Université Paris-Sorbonne


Centre de Linguistique Théorique et Appliquée

Center for Theoretical and Applied Linguistics






MIC Sorbonne 2010

Communication et contexte - Context-bound Communication

Colloque Interdisciplinaire - Interdisciplinary Conference



18-19 Novembre 2010






The usual claim is that only independent assertions have autonomous speech act, or speaker-deictic, status. Syntactically dependent assertions (such as with factive matrix predicates) do not have such illocutive-epistemic autonomy. It is also common to assume that strict V2 (such as in German, but not in English or French) provides unambiguous evidence for autonomous illocutive force. It will be shown that the evidence for these assumptions is based on the insertability of elements with epistemic force: i.e., on epistemic modal verbs, modal adverbials, and modal particles, all of which require licensing through clause-local speaker-deictic force which is indicative of the speaker's belief about the addressee's state of knowledge. The following syntactically dependent illustrate this illocutive (non-)autonomy.

(1)a   Felicitously, he regretted that he had done this.

    b  *He regretted that felicitously he had done this.

(2)a   Regrettably, she played with the lighter when mother fed the open fire.

    b *She played with lighter when regrettably mother fed the open fire.

Epistmic modal verbs, modal adverbials, and modal particles are all very dependable foreign conscience aligners. Since their Speaker deixis force is valid only for certain types of syntactically dependent clauses (non-factive verba dicendi, logical adverbial clauses, non-restrictive relative clauses) I investigate what the common property is for those subordinate clauses that license foreign conscience aligners and those that do not (factive verba dicendi, temporal-locative adverbials, restrictive relatives). This investigation leads us into the field of formal veridicality. The data which form the base for our linguistic conclusions and generalizations comprise German and French next to English.


La morphosyntaxe des numéraux est un domaine particulièrement mouvant du russe littéraire moderne. Il en résulte une telle complexité que l’emploi des numéraux semble échapper à toute logique si l’on s’en tient à la stricte synchronie. Parmi les apparentes incohérences du système, il en est une particulièrement remarquable : celle de l’accord de l’adjectif épithète d’un substantif dénombré avec 2, 3 ou 4 et situé entre le cardinal et le substantif qualifié à l’intérieur d’un syntagme numéral sujet au N.

Le cardinal exige que le substantif dénombré soit au G. sing. adnuméral ; l’adjectif, en revanche, se met au pluriel : au N. ou au G. si le substantif est féminin, et, si le substantif dénombré est masculin ou neutre, le plus souvent au G. — du moins en russe littéraire contemporain. Mais, à l’époque classique du russe littéraire moderne, le G. pluriel de l’adjectif est concurrencé par le N. pluriel, même avec un substantif masculin ou neutre.

Le propos de la présente communication est de montrer que la concurrence actuelle entre le N. plur. et le G. plur. de l’adjectif s’explique historiquement par des considérations qui ressortissent à la fois à la structure syntaxique de la proposition et à la structure métainformative de l’énoncé. La possibilité de choisir entre les deux formes de l’adjectif résulte de la confusion entre deux constructions distinctes en russe du xixe siècle :

1.     Ехали два Донские казака

2.     Deux cosaques du Don chevauchaient

construction personnelle à nexus (prédicat) accordé et

3.     Ехало два Донских казака.

4.     (≈ Les cosaques du Don qui chevauchaient étaient deux.)

proposition impersonnelle quantitative elliptique et remaniée dont la formule développée serait

Донских (Казаков) ехало два казака.

Des (cosaques) du Don, il en chevauchait deux cosaques.

La seconde construction n’était employée que dans un énoncé étendu dans lequel étaient sélectionnés deux centres d’intérêt : l’un topique  (exprimé par le syntagme elliptique au G. plur.) et l’autre focus (exprimé par le syntagme numéral). La construction personnelle était utilisée dans tous les autres cas. Mais, dans la seconde moitié du xixe siècle, les deux structures fusionnent en un unique modèle de phrase, d’où l’étrangeté des règles actuelles.



The sentence-initial position in Scandinavian languages can be considered to have an important meta-informative function (cf. that fact that it was labelled the ”foundation” (fundament) in Diderichsen’s (1946) positional grammar). In various previous studies it has been found that Scandinavian languages display a much higher percentage of non-subjects in clause-initial position than English or French does. This finding casts some doubt upon the commonly held view that Scandinavian languages are strictly SVO (with the subject as the global centre of attention and the object as the local one). In addition, it indicates that it may be more interesting to interpret these Scandinavian data in terms of extended utterances, where the topic constitutes the global centre of attention, and the focus the local one. The paper also discusses some additional problematic data taken from the corpus in question, where it would seem that also the third level in meta-informative centering theory, that of general vs. particular themes in texts and dialogues, could yield important insights when integrated into the analysis.


