Previous work has examined the influence of verb repetition on syntactic priming [1]. We report four experiments that used a confederate-scripting dialogue method [2] to examine whether syntactic priming is also affected by noun repetition. In all four experiments, we measured the proportion of Prepositional Object [PO] (e.g., The witch giving the parcel to the princess) and Double Object [DO] (e.g., The witch giving the princess the parcel) picture descriptions that participants produced after hearing a prime sentence.
Experiment 1 investigated whether syntactic priming was enhanced by repetition of all three NPs. We crossed two factors: Prime Type (PO vs. DO), and NP Overlap (Overlapping vs Non-overlapping NPs). In all conditions, the verb differed. Priming occurred in all four conditions, but was greatly enhanced when the NPs in prime and target overlapped.
Experiments 2 and 3 explored whether NP overlap affected priming when prime and target had syntactic differences. Experiment 2 comprised 4 conditions: PO, DO, PO-for (e.g., The witch buying the parcel for the princess) and Intransitive (e.g., The witch smiling). The PO-for and PO-to conditions displayed comparable priming effects, as predicted if the locus of syntactic priming involves a representation that is not specified for closed-class items [1,3]. In Experiment 3, the PO-for condition was replaced by a Low-Attached (LA) condition containing "to" (e.g., The witch burning the parcel to the princess). The LA-to and PO-to conditions showed comparable priming, with participants generally producing the PO target form. Since all accounts assume that PO and LA involve different structures, this result suggests that lexical overlap does not simply enhance the priming effect of a syntactic frame.
Experiment 4 explored whether the priming found for the LA-to structure was driven by residual activation of the prime's prepositional phrase. For example, participants might construct a target sentence on the basis of the activated phrase "to the princess". We replicated Experiment 3's design, except that only the beneficiary overlapped in prime and target. Priming was reduced throughout, and the LA sentence no longer primed the PO target. We hypothesise that this reflects the incremental nature of production: because structure is built up left to right, lexical overlap can influence syntactic processing in early word order positions only, before syntactic structure has been wholly determined. We are now investigating whether a repetition of the patient in the LA prime produces priming similar to the PO condition. We interpret these results in terms of theories of sentence production.
References
[1] Pickering, M. P., & Branigan, H. P. (1998) The representation of verbs: Evidence from syntactic priming in language production. Journal of Memory and Language, 39, 633-651.
[2] Branigan, H. P., Pickering, M. P., & Cleland, A. A. (2000). Cognition, 75, B13-B25.
[3] Bock, J. K. (1989). Closed-class immanence in sentence production. Cognition, 31, 163-186.