Sound level protrusions as physical correlates of sonority

Usage of acoustic cues in spoken term detection keyword spotting for zero low resource languages

“…Sonority has a long history in literature, it is dated as back as 1876 when Sievers first attributed it as sound fullness. He explained it as relative loudness of speech sounds [5]. Ladefoged defined it as loudness relative to other sounds having same length, stress and pitch [6].…”

Compiling sonority scales with obstruent vowels

“…Zec (1995) makes use of a more granular scale in which vowels are more sonorous than non-obstruent consonants, which are in turn more sonorous than obstruent consonants. Phonetic investigations of a combined seven languages 3 also show that obstruents are consistently less phonetically intense than non-obstruents (Parker 2008, Jany et al 2007). These two principles, if taken together, stipulate that if a language licenses syllabicity to obstruents (as do Liangshan Yi and Tashlhiyt), which must be lower-ranked on the scale than sonorants, then all sonorants will be licensed syllabicity.…”

“…1, exemplify this arrangement, which results in a contiguous zone of segment classes that can be licensed syllabicity (the permitted nuclei). Since sonority scales have been argued to be ranked with input from acoustic parameters such as intensity (Parker 2008), this could be taken to mean that all languages permit syllabicity in segments falling within a given range for an acoustic parameter. In this paper, I introduce obstruent vowels, phonetic obstruents that behave phonologically as permitted nuclei, and argue that they form a major exception to these generalizations, given their frequent separation from the contiguous range of other permitted nuclei in a given language.…”

“…rhotics vs. laterals or fricatives vs. affricates (Jany et al 2007). Intensity has recently been acknowledged as a good acoustic correlate for higher rank on the sonority scale (Parker 2008, Jany et al 2007. Figure 2 is a typical representation of this scale, after Blevins (1995), where a capitalized A (for vowels) or a coronal segment in each segment class stands in for the segment class in general, a convention I will use in the remainder of this paper: R for rhotics, L for liquids, N for nasals, S for fricatives, and T for stops.…”

“…Put another way, it appears that languages must license syllabicity to all segments ranked higher on the sonority scale than the lowest-ranked segment ( Figure 3), dividing the sonority scale into two zones: sufficient sonority and insufficient sonority. Since segments' intensity decreases with lower rank on the sonority scale, this can be taken to represent a cutoff in a parameter or set of parameters for a segment, dubbed "overall sound level" by Parker (2008).…”