Studying the interplay between pragmatics, semantics, and grammar requires a method that allows to measure the impact of factors of various types on the presence or absence of specific linguistic phenomena. Based on research executed in collaboration with Dirk Speelman, this talk will illustrate the importance of statistical corpus-based methods for such a multiperspectival approach. Specifically, I will illustrate the importance of corpus-based statistical methods by applying them to the description of 'doen' and 'laten' causatives in contemporary Dutch: given the existence of 'doen' and 'laten' as causative verbs, what is it that determines the choice between the two?


In Japanese, we find sentences with no overt subjects, objects or other elements that are usually assumed to be obligatory ingredients of grammatical constructions.  Syntacticians have postulated deletion processes and/or zero-form pronominal elements, and semanticists have proposed construal rules that would reconstruct “missing” elements in syntax.  On the other hand, computer scientist have noticed that such “underspecification” can be found in programming languages, and exploited subject-less sentences in Japanese as a guiding principle in the design of programming languages, as suggested in Nakashima (2010).

As a rather artificial and simplified example, suppose that we have a pseudo-English like language with which communicating agents interact, where communicating agents have different accessibility to various information within and outside of itself, such as their own internal states external states of other communicating agents or features of the environment in which they are communicating with each other.  Suppose that agent alpha transmits a message to agent beta, in text, speech or any other mode, that looks something like (1) and agent beta responds by (2).

(1-a) HUNGRY?

(1-b) HUNGRY.

Whatever “HUNGRY” might mean for human and non-human agents, as long as we assume that (1) is a question rather than a statement, and (2) is a statement rather than a question, and HUNGRY refers to some internal state of those communicating agents which is more readily accessed by the agents themselves rather than some external entity, it would be more natural to understand that with (1), alpha is asking agent beta if beta is HUNGRY or not, and with (2), beta is responding to alpha that beta is actually HUNGRY.  On the other hand, consider something like PALE, which is meant to refer to an external state of those agents, which can more readily be observed by some external entity rather than the agent themselves, in which case, the more natural interpretations would differ.

(2-a) PALE?

(2-b) PALE.

A similar consideration would perhaps account for some of the attributions of who is responsible for what in (3), this time considered as a real English expression.  (This example is mentioned in Peters (2010) but the main point of his presentation is rather independent of what is to be discussed here.)

 (3) You will arrive at the airport on time.

This can be construed as (a) a prediction, (b) an injunction and (c) a promise, among others. As prediction, it would be about the addressee’s expected timely arrival at the airport.  On the other hand, as an injunction or as a promise, this would probably refer to the intent on the part of the speaker to have the addressee take responsibility for or to be responsible for the addressee’s timely arrival at the airport, respectively. Syntacticians and semanticist may have considered postulating a hidden higher order predicate and its subject for interpretations like this but such approaches have become less attractive, at least in syntax. Under a rather extreme pragmatic view to be discussed in my presentation, on the other hand, those should emerge as natural default interpretation as a result of interactions of relative accessibility of information among communicating agents, to be overridden by various factors within the context.


(1)  Hideyuki Nakashima, “Situated Language: Case of Japanese,” a symposium presentation in PACLIC 24, November 6th, 2010.

(2)  Stanley Peters, “Listening In,” a symposium presentation in PACLIC 24, November 6th, 2010.



Results from eight experiments support the view that, regardless of strength of context, strengthened by multiple cues, when an end-product interpretation of an utterance does not rely on the salient (lexicalized and prominent) meanings of its components, it will not be faster than nor as fast to derive as when it does. To test this view, we examined salience-based (here, literal) interpretations and context-based (here, ironic) interpretations, at different temporal stages, in strong contexts inducing an expectation for an ironic utterance. In Experiments 1-3, expectancy was manipulated by introducing an ironic speaker in context mid-position who also uttered the target utterance. In addition, the speaker’s ironic intent was made explicit by an adverb. Findings show that ironic targets were slower to read than literal counterparts. Experiments 4-5 show that, even when participants were exclusively presented ironically biasing contexts and told that we were investigating irony interpretation, the expectancy for irony acquired through such conditions did not facilitate expectancy-based compared to salience-based interpretations in either short or long delays (750-1000 ms ISIs). Replication of Experiments 4-5 under the same conditions in Experiments 6-7, in which longer processing times were allowed (1500-2000 ms ISIs), did not change the pattern of results. In Experiment 8, participants were allowed even longer processing time (of 3000 ms ISI). Still, ironically related probes were not more accessible than salience-based and unrelated probes. Although salience-based, literally related probes began to decay, context-based (ironic) interpretations were still lagging behind.

