Beyond indefiniteness: Indefinites and
their discourse structuring potential
Workshop in honor of Hans Kamp on the occasion of the award of a Doctor Honoris Causa
degree from the University of Lorraine, Wednesday, October 8, 2014
Klaus von Heusinger (University of Cologne)
1. Introduction
Kamp (2014a, 1) argues for a three way division of indefinites:
“ […] the maximally three-fold distinction that we obtain from a single interpretation of
‘specific’ and a single interpretation of ‘incorporation’: that between the specifically used
indefinites, the incorporating ones and those that are neither.”
(1)
Division of indefinites according to Kamp (2014a):
a) (epistemically) specific indefinites
b) incorporated indefinites
c) non-specific and non-incorporated indefinites (“regular indefinites”)
Plan of the talk: I will elaborate on the distinction between specific and non-specific (nonincorporated) indefinites at four semantic-pragmatic levels:
(i)
(ii)
(iii)
(iv)
At the referential-semantic level, referential expressions are used to identify
referents, which are arguments of predicates or quantifiers, and thus they crucially
contribute to the truth conditions of a sentence;
At the discourse-representational level, referential expressions introduce discourse
referents that function as antecedents for anaphoric expressions. Such discourse
referents are located at different accessibility or embeddedness levels of the
discourse representation;
In the backward looking discourse perspective, referential expressions are
dependent in different ways on the given discourse - or in other words: the speaker
uses particular referential forms in order to match the assumed familiarity of the
hearer;
In the context-updating or forward looking potential perspective, referential
expressions not only introduce discourse referents at certain discourse levels, they
also activate them such that they take a certain position in a salience structure of
those elements at the same discourse level; they can also provide them with the
information that they will become topic in the upcoming discourse or function as the
antecedent of a long referential chain, which is their “discourse structuring potential”.
(i) and (ii) are rephrasing the analysis of Kamp, for (iii) I will argue that d-linked indefinites
(such as partitives) are not necessarily specific, and in (iv) I argue that specific indefinites
take a crucial function in the structure of the unfolding discourse. They are the heads of long
referential chains, often even longer than those of definites.
1
(2a)
Jemand mußte Josef K. verleumdet haben, denn ohne daß er etwas Böses
getan hätte, wurde er eines Morgens verhaftet. Die Köchin der Frau Grubach,
seiner Zimmervermieterin, die ihm jeden Tag gegen acht Uhr früh das
Frühstück brachte, kam diesmal nicht. Das war noch niemals geschehen. K.
wartete noch ein Weilchen, sah von seinem Kopfkissen aus die alte Frau, die
ihm gegenüber wohnte und die ihn mit einer an ihr ganz ungewöhnlichen
Neugierde beobachtete, dann aber, gleichzeitig befremdet und hungrig, läutete
er. Sofort klopfte es und ein Mann, den er in dieser Wohnung noch niemals
gesehen hatte, trat ein. Er war schlank und doch fest gebaut, er trug ein
anliegendes schwarzes Kleid… (Kafka, Der Prozess, Erstes Kapitel)
(2b)
Someone must have slandered Josef K., for without having done anything wrong
he was arrested one morning. The cook of Ms Grubach, his landlady, who
brought him breakfast every day at eight o’clock, did not come this time. That
had never happened before. K. waited a little while, and saw from his pillow the
old woman who lived opposite him and watched him with a curiosity very
unusual for her; but then, being at the same time disconcerted and hungry, he
rang the bell. Immediately there was a knock and a man he had never seen in
this apartment before came in. He was slim and yet firmly built, he wore a fitting
black dress. (Kafka, The process, first chapter)
(2c)
(i)
(ii)
(iii)
(iv)
the cook … who
his landlady …Ø
the old woman who … and … her
a man he had never seen … he… he…
Claim: Specific indefinites can be clearly distinguished from non-specific ones at different
levels. They show characteristics of definite noun phrases, and in certain respects of
demonstratives. At the discourse level, definites are used to “establish the object’s identity
through an evolving narrative” (Du Bois 1980, 204). In contrast to this function, we can
illustrate the discourse function of specific indefinites by the metaphor of Chekhov's gun
(http://en.wikipedia.org/wiki/Chekhov%27s_gun; 4 Oct 2014)
"One must never place a loaded rifle on the stage if it isn't going to go off. It's wrong to
make promises you don't mean to keep." Chekhov, letter to Aleksandr Semenovich
Lazarev (pseudonym of A. S. Gruzinsky), 1 November 1889
Note: In the following we illustrate the argument by means of indefinites headed by the
indefinite article a(n), i.e. we assume one form, but two functions. Languages, however, offer
various ways to also formally indicate the contrast between a specific and non-specific
interpretation. In Romance, the mood in the relative clause signals the specificity of the head
noun (Rivero 1975, 40 ex. (11) for Spanish; see also Leonetti 2012):
(3)
a
b
Quiere casarse con {la, una} muchacha que sea (SUBJ) rubia y con pecas.
‘He wants to marry {the, a} girl who may (NON-SPECIFIC only) be blonde and with
freckles.’
Quiere casarse con {la, una} muchacha que es (IND) rubia y con pecas
‘He wants to marry {the, a} girl who is blonde and with freckles’ (SPECIFIC only)
2
English and German can use the indefinite demonstrative this/dies for forcing a specific
reading of the indefinite (Ionin 2006, Deichsel 2013), while phonologically reduced [sm] is
often used to instantiate a non-specific reading.
(4)
a
b
c
Every student recited a poem of Pindar.
Every student recited thisindef poem of Pindar.
Every student recited sm poem of Pindar.
