Professional Documents
Culture Documents
Włodarczyk - Włodarczyk - Meta-Informative Centering in Utterances - Between Semantics and Pragmatics
Włodarczyk - Włodarczyk - Meta-Informative Centering in Utterances - Between Semantics and Pragmatics
Editors
Werner Abraham Elly van Gelderen
University of Vienna / Arizona State University
University of Munich
Editorial Board
Bernard Comrie Christian Lehmann
Max Planck Institute, Leipzig University of Erfurt
and University of California, Santa Barbara
Marianne Mithun
William Croft University of California, Santa Barbara
University of New Mexico
Heiko Narrog
Östen Dahl Tohuku University
University of Stockholm
Johanna L. Wood
Gerrit J. Dimmendaal University of Aarhus
University of Cologne
Debra Ziegeler
Ekkehard König University of Paris III
Free University of Berlin
Volume 143
Meta-informative Centering in Utterances
Between Semantics and Pragmatics
Edited by André Włodarczyk and Hélène Włodarczyk
Meta-informative Centering
in Utterances
Between Semantics and Pragmatics
Edited by
André Włodarczyk
Université Charles de Gaulle & Université Paris-Sorbonne
Hélène Włodarczyk
Université Paris-Sorbonne
Preface vii
Introduction ix
André Włodarczyk & Hélène Włodarczyk
The papers in this volume are a selection of lectures delivered at three consecutive
MIC Sorbonne workshops in Paris: Discourse coherence – text and theory (2008),
Context-bound Communication (2010) and New standards for language studies
(2012). Another selection relating specifically to Slavic languages has been pub-
lished in French in the Revue des Études Slaves (2009, vol. 80/1–2). The editors and
co-authors benefitted a great deal from discussions with the programme commit-
tee members and participants before, during and after the workshops.
Sachiko Ide (former President of the International Pragmatics Association,
Women’s University, Tokyo), Istvan Kecskes (President of the American Pragmatics
Association, New York State University, Albany), Masao Aizawa (Vice-President
of The Japanese Language National Institute, Tachikawa) and Yasunari Harada
(Director at the Institute for DECODE = Digital Enhancement of Cognitive
Development, Waseda University, Tokyo) have played an important role in the
dissemination and development of various aspects of the MIC theory.
Grants for a research exchange program with the Jagiellonian University
(Cracow) as well as invited lectures delivered by the editors at Waseda University
(Tokyo), The Institute of Slavic Studies of the Polish Academy of Sciences, Warsaw
University (Warsaw), The Japanese Language National Institute (Tachikawa), and
New York State University at Albany allowed us to discuss many MIC theoretical
issues.
We have been very privileged to have received help and advice from dis-
tinguished scholars. We wish to thank especially professors Werner Abraham
(Vienna University), Elisabeth Leiss (Ludwig-Maximilians-Universität München),
Radosław Katarzyniak (Wrocław University of Technology, Poland) and Robin K.
Burk (USMA, West Point, Dept. of Electrical Engineering & Computer S cience).
Last but not least, the editors also held discussions on various occasions with
their colleagues from CELTA Sorbonne research group: Jan Pekelder, Paul-Louis
Thomas, Stéphane Viellard and the late Jean Breuillard and Michel Viel, all of
them being professors at Paris-Sorbonne University, as well as Claude Delmas
(professor at Sorbonne Nouvelle University) and Claude Guimier (professor at
Caen University). They have all contributed a great deal in one form or another to
the realisation of our research plans.
A special and personal thank you goes also to our friend, Dr Rosemary
Masters, for her translations of a few chapters, careful rereading of, and insightful
comments on all the chapters of this book.
Meta-informative Centering in Utterances
Acknowledements
The publisher and editors wish to thank the Slavic Studies Institute (PAN – P
olish
Academy of Science) for permission to print the paper by André Włodarczyk
(“Roles and Anchors of Semantic Situations”, Études Cognitives – Studia kogni-
tywne 8, SOW, Warszawa 2008, p. 53–70) as well as the English translation of
the paper by the same author (“Les Cadres des situations sémantiques”, Études
Cognitives – Studia Kognitywne 5, Warszawa 2003, p. 35–51).
Introduction
At the beginning of the 21st century, the study of communication in human lan-
guages still remains under the influence of the theory of information structure
(Lambrecht 1994) which elaborates on the concepts of theme, rheme, and commu-
nicative perspective. The meta-informative centering (MIC) theory is an alternative
framework based on the concepts of centre of attention (CA), meta-information
and the meta-informative status of information. Importantly, the MIC theory is
built on associative semantics (AS) in which the concept of information is defined
as a compound relational structure, which to some extent makes it compatible
with the definition of information as often used in the field of computer (infor-
mation) science. In order to explain how it is possible for the hearer to interpret
a linguistic message (i.e. to build a mental representation of the situation spoken
about) a mapping must be established between the linguistic form and its content,
using some formalised representation of meaning. From the semantic point of
view, the logical motivation for the formal representation of situations consists in
reducing multi-argument relations solely to unary and binary ones, and combin-
ing them in compound configurations.
The origins of the MIC theory go back to Włodarczyk 1999 when we used the
concept of “centre d’intérêt” (Fr. centre of interest) for the first time. This French
term has since been replaced by that of “centre of attention” under the influence of
the American centering theory (Grosz & Sidner 1986), and it underwent a few revi-
sions and reassessments in the succeeding papers published in English. According
to the MIC theory, because of the linear order of speech, no judgment may be
uttered without selecting at least one centre of attention; thus centering is con-
sidered as a structuring operation not only at the discourse/text level but also at
André Włodarczyk & Hélène Włodarczyk
the level of utterance (cf. Chomsky’s binding theory) because of its necessarily
linear (sequential) nature. In the American centering theory, centres of attention
are defined at the discourse/text level: one constituent of an utterance is treated
as a forward or backward looking centre in order to maintain the cohesive flow of
information from one utterance to its successor. Forward and backward looking
centres make it possible to give an account of the relations which bind utterances
together into coherent discourses/texts. Although the centering theory precedes in
time the meta-informative centering theory, these theories were created separately
from each other. However, the results of the American theory can be straightfor-
wardly integrated into the MIC theory. Since the anaphoric and cataphoric moti-
vations of the meta-informative old and new status of the utterance during the
communication act match perfectly well the concepts of backward- and forward-
looking centres, the MIC theory is not an alternative to the American centering
theory. Rather, the latter is a complementary approach with respect to the MIC
which is more general.
Although the main characteristics of utterances consist in the fact that the
priority in their generation/understanding is given to attention-driven processes,
utterances are built choosing the most suitable ready-made verb valence schemata
which are used by the speaker in order to communicate the old or new status of
expressed content. Therefore, linguistic expressions which correspond to gram-
matical notions such as subject, object, topic and focus are defined as attention-
driven phrases (ADP); i.e. phrases governed by the centres of attention (1) in base
utterances where merely either an old or new status of information can be envis-
aged for both the subject and predicate in each utterance (but where no contrast
between the old and new status of information is expressed), and (2) in extended
utterances where the topic and focus are defined as dually opposed concepts;
(a) the topic being defined as governed by an old meta-informative status contrast-
ing with the new one of the comment and (b) the focus as governed by a new meta-
informative status contrasting with the old one of the background.
Although the importance of attention for cognitive functions of the brain has
recently also been emphasised by neuropsychologists (part 2 in this volume by
Franz J. Stachowiak), this faculty of the human brain has not yet been explicitly
used to account for linguistic objects in mainstream cognitive linguistics. In the
proposed theory, one kind of attention is a component of pragmatics and since the
semantic content of linguistic messages is not directly accessible, the interpreta-
tion of the truth value of utterances needs to be postponed and priority is given to
the resolution of the meta-informative status of communicated chunks of infor-
mation. In other words, in linguistic communication, truth valuation of the con-
tent of an utterance takes place only behind – so to speak – communication, i.e.
after the meta-informative old or new status has been assigned using such means
Introduction
The first part of this volume contains the presentation of essential issues of the
associative semantics (AS) and meta-informative centering (MIC) theories, the
second part is an attempt at a neuropsychological foundation of this theoretical
framework, and the third part deals with its application to several linguistic phe-
nomena in different languages. Due to its origins dating back to investigations
on typologically different languages and because it is attuned to contemporary
research in the field of computer science, the ASMIC theoretical framework exhib-
its both theoretical and experimental characteristics.
between morphology (tonic vs. atonic), prosody (stressed vs. unstressed) and the
pragmatic status (old vs. new) of different types of pronoun uses in discourse. As
pointed out by Jean-Christophe Pitavy concerning classical Greek: “the use of the
accented personal pronoun in conjunction with the verbal form allows the speaker to
construct an extended utterance.” Similarly, Perrine Vedrenne-Cloquet shows that
in Latin: “the pronouns ego and tu featured in discourse have a highly important role
in the construction of meta-informative coherence, since they can be used as topicali-
sations, but also as focalisations. Moreover, they can be used as additional markers:
their second position gives to the first position element a pragmatic emphasis which
it would not otherwise have.”
The three chapters about pronouns emphasise that reference to the old and
new meta-informative status makes it possible to distinguish between topicalised
and focalised uses of tonic pronouns. Moreover, the description of word order in
utterances containing topicalised and focalised pronouns provides evidence for
the fact that topic is the global CA of extended utterances. Olivier Azam con-
cluded from the observation of his Russian corpus that “topicalisation does indeed
appear to be incompatible with postpositioning”. Let us add (this is mentioned in
passing in Chapter 4) that the ellipsis of pronouns and zero-form pronouns (often
referred to indistinctly as pro-drop) should be thoroughly distinguished: the latter
can be attested to only in languages in which the person is relevantly marked by
verb endings (most Indo-European languages) whereas ellipsis is possible both
in Indo-European languages and in languages with no person category in verb
flexion (e.g. Japanese).
The volume is closed by a glossary of AS and MIC terms.
References
Grosz, Barbara J. & Sidner, Candace L. 1986. Attention, intentions, and the structure of dis-
course. Computational Linguistics 12(3): 175–204.
Lambrecht, Knud. 1994. Information Structure and Sentence Form. Topic, Focus, and the Mental
Representations of the Discourse Referents. Cambridge: CUP.
Sperber, Dan & Wilson, Deirdre. 1986. Relevance: Communication and Cognition. Oxford:
Blackwell.
Tomlin, Russell S. 1997. Mapping conceptual representations into linguistic representations:
The role of attention in grammar. In Language and Conceptualization, Jan Nuyts & Eric
Pederson (eds), 162–189. Cambridge: CUP.
Włodarczyk, André. 1999. La validation informative des énoncés et la quantification linguis-
tique. Études cognitives/Studia kognitywne 3: 121–133. Warszawa: SOW, PAN.
Włodarczyk, Hélène. 2012. L’ emploi des pronoms personnels en polonais (par contraste avec le
russe et le français). Études à la mémoire de J. Breuillard sous la dir. de S. Viellard, Revue des
Études Slaves, t.LXXXIII, fasc.2–3, 617–648. Paris: Institut d’Etudes Slaves.
part 1
Associative semantics
and meta-informative centering
Roles and anchors of semantic situations*
André Włodarczyk
Université Charles de Gaulle (Lille 3) / Université Paris-Sorbonne,
Centre de linguistique théorique et appliquée (CELTA)
* Research on this topic has been partly sponsored by a grant of the Hakuhodo Foundation
during the periods from October, 1st 2006 to January, 31st 2007 and from July, 15th 2007 to
September, 14th 2007 when the author was invited researcher at the National Institute of
Japanese Language (Tokyo).
André Włodarczyk
For this reason, infons can be seen as elements of semantic knowledge (declarative
know-what) as derived (in disorder) from linguistic interactions (communications)
of man. However, once they are arranged in a sequential order and completed by
sets of complementary infons they are enriched by ontological knowledge (proce-
dural know-how). It should be emphasized that this idea is similar to our earlier
proposal1 to organize declarative semantic attributes and their values into Feature
Structures according to their ontological grounding. It is therefore hypothesized
that ontology lies at the basis of any natural language semantics. What is seman-
tic proper for a given natural language is the selection and declarative organisa-
tion of ontological situations, their participants and anchors (locations in time and
space).
Language is a partial function from infons to expressions or vice versa (i.e.:
only a part of what the speaker means can be put into words, on the one hand,
and the hearer must complete every concept he picks up during the commu-
nication process, on the other hand). Thus, the theory of semantic situations
that is sketched hereafter seeks to clarify the fundamental composability ( partial
compositionality) of invertible functions which can be played by signs and their
combinations in order to enable their users to convey meanings about the enti-
ties of the World. Using the notion of composability we shall attempt to explain
the well-known and now largely accepted fact that linguistic expressions con-
vey only partially the information that the speaker has in mind. In this con-
nection, we shall use the couple of notions encoding/decoding in connection
with the communication of meaning both as it is conveyed in a linguistic mes-
sage (expression) uttered by a speaker and as it means something to the hearer
(a signified content).
As an immediate consequence of the above claim, it was necessary to revise
the current linguistic theory of predication. Indeed, it is no longer possible to
maintain that predication takes place when the content of a sentence is “com-
plete”. Instead, our theory postulates the neat separation of the truth conditions
(which are characteristic of information or situation with its roles or, more tra-
ditionally, argument structure relations) from the communicative new/given
statuses of expressions. Thus, it appears necessary to make the following distinc-
tion between two language levels: informative (relational in logical sense) level
and meta-informative (constitutive in syntactic sense) level. Indeed, in the MIC
(Meta-Informative Centering) theory, Predication is defined as uttering about
. Cf. the framework of the research project CASK (Computer-aided Acquisition of
Semantic Knowledge), Centre for Theoretical and Applied Linguistics (CELTA), Université
Paris Sorbonne.
Roles and anchors of semantic situations
a State of Affairs (SOA) with either new or given statuses. It becomes there-
fore clear that what is communication proper concerns the meta-informative
status of truth-conditional (relational) information (Włodarczyk A. & H 2006a,
2006b, 2008).
where:
The behaviour of more than one entities may include interaction with common
goals. Such interaction is known as cooperation and necessarily yields com-
munication. Humans are endowed with a particularly efficient communication
device: language. In the years 1980, a Polish mathematician J. Pogonowski (1979)
proposed a hierarchical reconstruction of structural linguistic theory in which
language is seen as a general system whose sorts are all its levels, objects – all its
(concrete and abstract) units and signature – all its syntagmatic, paradigmatic and
inter-level relations. Because natural languages are general systems themselves they
also exhibit properties of those systems. Consequently, they must be studied in
connection with the ontologies of their bearers (users). As we shall see later, in the
present theory the universe (Wld) contains objects (i.e.: animate and inanimate
André Włodarczyk
entities) which are related (Rel) through their behaviour (Bhv) to situation roles
within situation frames or environments (Env).
Currently Artificial Intelligence (AI) specialists agree that intelligent pro-
cessing of information requires utilisation of ontological knowledge; i.e.: such
knowledge which may be organized in structured and intelligent databases.
Indeed, ontologies are necessary for designing any information system which is
supposed to interact with humans. According to another tenet of the AI special-
ists, regardless of the overall architectures of information systems, such systems
often should contain more than one ontology. In order to cope with such com-
plex cases, at least two kinds of ontologies are postulated: (1) upper ontologies
on the highest levels of abstraction, and (2) domain ontologies on the lowest
(concrete) levels.
Animate Inanimate
Natural Artificial
Both natural and artificial entities may be either animate or inanimate. Thus,
the set A of animate entities intersects with the set X of natural entities and the
set Y of artificial entities respectively. Inanimate entities thus correspond to the
difference between natural and artificial entities X–A = {x | x ∈X and not x ∈A} as
regards inanimate natural entities and to the difference between artificial and ani-
mate entities Y–A = {x | x ∈Y and not x ∈A} as concerns inanimate artificial enti-
ties. Inanimate entities are those which belong either to X or Y but not to A. Such
a definition breaks away from the traditional logical classification but may be dealt
with in multiple inheritance systems or in any system able to process m embership
approximation.
André Włodarczyk
Purpose
Verbal
Desire
Visual
Belief
AGENTS
Human + + + + + + + +
Non human + + ± – – – – –
It was probably the theory of Case Grammar2 (Fillmore Ch. 1968, 1971) which
was most influential and very frequently referred to in linguistics. In particular,
. “Case grammar has not become a mainstream part of generative grammar, perhaps because
the phenomena it deals with do not yield to the kind of neat formal analysis that the paradigm
demands. There has been a tendency to push the issues off by arguing that they are ‘semantic’
Roles and anchors of semantic situations
rather than to deal with them in the syntax. However, case grammar has been widely discussed
and has influenced the design of many computer systems for natural language.” (Winograd T.
1983, p. 311)
. Note that in the proposal recently put forward by Pustejovsky J. [1995] the term “struc-
ture” is used instead of “component”.
André Włodarczyk
This h ypothesis is essential for the present theory of semantic situations which, as
it takes into account the partiality of meaning of linguistic expressions, entails the
need to use, for their formal representation, stratified structures (namely hyper-
graph structures) instead of simple tree structures. Thus, the semantic partiality
hypothesis may be seen as an alternate theory to that of surface/deep structures
because it makes it possible to elucidate incomplete contents, on the one hand,
and to take into account their basically heterogeneous character (conveyed by
linear language expressions), on the other hand.
The above compound infon can be developed as two elementary infons provided
that both of them hold in the same mental situation S1.
Now, if we want to represent more information which is associated with the main
information contained in the above utterance, we may wish to add some other
infons which hold in S1.
We call Roles the binary relations which correspond to elementary infons. Roles
are component relations within the frames of more general situations. Roles either
equal or are included in situations (Roles ⊆ Situations), i.e.: roles are either rela-
tions or proper subsets of relations. Every semantic situation is therefore com-
posed of infons which are partially determined by attributes from both (1) upper
ontologies and (2) domain ontologies. Situation participants of upper ontologies
are seen logically as semantic arguments and were represented in the theory of
situation semantics (Barwise, J. and Perry J. – 1983) as functions from roles to
entities named “anchors”. Note that the term anchor will be used here in quite a
different way.
In this semantic theory roles are defined as pairs of participation functions as
follows:
–– a type participation function from role types to the high level ontological
entities (TP: Role type –> Entity type)
–– an instance participation function from the role instances to the individual
entities (IP: Role instance –> Entity instance)
The instant roles inherit the properties from their types (abstractions: hypernyms,
super-classes). But, due to the partiality principle mentioned above, when generat-
ing or compiling a particular (concrete) role, language users either may be uncon-
scious of many inheritable abstractions or abstractions of entities may be taken
into consideration whenever their instances are indeterminate.
André Włodarczyk
The Role Component of semantic situations may contain from one to three
elements from the set of three kinds of roles whose generic names are: active,
passive and median. This theoretical shift6 has important consequences regarding
the very nature of roles which are defined here qualitatively rather than substan-
tially or relationally (as relations between terms). We must not forget that logical
arguments are first of all terms. However, such terms are not order-free. If we want
to make the sequential order of terms free, it is necessary to split each of them
into two parts: the role and its filler (often referred to as participant). We will
repeat this shift as regards our second component of semantic situations, which is
Anchor Component.
There are two kinds of participant types: Agents (animate entities) and F
igures
(inanimate entities).
Roles enumerated in the Table 2 represent the most usual ones. Here are some
agents or figures: (1) in active roles (Initiator, Causer, Enabler, Benefactor, Execu-
tor, Stimulant, Source, Instigator etc.), (2) in passive roles (Terminator, Causee,
Enabled, Beneficient, Executed, Experiencer, Goal, etc.) and (3) in median roles,
(Mediator, Instrument, Benefice, Motor, Means, Matter etc.). While active and
passive roles are considered here as primitive roles, median roles are definable by
introducing two embedded associated situations (a-situations) in which (a) the
participants of median roles fulfil passive roles in the first of the associated situa-
tions and (b) they fulfil active roles in the second associated situation. I distinguish
therefore role associations (see the above discussion about infons) as explicitly
expressed by linguistic utterances from situation associations which are implicitly
“responsible” for the creation of median roles (but note, however, that associations
of semantic situations are also often marked in languages; i.e.: have explicit mor-
phological markers, for example prefixing or suffixing verbs or using compound
verbs). Let us also add that there may be more than one median roles in the same
. Note however the rather exceptional idea of Tesnière L. (1959) who theorized about roles
using the abstract name “actant” (acting entity) and enumerated three of them: first actant,
second actant and third actant.
Roles and anchors of semantic situations
. Note, however, that due to the general principle of partiality of linguistic expressions both
typified participants and their instances may happen to be indeterminate.
André Włodarczyk
While HEAD contains shallow level valence FOOT corresponds to the standard
level valence. BODY may contain other situations which are thus associated with
the main (linguistically marked) situation. The above formalism enables the analy-
sis of the semantic situation (meaning) of the utterance #1 as follows:
. Note that in this approach the analogy goes from Roles to Anchors rather than backwards.
. CDL (Concept Description Language) is designed by Yokoi T. et al. (2005–2007).
Roles and anchors of semantic situations
Note that in the analysis the median role (fulfilled by ‘knife’) has been split into
two primitive roles (passive and q-active) each belonging to another associated
situation (use x y and causeDie x y).
Let us analyse now the difference between #3a and #3b beneath.
[SIT3: ‘make_happy’
HEAD make_happy x y
BODY
{Role1 x=Q-agent generic=active type=benefactor}
{Role2 y=agent generic=passive type=beneficient }
{Role1 < make_happy > Role2}
FOOT
{book < make_happy > Mary}]
FOOT
{Peter < SIT2:give1 > book*book < SIT2:reward > y = Mary}]
or
{Peter < SIT3:give1 > book*book < SIT3:make_happy > y = Mary}]
The noun “book” has been interpreted as fulfilling the median roles defined by two
pairs of associative situations “give1” and “reward” (meaning 1) as well as “give1”
and “make_happy”.
Here are a few other examples of semantic situations containing median roles
(see Table 4). It is important to note that the difference between “to give” and “to
receive” is meta-informative only, but the difference between “to sell” and “to buy”
concerns in addition the informative contents. This becomes obvious if we con-
sider that there might be another median role (fig-value object=money) attached
to both the pair “to sell” and “to buy”, but not to “to give” and “to receive”. It is
also for the same reason that the standard level meaning of both “to sell” and “to
buy” includes associations of their reciprocal situations (i.e.: “to sell” requires the
embedded a-situation [SIT1: ‘buy”], on the one hand, and “to buy” requires the
embedded a-situation [SIT1: “sell”], on the other hand).
consider such interpretation as standard level meaning, leaving room for the
procedural extensions giving rise to deep semantics.
We now present the analysis of an utterance containing the median role which
results from the association of more deeply embedded a-situations.
[SIT0: ‘buy’
HEAD buy x y z
BODY
{Peter participant=agent: }
{car participant=figure: }
{daughter participant=agent: }
[SIT1: ‘acquire’
HEAD acquire x y
BODY
{acquire x y
{Role1 x=buyer generic=active type=initiator }
{Role2 y=bought generic=passive type=terminator }
{Role1 < acquire > Role2}}
FOOT
{Peter < acquire > car}]
[SIT2: ‘offer’
HEAD offer x y z:
BODY
{Role1 x=donor generic=active type=benefactor}
{Role2 y=receiver generic=passive type=beneficient }
{Role1 < offer > Role2}
[SIT2.1: ‘give’
HEAD give x y
BODY
{Role1 x=giver generic=active type=owner }
{Role2 y=gift generic=Q-active type=ownedObject }
{Role1 < give > Role2}
FOOT
{Peter < give > car }]
[SIT2.2: ‘reward’
HEAD {reward x y }
BODY
{Role1 x=rewarder generic=Q-active type=initiator }
{Role2 y=rewarded generic=passsive type=terminator }
Roles and anchors of semantic situations
while the figure ‘car’ fulfils the median role because in the [SIT1 ‘buy:acquire’]
it fulfils a passive role as ‘bought’ in {Peter < buy:acquire > car} and it fulfils a
Q-active role in {car < {buy:offer:reward } > daughter}.
At the basis of ontological theories there are classifications. Theories and Logic
provide formal languages which can determine whether the given ontological
classifications are satisfiable with respect to the knowledge as conveyed by well-
formed expressions of a formal language. It is clear that such kind of classification
with a language is in fact a meta-classification (cf. Kent R. 1998).
However, when the formal language expressions are seen as type descriptors
and the uses of natural (human) language expressions are seen as instances, the
corresponding classifications are partial with respect to the ontological meta-
classifications from which the formal descriptions are derived. In such cases, the
ontological type description formal language is a semantic meta-language describ-
ing the given natural language.
References
Barwise, Jon & Perry, John. 1983. Situations and Attitudes. Cambridge MA: The MIT Press,
Bradford Books.
Devlin, Keith. 1991. Logic and Information. Cambridge: CUP.
Dick, Judith P. 1991. A Conceptual Case-Relation Representation of Text for Information
Retrieval. Ph.D. dissertation, University of Toronto.
André Włodarczyk
Fillmore, Charles J. 1968. The case for Case. In Universals in Linguistic Theory, Emmon Bach &
Robert T. Harms (eds), 1–90. New York NY: Holt, Rinehart & Winston.
Fillmore, Charles J. 1971. Types of lexical information. In Semantics, Danny Steinberg & Leon
A. Jokobovitz (eds), 370–392, Cambridge: CUP.
Ingarden, Roman. 1981. Spór o istnienie świata, Tom III : O strukturze przyczynowej realnego
świata (The Controversy over the Existence of the World, Vol. III: On the Causal Structure
of the Real World), translated from German into Polish by Danuta Gierulanka. Warszawa:
PWN.
Kent, Robert. 1998. The model theory of ontologic. 〈http://www.ontologos.org/Papers/Papers.
html〉
Pogonowski, Jerzy. 1979. Formal methods in linguistics. Buffalo Papers in Linguistics 1(3): 31–83.
Pustejovsky, James. 1995. Generative Lexicon. Cambridge MA: The MIT Press, Bradford Books.
Somers, Harold. 1987. Valency and Case in Computational Linguistics [Edinburgh Information
Technology Series 3]. Edinburgh: EUP.
Sowa, John F. 1999. Knowledge Representation: Logical, Philosophical, and Computational Foun-
dations. Pacific Grove CA: Brooks Cole.
Tesnière, Lucien. 1959. Éléments de syntaxe structurale, Paris: Klincksieck.
Winograd, Terry. 1983. Language as a Cognitive Process, Vol. 1: Syntax. Reading MA:
Addison-Wesley.
Włodarczyk, André & Włodarczyk, Hélène. 2006a. Subject in the meta-informative centering
theory. Études cognitives / Studia kognitywne 7: 39–64. Warszawa: SOW, PAN.
Włodarczyk, André & Włodarczyk, Hélène. 2006b. Focus in the meta-informative centering
theory. In La Focalisation dans les langues, André Włodarczyk & Hélène Włodarczyk (eds),
27–46. Paris: L’Harmattan, Coll. Sémantiques.
Włodarczyk, André & Włodarczyk, Hélène. 2008. The Pragmatic Validation of Ulterances. In
Études cognitives/Studia kognitywne 8: 117–128. Warszawa: SOW, PAN.
Yokoi, Toshio et al. 2005–2007. Concept Description Language – CDL. Core Specifications, Ver-
sion 1.0. Tokyo: Institute of Semantic Computing (ISeC).
Frames of semantic situations
André Włodarczyk
Université Charles de Gaulle (Lille 3) / Université Paris-Sorbonne,
Centre de linguistique théorique et appliquée (CELTA)
. François J. and Denhiere G. (1997, p. 125) list – from 1967 to 1994 – 34 different pro-
posals for the conceptual classification of verbal predicates; amongst which are the works of
W. Chafe, S. Dick, Ch. Fillmore, J. François, R. Jackendoff, R. Martin, Z. Vendler, A.R. Verkuyl
etc. In the domain of Slavonic linguistics alone one may cite amongst many others, Bulygina
T.V. (1982), Laskowski R. (1998), Piernikarski C. (1969), etc.
André Włodarczyk
participants, but we will not be dealing with these here. The aim of this study
is to define situations from the point of view of their “internal construction”
(i.e. without taking into account modality, tense, aspect etc.) It is indeed desirable
(1) to put forward a coherent system of classifying situations based on a small
group of well defined primitives and (2) to show the way in which different verbal
expressions, used in context, may inherit characteristics from partially organised
(hierarchised) situations, taking into account the primitives that have been
identified.
This research on semantic situations follows on from our work based on the
hypothesis of parallel processes which are supposed to characterise linguistic and
cognitive operations. Furthermore, as a working hypothesis, we postulate that
semantic situations (i.e. situations expressed by linguistic means) have a discrete
character (as indeed does everything that we perceive) in spite of the fact that
epistemic agents give the impression that they capture the continuum. We also
propose (as is frequently put forward) that it must be possible to classify all situa-
tions by using as small a group of primitives as possible, and that these, as such, do
not directly represent any situation.
Definition 1: A “situation” is any state or course of events contained within limits
called “frames”.
1.1 Space
Definition 2: There are two types of situation: static (the state of things) and
dynamic (the course of events or actions).
All situations have this in common, that they are located in space (their prop-
erties are plotted in three-dimensional space), and that they are distinctive in the
way in which they are concerned with time. State is distinguished from the three
other kinds of situation known generically as action situations (c.f. hereafter) by
the criterion of dynamism, that is, by the fact that the passing of time brings about
change therein.
Definition 3: The constituent elements of a static situation are its “place” (in the
universe of entities) and its “periphery” (places of untimed transition)2
. Transition is here used as it is understood in the network of bipartite graphs, known as
“Petri nets”, see later, in the section on representation and formalisation.
Frames of semantic situations
1.2 Time
All situations (static and dynamic) are surrounded in time by neighbouring situa-
tions: preceding or subsequent. As we will see later, neighbouring situations act as
transitions for the situations in question.
Proposition: Dynamic semantic situations are structured.
Definition 4: The constituent elements of a dynamic situation are its “moments”
(timed transitions) and its “stages” (the interval from one transition to the next).
From a conceptual point of view, all situations should only be characteristic.
A situation can be said to be characteristic when all the moments and all the stages
of its “life cycle” are taken into account, although semanticians often emphasise that
languages rarely express characteristic situations without selecting within them any
moment or stage (for example, in French, the verb “bouger” would represent the
characteristic situation expressing the abstract concept of “movement”), while the
language vocabularies are made up of more words which contain aspectual semes3
(for example, the French verb “arriver” would express the final phrase of the charac-
teristic situation “bouger”, or more precisely, of one of its hyponyms “se déplacer”).
Thus certain utterances express what we will call whole situations, whereas others
present situations that are analysed into moments and/or stages.
Not only are situations analysed – as we have said – into “internal” moments
and stages, but they are also related to neighbouring situations that are often taken
into account in linguistic expression. Indeed, what happens in the world is never
perceived “in a vacuum” but in relation to what comes before and follows on after-
wards. Many linguists consider that situations can be seen as subject to “life cycles”
with the result that within them we can see the following three – inner – stages:
“the starting stage” (begin), “the middle stage” (run) and the “final stage” (end). It
is to these three fundamental stages in particular that we add two segments, one
on either side, representing the preceding neighbouring situation, (before) and the
following neighbouring situation, (after). These two preceding and following situ-
ations must be considered as outer stages of the situation in question; they are
therefore its immediate preparation and consequence.4
. In fact the categories of tense and aspect in a given language involve semantic situations;
they allow the user, by diverse inflectional, derivational or syntactical means, to choose
different ways of looking at semantic situations.
. According to the triple of sequential processes (Hoare C.A.R., 1969), operations are
preceded by conditions (known as “preconditions”) and followed by consequences (known
as “postconditions”).
André Włodarczyk
MOMENTS
initial start enter exit finish terminal
< before > < begin > < run > < end > < after >
STAGES
Our definitions of semantic situations are cognitive in nature (and not onto-
logical or phenomenological). We distinguish between the four different situa-
tions: (1) states, (2) events, (3) ordinary processes and (4) refined processes. It should
be remembered that despite certain similarities, these terms are either used with
different meanings in other theories (states, events, processes) or are our own terms
(ordinary processes/refined processes).
In dynamic situations, entities undergo changes as a result of the fourth
dimension (time). All dynamic situations develop or progress (a) by moments:
start, enter, exit and finish and (b) by stages (ordinary and refined processes have
three distinct stages: begin, run and end.) The middle stage (run) does not appear
in events where the beginning is immediately followed by the end, because events
are perceived without internal duration.
1.3 Progression
Ordinary and refined processes are characterised not only by time but also by
progression,5 which is defined as the movement from one stage to another.
Although time is here a necessary condition, progression must be understood as
1.4 Granularity
The fourth concept (but the sixth dimension) which we feel we must bring into the
situations model concerns the cognitive granularity8 of stages. Here we must ascer-
tain (clarify) the way in which progressions evolve. Granularity can be defined as a
conjunction of selectors of semantic features (such as intensity, speed, size, weight,
strength etc.). For example, for situations where the granules involve the speed of
development, the semantic features can be either acceleration (ti = vj & ti+1 = vj+1,
i.e. in ti time, speed is vj and in time ti+1 speed is increased and is equal to vj+1) or
deceleration (ti = vj & ti+1 = vj–1, i.e. in ti time, speed is vj and in time ti+1 speed is
decreased and is equal to vj–1). Most situations with granulated stages are dual
pairs of selectors of semantic features concerning (a) the orientation of physical
movement (left/right,9 up/down,10 forwards/backwards), and (b) the orientation
of psychological movement (movement of the spirit), polarity (yes/no,11 more/
less) and intensity (strong/weak)12 etc. Although the concept of granularity is close
to that of discreteness, we will take care not to confuse them.
The regularity of the defining differences between state and actions (event, ordi-
nary and refined process) can be represented in a table showing their position in
a hierarchy (Table 1): the more complex a situation, the more defining features it
has. The hierarchical relationships thus elucidated can be compared to the order
of classic logic where propositional logic is zero order (order 0), predicate logic is
first order (order 1) etc. In the same way, states (static situations) can be seen as
being order 0, events, and ordinary and refined processes are successive orders.
Thus different situations are included within one another and organise themselves
in the following way:
Situation = State ⊆ Event ⊆ Ordinary Process ⊆ Refined Process
This formula reads as: “Situations are ordered: state ⊆ event ⊆ ordinary process ⊆
refined process, from the smallest to the largest”. Interlocked situations make up a
hierarchical order where each kind of situation inherits the properties of the one
that precedes it. The inclusions that we propose in order to define the hierarchical
order of types of situations are only concerned with their properties (dimensions).
Space (3D) + + + +
Time – + + +
Progression – – + +
Granularity – – – +
This table brings to mind certain existing classifications of verbs into semantic
groups. As both Vendler Z. (1967) and his successors realised, semantic situations
are not notions expressed by verbs alone. The problem is a little more complex
since we are here dealing with semantic types which sometimes have no direct
equivalent in the form of a linguistic expression, but which can be found amongst
the semes contained in (a) verbs themselves, (b) verbal periphrases, (c) presuppo-
sitions, (d) communicative acts, (e) knowledge etc. Verbs are not therefore neces-
sarily being dealt with for their properties as lexemes, but as verbs employed in
utterances for the purpose of designating situations. Depending on the context13
in which it is used, the same verb can express different kinds of semantic situations.
. By context, we mean not just the aspect-temporal inflections of a verb, but also the
participants and the anchors with which it is used.
Frames of semantic situations
SITUATION FRAME
– +
without time with time
– +
without progression with progression
– +
without granularity with granularity
It goes without saying that the inclusion relationships which we have shown
between the different kinds of semantic situations, allow situations to be formed
in two ways: successively or in parallel. When one situation follows another, their
make-up follows the law of transitivity in inclusion relationships. Winston M. E.,
Chaffin R. and Hermann D.J. (1987) identify the inclusion relationship of classes
which precedes the inclusion relationship of composition in the hierarchy, and this
in turn precedes the spatial (topographical) inclusion relationship. These authors
observed that mixed inclusion relationships are valid if, and only if, the conclusion
expresses the weakest relationship in the sense of the hierarchical order existing
between these inclusion relationships.
This formal property of inclusion relationships has an interesting application
in the description of semantic situations expressed in languages. In the case that
we are dealing with here, concrete verbs, even when they express situations of an
identical type (e.g. processes), must obey the transitivity of inclusion if their lexi-
cal meanings refer to inclusion of different sorts. The weakest inclusion is: state ⊆
refined process. That is why states often follow processes or, to put it another way,
processes are often in a composition relationship with states (c.f. result aspect) but
in the case of the inclusion state ⊆ event, it is obvious that it is events which should
be composed with states.
André Włodarczyk
Names are the names of the verbs (or adjectives) and verbal (or adjectival) phrases.
There are four types: states and events, ordinary and refined processes. So as to
comply with the theorising tradition in linguistics, however, we also distinguish
between static situations (states) and dynamic situations (actions: events and
processes).
Situation frames have dimensions which determine their type, to which the
parameters “moment” and “stage” are added.
Situation Frame = { Dimension, Moment, Stage }
initial | start | enter | exit | finish | terminal before | begin | run | end | after
States only have the “space” dimension, they are states of things but they can
be transformed into events or even processes. Events are characterised by the fact
that they have no “middle” element, and as a result, no “run” stage. Processes show
all moments and all stages, but Ordinary processes are distinguished from Refined
processes by the absence of granularity in their progression. The order of the differ-
ent types of situation frames can be seen in Table 2.
It is important to note that in spite of the presence of the parameters nec-
essary for the representation of aspect (understood as including “modes of
action”), the structure of the situation does not in itself show any characteristics
of aspect. It corresponds to the semantic nest which is the idealised meaning of
Frames of semantic situations
SITUATION TYPES
GRANULARITY
PROGRESSION
TERMINAL
BEFORE
INITIAL
FINISH
ENTER
AFTER
BEGIN
START
SPACE
TIME
EXIT
RUN
END
STATE + – – – – – – – – – – – – – –
EVENT + + – – – + – – + – + – – – +
ORDINARY PROCESS + + + – + + + + + + + + + + +
REFINED PROCESS + + + + + + + + + + + + + + +
The dimensions which we have noted (space, time, progression and granularity)
have allowed us to identify four kinds of situations and we will subsequently give
some language examples of these. In particular, we show that certain aspect pro-
cedures (prefixes, conjugated forms, syntactical constructions) allow us to change
the association of a verb from one semantic type to another, for example a verb can
be changed in this way from a verb of process into a verb of event. In this article,
we will simply give some examples without going into the details of the prob-
lem of links between aspect parameters and semantic types. We must nonetheless
emphasise the fact that the situation structures that we have defined, in particular
moments and stages, whilst remaining theoretical, must, independently of aspec-
tualisation, be considered inherent to the situations.
3.1 States
STATE is the simplest kind of situation, and is defined as a situation in space which
lasts for a greater or lesser period of time but which is not affected by the passing
André Włodarczyk
of time since it remains identical throughout. We therefore consider that the time
parameter is not applicable to states. In a state, it is not possible to distinguish
between successive stages. Although states can have a duration, we do not make
any distinction between the state-moment and the state-duration.
Mary is feeling bad.
Marie se sent mal.
Maria czuję sie niedobrze.
At that time all inhabitants of the district were in their beds.
A cette heure-là tous les habitants du quartier étaient couchés.
O tej porze wszyscy mieszkańcy tej dzielnicy leżeli w łożku.
3.2 Events
An EVENT, which is a momentary dynamic situation (that is, without progres-
sion), has a “start” and a “finish” that are so close that it is not possible to distin-
guish a run stage between them. Indeed, even if, in reality, nothing can happen
without occupying an interval of time, however minimal that may be, in linguistic
representation man conceives of events as instantaneous and does not take into
account the real time interval which they occupy. On the other hand, events can
be used to express the start or finish moment of a state, an ordinary process or a
refined process.
Depending upon the situation which precedes or follows an event, it is pos-
sible to identify three kinds of event. A type E1 event occurs and vanishes with no
relationship to a preceding or subsequent event, and can be thought of as a very
short moment of time within a state, no sooner starting than finishing.
Suddenly he shouted.
Nagle wrzasnął.
Soudain il poussa un hurlement.
The semantic features of the granules in certain refined processes can be quantita-
tive in nature. Because of the variation in the size of these features, it is possible
to subdivide these processes into three phases14 which make up their “life cycle”
(Figure 4).
MOMENTS
enter exit
< run > <e
initial start > nd finish terminal
gin >
< be
< before > STAGES < after >
Figure 4. Diagram of the life cycle of one specific case of a refined process
More generally, the “begin” stage is thus the phase in which the specific size
is increasing (represented by a rising segment), the “run” – the phase where the
. A diagram of this kind is used in numerous works on aspect, e.g. Sémon 1986; Barentsen
1995.
André Włodarczyk
The primary source of inspiration for defining the internal stages of a s emantic
situation was for us algorithmic in nature. In the course of working out the
order of situations, however, we thought it useful to consult theories of modal
logic: (1) of action (von Wright G.H. 1967), (2) of the structure of events
(Winskel G. 1982 & 1983; Kowalski R.A. & Sergot M.J. 1986) and (3) of situa-
tions (McCarthy J. & Hayes P.J. 1969). It was above all the relationship of these
theories to computer science which seemed to us to be able to contribute to
the development of theories of meaning in linguistics. This continuing dialogue
between logic and computer science is, moreover, equally profitable to logic
itself. As far as the contribution of logic and computer science to linguistics is
concerned, it is barely necessary to point out that generative linguistics owes its
appearance to the development of computer science techniques (in particular,
to the intensive research carried out on formal languages, including “Chomsky’s
hierarchy of languages”), and that since then, a great deal of research on natural
language processing has been inspired either directly by logic or by theories in
the area of computer science on the treatment of symbols.
In order to construct our concept of situation structure, however, we relied
above all on sequence structure as defined by Hoare C.A.R. (1969). It was no doubt
this same concept which inspired Moens M. (1987), and then Moens and Steedman
(1988), to propose the “tripartite structure of events” called “nucleus”. This nucleus
was then adopted by theoreticians of discourse representation: Kamp and Reyle
(1993), and other specialists of modes of action (Aktionsarten): B lackburn et al.
(1993) and Gagnon and Lapalme (1995).
As a result of the intensive use in computer science of both (1) Petri nets and
(2) the Event Structures15 of Winskel G. (1982 and 1983), and because these two
. The semantics of event structures was formulated by Winskel G. for the CCS – Calculus
of Communicating Systems of Milner R.A. (1980).
Frames of semantic situations
representations are useful for describing parallel structures,16 we can compare the
results of our research on situation with the research in this area prompted by
Petri nets and event structures. Petri nets were first applied to linguistic analysis
by Mazurkiewicz A. (1986, 2000). As we will see, however, our use of the Petri
model differs from that of Mazurkiewicz A. (1986) in that we do not deal with
inter-situational relationships (i.e. involving multiple situations), and from that of
Mazurkiewicz A. (2000) in that we identify two levels of analysis: micro-structural
(lower) and macro-structural (higher). In order to present macro-structural con-
cepts we will also use the formalism of “event structures”. Nevertheless, the latter
will be augmented in the following ways: (a) by adding the time component17 and
(b) by completing the structure of “events” (“semantic situations” in our terminol-
ogy) with the “terminal moment” and thus, with the “after” stage.
never follow one another. This formal constraint imposes the order of succession
(precedence) at the lower level which, taken at the higher level, led Mazurkiewicz
A. (1986, p. 10) to consider that there might be a natural order characterising the
relationship between places (interpreted as “states”) and transitions (interpreted
as “events”).
In our opinion, it is above all the assimilation of the primary concepts of Petri
nets with semantic situations (expressed by languages) which leads Mazurkiewicz
to propose a relationship of succession (precedence) at the higher level (though
removed at the lower level) and to conclude that this generates order (this rela-
tionship would be above all transitive and asymmetric) at the higher level also. To
realise that this is not in fact the case, one only has to remember that the relation-
ships (of succession or even of concurrence) between semantic situations (in daily
life) in fact depend upon expression or are simply unforeseeable.18
A “transition” in Petri nets is not in itself a dynamic situation. In order to
make transitions dynamic, computer scientists proposed extended Petri nets
(Murata 1989, 1991). In addition, and by this time through the study of aspect in
languages, researchers at Berkeley19 proposed introducing more enrichments into
the initial formalism: “our extensions to the basic Petri Net formalism include typed
arcs, hierarchical control, durative transitions, parametrization, typed (individual)
tokens and stochasticity”.
Independently of this, Mazurkiewicz A. (2000) also proposed the extension
of Petri nets by introducing the concept of “progression” (represented by a tri-
angle) which allowed him to time transitions. Hence the need to identify the fol-
lowing four kinds of progression: (1) open progression [state1 → progression →
state2], (2) closed progression [event1 → progression → event2], (3) progression
closed from behind [event → progression → state], (4) progression closed in front
[state → progression → event].20 Mazurkiewicz’s proposal has, however, profound
consequences for the very structure of Petri nets, in particular the need to intro-
duce a third graph, that of triangles. Thus extended, Petri nets would no longer
be systems of transitions and places (bipartite graphs),21 but of systems of transi-
tions, places and progressions (tripartite graphs). Unless the progression defined
by Mazurkiewicz A. (2000) is a “macro-stage” (cf. Grafcet) and consequently has
. This observation has already been made by Bojar B. (1986, p. 85)
. Chang N. (1997), Narayanan S. (1997), Chang N., Gildea D. and Narayanan S. (1998)
. N.B. [state → progression → event] ≡ [event ← progression ← state].
. Following the explanation given by Sowa J. (2000), Petri Nets are a fusion of Flow Chart
and State Machine structures, each of which can be represented by a graph. In particular, it is
the way in which this fusion was carried out which makes us think of bipartite graphs.
Frames of semantic situations
STATE t1 t2
EVENT m1 m2
ORDINARY PROCESS m1 m2
REFINED PROCESS m1 m2
. “… any timed Petri net can be translated by a t-timed net. Formally, a duration of sensiti-
zation di = [di min; di max] is associated with each ti transition in the net”.“In the case of t-timed
nets, the interval associated with the transitions characterises their duration. It is the time
during which the place tokens on entry are no longer present (they are reserved) but during
which the tokens produced are not yet visible in the place of exit” (Pradin-Chézalviel B. &
Valette R. – 2000).
André Włodarczyk
The model in Figure 6 compared to that in Figure 1 simply adds parallels, with-
out changing the time model itself. The internal time of a situation can be simply
modelled by linear time. But this is not the case with time in inter-situational rela-
tionships. Human experience shows that to take account of these relationships we
would have to use the branching model24 (but not, for example, the circular model).
. In the context of organic programming (GAEA system), complex situations are also
defined as being made up of primitive situations (Nakashima H. – 1998). cf. also Nakashima H.,
Ohsawa I. and Kinoshita Y. (1991) and Harada Y. and Nakashima H. (1995).
. This important observation comes from Kayser D. (personal communication).
. We call “extended event structures” the event structures of Winksel G. (1982 & 1983)
integrated into the event and configuration structures of Winkowski J. (1992).
Frames of semantic situations
structures, any action26 which takes place in time (timed action) is a quadruple
made up of:
The three defined times are ordered: t0 > t1 > t2. Taking inspiration from Hoare’s
logic of sequential processes, according to which all sequences are made up of a
prefix {ф} (expressing preconditions), an operation α and a suffix {ψ} (express-
ing postconditions or effect): {ф} α {ψ}, we add to the event structures a 4th time
(effect or resulting time: t3). Only those operations α whose preconditions {ф} are
true can give results prescribed by postconditions {ψ}.
In relation to “event structures”, in our model of semantic situations what we
call state is not an event structure because the predicate of the state expresses the
property of the objects or entities without the state being characterised by any
internal structure. What we have described as an event can be defined in the for-
malism of event structures as a situation in which the enabling time t0 coincides
with the start time t1, and the completion time t2 coincides with the effect time t3
(in short, t0 = t1 and t2 = t3). Thus an event can be defined as a particular case of
event structures which we would call critical situations. Ordinary processes, on the
other hand, can be described by event structures, on condition, however, that a
middle phase is added between the start and completion times.
Translated from French by Rosemary Masters
References
Banaszak, Zbigniew, Kuś, Janusz & Adamski, Marian. 1993. Sieci Petriego – Modelowanie, Ste-
rowanie i Synteza Systemów Dyskretnych, Zielona Góra: Wydawnictwo Wyższej Szkoły
Inżynierskiej.
Barentsen, Adrian. 1995. Trexstupenčataja model’ invarianta soveršennogo vida v russkom
jazyke. In Semantika i struktura slavjanskogo vida, Stanisław Karolak (ed.). Kraków:
Wydawnictwo Naukowe WSP.
Blackburn, Patrick; Gardent, Claire & de Rijke, Maarten. 1993. “Back and forth through time
and events.” Proceedings of the Ninth Amsterdam Colloquium, December 14–17, 1993.
Amsterdam: ILLC, University of Amsterdam, 161–75.
. In this paragraph we use the terms actions and events in their accepted sense in the
“theory of event structures” and not as we define them in our typology of semantic situations.
André Włodarczyk
Bojar, Bożenna. 1986. Werbalne wykładniki leksykalne relacji temporalnych w językach pols-
kim i bułgarskim. In Studia Gramatyczne bulgarsko-polskie, Tom I: Temporalność, 83–91.
Wrocław: Ossolineum.
Bulygina, Tatiana V. 1982. K postrojenju tipologii predikatov v russkom jazyke. In Semantičeskie
tipy predikatov, O.N. Seliverstova (ed.). Moskva: Nauka.
Chang, Nancy. 1997. A Cognitive Approach to Aspectual Composition. Ms, UC Berkeley Com-
puter Science Division and International Computer Science Institute. Report ICSI TR
97–034. 〈http://www.icsi.berkeley.edu/NTL/papers/cog.pdf〉
Chang, Nancy, Gildea, Daniel & Narayanan, Srini. 1998. A dynamic model of aspectual com-
position. Proceedings of the Twentieth Annual Meeting of the Cognitive Science Society
COGSCI-98. Madison WI.
Dowty, David. 1979. Word Meaning and Montague Grammar: The Semantics of Verbs and Times
in Generative Semantics and in Montague’s PTQ. Dordrecht: Reidel.
François, Jacques & Denhière, Guy (eds). 1997. Sémantique linguistique et psychologie cognitive,
Grenoble: Presses de l’Université de Grenoble.
