The role of animacy in sentence production: Evidence from French
Abstract
We present evidence from a sentence recall paradigm in French bearing on the role of conceptual information, specifically animacy, in sentence production. Previous studies have found that rates of inversion errors in production indicated a preference to produce more conceptually accessible DPs (on a scale such as that of Keenan and Comrie, 1977) earlier in a sentence. However, there is a debate as to the stage of sentence production at which accessibility plays a role. Models of sentence production have proposed separate stages for the assignment of lexical items to grammatical roles and for the linear ordering of constituents (see Bock and Levelt 1994). Previous studies have found evidence for an accessibility effect on grammatical assignment only (Bock and Warren 1985, Tanaka 2011) or an effect on both levels (see Branigan et al., 2007 for a review).
We studied the recall of three sentence types: transitives with voice alternations [1], sentences with coordinated DPs [2], and ditransitives [3]. Modern French does not have an equivalent to the English dative alternation, however word order variation in ditransitives is attested; postverbal arguments of ditransitive verbs may appear in NP-PP order or PP-NP order, although NP-PP order is considered to be canonical (see Thuilier 2012 for corpus evidence). Preliminary results show that for transitives (N=30), speakers were more likely to produce inversions from passive to active when doing so would avoid having an inanimate argument in subject position (39% of otherwise correct recalls) than when both arguments were animate (11.4%, p < .01 in a mixed-effects logistic regression analysis). Active-to-passive inversions were very rare (only 2 instances) and were therefore excluded from analysis. No significant differences were found for inversions in coordinations (N=32) based on animacy. For ditransitives (N=33), the results did not show a preference for ordering animate arguments before inanimates. On the contrary, there was an interaction (p < .01) such that ditransitives with inanimate themes had the most inversions from PP-NP to NP-PP order (22%) and the lowest percentage of inversions from NP-PP to PP-NP order (4.5%), with animate argument in between (15% and 12%, respectively).
Our results for transitives and coordinations support a role for animacy in grammatical role assignment but not for simple linear order. The ditransitive result, however, is not predicted by conceptual accessibility. Because of the free recall nature of the task, subjects were able to vary factors such as length and definiteness that were controlled in the presented items. We are in the process of coding the productions for these factors to gain insight into the nature of this surprising effect.
Animacy alternatives (), order alternatives [], recall prompt < >
(1) , (le voleur/le revolver) a été trouvé par le policier.
'At the end of the alley, the thief/revolver was found by the policeman.'
(2) [les traîtres et (les lâches/les échecs)|(les lâches/les échecs) et les traîtres]
'This young man has always avoided traitors and cowards/failures.'
(3) [(un agent commercial/un nouveau budget) à un décorateur|à un décorateur (un agent commercial/un nouveau budget).]
'The project manager assigned a commercial agent/new budget to a decorator.'
Origin | Files produced by the author(s) |
---|