Bilingualism and Developmental Linguistics research group

Speaker: Mits Ota (University of Edinburgh)

Title: Child consonant harmony – not!

Abstract: Developmental linguists and theoretical phonologists have long been vexed by young children's production of wordforms that appears to exhibit consonant place harmony (e.g., doggy > [gɔgi]). Most other well-known features of early word production are consistent with what is typically avoided in mature phonological systems, suggesting a shared underlying principle (e.g., consonant clusters are typologically marked AND children tend to reduce them to singletons). But long-distance place agreement of consonants is not a common phenomenon in adult phonology; if anything, most languages show the opposite tendency to avoid place-sharing consonants separated by a vowel. In this talk, I will argue that this type of parallelism does not apply to consonant harmony because child 'consonant harmony' is less to do with a phonological process than copying of segments to cope with fragile lexical memory of dissimilar consonants. I support this interpretation by reviewing evidence that 1) child 'consonant harmony' usually results in identical, not just place-sharing, consonants, 2) when learning wordforms, infants are challenged by dissimilar segments within words, and 3) older children and adults also resort to consonant copying during initial stages of wordform learning. Furthermore, I will discuss why this type of learning bias does not appear to dictate the shape of words in the adult lexicon.