Velar Softening in English: A Constraint-Based Approach

Abstract

This study examines velar softening, a morphophonological rule occurring in English words of Romance origin where velar stops (/k/, /g/) become sibilants ([s], [dʒ]) typically before a non-low front vowel. Traditional rule-based theories and previous constraint-based theories fail to account for the seemingly unpredictable nature of this phenomenon. This paper presents an alternative analysis using harmonic parallelism (classical Optimality Theory), arguing that velar softening is a direct input-output mapping rather than a step-by-step derivation. Following Lee (2004), I posit multiple input forms for alternating bases and show that velar softening can be explained through a set of constraints, including one that resolves diachronic opacity. To address a key counterexample, Isaa[k]ism, I further propose that lexical entries encode etymological origin (Romance vs. non-Romance), which can be represented in markedness constraints. Based on this analysis, the paper employs faithfulness and markedness constraints like Ident-IO[F], *SDOR+Suf1(R), *Sib+Suf2(R), SDOR-ɛ]w(R), and Max(SDOR). This OT analysis provides a more comprehensive and robust account of velar softening, capturing its patterns and exceptions without relying on a derivational approach.