Transparency and cumulative thinking are key ingredients for a more robust foundation for experimental studies and theorizing. Empirical sciences have long faced criticism for some of the statistical tools they use and the overall approach to experimentation – a debate that has in the last decade gained momentum in the context of the “replicability crisis.” Many solutions were proposed, from open data, code, and materials – rewarded with badges – over preregistration to a shift away from focusing on p values. There are a host of options to choose from; but how can we pick the right existing and emerging tools and techniques to improve transparency, aggregate evidence, and work together? I will discuss answers fitting my own work on language acquisition spanning empirical (including large-scale), computational, and meta-scientific studies, with a focus on strategies to see each study for what it is: A single brushstroke of a larger picture. My goal is, in all these efforts, to better understand how the lexicon develops across the life span – with an emphasis on early development.
Bio
How babies begin learning words is the core question that my research program tries to answer. It might seem like child’s play, but as soon as we are confronted with a language we do not understand, we can get an idea of the problems children face. It’s difficult to even find the boundaries between words, let alone decipher their meaning, just from observing native speakers. In my experimental work, I assess how different day-to-day experiences might influence infants’ learning trajectory, because this can shed light on what children pay attention to in their input and how they learn. I complement this by computational modelling, which allows me to specify the memory and learning mechanisms when simulating a little language learner. I also use modelling to characterize the input children receive and to check which information they can reasonably access. A third strand of research relies on meta-scientific tools to aggregate all existing evidence, as demonstrated on the online platform metalab.stanford.edu. My goal is to use this wealth of data to generate more robust theories. For the same reason I am a member of the governing board of ManyBabies (manybabies.stanford.edu), an international network of child laboratories that aims to assess influential experiments in developmental psychology. Before coming to the MPI, I worked at Ecole Normale Supérieure in Paris, France, with Alejandrina Cristia. My position was funded among others by a Marie Sklodowska-Curie individual fellowship to assess the impact of speaker variability in daily life on early language development. This research question was motivated by my PhD thesis, supervised by Paula Fikkert, Lou Boves, and Louis ten Bosch at Radboud University Nijmegen. In my dissertation I used computational modelling to assess the theoretical impact of learning from one versus multiple people.
