upon in the discussion of the Presupposed Equivalence in Section 4.
NES PRINCIPLE: An NES denies the existence of the thing(s) described by the subject NP in the
“Thematic location.”
We have seen examples with implicit Thematic locations associated with implicit observers. There are
also cases, like (10), in which the implicit Thematic location is simply “the actual world,” yielding a literal
denial of existence.
(10) Edinorogov ne suš…estvuet.
Unicorns-GEN-m-pl NEG exist-sg
‘Unicorns do not exist.’
In an NDS, the NP subject or “thing” is always in the Theme, so it carries a presupposition of
existence in the Reference location.
Just as the “Reference location” is associated with material in the Theme, so is “Reference time.” In
(11a), from Apresjan (1980), the contextually specified past time is associated with the Theme “Otec”; the
conversation must have been about “where Father was/ has been,” perhaps today, perhaps in his life. In
(11b), the reference time must be some given seaside occasion, since “at the sea” is the Theme.
(11)a. Otec ne byl na more.
Father-NOM-m-sg NEG was-m-sg at sea.
Father was not at the sea.
b. Otca ne bylo na more.
Father-GEN-m-sg NEG was-n-sg at sea.
Father was not at the sea. (“There was no Father there.”)
In example (1) above, Otvet is in the Theme and it is presupposed that it exists in the “Reference
location” or “universe of discourse.” The sentence asserts that it did not arrive. Sentence (2) asserts non-
existence of the answer in “Thematic location” but says nothing about whether it exists in any other location,

---

including the “universe of discourse.” This leaves the sentence open to pragmatic influences that may
support or inhibit the “insinuation” (Padu…eva 1997) that perhaps no answer exists at all.
3.4 Existence presuppositions for Thematic vs. Rhematic locations
Although most locative expressions are normally understood as involving locations which are presupposed
to exist, it is predicted that a Rhematic location might in principle not be presupposed to exist, while a
Thematic location must be presupposed to exist. This prediction seems to be confirmed by the difference
between the somewhat awkward (12a) and the totally impossible (12b).
(12) a.?Ivan [Rheme ne byl na lekcii.]
Ivan-NOM-m-sg NEG was-m-sg at lecture.
Lekcii ne bylo.

lecture-GEN-f-sg NEG was-n-sg
‘Ivan was not at the/a/his lecture. There wasn’t any lecture.’

b. *Ivana ne bylo [Theme na lekcii.]
Ivan-GEN-m-sg NEG was-n-sg at lecture.
Lekcii ne bylo.
lecture-GEN-f-sg NEG was-n-sg
‘Ivan was not at the/a/his lecture. There wasn’t any lecture.’
3.5 Summary scheme from a speaker’s perspective
We may summarize the analysis presented so far from a speaker’s perspective as follows: Suppose the
speaker’s intentions are as sketched in (13).
(13) Theme = Loc, Message = NEG(BE(Thing, [Loc]))3
Then according to our analysis, the existence of Loc is presupposed, and the assertion is the negation of
the being of Thing in that Loc.
The realization of these intentions, assuming that “Thing” is expressed by an NP that meets the relevant
syntactic requirements (such as non-oblique case), will involve (i) putting the NP expressing the Thing into
Genitive case, and (ii) optionally substituting a lexical V for byt’ to express BE, if there exist axioms (see
Section 4) supporting the equivalence of V and byt’.
4 Weak verbs and the sources of their existential “axioms”
From the NES PRINCIPLE in 3.3 it follows that an NES presupposes the equivalence (in the context of
3
W e enclose “Loc” in square brackets as an informal indication that Loc, as Theme, is outside the scope of negation;
see Peregrin (1995) for formalization and discussion.