2 Referential-semantic level: Uniqueness, presupposition, and denotational type
The referential-semantic perspective, which goes back to Frege and Russell, assumes that
definite descriptions express an existential and uniqueness condition, while indefinites make
just an existential contribution to the sentence, as in (5) (Heim 2011, 997), which
corresponds to the intuitive contrast between (6a) and (6b):
(5)
(6)
a
b
a
b
⟦the⟧ = λP. λQ. ∃x [∀y [ P(y) ⟷ x = y ] & Q(x) ]
⟦a⟧ = λP. λQ. ∃x [ P(x) & Q(x) ]
The book arrived.
A book arrived.
Russel (1905) argued that the existential and uniqueness condition is part of the semantics
and thus relevant for the truth conditions, while Frege (1892) and Strawson (1950) argued
that it is a presupposition. The Russellian analysis leads to the quantifier type <<e,t>,t> for
definite descriptions as well as indefinite descriptions, while the Fregean truth conditions for
definite descriptions can be accounted for by the type e (Heim 2011). This means that the
classical semantic analysis of definite descriptions and indefinite descriptions differs in
uniqueness, presuppositionality and denotational type: A definite description presupposes its
descriptive content and a uniqueness condition and is of type e, whereas an indefinite
description asserts its descriptive content, expresses a non-uniqueness implicature, and is
of the quantifier type <<e,t>,t> or of the predicative type <e,t>.
Specific indefinite express a “referential intention” of the speaker. Heim (1991, 518)
represents this reading as in (7):
(7)
a.
b.
⟦aquant N⟧ = λQ. ∃x. [N(x) & Q(x)]
⟦aref N⟧ = is defined only if there is a unique individual that the speaker
of the sentence has in mind, and this individual is N
This semantics allows to clearly distinguish between specific and non-specific indefinites.
The semantics of specific indefinites is similar to that of direct referential expressions (proper
names, demonstratives), rather than to definite noun phrases (see Fodor & Sag 1982,
Kratzer 1998, Ionin 2006). It is very controversial whether the contrast between a specific
and non-specific reding is semantic or pragmatic in nature and, if it is semantic, whether it is
the correct representation. We discuss the first point, but see for the second the discussion
about “intermediate scope behavior” (Farkas 1981 etc.).1
1
Note: The idea of modeling the “referential intention of the speaker” with “vicarious anchors” in
Kamp (2001/2010, 2014a,b) always leads to widest scope behavior of the indefinite, as in the
“referential approach” of Fodor & Sag 1981, or in the choice function analysis of Kratzer 1998.
Intermediate scope must be explained by independent mechanism.
3
Even if the speaker has a particular poem of Pindar in mind, the utterance of (8a) only
contributes an existential statement, as in (8d). Since the hearer cannot know which poem
the speaker has in mind, the speaker commits himself or herself only to the existential
statement, which would become true even if Ann would read a poem different from what the
speaker has in mind (see for discussion King 1988, Ludlow & Neale 1991, but Kratzer 1998).
(8)
a.
b.
c.
d.
Ann read a poem of Pindar.
Ann read thisindef poem of Pindar.
Ann read a certain poem of Pindar.
Ann read sm poem of Pindar.
This is the classical wisdom with regard to the contrast between a specific vs. a non-specific
reading in episodic contexts without operators. However, there are arguments that the
specific interpretation must be available:
1. Kamp (2014a, 13; quoting Karen Lewis 2012) discusses the possibility that a sentence as
the first in (9) might have two associated propositions (a singular and an existential one). In
the context of the second clause in (9), the first one seems to express the singular
proposition:
(9)
I went out to dinner with the woman from the bar last night. Can you believe it –
a woman went out to dinner with me!
2. Sæbø (2013) analyses specific indefinites that serve as antecedents for direct referential
terms in speech reports. Sæbø (2013, 267) provides the following example (10a) and (10b):
Suppose that Sæbø says (10a) to his wife and that his wife has no idea about the identity of
the “someone else”. Still, the “someone else” can later utter (10b).
(10)
a
b
I have met someone else.
He has told his wife he has met me.
The pronoun me in (10b) is more informative than the indefinite someone in (10a). Sæbø
argues that this is only possible if the indefinite in (10a) has a referential intention such that
in the reported speech that referential intention can be spelled out by a pronoun (see for
details Sæbø 2013).
3. But even in an approach that does not assume an immediate sentence semantic effect,
some authors assume a discourse effect, as illustrated by the position of Stalnaker (1998,
16), who holds that the difference between specific and non-specific indefinites is crucial for
discourse structure: “The account I am sketching suggests that this difference matters, not to
the interpretation of the indefinite expression itself, but only to the evaluation of subsequent
statements made with pronouns anaphoric to the indefinite expression.” (see for discussion
Dekker 2004).
3. Discourse-representational level: Anaphora and accessibility
In theories of discourse representation that model anaphoric references by linking the
anaphoric expression with its antecedent, both definite as well as indefinite descriptions
4
introduce discourse referents that serve as antecedents or anchors for anaphoric
expressions, as in (11). In absence of any operator there is no difference between specific
indefinites (c) and non-specific ones as antecedents. Definite and indefinite descriptions can
be clearly distinguished from quantifier expressions in (11e) and weak or incorporating
nouns in (11f), which do not license anaphoric links.
(11)
a
b
c
d
e
f
Ann reads the book. It is a mystery story.
Ann reads a book. It is a mystery story.
Ann reads this book. It is a mystery story.
Ann reads sm book. It is a mystery story.
Ann reads every book. #It is a mystery story.
Ann is book reading. #It is a mystery story.
Even though both types of indefinites allow for anaphoric links, there are clear contrasts.
