Phonological and morphological roles modulate the perception of consonant variants

Abstract

Abstract This study investigated how the perception of a sound is affected by its phonological and morphological roles within a word. We asked American English listeners (n = 24) to judge differences among phonetic variants of sounds [l], [n], [ɹ] in three word conditions: 1) at morpheme boundaries with a phonological process, such as [n] in down-ed, which triggers voicing agreement on the suffix, 2) internally without a process, such as [n] in mound, and 3) at morpheme boundaries alone, such as [n] in town-ship. We used Praat synthesis with different acoustic settings to create variants, e.g., [n]a, [n]b, [n]c, which were spliced into a base to produce three tokens, dow[n]a ed, dow[n]b ed, dow[n]c ed. Identical variants were used across conditions (e.g., in condition 2: mou[n]a d, mou[n]b d , mou[n]c d). On each trial, participants heard two tokens of the same word (e.g., dow[n]a ed – dow[n]b ed) and rated the difference between the target sound using a sliding scale with endpoints “0% (totally identical)” and “99% (totally different)”. Analysis with linear mixed-effects model revealed significant differences between ratings among all conditions, with the pattern township < downed < mound. These results suggest that a sound’s phonological and morphological roles within a word affect how people perceive it. We evaluate this finding in light of the differing predictions made by phoneme-based theories, which incorporate phonemes as a fundamental unit, versus exemplar theories, which argue that phonological units are emergent.