Less is more: Probabilistic reduction is best explained by small-scale predictability measures

TL;DR

This paper explores how small-scale predictability measures, like n-grams, can effectively explain probabilistic reduction in language models, emphasizing the importance of minimal context for cognitive phenomena analysis.

Contribution

It demonstrates that n-gram representations are sufficient for understanding probabilistic reduction, challenging the need for larger context in modeling cognitive language processes.

Findings

N-gram models suffice for probabilistic reduction analysis

Small-scale predictability measures are effective in cognitive language modeling

Whole utterances are not necessary for observing probabilistic effects

Abstract

The primary research questions of this paper center on defining the amount of context that is necessary and/or appropriate when investigating the relationship between language model probabilities and cognitive phenomena. We investigate whether whole utterances are necessary to observe probabilistic reduction and demonstrate that n-gram representations suffice as cognitive units of planning.