Why a phenomenology of vowel sounds is needed

In literature, there is an extensive and often controversial debate on the primary acoustic and perceptual cues of vowel quality, resulting in two main viewpoints that these cues are contained in either the formants, or, alternatively, in the spectral shape. However, in our understanding, one aspect is highly underestimated: the fact that any spectral representation of vowel-quality is directly or indirectly related to pitch and to the fine structure of the spectrum. Hence, a given formant pattern as well as a given spectral envelope is in many – if not all – cases ambiguous in terms of representing sounds of different vowel qualities, if speakers equal in size and gender or even single speakers produce the sounds on very different pitches. If they in addition vary vocal effort and phonation type, this ambiguity will be further enhanced. Neither of the two above-mentioned viewpoints can account for this issue, however. – The present paper (i) summarises the ongoing debate, (ii) describes the empirical evidence for spectral representation of vowel quality being related to pitch, spectral shape and spectral fine structure of a sound, (iii) concludes that existing approaches to determine the acoustic cues for vowel quality do not account for all recognisable vowel sounds, both conceptually and methodologically, and (iv) argues for the need of a phenomenology of the acoustics of vowels in terms of building up large-scale, language-specific sound descriptions, addressing all variations of production parameters and their possible extension relevant for perceived vowel quality.