Content uploaded by Helen Hint
Author content
All content in this area was uploaded by Helen Hint on Jan 23, 2017
Content may be subject to copyright.
Content uploaded by Helen Hint
Author content
All content in this area was uploaded by Helen Hint on Jan 23, 2017
Content may be subject to copyright.
Available via license: CC BY-NC-SA
Content may be subject to copyright.
THIRD PERSON PRONOUN FORMS IN ESTONIAN
IN THE LIGHT OF CENTERING THEORY
Helen Hint
University of Tartu
Third person pronoun forms and Centering Theory in Estonian
Abstract. This paper explains the distinctions between the Estonian 3rd person overt
pronoun and the zero person marker in spoken narratives. As both forms express the most
salient entities in discourse, the saliency criterion cannot distinguish them. The Center-
ing Theory is used to explore if the overt pronoun and zero have different effects on
discourse coherence, i.e. whether there is a difference between transition types relating
to zero and those signaling the overt pronoun. Additionally, factors such as grammatical
role, case and clause type affecting the choice of pronominal forms are studied to supple-
ment results from the Centering analysis. It is hypothesized that the use of the zero form
connects to the
CONTINUE
transition, while the overt pronoun combines with other Center-
ing-based transition types as well. Furthermore, results show that the zero form is more
restricted in its usage contexts and signals mainly nominative subjects in main clauses,
while the overt form can appear more widely in different linguistic environments.
Keywords: zero person marker, third person pronoun, Centering Theory, reference
resolution, spoken narrative, Pear Stories
DOI: http://dx.doi.org/10.12697/jeful.2015.6.2.05
1. Introduction
There are various forms available to a speaker who wants to refer
to a particular entity. Expressions such as “a lady”, “my mother”, “she”
and “Ülle” might all refer to the same person in different contexts.
It is assumed that the choice of referring expression depends on the
attentional/cognitive state of the referent: more reduced forms point to
more salient entities, whereas more elaborate phrases are needed for
less salient (new) entities (see e.g. Ariel 1990, Gundel et al. 1993). One
theory concerned with the relationship between the attentional state and
the form of referring expressions is Centering Theory (Grosz et al. 1995,
Walker et al. 1998). As a theory of focus of attention and anaphora, it is
applicable to analyze reference resolution.
This article employs the Centering framework to explore the distinc-
tions between the use of the Estonian 3rd person overt pronoun and
ESUKA– JEFUL 2015, 6–2: 105–135
106 Helen Hint
the zero person marker in spoken narrative discourse. The Pear Film
(Chafe 1980) is used to collect the narratives. In Estonian some contexts
allow the use of zero forms as anaphoric devices, although this is not
con sidered to be a default choice.1 Rather, the 3rd person short pronoun
ta ‘s/he’ (PL nad ‘they’) is the most expected choice for referring to
highly salient entities. Usually, zero can be replaced with an overt
pronoun without any changes in the grammaticality of the clause. The
opposite possibility is quite rare. It has been noted that in narrative
contexts it is possible to leave out the subject if it remains unchanged in
a longer sequence of sentences (Lindström 2005: 175). As yet there are
no systematic analyses regarding the distinction of zero and the overt
3rd person pronoun ta in Estonian.
This paper aims to fill the gap by exploring the relation between
Centering transitions and different forms of referential pronouns in
Estonian spoken narratives. The main reason for applying the Centering
Theory in the present work is that it is the only carefully implemented
framework that allows us to test the differences between referential
forms by following a rigid, fixed procedure. As far as I know, it is the
first attempt to apply the Centering framework to Estonian. However,
note that Centering is used here merely as an explanatory linguistic
tool and not as a computational method with practical application. To
give a more extensive picture of factors triggering the use of overt 3rd
person pronouns or the zero person marker, the Centering analysis is
supplemented by qualitative analysis considering various grammatical
features of corresponding utterances.
The questions asked are as follows.
1) Is there a difference between the transition types signaling zero
and those signaling the overt pronoun? It is hypothesized that the
use of zero forms strongly connects to the CONTINUE transition. The
relationship appears to be a cross-linguistic phenomenon (see Di
Eugenio 1998, Iida 1998). At the same time, the pronoun ta should
combine with other Centering-based transition types as well.
2) How do grammatical factors such as grammatical role, case and
clause type (main vs. subordinate) influence the choice of pronom-
inal forms, and how do they supplement the results from the
Centering analysis?
The organization of the article is as follows. Section 2 gives a brief
overview of Estonian 3rd person anaphoric expressions and the research
1 A similar phenomenon is observed for example in English as a diary-style null subject,
see e.g. Scott (2013).
ird person pronoun forms and Centering eory in Estonian 107
done in this area. The basic notions and ideas of Centering Theory
are explained in section 3. In section 4, the material and method of
the present study are introduced. Section 5 presents the results of the
analysis and answers the research questions. Finally, general discussion
and conclusions can be found in section 6.
2. The Estonian 3rd person anaphoric paradigm
Various anaphoric devices are available for referring back to 3rd person
entities in Estonian. These include
1) short and long forms of personal pronouns:
SG
ta/tema ‘s/he’ (
PL
nad/
nemad ‘they’);
2) demonstrative pronouns: SG see ‘this’ (PL need ‘these’) and SG too
‘that’ (PL nood ‘those’)
3) zero person marker (mostly zero subjects).
In this article, the term zero person marker stands for the absence
of a phonological form in grammatical person interpretation, as Siewi-
erska (2004: 22) has defined it. Note that Kibrik (2011: 121) includes
only overt devices under the notion of pronoun and opposes zero refer-
ence to overt forms, thereby excluding zero reference from pronouns.
However, as the present work considers zero forms in contexts similar
to overt pronoun usage, the term zero pronominal is deemed more
precise. Works discussing Estonian pronouns do not usually list the zero
form as a feature of the language’s pronominal system (e.g. Pajusalu
2005, 2009, Kaiser and Hiietam 2003). Estonian traditional grammar
describes the phenomenon as contextual ellipsis, stating that it serves as
an anaphoric substitute for a previously mentioned initial form (EKG II:
223). The term ‘subjectless clause’ has also been used (Lindström 2001).
It is true that the usage of referential zero person markers is some-
what restricted. According to the traditional view, only 1st and 2nd
person pronouns can be realized by zero in sentences where the predi-
cate verb form also carries a personal ending (EKG II: 223). However,
the 3rd person zero form (ellipsis) is considered acceptable in subse-
quent sentences sharing similar context and a common subject (EKG II:
225). It has also been noted that the 3rd person zero is mainly charac-
teristic of narrative contexts and spoken language (see Lindström 2005:
175, Kivik 2010). Lindström (2001) has studied subject ellipsis in narra-
tives and concludes that ellipsis (zero subject) is possible in contexts
where the agent remains the same for a longer period, and the utter-
108 Helen Hint
ances containing zero build up a whole episode in the narrative. Also,
entities expressed with zero are usually subjects (Ibid.). More recent
studies have also shown that the use of ellipsis is highly usual in Esto-
nian conversational data (see Vihman 2015). Therefore, Estonian may
be described as a language including both full and reduced pronominal
forms2 (Siewierska 2003, 2004).
Traditionally, the choice of referential expressions is explained
in terms of salience or the accessibility of the referent. According to
different accessibility hierarchies (e.g. Gundel et al. 1993, Ariel 1990,
Givón 1983, see also Kaiser and Trueswell 2008), more reduced forms,
i.e. those with the least phonetic content, refer to most salient enti-
ties, whereas fully specified longer forms refer to less salient entities.
Following this idea, two hypotheses may be drawn: firstly, Estonian
demonstrative pronouns refer to less salient entities compared to short
3rd person pronouns
3
, and secondly, the referent of the zero pronominal
should be more accessible than ta.
