Subglottal Resonances and Vowel Formant Variability: A Case Study of High German Monophthongs and Swabian Diphthongs
Andreas Madsack, Steven M. Lulich, Wolfgang Wokurek, Grzegorz Dogil · 2008
Recent studies have shown that subglottal resonances can cause discontinuities in formant trajectories (Chi & Sonderegger, 2007), are salient in speech perception (Lulich, Bachrach & Malyska, 2007), and useful in speaker normalization (Wang, Alwan & Lulich, 2008), suggesting that variability in the spectral characteristics of speech is constrained in ways not previously noticed. Specifically, it is argued that 1) for the same sound produced in different contexts or at different times, formants are free to vary, but only within frequency bands that are defined by the subglottal resonances; and 2) for sounds which differ by certain distinctive features, certain formants must be in different frequency bands. For instance, given several productions of the front vowel [ae], the second formant (F2) is free to vary only within the band between the second and third subglottal resonances (Sg2 and Sg3), but in the back vowel [a] F2 must lie between the first and second subglottal resonances (Sg1 and Sg2). The feature [+/-back] is therefore thought to be defined by whether F2 is below Sg2 ([+back]) or above it ([-back]; Stevens, 1998). This has been tested mostly in English, and only recently in Korean (Jung, 2008). We present new evidence that the same definition applies to both High German and Swabian German monophthongs, and suggest that it can also account for the contrast between the Swabian German diphthongs [aj] and [əj]. We made standard microphone recordings of twelve native speakers of German, including eight Swabians, and four fluent non-native speakers (whose native languages were Russian (1), Georgian (2), and Turkish (1)). The speakers read sentences which included both of the [aj] and [əj] diphthongs (non-Swabians produced them without a distinction), as well as sentences containing nonsense words designed to elicit monophthongs in a neutral phonetic environment. In the latter case, the carrier sentence was ‘Peter hat hVd gesagt’ (‘Peter said hVd’), where the vowel was any of the standard German monophthongs. We also made recordings from an accelerometer placed on the skin of the neck below the larynx, which measured subglottal acoustics. Front vowels, including the rounded front vowels, had F2>Sg2, and back vowels had F2 Sg1 whereas all other vowels had F1<Sg1. This is congruent with the hypothesis in Stevens (1998) that the feature [+/low] is defined by a relation between Sg1 and F1 (see Table 1).