The Effects of Pitch, Gender, and Prosodic Context on the Identification of Creaky Voice

Abstract

Rates of identifying creak in male and female speakers are similar, suggesting that listeners have a comparable ability to hear creaky voice in all speakers.

The versatility of creaky phonation: Segmental, prosodic, and sociolinguistic uses in the world's languages

“…Much of this work is summarized in Kreiman, Gerratt, Garellek, Samlan, and Zhang (2014) and Garellek (2019), who lay out a psychoacoustic model of voice quality which contains four major components: harmonic source spectral slope, inharmonic source noise, time‐varying source characteristics, and the vocal tract transfer function. For spectral tilt in creaky phonation, low H1 − H2 (the amplitude of the first harmonic minus the amplitude of the second harmonic) relative to modal vowels has been shown to be a reliable acoustic correlate of the vocal fold constriction that characterizes most creaky phonation (Andruski & Ratliff, 2000; Blankenship, 2002; Brunelle & Finkeldey, 2011; Davidson, 2019a; DiCanio, 2012; Garellek, 2015; Garellek & Keating, 2011; Miller, 2007). In some languages, other spectral tilt measures may also distinguish creaky from modal phonation, such as H1 − A3 (the amplitude of the first harmonic minus the amplitude of the harmonic closest to F3) in Itunyoso Trique (DiCanio, 2012), H1 − A2 (the amplitude of the first harmonic minus the amplitude of the harmonic closest to F2) in Jalapa Mazatec (Garellek & Keating, 2011), or H2 − H4 (the amplitude of the first harmonic minus the amplitude of the fourth harmonic) in American English (Garellek & Seyfarth, 2016).…”

“…Some harmonics used in spectral tilt measurements are illustrated in Figure 1. As for inharmonic noise, measures such as cepstral peak prominence, harmonics‐to‐noise ratio (HNR), and jitter have shown that creaky voice is usually less periodic than modal voice (Blankenship, 2002; Davidson, 2019a; Garellek, 2012; Garellek & Keating, 2011; Kuang & Liberman, 2016). One consideration regarding measures relying on F0 is that if creaky phonation causes the fundamental frequency to be too aperiodic to be accurately tracked, then it is not possible to correctly estimate spectral tilt or noise measures that are based on harmonics.…”

“…That is, perhaps creaky phonation is more salient when compared with a speaker's modal voice that is higher in pitch, as it often is for female speakers, as opposed to speakers with lower F0. Davidson (2019a) presented listeners with fully modal, fully creaky, and partially creaky utterances (which started as modal and ended as creaky), and found that listeners were nearly equally likely to identify creak in the fully and partially creaky utterances regardless of whether the speakers were male or female, or what their average modal F0 was. This suggests that listeners can identify creaky phonation in both men and women, so stronger negative evaluations of women are not because creak is less discernible in men.…”

Framing ‘female’ vulgarity: an example of the use of linguistic markers in an episode of NBC’s Parks and Recreation

“…Similarly to HRT, this prosodic feature is often found at the end of utterances, or on pre-final syllables (Henton & Bladon: 1988). Creaky voice may be recognized aurally (Davidson: 2018) and observable on a spectrogram. This phonation type may also be identified by calculating jitter and shimmer values of a voice, which respectively refer to the variability of fundamental frequency, and variability of the amplitude of sound waves (Wertzner et al: 2005).…”

The cross‐linguistic patterns of phonation types

“…The goals of the current paper are (a) to provide an update on the state of research since Gordon and Ladefoged's (2001) overview, maintaining their original focus on the acoustics of phonation as it is used for contrastive and allophonic purposes, and (b) to also expand further into studies of the perception of phonation, reflecting the considerable recent growth in that field. In the interest of space and in keeping with Gordon and Ladefoged's (2001) approach, we do not attempt to provide an extensive overview of the articulatory mechanisms involved in phonation (see Garellek, 2019a, 2019b; Kreiman & Sidtis, 2011; Laver, 1981), nor do we cover the many ways in which phonation is used as a marker of prosodic structure (Epstein, 2002; Redi & Shattuck‐Hufnagel, 2001), as an index of social factors (see Kreiman & Sidtis, 2011) such as gender (Davidson, 2019a, 2019b; Zimman, 2018; Becker, Khan, & Zimman, 2017; Podesva, 2013; Yuasa, 2010), sexual orientation (Podesva, 2007), or ethnicity (Szakay, 2012), or as a result of voice pathology (see Kreiman & Gerratt, 1996, 2005; Sapienza, Hicks, & Ruddy, 2011).…”