Preferences in Miniature Language Learning as Evidence of Biases in Language Use

Abstract

Miniature language learning has been widely used to study the causality of language universals. However, factors that have been shown to affect language acquisition in other paradigms are understudied in miniature language learning, which calls into question results that claim to uncover universal biases. In this dissertation, I ask how the production modality of the experiment and the similarity to participants’ first language influence miniature language learning behavior to better understand how preferences in miniature language productions are and are not evidence of general cognitive biases underlying language universals. Chapter 1 introduces the dissertation with a brief review of the concept of language universals and how miniature language learning has been used to study them. In Chapter 2, I ask whether typing productions rather than speaking productions leads to more regularization of the miniature language. I find no difference in the structure of productions between participants who type and speak their productions, suggesting that either modality can be used for miniature language learning studies. In Chapter 3, I ask if the structure of participants’ first language (L1) affects a well-known miniature language learning result that is claimed to be due to a general cognitive bias. I expose English, German, and Russian speakers to miniature languages and find that speakers of all three languages produce the same type of miniature language output. This finding supports the existence of an L1-independent cognitive bias to balance production effort against message uncertainty. In Chapter 4, I ask if I would find an L1 influence if the miniature languages were more like participants’ L1. I expose German speakers to miniature languages that vary in the degree of direct similarity to German. I find some evidence of L1 influence, but the pattern aligning with the bias to balance production effort against message uncertainty remains for participants learning the miniature languages closest to German, further supporting the idea that this general cognitive bias is responsible for a statistical language universal. Chapter 5 summarizes the results of the three studies I presented and discusses how my findings contribute to the discipline of Linguistics.