Also taking into consideration the animacy hierarchy as described by
Siewierska (2004: 149), stating that human entities are more salient than
other animate entities and animate entities outperform abstract ones,
the salience/accessibility view is compatible with the dif ferences in
Estonian personal pronouns and demonstratives. The personal pronoun
ta/tema is mostly restricted to human or animate referents recently
mentioned in the discourse. The demonstratives see and too typically
express inanimate referents, although in some contexts they may indi-
cate animate entities as well. When there are two similar animate refer-
ents in the discourse, ta is used to refer to the first and a demonstrative
to the second entity mentioned. (EKG II: 208–209, see also Pajusalu
2006) More recent studies have shown that ta refers to the most acti-
vated referents in discourse and demonstratives tend to refer to entities
not yet in the focus of attention of the addressee (Pajusalu 2006, 2009).
2 Siewierska (2003, 2004) has used the term reduced pronominal to denote pronominal
forms with no overt form, such as zero pronouns, as opposed to f ull pronominal forms,
such as personal pronouns. This term should not be confused with the notion of re-
duced referential device, which includes all possible pronominal forms as opposed to
lexically full referential devices (e.g. Kibrik 2011).
3 This conclusion takes into account only the short form of the 3rd person pronoun. The
long form tema is actually longer than demonstrative pronouns. Note that the usage
of long forms is largely guided by pragmatic factors, such as contrast (Pajusalu 2009,
Kaiser and Hiietam 2003). Therefore, it is not considered the common neutral choice
when referring to highly salient entities. Although tema refers to just as salient entities
as the short form ta, its special function must be taken into consideration and the posi-
tion of Estonian long pronoun forms on the salience scale needs further investigation.
ird person pronoun forms and Centering eory in Estonian 109
Additionally, the personal pronoun ta is sensitive mostly to the gram-
matical role of the referent, preferring subjects; the use of the demon-
strative see is more complex, expressing preferences in grammatical
role (object) as well as word order (postverbal arguments) (Kaiser and
Vihman 2010). The choice between a short and a long pronoun form is
considered to be pragmatic and factors such as contrast come into play
when choosing them (Pajusalu 2009, Kaiser and Hiietam 2003).
While the distinction between Estonian personal and demonstrative
pronouns is relatively well studied, the difference between the referen-
tial properties of the zero person marker and the personal pronoun ta/
tema is not so clear. The widely approved Givenness Hierarchy (Gundel
et al. 1993) states that (unstressed) personal pronouns and zero forms
may only refer to highly salient entities and both express the highest
status in focus. This leaves salience quite incapable of differentiating
overt and zero pronominal forms. The equal level of salience of these
forms is supported by the circumstance that zero and ta are inter-
changeable in many contexts. Two excerpts from narratives (examples
1 and 2) suggest that there is no difference in the cognitive salience of
ta and zero: both of them refer to the most salient entity in the ongoing
discourse.
(1) 1 ja= ja siss em (.) see: se= talumees ee korja-s pirn-e.
and and then this this farmer pick-3SG.PST pear-PL.PART
2 mm: ja ø= roni-s sinna ‘redeli otsa
and 3SG.ZERO climb-3SG.PST there ladder.GEN up there
3 jaa= ja siis økorja-s ästi hoolega= neid
and and then 3SG.ZERO pick-3SG.PST very carefully them.PL.PA RT
‘and and then umm this this farmer ee was picking pears; mm and he
climbed up there onto a ladder; and then was picking them very care-
fully’
(2) 1 ja= sis natukese aja pärast ää lähe-b mööda
and then a little.GEN time.GEN after go-3SG.PRS by
mingi: väike põnn (.) ästi suure jalgratta-ga.
some little kid very big.GEN bike-COM
110 Helen Hint
2 jaa: m (.) sis= ta alguse-s mõtle-b
and then 3SG.SHORT beginning-IN think-3SG.PRS
3 et= ta sõida-b nisama mööda
that 3SG.SHORT ride-3SG.PRS just by
‘and then in a little time ää some little kid with a very big bike goes by;
and m at the beginning he thinks; that he will just ride by.’
As the zero person marker and the pronoun ta express an equal
level of salience, some other factors must be crucial in differentiating
them. Recent cross-linguistic studies (see e.g. Kaiser and Trueswell
2008, Kaiser 2009, Brown-Schmidt et al. 2005, Byron et al. 2008) have
also shown that although salience has a clear effect on the choice of
referring expressions, various form-specific factors should be taken
into consideration in explaining the use of distinct referential forms. A
multiple-constraint approach may outperform the uniform dimension
(e.g. salience) perspective. Adapting this idea and the Centering Theory
framework in this article, it will be examined if zero and ta, although
equal in salience, have distinct effects on discourse coherence.
3. Centering theory: an overview
A brief introduction concerning the key terms of Centering Theory
is given in this section. The overview is based on two central detailed
works on Centering: Grosz, Joshi, and Weinstein (1995) and Walker,
Joshi, and Prince (1998).
The Centering framework explains discourse coherence within a
discourse segment by bringing together attentional state factors and the
choice of referring form (Grosz et al. 1995: 204, Walker et al. 1998). The
theory is concerned with the distribution and interpretation of referring
expressions in natural language (Gundel 1998: 190). The notion infer-
ence load is used to describe changes in the hearer’s attentional focus
during the task of identifying and tracking entities through the text. In
a coherent text, this task of tracking entities in a given discourse should
be made as easy as possible to reduce the inference load placed upon a
hearer. (Grosz et al. 1995, Walker et al. 1998) The choice of linguistic
expression directly influences the inference load and hence one of the
basic claims of Centering states that the perceived discourse coherence
ird person pronoun forms and Centering eory in Estonian 111
depends on the distinct inferential demands positioned on a hearer/
reader by the usage of different referring expressions and syntactic
forms (Grosz et al. 1995: 206–207). Thus the Centering model imitates
psychological processes involved in tracking entities through the
discourse (Ballantyne 2004: 50). The theory is mainly interested in the
interpretation of anaphoric expressions (e.g. pronouns, zero anaphora,
definite descriptions) and is focused on noun phrases (NPs). (Grosz et
al. 1995, Walker et al. 1998)
Centering as a model of analyzing discourse defines it as a coherent
sequence of utterances. Units of discourse are utterances, i.e. “the
uttering of a sequence of words at a certain point in a discourse”, not
sentences (Grosz et al. 1995: 208). In each discourse segment, local and
global coherence are exhibited. Local coherence is coherence among
utterances in a particular segment, while global coherence pertains
to coherence with other segments in ongoing discourse (Grosz et al.
1995: 204). Although Centering was originally concerned with the local
coherence of discourse segments (Grosz et al. 1995), there have been
attempts to apply the centering algorithm to global coherence as well
(Hedberg and Dueck 1999, Cristea et al. 1998, Di Eugenio 1998). Walker
(1998) even suggests that the within-segment approach of Centering
should be replaced with a version of Centering that accounts for global
discourse structure. The present work also deviates from the original
notion of Centering in that segment boundaries are not explicitly taken
into account and no global/local discourse distinction is made.
3.1. Centers
The basic term of the theory is center. Centers are semantic entities
in an utterance, part of a discourse model concerning the utterance in
a segment, that link the utterance to other utterances in a particular
segment (Grosz et al. 1995: 208, Walker et al. 1998: 3). As Grosz and
her colleagues (1995: 208) point out, centers are discourse constructs
and semantic objects, not words, phrases or syntactic forms. Centers
are considered to regulate the local information flow in discourse (Hu
and Pan 2001: 139).
Three kinds of centers are distinguished in the centering framework:
forward-looking, backward-looking and preferred centers (Grosz et al.
1995, Walker et al. 1998, see also Ballantyne 2004, Taboada and Hadic
Zabala 2008).
112 Helen Hint
1) All discourse entities mentioned in an utterance U
n
in a discourse
segment make up a set of forward-looking centers (Cf), which are
ranked by means of salience.
2)
The first, i.e. the most salient member of the Cf set in utterance Un is
called the preferred center (Cp). The Cp predicts what the backward-
looking center Cb of the following utterance will presumably be.
3) A special member of the Cf set, the backward-looking center (Cb),
is the entity which is in the center of attention in the utterance U
n
.
