Cristia 2016: Getting a handle on variation in infant laboratory research: A forward approach

Cristia, Alejandrina. 2016: Getting a handle on variation in infant laboratory research: A forward approach. Retrieved from http://bootphon.blogspot.com/2016/01/getting-handle-on-variation-in-infant.html. January 8, 2016.

Comments on the blog post, or the related paper “Infant artificial phonology learning: A meta-analytic approach” available here: https://osf.io/fm4qz/, can be made on the bootphon blog.

Extract of blog post

What is your new year’s work resolution? If I could make a suggestion it would be this: Keep a lab notebook.

How do infants learn their language? To be able to measure the effect of specific variables, many have turned to studying acquisition in vitro, through laboratory learning experiments. For instance, one recent line of work investigates how infants draw generalizations about the sound patterns they hear by having babies hear sets of words exhibiting some pattern, and thereafter presenting them with new words following that pattern and others violating it. If infants show a significant preference between the two types, then it must be because they extrapolated from the initial exposure. This compelling idea received its first empirical support from 2 experiments published by Chambers and colleagues in 2003. Since then, I and several others used the same method to carry out conceptual replications. And yet in a  meta-analysis of all these infant phonotactic learning studies harnessing the power of 600-plus infants, I found that the median effect size was very close to zero (see funnel plot below). Why?