Consonant and vowel rounding: Same acoustics, different visuals

Abstract

The phonological feature [±round] is normally associated with vowels, but this feature is also relevant for some consonant contrasts, such as the retroflex/non-retroflex sibilant contrast in Mandarin. One might assume that contrastive lip rounding would be the same gesture for consonants and vowels within language. However, because similar acoustic results can be achieved through different articulations, we hypothesize that the exact parameters of rounding might be variable for different segment types. An audiovisual production experiment was conducted with 30 Mandarin native speakers. Subjects produced 35 words with sibilants onsets (retroflexes /ʂ, ʈʂ, ʈʰ/, alveolars /s, ts, tsʰ/, palatoalveolars /ɕ, tɕ, tɕʰ/), and 40 words with high-front rounded/unrounded nuclei (/i/ and /y/). We found robust individual variation and that consonants and vowels were characterized with different visual properties: /y/ had small aperture in both vertical and horizontal directions and therefore smaller lip opening area than in /i/, while retroflexes had smaller horizontal but larger vertical apertures than alveolars or palatals. Acoustical consequences of rounding were similar: both spectral Center of Gravity and F1–F4 were lowered compared to unrounded counterparts, indicating that the same phonological feature may have multiple visual/articulatory correlates.