One concerns the embeddedness of the indefinite antecedent, another the semantics of
anaphoric expression. We focus on the first one.2
Definite descriptions introduce their discourse referents at a higher position in the discourse
or accessibility structure than indefinites, as can be shown by the non-availability of the
anaphoric link if the antecedent is embedded under negation or other operators. However, if
the indefinite a book is read specifically, as triggered by the specific adjective a certain in
(12c) or the long descriptive relative clause in (12d), the referent is accessible similar to the
discourse referent of the definite description in (12a), which is introduced at the highest
discourse level (Karttunen 1969 / 1976). Kamp (1981 / 2013) and Kamp & Reyle (1993)
model the contrast by assuming that the discourse referent in (12a, c-d) is introduced in the
main structure, while the one in (12b) is introduced in some embedded structure. Kamp
(2001/2010, 2014a,b) assumes that the “vicarious anchor” forces the discourse referent in
the highest box.
(12)
a
b
c
d
Peter does not read the book. Ann reads it.
Peter does not read a book. #Ann reads it.
Peter does not read a certain book. Ann reads it.
Peter does not read a book he has bought this morning. Ann reads it.
4. Backward discourse properties: information status and partitivity
In a discourse perspective, the main contrast between definites and indefinites is that
definites are familiar, i.e. coreferential to or identified with an already introduced discourse or
otherwise contextually retrievable term, while indefinites are novel, i.e. they introduce new
items (so far not used ones), as illustrated below:
(13)
a
b
Ann owns a Porschei. Bill has sold the Porschek.
Ann owns a Porschei. Bill has sold a Porschek.
2
(i = k)
(i ≠ k)
There is a long controversy about the semantics of anaphoric pronouns related to (non-specific)
indefinites: Dynamic binding approaches seem to restrict the restrictor set of the indefinite in an
unwelcome way, while E-type approaches assume a uniqueness condition, which is often too strong a
condition. If we assume the semantics (7b) for specific indefinites, these problems do not arise, but
they still hold for non-specific antecedents - and we cannot exclude them from being the head of
anaphoric chains. See Kamp (2014a, 39) for discussion.
5
Different pragmatic approaches to definiteness highlight the importance of the information
status on the proper use of referring expressions. At least since Prince (1981), it is known
that a categorical distinction between given vs. new information on the one hand and definite
vs. indefinite noun phrases as referring to known and new entities on the other hand falls too
short. A third information status, inferrable, inferred or (discourse) linked, was shown to hold
for definite and indefinite noun phrases. Discourse linked definite and indefinite noun
phrases are related to some anchor in the previous context. For example, in the first
sentence of (14), a book is the anchor, while the cover in (14a) and a page in (14b) are
linked in a part-whole relation to that anchor. The difference in selecting a definite or
indefinite article is determined via encyclopedic knowledge. That is, books typically have a
unique cover, while having more than one page.
(14)
a
b
Peter bought a book. The cover was red.
Peter bought a book. A page was missing.
Partitive indefinites are a subkind of inferred indefinites: Partitive indefinites pick out one
(unmentioned) referent from a discourse-familiar group. Obviously, such indefinites
presuppose existence and behave like strong quantifiers. Enç (1991) relates partitive
indefinites to Pesetsky’s (1987, 107) notion of d(iscourse) linking that accounts for the
different presuppositions of which vs. who.
(15)
a.
b.
Oda-m-a
birkaç
çocuk
gir-di.
room-1.sg.-Dat. several child
enter-Past
‘Several children entered my room.’ (Enç 1991, ex. 16)
Iki
kız-ı
tanı-yor-du-m.
two girl-Acc. know-Prog.-Past-1.sg.
‘I knew two girls.’
(Enç 1991, ex. 17)
The first sentence introduces a set of children, and the accusative case in the second
sentence indicates that the two girls are part of that set. The expression two girls
presupposes existence. Enç takes this observation as a strong indicator that such an
expression is specific and proposes that specificity can be derived from partitivity, or more
exactly from familiarity of the superset involved. Diesing (1992) and de Hoop (1995) take
partitivity as an instance of Milsark’s (1974) contrast between a weak (cardinal, non-specific)
and a strong (presuppositional, specific) interpretation. Partitive specificity is orthogonal to
referential (or scopal) specificity, as in (16a) and to epistemic specificity, as in (16b) (see
Abbott 1995, Farkas 1994, van Geenhoven 1998 for discussion).
(16)
a
b
John wants to marry one of Steve’s sisters. (He doesn’t care which one)
One of Steve’s sisters cheated on the exam. (We have to find out which one)
We can summarize this short discussion: 1. indefiniteness marking and information status
cross-classify: There are novel definites and an intermediate category of inferred indefinites
and definites. 2. d-linked indefinites in general and the subkind of partitive indefinites show
certain characteristics of definites and / or specific indefinites, but they are not specific in the
sense that they refer to a speaker intended referent. In the case of partitives, the restrictor
6
set is discourse given, but the particular referent is still not determined, neither by the
discourse nor by the speaker.
5. Forward looking discourse properties
Referring expressions not only introduce new discourse referents and link them to already
established ones, but they also activate these referents in different ways. In a series of
papers, we (Chiriacescu & von Heusinger (2010), Chiriacescu 2011a, von Heusinger &
Chiriacescu 2011) have further elaborated Givón’s factors of topic continuity, which resulted
in three measurable parameters. These parameters represent the “forward looking potential”
of different types of referring expressions:
1. Explicitness of the anaphoric expression investigates the type of referring
expression used to pick up the referent after its first introduction. The lexical and
descriptive material was shown to inversely correlate with the accessibility of the
discourse referent (Ariel 1990, Arnold 1998).
