Guest speaker: Paula Fikkert (Radboud University Nijmegen)

We are very pleased to welcome Paula Fikkert, a phonologist and acquisitionist from Radboud University Nijmegen. Her research is about phonological representations and the role they play in both perception and production throughout the process of first language acquisition.

She will be giving a talk at 2:30 PM on Tuesday, June 2 (the location is to be announced: watch this space for an update): "Is the devil in the detail? Abstract and detailed representations in perception and production."

Children learn to recognize words fast and reliably despite noise and variation in the input. They do this by extracting relevant phonetic features from the input and matching these onto phonological representations of words stored in the mind. How they learn to do this? Over the last four decades we have learned that infants are amazingly good at phonetic learning. However, our understanding of what happens when children construct their mental lexicon, which requires phonological learning, is as yet poor.

Phonological learning involves the construction of invariant phonological representations of words that are both abstract enough to allow fast recognition and handle phonetic, phonological and morphological variation automatically, and detailed enough to keep lexical items distinct. Moreover, these same phonological representations are used to initiate articulation for production. 
In this talk I will argue that a comprehensive theory of phonological acquisition should take both perception and production into account, as well as learning and development. I will present a large set of production and perception data addressing the nature of place and manner of articulation as well as laryngeal features. For each set of features asymmetries in children’s perception and production are attested. However, the asymmetries do not allow for one straightforward explanation, and are motivated differently for each set. I will discuss the consequences for a model of phonological acquisition. Most data will come from Dutch, but data from other languages, including German, English and Japanese, will be presented as well.