Cb is the highest-ranking entity from the set of Cf of the previous
utterance Un-1, which is realized also in the current utterance.
Therefore, Cb serves as the link between the present and previous
utterance in a particular discourse segment.
When applying Centering to a new language, the Cf ordering prin-
ciples must be specified. Researchers have applied different Cf ranking
criteria to different languages. The most traditional criterion has been
the grammatical function (subject > object(s) > other), where subjects
are classed higher than objects and objects higher than other func-
tions (Walker et al. 1998: 7). Several other criteria have been used in
different works related to Centering as well, such as topicality (Walker
et al. 1994), empathy (Di Eugenio 1998), the Givenness Hierarchy
(Ballantyne 2004) and animacy (see Taboada 2008). The description
of ordering principles followed in this article can be found in section 4,
Material and Method.
3.2. Transitions
By comparing the centers of utterances U
n
and U
n-1
, the transition
types are computed within the Centering framework. Two factors are
important in the typology of transitions: the comparison of Cb of Un
and Un-1, and the comparison of the Cb of Un with the Cp of Un. The
transition types are shown in Table 1, where Cb(Un-1) = [?] stands for
cases where Cb(Un-1) does not exist. (Walker et al. 1998: 5–6)
Table 1. Transition types in CT (Walker et al. 1998: 6)
Cb(Un) = Cb(Un-1)
or Cb(Un-1) = [?]
Cb(Un) ≠ Cb (Un-1)
Cb(Un) = Cp(Un)CONTINUE SMOOTH SHIFT
Cb(Un) ≠ Cp(Un)RETAIN ROUGH SHIFT
ird person pronoun forms and Centering eory in Estonian 113
U
n
and U
n-1
are linked by the
CONTINUE
transition when the Cb and
the Cp of the current utterance are the same and the Cb at hand simul-
taneously coincides with the Cb of the previous utterance. This means
that the speaker is talking about a certain entity and aims to keep it in
the center of attention in the next utterance(s) as well (Taboada and
Hadic Zabala 2008: 70–71, Walker et al. 1998: 5) A RETAIN transition
relates two utterances in cases where the Cb of utterances Un and Un-1
are the same, but the Cb and the Cp of the present utterance do not
coincide. If the Cb of the current utterance is not concurrent with the
Cb of the previous utterance, then there is a SHIFT relation between two
utterances: a SMOOTH SHIFT in cases where the Cb is the same as the Cp
of U
n
, and a ROUGH SHIFT in cases where the Cb is not the same as the
Cp in Un. Transitions help to explain how coherence is achieved: a text
is more coherent if it maintains the same centers (Taboada and Hadic
Zabala 2008: 71).
Example (3), an excerpt from a coherent discourse segment, helps
to understand these basic centering notions. In the first sentence, there
are two centers: boy (realized with a zero form) and basket. They are
listed in a set of Cf according to the grammatical function. The first
member of the Cf list is boy, so it is the preferred center Cp. As this is
not the first utterance of the segment, it has a backward-looking center
Cb, which is boy. The second utterance has three centers: boy, bike
and pear. The highest of them is boy, which is also the Cb of this utter-
ance and the Cp for the following utterance. Therefore, the transition
between the first and second utterance is
CONTINUE
. In the last utterance,
three Cf-s are realized: girl, boy and road. However, the first member
of the Cf list is no longer boy, instead it is girl, a grammatical subject of
this utterance. girl is also the Cp in this utterance. Still, the Cb is boy,
thus the transition is RETAIN.
(3) 1 ja= sis øvõttis terve selle korvi
and then 3SG.ZERO take.3SG.PST whole this.GEN basket.GEN
‘and then he took the whole basket’
Cf: boy, basket
Cb: boy – Cp: boy
transition: CONTINUE
114 Helen Hint
2 ja: øpani punuma koos= oma ratta-ga
and 3SG.ZERO put.3SG.PST pleach.mINF with his bike-COM
selle pirni-ga.
this.GEN pear-COM
‘and (he) turned tail with his bike with the pear’
Cf: boy, bike, pear
Cb: boy – Cp: boy
transition: CONTINUE
3 ää ja siis ee tee peal tuli ta-lle
and then road on come.3SG.PST 3SG.SHORT-ALL
vastu mingi: tüdruk
against some girl
‘ää and then ee on the road some girl came towards him’
Cf: girl, boy, road
Cb: boy – Cp: girl
transition: RETAIN
3.3. Rules and constraints
Besides transitions, the Centering framework involves rules and
constraints.
1) Rule 1, also known as a Pronoun Rule, states that if any element of
the set of Cf of the previous utterance Un-1 is realized as a pronoun
in the current utterance Un, then the Cb of the current utterance
Un must also be a pronoun. This indicates that when the speaker
continues to talk about the same entity, it is a salient one, which is
(usually) referred to by pronouns. Therefore, according to rule 1,
the least marked referential form in the utterance refers to the most
salient entity. (Walker et al. 1998: 4–5, Grosz et al. 1995)
2) Rule 2 ranks the transition types within Centering, stating that
CONTINUE is preferred to the RETAIN transition, which is preferred
to the SMOOTH-SHIFT transition, which is preferred to the ROUGH-
SHIFT transition (Walker et al. 1998: 4). The ordering of the transi-
tion types is based on the processing time they require from the
addressee and thus makes it possible to measure the coherence of a
ird person pronoun forms and Centering eory in Estonian 115
discourse segment, as the most coherent segments require less time
for processing (Walker et al. 1998: 6). It implies that in a coherent
discourse, CONTINUE transitions should predominate, while ROUGH-
SHIFT should occur rarely.
Constraints, as they are formulated by Walker and colleagues
(1998: 3), state that:
1) each utterance in a discourse segment has precisely one Cb;
2) every element listed in the Cf set must be realized in that utterance;
3) the Cb of the current utterance is the highest-ranked element in the
set of Cf of the previous sentence that is realized in the current utter-
ance.
The term realized element includes pronouns, zero pronominals, explic-
itly realized discourse entities and implicitly realized entities inferable
from the discourse situation (see Walker et al. 1998: 4). Constraint 1
is also regarded as the central claim of the theory (Poesio et al. 2004).
A classic example explaining Rule 1, the importance of Cf ranking
and Cb realization by a pronoun comes from Gordon et al. (1993: 313).
Passage (4) is given two possible final utterances, 3 and 3’.
(4) 1 Susan gave Betsy a pet hamster.
Cf: Susan, Betsy, hamster
2 She reminded her such hamsters were quite shy.
Cf: Susan, Betsy, hamsters
Cb: Susan
3 She asked Betsy whether she liked the gift.
3’ Susan asked her whether she liked the gift.
Cf: Susan, Betsy, gift = hamster
Cb: Susan;
Here, utterance 1 has no Cb as it is the first utterance of the segment.
Its Cf list is ranked according to grammatical role, yielding Susan as
the first element. In utterance 2, Susan is the Cb and again the highest-
ranking Cf. In utterance 3, the Cb (Susan) is preserved and realized by
a pronoun, and it is a natural-sounding coherent continuation of the
previous segment. However, in 3’ the Cb (Susan) is realized by a proper
name and a pronoun is used to refer to the second entity in the previous
116 Helen Hint
Cf list (Betsy). This results in a very unnatural result and the utterance
3’ reads poorly.
4. Material and method
In this study, a Centering analysis is applied to experimentally elic-
ited spoken narratives based on the Pear Film
4
(Chafe 1980). The design
of the research follows the one worked out by Wallace Chafe and his
coworkers. 16 female high-school or university students, aged 18–26,
all native speakers of Estonian voluntarily participated in the study. To
hedge against possible gender-based differences in pronoun use in this
study, only female subjects are included. The subjects are shown the
Pear Film: a short motion picture without explicit language about a man
who is picking pears and a boy who steals some pears from him. After
seeing the footage, the subject is asked to retell the content of the film
to a friend who has not seen it. All hearers are approximately the same
age as the speaker. Note that in the original experiment, the hearer was
not acquainted with the subject. However, I find it best to have the story
told to friends since it makes one feel more comfortable and therefore
provides more natural narratives. Having just seen the movie, all refer-
ents must be known for the speaker. Yet for the hearer every single one
of them is new at the beginning of the narrative. Therefore, the experi-
ment provides us with information about how the speaker estimates the
cognitive states of the hearer.
