Local versus Global Statistical Learning in Language
Luca Onnis, Shimon Edelman · 2019
A generally held assumption about human statistical learning is that learners keep track of the global statistics of the elements of interest across the entire set of stimuli they are exposed to. In naturalistic settings, this assumption is problematic because it requires that the cognitive system keep track of an exponentially growing number of relations while determining which of those relations are relevant and which are not. We investigated a more plausible assumption, namely, that statistical learning proceeds incrementally, using small windows of opportunity in which the relevant relations are assumed to hold over temporally contiguous objects or events. This local statistical learning hypothesis was tested on two learning tasks, one involving non-adjacent structures and the other -- novel word-to-world mappings. Our results suggest that human subjects make use of simple general-purpose cognitive heuristics that exploit temporal contiguity and contrast, leading to superior learning in tasks that belong on two different levels of language acquisition.