Débat sur l'Evidentialité

DEBAT 1 : de 11h30 à 13h

Scott DeLancey (University of Oregon) : The mirative and evidentiality
.The mirative is an evidential-like category which is used to mark information which is new or surprising to the speaker, regardless of whether the information source is first- or second-hand. A true evidential system is fundamentally concerned with the distinction between direct and indirect sources of information. These categories are closely related conceptually, representing the grammatical marking of deviation from prototypical knowledge. These grammatical categories are devices for marking various kinds of deviation from the unmarked, prototypical knowledge status: knowledge obtained by direct experience, which is fully consistent with the rest of the speaker's knowledge of the world.

Gilbert Lazard (Université de Paris) : La catégorie grammaticale du médiatif
Plusieurs langues du Proche-Orient et des Balkans (persan, géorgien, turc, arménien occidental, bulgare, albanais) possèdent une catégorie grammaticale du "médiatif". Elle a pour signifiant une série de formes verbales dérivées du parfait et pour signifié une référence (non spécifiée) à la source de l'information communiquée. Dans les principaux emplois cette source est l'inférence, le ouï-dire ou la constatation inattendue.
Les formes médiatives s'opposent aux formes non marquées par le fait que le locuteur n'énonce pas son message purement et simplement, mais médiatement à travers la connaissance qu'il en a prise et par rapport à laquelle il prend par là même de la distance. C'est en cela seulement que consiste leur valeur. Elles n'impliquent en elles-mêmes ni mise en doute ni supposition. Le médiatif n'est ni un dubitatif ni un présomptif, même si de telles nuances modales peuvent se développer accessoirement

Jan Nuyts (Universiteit Antwerpen) : Subjectivity as an evidential dimension in epistemic modal expressions
In the literature on epistemic modality, there is a recurrent suggestion that one should distinguish between two types of evaluations of the likelihood of a state of affairs, viz. `subjective' and `objective' ones. The distinction is poorly defined, however, and there is hardly any consensus on how it relates to linguistic facts. This paper argues that subjectivity/objectivity is not a matter of two types of epistemic modal evaluations, but involves a separate evidential(like) evaluation, which can (but must not) be expressed jointly with an epistemic evaluation. Different possible definitions of the notion will be considered - notably: the speaker's assessment of the quality of his/her evidence (strong vs. weak) vs. the speaker's indication whether (s)he is solely responsible for his/her views or considers them to be shared by others (including the hearer) - and the relation with other types of evidential dimensions will be discussed. Also the question how the distinction gets expressed in the West Germanic languages will be addressed. The analyses will draw on corpus data from Dutch and German, as well as on data obtained from a series of experiments inquiring into the status of subjectivity (in the alternative definitions) for native speakers of Dutch (to the extent that results will be available: the experiments are currently in full process - this experimental program is performed jointly with Wietske Vonk from the Max-Planck Institute for Psycholinguistics, Nijmegen, The Netherlands).

Vladimir Plungian (Moscou Institute of Linguistics) : The place of evidentiality within the universal grammatical space
Two main problems are discussed:
(1) the relationship between evidentiality and modality, and
(2) a classification of verbal systems with grammaticalized evidentiality in the world's languages.
It is argued that evidentiality may be treated as a modal subcategory in so far as it involves an epistemic judgement (the most frequent combination seems to be that of hearsay and lower certainty). Nevertheless, it is not always the case that an evidential meaning is related to an epistemic one. In this context, a tentative classification of evidential systems is worth considering which opposes primarily two types of evidential markers: "mixed" (i.e., having both epistemic and evidential values) and "pure" ones. Some less known data are analyzed (from verbal systems of Susu, Dogon and Samoyedic languages, among others).

DEBAT 2 : de 8h30 à 10h

Stanka Fitneva (Cornell University) : The cognitive salience of evidentiality : An example from Bulgarian
The goal of this paper is to argue that the linguistic marking of the source of knowledge is a cognitively salient grammatical category. The paper reports the results of two experiments in Bulgarian, traditionally considered to be making a distinction between first hand and second hand information. I propose that the evidential information encoded in Bulgarian is actually best described by two dimensions: how the speaker is related to the acquisition of the information he reports and what the initial mode of acquisition of the information is. V. Friedman (e.g., 1986), on the other hand, suggests that Bulgarian grammaticalizes the expression of the speaker's confirmation of the truth of his statement. In the experiments, 6 and 9 year olds children had to judge the relative reliability of information expressed with the grammatical forms subject of contention. The results are best explained by the two dimensional model of evidentiality, thus providing evidence for the cognitive salience of the grammatical marking of source of evidence.

Sergej Tatevosov : Epistemology in Tsakhur grammar : Pragmatically oriented view
I would like to discuss the semantic characteristics of the particle ji in Tsakhur (Northern Caucasus, Lezgic subgroup of Nakh-Daghestanian languages). This particle exhibits properties that are closely related to semantic domains of both evidentiality and mirativity, yet it can not be regarded as a clear-cut instance of either of these categories. The distribution of ji is motivated by both evidentiality and mirativity, but in a rather unusual way: there is a strong correlation between new knowledge and non-attested evidence as opposed to assimilated knowledge and attested evidence.
I suggest that ji refers to the event of acquiring information (IE) about the situation described (P) without specifying neither source nor novelty of information. Depending on whether IE occurs at the same time and place as P or not, ji yields two considerably different interpretations.

The situation is not personally witnessed by a speaker, the statement is based on the reported or inferring evidence; change of speaker’s knowledge is foregrounded (thus resulting in ‘new non-attested’ effect).
The situation is personally witnessed by a speaker, the meaning of the whole clause does not involve the change of knowledge but rather a result of this change, that is, speaker’s experience (interpretation being ‘assimilated attested’).

