Reanalyzing L2 Preposition Learning with Bayesian Mixed Effects and a Pretrained Language Model

TL;DR

This paper employs Bayesian and neural models to analyze Chinese learners' English preposition understanding, revealing interactions between ability, task, and stimuli, and exploring language model probabilities as predictors.

Contribution

It introduces a Bayesian mixed-effects approach combined with pretrained language models to analyze second language preposition learning data.

Findings

Bayesian models effectively handle data sparsity and learner diversity.

Interactions between ability, task type, and stimuli are identified.

Language model probabilities show potential as predictors of grammaticality.

Abstract

We use both Bayesian and neural models to dissect a data set of Chinese learners' pre- and post-interventional responses to two tests measuring their understanding of English prepositions. The results mostly replicate previous findings from frequentist analyses and newly reveal crucial interactions between student ability, task type, and stimulus sentence. Given the sparsity of the data as well as high diversity among learners, the Bayesian method proves most useful; but we also see potential in using language model probabilities as predictors of grammaticality and learnability.