Voicing of glottal consonants and non-modal vowels

Abstract

Variation in voicing is common among sounds of the world’s languages: sounds that are analyzed as voiceless can undergo voicing, and those analyzed as voiced can devoice. Among voiceless glottal sounds in particular, voicing is widespread: linguists often expect the voiceless glottal stop [ʔ] and fricative [h] to be fully voiced, especially between vowels. In this study, we use audio recordings from Illustrations of the International Phonetic Alphabet published in the Journal of the International Phonetic Association to explore the extent to which glottal consonants and non-modal (breathy and creaky) vowels differ in terms of percentage voicing and voicing intensity in three phrasal positions. We find that voiceless [h] is only slightly less voiced than voiced [ɦ] in initial position. Between two vowels, both [h] and [ɦ] are as voiced as breathy vowels. Glottal stops and creaky vowels are both characterized by high percentages of voicing, but they differ in voicing intensity: in all phrasal positions, glottal stops generally have periods of strong and weak voicing, whereas creaky vowels are strongly voiced. In contrast, vowels described as ‘rearticulated’, ‘checked’, or ‘glottalized’ show similar drops in voicing intensity to glottal stops. We interpret these results through an articulatory lens: glottal consonants and non-modal vowels are both modulations in phonation resulting from laryngeal constriction and vocal fold spreading. We argue further that, because voicing during [ʔ] and [h] is largely predictable from respiratory and prosodic constraints, many cases of [ʔ] and [h] can be considered phonetically underspecified for voicing.

Glottal stops do not constrain lexical access as do oral stops

“…In this view, laryngeal features are licensed by prosodic structure, and never analysed as segments, largely in line with the proposal made by [64]. In a similar vein, Garellek et al [36] suggested that while treating a glottal stop as a segment may be useful for phonological analysis, the actual phonetic realization of a glottal stop (or glottalization) can be better characterized as modulation of phonation gesture (with a range of phonetic implementation that may vary with prosodic structural factors), which poses challenges for treating it as segmental.…”

“…However, these considerations are contingent on whether a glottal stop with a full closure is indeed a hyperarticulated phonetic form that enhances its underlying representation as just discussed above [24][25][26][27]. In fact, as was noted by Davidson [35], it is not often the case that a glottal stop is realized with a full closure across languages [see also 36,37]. For example [35], showed that a phonemic glottal stop in Hawaiian is most often realized as creaky voice especially in a word-medial intervocalic context with flanking vowels being 'glottalized.'…”

“…[35] therefore leaves a question open as to "what the gestural target for this phoneme is"-i.e., a glottal closure versus a period of creaky voice. Garellek et al [36] suggested yet another option that the target is simply specified with [glottal constriction] which may or may not lead to full closure depending on the prosodic strength of the gesture. This debate, however, does not entirely rule out a possibility that the non-modal creaky voice in an intervocalic context can reflect a form of lenition of a full glottal gesture while a full glottal stop reflects an underlying gestural target.…”

“…To prevent clicks, there was a 5 ms ramp down into the silence and 5 ms ramp up to the original signal after the silence. In doing so, care was given not to cut off the portion of the formant transitions, so that the full silence was still consistent with realization of a full glottal stop which is often accompanied by some degree of glottalization during the vowel [36]. The same 120 filler trials were used as in [15], with the target words of those trials using neither a vowel-nor glottal stop-initial.…”

“…Given the marginally significant effect of target in both the reaction time and the fixation analyses, we performed an additional exploratory analysis using a general additive model (GAM) to investigate where in the time series the two conditions differed significantly (along the lines of [35]). Since a GAM analysis requires a continuous variable, we calculated the target preference with the Euclidean distance (cf., [36]) from the fixation position to the competitor and to the target (preference = distance(fixation,competitor)-distance(fixation,target)). We capped these values at 600 pixels because differences larger than that were due to the layout of the display on any given trial (i.e., the distance is larger if the target and competitor are at opposite sides of the screen but smaller if one is above the other).…”