Learning Through Prediction: A Case of Verb Bias Learning

Linguistic prediction, which emerges from acquired knowledge, is a pervasive process in language comprehension. In language acquisition theories, prediction has also been suggested as a key factor driving the implicit learning process. However, how prediction develops as learning unfolds and how it, in turn, drives the learning process remains unclear. This study examines the relationship between prediction and learning, with a focus on three key questions: (1) whether learning leads to prediction, (2) whether prediction motivates learning, and (3) whether individuals’ prediction skills are stable across tasks. We first replicated the malleability of verb bias in adults (Ryskin et al., 2017) and their ability to predict using verb semantics (Nation et al., 2003). Beyond replications, our results revealed that learners who successfully updated their verb biases showed a higher proportion of first fixation to the instruments than to the animals upon hearing an instrument-trained verb, indicating that individuals’ verb bias predictions were modulated by the success of learning, and they were able to use the newly learned verb bias knowledge to generate anticipatory eye movements after training. To understand whether prediction might in turn motivate learning, we found that the more divergent learners’ initial verb bias knowledge was from the received training type, the greater the learning effects occurred, linking prediction errors to learning outcomes. Finally, adults’ ability to predict linguistic items based on verb information remained stable across language tasks. Taken together, these results elucidate the dynamic interplay between prediction and learning, providing empirical support for prediction-based learning frameworks.