Using the Semantic Priming Project to understand variability in priming

The Semantic Priming Project was a large-scale effort to provide normed priming data of nearly 2000 concepts (Hutchison et al., 2013), and this data was combined with other lexical and relatedness variables in order to investigate how to predict the variability in priming effects. Word length, frequency, neighborhood/set sizes, and part of speech were used to predict priming effects, along with associative, semantic, and corpora-based relatedness measures. Across lexical decision and naming tasks, we found that priming was most commonly related to word frequency and neighborhood size at the lexical level, associative overlap and set size, semantic feature overlap, and a corpora-based pointwise mutual information measure. Predictive variables were mixed across stimulus onset asynchrony and type of prime-target relatedness portraying a medium effect size prediction, displaying the difficulty in capturing the variability in simple priming effects. Item versus subject level regression approaches will also be discussed.

语义启动项目是一项大规模的尝试，旨在提供近2000个概念的标准启动数据（Hutchison 等，2013年），并将这些数据与其他词汇和相关变量相结合，以探究预测启动效应变异性的方法。在预测启动效应时，使用了单词长度、频率、邻域/集合大小以及词性等因素，同时结合了联想、语义以及基于语料库的相关性度量。在词汇决策和命名任务中，我们发现启动效应最常与词汇层面的单词频率和邻域大小、联想重叠和集合大小、语义特征重叠以及基于语料库的点互信息度量相关。预测变量在刺激 onset 同步性和启动-目标相关性类型上存在混合，呈现中等效应大小的预测，显示出捕捉简单启动效应变异性的困难。此外，还将讨论项目层面与主体层面的回归方法。