Frequency maDDers, but phonology and formality maTTer, too.

Abstract

In cases of allophonic variation, the frequency of a variant has been claimed to influence phonological representations, affecting the perception and recognition of words. For example, Connine (2004) found stronger Ganong effects in words with frequent variants (e.g., intervocalic tap in the word pretty) than in words with infrequent variants (e.g., [t] in the word pretty). Specifically, in a phoneme‐monitoring task, listeners reported more PRETTY responses when listening to a pre[D]y‐bre[D]y continuum than a pre[t]y‐bre[t]y continuum. If this bias is due to variant frequency, we might expect this effect to generalize across variants, independent of other factors (context, within‐paradigm alternations). However, a diverse set of variants and alternate explanatory factors contributing to the effect have not yet been investigated. For example, the roles of paradigm variants (e.g., baiting‐bait versus pretty‐*pret) and filler type (e.g., testing for intervocalic [t] in the context of carefully versus casually articulated fillers) might also explain these effects. This study builds on Connine (2004) and examines both paradigmatic relations and filler type across two allophonic alternations (tapping, final [t] glottalization). Ultimately, the results show that variant frequency alone is insufficient to account for this effect. An integrated theory of speech perception is proposed.