The coding procedure largely followed the manual by Hadic
Zabala and Taboada (2006, see also Taboada 2008), as it takes into
con sideration the characteristics of spoken language. In this study,
an utterance is defined as an intonation unit (Taboada 2008: 181).
An ut terance is generally a tensed clause (Hadic Zabala and Taboada
2006). If the unit does not fill the requirements of a clause, it might be
considered an utterance if it contains entities (Ibid.). Only utterances
containing the 3rd person overt pronoun or the zero person marker are
coded and analyzed for this study. In total, 517 cases of pronoun use
(overt or zero form) were included in the study.
Besides the proper centering analysis, grammatical factors
pertaining to the use of pronoun forms under study, such as (1) case, (2)
grammatical role (subject, other), and (3) clause type (main or subordi-
4 The fi lm can be viewed at http://pearstories.org/.
ird person pronoun forms and Centering eory in Estonian 117
nate clause) are coded in order to gain a fuller picture of the differences
between these forms in Estonian spoken language.
In the analysis procedure, every narrative was first segmented into
utterances and for each utterance, the list of forward-looking centers Cf,
the b
ackward-looking center Cb and the preferred center Cp were coded.
As deciding on the Cb and the Cp of an utterance directly depends on the
ordering of the Cf list, it is crucial in adapting the Centering Theory to a
new language to determine what ordering best suits that language. The
most appropriate ordering for a language would result in the least viola-
tion of the Pronoun Rule. The most often used ranking principle follows
grammatical relations, according to which subjects are
ranked higher
than objects and objects are ranked higher than adverbials (Walker et
al. 1998). However, as the ordering of entities realized in a sentence
is considered to be language-dependent, different Cf templates occur
(see Taboada 2008). In this article the original grammatical function
ordering principle has been modified to fit Estonian data.
In Estonian, the grammatical subject does not always indicate the
most agentive argument in the clause (EKG II: 14). Besides proto typical
subjects, many subject-like arguments exhibit some subject properties
(see Metslang 2013). According to Erelt and Metslang (2006), clauses
can be divided into unmarked and marked basic clauses, based on
the realization of the main topic of the clause. Marked basic clauses
include existential clauses, possessive clauses, source-marking resulta-
tive clauses and experiential clauses. While in unmarked basic clauses,
the main topic is realized as a grammatical subject, in unmarked ones
it appears as an adverbial, an oblique or a direct object. (Erelt and
Metslang 2006) However, in all marked basic clauses, the topic has
some subject-like features (e.g. the clause-initial position of the expe-
riencer or possessor) and it is often an animate entity (Ibid.). Examples
of marked basic clauses can be found in (5) (a possessive clause) and (6)
(an experiential clause).
(5) Mari-l on raamat.
Mary-AD be.3SG book
‘Mary has a book.’
(6) Mari-le meeldi-b muusika.
Mary-ALL like-3SG music
‘Mary likes music.’
118 Helen Hint
As the ordering criteria are based on the salience of the referent, it is
clear that the topic must rank higher in the hierarchy than grammatical
subject. This is also supported by the fact that if utterance (6) were
followed by another utterance in a segment, it is most probably Mari
who would be mentioned by a pronoun, not music.
In this article, the notion of oblique subject is applied to account for the
non-nominative animate subject-like topic of marked basic clauses. It
is clear that sticking strictly to grammatical function in ordering the Cf
list in Estonian would result in more frequent violation of the Pronoun
Rule. The Cf ordering criteria followed in this article are as follows (7):5
(7) oblique
subject >grammatical
subject >grammatical
object2 >grammatical
object1 > other5
In ranking the Cf list of an utterance, it must also be decided what
to do with NPs containing more than one entity (e.g. possessives, such
as minu vend ‘my brother’, and conjoined NPs, such as Jaan ja Mari
‘John and Mary’). According to Di Eugenio (1998), if the possessee
is inanimate, the possessor is ranked before the possessee; and if the
possessee is animate, the possessor is ranked after the possessee. For
conjoined NPs, the individual entities are less salient than the whole
group denoted by the conjoined NP. Therefore, the group as a whole
must rank first and individual entities are subsequent, following the
linear order.
The decision as to whether there is a zero in an utterance is based
on the valency of the verb. For example, in the case of intransitive verbs
which take one argument, the utterance must contain at least one verb
and a subject. If there is no overt NP expressing the subject, then it
must be a zero form. In case of transitive verbs with two arguments, the
utterance must have at least a subject and an object along with the verb;
when one of them is not expressed with an explicit NP, it is considered
to be a zero. Only cases where the obligatory argument is missing are
considered to be instances of a zero form.
For the analysis, all zero forms appearing in the data are considered
to be informative regarding their cognitive status, therefore every single
zero form is included. The question of coordination and its influence
on the usage of zero forms is considered to be a syntactic issue and is
therefore neglected here.
5 When several NPs apper in fu nction other than oblique subject, grammatical subject or
object, linear order is followed in ordering.
ird person pronoun forms and Centering eory in Estonian 119
Cases where the interpretation of the zero person marker is ambig-
uous are excluded from the analysis. Example (8) illustrates such a situ-
ation. Here, the zero form in the second utterance may either refer to the
boy with a bike or to one of the boys who took the hat to the boy with a
bike. Also omitted are speakers’ comments not related to the narrative
discourse.
(8) 1 ja siis ee (.)üks poiste-st vii-s selle (.) ta-lle?
and then one boy.PL.GEN-EL bring-3SG.PST this.GEN 3SG.SHORT-ALL
‘an d th e n ee one of the boys took this to him’
2 ja siis õ (.) õ ø võttis= sea-lt kolm pirni.
and then 3SG.ZERO take.3SG.PST there-ABL three pear.PL.PART
‘and then õ õ (he) took three pears from there’
5. Results
The quantitative data for pronominal forms under examination is
presented in this section. Note that due to the infrequent occurrence of
plural forms, chi-square tests reported in this study are only applied to
compare the 3rd singular overt form ta and the 3rd singular zero form6.
There were 417 instances of
SG
pronominal forms altogether. Even so,
for the sake of the complete overview of the data, the contingency tables
present details for the plural forms also (100 instances altogether).
The initial assumption that the zero person marker signals CONTINUE
and the overt pronoun is used more in cases of other transitions is not
fully supported. It is true that there appears to be a tendency of zero
only being used with
CONTINUE
, while overt forms may express transi-
tions other than CONTINUE as well. However, for both the overt form as
well as zero, CONTINUE is the most exploited transition. Therefore other
differentiating aspects besides Centering transitions are also examined.
The overall distribution of pronominal forms with respect to the
Centering transition type is shown in Table 2. What the results clearly
indicate is that the ordering rule (Rule 2) applies to analyzed data and
therefore the narratives used for this study are examples of coherent
6 As the observations from the corpus used here are not totally independent, the chi-
square test results are used as a suggestive evidence, not as a strong statistic proof.
120 Helen Hint
text. The most frequent transition is CONTINUE (424 utterances), as the
theory predicts. At the other end are
SHIFTS
(35), which are rather rare
in the data. In between are RETAINS (38). In the data there were also 20
utterances without Cb and thus no transition type was applicable to
them. A small number of shifts is of course anticipated since the theory
itself predicts that full NPs are more likely to signal them. The cases
where no Cb can be determined might be considered as the beginnings
of new discourse segments, although this assumption needs further
investigation. Note that in Table 2, the SMOOTH- and ROUGH SHIFT are
not distinguished. This is due to the fact that it is very uncommon for
utterances containing pronouns to express ROUGH SHIFTS (see also Di
Eugenio 1998). Even Grosz and colleagues (1995) in their original paper
on Centering see center shifting as a uniform category.
