Preanalyzing L2 PrepositionLearning with Bayesian Mixed Effects and a Pretrained Language Model

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 previousfindings from frequentist analyses and revealnew and crucial interactions between studentability, task type, and stimulus sentence. Given the sparsity of the data as well as high diversityamong learners, the Bayesian method proves most useful; but we also see potential in using language model probabilities as predictors of grammaticality and learnability.