Gagnon, M. & G. Lapalme. 1995. “From Conceptual Time to Linguistic Time.” Computational
Linguistics 22: 91–127.
Grafcet, Collectif ADEPA. 1979. Le Grafcet, diagramme fonctionnel des automatismes séquentiels,
rapport de la commision ADEPA sur la normalisation du GRAFCET, Montrouge.
Harada, Yasunari & Nakashima, Hideyuki. 1995. Situated disambiguation with properly speci-
fied representation. In Semantic Amibuity and Underspecification, Kees van Deemter &
Stanley Peters (eds). Stanford CA: CSLI.
Hoare, Charles A.R. 1969. An axiomatic basis for computer programming. Comm. ACM 12:
576–583.
Kamp, Hans & Reyle, Uwe. 1993. Formal Logic and Discourse Representation Theory, Introduc-
tion to Modeltheoretic Semantics of Natural Language. Berlin: Springer Verlag. Series: Stud-
ies in Linguistics and Philosophy, Vol. 42.
Koseska-Toszewa, Violetta. 1986. O temporalności inaczej. In Studia Gramatyczne bułgarsko-
polskie, Tom I: Temporalność, 39–64.
Kowalski, Roger A. & Sergot, Marek J.A. 1986. Logic-based calculus of events. New Generation
Computing, Vol 4.
Laskowski, Roman. 1986. System temporalno-aspektowo-modalny języka polskiego a sieci
Petriego. In Studia Gramatyczne bułgarsko-polskie, Tom I: Temporalność, 23–37.
Laskowski, Roman. 1998. Uwagi o znaczeniu czasowników. In Gramatyka współczesnego języka
polskiego, Morfologia, wyd. 2 zmienione. T. 1, 152–157. Warszawa: P.W.N.
Maggiolo-Schettini, Andrea & Winkowski, Józef 1992. Towards an Algebra for Timed Behaviours
[Report 677. Institute of Computer Science of the Polish Academy of Sciences. January
1990]. Also Theoretical Computer Science 103: 335–36.
Mazurkiewicz, Antoni. 1986. Zdarzenia i stany: Elementy temporalności (Events and states: Ele-
ments of temporality). In Studia Gramatyczne bułgarsko-polskie, Tom I: Temporalność, 7–21.
Mazurkiewicz, Antoni. 2000. Oral communication on progression at the meeting of Polish-
French Research Group “Etudes cognitives”. Paris: LACITO.
Mccarthy, John & Hayes, Patrick J. 1969. Some philosophical problems from the standpoint of
artificial intelligence. In Machine Intelligence 4, Bernard N. Meltzer & Donald Michie (eds).
Edinburgh: EUP.
Milner, Robin A. 1980. A Calculus of Communicating Systems. Berlin: Springer.
Moens, Marc. 1987. Tense, Aspect and Temporal Reference, Ph.D. dissertation, University of
Edinburgh.
Frames of semantic situations
Moens, Marc & Steedman Mark 1988. Temporal ontology and temporal reference. Journal of
Computational Linguistics 14(2): 15–28.
Murata, Tadao. 1989. Petri nets: Properties, analysis, and applications. In Proceedings IEEE-89,
Vol. 77, 541–576.
Murata, Tadao. 1991. Planning with Petri nets. In Information and Control.
Nakashima, Hideyuki. 1998. GAEA Version 2.3 Manual, Revision 011, Cooperative Architec-
ture Project Team, Electrotechnical Laboratory (ETL), Japan.
Nakashima, Hideyuki, Ohsawa I. & Kinoshita Y. 1991. Inference with mental situations,
TR-91-7: A technical report of Electronical Laboratory, Tsukuba. 〈http://ci.etl.go.jp/
kyocho/Papers〉
Narayanan, Srini. 1997. Talking the talk is like walking the walk: A computational model of ver-
bal aspect. In Proceedings of the Nineteenth Annual Meeting of the Cognitive Science Society
COGSCI-97. Stanford CA: Stanford University Press.
Pawlak, Zdzisław. 1981. Information systems – Theoretical foundations. Informations Systems
6: 205–218.
Pawlak, Zdzisław. 1982. Rough sets. International Journal of Information and Computer Science
11: 344–356.
Pawlak, Zdzisław. 1991. Rough Sets: Theoretical Aspects of Reasoning about Data. Dordrecht:
Kluwer.
Petri, Carl A. 1966. Kommunikation mit Automaten, Schriften des IIM Nr 2, Institut für
Instrumentelle Mathematik. Bonn 1962, English translation: Technical Report RADC-
TR-65-377, Grifftiths Air Force Base, New York, Vol. 1, Suppl. 1.
Piernikarski, Cezar. 1969. Typy opozycji aspektowych czasownika polskiego na tle słowiańskim,
Ossolineum, Wrocław.
Polkowski, Lech & Skowron, Andrzej. 1999. Towards adaptive calculus of granules. In Comput-
ing with Words in Information /Intelligent Systems [Studies in Fuzziness and Soft Comput-
ing 30], Lofti A. Zadeh & J. Kacprzyk (eds), 201–228. Berlin: Physica-Verlag.
Pradin-Chézalviel, Brigitte & Valette, Robert. 2000. Accessibilité de marquage et logique linéaire
dans un réseau de Petri t-temporel, IUT A. Toulouse: Université Paul Sabatier.
Reiter, Raymond. 2001. Knowledge in Action – Logical Foundations for Specifying and Imple-
menting Dynamical Systems. Cambridge MA: The MIT Press.
Sémon, Jean-Paul. 1986. Postojat’ ou la perfectivité de congruence, définition et valeurs textu-
elles. Revue des Etudes Slaves, T. 58/4, Institut d'Etudes Slaves, Paris .
Sowa, John F. 2000. Knowledge Representation: Logical, Philosophical, and Computational Foun-
dations. Pacific Grove CA: Brooks Cole.
Suraj, Zbigniew & Szpyrka, Marcin. 1999. Sieci Petriego i PN-Tools (Narzędzia do konstrukcji i
analizy sieci Petriego), Rzeszów: Wydawnictwo Wyższej Szkoły Pedagogicznej.
Vendler, Zeno. 1967. Verbs and times. In Linguistics in Philosophy, Zeno Vendler, 97–121. Ithaca
NY: Cornell University Press. (Revised version of Vendler Zeno. 1957. Verbs and times. The
Philosophical Review 66: 143–160).
Von Wright, Georg H. 1967. The logic of action: A sketch. In The Logic of Decision and Action,
Nicholas Rescher (ed.). Pittsburgh PA: University of Pittsburgh Press.
Winkowski, Józef 1992. Modelling timed behaviours with the aid of event and configuration
structures. ICS PAS Reports, N0 715, 34.
Winskel, Glynn. 1982. Event Structure Semantics for CCS and Related Languages. Springer LNCS
140: 561–576.
Winskel, Glynn. 1983. Event structure semantics for CCS and related languages. Aarhus Univer-
sity, DAIMI PB-159, April.
André Włodarczyk
Winston, Morton E., Chaffin, Roger & Hermann, Douglas J. 1987. A taxonomy of part-whole
relation. Cognitive Science 11(4): 417–444. Amsterdam: Elsevier.
Włodarczyk, André. 2003. Linguistique dynamique: évolution du discours dans le temps. In
Etudes linguistiques romano-slaves offertes à Stanislaw KAROLAK, 497–510. Kraków:
Edukacja.
Zadeh, Lofti A. 1996. Fuzzy logic = computing with words. IEEE Transactions On Fuzzy Systems
4: 103–111.
Zadeh, Lofti A. 1997. Toward a theory of fuzzy information granulation and its certainty in
human reasoning and fuzzy logic. Fuzzy Sets and Systems 90: 111–127.
Grounding of the meta-informative
status of utterances
André Włodarczyk
Université Ch. de Gaulle (Lille 3) / Université Paris-Sorbonne, Centre de
linguistique théorique et appliquée (CELTA)
1. Introduction
Linguistic terminology and research methods are elaborated following the para-
digm shifts of social sciences, starting with the approaches known as normativist,
evolutionist, structuralist and mathematical but ending up with the computational
one. However, the recent computational shift did not bring the desirable system-
atic insight into the really hard problems that the research on natural languages
requires from linguists. For this reason we claim that by using computational
metaphors and tools as well as by taking into account discoveries in the field of
neurology, reconstructions of many linguistic notions can gain much in coherence
and adequacy in the future. Consequently, the proposed approach postulates the
need for an interdisciplinary collaboration between linguists, logicians, psycholo-
gists, neurologists and computer scientists.
Human communication uses a basically aerial channel so that the repre-
sentations resulting from both sequential and parallel mental operations must
be converted to/from a series of signs, i.e.: they must be encoded/decoded in
sequences of sounds (these being modelled/recognized due to the phonemic
abstractions). Obviously, as linguistic units cannot be uttered other than by
following a sequential order, the result of both sequential and parallel operations
on expressions is known as a syntactic construction. Undoubtedly, subject and
object actually exhibit the syntactic organization of linguistic expressions, but
this is far from being their primary role, hence syntactic functions of both sub-
ject and object are secondary in our theory. Let us recall that what – following
the Prague School t radition – linguists usually refer to as “information” (cf. the
notion of “Information Structure”) we name meta-information, reserving the
term information to refer to the semantic content of linguistic utterances. In our
approach meta-information means literally “information about another informa-
tion” because it is mainly responsible for the constitutive structure of predication
(aboutness) and ultimately determines the sequential ordering of the information
flow in utterances.
When modelling semantic situations, it is convenient to apply elements of
dependency1 of some kind in order to render the non-linear nature of meaning.
In addition to this, it seems that a separate meta-informative layer/dimension is
necessary in order to achieve the ordering of non-linear, mental representations
. Dependency is a (directed tree-like) relation first applied to syntax by S.D. Katsnelson
(1948) and by L. Tesnière (1953), both using the same term valence in order to represent the
relations between the units of an utterance.
Grounding of the meta-informative status of utterances
. Among them Slavic languages for which Mel’čuk I. (1988, 2003) proposed a dependency
syntax. The standard syntactic structure theory proposed by Chomsky N. (1957) was based
on constituency.
André Włodarczyk
As we put it above, predication in its specifically linguistic sense, i.e. as it has been
defined within the MIC theoretical framework, consists of uttering an information
about another information which takes the form of the subject of an utterance.
The first theoreticians of the subject-predicate relationship in natural languages
were obviously Plato and Aristotle, and many grammarians and linguists followed
them, giving rise to the foundation of what is known today as “traditional syn-
tax” preceding the theoretical effort made within the Generative Grammar frame-
work. The difficult (almost impossible) mutual understanding between linguists
and logicians concerning “predication” comes from the fact that logicians defined
predication as a formal relationship between terms, withdrawing from the original
concept of predication viewed as aboutness, i.e. in the way that fits the purposes of
artificially built formal, as opposed to natural, languages. In logic, predication may
be used in two ways. Firstly, predication is understood as the assertion P(a) where
P is a predicate and a is an individual term to which it applies. Thus the assertion
“Peter is a student” is interpreted using the set-theoretical framework (where an
element a belongs to a set) as “the individual called Peter belongs to the set of
students”. Secondly, in predicate calculus, predication is considered as the logical
formula of a relation: P(a, b) where the terms a and b fill in the argument places of
the predicate P; this is the way logicians interpret “Peter bought a book”: buy(Peter,
book). None of these formulae, neither P(a) nor P(a, b), takes account of the ‘linear
order’ (more properly, sequential order of constituents) of linguistic predicative
expressions (utterances) because they reflect the semantic relational information
itself, leaving aside the quite easily observable fact that what is explicitly stated in
the linguistic form is not semantic information but another information about it.
Thus, as such, logical formulae can prove to be more or less convenient for the
formal notation of semantic content but not for grasping the meta-informative
structure of natural utterances.
Since the last quarter of the 20th century, the genuinely linguistic view on predi-
cation has been making a come-back in a more and more formalised way, especially
in the fields of logical pragmatics and computer science, because it became clear
that a purely syntactic framework, and even the so-called “semantico-syntactic”
frameworks, were insufficient to account for natural language real utterances. The
syntactic and semantico-syntactic models designed on the basis of generative gram-
mar led to very complicated theories, often disconnected from linguistic intuition
and evidence. The most relevant objections to purely syntactic approaches came
not only from the semanticists but also from the logicians [Thomason R. H. – 1977]
who emphasised that only a formalised pragmatic theory (on the basis of Grice’s
pragmatics) might contribute to simplifying syntax and semantics.
Grounding of the meta-informative status of utterances
Table 1. Properties of selection and combination in the process of meaning creation
Bi-axial nature of discourse Semantics Pragmatics
Information Meta-information
Selection Property Comparison Centering
(paradigmatic oppositions) a ∈ A (attribute space) (distinguish e)
Combination Situation Configuration Predication
(syntagmatic relations) r(x) (relation on x) (say p about e)
inf(r) ≅ e m-inf(c) ≅ r
Figure 1. Information (inf) and meta-information (m-inf) can be seen as a composi-
tion of functions (m-inf ⋅ inf) from discourse to ontology or reference (with semantics as a
common term)
that this problem reveals the real nature of linguistic predication. We propose
therefore that more attention be paid, in linguistic studies, to old/new alternation
within the syntactic structure of utterances.
Nevertheless, since the notions of old and new have given rise to a number of
interpretations in the literature,4 it is crucial to emphasize the original contribu-
tion of the concept of meta-informative old/new status to linguistic studies of this
sort of problem. The old/new opposition is often seen as representing alternative
terms either for anaphoric/cataphoric or known/unknown. We consider that the
meta-informative status of an utterance, on condition that it is in accordance with
the verb schema (valency), depends on the pragmatic choice of the speaker and,
as it is shown below (Table 2), that it is motivated by communicative and cogni-
tive acts as well as by ontology. The reason for this is that oldness and newness are
neither values nor features of information, rather they concern the way information
is treated in communication. It is precisely this treatment that we call “assigning
a meta-informative old/new status to information (or its chunks)”. Consequently,
we propose replacing the notion of ‘old/new information values’ with that of ‘meta-
informative old/new status’ of any information being a part of an utterance or a text.
Let us look at a few important consequences of the proposed theoretical
shift. In order to elucidate the discourse grounding of newness, let us recall, as an
example, that the life and work of a famous scientist, say Albert Einstein, belong
in our time to the shared knowledge of most people with high school education.
However, in a text Einstein’s name may be freely introduced by the speaker with
either an old or new meta-informative status. In the following utterance: It is Albert
Einstein who discovered the law of Relativity, the subject Albert Einstein is focalised
as a new chunk of information contrasting with what is treated as the old informa-
tion: “there is a person who discovered the law of Relativity (supposedly although
you do know that this law has been discovered, what you do not know is the name of
that physicist) and I am telling you: that person’s name is Albert Einstein”. The aim
of the following example is to show that text or discourse itself is not a sufficient
condition of the meta-informative old or new status. A noun phrase which has
already appeared in a text/discourse may well be treated as having the new status
in a subsequent part of the same text/discourse just because the entity it refers to is
supposed to enter into a new situation (becomes a participant of a new role). For
instance, within the general theme of a text on Chopin’s life, the name “Chopin” (or
. For a review of the most frequently used meanings of the terms given and new, see [Prince,
E. – 1981]. She established that given has at least three different meanings: (1) predictability,
the speaker assumes that the hearer is able to predict the occurrence of an item within a sen-
tence (Halliday 1967), (2) saliency, assumed by the speaker to be present in the consciousness
of the hearer (Chafe1976) and (3) shared knowledge, known and accepted as true (Clark &
Haviland 1977).
Grounding of the meta-informative status of utterances
“Frederic”) may nevertheless occur with the meta-informative new status despite
the fact that this name has already been quoted many times and that Chopin is a
very famous composer, as in the following example: “It was Chopin whom Georges
Sand invited to the party.” This is an important point making it possible to explain
also why pronouns, although most often used as anaphoric or cataphoric devices
in a text, may be used to refer alternatively to old or new chunks of information:
the chapters concerning personal pronouns in this volume provide evidence for
the need to distinguish between the old/new status as ‘recoverable’ from the text
and as that which is grounded in the World and refinable by shared knowledge.
Obviously, although first and second person pronouns are unambiguously iden-
tified as the speaker and hearer, in every speech situation, they may be treated
as referring to either old or new meta-informative status of utterances. Similarly,
although the third person pronouns normally refer to the entities mentioned in
previous parts of the text, they may nonetheless appear in the text with either new
or old status. Many examples of such usage of personal pronouns can be found in
Latin, Greek or Russian utterances which are analysed in the present volume (see
chapters by Azam O., Pitavy J.-Ch. and Vedrenne P.). This suggests that in utter-
ances occurring in different contexts, personal pronouns may encode either old or
new meta-informative status: sometimes (a) as the subject or object of utterances
with totally new or old status and sometimes (b) as a focus (new information) con-
trasting with its background (the status of which is old) or a topic (old information)
contrasting with its comment (the status of which is new).
. Note that such discourse strategies are partly fixed by syntactic rules of languages but
here we focus only on the range of possible choices the speakers have at their disposal when
introducing their own points of view.
André Włodarczyk
. We understand ontology as the representation of the world in the speaker’s mind. The
ontology which is activated in the speaker’s mind and, at the same time, matches the speech/
text situation and its environment, is said to be grounded.
. The term potential is understood here as the capacity of an entity to play a role in a given
situation (not as an equivalent of possible which in modal theory is opposed to necessary).
André Włodarczyk
Σ-situations P-situations
(situation types) (situation instances)
Pairs in A differ from pairs in B by the fact that while the former concern
situations, the latter describe the entities (participants which enact roles). Saying
so, we combine the two views on ontology (namely, the situation-oriented view
of Wittgenstein and the entity-oriented view of Aristotle, see Wolniewicz 1969).
Indeed, ontological grounding is conditioned by the categorisation of the world.
It is well-known that Aristotle distinguished between substance and accidence,
giving rise to the definition of entity (object ontology). In modern ontology,
Wittgenstein L. has shifted the Aristotelian view of ontology from entities to facts.
“Aristotle’s ontology is an ontology of substances, Wittgenstein’s ontology is an
ontology of facts. But (…) both ontologies lay down conditions which in view of our
vocabulary appear to be identical.” And farther: “…let us note in passing that their
symmetric character seems to be itself an additional manifestation of the parallelism
discussed.”.[Wolniewicz B. – 1982]
Nonetheless, for Ingarden (1974), entities are no more isolated beings/objects but
consist of interrelated elements and should be viewed as structures. In the case of
living beings viewed as structured objects, they are said to be able to occur in vari-
ous environments or situations to which they adapt and with which they interact.
Such living beings are known as agents and their interaction with the universe as
well as between themselves is called their behaviour (Ingarden 1974). Structures
built of elements which develop their behaviour in environments can be described
in terms of the theory of general systems. Consequently, in Associative Semantics
theory the universe contains objects (i.e.: animate and inanimate entities) which
are related to situation roles through their behaviours within situation frames or
environments. In addition to the above, entities as well as facts may each be seen
either as types (classes) or as instances (tokens).
On the other hand, due to the fact that the First Order Logic8 (FOL) has
been derived from reflections on natural languages, linguists often study jointly
. The notions of types and instances (occurrences) of situations are sometimes compared
to the universally (∏) and existentially (∑) quantified arguments of logical predicates, though
in First Order Logic (FOL) the symbols are inverted (∏ denotes the universal quantifier
Grounding of the meta-informative status of utterances
two logical concept quantifiers (some/all) and descriptors (a/the) as being closely
related linguistic devices. But natural languages have the possibility of expressing
ontological features and relations using the markers of noun determination (quan-
tifiers, articles) as well as those of verb aspects and modalities.
New Old
Comment Background
Local Global
Π Σ
Object Subject
Figure 2. The formal concept analysis context9 of information centering in utterances
while ∑ denotes the existential one). We follow here the set-theoretical tradition in which the
symbols have been inverted.
. The Formal Concept Analysis (FCA) Context (Wille R. 1982) makes it possible to visu-
alise, in lattice form, heterarchies of concepts representing information systems (binary tables
or bipartite graphs with objects, attributes and assignments).
André Włodarczyk
It is a well-known fact that the semantic agent often corresponds “quite n aturally” –
one would be tempted to say – to the subject of a sentence. This statement gives
rise to a quite understandable confusion which consists of taking every subject for
an agent because of the impression that the scope of agentivity can be extended as
many times as needed due to the rhetorical operation called metaphor. Neverthe-
less, although many linguists claim that this is not the case, bearing in mind that
agent and subject belong to two different kinds of notions, semantic and syntac-
tic, respectively, some linguists distinguish between “syntactic” (“formal”) subjects
and “notional” (“logical”) ones. We argue however that the subject as represent-
ing a global centre of attention and the object defined as a local centre of atten-
tion, in fact indicate that their functions are essentially pragmatic, namely they
point to some chunks of information thus playing a meta-informational role in
communication. In our view, semantic denotations (information) and pragmatic
selections (meta-information) are parallel (more specifically probably concurrent)
operations.
Both the centered information (which is spoken about) and the about-
information are supposed to be always endowed with communicative old/new
status. The aboutness relating to (a) S-subjects and P-subjects, in base utterances,
and to (b) Topics and Focuses, in extended utterances, is expressed in the second
part of the utterance; i.e. as S-predicates and P-predicates, on the one hand, and as
comments and backgrounds, on the other. The aboutness of (general and particular)
themes are “summaries” of the units of texts/discourses which are bigger than an
utterance. Their relationship with the units they represent is syntagmatic as well as
paradigmatic.
Grounding of the meta-informative status of utterances
7. Conclusion
To sum up, meta-informative old and new status are motivated by (a) the flow
of information in an inter-utterance communication (anaphora/cataphora), (b)
concerns resulting from the discourse (known/unknown) and (c) beliefs pertaining
to ontological knowledge. Anaphora and cataphora are not simply syntactic devices.
They are operations driven by the centered information which is expressed either
within the limits of an utterance or in two adjacent utterances. Consequently, as such,
these statuses should not be confused with their grounding itself. In an older version
of the MIC theory, grounding types were not clearly distinguished. The anaphora/
cataphora pair was inserted into a dual list of totally ordered concepts without dis-
criminating between inter-utterance communicative, discourse (recently acquired
information) and ontological grounding. We did this in order to account for lin-
guistic morphemes10 expressing alternatively all these concepts. In fact, we were
not wrong to consider the above mentioned list of concepts as representing roughly
two “graduated scales of opposite values of old and new information” (Włodarczyk
& Włodarczyk 1998). Indeed, all these notional pairs can be perceived form a dual
perspective ranging over some strength criterion and its opposite.
We did not develop the operational strategy (internal mental representations
in the human brain) which underlie the information centering of utterances. We
think however that the three following operations seem to be quite useful for far-
ther inter-disciplinary approaches.
. Japanese noun particles (Włodarczyk A. 1980, 1996, 1998) and Polish and Russian verb
aspect (Włodarczyk H. 1994, 1997).
André Włodarczyk
Acknowledgment
References
Baader, Franz, Calvanese, Diego, McGuinness, Deborah L., Nardi, Daniele & Patel-Schneider,
Peter, F. (eds). 2003. The Description Logic Handbook: Theory, Implementation, and Applica-
tions. Cambridge: CUP.
Chafe, Wallace. 1976. Givenness, contrastiveness, definiteness, subjects, topics, and point of
view. In Subject and Topic, Charles Li (ed.). New York NY: Academic Press.
Chomsky, Noam. 1957[2002]. Syntactic Structures. Berlin: Mouton de Gruyter.
Clark, Herbert H. & Haviland, Susan E. 1977. Comprehension and the given-new contract. In
Discourse Production and Comprehension, Roy O. Freedle (eds). Hillsdale NJ: Lawrence
Erlbaum Associates.
Halliday, Michael A.K. 1967. Notes on Transitivity and Theme in English (Part 2). Journal of
Linguistics 3, p. 206.
Halliday, Michael A.K. & Greaves, William S. 2008. Intonation in the Grammar of English.
London: Equinox.
Ingarden, Roman. 1974. Der Streit um die Existenz der Welt, Vol. 3: Über die kausale Struktur der
realen Welt. Tübingen: Niemeyer.
Jakobson, Roman & Halle, Morris. 1956. Two aspects of language and two types of aphasic
disturbances. In Fundamentals of Language, Roman Jakobson. The Hague: Mouton.
Katsnelson S. D. 1948. O grammatičeskoj kategorii. Vestnik LGU 1948 N°2. (Second edition
1986, in Obscee i tipologičeskoe jazykoznanie. Leningrad).
Mel’čuk, Igor. 1988, Dependency Syntax: Theory and Practice. Albany NY: The SUNY Press.
Mel’čuk, Igor. 2003. Levels of dependency in linguistic description: Concepts and problems.
Dependency and Valency. An International Handbook of Contemporary Research, vol.
1, Vilmos Agel, Ludwig M. Eichinger, Hans Werner Eroms, Peter Hellwig, Hans Jurgen
Herringer & Henning Lobin (eds), 188–229. Berlin: Walter de Gruyter.
Prince, Ellen F. 1981. Toward a taxonomy of given-new information. In Radical Pragmatics
223–255. New York NY: Academic Press.
Tesnière, Lucien. 1953. Esquisse d'une syntaxe structurale. Paris: Klincksieck.
Thomason, R.H. 1977. Where pragmatics fits in. In Proceedings of the Texas Conference on
Performatives, Presuppositions and Implicatures, Andy Rogers, Robert Eugene Wall & John
Peter Murphy (eds), 161–166. Washington DC: Center for Applied Linguistics.
Von Wright, Georg H. 1963. The Logic of Preference. Edinburgh: EUP.
Wille, Rudolf. 1982. Restructuring lattice theory: An approach based on hierarchies of con-
cepts. In Ordered Sets, Ivan Rival (ed.), 445–470. Dordrecht: Reidel (Reprint in: ICFCA
‘09: Proceedings of the 7th International Conference on Formal Concept Analysis, Berlin,
Heidelberg, 2009, 314).
Grounding of the meta-informative status of utterances
Wisniewski Edward J. 1997. Understanding novel noun phrases. In Proceedings of IWHIT 1997,
The University of Aizu, 97–101.
Włodarczyk, André. 1980. 主題から主語へ、そして主語から主題へ – ハとガ (Shudai kara
shugo e, soshite shugo kara shudai e – wa to ga) (From Topic to Subject and from Subject
to Topic – wa and ga). In 言語月刊 (Language monthly), Vol. 9, No 8/80, 大修館 Tokyo:
Taishûkan.
Włodarczyk, André. 1996. La thématisation en japonais et l’article français. Studia Kognity-
wne: Sémantique et Linguistique Contrastive, Warszawa: Editions Slawistyczny Ośrodek
Wydawniczy.
Włodarczyk, André. 2003. Linguistique dynamique: Évolution du discours dans le temps. In
Etudes linguistiques romano-slaves offertes à Stanisław Karolak, 497–510. Kraków: Oficyna
Wydawnicza Edukacja.
Włodarczyk, André. 2008. Roles and anchors of semantic situations. Etudes cognitives/Studia
kognitywne VIII, 53–70. Warszawa: SOW.
Włodarczyk, André & Włodarczyk, Hélène. 1998. Graded Informative content of linguistic
messages. In Proceedings of the 16th International Congress of Linguists. Oxford: Pergamon.
(CD-ROM: Paper N0 0431).
Włodarczyk, André & Włodarczyk, Hélène. 2006a. Focus in the meta-informative centering
theory. In La Focalisation dans les langues. André Włodarczyk & Hélène Włodarczyk (eds).
Paris: Coll. Sémantiques, L’Harmattan.
Włodarczyk, André & Włodarczyk, Hélène. 2006b. Subject in the meta-informative centering
theory. Études cognitives/Studia kognitywne 7. Warszawa: SOW.
Włodarczyk, André & Włodarczyk, Hélène. 2008. The pragmatic validation of utterances.
Études Cognitives/Studia Kognitywne 8. Warszawa: SOW, PAN.
Włodarczyk, André & Włodarczyk, Hélène. 2012. Subjecthood and topicality are both prag-
matic issues. An International Workshop on Linguistics of BA and the 11th Korea-Japan
Workshop on Linguistics and Language Processing. December 10–11, 2011. Tokyo: Waseda
University.
Włodarczyk, Hélène. 1994. L’Aspect verbal slave et les domaines du donné et du nouveau. In
Etudes cognitives, Vol.1: Sémantique des catégories d’aspect et de temps, 113–130. Warszawa:
Polska Akademia Nauk, Instytut Slawistyki, Slawistyczny Ośrodek Wydawniczy.
Włodarczyk, Hélène. 1997. L’Aspect verbal dans le contexte en polonais et en russe. Paris: Institut
d’Etudes Slaves.
Wolniewicz Bogusław. 1969. A parallelism between Wittgensteinian and Aristotelian ontolo-
gies. In Boston Studies in the Philosophy of Science, Vol. IV, Robert S. Cohen & Marx W.
Wartofsky (eds), 208–210. Dordrecht: Reidel.
Wolniewicz, Bogusław. 1982, A formal ontology of situations. Studia Logica 41: 381–413.
Wooldridge, Michael J. 2000. Reasoning About Rational Agents. Cambridge MA: The MIT Press.
Attention-centered information in language
Hélène Włodarczyk
Université Paris-Sorbonne, Centre de linguistique théorique
et appliquée (CELTA)
1. Introduction
In current linguistic research, argument structure and verb valency theories are
attempts to capture the mappings between noun phases in the syntactic plane
and the roles enacted by the participants in the semantic plane of discourse. In
the MIC theory this correspondence is recognised as a more complex phenom-
enon, the solution of which cuts across another dimension of language, namely,
Hélène Włodarczyk
The function of attention in human cognition has been a major topic in experimen-
tal psychology since the very beginning of this science, and it is still a fundamental
topic of neuro-cognitive research (Bunting et al. 2008; Cowan 2008; Shelton et al.
2008; Stachowiak in this volume). Let us quote a statement from a psychological
study about attention: “… the experience of attending to an object becomes an
experience of being aware of that object when it is conjoined with attending to a
representation of the self. “(LaBerge 2003: 291). In linguistic studies this statement
from neuro-cognition can be related to the meta-informative level which neces-
sarily accompanies any utterance. This level is represented in the MIC theory by
the two properly meta-linguistic (“meta-informative”) “predicates” (in the sense
of relations): speak with (x, y) and speak about (x, z) which are necessarily added
to the predicate representing the situation spoken about, this situation a ppearing
Attention-centered information in language
as the z variable in the predicate speak about (x, z). The “representation of the
self ” mentioned by psychologists has its counterpart in the representation of the
speaker (and his point of view) in linguistic utterances.
In linguistics, we need a general concept to capture what is common to the
concepts of subject, object, topic and focus. In the MIC theory, these concepts are
put together under the heading attention-driven phrases (ADP) and are considered
as segments of linguistic utterances (on the level of syntax) pointing at “centres of
attention” (CA) defined as psychological phenomena (on the level of cognition).
Paradoxically, as we shall see below, without this common concept of ADP, the
specificity of each of them as subject, object, topic and focus is difficult to grasp.
As applied to language studies, the concept of attention centering was first
introduced within the framework of computational linguistics (known as center-
ing theory, Grosz et al. – 1986, 1995) with the aim of automatically discovering
coherent information from text documents. For this purpose, the terms forward-
and backward-looking centres have been coined, taking it for granted that informa-
tion is kept coherent in text, and more specifically in adjacent pairs of utterances.
In this theory, the concept of centre of attention concerns only the text/discourse
level and, even more importantly, attention focusing is not considered as one of
the most important operations underlying the structure of a single utterance. In
other words, the computational Centering Theory did not encompass definitions
of the subject and object as attention-driven phrases governed by the centres of
attention on the first meta-informative level (where no contrast is proposed by
the speaker between the old and new information),1 neither were the topic and
focus defined as dually opposed concepts, respectively the topic (as governed by
an old meta-informative status contrasting with the ‘new” one of the comment)
and the focus (as governed by a new meta-informative status contrasting with the
old one of the background). It should be stressed that the meta-informative cen-
tering (MIC) theory is not an alternative to the computational one. Rather, it is
a complementary approach; in the MIC approach, the anaphoric and cataphoric
motivations of the meta-informative old and new status of the utterance during the
communication act (cf. Chapter 3 in this volume), match perfectly the concepts of
backward and forward looking centres in the American centering theory, which can
therefore be integrated into the MIC framework.
Although the American theory precedes the European one in time, it is
important to state that these theories have separate origins. The origins of the MIC
theory go back to 1999 when we first used the concept of centre of interest (“centre
. The definition of old and new meta-informative status is introduced in Chapter 3 in this
volume.
Hélène Włodarczyk
Moreover, in the MIC theory, we consider yet another cause of partiality, namely
in natural utterances, there is an implication between local and global ADPs; a
local ADP is dependent on a global one which may remain implicit, i.e. when the
object is present, the subject may remain implicit and when the focus is present,
the topic may be omitted. This implication has very important consequences both
on the level of base and extended utterances (see below).
Base and extended utterances are defined as communicative discourse units. Each
utterance can contain more than one centered segment: we distinguish between
global and local CAs and between base and extended utterances, the latter includ-
ing second level ADPs (topic and focus) added or superimposed on first level
ADPs (subject and objects). Utterances in which the subject and the predicate
have the same meta-informative status (old or new) are said to be base utterances:
we call this the first meta-informative level.
. Two reasons for subject omission in Japanese are taken into account in the chapter here-
after by Nariyama Shigeko: (1) the morpho-syntactic ground: the Japanese verb has no per-
sonal category (no personal endings) (2) the meta-informative ground: the subject is omitted
when it is old.
Attention-centered information in language
Table 1. Three layers of linguistic information (from Włodarczyk A. & H. 2006a)
informative layer typical semantic units: situation schemata
meta-informative layer linguistic units uttered in a context:
1st m-inf. level: base utterances
2nd m-inf. level: extended utterances
abstract cognitive layer arrangements of sets of utterances:
texts, monologues, dialogs, discourses etc.
In the MIC theoretical framework, the theme is not merely a synonym of topic
(Table 2). As a matter of fact, we define topic and theme by reference to the repre-
sentation layer of linguistic information to which they respectively belong. We use
the word “theme” as a term referring to texts or discourses as organised, linguisti-
cally coherent sets of utterances. Thus, making reference to the meta-informative
status of base and extended utterances respectively, it is possible to capture and
better explain the difference and, at the same time, the similarity which character-
ise the subject and the topic.
Global Local
. Topicalisation and focalisation of subordinate clauses in Polish have been investigated in
the MIC framework by Miladi L. 2006, 2009.
Hélène Włodarczyk
of the second meta-informative level are often realized as left or right dislocated
constituents, i.e. they occur out of the clause boundaries on the left (in utterance
(2), three topicalised noun-phrases are separated from the clause by the dash ) or
on the right (in (3), the topic is placed at the end, after the comma).
Transport of the mails, transport of the human voice, transport of flickering
(2)
pictures – in this century as in others our highest accomplishments still have
the single aim of bringing men together. (Antoine de Saint-Exupéry, Wind,
Sand and Stars)
(3) They bought a new car, Susan and Tom.
. See the book chapter “Extensions de l’énoncé” in [Cotte et al. – 2004].
Attention-centered information in language
(of a base utterance) pointing at the main centre of attention (in the mental rep-
resentation of a situation).
Object is also defined as an attention-driven phrase referring to a centre of
attention but, as it is a local one in the constituent structure of the utterance, it is
secondary with respect to the subject. Thus, the object is dependent on the subject
and its occurrence in an utterance entails a subject. Talmy Givón, among contem-
porary linguists, also takes into account the importance of attention and, in this
regard, his approach is close to our point of view: “the subject and DO (direct
object) may be viewed as the grammaticalized primary and secondary topic of
the discourse at the time when the clause in which they appear is being processed.
(Givón 2001: 198)” In fact, what Givón calls the pragmatic function “topic” we call
more generally the “centre of attention”. However this author relates the necessity
of selecting a “primary and a secondary topic” exclusively to the cognitive limita-
tions imposed by attention in the human brain, but not explicitly to the sequential
(“linear”) nature of units in linguistic utterances.
In an utterance the subject of which has been dropped5 – because it has an “old”
meta-informative status and can be easily recovered from the context – the pres-
ence of the object allows us to infer the existence of the absent subject. However, the
possibility of allowing the subject be completely absent (without any prosodic, mor-
phemic or syntactic marker) or to represent it by a pronoun, is language-specific.
(4) Mary entered the store and bought a book.
. N.B.: We treat in two different ways the possibility of dropping an “old” status subject in
the third person, noun or pronoun, and the use of the zero form of a subject pronoun in the
first or second person in languages where verbs exhibit explicit personal endings making it
possible to identify the subject as the speaker or hearer (Włodarczyk, H. 2012).
Hélène Włodarczyk
Polish utterance (6) may be uttered for instance as the continuation of (7) hereaf-
ter, which is a translation of English (4):
(7) Pol.
Mary weszła do sklepu
N nom fem sg Vpast 3pers sg fem prep n gen mas sg
Mary entered into the store
i
kupiła książkę.
Conj Vpast 3pers sg fem n acc fem sg
and bought a book
. It is however sometimes possible to postpone the semantic interpretation of the subject.
In such cases, the speakers just communicate information about an entity which appears to
them as their global centre of attention. The subject appears first as a pseudo-active partici-
pant until a deeper semantic interpretation is possible, e.g.: the utterance “A car (pseudo-active
role) hit a pedestrian.” can be interpreted as “A driver (active role) hit a pedestrian with his car.”
(Włodarczyk & Włodarczyk 2012).
Attention-centered information in language
(8) Pol.
Jest ojciec w domu.
V 3pers sg pres n mas sg nom prep n mas sg loc
is father at home
“Father is at home.”
(9) Pol.
Nie ma ojca w domu.
neg V 3pers sg pres n mas sg gen prep+n mas sg loc
not is father at home
“Father is not at home.”
(10) Pol.
Student kupił książkę.
n mas sg nom V 3pers sg past n fem sg acc
student bought book
“The student bought a/the book.”
(11) Pol.
Student nie kupił książki.
n mas sg nom neg V 3pers sg past n fem sg gen
student bought book
“The student did not buy a/the book.”
As a matter of fact, the same participant of the semantic situation can be alterna-
tively introduced into the discourse by the speaker either as the subject or as the
object of the utterance, depending on his/her viewpoint or, more precisely, on the
attention he/she pays to it. Traditional occidental grammarians were well aware
of the interchangeability of subject and object in the utterance. The passive voice
is one of the devices which natural languages use to realize such interchangeabil-
ity without changing the information content of the utterance,7 e.g. “The Greeks
defeated the Persians.” (active construction) vs “The Persians were defeated by the
. Nor its truth validity (see Włodarczyk & Włodarczyk 2008b).
Hélène Włodarczyk
(12) Ger.
Die Regierung betrachtet
n fem sg nom/acc V 3pers sg pres act
the government considers
das Volk als korrupt.
art n neu sg nom/acc conj + adj
the people as corrupt
“The people consider the government corrupt.”
But, if we interpret this utterance as an extended one, we have two possible meta-
informative glosses:
which can be interpreted as (literally): “As for the government, people consider it
as corrupt.” If we put a pitch accent on the first NP we can interpret the utterance
as containing a focus:
Consequently, utterance (12) would be then translated into English by: “It is the
government that the people consider as corrupt.”
The following utterance (13) exhibits a different word order.
(13) Ger.
Das Volk betrachtet die Regierung als korrupt.
n neu sg nom/acc V 3pers sg pres active n fem sg nom/acc conj + adj
the people considers the government as corrupt
“The people consider the government corrupt.”
Furthermore, with a stress on the first NP in utterance (13), we get the following
interpretation:
In utterances (12) and (13) both noun phrases have homonymic forms of nomi-
native and accusative cases because they are feminine and neuter nouns respec-
tively: Die Regierung (the government) and das Volk (the people). Nevertheless,
whatever be the word-order (either 12 or 13), any German native speaker prefers,
even out of context and in a written form with no prosodic marker, the read-
ing in which the people fill the active role and the government the passive role
because, as the author of the dissertation puts it, “… it is part of our knowledge
of the world that usually people consider the government corrupt and that it is not
the other way round.” (Mohr 2005). It is necessary to add that, in oral speech,
prosodic markers enable us to interpret both 12 as 13 either as base utterances
(with a subject+predicate structure) or as extended ones (with a topic+comment
or focus+background structure).
As a matter of fact, the MIC theory makes it possible to predict that, in written
German, each of these two utterances has theoretically possible various interpre-
tations among which the adressee can only choose when s/he takes into account
the common sense knowledge which makes it possible to reject the option of con-
sidering the two NPs alternatively as filling the active or passive role. The differ-
ent readings of each sentence depend on the prosody which makes it possible to
distinguish between base and extended utterances (with topic and/or focus). In
the above German examples, knowledge of the world makes it possible to interpret
utterance (12) as an extended utterance even in its written form, without prosodic
markers. Without prosodic markers (in the written form of the utterance consid-
ered out of context), it is impossible to decide whether it is a topic-comment utter-
ance in which the object NP “die Regierung” is topicalised or a focus-background
utterance in which the object NP is focalised (and therefore put in the first posi-
tion) or whether it is a focus-background utterance in case there is a stress on the
first NP. In German when the topicalised or focalised object appears in the first
position, as a consequence, the subject (here: das Volk) appears in the postverbal
position because German syntax puts the verb in the second place. Depending on
intonation, utterance (12) is interpreted as a base (subject-predicate) utterance or
as an extended (focus (here: focalised subject) – background) utterance. Thus the
different possible interpretations (and prosodic realisations) of a linear sequence
of morphemes depend ultimately on the text structure which in turn depends on
centering as cohesion device.
Moreover, let us point at the fact that in case marking languages (for Slavic
languages, see Włodarczyk H. 2009) any change of word order is sufficient to lead
to the extension of the utterance, i.e. the topicalisation or focalisation of the “dis-
placed” constituent. In German not only the first position in the utterance but also
the “after last” position (Nachfeld) can be used for topicalization or focalization
(Vinckel 2004).
Hélène Włodarczyk
Fr. L’effort pour coordonner les données tactiles et visuelles, reste l’un
(14)
des plus prodigieux de ma vie […]. Cet effort, chaque enfant en bas âge
l’accomplit.
“The effort to coordinate tactile and visual data remains one of the most tre-
mendous efforts in my life […] Such an effort, each infant has to accomplish
it.”
In (14) the Object NP “cet effort” is “left dislocated” (in our terms “topicalised”)
but Gaatone stressed that it could just as well be the subject of an utterance in
passive voice, as in (15) below.
(15) Fr. Cet effort est accompli par chaque enfant en bas âge.
“This effort is accomplished by each infant.”
In our view, both utterances (14) and (15) fit in the text coherence but the choice
between them makes it possible to put more or less emphasis on the noun phrase
“cet effort”. The topicalised NP is more highlighted than the subject of the utter-
ance in passive voice and, last but not least, it has an old meta-informative status
which can be motivated by anaphora (cf. Chapter 3 in this volume). This is prob-
ably the reason why the topicalised object was used by the author of the original
version of this utterance and why, in fact, utterances (14) and (15) are not simply
equivalent.
. Polish scholars insist that in Polish (and other Slavonic languages) the word order is “free”
but not “arbitrary”: “Układ wyrazów w zdaniu w języku polskim jest swobodny, jakkolwiek
niezupełnie dowolny.” (“The word order in a Polish sentence is free, however it is not com-
pletely arbitrary”.) (Szober 1967). We claim, however, that each different order has a different
meta-informative interpretation.
. In Polish, the use of the particle to after a topicalised NP and/or before a focalised NP
is not obligatory. More research should be conducted in order to demonstrate on large data
whether the use of to is really characteristic of oral communication and less frequently used
in writing (literary) style. Actually utterance (17) has two different readings depending on the
intonation: either only the first NP is topicalised and the rest of the utterance is the comment,
or the first topicalised NP is followed by a focalised NP as we interpreted it in the gloss we
proposed.
Hélène Włodarczyk
(18) Pol.
Książkę podarowano
n fem sg acc V 3pers neu sg past imprs
book [it] was offered
Jankowi na urodziny.
n mas sg dat prep + n -hum pl acc
to Janek for birthday
Lit. “The book [someone] offered to Janek for his birthday.”
(19) Pol.
Książka ta została podarowana
n fem sg nom det fem sg nom V Passive 3pers fem sg past
book this was offered
Jankowi na urodziny (przez ojca).
n mas sg dat prep + n -hum pl acc prep + n mas sg acc
to Janek for birthday by father
“As for the book, it was offered to Janek for his birthday (by his father).”
in utterances (21) if put in contrast with “livre” in French utterance (20) and
“książka” Polish (19).
(21) Pol.
Samochód Janek
n mas sg acc n mas sg nom
car Janek
dostał na Boże Narodzenie.
V 3pers sg mas past act prep+n neu sg acc
got for Christmas
Lit. “The car, Janek got it for Christmas.”
(22) Fr. La voiture, Janek l’a eue pour Noël.
Lit. “The car, Janek got it for Christmas.”
Translators must be aware of the different respective uses of subject and topic as
text coherence devices, in order to render not only the semantic content of the
original text but also its pragmatic force.
In the next section, we will briefly consider the problems of explicit and non-
explicit subjects, of impersonal utterances and the relationship between the sub-
ject and the topic.
Hélène Włodarczyk
In structural linguistics, (e.g. Martinet 1962, 1967), subject is defined as the obliga-
tory NP that cannot be deleted from an utterance without resulting in agram-
maticality. In our view, this criterion is difficult to apply because of the frequent
ellipsis that may affect any constituent of an utterance which bears the old meta-
informative status in a coherent text. For this reason, we prefer to put forward the
morphological factor. Subject is explicit when it is marked by a morpheme in the
verb ending, regardless of which semantic role it refers to, that is to say that subject
may be explicit not only in Indo-European nominative languages but also in erga-
tive languages (provided that we deal with a language of this group exhibiting verb
morphemes pointing at the participants of the speech-act).
However, in languages like contemporary English or French, in which the per-
sonal inflection of the verb is not clearly distinctive (neither in spoken nor written
language), the subject must be made explicit by the use of a personal pronoun. As
a matter of fact, in contemporary English, only the 3rd person singular form of
verbs in the present tense is distinct from all other persons in both oral and writ-
ten language.
(26) I | you | we | you | they think [as opposed to] he thinks.
. In some languages verbs also have personal affixes for expressing not only subject but
also objects with the speech act participants, e.g. in Basque “there is agreement between the
case-markers of the NPs and the personal affixes in these three cases [the absolutive, the erga-
tive and the dative case]”, Brettschneider G. “Typological Characteristics of Basque” in Plank
1979: 372.
Attention-centered information in language
In oral contemporary French, in which 1st person plural form “nous pensons” is
frequently replaced by “on pense”, the only personal distinction concerns the 2nd
person plural form:
(28) Fr. Je/tu/il/on/ils pense(s) {pãs} [as opposed to] vous pensez {pãsε}.
In languages in which verbs have distinct morphemes for each person, the pres-
ence of the subject in the utterance may be marked only by the personal form of
the verb. This is the case in many Indo-European languages like Latin, Spanish and
Italian. In Polish15 where the personal morphemes of the verb are distinctive in all
moods and tenses, the first and second person pronoun in the nominative form
do not appear in the utterance unless the speaker intends to put an emphasis on
them, i.e. until s/he treats them not as a simple subject but either as a topicalised
or focalised subject (cf. (29) to (32)).
Pol. (from a novel by Maria Nurowska, Panny i wdowy,
(29)
Warszawa 1992, p.117)
Można sie było tego spodziewać.
Ø AS -hum V past 3pers neu sg pro neu sg gen V inf.
[it] possible was this to foresee
Ty zawsze na siebie
pro 2pers sg nom Top adv prep+refl pro Ø pro 2pers sg nom
you always on yourself [you]
bierzesz winę.
V 2pers sg pres n fem sg acc
take [the] fault
“It was easy to foresee. It’s always you who take the fault on yourself.”
(30) Pol. (from a novel by M.Musierowicz, Pulpecja, Signum, Kraków 1993, p.133)
Coś mi tu nie
Indef pro neu sg nom pro 1pers dat adv neg
something for me here does not
pasuje w tym pokoju.
V pres 3pers sg g prep det n mas sg loc
fit in this room
“— There is something I don’t like in this room. Look at the curtains, how
they hang.”
. In this respect, Polish differs from Russian where personal morphemes have disappeared
in past tense, thus speakers have to use personal pronouns for marking subjects. Compare the
Polish verb paradigm: widziałem, widziałeś, widział with the Russian one: я видел, ты видел,
он видел (I saw, you saw, he saw).
Hélène Włodarczyk
To nie ja je tak
MI Part neg pro 1pers 3pers pro pl acc adv
it is not me them this way
powiesiłam tylko Julia.
V past 1pers sg part n sg nom
hung only Julia
“— It isn’t I/me16 who hung them but Julia.”
The same usage occurs in Italian where the personal pronoun is used only when it
is necessary to treat the subject as topic or focus.
(31) It. Pago …
V pres 1pers sg
“I pay…”
(32) It. Io pago …
pro 1pers sg V pres 1pers sg
“It’s me who pays …”
As a matter of fact, many languages exhibit (like Italian) two different series of
stressed and unstressed forms of personal pronouns in all cases of the declension
(see Chapters 12, 13 & 14 in this volume). Even in English or French where declen-
sion has almost entirely disappeared in nouns, there still exist different forms for
subject or object pronouns: I/me, he/him etc. Moreover, stressed forms sometimes
differ from the unstressed one; e.g.: in French moi/je, toi/tu etc. Stressed forms
are used in order to topicalise or focalise pronouns.17 In languages like Italian or
Polish, the unstressed nominative 1st and 2nd person pronouns are simply omit-
ted (this kind of ellipsis can be considered as a “zero form” of pronouns). The pos-
sibility of omitting the personal pronoun which points to the subject function in
languages with personal morphemes in verbs is considered a universal device (e.g.
Encyklopedia 1999: 414). This omission also takes place in typologically different
languages and concerns not only subject pronouns but also object pronouns in
languages with two-fold (subject and object) personal morphemes in verbs (for
example in Basque). It is necessary to underline that the absence of the subject
in languages with verb personal endings should not be interpreted as a case of
implicit subject but rather as the use of the unstressed (zero) form of the 1st and
. It seems that, from the MIC perspective, the conditions of use of either I or me as the
form of the topicalised 1st person pronoun in English should be reconsidered.