It is important to note that the zero pronominal is used even more
frequently than the overt form. In singular forms, it can be observed
that the frequency of the 3rd person pronoun ta and the zero form is
roughly equal in the data (204 and 213, respectively). The occurrence
of plural pronouns is modest (nad is used in 34 cases and plural zero in
66 cases), but a significant fact is that the plural zero is used more often
than the plural overt form. Note also that while for the
CONTINUE
tran-
sition all observed pronoun options are possible, other transitions are
more constrained regarding the alternative forms. Form-wise, it appears
to be significant that the overt pronoun ta exhibits more variation in
transition types compared to zero. The zero form, on the other hand, is
mainly restricted to the
CONTINUE
transition (X
2
(3, N = 417) = 44,686,
p < .01).
Table 2. Distribution of transitions by pronoun form
FORM CONTINUE RETAIN SHIFT No Cb TOTAL
ta 140 30 17 17 204
zero (SG) 197 3 11 2 213
SG forms total 337 33 28 19 417
nad 2553134
zero (PL) 62–4–66
PL forms total 87571100
TOTAL 424 38 35 20 517
ird person pronoun forms and Centering eory in Estonian 121
In (9) a typical example of using the zero person marker in a narra-
tive is given. All the transitions are
CONTINUES
, there is only one animate
entity referred to in these utterances and the zero form is always in the
nominative case and the grammatical subject of the clause.
(9) 1 emm seal= oli üks mees kes korja-s pirne,
umm there be.3SG.PST one man who pick-3SG.PST pear.PL.PART
pirni-puu otsast, redeli-ga.
pear-tree from upon ladder-COM
‘umm there was a man who was picking pears from the pear tree,
with a ladder’
Cf: man, pears, tree, ladder
No Cb – Cp: man
transition: NONE
2 jaa: økorja-s pirne rahulikult.
and 3SG.ZERO pick-3SG.PST pear.PL.PART calmly
‘and (he) was picking pears calmly’
Cf : man, pears
Cb: man – Cp: man
transition: CONTINUE
3 jaa øpani= nad siis= tõi
and 3SG.ZERO put.3SG.PST 3PL.SHORT then bring.3SG.PST
pirni-d ilusti alla oma korvi,
pear-PL neatly down his basket.ILL
‘and (he) put them then brought the pears down neatly to his basket’
Cf: man, pears, basket
Cb: man – Cp: man
transition: CONTINUE
4øjättis nad sinna (.)
3SG.ZERO leave.3SG.PST 3PL.SHORT there
‘(he) left them there’
Cf: man, pears
Cb: man – Cp: man
transition: CONTINUE
122 Helen Hint
A similar excerpt with CONTINUE transitions is given in (10), but here
the overt pronoun ta appears in contexts where zero was used in the
previous example.
(10) 1 sis ää ø jäi seisma sinna: (.) puu alla.
then ää 3SG.ZERO stay.3SG.PST stand.mINF there tree under
‘then umm (he) stopped under that tree’
Cf: boy, tree
Cb: boy – Cp: boy
transition: CONTINUE
2s= ta vist taht-is alguse-s võtta
then 3SG.SHORT probably want-3SG.PST beginning-IN take.dINF
ühte pirni= aga
one.PART pear.PART but
‘then he probably at first wanted to take one pear but’
Cf: boy, pear, beginning
Cb: boy – Cp: boy
transition: CONTINUE
3s= ta ütle-s= et no kui juba sis juba.
then 3SG.SHORT say-3SG.PST that well if already then already
‘then he said that if already so, so be it’
Cf: boy
Cb: boy – Cp: boy
transition: CONTINUE
4 ja sis ta võttis kogu korvi.
and then 3SG.SHORT take.3SG.PST whole basket.GEN
‘and then he took the whole basket’
Cf: boy, basket
Cb: boy – Cp: boy
transition: CONTINUE
Yet another display of a sequence where the overt pronominal form is
used is presented in (11). Here other transitions besides CONTINUE are
also present, hence the occurrence of zero would be unlikely in contexts
similar to this one.
ird person pronoun forms and Centering eory in Estonian 123
(11) 1 sis= øsõitis seal mingi kruusatee peal.
then 3SG.ZERO ride.3SG.PST there some gravel road on
‘then rode there on some gravel road’
Cf : boy, gravel road
Cb: boy – Cp: boy
transition: CONTINUE
2 sis= ta-lle tuli vastu üks tüdruk, ka ratta-ga.
then 3SG.SHORT-ALL come.3SG.PST against one girl also bike-COM
‘then a girl came towards him, also with a bike’
Cf: girl, boy, bike
Cb: boy – Cp: girl
transition: RETAIN
3 ja siis ää se= poiss jäi ‘vaatama kuidagi
and then this boy stay.3SG.PST look.mINF somehow
‘and then ää the boy kept looking somehow’
Cf: boy
Cb: boy – Cp: boy
transition: CONTINUE
4 ja= sis ee se= tüdruk ee lükka-s= ta-lt mütsi
and then this girl push-3SG.PST 3SG.SHORT-ABL hat.GEN
pea-st ära. (.)
head-EL off
‘and then ee the girl ee pushed his hat off his head’
Cf: girl, boy, hat
Cb: boy – Cp: girl
transition: RETAIN
Although some difference in preferred transition types of overt
forms and zeros can be observed, as in examples (9) and (11), there
is still a coinciding tendency for both to code mainly CONTINUES, as
examples (9) and (10) illustrate. This indicates that discourse coherence
factors alone cannot completely explain why one form is chosen instead
of other. Hence other possible defining factors must be considered as
124 Helen Hint
well. The important aspects that need closer examination, as the exam-
ples also indicate, are grammatical role and case, as well as clause type.
A prototypical grammatical subject is expressed with the unmarked
nominative case in Estonian (Metslang 2013: 224), while other gram-
matical roles tend to exhibit other cases. Studying the behaviour of
subjects and subject-like arguments in Estonian written texts, Helena
Metslang (2013: 243) has found that zero-anaphora tends to express
more prototypical subjects and therefore helps to distinguish subjects
from objects. As it is often difficult to see the case marking in zero
forms, it is tempting to assume that referents in grammatical roles other
than subject cannot be expressed with zero. However, Lindström and
her colleagues (2014) have observed that it is very common in Esto-
nian to omit the experiencer marked with an oblique case (adessive
or allative) in certain experiential constructions. The distribution of
pro nominal forms by grammatical role presented in Table 3 also shows
that although the non-nominative arguments occur rarely in the data,
they are still possible in Estonian.
Table 3. Pronoun forms by grammatical role
FORM SUBJECT OTHER TOTAL
ta 109 95 204
zero (SG) 207 6 213
SG forms total 316 101 417
nad 30 4 34
zero (PL)66–66
PL forms total 96 4 100
TOTAL 412 105 517
As expected, zero pronominals almost exclusively (207 out of 213
in SG; 66 out of 66 in PL) express grammatical subjects. By contrast,
overt forms are not restricted to one grammatical relation: the overt
pronoun ta is used 95 times in a grammatical role other than subject.
The dif ference between ta and SG zero appears to be significant (X2(1,
N = 417) = 108.675, p < .01).
This data is strongly related to the results shown in Table 4, which
presents the distribution of Estonian cases used with the pronoun forms
ird person pronoun forms and Centering eory in Estonian 125
under examination
7
. The nominative, genitive and partitive are gram-
matical cases, while all others are semantic cases expressing oblique
arguments. Subjects most often take the nominative case, but partitive is
also possible in some constructions. Singular objects are expressed with
the partitive or genitive, plural objects with the partitive or nominative.