Evidence so far has failed to demonstrate that strong contexts, anticipating an ironic utterance, can facilitate ironic interpretations immediately compared to salience-based nonironic interpretations, thus replicating Giora et al.’s (2007), and Giora et al.’s (2009) findings.


Giora, Rachel, Ofer Fein, Laadan, Dafna, Wolfson, Jon, Zeituny, Michal, Kidron, Ran, Kaufman Ronnie, Shaham, Ronit. (2007). Expecting irony: Context vs. salience-based effects. Metaphor and Symbol 22(2): 119 - 146.

Giora, Rachel, Ofer Fein, Ronie Kaufman, Dana Eisenberg, and Shani Erez. (2009). Does an "ironic situation" favor an ironic interpretation? In G. Brône & J. Vandaele (Eds.) Cognitive poetics. Goals, gains and gaps, 383-399.



One of the seminal questions still awaiting resolution in linguistic theory is that of the nature and status of linguistic universals (see e.g. Evans and Levinson 2009; von Fintel and Matthewson 2008). In this paper I present a proposal of pragmatic universals understood as universals of processing, and demonstrate how a theory of utterance interpretation called Default Semantics (Jaszczolt 2005, 2010) can account for them by shifting the requirement of compositionality to the level of the representation of the interaction of the output of these processes. I begin with some pertinent examples of linguistic diversity and propose the types of processes that in each case secure the intended interpretation.

In Default Semantics, the requirement of efficiency of communication is explained as the trade-offs between (i) lexicon and grammar; (ii) processing of the utterance in context; and (iii) assigning default interpretations, where defaults are understood as salient, automatically reached meanings available for the particular speaker-addressee pair and the particular context. The diversity and processing universals are illustrated with examples from Swahili, Wari’ and English.


Select references:

Evans, N. and S.C. Levinson. 2009. ‘The myth of language universals: Language diversity and its importance for cognitive science’. Behavioral and Brain Sciences 32. 429-92.

Jaszczolt, K. M. 2008. ‘Psychological explanations in Gricean pragmatics and Frege’s legacy’. In: I. Kecskes and J. Mey (eds). Intentions, Common Ground, and the Egocentric Speaker-Hearer. Berlin: Mouton de Gruyter. 9-45.

Jaszczolt, K. M. 2009. Representing Time: An Essay on Temporality as Modality. Oxford: Oxford University Press.

Jaszczolt, K. M. 2010. ‘Default Semantics’. In: B. Heine and H. Narrog (eds). The Oxford Handbook of Linguistic Analysis . Oxford: Oxford University Press. 215-246.

Jaszczolt, K. M. and J. Srioutai. (forthcoming). ‘Communicating about the past through modality in English and Thai’ In: F. Brisard and T. Mortelmans (eds). Cognitive Approaches to Tense, Aspect and Modality’. Amsterdam: J. Benjamins.

Mauri, C. and J. van der Auwera forthcoming. ‘Connectives’. In: K. M. Jaszczolt and K. Allan (eds). The Cambridge Handbook of Pragmatics. Cambridge: Cambridge University Press.

Szabò, Z. G. 2000. ‘Compositionality as supervenience’. Linguistics and Philosophy 23. 475-505.

von Fintel, K. and L. Matthewson. 2008. ‘Universals in semantics’. The Linguistic Review 25. 139-201.



BDI agents are a well known concept in modern computer science. BDI models (and BDI architectures) are technically oriented conceptualizations and realizations of intentional approach to systems' modeling. Within this intentional approach mentalistic concepts of beliefs, desires and intentions are used to describe states and behaviors of individual agents (systems). Multimodal logics are used in BDI models to capture in a formal way many substantial aspects of beliefs, desires and intentions. Formulas of multimodal languages, used in these logics, are intentionally interpreted and treated as formal representations of natural language sentences. This fact seems to be the main reason for high popularity of BDI models in computer science community.

   In this presentation we propose an original approach to integrate BDI models with actual agency. In particular:

a) We assume that the multimodal language used to describe knowledge in our BDI agents consists of three operators of knowledge (Know), belief (Bel) and possibility (Pos).

b) We assume that formulas Know(p), Bel(p) and Pos(p) are generated by the agent herself and interpreted as formal representations of spoken sentences "I know that p", "I believe that p" and "It is possible that p".

