Weighted scalar constraints capture the typology of loanword adaptation

Abstract

This paper discusses three basic ways in which loanwords pattern differently than native vocabulary, with a particular focus on the implicational relationships that hold among generalizations that apply at different degrees of nativization. We argue that the overall typology and the effects of the core-periphery structure are best modeled if constraints are weighted as in Harmonic Grammar (Legendre, Miyata & Smolensky 1990), and violation scores are scaled according to degree of nativization. The implicational patterns of repair versus non-repair are predicted from basic patterns of interaction among scalar constraints, obviating the need for the kinds of ranking metaconditions required in ranked-constraint alternatives.