2. Referential persistence measures the frequency with which a referent is
anaphorically re-mentioned in the subsequent discourse (Givón 1983, Gernsbacher
& Shroyer 1989, Arnold 1998, Kehler et al. 2008).
3. Topic shift potential measures the distance in matrix sentences between a nontopical referent and its subsequent mention in topic position. (Note: Givón’s topic
continuity is different from this parameter, as it measures the duration of being a topic
and not the first usage as a topic)
5.1 Explicitness of the anaphoric expression
Explicitness of the anaphoric expression measures the DP-type of the linguistic expression
that picks up the referent after its first introduction. The lexical and descriptive material is
said to be inversely related to the prominence of the discourse referent, as illustrated in
Table 1. The two referents Benjamin and a student behind a portable toilet are introduced in
S0 and picked up in S1 by pronouns. However, in S2 only the second referent is picked up
by the full definite description the student.
S-1
A popular open-air rock concert took place in Cologne Ref1
Ref2
this evening.
S0 Benjamini saw a studentk behind a portable toilet.
S1 Hei went up to himk and wanted to greet himk.
pron
pron
S2 As hei moved closer, hei noticed that the studentk was pron
def
not doing so well.
NP
S3 Hek had apparently drunk too much.
-pron
S4 Despite it being Benjamini’s favorite open-air rock proper def
concert, hei helped the studentk getting up.
name
NP
S5 Hei accompanied himk home.
pron
pron
Table 1. Explicitness of the anaphoric expression (sample story with annotations)
5.2 Referential persistence
Referential persistence measures the frequency with which a referent is anaphorically rementioned in the subsequent discourse (Givón 1983, Gernsbacher & Shroyer 1989, Arnold
1998, Kehler et al 2008), as illustrated in Table 2:
7
S-1
A popular open-air rock concert took place in Cologne Ref1
Ref2
this evening.
S0 Benjamini saw a studentk behind a portable toilet.
S1 Hei went up to himk and wanted to greet himk.
1
1
2
2
S2 As hei moved closer, hei noticed that the studentk was 2
3
1
3
not doing so well.
S3 Hek had apparently drunk too much.
0
3
1
4
S4 Despite it being Benjamini’s favorite open air rock 2
5
1
5
concert, hei helped the studentk getting up.
S5 Hei accompanied himk home.
1
6
1
6
Table 2. Referential persistence of anaphoric expressions (sample story with
annotations - per sentence and cumulative)
5.3 Topic shift potential
Topic shift potential measures the distance in sentences between the mention of a referent
in non-topical position and its first occurrence as a topic in the subsequent discourse. Topic
shift potential measures the distance with which a non-topical referent is mentioned again as
a topic for the first time in the subsequent discourse.
S-1
S0
S1
S2
A popular open-air rock concert took place in Cologne this evening.
Topic shift
Benjamini saw a studentk behind a portable toilet.
–
Hei went up to himk and wanted to greet himk.
–
As hei moved closer, hei noticed that the studentk was not doing so
–
well.
S3
Hek had apparently drunk too much.
+
S4
Despite it being Benjamini’s favorite open air rock concert, hei helped
+
the studentk getting up.
S5
Hei accompanied himk home.
+
Table 3. Topic shift potential (sample story with annotations; only first topic shift is counted)
There is some work on the effect of indefinite this with respect to the parameters introduced
above (Gernsbacher & Shroyer 1989; Chiriacescu 2011b), and some work on the discourse
effects of certain stages of the grammaticalization path of indefinite articles (Givón 1981).
For example, Strube & Hahn (1996) argue that definite noun phrases have a higher
probability of becoming topic in the subsequent discourse than indefinite noun phrases. Up
to this point, there has been no such differentiation with respect to information structure.
6. Sentence continuation experiment
We used a multi-sentence story-continuation task (adapted from Gernsbacher & Shroyer
1989, and Chiriacescu & von Heusinger 2010) in which native speakers of German (n=96)
read stories (n=16) consisting of two sentences. Participants were asked to provide five
natural-sounding sentences as continuation for each story fragment. We manipulated the
definiteness and information status of target referents in the second sentence, which
resulted in four conditions, as illustrated in Table 4: (i) definite and linked NP, e.g. (2a); (b)
indefinite and linked NP, e.g. (2b); (c) definite and brand-new NP, e.g. (2c); and (d) indefinite
and brand-new NP, e.g. (2d). The first sentence of each story set the scene and introduced
8
the contextual anchor for conditions (2a-b). The second sentence introduced two (human)
referents of the same gender. The first referent in each experimental item was the sentence
aboutness topic (e.g., Benjamin in (2)). The second referent was the manipulated critical
referent, which appeared in direct object position in (e.g., the muscular drummer in (2)).
Importantly, a norming study (n=96) confirmed that association strengths to contexts were
similar for referents in (2a) and (2b), ts < 0.3; ps > .7. For our analyses, two independent
judges coded for the aboutness topic of the 5 sentence-continuations provided by the
participants.
(1) A popular open-air concert took place in Cologne this evening.
(2) Benjamin saw____
(a) the muscular drummer.
[def+linked]
(b) a muscular guard.
[indef+linked]
(c) the student behind a portable toilet.
[def+brand new]
(d) a student behind a portable toilet.
[indef+brand new]
Table 4: Sample experimental items (translated from German)
Previous work has shown that next mention biases are generally modulated by accessibility
and that definite and given NPs correspond to higher rates of immediate (i.e. local)
subsequent mention compared to their indefinite and brand new counterparts (Grosz et al.
1995, Strube & Hahn 1996).
In light of these observations we predicted that:
a) definiteness triggers higher referential persistence
b) indefinite NPs have (at best) an effect on non-local referential persistence (nonadjacent matrix sentences),3 and
c) givenness has an equally strong effect on both types of referring expressions.