   Although the classical BDI models tend to capture the same approach to modelling agency, they do not seem to be adequate when we seriously treat the assumed commonsense meaning of Know Bel, and Pos operators. In particular, the classical approaches seem not to take into account the substantial difference between the spoken formulas Know(p), Bel(p) and Pos(p) and the content of internal knowledge (encapsulated in the agent) that is labeled by Know, Bel, and Pos. Some remarks on this issue are given. In order to make the difference between "spoken" and "not spoken/underlying", we apply in our approach to modeling agency the well known concept of semiotic triangle:


In this approach the agent becomes a complex autonomous and active entity that encapsulates everything that is required to realize actual relation between object of the world and their descriptions (spoken modal formulas).

   To ensure proper usage of Know(p), Bel(p) and Pos(p) by our BDI agent, we define the so-called epistemic satisfaction relation that substantially differs to Tarskian satisfaction relation based on classical definition of truth:

   The major practical target of our research is to show in what way to design an artificial BDI agent that fulfills the following requirements:

(a) If the agent produces modal formulas Know (p) Bel(p) and Pos (p), these formulas should be consequently treated as representations of following sentences "I know that p", "I believe that p", and "It is possible that p", respectively.

(b) The sets of formulas produced by the agent should be acceptable from the commonsense point of view. For instance, it should be forbidden for the BDI agent to produce both Know(p) and Bel(p) as elements of the same state of knowledge.

Sets of requirements are given to define acceptable distributions of modal formulas Know(*), Bel(*) and Pos(*) and related proofs are given to show that these sets of requirements can be fulfilled.

   At the end the relation between the semantics of Know, Bel and Pos given by our original epistemic satisfaction relation and the well known possible-worlds semantics is briefly discussed, as well as some interesting directions of further research are mentioned.



The paper seeks answer to the question: why exactly we say what we say the way we say it. Although Giora (1997; 2003) argued that cognitively prominent salient meanings, rather than literal meanings, play the most important role both in production and comprehension of language, most attention in pragmatics research has been paid to comprehension rather than production. This paper argues that salience plays as important a role in language production as in comprehension, and discusses how salience of an entity can be interpreted as a measure of how well an entity stands out from other entities and biases the preference of the individual in selecting words, expressions and complex constructs in the process of communication. The role of perceptual and linguistic salience in language production involves a relation between prominence of entities in a ranking, and preference of a choice among alternatives.

   From the perspective of the interlocutor three theoretically-significant categories are distinguished: inherent salience, collective salience and emergent situational salience. Inherent salience is largely equivalent to cognitive status. It is characterized as a natural built-in preference in the general conceptual and linguistic knowledge of the speaker, which has developed as a result of prior experience with the use of lexical items and situation-bound utterances, and changes both diachronically. Inherent salience is affected by collective salience and emergent situational salience. The former is shared with the other members of the speech community, and changes diachronically. Emergent situational salience that changes synchronically refers to the salience of specific objects or linguistic elements in the context of language production, and may accrue through such determinants as vividness, speaker motivation and recency of mention. In an actual situational context individual salience is affected and shaped both by collective and situational salience. When the speaker is faced with the choice of a word or an expression, a ranking of the available choices is obtained on the basis of the degree of salience of entities in the generation context. The word or phrase then is selected for utterance on the basis of maximum salience.



The central aim of this paper is to show that the function of verb-first (V1) in Old Icelandic declaratives is essentially of an aspectual nature. V1, in contrast to V2, perfectivizes verbs, which otherwise would be aspectually neutral. This thesis was first unfolded in Leiss (2000). The main line of argument is that, in Old Icelandic, the definite article never appeared in sentence initial (thematic) position, but preferably in rhematic position, with the function to signal (marked) definiteness in a structural syntactic area where usually indefiniteness is unmarked and presupposed. It is further argued that the sentence initial position (thema position) is inherently definite and therefore is, especially in ‘young article languages’, rather resistant against the overt use of the definite article. In other words, the sentence initial position can be defined as a syntactic slot where definiteness is encoded silently. In this area of silent encoding definiteness, otherwise unspecified noun phrases are converted into definite DPs. When verbs are put into this silent definiteness slot, as is the case in Old Icelandic, they are specified as ‘definite verbs’. As definiteness is the nominal counterpart of verbal perfectivity and, further, since perfectivity can be defined as verbal definiteness (central thesis in Leiss 2000; see also Abraham/Leiss 2007), verbs in V1-position are identifiable as perfective verbs. The so-called narrative function of V1 (Van den Toorn 1959, Hallberg 1965, Önnerfors 1997:100-123) will be explained as being derived from its grammatical core function, which is verbal perfectivity. Thus, V1 has the same narrative function as, say, the perfective ‘passé simple’ in French and other Romance languages, which is a typical foregrounder in narrative discourse. In a general outlook it will be shown that the historical present and V1-verbs were functional on a par in Old Icelandic. Evidence comes from Van den Toorn (1959; 1961)) who independently studied the distributions of both V1-sentences and the historical present: There are certain sagas which show a high frequency in the use of V1, simultaneously avoiding the historical present, and vice versa. V1 and historical present thus appear in complementary distributions in the different sagas. Our final claim is that the central function of V1 and the historical present in Old Icelandic was to compensate for the loss of aspectual prefixes in Old Norse in a very specific way. This result encourages one to take a new look at V1-languages in a cross-linguistic and typological perspective.



