Markedness and the Learning of Lexical Stress Patterns


The Unniversity of Texas at Austin
Megan Crowhurst (Lead)

University of Agder
Allison Wetterlin and Linda Wheeldon


Research in metrical phonology has produced a well-articulated typology of stress patterns and theories of metrical markedness (e.g. Prince 1983; Hayes 1995; Halle & Vergnaud 1987; Kager 1992). Despite this solid base, however, little experimental research with humans has explored the relationship between markedness and the formation of inferences about stress. In particular, little or no research has studied how adults learn contrasting attested patterns whose markedness we cannot fully evaluate. The present study examines the relationship between competing stress-attracting word edges in contingent e/e (edge/same-edge) and e/-e (edge/opposite-edge) patterns. We do not yet understand the relative markedness of the patterns in the {e/e, e/-e} typology. Are e/e patterns unmarked relative to e/-e patterns? Or, are left-edge patterns unmarked? And can ease-of-learning evidence bear on these issues? We are conducting an artificial language learning study in which participants hear and repeat nonce trisyllables with varied stress patterns (training) and then are tested on novel stimuli designed to investiagate any generalisations they have made about the learned stress pattern. The results of this study promise to be interesting for both theoretical and methodological reasons, as they can bear not only on metrical markedness, but are also expected to encourage us to learn about markedness by studying covert phonological generalisations.