To test for statistical significance, we conducted separate generalized linear regression
models in the same way we did for sentence-based referential persistence. Because our
models again failed to converge with the inclusion of random intercepts for items, these
intercepts were excluded from our final models.
3
Gernsbacher & Shroyer (1989) and Chiriacescu (2011) have shown this for indefinite-this, Deichsel
and von Heusinger (2011) for indefinite dies in German. A high topic shift potential for specifically
used indefinites headed by an indefinite article is suggested by Wright & Givón (1987), Stalnaker
(1998), von Heusinger (2011) and Kamp (2014a).
9
Condition
Example
[def+linked]
[def+brand-new]
the muscular drummer
the student behind a
portable toilet
a muscular guard
a student behind a portable
toilet
[indef+linked]
[indef+brand-new]
Mean S1
Mean S1 - S3
Mean S1 - S5
4.31 (0.71)
5.12 (0.52)
11.69 (1.54)
14.06 (1.75)
16.81 (2.21)
21.94 (2.09)
5.44 (0.48)
5.19 (0.68)
13.50 (1.03)
14.19 (1.49)
20.62 (1.60)
21.75 (1.97)
Table 5. Mean number of times across items that the referent in direct object position has been mentioned in the first sentence
of the six continuations (S1), up to the third sentence of the six continuations (S1 - S3), and up to the fifth sentence of the six
continuations (S1 - S5); standard errors (SE) are presented in parentheses; def = definite article; indef = indefinite article; link =
linked direct object; unlink = unlinked direct object.
As expected, there were no reliable differences between conditions for first sentences (S1),
zs < 1.5, ps > .15.
However, for analyses on persistence scores including the first three sentences of
continuations (S1 – S3), we found a marginal effect of Linking, z = 1.87, p = .06, confirming
that, overall, unlinked referents were mentioned more often than linked referents.
Finally, analyses including all five sentences of continuations (S1 – S5) resulted in significant
main effects of Definiteness, z = 2.49, p < .05, and Linking, z = 3.28, p < .01, which, in turn,
qualified a marginal Definiteness X Linking interaction, z = - 1.91, p = .06. This interaction
indicates that, for unlinked referents, definiteness marking did not make a significant
difference for the likelihood that the referent will be mentioned again in subsequent
discourse. In contrast, for linked referents, referents with an indefinite NP (a muscular guard)
were mentioned more often than referents with a definite NP (the muscular drummer).
With respect to our three predictions, we can say:
a) Contrary to the prediction, definite NPs have a lower referential persistence than
indefinites
b) Indefinite noun phrases are more likely to be picked up anaphorically
c) Contrary to the prediction, linked noun phrases are less likely to be rementioned in
the unfolding discourse
d) These results are primarily triggered by the low frequency of anaphoric expressions
to [def+linked] noun phrases, which is again an unexpected result.
e) It is still an open question whether indefiniteness marking or brand-new information
status determined the result.
7. Referential intention and Chekhov’s gun
Overview of the discussion so far
- specific indefinites are referential expressions with a referential intention (of the speaker)
- specific indefinites show a referential behavior similar to proper names and
demonstratives, rather than to definites
- specific indefinites show wide scope behavior (open problem: intermediate scope)
- sentences with specific indefinites (can) express a singular proposition
10
- sentences with specific indefinites often only express the existential proposition
- the discussion of anaphoric pronouns did not help towards the understanding of the
contrast between specific and non-specific indefinites
- d-linked indefinites, such as inferred indefinites or partitive indefinites, show certain
characteristics of specific indefinites, but they are not necessarily specific (although they
can)
- forward looking properties include the discourse structuring potential, i.e. the potential to be
mentioned again in the unfolding discourse.
Du Bois (1980, 221):
The opening of a new file with an a-form mention tends to raise the expectation that the
file will continue to be used, as more information is added to it. Given this expectation, it
would be useful to signal cases where little or no further use will be made of the file.
There appears to be a slight tendency to use an indefinite pronoun (someone,
somebody) rather than an a-form to mark the introduction of an unimportant character
who will not be spoken about much.
Kamp (2014a, 37):
Usually multi-sentence assertions in which a pronoun in one sentence is anaphoric to an
indefinite noun phrase in a previous sentence, the speaker is using the indefinite
specifically in the sense of Section 2.
Weinrich (2007, 410):
Von einem kataphorischen (“unbestimmten”) Artikel erhält der Hörer die
Gegenanweisung. Er soll nach geeigneten Determinantien für das begreffende Nominen
nicht in der Vorinformation suchen, sondern diese von der noch unbekannten
Nachinformation erwarten. (…) Im Verhältnis zum anaphorischen Artikel hat der
kataphorische Artikel eine Frequenz wie 1 : 7. Nur durch sein relativ seltenes
Vorkommen kann er nämlich die Aufgabe erfüllen, die Aufmerksamkeit des Hörers für
die zu erwartende Nachinformation zu wecken.
11
7.1 Text based coherence constraint
In order to explain this function of indefinites (or brand-new discourse referents), we could
postulate a general coherence constraint on discourse segments of the kind of Asher &
Lascarides (2003, 230):
Formulating an update constraint to ensure that the resultant logical form
reflects the maximally coherent discourse is tricky not least because it is unclear
how one should 'measure’ or represent the degree of coherence. This constraint is
also quite different from other constraints on update. It has more the flavour of
optimality theory where alternative consequences from a set of rules get ranked
There should not be too many discourse referents that are not used in the unfolding text.
However, we have then to define which expressions introduce such discourse referents, as
there are many noun phrases that introduce discourse referents that are never taken up. So
we could say that this only holds for referential expression in narratives (vs. descriptive
mode, see Du Bois 1980), or we say it only holds for specific noun phrases.