Abraham, Werner / Leiss, Elisabeth (2007): On the interfaces between (double) definiteness, aspect, word order in Old and Modern Scandinavian. In: Working Papers in Scandinavian Syntax 80 (2007), 17-44.

Hallberg, Peter (1965): Om språkliga författarkriterier I isländska sagatexter. In: Arkiv för nordisk fililogi 80, 157-186.

Leiss, Elisabeth (2000): Artikel und Aspekt. Die grammatischen Muster von Definitheit. Berlin, New York: de Gruyter (Studia Linguistica Germanica; 55).

Hinterhölzl, Roland / Petrova, Svetlana (eds.) (2009): Information structure and language change. New spproaches to word order variation in Germanic. Berlin, New York: Mouton de Gruyter 2009 (Trends in Linguistics. Studies and Monographs; 203).

Önnerfors, Olaf (1997): Verb-erst-Deklarativsätze. Stockholm: Almqvist & Wiksell International (Lunder germanistische Studien; 60).

Schaller, Stephanie (2009): Die Verberststellung bei Deklarativsätzen im Sprachwandel des Deutschen. Untersuchung des Zusammenhangs zwischen Aspekt und Verb-erst-Stellung im Althochdeutschen. Master’s thesis: University of Munich 2009.

Van den Toorn, M[aarten] C[ornelis] (1959): Zur Verfasserfrage der Egilsaga Skallagrimmssonar. Köln, Graz 1959 (Münstersche Forschungen; 11).

Van den Toorn, M[aarten] C[ornelis] (1961): Zeit und Tempus in der Saga. In: Arkiv för nordisk fililogi 76, 134-152.



In this paper I investigate the relationship between gender and focalization. I try and contribute to the identification of the meanings that are likely to be focalized. Some meanings resist focalization ; amongst them is natural gender. In texts we can find such utterances as She was an extraordinary woman but not *She was a woman# (this information being already conveyed by she). In a novel, when a character is introduced, there is generally no explicit mention of gender. The information is of course conveyed, but indirectly, through the character’s first name, for instance.

I also note a convergence between this discourse phenomenon and the structure of the lexicon. There is indeed no noun which expresses natural gender only (the noun woman, for instance, denotes a feminine animate, but further categorizes the person as an adult).

In order to account for this, one cannot argue that natural gender is of no interest to speakers, or that it is not relevant for them. It is in fact so relevant that it is the first distinction to be made in a person’s life, and is not questioned afterwards.

These observations will lead me to discuss some manifestations of grammatical gender ; I will address the issue of pronouns.

Genre naturel et focalisation en anglais

La communication portera sur la question du genre en lien avec la focalisation. Elle sera une contribution à l’identification des objets focalisés et non focalisés.

En effet, toutes les significations ne sont pas focalisables ; le genre naturel est de celles-là. On trouve dans les textes des énoncés du type she was an extraordinary woman, mais pas *she was a woman# (l’information étant déjà donnée dans she). Dans un roman, lorsqu’on présente un personnage, le plus souvent on n’en dit pas le genre. L’information est communiquée, certes, mais de façon indirecte, par le prénom par exemple.

On observe en outre une convergence entre ces phénomènes discursifs et la structure du lexique : il n’existe pas de nom qui dise le genre et seulement le genre (woman, par exemple, réfère à un animé humain féminin mais ajoute le trait sémantique « adulte).

Pour expliquer ces phénomènes on ne peut avancer que la distinction n’intéresse pas les locuteurs, ou qu’elle n’est pas pertinente pour eux. Elle est au contraire tellement pertinente qu’elle est, dans la vie des êtres humains, la première à être faite, et n’est pas remise en question par la suite.

Ces quelques faits nous permettront d’aborder certaines manifestations du genre grammatical en anglais, en explorant notamment le problème des pronoms.



