Language learning as uncertainty reduction: The role of prediction error in linguistic generalization and item-learning

Discriminative theories frame language learning as a process of reducing uncertainty about the meaning an utterance by discriminating informative from uninformative cues via mechanisms prediction error and cue competition. Previous work showed that discriminative is affected order in which information presented during learning. Specifically, suffixes, where complex stems precede affixes, promotes better generalization than prefixing, tends to promote item-learning instead. We explored this two large-scale web-based artificial experiments with adult learners (total N = 434), well computational simulations implementing model. While we did not find overall benefit suffixing over prefixing generalization, consistent our theoretical predictions, found participants prefix condition were unable discriminate between frequent, but low-frequency, cues. This resulted them being more likely show incorrect overgeneralization feature for low frequency test items suffix condition. item (although there was type-frequency items), discuss terms methodological limitations empirical paradigm. Taken together, these results underline crucial role plays linguistic have implications how interacts item-learning.