Ecole Normale Supérieure de Paris
A meta-analytic view on language acquisition: Gaining new insights from "old" data
A theory is only as strong as the evidence that supports it. Ideally, we know the answers to the following questions: Is the phenomenon real? Does it generalize across different populations, stimuli, and languages? Single studies are not made to respond to such questions, but their answers are crucial for theory development. Meta-analysis is a well-established tool to aggregate results from studies on a phenomenon, and they can provide novel insights that are not only theoretically but also practically of high relevance. Nonetheless, meta-analyses are not commonly used in language acquisition research.
In this talk I will demonstrate how to gain new insights by accumulating data that previously have been examined only in isolation. To this end, I will first outline general benefits of meta-analyses and then walk through an example, addressing a specific research question. Beyond their contribution to building more robust theories, meta-analyses also can be tremendously useful when planning experiments by guiding sample size decisions. In addition, they offer new interpretations for unexpected outcomes, especially null results.
A next step in theory building is a meta-meta-analytic view. Do developmental trajectories suggest a particular pattern of acquisition (such as the long-held belief that a native phonology emerges before substantial lexical acquisition takes place)? Is general cognitive maturation alone sufficient to explain observed patterns of child language development or do trajectories differ with the level of linguistic representation? Finally, and more practically, do the methods we use make a difference in outcomes? These questions can be answered with the help of a standardized collection of meta-analyses. MetaLab is exactly such a database in the form of a dynamic, ever growing collection of meta-analyses on early language acquisition. This database covers phenomena on various linguistic levels, ranging from the acquisition of phoneme categories to the role of pointing in vocabulary development. Results from studies testing over 10,000 participants are contained in the database, and various easy to use visualization tools allow for a quick exploration of the data. It is also possible to contribute new data, be it single studies (including null-results) or a whole collection on a phenomenon. This way, interested researchers can contribute to making language acquisition research more reliable and robust.
The MetaLab website: metalab.stanford.edu
Return to the list of speakers