Ordering preferences for postverbal complements in French

Résumé En

This chapter deals with the factors, which influence postverbal complement ordering in French. Existing studies (Blinkenberg, 1928-1933 Berrendonner 1987, Schmitt 1987, Abeillé Godard 2004, 2006) have proposed that the weight and the discourse status (given or new) of the complements play a role, as well as the verb meaning, but they are not based on quantitative data. To remedy this lacune, we annotated extracts from 4 corpora (two written and two spoken) for several factors: complement length, animacy, definiteness, pronominality, collocationality of verb-preposition combinations, verb lemma and verb semantic class (following Dubois and Dubois Charlier 1997’s classification). We first observe a general preference for direct before indirect complements, in all 4 corpora. We also observe that the following factors may contradict this general tendency: the length of the NP (relative to that of the PP) and the verb lemma (with its semantic class) may reverse the preference and drive an indirect before direct complement ordering. In order to evaluate the respective weights of these factors, and to abstract away from the specificity of each corpus, we have also constructed a multifactorial statistical model, following the methodology of Bresnan et al. (2007) and Bresnan and Ford (2010). In this model, the three factors that appeared as statistically significant are the relative length of the complements, the verb combined with its semantic class and the corpus. Contrary to similar studies performed on English or German, pronominality, definiteness and animacy do not play a significant role. In order to complement this result, we conduct two psychological experiments, asking subjects to judge sentences (identical or similar to the ones in the corpora) with equal complement length, and one complement ordering or the other, randomly presented with distractors. Contrary to English and German, again, animacy did not play a significant role, nor did definiteness.