. Out of context and in a written text (without intonation markers) it is not easy to
interpret a stressed pronoun as a topic or a focus.
Attention-centered information in language
2nd person pronoun in the nominative case. We treat in a radically different way
the absence of subject in languages with no personal inflection (see below: implicit
subject).
From a diachronic perspective, it is worth adding that, concerning personal
inflection, modern personal morphemes of Indo-European verbs are former per-
sonal pronouns that were agglutinated to the verb radical.18
In the situation expressed by the verb to eat (utterances 33, 34), the object noun
fish, referring to a passive participant, implies that animate being(s)19 are involved
in the situation as active agent(s). But it is also possible to express the subject
explicitly as in utterance (35).
. In a more recent time, in the history of Slavic languages, Polish reinforced the plural first
person morpheme of the verb by adding -y to the -m inherited from Proto-slavic, e.g. Russian
идём (let’s go | we go), pol. idziemy (we go). Thus, the morpheme -my in modern standard
Polish is completely homonymous with the pronoun my (we).
. In Japanese the distinction between singular and plural (excepts for the plural form -tachi
(達) which is used almost exclusively for humans) is not marked by grammatical morphemes.
Hélène Włodarczyk
In Japanese, although there can be no agreement between the verb and the
subject, the honorific morpheme of the verb is, however, chosen in agreement
with the subject,20 be it only implicit or explicitly expressed in the utterance. In
Example (36), it is possible to infer from the honorific form that the one who has
arrived is a person to whom the speaker owes respect.
(36) Jap. 鈴木先生がいらっしゃいました。
(Suzuki sensei ga irasshaimashita.)
Subject Predicate
Suzuki sensei ga irasshaimashita.
proper Noun professor nom. part V past polite deferent
professor Suzuki has arrived
“Professor Suzuki arrived.”
Hereafter is a short dialog in Japanese with its English translation. None of the
following Japanese utterances has a subject neither is there any personal form of
the verb which would indicate the subject but in translation subject pronouns in
English or personal verb forms in Polish must be added.
. Verb appreciative honorifics agree with the subject [Włodarczyk A. 1996].
Attention-centered information in language
As the translations of the above examples suggest, Japanese utterances where the
implicit subject has the same referent as the explicit topic, can often be trans-
lated into English by a basic utterance with a subject, (utterances 39, 40, 41),
instead of the literal translation with a topic. In other cases, the topic remains
present in the English translation alongside the co-referring subject or object
(utterances 42 and 43). But the choice of an utterance with or without a topic
in the translation from Japanese into a language with an obligatorily explicit
subject ultimately depends on the text coherence (on the place of a given utter-
ance within the whole text), and this cannot be shown on the basis of isolated
utterances taken out of context. Further on, translation with or without a topic
depends also on stylistic conventions which vary from language to language (see
§ 4.4. below).
On the other hand, it is necessary to stress the fact that in Japanese, notwith-
standing a widespread opinion among linguists, the particle wa is not exclusively
a marker of topic. Consequently, in Japanese discourse, topic-comment utterances
are not as frequent as occurrences of wa. As a matter of fact, the particle wa is also
used to mark the subject bearing the old meta-informative status matching a predi-
cate of the same meta-informative status (cf. Włodarczyk A. 2005). In base utter-
ances expressing general laws or observations, the wa particle can also be used as a
marker of generic, general, potential or habitual subject (Włodarczyk A. 1980, 1996,
1998). In utterance (44) below, pronounced without a pause after the wa particle,
both sora (sky) and aoi (is blue) bear the old meta-informative status motivated by
the situation type, i. e. the whole situation spoken about is considered as general.
Utterance (44) thus means “it is the characteristic of the sky to be blue” and not
“[today] the sky is blue”.
Attention-centered information in language
The wa particle occurs in entirely old base utterances in which the old meta-
informative status of subject and of the verb is motivated by the generic o
ntological
status.
into subject- and topic-prominent types (Li & Thompson 1976). In European gram-
mars, from traditional up to more recent ones, topicalisation and focalisation were
either not paid any attention at all (when marked only by prosody or word-order),
or considered as typical of colloquial speech when marked by dislocation and cleav-
ing (Blanche-Benveniste et al. 1990; Miller & Fernandez-Vest 2006). Thanks to the
possibility of exploring large corpuses it will be possible to reappraise ready-made
ideas in linguistics according to which in languages like French and English cleav-
ing and dislocation (markers of extended utterances in RWO languages) belong
only to colloquial speech. This view takes its origin in the syntax of European lan-
guages, itself rooted in classical (Greek and Latin) grammatical tradition which is
focused on base utterances, namely on the subject-predicate (or NP+VP) structure,
leaving aside extended utterances. The classical European grammatical tradition
was continued in syntactic studies during the 20th century (up to Noam Chomsky’s
generative syntax), therefore most syntactic theories still lack concepts and terms to
deal with extended utterances. In the MIC theory, subject and topic structures are
both meta-informative: the subject-predicate structure belongs to the first meta-
informative level whereas the topic-comment belongs to the second one. The prob-
lem will have to be thoroughly explored on large corpora and it will be necessary
to observe whether one or the other of two types of utterances is really (i.e.: with
statistical proof) more frequently used in some languages than in others. To be of
some evidential value such exploration must be conducted both on original utter-
ances in their discourse context and with huge parallel corpora. In order to achieve
this task, the MIC theory offers reliable formalised criteria for identifying topics
and distinguishing simple subjects from topicalised ones.
. For more arguments in favour of the anonymous subject hypothesis, especially in Russian
and Polish, please see Włodarczyk, H. 1994, 1996. That hypothesis was elaborated and tested
on a computer using Prolog. This program contains rules generating and analysing different
kinds of ‘impersonal’ utterances in Polish and Russian, these rules being homologous to the
ones which generate utterances with a subject.
Attention-centered information in language
displays distinctive p
ersonal morphemes (e.g. Slavonic languages, but also some
Romance languages), those morphemes are the only mark of the presence of this
anonymous subject in the utterance, as is the case in Polish (46) and Latin (47)
utterances.
(46) Pol. Pada.
(47) Lat. Pluit.
“It is raining.”
Although German verbs have personal morphemes in the present tense, a neutral
pronoun must be used to mark the anonymous subject.
(50) Ger. Es regnet.
Evidence for the hypothesis that in languages with explicit subjects, impersonal
utterances do entail an anonymous subject is corroborated by the existence of
two distinct semantic types of anonymous subjects: one refers semantically to
the set of animate non-human entities (–hum), the other one to human beings
(+hum).23 This distinction can be observed in many European languages, for
instance: English, French, German, Russian, Polish. As a matter of fact, besides
the -hum anonymous subject (like the one in utterances 46–50) above which is
used in order to refer to -hum entities as, for instance, atmospheric phenom-
ena, there exists another type of anonymous subject referring to human beings.
Furthermore, within the category of human anonymous subjects, languages
(54) Pol.
Nigdy nie wiesz, co może się zdażyć.
adv Ø 2pers G neg V 3pers sg G rel pro 3pers V 3pers V inf
sg +hum +hum +spk pres neu sg nom sg G
Never [you] do not know what may happen
On the other hand, the speaker may insist on the fact that she herself (he himself)
does not belong to the mentioned group (cf. Polish utterances 55–56 below).
(55) Pol.
Zatrzymano go na drodze.
Ø3pers neu V 3pers sg neu pro 3pers masc prep + N fem
sg +hum +hum -spk sg acc sg loc
[someone] stopped him on [the] road
“He was stopped (by the police) on the road.”
. It is perhaps worth mentioning that in English, as in most European languages, the
masculine personal pronoun he was until recently (Halliday & Hasan 1976), the unmarked
gender form for the generic human being, and it was therefore not necessary to use the
unpronounceable form s/he.
Attention-centered information in language
(56) Pol.
Zatrzymali go na drodze.
Ø3pers pl V 3pers pl pro 3pers prep + N
+hum +hum -spk masc sg acc fem sg loc
[They] stopped him on [the] road
Lit. “He was stopped (by the police) on the road.”
Utterances with +hum anonymous subject in Slavic languages, like Polish utter-
ance (57) can often be rendered into English or French by a passive utterance
in which the participant playing the active role remains implicit as soon as
the object of the passive (by whom) is not expressed: “He was stopped (by the
police) on the road.” Moreover, as has already been pointed out for Polish, lan-
guages may use several different types of anonymous subject referring either
to a generic human being or to a more limited group of persons: the choice
between them is determined by semantic nuances as well as pragmatic and sty-
listic conditions.25
Only when taking into account the semantic distinction between the anony-
mous +hum and –hum subject, is it possible to understand the communicative
purposes for which impersonal utterances are used. Such utterances allow us to
identify the global CA very generally, merely as a participant (of the situation)
belonging to a class of entities (+hum or -hum) without selecting a particular ele-
ment of one of these classes. As is the case with an explicit subject, the semantic role
of the participant referred to by the anonymous subject may be various and it has
to be reconstructed by the hearer when interpreting the utterance (Włodarczyk H.
1996). Just as the semantic role pointed to by an explicit subject, the one to which
the anonymous subject points, is not expressed in the utterance; it is a part of
covert information.
In languages with explicit subjects, the anonymous subject corresponds to a
linguistic sign with both (a) the form (the “morphonological zero form” inclusive,
e.g. in Polish, Russian, Italian; or the invariant form of an indefinite pronoun as
man, es in German, on, il in French) and (b) the content (it refers to an indefinite
either human or non-human entity). But anonymous subjects differ from indefi-
nite pronouns (such as someone or something) in that they may occur only in the
position of the subject of an utterance (or exist only in the nominative case-form
in languages with declension). The presence of an anonymous subject in an utter-
. About Polish, Russian and French, see Włodarczyk, H. 1996. We distinguish several types of
+hum and –hum AnS in Russian and Polish, for a detailed discussion see Włodarczyk, H. 1994.
Hélène Włodarczyk
“On the other hand, TCs [there Clauses] are close to inverted constructions in
that the information presented in the postposed segments (e.g. the Object or local
CA) is consistently less familiar within the discourse than that presented in the
preposed segments (e.g. the Subject or global CA). The use of TCs is, in this sense,
the result of the speaker’s choice to postpone unfamiliar information and place
it in the local CA position of the utterance for the sake of maintaining discourse
coherence.”(Martinez-Insua 2011: 103–104)
Thus impersonal constructions can be used in the so called strict word-order lan-
guages (like French or English) to postpone until the end of an utterance with an
entirely new status the introduction of a NP with a new meta-informative status.
The French constructions with the pronoun il or the English there constructions
used to express an anonymous subject make it possible to treat the participant
introduced after the verb in both languages as a local CA (i.e. as a forward-looking
centre) although the use of a direct construction with this participant worded as
an explicit subject would imply that the participant is treated in the utterance as
a backward-looking centre, relying on information previously introduced into the
discourse.
To return to the problem of the anonymous subject, let us stress that there
is in fact no significant difference between those anonymous subjects which
have a “morpho-phonological zero form” and those which have the form of an
Attention-centered information in language
indefinite pronoun occuring in the nominative case only. On the contrary, the
semantic distinction between +hum and –hum anonymous subjects is attested
cross-linguistically regardless of their form (either a morpheme or a zero form).
Necessarily, however, impersonal utterances with an anonymous subject occur
only in languages with an explicit subject. In such languages however, utterances
with an anonymous subject must not be confused with utterances in which, due
to the text coherence, the subject undergoes ellipsis, e.g. utterance (60) used in
answer to (59).
It is worth adding that in languages with an implicit subject (like Japanese) the
concept of the anonymous subject is simply absurd since there is no need (gram-
matical obligation) to replace the subject by some abstraction.
. The only exception we admit for Russian and Polish concerns genitive NPs (with a parti-
tive meaning), in fact this genitive is an ‘adnumeral’ form of the nominative or accusative cases
with numerals or with negation. For details see Włodarczyk, H. – 1992, 1993, and 1996. The
concept of ‘adnumeral’ or “second” genitive was introduced into French linguistic studies on
slavic languages by Menantaud 2008.
Hélène Włodarczyk
cases not as objects but as subjects is motivated by the fact that such NPs are very
often topicalised (“fronting” in Fisher’s terms) and that topic is taken for a subject
because it shares with the subject the property of being a global CA.
Let us compare the following Polish utterances hereafter: (60) with a subject,
(62) with an anonymous subject and (63) with both an anonymous subject and a
topicalised object.
(61) Pol.
Śnieg zasypał drogi.
N sg masc nom V 3pers sg past N pl fem pl acc
snow covered roads
“Snow covered the roads.”
(62) Pol.
Śniegiem zasypało drogi.
N sg masc instr Ø 3pers sg neu -hum V 3pers sg past N pl fem. pl acc.
[by] snow [it] covered roads
“The roads have been covered by snow.”
In (62), scholars admitting the existence of “oblique subjects” consider that the
noun snow (śniegiem) may be considered as an “instrumental subject” due to its
semantic quasi-active role and probably because of its being topicalised. In the MIC
framework however, we claim that the noun śniegiem is not a subject but a topi-
calised (indirect) object, and that the subject has the zero form of a -hum anony-
mous subject (expressed by “it” in English).27 Notice that though, in utterance (63),
it is the direct object NP drogi (roads) corresponding to a semantic passive role
which is topicalised, this NP cannot be interpreted as an “oblique subject” either.
(63) Pol.
Drogi zasypało śniegiem.
N fem pl acc Ø 3pers neu sg -hum V 3pers sg past N sg mas instr
roads [it] covered [by] snow
“The roads have been covered by snow.”
. Polański (2003) defends the same point of view, considering that NPs in oblique cases
appearing in Polish impersonal utterances should not be treated as subjects.
Attention-centered information in language
case) cannot be interpreted as the so-called “oblique subject”. For this reason the
concept “oblique subject” looses its theoretical validity.
languages (e.g. Germanic, Slavic languages, Turkish etc.) The latter are considered
to have different degrees of word-order freedom. It should be added that, even in
FWO languages there exists a canonical WO of the base utterance. Any change
with respect to this canonical WO produces a pragmatic effect (among which are
topicalisation and focalisation),28 consequently the “freedom” of FWO languages
is not arbitrary.
In the MIC framework, both so called RWO and FWO languages can be
dealt with in a consistent way. According to this theory, in both types of lan-
guages word-order is a meta-informative marker used to highlight attention-
driven phrases and to distinguish between old and new information. In order to
determine the canonical WO in a language we must only take base utterances
into consideration. In French and English, WO is used mainly to distinguish
between the subject and the object of a base utterance. In order to extend an
utterance, French or English speakers much refer to explicit segmental markers
(adverbs or particles) which lead to important changes in the syntactic structure
of the utterance: cleft sentences, dislocation, etc. In FWO (for example in Slavic)
languages morphological case markers of subject and object make it possible to
use the word order on both (1) the first meta-informative level – in order to dis-
tinguish entirely new from entirely old base-utterances and (2) the second meta-
informative level where the mere dislocation of the subject or object can induce
its topicalisation or focalisation (Włodarczyk H. 2009). However, even in this
type of languages, the change in word-order often depends on explicit segmental
markers (particles, such as “to” (lit. “this”) in Polish and Russian).
Thus, in the MIC approach, we replace the typological distinction between
RWO and FWO languages by the distinction between languages in which the
word order concerns mainly the first meta-informative level, i. e. the base utter-
ance (like in French and English), and those in which the word order is used
mainly as a marker of the second meta-informative level (like in Slavic languages).
But this classification must be understood as a weighed heterarchy because there
is probably no language in which word order is used exclusively for marking the
meta-informative status in base utterances. In addition, let us emphasize the fact
that besides word order, intonation is the most important marker of the second
meta-informative level (it is a favourite expressive device for marking the topic or
focus). Since, however, it may be sometimes difficult to perceive intonation (even
in oral communication) and since intonation is not sufficiently marked in most
writing systems, most languages use additional, explicit meta-informative m arkers
. For the MIC approach to these problems in Polish and Russian see Włodarczyk, H. 2009.
Attention-centered information in language
of the results of topicalisation and focalisation: these are mostly particles and spe-
cial syntactic constructions.
6. Conclusion
Acknowledgement
Abreviations
/ or lat. Latin
1pers first person loc locative
2pers second person mas masculine
3pers third person mi meta-informative
acc accusative n noun
act active voice neg negation
adj adjectif neu neuter
adp attention-driven phrase nom nominative
art article np noun phrase
ca centre of attention num numeral
conj conjunction part particle
dat dative pas passive voice
det deictic pf perfective
en. English pl plural
fem feminine pol. Polish
fr. French pp prepositional phrase
g gender as a variable (fem, mas or neu) prep preposition
ge. German pres present
gen genitive pret preterite
±hum human/non-human pro pronoun
imprs impersonal refl reflexive
inf infinitive rel relative (pronoun)
instr instrumental sg singular
ip imperfective ±spk speaker involved or not
it. Italian v verb
jap. Japanese
References
Blanche-Benveniste, Claire, Bilger, Mireille, Rouget, Christine & van den Eynde, Karel. 1990. Le
français parlé: Études grammaticales. Paris: Éditions du CNRS.
Brettschneider, Günter. 1979. Typological characteristics of Basque. In Ergativity: Towards a
Theory of Grammatical Relations, Frans Plank (ed.), 371–384. London: Academic Press.
Bunting, Michael F., Cowan, Nelson & Colflesh, Gregory J. H.. 2008. The deployment of atten-
tion in short-term memory tasks: Tradeoffs between immediate and delayed deployment.
Memory & Cognition 36(4): 799–812.
Chafe, Wallace L. 1994. Discourse, Consciousness and Time. Chicago IL: The University of
Chicago Press.
Corbetta, M. 1998. Frontoparietal cortical networks for directing attention and the eye to visual
locations: identical, independent, or overlapping neural systems. Proc. Natl. Acad. Sci. USA
95, 831–838.
Cowan, Nelson. 2008. What are the differences between long-term, short-term, and working mem-
ory? In Progress in Brain Research, Vol. 169: Essence of Memory, Wayne S. Sossin, Jean-Claude
Lacaille, Vincent F. Castellucci & Sylvie Belleville (eds), 323–338. Amsterdam: Elsevier.
Attention-centered information in language
Dryer, Matthew S. 2007. Word order. In Clause Structure, Language Typology and Syntactic
Description, Vol. 1, 2nd edn, Timothy Shopen (ed.). Cambridge: CUP.
Encyklopedia językoznawstwa ogólnego, 2nd edn. 1999. Kazimierz Polański (ed.), Ossolineum,
Wrocław – Warszawa – Kraków.
Fischer, Susann. 2004. The diachronic relationship between quirky subjects and stylistic front-
ing. In Non-nominative Subjects [Typological Studies in Language 60], Peri Bhaskararao &
Karumuri V. Subbarao (eds), 193–212. Amsterdam: John Benjamins.
Gaatone, David. 1998. Le Passif en Français. Paris: Editions Duculot, Champs Linguistiques.
Givón, T. 1988. The pragmatics of word order: Predictability, importance and attention. In Stud-
ies in Syntactic Typology [Typological Studies in Language 17], Michael Hammond, Edith
A. Moravcsik & Jessica Wirth (eds). Amsterdam: John Benjamins.
Givón, T. 1988. 1994. The pragmatics of voice: Functional and typological aspects. Voice and
Inversion [Typological Studies in Language 28], T. Givón (ed.), 3–44. Amsterdam: John
Benjamins.
Givón, T. 2001[1984]). Syntax: An Introduction, Vol.1. Amsterdam: John Benjamins.
Grice, H. Paul 1975. Logic and conversation. In Syntax and Semantics, Vol 3, Peter Cole & Jerry
L. Morgan (eds). New York NY: Academic Press.
Grosz, Barbara J. & Sidner, Candace L. 1986. Attention, intentions, and the structure of d
iscourse.
Computational Linguistics 12(3): 175–204.
Grosz Barbara J., Joshi, Arawind K & Weinstein, Scott. 1995. Centering: A framework for
modeling the local coherence of discourse. Computational Linguistics 21(2): 203–226.
Grzegorczykowa, Renata; Laskowski, Roman & Wróbel Henryk. 1998. Gramatyka współczesnego
języka polskiego, Morfologia, wyd. 2 zmienione, T. 1–2. Warszawa: P.W.N.
Halliday, Michael A.K. & Hasan, Ruqaiya. 1976. Cohesion in English. London: Longman.
Halliday, Michael A.K & Greaves, William S. 2008. Intonation in the Grammar of English.
London: Equinox.
Haspelmath, Martin. 2001. Non-canonical marking of core arguments in European languages.
In Non-Canonical Marking of Subjects and Objects [Typological Studies in Language
46], Alexandra Y. Aikhenvald, Robert M.W. Dixon & Masayoki Onishi (eds), 53–83.
Amsterdam: John Benjamins.
Hayashi, A., Tomlin, Russell S. & Yokota, T. Attention detection and Japanese (Japanese Fish
Film Paper) 〈http://logos.uoregon.edu/tomlin/in_progress.html〉
Ikegami, Yoshihiko. 2007. Subjectivity, ego-orientation and subject-object merger. In Japa-
nese Linguistics European Chapter, Ikegami Yoshihiko, Viktoria Eschbach-Szabo & André
Włodarczyk (eds). Tokyo: Kurosio.
James, William. 1890. The Principles of Psychology, 2 vols. Dover Publications 1950.
Jespersen, Otto. 1924, The Philosophy of Grammar. London: George Allen & Unwin.
Kuroda, Shigeyuki. 1965. Subject, in Japanese Generative Grammar – Syntax and Semantics,
Vol. 5, Masayoshi Shibatani (ed.), 1–16. New York NY: Academic Press.
LaBerge, David. 2003. Attention, awareness and the triangular Circuit. In Essential Sources in the
Scientific Study of Consciousness, Bernard J. Baars, William P. Banks & James B. Newman
(eds). Cambridge MA: The MIT Press, Bradford Book.
Lambrecht, Knud. 1994. Information Structure and Sentence Form. Topic, Focus, and the Mental
Representations of the Discourse Referents. Cambridge: CUP.
Langacker, Ronald W. 1987. Foundations of Cognitive Grammar, Vol. I: Theoretical Prerequisites.
Stanford CA: Stanford University Press.
Langacker, Ronald W. 1991. Foundations of Cognitive Grammar, Vol. II: Descriptive Application.
Stanford CA: Stanford University Press.
Hélène Włodarczyk
Li, Charles N. & Thompson, Sandra A. 1976. Subject and topic: A new typology of language. In
Subject and Topic, Charles N. Li (ed.), 457–61. New York NY: Academic Press.
Martinet, André. 1962, A Functional View of Language. Oxford: Clarendon. (In French: Langue
et fonction, Denoël, Paris 1969).
Martinet, André. 1967, Éléments de linguistique générale. Paris: Armand Colin.
Martínez-Insua, Ana E. 2011. When discourse matches syntax: On meta-informative centering
theory and discourse coherence in the recent history of English. International Journal of
English Studies (IJES) 11(2): 91–111.
Menantaud, Henri. 2008. Génitif premier et génitif second: Faut-il diviser le génitif polonais
en “sous-cas”? L’Enseignement du polonais en France: 2èmes Assises, 49–54. Paris: Institut
d’Études Slaves.
Miladi, Lidia. 2006. Topicalisation et focalisation des complétives et/ou des infinitives dans les
constructions verbales du polonaise. Actes du colloque international ‘Constructions verbales
et production de sens’ [Collection «Recherches en linguistique étrangère], 224–241. Besan-
çon: Presses universitaires de Franche-Comté.
Miladi, Lidia. 2009, Les fonctions discursives de la particule énonciative to du polonais dans
les constructions à segment détaché à gauche. Revue des Etudes Slaves, numéro spécial:
La cohérence du discours dans les langues slaves: Linguistique théorique et textuelle,
LXXX(1–2): 87–103.
Miller, Jim & Fernandez-Vest, Jocelyne. 2006. Spoken and written language. In Eurotype: Typol-
ogy of Languages in Europe Pragmatic Organisation of Discourse in Languages in Europe,
Giuliano Bernini & Maria Schwartz (eds), 9–64. Berlin: Mouton de Gruyter.
Mohr, Sabine. 2005. Clausal Architecture and Subject Positions: Impersonal Constructions in
the Germanic Languages [Linguistik Aktuell/Linguistics Today 88]. Amsterdam: John
Benjamins.
Morel, Marie-Annick & Laurent, Danon-Boileau. 1998. Grammaire de l’intonation, L’exemple du
français. Paris: Faits de Langues, Ophrys.
Myachykov, Andriy, Garrod, Simon & Scheepers, Christoph. 2009. Attention and syntax in sen-
tence production: A critical review. Discours, Revue de linguistique, psycholinguistique et
informatique 4: Linearization and Segmentation in Discourse. 〈http://discours.revues.org/
index7594.html〉
Oakley, Todd, 2009. From Attention to Meaning. Explorations in Semiotics, Linguistics, and
Rhetoric. Frankfurt: Peter Lang.
Polański, Kazimierz (ed.). 1999 (wydanie 2). Encyklopedia językoznawstwa ogólnego. Wrocław.
Warszawa. Kraków. Zakład Narodowy im. Ossolińskich.
Polański, Kazimierz. 2003. Na marginesie dyskusji o podmiocie. Etudes linguistiques romano-
slaves offertes à Stanisław Karolak, 409–416. Kraków: Oficyna Wydawnicza Edukacja.
Paul, Hermann. 1880[1995]. Prinzipien der Sprachgeschichte, 10th edn. Tübingen: Niemeyer.
Shelton, Jill T., Elliott, Emily M. & Cowan, Nelson. 2008. Attention and working memory: Tools
for understanding consciousness. Psyche, 14(1).
Talmy, L. 2000. Toward a Cognitive Semantics. Cambridge, MA: The MIT Press.
Tomlin, Russell S. 1995. Focal attention, voice, and word order: An experimental, cross-
linguistic study. In Word Order in Discourse [Typological Studies in Language 30], Pamela
A. Downing & Michael Noonan (eds), 517–554. Amsterdam: John Benjamins.
Tomlin, Russell S. 1997. Mapping conceptual representations into linguistic representations:
The role of attention in grammar. In Language and Conceptualization, Jan Nuyts & Eric
Pederson (eds), 162–189. Cambridge: CUP.
Attention-centered information in language
Włodarczyk, Hélène.1994. Les phrases à sujet anonyme et verbe réfléchi en polonais et en russe,
Organisation sémantique de la phrase dans les langues slaves, Revue des Études Slaves T.
LXVI (3): 669–680, Paris: Institut d’Études Slaves.
Włodarczyk, Hélène. 1996. Les traits sémantiques du sujet “anonyme” en polonais, russe et fran-
çais, Semantyka a konfrontacja językowa: 179–198. Warszawa: SOW, PAN.
Włodarczyk, Hélène. 1998. Wykładniki wartości informacyjnej wypowiedzenia w j. polskim i
francuskim (aspekt, okreslonosc, modalnosc), Congrès des Slavistes Cracovie 1998, Revue
des Etudes Slaves T. LXX(1): 53–66. Paris: Institut d’Études Slaves.
Włodarczyk, Hélène. 1999. Les marqueurs de la validation des énoncés en français et polonaise.
Études cognitives/Studia kognitywne 3: 135–162. Warszawa: SOW, PAN. (Erratum: dans cet
article, il faut lire S pour P et P pour S).
Włodarczyk, Hélène. 2003. L’Interprétation dynamique des centres d’intérêt dans les dialogues
français et polonais. In Études linguistiques romano-slaves offertes à Stanisław Karolak,
511–528. Kraków: Oficyna Wydawnicza Edukacja.
Włodarczyk, Hélène. 2004. Les centres d’intérêt de l’énoncé en polonais et en français. In Énon-
cer, l’ordre informatif dans les langues, Pierre Cotte, Martine Dalmas & Hélène Włodarczyk
(eds), 33–48. Paris: L’Harmattan, Coll. Sémantiques.
Włodarczyk, Hélène. 2006. Struktura meta-informacyjna wypowiedzenia w języku polskim, III
Kongres Polonistyki Zagranicznej, Poznan 2006; Postcriptum 2006/2 Nr 52: 181–185. Kato-
wice: Szkoła Języka i Kultury Polskiej Uniwersytetu Śląskiego.
Włodarczyk, Hélène. 2008. Du bon (et du mauvais) usage des formes accentuées et atones
des pronoms personnels en polonaise. L’Enseignement du polonais en France, IIes Assises:
39–48. Société française d’Etudes Polonaises. Paris: Institut d’Études Slaves.
Włodarczyk, Hélène. 2009. Les centres d’intérêt de l’énoncé et la cohérence textuelle en polonais
et russe. Revue des études slaves LXXX(1–2): 13–32. Paris: Institut d’Études Slaves.
Włodarczyk, Hélène. 2012. L’emploi des pronoms personnels en polonais – par contraste avec le
russe et le français. In Études à la mémoire de J. Breuillard sous la dir. de S. Viellard, Revue
des Études Slaves, t. LXXXIII(2–3): 617–648. Paris: Institut d’Études Slaves.
part 2
Neuropsychological evidence
for the MIC theory
Semantic and episodic memory by reference
to the ontological grounding of the old
and new meta-informative status
Franz J. Stachowiak
Charles Sturt University Albury, NSW, Australia
The present paper argues for a model of language production and comprehension
in which a MIC-component interacts with working memory, episodic memory,
and semantic memory as resources for the meta-informative processing of
new and old information. This component is also interconnected with the
grammatical encoding system which produces “surface” structures. Focused
attention plays a decisive functional role in this system as it directs the processes
which transform intentions into messages. Although MIC theory is essentially
a linguistic theory it offers itself to direct mapping with notions from cognitive
and clinical neuropsychology. Levelt (1989) dedicates pioneering work to the
attention resources, the focus of the message and the macro-planning of the
information in the early stages of speech production, however, his model remains
vague with respect to the formal status of the cognitive systems involved. MIC
theory complemented by findings from cognitive neuropsychology can fill these
gaps. Despite ample evidence for the relative independence of the different
memory systems (e.g. double dissociations between episodic and semantic
memory; particular vulnerability of episodic memory in medial temporal lobe
impairment and diseases such as Alzheimer’s), recent research accentuates the
interaction between these systems in building up generic “semantic” knowledge
from autobiographical experience and vice versa facilitating the retrieval of
information from episodic memory by contributing generic information from
semantic memory. Input of information into the MIC-component from episodic
memory is more likely to be attributed “new status” than inputs from semantic
memory, although attention processes can redirect the meta-informative status
of that information at any time depending on outcomes from ToM (Theory of
Mind), which is attributed an active role in this model. ToM is the capacity of
humans to understand mental states of others including the ability to judge the
contents of their respective memory systems. In verbal interaction estimates of
the communication partners’ memory systems’ contents form an input to the
MIC component of the speaker who selects CAs. This ability is seen in early
childhood and understood as a key factor in Tomasello’s (2000) social-pragmatic
theory of word learning and language acquisition. The model proposed here will
Franz J. Stachowiak
1. Introduction
The present paper examines the role of semantic and episodic memories as
ontological source or grounding of old vs. new information. It starts from the
assumption that these memory systems (in connection with the working memory
system) are continuously being addressed (storage and retrieval processes) dur-
ing the production and perception of utterances. It needs to be noted, however,
that the newness or oldness of information is dependent on the speaker’s wish (or
decision or compulsion) which meta-informative status to attribute to it. Gen-
eral public knowledge of a name like “Albert Einstein”, which would “normally”
be considered as old information as it stands for a famous figure in science, can
nevertheless acquire a meta-informative status as “new”. To use an example from
A. Włodarczyk (Chapter 3 in this volume): “In the following utterance: It is Albert
Einstein who discovered the law of Relativity, the subject Albert Einstein is focalised
as a new chunk of information contrasting with what is treated as the old infor-
mation: there is a person who discovered the law of Relativity (supposedly although
you do know that this law has been discovered, what you do not know is the name
of that physicist) and I am telling you: that person’s name is Albert Einstein.” This
oscillation between new and old information and especially the drawing of atten-
tion to something as new on the background of known information is reminiscent
of the classical distinction of Gestalt-Psychology between figure and ground. The
focalization on Albert Einstein in the sentence above makes only sense in the con-
text of someone for whom this information is new. The speaker assumes that the
addressee has attributed this fact to another person, for instance to “Otto Hahn”
or “Madame Curie” or does not know at all, who was the discoverer of Relativ-
ity. In a TV quiz with multiple choice questions, in which the three names are
given as possible solutions to the question “Who discovered Relativity?” this sen-
tence would be understood as the solution to a trivial question, either confirming
a correct guess or correcting a wrong answer. In this case “Albert Einstein” would
be given the status of new information although the name had been mentioned
before in that context, namely as one of the possible solutions.
Interestingly, the discovery of Relativity was an event in Albert Einstein’s life,
which must have left strong traces in his and his colleagues’ episodic memories,
but the important fact that it happened became general knowledge, i.e. semantic
knowledge of billions of traditionally educated people. In contrast, most private
events of Einstein’s life are not known to the public. The fact that he got divorced
from his wife Mileva in order to marry his cousin Elsa, cannot be considered as
general knowledge. However, if this information is predicated to his name, new
epistemic knowledge is produced. The MIC theory distinguishes clearly between
two kinds of situations: Σ-situations: generic > general > potential > habitual on
Franz J. Stachowiak
the one hand and Π-situations: specific > particular > actual > occasional on the
other. The events of the divorce and the remarriage belong to the Π-type of situa-
tion. In neuropsychological terms one could speak of information retrieved from
episodic memory, whereas the names and well known facts such as the discovery
of Relativity are said to be stored in semantic memory. However, there is a certain
problem in linking the linguistic approach to the neuropsychological one in our
example. The fact that Einstein married twice may have been part of his own and
his wives’, friends’ and family’s episodic memories but not of other persons’ except
biographers’, historians’ or simply curious people who find this information in the
internet. When distinguishing between old and new information we make refer-
ence not only to our own memory, but also to what we assume to be stored in the
memories of our communication partners.
This in mind, let us first examine what the nature and structure of episodic
vs. semantic memories is and how they interact in forming our autobiographi-
cal memories, world knowledge or encyclopaedic knowledge (to use some of the
usual terms). Later, on that basis, we will consider how they interact with what
I would like to call the MIC-component of the pragmatic competence of a speaker/
listener.
The term “episodic memory” was introduced by Tulving (1972), who assumed
it to be a late evolutionary development out of semantic memory and unique to
humans, the essence of which “lies in the conjunction of three concepts – self,
autonoetic awareness, and subjectively sensed time” (Tulving 2002: 5). Episodic
memory pertains to the learning, storing and retrieval of autobiographical facts,
i.e. unique personal experiences, to events in our lives, occurrences, careers,
encounters, talks with others, accidents, etc. The question “Where were you and
what did you do when September 11, 2001 happened?” is a question about an
episode in one’s personal life, which might be easy to retrieve as it must have
been connected with world breaking news at the time the incidence happened.
Ethnographic analysis (Reeves, Kuper & Hodges 2008) offers ideas on the orga-
nizational categories which determine the structuring (and restructuring) of
episodic memory:
Following Tulving (2002) these aspects would have to be related to the autobio-
graphic self, but as will be shown later, should also be related to the “self ” of others
from the viewpoint of an observer.
Using MIC terminology, “September 11, 2001” would be a Π-situation.
In contrast: semantic memory, although it is – like episodic memory – a form
of long term declarative (factual) memory, stores knowledge that can be learnt
independently of reference to time and place. According to the classical view it
is not relevant where and when it was learnt. Evolution has given animals trait
memory, which might be an early form of semantic memory in Tulving’s sense, but
humans, of course, build up knowledge from learning. The fact of the discovery of
Relativity can be learnt at school without any personal involvement. “Encyclopae-
dic knowledge of information such as the features of objects (e.g. apples are usu-
ally red), categories (e.g. oranges and bananas are both types of fruit), historical
events, mathematical tables, cognitive maps, and similar types of information are
considered to be stored in semantic memory systems of the brain” (Dickerson &
Eichbaum 2010: 87).
The term semantic memory is not to be interchanged with what “semantics”
stands for in linguistics. The subjective lexicon can certainly be considered as part
of semantic memory. However, the semantic system relevant for language pro-
duction and comprehension is a functionally autonomous system. The meaning
of words, the selectional restrictions for the combining of word meanings, the
phrasal semantics etc. are relevant for the functioning of the language system and
can be disturbed in the case of aphasia without a general loss of semantic memory
or encyclopaedic knowledge. Thus a person’s knowledge about trees in general is
wider than the meaning of the word “tree”, and semantic disturbances in aphasia
compromise primarily the linguistically relevant semantic network (or feature sys-
tem or decision table system (Miller and Johnson-Laird 1976) depending on the
model you want to use) related to the “lemma” (Levelt 1989) of the lexical entry
“tree”, but not the encyclopaedic or generic knowledge about trees in its entirety.
Interestingly, closed class words (function words such as articles, prepositions etc.)
can be severely affected in Broca’s aphasia, i.e. retrieval of them becomes very dif-
ficult in oral speech and reading, whereas open class words do not pose major
difficulties for these patients. Frontal lesions as in the case of Broca’s aphasia hit
function words as they play a major role in the syntactic processing of utterances.
Franz J. Stachowiak
“Thus, many different studies have found that children begin to develop joint
attentional skills at around 9 to 12 months of age, including such things as
following the gaze direction and gestures of adults, imitating adult actions on
objects, and directing adult attention to outside objects using various kinds of
gestural signals (see Tomasello 1995a, for a review). Most children also show their
first signs of comprehending language at this same age, with the first linguistic
productions coming soon after”. (Tomasello 1999: 406)
At the same time these theories offer the solution to the problem that episodic
memory is considered as related to the self and to autobiographical events. When
distinguishing between old and new information we try to match our own episodic
and semantic memories with those of others, and narratives about the actions of
others are based on the original contents of their episodic memories which were
either transmitted to others or extrapolated by communication partners.
6. C
ognitive communication disorders as disturbances
of the MIC-component
There is clinical evidence from diverse syndromes that the system of meta-
informative shaping of an utterance can be disturbed so that communication with
these patients becomes rather cumbersome. Cognitive communication disorders
following from right hemisphere lesions, caused for instance by Traumatic Brain
Injury (TBI), stroke or tumour result from the underlying cognitive deficits caused
by the neurological impairment, i.e. from disorders of attention, memory, organi-
zation, information processing, problem solving, and executive functions. These
difficulties in communicative competence are distinct from aphasia. Phonological,
grammatical or lexical-semantic problems are generally not symptoms of cogni-
tive communication disorders. Impaired attention, difficulties processing com-
plex information, difficulties learning new information, problems distinguishing
between relevant and irrelevant information, inefficient retrieval of stored infor-
mation, problem solving difficulties and inappropriate social behaviour due to lack
of empathy and failing to read others’ minds as well as inflexibility and impulsivity
have an impact on the way these patients communicate. Abusamra, Coré, Joanette
and Ferreres (2009) found that the expression of patients with right hemisphere
damage is often less informative than that of control subjects and that this infor-
mation is organized more simply. Characteristic is “lack of coherence and strong
tendency to tangential discourse” (p. 74) and “what distinguishes patients with
RHD from aphasic patients is their tendency to make inappropriate comments or
to stray from the topic of the story.” (p. 74).
Semantic and episodic memory w.r.t. old and new meta-informative status
The patient obviously refers the question to his own family and seems to approach
a correct description of why he considers his wife a witch, underlining this with
information from his episodic memory, but when asked again he drifts away from
the topic and refers to generic information pertaining to religion etc, which is
semantically related but rather to the literal meaning of “witch” than to the meta-
phorical meaning which the examiner has in mind. In the end the patient takes
only up what has been said in the context and confirms the question (contextual
grounding).
Similarly schizophasia is characterized by a problem to uphold attention to
relevant information, patients often produce chains of associations to what was
said in the context and drift away from the topic. “Among the more noticeable
abnormalities in the speech samples she (Chaika) collected were ‘production of
new sentences according to phonological and semantic features of previously
uttered discourse rather than according to topic’ and ‘failure to monitor own
Franz J. Stachowiak
speech.” (Mitchell & Crow 2005: 968). The following example from Ian Thompson
(2001: see also discourse in neuropsychiatry) shows this typical kind of sheering
off from the centre of attention:
“Parents are people that raise you. Anything that raises you can be a parent. Parents
can be anything, material, vegetable or mineral, that has taught you something.
Parents would be the world of things that are alive, that are there. Rocks, a person
can look at a rock and learn something from it, so that would be a parent.”
As outlined above and in my next chapter (in this volume) clinical evidence shows
the relative functional autonomy of the modules depicted: aphasia can destroy the
grammatical component and the linguistic semantic component, cognitive com-
munication disorders can compromise the MIC-component as a result of atten-
tion and memory problems. Different brain injuries can impair working memory,
episodic memory and semantic memory in rather isolated ways leading to various
dissociations of disturbances. In the intact brain all these systems interact, as mod-
ern imaging technology proves. A research gap needs to be closed by investigating
the relationship between episodic memory and verbal behaviour. How are narra-
tives structured which have to rely on variable contents of episodic and semantic
memory?
An interesting study – however without the view of a combined linguistic and
neuropsychological model as presented here – has been undertaken in cognitive
neuropsychology. In an experiment in which subjects had to recall three different
types of story: a previously unknown story, a well-known fairy tale, and a modi-
fied well-known fairy tale, Dalla Barba, Attali and La Corte (2010) were able to
show that the interference of strongly represented, overlearned information in epi-
sodic memory recall is implicated in the production of confabulations of old age
people. This effect turned out to be particularly prominent when the to-be remem-
bered episodic information showed strong semantic similarities with preexisting,
Semantic and episodic memory w.r.t. old and new meta-informative status
• Act–Single actions
people undertake
• Time–The sequencing
Verbal context| of events that occur
• Goal–Things that
people are trying to
MIC-COMPONENT accomplish
SEMANTIC MEMORY -
Ground
“It is ...
LINGUISTIC SEMANTICS
Lexicon etc
verlearned information. One of the three narratives was a modified version of the
o
well known (overlearned) fairy tale “Little red riding hood”. In this version Little
Red Riding Hood is not eaten by the wolf. The new version was as follows:
As expected old age subjects who had to recall this story produced more con-
fabulations in this modified story than in a completely new everyday story. The
confabulations consisted in saying that “the wolf eats Little Red Riding Hood,”
as it is known from the original version of the story, rather than recalling the
modified version in which Little Red Riding Hood is not eaten by the wolf.”
(ib. P. 658). In the above model the confabulatory recall could be explained as
due to vitiated attention processes related to the distinction between new and
old information. The grounding of the activity of the MIC-component would
be compromised by the uninhibited retrieval of old information, which should
have a structural effect on the distribution of information in the stories as they
were retold by the subject.
Another way of gathering evidence for the relationship between types of
memory and the meta-informative status of the information retrieved from mem-
ory in a certain narrative task has been attempted in a small pilot study conducted
for this presentation: Eight 3rd year students were asked to give a short report
about their memories of the first day at Charles Sturt University.
Task: Please write a report of your first visit to Charles Sturt University.
Describe this event with the details you remember. Describe in a few sentences how
memory influences your verbal report, for instance if you can’t remember details.
The responses were evaluated using the organizing categories of episodic mem-
ory described above (Reeves, Kuper & Hodges 2008). The distribution of infor-
mation related to these categories was classified as new or old (meta-informative
status) and further sub-classified according to sentence initial or final position.
For the sake of consistency and simplicity the same categories were assumed for
semantic-generic information and applied according to the same scheme. The
centres of attention (or the relevant parts of information) in the different sen-
tential positions were identified and cross-classified according to the categories
mentioned (see Figure 3). The following example (Figure 2) shows the attempts
to classify the different phrases. The scheme used (Figure 2) obviously needs
Semantic and episodic memory w.r.t. old and new meta-informative status
Figure 2. Example of student report with classification remarks between lines
This simple count demonstrates that in the specific task, which asked for
information from episodic memory, most of the relevant new information appears
in sentence final position (the expected place for new information). This applies
to all categories used. But it is obvious that the categories “Act – Single Actions
people undertake”, “Events – activities that people carry out” and “Feeling – emo-
tions felt and expressed” relate to information that is expressed by the verb phrase.
The category “Activitiy – a set of related activities that occur” did not apply as no
activities were reported that were not directly related to the main event, the first
day at the University. Interestingly information from all other categories could
appear in this position, but also in sentence initial position, where it seems to get
the status of focused attention. The actor “usually the first person “I”, which was
considered as old information, appeared most frequently in initial subject posi-
tion (26 instances), as can – of course – be expected. Generic information gets the
status of new information in several instances and relates new information to old
one as described above.
These data are, of course, too sparse to arrive at sound conclusions, but they
seem to indicate that the mapping between information stored in episodic and
semantic memory on the one hand and centres of attention on the other as a result
of the meta-informative processing of utterances – is guided by structural prin-
ciples. There is no 1:1 mapping of memory entries into fixed positions in the meta-
informative structuring of utterances, in contrast, there is a lot of flexibility. In
FWO languages such as Jiwarli1 (Austin 2001) any kind of information from the
above categories can be moved into initial position as centre of attention. However
some types of information seem to lend themselves to predilected positions on the
way from intentions into utterances.
References
Abusamra, Valeria, Côté, Claude H., Joanette, Yves & Ferreres, Aldo. 2009. Communication
impairments in patients with right hemisphere damage. Life Span and Disability XII(1):
67–82.
Austin, Peter K. 2001. Word order in a free word order language: The case of Jiwarli. In Forty
Years On: Ken Hale and Australian Languages, Jane Simpson, David Nash, Mary Laughren,
Peter Austin & Barry Alpher (eds), 205–323. Canberra: Pacific Linguistics.
Battaglia, Francesco P. & Pennartz, Cyriel M.A. 2011. The construction of semantic memory:
Grammar-based representations learned from relational episodic information. Frontiers in
Computational Neuroscience 5: 1–22.
Dalla Barba, Gianfranco, Attali, Eve & La Corte, Valentina. 2010. Confabulation in healthy
aging is related to interference of overlearned, semantically similar information on epi-
sodic memory recall. Journal of Clinical and Experimental Neuropsychology 32(6): 655–660.
Dennet, Daniel. 1977. The Intential Stance, 6th edn. Cambridge MA: The MIT press.
Dickerson, Bradford C. & Eichenbaum, Howard. 2010. The episodic memory system: Neurocir-
cuitry and disorders. Neuropsychopharmacology Reviews 35: 86–104.
Gallese, Vittorio. 2001. The “Shared Manifold” hypothesis: From mirror neurons to empathy.
Journal of Consciousness Studies 8(5–7): 33–50.
Greve, Andrea, Van Rossum, Marc C.W. & Donaldson, David I. 2007. Investigating the func-
tional interaction between semantic and episodic memory: Convergent behavioral and
electrophysiological evidence for the role of familiarity. NeuroImage 34: 801–814.
Kensinger, Elizabeth A., Ullman, Michael T. & Corkin, Suzanne. 2001. Bilateral medial temporal
lobe damage does not affect lexical or grammatical processing: Evidence from amnesic
patient H.M. Hippocampus 11: 337–346.
Klein, Stanley B., Cosmides, Leda, Gangi, Cynthia E., Jackson, Betsy, Tooby, John & Costabile,
Kristi A. 2009. Evolution and episodic memory: An analysis and demonstration of a social
function of episodic recollection. Social Cognition 27: 283–319.
Levelt, Willem J.M. 1989. Speaking. From Intention to Articulation. Cambridge MA: The MIT
Press.
Miller, George A. & Johnson-Laird, Philip N. 1976. Language and Perception. Harvard Univer-
sity Press. Cambridge, Mass.
Mitchell, Rachel L.C. & Crow, Tim J. 2005. Right hemisphere language functions and schizo-
phrenia: The forgotten hemisphere? Brain 128: 963–978.
Nuxoll, Andrew M. & Laird, John E. 2007. Extending Cognitive Architecture with Episodic
Memory. In the Proceedings of the 21st National Conference on Artificial Intelligence (AAAI).
Prince, Steven E., Tsukiura, Takashi & Cabeza, Roberto E. 2007. Distinguishing the neural
correlates of episodic memory encoding and semantic memory retrieval. Psychological
Science 18(2): 144–151.
Reeves, Scott, Kuper, Ayelet & Hodges, Brian D. 2008. Qualitative Research Methodologies:
Ethnography. BMJ 337:a1020.
Scoville William B. & Milner, Brenda. 1957. Loss of recent memory after bilateral hippocampal
lesions. Journal of Neurology, Neurosurgery & Psychiatry 20: 11–21.
Thompson, Ian. 2001. Discourse analysis in neuropsychiatry. Communication and Mental
Illness, Jenny France & Sarah Kramer (eds), 393–406. London: Jessica Kingsley Publishers.
Tomasello, M. 1999. The Cultural Origins of Human Cognition, Harvard University Press. Cam-
bridge, Mass.
Tulving, Endel. 1972. Episodic and semantic memory. In Organization of Memory, Endel
Tulving & Wayne Donaldson (eds), 381–403. New York NY: Academic Press.
Tulving, Endel. 2002. Episodic memory: From mind to brain. Annual Review of Psychology
53: 1–25.
Tracing the role of memory and attention for
the meta-informative validation of utterances
Franz J. Stachowiak
Charles Sturt University Albury, NSW, Australia
1. Introduction
In the following I will try to establish a connection between some aspects of the
MIC theory and neuropsychological findings about memory and attention. I will
argue that verbal processing of information is strongly influenced by or even rests
on the capacity and mode of operation of working memory and other types of
memory and is intricately related to attentional processes, which play a role in
directing the interest of a communication partner in spoken or written language.
In fact I assume that the basic slot structure for marking the information as new or
old in verbal utterances derives from operational principles, limits and strategies
(to overcome these limits) of working memory as a gate to long term memory. In
particular I will support the following claim by Hélène and André Włodarczyk
(MIC 2010 conference paper):
We claim that it is impossible to produce a (pragmatically) felicitous utterance
without introducing the information it conveys as either old or new, this
distinction being expressed as much by the prosody and the syntactic structure
of the utterance as by the choice of its grammatical and lexical units. As a matter
of fact, the meta-informative (pragmatic) validation of an utterance as old or new
is different from its informative (semantic) validation as true or false.