7.2 Text planning constraints
Alternatively, we can also assign a function to the intentional structure of a text, i.e. the
planning of the speaker of a text (see Grosz and Sidner 1986). We could then generalize the
idea of “referential intention” from the intention to identify an object or individual in the world
toward the intention to identify a discourse character that is or will be identified in the text
world. In this view, the use of an indefinite (or brand-new discourse referent) signals that the
unfolding discourse contributes more properties to that referent.
7.3. Indefinites and more
Indefinites have been a fascinating and very productive research topic and they will remain
such in future.
Appendix: Results and discussion for topic shift potential
Table 5 presents the mean percentages of cases in which the direct object was made topic
in the first continuation sentence (S1), by the end of the third sentence continuation (S1 –
S3), and by the end of the fifth continuation sentence (S1 – S5). As it can be observed, in
the first sentences, participants tended to make [def+linked] referents topic less often than
[def+brand-new] and [indef] referents. While this observation also holds when including the
first three continuation sentences (mean S1 – S3), by the end of the fifth sentence
continuation, there is a clear tendency for [brand-new] referents to become topic more often
than [+linked] referents, with no strong difference between [def+linked] and [indef+linked]
referents.
12
Condition
Example
Mean S1
[def+linked]
[def+brand-new]
Mean S1 S3
62.44 (5.94)
70.81 (6.53)
Mean S1 S5
69.50 (5.47)
81.25 (6.23)
the muscular drummer
33.25 (4.82)
the student behind a portable
49.00 (6.69)
toilet
[indef+linked]
a muscular guard
50.00 (6.97)
69.81 (5.92)
73.88 (6.25)
[indef+brand-new] a student behind a portable
40.56 (6.96)
72.81 (6.39)
81.12 (5.03)
toilet
Table 5. Mean percentages for which direct objects of the second sentence had been made topic by
the first sentence (S1), the third sentence (S1 – S3), and the final sentence (S1 – S5); standard
errors (SE) are presented in parentheses; def = definite article; indef = indefinite article; linked =
linked direct object; brand-new= unlinked direct object.
We tested whether the observed differences were statistically reliable by performing logistic
regression for: (i) the first sentence continuation, (ii) the first three continuation sentences,
and (iii) all five continuation sentences. Again, we included random intercepts for items but
not for participants. Note that scores follow a binomial distribution because direct objects
were either topic of a sentence (score of 1) or not (score of 0).
(i) For the first sentence continuations, results of the final model revealed significant main
effects of Definiteness, z = 4.77, p < .0001, and Linking, z = 2.79, p < .01. This led to a
reliable Definiteness X Linking interaction, z = -11.25, p < .0001, confirming that, in their first
continuation sentences, participants made linked referents denoted by definite NPs (i.e. the
croupier) the topic of the discourse less frequently than referents of the other three
conditions (i.e. a gambler, the/a police officer).
(ii) When we included the first three continuation sentences in our analyses, the final model
revealed a main effect of Linking, z = 2.02, p < .05, that was accompanied by a Definiteness
X Linking interaction, z = -2.41, p < .05. The main effect of Definiteness was marginal, z =
1.74, p = .08, however. In other words, similar to the first sentence continuations, [brandnew] or unlinked referents became the topic more often than [linked] referents and
definiteness affected both linked and unlinked referents. Note that overall, effects in terms of
subsequent topic shift occurred up to the third sentence continuation, which means that less
topic shifts occurred after this point.
(iii) Including all five sentence continuations provided by the participants in our analysis
made the effect of definiteness disappear, z = 1.05, p = .29. However, like for the former
analyses, unlinked or [brand-new] referents were made topic significantly more often than
[linked] referents, z = 8.6, p < .0001. The Definiteness X Linking interaction remained
significant as well, z = -2.63, p < .01, suggesting that, throughout continuations, [def+linked]
referents displayed the lowest topic shift potential.
indef-unlinked
S-1
S0
S1
S2
S3
S4
S5
In Köln fand an diesem Abend ein beliebtes Open-Air Rockkonzert statt.
Benjamin sah einen Schüler hinter dem Dixiklo.
Er ging zu ihm und wollte ihn begrüßen.
Als er näher kam, sah er, dass es dem Schüler nicht gut ging.
Er hatte anscheinend zu viel getrunken.
Obwohl es Benjamins Lieblings Open-Air-Rockkonzert war half er dem Schüler auf.
Er begleitete ihn nach Hause.
13
S-1
S0
S1
S2
S3
S4
S5
A popular open-air rock concert took place in Cologne this evening.
Benjamini saw a studentk behind a portable toilet.
Hei went up to himk and wanted to greet himk.
As hei moved closer, hei noticed that the studentk was not doing so well.
Hek had apparently drunk too much.
Despite it being Benjamini’s favorite open air rock concert, hei helped the studentk getting up.
Hei accompanied himk home.
def_unlinked
S-1
S0
S1
S2
S3
S4
S5
In Köln fand an diesem Abend ein beliebtes Open-Air Rockkonzert statt.
Benjamin sah den Schüler hinter dem Dixiklo.
Dem Jungen war offensichtlich schlecht.
Benjamin, der sein Mathelehrer war, sah sich genötigt ihm zu helfen.
Also half er dem Jungen sich ins Nothelferzelt zu begeben.
Die Notärztin verabreichte ihm eine Infusion.
Danach konnten beide zurück zum Konzert gehen.
S-1
S0
S1
S2
S3
S4
S5
A popular open-air rock concert took place in Cologne this evening.
Benjamin saw a student behind a portable toilet.