Therefore, ji refers not only to a dynamic event of acquiring information (=change of speaker’s knowledge) but also to a result state of such an event (=acquired knowledge itself). The difference between two interpretations is in a certain sense parallel to that between perfect and resultative (Nedyalkov 1988): ji can represent an event of acquiring information either in a perfect-like (involving both result state of this event and its preceding development) or resultative-like (result state only) manner.
Nedyalkov, Vladimir P., ed. 1988. Typology of resultative constructions. Amsterdam: Benjamins

Frank Brisard (University of Antwerp) : Conceptual crossroads : On the corner of time and evidentiality
Taking some classic remarks about the presumed semantic distinctions between will and be going to (gonna) as the point of departure, I will demonstrate the fundamental difficulties involved in discussing tense forms with a considerable overlap in temporal denotation from the perspective of temporal organization alone. The model I will use to demonstrate this adopts a phenomenologically and cognitively inspired stance on the human activity of interpreting time and applies this to the grammatical expression of temporal categories. Together with the idea that evidentiality concerns the linguistic expression of epistemology, this brings us to an appreciation of the tense paradigm as functionally equivalent to many other structures in language dealing with the information status of linguistic contents. Concretely, the forms will and be going to will be calibrated on a conceptual continuum comprising both degrees of givenness and of (perceptual) presence/absence as the main criteria from which temporal denotations follow. Will, it is claimed, only covers the realm of "projected reality", where events are neither given nor present but based on premises that are necessarily given (a meaning very much apparent in the conditional construction), whereas be going to concentrates on states of affairs that are non-present yet given (or "evoked reality"). The latter definition, though seemingly at odds with intuitive beliefs about the internal structure of the future, merely takes into account the grammatical form of the construction, which figures the verb to be in the present tense, and aims at harmonizing this grammatical fact with the obvious referential force of the construction. The explanations will be formulated in terms of the pragmatic need on the part of the language user to seriously constrain the potential for referring to states of affairs situated in the future, whereby the future is seen as taking part in the even more extensively construed realm of irreality. The data for the analysis are taken from a combined British/American English corpus.

Métiyé Meydan (Université de Marne la Vallée) : Un ou des suffixe(s) évidentiel(s) -mis en turc

Anaïd Donabédian (INALCO/CNRS-Paris) : Evidentialité et hiérarchie communicative en arménien occidental
Comme un certain nombre de langues du continuum balkanique (albanais, macédonien, bulgare, turc), l’arménien occidental connaît cette catégorie dont le nom fluctue autant que les approche linguistiques qu’elle a suscitées : testimonial, admiratif, évidentiel, médiatif, inférentiel, etc. Toutes ces approches définissent cette catégorie en accordant un rôle central à la notion de « source de savoir », incarnée tantôt par des emboîtements énonciatifs (informations de seconde main), tantôt par des processus de raisonnement logiques (inférence, abduction, déduction).
Une recherche contrastive fondée sur une analyse détaillée de corpus en arménien et en russe (Donabédian, Bonnot, Seliverstova : 1997) a mis en évidence la coïncidence des valeurs pragmatiques des énoncés évidentiels en arménien avec des énoncés russes accentuellement marqués (et non pas, comme on aurait pu l’attendre, avec des particules issues de ‘dire’ ou renvoyant à la source du savoir).. Ces résultats confirment le lien très étroit entre les différentes valeurs pragmatiques de cette catégorie, déjà mis en lumière pour le turc par (Slobin&Aksu : 1982 et 1986), qui suggèrent la primauté de la valeur admirative sur les autres valeurs pragmatiques, primauté très clairement mise en évidence par (Basturk, Danon-Boileau, Morel : 1996) et (DeLancey :1997).
La thèse que nous défendons, à la suite de ces travaux, est que la valeur modale de l’évidentiel constitue un type particulier d’assertion dans lequel un modus [X] préexiste à un dictum non stabilisé [P/P’]. Cette analyse permet de rendre compte de l’ensemble des valeurs pragmatiques typiques (admiratif, inférentiel, ouï-dire), ainsi que de valeurs secondaires, discursives et textuelles, de l’évidentiel (causalité, polémique, jugement de valeur, style indirect libre, etc.).
La description du fonctionnement de l’évidentiel met en jeu des phénomènes intonatifs (notamment la place de l’accent de phrase), et des contraintes concernant la segmentation des énoncés évidentiels en thème et rhème. Or, à ce jour, même si certains travaux évoquent le rôle de l’intonation dans la répartition entre les valeurs de l’évidentiel, aucune étude systématique portant sur les schémas intonatifs de ces énoncés n’a été conduite.
Le matériau de base de notre étude empirique est constitué de données attestées et enregistrées, avec des contextes longs, permettant de mettre en évidence leur structure communicative.
L’étude montre que la segmentation thème/rhème des énoncés évidentiels est extrêmement contrainte, mettant en œuvre le contexte (hiérarchie communicative), la place de l’accent de phrase et le schéma intonatif, ce qui n’est pas sans rappeler les contraintes de même ordre propres à l’interrogation et à l’injonction. Cela confirme la thèse de l’interprétation strictement modale que nous proposons pour l’évidentiel en arménien, et dont il conviendrait d’étudier la validité pour les autres langues citées ci-dessus.

Patrick Dendale (Université d’Anvers-UA & (CELTED (Metz) Fax. +32 3 239.03.25	Liliane Tasmowski (Université d’Anvers-UA) Tél. +32 3 820.28.21
Université d’Anvers-UA Département de linguistique Universiteitsplein 1 B-2610 Wilrijk Belgique Fax. +32 3 820.28.23