When is semantic priming automatic? Instrument and location participant role priming as a case study
Su, Chung-I Erica
MetadataShow full item record
Recently, semantic priming has been used to argue for a close semantic relationship between verbs and their typical participant role fillers (Ferretti, et al., 2001; McRae, et al., 2005). This suggests that verbs automatically provide more situational knowledge than is typically assumed in many linguistic frameworks (see Elman, 2009 for discussion). Moreover, it would provide a mechanism for how some linguistic and real-world knowledge can be rapidly coordinated to immediately guide on-line language parsing and interpretation. However, it has yet to be conclusively demonstrated that the semantic priming results reported by Ferretti et al. are due to automatic activation. They may instead be due to strategic processes. In this study, I attempt to answer three questions. First, was the instrument priming effect observed by Ferretti et al. and in Experiment 1 of this dissertation (which was a direct replication of Ferretti et al.'s instrument priming study) the consequence of automatic activation of typical instrument role fillers by instrument verbs? The results of Experiments 2-3, which employed experimental paradigms that are argued to be immune to strategic influences, suggest that Ferretti et al.'s instrument priming effects was not due to automatic activation of typical instrument role fillers by instrument verbs. The second question is: Was the instrument priming effect observed by Ferretti et al. and in the replication (Experiment 1) influenced by strategic processes? The results of Experiments 4a-c, which employed baseline primes that were unlikely to contribute to any instrument expectation strategy, all showed that the magnitude of instrument priming obtained in Experiment 1, was significantly reduced when unrelated instrument verb primes that contributed to strategy formation were replaced with more neutral primes. These results clearly indicate that the instrument priming effect observed by Ferretti et al. was not due to automatic processes, but instead was due to the use of two response strategies that were promoted by the confluence of task and materials variables. The results of Experiments 5a-b and Experiment 6 provide evidence for this claim. The results reported on in this dissertation are important theoretically because they undermine the representational and processing claims of one widely-cited view of verb participant role information, namely the verb-specific concepts theory (e.g., Ferretti, et al., 2001). According to the verb-specific concepts view, verbs and their typical participant role fillers overlap in representations, and hence recognition of verbs should automatically activate their typical participant role fillers, and vice versa. Because this view has been used to support sentence processing theories (e.g., Elman, 2009), the results reported in this dissertation have implications for both the organization of lexical semantic representations and sentence processing. Additionally, the experiments in this dissertation make a number of methodological contributions. Among these contributions is that they provide a direct comparison of three types of neutral baseline for semantic priming studies. The results of the neutral baseline experiments show that these three baselines are roughly comparable. These experiments also delineate a set of factors commonly present in semantic priming studies that promote the development and use of response strategies.