The boy was feeling sick apparently.
Benjamin, who was his maths teacher, felt compelled to help him.
So he helped the boy to proceed toward the emergency tent.
The emergency doctor gave him an infusion.
Afterwards, both were able to return to the concert.
indef_linked
S-1
S0
S1
S2
S3
S4
S5
In Köln fand an diesem Abend ein beliebtes Open-Air Rockkonzert statt.
Benjamin sah einen muskulösen Ordner.
Er beobachtete ihn.
Er ging hinüber und kam mit ihm ins Gespräch.
Er fragte ihn nach seiner Nummer.
Der Ordner warf ihm ein Lächeln zu.
Der Ordner verwies ihn des Konzertes.
S-1
S0
S1
S2
S3
S4
S5
A popular open-air rock concert took place in Cologne this evening.
Benjamin saw a muscular steward.
He watched him.
He went over to him and got into a conversation with him.
He asked him for his phone number.
The steward smiled at him.
The steward expelled him from the concert.
def_linked
S-1
S0
S1
In Köln fand an diesem Abend ein beliebtes Open-Air Rockkonzert statt.
Benjamin sah den muskulösen Schlagzeuger.
Er hatte ihn irgendwie anders in Erinnerung gehabt.
14
S2
S3
S4
S5
Ein wenig neidisch war er schon.
Solche Muskeln erfordern bestimmt viel Training.
Das wär nichts für ihn, viel zu anstrengend.
Der Bassist dagegen sah aus wie eine Bohnenstange.
S-1
S0
S1
S2
S3
S4
S5
A popular open-air rock concert took place in Cologne this evening.
Benjamin saw a muscular drummer.
He remembered him somewhat differently.
Granted, he was a bit envious.
Such muscles certrainly require a lot of workout.
That wasn’t his thing, way too exhausting.
The bass player however looked like a rake.
References
Abusch, Dorit. 1994. The scope of indefinites. Natural Language Semantics 2. 83–135.
Arnold, Jennifer E. 1998. Reference Form and Discourse Patterns. Stanford, CA: Stanford University
dissertation.
Arnold, Jennifer E. & Maryellen C. MacDonald. 1999. The effects of reference specificity and utterance
contribution on pronoun resolution. In Proceedings of the Twenty-First Annual Meeting of the Cognitive
Science Society, 31–36.
Bittner, Maria. 2014. Temporality. Universals and Variation. Chichester/West Sussex, UK: Wiley-Blackwell.
Chierchia, Gennaro. 2001. A puzzle about indefinites. In Carlo Checchetto & Gennaro Chierchia (eds.),
Semantic Interfaces: Reference, Anaphora and Aspect, 51–89. Stanford, CA: CSLI Publications.
Chiriacescu, Sofiana. 2011. The Discourse Structuring Potential of Indefinite Noun Phrases. Special Markers in
Romanian, German and English: Universität Stuttgart dissertation.
Chiriacescu, Sofiana & Klaus von Heusinger. 2010. Discourse prominence and pe-marking in Romanian.
International Review of Pragmatics 2(2). 298–332.
Deichsel, Annika. 2013. The Semantics and Pragmatics of the Indefinite Demonstrative dieser in German.
Universität Stuttgart: dissertation.
Dekker, Paul. 2004. The pragmatic dimension of indefinites. Research on Language and Computation 2. 365–
399.
Diesing, Molly. 1992. Indefinites. Cambridge, MA: MIT Press.
Donnellan, Keith. 1966. Reference and definite descriptions. Philosophical Review 75. 281–304.
Du Bois, John W. 1980. Beyond definiteness: The trace of identity in discourse. In Wallace Chafe (ed.), The
pear stories. cognitive, cultural, and linguistic aspects of narrative production, vol. 3 Advances in
Discourse Processes, 203–274. Norwood, N J: Ablex Publishing Corporation.
Enç, Mürvet. 1991. The semantics of specificity. Linguistic Inquiry 22(1). 1–25.
Farkas, Donka. 1981. Quantifier scope and syntactic islands. In Roberta Hendrick, Carrie Masek &
Mary Frances Miller (eds.), Papers from the Seventeenth Regional Meeting Chicago Linguistics Society
(=CLS) 17, 59–66. University of Chicago.
Farkas, Donka. 1994. Specificity and scope. In Lea Nash & Georges Tsoulas (eds.), Actes du Premier Colloque
Langues & Grammaire, vol. 1, 119–137. Paris: Université Paris-8.
Farkas, Donka. 2002. Specificity distinction. Journal of Semantics 19. 213–243.
Fodor, Janet & Ivan Sag. 1982. Referential and quantificational indefinites. Linguistics and Philosophy 5. 355–
398.
Frege, Gottlob. 1892 (1980). Über Sinn und Bedeutung. In Günther Patzig (ed.), Funktion, Begriff, Bedeutung.
Fünf logische Studien [zuerst erschienen in: Zeitschrift für philosophie und philosophische kritik, nf 100,
25-50.], 40–65. Vandenhoek and Ruprecht.
van Geenhoven, Veerle. 1998. Semantic Incorporation and Indefinite Descriptions. Semantic and Syntactic
Aspects of Noun Incorporation in West Greenlandic. CSLI Publications.
Gernsbacher, Morton Ann & Suzanne Shroyer. 1989. The cataphoric use of the indefinite this in spoken
narratives. Memory & Cognition 17(5). 536–540.
Givón, Talmy. 1983. Topic continuity in discourse: An introduction. In Talmy Givón (ed.), Topic Continuity in
Discourse: A Quantitative Cross-language Study, 1–42. Amsterdam: Benjamins.
