Acquiring a morphological system is a particularly difficult task because many morphemes never occur in isolation, especially grammatical morphemes. As such, identifying which sound sequences are morphemes presents a problem even greater than that faced in segmenting the sound stream into words. A stochastic approach to the process of discovering which sequences are morphemes, the initial step in morpheme acquisition, was implemented based on an analysis of frequently occurring sound sequences in adult speech directed towards children acquiring Spanish as a first language. Analyzing the child-directed speech with this method revealed that the frequent sound sequence approach was successful in discovering a number of morphemes with reasonable accuracy. When probabilistic syntactic information was added based on what is called the frames approach (Mintz 2003), the token accuracy, as well as the number of unique morphemes uncovered, increased dramatically. This was found to be specific to the frames approach and not simply a result of the added syntactic knowledge. These results suggest that children can effectively use frequent sound patterns as part of a larger probabilistic learning approach to begin to decode the morphological system.
Selected Proceedings of the 7th Conference on the Acquisition of Spanish and Portuguese as First and Second Languages
edited by Carol A. Klee and Timothy L. Face Table of contents
ISBN 978-1-57473-409-6 library binding
vi + 194 pages
publication date: 2006
published by Cascadilla Proceedings Project, Somerville, MA, USA