Nasal-lateral assimilations: typology and structure

Abstract

In this work, I show that assimilatory nasalization and lateralization in sequences of nasals and laterals (NL and LN) are driven neither by sonority nor by feature sharing, but by a markedness constraint penalizing non-identical but phonologically similar adjacent segments – which I formalize here in an asymmetric implementation of Agreement by Correspondence (ABC; Walker 2000, 2001; Hansson 2001; Rose & Walker 2004). Typological evidence from a survey of 46 languages provides consistent implicational generalizations regarding the distribution of targets of assimilation; penalties on similarity correctly predect the relative lack of assimilation in heterorganic (ML, LM) sequences and in stop-lateral (TL) sequences. I further show that implementions of ABC in which correspondence is symmetric do not predict the observed preference for NL assimilation over LN assimilation; implementing the correspondence relation asymmetrically solves this problem. Analyses predicated on other considerations do not correctly predict the typology.