Trial-Level and Contiguous Syntactic Adaptation: A Common Domain-General Mechanism at Play?

Abstract

Garden-path sentences generate processing difficulty due to a more preferred parse conflicting with incoming parsing information. A domain-general cognitive control mechanism has been argued to help identify and resolve these parsing conflicts. This cognitive control mechanism has been argued to underlie adaptation to garden path processing at the trial level (conflict adaptation) and contiguously over the experiment (syntactic adaptation) in independent literature. The strongest evidence for its domain generality comes from garden-path processing being facilitated when preceded by a non-syntactic conflict (e.g., Stroop). This has been reliably observed in the visual world paradigm, which, like Stroop, requires irrelevant visual information to be suppressed. We tested the domain generality of conflict adaptation and its relationship to contiguous syntactic adaptation across four experiments (n = 562). To eliminate the visual object confound, the Stroop task was followed by a sentence-reading task. We observed Stroop and ambiguity effects, but no conflict adaptation in each experiment. Contiguous syntactic adaptation was replicated and most compatible with the parser changing its expectations and/or improving revision. While the data largely fail to support a domain-general cognitive control mechanism, a language-specific one could operate in both trial and contiguous syntactic adaptation and is worth future exploration.