Tooley Kristen M, Brehm Laurel
Department of Psychology, Texas State University, 601 University Drive, San Marcos, TX, 78666, USA.
University of California, Santa Barbara, CA, USA.
Psychon Bull Rev. 2025 Jan 30. doi: 10.3758/s13423-025-02643-3.
Structural priming effects are widespread and heavily relied upon to assess structural representation and processing. Whether these effects are caused by error-driven implicit learning, residual activation, a combination of these, or some other learning mechanism remains to be established. The current study used preexisting data and a novel data analysis approach that links processing at the prime to later processing at the target to better understand the nature of structural priming. This novel analytic approach was applied to total reading times from a previously published structural priming study in comprehension, which provided processing measures of the structurally critical regions of prime reduced-relative clause sentences. These were then used as predictors in a series of hierarchical linear models where analogous processing measures at the target sentence regions served as outcome variables. Separate sets of models were run for prime-target pairs that had the same structure (i.e., abstract priming) and those that had the same structure and initial verb (i.e., a lexical boost). Prime-to-target processing relationships were observed for both types of prime-target pairs, but showed very different patterns. This provides support for the claim that abstract priming effects and the lexical boost are caused by different mechanisms. Additionally, the observed effects were positive and so do not support the error-driven learning prediction that processing difficulty at the prime should lead to greater facilitation at the target. Overall, this novel method provides a new tool for investigating structural priming and processing.
结构启动效应广泛存在，并且在评估结构表征和加工过程中被大量依赖。这些效应是由错误驱动的内隐学习、残余激活、两者的结合，还是其他某种学习机制所引起，仍有待确定。本研究使用了已有的数据以及一种新颖的数据分析方法，该方法将启动项的加工与后续目标项的加工联系起来，以更好地理解结构启动的本质。这种新颖的分析方法被应用于先前发表的一项关于阅读理解的结构启动研究中的总阅读时间，该研究提供了启动项简化关系从句句子的结构关键区域的加工指标。然后，这些指标被用作一系列分层线性模型中的预测变量，其中目标句子区域的类似加工指标作为结果变量。针对具有相同结构的启动项 - 目标项对（即抽象启动）以及具有相同结构和初始动词的启动项 - 目标项对（即词汇促进），分别运行了几组模型。在这两种类型的启动项 - 目标项对中都观察到了启动项到目标项的加工关系，但呈现出非常不同的模式。这为抽象启动效应和词汇促进是由不同机制引起的这一观点提供了支持。此外，观察到的效应是正向的，因此不支持错误驱动学习的预测，即启动项的加工难度应导致目标项有更大的促进作用。总体而言，这种新颖的方法为研究结构启动和加工提供了一种新工具。