It can be seen that pronominal forms are mainly used in the nomina-
tive case, which is not surprising considering the fact that the nomina-
tive is the typical case for marking the grammatical subject. Note that
some instances of exterior locative cases (allative and adessive) belong
to marked basic clauses, where the experiencer of the clause is not the
grammatical subject of the clause. In these clauses, the oblique subject
expressed with a locative case is nevertheless considered to bear the
most important and active role in a sentence. Note also that the allative
and adessive cases can be expressed with the zero form. This result is
especially interesting as it poses a contradiction to the cross-linguistic
observation that there are probably no languages with zero obliques
(Siewierska 2003). Other studies have also shown that zero obliques
are quite usual in Estonian experiential clauses (Lindström et al. 2014).
Table 4. Pronoun form by case
FORM NOM GEN PART EL ALL AD ABL COM TOTAL
ta 114 5 11 1 29 40 3 1 204
zero (SG) 207–11 2 2 ––213
SG forms total 321 5 12 2 31 42 3 1 417
nad 31––– 3 – ––34
zero (PL) 66––– – – ––66
PL forms total 97––– 3 – ––100
TOTAL 418 5 12 2 34 42 3 1 517
Example (12) is a typical instance of a zero person marker, as it
expresses the grammatical subject in the nominative case. Similar
conditions apply for the pronoun ta in example (13), but examples (14)
and (15) show ta in contexts where zero is most unlikely to appear.
7 Note that in the data, not all possible Estonian cases (14) are present with pronominal
forms.
126 Helen Hint
(12) ja= sis økuidagi läks selle ratta-ga minema
and then 3SG.ZERO somehow go.3SG.PST this.GEN bike-COM away
‘and then (he) somehow went away with this bike’
(13) ja= sis= ta tühjenda-s selle: nagu need
and then 3SG.SHORT empty-3SG.PST this.GEN like this.PL
põlle-tasku ära
apron-pocket away
‘and then he emptied it, like these in an apron pocket’
(14) ee ja= sis sõida-b mingisugune ee: õ ‘kooli-tüdruk
ee and then ride-3SG some-kind-of school-girl
ta-lle vastu.
3SG.SHORT-ALL against
‘ee and then some kind of ee õ schoolgirl rides towards him’
(15) jaa (...) mm ta-l= oli juba umbes kaks
and 3SG.SHORT-AD have.3SG.PST already about two
korvi korja-tud
basket.PART pick-PASS.PPTCP
‘and (...) umm he had picked about two baskets already’
The preference to use the zero form for mostly nominative subjects is
also supported by the fact that a wide majority of zero forms expresses
the backward-looking center Cb in an utterance. As can be seen in
Table 5, besides marking the Cb in an utterance (146 utterances), the
overt form ta is related to instances where the pronoun is not the Cb
of the current utterance (40 utterances) or there is no Cb in the utter-
ance (18 utterances). However, except for some marginal cases, the zero
form tends to express only Cb-s (208 utterances) (X2(2, N = 417) =
55,327, p < .01). According to the Centering Theory, the Cb is the most
central entity in an utterance and each utterance can contain only one
Cb. However, in actual language use, often more than one pronoun may
appear in one utterance. In these cases, the zero form is reserved for the
Cb and the overt form refers to the less central entity, a non-Cb.
ird person pronoun forms and Centering eory in Estonian 127
Table 5. Pronoun form by Cb
FORM PRO = Cb PRO ≠ Cb NA TOTAL
ta 146 40 18 204
zero (SG) 208 3 2 213
SG forms total 354 43 20 417
nad 3121 34
zero (PL) 642– 66
PL forms total 95 4 1 100
TOTAL 449 47 21 517
Example (16) illustrates cases like these. Here, the zero form is
reserved for the Cb of the utterance (three boys), whereas the less
central entity, i.e. an entity other than the subject (the boy with a bike),
is referred to by using an overt pronoun.
(16) ja siis veel øand-si-d ta-lle
and then furthermore 3PL.ZERO give-PST-3PL 3SG.SHORT-ALL
mütsi tagasi,
hat.GEN back
‘and then (they) also gave him the hat back’
Also, in utterances where there is no Cb, a pronoun can sometimes
be used. As example (17) suggests, in such cases the general discourse
topic is realized as a pronoun. Here, utterance 4 has no Cb, as in utter-
ance 3 neither man nor pears are mentioned. However, as utterance 3
does not realize any animate entities, it is clear that ta in utterance 4
refers back to the man, who is the general discourse topic. The same
applies to utterance 6. Although here the previous utterance 5 intro-
duces a new animate entity into discourse, it is clear from the content
of the film that the pronoun ta in utterance 6 cannot refer to the man
with a goat. It has been assumed that the absence of a backward-looking
center may indicate a segment boundary (Passonneau 1998, see also
Walker 1998), and the use of zero in these cases is unlikely. However,
Walker (1998) has claimed that pronouns actually can be used in the
initial utterance of the next segment. In Estonian, an overt pronominal
is used in such cases.
128 Helen Hint
(17) 1 õ seal= oli (.) mees: farmer= või keegi. seline (.)
there be.3SG.PST man farmer or somebody such
mehiko tüüpi.
mexican type
‘õ there was a man, a farmer or somebody, like Mexican type’
Cf: man, film
Cb: film – Cp: man
2 jaa: ta korja:-s (.) puu-l (.) ta=
and 3SG.SHORT pick-3SG.PST tree-AD 3SG.SHORT
korja-s pirn-e?
pick-3SG.PST pear-PL.PART
‘and he was picking up on a tree, picking pears’
Cf: man, pears, tree
Cb: man – Cp: man
transition: SMOOTH SHIFT
3 ää jaa: (.) ja se= oli= igav töö.
and and it be.3sg.pst boring work
‘ää and and it was boring work’
4 ja= siis ta (.) ää korja-s pirn-e
and then 3SG.SHORT pick-3SG.PST pear-PL.PART
‘and then he ää was picking pears’
Cf: man, pears
Cb: ø – Cp: man
5 ja= si vahepeal= sea-lt jaluta-si-d mingisugune tüüp
and then meanwhile there-ABL walk-PST-3PL some-kind-of guy
jaluta-s vel kitse-ga mööda?
walk-PST.3SG as well goat-COM by
‘and then meanwhile walked some kind of guy, walked by with a
goat as well
Cf: man, goat
Cb: ø – Cp: man
ird person pronoun forms and Centering eory in Estonian 129
6 aga: (.) ja= sis no= ta= oli juba ‘kolm
but and then well 3SG.SHORT be.3SG.PST already three
selist suur-t ä korvi= oli pirne.
such.PART big-PART basket.PART be.3SG.PST pear.PL.PART
‘but and then well he had already three such big ä baskets of pears
he had’
Cf: man, baskets, pears
Cb: ø – Cp: man
Differences in pronoun use between clause types (main and subordi-
nate) were also examined. Table 6 presents the distribution of pronoun
forms by clause type. 169 instances of the overt form ta appeared in
main clauses, 35 in subordinate clauses. The zero form was used in the
main clause 209 times, but only 4 times in subordinate clauses. There
appears to be a significant relationship between the pronominal form
and the clause type (X
2
(1, N = 417) = 28,693, p < .01): the zero person
marker is more likely to occur only in the main clause than is the overt
pronoun ta. In the subordinate clause, the overt form is preferred to
zero.
Table 6. Pronoun form by clause type
FORM Main clause Subordinate clause TOTAL
ta 169 35 204
zero (SG) 209 4 213
SG forms total 378 39 417
nad 31 3 34
zero (PL)65166
PL forms total 96 4 100
TOTAL 474 43 517
However, it must be kept in mind that the preference for overt forms
in subordinate clauses might be connected to the circumstance that in
subordinates, a non-nominative case may be the reason triggering the
use of the overt form, as in example (18).
130 Helen Hint
(18) ja= sis ø avasta-s et ta-l= on
and then 3SG.ZERO discover-3SG.PST that 3SG.SHORT-AD be.3SG
üks korv kadunud= ja:
one basket missing and
‘and then he discovered that he has one basket gone missing’
6. Discussion and conclusion
This article has examined how the usage of the Estonian 3rd person
overt pronoun and the zero person marker in spoken narrative discourse
differ from each other. The starting point was that the overt pronoun
and zero express equal levels of salience, therefore other factors must
explain the distinction between them. The Centering Theory was
applied to see if and how Centering transitions may help to interpret
the difference between those two forms. Although a certain disparity
in the transition types was found, there was still an overlapping area
in the usage of the two forms, indicating that the effect on discourse
salience is not completely different among the overt and zero forms.
