The grammaticality asymmetry in agreement attraction reflects response bias: Experimental and modeling evidence

Cue-based memory retrieval models have come to dominate theories of how linguistic dependencies such as subject-verb agreement are formed in comprehension. A cornerstone of the arguments in favor of such models is the grammaticality asymmetry in agreement attraction—the observation that non-subject noun phrases may interfere with the computation of agreement in ungrammatical, but not grammatical, sentences. This asymmetry is most often accounted for via the interaction of categorical number marking and the dynamics of retrieval interference. In the present paper, we challenge this interpretation of the grammaticality asymmetry, and instead argue that it is an artifact of response bias. In three forced-choice judgment experiments, we show that neutralizing response bias eliminates the grammaticality asymmetry in speeded acceptability judgments. We argue that this finding, together with the response time patterns in our experiments, favors an account that attributes attraction effects to a continuous representation of number, rather than to the dynamics of cue-based retrieval. We implement a model of grammaticality judgments, which links the continuous representation of subject number to the rate of evidence accumulation in a diffusion process. This model accounts for the presence or absence of the grammaticality asymmetry through shifts in the decisional starting point (i.e. response bias). The results have implications for the representation of number in language production and comprehension, the role of memory in the processing of linguistic dependencies, and highlight the importance of monitoring for response bias effects in judgment tasks.

基于线索的记忆检索模型已成为理解中语言依存关系（如主谓一致）形成理论的支配性观点。支持此类模型论点的基石之一在于一致吸引中的语法对称性——即观察到非主语名词短语可能干扰不合法句子中的一致性计算，而在合法句子中则不会。这种不对称性通常通过类别性数量标记与检索干扰动态的相互作用来解释。在本研究中，我们挑战了这种对语法对称性的解释，相反我们认为它是一种响应偏差的产物。在三项强制选择的判断实验中，我们表明通过消除响应偏差消除了快速接受度判断中的语法对称性。我们认为这一发现，结合我们实验中的响应时间模式，支持将吸引效应归因于数量的连续表征，而不是基于线索检索的动态。我们实施了一种语法判断模型，该模型将主体数量的连续表征与扩散过程中的证据积累速率相联系。该模型通过决策起点（即响应偏差）的变化来解释语法对称性的存在与否。这些结果对语言生产与理解中数量的表征、记忆在语言依存关系处理中的作用产生了影响，并突出了在判断任务中监测响应偏差效应的重要性。