Umeå University's logo

Formant centre frequencies are regularly used as acoustic measure of vowel quality because of the well established correlation between them and properties of vowel production(Fant, 1960). While it is recognised that it is important transform formant frequencies, as measured from the spectrogram, into relevant psychoacoustic scales when addressing issues of vowel perception, production studies are still largely based on the the raw, Hz-scaled, formant measurements. As a consequence, systematic studies of vowel articulation proficiency due to clinical conditions or effects of treatment are not easily afforded. 

A number of derived measures have been proposed to capture whole-system effect of vowel production. Previous research has primarily used the calculated total area of formants measured from either [a], [u], [i] and [æ] or only [a], [u] and [i] productions, joined together to form a quadrilateral or a triangle (Vowel space area, VSA)(Kent & Kim, 2003). In the reduced articulatory range of hypokinetic dysarthria, VSA is expected to be reduced by the dysarthria, and increased due to successful treatment the patient’s articulatory range. 

However, VSA has been proposed to be not powerful enough to capture effects on vowel articulation found specifically in patients with Parkinson’s disease (PD), and therefore not able to establish treatment effects.

Two alternative measures have been proposed previously that are more focused towards specifically towards finding expansion and reductions in vowel spaces, the Vowel articulation index (VAI)(Roy, Nissen, Dromey, & Sapir, 2009) or Formant Centralization Ratio (FCR)(Sapir, Ramig, Spielman, & Fox, 2010). Both VAI and FCR has been shown to be able to establish significant reductions vowel articulation due to PD. Similar to VSA, however, the VAI and FCR measures do not given any detailed insight into the nature of the change in articulation. Further, all three measures reduces all obtained vowel formant measurements into a single metric, resulting in a substantial loss of statistical power and a considerable reduction in quality of intra-speaker models of articulatory proficiency. 

In response to the problematic properties of previous metrics, a revised view of formant frequencies more suitable for studies of changes in vowel articulation has recently been outlined, along with a new metric (Vowel Formant Dispersion, VFD). The VFD view of vowel formants are more closely connected to properties of vowel articulation and therefore affords more more detailed interpretations to be drawn concerning speakers’ production proficiency and offers increased reliability of within speaker estimates of this proficiency. This paper presents a detailed view of how VFD may applied to clinical populations where vowel production is affected by the condition (e.g. hypokinetic dysarthria) or where production should be affected (e.g. transexual patients or in successful treatment of dysarthric patients). It will be shown that most aspects of the study of vowel articulation proficiency will be significantly enhanced by the VFD method.