Selective signal strengthening: speakers only enhance cues most likely to improve listener perception

Abstract

A critical issue in models of speech production concerns the modularity of processing: for example, can information from morphology influence the articulation of phonologically identical segments? A phenomenon that has the potential to help settle this question is phonetic enhancement of affixes; that is, longer or clearer articulation in affixes, compared to their non-affix counterparts. The occurrence of phonetic enhancement suggests that morphological information can indeed influence segment articulation. However, to date, the literature on phonetic enhancement of affixes has been mixed: while numerous studies have found phonetic enhancement, others have failed to find the effect. If it is the case that morphology interacts with phonetics, these divergent findings require some explanation. In the present study, we first show that this discrepancy in the literature can be accounted for with a division between consonants and vowels: phonetic enhancement occurs only in vowels. Next, we present two analyses of spoken German. We compared 1) duration and 2) articulatory trajectories between segments in German consonantal vs. vowel suffixes (-t and -er). For vowels, we found 1) significantly longer durations and 2) significantly larger articulatory trajectories in suffixes compared to segmentally identical non-suffixes. Importantly, however, no effect was found for consonants for either duration or articulatory trajectory. We interpret these results in terms of an inherent difference in audibility of consonants versus vowels. We suggest that while morphology can affect segment articulation, speakers enhance only those segments where the enhancement will help reduce listeners’ uncertainty about the message.

Supplementary materials

The supplementary materials of the study is available here.