In a classical sense, pragmatics deals with language uses in context, but is also concerned with context-dependent aspects of language structure (Levinson 1983). Several questions arise from these apparent contradictory domains of investigation: How do linguistic information contribute to pragmatic meaning? Is a contextual framework compatible with a linguistic-driven one? How does linguistic meaning interact with contextual information? How does pragmatics interface with semantics? And more generally: How does cognition play a role in context construction and conceptual information? These questions are more crucial if the following assumptions are stated as acceptable: Grammar differs from usage; linguistic meaning is of another kind than pragmatic meaning; words encode concepts, but concepts are not equivalent to meaning; verbal communication is the result of both a coding/decoding process, and an inferential one.

In this communication I will argue that these basic assumptions allow escaping the paradox of pragmatics. I will give empirical arguments linked to the computation of temporal order and causal relations in discourse and the scope of negation in its descriptive and metalinguistic uses.

The conclusion I defend is that pragmatic meaning is the result of a complex interplay between linguistic information and contextual information. More precisely, the semantics of linguistic expressions is not common sense: tenses have directional orientation, causal connectives bear direction features, and the semantics negation involves wide scope. The combination of linguistic and pragmatic information does not lead to a paradox, because contextual information is stronger than linguistic information. In cases of conflict of information, pragmatic accommodation tries to solve it by imposing pragmatic meaning.




It is well known that in active Dutch sentences the ±definite nominal subject (hence SUB) usually precedes the ±definite nominal direct object (hence DOB) if both are placed in the middle field, i.e. between the finite verb and the position of the final verb(s) (which may remain empty) (Nieuwborg 1968, Godin 2005). Nonetheless, some types of violations of this general tendency are reported in the literature (Broekhuis 1997, Haeseryn e.a. 1997). In this paper, we shall focus on utterances like (1).

(1)     toen  [overviel]Vfin    de   soldaten     een       hevig     heimwee         [Ø]Vinf

         then    came over    DEF  soldiers   INDEF   intense  homesickness

   “(and) the soldiers were suddenly overcome by an intense longing for home”

According to Haeseryn e.a., SUB-DOB-permutation is optional if SUB and DOB refer respectively to non-living and living entities (1997: 1322). Nieuwborg points out that there is a tendency to place SUB at the end of the middle field if the main verb and/or SUB is/are abstract and emotionally connoted (1968: 119). In this contribution we would like to present the idea that these analyses, which are merely based on the semantic level lack explanatory power. We will show that better results can be achieved by (a) introducing two additional parameters: ±definiteness and ±complexity, (b) using the pragmatic or meta-informative level as an additional explanatory device (Wlodarczyk & Wlodarczyk 2006: 7-9).


Broekhuis, H. (1997), ‘Twee typen subject.’ In: Nederlandse Taalkunde 2, 1, 35-51.

Godin, P. (2005), ‘Achteropplaatsing van het substantivisch onderwerp in het Nederlands (met bijzondere aandacht aan rouwberichten)’. In: Ph. Hiligsmann e.a., Woord voor woord, Zin voor zin, Liber Amicorum voor S. Theissen. Gent, KANTL, 147-163.

Haeseryn, W. e.a. (1997), Algemene Nederlandse Spraakkunst, Groningen-Deurne: Martinus Nijhoff-Wolters Plantyn.

Nieuwborg, E. (1968), De distributie van het onderwerp en het lijdend voorwerp in het huidige geschreven Nederlands in zijn A.B.-vorm. Antwerpen: Plantyn.

Wlodarczyk, A. & H. Wlodarczyk (2006), Subject in the Meta-Informative Centering Theory. In: Etudes cognitives/Studia kognitywne VII, SOW, PAN, Warsaw, 7-32.



When expressing processes that involve both an agent and a patient, providing the grammatical properties of the constituents involved allow this alternation, speakers of English may choose between active and passive voice, as in:

(1)  Yourcenar wrote that novel in 1951

(2)  That novel was written in 1951 (by Yourcenar)

The present article aims to explore the relation between pragmatics and passive voice as a meta-information strategy (or information-packaging construction, in Huddleston & Pullum’s 2005 terms) in the recent history of English. Whereas (1) and (2) illustrate variation between active and passive voice with referential items (that novel as the local centre of attention (CA) of the active and global CA of the passive, and Yourcenar as the agent in both syntactic variants), examples (3) and (4) evince the same operation with a clausal local CA in the active and an expletive global CA in the passive (followed by an extraposed that-clause):

(3)  People assumed that Yourcenar was the author of the novel

(4)  It was assumed that Yourcenar was the author of the novel

This paper focuses on passive constructions such as (4), where an anonymous subject (AS) is marked by the impersonal pronoun it (Włodarczyk & Włodarczyk 2006a). More specifically, we will analyse the confluence of (i) the conveyance of a different thematic meaning with respect to the active voice, (ii) the markedness that meta-information strategies display from both a pragmatic and a syntactic point of view, and (iii) the ‘agentlessness’ brought about by the informatively empty global CA (the AS), which implies that the active notional subject of the sequence cannot be syntactically recovered but only pragmatically retrieved (Seoane 1993). Similarly, we will approach the meta-informative (pragmatic) validation of this type of passive constructions as utterances introducing old or new information into their contexts of occurrence.

