Abstract:

In sentence production, syntactic priming effect refers to a phenomenon that speakers or writers tend to reuse syntactic structures that they have recently processed, and the repeated syntactic structures elicit shorter sentence production latencies than the novel ones. Different theories have been proposed to account for the syntactic priming effect. Among them, both the Residual Activation Theory and the Implicit Learning Theory aim to address speakers’ syntactic choices ratio, but with different theoretical bases on memory and learning, respectively. In addition, the Two-stage Competition Model attempts to explain the syntactic priming in syntactic choices ratio as well as production latencies. Experimental works have demonstrated that syntactic priming stably exists even after excluding the semantic and lexical relevance between the prime sentence and the target sentence, which suggests that syntactic priming actually reflects syntactic processing in sentence production.
In the present study, we explored the cognitive mechanism of syntactic processing in Mandarin spoken sentence production using a syntactic priming paradigm. In this paradigm, participants were firstly instructed to read the prime sentence aloud and then judge whether the sentence’s description matched the prime picture. Then, participants were instructed to describe the target picture using the verb at the bottom. The dependent variables included not only the syntactic choices ratio but also the sentence production latencies. In experiment 1, we manipulated syntax structure of the prime sentence (double object dative vs. prepositional object dative) and verb overlap between the prime sentence and the target sentence (no verb overlap vs. verb overlap). In experiment 2, we added a within-subjects factor of lag by manipulating whether the target picture was presented immediately after the prime sentence, or after two or six intervening descriptions (Lag 0 vs. Lag 2 vs. Lag 6).
For the syntactic choices ratio, results indicated that participants produced significantly more double-object sentences following double-object primes than following prepositional-object primes, and the syntactic priming was much stronger in the verb overlapping condition than in the verb non-overlapping condition. Furthermore, we found that the syntactic priming effect persisted when the prime sentence and the target sentence did not share the verb, whereas the lexical boost effect (i.e. a stronger syntactic priming effect caused by the verb overlap between the prime sentence and the target sentence) disappeared after six intervening descriptions. For the sentence production latencies, results indicated that the repetition of syntactic structures significantly speeded up target sentence production. Besides, we found that verb overlap slowed down the sentence production as the lag between primes and targets increased (also known as the lexical suppression effect) only when the syntactic structure was repeated.
Unifying the different findings in the syntactic choices ratio and the production latencies, we suggest that, apart from the stage where a sentence is planned specifically, sentence production comprises a syntactic construction stage in which speakers map the thematic roles to one structural alternative. These findings are consistent with the Two-stage Competition Model. In methodology, we also suggest that syntactic choices ratio is sensitive to sentence generation process.

Key words: sentence production, Two-stage Competition Model, syntactic priming effect, lexical boost effect