Modeling misretrieval and feature substitution in agreement attraction: A computational evaluation

Abstract

We present a self-paced reading study investigating attraction effects on number agreement in Eastern Armenian. Both word-by-word reading times and open-ended responses to sentence-final comprehension questions were collected, allowing us to relate reading times and sentence interpretations on a trial-by-trial basis. Results indicate that readers sometimes misinterpret the number feature of the subject in agreement attraction configurations, which is in line with agreement attraction being due to memory encoding errors. Our data also show that readers sometimes misassign the thematic roles of the critical verb. While such a tendency is principally in line with agreement attraction being due to incorrect memory retrievals, the specific pattern observed in our data is not predicted by existing models. We implement five computational models of agreement attraction in a Bayesian framework, finding that our data are better accounted for by an encoding-based model of agreement attraction, rather than a retrieval-based model. A novel contribution of our computational modeling is the finding that the best predictive fit to our data comes from a model that allows number features from the verb to overwrite number features on noun phrases during encoding.

Semantic Attraction in Sentence Comprehension

“…However, the fact that agreement attraction effects occur in production and in comprehension and that they largely (but not entirely) pattern together has led researchers to consider these accounts also as explanations for comprehension, where they had some considerable success. For instance, both accounts explain the singular–plural asymmetry (Bock & Cutting, 1992; Bock & Eberhard, 1993; Wagers et al., 2009) and the finding that comprehenders, when asked, tend to report the correct subject but with incorrect number marking (Avetisyan et al., 2020; Paape, Avetisyan, Lago, & Vasishth, 2021; Patson & Husband, 2016; Schlueter et al., 2019).…”

Estimating the True Cost of Garden Pathing: A Computational Model of Latent Cognitive Processes

“…Our results also highlight the need for more sophisticated data analysis in psycholinguistics that goes beyond comparing condition means. One fruitful approach that we used in the present work is to establish a linking hypothesis between online measures such as reading times and offline measures such as acceptability judgments or answers to comprehension questions and to analyze both within a single latent‐process model (e.g., Ferreira & Yang, 2019; Paape et al., 2021; Nicenboim & Vasishth, 2018). Another promising avenue is to model reading times and reaction times with mixture distributions (e.g., Vasishth, Chopin, Ryder, & Nicenboim, 2017; Vasishth et al., 2017; Lissón et al., 2021), that is, to assume that the data points within conditions may be generated by different processes with different mean latencies and potentially different amounts of variability.…”

“…The model structure is shown in Figure 1. For other recent work that applies MPTs to sentence processing, see Logačev and Dokudan (2021) and Paape, Avetisyan, Lago, and Vasishth (2021). Our Bayesian MPT model is implemented in Stan (Stan Development Team, 2021) and fitted via the rstan package (Stan Development Team, 2020) in R (R Core Team, 2021).…”

Individual Differences in Cue Weighting in Sentence Comprehension: An Evaluation Using Approximate Bayesian Computation

“…Parameter estimation for the LV05 model is difficult because the model is nondeterministic and its likelihood cannot easily be expressed analytically, unless we drastically simplify the model (e.g., in Lissón et al, 2021;Nicenboim & Vasishth, 2018;Paape et al, 2020). Therefore, we cannot employ the standard Bayesian estimation method for estimating parameters of the LV05 model.…”