Contenu du sommaire : L'énoncé dans les traditions linguistiques : logos, vākya, kalām, oratio et les autres

This collective introduction sketches out the history of the reflection on the utterance in five cultural areas which played a conspicuous role in the constitution of metalinguistic knowledge: the Sanskrit, Tamil, Greek, Latin, Arab and French traditions. The study of linguistic sequences regarded as meaningful units of communication operating above word level is informed by objectives and interpretative frameworks which sensibly differ according to the traditions and periods taken into account. Yet intersecting points linking these traditions can be identified when attention is paid to the context of the reflection on the utterance, the characteristics of the metalanguage used and recurring stumbling blocks in analysis.

This paper focuses on the early interactions, in medieval India, between the traditions of Pāṇinian grammar (Vyākaraṇa) and Vedic exegesis (Mīmāṃsā) on the definitions of the sentence (vākya). The first section discusses the origins of the debate in ancient grammar, and highlights the existence of a conflict between two of its most prominent authorities: Pāṇini (4th century BC) and Kātyāyana (3rd century BC). A second section deals with the contribution to this debate made by Bhartr̥hari (5th century AD), the author of the Vākyapadīya (“A Treatise on the Word and the Sentence”). It is shown that Bhartr̥hari, even while claiming to reconcile Pāṇini and Kātyāyana, introduces into Vyākaraṇa a definition of the sentence on a semantic (and not merely formal) basis, which he borrows from Vedic exegesis. This borrowing appears as a preliminary to his own theorisation of the sentence as “undivided” (akhaṇḍa) in the Vākyakāṇḍa.

The Ancient Tamil grammatical tradition does not provide in its terminology the means to distinguish between individual words and the larger linguistic entities which the French language can refer to by means of syntagme (english phrase or syntagma) and phrase (english sentence). In practice, however, traditional Tamil grammarians were nevertheless capable of making precise statements concerning elements of Ancient Tamil syntax by making use of terms with a broad linguistic spectrum, such as kiḷavi (or col), or by metalinguistically deploying the flexible “citation-verb” eṉṟal, or by means of deictic variables and the ceytal verb (“to do”). This article examines in detail a few examples, extracted from the vast traditional Tamil grammatical corpus. In doing so, it puts them in their intellectual context, which could be called “The Indian Metalinguistic Area”, where this literature has coexisted for a very long time with Sanskrit grammatical literature.

The Arabic grammatical tradition has two different terms in order to refer to complex linguistic units: kalām, which belongs to the common stock of the language, denotes the utterance as defined in terms of semantic autonomy and communicational relevance, implying a speaker and an addressee; jumla, which appeared at a later period, refers to the sentence as a syntactic construction based upon a predicative link. The paper investigates the way in which the relationship between these two notions was considered by the Arabic grammarians between the late 8th and the mid-14th centuries.

The ancient Greek logos, that we may translate by “statement”, can refer to any signifying linguistic whole that can be analysed into constitutive parts which are themselves signifying words, in fact, that the tradition will define as “parts of logos”. Plato declares a sequence such as “the man learns” to be “primary and minimal” form of logos, described as the intertwining of a noun (onoma) and a verb (rhèma). This complex object, subject to a syntax, possesses the sui generis attribute of saying something, true or false, about the world: Aristotle will call it logos apophantikos “declarative statement” –and it is through leaning on this base that the Greek logical-grammatical tradition will shed light on the diversity of these statements which are other than simple and declarative. The Stoics have made here a decisive contribution whose fecundity one can glimpse in the surviving works of Apollonius Dyscolus.

The sentence (oratio) is at the heart of the concerns of Priscian in introducing syntax in the Latin grammatical tradition. One important question is that of the relationship between sentence and word (dictio), the latter having been analyzed precisely as “a sentence part” (pars orationis). The medieval notions of suppositum and appositum (the origins of which terms can be found in Priscian's text) make it possible to have a function-based syntax, these functions being more or less abandoned by the time of the first Humanist grammar. Although Scaliger (1540) deals with the sentence only indirectly (the De causis only touches upon syntax very briefly), the oratio is frequently mentioned in connection with other linguistic elements, among them dictio which is more than just a part of the sentence. Sanctius (1587), on the other hand, proposes a clear and unified, but also restrictive theory regarding the sentence, based on a small number of rigid principles.

French General Grammar (17th-18th centuries) made the utterance, in the form of the proposition, the relevant level of analysis for the attempt to establish a general theory of language and languages. The renewal of the analytical traditions relevant to the utterance is achieved by a representation of the proposition as a multidimensional object capable of several “points of view”, accounts of which are offered by the numerous propositional typologies developed by the general grammars of the 18th century. This study sets out to illuminate the underlying principles and conditions under which points of view developed as an interpretative motif, from its first formulation in the grammar and logic of Port-Royal up to the start of the 19th century.

The question of the division between syntax and discourse remains a live issue for modern linguistics. Benveniste and Culioli reject such a division. We will review their arguments, as well as the caracterization they give for such products of both syntax and discourse. The sentence is defined by Benveniste as a predication unit; for Culioli it is organized by a “schem of lexis”, that is a predicative schem with a more temporal dimension; in both case, syntax is given as being based on cognition. As for the formal aspect of the utterances analyzed, they are viewed as layouts of morphosyntactic units, rather than as structured configurations. Both linguists apply a heuristic approach linked for Culioli to the search for proliferating embedded structures that call for a dynamic conception of cognition in terms of mental gestures.