The paper will present and discuss data taken from the electronic Penn Parsed Corpora of Historical English, including the Penn-Helsinki Parsed Corpus of Early Modern English and the Penn Parsed Corpus of Modern British English (more than two million words in total).



Huddleston, Rodney and Geoffrey K. Pullum. 2005. A student’s introduction to English Grammar. Cambridge: Cambridge University Press.

Seoane-Posse, Elena. 1993. “The passive in early Modern English”. Atlantis. Revista de la asociación española de estudios anglo-norteamericanos, vol. XV, 1-2: 191-213.

Włodarczyk, André and Hélène Włodarczyk. 2006a. “Subject in the Meta-Informative Centering Theory”. Études cognitives / Studia kognitywne VII, SOW, PAN, Warsaw: 7-32.



Initially, theories of Dynamic Semantics, e.g. DRT, were not compositional in their nature, if we understand compositionality strictly in the Montagovian sense. Most compositional versions of Dynamic Semantics depended on the prior identification of the anaphoric links performed already on the syntactic level. However, anaphora and co-reference, in general, are complex phenomena that should be considered across several layers of the natural language description.

Self-organising Logic of Structures (henceforth SLS) has been proposed as a basis for the strictly compositional semantic representation of Noun Phrases (NPs) perceived as elements of the discourse. SLS-based semantic representation is not depended on any prior syntactic-based identification of anaphoric links, e.g. expressed by the means of syntactic indexes. In SLS meaning is represented as a relation on states (hearer’s information states): input and output. A state includes: discourse markers, valuation functions and links between discourse markers. SLS is a kind of variable free logic and anaphora or co-reference are directly represented by the links. The anaphora resolution process is perceived as an intrinsic part of the semantic interpretation of the relevant SLS operators. Lack of variables makes the notion of scope irrelevant as a basis for the representation of the multi-quantifier sentences. Instead, two notions of cardinality dependency and independency of two Generalised Quantifiers are formalised. Interpretation of several NPs in a sentence is formulated in terms of possible relation structures that can be expressed and validated by the comparison with a class of situations represented by the verb and constrained by the context of interpretation.

As a result, practically all possible readings of NPs, concerning quantification and plurality (including branching and cumulative interpretations) can be expressed with the help of one SLS dynamic formula.

SLS mechanisms have been applied to the interpretation of several semantico-pragmatic phenomena like: existential presupposition, reference, referential indefinites, anaphora, quantification, or quantification variety.

SLS introduces a mechanism of a compositional, linear, incremental interpretation of the natural language expressions on the basis of binding with the preceding expressions by the virtue of their properties. However, in SLS, this process is not constrained on the level of the meta-information description of the communication process. Possible ways of embedding SLS as a part of a more complex utterance interpretation along the lines of the MIC theory will be discussed.



L’énoncé verbal allemand assertif indépendant possède une première position qui peut accueillir tout syntagme de l’énoncé (mais un seul), quelle que soit sa fonction syntaxique et son statut discursif (focus, topique, non marqué). Le placement d’objets en première position est relativement rare (environ 3% des cas dans notre corpus). De fait, certains auteurs affirment que ce choix indique toujours un focus (sur l’objet ou sur un autre élément), alors qu’il est admis que le sujet en première position peut certes être focalisé, mais donne souvent lieu à des énoncés non marqués.

Nous proposons d’étudier l’occupation de la première position par les objets et les sujets au moyen d’un analyseur syntaxique. En particulier, nous voulons observer l’influence de différents paramètres sur la position des objets et des sujets :

la nature des syntagmes (syntagme nominal, pronom),

le caractère défini, indéfini, démonstratif du syntagme, la personne pour les pronoms personnels,

le temps, la voix et la phase du verbe,

le lemme verbal.

Puisque l’allemand permet de placer en première position ou non aussi bien le sujet d’un verbe passif que l’objet accusatif d’un verbe actif, on peut s’interroger, en particulier, sur le rôle de la diathèse dans la structure informationnelle de l’énoncé.

L’extraction automatique permettra notamment de faire des observations chiffrées sur de grands corpus. Elles seront complétées par l’analyse des énoncés intéressants, pour déterminer la focalisation et la topicalisation.



In my talk 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.