Franz J. Stachowiak
In a seminal article, in which they examine the role of memory in complex cogni-
tive tasks, Ericsson and Kintsch (1995) have presented a model of text compre-
hension, in which they assume an interaction of two kinds of processes: the short
term activation of information while text segments are being processed and the
storage and retrieval of information that forms episodic text structures in long
term memory. New information from the text is continually integrated into these
episodic text structures, but relevant parts of them need to remain accessible dur-
ing reading or auditive comprehension. Ericsson and Kintsch claim that the acces-
sible portions of this structure in long term memory serve as extended working
memory, called “long term – working memory”. In order to overcome the limits
of working memory capacity, when massive rapid information flow takes place,
we need to hold only a few relative parts of the information in working memory,
which serve as cues to retrieve the entire information and expand the episodic text
structures. Sometimes texts are not expressed in such a way that understanding
is smooth and easy. Therefore “items from the episodic text memory under con-
struction must be reinstated in the focus of attention to ensure the coherence of the
memory representation”. (Ericsson & Kintsch 1995: 30). Various devices are used to
achieve this such as “anaphoric and cataphoric elements in the episodic model of the
current text, generic lexical knowledge, as well as contextual features”:
In fact, language possesses a variety of graded syntactic devices to indicate to the
processor just where in the text information that is to be reactivated is located.
These are not only anaphoric cues but also cataphoric cues (e.g. referents marked
with an indefinite “this” will recur as a central concept in the subsequent text).
Syntax also instructs the reader when not to look for prior links but to start
a new thematic unit by means of a variety of switching devices. For example,
a plain “and” signals continuity (only 16% of the occurrences of “and” were
associated with switches in topic in Givón’s corpus), whereas an “And” following
a period signals the beginning of a new thematic unit (100% switches). Thus,
the cues present in a reader’s focus of attention not only make possible retrieval
from LT-WM but also indicate to the reader when to attempt such retrieval and
when not to. (Ericsson & Kintsch 1995: 30)
Although due reference is made to these syntactic devices Ericsson and Kintsch
pursue a psychological line of argument bypassing the detailed linguistic dis-
cussion with respect to universally relevant notions such as topic – comment
structure and the different pragmatic strategies of marking information as new
or old.
In the following I will argue that a detailed linguistic analysis of the infor-
mational structure of texts, as proposed in the MIC theory, is a necessary pre-
condition for the further elaboration also of the approach taken by Ericsson and
Kintsch. In fact, it can be shown that a clear distinction between semantic and
Memory and attention in MIC
pragmatic aspects of the content side of the information structure and their delin-
eation from “syntagmatic and paradigmatic aspects of form” can lead back to basic
neuropsychological foundations that have not been identified in Ericsson’s and
Kintsch’s paper.
German is considered a Free Word Order Language (FWL). The first Example (1)
shows that German seems to allow the rather free movement of the constituents
of a sentence into or towards positions in which they strike the attention of the
recipient. My notion of CA thus implies not only the functional aspect of bringing
something to someone’s attention, by marking it as “interesting, new, important,
surprising” (for instance by primary or secondary stress in almost any position in
the sentence) but also the local aspect of finding the optimal place in a linear order,
which by its mere position supports the recipient of the utterance in processing the
information. The classical topic-comment distinction is covered by my consider-
ations. This distinction is treated in the MIC theory as the distinction between
chunks of old and new information to which the relevant feature of prominence
is added. A chunk of information which is both old and prominent is called a
topic, whereas a chunk of information which is both new and prominent is called
a focus. Non-prominent old and new chunks of information are called respectively
the comment and the background.
Let us consider the German sentence (1)
Mein Mann kann morgens gut Kaffee kochen.
My husband can in the morning well coffee cook.
msgposs nom 3sgprs adv adv n acc inf
This sentence exhibits the default word order of a sentence with the finite verb
“kann” in second position (Gerdes & Kahane 2006). At least 20 different accept-
able sentences can be formed from the same set of words and constituents. These
different possibilities can be explained in the MIC framework as different prag-
matic (meta-informative) structuring of the same propositional information,
expressed as a base utterance or extended utterance (Włodarczyk & Włodarczyk
2008). A base utterance is entirely new or old; it contains only a subject (and even-
tually an object) as centre(s) of attention. An extended utterance is characterised
by the opposition (1) between old information (background) and new information
(focus) or (2) between new information (comment) and old information (topic).
Some extended utterances contain both a topic and a focus; in such cases, the
topic comes at the beginning and the focus may follow it immediately or appear
Franz J. Stachowiak
the very far end of a sentence and gets particular attention there either as a focus
or a postposed topic.
However, very few decisions, in fact one or two fixed positions and some
inflections can considerably reduce the number of possible word orders in
German, even down to only a single one. If some morpho-syntactic processing is
under way and a few word positions have been determined, the pragmatic free-
dom of arranging the linear order of the words diminishes or disappears. Thus if
a sentence has been started with “Gut Kaffee…” as in (1.5), only one particular
linear word order can follow. A similar phenomenon seems to occur in Russian, as
was pointed out by Myachykov and Tomlin (2008) in a study on the relationship
between perceptual priming and structural choice in Russian sentence produc-
tion. Despite the morphological richness of Russian and its rather free word order,
which allows speakers to preferably assign the sentential starting point, but not the
subject, to the perceptually primed referent (in a visual task) alternating between
the agent-initial and the patient-initial structural alternatives, Russian speakers
were more reluctant than speakers of English (despite the narrower inventory of
structural options in English) to alternate structure as a function of perceptual
prime. “This tendency may result from the necessity to maintain early commit-
ments to the case-marked noun forms, which effectively binds structural selection
to a much smaller number of available alternatives than the normative grammar
of Russian suggests” (Myachykov & Tomlin 2008: 1). This confirms the position
adopted in the MIC Theory that in language production the pragmatic decisions
come before the grammatical ones, but it is also evident that the grammatical ones
restrict the pragmatic ones. In language comprehension automatic grammatical
parsing comes first and is resolved in very short time – between 300–500 ms after
hearing the sentence (Friederici & Jacobson 1999; Friederici, Hahne; Saddy 2002).
Thus unexpected deviations from canonical word order or violations of phrase
structure are detected rather early and checked with respect to semantic and prag-
matic functions some 300 msec. later. I assume that on a scale between conscious-
ness and automaticity, pragmatic decisions are rather more conscious, whereas the
grammatical ones are clearly more automatic.
It must be emphasized that in the MIC theory, the basic syntactic structure
(represented as NP-VP in a constituency tree) corresponds to the base (entirely
new or old) utterance schema: a subject and what is predicated about it, i.e. a verb
with eventually an object. This syntactic structure reflects directly the pragmatic
structuring of a base utterance which consists in the choice of a global centre of
attention, i.e. a subject, and eventually a local one, the object. This makes it pos-
sible for speakers to produce and hearers to interpret base utterances very quickly,
even before they analyse the situation on the deep semantic level, i.e. even before
they understand clearly, which participant plays an active role and which a passive
Franz J. Stachowiak
one. On the other hand, it seems probable that extended utterances (consisting
of two different old and new information chunks) require more conscious and
deliberate pragmatic decisions which result in non-canonical word order. Further
psycholinguistic testing is necessary to corroborate this hypothesis.
Obviously the initial and final fields in the topological model of the German
sentence (cf. Zifonun et al. 1997) are important slots for the strategic position-
ing of the centers of attention in distributing the information that one wants to
convey in the most effective and successful way. Apparently new and given infor-
mation can appear in both locations. It depends on how one presents them meta-
informatively and how one wants to locate them with regard to the preceding and
subsequent information.
The languages of the world apply a whole range of different grammatical
devices to mark the centers of attention or the highlights of information (Gundel
et al. 2004). Constructions such as topic and focus fronting, cleft sentences, dis-
location to the periphery, topic markers (as in Japanese and Korean), sentence-
linking features such as ellipsis, anaphora, concord, word repetition, pro-forms,
adverbial contrasts and, last but not least, prosodic features are applied, but con-
sidering the diversity of all these devices it becomes apparent that there is no one
to one mapping between these techniques and pragmatic functions or subfunc-
tions of manipulating the distribution of information in the chain of speech.
Of course, it need not be elaborated here that – if we take the sentence or an
utterance as a cut out section of a longer text or stretch of speech – the first element
of this section, prototypically refers backwards or hooks up to what has been men-
tioned (“anaphoric grounding”), what is already known (“discourse grounding”)
or what is referred to (“ontological referential grounding”) in a more general sense
(see Table 1 below). The final element of the cut out section prototypically refers
to what is to follow in the text or communication (“cataphoric grounding”), to
information not yet known (“discourse grounding”) or it refers to specific, particu-
lar, actual or occasional aspects of ontological knowledge as stored in long term
memory. On the phonetic level this backward and forward processing is paralleled
by the way in which the realization of speech sounds is influenced by the preced-
ing and following sounds (Coarticulation).
Similarly – although on a higher cognitive level – the flow of information needs
to be handled in such a way that our “cognitive apparatus” can deal with it effec-
tively. André and Hélène Włodarczyk have described this as a meta-informative
process, depicted in their model of the grounding of the old or new status of infor-
mation (see Table 1) presented in the meta-informative mode. This corresponds
quite well with approaches in cognitive neuropsychology, and I would like to com-
plement their scheme by some further distinctions. Their notions of intermediate
and permanent memory point clearly to the fact that the marking of information
Memory and attention in MIC
as new or old involves online processes as well as the retrieval of stored memory
contents from the different facets of memory. I think that it is important to relate
this psycholinguistic aspect to the structural aspects which play a role in the lin-
earization of speech. Before going into the details of neuropsychological research
on memory and attention I would like to briefly discuss some basic aspects of the
processing of old and new information and the role of human memory in this.
Like other animals we are continuously confronted with an enormous flow of infor-
mation via the different senses such as hearing, sight, touch, smell, taste etc. With-
out the possibility of encoding and storing information in memory and retrieving
it in order to match old and new information and thus to learn from experience,
coping with the environment would be much more difficult. Of course, relevant
information for the survival of a species is coded in genetic memory, but in many
species, especially in primates and in particular in humans, different brain based
memory systems are operating in order to collect, evaluate and mark new relevant
information for social and cognitive behaviors and decisions.
Ernst Pöppel (1998), who has done pioneering work on the temporal struc-
ture of perception and cognitive processing found that the flow of information
Franz J. Stachowiak
input can be handled only by allowing it within certain time windows, which form
oscillating frames for parcelling the information into processable bits: “Humans
integrate information in a time window of approx. 3 seconds and generate a state
of awareness that can be referrred to as “subjective present”. This applies also to the
processing of language:
”Erkennbar wird unser “Gegenwartsfenster” beim Sprechen. Unsere Sprache ist
rhythmisch gegliedert, wobei Satzaussagen – im Übrigen unabhängig von der
Sprache und auch vom Alter – nur wenige Sekunden dauern, jeweils getrennt durch
Planungspausen für die folgende Aussage“. (Pöppel 1998)
“The window of our subjective present becomes recognizable in speaking. Our
language is structured rhythmically, and predications (propositions) last –
independently of language and the speaker’s age – only a few seconds, each of
them separated by planning pauses for the next statement”.
4. H
uman memory as a complex differentiated system and its relation
to text comprehension
Human memory is not a unitary system. Depending on the duration of the reten-
tion of information and the storage capacity, various kinds of memory must be
differentiated (see Figure 1). Sensory or iconic memory retains information only for
the blink of an eye, a very short moment. Short term memory allows for memory
in the range of up to 15 seconds and passes information on to long term mem-
ory where it can possibly be stored for a whole life. Modern neuropsychological
research deals with the so called working memory (Cowan 1999, 2005), a variant
model of short term memory. It is understood as a system that temporarily (in
the range of 1–3 seconds ) retains and manipulates information during cogni-
tive tasks such as learning, understanding, reasoning or calculation operations.
Working memory stores auditive information, for instance when comprehending
a verbal utterance, but also visual information, for instance when making deci-
sions in a complicated traffic situation. Working memory is very closely related
to attentional processes (Cowan et al. 1999). It is directly relevant to the notion of
center of attention in the MIC theory.
In long term memory information is often retained throughout a whole life.
It classifies and integrates information in a system of concepts. This pertains to
information, which is relevant for our personal biography (episodic memory)
such as occurrences, experiences, careers, encounters, accidents etc., but also to
information which is indispensable for our general knowledge, our education and
intelligent behavior (semantic memory). Semantic memory comprises knowledge,
as for instance the fact that I know that Paris is the capital of France. Episodic
Memory and attention in MIC
MEMORY
memory refers for instance to my memories of my very first stay in Paris when
I was 19 years old. A distinction has also to be made between explicit and implicit
memory. The former, also called declarative knowledge, signifies entries in long
term memory, which can be recalled purposely and consciously, as for instance a
mathematical formula such as a² + b² = c². In contrast, the results of procedural
learning, as for instance one’s competence to ride a bicycle or peel a potato, are
considered as implicit knowledge.
Semantic memory implies explicit knowledge. It is closely related to language
and word meanings. Our vocabulary is part of semantic memory. That a poodle
is a dog and that a dog is a mammal is part of my world knowledge, which I can
retrieve deliberately and “declare” as a matter of fact, but it is also part of my concep-
tual and notational knowledge. In the same way general knowledge, as for instance
expressed in “take-home-messages such as “The year has 12 months”, is considered
as part of semantic memory. Semantic memory exceeds episodic memory, as it
is beyond the scope of autobiographical memory contents. In our brain seman-
tic memory is organized as a conceptual network. However, semantic memory
interacts with episodic memory or is even extracted as a kind of yield or upshot
from episodes in life. It is quite obvious that verbal utterances frequently refer to
contents of the different kinds of memories in order to give them – depending
on the context – the status of new or old information. The opposition between Σ
(types) and Π (occurences) in the ontological grounding of the meta-informative
old or new status (Włodarczyk A. Chapter 3 in the present volume), could there-
fore be related directly to these forms of long term memory. If the distribution of
new and old information in an utterance serves the purpose of matching old and
new memory entries and of learning, it can be assumed that relevant information
is passed on into episodic memory and from there to semantic memory, while at
the same time information from semantic memory is fed into working memory in
order to decode a message semantically.
Of central importance to the MIC theory is the notion of “working memory”.
The best known model was developed by Baddeley and Hitch (1974), which has
attracted an immense amount of neuropsychological research.
Working memory refers to our capacity to simultaneously process and store
information. According to Baddeley’s model (Figure 2), working memory con-
sists of several subcomponents which play an active role in the process of holding
new information while processing or changing it in the visual and the verbal sys-
tem (for instance in text production and and comprehension). The central execu-
tive serves the attentional control of information that is actively dealt with in the
visual or verbal subsystems and it controls the transfer of information from and
to long term memory. In other words, this component is a good candidate for
a mechanism that is engaged in pragmatic decisions on the meta-informative
Memory and attention in MIC
Central
executive
5. W
orking memory capacity and processes from the basis
of the meta-informative processing of linguistic utterances
How much information can be held in working memory? In a very careful lit-
erature review, in which he brought together a wide variety of data on capac-
ity limits, Nelson Cowan (2001) came to the conclusion that a single, central
capacity limit between three and four unrelated chunks, which can be attended
at one time, characterizes human short term memory. Chunks are understood
as a collection of concepts that have strong associations to one another and
much weaker associations to other chunks currently in use. With respect to
language processing this implies that on average not more than four centers of
attention, about which the speaker intends to predicate, can be processed at the
same time.
The objective of many studies on the limits of working memory was to obtain
data, which allow working memory span to be used as a measure to predict the
performance of tested individuals in a whole range of cognitive tasks such as
reading comprehension, text recall, reasoning and problem solving. The following
figure (Figure 3) shows the decline in performance depending on the number of
intervening items. After 12 intervening items the retention rate declines towards
zero.
Hence, one can conclude that the interference from subsequent items causes
the deletion of information in short term memory. It is also evident that this study
falls short of accounting for memory processing of complex information in cog-
nitively higher demanding tasks. Scanning a brief list of digits contrasts sharply
with a task in which the information presented has some relationship to other
1
0,9
0,8
0,7
Proportion correct
0,6
1 digit/s
0,5
4 digit/s
0,4
0,3
0,2
0,1
0
1 2 3 4 5 6 7 8 9 10 11 12
Number of intervening items
Figure 3. Proportion of correct recall of items as a function of the number of intervening
items and presentation rate. Presentation rate was either one digit per second or four digits
per second (Acc. to Waugh & Norman 1965)
Rehearsal
Long-term
Information, Short term & memory
Input, working Transfer
Text memory Episodic
text
structures
Forgetting,
Decay
This model does not explicitly specify a component that attributes relevance
or importance to the information that is being processed. The Central Executive
Component in Baddeley’s working memory model would, however, be a candidate
for performing the task of meta-informative centering in the sense of the MIC
Theory (see further elaboration below).
One of the best known psychological findings to which the linguistic distinc-
tion between old and new information in discourse can be related is the serial
position effect. This refers to the fact that in free list recall – i.e. when subjects
are asked to recall the items of a list of digits, words or other symbols in free
order – the accuracy of the recall depends on the position of the items in the list.
In immediate recall the last items of a list are usually better remembered than
the first ones (Recency Effect). The first items of the list are better recalled than
the middle ones (Primacy Effect). An explanation of the recency effect is that in
short-term memory the processing of new items overwrites old ones. So at the
time of recall the last ones have not been overwritten and are still held active in
working memory. The primacy effect is explained with the hypothesis that initial
items are more readily encoded into long-term memory because they could be
rehearsed before new items came to be processed. Thus in recall the first items
are retrieved from long term memory. The first item in a list can also be processed
by itself, whereas the following ones have to be incrementally dealt with together
with the subsequent items. The explanations are supported by the finding that
the recency effect is eliminated when an interfering task is given before recall. If
for instance an arithmetical task is to be solved before the list items are recalled,
the recency effect disappears, as the new task requires attention and processing
capacity. Fast presentation of the items reduces the primacy effect. Later recall
increases it.
The following figure (Figure 5) shows the serial position effect.
Memory and attention in MIC
100
50
0
Primacy Intermediate Recency
Position in sequence
Figure 5. Serial position effect. The percentage of items recalled as a function of their
position in the list (Based on Murdock 1962)
If the explanations given above are correct, the distinction between primacy
and recency effect emphasizes the concept of transfer between working memory
and long term memory. Further findings add more details. Forward serial recall
generates a stronger primacy effect, whereas backward recall is characterized by
an increased recency effect. Although comprehending a verbal text is a much more
complex task, backward recall in list learning experiments resembles to some
degree the recall of segments of text that have already passed through working
memory but need to be checked again in order to fully understand the nature
of information to be conveyed. This could be the process of scanning backwards
from the comment towards the topic in a sentence. It is also conceivable that rel-
evant information is shifted to the end of a sentence in order to profit from the
recency effect especially if too much information had to be processed in the pre-
ceding paragraphs. Oberauer (2003: 471) holds that “one mechanism for generating
both primacy and recency effects is retroactive interference. Retroactive interference
during encoding (input interference) naturally produces a recency effect … Retroac-
tive interference during a series of outputs (output interference), on the other hand,
provides an advantage for the items retrieved first, because the early items interfere
with the later items in the output sequence”.
Although these studies do not investigate meta-informative strategies in sen-
tence and text processing directly – a task that calls for more linguistically based
research – it appears that some basic neuropsychological preconditions for dealing
with new and old information in texts can be traced in these results on working
memory capacity. The limitation of working memory capacity is a factor in a cog-
nitively demanding task such as text production and comprehension. It imposes
the necessity to select centers of attention as bootstrapping devices to cope with
Franz J. Stachowiak
In the preceding chapters it has already become apparent that attentional pro-
cesses are an integral part of working memory functions. Tasks that engage work-
ing memory at short notice, require attentional behaviour in the form of alertness,
focussed attention and also divided attention. Tasks that require divided attention
implicate at least two sources of attraction or stimuli that need to be regarded
simultaneously or sequentially in a fast alternation of the focus of attention.
The traditional models of attention (Broadbent 1958) assumed that only one
piece of information at a time would pass a perceptual filter and then be processed
semantically. The most critical part of this model was that it did not allow for
parallel processing of stimuli. Problems in the older models were overcome by
the introduction of a pertinence factor by Norman (1968). Pertinence refers to
the level of significance or salience, which the incoming information has for the
receiver. Certain kinds of inputs, such as the sound of our name, can be personally
important to us and are attributed the status of high level pertinence, others will
have changing levels of pertinence depending on the requirements and estimation
of ongoing information flow in a given context. Figure 6 shows the components of
Norman’s model and their interaction.
Sensory inputs
Processing
Storage
Pertinence
Selection
Attention
This means that pertinence can be given to a percept in such a way that it
catches the attention of the receiver. Relevant chunks of information in human
language need to be marked in such a way as to catch the attention of the commu-
nication partner. Word order change and intonation may be primary instruments,
but equally and additionally effective are “adverbial” lexical focus markers such as
“eben” (“just”), “gerade” (“exactly”), “insbesondere” (“especially”) which can even
be used cumulatively in order to draw the attention to that part of the information
that is considered as particularly relevant by its producer, as the following text
passage from a talk by my colleague Henning Lobin at the conference “Digital
Science” (20.09.2010 in Cologne) shows:
2. “.. Und auch hier wird gerade im BMBF ja eine Art Integration all dieser
Förderungen diskutiert, eh.. eine Zusammenführung dieser eh..Fördervorhaben, die
eh.., wenn man sie alle zusammenrechnet, ein unglaublich hohes Fördervolumen
auch im Moment aufweisen, um eben gerade insbesondere im Bereich der
Gesellchaftswissenschaften, Geistes- und Kulturwissenschaften so etwas wie ein
E-Science Initiative ins Leben zu rufen”
literal translation
“… and also here is in the BMBF (Federal Minsitry of Education and Research)
surely a kind of integration (of) all those eh – fundings being discussed, eh… a
joining (of) those support projects, which eh…, if one them all together counts,
an unbelievably high funding volume also at the moment exhibit, in order to just
exactly especially in the area of the social sciences, arts and cultural sciences (so)
something like an E-Science initiative into life to call ”.
The important point that he is making in this part of his talk – after speaking
about other research areas – is that the cultural sciences will get a lot of funding
within an initiative to support digital science. The relevant new information is
located in the right periphery, which starts with the final conjunction “um” and is
followed by the adverbial focus markers “eben gerade insbesondere” in cumula-
tive order calling for intensified attention to the chunks of information given in
the final position, which guarantees the recency effect in the working memory of
the listeners.
In order to further substantiate the usefulness of Norman’s pertinence model
of attention for the concept of “Center of attention” in the MIC Theory, it is appro-
priate to recall John Morton’s influencing Logogen model (Morton 1969), in which
he introduced the notion of “threshold”. A logogen is an abstract unit that contains
information about a lexical item such as its sound, form and meaning and con-
nected images. In a language task, for instance in comprehending a word or using it
in a naming task the logogen will be activated by verbal or other stimuli and context
information, if the incoming information is consistent with the information stored
with the logogen and if a certain threshold level is overcome. Transferred to the
Franz J. Stachowiak
notion of “Center of Attention” one can similarly assume that certain information
is necessary and certain “verbal tricks” must be applied in order to reach the level
at which pertinence is given to a partular constituent of a sentence. Focus markers
in this sense serve to reach the threshold at which focussed attention is activated.
Moving the subject of a sentence, which is also its topic and the agent of a
verbalized action into the very last position of a sentence captures the attention of
a hearer or reader, because this “trick” is – at first – against his or her expectations.
Thus it can be expected that in a FWO language the positions of new and old infor-
mation are exchanged for the purpose of attention capture. Neurologically this
leads to the activation of certain neural circuits in the brain. Experiments using
imaging technology seem to indicate that prefrontal, frontal, parietal and tempo-
ral areas of the brain are activated when attention is captured (Watkins et al. 2007).
7. Th
e MIC component in models of language production
and comprehension
“The speaker agrees (a) to use given information to refer to information she thinks
the listener can uniquely identify from what he already knows and (b) to use new
information to refer to information she believes to be true but is not already known
to the listener”. (1977: 92)
In order to fulfill this contract the speaker’s task is to apply all the necessary means,
which her language offers in order to make it easy for the listener to identify what
is new and what is old information. The MIC theory investigates the devices that
the languages of the world offer to their speakers in order to achieve this goal.
Linguistic research shows that these devices depend on different structural pos-
sibilities which are determined by properties such as free versus restricted word
order and other typological aspects. In a model of language production a compo-
nent needs to be identified which represents the functions that a speaker fulfills in
determining the centers of attention (CA) in the linear order of a given sentence
in a given context.
Memory and attention in MIC
–– Identify the given and new information in the linear structure of an utterance
–– Search your memory for information that matches the “given” information
(antecedent)
–– Integrate the new information into memory by substituting or equalizing the
new information with the antecedent knowledge
–– If there is no antecedent in memory, form bridging assumptions or implica-
tures in order to get the message
–– Rely on your world knowledge, beliefs and imagination and external situ-
ational information to decode the message
inf(r) ≅ e m-inf(c) ≅ r
References
Baddeley, Alan D. & Hitch, Graham J. 1974. Working memory. In The Psychology of Learning
and Motivation, Vol. 8, Gordon H. Bower (ed.), 47–89. London: Academic Press.
Baddeley, Alan D. 2000. The episodic buffer: A new component of working memory? Trends in
Cognitive Sciences 4(1): 417–423.
Baddeley, Alan D. 2003. Working memory and language: An overview. Journal of Communica-
tion Disorders 36(3): 189–203.
Broadbent, Donald. 1958. Perception and Communication. London: Pergamon Press.
Clark, Herbert H. & Clark, Eve V. 1977. Psychology and Language: An Introduction to Psycholin-
guistics. New York NY: Harcourt Brace Jovanovich.
Cowan, Nelson. 1999. An embedded-processes model of working memory. In Models of Work-
ing Memory, Akira Miyake & Priti Shahi (eds), 62–101. Cambridge: CUP.
Cowan, Nelson. 2001. The magical number 4 in short-term memory: A reconsideration of
mental storage capacity. Behavioral and Brain Sciences 24: 87–185.
Cowan, Nelson. 2005. Working Memory Capacity. New York NY: Taylor and Francis.
Cowan, Nelson, Nugent, Lara D., Elliot, Emily, Ponomarev, Igor, & Saults, John Scott. 1999. The
role of attention in the developement of short-term memory: Age differences in the verbal
span of apprehension. Child Developement 70: 1082–97.
Memory and attention in MIC
Ericsson, Anders K. & Kintsch, Walter. 1995. Long-term working memory. Psychological Review
102: 211–245.
Friederici, Angela D. & Jacobsen, Thomas. 1999. Processing grammatical gender during lan-
guage comprehension. Journal of Psycholinguistic Research 28(5): 467–484.
Friederici, Angela D., Hahne, Anja, & Saddy, Douglas. 2002. Distinct neurophysiological
patterns reflecting aspects of syntactic complexity and syntactic repair. Journal of Psycho-
linguistic Research 31: 45–63.
Garrett, Merrill F. 1980. Levels of processing in sentence production. Language Production:
Vol. 1, Speech and Talk, Brian Butterworth (ed.), 177–220. San Diego CA: Academic Press.
Gerdes, Kim & Kahane, Sylvain. 2006. A polynomial parsing algorithm for the topological model:
Synchronizing constituent and dependency grammars, illustrated by German word order
phenomena. In Proceedings of the 21st International Conference on Computational Linguis-
tics and the 44th annual meeting of the Association for Computational Linguistics, 1097–1104.
Gundel, Jeanette K. & Fretheim, Thorstein. 2004. Topic and focus. In Handbook of Pragmatic
Theory, Larry Horn & Gregory Ward (eds), 174–196. Malden MA: Blackwell.
Haviland, Susan E., & Clark, Herbert H. 1974. What’s new? Acquiring information as a process
in comprehension. Journal of Verbal Learning and Verbal Behavior 13: 512–521.
Morton, John. 1969. Interaction of information in word recognition. Psychological Review 76:
165–178.
Murdock, Bennet B. Jr. 1962. The serial position effect of free recall. Journal of Experimental
Psychology 64: 482–488.
Myachykov, Andriy & Tomlin, Russell. 2008. Perceptual priming and structural choice in
Russian sentence production. Journal of Cognitive Science 6(1): 31–48.
Norman, Donald E. 1968. Toward a theory of memory and attention. Psychological Review 75:
522–536.
Oberauer, Klaus. 2003. Understanding serial position curves in short-term recognition and
recall. Journal of Memory and Language 49: 469–483.
Pöppel, Ernst. 1998. Wie kommt die Zeit in den Kopf. Zeitschrift für KulturAustausch 3: 29–31.
Stachowiak, Franz J. 1985. Metaphor production and comprehension in aphasia. In The Ubiquity
of Metaphor [Current Issues in Linguistic Theory 29], Wolf Paprotté & René Dirven (eds),
559–599. Amsterdam: John Benjamains.
Stachowiak, Franz J., Huber, Walter, Poeck, Klaus & Kerschensteiner, Max. 1977. Text compre-
hension in aphasia. Brain and Language 4: 177–195.
Telljohann, Heike, Hinrichs, Erhard W., Kübler, Sandra, Zinsmeister, Heike & Beck, Kathrin.
2006. Stylebook for the Tübingen Treebank of Written German (TüBa-D/Z). Tübingen: Semi-
nar für Sprachwissenschaft, Universität Tübingen.
Watkins, Susanne, Dalton, Polly, Lavie, Nilli & Rees, Geraint. 2007. Brain mechanisms mediat-
ing auditory attentional capture in humans. Cerebral Cortex 17(7): 1694–1700.
Waugh, Nancy C. & Norman, Donald A. 1965. Primary memory. Psychological Review 72(2):
89–104.
Włodarczyk, André & Włodarczyk, Hélène. 2008. Roles, anchors and other things we talk
about: Associative semantics and meta-informative centering theory. Intercultural Prag-
matics 5(3): 345–365. Berlin: Mouton de Gruyter.
Zifonun, Gisela, Hoffmann, Ludger, Strecker, Bruno & Ballweg, Joachim. 1997. Grammatik der
deutschen Sprache. Berlin: De Gruyter.
part 3
This chapter focuses on the it-cleft construction and aims at, first, profiling it
from the perspective adopted in Meta-Informative Centering Theory (hereafter,
MIC) and, second, describing the major tendencies yielded by a diachronic
corpus-based analysis. It-clefts deserve a specific place in an account of English
constructions because of at least two specific characteristics: the expletive nature
of its introducer and the difficulty of accounting for the postverbal subordinate
clause by resorting to grammatical concepts such as relativisation, detached
postmodification, or even some kind of right dislocation.
The English it-cleft will be presented here as a device of focalisation used for
establishing a meta-informative contrast with the second part of the utterance,
making it possible, in a Strict Word Order (SWO) language as English, to put this
focalised constituent at the front of the utterance (thus contradicting the neutral
order: given-before-new). The study presented here illustrates the quite recent
consolidation of the mechanism of clefting as a thematising meta-informative
strategy in the English language. This consolidation of the it-cleft construction
is corroborated both quantitatively and qualitatively. By means of this system,
the speaker focuses on the referent of the Local CA constituent that occupies the
X-position and places heavy (normally available) information in final location
within the background rightmost clause.
1. Introduction
In the literature the label ‘cleft’, as suggested in Jespersen (1909–49, 1937), is com-
monly used with structures such as (1) to (5), which, irrespective of their syntactic
* The research reported in this article was funded by the Spanish Ministry of Science and
Innovation and the European Regional Development Fund (grant no. FFI2009-11274/FILO),
and Autonomous Government of Galicia (Directorate General for Scientific and Technological
Promotion, Grant No. CN2011/011).
Ana Elina Martínez-Insua & Javier Pérez-Guerra
The construction1 in (1) is called basic pseudo-cleft or basic wh-cleft, (2) illustrates
the reversed or inverted pseudo- or wh-cleft, the structure in (3) is known as cleft
or it-cleft, the one in (4) is known as th-cleft, and the example in (5) illustrates
a pronominal cleft. The treatment of the constructions in (1) to (5) as clefts is
grounded not only on the formal similarities between the constructions but also
on semantic, pragmatic and communicative considerations. To illustrate this, it
is commonly assumed that it- and pseudo-clefts share the same truth-conditions
or, as Prince (1978: 884) puts it, the “same objective information”, which implies
that they are semantically identical. In this vein, Akmajian (1970: 149) maintains
that it-clefts and pseudo-clefts “are synonymous, share the same presuppositions,
answer the same questions, and in general they can be used interchangeably”.
Declerck (1988: 209), however, claims that “there are numerous pragmatic factors
that may induce the speaker to prefer one type of cleft to another in a particular
context” [our italics] (see Traugott 2008: Section 3 for semantic and informative
differences between it- and pseudo-clefts).
It-clefts deserve a specific place in an account of English constructions because
of at least two specific characteristics: the expletive nature of its introducer and the
difficulty of accounting for the postverbal subordinate clause by resorting to gram-
matical concepts such as relativisation, detached postmodification, or even some
kind of right dislocation. First, on strictly syntactic grounds, basic pseudo-clefts
can be analysed as (unmarked) sentences in which a relative clause,2 be it headless
or headed, functions within the subject of the sentence. Second, with the same
. For the consideration that it-clefts constitute a grammatical construction type, see, among
others, Davidse (2000), Lambrecht (2001) and Traugott (2008).
. Although the majority of the scholars agree with the consideration that the wh-clause in
a pseudo-cleft is a relative clause (see references in favour of this view in Dikken et al. 2000),
there are some who think differently. To cite an example, Quirk et al. (1985: 1388) ground their
disagreement on the grammaticality of why pseudo-clefts, which have no relative counterpart,
as in ?Why we left was because we were tired. As we see it, the unacceptability of such examples
is due to the fact that headless why ‘relative clauses’ do not exist any longer, which corroborates
It-clefts in the meta-informative structure of the utterance in Modern and Present-day English
the analysis of the wh-components of pseudo-clefts as relative clauses. In fact, why pseudo-
clefts were possible in older English, as shown in (i), retrieved from the Helsinki Corpus:
(i) And why he is moyst, is, that it should be the more indifferenter and abler to euery
thing that shoulde be reserued or gotten into him: (Vicary: 33).
Ana Elina Martínez-Insua & Javier Pérez-Guerra
tree structure). The specific labels for one and the others are ‘Global CA’ and
‘Local CAs’, respectively.
Centering and predication are crucial notions within the MIC Theory. While
centering involves selecting and highlighting one entity among others, predication
implies saying something about the entity selected by the speaker as the global
CA of the message. Thus, predication takes place when speakers produce linguis-
tic expressions in which “some distinguished segments are highlighted as centres
of attention” (Włodarczyk, this volume). From such a perspective, “no utterance
can be formed in a natural language without choosing a CA and assigning to it
a meta-informative Old or New status” (Włodarczyk, this volume). This implies
that the meta-informative status of the CAs, as treated in this framework, con-
cerns their oldness or newness. In other words, the meta-informative status of
information concerns the way in which it is treated in the communicative event,
the way speakers introduce it into their discourse. Ultimately, the old/new meta-
informative status originates in the discourse strategies (partly fixed by syntactic
rules) established and employed by the speaker (Włodarczyk, this volume).
MIC has already been presented elsewhere as an appropriate framework
for the analysis and study of English thematically marked sentences, especially
those where an Anonymous Subject (henceforth, AnS) occupies the Global
CA position that another semantically full constituent would occupy in the
unmarked version of the sentence (see, Martínez-Insua 2011, Martínez-Insua &
Pérez-Guerra 2012). This chapter aims at highlighting the appropriateness of
this theoretical framework for the description of clefting as a focusing meta-
informative strategy.
This section tries to characterise it-clefts against the background of the MIC
approach to language, bearing in mind MIC’s distinction between Global and Local
CAs. In Section 3.1 we describe the meta-informative organisation of the it-cleft
sentence and pay special attention to the distribution of the CAs. In Section 3.2
we describe the structural and grammatical features of the construction.
(i) Sentences where the expletive and semantically empty AnS it3 fills the Global
CA slot, while a semantically full constituent functions as Local CA and is
followed by a clause that constitutes its background. The meta-informative
layout of the it-cleft would then be described as in (6):
(6) [ItAnS]Global CA [be] [Xi]Local CA (FOCUS) [introducer +
[clause …n.p.i…]]BACKGROUND
where the rightmost clause is the background contrasting with the focus expressed
in the cleft part of the utterance. This clause contains a null pointer [n.p.i] left by
the focalisation of one of its constituents [Xi] which acts as the Local CA.
(ii) Sentences where the expletive and semantically empty AnS it fills the Global
CA slot, while the semantically full Local CA is a clausal constituent whose
unmarked meta- informative layout has been altered. As a result of such word-
order alteration in the (main) clause, one of the components of the (subordi-
nate) clause has been focalised and brought to initial position, leaving a null
pointer behind. The meta-informative layout of the it-cleft is then described
as in (6′):
(6′) [ItAnS]Global CA [be] [[Xi] [introducer + [clause …n.p.i…]]]Local CA
where the Local CA is made up of a (rightmost) clause from which one of the
constituents has been topicalised and placed before the introducer [Xi], leaving a
null pointer behind [n.p.i].
The first one of these two characterisations, where the rightmost clause is
taken as the background of the utterance, seems to be the most appropriate one, as
we see it. The role of the rightmost clause as a second-level CA is in line with the
role of constituents such as a book, in Mary gave John a book, to John in Mary gave
a book to John, or with his key in John opened the door with his key. Knowing that
MIC does not allow for the presence of “neither double Local nor two-members
single Local CAs” (Włodarczyk & Włodarczyk, personal communication), the
rightmost clause has to be interpreted as a second-level CA (Włodarczyk &
Włodarczyk 2006: 8). In other words, according to the MIC theory, which bases
its premises on actual surface structure, we argue that a link construction such
. From a contemporary synchronic perspective, the expletive nature of the so-called ‘intro-
ducer’ (it) is justified by the fact that agreement does not hold between the two main constitu-
ents of the copular construction (it as the grammatical subject and the focus or ‘X’), as in It is
my parents that I never trust (see, in this respect, Pérez-Guerra 1999: Chapter 4, Section 3.2).
By using historical data, a number of scholars maintain that it keeps referential (cataphoric)
status (see Pérez-Guerra 1999: 170–171).
It-clefts in the meta-informative structure of the utterance in Modern and Present-day English
as the it-cleft construction (see Section 4.1 in this respect) consists of the linking
of two second-level (focus and background) CAs by means of a syntactic design
governed by an almost semantically bleached linking verbal operator, namely the
expletive it.4
(i) The Local CA ([X] in the representation in (6) above and the null pointer
are coreferring. For this condition to hold, X must materialise, in Givón’s
(1984: 731) words, an entity; otherwise referentiality would be out of the ques-
tion.6 The characterisation of X as an entity-denoting constituent excludes
sentences like (7), adapted from Declerck (1988), from the class of clefts:
(7) It is always expensive what Cambridge University Press sells.
Example (7) illustrates the extraposition of the Subject what Cambridge University
Press sells and the filling of the empty Global CA slot with the AnS it.
(ii) The ‘introducer’ in the rightmost clause can be one of the following: Ø
(Visser’s 1970: Chapter I apo koinou constructions), that, who or which. In the
literature, other expressions – for instance, what, when (see Declerck 1997)
or where – are frequently included in the set of possible introducers of the
rightmost clauses of it-clefts. From a syntactic perspective, such (apparent)
clefts with wh-forms are not syntactically different from extraposed headless
relative clauses or pseudo-clefts, as shown in (8) and (9) vs (10) (similarly,
Delahunty 1982: 268ff & Ball 1994a: 181):
. This proposal is, in fact, in keeping with syntactic solutions which rely on the fact that the
syntactic relation holding between the focus and the clause is subordination (see Section 3.2
in this respect).
. Notice the markedly syntactic nature of the criteria and terminology employed in the
following list.
. Example (i) will only be acceptable if a genius is understood as an entity-denoting
‘nominal,’ that is, in an identifying (probably contrastive) way:
If a genius is interpreted predicatively, the sentence becomes unacceptable. Notice that only
under very special circumstances can predicative nominals undergo clefting (see (iii) in the
main text).
Ana Elina Martínez-Insua & Javier Pérez-Guerra
(iii) The null pointer within the rightmost clause may fulfil a large array of func-
tions: Global CA (in (11)), Local CA (in (3) above and in (12)), (non-sentence)
adverbial (in (13) and (14), adverbial complement or obligatory adverbial
(in (15)), prepositional complement of a verb (as in (16)), of an adjective
((17)) or of a noun ((18)),7 complement of a preposition (in (19)), predica-
tive complement of the subject or of the object in very special environments
(examples (20) and (21),8 taken from Declerck 1988), predicative9 (in (22)), or
particle of a phrasal verb (in (23)).
(11) It is a gap that occurs in initial position.
(12) It is {#Ø/10 to} me that he dedicated the book.
(13) It was with much attention that I checked the last proofs of the article.
(14) It was only reluctantly that he agreed to help me with them at all.
(15) It is to Boston that she went.
(16) It is to my article that she was referring.
(17) It was about that Minister that the President was angry.
. See Dik (1997: 309–310) for the analysis of clefts with PP-foci as non-prototypical clefts.
. In examples (20) and (21), pretty and a teacher are clearly contrastive and thus the meaning
of the construction is identifying, as required by condition (i) in the main text, otherwise the
clefting of a predicative complement would be disallowed.
. Inflected verbs cannot be cleft (see (i) below) and (uninflected) VPs are accepted as Local
CAs in clefts only in informal Irish English (as in (22) in the main text; see Givón 1984: 731
for a detailed account):
Disjuncts and conjuncts cannot occur in the Local CA position of clefts, as shown
in examples (24) and (25), respectively:
(24) *It was frankly that he didn’t expect to discover the philosopher’s stone
in his study.
(25) *It is nonetheless that conjuncts can be cleft in languages other than
English.
. The acceptability of the clefts in (i), (ii) and (27) in the main text, including, respectively,
an -ing, a that- and an infinitive clause in X-position, indicates that the restriction on clausal
clefting is maybe due not to the theoretical organisation of the construction but to perfor-
mance preferences:
Delahunty (1982, 1984) justifies the acceptability of that-clauses by contending that they are
NPs. In fact, only that-clauses which are equivalent to entity-NPs, namely factive that-clauses,
are eligible for clefting. *It was that he was going to finish the book in two years that he thought
is ungrammatical because it includes a non-factive cleft clause in X-position.
Ana Elina Martínez-Insua & Javier Pérez-Guerra
(v) The rightmost clause can be either finite (supposedly a that-clause), as in (3)
above, or nonfinite (-ing or infinitive clause), like the one in (27):
(27) [Within the United States,] it is Robinson to appear like a Jones. (Gibb:115)
Despite their markedness, it-clefts, like other thematic systems (e.g. most there-
sentences), normally have the truth-conditions of their unmarked counterparts,
even though minor informative differences between both versions can be detect-
ed.13 In this section, devoted to the study of certain semantic and pragmatic
aspects of the constructions under analysis, we pay attention, first, to the iden-
tifying nature of it-clefts (Section 4.1) and, second, to the consequences which
the meta-informative layout of an it-cleft has for the informative design of the
sentence (Section 4.2).
. Visser (1970: 50) claims that examples of the type they are my brothers that… were clefts
from the sixteenth to the nineteenth centuries.
. Atlas and Levison (1981: 2), when they deal with the cleft in (i) and the noncleft in
(ii), point out that “[(i)] exhibits presuppositional behaviour that [(ii)] does not, namely, the
preservation of the inference to [(iii)] under denial and questioning of [(i)]” (see Halvorsen
1978: 18–19 for a similar account):
It is obvious that (7) has not got an identifying meaning, since expensive is not a
value but a predication.
Let us now deal with the clefts in (28) to (31), which are taken as attributive
in the literature:
(28) It is a poor heart that never rejoices.
(29) It’s a wise child that knows its own father.
(30) It would be a brave man who marries her, won’t it?
[taken from Declerck (1988)]
(31) Was it an interesting meeting you went to last night?
Regarding proverbial examples like (28) and (29), and less proverbial ones such as
(30), whose reading is clearly attributive, we agree with Declerck that these exam-
ples are not clefts but sentences involving some sort of relative clauses. Out of the
several reasons which Declerck adduces in favour of his proposal, we will, just for
the record, mention two. On the one hand, these examples cannot be uncleft ((28′)
to (30′)); on the other, the selection of relative pronouns is closer to clear relative
clauses than to the usual pattern of clefts ((28″) and (29″) vs. (32)):
(28′) A poor heart never rejoices. [the meaning is different from that of (28)]
(29′) A wise child knows its own father. [the meaning is different from that of (29)]
(30′) A brave man would marry her. [the meaning is absolutely different from
that of (30)]
(28″) It is a poor heart that/which/who never rejoices.
(30″) It would be a brave man that/who marries her, won’t it?
(32) It will be John that/#who will marry her, won’t he?
As far as (31) above is concerned, Declerck concludes that, owing to its hybrid nature
between the identifying and the attributive subclasses, this example illustrates a
new type of it-be construction. Most of the features cited by Declerck evincing the
attributive reading of the construction are semantic, whereas those signalling the
identifying interpretation are grammatical or syntactic. From both a grammatical
and a semantic point of view, the (supposedly) new class represented by (31) eas-
ily finds its way in the general class of (identifying) clefts. As we see it, Declerck’s
judgements are influenced by the obvious attributive nature of the adjective inter-
esting premodifying X in (31), which he extends up to the whole construction.
Declerck mentions the fact that the Local CAs (X-segments) in examples like (31)
can be quantified either by inserting no or by grading the adjective, and adds that
this is a consequence of the attributive character of the sentence. Adjectives, both
in attributive and in identifying contexts, can be expanded by constituents which
are modifiers typical of attributive items and not of the contexts in which these
adjectives are involved. In other words, the possibility of either inserting no before
the adjective or of grading interesting in (31) simply demonstrates the attributive
meaning of interesting, not of the whole example. Moreover, the supposedly ‘odd’
example (33) given by Declerck is perfectly acceptable in English, at least by the
native speakers consulted:
(33) It is John that/who is going to lead and an interesting subject that is going to
be discussed.
If our informants are correct in their predictions, the coordination of the clear cleft
It is John who is going to lead and (It is) an interesting subject that is going to be dis-
cussed in (33) corroborates the non-sylleptical semantic equivalence of the struc-
ture, or, in other words, the inclusion of (31) in the class of (identifying) clefts.
Declerck also compares (31) with other structures comprising verbs other
than be which are clearly attributive:
It-clefts in the meta-informative structure of the utterance in Modern and Present-day English
(34) [A couple of weeks ago it seemed clear which candidate deserved voting for,
but now many people think that] it has become a difficult choice they have
to make.
(35) It used to be a really good painter who painted their portraits.
We agree with the attributive nature of these examples, which is demonstrated
by their impossibility to be formulated as non-clefts, as shown in (34′) and
(35′):
(34′) They have to make a difficult choice. [not equivalent to (34)]
(35′) A really good painter {used to paint / painted} their portraits.
[not equivalent to (35)]
By contrast, ‘identifying’ (31) has a corresponding non-cleft:
(31′) Did you go to an interesting meeting last night?
Once the contrastiveness reading of (31) has been accepted, the meaning of
exhaustiveness can be justified. Finally, as we see it, tense restrictions affect not
only attributive sentences, as Declerck tries to show in (36), but also identifying
ones like (37):
(36) It would be/*is a more interesting subject that we would be discussing if
John had his way.
(37) It would be/??is Linguistics that we would be discussing if John had his way.
Summing up, we have tried to demonstrate that Declerck’s mixed type of identify-
ing/attributive clefts belongs to the general one of (identifying) clefts. Apart from
semantics, already discussed, all the grammatical factors (existence of non-cleft
counterparts, intonation and even pronominal connection) in Declerck’s study
also point towards the aforementioned conclusion:
(38) Was it a MEETING that you went to last night?
(39) It was an excellent description of himselfi/*himi that Johni wrote.
Ana Elina Martínez-Insua & Javier Pérez-Guerra
We have agreed with Declerck about the exclusion of certain it-be structures from
the class of clefts. To attempt the analysis of these examples is beyond the aims of
this chapter. It is their exclusion from the class of clefts that must remain uncon-
troversial for the purposes of the corpus study.
We contend that even though the content of the that-clause in (40) is clearly not
shared by the audience, it is – and here lies the justification of the cleft meta-
informative organisation – presented as such. Harold (1995: 158) claims that the
. See Croft (1991: 119), who claims that “the information in the backgrounded clause
refers back to an already open cognitive file, namely, a file established under ‘generally known
facts’. Thus, even if the backgrounded clause is new information to the hearer, a cognitive
file has been opened for it”. Similarly, this new/given duality is supported by Engelkamp and
Zimmer (1983: 64) when they say that “cleft sentences are used in particular communicative
situations, i.e. when the speaker fills an ‘active’ [author: givenness dimension] gap [author:
newness dimension] in the hearer’s knowledge” [our italics].
It-clefts in the meta-informative structure of the utterance in Modern and Present-day English
information in the that-clause of a cleft “is generally not in the listener’s mind. It
may be known or inferable or it may be totally new”. In this latter case, as pointed
out before (see also Declerck 1988), by means of the clefting device, the new infor-
mation is presented as if it were old. This leads us to conclude that the aim of
English clefts is to trigger the interpretation of the that-clause after the Local CA
as given, be it a carrier of old information or not.
Clefts are also claimed to bring forward to the attention of the hearer/reader a
certain (post-be) theme of the discourse as the focus of attention (the Local CA),
which makes clefts be considered meta-informative devices for focus-marking
(Rochemont’s 1986 ‘constructional focus construction’). This ‘presentative’ func-
tion is outlined in, for instance, Hetzron (1975: 362–364), Van Oosten (1986: 59)
or Declerck (see the section devoted to so-called ‘stressed-focus clefts’ in his 1988
study). Consequently, the meta-informative layout of the cleft makes the hearer
interpret George Orwell as focal in, for Example (40) above. Once it has been char-
acterised as such, the referent of the Local CA may become the ensuing discourse
theme.
The principle of end-focus (as part of the general one of given-before-new),
when applied to the it-cleft pattern, predicts that, on the one hand, the rightmost
clause of the it-cleft must contain information either unavailable or less predict-
able than the information conveyed by other constituents preceding the final
clause, and, on the other, which is in fact a consequence of the previous one, the
referentiality potential of the segment in postverbal Local CA should be, commu-
nicatively speaking, more referring than that of the final clause.
