Feature encoding modulates cue-based retrieval: Modeling interference effects in both grammatical and ungrammatical sentences

Abstract

Studies on similarity-based interference in subject-verb agreement dependencies have found a consistent facilitatory effect in ungrammatical sentences but no conclusive effect in grammatical sentences. Existing models propose that interference is caused either by a faulty representation of the input (encoding-based models) or by difficulty in retrieving the subject based on cues at the verb (retrieval-based models). Neither class of model captures the observed patterns in human reading time data. We propose a new model that integrates a feature encoding mechanism into an existing cue-based retrieval model. Our model outperforms the cue-based retrieval model in explaining interference effect data from both grammatical and ungrammatical sentences. We argue that our integrated encoding and retrieval model can provide a basis for experimental and modeling work on understanding interference effects in sentence comprehension.

The Unexpected Lightness of the Main Verb: An Eye-Tracking Study on Relative Clauses and Trace Reactivation

“…It can however be accounted for by taking into consideration the Activation-based Retrieval Model of sentence comprehension (proposed by Lewis and Vasishth 2005; further developed by Lewis, Vasishth and Van Dyke 2006;Vasishth et al 2008;Yadav, Smith and Vasishth 2021;among others). According to the model, one of the working memory constraints on sentence processing is represented by the fluctuation in the activation levels of the encoded (linguistic) items: once a constituent is encountered and encoded in working memory, it naturally starts to decay as a function of time elapsed from its encoding.…”

Modeling Misretrieval and Feature Substitution in Agreement Attraction: A Computational Evaluation

“…A further avenue to be explored is whether additional augmentations of the Lewis and Vasishth (2005) model, hybrid approaches involving both encoding and retrieval components (Yadav, Smith, & Vasishth, 2021), or future deep learning models may be successful in predicting the entire range of empirical data. For instance, the Misidentification stage of our encoding model can be interpreted as a retrieval process.…”

Number Agreement Attraction in Czech Comprehension: Negligible Facilitation Effects

“…Agreement attraction effects both in production and comprehension have been explained using several theoretical approaches. Among the most discussed in the literature are the Marking and Morphing model (Bock et al, 2001 ; Eberhard et al, 2005 ) and Cue-based Retrieval models (Engelmann et al, 2019 ; Lewis & Vasishth, 2005 ; Logačev & Vasishth, 2016 ; Parker et al, 2017 ; Vasishth et al, 2019 ; Yadav et al, 2021 ). In recent years, the Self-organized Sentence Processing account (Smith et al, 2018 ; Villata et al, 2018 ) has been also proposed to explain agreement attraction effects.…”