Cue-based Features: Modeling Change and Variation in the Voicing Contrasts of Minnesotan English, Afrikaans, and Dutch

Abstract

Phonological voicing in obstruents is signaled by numerous acoustic cues, both spectral and temporal. Voicing contrasts have been featurally described as [±voice], [±spread glottis], fortis versus lenis, or a combination of features such as [±spread] and [±slack] vocal folds, depending on the cues utilized in a particular language. Describing obstruent voicing contrasts with only cues or features, to the exclusion of the other, misses larger cross-linguistic patterns. This dissertation proposes cue-based features to capture phonological voicing contrasts.

Cue-based features are developed from the results of three experiments examining voicing contrasts in Minnesotan English, Afrikaans, and Dutch. In Minnesotan English, two production tasks show that fully devoiced obstruents (underlyingly voiced but produced with no vocal fold vibration) are significantly different along other phonetic dimensions from both voiced obstruents and underlyingly voiceless obstruents, creating a class that is intermediate between [+voice] and [-voice]. Production and perception tasks in Afrikaans and Dutch examine two plosive voicing phenomena in different demographic groups. In word-final obstruents, both languages neutralize glottal pulsing, but other cues weakly maintain the contrast. In word-initial position, underlyingly voiced plosives are variably devoiced, but differences in the following vowel’s f0 maintain the contrast. The results of these experiments confirm that acoustic cues that are not necessarily linked to the definition of a distinctive feature have a significant role in the realization of phonological contrast.

The cue-based feature approach is a two-part model that represents the phonetics/phonology interface. This work builds on the ideas of Kingston and Diehl (1994), Reetz (2000), Purnell et al. (2005b), Boersma (2007, 2008), Kingston et al. (2008), and Lahiri and Reetz (2010), but differs in assumptions about the interface and what contributes to phonological contrast. Cue-based features map individual cues to single features. Each cue has its own weighting, and all cue weights are cumulative within a single feature. The total of all cue weights for a feature must reach a threshold to signal [±feature]. This model recognizes the contribution of acoustic cues while still maintaining one phonological feature label, [voice], and can model phonological change and variation.