Therefore additional factors such as grammatical role, grammatical
case, the possibility to realize the backward-looking center, and clause
type were studied.
The Estonian data is compatible with previous findings that the use
of zero for
CONTINUE
is a robust cross-linguistic phenomenon (see e.g.
Di Eugenio 1998) and fuller forms are used for shifting Cb (see e.g.
Hedberg and Dueck 1999). The results reported in the previous section
indicate that the main difference between the Estonian overt form
and zero lies in the fact that the use of the latter is restricted to only
par ticular contexts, whereas the overt form can be used more freely in
all possible semantically appropriate contexts. The zero form is applied
to signal CONTINUE transitions, taking the role of a grammatical subject
in the nominative case. It also must realize the backward-looking center
Cb of an utterance, and is restricted to main clauses only. While all
these traits are also very common to the overt form, other grammatical
surroundings are likewise fully acceptable with the overt form ta: it can
be found in all possible cases in Estonian, it can take any grammatical
role in a sentence and it can occur in subordinate clauses.
Note that as the majority of data represents singular ta and zero,
conclusions are valid primarily for these forms. Table 2 indicates that
ird person pronoun forms and Centering eory in Estonian 131
plural forms behave slightly differently from singular forms, as in
plural, zero forms are used noticeably more than overt forms. However,
due to the scarce occurrence of plural forms, this cannot be claimed
with full certainty and more language data is needed to analyze plural
forms more reliably.
As the zero form can only be employed in the aforementioned condi-
tions, it is probably more appropriate to apply the term zero subject
instead. In previous research (Lindström 2005, Kivik 2010) this defini-
tion is indeed preferred. Moreover, the likelihood of reduced pronominal
forms to realize higher arguments, i.e. subject rather than object1 and
object1 rather than object2, appears to be a cross-linguistic tendency
(Siewierska 2003). However, the scarcity of zero objects in the given
study alone is not a sufficient argument for neglecting the possibility of
zero object in Estonian. Rather, more supplementary and diverse data
is needed to research the phenomenon.
The present study suggests that the initial hypothesis about the equal
level of salience of zero and overt forms is valid, as the CONTINUE transi-
tion, which is the most frequent transition for both forms, is generally
used when referring to highly salient entities. Nevertheless, con sidering
the fact that the zero person marker is so restricted in its use, it is
possible to see it as representing only the highest point of salience, as it
has been argued that entities mentioned in main clauses are more salient
than entities in subordinate clauses, and subjects are more salient than
objects (see e.g. Kaiser 2000, Yoshida 2011). Still, the overt form is also
completely acceptable in similar situations, so it is not possible to fully
distinguish the two forms.
One possible way to account for why the speaker might use zero
instead of the overt pronoun is the need to avoid the “repeated-name
penalty” (Gordon et al. 1993). Originally, the repeated name penalty
was reported to account for cases where a definite description was used
instead of a pronoun to refer back to Cb (Ibid.). However, in Estonian,
where there are two possible equally salient devices for referring back
to an already-mentioned entity, it might be probable that the overt form
is avoided in order to reduce processing time. It is also likely that the
use of a repeated pronoun instead of zero where it would be acceptable
indicates a new discourse segment boundary. (Ibid.)
This study has concentrated on spoken monologues in narrative
contexts. High occurrences of zero person markers in the data clearly
indicate that this form cannot be omitted from the description of the
Estonian pronominal system. However, as the possible contexts for the
132 Helen Hint
zero pronominal are undoubtedly restricted, further investigation is
needed to fully determine its place in the system.
Acknowledgements
I wish to thank Elsi Kaiser, Renate Pajusalu, and Virve Vihman
for their valuable help and comments regarding this work. I am also
grateful to the anonymous reviewers for the highly relevant feedback.
This research was supported by the European Social Fund’s Doctoral
Studies and Internationalisation Programme DoRa, which is carried
out by the Archimedes Foundation, and by Estonian Research Council
grant PUT701.
Address:
Helen Hint
Eesti ja ü ldkeeleteaduse instituut
Tart u Ü li kool
Jakobi 2, 51014 Tartu
E-mail: helen.hint@ut.ee
References
Ariel, Mira (1990) Accessing noun-phrase antecedents. London and New York: Rout-
ledge.
Ballantyne, Keira Gebbie (2004) “Givenness as a ranking criterion in centering theory:
e
vidence from Yapese”. Oceanic Linguistics 43, 1, 49–72.
Brown-Schmidt, Sarah, Donna K. Byron, and Michael K. Tanenhaus (2005) “Beyond
salience: interpretation of personal and demonstrative pronouns”. Journal of
Memory and Language 53, 292–313.
Byron, Donna K, Sarah Brown-Smith and Michael K. Tanenhaus (2008) “The overlap-
ping distribution of personal and demonstrative pronouns”. In Jeanette K. Gundel
and Nancy Hedberg, eds. Reference: interdisciplinary studies, 143–175. Oxford:
Oxford University Press.
Chafe, Wallace, ed. (1980) The pear stories: cognitive, cultural, and linguistic aspects
of narrative production. Norwood, N.J.: Ablex.
Cristea, Dan, Nancy Ide and Laurent Romary (1998) “Veins theory: a model of global
discourse cohesion and coherence”. Proceedings of ACL/COLING98, 281–285.
Di Eugenio, Barbara (1998) “Centering in Italian”. In Marilyn A. Walker, Aravind K.
Joshi, and Ellen F. Prince, eds. Centering theory in discourse, 115–137. Oxford:
Oxford University Press.
ird person pronoun forms and Centering eory in Estonian 133
EKG II = Erelt, Mati, Reet Kasik, Helle Metslang, Henno Rajandi, Kristiina Ross,
Henn Saari, Kaja Tael, Silvi Vare (1993). Eesti keele grammatika II. Süntaks. Lisa:
kiri. Tallinn: Eesti Teaduste Akadeemia Keele ja Kirjanduse Instituut.
Erelt, Mati and Helle Metslang (2006) “Estonian clause patterns – from Finno-Ugric
to Standard Average European”. Linguistica Uralica 42, 4, 254–266.
Givón, Talmy (1983) Topic continuity in discourse: a quantitative cross-language
study. Philadelphia and Amsterdam: John Benjamins.
Gordon, Peter C., Barbara J. Grosz, and Laura A. Gilliom (1993) “Pronouns, names,
and the centering of attention in discourse”. Cognitive Science 17, 311–347.
Grosz, Barbara J., Aravind K. Joshi, and Scott Weinstein (1995) “Centering: a frame-
work of modelling the local coherence of discourse”. Computational Linguistics
21, 2, 203–225.
Gundel, Jeanette (1998) “Centering theory and the givenness hierarchy: towards a
synthesis”. In. Marilyn A. Walker, Aravind K. Joshi, and Ellen F. Prince, eds.
Centering theory in discourse, 183–198. Oxford: Oxford University Press.
Gundel, Jeanette, Nancy Hedberg, and Ron Zacharski (1993) “Cognitive status and the
form of referring expressions in discourse”. Language 69, 2, 274–307.
Hadic Zabala, Loreley and Maite Taboada. (2006) “Centering theory in Spanish:
coding manual”. Manuscript. Simon Fraser University.
Hedberg, Nancy and Sandra Dueck (1999) “Cakchiquel reference and centering
theory”. Proceedings of the workshop on structure and constituency in the languages
of the Americas. Universit y of British Columbia working papers in linguistics, 5
9–74.
Hu, Jianhua and Haihua Pan (2001) “Processing local coherence of discourse in
Centering Theory”. Proceedings of the 15th Pacific Asia Conference on Language,
I
nformation and Computation, 139–150. Hong Kong: City University of Hong Kong.