Heim, Irene. 1991. Artikel und definitheit. In Arnim von Stechow & Dieter Wunderlich (eds.), Semantik: ein
Internationales Handbuch der Zeitgenössischen Forschung, Bd. 6, 487–535. Berlin: de Gruyter.
15
Heim, Irene. 2011. Definiteness and indefiniteness. In Klaus von Heusinger, Claudia Maienborn & Paul Portner
(eds.), Semantics. An International Handbook of Natural Language Meaning, vol. 2, 996–1025. Berlin: de
Gruyter.
von Heusinger, Klaus. 1997. Salience and definiteness. The Prague Bulletin of Mathematical Linguistics 67. 5–
23.
von Heusinger, Klaus. 2002. Specificity and definiteness in sentence and discourse structure. Journal of
Semantics 19. 245–274.
von Heusinger, Klaus. 2011. Specificity. In Klaus von Heusinger, Claudia Maienborn & Paul Portner (eds.),
Semantics. An International Handbook of Natural Language Meaning, vol. 2, 1025–1058. Berlin: de
Gruyter.
de Hoop, Helen. 1995. On the characterization of the weak-strong distinction. In Elke Bach, E. Jelinek,
Angelika Kratzer & Barbara Partee (eds.), Quantifcation in Natural Languages, 421–450. Netherlands:
Kluwer Academic Publishers.
Ionin, Tania. 2006. This is definitely specific: Specificity and definiteness in article systems. Natural Language
Semantics 14. 175–234.
Kamp, Hans. 1981. A theory of truth and semantic representation. In Jeroen Groenendijk, Theo Janssen &
Martin Stokhof (eds.), Formal Methods in the Study of Language, Proceedings of the Third Amsterdam
Colloquium, 277–322. Amsterdam: Mathematical Center. Reprinted in: J. Groenendijk, T.M.V. Janssen,
M. Stokhof (eds.), Truth, Interpretation and Information, Selected Papers from the Third Amsterdam
Colloquium, Foris, Dordrecht, 1984, 1–41, and reprinted in: K. von Heusinger, A. ter Meulen (eds.), The
Dynamics of Meaning and Interpretation. Selected Papers of Hans Kamp, Brill, Leiden, 2013, 329–369.
Kamp, Hans. 2001-2010. Discourse structure and the structure of context. University of Stuttgart.
Kamp, Hans. 2014a. Dividing the province of indefinite noun phrase uses into three parts. Ms. Universität
Stuttgart / University of Texas at Austin.
Kamp, Hans. 2014b. Using proper names as intermediaries between labelled entity representations. to appear in
Erkenntnis .
Kamp, Hans & Agnes Bende-Farkas. 2006. Specific Indefinites: Anchors and Functional Readings.
Kamp, Hans & Uwe Reyle. 1993. From Discourse to Logic. Introduction to Modeltheoretic Semantics of
Natural Language, Formal Logic and Discourse Representation Theory, vol. 42 Studies in Linguistics and
Philosophy. Dodrecht: Kluwer Academic Publishers.
Karttunen, Lauri. 1969. Discourse referents. In International Conference on Computational Linguistics
COLING 70, 1–37. Stockholm: Research Group For Quantitative Linguistics. Reprinted in: James D.
McCawley (ed.) 1976. Syntax and Semantics, 363–385. New York: Academic Press.
Kehler, Andrew, Laura Kertz, Hannah Rohde & Jeffrey L. Elman. 2008. Coherence and coreference revisited.
Journal of Semantics 25. 1–44.
King, Feffrey C. 1988. Are indefinite descriptions ambiguous? Philosophcal Studies 53. 417–440.
Kratzer, Angelika. 1998. Scope or pseudoscope? Are there wide-scope indefinites? In Susan D. Rothstein (ed.),
Events and grammar, 163–196. Dordrecht: Kluwer.
Ludlow, Peter & Stephan Neale. 1991. Indefinite descriptions: In defense of Russell. Linguistics and Philosophy
14(2). 171–202.
Milsark, Gary. 1974. Existential Sentences in English. Cambridge, MA: Massachusetts Institute of Technology
dissertation.
Neale, Stephen. 1990. Descriptions. Cambridge, MA: MIT Press.
Onea, Edgar & Ljudmilla Geist. 2011. Indefinite determiners and referential anchoring. International Review of
Pragmatics 3(2). 194–227.
Partee, Barbara. 1970. Opacity, coreference, and pronouns. Synthese 21. 359–385.
Pesetsky, David. 1987. Wh-in-situ: Movement and unselective binding. In Eric J. Reuland & Alice G. B. ter
Meulen (eds.), The Representation of (In)definiteness, 98–129. Cambridge, MA: MIT Press.
Prince, Ellen F. 1981. Toward a taxonomy of given-new information. In P. Cole (ed.), Radical pragmatics, 223–
255. New York: Academic Press.
Rivero, Marìa-Luisa. 1975. Referential properties of spanish noun phrases. Language 51. 32–48.
Russell, Bertrand. 1905. On denoting. Mind 14(56). 479–493.
Sæbø, Kjell Johan. 2013. Reports of specific indefinites. Journal of Semantics 30. 267–314.
Stalnaker, Robert. 1998. On the representation of context. Journal of Logic, Language, and Information 7. 3–19.
Strawson, Peter Frederick. 1950. On referring. Mind 59. 320–344.
Strube, Michael & Udo Hahn. 1999. Functional centering: grounding referential coherence in information
structure. Computational Linguistics 25(3). 309–344.
de Swart, Henriëtte. 2013. Indefinites. Oxford Bibliographies .
Wright, S. & Talmy Givón. 1987. The Pragmatics of indefinite reference: Quantified text-based studies. Studies
in Language 11(1). 1–33.
16