This section accounts for the data employed in our analysis of it-cleft sentences in
the history of English. In Section 5.1 we account for the frequency of the construc-
tion from Late Middle English to the present. Section 5.2 deals with the informa-
tive status of the major constituents in the it-clefts in the database.
Table 1. The corpus (raw data and normalised frequencies [n.f.] per 100,000 words
and 1,000 clauses)
Period words clauses it-clefts n.f./100,000 words n.f./1,000 clauses
It-clefting has not been a productive thematic mechanism in the history of the
English language. As Ball (1994b: 610) observes, a few “clefts and cleft-like con-
structions are attested in (…) OE and Early Middle English”. Whereas in LME the
number of it-clefts is not significant at all in the corpus employed (0.63), at the end
of the Modern period, their frequency is slightly above two out of one thousand
clauses. In PDE the proportion of clefts reaches and exceeds the figure of 5 out of
one thousand clauses.
LME 1 2
EModE 3 8 9
LModE 1 2 10
PDE 4 17 15
LME 1 1 1
EModE 9 5 6
LModE 7 4 2
PDE 15 8 13
especially from EModE onwards. These results make it possible to corroborate the
informative characterisation of it-clefts as focusing meta-informative strategies.
In Section 1 we defined the concept ‘cleft sentence’, which was restricted to the meta-
informative pattern known as ‘it-cleft’ since pseudo- or wh-clefts were regarded
as identifying or attributive copulative constructions comprising sentence-initial
wh-clauses realising the function Global CA, and pronominal and th-clefts were
treated as examples of non-restrictive relativisation or even (pseudo-) right dislo-
cation. In Sections 1 and 2 we described the basic structure of an it-cleft as follows:
Global CA in the form of the AnS it followed by a link verb, then the Local CA
and finally either a nonfinite or a finite non-meta-informatively-centered clause,
the latter being preceded by one of the following introducers: Ø, that, which or
who. Several conditions and remarks seem in order here concerning this structure,
which were discussed in detail in the course of this paper:
(i) existence of a null pointer within the rightmost clause: One of the constituents
of the rightmost clause is promoted to Local CA position.
(ii) categories in Local CA position: Most categories can occur in Local CA posi-
tion, non-contrastive adjective phrases (APs) and verbal constituents being
excluded.
(iii) syntactic functions related to the Local CA segments: The Local CA segments
can be associated with null pointers fulfilling most syntactic functions within
the rightmost clause, with the exception of verb-based functional roles, dis-
juncts and conjuncts. The usual functions realised by the null pointers of
the sentence-final clauses of the corpus were subject and adjunct, which was
accounted for on the basis that the constituents fulfilling these functions can-
not easily be focalised by means of other strategies.
(iv) information: The information conveyed by the Local CA segment is unavail-
able (either new or, as is claimed in the literature, ‘presented as new’ by the
speaker) to the hearer in the majority of the cases. Such an assertion, which
was corroborated by the data, characterised the cleft construction as a focusing
meta-informative device, not conditioned by the given-before-new principle.
Sources
References
Akmajian, Adrian. 1970. On deriving cleft sentences from pseudo-cleft sentences. Linguistic
Inquiry 1: 149–68.
Ariel, Mira. 1996. Referring expressions and the ±coreference distinction. In Reference and Ref-
erent Accessibility, Thorstein Fretheim & Jeanette K. Gundel (eds), 13–35. Amsterdam: John
Benjamins.
Atlas, Jay David & Levison, Stephen C. 1981. It-clefts, informativeness, and logical form. In
Radical Pragmatics. Peter Cole (ed.), 1–61. New York NY: Academic Press.
Ball, Catherine N. 1994a. Relative pronouns in it-clefts: The last seven centuries. Language Vari-
ation and Change 6: 179–200.
Ball, Catherine N. 1994b. The origins of the informative-presupposition it-cleft. Journal of Prag-
matics 22: 603–628.
Biber, Douglas, Finegan, Edward & Atkinson, Dwight. 1994. ARCHER and its challenges: Com-
piling and exploring ‘A Representative Corpus of Historical English Registers’. In Creating
and Using English Language Corpora, Udo Fries, Gunnel Tottie & Peter Schneider (eds),
1–14. Amsterdam: Rodopi.
Ana Elina Martínez-Insua & Javier Pérez-Guerra
Brömser, Bernd. 1984. Towards a functional description of cleft constructions. Lingua 62:
325–348.
Croft, William. 1991. Syntactic Categories and Grammatical Relations. The Cognitive Organiza-
tion of Information. Chicago IL: University of Chicago Press.
Daneš, Frantisek. 1974. Functional sentence perspective and the organization of the text.
In Papers on Functional Sentence Perspective, Frantisek Daneš (ed.), 106–28. Prague:
Academia.
Davidse, Kristin. 2000. A constructional approach to clefts. Linguistics 38(6): 1101–1131.
Declerck, Renaat. 1988. Studies on Copular Sentences, Clefts and Pseudo-clefts. Leuven: Foris.
Declerck, Renaat. 1997. When-clauses and Temporal Structure. London. Routledge.
Delahunty, Gerald Patrick. 1982. Syntax and Semantics of English Cleft Sentences. Bloomington
IN: Indiana University Linguistics Club.
Delahunty, Gerald Patrick. 1984. The analysis of English cleft sentences. Linguistic Analysis 13:
63–113.
Dik, Simon C. 1997. The Theory of Functional Grammar, Part 2: Complex and Derived Construc-
tions, edited by Kees Hengeveld. Berlin: Mouton de Gruyter.
den Dikken, Marcel, Meinunger, André & Wilder, Chris. 2000. Pseudoclefts and ellipsis. Studia
Linguistica 54(1): 41–89.
Engelkamp, Johannes & Zimmer, Hubert D. 1983. Dynamic Aspects of Language Processing.
Focus and Presupposition. Berlin: Springer.
Enkvist, Nils Erik. 1979. Marked focus: Functions and constraints. In Studies in English Linguis-
tics for Randolph Quirk, Sidney Greenbaum, Geoffrey Leech & Jan Svartvik (eds), 134–152.
London: Longman.
Firbas, Jan. 1992. Functional Sentence Perspective in Written and Spoken Communication.
Cambridge: CUP.
Givón, Talmi. 1984. Syntax. A Functional-typological Introduction. Amsterdam. John Benjamins.
Grosz, Barbara J. & Sidner, Candace L. 1986. Attention, intentions, and the structure of
discourse. Computational Linguistics 12(3): 175–204.
Grosz, Barbara J., Joshi, Aravind K. & Weinstein, Scott. 1995. Centering: A framework for
modeling the local coherence of discourse. Computational Linguistics 21(2): 203–226.
Halliday, Michael A.K. & Matthiessen Christian M.I.M. 2004. An Introduction to Functional
Grammar. London: Arnold.
Halvorsen, Per-Kristian. 1978. The Syntax and Semantics of Cleft Constructions [Texas Linguistic
Forum 11]. Austin TX: University of Texas.
Harold, Bruce B. 1995. Subject-Verb word order and the function of early position. In Word
Order in Discourse [Typological Studies in Language 30], Pamela Downing & Michael
Noonan (eds), 137–161. Amsterdam: John Benjamins.
Hetzron, Robert. 1975. The presentative movement or why the ideal word order is VSOP. In
Word Order and Word Order Change, Charles N. Li (ed.), 347–388. Austin: University of
Texas Press.
Huddleston, Rodney. 1984. Introduction to the Grammar of English. Cambridge: CUP.
Jespersen, Otto. 1909–1949. A Modern English Grammar on Historical Principles. Vols. I–VII.
Heidelberg: Ejnar Munksgaard.
Jespersen, Otto. 1937. Analytic Syntax. London: Allen and Unwin.
Lambrecht, Knud. 1994. Information Structure and Sentence Form. Cambridge: CUP.
Lambrecht, Knud. 2001. A framework for the analysis of cleft constructions. Linguistics 39(3):
463–516.
It-clefts in the meta-informative structure of the utterance in Modern and Present-day English
Shigeko Nariyama
The University of Melbourne, Australia
This paper discusses two ellipsis-related issues using the MIC theory as the
backbone of explanation: (1) ellipsis as a discourse coherence marker, which
answers the question, why use ellipsis (i.e. the pragmatic issue); and (2) referent
identification of ellipsis, which answers the question, how do we know
who/what a zero encoding of ellipsis refers to? (i.e. the semantic issue.) The
former is an issue also related to the speaker’s cognition whether or not to utilize
ellipsis for enhancing a discourse coherence based on the speaker’s assumption
of the addressee’s knowledge and ability to make sense of the zero encoding. On
the other hand, the latter deals with a matter stemming from the other side of
the same issue; namely, the addressee’s cognition to interpret and retrieve the
referential identity of ellipsis. The paper examines the mechanisms of referent
identification of ellipsis found in Japanese, the language that is said to utilize
ellipsis with extremely high frequency but without conventional cross-referencing
systems. It shows that Centre of Attention plays a central role in providing an
adequate and consistent explanation of the mechanism of ellipsis. It also presents
the importance of implicit lexical knowledge of Japanese that contributes to
referent identification of ellipsis.
1. Introduction
issue for languages, such as Japanese and other East Asian languages, as the gram-
mar of these languages freely allows zero encoding of entities without recourse
to the conventional cross-referencing systems. For instance, Romance languages
commonly rely their referent identification on subject-verb agreement features
in terms of person, number and gender. In contrast, the following Japanese sen-
tence taken from an article contains at least four ellipses without any conventional
agreement features displayed elsewhere in the overt part of the sentence. Nonethe-
less, the meaning of the sentence is unambiguous to the speakers of Japanese who
will all interpret the sentence as the given English translation.
(1) 妻に話すと同意してくれた。
Tuma ni hanasu to, douisite kureta.
wife to talk when, agree gave
‘When (I) talked to (my) wife, (she) agreed with (me)’.
This paper discusses two ellipsis-related issues using the MIC theory as its
backbone of explanation: (1) Ellipsis as a discourse coherence marker, which
answers the question, why use ellipsis (i.e. the pragmatic issue); and (2) Referent
identification of ellipsis, which answers the question, how do we know who/what
a zero encoding of ellipsis refers to? (i.e. the semantic issue.) The former is an issue
also related to the speaker’s cognition whether or not to utilize ellipsis for enhanc-
ing a discourse coherence based on the speaker’s assumption of the addressee’s
knowledge and ability to make sense of the zero encoding. On the other hand,
the latter deals with an issue from the other side of the same issue; namely, the
addressee’s cognition to interpret and retrieve the referential identity of ellipsis.
The remaining of this paper is organized as follows to address these two issues.
Section 2 observes the first issue – the role of ellipsis in discourse and its extent of
contribution towards creating a discourse coherence. Section 3 examines the second
issue – the mechanisms of referent identification of ellipsis found in Japanese, the
language that is said to utilize ellipsis at the extremely high frequency nonetheless
without conventional cross-referencing systems. It shows that Centre of Attention
plays the central role in providing an adequate and consistent explanation on the
mechanism of ellipsis. It also presents the importance of implicit lexical knowledge
of Japanese that contributes to referent identification of ellipsis.
One of the major functions of ellipsis, or anaphors more generally, that is to say,
the use of reduced referential form instead of repeating the exact same words
Subject ellipsis in Japanese
over and over again, is pertained to creating discourse coherence (e.g. Halliday &
Hassan 1976; Fox 1996). This claim is plausible, in that the use of ellipsis creates
cognitive states of coherence on the grounds that the addressee has to do context
search for the interpretation of the missing referent based on reduced information
provided, and this search links the current sentence to other sentences and meta-
information, and thus creates coherence. This is vindicated by the converse. Fox
(1996) reports that full expressions (not using reduced referential form, including
ellipsis) for inferable referents create paragraph/story boundary and discontinue
coherence.
Speakers in general are not conscious of making their sentences linguistically
or cognitively coherent. Rather they simply do not want to repeat the same words
or what is understood and obvious to the addressees, or such a sentence will sound
too tedious and uneconomical. This works fine in communication because of the
following. As will be elaborated in Section 3 in the discussion of the role of topic in
discourse, the global topic and old information, i.e. the entities with an old meta-
informative status and Centre of Attention, are most commonly encoded as ellip-
sis, and this reduced encoding leads to achieving the optimal coherence. This is an
utterly logical consequence, in that discourse is about talking about ‘something’,
i.e. ‘topic’ and linking sentences with the topic as the pivot, which leads to topic
continuity and hence coherence (Hinds 1983). As a consequence, the speaker is
able to encode the topic as ellipsis and the addressee can identify the ellipsis as
referring to the topic (see Section 3).
Note the differences between coherence and cohesion. Halliday and Hasan
(1976) in their book on Cohesion in English differentiate coherence from cohe-
sion, both of which function to enhance readability of sentences, i.e. ‘easy to
understand/read well’. According to their classification, coherence is created by
meta-linguistic means, such as commonsense knowledge, cognitive state, and
logical argumentation (cause and effect, question and answer, and so forth),
while cohesion is enhanced using grammatical means, such as anaphora and
ellipsis, substitution, and information structures including new/old information
and theme and rheme. That is to say, Halliday and Hasan regard ellipsis as part
of cohesion.
However, as Section 3 will show in explaining the mechanism of referent iden-
tification found in Japanese ellipsis, understanding referential identity of ellipsis
requires not only grammatical means hence establishing cohesion, but also meta-
linguistic means thereby requiring no antecedent either endophorically (within
texts) or exophorically (outside texts), hence establishing coherence. In the regard,
the MIC theory correctly and comprehensively encompasses coherence and cohe-
sion as meta-informative.
Shigeko Nariyama
Japanese is taken as a pilot language in this study for its abundant use of ellipsis
nonetheless without familiar agreement features, such as subject-verb agreement
and clitic pronouns. According to a 1955 research report by the National Language
Institute for Japanese Language, the subject was omitted 74% of the time in con-
versation, 37% in written texts, and 20% in novels in Japanese.1 A recent report on
the topic in 2008 by Ide (forthcoming) showed that the rate of omission was 68.7%
for the subject and 40.1% for the object. In comparison, the rates of its English
counterpart were 15.4% and 8.0% respectively.2
How are speakers of Japanese able to understand the referential identity of
so many ellipses contained in a single sentence in the discourse without cross-
referencing systems, as in (1)? Nariyama (2003) elucidated the mechanisms of
Japanese ellipsis with its approach ranging from lexical semantics to pragmatics,
which are very different from agreement systems found in European languages.
She classified the mechanisms as the interplay of the following three tiers of
device of restrictions: (1) Argument-inferring predicates (Word-level devices),
(2) Argument structure (Sentence-level devices that specify how to structure
a sentence), and (3) Discourse topic oriented structure (Inter-sentential-level
devices that specify how to sequence sentences coherently following one
another). These devices impose restrictions on the selection of words and sen-
tence structures and how to sequence sentences. These restrictions, thus in turn,
signal the referential identity of ellipsis. More recently, Nariyama (2009) added
various types of implicit lexical knowledge to Predicate devices that contribute
to identifying the referent of ellipsis. These devices utilize both the informative
and meta-informative means.
Each device is briefly summarised below. For simplicity and logical argu-
mentation of this paper, these devices are explained in the order of Argument
structure (Sentence-level) (in Section 3.1), Discourse topic oriented structure
(Discourse level) (in Section 3.2), and Argument-inferring predicates (Word-
level) (in S ection 3.3).
. See Nariyama (2003: 23–26) for the rates of the subject omission in other genres and the
rates of the object ellipsis.
. These results are based on experiments using a corpus called Mister O corpus. Subjects
participating in the experiments are asked to look at a series of pictures containing a story line
with no speech. In task 1, they were asked to construct a story. In task 2, they formed pairs to
make conversations using questions prepared by the researchers. The two tasks were done in
Japanese and also in English.
Subject ellipsis in Japanese
Subject ← → Non-Subject
Person/Animacy [1st Person > 2nd P > 3rd P > animate > inanimate]
. I thank André Włodarczyk for personally revealing his unpublished work on this issue
in the MIC theory. It shed light on how closely my own research is related to the MIC theory
and how it can be represented using the MIC theory that captures a wider scope of language
mechanisms.
Shigeko Nariyama
The differences among the three sentence types are captured in the principle of
ellipsis – The higher the referent is in terms of the person/animacy and/or dis-
course salience, the more prone it is to ellipsis. This is summarised in Table 2
(Nariyama 2003: 242).
This principle also helps the addressees to identify what ellipses refer to. That is
to say, when two arguments are omitted as in (4c), the interpretation always f ollows
the principle of ellipsis – the argument high on the person/animacy and/or dis-
course salience in Japanese (e.g. ‘I found him’), and the meaning cannot be inter-
preted in the inverse order. One exception to this is when a referent is focused on
providing new information after a Wh-question, as in “Who did it?” “I did ø”.
Shigeko Nariyama
Thus, the two principles, the principle of argument structure and the principle
of ellipsis, jointly claim – (1) an argument high on the animacy hierarchy and/or
discourse salience is made into the subject, and (2) this subject is most prone to
ellipsis. In other words, a referent with old meta-informative status is made into
the subject and this subject is most prone to ellipsis.
This explains why the subject ellipsis occurs more frequently than the non-
subject ellipsis in actual texts. Table 3 shows statistically from analysing 3718
naturally occurring sentences from various genres. 93.5% of those sentences with
ellipsis have subject ellipses and overt object, and the rest 6.5% have non-subject
ellipses. Among 6.5%, 4.6% of those sentences have both arguments omitted, and
only 1.9% of those sentences violated the principle. This violation is the result of
placing a focus on the overt subject, such as the answer to a wh-question.
(5a) and (5b) show that ellipted subjects refer to the topicalised subject. The ellipses
in (5a) refer to the topic. On the other hand, (5b) does not have the mention of the
topic, and hence the ellipses do not refer to the non-topicalised subject, instead
refer to the global/discourse topic that must have been mentioned prior to (5b)
or the first person as the default topic. In other words, sentences in Japanese are
sequenced with the topic marked by wa as the pivot and the subsequent mentions
of the same topic are ellipted. This omitted referents referring to the topic makes
the sentences coherent and allows the hearer to infer the referent correctly and
easily. This is illustrated in Table 4.
Table 4.
Basic rule of Discourse Structure: ‘X-wa … . øx … .’
involving conjuctives indicating that the adjoining subjects are the same or
different (see Section 3.3.5, and also Nariyama 2003: §5.5 for the detailed treat-
ment of other means in complex sentences).
(i) ‘[ øx …,] X-wa ….’ (cataphora: 13%)
or ‘X-wa [ øx …,] ….’ (anaphora: 87%)
(ii) ‘X-wa …. [ Y-ga …,] øx ….’
or ‘X-wa …. [ øx …,] Y-ga ….’
or ‘X-wa …. [ øx …,] øx ….’
(iii) ‘{[ øy … SS conjuctive,] Y-ga … }.’
(7) ジョンとメリーがけんかしている。(?)気が強いな。
John to Mary ga kenka shiteiru. (?) kiga tsuyoina.
and sub argue doing temper strong
‘John and Mary are having an argument. (He/She) is stubborn.’
Furthermore, some verbs are found to induce the same/different subjects in com-
plex clauses. That is to say, some verbs semantically restrict the subject in the
embedded clause to be the same as [SS] as in (13), or different from [DS] the
matrix subject as in (14).
This switch-reference function reflected in some verbs help identify the referents.
57 same-subject verbs and 273 different-subject verbs have been identified in
Japanese (Nariyama 2009: 72).
This paper has observed various aspects of ellipsis in light of discourse coher-
ence and referent identification, and has shown that the notion of Centre of Atten-
tion in the MIC theory captures the core mechanisms of ellipsis. It is claimed
that anaphora including ellipsis is universal and that no human language has
been found to repeat the same referent (Van Valin 1987: 513). Statistically, Gil-
ligan (1987) reported the finding from examining 100 languages that 93 languages
of those permit ellipsis in finite clauses. As such, studies of ellipsis in relation to
coherence constitute an important cross-linguistic study and a useful test base for
cross-lingual comparative studies.
Shigeko Nariyama
References
Fox, Barbara. 1996. Studies in Anaphora [Typological Studies in Language 33]. Amsterdam:
John Benjamins.
Gilligan, Gary. 1987. A Cross-linguistic Approach to the Pro-drop Parameter. Ph.D. disserta-
tion, University of Southern California.
Halliday, Michael A.K. & Hasan, Ruqaiya. 1976. Cohesion in English. London: Longman.
Hinds, John. 1983. Topic continuity in Japanese. In Topic Continuity in Discourse: A Quanti-
tative Cross-language Study [Typological Studies in Language 3], T. Givón (ed.), 43–93.
Amsterdam: John Benjamins.
Ide, Sachiko. Forthcoming. Research project entitled Empirical and theoretical studies on culture,
interaction, and language in Asia.
Kuno, Susumu. 1973. The Structure of the Japanese Language. Cambridge MA: The MIT Press.
Nariyama, Shigeko. 2002. The WA/GA distinction and switch reference for ellipted subject iden-
tification in Japanese complex sentences. Studies in Language 26(2): 369–431.
Nariyama, Shigeko. 2003. Ellipsis and Reference-tracking in Japanese [Studies in Language Com-
panion Series 66]. Amsterdam: John Benjamins.
Nariyama, Shigeko et al. 2005a. Extracting representative arguments from dictionaries for
resolving zero pronouns. In Proceedings of Machine Translation Summit X, 3–10. Phuket
Thailand.
Nariyama, Shigeko et al. 2005b. Building a cross-lingual referential knowledge database using
dictionaries. In Proceedings of Recent Advances in Natural Language Processing, 354–360.
Sofia Bulgaria.
Nariyama, Shigeko. 2009. How can we know ‘who did what to whom’ in Japanese? The Grammar
of omission: Less is more, 「誰が? 誰に? 何を?:日本語の省略がわかる本」Meiji Shoin
明治書院 Tokyo.
Silverstein, Michael. 1976. Hierarchy of features and ergativity. In Grammatical categories in
Australian languages. Robert MW Dixon (ed), 112–171. Canberra: Australian Institute of
Aboriginal studies.
Stirling, Lesley. 1993. Switch-reference and Discourse Representation, Cambridge: CUP.
Van Valin, Robert. 1987. Aspects of the interaction of syntax and pragmatics: Discourse corefer-
ence mechanisms and the typology of grammatical systems. In The Pragmatic Perspective
[Pragmatics & Beyond Companion Series 5], Jef Verschueren & Marcella Bertuccelli Papi
(eds), 513–531. Amsterdam: John Benjamins.
Watanabe, Yasuko. 1989. The Function of “WA” and “GA” in Japanese Discourse. Ph.D. disserta-
tion, University of Oregon at Eugene.
Wilkins, David. 1988. Switch-reference in Mparntwe Arrernte: Form, function, and problems of
identity. In Complex Sentence Constructions in Australian Languages [Typological Studies
in Language 15], Peter Austin (ed), 141–175. Amsterdam: John Benjamins.
Subject ellipsis in Japanese
Miki Saijo
Tokyo Institute of Technology
1. Introduction
The purpose of this paper is to show how various topics and focuses of plural par-
ticipants brought into a Japanese conversation can be controlled to produce a con-
sistent flow of conversation, explaining it with reference to the MIC Theory. The
data to explain this issue is a staff review meeting of science café held in Hiroshima
(hereafter “Hiroshima data”). There were six individuals,1 and each participant
stated his or her opinion prompted by the chair of the meeting, forming a multi-
party conversation.
. Four persons appear in the example discourse used for the analysis.
Miki Saijo
This data is an excerpt from a review meeting of “Science Café Hiroshima”. Science
Café is a talk event between scientists and non-scientists so that they can get to
know each other. The city of Hiroshima is one of the only two cities in the world
that experienced nuclear bombing in August 1945. The damage from nuclear
bombs was so extensive and tragic that citizens started up activities to fill the gap
between average citizens and scientists in order to prevent science from again
becoming something that citizens could not control. Science Cafe Hiroshima is
one of such activities that were managed by citizen volunteers to make it pos-
sible for scientists and non-scientists to communicate. The theme of the particular
Science Café held in June 2007 was the role that hydrogen plays as a future energy
source. A scientist presented a speech about challenges for using hydrogen energy
as an energy source for electric power generation and motor vehicles, which was
then followed by a workshop-style discussion on how to use hydrogen energy.
After the presentation and workshop, the participants had a chance to observe
RX-8, one of the only six cars in the world equipped with a hybrid engine which
operates on hydrogen and gasoline. We will refer to this car, RX-8 (see Figure 1),
in the data, which will be shown later.
4. Data
01 A : Yappari RX-8 (âru ekkusu eito) tte ichidowa untensite mitaidesuyo, ne.
02 B : Soreto Are nenryo, suisono jite, jidôsha janakute tada RX-8ga kitadakede mou.
03 C : Sô desuka.
04 D : Sugoi Gan gan totteta monne.
05 A : Sugoi nanka so, nanka, otakunakanjide, (Warai)Tôkyô môtâshô mitaina kanjidesuyone.
06 B : Konna kamera motte.
07 D : Yokuiru otakuno shônen.
08 C : Â sôdesune.
to drive RX-8), and the presence of “yone (wasn’t it?)” allows us to assume that
the person A wished to share his presumption with others, so that the subject of
the descriptive part “untensite mitai (wanted to drive)” has to be “minasan (you)”
rather than “I.” We also note that the utterance has no object. Instead, we find a
marker of reference “tte’ for introducing the topic. This part (RX-8tte) showed
a strong prominence in the actual speech, showing that it is indeed the topic.
In 02 utterance, “minna kôfunshite imashita (everyone was excited).“is added.
“Mô” which is an adverb meaning “already (or readily)” is placed here to modify
“kôfunshite imashita (was excited).” Although this utterance has neither subject
nor verb, the exclamation mark is enough to convey B’s excitement to C. The cor-
rectness of the assumption is clear from the fact that the person C acknowledged
the receipt of the information by responding, “Sô desuka (Is that so?)” in 03. How-
ever, nobody knows if the information is true or false. In 04 utterance, “Monne”
consists of a sentence final participle “mon” that expresses information sharing
to which is added another final particle “ne,” and it represents D’s thinking that
B’s picture taking of RX-8 is clear evidence of his excitement. In 05 utterance, a
noun phrase “anataga shahin wo totteiru sugataga (the view of your taking pic-
tures)” is added. Although it is the topic of this utterance, it is omitted. “Otakuna
kannjide (the geeky feeling)” and “Tôkyô môtâshô (the Tokyo Motor Show)” are
both expressed with strong prominence, but they are not presented as the topic
but rather as the speaker’s comments. This ends the topic introduced in 01, RX-8.
The topic of the following utterances is geeky feeling or geek. The features of the
excerpt are as follows:
Subject is implicit.
Object, verb and adjective may remain implicit.
Noun phrase which shows topic remains implicit.
By prominence on Noun Phrase, participants can infer the implicit topic.
Sentence final particle shows speaker’s attitude to conversation.
Miki Saijo
5. C
entre of attention and peripheral attention in multi-party
conversation
. Professor Sachiko Ide kindly suggested this point to the author in MIC Conference 2008.
Centre of attention in multi-party conversation in Japanese
CA and the topic in this utterance. Simultaneously, 01 expresses the wish to share
the comment “unten shitemitai” with the participants using the sentence final par-
ticle “yone,” marking it as a peripheral attention. In this paper, we consider that the
predicates, comments as new information and backgrounds as old information, are
peripheral attentions (PA). In 02, the person B establishes RX-8 as the CA by the
repetition of the word, but converts the PA of 01 to a focus conveying the situation
of “minna kôfunshite ita” using the adverb “mou”. In response to the implicit PA in
02 that everyone was excited or moved, a topic, the pictures of RX-8, is presented
in 04 as an expansion of the focus of RX-8, and the fact that the person B is taking
a lot of pictures is presented as PA. In 05, the phrase, the pictures of RX-8, is fur-
ther extended and a topic of “the person B taking pictures of RX-8” is introduced.
But it is not shown explicitly. The CA can be restored from the PAs expressed as
comments such as “geeky feeling” and “like the Tokyo Motor Show.” Figure 6 is a
graphical representation of the topics, focuses, comments and backgrounds in the
chain of utterance from 01 through 08 and how they shift. In the Hiroshima data,
one topic is shown as the individual CA and in the same utterance PAs are also
presented. And the next speaker selects one or more preceding attentions, and adds
comments on them as new information. In this way, in the Hiroshima data, plural
speakers extended a shared topic in a discourse.
Looking at the issues in this way, there is a noticeable similarity (see
Figure 7) between the CA structure of the extended utterance indicated by
Włodarczyk 2007 and the CA structure of the sequence of utterances found in
the Hiroshima data.
Miki Saijo
01A
RX-8 Want to drive Wasn’t it?
CA: topic PA: comment PA: comment
02B
RX-8 Already (everyone was exited)
CA: focus PA: background
03C
Is that so?
PA: background
04D
Photo of RX-8 Taking photo Weren’t you?
CA: topic PA: comment PA: background
05E
View of Your taking Geeky Like being in Tokyo Isn’t it?
Photo CA: Topic PA: comment motor show? PA: comment PA: background
06B
Having a camera
PA: comment
07D
View of Your Boy, a geek
taking photo CA: focus PA: comment explanatory notes
gradation: new
08C white: old
That’s right explicit: solid line
PA: background implicit: dotted line
Figure 6. Graphical representation of topics, focuses, comments, backgrounds in the excerpt
general shared
CA focus background CA focus background
local individual
CA PA
Włodarczyk (2007)
6. Conclusion
RX-8, which they can recall, as the CA, explicitly or implicitly, and expanded
the discourse by presenting their comments about it as the PA. Moreover, the
particular topic-comment and the central and peripheral structure of attention
are identical to the structure of the expanded utterances of the MIC Theory in
the utterance level. It is suggested here that the participants can maintain con-
versation despite the fact that utterances in a multi-party conversation in Japa-
nese often fail to indicate major elements grammatically because the participants
keep repeating the previous speaker’s utterance, or expanding it by comments,
or inviting other parties’ empathy using final particles, so that they adjust their
own CA with other parties’ CA. The technique used in this case, i.e. reviewing the
structure of conversation by supplementing the grammatically missing elements
using markers of other parts of the utterances in order to correlate the newness
vs. oldness of information as well as central vs. peripheral attentions, carries the
risk of arbitrary manipulation, that it cannot be restored if there are too many
missing parts. Also, we note that the excerpt is indexical in that the shared CA
is the vehicle that all the participants saw immediately prior to the conversation
so that there was no negotiation among the participants on that point. Therefore,
it is our intention to analyze in our future research, possibly co-working with
researchers in the field of information processing, whether we can see the struc-
ture of the information management by participants even in a case where there
is no clear-cut reference in the participants’ awareness by assuming a shared CA
in discourse. In addition, we wish to bring the research on self utterance in long,
coherent discourse and expression with references on other parties’ utterances
(Saijo 1999) into this framework to tackle the issue of consistency of discourse in
a multi-party conversation.
References
Bono, Mayumi & Takanashi, Katsuya. 2007. What is necessary in analysis of multi-party interac-
tion? Trends and current state of interaction studies. Transaction of the Japanese Society for
Artificial Intelligence, 703–710. (In Japanese)
Morikoto, Ikuyo, Saijo, Miki, Nohara, Kayoko, Takagi, Kotaro, Otsuka, Hiroko & Okumura,
Manabu. 2006. How do ordinary Japanese reach consensus in group decision making? Iden-
tifying and analyzing “Naïve Negotiation.” Group Decision and Negotiation 15: 157–169.
Nambu, Misako, Harada, Etsuko, Suto, Satoru, Shigemori, Masayoshi & Uchida, Kaori. 2006.
Risk sharing communication in medical settings: Analyses of nurses’ conversation. Cogni-
tive Sudies 13(1): 62–79. (In Japanese).
Saijo, Miki. 1999. The Role of Meta-linguistics in Discourse. Tokyo: Kazamashobo. (In Japanese)
Włodarczyk, André & Włodarczyk, Hélène. 2006a. Focus in the meta-informative center-
ing theory. In La focalisation dans les langues, 1–27. Paris: Collection Semantiques,
L’Harmattan.
Miki Saijo
Włodarczyk, André & Włodarczyk, Hélène. 2006b. Subject in the meta-informative centering
theory. In Etudes cognitive. Studia kognitywne VII, 7–32. Warszawa: SOW, PAN.
Włodarczyk, A. 2007. 「日本語の二重主語文-発話におけるメタ情報の中心化理論か
ら」, Nihongo no ’nijû-shugo-bun’ – hatsuwa ni okeru meta-jôhô no chushinka-riron kara
(The ‘Double Subject’ Theory – from the viewpoint of the meta-informative centering in
utterances). http://perso.numericable.com/andre.wlodarczyk/
Verbal aspect in Slavic languages
between semantics and pragmatics
Hélène Włodarczyk
Université Paris-Sorbonne, Centre de linguistique théorique
et appliquée (CELTA)
1. Introduction
The ASMIC theory is of great help in dealing with the blurred borderline
between semantics and pragmatics in aspect use, putting forward a tentative way
out of some endless debates on Slavic aspectology: the problem of aspect pairs, the
need to revisit the distinction between aspect and Aktionsart, the amazing differ-
ences in the usage types of imperfective (IPF) verbs in Slavic compared to that of
the imperfect tense in French or progressive forms in English, etc.
2. O
ntology and Semantics: Explaining specific meanings
with universal concepts
Does it make sense to ask again and again what is universal and what is specific in
the verbal aspect of a given language? We consider that aspect belongs to universal
linguistic categories that can be defined by bundles of univocally defined attribute-
feature structures. Such attributes and their different values belong to the ontology
of human languages. The semantic content of a linguistic expression is a function
mapping this expression onto ontological concepts. In order to be able to explain
the semantic content of various aspectual expressions in different languages, we
need a general set of ontological abstract attribute-feature structures from which
we can choose specific subsets of semantic feature structures. We consider lin-
guistics as a theoretic-experimental science which consists of explaining specific
meanings with universal concepts, the relevance of which can be corroborated by
their use in the description of large collections of data (Włodarczyk H 2009). The
bundles of features serve as descriptions of individual uses of aspect forms in con-
text: all uses which could be described by the same feature bundles are considered
as occurrences of the same type: thus they are reduced to one type of use, we call
a usage type of aspect.
It is worth emphasizing that aspect categories in different languages differ
both in form and content so that important discrepancies must be taken into
account. As concerns form: on the one hand, the forms expressing aspect mean-
ings greatly differ from one language family to another; on the other hand, the
boundaries between lexicon and grammar and the degree of grammaticalisation
of aspect meanings are different in various languages. Let us take as an example the
contrast between French or English on the one hand and Slavic languages on the
other. InFrench and English, aspect morphemes belong to verbal flexion and are
amalgamated with tense forms; in Slavic languages, aspect morphemes are deri-
vational and therefore they are present in different tenses and moods of the con-
jugation even in non finite forms. Moreover, several aspectual meanings currently
conveyed by prefixes in Slavic languages are expressed by semi-auxiliary aspectual
verbs in French and English, e.g. the beginning of a situation can be expressed
Hélène Włodarczyk
by a prefix in Polish and by auxiliary verbs in French and English, cf. utterances
(1, 2, 3) hereafter.
(1) Pol. Nagle zaśpiewali pieśń ludową.
(2) Fr. Soudain, ils se mirent à chanter une chanson populaire.
(3) En. Suddenly they began to sing a folk song.
There exists a long-lasting debate concerning the borderline between the seman-
tics and pragmatics of aspect. As a matter of fact, it is always difficult to distinguish
the two domains from one another because the paradox of verbal aspect is that its
meaning seems to be completely pragmatic from the very beginning, since aspect
is literally the “viewpoint” of the speaker on the situation spoken about.
On the other hand, in structural linguistics, most contextually-bound uses
of semantic categories were regarded as belonging to the pragmatic level, taking
it for granted that systemic meanings, identified as the system of semantic oppo-
sitions characteristic of a given language or family of languages (be they gram-
matical or lexical), belong to “langue”, whereas on the contrary all uses in context
were considered as pertaining to “parole”, thus as belonging to the pragmatic level
defined as the “usage” of the linguistic system by a speaker in a speech act. Many
structuralist linguists insisted on the fact that a general definition of the mean-
ing of the category of aspect (on the level of “langue” or language system) is not
sufficient to give account of its numerous uses in context (on the level of “parole”
or language usage), therefore they opposed the general and particular meanings
(Jakobson 1932, 1936), the primary and secondary meanings (Kuryłowicz 1972,
1977), or the semantic potential and particular meanings of a grammatical category
(Bondarko 1971a, 1971b). But it is important to underline that not all particular
or secondary contextually-bound meanings are strictly pragmatic, many of these
uses are semantic.
Indeed, the development of pragmatic studies in logic and linguistics in the
20th century led to a quite new understanding of langue as the linguistic system
and parole as its usage. In the ASMIC framework integrated into the methodology
Verbal aspect in Slavic languages between semantics and pragmatics
. Information on the SEMANA software designed for interactive linguistic research
(making use of artificial intelligence tools to build databases) is available on CELTA’s website:
http://celta.paris-sorbonne.fr/anasem/indexIL.html
. As we will explain below, generally only the PF verbs derived from IPF dynamic telic verbs
with the sense of “reaching the finish moment” are considered grammatical aspect partners.
Hélène Włodarczyk
of this abstract binary opposition between PF and IPF were granted the status of
“grammatical aspect”.
In the history of Russian aspect theories, V. V. Vinogradov (1947–1972:
379–390) pointed to A. A. Potebnia (1835–1891) as one of the first aspectologists
who separated aspect as a binary opposition from Aktionsart which he called
“duration degrees” (stepeni dlitel’nosti). Potebnia based his conception on
diachronic observations; the typically Slavic aspect opposition developed gradu-
ally on the semantic ground of Aktionsart which existed in other Indo-European
languages, in which however the same binary aspect opposition does not prevail.4
In Slavic languages, the Perfective/Imperfective opposition is the consequence of
the clustering into only two classes of all Aktionsart verbs based on their abstract
common properties regarding situations (Netteberg 1953, Włodarczyk H. 1997).
The need to treat aspect and Aktionsart as manifestations of one category was
put forward by many linguists in the last quarter of the 20th century5 (among oth-
ers, by Karolak 1997). The binary category of aspect in Slavic languages emerged
out of the great semantic diversity of the derivational lexical markers that are
used as aspect morphemes: prefixes and suffixes. The use of prefixes and suffixes
to express aspectual meanings is a potentiality within many Indo-European lan-
guages but aspectual categorisation developed diversely in different subfamilies. In
Latin (as pointed out by Veyrenc J. [1968] in his Russian grammar), there existed
a frequentative suffix (jacere → jactare) which however did not evolve (transform)
into an imperfectivising morpheme. Likewise, prefixes could be used to express
a more precise, more definite in time (hence non-durative and non-repeated)
meaning than that of the simple verb (jacere → ejacere ‘throw away’). In modern
Germanic languages this sort of evolution concerns the use of prepositions to cre-
ate verbal affixes, be they prefixed or postfixed or used as so-called verbal particles
to create compound or phrasal verbs. In English, the meanings of verbs with par-
ticles (called “phrasal verbs”) pertain to aspectual semantics (Brinton 1988) and
it would be of great utility for teaching Slavic languages to English speakers to
. However it cannot be excluded that the phenomen of reducing multiple Aktionsarts to a
binary category is cyclic. Diachronic studies of verbal categories in Germanic languages (Leiss
1992, 2002) suggest that the history of aspectuality in Germanic languages has already passed
through an early phase (in Gothic) of binary aspect opposition expressed by pairs of perfec-
tive and imperfective verbs. Moerover, I notice that the use of the ga- prefix (ge- in modern
German) with the original meaning of “together” to form perfective aspect partners in Gothic
(Leiss 2002) can be compared with that in Slavic languages of the po- perfectivising prefix
with the original meaning of distribution and totalisation.
. As concerns the history of the concept of Aktionsart in Slavic linguistics, see Włodarczyk,
H. 1997.
Verbal aspect in Slavic languages between semantics and pragmatics
resultative aspectual meaning derived from IPF telic action verbs (in the MIC
framework these verbs are called “ordinary process” verbs, see Section 6 hereaf-
ter). Indeed, the “resultative” meaning seems to be considered the prototypical
meaning of the PF aspect because it is very frequent. Nevertheless one should
not forget that not all perfective verbs express the result of a telic active situation;
using the semantic features designed in a previous study,7 we describe “resultative”
PF verbs as verbs denoting the finish moment of an ordinary process and/or, in an
appropriate context, the after stage which starts after the finish moment. However,
the frequency of ordinary process verbs (generally called telic action verbs) in text
corpuses is so important that in Slavic aspectology it kept out of sight the existence
of less frequent state or refined process verbs whose perfective derived partners
express most frequently either the beginning (the start moment or begin stage)
or the interruption of a state or process in one of their stages. This is the reason
why in Slavic grammars static and refined process verbs are mostly considered as
imperfectiva tantum; as a matter of fact since these types of verbs have no differ-
entiated stages nor moments, they lack a PF partner pointing at a finish moment
(cf. the list of imperfectiva tantum verbs in the Polish grammar of the Academy
of science, Laskowski 1998). Moreover, the other way round, all perfective verbs
with any other meaning than that of result are considered as perfectiva tantum, as
a consequence such verbs are cast aside from the category of “grammatical aspect”
into that of “lexical aspect” (Aktionsart). As we shall see in Section 7 below, the
description of Slavic verbs based both on the concept of hypercategory and on the
universal semantic features proposed in the ASMIC framework (Chapter 2 in this
volume and Section 6 hereafter) makes it possible to reduce the huge number of
perfectiva and imperfectiva tantum to a few peripheral verbs. In our view Imperfec-
tiva tantum stricte sensu are IPF verbs with the meaning of abstract static relations,
e.g. Pol. znaczyć IPF, Rus. znachit’ IPF, “to mean”; Pol. należeć, Rus. prinadlezhat’
IPF, “to belong”). In our approach, on the other hand, perfectiva tantum are mostly
those event verbs which lack suffixal IPF derived partners. However, there exist
event PF verbs from which it is possible to derive an IPF partner denoting the
habitual repetition of the event. The list of Polish and Russian perfectiva tantum
verbs in the ASMIC restrictive acceptance is even shorter than that of imperfectiva
tantum in the same restrictive acceptance. In this perspective, it is possible to give
an exhaustive enumeration of both imperfectiva and perfectiva tantum verbs for
each Slavic language. Nevertheless, due to historically motivated morphological
. All terms in italics are the labels of the semantic features designed for the aspect category
in Włodarczyk & Włodarczyk 2003. We give a brief summary of theses features in §6 imme-
diately hereafter.
Verbal aspect in Slavic languages between semantics and pragmatics
restrictions, the list is not exactly the same in all Slavic languages8 although the
(im)possibility of deriving a certain type of aspectual partner relies basically on
the same semantic conditions in all Slavic languages.
6. Universal semantic features for the description of aspect usage types
To describe the different usage types of aspect forms we use three kinds of seman-
tic features: Situation Type, Situation Internal View and Situation External View.9
First, we interpret each contextual use of a verb form by reference to a formalised
classification of situations among which we distinguish states and actions, which
are further divided into events and processes that are of two sorts: ordinary or
refined, i.e. consisting in repeated quanta of the same process (Table 1 of Chapter 2
in this volume, repeated hereafter).
Space (3D) + + + +
Time – + + +
Progression – – + +
Granularity – – – +
. Verbs recently borrowed from non Slavic foreign languages are very often either imper-
fectiva tantum or biaspectual verbs (i.e. the same form may be used alternatively in different
contexts as perfective or imperfective).
. In former versions of this theory of aspect (Włodarczyk & Włodarczyk 2003, 2006) we
used the term “situation control” in place of situation external view. We decided to change this
term because, in linguistics, “control” is mainly used to refer to the ability of an active agent to
“control” a situation consciously and voluntarily.
Hélène Włodarczyk
and the role they play in the situation. The classification we use keeps apart frames
and participants playing roles in situations; however the relevance of participants
(and even anchors) for aspectual interpretation is taken into account in the analy-
sis of the context in which a verb occurs. The parameters used in Table 1 organise
situations into an ordered set.
In addition to situation types, we describe aspect uses with two sorts of seman-
tic features concerning the internal and external view of the situation. The internal
view of the situation is the point of view consisting of the selection of a moment
or stage (Figure 1).
MOMENTS
initial start enter exit finish terminal
< before > < begin > < run > < end > < after >
STAGES
The external view of the situation is made up of three types of features con-
cerning (1) the quantification of the situation (is it repeated or not), (2) the modi-
fication of the flow and the modification of intensity and (3) the composition of
several situations into one situation.10
For the description of Polish and Russian verbs we distinguish five different
flow modifications as follows:
. A description of these features can be found in Włodarczyk & Włodarczyk 2003, 2006.
Verbal aspect in Slavic languages between semantics and pragmatics
EXTERNAL VIEW
What we call external view corresponds only partly to the verb nuances which
are generally named Aktionsart in Slavic aspectology. Our concept of external view
is slightly narrower in that we consider that any verb nuance called an Aktionsart
but concerning the internal flow of the situation should be classified as belonging
to the internal view, i.e. to aspect in the narrow sense. For instance, we consider
that the Aktionsart verbs known as “ingressive” verbs, which draw attention to the
beginning of the situation (either the start or enter moments), belong to the inter-
nal view. On the other hand, we claim that any semantic nuance which modifies
the flow of a situation, its normal sequential cycle, belongs to the “external” view.
Among such nuances we put first of all any of the five modifications of the normal
flow of a situation listed above: interrupt, resume, off-and-on, keep and trans. The
modifications of intensity (strong or weak, or increase) are generally described as
Aktionsart (“augmentative” or “attenuative”) but it must be emphasized that they
always combine with the internal view, i.e. with the selection of a moment or stage.
Moreover, in Slavic languages, an intensity modification is also very often associ-
ated with a flow modification. In the third type of internal view parameters we put
the composition into one complex situation of several similar situations. On the
one hand, we use the composition parameter to give an account of verbs which
are considered in Slavic aspectology as belonging to the “distributive Aktionsart”:
the same situation is repeated several times by the same active participant but on
different passive participants or by different active participants in different places
or times, cf. utterances
(4) Pol. Wzsyscy pasażerowie już powychodzili PF.
(5) Rus. Vse pasazhyry uzhe povychodili PF.
“All the passengers have already come out one after the other.”
On the other hand, composed situations concern verbs in which the prefix points
at one situation and the root at another one, e.g. in the following verbs: Pol.
wyciągnąć PF and Rus. vytjanut’ PF, “to pull out”, the prefix points at the path of
Hélène Włodarczyk
motion (out of something) and the root at the manner of motion (pulling). All
verbs expressing a composed situation are perfective because they combine the
composition parameter with the selection of the finish moment of the last situa-
tion which is composed into the complex one.
We must emphasize that our approach differs from the Slavic theory of
Aktionsart in that we treat the external view parameters as ontologic concepts not
as classes of derived verbs. As a matter of fact, in order to describe the semantic
content of verbs which are generally considered in Slavic aspectology as Aktion-
sarts we use both internal and external view parameters. As an example let us look
at the description of so-called “augmentative” Aktionsart verbs.
Pol. Nie wyspałem się. Wczoraj póżno w nocy u sąsiqdów rozpłakało się
(6)
dziecko i już nie mogłem zasnąć.
Rus. Ja ne vyspal’sja. Nochju u sosedov rebënok razplakalsja i ja uzhe potom
(7)
nie mog usnut’.
“I could not get enough sleep. Last night my neighbours’ child began crying
loudly and I could not fall asleep.”
We describe the complex meaning of the verb “rozpłakać się PF, (generally called
“augmentative” Aktionsart) using the following parameters:
Type of situation: refined process
Internal view: enter moment
External view: not repeated, increasing intensity, flow modification: keep.
The chosen parameters give a much more precise account of the meaning of the
verb than the traditional label of “augmentative” Aktionsart. As a matter of fact
such a verb denotes not only the increasing intensity of the situation but also the
fact that the situation has reached the enter moment (last moment of the begin
stage and first moment of the run stage) and that the run stage is continuing
beyond its normal duration.
All the nuances described by the three sorts of parameters we mentioned (sit-
uation types, internal and external view) are mostly expressed in Slavic languages
by amalgamating morphemes (prefixes and suffixes) in which various types of
these meanings are combined, as in the Examples (6) and (7) on which we have
just commented.
We use these three types of features to give an account of the numerous con-
textually-bound semantic usage types of the aspect category: aspect features make
it possible to typify individual uses (occurrences) into usage types. Each aspect use
can therefore be described by a semantic feature bundle consisting of two parts:
situation internal view and situation external view. All uses described by the same
feature bundle belong to the same usage type. The situation type is considered as
Verbal aspect in Slavic languages between semantics and pragmatics
a condition for the usage of aspect. Thus we proposed the following formalised
definition of aspect usage:
Aspect Usage = {Situation Internal View, Situation External View} condition:
Situation Type
In Example (8), the after stage is the state when there are no tickets left, the process
leading to this stage is that of selling tickets. Likewise, other contextual meanings
related to the semantic features of the aspect category (the internal and the exter-
nal view features) belong to the semantics of the category.
In order to test the above mentioned features, we are conducting an experi-
ment which consists of describing aspect uses in Polish with the SEMANA soft-
ware12 comprising a database builder and computational data analysis tools (formal
concept analysis, rough set analysis, statistical data analysis). The computer-aided
method of acquisition of semantic knowledge (Włodarczyk A. 2007, 2009) con-
sists of collecting numerous contextual uses of a linguistic category and describ-
ing each of them by a feature bundle. The interactive method makes it possible
to add or remove from the database builder a feature or one of its values, when
it appears necessary for the relevant description of the corpus of examples. The
linguist c ontinues to describe more and more samples until no new configuration
. It is necessary to nuance the choice of the speaker: when aspect belongs to grammaticized
resources of a given language, the “choice” of a moment or stage in the situation analysis is
very often restricted by stereotyped communication habits which impose on a speaker’s mind.
. Information about interactive linguistics and the SEMANA software can be found at
〈http://celta.paris-sorbonne.fr/anasem/indexIL.html〉
Hélène Włodarczyk
of features is added; when this is achieved, the “general” meaning of the category
will be defined as the common core of features used to describe the different uses.
