Retour vers liste

Détail de la contribution

Auteur: Patricia AMARAL

Co-Auteur(s): Kirby VARNADOE-RUSS

An experimental approach to Levinson's M-based implicatures

Abstract/Résumé: The Gricean distinction between “what is said” and “what is implicated” is assumed in most theories of utterance interpretation. However, there is ongoing debate concerning its robustness in experimental settings (see Doran et al. 2012 for discussion). In this paper, we follow Levinson’s (2000) theory of Generalized Conversational Implicatures (GCIs) and test experimentally the generation of M(anner)-based implicatures in Spanish. We focus on three types of expressions: double negatives, periphrastic causatives, and the conjunction y ‘and’ followed by an additive adverb, as in (1)-(3). The choice of the double negative in (1), the periphrastic causative in (2), and the adverb with the conjunction ‘and’ in (3), instead of their unmarked counterparts (in parenthesis), licenses the inference that the speaker does not want to convey the meaning associated with the unmarked form, triggering the implicatures (+>) in (1’),(2’) and (3’). Our results indicate that there are significant differences among these expressions with respect to the generation of M-based GCIs. (1) No es imposible (Es posible) que llueva hoy. ‘It is not impossible (it is possible) that it will rain today.’ (1’) +> The possibility that it rains today is pretty low. (2) Pepe hizo parar (paró) el coche. ‘Pepe made the car stop (stopped the car). (2’) +> Pepe stopped the car in an indirect manner or with difficulty. (3) Ayer Roberto fue a la biblioteca y también (y) trabajó en su libro. ‘Yesterday Roberto went to the library and he also (and) worked on his book.’ (3’) +> Roberto went to the library and he worked on his book elsewhere (not in the library). We tested 35 native speakers of Spanish from Mexico using Survey Monkey. Subjects were presented with a questionnaire containing 12 critical items (4 per type of expression) and 24 fillers. For each sentence, subjects were given four possible interpretations: (i) the normal interpretation, (ii) the abnormal (marked) interpretation, and (iii) two unrelated sentences. Our results show that overall subjects are significantly more likely to choose an abnormal (marked) interpretation to a marked form than a normal interpretation (p<0.05). A comparison between the types of expressions shows that the periphrastic causatives constitute the class with higher significance with respect to the predicted choice of a marked interpretation. We discuss the differences and interpret this finding as an indication that M-based implicatures seem to be preferred with sets of contrasting paradigmatic options that allow for a clear-cut notion of markedness. This paper contributes to empirical study of GCIs and brings data from Spanish to a theoretical discussion that has been based mostly on English data.