Meta-analysis: Cumulating language acquisition research
There is a large literature on various aspects of language acquisit
ion. However, it can be hard to draw conclusions from just looking at the results of different studies. For instance, imagine Study A that has found babies can recognize words at 8 months of age, and Study B that finds that 8 months old babies cannot do so, but only recognize words at 12 months. How do we interpret these findings?
Meta-analysis is a tool that allows us to quantitatively put together the evidence from two or many studies. For instance, we might see that Study A only tested 15 participants, while Study B tested 30. This means that Study B should be given more weight – something that meta-analysis can do. Another thing it allows us to do is to look at factors that might influence found effects – for instance, we might see that Study A was conducted in the USA, while Study B was conducted in Japan, or that Study A used very easy words and Study B used more difficult ones. By adding such factors to the weighted meta-analytic model, we can understand how strong evidence from existing studies is, and which factors play a role in explaining them. It thus can give us more general insights on a phenomenon than a single study could.
You can explore some meta-analyses on MetaLab: https://langcog.github.io/metalab/
Example data from MetaLab: Word segmentation effect sizes by infant native language.