Abstract:
This study examines the phonetic variation of the Korean liquid, which can surface as [ɾ],
[n], [l], [ll], or ∅ in overlapping environments. Previous studies on this topic have not
analyzed the full range of the alternations and/or have not explicitly acknowledged that
the Korean lexicon may be divided into distinct lexical strata. Since environment alone
cannot predict how the liquid may appear on the surface, I propose that there are two
sublexical systems (“Native/Sino” and “Loanword”) within the Korean language based
on the etymology of a given word that affect a speaker’s production of the liquid. I
develop an optimality-theoretic (OT) analysis in which markedness constraints are more
highly ranked in words belonging to the “Native/Sino” sublexicon and faithfulness
constraints are more highly ranked in words belonging to the “Loanword” sublexicon. In
analyzing the Korean lexicon as being split into two distinct lexical strata, this study finds
that in loanwords with two surface forms in free variation, one has become more
nativized and is less marked while the other is more faithful to the input pronunciation.