Usage types are obtained when reducing all identical uses to only one by deleting
duplicates.
This method is currently being applied to the description of context uses of PF
and IPF verbs in Polish in order to discover experimentally the configuration of
features which characterize, respectively, the PF and the IPF aspect (Włodarczyk H.
2009). Each configuration characteristic of one of the two aspects consists of a core
and a periphery. The core can be interpreted as corresponding to the most typi-
cal and frequent usage types, the periphery to less frequent and less typical usage
types. With respect to internal view parameters (the analysis of the situation into
moments and stages), IPF verbs mostly denote stages, whereas moments charac-
terise more frequently the usage of PF verbs. However it is important to note that
some PF verbs (mostly those derived with a prefix which are ususally classified as
“Aktionsart verbs”) point also at stages but these stages are additionally character-
ised by one parameter of the external view (repetition, modification of flow and
intensity or composition of situations). For instance, the combination of the inner
stage of an ordinary process (the run stage) with the interrupt parameter leads to
verbs which do not point at the finish moment but nevertheless are PF verbs, called
“limitative” Aktionart verbs, e.g. the verbs Pol. poczytać, Rus. pochitat’ (“to read for
a while”) are both perfective but they are used to denote ordinary processes that
are interrupted in their run stage, hence before they reach their finish moment.
The combination of the run stage with the interrupt parameter is comparable to
the combination of the same interruption parameter with the parameter of unana-
lysed situations (considered as wholes) expressed by IPF verbs denoting states and
refined processes. Since IPF verbs denoting states and refined processes simply do not
have any finish moment, it is impossible to derive from them a PF verb pointing at
the finish moment; however it is possible to derive from these IPF verbs aspectual
PF partners with the meaning of interruption of the whole situation (see §7.1 and
7.2 hereafter). Thus, when collecting samples of aspect uses (Włodarczyk H. 2009)
we were able to record that not all verbs denoting a stage are imperfective verbs,
some are perfective but, in that case, they combine the stage meaning with another
parameter, e.g. interrupt, resume, intensity, composition of situations.
It is worth underlining that the concept of hypercategory can serve as formal foun-
dation for the notion of “verb arsenal” (“arsenal verbal”) proposed by [Sémon J.-P.
1986] or that of “verb clusters” proposed by [Janda 2007] to replace the traditional
Verbal aspect in Slavic languages between semantics and pragmatics
concept of aspect pair in which an imperfective verb is associated with one derived
perfective partner only. As a matter of fact, an aspect pair is only a particular real-
ization of the more general case of aspectual derivation which consists in an aspect
cluster: one simple IPF verb has generally more than one PF partner (Włodarczyk
& Włodarczyk 2001; Włodarczyk H. 2009). As a consequence of the definition of
PF Aspect as a hypercategory, we consider that all those prefixed PF verbs that can
be derived from one simple IPF verb – adding any aspectual nuance13 (beginning,
limitation, intensity, etc.) to the root verb – should all be considered as aspectual
partners of the root verb. We call such a group of verbs an aspect cluster, e. g. from
the simple Russian verb pisat’ IPF (to write) it is possible to create several derived
perfective verbs: napisat’ PF (to write a text up to its end), this verb belonging
to the semantic class of so called “resultative” perfectives ; popisat’ PF (to write
during a certain time), this verb belonging to so called “limitative” perfectives;
razpisat‘sja PF (to write too much), this verb being an example of so-called “inten-
sive” perfectives.14 As we mentioned above, in the traditional treatment of aspect
as opposed to Aktionsart, only resultative perfectives are considered as “real” or
“proper” aspectual partners of the simple IPF verb from which they are derived
with a prefix. On the contrary, we consider that it is possible to derive from an
IPF simple verb more than one prefixed PF verb and that all derived PF verbs can
be considered as grammatical PF partners on condition that the prefix does NOT
change the lexical meaning of the IPF root verb but adds to it only an aspectual
nuance. We consider as aspectual those nuances which correspond to one of the
values of the aspect parameters presented in Section 6 in this chapter. The refer-
ence to the different values of the aspect parameters we have defined makes it
possible to distinguish among derived PF verbs those which can be considered
as pertaining to grammatical aspect, as opposed to the lexical classes of derived
verbs formed with prefixes having not only an aspectual perfectivising mean-
ing but adding also various (spatial or abstract) meanings to the root verb. As
an example, we consider a derived PF verb the prefix of which adds the sense of
the start moment to the meaning of its IPF simple verb as a “grammatical aspect”
partner; this is the case of verbs Pol. zapłakać PF and Rus. zaplakat’ PF, “begin to
cry”, formed by adding the prefix za- to IPF simple verbs Pol. płakać IPF and Rus.
plakat’ IPF, “to cry”. However the same prefix za- can be used to create a derived
. In this approach, we define as “aspectual nuance” one of the aspect parameters defined in
the MIC theory (Section 6 in this chapter).
. We use the expression “so-called” because in our approach based on the description of
aspect semantics using feature structures, we replace each Aktionsart name or aspectual label
with a bundle of features.
Hélène Włodarczyk
PF verb with a new lexical meaning from another simple IPF verb; eg. in Polish
zamówić PF, “to order”, from verb mówić IPF “to speak”; in Russian zakazat’ PF
“to order (something)” from verb kazat’ IPF “to give orders (to some one)”. Thus
aspect clusters are a subclass of all derived PF verbs created by the adjunction of
prefixes to simple verbs.
Aspectual clusters can be made up of only one PF partner or up to six or seven
PF partners with various prefixes. E.g. the simple Polish verb spać IPF (to sleep) has
several perfective partners which are used in different contexts depending on the
semantic feature added by the prefix: pospać PF (to sleep during a certain limited
period of time), wyspać się PF (to sleep as much as needed to feel well), zaspać PF (to
miss something because of sleeping too long, oversleep), odespać PF (to recover sleep-
ing, sleep off), dospać PF (to keep sleeping until a certain moment), przespać PF cały
film (to spend a whole period of time sleeping, sleep over). It is sometimes possible to
further derive an IPF partner by suffixation from a prefixed PF verb belonging to
an aspect cluster. E.g. the verb wysypiać się IPF may be derived from wyspać się PF
“to get enough sleep” and used as its IPF partner in contexts of habitual repetition
as in the following utterance.
(9) Pol.
On się wysypia tylko podczas weekendu.
he Pr 3pers N sg sleeps enough V IPF pres 3pers sg only during week-end
“He gets enough sleep only during the week-end.”
situation types, situation internal view and situation external view we can dis-
tinguish precisely the different possible semantic types of PF partners that can
be derived from a simple IPF verb in Slavic languages (see Piernikarski 1969),
depending on the type of semantic situation to which the simple verb refers (in a
given context). The four kinds of semantic situations we defined, i.e. state, event,
ordinary process and refined process, allow different semantic types of derived PF
in Slavic languages. It is generally assumed in the Polish and Russian aspectologi-
cal literature15 that “real” or “exact” perfective partners (having exactly the same
lexical meaning as their imperfective root) are prefixed PF verbs (often called
“resultative perfectives”) pointing at the finish moment of ordinary process verbs.
However, as shown above in §5, in the derived PF clusters of ordinary process IPF
verbs, the derived PF verb pointing at the finish moment, called “result” of the pro-
cess is the most frequent (Włodarczyk & Włodarczyk 2001), but generally NOT
the only possible one. As mentioned above in §5, such resultative verbs cannot
be derived either from stative verbs or from refined process verbs which, for this
reason, are generally considered as imperfectiva tantum. If we rely both on the
formalised list of aspectual semantic features defined in the AS theory and on the
two step concept of aspect as a hypercategory, it is possible to put an end to end-
less debates16 concerning the semantic identity of a derived aspectual partner with
the base verb. As a matter of fact no derived verb is completely identical with the
base verb and each type of semantic situation that can be expressed by a base verb
requires different aspect partners.
postat’ PF, “to stand”. Verbs derived with the prefix po- and conveying the mean-
ing that a situation is limited in duration were considered as “non grammatical,
limitative” Aktionsart derived from “imperfectiva tantum stative verbs”. These
“limitative Aktionsart” verbs, be they derived from stative, ordinary or refined
process IPF verbs, were for the first time treated as “congruency perfectives”17 by
(Sémon 1986) and named “aspectual Aktionsart” by (Mehlig 1994).
(10) Pol. pokołysać (się) PF, Rus. pokachat’(sja) PF, “to swing for a certain time”;
Pol. pomachać rękoma PF, Rus. pomachat’ rukami PF, “to wave arms for a
certain time”.
(11) Pol. machnąć PF, Rus. maxnut’ PF, “to wave once”.
It goes without saying that other, although less frequent, PF verbs are sometimes
derived from refined process verbs, e.g. one can derive from the Pol. IPF verb
machać, Rus. machat’ (“to wave“) the PF derived verb with intensity meaning Pol.
namachać się PF, Rus. namachat‘sja PF (“to wave a lot and get tired of it“).
Discussion of the lexical or grammatical status of derived PF verbs of types
(a) and (b) has occupied the community of Slavic aspect specialists for at least
half a century. As far as I know, the proposal to treat the so-called limitative verbs
(with prefix po-) as PF partners appeared first in (Sémon 1986). The treatment of
“semelfactive” verbs (with suffix -nu- in Russian) as perfective partners is argued
for in (Xrakovskij 1997).
. Such derived prefixed verbs are used as PF partners of the simple IPF verb when it is
impossible to derive a resultative partner.
. Such verbs are generally called “semelfactive Aktionsart” verbs.
Verbal aspect in Slavic languages between semantics and pragmatics
As has already been pointed out, we admit that the properly pragmatic usage
types of the aspect category are related to the interaction between speaker and
hearer, especially to the meta-informative (MI) old or new status, which a speaker
assigns to an utterance. In Polish and Russian, the relation between verbal aspect
and meta-informative status can be observed in the contextual uses of IPF and
PF verbs in old and new utterances. Namely, it has been observed (Forsyth 1970;
Bondarko 1971b) that PF verbs are more frequently used in a new utterance and
IPF verbs in an old one. More precisely, we observed (Włodarczyk H. 1997) that
IPF verbs appear in both old and new contexts whereas the PF verbs are used
almost exclusively in new context, the only exception being the expressive usage
type of a PF verb in an old MI context (see §8.5 hereafter). However, the configu-
rations of semantic features which are characteristic of IPF and PF aspect respec-
tively remain valid whatever the meta-informative status of the utterance in which
a verb form occurs. The interaction between the semantic and pragmatic usage
types of aspect is extremely complex partly because the meta-informative status
of a verb phrase depends on the type of utterance in which it occurs, either a base
utterance (entirely new or old) or an extended utterance (with a contrast between
new and old constituents).
Hélène Włodarczyk
As shown in Table 2, one of the possible types of grounding of the old or new
meta-informative status is ontological knowledge. When applied to the usage
of aspect in old and new utterances, the ontological reference accounts for the
usage of verbal aspect which opposes situations seen as types to situations seen
as occurrences or tokens (cf. Table 3 from Chapter 3 in this volume repeated
. The implication of old/new status for the usage of IPF past forms in Polish in contrast
with French imparfait forms is discussed in [Walkiewicz & Włodarczyk 2012].
Verbal aspect in Slavic languages between semantics and pragmatics
here for convenience). The opposition of types and occurrences concerns both
situations and their participants, above all those playing an active role, expressed
in most utterances by the subject of the verb. The number form (singular or
plural) of the subject noun denoting the active participant and its determination
(in/definiteness) in the utterance are indices leading to the interpretation of the
situation as type or token.
8.2 U
sage types of IPF and PF verbs in ontology-motivated
old and new utterances
We use the concepts presented in Table 3 to explain the usage of IPF and PF verbs
in ontology-motivated old and new utterances (cf. Włodarczyk H. 1997). As con-
cerns the ontological grounding of the old/new meta-informative status, perfective
verbs appear almost exclusively in Π contexts whereas imperfective verbs are used
mostly (but not exclusively) in Σ contexts. There is an obvious relation between
the semantic and pragmatic usage types of PF and IPF verbs, namely in Slavic
languages, PF verbs – most of which are derived by prefixation from a simple
Hélène Włodarczyk
IPF verb – denote a more specific point of view on a situation spoken about than
their IPF partner. As a matter of fact, a perfectivizing prefix (a) adds information
about one chosen moment of the situation (the start, enter, exit or end moment)
whereas the IPF verb may be used to refer to the whole situation without analys-
ing it into parts, and (b) very often the prefix also adds information about what
we call “external” aspect, especially the modification of the flow (interrupt, resume,
trans)20 or of the intensity of the situation, eventually a prefix makes it possible
to compose several distinct situations into one composed situation (we call this
the composition parameter, Section 6 in this chapter). By contrast, the IPF verbs
contain less specification about the same situations than their derived PF part-
ners, thus they can be used easily in generic and general utterances (often called
“gnomic statements” by linguists.)
Beginning with the first ontological motivation, IPF verbs can appear in utter-
ances referring to generic (12 and 13) and general (14, 15) situations.
(12) Pol. Sól rozpuszcza się (IPF) w wodzie.
(13) Rus. Sol’ raspuskaetsja (IPF) v vode.
“Salt dissolves in water.21
In utterances (12) to (15) IPF verbs used in generic or general statements denote sit-
uations seen as wholes: either non modified (by any of the external aspect param-
eters) states or refined processes, or ordinary processes seen as wholes (including
all their moments and stages).
On the other hand, we find PF verbs used in utterances (16, 17, 18, 19, 20) to
point at specific and particular situations.
(16) Pol. Polski poeta Adam Mickiewicz zmarł w Istambule w 1855 r.
“The Polish poet Adam Mickiewicz died in Istanbul in 1855.”
Rus. 28/02/2013 – Proslavlennyj latyshskij poet Imats Ziedonic skonchalsja v
(17)
sredu v vozraste 79 let.
“2013-02-28 – The famous Latvian poet Imats Ziedonic passed away on
wednesday at the age of 79.”
. Two of the flow modification parameters: the off-and-on parameter of intermittence and
the keep parameter do not concern PF verbs but IPF verbs.
. When two Polish and Russian utterances can be approximatively translated by one and
the same English utterance we give it after the second example.
Verbal aspect in Slavic languages between semantics and pragmatics
This specific or particular pragmatic usage combines with the different semantic fea-
tures that can be expressed by PF verbs: moments, modification of flow or intensity,
composition. In utterances (16) to (20) all PF verbs can be characterised semanti-
cally by the finish moment and the non repetition (uniqueness) of the situation.
On the other hand, IPF verbs are used in utterances referring to habitual sit-
uations, i.e. situations which are supposed to be repeated regularly as habits an
indefinite number of times.
(21) Pol. Ekonomiści czytają (IPF) prasę codziennie.
(22) Rus. Ekonomisty chitajut (IPF) pressu kazhdyj den’.
“Economists read the press every day.”
In habitual contexs, IPF verbs keep their semantic content: they point at the whole
situation (utterances 21 and 22) or at its run stage (in the sense of ongoing process).
In opposition, PF verbs occur in utterances denoting occasional situations, i.e.
situations which occur only on particular occasions.22 In utterances (23, 24), the
temporal adverb “yesterday” pointing only at one particular day, the singular form
of the noun “economist” and the particular noun “article” opposed to the general
noun “press” from utterances (21, 22), trigger an occasional reading of the situation.
(23) Pol. Wczoraj ekonomista przeczytał (PF) ciekawy artykuł.
(24) Rus. Vchera ekonomist prochital (PF) interesnuju stat’ju.
“Yesterday the economist read an interesting article.”
The occasional pragmatic usage combines with the semantic parameter charac-
teristic of the PF verb in the given context, in utterances (23) and (24) the finish
moment of the process of reading.
It has also been noticed that IPF verbs may be used to denote potential situa-
tions, i.e. processes which some agents are able to perform:
. Occasional situations can be repeated a definite number of times (for instance: yesterday
he met his cousin and kissed her three times.).
Hélène Włodarczyk
In such utterances with potential meaning, IPF verbs refer to the whole situation
spoken about.
The opposite usage of PF verbs appears in contexts where the situation they
refer to is depicted as actual. But since in Polish and Russian, PF verbs cannot be
used to refer to present but only to past or future situations, their actual usage is
restricted to past or future tenses.23 For instance, when asked a few days after a
conversation in which the decision was taken that one of the interlocutors should
write a letter to their common director, utterances (27) and (28) in which the PF
verb “to write” is used in the past are questions aiming at making sure that the
intended action to write a letter was actually achieved.
(27) Pol. Napisał (PF) pan list do dyrektora?
(28) Rus. Vy napisali (PF) pis’mo direktoru?
“Did you write the letter to the director?”
In Examples (27) and (28) the PF verbs point at the finish moment and can be
given a resultative interpretation.
Obviously, the pragmatic usage types of aspect are extremely context-
dependent but the influence of context is even more decisive for IPF than PF verbs.
Since the IPF is the unmarked member of the aspect opposition (Jakobson 1932),
its ontology-based usage types (the different nuances of the ∑ motivations) appear
alternatively depending on the context which makes it possible to enhance one of
them. Thus, the generic interpretation of a verb form depends on the generic inter-
pretation of the co-occurring subject and object noun phrases; the habitual inter-
pretation is triggered mostly by temporal adjuncts denoting regular repetition; the
potential interpretation can be caused by the semantic content of the verb itself
(an action which an agent may be able to perform). On the other hand, since the
PF is the marked member of the aspect opposition, its ontology-based uses are not
so much context-dependent; as a consequence, the four ∏ meanings of situation
tokens co-appear with a PF verb, e.g. the specific, particular, actual and occasional
meanings in (16, 17, 18, 19, 20), although one of them may be highlighted by the
context, for instance by an adjunct, e.g. the occasional meaning of the PF verb in
utterances (19, 20, 23, 24). Thus, it is important to emphasize that such examples
as (18, 19, 20) pertain simultaneously to particular, actual and occasional mean-
ings, one of them being possibly highlighted by a constituent of the utterance.
. As we will show in 8.4 below, in the present tense, only IPF verbs can be used with actual
meaning.
Verbal aspect in Slavic languages between semantics and pragmatics
Although the four above mentioned ∑/∏ oppositions serve as motivation for
the old or new MI status, they concern the ontology of the situations referred to,
i.e. ultimately the semantics of the situations. This seems to be one more reason
why it is difficult to draw a clear line between semantics and pragmatics since the
meta-informative level itself is partly motivated by the informative level (as mental
reference to ontological knowledge stored in long term memory).
(29) Pol. A – Opowiedz mu jakąś bajkę, on nie może usnąć. B – Dobrze, którą
mam opowiadać (IPF)?
“A – Tell him a tale, he cannot get to sleep. B – Well, which tale should
I tell (IPF)?”
Examples of the cataphoric use of PF verbs can be found at the beginning of com-
municative exchanges, discourses or texts, as in utterances (31) and (32) hereafter
which are respectively the first sentence of a short story and a novel.
(31) Pol. Przyjechał (PF) kiedyś do mnie mój daleki kuzyn, misjonarz.
(S. Mrożek, Z gawęd wuja, p. 123)
“Once a cousin of mine came to see me, he was a missionary.”
Hélène Włodarczyk
. This can be compared to [Weinrich 1973, 1989] who distinguishes in French tenses used
in the foreground or background.
. Examples of cultural differences of the known/unknown motivation in aspect usage types
are given in Włodarczyk, H. 1997 p. 187.
Verbal aspect in Slavic languages between semantics and pragmatics
By contrast, the use of an IPF verb in an utterance with old MI status can be
motivated by the fact that the speaker considers that the situation s/he speaks
about belongs to the knowledge of the hearer. For instance, it is possible to use
an IPF verb as referring to a known situation to remind the hearer of a fact
which the speaker considers to be a part of the hearer’s knowledge. Let us take
as an example the following conversations (33 in Polish and 34 in Russian) in
which the IPF verb “to say” (Pol. mówiłem, Rus. ja govoril) is used to suggest that
person A should keep in mind a piece of knowledge already communicated by
person B.
(33) Pol. A – Dlaczego nie zaprosiłeś Piotra? B – Mówiłem (IPF) ci, że wyjechał
na dwa miesiące.
(34) Rus. A – Pochemu ty Petra ne priglasil? B – Ja zhe govoril (IPF) tebe, chto on
ujechal na dwa mesjaca.
“A – Why didn’t you invite Peter? B – I already told (simple past) you that
he left for two months.”
Another condition of the cognitive motivation of the IPF aspect may be the refer-
ence to common knowledge: it is possible to use an IPF verb to refer to a situation
which is assumed to be known in a given environment. For instance, a waiter in
a big restaurant (with numerous staff) may ask (using an IPF verb) guests already
seated at a table whether they have already ordered something. In this case, all the
usual episodes of a restaurant visit, including the “necessary” phase of placing an
order is perceived as known by its participants.
(35) Pol. Czy panowie już zamawiali (IPF)?
(36) Rus. Vy uzhe zakazyvali (IPF)?
“Have you already ordered (present perfect)?”
Another example of the known motivation of the use of an IPF verb form is given
in utterances (37) and (38) in which the IPF verb “to open” is used because the
act of opening the window is assumed to be the cause of the sensation of cold in a
room where the window is already closed.
(37) Pol. Ale tu zimno, otwierałeś (IPF) okno?
(38) Rus. Kak zdes’ xholodno! Ty otkryval (IPF) okno?
“It’s so cold here, did you open (simple past) the window?”
In such known usage of the IPF, the situation expressed by the verb consists in old
information (has the old MI status), because either what is new in the utterance is
only the positive vs negative sign accompanying the verb (did the situation actu-
ally take place or not?) or the new information concerns only one of the partici-
pants or anchors but not the situation itself.
Hélène Włodarczyk
In the anaphoric and known usage types of IPF aspect illustrated by utter-
ances (29, 30 and 35 to 38) we observe at the same time a shift in the point of view
in relation to the situation, a change consisting of the selection of another part
of the situation to that selected by the PF partner. A PF verb points at the finish
moment of a process and frequently conveys the meaning of a resulting state (we
call it the after stage) beginning after a process has reached its finish moment. On
the contrary the IPF aspect used with an anaphoric or known meaning induces at
the same time a shift of the attention from the finish moment of a process to its
run stage, leaving unexpressed or implicit the finish moment – which leaves open
diverse interpretations: either the finish moment has not been actually reached
(because the process failed or was interrupted) or it has been cancelled by the
moment we call terminal, consisting in putting an end to the resulting state initi-
ated by the finish moment of a process.
The anaphoric and known discourse motivations of the old status account for
usage types of the IPF aspect which are generally regarded as “untypical” because
they seem to contradict what is considered its most typical semantic usage, namely
the usage of the IPF pointing at ongoing processes.26 The reference to the meta-
informative status makes it possible to give a more coherent account of those prag-
matic usage types of IPF verbs which perplex non native speakers27 for whom it
is hardly possible to puzzle out a logical link between such usage types and the
traditional concept of the imperfective aspect as durative. As a matter of fact, the
usage of IPF aspect in old utterances motivated by discourse or knowledge seems
to override its ontology-based usage types denoting generic, general, potential or
habitual situations. Thus, in an old utterance with anaphoric or known motiva-
tion, IPF verbs denote specific, particular or actual situations. Thus, IPF verbs
can express both typical (∑) and individual (∏) situations. In their anaphoric and
known usage types IPF verbs point at particular tokens of situations (instead of
types), namely they express particular, actual and occasional situations just as PF
verbs but without the specific semantic features added by PF morphemes, mostly
prefixes adding a precision concerning the analysis or mode of the situation
referred to by the verbal root (e.g. the start or finish moment, the interruption or
resuming, etc.). In former studies this use of the IPF verb was explained by the fact
. At the same time these pragmatic usage types have no regular correspondents in the
usage types of the French IMP or English progressive tenses. However a similar anaphoric or
known usage of the English progressive form has been described by Adamczewski H. 1978. It
seems that what is called “imparfait pittoresque” in French grammars is probably an anaphoric
or known usage (Włodarczyk H. 1995).
. As regards the non-correspondence between such uses of the IPF aspect in Slavic and
the use of imparfait in French see (Walkiewicz & Włodarczyk 2012).
Verbal aspect in Slavic languages between semantics and pragmatics
that the IPF verb is the most abstract name of a situation without any of the par-
ticular aspect features which can be added by a perfectivizing morpheme.28 The
discourse-grounded anaphoric and known usage types of imperfective verbs refer
to the same situations as those denoted by a perfective verb expressing a particular,
actual or occasional situation, but they appear in case this situation had already
been mentioned in the previous discourse or belongs to the knowledge universe of
the discourse participants.
8.4 Usage of Slavic imperfective verbs in both old and new contexts
As regards the different usage types of the Slavic imperfective aspect, we must
bear in mind that they should not be confused with the usage types of the French
imparfait or with that of the English progressive forms. The non-correspondance
originates mostly in the fact that some semantic meanings of aspectual forms in
different languages are comparable whereas some pragmatic usage types remain
different. Without deeper insight into the problem of the borderline between
what is semantic and what is pragmatic it is practically impossible to compare the
usage types of aspectual forms in different languages. One of the main theoretical
problems in the history of Slavic aspect studies has its origin in the fact that the
imperfective aspect has been globally identified with the Imperfectum of classical
languages (e.g. Lomonossov’s Russian grammar).29 This concerns first of all what
is considered the primary (or most important) usage of the imperfective aspect in
Slavic. A huge aspectological tradition considers as the primary meaning of the
imperfective aspect its usage denoting an actual ordinary process in its run stage
or an actual state lasting through a certain period; this usage is called the “dura-
tive” or “ongoing process” meaning of the IPF and it is currently opposed to the
“momentary” meaning of the PF (Karolak 1997). In the history of aspect studies
this opposition is represented by the spatial metaphor of a line (representing the
IPF) opposed to a point (representing the PF). However investigations into Polish
and Russian aspect usage in context (Włodarczyk H. 1997) led us to the conclusion
that the usage of the Imperfective aspect in Slavic languages to denote an actual
ongoing process is only one of its possible different meanings in context. It should
not be considered as the primary meaning of the imperfective if we admit that the
imperfective is the unmarked member of the privative aspect opposition. What can
be stated about the unmarked member is that it exhibits many different contextual
usage types, some of which are more frequent than the others. As the unmarked
. Forsyth J., 1970 calls this use of the IPF aspect simple denotation, Bondarko 1971b general
factual (obobshchenno-fakticheskoe znachenie).
. On the history of Slavic aspectology cf. Włodarczyk 1997; Archaimbault 1999.
Hélène Włodarczyk
member, the imperfective aspect has no primary usage, more precisely, no default
usage appearing in a minimal context. To explain it, let us draw attention to the fact
that he following utterances cannot be interpreted without further context:
(39) Pol. Ojciec czyta gazetę.
(40) Rus. Otec chitaet gazetu.
N mas sg nom V IPF pres 3pers sg N fem sg acc
“Father reads (or is reading) the newspaper.”
In such a short utterance, there is no reason to choose the actual durative meaning
rather than the habitual or potential meaning of the IPF verb. As a consequence,
out of context, it is impossible to choose the simple or progressive present in the
English translation.
The actual durative reading becomes explicit only in the following utterances
(41) and (42):
(41) Pol. Nie przeszkadzaj! Ojciec czyta gazetę.
(42) Rus. Ne meshaj! Otec chitaet gazetu.
neg. V Imperative N mas sg nom V IPF pres N fem sg acc
2pers 3pers sg
“Do not disturb! Father is reading the newspaper.”
On the other hand, the habitual meaning can be made explicit when the IPF verb
is used with an adjunct phrase specifying the habitual repetition:
(43) Pol. Ojciec czyta gazetę codziennie.
(44) Rus. Otec chitaet gazetu kazhdyj den’.
N mas sg nom V IPF pres 3pers sg N fem sg acc Prep P or Adverb
“Father reads the newspaper every day.”
However, an actual situation, on the other hand, may also be expressed by an IPF
verb form:
Hence, the same IPF Slavic verb form may be used to point alternatively at an
actual occurence of a situation (47, 48) – then it has to be translated into English
by a progressive form – or at a potential sitation (45, 46) – which translates into
English by a non progressive verb form.
The actual meaning is a pragmatic contextual usage of the imperfective aspect
which can be used regularly to refer to individual situations (situation tokens)
seen as unanalysed whole or as actually ongoing processes or lasting states (in any
tense: past, present or future). But, on the contrary, as was shown in Section §8.2.,
the ontological motivation makes it equally possible (in other contexts) to use
imperfective verb forms to refer to typical (generic, general, habitual or potential)
situations.
When comparing IPF verb forms in Slavic with imparfait tense forms in
French we propose to keep in mind the definition by Jerzy Kuryłowicz (1972),
according to whom the imparfait tense (in Latin and Romance languages) is a
marked form aiming at transferring into the past the point of view of the Pres-
ent tense in its actual (opposed to atemporal) reading, i.e. a situation going on at
speech time. To explain the actual meaning of progressive tense forms in English,
we must use another framework, the theory proposed by Reichenbach (1947) who
introduced a reference point (R) distinct both from the speech moment (S) and
the moment of the event (E) talked about. A reference point (R) anterior to the
speech act moment (S) and simultaneous with the moment of event (E) is neces-
sary to account for the French imparfait or English progressive simple past. This
reference point is nothing but another situation’s moment. On the contrary, in
Slavic languages, imperfective past verb forms are defined only by reference to the
speech act moment but not to another event: they are opposed to PF past forms
in that PF verb forms do not point either at unanalysed situations or at stages
but refer either to selected moments or modified stages. The interpretation of IPF
verbs as denoting actually ongoing situations depends on information added in
the context, i.e. information about another situation which is simultaneous and
serves as a reference point.
(49) Pol. Gdzie diabeł nie może, tam babę pośle (PF). (J. Krzyżanowski, Mądrej
głowie dość dwie słowie, t. I, p. 48)
Lit. “Where the devil is helpless he sends a woman.”
(50) Rus. Mat’ i vysoko podymet (PF), da ne bol’no opustit (PF) ruku. (Dal’, Po-
slovicy russkogo naroda)
Lit. “Although a mother raises her hand it does not hurt when she lets it
down.”
The general usage of the PF can be triggered by the impersonal use of the second
person singular denoting an indefinite anonymous human subject with the mean-
ing “anyone ” (cf. Chapter 4 in this volume).
(51) Pol. Coś na Bóg spuścił (PF 2pers sg), to już stracono (PF).
Lit. “What you left to God is already lost.”
(52) Rus. Bez truda ne vytashchish’ (PF 2pers sg) rybku iz pruda. 〈http://www.
mista.ru/pogovorki.htm〉
Lit. “Without effort you cannot pull a fish out of a pond.”
In the same way, a PF verb used with a potential meaning always contains an
expressive nuance emphasizing that the action is not easy to perform but that the
subject has such qualities that he is able to perform it at any moment.
(53) Pol. On jest bardzo oczytany: on odpowie (PF) na każde pytanie.
“He is well-read: he will (can?) answer any question.”
Verbal aspect in Slavic languages between semantics and pragmatics
(54) Rus. On vam ljubuju spravku dast (PF), vsë ob”jasnit (PF) – ochen’
erudirovannyj chelovek.
“He can give you any information, explain anything; he is a very learned
person.”
A PF verb used in a habitual context also adds some expressivity to the utter-
ance: the situation expressed by a PF verb is not so regularly repeated as when it is
expressed by an IPF, it can thus be called a sporadic repetition. This usage of a PF
verb is often used with the expressions Rus. byvaet, byvalo and Pol. bywa, bywało
(it occurs/occurred that).
(55) Pol. Chociaż jestem bardzo zapracowany, bywa, że wyjdę (PF) na krótki
spacer z przyjaciółmi.
Rus. Xotja u menja mnogo raboty, no i s druzjami inogda byvaet vyjdu (PF)
(56)
poguljat’
“Although I am very busy, I will sometimes go out for a short walk with
friends.”
In all these types of uses the present tense forms of PF verbs do not have their
usual future meaning but that of atemporal present characteristic of general
statements.
9. Conclusion
The pragmatic usage types of the aspect category based on ontology (∑ and ∏
uses) are close to their semantic usage and comparable in different languages
(and therefore easier to grasp for non-native speakers) whereas the usage types
based on discourse (anaphoric/cataphoric) and knowledge (known/unknown) are
more language specific. As was shown in this chapter, the different pragmatic and
semantic usage types of PF and IPF verbs can combine diversely in context. As
regards PF verbs their different usage types do not override. On the other hand
some usage types of IPF verbs can override each other and this is the cause of
non-native speakers’ difficulties with some usage types of the IPF verbs classifed
as “untypical” – although their real frequency remains to be studied on large
corpuses.
Beyond the description of verbal aspect itself, the ASMIC theory can serve as
tertium comparationis to bring together verbal and nominal categories which are
involved in the marking of the old and new meta-informative status, thus allowing
cross-category comparisons in the pragmatic dimension of language. Former stud-
ies comparing the nominal and verbal categories were based on semantic investi-
gations, bringing together nouns and verbs in the perspective of heterogenous vs
Hélène Włodarczyk
homogenous types of situations (Mehlig 1994, 1996); the semantic (often gram-
maticized) behaviour can be compared to that of count and mass nouns. The rela-
tion between the aspect category and the meta-informative old or new status can
serve as a basis for investigating what is common between article and aspect (Leiss
2000)30 more specifically in the pragmatic dimension (Włodarczyk H. 1997), and
thus opens up a wide range of possibilities not only in theoretical linguistics but
also in applied linguistics where reference to the motivations of the old/new meta-
informative status may serve to improve the translation and teaching methods of
Slavic languages in non-Slavic environments (Włodarczyk H. 1998). This is an
important application for the contrast between Slavic and non-Slavic languages.
As a matter of fact, when contrasting aspect usage in two modern languages,
one should take into account both semantic and pragmatic properties of aspect
forms. As an example, French imparfait and Slavic past imperfective are both fit
alternatively for old and new contexts. However, although the Slavic past imper-
fective occurs in similar old and new contexts as the French imparfait, it is used
in a wider range of new contexts (cf. Walkiewicz & Włodarczyk 2012). In a larger
typological perspective, taking into account both the semantic and pragmatic fea-
tures of aspect may help in understanding and comparing some aspect uses in
different languages31 which seem puzzling when considered only semantically.32
Let us emphasise that in all types of languages, aspect is an essential tool of the
meta-informative structure since it compels speakers to choose a point of view
(a centre of attention) in relation to a situation they talk about.
References
Abraham, Werner. 2013. The developmental logic of the analytic past in German and Polish
An issue of universalism or areal contact? In: Lohndal, Terje (ed.), In Search of Univer-
sal Grammar: From Old Norse to Zoque, 175–194. Linguistik Aktuell/Linguistics Today.
Amsterdam: John Benjamins.
Adamczewski, Henri. 1978. Be+ing dans la grammaire de l’anglais contemporain. Paris: Librairie
Honoré Champion.
Agrell, Sigurd. 1908. Aspektänderung und Aktionsartbildung beim polnischen Zeitworte. Lund:
Håkan Ohlssons Buchdruckerei.
Archaimbault, Sylvie. 1999. Préhistoire de l’aspect verbal. L’émergence de la notion dans les
grammaires russes. Paris: CNRS Editions, Coll: Sciences du Langage.
Avilova, Natalia S. 1976. Vid glagola i semantika glagol’nogo slova. Moskva: Nauka.
Bondarko, Aleksandr V. 1971a. Grammaticheskaja kategorija i kontekst. Leningrad: Nauka.
Bondarko, Aleksandr V. 1971b. Vid i vremja russkogo glagola. Moskva: Prosveshchenie.
Brinton, Laurel J. 1988. The Development of English Aspectual Systems. Aspectualizers and Post-
verbal Particles. Cambridge: CUP.
Czarnecki, Tomasz. 1998. Aspektualität im Polnischen und Deutschen. Bedeutungen und Formen
in einer konfrontativen Übersicht. Gdańsk: Wydawnictwo Uniwersytetu Gdańskiego.
Czochralski, Jan A. 1975. Verbalaspekt und Tempussystem im Deutschen und Polnischen, Eine
konfrontative Darstellung. Warszawa: PWN.
Forsyth John. 1970. A Grammar of Aspect, Usage and Meaning in the Russian verb. Cambridge:
CUP.
Janda, Laura A. 2007. Aspectual clusters of Russian verbs. Studies in Language 31(3): 607–648.
Jakobson, Roman. 1932. Zur Struktur des russischen Verbum. Charisteria Guilelmo Mathesio…
oblata: 74–84. Prague: Cercle linguistique de Prague.
Jakobson, Roman. 1936. Beitrag zur allgemeinen Kasuslehre (Gesamtbedeutung der russischen
Kasus). TCLP VI: 240–288. Prague.
Kardela, Henryk. 1997. Telicity as a perfectivising category: Notes on aspectual distinctions in
English and Polish, Language History and Linguistic Modelling: A Festschrift for Jacek Fisiak
on his 60th birthday, Vol.2: Linguistic Modelling, Raymond Hickey & Stanislaw Puppel
(eds). Berlin: Mouton de Gruyter.
Karolak, Stanisław. 1997. Arguments contre la distinction: aspect/modalité d’action ( Aktionsart).
Etudes cognitives/Studia kognitywne Vol. 2: 175–192. Warszawa: Slawistyczny Ośrodek
Wydawniczy (SOW) Polska Akademia Nauk (PAN).
Kątny, Andrzej. 1994. Zu ausgewählten Aktionsarten im Polnischen und deren Entsprechungen im
Deutschen, Rzeszów: Wyd. Wyższej Szkoły Pedagogicznej.
Kuryłowicz, Jerzy. 1972. Miejsce aspektu w systemie koniugacyjny. Symbolae Polonicae in
Honorem Stanislai Jodłowski, 93–99. Wrocław, Warszawa, Kraków, Gdańsk: WPAN
Ossolineum.
Kuryłowicz, Jerzy. 1977. Problèmes de linguistique indo-européenne, Warszawa: WPAN
Ossolineum.
Laskowski, Roman. 1998. “Uwagi o znaczeniu czasowników”. In Gramatyka współczesnego
języka polskiego, Morfologia, wyd. 2 zmienione, T. 1: 152–171. Warszawa: PWN.
Leiss, Elisabeth. 1992. Die Verbalkategorien des Deutschen. Ein Beitrag zur sprachlichen Katego-
risierung. Berlin: Walter de Gruyter.
Leiss, Elisabeth. 2000. Artikel und Aspekt: die grammatischen Muster von Definitheit. Berlin:
Walter De Gruyter.
Leiss, Elisabeth. 2002. Der Verlust des aspektuellen Verbpaare und seine Folgen im Sprach-
wandel des Deutschen: ein Überblick. In Grammatische Kategorien aus sprachhistorischer
und typologischer Perspektive. Akten des 29 Linguistenseminars, Kyoto 2001, Mitsonobu
Yoshida (eds), 9–25. München: Iudicium.
Mehlig, Hans R. 1994. “Gomogennost’ i geterogennost’ v prostranstve i vremeni”. Revue des
Etudes Slaves 66: 595–606.
Hélène Włodarczyk
Mehlig, Hans R. 1996. Some analogies between the morphology of nouns and aspect in Russian.
Folia Linguistica XXX: 87–109.
Netteberg, Kristine. 1953. Études sur le verbe polonais. Copenhagen: Rosenkilde og Bagger.
Piernikarski, Cezar. 1969. Typy opozycji aspektowych czasownika polskiego na tle słowiańskim.
Wrocław: Ossolineum.
Piernikarski, Cezar. 1975. Czasowniki z prefiksem po- w języku polskim i czeskim. Warszawa:
PWN.
Potebnja, Aleksandr A. 1941. Iz zapisok po russkoj grammatike (1899), t. 4. Moskva - Leningrad:
Izd. A.N.S.S.S.R.
Reichenbach, Hans. 1947. Elements of Symbolic Logic. New York NY: Macmillan & Co.
Rosch, Eleanor H. 1973. Natural categories. Cognitive Psychology 4(3): 328–350.
Sémon Jean-Paul. 1986. Postojat’ ou la perfectivité de congruence, définition et valeurs textu-
elles. Revue des Etudes Slaves 58(4): 609–636.
SFPW: Słownik frekwencyjny polszczyzny współczesnej, 1980. red. Kurcz, Ida, Lewicki, Andrzej,
Sambor, Jadwiga, Szafran, Krzysztof, Worończak, Jerzy. (eds). Kraków: Instytut Języka
Polskiego PAN.
Ter Meulen, Alice G.B. 1995. Representing Time in Natural Language: The Dynamic Interpreta-
tion of Tense and Aspect. Cambridge MA: The MIT Press, A Bradford Book.
Vendler, Zeno. 1967. Verbs and times. On Linguistics and Philosophy, Ithaca, Zeno Vendler,
97–121. New York NY: Cornell University Press. (Revised version of Vendler, Zeno. 1957.
Verbs and times. The Philosophical Review 66: 143–160).
Veyrenc, Charles Jacques. 1968. Grammaire du russe. Paris: Presses universitaires de France
(coll. Que sais-je?, n0 1278).
Veyrenc, Charles Jacques. 1980. Études sur le verbe russe. Paris: Institut d’études slaves.
Vinogradov, Viktor Vladimirovitch. l947, 2nd ed. l972. Russkij Jazyk. Grammaticheskoje uchenije
o slove. Moskva - Leningrad: Izd. A.N.S.S.S.R.
Walkiewicz, Aleksandra & Włodarczyk, Hélène. 2012. Revisiting the correspondence between
the Polish Imperfective Aspect and the French Imparfait in the Distributed Grammar (DG)
framework. In Studies in Polish Linguistics 7, 5–36. Kraków: Institute of Polish Language,
Polish Academy of Science, Lexis.
Weinrich, Harald. 1973. Le Temps. Paris: Le Seuil.
Weinrich, Harald. 1989. Grammaire textuelle du français. Paris: Didier/Hatier.
Włodarczyk, André. 2003. Les cadres des situations sémantiques. Études cognitives/Studia
kognitywne V, 35–52. Warszawa: SOW.
Włodarczyk, André. 2007. CASK – Computer-aided acquisition of semantic knowledge project
(in Japanese). Japanese Linguistics 21. Tokyo: The National Institute for Japanese Language.
English version downloadable at 〈http://celta.paris-sorbonne.fr/anasem/papers/〉
Włodarczyk, André. 2009. Interactive discovery of ontological knowledge for modelling
language resources. In Representing Semantics in Digital Lexicography. Proceedings of
MONDILEX, the 4th Open Workshop, Violetta Koseska-Toszewa, Ludmilla Dimitrova &
Roman Roszko (eds). Warszawa: SOW. 〈http://celta.paris-sorbonne.fr/anasem/papers/〉
Włodarczyk, André & Włodarczyk, Hélène. 2001. La préfixation verbale en polonais I. Le statut
grammatical des préfixes, II: L’Aspect perfectif comme hyper-catégorie. Études cognitives/
Studia kognitywne 4: 93–120. Warszawa: SOW, PAN.
Włodarczyk, André & Włodarczyk, Hélène. 2003. Les paramètres aspectuels des situations
sémantiques. Études cognitives/Studia kognitywne 5: 11–34. Warszawa: SOW, PAN.
Verbal aspect in Slavic languages between semantics and pragmatics
Włodarczyk, André & Włodarczyk, Hélène. 2006. Semantic structures of aspect (A cognitive
approach). In Od fonemu do tekstu, prace dedykowane Profesorowi Romanowi Laskowskiemu,
389–408. Kraków: Instytut Języka Polskiego Polskiej Akademii Nauk. Lexis.
Włodarczyk, Hélène. 1994. L’Aspect verbal slave et les domaines du donné et du nouveau.
Études cognitives, Vol.1: Sémantique des catégories d’aspect et de temps, 113–130. Warszawa:
SOW, PAN.
Włodarczyk, Hélène. 1995. L’apport de la linguistique théorique à la traduction du polonais en
français: Le problème du temps et de l’aspect au passé. Les contacts linguistiques franco-
polonais, 13–27. Lille: Presses Universitaires de Lille. Travaux et recherches, Collection
UL3.
Włodarczyk, Hélène. 1996. Les traits sémantiques du sujet ‘anonyme’, en polonais, russe et fran-
çais. Semantyka a konfrontacja językowa: 179–198. Warszawa: SOW, PAN.
Włodarczyk, Hélène. 1997. L’Aspect verbal dans le contexte en polonais et en russe. Paris: Institut
d’Études Slaves.
Włodarczyk, Hélène. 1998. Wykładniki wartości informacyjnej wypowiedzenia w j. polskim i
francuskim (aspekt, okresloność, modalność), Congrès des Slavistes Cracovie 1998. Revue
des Études Slaves T. 70(1): 53–66. Paris: Institut d’Études Slaves.
Włodarczyk, Hélène. 2003. L’Aspect perfectif comme hypercatégorie (approche cognitive), com-
munication au XIIIe congrès des slavistes à Ljubljana en août 2003. Revue des Études Slaves
74(2–3):327–338. Paris: Institut d’Études Slaves.
Włodarczyk, Hélène. 2008. La place du temps dans la théorie cognitive de l’aspect (perfectifs
d’achèvement et d’interruption en polonais et en russe). In Le Temps construit, Mélanges
offerts à Jean-Paul Sémon, Jean Breuillard (ed.): 109–131. Paris: Institut d’Études Slaves.
Włodarczyk, Hélène. 2009. From ontological attributes to semantic feature structures – Experi-
mental research on aspect in Polish. Cognitive studies – Études Cognitives 9: 15–32.
Warszawa: SOW, PAN.
Xrakovskij, Viktor S. 1997. Mul’tiplikativy i semel’faktivy (problema vidovoj pary). Semantika i
struktura slavjanskogo vida II, Stanisław Karolak (ed.), 227–240. Kraków: Wyd. Naukowe
WSP.
The position in the utterance and the melodic
realisation of object and reflexive pronouns
in classical modern literary Russian
Olivier Azam
École Normale Supérieure, Paris / Université Paris-Sorbonne, Centre de
linguistique théorique et appliquée (CELTA)
The numerous studies on word order and intonation in modern Russian have
primarily focused on the “weightiest” elements in the utterance, appearing
to almost systematically ignore the pronouns. Yet these are among the most
common words in a language. The absence of rules governing their use, in a
language in which word order, though unfixed, is in no way arbitrary, poses a
serious problem for the non-native speaker.
By focusing on the position of object and reflexive pronouns, the current
study aims to respond to this absence of critical work by presenting, in the
light of the metainformative theory of centres of attention, the criteria which
determine the enunciator’s choice of a particular word order and prosodic
realisation. We generally distinguish two groups of pronouns complements
(differentiated only by the prosodic criteria in modern Russian, though they
are also morphologically distinct in Old Russian.) In place of this traditional
binary opposition (tonic pronouns versus atonic pronouns) which has limited
practical use, the current study proposes a more precise opposition of three types
of expressed pronouns – stressed, tonic and atonic – to which must be added
elliptical pronouns, particularly common in Russian. This new distinction allows
a more nuanced analysis of utterances, which in turn renders it possible to prove
the degree to which pragmatic strategy, word order, intonational curve and the
tonic of the latter are closely linked in modern lliterary Russian.
Although the issue of word order in Russian has been widely explored in linguis-
tics for many decades, and even though some rules, however clumsily formulated,
which govern this order have become a kind of mantra which is reproduced even
in handbooks for foreigners, it is evident that the problem of placing pronouns
in the utterance is almost never tackled in Russian linguistics. When it comes to
Olivier Azam
e lement order in the sentence, it is almost always a question of nouns and nominal
groups, so that it is very difficult to determine whether specific factors such as
rhythm or the balance of the sentence may play a role in the choice of pronoun
position in the utterance.
For foreigners it is striking that Russian uses far fewer object pronouns than
French, German or English. In this way, such commonplace sentences as “give it to
me” or “I told you it” can almost never be literally translated into Russian. Rather
than using two object pronouns (a possible, though rare, construction), Russian
only expresses one of these complements, judging that the context will suffice to
establish the second, or else it will replace the object pronoun with a demonstra-
tive. It may also express one of the complements by means of a substantive. The
primary difficulty is therefore to determine whether or not it is necessary to use
an object pronoun. This is a specific question which will be the subject of a spe-
cial study. Within the limits of this article, which forces me to limit my analysis
to those governing a single object or reflexive pronoun,1 my aim is to resolve the
double bind presented by the following stage, once it has been established that the
pronoun should be used: where should it be placed within the utterance and how
should one realise it on the melodic plane, or in other words, how should it be
pronounced?
2.1 The choice of the Gospels and the oral use of a written corpus
In these circumstances the choice of the Gospels as a corpus might initially seem
surprising. It undoubtedly presents some minor drawbacks.
The first, of which one must simply take note, is that the most widely accepted
modern Russian translation of the New Testament is the so-called Synodal Trans-
lation, which dates from the beginning of the nineteenth century. Now it seems
apparent that one of the aspects of the language which has significantly altered over
the past one hundred and fifty years is the position of object pronouns, de facto if
not de jure (I will return to this distinction), and even if this evolution appears less
radical than that which has affected the use of possessives (see Ch. Bonnot 2008).
In spite of all this, the Synodal Translation is indisputably w ritten in m
odern
. I will also leave aside particular structures whose word order is more or less fixed in the
corpus, such as participial phrases or non-nominative clauses (zero subject or impersonal
clauses such as instrumental clauses) in which the choice of the position of the pronoun is
subject to specific restraints.
Object and reflexive pronouns in modern Russian
. It is precisely this disjunction between the language of the Russian Synodal Bible and
the most contemporary usage which I wished to emphasise by deliberately conserving, to the
point of transliteration, the original spelling which is used as a visual signal. From a desire
to simplify I have however chosen not to transliterate the hard sign at the ends of words. It
will suffice to remember that all words which end, in Soviet and post-Soviet spelling, with a
consonant (other than yod) which is not followed by a soft sign, until the 1917–1918 reform,
finished with a hard sign.
. I have used a play on typographical conventions to illustrate certain prosodic functions:
the enclitic pronoun is in exposant; the simply tonic pronoun is rendered in italics, the stressed
pronoun in bold. Elements other than the pronoun which are emphasised may be in bold
too. Finally, where it has seemed necessary to give the wider context, the part of the quota-
tion which constitutes the actual example is underlined. In order to render this article more
concise, the examples have not been translated. Since each verse is accompanied by its Bible
reference (the number of the verse in bold at the beginning of the quotation, the chapter and
name of the evangelist in brackets at the end of the example) the reader can easily consult his/
her Bible translation of choice.
Olivier Azam
in no way undermine the conclusions elaborated here and only demonstrate that
certain pronominal positions are ambiguous and open to interpretation.
In spite of the limited drawbacks consonant with the choice of the Gospels,
this oral use of a prepared reading of texts which, though undoubtedly written,
nevertheless contain a large proportion of direct speech, seemed a more valid
option than to rely on ad hoc examples or on the point-blank interrogation of
Russophones on this or that utterance when the intonation does not seem evident.
Moreover the Gospels boast other advantages: three out of the four are synoptic.
The observation of pronouns can therefore be conducted in near-identical, yet
slightly differing contexts, which sometimes change nothing and sometimes add
or lack an infinitesimal detail which itself has an effect on the use of the pronouns.
These very similar contexts allow a considerable degree of refinement within the
analysis.
2.2 Elliptical, atonic, tonic and stressed object pronouns in modern Russian
By object pronoun (without further precision) I mean all object pronouns which
have a strong rection, in other words all object pronouns whose presence, at least
elliptical (often used in Russian), is necessary in order for the utterance to be com-
plete.4 This study will show that with regard to these object pronouns, whether or
not they are introduced by a preposition has no effect, in Russian, on their position
within the utterance.
The only object pronouns studied in this article are the reflexive and personal
pronouns.5 It is generally accepted that in modern Russian pronouns can be atonic
or stressed, though it must rightly be clarified that the pronoun may also not be
expressed at all. In reality, this study shows that the object pronoun in modern
Russian may occur in not three but four distinct forms: (1) it may not occur at all,
this would be termed an elliptical pronoun; (2) it may be present in the utterance as
an atonic or more precisely enclitic pronoun (since the enclitic pronoun retains its
lexical stress); (3) it may occur in a simply tonic form and finally (4) it may occur
in its stressed tonic form.
By enclitic pronoun I mean precisely that which Jean Breuillard refers to as
“the melodic enclitic position of the pronoun,” which is defined by the fact that the
unit in an enclitic position “cannot be spoken in a higher tone than the unit which
precedes it” (J. Breuillard 2008: 55).
. Thus in an utterance such as On podošël k nemu (he came towards him), we would be
dealing with an object pronoun (k nemu) just as in On uznal eë (he recognised her).
. For the sake of convenience the term “pronoun” without further precision will henceforth
refer to object and personal pronouns. In all other cases the type of pronoun will be specified.
Object and reflexive pronouns in modern Russian
. “Nexus” therefore has a different meaning here to that understood by Otto Jespersen, who
first used the term in linguistics (see A Modern English Grammar, 1909).
. This is also the meaning of the term “predicate” in the M.I.C. theory.
Olivier Azam
e xtralinguistic knowledge the speaker may have, combines at once both actual
information (that which has actually been spoken or written) and meta-informa-
tion. Yet it is at the same time the field of incidence as the speaker perceives it and
the knowledge s/he ascribes to the listener, (that is to say, still with reference to J.-P.
Sémon, to the subjective representation s/he constructs of the objective recipient
of the utterance) which motivate the pragmatic choices of the speaker, choices
which, in the Slavic languages, in turn govern word order and therefore have a
direct influence on syntax.
Finally the M.I.C. theory also has the merit of recalling the degree to which
the nebulous notion of “theme” is dangerous: too often it can be confused with
that of the subject.
As all of these criticisms seem entirely justified, M.I.C. terminology will be
adopted here. Thus, rather than “indivisible” utterances and “entirely rhematic”
sentences, I will discuss base utterances (M.I.C.), which is to say utterances in
which the nexal relations (established by the nexus with the different elements
which it controls and between these different elements themselves) are new or old.
The other utterances, those it was maintained could be divided into “theme” and
“rheme” are extended utterances in the M.I.C. theory: from a meta-informative
perspective, the speaker foregrounds a centre of attention through focalization or
on the contrary through topicalization, and sometimes through both.8 Focaliza-
tion consists in treating a centre of attention as new while a topicalized centre
is treated as given. However, in written texts, the distinction between focaliza-
tion and topicalization is not always easily grasped: for lack of a dedicated term
which would subsume both notions within the M.I.C. theory, I will simply refer to
“emphasis” when this distinction is dispensable. In the majority of cases however,
intonation allows one to distinguish between the two. Reusing the two adjectives
proposed by J.-P. Sémon (Sémon 1997) I will refer to tensive intoneme for the
topic and detensive intoneme for the focus.9
. In this case, the focalization of one element is accompanied by the topicalization of
another.
. In Russian the most common tensive intoneme (which the 1982 Grammar Book of the
Academy terms IK3) is not only ascending: the sharp rise produced on the syllable which
carries the centre of the intonation contour is followed, in the very next syllable, by a fall in the
voice which returns to a lower level than before the rise. The other intonation contours which
are rarer variants of IK3 (IK4 and in a more pompous style, IK6) are in fact mostly (IK4) or
entirely (IK6) ascending. In the vocabulary of the Grammar Book of the Academy, tensive
intonemes express nezaveršënnost’ (incompletion) and detensive intonemes zaveršënnost’
(completion). IK1 (neutral) or IK2 (a contrastive intoneme, an expressive variant of the pre-
vious term) in an enunciative sentence.
Object and reflexive pronouns in modern Russian
It is important to note that it is not only the constituents of the utterance, but
also the relationships maintained between these constituents through the nexus,
which can be new or old (already known). This precision is particularly important
when dealing with pronouns, since by definition a pronoun replaces a noun which
the recipient of the message should be able to restore. Except in the case of an
unusual effect, the referent of the pronoun is therefore always known. If the new
character, which means an utterance is extended, was only related to the identity
of its constituents, then the part of the utterance represented by the pronoun could
never, as it were, constitute the new element.
The definition of the canonical place of the object pronoun will serve as the prem-
ise for this study, which will present all of the potential variants in word order in
relation to this place.
certainty that the pronoun is in its canonical place when it is in a melodic enclitic
position directly after the verbal nexus which governs it.
(0) 15. gorjačka ostavila ee; i ona vstala i služila im. (Mt 8)
Nevertheless it may sometimes occur that the pronoun in a melodic enclitic posi-
tion is itself preceded by a subject pronoun which is also enclitic (modern Russian
like Slavonic Russian allows two successive enclitics).
(0bis) 6. Nа vsjakij prаzdnik otpuskаl on im odnogo uznikа. (Mc 15)
The phenomenon is exceptional: in this case the presence of the subject pronoun
would be disregarded and it would be considered that the object pronoun still
occupied its canonical place.
Having posed the question of the canonical place of the pronoun it remains
to stipulate the conditions of its use. As I have stated, it is a default position that
the speaker will use whenever s/he does not feel the need to displace the pronoun
or to modify its melodic realisation, that is to say, in concrete terms, in two types
of utterances:
3.2 Th
e displacement of the pronoun in an utterance containing a verb
governing a single expressed complement
In an utterance which only contains a single object complement of strongly
expressed rection, the pronoun, when it is not in its “canonical place,” may occupy
two locations:11 it can either be preposed, or postposed and stressed. The latter will
only be studied in the fifth segment, and I will begin here with the case of prepo-
sitioning (points 3 and 4).
When preposed to its governing verb, the pronoun complement in classical
Russian is always part of an extended utterance, never a base one.
. By the displacement of the object pronoun I therefore mean not only a change of place
but also any deviation in relation to the two criteria which define its canonical place: the dis-
placement and/or change in its melodic materialisation.
. When the context is sufficiently clear I will use the term place (“last place,” “change of
place”) when dealing simply with the “geographical” location of the pronoun in the linearity
of the utterance; but in ambiguous cases and if I wish to specify that I am not dealing with its
melodic realisation, I will rather use the words location or position.
Object and reflexive pronouns in modern Russian
The pronoun complement can remain in its canonical position when it appears in
an extended utterance in which it does not constitute the new element. However
it so happens that this is situated after the verb, and therefore towards the end of
the utterance.
Even if it is perfectly possible to place the new emphasised element of an
extended utterance after the postposed enclitic personal object pronoun (which
occurs when the pronoun is left in its canonical place) the speaker can chose to ret-
rograde the pronoun before the verb – the term is deliberately pejorative – in order
to completely free up the end of the sentence which is, as we know, the place of hon-
our, since in a neutral word order12 in Russian the old always precedes the new.
Thus displaced, the pronoun is no longer in the position of melodic enclitic. It
becomes tonic according to the definition given earlier without, however, becom-
ing stressed. It does not carry the centre of a neutral intonation contour at the
end of the sentence (of IK1 type), nor, a fortiori, is it the carrier of the centre of a
contrastive intoneme of type IK2 since the aim of its preposition here is precisely
to “render it forgettable” – to speak coarsely – in favour of a more important
element.
Examples 1–6 illustrate this kind of prepositioning of the tonic pronoun, in
which one can distinguish different cues for the emphasis of an element other than
the pronoun which justifies its displacement.
. In speaking of word order here I am only referring to the higher level of m
eta-informative
analysis (the paleophoric – old or already known – versus the neophoric group) and not to the
order of terms within each group (here the pronoun would be part of the paleophoric group).
Olivier Azam
Clearly if the focalisation affects the pronoun itself and not the infinitive then the
pronoun can be displaced; it will be stressed and pushed in front of the modal or
other verb controlling the infinitive (see eg. 22) or conversely stressed and post-
posed to the infinitive (see eg. 49).
4.2 P
reposed tonic pronoun in an interrogative utterance of empty
emphatic term
An even more telling indication of the emphasis of an element other than the
pronoun is the use of an empty emphatic term (an interrogative pronoun, adjective
or adverb formed from the root 〈*k〉) in an interrogative utterance, often accom-
panied by the preposition of the tonic pronoun.
(2) 46. Bože Moj, Bože Moj! dlja čego Ty Menja ostаvil? (Mt 27)
In this case of course one cannot assume a concern on the part of the speaker to
free up the last place for something else – since this location will remain unoccu-
pied – but perhaps a desire to disambiguate the utterance, above all if it is not spo-
ken but intended to be read: certainly the use of an empty emphatic does not, for
all that, exclude the possibility that another element of the utterance could carry
the highest meta-informative charge. Now if the pronoun were left in its habitual
location, one might believe (if one had not heard the original utterance spoken)
that it was the neophoric element and that it should be stressed.
In this corpus I have pointed out the preposed tonic pronouns used with
empty emphatic terms which mean that the focalization affects the action (eg. 2),
but also the object complement (Jn 8, 53), the means (Mc 9, 50) or the manner
(Mt 22, 17).
4.3 P
reposed tonic pronoun in an utterance with a topicalized or focalized
subject
Remaining with the interrogative utterance, but this time with a full emphatic
term, the emphasis may affect the subject followed by the interrogative particle li:
Object and reflexive pronouns in modern Russian
The emphasis which encourages the prepositioning of the object pronoun can also
focus on a stressed personal subject pronoun, which is particularly evident in a
negation when it directly affects the nominative pronoun, particularly in the case
of two independent elements coordinated by a which contain the same nexus: the
identity of the subject, denied in the first clause, is corrected by the second. In both
clauses, the subject pronoun carries the nuclear stress and the object pronoun is
preposed.
(4) 16. Ne vy Menja izbrаli, а Ja vаs izbrаl (Jn 16)
The emphasis of the subject can also take the form of the topicalization of the sub-
ject by means of the correlation kto…, (tot). Here again the pronoun is preposed
and tonic in both clauses:
(5) 11. On otvěčаl im: Kto menja iscělil, Tot mně skаzаl: voz′mi postel′ tvoju
i xodi. (Jn 5)
The use of i as a discourse or pragmatic marker affecting the subject (in the
following example i oni) is another means of underlining its emphasis, once again
through topicalization.
(7) 12. kogdа dělаeš′ oběd ili užin, ne zovi druzej tvoix, ni brаt′ev tvoix, ni
rodstvennikov tvoix, ni sosědej bogаtyx, čtoby i oni tebja kogdа ne pozvаli,
i ne polučil ty vozdаjanija. (Lc 14)
When the focalized subject is a nominal group and it appears in its normal place –
in last place in a neutral word order – the pronoun is naturally displaced and
preposed.
(8) 54. esli Ja Sаm Sebja slаvlju, to slаvа Moja ničto. Menja proslаvljaet Otec
Moj, o Kotorom vy govorite, čto On Bog vаš. (Jn 8)
In an expressive word order, the focalized subject nominal group can be brought
to the beginning of the utterance and affected by a contrastive detensive intoneme:
(10) 25. Otče prаvednyj! i mìr Tebja ne poznаl (Jn 17)
It is the wider context and above all the oral use of the corpus (the audition of the
recordings) which allows one to dissipate the ambiguity. In Jn 17, 25, the subject
is focalized and this insistence is strengthened by i which is not the coordinator
but has an adverbial value and means “also, and even”. In verse Jn 1, 10, on the
other hand, i is a coordinator with a consecutive value. In the second quotation
the subject is not focalized but topicalized, the focus affecting the nexus or, more
precisely, the nexus and the sign (“and though the world was made through him,
the world did not recognise him”). Whether there is focalization of the sign and
the nexus or focalization of the subject as in the first example, however, the result
is the same for the pronoun, which is preposed and tonic.
4.5 P
reposed tonic pronoun in an utterance with a focalized object
complement
It is hardly surprising that an object complement would be focalized: even the
recourse to a predicative construction betrays the importance the speaker places
on the ascribed quality (expressed here by the adjective odin “alone/only”):13
(13) 32. vy rаzsěetes′ kаždyj v svoju storonu, i Menja ostаvite odnogo; (Jn 16)
. In this example the emphasis could have affected the pronoun (through topicalization of
the object and the non-focalization of its complement): the pronoun would then have been
Object and reflexive pronouns in modern Russian
Far more exceptional is the use of a cardinal number object complement, a con-
struction impossible in English:
(14) 70. Iisus otvěčаl im: ne dvěnаdcаt′ li vаs izbrаl Ja? no odin iz vаs diаvol.
(Jn 6)
4.6 O
ther focalizations which entail a recourse to the preposed tonic
pronoun
The focalization can affect the temporal (eg. 15) or locative phrase (eg. 16).
(15) 18. Esli mìr vаs nenаvidit, znаjte, čto Menja prežde vаs voznenаviděl.
(Jn 15)
(16) 7. i pojdite skorěe, skаžite učenikаm Ego, čto On voskres iz mertvyx i
predvаrjaet vаs v Gаlileě; tаm Ego uvidite. Vot, ja skаzаl vаm. (Mt 28)
We have just seen one of the cases in which the object or reflexive pronoun can be
preposed: the speaker, at liberty to leave the pronoun in its canonical place, decides
to move it back before the nexus: the pronoun is then tonic but not stressed. This
occurs in an extended utterance whose informative charge is focused on an ele-
ment other than this pronoun.
The second case of prepositioning of the pronoun is also found in extended
utterances, but this time in those whose meta-informative charge is focused on the
pronoun itself. The speaker can then make the choice to prepose the pronoun to an
end diametrically opposed to that outlined in point 3: here the aim of the preposi-
tioning is not to “retrograde” the pronoun but on the contrary to emphasise it. In
this case, the speaker will not merely prepose the pronoun, s/he will also stress it,
that is to say will focus the intonation contour on it. The latter is usually contras-
tive, and it can be verified (infra point 6) that when the stressed pronoun precedes
the governing verb, the intoneme is usually tensive.
Examples 17 to 32 demonstrate the use of the preposed and stressed object
pronoun as well as the diversity of cues which can underline its emphasis. It is
unsurprising that many among them have already been mentioned before in rela-
tion to the emphasis, by focalization or topicalization, of an element other than the
pronoun. Naturally in the absence of formal cues, there remain many cases where
stressed and the sentence would have meant “As for myself, you shall leave me alone”. This was
not the choice of the readers of these texts however.
Olivier Azam
only the field of incidence allows us to determine that the meta-informative charge
is focused on the object pronoun.
Since this study is concerned with recordings rather than on the written
Gospel text, the classification of examples reflects finally the choices which the
readers were forced to make when faced with utterances which could be inter-
preted in two different ways. Here, however, ambiguity and hesitation can only
affect the choice between an extended utterance with a stressed pronoun and an
extended utterance whose informative charge is focused on an element other than
the pronoun: prepositioning to the nexus means confusion with a base utterance
is not possible.
5.1 Th
e “placing in parallel” and prepositioning of the stressed complement
pronoun
All kinds of “placing in parallel” – ranging from basic constriction to opposition
and passing through several degrees of contrastive comparison and confrontation –
encourage the emphasis or “selection of Centres of Attention” (M.I.C.) within the
utterance, and, among the different possible selections, the emphasis of the pro-
noun through focalization or topicalization.
This recourse to the stressed preposed pronoun is particularly frequent when
this type of “placing in parallel” occurs between two clauses (most often coordi-
nated or juxtaposed, more rarely dependant) within the compound sentence.
5.2 U
se in a compound sentence which contains independent elements
coordinated by no (topicalization or focalization of the object)
Thus the preposed stressed focalized pronoun is readily used in an independent
element coordinated by the adversative conjunction no with another independent
element of a different sign which corrects the information contained in the first
clause and more precisely the identity of the referent of the complement.
(17) 44. Iisus že vozglаsil i skаzаl: věrujuščij v Menja ne v Menja věruet, no v
Poslаvšаgo Menja; (Jn 12)
(18) 37. kto primet odno iz tаkix dětej vo imja Moe, tot prinimаet Menja; а kto
Menja primet, tot ne Menja prinimаet, no Poslаvšаgo Menja. (Mc 9)
Elsewhere, the focalization of the object pronoun affected by the negation can
even be reinforced by the use of the particle že (Mc 10, 40; Jn 17, 20).
It is also possible for the preposed stressed pronoun to appear in the clause
introduced by this aforementioned no. In the example given here the opposition
introduced by the adversative conjunction does not focus on the identity of the
object referent but on the entirety of the preceding clauses and the object pronoun
Object and reflexive pronouns in modern Russian
is not focalized but very clearly topicalized: the English translation of Lc 24, 24
(spoken with a distinct pause between ego and ne) would be “him they did not see.”
(19) 24. i pošli někotorye iz nаšix ko grobu i nаšli ták, kák i ženščiny govorili;
no Ego ne viděli. (Lc 24)
5.3 U
se in an extended utterance containing independent elements
coordinated by a (focalization or topicalization of the object)
The use of the adversative conjunction a assumes both topicalization and focal-
ization, since there is a reprise of an element from the preceding clause and the
correction of another element, the corrected element being focalized in the clause
introduced by a.
In this corpus, the preposed stressed pronoun only constitutes the focus of
a clause introduced by a in one example (whose topic is precisely a first person
preposed and stressed personal subject pronoun which is followed by a pause):
(20) 11. Ja uže ne v mìrě, no oni v mìrě, а Ja k Tebě idu. (Jn 17)
Everywhere else, the object pronoun itself constitutes the topic of the proposition
introduced by a. In the examples given the focalized element of this second clause
can be:
5.4 U
se in a compound sentence containing independent elements
coordinated by i (topicalization of the object)
In the following example where the conjunction i is used in a very similar man-
ner to a, the focalization of izbavim in the second clause is accompanied by the
topicalization of vas.
. Compare on the one hand with Example 1, in which the pronoun is tonic and situated
before the infinitive (but after the modal) and, on the other hand, with Example 49, where the
pronoun is accented and situated after the infinitive.
Olivier Azam
(24) 14. i esli slux ob ėtom dojdet do prаvitelja, my ubědim ego i vаs ot
neprijatnosti izbаvim. (Mt 28)
In spite of the slight nuance introduced by the change of conjunction (the idea
of opposition is toned down) for this analysis the example should be considered
together with those of paragraph 4.3.
5.5 U
se in a compound sentence containing juxtaposed independent
elements which have the same verb (focalization of the object reinforced
by i used as a discourse marker)
In spite of appearances the following example has only the prepositioning and stress
of the object pronoun in common with the previous one. Indeed here, even though
it is found at the beginning of the clause, i is used as a particle and not as a conjunc-
tion, the clauses are juxtaposed rather than coordinated. In this kind of coordinated
compound sentence where the two independent elements contain the same verb or
two synonymous verbs, the pronoun is no longer the topic but the focus:
(25) 1. Dа ne smuščаetsja serdce vаše; věrujte v Bogа, i v Menja věrujte. (Jn 14)
We will observe (infra point 5) that in this example it can also be postposed and
stressed.
However subordination can also allow the focalization of the object pronoun
without necessarily producing a topicalization of the subject. This is notably the
case in the setting of a conditional clause:
(28) 8. Ja skаzаl vаm, čto ėto Ja; itаk, esli Menja iščete, ostаv′te ix, pust′ idut;
(Jn 18)
5.7 F
actors which underline the focalization of the preposed pronoun
within the propositional frame
The justification for a recourse to the preposition of the stressed pronoun can be
found within the proposition itself, and several cues allow one to determine that
the pronoun is focalized here. These cues can be:
a. The presence of the adjective odin (“alone”) which serves to underline the
singularisation of the object;
(29) 10. Gospodu Bogu tvoemu poklonjajsja, i Emu odnomu služi. (Lc 4)
b. The presence, just before the pronoun, of an i used as an adverb and meaning
“also”:
(30) 45. Učitel′! govorja ėto, Ty i nаs obižаeš′. (Lc 11)
5.8 N
o formal cue underlining the emphasis of the preposed pronoun
complement; this is deduced from the field of incidence
It can also occur however that no formal cue underlines the emphasis of the pre-
posed pronoun in the written text. The only formal indication of its emphasis is its
oral accentuation: the detensive contrastive intonation contour is focused on it. In
a written text intended to be read aloud such as this corpus, the readers’ choice to
stress the pronoun (illustrated by the example below) suggests a decision based on
a prior meta-informative analysis of the field of incidence.
(32) 5. Sej est′ Syn Moj Vozljublennyj, v Kotorom Moe blаgovolenie; Ego
slušаjte. (Mt 17)
We have just seen (supra point 4) that in an extended utterance whose principle
meta-informative charge is focused on the pronoun (whether by topicalization or
focalization), the speaker can choose to accentuate it by preposing it to the nexus.
Olivier Azam
(33) 39. Est′ že u vаs obyčаj, čtoby ja odnogo otpuskаl vаm nа Pаsxu; xotite li,
otpušču vаm Cаrja Iudejskаgo? 40. Togdа opjat′ zаkričаli vsě, govorja:
ne Ego, no Vаrаvvu. Vаrаvvа že byl rаzbojnik. (Jn 18)
If the verb otpustit’ were reestablished in verse 40, where would the complement be
positioned in relation to it? As long as the pronoun remained stressed, the meta-
informative analysis would remain unchanged whether it came before or after.
More troubling still, chiasmi, often considered as simple stylistic forms, also
appear to confirm that only the accentuation of the pronoun is relevant. There
are numerous chiasmi in the Gospels and certain extracts which contain them,
already used to illustrate the prepositioning of the stressed pronoun, could equally
be used here to illustrate how it remains postposed. One can add the following
examples to the previously cited chiasmi (eg. 17–18 and 27–28):
(34) 37. Vse, čtó dаet Mně Otec, ko Mně pridet; i prixodjaščаgo ko Mně ne
izgonju von; (Jn 6)
. That is to say the fact that the pronoun is stressed in the sense I have defined it.
Object and reflexive pronouns in modern Russian
(35) 20. Pomnite slovo, kotoroe Ja skаzаl vаm: rаb ne bol′še gospodinа svoego.
Esli Menja gnаli, budut gnаt′ i vas; esli Moe slovo sobljudаli, budut
sobljudаt′ i vаše. (Jn 15)
(36) 48. i skаzаl im: kto primet sie ditja vo imja Moe, tot Menja prinimаet; а kto
primet Menja, tot prinimаet Poslаvšаgo Menja; ibo kto iz vаs men′še vsěx, tot
budet velik. (Lc 9)
(37) 16. Slušаjuščij vas Menja slušаet, i otvergаjuščijsja vas Menja otvergаetsja; а
otvergаjuščijsja Menja otvergаetsja Poslаvšаgo Menja. (Lc 10)
These chiasmi illustrate the two possible combinations: the preposed then post-
posed stressed pronoun (which is symbolized by A+P) and the postposed then
preposed stressed pronoun (P+A). As I will show however chiasmi are rarely lim-
ited to their most basic expression. In general they appear in complex combina-
tions such as P+A | P+P (supra eg. 27 Jn 13, 20); A+P+P (Lc 9, 48) or even in a
series of parallel and sequential chiasmi (P+A | P+A followed by a P+P combina-
tion in Lc 10, 16) or symmetrical and embedded chiasmi of P+A+P type in Jn 12,
44 (supra eg. 17) or P+A | A+P – a chiasmus within a chiasmus – in Mc 9, 37
(supra eg. 18). An important detail in this corpus is that the only isolated chiasmi
are of A+P type and they are comparatively plentiful (Jn 18, 8; Jn 15, 20 and Jn 37,
6): this frequency and the absence of isolated chiasmi of P+A type can be easily
explained by the ambiguity of the postposed pronoun in the written text as men-
tioned. When such a pronoun appears in a clause which follows a clause where it
was preposed and stressed, the reader, undoubtedly deprived of acoustic support
but alerted by the first prepositioning of a stressed pronoun, could easily see the
effect of the chiasmus and deduce from this that the postposed pronoun of the sec-
ond clause should also be stressed. On the other hand, if the context is not abso-
lutely unequivocal, it is harder to immediately discover a chiasmus of P+A type
on the first reading: glancing over the first clause, the reader may not notice that
the postposed pronoun is not in its canonical place but stressed, and only if the
chiasmus is involved in a wider play of oppositions (Jn 13, 20; Lc 10, 16; Jn 12, 44;
Mc 9, 37) will the reader discover, sometimes a posteriori, that the first postposed
pronoun should also be stressed.
In spite of these common sense restrictions on the use of isolated chiasmi
(P+A), the very existence of the chiasmi and their diversity appears to confirm
what the utterances with elliptical nexus and stressed pronoun suggested (eg. 33):
the relative place of the latter in relation to the nexus is not important as long as
the pronoun is stressed. It would therefore appear that we are dealing here with two
almost identical variants, what traditional Russian grammar complacently terms
ravnopravnye varianty. Linguistics will not be satisfied by the idea that there could
be no difference in a language between two distinct forms of expression however,
nor will it settle for admitting that the difference could be merely “stylistic”.
Olivier Azam
Here, however, a second cue is added to the first: the coordination of the pro-
nominal complement with a nominal group of the same function i o dětjax vašix
(see. infra 5.5.d). Again in this example one finds a chiasmus. There are several
other examples of stressed postposed pronouns with no in the chiasmi already
quoted.
6.2 U
se in a compound sentence containing independent elements
coordinated by a (topicalization of the object)?
Though the examples are relatively numerous when the stressed pronoun is pre-
posed, there is only one example where the pronoun is apparently postposed.
(39) 28. Tаm budet plаč i skrežet zubov, kogdа uvidite Avrааmа, Isааkа i Iаkovа i
vsěx prorokov v Cаrstvii Božiem, а sebja izgonjaemymi von. (Lc 13)
In fact the postpositioning is illusory: if one considers the entirety of the meta-
sentence (or coordinated compound sentence) in linear form, uvidite does indeed
precede sebja, but they do not appear in the same clause; uvidite is not reused in
the second clause and sebja is the complement of an elliptical nexus. As above
in the example of Jn 18, 39–40 (eg. 33) it is absolutely impossible to say whether
Object and reflexive pronouns in modern Russian
sebja should precede or follow the nexus in the case of the restoration of uvidite.
As we will see later (see infra point 6) sebja will in fact probably be preposed, from
whence the question mark which accompanies the title of this subheading.
6.3 U
se in a compound sentence containing juxtaposed independent
elements containing common elements (focalization of the object)
The common element here is the noun mir (“peace”):17
(40) 27. Mir ostаvljaju vаm, mir Moj dаju vаm; ne tаk, kаk mìr dаet, Ja dаju vаm.
(Jn 14)
6.4 U
se in a compound sentence whose clauses are linked
by subordination
Here again the topicalization of the subject often accompanies the focalization
of the object expressed by the stressed postposed pronoun. This topicalization is
obtained by the use
(41) 40. Kto prinimаet vаs, prinimаet Menja; а kto prinimаet Menja, prinimаet
Poslаvšаgo Menja; (Mt 10)
As seen with the preposed stressed pronoun, subordination can also allow its
focalization when it is postposed without necessarily producing a topicalization of
the subject. This is particularly evident with a conditional clause:
Of course here as when the pronoun is preposed (see eg. 4.6), the use of a con-
ditional clause in itself does not provoke a recourse to the stressed form of the
pronoun, but it demonstrates that the presence of the former generally indicates
focalization. Here (eg. 43) the condition which must be fulfilled is that the action
expressed by the subordinate verb affects a specific object (the conditional clause
then underlining the importance of the selection of the object by focalization in
order to accomplish the action described in the main clause).
6.5 F
actors underlining the focalization of the postposed pronoun
within the propositional frame
The pronoun is still focalized here. The cues which accompany its focalization
can be:
c. double negative ni
The place of the negation just before the complement pronoun is an obvious sign
of its focalization. With ni – a double negation – however, the cue itself is also
double because ni implies in addition the coordination of the stressed pronoun
with a nominal group of the same function (see 5.5.d).
(46) 19. vy ne znаete ni Menja, ni Otcа Moego; (Jn 8)
d. When the stressed pronoun is postposed, one often finds a cue as to its focal-
ization which was not found when it was preposed: the presence of a second
complement in the same case.
This second complement can be coordinated with the stressed pronoun. It may be
a question of a second stressed pronoun:
(47) 9. i zvаvšij tebja i ego, podošed, ne skаzаl by tebě: ustupi emu město; i togdа
so stydom dolžen budeš′ zаnjat′ poslědnee město. (Lc 14)
Or else a coordinated nominal group:
(48) 33. tаm rаspjali Ego i zloděev, odnogo po prаvuju, а drugаgo po lěvuju
storonu. (Lc 23)
Finally, the second accusative noun phrase could simply be apposed to the stressed
complement pronoun:
(49) 40. A teper′ iščete ubit′ Menja, Čelověka, skazavšago vam istinu, kotoruju
slyšal ot Boga. Avraam ėtogo ne dělal. (Jn 8)
Object and reflexive pronouns in modern Russian
It is evident in this last example that the pronoun is the complement of an infini-
tive and its place should be compared with the one it occupies in Example 1, on the
one hand, and 22, on the other.
6.6 N
o formal cue underlining the emphasis of the postposed pronoun:
This is deduced from the field of incidence
As when the stressed complement pronoun is preposed, it can also occur that no
textual cue underlines its emphasis when it is postposed, with a greater concurrent
risk however. This risk is that the field of incidence may not be unequivocal enough
and will lead to an erroneous meta-informative analysis which would result in a
pronoun which should be stressed being read as a postverbal enclitic pronoun. In
fact it is precisely when focalization is not accompanied by easily identified textual
cues such as those I have described that readers hesitate and their interpretations
most often diverge.
As regards the classification of quotations under this heading I have therefore
followed a single reader (the same for all the examples), but this choice and the
divergences noted do not undermine the findings. On the contrary, they serve to
prove (1) that written utterances with postposed complement pronouns are struc-
turally ambiguous if no textual cue confirms that the pronoun focuses on itself
the highest meta-informative charge; (2) that it is indeed the intonation, (and not
the place of the pronoun) which plays the most important role, since without it,
doubt over the emphasis of the pronoun remains and (3) that the postpositioning
of a stressed pronoun is possible in spite of everything, even in the absence of cues
and despite the risk of confusion when written between the stressed postposed
pronoun and the pronoun in its canonical place.
In the following example the focalization of the object goes hand in hand with
the topicalization of the subject, affected by a tensive intoneme and separated from
the nexus by a pause:
In Mc 2, 14, on the other hand, the reader’s decision to focalize the complement
was less foreseeable:
(51) 14. Proxodja, uviděl On Levija Alfeevа, sidjaščаgo u sborа pošlin, i govorit
emu: slěduj zа Mnoju. I on, vstаv, poslědovаl zа Nim. (Mc 2)
The reader could make this choice because he was reading a written text… As
soon as the text is spoken, however, no further doubt is possible: thanks to the
intonation the structurally ambiguous utterance becomes an extended utterance
with focalization of the pronoun.
Olivier Azam
(52) 28. Vy slyšаli, čto Ja skаzаl vаm: idu ot vаs i pridu k vаm. (Jn 14)
One might indeed ask why, in the first clause (idu ot vas) the reader decided to
accentuate the pronoun, when he has the nuclear stress fall on the verb in the
second where the pronominal complement (k vam) is no more than an enclitic
occupying its canonical place. The reason for this choice is based on the change of
verb. What distinguishes the two clauses – and thus decides where the emphasis
is placed – is the direction of the motion. Now with a displaced verb without a
preverb such as idu the mono-oriented displacement can occur in several direc-
tions; this would then be expressed by the complement ot vas which would be
focalized, from whence the accentuation of the pronoun. On the other hand, in the
second clause, the direction of the mono-oriented displacement is already given
by the verb itself or, more precisely, by its preverb. The complement k vam there-
fore teaches us nothing more than 〈pri〉 already taught us, and there is therefore no
reason to focalize it.
7. T
opicalization or focalization of the preposed and postposed stressed
pronoun?18
As feared, the classification of cues which accompany the prepositioning and post-
positioning of the object pronoun are surprisingly similar… And yet on closer
examination it is possible to distinguish a fundamental difference: while the
examples of the prepositioning of the stressed object pronoun gathered together
examples of topicalized and focalized pronouns, it is striking to find that there are
no examples of a postposed topicalized pronoun. In this respect, a comparison of
points 4.2 and 5.2 is revealing: when used in a compound sentence which contains
independant clauses coordinated by a, all of the stressed pronouns are topical-
ized and are relatively numerous. Yet I have been able to establish that with a no
stressed postposed pronoun is found. It is scarcely likely that the total absence of
postposed stressed topicalized pronouns in the entirety of the Gospels is down
to chance; undoubtedly we hold the key here to the principle difference between
. I would particularly like to thank Ch. Bonnot and H. Włodarczyk for having set me on
this path by suggesting that I explore the distinction between emphasis through topicalization
and focalization.
Object and reflexive pronouns in modern Russian
. The concept of “expressivity” in a variant of word order is of course only admissible if
this term is understood in a relative sense: “expressive” here means “more expressive” than
another, since all word orders necessarily “express” something.
. I will willingly take as a comparative element an extended utterance whose focus car-
rying the contrastive accent otec is a noun filling a different syntactic function – here, that
of subject – in order to underline that neither the syntactic function of the pronoun, nor its
pronominal nature have any bearing here on my argument.
Olivier Azam
beginning of the utterance – and a fortiori a contrastive accent at the end of the
sentence – is less usual than a detensive intoneme at the beginning of the utterance,
the surprise it generates is therefore of a nature to more easily capture the attention
of the listener and fix it on the pronoun which carries the centre of this detensive
contour: this is what leads me to conclude that the prepositioning of the pronoun
constitutes the “most expressive” variant of word order when the stressed pronoun
is focalized.
8. C
onclusion: The position and melodic realization of pronouns in base
and extended utterances
The ultimate goal of the analysis of this corpus was to define the canonical or
default place of the pronoun: that which it occupies when it is placed just after the
nexus in the position of a melodic enclitic. I have in the first place asked the reader
to accept the veracity of this conclusion by presenting it as a principle which was
taken as the starting point and basis of this study.
Undoubtedly, by definition, a principle cannot be proven. It is perfectly pos-
sible to prove however that its negation is false: for this the reader must simply
examine each of the other possible realizations of the pronoun successively (pre-
posed tonic, preposed or postposed stressed) and to reexamine the entirety of this
study in order to determine that none of them could be seen as the default position.
Having established the definition of canonical place, from then on it was pos-
sible to present an algorithm which follows the rules which allows one to deter-
mine both the positioning and the melodic realization of the personal and reflexive
pronouns in an utterance which only contains a single expressed complement.23
. Although it seems reasonable to believe that the rules given below can be applied to the
entirety of Russian classical literature, their validity will nevertheless only be vouched for in
relation to the corpus studied here: the Synodal Translation of the Gospels.
Olivier Azam
b. where the emphasis is focused on the pronoun, the latter is not just tonic
(that is, non-enclitic) but stressed (bearing the centre of a contrastive
intoneme).
–– If the emphasised pronoun constitutes the topic of the utterance, it will be
preposed, bearing a tensive intoneme and usually followed by a more or
less perceptible pause.
–– If the pronoun constitutes the focus of the utterance, it will
–– bear a detensive intoneme and usually be postposed to the nexus (in
an unmarked word order)
–– but still bearing the centre of a detensive intoneme, it can also be pre-
posed to the nexus (in a more expressive word order), and in this case
it cannot be followed by any pause.
Translation from French by Dúnlaith Bird
References
Bonnot, Christine. 2008. Un cas ‘d’inversion’ de l’ordre canonique en russe moderne. La post-
position du pronom possessif épithète. In Questions de linguistique slave. Études offertes
à Marguerite Guiraud-Weber, 39–54. Aix-en-Provence: Publications de l’Université de
Provence.
Breuillard, Jean. 2008. Un cas d’enclise du sujet en russe. À propos des phrases du type Perevo-
dila Irina bystro. In Questions de linguistique slave. Études offertes à Marguerite Guiraud-
Weber. 55–65. Aix-en-Provence: Publications de l’Université de Provence.
Krylova, Ol′ga & Xavronina, Serafima. 1986. Porjadok slov v russkom jazyke. Moskva: Russkij
jazyk.
Lehfeldt, Werner. 2003. Akzent und Betonung im Russischen, 2nd edn. München: Otto Sagner.
Russkaja grammatika.Tom I. 1982. 99–122 (§155–171). Moskva: Nauka.
Sémon, Jean-Paul. 1995–1996. Interrogation et informativité. Paris: Cours polycopié de prépara-
tion à l’option linguistique de l’agrégation de russe à l’université Paris-Sorbonne.
Sémon, Jean-Paul. 1996. Imparfait français et prétérit imperfectif russe. In Études russes:
mélanges en l’honneur de Louis Allain, 205–216. Lille: Presses universitaires du Septentrion.
Sémon, Jean-Paul. 1997. Lexique de la chronopoïèse. Paris: service de reprographie de l’université
Paris-Sorbonne.
Włodarczyk, André. 2004. Centres d’intérêt et ordres communicatifs. In Enoncer, l’ordre infor-
matif dans les langues, Pierre Cott, Martine Dalmas & Hélène Włodarczyk (eds). Paris:
L’Harmattan.
Włodarczyk, André & Włodarczyk, Hélène. 2006. Focus in the meta-informative centering
theory. In La focalisation dans les langues, André Włodarczyk & Hélène Włodarczyk (eds).
Paris: L’Harmattan, collection Sémantiques.
Włodarczyk, Hélène. 2008. Du bon (et du mauvais) usage des formes accentuées et atones des
pronoms personnels en polonaise. In L’Enseignement du polonais en France, IIes Assises.,
39–48. Paris: Société française d’Etudes Polonaises, Institut d’Etudes Slaves.
Accented and unaccented pronouns
in Ancient Greek
A pragmatic choice by the speaker
Jean-Christophe Pitavy
Université Jean Monnet, Saint-Etienne
seems to confirm that the use of the accented personal pronoun in conjunction
with the verbal form allows the speaker to construct an extended utterance.
The verification of this mechanism requires a meticulous examination of the
meta-informative structure of a very large number of examples of usage. There are
obviously exceptions, notably – and I have cited some of these – as regards certain
formulaic structures or those which are inserted at the level of meta-informative
commentary. P1 personal pronouns which engage the speaker as the necessary,
privileged reference point, are commonly introduced in such cases. Moreover, it
is not enough to examine the meta-informative structure of each usage in order
to decide whether we are dealing with topicalisation or focalisation. We must
establish more clearly the relationship between the choice apparently made by
the speaker and the processes supporting this structuring: certain parameters –
purely syntactical (the order of words) and pragmatic (particles) – have been
called upon in relation to pronominal marking. Beyond, however, presenting the
elements of the complexity of the situation in ancient Greek (elements arising
from both the uses of the language and the historical and material conditions
affecting access to the corpus), the aim has been to show the advantages of the
Meta-Informative Centering Theory in evaluating the use of certain strong
pronominal forms. Where it is recognised that these pronouns form part of an
extended utterance, it has proved possible and easier to differentiate between
focalisation and topicalisation.
0. Introduction
. For the framework and corresponding terminology (centre of attention, focus, topic,
extended utterance…), I rely, unless otherwise stated, on the Meta-Informative Centering
theory as presented in Włodarczyk (2004) and Włodarczyk and Włodarczyk (2006).
. In this study I will not enlarge upon the particular conditions of usage and description
which apply to a language that is no longer spoken, and for which we cannot use the assess-
ments and other tests of acceptability available for modern spoken languages. These difficul-
ties can be partially compensated for by taking into account the variety of uses shown and by
the special place accorded to hypothesis.
To this we must add the representative nature, with regard to usage, of the literary
corpuses and the problem of the very value of these texts as testimony, since they have been
transmitted to us over a long period of time and have been subject to numerous accidents
Accented and unaccented pronouns in Ancient Greek
investigate the organisation of the different means offered by the language to mark
the category of person, inviting the elaboration of a theoretical hypothesis.
I will draw upon certain points – of two kinds, in language and in discourse –
as regards the use of accented pronouns in classical Greek that can be ascertained
from the corpuses. The linguistic field at the level studied here is characterised by
complex organisation and uneven illustration.
The corpus used consists of a collection of classical Greek prose: Plato’s dia-
logues and a choice of orators (mainly Lysias and Demosthenes). As the study
deals with P1 and P2 pronouns,3 it is obviously Plato’s texts which will provide
most of the examples.
along the way. I will not touch upon these general points, other than the last one, which has
consequences for the study of certain pronouns.
. Cf. Abbreviations used at the end of this chapter.
. One may add that these referents illustrate the “meta-informative” meaning of a feature as
new, for example, since the speaker, an indispensable reference point for the discourse, cannot
be unknown.
Jean-Christophe Pitavy
1.2 Hypothesis
The initial idea is to use the propositions of the theory of centres of attention to
account for certain uses of accented pronouns in Ancient Greek. Rather than pro-
posing that the accented (tonic) form represents a focalisation (or an emphatic
structure, a different though similar operation), or inversely, that we are deal-
ing with topicalisation, in fact rather than considering that the meta-informative
structure is unrelated to the pronominal morphology, I prefer to acknowledge, by
following the distinctions of the theory mentioned earlier, that the accented pro-
noun may represent either a centre of attention referring to a topicalised speech
participant; or a centre of attention referring to a focalised speech particpant.
As a basic principle it is proposed that since the choice of the accented rather
than the unaccented form is not without importance, this choice is not made with-
out consequence or purpose, but is motivated by the meta-informative structure
of the utterance. As in many languages, it is not always easy to determine the fac-
tors behind the use of accented forms rather than their unaccented equivalents.
I suggest that, when the speaker has a choice between two forms (in cases where
the corpus allows us to see differentiated and alternating forms), the use of the
accented form corresponds in certain cases to an extended utterance structure,
and thus an utterance that is marked from the point of view of the opposition
between one part referring to “old” information and one part referring to “new”
information. In other words, this sets up a formal contrast – one that is, as it hap-
pens, morpho-phonological and syntactical. Working out whether one is dealing
with a topic or a focus brings into play numerous criteria which go beyond the
simple use of pronouns.
The different processes pose numerous and varied problems which clearly cannot
here be given the development that they deserve, as this would risk modifying
considerably the purpose of this study which concentrates on pronominal mor-
phology. Furthermore, we cannot take all of these processes into account in the
same way. Intonation marks in Ancient Greek, for example, are not at all well
understood. The corpuses at our disposal include signs of accentuation, added
Accented and unaccented pronouns in Ancient Greek
to the texts several centuries after their composition.5 These signs are thus only
the tip of an iceberg, about which we know almost nothing and for which we are
reduced to hypotheses about the system and its likely consequences in discourse.
At the same time, it goes without saying that the different processes are interde-
pendent and that for a variety of reasons none of them will be completely ignored.
Intonation, for example, is clearly linked to the other processes mentioned: the
word order (the place of unaccented units is determined by those that have an
accent), particles and the use of certain morphemes. As we will see, accents are a
deciding factor at certain levels of morphosyntactical organisation (certain mor-
phemes are made up on two series of forms, accented and unaccented).
Other criteria, which I would like to illustrate here, invite us to take account
of the relationships between the centres of attention organised in the utterance in
question and the information available co-textually. In my view this kind of exter-
nal relationship for the extended structure concerns the coherence of the discourse.
P1 and P2 pronouns, which have two series of forms which I will provisionally
call “weak” and “strong” at this initial stage;
P3 and P6, which have no specific personal pronoun;6
. It is generally considered that the accentuation and punctuation of classical texts appeared
at the same time, both normally attributed to the Alexandrian grammarian Aristophanes of
Byzantium, in the 3rd century BCE. See Ildefonse (1997: 21).
. This is true for the language system. The situation in discourse is in reality more complex.
For different grammatical functions classical Greek, like other Indo-European languages such
as Latin, relies on various forms of demonstratives (deictic and anaphoric forms from the
Jean-Christophe Pitavy
P4 and P5, for which the contrast between accented and unaccented forms is a
theoretical one; it is assumed to exist at the language level by grammarians, but is
badly represented in the corpuses (where accented forms mainly dominate). This
contrast, virtually neutralised in discourse in classical prose, is inaccessible and
thus cannot be used in analysis.
2.2.1 Similarities
P1 and P2 have two characteristics in common:
2.2.2 Differences
For P2 (Table 1), the opposition is only marked from the point of view of the tone
(a syllable of identical structure is either accented or unaccented):
accusative se sé
genitive sou soû
dative soi soí
differentiates the two series. Originally accents were not noted in the written form:
texts have thus often been given accents several centuries after their composition,
according to criteria imposed by successive editors. This historical peculiarity
means that we cannot trust the available forms of this pronoun as much as we can
trust the others. The contrast between the two series must be retained in discourse
by means of other, more or less complex criteria, relying upon the presence of a
certain kind of constituent, for example, or the position of the pronoun.
For a hypothetical understanding of the use of one kind of form, we can now
formulate the principle of the symmetry of value and function between the equiva-
lent pronouns of different persons. We can work out the distribution of unaccented
pronominal forms based on the more trustworthy distribution of the P1 pronoun.
In fact, for P1 (Table 2), the contrast from the point of view of tone is coupled
with that of the series of forms that I will here call “short” and “long” (this termi-
nology is not commonly used). This difference, which is actually morphological
(formal structure, number of syllables) is usually better preserved in the texts that
have come down to us:
accusative me emé
genitive mou emoû
The opposition between the two series is normally more reliable in the
corpuses.
For both personal pronouns, the unaccented form is obviously clitic (Creissels
1995: 24–28; Givón 1984: 83; Riegel et al. 1994: 200–204; Schiering 2006: 17–20;
Zwicky 1996: xii, xvii). There is therefore for the same information (the person)
“strong” and “weak” marking, in spite of the different means used:
P1: “long” accented form v. short clitic form;
P2: accented form v. clitic form.
In both cases (Table 4), there is an accented (“strong”) pronoun form in one of
the contrasted terms instead of minimal marking (verbal inflection for the subject,
clitic form for the object):
An important difference between the nominative and other cases can be seen
in the fact that there is a redundant structure for expressing the subject: the subject
constituent is marked twice, that is by the verb form and a separate pronoun.7
We may have reservations about likening the marking potential in one case to
that of another, but
. A classic and well established analysis of pro-drop languages, suggest that the use of the
subject pronoun in this case is deliberate because there is “emphasis”.
Accented and unaccented pronouns in Ancient Greek
We therefore find ourselves dealing with a typical case, well attested in languages
in which one can choose not just a base utterance, but an extended one. Here, the
pronoun is the ingredient which allows the utterance to be extended.
Italian, a pro-drop language, is known for its pragmatic exploitation of the
use of the subject pronoun, combined with other parameters such as word order.
It thus sets up, for example, a contrast between three types of utterance in the
following way:
(2) (a) l’ho fatto, “I did it”, simple utterance;
io, l’ho fatto, extended utterance in which the subject pronoun is moved
(b)
to the left and plays the role of topic, roughly equivalent to the English
“As for me, I did it” (French “moi, je l’ai fait”);
Jean-Christophe Pitavy
l’ho fatto io, extended utterance in which the same pronoun is moved,
(c)
this time to the right, and plays the role of focus “I’m the one who did
it” , “I did it” (“c’est moi qui l’ai fait”).
We can of course compare the different solutions adopted by English and French,
which are non pro-drop language and combine a separated pronoun with syntac-
tical structures that also involve the order of the constituents (cf. Dik 1989: 131;
Lambrecht 1994: 115–116).
It goes without saying that these different processes are not enough in them-
selves: it is not enough to use structure (b) or (c) for valid topicalisation or focali-
sation. It is the coherence with the elements of the universe of discourse which
makes one or another type of meta-informative structure possible.
If we therefore say that the use of the strong form of a personal pronoun allows
for either the topicalisation or the focalisation of the constituent representing the
speaker or the addressee, the first criterion that determines the nature of the meta-
informative structuring is a recognition of the relationships of coherence with the
co-text or the immediate context.
Socrates proposes a change of role both for Protagoras (he is to ask the questions
rather than answer them) and for himself. The new information, for each one
of the stages, is not the identification of who must reply (answer: me), but what
the speaker (I) must do. The recognition of the semantic-pragmatic relationship
of this utterance to the co-text thus allows us to interpret the value of the two
Accented and unaccented pronouns in Ancient Greek
personal markers as representing the two parts of the topic.8 This is also the case
in the following example:
The information that forms the basis of the topic is given in the preceding sen-
tence, and thus constitutes a shared topic (P4: “we [fell in love]”) which is split
into two parts in the following sentence: the pronoun “we” is taken up by the two
pronouns (egṑ “I”, hoûtos “the latter”). Each member of this pair (he, I) has his own
comment.
The following example is a little more complex:
SŌ. […] eboulómēn gàr án moi toùs lógous ménein kaì akinḗtōs hidrûsthai
(5)
mâllon ḕ pròs têi Daidálou sophíai tà Tantálou khrḗmata genésthai. Kaì
toútōn mèn ádēn;
epeide dé moi dokeîs sù truphân,
conj. ptc P1d you-seem P2N to weaken
autós soi sumprothumḗsomai deîxai hópōs án me didáxē(i)s perì toû hosíou.
(Plato, Euthyphro, 11e1)
“So.: […] I would prefer my arguments to remain in place, that they should
be firmly fixed, rather than receive in addition to the wisdom of Dedalus,
the wealth of Tantalus. But enough of that. Since it seems to me that you
are weakening, I will join my efforts to yours to show you how you could
teach me about holiness.”
The subject pronoun is here used to return to the addressee (Euthyphro), after an
expression putting an end to further development. One can thus assume that he
is putting together a new topic: the sentence reintroduces Euthyphro (you), who
becomes (again) an object of worry for the speaker (Socrates).
. Note the parallel structure used for each of the participants (he, I):
These few examples may be enough to illustrate the fact that it is not the use
of the subject pronoun itself, together with the verb form, which determines the
value of topic or focus. Such uses of pronouns seem to show that the pronoun
marks the topic according to the meta-informative structure – in this case the
available co-textual information about the referent corresponding to the pronoun.
The addressee (sú “you”) appears as the element modifying the descriptions in
the preceding sentence (spoken by Socrates). Cebes corrects their roles, making
Socrates the one who must reply (the verb is P2 singular). He does not tell us that
there will be a reply to the question, but he indicates who will be responsible for
discovering it.
This kind of focalisation is common in Plato’s dialogues, notably when dealing
with the delicate task of replying or defining who should take on a specific role. In
the following example, on the other hand, there is no informative change as far as
either the process or the role of one of the participants is concerned:
(7) SŌ. […] ei gàr ek mèn tôn állōn tà zônta gígnoito, tà dè zônta thnḗiskoi, tís
mēkhanḕ mḕ oukhì pánta katanalōthênai eis tò tethnánai? – Oudè mía moi
dokeî, éphē ho Kébēs, ô Sṓkrates,
allá moi dokeîs pantápasin alēthê légein.(Plato, Phaedo, 72 d 5)
conj P1D you-seem absolutely true to say
“So. […] if in fact living creatures come from other living creatures, and
living creatures die, is there any escape from the final result that all things
would end in death? – None, in my opinion, Socrates, said Cebes, on the
contrary, you seem to say what is perfectly true.”
Accented and unaccented pronouns in Ancient Greek
The verb form is used with no personal subject pronoun (Ø dokeîs légein “you
seem to say”). We have here a confirmatory utterance: in the final phrase, Cebes
brings nothing “new”, either about what is happening or about the role that
Socrates (“you”) is playing.
The case of an utterance involving focalisation because the variable of a con-
stituent is changed, can be illustrated by the following example:
(8) SŌ. Tí oûn tis ereî pálin ex arkhês epistḗmēn? ou gár pou aperoûmén gé pō? –
THEAI. Hḗkista, eánper me sú
conj neg P2N
ge apagoreúēis. (Plato, Theaetetus, 200 d 6)
ptc you-give-up
“So.: Therefore, going back to the beginning, how can we define science?
For we are not going to give up, are we? – theaetetus: Certainly not, at
least as long as you yourself don’t give up.”
The change (new information) involves the agent of the process “giving up” (apag-
oreúein: the two speakers use different forms of the same verb). The change is from
P4 to P2. Theaetetus introduces a stark contrast between the continuation of the
conversation between the two of them and the decisive, marked role played by
Socrates alone.
This change of person in what is an unchanging process can be seen in the
following injunctions:
(9) SŌ. Ê kaì hēúreté ti toioûton?
THEAI. Émoige dokoûmen; skópei dè
P1D+ptc we seem see ptc
kaì sú.(Plato, Theaetetus, 147e4)
also P2N
“So.: And have you found something [comparable to the roots]? – THE.:
I believe in fact that we have found something; but see if you also agree.”
The unchanging element here is the judgement asked for (in Socrates’ question:
“have you found…?”) and expressed (in Theaetetus’ reply). But this reply includes
a variable which corresponds to the person responsible for the stated point of view
(the subject of the judgement: “I believe…”). Theaetetus’ reply thus creates a con-
trast between two opinions: that of the speaker (émoige “[it seems] to me”), and
the one that is demanded of Socrates (skópei sú “you, see”).
(10) SŌ. Tò erōtṓmenon ou pánu moi phaínetai apokrínesthai. –
GOR. Allà sú, ei boúlei, eroû autón. (Plato, Gorgias, 448d6)
conj P2N if you-want ask P3A
“So.: I have the impression that [Polus] is not replying to what he is asked. –
GOR.: Well, question him yourself, if you want.”
Jean-Christophe Pitavy
In both cases, the person in discussion with Socrates focalises him in relation to
a previously established background that is maintained (examination of a result,
questions). In the following example, P1 is opposed to P3 as the essential piece of
information modified in relation to an identical predicate:
(11) Kakeînos adikeîn mèn hōmológei, ēntebólei dè kaì hikéteue mḕ apokteînai all’
argúrion práxasthai. Egṑ d’ eîpon hóti
“ouk ego se apoktenô, all’ ho póleōs nómos…”
neg P1N P2acc I-will-kill but art city Gsg law Nsg
(Lysias, On the murder of Eratosthenes, 25–26)
“And recognising that he was guilty, he asked me, begged me not to kill him,
but only to ask him for money. I said to him “It is not I who will kill you,
but the law of the City.”
The speaker can also produce an extended utterance to emphasize the importance
of a role (played by the speaker himself, in the following example):
These different examples are obviously carefully chosen; in many cases, the use of
the strong pronoun cannot be so easily interpreted as far as its focal or topical role
is concerned. One can adduce two sets of reasons for this:
. See the expressions meaning “in my opinion, in your opinion” given earlier, Table 5 and
Example (2) and Pitavy 2006b: 233.
Accented and unaccented pronouns in Ancient Greek
–– strong pronouns are only one of the marks in the construction of an extended
utterance, and we must take account of the role played in the m
eta-informative
structure by other processes.
4.1 Intonation
The choice of a pronoun from one or other series is morphological and the dif-
ference in accentuation (clitic forms) indicates a prosodic contrast. The use of
this criterion is extremely limited, and it is impossible, for example, to take into
account criteria which are either unknown or totally hypothetical (since we have
no available data). Strictly speaking, the only indication we have of intonation and
rhythmical organisation comes from the written accent, which involves the prob-
lems that have already been mentioned.
4.2 Syntax
Clitic forms, which are attached to another constituent, clearly do not behave in
the same way as Tonic forms (called separate). Accented Greek pronouns are used
as noun phrases, which means they are “autonomous” in relation to the verb and
that they can easily be displaced, to the left or to the right of the verbal nucleus.
It is reasonable to suppose that the use of the same accented form, as in any
noun phrase, would not have the same value if the pronoun preceded the verb
rather than followed it.
. Cf. Lambrecht (1994: 6), Dik (1983) for Greek word order and Dik (2003)
Jean-Christophe Pitavy
I will here deal with two problems as regards the syntax of pronouns: clitic
forms, and the place of accented forms.
. In Greek, as in a number of languages, clitics are considered to be excluded from first
place in the sentence and typically appear in second place – these are “Wackernagel” clitics, a
reference to the “law of position” set out in Wackernagel (1892).
. This usage brings to mind phrases with prepositional accentuation to be found in certain
Slavonic languages (for example, in Russian из дому íz domu “outside one’s home”, гулять
по полю guljat’ pó polju “walk in the field”; Polish: na dół “downstairs”, wypić do dna “drink
in one go”, etc.). This kind of proclitic attraction is presented as being disappearing in certain
descriptions, cf. among others Comrie and Stone (1978: 82–85) and Proeme (1994: 361). In
fact the situation in Ancient Greek is very different: whereas in Slavonic languages this place-
ment of the stress on the first clitic takes place in Prep+N sequences, in Greek, where the
phenomenon is extremely rare and is attested to in accentuated forms, this accentuation is
normal, since it characterizes a sequence of clitics (Prep+Pro). In this case it is the use of
an enclitic pronoun which is noteworthy. Being proclitic, the preposition governs, morpho-
phonologically, a form bearing the accent for the PP that it forms with it. As there is no reason
Accented and unaccented pronouns in Ancient Greek
These cases are, however, rare and do not provide grounds for suggesting that the
(almost non-existent) choice between prep + accented form and prep + clitic is a
significant one.
to think that the simple use of a prepositional construction involves the use of an extended
structure, and as examples of post-prepositional clitics are extremely rare, we must conclude
that the speaker has no other choice of pronominal form and that, independently of the meta-
informative conditions, the morphological contrast is neutralised in the sentence and in the
utterance.
. On word order in Ancient Greek, see Dik (1995) for data taken from the works of
Herodotus.
. Dik (2003), who proposes the absence of emphatic value in certain postposed personal
subject pronouns.
Jean-Christophe Pitavy
4.2.3.1 The verbs “to say” and “to think” + pronoun. The P1 pronoun egṓ pre-
cedes légō “I say” in 73 cases,17 whereas the reverse word order (légō egṓ) is only
found once.18
The P2 pronoun sú is used in Plato 85 times19 before légeis “you say”, whereas
it only appears after the same verb form 5 times.
. In the Example (5) dokeîs sù truphân, we may consider the subject pronoun anteposed
to the modal auxiliary (“to seem + infin.”). In all the examples of sú with dokeîs + inf., the
pronoun always precedes the infinitive.
. As we are only talking about the overall illustration of a corpus which merits a much
more developed study, I here consider anteposed any pronominal form placed in the p reverbal
zone, whether it is immediately next to the verb or not, and the same is true for postposition
cases of the same forms, in the postverbal zone.
. Plato: 56 instances; Lysias: 3 and Demosthenes: 14.
. Plato, Letters.
. Plato: 82 instances; Lysias: 1; Demosthenes: 2.
Accented and unaccented pronouns in Ancient Greek
The verb oîmai “to believe, think”, almost systematically used with a subject
pronoun, notably in discursive comments (see above), has the pronoun egṓ before
the verb 78 times in all of these authors,20 whereas the reverse order (egṓ oîmai) is
only found 15 times in Plato and only once in Demosthenes.
Expressions involving a verb of opinion, with the personal pronoun in the
dative (of the kind dokeî + P1/P2 “it seems to me / you”) confirm this quantitative
data in Plato and the orators.
Even if we take the null hypothesis, that is that the use of the (accented) sub-
ject pronoun together with the verb is not significant, the preverbal position is
dominant. If we now assume that, on the contrary, the accented form corresponds
to a marked usage, this value would be largely confirmed by the place of the pro-
nominal constituents. In other words, when it is used, the “strong” form of the
pronoun contributes to an extended structure, just as anteposition does in relation
to the verb.
(14) «SŌ. Éti dḕ tríton skepsṓmetha ei sundokeî soí te ka hēmîn. – CRIT. Tò poîon
dḕ toûto? – SŌ. Egṑ dḕ phrásō.
Dokeî gàr dḕ emoí te kaì tôide,
it-seems in-fact ptc P2D as-well-as DEMD
perì hósōn estìn epistḗmē…»(Plato, Laches, 198 c 11)
“So. We still have to examine a third point in order to know whether you
agree with us. – critias: what point? – So. I will explain. He and I both
have the impression that everything for which there is knowledge…”
Here we have the one single occurrence in the entire corpus of the postposition of
the accented pronoun, in a construction in which the pronoun is usually placed in
the preverbal position: emoì (gàr) dokeî. As I mentioned earlier, subject pronoun
forms and accented forms are used in Greek as constituents. Here, therefore, we
must consider the personal pronoun coordinated with a demonstrative pronoun to
have the same syntactic function (tôide “to the latter”). The use of the unaccented
form of the personal pronoun is not possible: one cannot have *moi coordinated
with a NP in the dative or a demonstrative pronoun in the same case, both struc-
tures having the status of constituent. In this case, the postverbal position can be
analysed as being independent of a possible choice of the accented form by the
speaker. The utterance is opened by the verb form placed at the start and u
nderlined
by gár (focalising value), and since the complement in the dative includes a demon-
strative pronoun (tôide), the personal pronoun is automatically accented.21
With the Example (9), given earlier, we can analyse a comparable situation
with the marker kaí:
The corpus contains many examples of imperatives with a subject pronoun pre-
ceded by the inclusion marker kaí (“also”). In fact this adverbial marker always
precedes the constituent to which it refers, which is autonomous in relation to the
verb (NP, adverb, tonic forms). The unaccented forms me, se, moi, soi etc. and,
with all the more reason, the zero nominative pronoun, cannot therefore be accen-
tuated by kaí. In the Example (9), since the subject is focalised it is bound to be
marked by the tonic form, independently of its position in relation to the verb.
4.2.3.3 Syntax and clause type. The corpus offers another, particularly interest-
ing example of the same syntactical construction carried out in two different or-
ders and presenting a contrast that is both formal and meta-informative:
It is worthy of note that the second expression, which in formal terms includes
the same constituents as the first, shows an inverse word order. In Socrates’
reply, we clearly understand from the micro-context that it is the subject,
“you” which is focalised by the anteposed accented subject pronoun, whereas
. The corpus has only one other example of P1D paired with other pronoun and noun
forms in the dative, a complement of the same verb (dokeî), but in the preverbal position: SO.
“ … Emoì kaì humîn kaì hólēi tê pólei oú moi dokeî…” “for me, for you, and for the whole City,
it doesn’t seem to me that…” (Plato, Apology, 34 e 2).
Accented and unaccented pronouns in Ancient Greek
4.3.1 Dḗ
In the corpus of Attic Greek prose writers, for example, the insistence particle dḗ
“really”, is above all used with subject pronouns (twenty occurrences evenly dis-
tributed between the two persons: 9 egṑ dḗ and 11 sù dḗ) while it is rarely used with
complement forms: 1 emè dḗ ; 1 emoû dḗ; 1 emoì dḗ; 3 soì dḗ; 1 sè dḗ. For example,
we can cite:
. It would of course be possible in a different situation where the addressee is emphasised
in contrast to the other participants in the debate, but that is not the case here.
. Out of the 5 occurrences of the sequence légeis sú, 4 are interrogative (Symposium,
Charmides, Euthydemus, Republic).
. This seems to be confirmed by the fact that in our corpus, sequences of accented, focalised
interrogative pronouns + accented personal pronoun of the type tí émoi, are not found, whereas
one can find without difficulty the variant with the unaccented personal pronoun (tí moi).
Jean-Christophe Pitavy
These three devices used together allow for the construction of an extended utter-
ance involving either a topic or a focus.
4.3.2 Ge
With the particle ge (cf. Denniston 1934: 114–115), the incidence of morpho-
syntax (the pronoun form) with pragmatics (the use of ge) is even clearer,
although it must be specified that this particle is enclitic, that is to say it is
. Here I am of course putting forward the hypothesis that the word soi in soì dḗ should be
interpreted as tonic (“strong”), not clitic, for a number of reasons. Syntactically a clitic cannot
be used in first place (this confirms Wackernagel’s law). From the point of view of the use of
pragmatic particles, accented forms are normally used with dḗ or ge (see the summary given
at the beginning of Section 4.3.1.). These parameters thus allow us to confirm the accentuation
given by the texts.
Accented and unaccented pronouns in Ancient Greek
Conclusion
A great deal of data, a small sample of which has been presented here, seems to
confirm that the use of the accented personal pronoun at the same time as the
verb form enables the construction of an extended utterance. A verification of
this principle would demand a meticulous examination of the meta-informative
structure of a greater number of uses. There are obviously exceptions, and I have
quoted some of these, notably relating to certain formulaic structures or those that
are used at the level of meta-informative comment. P1 personal pronouns which
involve the speaker as a privileged, constituent reference point, are particularly
common in such uses.
Secondly, it is not enough to look to the meta-informative structure of each
usage in order to decide whether we are dealing with topicalisation or focalisation.
Jean-Christophe Pitavy
Abbreviations used
References
Denniston, John Dewar. 1934. The Greek Particles. Oxford: The Clarendon Press.
Dik, Helma. 1995. Word Order in Ancient Greek. A Pragmatic Account of Word Order Variation
in Herodotus. Amsterdam: J.C. Gieben.
Dik, Helma. 2003. On unemphatic “emphatic” pronouns in Greek: Nominative pronouns in
Plato and Sophocles. Mnemosyne 56(5): 535–550. Leiden: Brill.
Dik, Simon Cornelis. 1978. Functional Grammar. Amsterdam: North-Holland.
Dik, Simon Cornelis. 1989. The Theory of Functional Grammar. Part 1: The Structure of the
Clause. Dordrecht: Foris Publication.
Ducrot, Oswald. 1984. Le Dire et le Dit. Paris: Éditions de Minuit.
Gildersleeve, Basil Lanneau & Miller, Charles William Emil. 1911. Syntax of Classical Greek from
Homer to Demosthenes. New York NY: American Book Company.
Givón, T. 1984. Syntax. A Functional-Typological Introduction, 1. Amsterdam: John Benjamins.
Halliday, Michael A.K. & Hasan, Ruqaya. 1976. Cohesion in English. London: Longman.
Ildefonse, Frédérique. 1997. La naissance de la grammaire dans l’Antiquité grecque. Paris: Vrin.
Kühner, Raphael & Gerth, Bernhard. 1898–1904. Ausführliche Grammatik der griechischen
Sprache. Satzlehre. Hanover: Hahnsche Buchhandlung.
Lambrecht, Knud. 1994. Information Structure and Sentence Form. Topic, Focus and the Mental
Representations of Discourse Referents. Cambridge: CUP.
Langacker, Ronald. 1987. Foundations of Cognitive Grammar, Vol. 1: Theoretical Prerequisites.
Stanford CA: Stanford University Press.
Lyons, John. 1968. Introduction to Theoretical Linguistics. Cambridge: CUP.
Moeschler, Jacques & Reboul, Anne. 1994. Dictionnaire encyclopédique de pragmatique. Paris:
Éditions du Seuil.
Nølke, Henning. 2001. Le regard du locuteur 2. Pour une linguistique des traces énonciatives.
Paris: Kimé.
Perdicoyanni-Paléologou, Hélène. 2002. The endophoric Use of ὁ, ὅδε, ἐκεῖνος, αὐτός in Ori-
gen’s Homilies on Jeremy, as translated by St Jerome (is, hic, iste, ille). Belgisch Tijdschrift
voor Filologie en Geschiedenis 80: 35–50.
Pitavy, Jean-Christophe. 2006a. Cohérence discursive, ellipse: Du nouveau en questions
et réponses? In Cohérence et discours, Frédéric Calas (ed.), 251–262. Paris: Presses de
l’Université Paris-Sorbonne.
Pitavy, Jean-Christophe. 2006b. Au sujet du pronom sujet: ἐγώ et la liberté du locuteur en
grec. In Ἐν κοινωνίᾳ πᾶσα φιλία. Mélanges offerts à Bernard Jacquinod, Jean-Luc Breuil,
Christophe Cusset, Florence Garambois, Nicoletta Palmieri & Éric Perrin-Saminadayar
(eds), 227–242. Saint-Étienne: Presses universitaires.
Proeme, Henk. 1994. On word stress in Polish prepositional phrases with a monosyllabic pro-
noun or noun. In Dutch Contributions to the Eleventh International Congress of Slavists,
Bratislava, August 30-Septembre 9, Adriaan Barentsen, B.M. Groen, Rob Sprenger (eds),
Amsterdam: Rodopi.
Riegel, Martin, Pellat Jean-Christophe & Rioul René. 1994. Grammaire méthodique du français,
Paris: Presses universitaires de France.
Schiering, René. 2006. Cliticisation and the Evolution of Morphology: A Cross-linguistic Study
on Phonology in Grammaticalisation. Ph.D. dissertation, University of Konstanz.
Schwyzer, Eduard & Debrunner, Albert. 1939–71. Griechische Grammatik, Munich: C.H. Beck.
Wackernagel, Jacob. 1892. Über ein Gesetz der indogermanischen Wortstellung. Indoger-
manische Forschungen I: 333–436.
Jean-Christophe Pitavy
Włodarczyk, André & Włodarczyk Hélène. 2006. Focus in the meta-informative centering the-
ory. In La focalisation dans les langues, André Włodarczyk & Hélène Włodarczyk (eds),
27–46. Paris: L’Harmattan, Coll. Sémantiques.
Włodarczyk, André. 2004. Centre d’intérêt et ordres communicatifs. In Énoncer, l’ordre infor-
matif dans les langues, Pierre Cotte, Martine Dalmas & Hélène Włodarczyk (eds), 13–32.
Paris: L’Harmattan, Coll. Sémantiques.
Zwicky, Arnold M. 1995. What is a clitic? In Clitics: A Comprehensive Bibliography 1892–1991
[Library and Information Sources in Linguistics 22], Joel A. Nevis, Brian D. Joseph,
Dieter Wanner & Arnold M. Zwicky (eds). Amsterdam: John Benjamins.
Personal subject pronouns
and the meta-informative centering
of utterances in classical Latin
Perrine Vedrenne-Cloquet
Université Paris-Ouest
1. Introduction
This article deals with the role played by the personal pronouns ego and tu in the
meta-informative organisation of utterances. In Latin, a verb with a personal end-
ing is sufficient for the grammaticality of an utterance.1 The first and second per-
son subject pronouns with the same reference as an accompanying personal verb
form thus constitute a redundant expression of the subject. Their presence in this
sort of context, where there is no ambiguity over identifying the subject, is usually
analysed in normative grammar in terms of its expressive or stylistic contribution.
It seems to us that this question should be studied as a problem of informative
structure (Pinkster 1990), and more precisely, from the angle of the meta-infor-
mative centering theory (Włodarczyk 2003, Włodarczyk & Włodarczyk 2006a,
2006b, 2008) the terminology of which we will employ here.2
In a verb form with a personal ending, such as credo, the verbal predicate
is assigned to a personal entity which becomes the global centre of attention
(CA), and thus takes on the role of “subject”. This term is to be understood in its
Aristotelian sense (“what we speak about”), since it is hardly possible to treat a
personal ending as a grammatical subject.
When the personal pronouns ego and tu are associated with a verb that has
an ending corresponding to the person in question, we may hypothesise that this
double reference to the person takes on a meta-informative value.3 Or rather, if in
an utterance with no pronoun the personal entity already has a meta-informative
value, then in an utterance containing ego or tu, this pronoun is the result of
a supplementary informative prominence (a “meta-meta-informative” value).
According to the MIC theory, two levels must therefore be distinguished:
–– Level one (credo): pronoun Ø + verb with a personal ending: subject + predi-
cate corresponding to a base utterance not divided into old and new,
–– Level two (ego credo): full pronoun + verb with a personal ending: extended
utterance in which the pronoun is either topicalised (old meta-informative
status) or focalised (new meta-informative status) and contrasts with the rest
of the utterance which has the opposite meta-informative status.
We will see that it is in fact impossible to attach just one meta-informative role to
the pronouns ego and tu: depending upon the context, they may appear as Topic
or Focus.
The Topic value of ego and tu is quite legitimate: it emphasises information
about the actors in the situation spoken about in the utterance. These pronouns
point to the participants of the speech act; the fact that they have an old meta-
informative status is therefore not surprising in theory. Moreover, in an extended
utterance, as an amplification of the subject of the base utterance, they fulfil as
Topic the same global CA role as the subject. We will thus see, first and foremost,
to what extent this Topic value is relevant to the construction of the informative
cohesion of the text. These pronouns may, however, also be used as Focus, whether
. Thus, in parallel with meta-informative centering theory, Touratier gives these pronouns,
when they accompany a personal verb, the syntactic function of extraposition, a function
which acts as a bridge between the syntactic and informative levels, since according to him
it makes “du contenu du syntagme extraposé à gauche, le thème ou le support informatif du
reste de l’énoncé” (the content of the phrase moved to the left, into the theme or informative
basis for the rest of the utterance) (1994: 35). Other syntactical presentations take into account
the informative organisation of the sentence and more or less correspond to extraposition as
defined by Touratier: cf. extraposition to the left (topical extraposition ), or Top and Foc posi-
tions on the left periphery of the sentence (Rizzi 1997).
Personal subject pronouns and the meta-informative centering of utterances in classical Latin
the status of new information of this CA emerges from the text structure or is the
result of encyclopaedic knowledge.
These two values of Topic and Focus will be illustrated by a detailed analysis
of various occurrences which may seem fastidious to non-latinists, but is justified
by the importance of context for understanding the specific role of these pronouns
in informative coherence. Finally, the meta-informative role of ego and tu explains
certain usages in context, where these pronouns, rather like particle-connectors,
act as markers of the highlighting of other elements of the utterance which are
treated as topics. We will give examples of this usage, after listing the other means
available to Latin for marking informative salience.
2. Topicalisation
The topicalising value of the personal pronouns ego and tu can be seen clearly in
a great number of usages. Detailed analysis, however, shows that ascertaining this
value is not in itself enough to explain fully the informative role of the pronouns
in question. In view of their particular enunciative dimension, the enunciative
structure must be integrated with the informative structure. This can be observed
in the following example, in which thematic progression relies on the topicalisa-
tions of Pansam and ego:
(1)
Gratulor nobis Q. filium exisse: molestus non erit. Pansam bene loqui
credo; semper enim coniunctum esse cum Hirtio scio; amicissimum Bruto
et Cassio puto, si expediet – sed quando illos uidebit ? – Inimicum Antonio
quando aut cur ? Quousque ludemur ? Ego autem scripsi Sextum aduentare,
non quo iam adesset, sed quia certe id ageret ab armisque nullus discederet.
[Att. 15,22 ]
“I congratulate ourselves that young Quintus has gone out of town: he
won’t be a nuisance to us. I believe Pansa is using satisfactory language. For
I know that he has always been closely united with Hirtius. I think he will
be a very warm friend to Brutus and Cassius if it turns out to be expedient.
But when will he ever see them? And that he will be opposed to Antony –
but when and on what grounds? How long are we to be fooled? However,
I wrote you word that Sextus Pompeius was coming, not because he was
actually near, but because he was certainly contemplating that move and
because he was not showing any signs of abandoning arms.”4
. The English translation of Latin examples are taken from http://perseus.uchicago.edu/
Perrine Vedrenne-Cloquet
This first part of a letter addressed to Atticus, in which Cicero writes about young
Quintus, his nephew, then Pansa, sets out an initial predication (Pred1: gratulor
nobis; credo; scio; puto) and a second predication (Pred2: Quintum filium exisse;
Pansam bene loqui; coniunctum esse cum Hirtio; amicissimum Bruto et Cassio). In
the first utterance (gratulor nobis Q. filium exisse), the verb gratulor is placed first –
in the opening line the presence of the speaker is stated from the outset, in a sub-
jective affirmation.5 But subsequently the second predication is in the word-initial
position: here it is word order which plays its part in the topicalisation of Pansam
(“I believe Pansa is using satisfactory language”), and underlines the thematic pro-
gression. The verbs in Pred1 (credo, scio, puto) are placed at the end, after Pred2
(Pansam bene loqui; coniunctum esse cum Hirtio; amicissimum Bruto et Cassio),
benefiting informative cohesion.
Then after several interrogative propositions, the order is inverted: the utter-
ance ego autem scripsi Sextum aduentare follows the same syntactical model as
previous utterances, in which the first person verb governs an infinitive propo-
sition, but the word order and the relative importance are different: Pred1 (ego
autem scripsi) is at the head, and this time topicalisation concerns the first person
subject, in the form of the pronoun ego, the connecting word autem underlining
the change of Topic6 (“However, I wrote you word…”). The form ego scripsi is in
effect the opposite of the forms credo, scio, puto which have no personal subject
pronoun.
The distinction between the first person verb forms (credo, scio, puto) and
the expression of the first person reinforced by the pronoun ego in ego scripsi
therefore appears to be motivated by the choice of a different topicalisation: that
of the topic of Pred2 (Pansam) in the first case, and of the topic of Pred1 (ego) in
the second case.
But the parallel initially established between these forms is illusory: scripsi, in
the perfect tense refers to another utterance in the first person, temporally distinct
from the first person of credo, scio and puto. Furthermore, credo, scio and puto are
meta-informative markers: unlike scripsi, they belong to the plane of the speech
act. The first person occurrences are thus placed on two different speech levels: the
. The letters of Cicero begin for the most part with an utterance in the first person: most
often, an indication of time and/or place is followed by the comment accepi litteras or perueni,
etc. Moreover we find in volume IX of Correspondence at the very beginning: “Spero tibi”
(Att. 14,5,1), “O mi Attice, uereor” (Att. 14,12,1), “Quam uellem” (Att. 15,4a), “Gratulor”
(Att. 15,22), “Mirifice torqueor” (Att. 15,23,1), and “Gaudeo” (Att. 5,27,1), and with no subjec-
tive marking, “Narro tibi” placed in parataxis (Att. 15,16a,1, Att. 15,21,1).
. Cf. pragmatic analyses of autem, and in particular the study by Kroon 1995.
Personal subject pronouns and the meta-informative centering of utterances in classical Latin
first person of credo concerns the speech act, whereas the first person of ego scripsi
refers to the situation spoken about in the utterance. Here, ego does not so much
refer to the speaker as to the person who in an earlier letter wrote “Sextus quem
advent are aiunt” (this is in fact the letter Att. 15,21,1). Because of the influence
of the pronoun ego, the first person is topicalised in the same register as Pansam:
“I say [with modalization] about Pansa that…” and “I say [without modalization]
about myself that in the past…” The topicalisation of the first person by the pro-
noun ego here gives prominence to what, in this first person subject, does not
stem from the speech act plane. It is important therefore to adjust both the syn-
tactical and the pragmatic organisation so as to explain as fully as possible the
meta-informative organisation of the message.
The first person can thus be the object of topicalisation in order to distinguish
between different speech participant instances in an utterance. Through this use of
ego as contrasting topic, the speaker is contrasted with Pansa, by no longer simply
taking him as the mark of the person who says something, but as the mark of the
person about whom something is said, with referential identity between the two.
Topic value thus plays its part in discourse structure by enabling a distinction to
be made in the use of a first person form, between what belongs to the level of the
utterance and what concerns the level of the speech act.
3. Focalisation
(2) De meo itinere uariae sententiae; multi enim ad me. Sed tu incumbe,
quaeso, in eam curam; magna res est. [Att. 15,25]
“As to my journey various opinions are expressed: for I have a great number
of visitors. But pray throw yourself heart and soul into that question. It is a
serious matter.”
In line with what has been noted above, we postulate that the 2nd person is
emphasised by the personal pronoun at the start: as for you (in bold), I give an
Perrine Vedrenne-Cloquet
(3) Auriculam Mario grauiter miraris olere. Tu facis hoc: garris, Nestor, in
auriculam. [Mart., Ep. III, 28]
“You’re amazed that Marius’ ear smells nasty. It’s you that make it so: you
chatter, Nestor, into his ear.”
In the utterance Tu facis hoc, it is difficult to deny the New meta-informative status
of the pronoun tu as focalised CA, when the remainder of the utterance, as “back-
ground”, is made up of the anaphoric pronoun hoc, with no prominence at all, and
of the verb facis, which is in itself also anaphoric.
Here is another example, outside the corpus, taken from one of Ovid’s Heroi-
des, which shows that the distribution of meta-informative new/old information
can also depend upon encyclopaedic knowledge:
(4) Non ego cum Danais arma cruenta fero; [Ov., Her. V, v.156]
“I do not bring a bloody war with the Greeks.”
In this fictional epistle, Oenone is addressing Paris and setting herself more or less
implicitly in contrast to Helen, the woman for whom he has left her. Here, ego is
clearly focalised: from the first polemical negative non, the various participants –
at different levels of coding and de-coding, the author, the fictional speaker, the
Personal subject pronouns and the meta-informative centering of utterances in classical Latin
fictional listener, the readers – pick up the meaning X cum Danais arma cruenta
fert (“X does bring a bloody war with the Greeks”), which is given as old informa-
tion in so far as it refers this time not to an anaphora within the text, but to the
knowledge held by the participants outside the text itself.
If the pronouns ego and tu are used to put meta-informative emphasis on the
subject when they are associated with a verb that has a personal ending, the study
of their place in discourse forces us to distinguish a number of nuances: we can
thus recognise a topicalising and a focalising value, on condition that we take into
consideration the overall coherence of the discourse. As has already been said,
we must not, however, loose sight of the specific pragmatic (meta-informative)
usage of these pronouns if we want to give as accurate an account as possible of
their informative use. But the part played by what are called personal “subject”
pronouns in the informative coherence of a text is more complex than it appears,
because the forms ego and tu can be used to modify the meta-informative status
of an element by detaching it from the rest of the utterance (Amacker 2001, Griffe
2007). Only the fact that these forms, associated with an accompanying personal
verb form, are necessarily felt to carry a meta-informative emphasis, explains that
they can, in second place in the utterance, underline the first-placed element, and
thus give it an informative prominence that it would not ordinarily have.
In order to shed light on the defining role that ego and tu can play in an utterance,
it is useful to review the means available in Latin for emphasising prominence in
the absence of markers of prosody in the written form of the language.
Generally speaking, prominence is bestowed by dislocation variously alter-
ing the order of words. Let us look at the three examples of increasing syntactical
change proposed by Colette Bodelot:7
Nausea is placed before the adverb jam, hosting the enclitic interrogative particle -ne.
–– Absence of grammatical agreement: the element that has been moved to the
left is in the nominatiuus pendens, and its anaphoric indicator is in the case
demanded by the syntactical structure of the clause.
(7) Plaut. Men. Arg. 1–2: Mercator Siculus, quoi erant gemini filii,/Ei surrupto
altero mors optigit.
“A Sicilian merchant, who had twin sons, had one of them taken away and
died shortly afterwards.”
Mercator Siculus is in the nominatiuus pendens, and the indicator, carrying the
syntactical function, is in the dative: ei.
Topicalisation can therefore be signalled by different kinds of extraction:
word-initial position, word-initial position with anaphora and grammatical agree-
ment, and word-initial position with anaphora but no grammatical agreement.
Furthermore, connectors such as autem or quidem can also serve as markers. But
a study of certain collocations tends to show that the pronouns ego and tu are in
their turn able to signal the topicalisation of other elements: always supposing that
they are themselves sufficiently clearly prominent to become as a consequence
discriminating indicators of changes in word order.
We can thus see constructions which have an anaphoric pronoun in first place,
marked morphologically by a particular case (usually the accusative or the dative),
and a 1st or 2nd person personal pronoun in second place: this morpheme then
brings about the detachment of the element that precedes it from the rest of the
utterance. Without that detachment and in the absence of any intonation criterion,
the prominence of the anaphora would not be evident. In the same way, in the
examples taken from C. Bodelot, the particule -ne, normally attached to the first
word, is attached to the next and thus detached from Nausea.
(8)
sed tamen idem σύνταγμα misi ad te retractatius, et quidem αρχέτυπον
ipsum crebris locis inculcatum et refectum. Hunc tu tralatum in
macrocollum lege arcano conuiuis tuis.8 [Att. 16,3,1]
. Cf. also Huic ego locum in proximo conduxi et, ut possum, ex meis angustiis illius sustento
tenuitatem. [Fam. 16,21,4, letter from young M. Tullius to Tiro]; Huic ego litteras ipsius arbi-
tratu dabo; eae te ne mouerint. [Att. 16,1,6]; Eas ego oportet perspiciam, corrigam; [Att. 16,5,5].
Personal subject pronouns and the meta-informative centering of utterances in classical Latin
The anaphoric pronoun Hunc takes up αρχέτυπον (“the original text”), tralatum
is the attributive of hunc, the complement of the object of the verb lege. The per-
sonal pronoun, in the same person as the governing verb, separates the comple-
ment of this verb and its attributive. In other words, an element belonging to the
first predication (tu lege) separates hunc from the second predication element tra-
latum.9 The intercalation of the personal pronoun between hunc and tralatum in
macrocollum signals the topicalisation of hunc by detaching it from the rest of the
second predication: without this intercalation, the string hunc tralatum in macro-
collum is unremarkable and nothing in the written form marks any dislocation.
The sentence in question could be translated as follows, if one wanted to retain
the informative prominence: “This, (you) read it copied on large paper privately
to your guests.”
The same kind of collocation appears in more complex utterances, such as:
(9) quodque ex istis fructuosis rebus receptum est, id ego ad illud fanum
sepositum putabam.10 [Att.15,15,3]
“while the receipts from those paying properties I thought had been put
aside for building the shrine.”
Literally: “and what has been received from those paying properties, that,
I thought had been put aside for building the shrine.”
. Attributive and predication since we have here a case of the attributive of the object where
“la caractérisation de l’objet N1 par son attribut est toujours le propos de l’énoncé.” “the char-
acterisation of the object N1 by its attributive is always the purpose of the utterance.” (Riegel
et al. 1994: 241)
. Cf. also Quod sine ulla dubitatione apud consules obtinuimus propter summam boni-
tatem et aequitatem causae, id tu nos obtinuisse non modo facile patiare sed etiam gaudeas.
[Att. 16,16b,2] “what we have obtained from the consuls without any hesitation, owing to their
great kindness and the justice of our cause, you will not only acquiesce, but even rejoice, in
our having secured.”
Perrine Vedrenne-Cloquet
apply exclusively to NPs, but also to clauses. However, after this first topicalisation
(of the relative clause),11 the meta-informative status of the anaphoric pronoun
id is in turn modified by ego: its position between id and the rest of the utter-
ance disturbs the normal word order id ad illum fanum sepositum putabam. The
intercalation of the personal pronoun between id and ad illum fanum sepositum
detaches id and signals its topicalisation: “what has been received, that, I thought
had been put aside.” Thus id takes up the topicalisation and is itself the object of
a second topicalisation: it is an informative marker which is in its turn pushed to
the front.
This analysis is furthermore corroborated by C. Touratier’s analysis of the
phenomenon of correlation, at a more syntactical level and using different
terminology:
We must note, however, that outside this general context, one could view id as
focalised. Indeed, taken separately, the utterance could equally well be translated as
“what has been received, it is that which I though had been put aside.” In this case,
id, taken from the preceding topicalisation, would then undergo focalisation and
would become the new meta-informative contributor of the rest of the u tterance.
It would then serve as a pivot, a junction point between the topicalisation and focal-
isation of the same content, but perceived informatively in two different ways.
. Bortolussi (2007: 37) has established that in Latin there can only be “thematization” of
the element in word-intial position when it is repeated, whereas word-initial position without
repetition can also be focalising.
Personal subject pronouns and the meta-informative centering of utterances in classical Latin
There thus seems to be nothing here, other than the context, which allows us
to distinguish between the topicalisation and the focalisation of the anaphora.
More generally speaking, without the help of the intonation and interpretation
provided by the wider context, it is often difficult to distinguish clearly between
the meta-informative status of the pronouns ego and tu: topic, focus or additional
meta-informative marker.
5. C
onclusion
Corpus
Cicéron, Correspondance, t. IX (de septembre 45 au 6 août 44 a. C.), texte établi, traduit et annoté
par Jean Beaujeu 1988. Paris: Les Belles Lettres.
Perrine Vedrenne-Cloquet
References
Amacker, René. 2001. Indices de thématisation segmentée non initiale chez Cicéron. In De
lingua latina nouae quaestiones: Actes du Xe Colloque international de linguistique latine:
Paris-Sèvres, 19–23 avril 1999, Claude Moussy, Michèle Fruyt, Jacqueline Dangel, Lyliane
Sznajder & Léon Nadjo (eds), 185–200. Louvain: Peeters.
Bodelot, Colette. 2007. Syntaxe liée vs. syntaxe non liée ou de l’utilité de distinguer une macro-
syntaxe de la (micro-)syntaxe en latin. In Eléments ‘asyntaxiques’ ou hors structure dans
l’énoncé latin, Actes du colloque international de Clermont-Ferrand, 16–17 septembre 2005,
Colette Bodelot (ed.), 15–33. Clermont-Ferrand: Presses Universitaires Blaise Pascal.
Bortolussi, Bernard. 2007. Ordre des mots et thématisation en latin. LINX 55: 33–47.
Griffe, Michel. 2007. L’extraposition en latin. In Eléments ‘asyntaxiques’ ou hors structure dans
l’énoncé latin, Actes du colloque international de Clermont-Ferrand, 16–17 septembre 2005,
Colette Bodelot (ed.), 37–54. Clermont-Ferrand: Presses Universitaires Blaise Pascal.
Kroon, Caroline. 1995. Discourse particles in Latin. A study of nam, enim, autem, vero and at.
In Amsterdam Studies in Classical Philology, 4. Amsterdam: J.C. Gieben.
Pinkster, Harm. 1990. Ego, tu, nos. Opmerkingen over het gebruik van subjektpronomina in het
bejzonder in Cicero De Oratore II. Lampas 19: 309–322.
Riegel, Martin, Pellat, Jean-Christophe & Rioul, René. 1994. Grammaire méthodique du français.
Paris: Presses Universitaires de France.
Rizzi, Luigi. 1997. The fine structure of the left-periphery. In Elements of Grammar: Handbook of
Generative Syntax, Liliane Haegeman (ed.), 281–337. Dordrecht: Kluwer.
Touratier, Christian. 1994. Syntaxe Latine. Louvain: Peeters.
Włodarczyk, André. 2003. Les homotopies du topique et du focus. In Ordre et distinction dans
la langue et le discours. Actes du colloque international de Metz (18,19 et 20 mars 1999),
Bernard Combettes, Catherine Schnedecker & Anne Theissen (eds), 513–526. Paris:
Honoré Champion éditeur.
Włodarczyk, André & Włodarczyk, Hélène. 2006a. Focus in the meta-informative centering
theory. In La Focalisation dans les langues, André Włodarczyk & Hélène Włodarczyk (eds),
27–46. Paris: L’Harmattan, Coll. Sémantiques.
Włodarczyk, André & Włodarczyk, Hélène. 2006b. Subject in the meta-informative centering
theory. Études cognitives/Studia kognitywne 7: 39–64. Warszawa: SOW, PAN.
Włodarczyk, André & Włodarczyk, Hélène. 2008. The pragmatic validation of utterances.
Études Cognitives/Studia Kognitywne 8: 117–128. Warszawa: SOW, PAN.
Glossary of defined terminology
Terms with an asterisk (*) in the explanatory notes are also glossary entries
Aboutness Two kinds of aboutness: (a) the spoken-aboutness (resemblance) consists of
(Speak- creating *Centres of Attention (CA) and (b) the speak-aboutness (contiguity)
Aboutness) consists of combining the created *Centres of Attention (CA) with everything
(Spoken- which is predicated about them.
Aboutness)
Agent An animate entity seen fit to play active roles in semantic situations as
expressed by utterances. In the Associative Semantics (AS) theory, it is possible
for an animate entity to play more than one active role. Moreover, agents can
also play non active roles (passive or median).
Anchor Spatio-temporal component (or operation) of semantic situations. One of
their three fundamental constitutive components (the two others being frames
and roles).
Associative The Associative Semantics (AS) theory is an ontology-based semantic theory
Semantics of information as expressed by natural language utterances with clear-cut
(AS) distinctions between (a) the universal (or at least common) ontology and
(b) the semantics of natural languages. In this theory, roles and anchors are
seen as elementary (unary) relations. Roles are defined ontologically as active
and passive unary relations of associated semantic situations. The combination
of their realizations gives rise to median roles such as “instruments” or
“means” among others. The main principle of AS is that one participant may
enact more than one role.
Attention- That phrase of a *Base or *Extended Utterance which can be assigned a
driven Phrase meta-informative *Old or *New Status. In order to communicate judgments,
(ADP) speakers need to select some information stored in their memory (thus
creating Centres of Attention (CA). While the Centre of Attention (CA) is
determined functionally as a mental concept which results in the selection
of a chunk of information, the Attention-driven Phrase (ADP) is defined
as that part of an *utterance (linguistic expression) which bears the selected
information with an explicit meta-informative pointer to it. The segment
of an expression is considered to be centered (corresponding to a Centre
of Attention) if it represents an entity, anchor or situation which has been
selected (see *Selection) among other entities, anchors or situations. An ADP
often contains (syntactic, morphological, prosodic) meta-informative markers.
If an utterance expresses more than one Centre of Attention (CA), one of them
is global (main) and the other one is local (secondary).
Example: The utterance “John and Mary dance” contains a sole (though
collective) ADP although the semantic information contains two participants
“John” and “Mary” both selected as Centres of Attention. Indeed, semantically
we have two associated (joint concurrent) situations: [John dances] & [Mary
dances] developing the schema with a dual existential quantifier: Exist2(x)
dance(x). Note that the utterance “John dances with Mary” contains only one
entity (“John”) which has been centered. See: *Global (or main) Attention-driven
Phrase (G-ADP) and *Local (or secondary) Attention-driven Phrase (L-ADP).
Glossary of defined terminology
Background The background is the speak-about part of the *Extended Utterance the
spoken-about part of which is known as the *Focus. As a rule, the *Status of the
spoken-about part of an *Extended Utterance is opposite to its speak-about. In
the case of the Focus-Background relationship, the speak-about has an *Old
meta-informative *Status. Antonym of *Comment.
Centre of Attention-centered information is that piece of knowledge to which humans
Attention are attuned by periods of time. In neurology humans are able to pay attention
(CA) to more than one concept at the same time provided that the selected *Centres
of Attention concern concepts which belong to different levels (regions
of memory). In linguistic predication two *Centres of Attention (CA) are
expressed by two different *Attention-driven Phrases (ADP) belonging to
two different levels of the constituency tree (structure of meta-information):
*Subject and *Object.
Combination Higher level operation on two selected chunks of information between which a
relationship can be established.
Comment The comment is the speak-about part of the *Extended Utterance the
spoken-about part of which is known as the *Topic. As a rule, the *Status of the
spoken-about part of an *Extended Utterance is opposite to its speak-about. In
the case of the Topic-Comment relationship, the speak-about has a *New meta-
informative *Status. Antonym of *Background.
Default Role The Default Role Mapping principle establishes the default relationships
Mapping between *Centre of Attention (CA) and *Agent (or *Figure) as expressed by an
*Utterance.
In Accusative languages, there is a default relation linking the *Subject with the
*Active Role (enacted by an *Agent or *Figure) while in Ergative languages, the
default role mapping links the *Subject with the Passive Role enacted by an
*Agent or *Figure).
Figure Inanimate entity. This kind of entity can enact both active and passive roles.
Active roles played by *Figures are said to be *Quasi-active (*Q-Role).
Focus The Local *Attention-driven Phrase (the spoken-about part) of an *Extended
utterance, the meta-informative Status of which is *New, whilst its speak-about
part (*Background) has an opposite (contrary) Status (*Old). Antonym of
*Topic.
Frame Representation of the space of states and actions. There are four kinds of
(Situation frames: 1 static and 3 dynamic frames: states are semantic situation frames
Frame) delimited by [+Space] only, events are semantic situation frames delimited
by [+Space] and determined by [+Time] (N.B.: events lack a middle stage.
For example: “to cough” as in “John just coughed”). Ordinary processes are
semantic situation frames delimited by [+Space] and determined by [+Time]
and [+Progression] and refined processes are semantic situation frames
delimited by [+Space] and determined by [+Time], [+Progression] and
[+Granularity].
Global The main *Attention-driven phrase (ADP) of a *Base Utterance which
Attention- expresses the hierarchically uppermost *Centre of Attention (CA) is defined
driven Phrase as a Global ADP of that utterance (*Subject). The ADP of an *Extended
(Global ADP) Utterance which expresses the hierarchically uppermost *Centre of Attention
(CA) is defined as a Global ADP of that utterance (*Topic). N.B.: The *Centres
of Attention may coincide in the expression plane, i.e. they may correspond to
the same ADP.
Glossary of defined terminology
Q-Role Agents typically fit dynamic situations (actions) well, whereas figures fit the
Quasi-active static ones (states). When it is not so, shallow level (i.e. a partially specified
Role semantic level) is introduced. There is an important feature of situation
Q-active Role participants which characterizes the shallow level: in active roles, the
inanimate entities (figures as opposed to agents) are semantically interpreted as
if they were animate. Such figures are said to enact quasi-active roles (Q-active
roles or Q-roles). In the cases of agentivation or figuration, for example, the
Q-initiator will be said to designate the figure in an active role (which normally
fits agents) and Q-source will be said to designate the agent in an active role
(which normally fits figures). See also *Role.
Role One of the three fundamental constitutive components of semantic
*situations. Two elementary (active and passive) and one derived (median)
roles are distinguished and defined as unary relations of associated semantic
situations. N.B. Due to the definition of semantic situations as associations,
one participant (role filler) may play more than one role in the given semantic
situation. See also *Roles and Voice.
Roles and In Accusative languages, by default, in the active voice utterances the *Subject
Voice corresponds to the *Active or *Q-Active Role while in the passive voice
utterances, the *Subject corresponds to roles which are neither active nor
q-active.
Selection Lower level operation on that chunk of information which is chosen on the
lower constitutive level of pattern recognition or pattern production.
Situation Situations are defined regardless of their participants. Their three fundamental
components are: (1) Frame (their instances are relations), (2) Role (their
instances are Participants) and (3) Anchors (their instances are spatio-temporal
locations).
Status See: Meta-informative *Status.
Subject Main meta-informative “spoken-about” part of a *Base Utterance, i.e. the
Attention-driven phrase (ADP) pointing at the global Centre of Attention
(CA). The meta-informative *Status of the subject may be either *New or *Old
as is the meta-informative *Status of a whole *Base utterance.
Theme That part (or set of subsets) of a text/discourse which expresses the global or
local *Centre of Attention (Concern).
Topic The Global *Attention-driven Phrase (the spoken-about part) of an *Extended
utterance, the meta-informative Status of which is *Old while its speak-about
part (*Comment) has an opposite (contrary) Status (*New). Antonym of
*Focus.
Utterance An utterance is a proposition uttered in a context and validated as bearing
information with *Old or *New Status. When used in an utterance, the
objective (denotative) oldness or newness of information motivates (see
*Motivation) the meta-level segmentation of communicated information
giving rise to the creation of pragmatic (connotative) content. Consequently,
an utterance can be assigned either *Old or *New meta-informative *Status
regardless of its truth-propositional semantic valuations.
Utterance The *Base Utterance has either a *New or an *Old meta-informative Status. It
(Base is not divided into parts (segments) with respect to the *New or *Old *Status,
Utterance) i.e.: the *Status of a *Base utterance cannot but be either entirely *New or
entirely *Old.
Glossary of defined terminology