TIPA. Travaux interdisciplinaires sur la parole et le langage (Dec 2014)
Les verbes à objet implicite dans le français parlé :
Abstract
An element X is said to be more prominent than an element Y whenever W receives more emphasis, and is more salient than Y. On a phonological level, the highest degree of prominence of a linguistic item X compared to another linguistic item Y is the case where X is phonologically realized and Y is not, as illustrated for example by the phenomenon of verbs with implicit objects. Indeed, some verbs are semantically transitive and hence necessarily require a semantic object, but can nevertheless appear without a direct object that is phonologically realized : the semantic object is then implicit. Thus, a verb like manger ‘eat’ for example can be used without an explicit direct object, as in (1) :1. Jean est en train de manger.‘John is eating’The utterance is most naturally understood as meaning that John is eating something, a thing whose precise identity does not matter : manger is here interpreted with an implicit indefinite object (or IIO). But if I utter (2) while throwing a ball to someone,2. Attrape !‘Catch (it)’my utterance will then be understood as an incentive to catch not something or other, but precisely the ball I am throwing : attrape ‘catch’ is then interpreted with a definite implicit object (or DIO). However, other verbs prima facie seem to have more difficulties admitting implicit objects. Thus, the utterances in (3) and (4) for example do not sound fully acceptable ; the object omission here appears quite difficult with the verb faire ‘make’ :3. Marie est en train de faire‘Mary is making’4. Fais!‘Make!’Which factors can explain the acceptability of the omission of the object complement, and the definite or indefinite interpretation of the implicit object ?Several approaches of a lexicalist kind have been defended in the literature – among which many for the English language – on which the ability of a verb to admit IIOs or DIOs corresponds to a lexical property of the verb. Thus, Fillmore (1986) first and then more recently Gillon (2012) have contended that this corresponds to a property that is determined within the lexicon in a fully arbitrary way, and cannot be inferred from other semantic properties of the verb. Mittwoch (1971, 1982) and Noailly (1998) prefer to analyze the ability of admitting IIOs or DIOs as the result of aspectual properties of the verb ; thus, verbs admitting IIOs would be atelic verbs, while verbs admitting DIOs would be telic. Finally, Rappaport Hovav and Levin (1998) develop a so-called ‘event structure’ approach, according to which causative verbs do not admit object omission due to their complex event structure. Yet, an obvious limit of the lexicalist approach is that such analyses are often based on written corpora, or even on invented examples, that might thus be possibly biased, the latter because they rely on introspection, the former because they reflect a very constrained use of the language, that can only imperfectly illustrate the phenomenon as it is realized in spoken French.In this paper, I rely on a corpus of Parisian Spoken French from the 2000s, the CFPP 2000, to show that lexicalist approaches are unable to account for the phenomenon of object omission as it is realized in spoken French. I argue that object omission is guided by communicational criteria, according to the following principle :The phonological prominence of the verb on the semantic object (that is, the phonological realization of the verb and non-realization of the semantic object) reflects an informational prominence : a semantic object is left implicit if and only if the action described by the verb is more salient than the semantic object on an informational level. Based on the corpus, I distinguish three cases. First, as has been put forward by Lambrecht & Lemoine (2005), whenever a DIO is pronominalizable, it corresponds to a topical information, by contrast with the verb that is part of the comment : the semantic object is what is talked about, the given information, whereas the verb brings the genuine informational import about the topic, hence the informational prominence of the verb on the object. This kind of implicit object thus supports Butt & King’s (2000) hypothesis, according to which a nominal constituent can be left implicit if and only if it corresponds to old information. By contrast, whenever the DIO does not receive an explicit antecedent and hence cannot be pronominalized, it no longer necessarily corresponds to old information, pace Butt & King’s (2000) claim. But Lambrecht & Lemoine’s analysis, according to which the DIO is part of the focus information, might be too strong as well. Rather, I argue that Butt & King’s notion of ‘completive information’ might be more suitable to describe non pronominalizable DIOs. Furthermore, Lambrecht & Lemoine overlook another type of non pronominalizable DIOs, which, although less common, also appear in our corpus, namely DIOs that are anaphoric, but receive as antecedent another implicit object rather than an explicit antecedent : although non pronominalizable, these DIOs are closer to the preceding case of pronominalizable DIOs. In each of these cases however, the data remain consistent with our main hypothesis, according to which a verb is informationally prominent over the implicit object : indeed, whether it is topical information or completive information, they remain informationally secondary with regard to the focus. Finally, in the case of IIOs, the informational prominence of the verb first and foremost result from the fact that the semantic object is of an indefinite kind : some clues might possibly be inferred from the context regarding the semantic value of the implicit object, but its precise identity does not matter, and it is the verb that bears all the informational weight.
Keywords