Such concepts as subject, topic, theme, focus are most often used in order to point out to the selected chunks of information as expressed by linguistic utterances. The MIC theory, rather than pretending to replace all the other explanations of the above concepts, constitutes a homogeneous framework aiming at elucidating them altogether. This model-theoretical integration has been made by evoking a psychological notion of "centre of attention" which has previously been explored in the Centering Theory of Grosz B. (1986). In a more general setting, linguists succeeded to coin the term “salience”. Although, this term seems to cover today various pragmatic phenomena, it is being used by the most prominent specialists of that question.

In order to represent linguistic relationships of all kinds (syntactic, semantic, pragmatic, etc.), it is common to use the definition of the formal predicate, which is borrowed from elementary logic. We propose a theoretical framework that is general enough to grasp the problem of linguistic predication efficiently. Our model is inspired by the first definition Aristotle gave of a “predicate” (before he abandoned the notion in favour of that of syllogism).

Broadly speaking, we distinguish utterance constituents relying on centres of attention at three levels:

(a)   base utterances level (with subjects and objects),

(b)  extended utterances level (with topics and focuses) and

(c)   discourse level (with the general theme and the particular themes).

As it happens, the formal (abstract) definition of “predicate” in logic strays away from its original definition and loses the etymological meaning of the expression “to predicate”. The approach we propose gives back to the term “predicate” its original linguistic meaning and makes it possible to build an adequate model of linguistic predication.

As such, the old and new meta-informative statuses should not be mixed up with their groundings. In an older version of the MIC theory (Wlodarczyk A. & H. 1998), the above groundings were not clearly distinguished. We distinguish now between at least three kinds of such grounding motivations of meta-informative old and new status as follows:

(a) the communicative grounding is overt (speech act bound)

(b) the epistemic refinement is related to the process of knowledge acquisition

(c) the ontological (referential) grounding depends on the speaker’s knowledge

In our view, in knowledge stored in long term memory, situations are typed (S) as generic, general, habitual or potential, and the instances (P) of these types can be determined as their dual counterparts : specific, particular, occasional or actual respectively. Entities are continuant as types (S) but, seen as tokens, they are occurrent (P). Situations are persistent as types (S) but, seen as tokens, they are transient (P). The two opposed pairs of notions, generic vs. specific and general vs. particular, directly concern situations, whereas the  other two pairs, potential vs. actual and habitual vs. occasional, are entity-oriented (agents or figures or even other situations seen as agents or figures) which take part in situations. “Potential” is understood here as a 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).





It is generally assumed that word-order concerns the syntactic constituents of the utterance. Problems however occur when one searches for universal definitions of the basic syntactic constituents : verb, subject and object. In the MIC framework (Wlodarczyk A., Wlodarczyk H. 1997, 2003, 2006, 2008, 2009), both subject and object are defined on the meta-informative level as expressing the global and local centres of attention respectively. In human languages, because of the linear order of speech sounds, no judgment may be uttered without pointing at one (or more than one) centre of attention. Thus, we consider centering as a structuring operation not only within the boundaries of an utterance but also within those of the text. Moreover, the linear ordering of constituents in an utterance depends on their meta-informative Old or New status which relies heavily on the pragmatic use of nominal determination and verbal aspect (Stark et al. 2007). We distinguish between base utterances and extended utterances (which should not to be identified with the syntactic notions of “simple clauses” and “compound sentences”). In our pragmatic setting, extended utterances are characterized therefore by the contrast between Old and New status of information: an ‘old’ topic in contrast with a ‘new’ comment and/or a ‘new’ focus in contrast with an ‘old’ background.

Consequently, the MIC approach makes it possible to consider both rigid word order (RWO) and free word order (FWO) languages in a homogenous way. As a matter of fact, we claim that, in both types of languages, word order is among others a meta-informative device which relies on the psychological precondition known as centre of attention. In the former type of languages, word order concerns mainly the first meta-informative level (i.e.: the place of subject and object in base utterances) and in the latter type of languages, it mainly concerns the second meta-informative level (i.e.: the place of topic and focus in extended utterances).

Moreover, we would like to emphasize that after the word-order, it seems to be intonation which is the most universal marker of meta-information.


Grosz B. J., Joshi A.K., Weinstein S. (1995) “Centering: a framework for Modeling the Local Coherence od Discourse”, Computational Linguistics, Vol. 21, Nr 2, 1995, p. 203-226

Stark E., Leiss E., Abraham W., 2007, Nominal Detrmination: typology, context constraints and historical emergence, John Benjamins, Amsterdam/Philadelphia.

References to Wlodarczyk A. and Wlodarczyk H. on the web site of CELTA: