Journal of Semantics, 37, 2020, 311–366
doi:10.1093/jos/ffaa002
Advance Access Publication Date: 30 May 2020
Articles
Definiteness, Uniqueness, and Maximality
in Languages With and Without Articles
Radek Šimík
Charles University
Christoph Demian
Humboldt-Universität zu Berlin
First version received 21 July 2018; Second version received 17 February 2020; Accepted 2 March
2020
Abstract
We present a number of experiments testing influential hypotheses about the
meaning of definite descriptions (in languages with articles, represented here by
German) and bare nominals (in articleless languages, represented here by Russian).
Our results are in line with the commonly entertained hypothesis that definite
descriptions convey uniqueness (if singular) or maximality (if plural), but fail to
support two hypotheses about bare nominal interpretation, namely that singular bare
nominals convey uniqueness (Dayal 2004) and that topical bare nominals convey
uniqueness/maximality (Geist 2010, among many others). Uniqueness or maximality
inferences are expected to arise via covert type-shifting under these approaches. Our
results are compatible with what we take to be the null hypothesis, namely that bare
nominals in articleless languages are existential and free of presuppositional semantics,
even if they correspond—in their use—to definite descriptions (Heim 2011).
1 INTRODUCTION
The semantic notion that underlies definite descriptions is often assumed to be a linguistic
universal, whether it corresponds to familiarity (Krámský 1972), identifiability (Lyons
1999), or uniqueness/maximality (Chierchia 1998; Dayal 2004; Geist 2010; Schwarz
2013; among many others). An implication of this assumption is that languages without
definite articles (henceforth articleless languages) can convey the same meaning as definite
descriptions do, albeit with different formal means. Consider the examples below. While
English obligatorily distinguishes between definite and indefinite nominals (the cat vs.
a cat), Russian can resort to a bare nominal (koška), which is considered ambiguous and
© The Author(s) 2020. Published by Oxford University Press. All rights reserved.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial
License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and
reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact
journals.permissions@oup.com
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
312 R. Šimík & C. Demian
as such can correspond to a definite or indefinite. Various factors are believed to affect
the correspondence, including syntactic position, case, grammatical number, word order,
prosody, givenness, or topicality.
(1) a. English
The / A cat slept.
b. Russian
Koška spala.
cat slept
(i) ‘The cat slept.’
(ii) ‘A cat slept.’
We have devised an experimental design, based on the Covered Box paradigm (due to Jesse
Snedeker; see Huang et al. 2013; Pearson et al. 2011; Romoli et al. 2011), to test some of the
prominent or commonly entertained hypotheses about the meaning of definite descriptions
in languages with articles and bare nominals (henceforth bare NPs) in articleless languages.
The overarching research question is whether bare nominals in articleless languages can
convey the same meaning as definite descriptions do, albeit with different formal devices,
which we will call definiteness correlates. It has been standard to answer this question
in the affirmative. The aim of our paper is to see whether this standard answer stands up to
empirical scrutiny. We will see that it does not.
The definiteness correlates we investigate are grammatical number (building on Dayal’s
2004 work) and topicality; in the latter case, we approach topicality via word order and
prosody (building on Geist 2010). On the semantic side, we concentrate on uniqueness (for
singulars) and maximality (for plurals)—concepts that are broadly agreed to underlie the
morphological category of definiteness. We use German as a representative of a language
with articles and Russian as a representative of an articleless language. The particular
hypotheses we test are (i) that German definite descriptions (as opposed to indefinites)
convey uniqueness/maximality, (ii) that Russian singular bare NPs convey uniqueness (as
opposed to plural bare NPs, which do not necessarily convey maximality), (iii) that Russian
preverbal (topical) bare NPs (as opposed to postverbal ones) convey uniqueness/maximality,
(iv) that Russian unstressed (topical) bare NPs (as opposed to stressed ones) convey
uniqueness/maximality. Finally, (v) we performed a crosslinguistic comparison, in order to
see whether German definiteness conveys uniqueness/maximality more clearly than Russian
definiteness correlates.
The experimental results confirm hypothesis (i), but fail to confirm hypotheses (ii)–(iv).
This means that definite articles interact with uniqueness/maximality more strongly than
definiteness correlates, which might as well not interact with uniqueness/maximality at all.
The results are problematic for the standard view that the semantic notions of uniqueness
and maximality are operative in articleless languages (in the form of covert type-shifts)
and that they can be activated by means of what we call here definiteness correlates. On the
other hand, they are compatible with what could be considered the null hypothesis (recently
briefly considered by Heim 2011), namely that the lack of definite articles translates to the
lack of definiteness-related semantics.
Finally, even though hypothesis (i) was confirmed by our results, explorative comparisons
with filler experiments demonstrate that the effect brought about by uniqueness/maximality
violations is weaker than one might expect—much weaker than the effect caused by
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
Definiteness, uniqueness, and maximality in languages with and without articles 313
scalar implicature violations, for instance, and comparable to the effect caused by violations
of pragmatic exhaustive inferences.1
The paper is organized as follows. In section 2 we provide the necessary background on
the interpretation of definite descriptions and bare NPs. Section 3 introduces a number of
approaches to the semantics of bare NPs—Heim’s (2011) hypothesis that all bare NPs correspond
to indefinites, Dayal’s (2004) proposal that the correspondence between bare NPs
and (in)definites depends on grammatical number, and finally Geist’s (2010) formalization
of the traditional view that topical bare NPs correspond to definite descriptions. Section 4
describes in detail the experimental design we used for testing the selected hypotheses.
Section 5 discusses the individual (sub)experiments—one per each hypothesis. Section 6
discusses the filler experiments. The paper is concluded by a general discussion (section 7).
2 BACKGROUND
2.1 Definite descriptions
According to a broadly accepted theory (ever since Frege 1892 and Russell 1905), a singular
definite description conveys that there is exactly one individual that satisfies the description.
The majority view is that this implication is a presupposition (Elbourne 2013; Frege 1892;
Strawson 1950), understood as a definedness condition.2 For instance, the meaning of the
cat is defined if and only if there is exactly one cat; if defined, the cat denotes / refers to
that cat (we use smallcaps for metalanguage). We follow standard notation and use the
ι-formula ιx cat(x) (type e) to encode this presupposition and the determinate meaning.3
The pertinent presupposition will be referred to as the uniqueness presupposition, but
sometimes also as the uniqueness inference (where the nature of the inference is not at
stake).
(2) a. The cat slept.
b. slept(ιx cat(x))
This so called uniqueness theory of singular definite descriptions is usually coupled with
the maximality theory of plural definite descriptions, according to which a plural definite
description denotes / refers to the maximal plural entity that satisfies the description (Link
1983; Sharvy 1980). This meaning is captured by the σ-formula in (3b).
(3) a. The cats slept.
b. slept(σx cat(x))
1 The term “filler experiments” refers to small subexperiments that did not test the core hypotheses and
contained items functioning as fillers/controls to the items in the main experiments. The structure of
our experimental design will become evident in section 4.
2 Various other hypotheses have been explored, including that uniqueness/maximality is asserted (King
2001; Russell 1905), conversationally implicated (Ludlow & Segal 2004; Szabó 2000), or conventionally
implicated (Horn & Abbott 2012).
3 We adopt Coppock & Beaver’s (2015) terminological convention and distinguish between a DETERMINATE
interpretation (denotation of type e expressions) and an INDETERMINATE interpretation (interpretation
as an existential quantifier). We sometimes abbreviate “NP with (in)determinate interpretation”
as “(in)determinate NP”.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
314 R. Šimík & C. Demian
Combining uniqueness and maximality in a unified theory is attractive because uniqueness
is a special case of maximality: if there is exactly one entity that satisfies the description, it
is also the maximal one.4
2.2 Bare noun phrases in articleless languages
It is standard to assume that in Russian it is possible to express the above meanings by (4a)
and (4b), respectively.5 Some scholars would argue that the indicated interpretation is not
just “available”, but—due to the preverbal position of the subject—necessary; we will turn
to this issue shortly.
(4) a. Koška spala.
cat slept
Available: ‘The cat slept.’
b. Koški spali.
cats slept
Available: ‘The cats slept.’
These examples clearly demonstrate that definite descriptions in languages with articles
correspond, at least in some cases, to bare NPs in articleless languages. But bare NPs can
apparently also correspond to indefinites. Examples of that are provided in (5), where the
indefinite paraphrase is adequate, although arguably not the only one available.
(5) a. Na kovrike ležala koška.
on mat lay cat
Available: ‘There was a cat lying on the mat.’
b. Na kovrike ležali koški.
on mat lay cats
Available: ‘There were cats lying on the mat.’
Table 1 provides an overview of the commonly entertained meanings of bare NPs in
articleless languages. As is clear from this overview, some meanings assigned to bare NPs
4 Some recent approaches to definite descriptions take the uniqueness and/or maximality approach to
definite descriptions as a starting point and argue for various refinements. For instance, Coppock &
Beaver (2015) argue that singular definite descriptions presuppose “weak uniqueness” (cardinality
equal to or smaller than 1) and, on top of that, are by default predicative rather than determinate
(a position first argued for by Graff 2001); von Fintel et al. (2014) argue that maximality should be
replaced by maximal informativeness; Schein (2019) argues that the part-of relation, considered
essential for plural and mass definite descriptions, should be replaced by an overlap relation. These
refinements pay off in some special circumstances but do not differ significantly from what we call
here the standard account, as far as the canonical uses of definite descriptions are concerned.
5 Russian potentially stands for a sizeable group of languages; 40% of the more than 600 investigated
languages in Dryer (2013) have no definite articles. For recent theoretically informed discussion
of some of them, see Cisneros (2019); Collins (2019); Dayal (2011); Jenks (2018); Modarresi (2014);
Schwarz (2013); and the references cited therein.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
Definiteness, uniqueness, and maximality in languages with and without articles 315
Table 1 Bare NP meanings
sg indef 1 Property λx[cat(x)] (the property of being) a cat
2 Quantifer λQ∃x[cat(x) ∧ Q(x)] (the set of properties that at least) some cat (has)
def 3 Entity ιx cat(x) the (single) cat
4 Kind ∩cat the species / taxonomical kind called “cat”
pl indef 1 Property λx[cats(x)] (the property of being) cats
2 Quantifer λQ∃x[cats(x)∧Q(x)] (the set of properties that at least) some cats (have)
def 3 Entity σx cat(x) (all) the cats
4 Kind ∩cats the abstract representative of (all) the cats
correspond to indefinite NPs, others to definite NPs.6 It is commonly assumed that not
all of these meanings are always equally available as denotations for bare NPs, despite
them being interrelated via general type-shifting principles (postulated by Partee 1987
and Chierchia 1998, among others). Various restrictions on these meanings or the typeshifts
that relate them have been proposed, giving rise to accounts that make different
predictions—sometimes complementary, sometimes contradictory. We will now turn to
selected accounts, whose predictions we tested experimentally.
3 BARE NP THEORIES AND FACTORS AFFECTING BARE NP
INTERPRETATION
3.1 Bare NPs are always indeterminate (Heim 2011)
Heim’s (2011:1006) response to the observation that bare NPs in articleless languages
can either correspond to definites or to indefinites is the following: bare NPs always
denote existential quantifiers, but—due to the lack of definite articles in the grammatical
system—can also be used in situations where a definite description would be appropriate in
a language with articles.7 In order to understand Heim’s (2011) position, we need to say
a little bit about how (the meanings of) indefinite and definite NPs relate to one another. The
basic recognition is that a sentence with a definite NP asymmetrically entails (is logically
stronger than) a sentence with a corresponding indefinite: whenever (6a) is true, (6b) is true,
too, but crucially not conversely.
(6) a. The cat slept on the couch.
b. A cat slept on the couch.
6 The table presents a standard, but simplified view. Definite descriptions are sometimes considered
to be predicative (Coppock & Beaver 2015; Graff 2001) or quantificational (King 2001; Neale 1990;
Russell 1905). Also, plural indefinites (= bare plurals) can have kind interpretations in languages
like German or English (see, e.g., Chierchia 1998). What will be important to us is the presence of
uniqueness/maximality inferences—something that is usually preserved independently of a particular
analysis.
7 See Borik (2016) (recently further developed by Seres & Borik to appear) for a related proposal and
for empirical arguments from Russian.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
316 R. Šimík & C. Demian
The reason why replacing a definite by an indefinite often leads to infelicity—as e.g. in
(7)—is pragmatic: using a form with weak meaning (indefinite NP) implicates that the
corresponding form with strong meaning (definite NP) is not supported in the common
ground. Consequently, while (7a) presupposes that there is exactly one seat (7a-ii), (7b)
implicates that this presupposition is not true (7b-ii), which, in combination with the
assertion, leads to the strengthened implication that there are at least two seats (7b-iii).
We call this the anti-uniqueness implication of indefinite NPs.8
(7) The bicycle was fine after (Heim 2011:1005)
a. the seat was replaced.
(i) Assertion: replaced(ιx seat(x))
(ii) Presupposition: |seat| = 1
b. # a seat was replaced.
(i) Assertion: ∃x[seat(x) ∧ replaced(x)]
(ii) Implicature: |seat| = 1
(iii) Assertion+Implicature: |seat| > 1
Crucially, the implicature is not always drawn: (6b) does not imply that there were multiple
cats. This supports the stance that the anti-uniqueness implication of indefinites is pragmatic
in nature.9
Turning back to articleless languages, the idea is that, in the absence of a definite article
in the grammar or a language, there is nothing that could generate the anti-uniqueness
implication and the bare NP—by hypothesis a presupposition-free existential quantifier—
remains felicitous in situations where uniqueness is satisfied in the common ground and
where a definite description would have to be used in a language with articles.
(8) ‘The bicycle was fine,’
posle togo kak zamenili sedlo.
after that as replaced seat
‘ . . . after we replaced the seat.’
(i) Assertion: ∃x[seat(x) ∧ replaced(x)]
(ii) No implicature generated due to the absence of the definite article!
Figure 1 visualizes the pattern as predicted by Heim (1991, 2011): Definite NPs are only
applicable in situations where uniqueness is satisfied (+u). Indefinite NPs are applicable in
situations where uniqueness is not satisfied (−u)—if the implicature is drawn (+i)—or in
both kinds of situations—if the implicature is not drawn (−i). Finally, bare NPs in articleless
languages behave like English indefinites with the difference that the pertinent implicature
is never drawn because there is no definite article to trigger it.
Heim’s (2011) idea that bare NPs are essentially broadly applicable indefinites leaves
plenty of issues open (as we will see shortly, the empirical picture is complex), but it is
8 This type of account goes back to Hawkins (1981), who, however assumed the Russellian (1905)
semantics of definite descriptions. For reanalyses in terms of the Fregean (1892) analysis, see Heim
(1991) and Sauerland (2008).
9 The question why the anti-uniqueness implication arises in some cases (such as (7b)) but not in
others (such as (6b)) has not been, to the best of our knowledge, systematically investigated. One
possibility, briefly mentioned in 5.1, is that indefinites that introduce new discourse referents do not
really compete with definites and hence do not trigger this implicature.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
Definiteness, uniqueness, and maximality in languages with and without articles 317
Figure 1 Definite, Indefinite ±I(MPLICATURE), Bare
useful for us as the null hypothesis, whereby a single form (bare NP) corresponds to a single
meaning (existential quantifier) and where the lack of definite articles translates to the lack
of uniqueness or maximality implications.
3.2 Bare NPs correspond to definites (if singular) (Dayal 2004)
Dayal (2004) follows Chierchia (1998) in assuming that nominals in Russian are “born”
either as property-denoting or as kind-denoting (denotation 1 and 4 in Table 1, respectively).
The former meaning is directly applicable to bare NPs in predicative positions (9)
and the latter meaning is applicable to bare NPs such as those in (10).10
(9) a. Murka koška.
M. cat
‘Murka is a cat.’
λx[cat(x)](Murka) = cat(Murka)
b. Vasilisa i Murka koški.
V. and M. cats
‘Vasilisa and Murka are cats.’
λx[cats(x)](Vasilisa ⊕ Murka) = cats(Vasilisa ⊕ Murka)
(10) a. Koška byla priruˇcena 9000 let nazad.
cat was domesticated 9000 years ago
‘The cat was domesticated 9000 years ago.’
was domesticated 9000 years ago(∩cat)
b. Koški rodstvenniki tigrov.
cats related tigers.gen
‘Cats are related to tigers.’
related to(∩tigers)(∩cats)
The use of bare NPs in argument positions of entity-selecting predicates, as in (11), requires
more work in this approach. There are two hypothetical options: either the property
cat/cats shifts to an entity (iota-shift), yielding ιx cat(x) (the single cat there is) and
σx cat(x) (all the cats there are), respectively, or to an existential quantifier (ex-shift),
yielding λQ∃x[cat/cats(x)∧Q(x)]. These two shifts lead to the meanings of definites (11a)
and indefinites (11b), respectively.
10 An anonymous reviewer remarks that s/he finds it difficult to get the kind-reading for a singular NP.
Our consultant confirms that the plural is better than the singular, but does not consider the singular
unacceptable.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
318 R. Šimík & C. Demian
(11) Na stupenkax {sidela koška / sideli koški}.
on stairs sat.sg cat sat.pl cats
a. ‘{The cat was / The cats were} sitting on the stairs.’
sitting on the stairs(ι/σx cat(x) iota-shift
b. ‘{A cat was / Cats were} sitting on the stairs.’
∃x[cat/cats(x) ∧ sitting on the stairs(x)] ex-shift
Dayal (2004) argues, however, that this approach overgenerates: the range of truthconditions
derivable by the ex-shift is broader than attested. This is exemplified by (12)
(from Dayal 2004:405), where the bare NP—kot ‘cat’ or myš’ ‘mouse’ respectively—either
receives (i) a determinate reading or (ii) a narrow scoping existential reading, but crucially
not (iii) a wide scoping existential reading (a generalization corroborated by Geist 2010,
see below; yet, see Borik 2016 for a conflicting view). If ex-shift were a general interpretive
option, bare NPs should scope similarly as English indefinites do (see e.g. Abusch 1993).11
(12) a. Kot ne sidit na stule.
cat neg sits on stool
(i) ‘The cat is not sitting on the stool.’
(ii) ‘There isn’t any cat sitting on the stool.’ ¬ > ∃
(iii) *‘A cat is not sitting on the stool.’ *∃ > ¬
b. Mne kažetsja, ˇcto v komnate myš’.
me seems that in room mouse
(i) ‘The mouse seems to me to be in the room.’
(ii) ‘It seems to me that there’s a mouse in the room.’ seem > ∃
(iii) *‘A mouse seems to me to be in the room.’ *∃ > seem
Dayal’s (2004) response to this situation is to ban the ex-shift altogether: bare NPs in
articleless languages either refer to kinds (directly or by nom-shift, turning properties to
kinds) or to particulars (by iota-shift).12 This solves the problem with wide-scope readings
(iii), but creates a new one, namely how to account for narrow-scope readings (ii). This is
where Dayal’s (2004) story diverges for singulars and plurals and where our treatment of
the singular number as a definiteness correlate steps in. Consider her example (13) (Dayal
2004:406):13
(13) a. #Sobaka byla vezde.
dog.sg was everywhere
‘The dog/A particular dog (the same one) was everywhere.’
11 Some authors have taken issue with Dayal’s (2004) empirical generalizations about Russian. See,
e.g., Bronnikov (2006) or Borik (2016).
12 This stipulation is dubbed the REVISED MEANING PRESERVATION (Dayal 2004:419, revising Chierchia’s
1998 proposal) and it states that NOM-shift and IOTA-shift are preferred over EX-shift, which is thought
of as more complex because it introduces quantification. Since the two preferential shifts cover the
whole range of possible sorts (kinds and particulars), the EX-shift simply never applies.
13 Dayal’s (2004) translation of (13a) apparently contradicts her conclusion that bare NPs—if
(apparently) they correspond to indefinites—cannot be specific. We believe that the translation
‘a particular dog’ (as an alternative to ‘the dog’) reflects Dayal’s temporary agnosticism concerning
the interpretation of the bare NP; the only translation compatible with her final proposal is ‘the dog’
(modulo the issue of familiarity; see below).
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
Definiteness, uniqueness, and maximality in languages with and without articles 319
b. Sobaki byli vezde.
dogs.pl were everywhere
‘There were dogs (different groups) everywhere.’
Dayal’s (2004) interpretation of these facts (see also Dayal 2011) is the following. The
universal quantification in byla/i vezde ‘was/were everywhere’ is an integral part of the
predication and as such cannot be “underscoped” by genuine quantifiers. The reason
why sobaki ‘dogs’ appears to take narrow scope is that the existence quantification
stems from the very predicate—via the process of derived kind predication (DKP)
(see Chierchia 1998 for a formal definition), introducing existential quantification over
particular instances of a kind. DKP is applied in situations where a kind is fed into
a predicate that expects a particular entity: it is incoherent to say that a kind is located
in some particular position; see (14a). In this case, DKP salvages the computation by
introducing existential quantification over the instances of the kind denoted by sobaki
‘dogs’, see (14b) (∪∩dogs is the function that characterizes the set of instances of ∩dogs).
(14) a. ∀x[place(x) → in(x)(∩dogs)] sort clash
b. ∀x[place(x) → ∃y[∪∩dogs(y) ∧ in(x)(y)]] resolution by DKP
Let us now turn to singulars. Dayal (2004, 2011) proposes that while singular bare NPs
can denote kinds (cf. (10a)), they do not allow access to kind instances.14 For this reason,
the DKP is inapplicable to singulars and—after the exclusion of the ex-shift—the iotashift
(from properties; see (15b)) remains the only option to turn a bare singular into an
individual-denoting expression, see (15c). That, however, derives the absurd reading that
a single entity was in multiple places at the same time—a reading that is indeed attested.
(15) a. ∀x[place(x) → in(x)(∩dog)] sort clash
b. ∀x[place(x) → in(x)(dog)] type clash
c. ∀x[place(x) → in(x)(ιy dog(y))] resolution by iota-shift
In sum, bare plurals can receive narrow-scoping existential readings via DKP, but bare singulars
cannot—they refer to particular entities and thus correspond to definite descriptions.
This still leaves us with the question of how apparently narrow-scoping existential bare
singulars, cf. (12a-ii) and (12b-ii), are to be derived. Dayal (2004) offers the following
tentative answer: Apparently narrow-scoping existential bare singulars are determinate,
after all, only with two provisos: (i) determinate bare NPs, as opposed to definite
descriptions, do not presuppose that their referent is familiar (Dayal 2004:408–409); (ii) the
uniqueness presupposition—introduced by the iota-shift—might not be able to project
across operators such as negation (which could in turn be related to the lack of familiarity;
14 Dayal (2004) formulates this as a singleton requirement on the instantiation set. Dayal (2011:1100),
perhaps more intuitively, compares this to the difference between the players (≈ bare plural) and
the team (≈ bare singular) in sentences like The players / *The team live(s) in different cities (see
Barker 1992). We tentatively adopt the latter conceptualization of the issue.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
320 R. Šimík & C. Demian
see Dayal 2004: p. 405, fn. 10).15 The bottom line important for our purposes is that bare
singulars are expected to presuppose uniqueness (unless under operators), but bare plurals
are not expected to presuppose maximality; their determinate interpretation can be sidestepped
by DKP.
3.3 Bare NPs are ambiguous (if not topics) (Geist 2010)
Compared to Dayal (2004), Geist (2010) assumes a more permissive system, one where
the ex-shift is available, alongside the iota-shift.16 In other words, there is no absolute
prohibition on the indeterminate (= existential) interpretation on bare NPs. Yet, Geist
(2010) acknowledges that the distribution and interpretation of indeterminate bare NP is
more restricted than one would expect under this hypothesis. Firstly, Geist (2010) confirms
the observation that bare NPs—if indeterminate—cannot receive wide scope and/or specific
interpretations; see (12).17 This restriction is argued to follow from the failure to reduce the
domain of quantification to a singleton (Schwarzschild 2002). This process of what Geist
(2010) calls “referential anchoring” (following von Heusinger 2002) is claimed to only be
possible if there is an explicit determiner, also called “specificity marker”; see Ionin (2012).
An example of this is an unstressed version of the numeral odin ‘one’ (Geist 2010:211):18
(16) a. Džon xoˇcet ženit’sja na fracuženke.
John wants marry.inf on French.woman
‘John wants to marry a French woman (not a particular one).’
b. Džon xoˇcet ženit’sja na odnoj fracuženke.
John wants marry.inf on one French.woman
‘John wants to marry a French woman (a particular one).’
In addition, Geist is concerned with the observation that preverbal bare NPs—in contrast
to postverbal ones—cannot correspond to indefinites (Geist 2010:200).
(17) a. Kniga ležit na tom stole.
book.nom lies on that table
(i) ‘The book is lying on that table.’
(ii) Unavailable: ‘A book is lying on that table.’
b. Na tom stole ležit kniga.
on that table lies book.nom
(i) ‘The book is lying on that table.’
(ii) ‘A book is lying on that table.’
15 Nathan Klinedinst (JoS editor) uncovers caveats in Dayal’s reasoning. Firstly, if uniqueness fails
to project across negation, it is unclear how the reading (12a-i) is derived. Secondly, the status of
a determinate but narrow scoping (relative to negation) bare NP is questionable and it is unclear why
it should be paraphrased by a narrow scoping indefinite, as in Dayal’s (12a-ii). In our experiments, we
do not deal with scopal issues, so we refrain from trying to solve these caveats.
16 A comparable system is devised by Chierchia (1998).
17 Geist (2010:210) also notices that this restriction is not absolute and can be lifted by (post)modifying
the NP; she attributes this observation to Birkenmaier (1979). See also Borik (2016), who gives
examples of unmodified specific bare NPs.
18 As kindly pointed out by an anonymous reviewer, the bare NP in (16a) also affords the determinate
reading.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
Definiteness, uniqueness, and maximality in languages with and without articles 321
This effect of word order on the interpretation of bare NPs has been a well-known puzzle
for at least half a century and has been discussed by many authors and for many articleless
languages (Russian: Brun 2001; Chvany 1973; King 1995; Titov 2017; Czech: Biskup 2011;
Hlavsa 1975; Krámský 1972; Kuˇcerová 2007; Šimík & Burianová forthcoming; Polish:
Szwedek 1974, 2011; Chinese: Jenks 2018; Li & Thompson 1976; Matthews & Yip 1994;
Tagalog: Collins 2019; among many others).19 Geist (2010) captures the interaction as
a “restriction on indefiniteness” (rather than a requirement to be interpreted as a definite)
and derives it from a number of premises, listed in (18). Some of these are independently
motivated and argued for, others less so (particularly (18c)). The conclusion that one can
draw from the premises is in (19): Even though bare NPs can be indeterminate in general,
they must be non-specific, which in turn is incompatible with the requirement that topics
are specific.
(18) a. Bare NPs can be indeterminate (via ex-shift) or determinate (via iota-shift).
b. Indeterminate bare NPs must be non-specific (modulo modification; see fn. 17)
c. Preverbal bare NPs are (usually) topics.
d. Topics must be specific (in the sense of being determinate; Reinhart 1981).
(19) Preverbal bare NPs (if topical) must not be indeterminate.
Not all preverbal bare NPs are topical. Geist (2010) notices some cases where this is not
so. The case relevant for us involves so called thetic statements, understood by Geist (2010)
as statements with no topic–comment division (Sasse 1987). In a thetic statement, typically
involving an intransitive predicate, a bare NP argument is free to stand pre- or postverbally.
In both positions, it is the prosodically most prominent element of the sentence—it carries
so called sentence stress (marked by smallcaps)—and can be interpreted indeterminately,
see (20) (from Geist 2010:204).
(20) a. Meteorit upal. b. Upal meteorit.
meteorite came.down came.down meteorite
‘A meteorite came down.’ ‘A meteorite came down.’
In sum, Geist (2010) proposes that Russian bare NPs can be indeterminate or determinate.
If indeterminate, a bare NP is necessarily non-specific and as such cannot be the topic of
a sentence. Topicality is correlated with word order and prosody—topics are preverbal (or
possibly clause-initial) and do not carry sentence stress.20
3.4 Summary
We presented three types of accounts of the interpretation of bare NPs in articleless
languages, paying special attention to Russian. Heim’s (2011) idea is that bare NPs
19 It is largely an open question which factor exactly is responsible for the interaction: it could
be related to information structure (within that to topichood or givenness), syntax (within that to
a dedicated topic- or subject-position, or possibly to clause partitioning, cf. Diesing 1992 or Kuˇcerová
2007), or some other principle (Titov 2017). See Šimík & Burianová (forthcoming) for an overview and
corpus-based evidence favoring clause-initiality (as opposed to preverbal position or subjecthood)
as the source of bare NP determinacy.
20 Geist (2010) devotes a whole section to indeterminate topics (see Endriss 2009 for in-depth discussion),
showing that indeterminate topics are possible, but they must be either modified by the
specificity marker odin ‘one’ or at least involve another (post)modifier; see also footnote 17.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
322 R. Šimík & C. Demian
Table 2 Bare NPs = uniqueness/maximality predicted, ✗ = uniqueness/maximality not
predicted
Heim (2011) Dayal (2004) Geist (2010)
sg preverbal unstressed ✗
postverbal stressed ✗ ✗
preverbal stressed ✗ ✗
pl preverbal unstressed ✗ ✗
postverbal stressed ✗ ✗ ✗
preverbal stressed ✗ ✗ ✗
in articleless languages are always indeterminate, but due to the lack anti-uniqueness
implicatures, nothing prevents them from being used in situations where a definite
description would be appropriate. In contrast, Dayal (2004, 2011) proposed that bare
NPs are never genuinely existential. Bare singulars are always determinate and presuppose
uniqueness (unless embedded under negation or other operators). Any “indefiniteness
intuition” reported by speakers or reflected in translations is due to the fact that bare
NPs do not presuppose the familiarity of their referent. Bare plurals can receive quasiexistential
narrow-scoping readings via derived kind predication (Chierchia 1998), which
boils down to existential quantification over the instances of the kind denoted by the bare
plural. Finally, we discussed the proposal of Geist (2010), which brings into discussion
the traditional generalization that preverbal (clause-initial) bare NPs cannot correspond to
indefinites. We showed that Geist (2010) derives this generalization by relying on two main
premises: that preverbal bare NPs are topics (unless stressed, as in thetic statements) and
that topics must be specific. Since indeterminate bare NPs are always non-specific, they are
ruled out from the preverbal position / topical function.
For clarity, and in anticipation of our experiments, we represent the predictions in
Table 2, where the individual rows correspond to the properties of the grammatical subject
in our target utterances and, at the same time, to the individual experimental conditions.
The prediction profile is as follows: Heim (2011) does not predict any uniqueness/
maximality implications in bare NPs (null hypothesis); Dayal (2004) predicts uniqueness
in singulars, but not maximality in plurals (tested by experiment NUM; section 5.2);
Geist (2010) predicts uniqueness/maximality in preverbal unstressed bare NPs, but not in
stressed (pre- or postverbal) bare NPs (tested by experiments WO and PROS; section 5.3
and section 5.4, respectively).21 Note that even though the prediction profile is different
for each theory (and our experimental results can therefore non-trivially support Dayal’s
theory without supporting Geist’s theory, and vice versa), the individual predictions are not
necessarily in conflict with one another. It might well be the case that both predictions
get borne out by the results if, for instance, uniqueness/maximality interacts both with
21 There is no theory that distinguishes between the conditions POSTVERBAL STRESSED and PREVERBAL
STRESSED. Including both conditions allows us to make explorative inferences about the potentially
different contribution of the word order vs. prosodic factor.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
Definiteness, uniqueness, and maximality in languages with and without articles 323
Table 3 Definites vs. indefinites = uniqueness/maximality
predicted, ✗ = uniqueness/maximality not predicted
Standard account
def sg unstressed
stressed
pl unstressed
stressed
indef sg unstressed ✗
stressed ✗
pl unstressed ✗
stressed ✗
grammatical number and with word order and/or prosody. That, in turn, would suggest
that the eventual theory of definiteness correlates must be multifactorial.
For completeness, Table 3 captures the prediction of the standard uniqueness/maximalitybased
account of definite descriptions (experiment DEF; section 5.1). The prediction
is rather trivial and only concerns the definiteness factor, but it is not ruled out that
grammatical number or prosody affect the uniqueness/maximality implications in various
ways.22 Moreover, the inclusion of these factors achieves a better parallelism between the
Russian and the German experimental design.
4 OVERALL EXPERIMENTAL DESIGN
We devised an experiment where definiteness (correlates) and uniqueness/maximality
are treated as independent variables (linguistically and visually presented, respectively),
expected to interact in their effect on the dependent variable—the choice of visual vs.
covered picture. More particularly, we tested (i) the interaction between definiteness and
uniqueness/maximality in German and (ii) the interaction between definiteness correlates
(grammatical number, word order, and prosody) and uniqueness/maximality in Russian.
The existence of the latter interaction, esp. if it is comparably strong as the former interaction,
would lend support to the commonly entertained hypothesis that bare NPs in articleles
languages can convey the same meaning as definite descriptions in languages with articles,
provided certain conditions are met (i.e., if definiteness correlates are “activated”). We concentrate
on uniqueness and maximality as the most agreed-upon semantic notions behind
definite descriptions and corresponding bare NPs, singular and plural respectively. The core
experiments are supplemented by a number of filler experiments, testing other kinds of
presuppositions and implicatures. The fillers give us some confidence that the overall design
22 It is well-known, for instance, that in some contexts, maximality in plural definite descriptions can be
easily violated; see Malamud (2012) or Križ (2016) for discussion.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
324 R. Šimík & C. Demian
Table 4 Overview of experiment batches
Experiment batch Number of items ID numbers
1 MAIN 24 1–24
2 EXH 16 25–40
3 SCAL 8 41–48
4 BOTH 6 49–54
5 FOC 6 55–60
6 ONLY 4 61–64
7 ALSO 4 65–68
8 REST 4 69–72
total 72
worked well and at the same time provide grounds for various explorative comparisons,
which could in turn feed further hypothesis formulation and experimentation.23
4.1 Overall experimental setup and participant task
4.1.1 Experiment batches and number of items The whole experimental setup has two
languages mutations—German and Russian. The overall design for these two languages is
the same, although some factors could only be manipulated in a single language, namely
definiteness in German and position in the clause in Russian; these two factors correspond
to one another on the hypothesis that word order is a definiteness correlate (Geist 2010).
The items were constructed for German first and subsequently translated to Russian.
The overall experimental setup consists of 7 experiment batches and a few additional
fillers, totaling at 72 items, distributed as indicated in Table 4. The main batch (MAIN)
tested the interaction between uniqueness/maximality and definiteness (in German) and
definiteness correlates (in Russian). EXH tested how pragmatic exhaustification depends
on various factors such as word order, number, and definiteness; SCAL tested the scalar
implicature of ‘some’ (as compared to ‘all’); BOTH tested the cardinality inferences of
‘both’ as opposed to ‘two’; FOC tested the exhaustification brought about by “free” (nonassociated)
focus; ONLY tested the exhaustivity assertion brought about by focus associated
with ‘only’; finally, ALSO tested the additive presupposition of ‘also’.
The individual batches functioned as fillers for each other. This means that the items
from the MAIN batch were accompanied by twice as many (48) filler items. The role of the
fillers was (i) to distract the participants’ attention from the manipulations in MAIN, (ii) to
balance the overall proportion between the two values of the output variable (selection of
visible vs. covered picture; see below), and (iii) to provide additional data from which one
can draw post-hoc comparisons with the core experimental results.
23 The experiment (the MAIN batch) was preregistered in January 2018 at the Open Science
Framework; see https://osf.io/t7crf. All experimental materials, software, results, and analyses
are avilable at https://osf.io/b6dsh/?view_only=a0629414701243b080dbb0ec45b24069. The
experiment was preceded by a pilot on German, Czech and, with a modified design, on Polish (all
in Summer 2017). The core results of these pilot experiments are consistent with the German and
Russian ones reported here, lending independent empirical support to the present conclusions.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
Definiteness, uniqueness, and maximality in languages with and without articles 325
The overall number of items in the design (72) corresponds to the number of items that
every participant saw (24 from MAIN plus 48 from the other batches). The conditions
(see Tables 6 and 7 for conditions involved in MAIN) were distributed across lists using
Latin Square design and the order of presentation of individual batches and conditions
within them were pseudo-randomized in such a way that (i) two items from MAIN were
interrupted by at least one item from another batch and (ii) the same condition (from any
batch) was never presented twice in a row (even if interrupted by an item from another
batch; e.g., the order cMAIN
5 > cEXH
3 > cMAIN
5 , where cn is a specific condition, never came up).
Every participant saw every item exactly once, i.e., no item was presented in two different
conditions to one and the same participant.
4.1.2 Procedure and task The experiment is based on the so called covered box paradigm
(Huang et al. 2013), which has been used in two ways: (i) to detect less preferred or only
latently present interpretations and (ii) to detect less salient aspects of meaning such as
presuppositions. An example of the former approach is Zehr et al. (2017), who tested the
availability of local presupposition accommodation of again, which is clearly dispreferred
to the default global accommodation. An example of the latter approach is Zehr et al.
(2016), who tested for the presence of various presuppositions projected across negation
by comparing true but presupposition-violating conditions with a plainly true condition.
Our use of the paradigm lines up with the latter approach. In all our experimental items,
the visible picture is a truthful depiction of what the target sentence asserts, but, in the
critical conditions, the picture is in contradiction with what the target presupposes (by
hypothesis). By presenting the picture in combination with an additional covered picture
and by instructing the participants to be wary of slight mismatches, we hoped to encourage
them to concentrate not just on the target’s assertion, but on its less salient meaning
components. More particularly, we expected a higher proportion of covered picture choices
in cases where the visible picture contradicted the target’s presupposition as compared to
those where the presupposition matched the visible depiction.
In the present use of the paradigm,
1. the participant is presented with two “covered” pictures—one on the left, the other on
the right side of a computer screen (the participant is instructed that these pictures are
turned face down and hence are not visible).
2. While these pictures are covered, the participant listens (over headphones) to the audio
stimulus—a short narrative uttered by a single speaker, which consists of two to three
sentences, whereby the initial sentence(s) provide(s) a context in which the last (target)
sentence is supposed to be interpreted (below: ‘We threw away a lot of things during
the cleaning up of our cellar. The other/second trashbin is now full, too.’ – an item from
ALSO, item ID 67, target sentence boldfaced).
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
326 R. Šimík & C. Demian
3. After the audio stimulus, one of the pictures turns, as if randomly, face up and the
participant is asked to judge whether the visible picture is the correct depiction of the
last sentence; the participant has been instructed that exactly one of the two pictures is
the correct depiction, the other one being an imperfect match.24
4. The participant proceeds by pressing the left or right arrow, depending on which picture
she believes to be the correct one—the visible one or the one that remained covered
(below: participant pressed left arrow). The chosen picture is framed so as to confirm
the choice. If the covered one is chosen, it does not get disclosed.
5. The participant is asked to proceed by pressing spacebar. After it is pressed, the whole
procedure is repeated, starting with point 1.
The task was explained prior to the experiment using two examples: one in which
the visible picture was “correct” (selection of visible) and another one where it was
“incorrect” (selection of covered). The main dependent variable was picture choice
(covered vs. visible). The left- vs. right-sided picture disclosure was pseudo-randomized,
varied within items/conditions/subjects, and was balanced in frequency for each single
participant. Another variable we measured was reaction time but we did not include it in
the predictions of our hypotheses. The judgment was not speeded. The pace was individual
and the whole experiment took between 15 and 30 minutes.
4.1.3 Software, participants, and administration The experimental software was developed
by Christoph Demian, using Java. The experiments were carried out in universitybased
computer pools in Berlin (German) and in Moscow (Russian), mostly in January
24 The pictures were made by Christoph Demian.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
Definiteness, uniqueness, and maximality in languages with and without articles 327
2018, some remaining participants took part (in their homes) between February and July
2018. Each participant received a compensation of 5EUR (Berlin) or 300RUB (Moscow).
The participants were native speakers of the respective languages and were mostly university
students (of a variety of study programs). We did not control for age or gender, as these were
not expected to play any role. Data from 48 participants per language entered the analysis.
Participants were controlled for response quality: we only included those who performed in
the expected way in at least 6 out of 8 selected (filler) items (= 75% success threshold).25
For Russian, only 2 participants did not pass; for German, 3 did not pass. The excluded
participants were replaced by new ones, so as to arrive at the total of 48 participants
per language and thereby to achieve the same number of data-points for each individual
condition per experiment batch per language.
4.2 Design of MAIN
The MAIN experiment batch consisted of 24 items, which were manipulated in various
ways, so as to test for the interaction between uniqueness/maximality and definiteness
(German) or definiteness correlates (Russian). This batch provides data substrate for five
different subexperiments, each cutting the dataset in a slightly different way and thereby
addressing a different hypothesis. The hypotheses correspond to the predictions summarized
in section 3.4: (i) definite descriptions convey uniqueness/maximality (in German);
(ii) grammatical number of bare NPs affects the implications of uniqueness vs. maximality
(in Russian), (iii) word order in utterances with bare NPs affects uniqueness/maximality
implications (in Russian), (iv) prosody in utterances with bare NPs affects uniqueness/maximality
implications (in Russian). Finally, (v) we intended to compare the strength of the
expected interactions in languages with and without articles.
4.2.1 Manipulated factors and their levels Table 5 provides the complete overview of
the manipulated factors (typeset in smallcaps) and their levels (typeset in sans serif). The
word order factor, or the combined word order × prosody factor (manipulated in
Russian), stands in a correspondence relation to the definiteness factor (manipulated
in German) on one of the hypotheses (Geist 2010). For German, there are a total of 16
unique conditions (24); for Russian, due to the fact that word order and prosody do
not cross (condition PRED S & initial stress is not used because of its infelicity in the
kind of context used in the design), there are 12 unique conditions (3 × 22). Since there
were 24 items of MAIN, a Russian participant was exposed to each unique condition two
times (24 = 2 × 12). The situation was more complex for German, where a participant
was exposed to each unique condition at least once and half of the conditions twice
(24 = 16 + 8). Consequently, two German participants were needed in order to arrive
at the same number of datapoints for each unique condition (2 participants yielded 3
datapoints per condition). In total, we collected 96 datapoints per condition in Russian and
72 datapoints per condition for German. Finally, we would like to point out that not every
factor (and hence not all conditions) entered the analysis in each of the subexperiments, as
will become evident in section 5. Each individual subexperiment involved a 2 × 2 design,
25 The items used for this selection were all 4 items from ONLY and 4 items (in selected conditions) from
SCAL. These items/conditions were selected because they represent very clear contradictions (→
selection of covered) or very clear verifications (→ selection of visible).
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
328 R. Šimík & C. Demian
Table 5 MAIN: Fixed factors and their levels
Factor Levels Manipulation Comments
1 language German
Russian
within items /
between subjects
2 word order S PRED
PRED S
within items /
within subjects
only within Russian
3 definiteness definite
indefinite
within items /
within subjects
only within German
4 prosody final stress
initial stress
within items /
within subjects
no crossing with word order
5 number singular
plural
within items /
within subjects
6 picture +unique/maximal
−unique/maximal
within items /
within subjects
Table 6 A German item (ID 9) in all its unique linguistic conditions
Context: Die Lokomotive musste anhalten.
Target Definiteness Pros. prom. Number
1 Der Waggon hat sich abgekoppelt. def final stress sg
2 Ein Waggon hat sich abgekoppelt. indef final stress sg
3 Der Waggon hat sich abgekoppelt. def initial stress sg
4 Ein Waggon hat sich abgekoppelt. indef initial stress sg
5 Die Waggons haben sich abgekoppelt. def final stress pl
6 Waggons haben sich abgekoppelt. indef final stress pl
7 Die Waggons haben sich abgekoppelt. def initial stress pl
8 Waggons haben sich abgekoppelt. indef initial stress pl
with one factor being constant across subexperiments—picture—and the other differing
with the subexperiment—definiteness (experiment DEF), number (experiment NUM),
word order (experiment WO), and prosody (experiment PROS).26
4.2.2 Sample token set and the core properties of MAIN items One of the 24 MAIN items
(item ID 9) in all the conditions resulting from manipulating linguistic factors is shown in
Table 6 for German and Table 7 for Russian. Glossed examples in one of the conditions
are provided in (21). The context is the same for all conditions and is constructed in such
a way that it raises no expectations about the cardinality of the referents of the subject NP
in the target sentence. In our example item, for instance, the context provides no cue as
26 From this perspective, there were fewer unique conditions per subexperiment (4 = 2 × 2) and
a correspondingly larger number of datapoints per condition, particularly 288 ( = (24/4) ∗ 48) for
DEF and NUM and 192 ( = (16/4) ∗ 48) for WO, PROS. The lower number for WO and PROS is caused
by the exclusion of the PROSODY factor in WO (only the level final stress entered the analysis) and the
factor WORD ORDER in PROS (only the level S PRED entered the analysis).
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
Definiteness, uniqueness, and maximality in languages with and without articles 329
Table 7 A Russian item (ID 9) in all its unique linguistic conditions
Context: Lokomotiv dolžen byl ostanovit’sja.
Target Word order Pros. prom. Number
1 Otcepilsja vagon. PRED S final stress sg
2 Vagon otcepilsja. S PRED final stress sg
3 Vagon otcepilsja. S PRED initial stress sg
4 Otcepilis’ vagony. PRED S final stress pl
5 Vagony otcepilis’. S PRED final stress pl
6 Vagony otcepilis’. S PRED initial stress pl
to whether there is one or more carriages or in fact whether there are any at all. This is
important because the expected uniqueness/maximality inference must arise from the form
of the target NP in German (definite vs. indefinite) or from the form of the whole sentence
in Russian (e.g. S PRED vs. PRED S). This also means that uniqueness or maximality—if
they are present—are always accommodated, as there is no indication in the prior context
that they are satisfied. A related aspect of the design is that all parts of the target sentence
are “new” (in the technical sense of the word, i.e., “non-given”; for discussions of givenness,
see Rochemont 2016; Schwarzschild 1999; Wagner 2012; a.o.) and that the whole target
sentence represents the focus (assuming standard notions of information structure; see
Krifka 2008).27 The reason for this is twofold. First, we wanted to avoid any potential
interactions between definiteness and givenness or focusing (backgrounding). Second, it
was a crucial part of the design that the target sentence in some of the conditions (namely 1,
3, 4, and 6 in Russian) was thetic (lacking the topic–comment division) and thetic sentences
happen to be all new and broad focus.28, 29
(21) a. Die Lokomotive musste anhalten. Der Waggon hat sich abgekoppelt.
the locomotive had.to stop.inf the carriage has refl disconnected
27 An anonymous reviewer suggests that the target NP (‘carriage’) is in fact “given” due to its partwhole
bridging relation to the NP introduced in the context (‘locomotive’). What is important for our
purposes is that bridging only gives rise to givenness (as defined e.g. in Schwarzschild 1999) if it goes
from a part to its whole, but not conversely (Allerton 1978). In our items, the bridging went from the
whole to its (not necessarily unique) part.
28 We agree with an anonymous reviewer that word order and prosodic manipulations inuitively correlate
with information structure more than with definiteness or determinacy (and these correlations
have been successfully captured experimentally; see e.g. Šimík & Wierzba 2017). But this is precisely
the reason why prominent information structure manipulations, esp. the manipulation of focus and
givenness, must be avoided: focus/givenness would most likely override any uniqueness/maximality
implications (as the reviewer him/herself remarks). See also Titov (2017) for relevant discussion.
29 An anonymous reviewer notes that s/he finds the target sentence in condition 2 in Russian rather
infelicitous in the context provided because vagon was not previously mentioned. What is important
is that the reduced acceptability – if it is real – of the S PRED & final stress condition holds
independently of the PICTURE factor, so it is unlikely that it would affect the expected interaction.
Concerning the reported intuition, we could replicate it with some speakers but not others. What is
also worth mentioning is that we ran a pilot experiment on Polish, similar to the present one, but with
the dependent measure being acceptability, and there, the pertinent condition did not come out as
less acceptable.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
330 R. Šimík & C. Demian
Figure 2 An item (ID 9) in both PICTURE conditions (relative to NUMBER)
b. Lokomotiv dolžen byl ostanovit’sja. Otcepilsja vagon.
locomotive necessary was stop.inf.refl disconnected.refl carriage.nom
‘The locomotive had to stop. The/A carriage got disconnected.’
Each of the conditions is further combined with one of two possible pictures (two levels of
the picture factor): either +unique/maximal or −unique/maximal. In the former picture,
all the depicted potential referents of the subject satisfy the predicate (all carriages in the
picture are disconnected; in sg, this happens to be a single carriage); in the latter picture,
not all the depicted potential referents of the subjects satisfy the predicate (not all carriages
in the picture are disconnected); see Figure 2.
Let us now go through a step-by-step example of how the experimental item was
presented during the experiment (in a particular condition) and how the participant might
have reacted and why (see also the general procedure above).
1. The participant is presented with two “covered” pictures and at the same time listens
to the stimulus in (22), corresponding to condition definite & final stress & plural:
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
Definiteness, uniqueness, and maximality in languages with and without articles 331
(22) Die Lokomotive musste anhalten. Die Waggons haben sich abgekoppelt.
the locomotive had.to stop.inf the carriages have refl disconnected
‘The locomotive had to stop. The carriages got disconnected.’
2. After the stimulus is presented, one of the pictures gets disclosed, as depicted below. In
this picture, there are three carriages, two of which got disconnected and one remained
connected, corresponding to condition −maximal (= it does not hold that all the
carriages satisfy the predicate).30
3. At this point, it is up to the participant to decide which of the two pictures she
believes to contain the correct depiction of the target sentence (Die Waggons haben
sich abgekoppelt.) Our hypothetical participant decides that the visible picture is not
appropriate and selects the covered picture by pressing the left arrow. The selected
picture gets framed, in confirmation of the participant’s choice, but does not get
disclosed. (The participant now proceeds to the next item by pressing spacebar.)
We expect the picture choice to be influenced by the interaction between a linguistic factor
(here: definiteness) and the visual factor (picture). In the case at hand, the choice of the
covered picture is assumed to be motivated by the dissatisfaction with having a definite
description (die Waggons) refer to a non-maximal referent (two out of three carriages).
30 In this particular case, the predication is collective, which makes item 9 stand out. All the other
23 items involved distributive predication (pigs running away, light bulbs bursting, etc.). We did not
intend to manipulate collectivity/distributivity and only realized the special status of item 9 post-hoc.
Given the highly homogeneous nature of the items (23 out of 24 were distributive), we decided to
keep the data-points from item 9 in the dataset. Any effect of collectivity would be negligible anyway.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
332 R. Šimík & C. Demian
5 MAIN EXPERIMENTS
The MAIN experiment batch was designed to test the interaction between uniqueness/maximality
and various linguistic factors and to compare the strength of that interaction in
a language with and a language without articles. The MAIN experiment batch provides
data for five related experiments. Each of these experiments concentrates on a different
linguistic factor by either ignoring or excluding the manipulation of other factors. For
instance, if we are interested in testing the interaction between picture and definiteness,
we can disregard prosody by aggregating over the two conditions within the prosodic
manipulation (final vs. initial stress). Or, if we are interested in testing the interaction
between picture and prosody, without the interference of word order, we only consider
the prosodic manipulation (final vs. initial stress) within the relevant word order level
(namely S PRED), excluding the irrelevant level (PRED S) from the dataset.
5.1 DEF: Definiteness and uniqueness/maximality
The DEF experiment tests the following hypothesis:
DEF: Hypothesis
Definite descriptions in German (a language with articles) convey uniqueness (singular) or
maximality (plural).
Note that the hypothesis concerns uniqueness/maximality, but not their (not-)at-issue
nature, which motivates the neutral term “convey”. Indeed, even though we subscribe to
the presuppositional analysis of definite descriptions, our experiment is not designed to
distinguish it from other types of accounts.
Prediction and result If definite descriptions convey uniqueness (if singular) or maximality
(if plural), we expect to see an interaction between the factors definiteness and picture.
More particularly, we expect a higher proportion of covered picture choice in condition
definite & −unique/maximal, as compared to condition definite & +unique/maximal; no
such contrast is expected for condition indefinite. Less formally, participants should prefer
the covered picture if they are exposed to a definite description followed by a picture that
contradicts the uniqueness or maximality which the definite description—by hypothesis—
conveys. No such preference is expected for indefinites.31 This prediction is represented
in Figure 3 and the corresponding result in Figure 4. The columns indicate the proportion
of covered picture selection. What is relevant in the prediction is the difference between
definite & +unique/maximal and the other conditions, not the absolute proportion. The
result confirms the expected pattern: the definite & −unique/maximal condition is singled
out among the others in that the covered picture was selected significantly more often. More
31 In fact, if the anti-uniqueness implication is activated (see section 3.1), one would expect an effect
of PICTURE within the indefinite condition: the covered picture should be selected more in the
+unique/maximal condition than in the −unique/maximal condition. No such effect is observed in
the results.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
Definiteness, uniqueness, and maximality in languages with and without articles 333
Figure 3 DEF prediction
Figure 4 DEF result
technically, we see a significant interaction between picture and definiteness (z = 4.858,
p < .0001).32 Thus, our hypothesis is supported by the results:
32 All statistical analyses were performed using the Generalized linear mixed model by maximum
likelihood (Laplace Approximation), using the binomial setting and random intercepts for subjects
and items (unless explicitly stated otherwise). We used the lme4 package (Bates et al. 2015) of R
(R Core Team 2017).
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
334 R. Šimík & C. Demian
Figure 5 DEF result divided by NUMBER
DEF: Finding
Definite descriptions in German convey uniqueness/maximality.
It is good to point out that the data the result is based on involve aggregation over
different levels of factors not directly relevant to the present subexperiment, esp. over both
singulars and plurals and over stressed and unstressed subjects. Figures 5 and 6 demonstrate
that the crucial interaction is found in all the relevant data subsets, which means that
definite description convey uniqueness/maximality irrespective of grammatical number and
irrespective of the prosodic pattern.33
Discussion The result of DEF is in line with the common hypothesis that definite
descriptions convey uniqueness or maximality and it is robust across the different levels of
number (sg vs. pl) and prosody (final vs. initial stress). What seems and later proves to be
striking—anticipating the results of filler experiments (section 6) and the general discussion
(section 7)—is that the interaction between definiteness and uniqueness/maximality is
relatively weak: in the majority of the critical cases (58%), participants showed satisfaction
with definite descriptions used in situations where uniqueness or maximality were violated.
One could hypothesize that the weakness of the interaction is caused by the not-at-issue
nature of the uniqueness/maximality inference conveyed by definite descriptions. We will
come back to this point in the general discussion.
The result further clearly indicates that indefinite NPs in our design do not trigger the
anti-uniqueness implicature (Heim 1991, 2011; Sauerland 2008). More particularly, the
participants showed no dissatisfaction with encountering an indefinite NP followed by
a picture in which uniqueness/maximality is satisfied. We hypothesize that this is because the
33 We provide no statistics for post-hoc comparisons, relying only on the visual patterns, which are,
however, quite telling.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
Definiteness, uniqueness, and maximality in languages with and without articles 335
Figure 6 DEF result divided by PROSODY
indefinite functions as an introducer of a referent in the +unique/maximal condition and
therefore does not give rise to the anti-uniqueness implication. In other words, the indefinite
fulfills Heim’s (1982) novelty function (remember that the target referents are always
discourse-new in our design), thereby overriding any uniqueness/maximality inferences.
Related experimental evidence with a similar conclusion was recently offered by Bade &
Schwarz (2018).34
5.2 NUM: Grammatical number and uniqueness/maximality
The NUM experiment tests the following hypothesis:
NUM: Hypothesis
Singular bare NPs in Russian (an articleless language) convey uniqueness, as opposed to plural
bare NPs, which do not necessarily convey maximality (Dayal 2004, 2011).
If singular bare NPs convey uniqueness, but plural bare NPs do not necessarily convey
maximality, we expect to see an interaction between the factors number and picture.
More particularly, we expect a higher proportion of covered picture selection in condition
singular & −unique/maximal, as compared to condition singular & +unique/maximal;
no such contrast is expected for condition plural. Less formally, participants should prefer
the covered picture if they are exposed to a singular bare NP followed by a picture
that contradicts the uniqueness which the bare NP conveys (by hypothesis). No such
preference is expected for plural bare NPs. This prediction is represented in Figure 7 and the
34 An anonymous reviewer objects that “[i]f the indefinite is preempted by the definite article, then
an anti-uniqueness inference is predicted to arise, regardless of whether the indefinite introduces
a novel referent.” This statement is certainly true, the problem is that its protasis is apparently not
verified in the case at hand—the indefinite is just as felicitous as the definite and hence is clearly
not “preempted by the definite”.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
336 R. Šimík & C. Demian
Figure 7 NUM prediction
Figure 8 NUM result
corresponding result in Figure 8. As above, the columns indicate the proportion of covered
picture selection. The result does not confirm the prediction. Even though we find the
expected kind of interaction (between number and picture; z = −3.374, p < .001), it
goes in the unexpected direction: it is the plural condition that yields more covered picture
selection in the −maximal condition (as opposed to the +maximal condition); the singular
& −unique condition does not differ from the +unique baseline. In other words, plural
bare NPs exhibit a very slight tendency toward a maximizing interpretation, but singular
bare NPs exhibit no tendency toward expressing uniqueness. In sum:
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
Definiteness, uniqueness, and maximality in languages with and without articles 337
NUM: Finding
We found no evidence for the hypothesis that singular bare NPs in Russian convey uniqueness.
Unexpectedly, we found that plural bare NPs convey maximality, albeit very weakly so.
For the moment, it remains unclear if the observed effect of plurals on maximality is to be
attributed to some other factor, such as word order or prosody, because as before, we
aggregated over the different levels of these factors. It could be, for instance, that plural
bare NPs only trigger maximality if the NP is topical. We will turn to this issue shortly.
Discussion Dayal’s (2004) hypothesis that singular bare NPs in articleless languages
presuppose uniqueness clearly does not receive support from the present results. We
should point out that even though Dayal admits that in some cases uniqueness might
be absent, this is not to be expected in the case of our experiment: all of our items
involve episodic statements and the target bare NPs are never embedded under negation
or some intensional operator, which could, according to Dayal, lead to a suspension of the
uniqueness presupposition. Our experimental result is thus in line with Heim’s (2011) null
hypothesis that bare NPs are simply indeterminate. The effect of plurarity on maximality
will be discussed in the general discussion (section 7).
5.3 WO: Word order and uniqueness/maximality
The WO experiment tests the the following hypothesis:
WO: Hypothesis
Preverbal (≈ topical) bare NPs in Russian convey uniqueness/maximality, as opposed to
postverbal (≈ non-topical) bare NPs (Geist 2010, among others).
For the purpose of this subexperiment, we included only a proper subset of the data
gained from MAIN, in particular conditions S PRED & final stress and PRED S & final
stress. Data from condition S PRED & initial stress did not enter the analysis. In other
words, the factor under investigation was word order (S PRED vs. PRED S); prosodic
prominence remained constant (sentence final = default). If topicality (expressed by word
order) conveys uniqueness/maximality, we expect an interaction between the factors word
order and picture. More particularly, we expect a higher proportion of covered picture
selection in condition S PRED & −unique/maximal, as compared to condition S PRED
& +unique/maximal; no such contrast is expected for condition PRED S. Less formally,
participants should prefer the covered picture if they are exposed to a sentence with
a preverbal bare NP followed by a picture that contradicts the uniqueness/maximality
brought about by the preverbal position (by hypothesis). No such preference is expected
for sentences with postverbal bare NPs because these are hypothesized to be underspecified
for (in)determinacy. The prediction is represented in Figure 9 and the corresponding result
in Figure 10. The result does not confirm the prediction: there is no significant interaction
between word order and picture (z = −.302, p = .76).
WO: Finding
We found no evidence for the hypothesis that preverbal bare NPs in Russian convey uniqueness
or maximality.
While the expected interaction is not confirmed, we find a main effect of picture (z =
−3.443, p < .001). In other words, participants exhibited a very slight tendency towards
interpreting bare NPs as one would interpret definites (independently of the manipulation
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
338 R. Šimík & C. Demian
Figure 9 WO prediction
Figure 10 WO result
of word order). A post-hoc analysis, whose results are shown in Figure 11, shows that
this effect is to be attributed to plural bare NPs only (i.e., there is an interaction between
number and picture within both S PRED and PRED S), in line with the results of NUM.