---

the given S) of the “existence predicate” and the predicate “literally” corresponding to the verb of this S
or, roughly speaking, of the verb byt’ and this verb. That is, NES’s such as those in (5) above have the
same truth and falsity conditions as their counterparts with byt’. (We ignore the subclass of verbs of
“appearing” to avoid aspectual complications.) We state this presupposed equivalence in (14) below; its
source is discussed in Section 4.7.
(14) PRESUPPOSED EQUIVALENCE:
V (THING, LOC) <==> BE (THING, LOC)
But how is this equivalence possible with verbs whose literal meaning is clearly not simply “exist” or “be”?
The usual answer is that in NES’s, “weak” verbs are used: verbs which have become semantically empty,
at least when occurring in combination with the given subject NP.
In the framework of our paper the question concerning properties of these verbs and reasons for their
“weakness” can be approached in the following way: Suppose we assume that the lexical verbs have their
normal meanings, whatever those are, so that it is not the case that they are simply semantically equivalent
to byt’ in these sentences. Then we ask: what types of further axioms can we find holding for the given S
in the given context (i.e., contained in the theory of the given S in the given context), whose presence could
contribute to making this equivalence a “locally valid” theorem? What is the nature and what are the sources
of such axioms?
Such a question is in principle open-ended, but let us list a few cases. Some are discussed in the litera-
ture, some can be found lying on the surface; we are trying to offer a slightly different perspective to help
integrate existing insights.
For each case below, we give an example NES followed by the equivalence statement presupposed
by the NES PRINCIPLE; then we show what sorts of supplementary axioms might be operative in the
given context to yield the equivalence.
4.1 Dictionary axioms
(15) a. NES: Ne belelo parusov na gorizonte.
NEG shone-white-n-sg sails-GEN-m-pl on horizon
‘No sails were shining white on the horizon.’
b. Presupposed Equivalence:
Na gorizonte belel parus <==>
On horizon shone-white-m-sg sail-NOM-m-sg <==>
Na gorizonte byl parus
On horizon was-m-sg sail-NOM-m-sg
‘A sail shone white on the horizon.’ <==> ‘There was a sail on the horizon.’
c.
‘Dictionary axiom’ (part of lexical semantics):
belet’ <==> byt’ belym (in the field of vision)

---

to shine-white <==> to be white
d. Dictionary or encyclopedic axiom; ‘common knowledge’:
Parus kak pravilo belyj.
Sail-NOM-m-sg as a rule white-NOM-m-sg
‘Sails as a rule are white.’
With (15c) and (15d) we can almost satisfy the presupposition of equivalence in (15b); the equivalence
only holds under further conditions such as the existence of a potential or actual observer whose field of
vision includes the relevant location. In a context in which such further conditions can be consistently
assumed to be met, (15c-d) together support the equivalence in (15b). Since (15c-d) are common
knowledge, (15b) is entailed in normal contexts, and genitive of negation is therefore a normal choice with
that combination of verb and subject.
4.2 Dictionary + contextual axioms
Modification of the previous example:
(16) a. NES: Ne belelo domov na gorizonte..
NEG shone-white-n-sg houses-GEN-m-pl on horizon
‘No houses were shining white on the horizon.’
b. Presupposed Equivalence:
Na gorizonte beleli doma <==>
On horizon shone-white-m-pl houses-NOM-m-pl <==> Na gorizonte byli doma
On horizon were-m-pl houses-NOM-m-pl
‘Houses shone white on the horizon.’ <==> ‘There were
houses on the horizon.’
c. ‘Dictionary axiom’ (part of lexical semantics):
belet’ <==> byt’ belym (in the field of vision)
to shine-white <==> to be white
d. Doma kak pravilo belye. (Normally FALSE)
house-NOM-m-pl as a rule white-NOM-m-pl
‘Houses as a rule are white.’
In contrast to the case above, houses normally come in a great variety of colors and are not normally
presumed to be white. But if the context includes the information that in this region, (16d) holds,
then the dictionary axiom (16c) together with the contextual axiom (16d) will play the same role as the
axioms (15c-d) in the previous example; together they entail the equivalence (16b).
Note: From the speaker’s point of view, the axioms generate the equivalence (16b), thereby
“bleaching” the verb belet’ and licensing the use of the genitive of negation.

---