Speaker: Dr Giulia Bovolenta, Research Associate, University of York

Title: Different types of violation expectation and their effects on syntactic structure learning

Abstract:

In this talk, I will present a series of artificial language learning experiments that investigate the role of prediction error in learning a foreign language. There are multiple ways in which violating expectations could support the acquisition of new syntactic structures:

On a linguistic level, encountering a syntactic structure with verbs not normally used with it can lead to enhanced structural priming (i.e., an increased likelihood to produce or to expect that specific structure again). This ‘inverse frequency’ priming can persist over time, suggesting a form of implicit learning. On a more general level, prediction error has been shown to promote the formation of new memories: stimuli that do not conform to prior experience are remembered better than those that do. If individual sentences are better remembered for violating expectations, they might provide a stronger base on which to form abstract syntactic representations, offering another potential route into the acquisition of new structures.

I will present a series of experiments where participants learned an artificial language with two competing syntactic structures, active and passive. Prediction error was manipulated both at the linguistic level (manipulating the co-occurrence statistics between structures and individual verbs) and at the event level (making entire sentences surprising). Results suggest that multiple error-based learning mechanisms may be active at the same time, with different outcomes for learning.

Email Mingtong LiLink opens in a new window for Teams link.