Discussion The result fails to confirm the expected effect of word order on uniqueness/
maximality of bare NPs. There are multiple ways of interpreting this result. The most radical
one is that the intuition expressed by so many authors in the past (see section 3.3) is a mere
illusion: there simply is no correspondence between word order manipulation in articleless
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
Definiteness, uniqueness, and maximality in languages with and without articles 339
Figure 11 WO result divided by NUMBER
languages and (in)definiteness in languages with articles. We find this conclusion unlikely,
not just because we trust other linguists’ (and, in fact, our own) native intuitions, but also
because the existence of this correspondence recently received support in a corpus study on
Czech (Šimík & Burianová forthcoming).35 We can think of a number of reasons why we
failed to find the expected interaction.
First, it is possible that the word order manipulation failed to be interpreted by the
participants as a manipulation of topicality—the hypothesized factor behind the expected
definiteness effect. This could, for instance, be the case if theticity is much more contextdependent
than sentence-form-dependent. In our design, the target sentence was all new (it
contained no given element) and even though topics can be new (a fact recognized ever since
Reinhart 1981), there is a possibility that new topics are less likely to be perceived as topics
per se. Needless to say, this state of affairs would go counter the common assumption,
as well as Geist’s (2010) assumption, that preverbal unaccented bare NPs are topical,
independently of context.
Second, it is possible that the pertinent word order manipulations do correspond to
definiteness manipulations, but uniqueness/maximality remain unaffected by it. This, of
course, goes counter to Geist’s (2010) but also Dayal’s (2004) assumption about how bare
NPs are construed. These authors, among many others, assume that uniqueness/maximality
are involved in the derivation of determinate bare NPs. Setting these approaches aside,
there are, of course, quite a few alternative notions that have been proposed to capture the
meaning of definiteness, including familiarity (used in both functional and formal tradition
and understood in various ways, see e.g. Christophersen 1939; Heim 1982; Roberts 2003),
35 Notice also that we do not expect there to be a relevant difference between Russian and Czech. As
already pointed out in footnote 23, a pilot experiment on Czech—with a very similar design to the
present one—yielded analogous results.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
340 R. Šimík & C. Demian
identifiability (e.g. Givón 1984; Lyons 1999), non-ambiguity (e.g. Löbner 1985), or
salience (e.g. von Heusinger 1997). These notions, at least in some approaches, can do
without uniqueness/maximality: a bare NP can refer to a familiar/identifiable/salient/etc.
individual (and thus be interpreted as determinate) without presupposing its uniqueness
or maximality. Considering this on the background of our experiment, it seems coherent
to assume that after the picture is disclosed, the participant can identify the referent that
satisfies the predicate, irrespective of whether uniqueness/maximality is satisfied or not.
A different experimental design would be needed to test whether any of the abovementioned
alternative notions is suitable to characterize determinately used bare NPs in
articleless languages or, for that matter, definite descriptions in languages with articles.
Notice that while we have been able to detect uniqueness/maximality as a component of the
meaning of German definite descriptions, it is possible that it is not the only component.36
Third, it is possible that the hypothesized interaction exists, but our experimental design
was not sensitive enough to capture it. This is not completely implausible, given that the
corresponding interaction found in DEF was present, but relatively weak. As we will see in
section 6, however, the design was powerful enough to even capture various conversational
implicatures. If uniqueness/maximality is a presupposition, which is considered more salient
than an implicature, the hypothesized interaction should be visible.
5.4 PROS: Prosody and uniqueness/maximality
The PROS experiment tests the following hypothesis:
PROS: Hypothesis
Unstressed preverbal (≈ topical) bare NPs in Russian convey uniqueness/maximality, as opposed
to stressed preverbal (≈ non-topical) bare NPs (Geist 2010, among others).
For the purpose of this experiment, we included data from conditions S PRED & initial
stress and S PRED & final stress. Data from condition PRED S were not included.
In other words, the factor under investigation was prosody (initial stress vs. final
stress), word order remained constant—the subject was always preverbal. If topicality
(expressed by the placement of prosodic prominence) conveys uniqueness/maximality,
we expect an interaction between the factors prosody and picture. More particularly,
we expect a higher proportion of covered picture selection in condition final stress &
−unique/maximal, as compared to condition final stress & +unique/maximal; no such
contrast is expected for condition initial stress. That is, participants should prefer the
covered picture if they are exposed to sentence with an unstressed (topical) bare NP
followed by a picture that contradicts the uniqueness/maximality brought about by the
lack of stress (by hypothesis). No such preference is expected for sentences with stressed
bare NPs because such sentences are hypothesized to be thetic, the bare NP in them nontopical,
and hence without any impact on uniqueness/maximality inferences. The prediction
is represented in Figure 12 and the corresponding result in Figure 13. The result does not
confirm the prediction: there is no significant interaction between prosody and picture
(z = .131, p = .9).
36 We are aware of one experimental study that compares uniqueness and salience in their capacity
to stand for definiteness, namely Bosch et al. (2011). These authors conclude that “[s]alience is only
a poor substitute for uniqueness”.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
Definiteness, uniqueness, and maximality in languages with and without articles 341
Figure 12 PROS prediction
Figure 13 PROS result
PROS: Finding
We found no evidence for the hypothesis that unstressed preverbal bare NPs in Russian convey
uniqueness or maximality.
As in WO, also here we find a main effect of picture (z = 5.696, p < .0001)—a very
slight but general tendency to interpret bare NPs similarly as one would interpret definites
(independently of the manipulation of prosody). And also in this case, we see that the effect
is to be attributed to plural bare NPs; see Figure 14.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
342 R. Šimík & C. Demian
Figure 14 PROS result divided by NUMBER
Discussion The result of this experiment reinforces the result of the WO experiment:
we find no evidence for the hypothesis that uniqueness/maximality depend on topicality,
whether manipulated by word order (WO) or by prosody (PROS). At the same time, the
same interpretational caveats apply. It remains an open question (i) whether we succeeded
at manipulating topicality (rather than just prosody), (ii) whether uniqueness/maximality
are involved in determinate interpreations of bare NPs, and (iii) whether the experimental
design is sensitive enough to capture the intended interaction. For more discussion, we refer
the reader to the discussion of the WO results.
5.5 ART: Presence/Absence of articles and uniqueness/maximality
The ART experiment seeks to answer the following question:
ART: Question
Do definite articles in German interact with uniqueness and/or maximality more strongly than
definite correlates in Russian do?
Technically speaking, we expect a three-way interaction between the type of definitenessmarking
strategy (articles in German vs. definiteness correlates in Russian), the presence
or absence of the relevant marker, and uniqueness/maximality. If German articles encode
uniqueness/maximality better than Russian definiteness correlates, we expect a more
pronounced increase in covered picture selection in the critical condition (def & +u/m), as
compared to the control conditions, than in the relevant corresponding critical conditions
in Russian (depending on the definiteness correlate: sg & +u/m, S PRED & +u/m, and
final stres & +u/m). In order to perform the statistical tests, we re-coded the data so as
to bring the individual definiteness correlate factors in Russian on a par with definiteness
in German, resulting in the following correspondences: (i) number: sg ≈ def, (ii) word
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
Definiteness, uniqueness, and maximality in languages with and without articles 343
Figure 15 DEF ≈ NUMBER
order: S PRED ≈ def, and (iii) prosody: s PRED ≈ def. The comparisons are visualized
in Figures 15, 16, and 17, respectively. For all three tests, the expected interaction came out
as significant: (i) z = 5.353, p < .0001, (ii) z = 3.064, p < .003, (iii) z = 2.725, p < .007.37
We can conclude:
ART: Finding
Definiteness in German interacts with uniqueness/maximality more strongly than any of the
definiteness correlates in Russian.
Discussion The result is not surprising in the light of the previous experiments, where
we found an interaction between definiteness and uniqueness/maximality, but not between
definiteness correlates and uniqueness/maximality, except in the NUM experiment, where,
however, the interaction went in the unexpected direction. Interestingly, upon performing
a post-hoc test, based on a re-coding of the number-related data, such that pl ≈ def, we find
no significant interaction between the strategy, presence of the marker (here: def and pl),
and uniqueness/maximality (z = 1.336, p = .18). This means that, as far as our experiment
is concerned, definiteness in German behaves on a par with plurality in Russian. More
theoretical and empirical work must be done in order to understand this puzzling post-hoc
result.
37 As before, the tests were performed using random intercepts for subjects and items, the exception
being test (i), which only used random intercepts for items, due to non-convergence of the more
complex model.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
344 R. Šimík & C. Demian
Figure 16 DEF ≈ WORD ORDER
Figure 17 DEF ≈ PROSODY
6 FILLERS
The fillers to the MAIN items were distributed among a number of experiment batches,
listed above in Table 4, and they tapped into a whole range of inferences: from purely pragmatic
or focus-induced exhausitifcation (EXH, FOC), through scalar implicatures (SCAL),
determiner- or particle-triggered presuppositions (BOTH, ALSO), to at-issue exhaustive
assertions (ONLY). The purpose of including all these inference types was to provide
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
Definiteness, uniqueness, and maximality in languages with and without articles 345
grounds for comparing the size of the effect brought about by uniqueness/maximality
violations (MAIN), which are—by hypothesis—contradictions of presuppositions, and the
effects brought about by various other types of violations (contradictions of implicatures,
presuppositions, or assertions). Below, we provide the results of most of the filler
experiments, but no associated predictions or formal analyses of the results. The discussion
is informal and explorative, but we consider it important for the general discussion and
valuable in that it generates new questions, which could feed new experimentation.
6.1 ONLY
The ONLY experiment tested the participants’ sensitivity to the contradictions of the
exhaustivity assertion introduced by the focus sensitive particle ‘only’ (nur in German,
tol’ko in Russian) or, more particularly, situations where a sentence of the form ‘only x
is such that P(x)’ is accompanied by a picture depicting that P(x), but also that P(y) (where
x = y). The manipulated factor picture had two levels +exhaustive and −exhaustive. In
the former case, the picture matched the exhaustivity of the target sentence with ‘only’; in
the latter case, it did not: there were additional elements in the picture which contradicted
the exhaustivity of the target sentence. The factor was manipulated between items; three (of
the four) items were −exhaustive and one was +exhaustive.38 An example item (ID 61) in
the −exhaustive condition is provided below—the German and Russian linguistic stimuli
are in (23) and the visual stimulus is in Figure 18.39
(23) a. Hanna und Nils bekamen die Aufgabe Obst zu malen. Hanna hat nur einen
H. and N. got the task fruit to paint H. has only an
Apfel gemalt.
apple painted
‘Hanna and Nils got the task to paint fruit. Hanna only painted an apple.’
b. Anja i Kolja poluˇcili zadanie narisovat’ frukty. Anja narisovala tol’ko
A. and K. got task draw.inf fruit A. drew only
jabloko.
apple
‘Anja and Kolja got the task to draw fruit. Hanna only drew an apple.’
Figures 19 and 20 show the results of ONLY for both languages. The manipulated variable
picture clearly has a strong effect on the outcome variable: the participants selected the
covered picture in the −exhaustive condition in 97% of the cases, which is expected,
given the fact that a contradiction arises in the assertion.
6.2 ALSO
The ALSO experiment tested the participants’ sensitivity to contradictions of the additive
presupposition introduced by the focus sensitive particle ‘also’ (auch in German, tože in
38 The disproportion was motivated by the aim to achieve a rough balance of matching and nonmatching
visible pictures within the whole experiment. The same disproportion is present in the
batches ALSO and REST.
39 Note that while the assertion is false, in view of the picture, what is sometimes considered the
presupposition, namely ‘Hanna painted an apple’, is true. This tension was present in most of the
experimental manipulations within the whole design.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
346 R. Šimík & C. Demian
Figure 18 Visual stimulus for item 61 from ONLY (condition −exhaustive)
Figure 19 ONLY result German
Russian). The manipulated factor picture had two levels +additive and −additive. In the
former case, the picture reflected the additivity expressed in the target sentence with ‘also’;
in the latter case, it did not: the presupposition introduced by ‘also’ was not verified in the
picture. As in ONLY, the factor was manipulated between items, whereby three (of the four)
items were −additive and one was +additive. An example item (ID 66) in the −additive
condition is provided in (24) and Figure 21.
(24) a. Als Julia und Thomas spazieren gingen, is es deutlich kälter geworden.
when J. and T. walk.inf went is.aux it clearly colder become
Thomas hat auch eine Mütze aufgesetzt.
T. has also a cap put.on
‘When Julia and Thomas went for a walk, it turned quite cold. Thomas put
on a cap, too.’
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
Definiteness, uniqueness, and maximality in languages with and without articles 347
Figure 20 ONLY result Russian
Figure 21 Visual stimulus for item 66 from ALSO (condition −additive)
b. Julja i Saša guljali, kogda stalo xolodnee. Saša tože nadel šapku.
J. and S. walked when became colder S. also put.on cap
‘Julja and Saša were walking and it turned cold. Saša put on a cap, too.’
Figures 22 and 23 show the results. The manipulated variable picture has a strong effect
on the outcome variable: the participants selected the covered picture in the −additive
condition in 96% (German) and 86% (Russian) of the cases. What is striking is that
the effect size is comparable to the one observed for ONLY, despite the fact that what
is contradicted here is a presupposition (rather than an assertion). It is also worth
mentioning that the effect is apparently much stronger than the effect observed for
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
348 R. Šimík & C. Demian
Figure 22 ALSO result German
Figure 23 ALSO result Russian
uniqueness/maximality presupposition contradictions in DEF (which led to 42% covered
picture selection).
6.3 BOTH
The BOTH experiment tested the participants’ sensitivity to contradictions of the cardinality
assertion/presupposition/implicature associated with the numerals ‘both’/‘two’ (beide/zwei
in German and oba/dva in Russian). The two manipulated factors (both within items) were
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
Definiteness, uniqueness, and maximality in languages with and without articles 349
Figure 24 Visual stimuli for item 49 from BOTH (left to right: conditions 2-pic, 2+1-pic, 3-pic)
numeral type (both vs. two) and picture (2-pic vs. 2+1-pic vs. 3-pic), giving a total of
6 conditions. One item (ID 49) in all its conditions is exemplified in (25) and Figure 24.
(25) a. Das Weibchen flog zu ihrem Nachwuchs zurück. Beide / Zwei Jungvögel waren
the female flew to its offspring back both two chicks were
im Nest.
in.the nest
‘The female bird flew back to its offspring. Both / Two birds were in the nest.’
b. Samka letela nazad k svoemu potomstvu. Oba / Dva ptenca byli v gnezde.
female flew back to its offspring both two chicks were in nest
‘The female bird flew back to its offspring. Both / Two birds were in the nest.’
Figures 25 and 26 show the results. The manipulated variables and their interactions have
very clear and very similar effects on picture selection in both languages. Leaving a formal
analysis aside, let us just mention a few immediately clear impressions. In both languages,
the conditions both & 2-pic, two & 2-pic, and two & 2+1-pic give little excuse to select
the covered picture. This is expected because none of these three conditions violates any
sensible inference of the associated sentence. Assuming that ‘both’ is a universal quantifier
presupposing a restrictor with cardinality 2 (Schwarzschild 1996), the picture in condition
both & 3-pic contradicts the presupposition of the associated sentence (it does not hold
that there are two chicks) and the picture in condition both & 2+1-pic contradicts both
the presupposition and the assertion (it does not hold that all the chicks are in the nest). As
one would expect, both conditions receive a high proportion of covered picture selection.40
What we find interesting and in some sense surprising is condition two & 3-pic, which
exhibits a contradiction of a scalar implicature (the implicature being ‘There are not more
than two birds in the next’). This contradiction produces an effect as strong as or even
stronger than the presupposition and/or assertion contradictions.
6.4 SCAL
The SCAL experiment tested the participants’ sensitivity to contradictions of the implicature
triggered by the determiner ‘some’ (einige in German and nekotorye in Russian), in
comparison to contradictions of the assertion conveyed by the determiner ‘all’ (alle in
40 The small but crosslinguistically consistent decrease in covered picture selection in condition both
& 2+1-pic as opposed to both & 3-pic might be due to a successful accommodation of the nucleus
into the restrictor, resulting in ‘Both chicks that are in the nest are in the nest’, which, albeit trivial,
does not lead to contradictions in either assertion or presupposition.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
350 R. Šimík & C. Demian
Figure 25 BOTH result German
Figure 26 BOTH result Russian
German and vse in Russian). The two manipulated factors (both within items) were
determiner (some vs. all) and picture (+all vs. −all), giving a total of 4 conditions.
One item (ID 47) in all its conditions is exemplified in (26) and Figure 27.
(26) a. Mit dem schwarzen Auto kamen drei Gangster. Alle / Einige trugen einen
with the black car came three gangsters all some wore a
Hut.
hat
‘Three gangsters arrived in the black car. All / Some wore a hat.’
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
Definiteness, uniqueness, and maximality in languages with and without articles 351
Figure 27 Visual stimuli for item 47 from SCAL (left +all, right −all)
Figure 28 SCAL result German
b. Na ˇcërnoj mašine priexali tri gangstera. Vse / Nekotorye byli v šljapax.
in black car came three gangsters all some were in hats
‘Three gangsters arrived in the black car. All / Some wore a hat.’
Figures 28 and 29 show the results. There is a very clear interaction between the two
manipulated factors. The high proportions of covered picture selection in conditions all &
−all and some & +all indicate the participants’ sensitivity to the assertion contradiction
in the former case and to the scalar implicature contradiction in the latter case. The
slight difference between the two conditions in German goes in the expected direction: the
violation of assertion contradiction is more pronounced than the violation of implicature
contradiction, reflecting the fact that drawing a scalar implicature is optional. In Russian,
the results for these conditions are virtually indistinguishable, indicating a very high
sensitivity to implicature contradiction.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
352 R. Šimík & C. Demian
Figure 29 SCAL result Russian
6.5 EXH
The EXH experiment tested the participants’ sensitivity to contradictions of exhaustivity
implicatures and their dependency on factors similar to the ones manipulated in MAIN.
The aim was to provide material for explorative comparisons between the effects of the
relevant factors on uniqueness/maximality, notions that are deemed to be presuppositional,
and those on exhaustivity, which is—at least in the case at hand—conversational (based on
the relevance implicature). The manipulated factors were number (sg vs. pl; manipulated
between items), word order (S PRED vs. PRED S; within items), definiteness (def vs.
indef; within items; manipulated only in German), determinedness (bare vs. determined;
within items; manipulated only in Russian and distinguishing between bare NPs and
NPs determined by the indefinite determiner kakoj-to), and picture (+exhaustive vs.
−exhaustive), totalling at 16 unique conditions per language.41 One item (ID 29) is shown
in all its conditions in (27)/(28) and in Figure 30. (All the illustrated conditions are in
condition sg, since number was manipulated between items.) The picture on the left
represents the +exh condition because the corresponding linguistic stimulus mentions all
the items ready for painting (here: the paint roller is all there is). The picture on the right
represents the −exh condition because the linguistic stimulus does not mention all the items
(the bucket is not mentioned, even though it is relevant).
(27) Wir hatten vor zu streichen.
we had before to paint
‘We were planning to paint.’
41 The factor NUMBER corresponds to a loose notion of semantic number in this experiment, such that
‘home shoes’ (item 25) count as singular (corresponding to ‘a pair of shoes’). Yet, most of the items
were not problematic in this way.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
Definiteness, uniqueness, and maximality in languages with and without articles 353
Figure 30 Visual stimuli for item 29 from EXH (left +exh, right −exh)
a. Die / Eine Farbrolle lag schon auf dem Boden bereit. S PRED & def/indef
the a paint.roller lay already on the floor ready
‘The / A paint roller was ready on the floor.’
b. Auf dem Boden lag schon die / eine Farbrolle bereit. PRED S & def/indef
on the floor lay already the a paint.roller ready
‘The/A paint roller was ready on the floor.’
(28) My xoteli pokrasit’ steny.
we wanted paint.inf walls
‘We wanted to paint the walls.’
a. (Kakij-to) Valik uže ležal na polu.
some paint.roller already lay on floor S PRED & bare/det
‘The/A paint roller was ready on the floor.’
b. Na polu uže ležal (kakij-to) valik.
on floor already lay some paint.roller PRED S & bare/det
‘The/A paint roller was ready on the floor.’
The results, provided in Figures 31 and 32, show in both languages a consistent effect of
picture: condition −exh yields more covered picture choices than +exh. This reflects
the tendency to pragmatically strengthen the interpretation from, say, ‘the/a paint roller
was ready’ to the exhaustive ‘only the/a paint roller was ready’. The fact that the effect is
present across all individual conditions suggests its default status, but it is also apparent
that individual factors can further contribute to drawing this exhaustivity implicature
and that this can happen in a cumulative fashion, particularly in German. (The Russian
pattern is less clear and consistent, so we leave it aside.) The first thing to notice is that
exhaustification is consistently more pronunced in PRED S than in S PRED. A possible
explanation is optional accommodation of focus on subject in PRED S (allowed by the fact
that in this condition, the subject carries sentence stress); narrow focus, in turn, is likely to
contribute to its exhaustive interpretation. We further see that plurals are more liable to get
exhaustified than singulars and within plurals, indefinites (thus: bare NPs) are more liable
to exhaustification than definites.
7 GENERAL DISCUSSION
Our experiments bear implications for three areas of linguistic research: the semantics of
definite descriptions, the semantics of bare NPs, and cross-linguistic NP semantics. We will
discuss these implications in turn.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
354 R. Šimík & C. Demian
Figure 31 EXH result German
Figure 32 EXH result Russian
7.1 The semantics of definite descriptions (DEF)
The basic prediction of the uniqueness/maximality theory of definite descriptions was
borne out by the DEF experiment, which exhibits consistent sensitivity to violations of the
uniqueness/maximality inference—by hypothesis a presupposition—of definite descriptions.
What came as a surprise was the relative weakness of the expected interaction: in only 42%
of the critical cases (definite & −unique/maximal) did the violation lead to the expected
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
Definiteness, uniqueness, and maximality in languages with and without articles 355
picture selection (covered); in other words, the participants tolerated the violations in the
majority of the critical cases. Our filler experiments confirmed the surprising nature of
this result. Other presupposition violations were recognized as such in more than 90%
of the cases, whether the presuppositions were triggered by a determiner (BOTH) or
by a focus-sensitive particle (ALSO). These violations led to effects comparable to those
of assertion contradictions (ONLY). Interestingly, also violations of scalar implicatures
(SCAL) induced a much stronger effect than the one observed in DEF. If anything, the
strength of the interaction in DEF matches the effect found in EXH, which tested pragmatic
exhaustive implicatures. This means that something like The boat sank was felt to be
similarly (in)appropriate as a description of a picture in which there are two boats, one
of which sank (violation of uniqueness), as (in)appropriate it was to use it as a description
of a picture where the (single) boat sank, but a/the water-scooter sank, too, which violates
the Grice’s (1975) maxim of quantity because the sunken water-scooter should also have
been mentioned.
We can think of a number of reasons why the interaction in DEF was relatively weak,
namely (i) that the design worked differently than intended, (ii) that uniqueness/maximality
are not presuppositions but implicatures, (iii) that the discourse structure of the items was
such that it backgrounded the relevance uniqueness/maximality inferences significantly,
(iv) that some definite NPs were construed as weak definite NPs, and, finally, (v) that the
grammatical nature of definiteness renders its semantics less salient. We will discuss these
reasons one by one. Notice, by the way, that they are not incompatible with one another,
they might in fact act in combination to produce the resulting weak interaction.
According to Coppock & Beaver (2015), definite NPs can have both determinate
and indeterminate readings. Only in the former case do they presuppose uniqueness (or
maximality). Even though the conditions under which indeterminate readings of definite
NPs arise in their account are quite different from those used in our experimental design,
it cannot be ruled out that at least some definites were interpreted indeterminately by some
participants. An anonymous reviewer pointed out that this interpretation might in fact be
prompted by the covered box design. As stated in section 4.1, the covered box paradigm has
not just been used for tapping into presuppositions and implicatures (our intention here),
but also into dispreferred interpretations. It is possible that participants selected the visible
picture in the critical condition precisely because it matched the dispreferred (indeterminate)
interpretation of the target definite NP. We consider this interpretation implausible in the
light of the filler results. Consider SCAL, for instance (section 6.4). Indefinite ‘some’ NPs are
in principle compatible with two readings: the logical reading, compatible with a situation
where all entities satisfy the nucleus (‘some and possibly all’), and the pragmatically enriched
reading, amounting to ‘some but not all’. It seems clear that the latter reading is preferred
and the former dispreferred. If dispreferred readings were prompted by our design, we
would expect a relatively high proportion of visible picture choices in the critical some &
+all condition. Yet, the visible picture was chosen in only 25% of the cases (cf. the 86%
of visible choices in the some & −all condition), suggesting that the participants mostly
stuck to the pragmatically enriched meaning, despite being prompted by the visible picture
towards the (dispreferred) logical interpretation.
The similarly small effect size found in DEF and in EXH could suggest a conversational
source of the uniqueness/maximality implication conveyed by definite descriptions
(assuming a correlation between inference cancellability and comparatively small effect size;
see e.g. De Veaugh-Geiss et al. 2018). That in turn would be in line with the proposals
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
356 R. Šimík & C. Demian
of Szabó (2000) and Ludlow & Segal (2004). The problem with this reasoning is the
striking difference in effect size between DEF and EXH on the one hand and SCAL on
the other. If the effects in all these experiments were caused by violations of cancellable
conversational implicatures, one might expect them to have similar effect sizes. On the other
hand, the pattern would be expected if definiteness-related uniqueness/maximality were
conversational and quantifier-related scalar inferences conventional, as argued by Chierchia
(2006) and many others since then.
Another factor that might have affected the effect sizes in DEF is the discourse relation
between the context and the target sentence. It is a notable property of the MAIN items
(as opposed to most of the filler items) that the context is incomplete without the target
because it raises an implicit and in some cases even explicit question.42 For instance, in
item 1 of MAIN, the participant first learns that something happened at the butcher’s (=
context), raising the implicit question ‘What was it that happened?’. Immediately after that,
the participant learns that a/the pig(s) ran away (= target), which answers this question.
The question raised by the context would be answered irrespective of how many pigs there
are and how many of them ran away. It is the fact that some pig(s) ran away (or in fact that
something notable happened) that seems to matter most. Importantly, this consideration
holds of every target sentence and every picture of MAIN, irrespective of the condition. The
high proportion of ignored violations of uniqueness/maximality could therefore reflect the
fact that some relevant aspect of both the target sentence and the picture (= some pig ran
away) jointly and successfully served as an answer to the implicit or explicit question raised
by the context. This effect might have been so strong that it backgrounded the importance
of the targeted uniqueness/maximality inference. Even if this speculation were on the right
track, however, it would still come as a surprise that a pragmatic factor of this kind
should affect the uniqueness/maximality inference to such an extent, esp. if this inference is
a presupposition.
Manfred Krifka (p.c.) suggested that what could be behind the relative weakness of the
targeted interaction is a potential weak-definite nature of the definite descriptions used
in our experiment. The hallmark of weak definites (as in listen to the radio or go to
the grocery store) is that they need not satisfy the uniqueness/maximality condition (see
Aguilar-Guevara 2014 for in-depth discussion). Manfred Krifka was responding to item 15
of MAIN, in which reception guests are startled (= context) by the burst of a/the light bulb
(= target). The implication is that the light bulb might be a weak definite description, at
least in relation to an ordinary “room situation”. We find it relatively plausible that the
“weakness” of the definiteness used in the MAIN experiment could have contributed to
the weakness of the expected interaction. On the other hand, none of the items we used
involved a prototypical weak definite.
The last piece of speculation we can offer stems from a comment by Ellen Brandner
(p.c.), who reminded us that articles, whether definite or indefinite, are not merely semantic
linguistic “devices”, but also—and perhaps very importantly so—grammatical ones. They
42 Using two recent approaches, we could also say that the contexts in MAIN had a very salient
inquisitive content (Ciardelli et al. 2013) or that they introduced a salient question under discussion
(Beaver et al. 2017; Roberts 2012).
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
Definiteness, uniqueness, and maximality in languages with and without articles 357
often represent the minimal pieces of morphology that make noun phrases usable in
sentences—without them, many sentences would simply be ill-formed. On top of that,
articles are often the main carriers of information about case, number, and gender, German
being a paradigmatic example of this. Our surprisingly weak results could therefore point
towards the need of a multi-factorial analysis, where the grammatical load of words reduces
the salience of the semantic inferences they convey.
Be it as it may, the results of DEF clearly suggest that the hypothesized uniqueness/maximality
of definite description is a more fragile phenomenon than one might
expect: it can “disappear” or be ignored relatively easily, possibly as a consequence of
various pragmatic processes or of their grammatical function. To conclude this part of the
discussion, let us mention that our results resonate with those of De Veaugh-Geiss et al.
(2018), who also found the uniqueness/maximality of definite descriptions (particularly
definite pseudoclefts) to be weaker than expected. De Veaugh-Geiss et al. (2018) resorted to
anaphoricity as a potential explanation of the lacking uniqueness/maximality. Our results
show the need to consider a broader set of confounding factors, as anaphoricity in the
narrow sense was not involved in our experiment and the bridging relation (also called
“associative anaphora”; cf. ‘carriage’–‘locomotive’) was present independently of the target
definiteness.
7.2 The semantics of bare NPs in an articleless language (NUM, WO, PROS)
The claim that definiteness correlates bring about presuppositional semantics of bare NPs
was not borne out by our experiments. We did not find singular bare NPs to convey
uniqueness to a greater extent than plural bare NPs convey maximality (Dayal 2004), nor
did we find the effect of topicality (conveyed by word order or prosody) on uniqueness or
maximality (Geist 2010, among many others).
Potential explanations for the missing effect of topicality (WO, PROS) were discussed in
quite some detail in the discussion of section 5.3. Apart from the null hypothesis, to which
we will turn shortly, they include (i) the possibility that the covered box paradigm did not
work as intended, (ii) the possibility that our design failed to manipulate topicality, (iii) the
possibility that uniqueness/maximality presupposition are not involved in determinate bare
NPs in articleless languages, and (iv) the possibility that our experiment was not sensitive
enough to capture the intended interaction.
Option (i), namely that our covered box design prompted participants to choose
less preferred (= indeterminate) readings (= visual pictures in the critical conditions),
applies equally to definite descriptions and bare NPs (assuming Coppock & Beaver’s
2015 analysis), with the difference that in the latter case, the choice of the less preferred
reading would completely override the preferred (determinate) one. Also here, it should be
mentioned that this view does not line up with what we saw in the fillers. In SCAL, for
instance, the less preferred interpretation (the logical one) was chosen only in 10% of the
cases. The reader is referred to section 7.1 for more detailed discussion.
Concerning (ii), in the light of our discussion in section 7.1, it is indeed not ruled out
that the effect of the discourse relation between the context and target was so strong that it
overrode any attempted topicality manipulation and the expected interaction—although
potentially there—simply dropped too “low” to be visible (floor effect). On the other
hand, the filler results demonstrate very high sensitivity to presuppositional and implicature
phenomena, sometimes even more strongly so than for German, which makes the floor
explanation hard to swallow.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
358 R. Šimík & C. Demian
Explanation (iii) seems more likely: at least some commonly entertained semantic/pragmatic
concepts used for modelling definiteness (identifiability, salience, etc.) could probably
be adapted for bare NPs and thereby be compatible with the null result. The concept of
identifiability is a case in point: the participants were always capable of identifying the
relevant referent, irrespective of the ±unique/maximal manipulation.
While trying to look for potential explanations for the missing interaction, we should
not forget that the results are compatible with the null hypothesis, which we identified, for
purposes of the present work, with Heim’s (2011) suggestion that bare NPs in articleless
languages are always indeterminate (see section 3.1).43 While this hypothesis goes counter
to the common conjecture that the semantics underlying definite descriptions is universal
and expressible, in one way or another, in every language, it is elegant in that it does not
need to resort to stipulating covert type-shifting operations. The details of the hypothesis
are still to be explored and there could be a need for supplementary assumptions, explaining
particular empirical patterns. Examples of this are the scope facts discussed in sections 3.2
and 3.3, particularly the claim that indefinite bare NPs cannot take wide scope. Yet, these
facts were contested by Borik (2016) and Seres & Borik to appear. As far as we can tell,
many empirical issues remain open.
Regarding the unexpected result of the grammatical number manipulation in the NUM
experiment, where we found a slight tendency of plural bare NPs to refer to the maximal
referent, we can offer the following speculations. There are two potentially relevant ways
in which singular NPs differ from plural ones:44 (i) the ease (or difficulty) with which
43 An anonymous reviewer points out that some other approaches are also compatible with the results,
mentioning Chierchia (1998) and Coppock & Beaver (2015). It is true that these authors do not identify
conditions under which indeterminate interpretations are ruled out. At the same time, they keep
the option—in the case of Coppock & Beaver (2015) even preferentially so—that bare NPs may
have determinate interpretations. These are assumed to be derived using the IOTA-shift, which in
turn introduces the relevant uniqueness (or maximality) presupposition. In that sense, Chierchia’s
(1998) or Coppock & Beaver’s (2015) proposals can be seen as providing a certain baseline, to which
an account like Geist’s (2010) functions as a natural extension in the sense that it attempts to identify
the conditions under which determinate interpretations come out as the only possible ones (without
implying that Chierchia 1998 or Coppock & Beaver 2015 would subscribe to such an account). In
that sense, Heim’s (2011) proposal is stronger, as it explicitly removes the presupposition-introducing
IOTA-shift from the grammatical system of articleless languages, at least for purposes of bare NP
interpretation.
44 An anonymous reviewer suggests that plural bare NPs do not lend themselves to partitive readings,
which in turn might increase the tendency for the maximizing interpretation. However, plurals do
not differ from singulars in this respect because bare singulars also do not readily permit partitive
readings.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
Definiteness, uniqueness, and maximality in languages with and without articles 359
uniqueness vs. maximality is satisfied and (ii) the option for bare plurals to have maximizing
meanings, even in languages with articles. Let us discuss these in turn.45
It is well-known that it is easier to satisfy maximality (for plurals) than it is to satisfy
uniqueness (for singulars). The reason is that maximality follows from existence: any
non-empty set of plural entities is automatically a set containing an entity that contains
all the others—the maximal entity. This is not the case for singulars: a non-empty set
of atomic entities is not automatically a set containing exactly one entity. Now, if bare
NPs in articleless languages—for some reason—presupposed existence, that presupposition
would amount to the presupposition of maximality (with bare plurals), but not to the
presupposition of uniqueness (with bare singulars). If, in addition, there was a tendency
for bare NPs to refer to maximal witnesses, bare plurals would effectively correspond to
definite descriptions. The problem with this proposal is obvious: it is unclear why bare
NPs in articleless languages should be presuppositional. In the light of the generalizations
discussed in sections 3.2 and 3.3 (but cf. Borik 2016), it seems clear that any such proposal
would have to be supplemented with the prohibition of presupposition projection across all
sorts of scoping operators (something tentatively entertained by Dayal 2004 for the case of
bare singular uniqueness presupposition; see section 3.2).
Concerning the other speculation, it has been known, at least since Condoravdi (1992),
that bare plurals in English can sometimes correspond to definite descriptions (Condoravdi
refers to this meaning as “quasi-universal”; see also Cohen 2005; Condoravdi 1997; Solt
2009).46 In (29), definite descriptions can apparently freely alternate with bare plurals
without any notable difference in truth conditions or presuppositions. Note also that these
bare plurals are unlikely to be existentials derived by derived kind predication, given the
assumptions introduced in Chierchia (1998), because students [on the campus] and trees [in
the desert] do not denote natural or well-established kinds; in Dayal’s (2013) terminology,
they are “indexical”, in the sense that they include a (covert) constituent referring to
a particular entity (a particular campus/desert).
45 An anonymous reviewer points out that something like weak definiteness, discussed in section 7.1
German, could also be at play in the case of Russian bare singulars. This suspicion is supported by
the fact that weak definites indeed typically correspond to bare singulars in articleless languages.
From this perspective, one could hypothesize that the result we see for plurals (namely slight
tendency towards maximality) is the “default” and that the lack of uniqueness in singulars is caused
by their “weak definite” character. The reviewer further points out that if weak definiteness are
understood in terms of pseudo-incorporation (Carlson et al. 2006; Schwarz 2014; a.o.) and if bare
singulars in articleless languages pseudo-incorporate particularly easily (cf. Sa˘g 2018 for Turkish),
then bare singulars in our experiment might have been pseudo-incorporated (and hence without any
salient uniqueness implications) across conditions. There is no agreement concerning the existence
of pseudo-incorporation of bare singulars in Russian; Geist (2010) argues that it does not exist,
Mueller-Reichau (2015) argues that it does (albeit with some caveats left open). While acknowledging
that pseudo-incorporation (possibly a correspondent of weak definiteness in articleless languages)
might have affected our results, we believe that a separate study would be needed to resolve
the issue. (One possible experimental manipulation—kindly suggested by the reviewer—would be
inserting adverbials to break the adjacency between the subject and the predicate.)
46 This observation could in fact have a source in the uniqueness vs. maximality asymmetry discussed
in the previous paragraph. As far as we know, this hypothesis has not been explored in the literature.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
360 R. Šimík & C. Demian
(29) a. In 1985 there was a ghost haunting the campus.
(The) Students were aware of the danger. (Condoravdi 1992:19)
b. The desert was barren, with barely enough water to sustain plant life.
(The) Trees were few . . . (Solt 2009:131)
Whatever the ultimate explanation, it cannot be a priori ruled out that a similar effect carries
over to articleless languages, causing the tendency to interpret bare plurals as maximizing.
This idea has at least two caveats. First, while in English, the maximizing interpretation
of bare plurals is arguably contextually and otherwise restricted (see the literature cited),
this could not be the case in Russian. Second, we would expect to see a comparable effect
of number in German, which we do not: there is no visible tendency of bare plurals (the
indefinite & plural condition) towards a maximizing interpretation.
An anonymous reviewer, leaning on Dayal (2013), suggests another possible view of the
tendency towards a maximizing interpretation in Russian bare plurals. According to Dayal’s
(2013) updated definition of the nom-shift, some “indexical” bare plurals—including those
in (29)—can be shifted to kinds, after all, and hence can be interpreted existentially via
derived kind predication (DKP).47 The relevant condition introduced by the nom-shifter
is that the cardinality of the extension of the property (weakly) varies with situations. In
(29a), for instance, the number of students in the campus is not constant in all situations.
Not all “indexical” bare plurals are like this, however, as illustrated in (30): in order for the
car I bought to be considered a “(complete) car”, there must always be 4 wheels; wheels [of
the car I bought] thus cannot be shifted to a kind (and later be input to DKP).
(30) I bought a car. The wheels / *Wheels need to be replaced. (Dayal 2013:57)
The anonymous reviewer is right in that bare plurals in our target utterances are “indexical”
because their denotation is restricted by a particular referent introduced in the context.
Examples include ‘carriages [towed by the locomotive]’ (item 9), ‘pots [on the stove]’ (item
16), or ‘cacti [on the window sill]’ (item 23). The question then is if the bare plurals we
used are more like ‘students [on the campus]’ (licit input to nom) or more like ‘wheels [of
the car]’ (illicit input to nom). If the latter holds, i.e., if the cardinality remains constant
across situations, then the existential reading derived by DKP is unavailable and the only
remaining possibility is the maximizing interpretation, correctly capturing the tendency
we observed in our results. We do not consider this explanation plausible, however. As
mentioned in section 4.2.2, it was an essential property of our design that the context
never introduced any expectations concerning the cardinality of the target NP extension.
In other words, the cardinality of the extension of the target NP naturally varies with the
evaluation situation (note, by the way, that the evaluation situations correspond to the
different levels of the picture factor in our experiment), the target NP is thus a licit input
to nom, and consequently lends itself to an indeterminate reading. We conclude that Dayal’s
(2013) refined version of nom-shift provides no handle on the unexpected interaction in the
results of NOM.
47 The intuitively maximizing interpretation arises indirectly (i.e., not via the IOTA-shift), as a consequence
of domain widening; see Dayal (2013) for discussion.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
Definiteness, uniqueness, and maximality in languages with and without articles 361
7.3 Definiteness crosslinguistically (ART)
Our experiments are as far as we know the first ones to compare the semantics of definite
descriptions with what we call here definiteness correlates in articleless languages. We
found a difference between the two: while we have been able to prove the relevance
of uniqueness/maximality for the semantics of definite descriptions (DEF), definiteness
correlates apparently do not necessitate such concepts (NUM, WO, PROS). Although many
questions remain open, one conclusion can be drawn quite safely: There is no one-to-one
relation between definiteness and definiteness correlates. Even if uniqueness/maximality
prove to be relevant for articleless languages (something we failed to find support for
in this paper), their effect is demonstrably smaller than the one found in languages with
articles, as shown by our crosslinguistic tests (ART). This conclusion is backed by the filler
experiments, where we saw a striking similarity between German and Russian with regard
to various presuppositions and implicatures.
The questions that remain open include the following: Which factors other than
grammatical number, word order, or prosody (the latter two standing for topicality) could
trigger a bare NP interpretation comparable to the one of definite descriptions? (Candidates
include verbal aspect, see Krifka 1989, but also Filip 1993, or nominal case, see e.g. Partee
& Borschev 2004.) Which notions other than uniqueness/maximality could lie behind the
recurring “definiteness intuitions” in the literature on articleless languages? (Candidates
include familiarity or identifiability.) While these and many other questions remain to
be tackled in future research, we hope that the present contribution on uniqueness and
maximality serves as a good starting point.
Acknowledgements
The materials in this paper were presented, in various stages of development, at an SFB
colloquium in Cologne, an SFB colloquium in Tübingen, a colloquium in Prague, in a guest
lecture series in Stuttgart, at a workshop in Barcelona at the occasion of Veronika King’s PhD
defence, at the 4th Expert meeting of the Time in translation project in Utrecht, at the Colloquium
Semantics and Pragmatics Exchange in Düsseldorf, in the ZAS Semantics Circle in Berlin, in a
linguistic lecture series in Bochum, and in the Slavic colloquium in Berlin. We profited greatly
from the feedback we got at these occasions. Thanks go to: Ana Aguilar-Guevara, Nadine Bade,
Petr Biskup, Ellen Brandner, Jan Chromý, Jakub Dotlaˇcil, Hana Filip, Berit Gehrke, Ljudmila
Geist, Jana Häussler, Halima Husi´c, Daniel Hole, Katja Jasinskaja, Veronika King, Manfred
Krifka, Bert Le Bruyn, Denisa Lenertová, Maria Martynova, Roland Meyer, Fereshteh Modaressi,
Uli Sauerland, Natalia Slioussar, Petra Schumacher, Stephanie Solt, Henriëtte de Swart, Luka
Szucsich, Barbara Tomaszewicz, and Martin Wągiel. Special thanks goes to Florian Schwarz,
whose careful feedback was crucial for us. We also greatly profited from the feedback of a number
of anonymous reviewers, the handling editor Roger Schwarzschild, and the proofreader Nathan
Klinedinst. All remaining errors are our own. Finally, we are grateful to the following people for
helping us with the experimental practicalities: Darja Benert (Russian recordings and translation),
Marta Wierzba (German recordings), Sandra Hoppe (Polish recordings for a pilot experiment),
Maria Nagornaya (experiment administration in Moscow), and Fabienne Salfner (and via her the
Priority Program Experimental Pragmatics, for experiment administration in Berlin).
Funding
The research was supported by the German Research Foundation (DFG), via the project
Definiteness in articleless Slavic languages (https://gepris.dfg.de/gepris/projekt/279721764).
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
362 R. Šimík & C. Demian
References
Abusch, D. (1993), ‘The scope of indefinites’. Natural Language Semantics 2: 83–135.
Aguilar-Guevara, A. (2014), Weak Definites: Semantics, Lexicon, and Pragmatics. Ph.D. thesis,
Utrecht University.
Allerton, D. J. (1978), ‘The notion of “givenness” and its relations to presupposition and to
theme’. Lingua 44: 133–68.
Bade, N. & F. Schwarz (2018), ‘Antipresuppositions are weak and slow: evidence from eyetracking
data’. Presented at the 42nd Annual Penn Linguistics Conference (PLC 42), March
2018, University of Pennsylvania.
Barker, C. (1992), ‘Group terms in English: representing groups as atoms’. Journal of Semantics
9: 69–93.
Bates, D., M. Maechler, B. Bolker & S. Walker (2015), ‘Fitting linear mixed-effects models using
lme4’. Journal of Statistical Software 67: 1–48.
Beaver, D., C. Roberts, M. Simons & J. Tonhauser (2017), ‘Questions under discussion: where
information structure meets projective content’. Annual Review of Linguistics 3: 265–84.
Birkenmaier, W. (1979), Artikelfunktionen in einer artikellosen Sprache: Studien zur nominalen
Determination im Russischen. Fink. München.
Biskup, P. (2011), Adverbials and the Phase Model. John Benjamins. Amsterdam.
Borik, O. (2016), ‘Constraints on the position and interpretation of bare singular indefinites in
Russian’. Linguistica 56: 9–23.
Bosch, P., A. Alexeyenko, K. Brukamp, M. Cieschinger, X. Deng & P. König (2011), ‘Definite
reference: salience is only a poor substitute for uniqueness’. In Proceedings of the PRE-CogSci
2011.
Bronnikov, G. (2006), A critique of Dayal (2004). Term Paper, University of Texas at Austin.
Brun, D. (2001), ‘Information structure and the status of NP in Russian’. Theoretical Linguistics
27: 109–35.
Carlson, G. N., R. Sussman, N. Klein & M. Tanenhaus (2006), ‘Weak definite noun phrases’. In
Davis, C., Deal, A. R. and Zabbal, Y. (eds.), NELS 36: Proceedings of the 36th Annual Meeting
of the North East Linguistic Society. GLSA Publications. Amherst, MA. 179–96.
Chierchia, G. (1998), ‘Reference to kinds across languages’. Natural Language Semantics 6: 339–
405.
Chierchia, G. (2006), ‘Broaden your views: implicatures of domain widening and the “logicality”
of language’. Linguistic Inquiry 37: 535–90.
Christophersen, P. (1939), The Articles: A Study of Their Theory and Use in English. Munksgaard.
Copenhagen.
Chvany, C. V. (1973), ‘Notes on root and structure-preserving in Russian’. In Corum, C. W.,
Smith-Stark, T. C. and Weiser, A. (eds.), You Take the High Node and I’ll Take the Low Node.
Chicago Linguistic Society. Chicago, IL. 52–290.
Ciardelli, I., J. Groenendijk & F. Roelofsen (2013), ‘Inquisitive semantics: a new notion of
meaning’. Language and Linguistics Compass 7: 459–76.
Cisneros, C. (2019), ‘Definiteness in Cuevas Mixtec’. In Aguilar-Guevara, A., Loyo, J. P. and
Maldonado, V. V.-R. (eds.), Definiteness Across Languages. Language Science Press. Berlin.
39–81.
Cohen, A. (2005), ‘More than bare existence: an implicature of existential bare plurals’. Journal
of Semantics 22: 389–400.
Collins, J. N. (2019), ‘Definiteness determined by syntax: A case study in Tagalog’. Natural
Laguage & Linguistic Theory 37: 1367–420.
Condoravdi, C. (1992), ‘Strong and weak novelty and familiarity’. In Barker, C. and Dowty, D.
(eds.), SALT 2: Proceedings from the 2nd Conference on Semantics and Linguistic Theory.
Ohio State University Press. Columbus, OH. 17–37.
Condoravdi, C. (1997), Descriptions in Context. Garland. New York.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
Definiteness, uniqueness, and maximality in languages with and without articles 363
Coppock, E. & D. Beaver (2015), ‘Definiteness and determinacy’. Linguistics and Philosophy 38:
377–435.
Dayal, V. (2004), ‘Number marking and (in)definiteness in kind terms’. Linguistics and Philosophy
27: 393–450.
Dayal, V. (2011), ‘Bare noun phrases’. In von Heusinger, K., Maienborn, C. and Portner, P. (eds.),
Semantics: An International Handbook of Natural Language Meaning, vol. 2. de Gruyter.
Berlin. 1088–109.
Dayal, V. (2013), ‘On the existential force of bare plurals across languages’. In Caponigro, I. and
Cecchetto, C. (eds.), From Grammar to Meaning: The Spontaneous Logicality of Language.
Cambridge University Press. Cambridge. 49–80.
De Veaugh-Geiss, J. P., S. Tönnis, E. Onea & M. Zimmermann (2018), ‘That’s not quite it: an
experimental investigation of (non-)exhaustivity in clefts’. Semantics and Pragmatics 11: early
access.
Diesing, M. (1992), ‘Bare plural subjects and the derivation of logical representations’. Linguistic
Inquiry 23: 353–80.
Dryer, M. S. (2013), ‘Definite articles’. In Dryer, M. S. and Haspelmath, M. (eds.), The World
Atlas of Language Structures Online. Max Planck Institute for Evolutionary Anthropology.
Leipzig.
Elbourne, P. D. (2013), Definite Descriptions. Oxford University Press. Oxford.
Endriss, C. (2009), Quantificational Topics: A Scopal Treatment of Exceptional Wide Scope
Phenomena. Springer. Berlin.
Filip, H. (1993), Aspect, Situation Types, and Noun Phrase Semantics. Ph.D. thesis, University of
California at Berkeley.
Frege, G. (1892), ‘Über Sinn und Bedeutung’. Zeitschrift für Philosophie und philosophische
Kritik 100: 25–50.
Geist, L. (2010), ‘Bare singular NPs in argument positions: restrictions on indefiniteness’.
International Review of Pragmatics 2: 191–227.
Givón, T. (1984), Syntax: A Functional-Typological Introduction, vol. 1. John Benjamins.
Amsterdam.
Graff, D. (2001), ‘Descriptions as predicates’. Philosophical Studies 102: 1–42.
Grice, H. P. (1975), ‘Logic and conversation’. In Cole, P. and Morgan, J. L. (eds.), Syntax and
Semantics, Vol. 3: Speech Acts. Seminar Press. New York. 41–58.
Hawkins, J. A. (1981), Definiteness and Indefiniteness: A Study in Reference and Grammaticality
Prediction. Croom Helm. London.
Heim, I. (1982), The Semantics of Definite and Indefinite Noun Phrases. Ph.D. thesis, University
of Massachusetts, Amherst, MA.
Heim, I. (1991), ‘Artikel und Definitheit’. In von Stechow, A. and Wunderlich, D. (eds.),
Semantik: Ein internationales Handbuch der zeitgenössischen Forschung. de Gruyter. Berlin.
487–535.
Heim, I. (2011), ‘Definiteness and indefiniteness’. In von Heusinger, K., Maienborn, C. and
Portner, P. (eds.), Semantics: An International Handbook of Natural Language Meaning, vol.
2. de Gruyter. Berlin. 996–1025.
Hlavsa, Z. (1975), Denotace objektu a její prostˇredky v souˇcasné ˇceštinˇe. Academia. Praha.
Horn, L. R. & B. Abbott (2012), ‘ the, a : (In)definiteness and implicature’. In Kabasenche, W.
P., O’Rourke, M. and Slater, M. H. (eds.), Reference and Referring. MIT Press. Cambridge,
MA. 325–55.
Huang, Y. T., E. Spelke & J. Snedeker (2013), ‘What exactly do numbers mean?’ Language
Learning and Development 9: 105–29.
Ionin, T. (2012), ‘Pragmatic variation among specificity markers’. In Ebert, C. and Hinterwimmer,
S. (eds.), Different Kinds of Specificity Across Languages. Springer. Dordrecht. 75–103.
Jenks, P. (2018), ‘Articulated definiteness without articles’. Linguistic Inquiry 49: 501–36.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
364 R. Šimík & C. Demian
King, J. C. (2001), Complex Demonstratives: A Quantificational Account. MIT Press. Cambridge,
MA.
King, T. H. (1995), Configuring Topic and Focus in Russian. CSLI Publications. Stanford, CA.
Krámský, J. (1972), The Article and the Concept of Definiteness in Language. Mouton. The
Hague.
Krifka, M. (1989), Nominalreferenz und Zeitkonstitution: Zur Semantik von Massentermen,
Pluraltermen und Aspektklassen. Wilhelm Fink. München.
Krifka, M. (2008), ‘Basic notions of information structure’. Acta Linguistica Hungarica 55:
243–76.
Križ, M. (2016), ‘Homogeneity, non-maximality, and all’. Journal of Semantics 33: 493–539.
Kuˇcerová, I. (2007), The Syntax of Givenness. Ph.D. thesis, MIT, Cambridge, MA.
Li, C. N. & S. A. Thompson (1976), ‘Subject and topic: a new typology of language’. In Li, C.
N. (ed.), Subject and Topic. Academic Press. New York. 457–89.
Link, G. (1983), ‘The logical analysis of plural and mass terms: a lattice theoretical approach’.
In Bäuerle, R., Schwarze, C. and von Stechow, A. (eds.), Meaning, Use, and Interpretation of
Language. de Gruyter. Berlin. 302–23.
Löbner, S. (1985), ‘Definites’. Journal of Semantics 4: 279–326.
Ludlow, P. & G. Segal (2004), ‘On a unitary semantical analysis for definite and indefinite
descriptions’. In Reimer, M. and Bezuidenhout, A. (ed.), Descriptions and Beyond: An
Interdisciplinary Collection of Essays on Definite and Indefinite Descriptions and Other
Related Phenomena. Oxford University Press. Oxford. 420–36.
Lyons, C. (1999), Definiteness. Cambridge University Press. Cambridge.
Malamud, S. A. (2012), ‘The meaning of plural definites: a decision-theoretic approach’.
Semantics & Pragmatics 5: 1–58.
Matthews, S. & V. Yip (1994), Cantonese: A Comprehensive Grammar. Routledge. London.
Modarresi, F. (2014), Bare Nouns in Persian: Interpretation, Grammar, and Prosody. Ph.D.
thesis, University of Ottawa and Humboldt University Berlin.
Mueller-Reichau, O. (2015), ‘Pseudo-incorporation in Russian? Aspectual competition and bare
singular interpretation’. In Borik, O. and Gehrke, B. (eds.), Syntax and Semantics 40: The
Syntax and Semantics of Pseudo-incorporation. Brill. Leiden. 262–95.
Neale, S. (1990), Descriptions. MIT Press. Cambridge, MA.
Partee, B. (1987), ‘Noun phrase interpretation and type-shifting principles’. In Groenendijk, J.,
de Jongh, D., and Stokhof, M. (eds.), Studies in Discourse Representation and the Theory of
Generalized Quantifiers, GRASS 8. Foris. Dordrecht. 115–43.
Partee, B. & V. Borschev (2004), ‘The semantics of Russian genitive of negation: the nature and
role of perspectival structure’. In Young, R. B. (ed.), SALT 14: Proceedings from the 14th
Conference on Semantics and Linguistic Theory. Cornell University. Ithaca, NY. 212–34.
Pearson, H., M. Khan & J. Snedeker (2011), ‘Even more evidence for the emptiness of plurality:
An experimental investigation of plural interpretation as a species of implicature’. In Li, N. and
Lutz, D. (eds.), SALT 20: Proceedings from the 20th Conference on Semantics and Linguistic
Theory. Linguistic Society of America. Washington, DC.489–507.
R Core Team (2017), R: a language and environment for statistical computing. R Foundation for
Statistical Computing. Vienna.
Reinhart, T. (1981), ‘Pragmatics and linguistics: an analysis of sentence topics’. Philosophica 27:
53–94.
Roberts, C. (2003), ‘Uniqueness in definite noun phrases’. Linguistics and Philosophy 26:
287–350.
Roberts, C. (2012), ‘Information structure: towards an integrated formal theory of pragmatics’.
Semantics & Pragmatics 5: 1–69.
Rochemont, M. (2016), ‘Givenness’. In Féry, C. and Ishihara, S. (eds.), The Oxford Handbook
of Information Structure. Oxford University Press. Oxford.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
Definiteness, uniqueness, and maximality in languages with and without articles 365
Romoli, J., Y. Sudo & J. Snedeker (2011), ‘An experimental investigation of presupposition
projection in conditional sentences’. In Ashton, N., Chereches, A. and Lutz, D. (eds.), SALT 21:
Proceedings from the 21st Conference on Semantics and Linguistic Theory. Linguistic Society
of America. Washington, DC. 592–608.
Russell, B. (1905), ‘On denoting’. Mind 14: 479–93.
Sa˘g, Y. (2018), ‘Pseudo-incorporation as sub-event kind formation’. Presented at the Workshop
on Altaic Formal Linguistics (WAFL) 14, October 2018, MIT.
Sasse, H.-J. (1987), ‘The thetic/categorical distinction revisited’. Linguistics 25: 511–80.
Sauerland, U. (2008), ‘Implicated presuppositions’. In Steube, A. (ed.), The Discourse Potential
of Underspecified Structures. de Gruyter. Berlin. 581–600.
Schein, B. (2019), ‘Nip & tuck for definite description’. Linguistics and Philosophy 42: 177–206.
Schwarz, F. (2013), ‘Two kinds of definites crosslinguistically’. Language and Linguistics
Compass 7: 534–59.
Schwarz, F. (2014), ‘How weak and how definite are weak definites?’ In Aguilar-Guevara, A.,
Le Bruyn, B. and Zwarts, J. (eds.), Weak Referentiality. John Benjamins. Amsterdam.
213–35.
Schwarzschild, R. (1996), Pluralities. Kluwer. Dordrecht.
Schwarzschild, R. (1999), ‘Givenness, AvoidF and other constraints on the placement of accent’.
Natural Language Semantics 7: 141–77.
Schwarzschild, R. (2002), ‘Singleton indefinites’. Journal of Semantics 19: 289–314.
Seres, D. & O. Borik (to appear), ‘Definiteness in the absence of uniqueness: The case of Russian’.
To appear in Advances in formal Slavic linguistics 2018 Berlin: Language Science Press.
Sharvy, R. (1980), ‘A more general theory of definite descriptions’. The Philosophical Review 89:
607–24.
Šimík, R. & M. Burianová (forthcoming), ‘Definiteness of bare NPs as a function of clausal
position: a corpus study of Czech’. In Ionin, T. and MacDonald, J. E. (eds.), Formal Approaches
to Slavic Linguistics (FASL) 26: The Urbana-Champaign Meeting 2017. Michigan Slavic
Publications. Ann Arbor, MI.
Šimík, R. & M. Wierzba (2017), ‘Expression of information structure in West Slavic: modeling
the impact of prosodic and word-order factors’. Language 93: 671–709.
Solt, S. (2009), The Semantics of Adjectives of Quantity. Ph.D. thesis, The City University of
New York.
Strawson, P. F. (1950), ‘On referring’. Mind 59: 320–44.
Szabó, Z. G. (2000), ‘Descriptions and uniqueness’. Philosophical Studies 101: 29–57.
Szwedek, A. (1974), ‘A note on the relation between the article in English and word order in
Polish (part 1 and 2)’. Papers and Studies in Contrastive Linguistics 2: 213–25.
Szwedek, A. (2011), The Thematic Structure of the Sentence in English and Polish: Sentence
Stress and Word Order. Peter Lang. Frankfurt am Main.
Titov, E. (2017), ‘The canonical order of Russian objects’. Linguistic Inquiry 48: 427–57.
von Fintel, K., D. Fox & S. Iatridou (2014), ‘Definiteness as maximal informativeness’. In Crniˇc,
L. and Sauerland, U. (eds.), The Art and Craft of Semantics: A Festschrift for Irene Heim, vol.
1. MITWPL. Cambridge, MA. 165–74.
von Heusinger, K. (1997), Salienz und Referenz: Der Epsilonoperator in der Semantik der
Nominalphrase und anaphorischer Pronomen. Akademie. Berlin.
von Heusinger, K. (2002), ‘Specificity and definiteness in sentence and discourse structure’.
Journal of Semantics 19: 245–74.
Wagner, M. (2012), ‘Focus and givenness: a unified approach’. In Kuˇcerová, I. and Neeleman,
A. (eds.), Contrasts and Positions in Information Structure. Cambridge University Press.
Cambridge. 102–47.
Zehr, J., C. Bill, L. Tieu, Romoli J. & F. Schwarz (2016), ‘Presupposition projection from the
scope of none: universal, existential, or both?’ In Moroney, M., Little, C.-R., Collard, J.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022
366 R. Šimík & C. Demian
and Burgdorf, D. (eds.), SALT 26: Proceedings from the 26th Conference on Semantics and
Linguistic Theory. Linguistic Society of America. 754–74.
Zehr, J., A. Hirsch, H. A. Bacovcin & F. Schwarz (2017), ‘Priming local accommodation of
hard triggers in disjunction’. In Lamont, A. and Tetzloff, K. (eds.), NELS 47: Proceedings of
the Forty-Seventh Annual Meeting of the North East Linguistic Society. GLSA Publications.
Amherst, MA.
Downloadedfromhttps://academic.oup.com/jos/article/37/3/311/5848470byMasarykovaUniverzitauseron22March2022