Iida, Masayo (1998) “Discourse coherence and shifting centers in Japanese texts”. In
Marilyn A. Walker, Aravind K. Joshi, and Ellen F. Prince, eds. Centering theory
in discourse, 161–180. Oxford: Clarendon Press.
Kaiser, Elsi (2000) “Pronouns and demonstratives in Finnish: indicators of referent
salience”. In Paul Baker, Andrew Hardie, Tony McEnery, and Anna Siewierska,
eds. Proceedings of the 3rd discourse anaphora and reference resolution collo-
quium, 20–27. (Technical Papers, 12.) Lancaster: University Center for Computer
Corpus Research on Language.
Kaiser, Elsi (2009) “Investigating effects of structural and information-structural
factors on pronoun resolution”. In Malte Zimmermann and Caroline Féry, eds.
Information structure from different perspectives, 332–353. Oxford: Oxford
University Press.
Kaiser, Elsi and John C. Trueswell (2008) “Interpreting pronouns and demonstra-
tives in Finnish: evidence for a form-specific approach to reference resolution”.
Language and Cognitive Processes 23, 5, 709–748.
Kaiser, Elsi and Katrin Hiietam (2003) “A typological comparison of third person
pronouns in Finnish and Estonian”. Nordlyd (Tromsø) 31, 4, 654–667.
Kaiser, Elsi and Virve-Anneli Vihman (2010) “On the referential properties of Esto-
nian pronouns and demonstratives”. In Hans Götzsche, ed. In Memory, Mind and
Language, 193–205. Newcastle: Cambridge Scholars Publishing.
134 Helen Hint
Kibrik, Andrej A. (2011) Reference in discourse. New York: Oxford University Press.
Kivik, Piibi-Kai (2010) “Personal pronoun variation in language contact: Estonian in
the United States”. In M. Norde, C. Hasselblatt, and B. de Jonge, eds. Language
contact – new perspectives, 63–86. Philadelphia and Amsterdam: John Benjamins.
Lindström, Liina (2001) “Verb-initial clauses in narrative”. In Mati Erelt, ed. Estonian:
Typological Studies V, 13 8 –168. Ta rt u.
Lindström, Liina (2005) Finiitverbi asend lauses. Sõnajärg ja seda mõjutavad tegurid
suulises eesti keeles. (Dissertationes philologiae Estonicae Universitatis Tartuensis.)
Tartu: Tartu University Press.
Lindström, Liina, Kristel Uiboaed, and Virve-Anneli Vihman (2014) “Varieerumine
tarvis-/vaja- konstruktsioonides keelekontktide valguses”. Keel ja Kirjandus 8–9,
609–630.
Metslang, Helena (2013) “Coding and behaviour of Estonian subjects”. Journal of
Estonian and Finno-Ugric Linguistics 4, 2, 217–293.
Pajusalu, Renate (2005) “Anaphoric pronouns in spoken Estonian: crossing the para-
digms”. In Ritva Laury, ed. Minimal reference: the use of pronouns in Finnish and
Estonian discourse, 107–134. Helsinki: Suomalaisen Kirjallisuuden Seura.
Pajusalu, Renate (2006) “Death of a demonstrative: person and time: the case of Esto-
nian too”. Linguistica Uralica 42, 4, 241–253.
Pajusalu, Renate (2009) “Pronouns and reference in Estonian”. Sprachtypologie und
Universalienforschung: Estonian in typological perspective 62, 1, 122–139.
Passonneau, Rebecca J. (1998) “Interaction of discourse structure with explicitness of
discourse anaphoric noun phrases”. In Marilyn A. Walker, Aravind K. Joshi, and
Ellen F. Prince, eds. Centering Theory in Discourse, 327–358. Oxford: Clarendon
Press.
Poesio, Massimo, Rosemary Stevenson, Barbara di Eugenio and Janet Hitzeman
(2004) “Centering: a parametric theory and its instantiations”. Computational
Linguistics 30, 3, 309–363.
Scott, Kate (2013) “Pragmatically motivated null subjects in English: a relevance
theory perspective”. Journal of Pragmatics 53, 68–83.
Siewierska, Anna (2003) “Reduced pronominals and argument prominence”. In
Miriam Butt and Tracy Holloway, eds. Nominals: inside and out. Stanford: CSIL
Publications, 119–150.
Siewierska, Anna (2004) Person. Cambridge: Cambridge University Press.
Taboada, Maite (2008) “Reference, centers, and transitions in spoken Spanish”. In
Jeanette K. Gundel and Nancy Hedberg, eds. Reference: Interdisciplinary studies,
176–215. Oxford: Oxford University Press.
Taboada, Maite and Loreley Hadic Zabala (2008) “Deciding on units of analysis within
Centering Theory”. Corpus Linguistics and Linguistic Theory 4, 1, 63–108.
Vihman Virve-Anneli (2015) “Pick it up: a look at referential devices in Estonian
child-directed speech”. Journal of Estonian and Finno-Ugric Linguistics 6, 2,
63–83 (this volume).
Walker, Marilyn, Aravind Joshi and Ellen Prince (1998) “Centering in naturally occur-
ring discourse: an overview”. In Marilyn A. Walker, Aravind K. Joshi, and Ellen F.
Prince, eds. Centering theory in discourse, 1–28. Oxford: Clarendon Press.
ird person pronoun forms and Centering eory in Estonian 135
Walker, Marilyn, Masayo Iida and Sharon Cote (1994) “Japanese discourse and the
process of Centering”. Computational Linguistics 20, 2, 193–232.
Walker, Marylin A. (1998) “Centering, anaphora resolution, and discourse structure”.
In Marilyn A. Walker, Aravind K. Joshi, and Ellen F. Prince, eds. Centering theory
in discourse, 401–435. Oxford: Clarendon Press.
Yoshid a, Et su ko (2 011) Referring expressions in English and Japanese: patterns of use
in dialogue processing. Philadelphia and Amsterdam: John Benjamins.
Abbreviations
3 – third person
ABL – ablative
AD – adessive
ALL – allative
Cb – backward-looking
center
Cf – forward-looking center
COM – comitative
Cp – preferred center
dINF – da-infinitive
EL – elative
GEN – genitive
IN – inessive
mINF – ma-infinitive
NP – noun phrase
PART – partitive
PAS S – passive
PL – plural
PPTCP – past participle
PRO – pronoun
PRS – present
PST – past tense
SG – singular
SHORT – short pronoun
ZERO – zero person marker
U – utterance
Kokkuvõte. Helen Hint: Kolmanda isiku pronoomenivormid eesti keeles
tsenderdamisteooria valguses. Artikkel käsitleb eesti keele 3. isiku eksplit-
siitse pronoomeni ja pronoomeni nullvormi erinevusi suulistes narratiivides.
Mõlemad vormid väljendavad diskursuse kõige esilduvamaid entiteete, mis-
tõttu ei saa nende erinevust selgitada pelgalt esilduvuse mõistele toetudes.
Seega on artiklis kasutatud tsenderdamisteooriat (the Centering Theory), et
näha, kas nullvorm ja eksplitsiitne 3. isiku pronoomen on seotud erinevustega
diskursuse sidususes. Täpsemalt vaadeldakse, kas need kaks vormi erinevad
tsenderamise üleminekute poolest. Põhjalikuma ülevaate saamiseks nulli ja
ta erinevustest on võrreldud ka neile vormidele iseloomulikke grammatilisi
faktoreid: grammatiline roll, kääne, lausetüüp. Analüüsist selgub, et nullvormi
kasutuskontekstid on kitsamad, piirdudes JÄTKA MISE üleminekuga, nomina-
tiivse subjektiga pealausetes, samas kui eksplitsiitne vorm esineb mitme-
sugustes kontekstides ja ümbrustes.
Märksõnad: pronoomeni nullvorm, 3. isiku pronoomen, tsenderdamisteooria,
referents, suuline narratiiv, pirnilood
