Acoustic characteristics of Swedish dorsal fricatives

The precise articulatory description of the "voiceless dorsopalatal/velar fricative" or "simultaneous" {∫x}, usually transcribed as {h} in IPA notation, is somewhat controversial [5]. The controversy arises from highly variable productions among different speakers and the absence of a phonemic contrast with any other dorsovelar fricative in the language. Moreover, the sound is attested only in some dialects of Swedish [8]. The present study attempts to acoustically differentiate anterior and posterior variants of the fricative {S h x} along with the posterior phoneme {h} and the anterior phoneme {ç}, for comparison. A single speaker who controls various regional dialects of Swedish participated in the study. In Experiment 1, the speaker produced VCV nonsense words with balanced V and alternating C. Multitaper spectral estimates were calculated and measured [1, 2, 11]. The results indicate that {∫} can be reliably differentiated from {x} and {h} based on center of gravity measures. However, center of gravity differences between {x} and {h} are not revealing, confirming Lindblad's x-ray tracings [8]. The most robust acoustic difference between the two dorsal fricatives is the presence of erratic, high-amplitude, low-frequency transients during {x}. These may be attributed to relatively unpredictable explosions of saliva and ephemeral lingual contact with the uvula or soft palate, more likely for slightly-retracted {x} than for {h}. Based on the current acoustic model of fricative production, it seems unlikely that {h} is produced with a second simultaneous constriction that is of any acoustic relevance. Further experimentation attempts to discover the acoustic correlates that may be exaggerrated when speakers actively distinguish two anterior and two posterior fricatives. The prospectus for a future sociophonetic study is contemplated.