ASA 129th Meeting - Washington, DC - 1995 May 30 .. Jun 06
5pSC23. About the fourth angle of the ``vowel triangle.''
ENST., 46 rue Barrault, Paris, France
The vowel triangle can be derived from the acoustic theory [R. Carre et
al., 2924 (A) (1994)]. The area function was deformed in order to
increase or decrease F1 or F2. In this new study, the acoustic criterion is to
increase (or decrease) both F1 and F2 or to increase F1 (or F2) and decrease F2
(or F1). At the end of the deformation process, the area functions produce the
most ``compact'' (F1 and F2 close as in /a/) or ``diffuse'' (F1 far from F2 as
in /i/) or ``grave'' (F1 and F2 both low as in /u/) sounds. But what about an
``acute'' vowel (F1 and F2 both high) which could correspond to the fourth
angle? It can be automatically converged to a pattern where F1 is around at
1000 Hz and F2 at 2000 Hz. But in this case the ratio between the maximum value
of the cross-sectional area divided by the minimum value becomes around 80.
This ratio is only 20 (between 0.5 and 10 cm[sup 2]) for the ``compact'' versus
``diffuse'' axis (from /a/ to /i/). Thus reaching this fourth angle is too much
costly in deformation amplitude and thus poorly efficient.