Examples of inversely filtered9 (estimated derivative of the glottal volume velocity waveform) glottal flows are given in Fig.12 together with a corresponding spectra. A sustained vowel /a/ was produced in five different qualities by the same speaker with approximately the same vocal effort. The characteistics of the respective voice qualities in the spectra of the glottal flows are very distinct. The results are similar to those found by Ní Chasaide and Gobl (in: Hardcastle & Laver, 1996:449). Considering Fant's speech production model, it is obvious why the same phoneme with unmodified articulation is perceived with different voice coloring .
Physiologically, the frequency of vocal fold vibration in modal voice is controlled by adjusting the effective mass and the stiffness of the vocal folds. Contraction of the CT (the cricothyroid muscle) elongates the vocal folds, resulting in a decrease in vibrating mass and an increase in stiffness of the vocal folds. Contrary to this, a contraction of the TA (thyroarytenoid) muscle leads to the thickening of the vocal folds, which in turn causes an increase in vibrating mass. The stiffness of the vocal fold body increases while that of the cover decreases (Hirose, 1996:134). In non-modal phonation types pitch change is also achieved by the actions of these two muscles, although the degree of activity of TA and CT depends on the phonation mode (Titze, 1994:269-271). Despite the fact that the pitch raising mechanism has already been described (ibid.), it is still under discussion how the lowering of pitch is achieved. It is hypothesized however, that the external laryngeal muscles may be making some contribution by lowering the larynx (Hirose, 1996).
Changes in fundamental frequency are usually related to a change in voice quality, albeit some authors of earlier works suggested that pitch and voice quality are controllable separately (Ní Chasaide & Gobl, 1996:455). In tonal languages a particular voice quality is quite often associated with a specific tone (Maddieson & Hess, 1987; Laver, 1994:477). One of the hypothetical tonogeneses is the evolution of tones from earlier voice quality contrasts (Fromkin,1978; Traill & Jackson, 1988; Hombert, 1978).
Different degrees in voice effort are responsible for changes in loudness. The main cause for this is a stronger airstream resulting in an increase of subglottal pressure, although the strength of vocal tract excitation depends mostly on the first derivative of the glottal flow .
Intonation also causes changes in voice quality . Pierrehumbert (1989) relates some factors of voice to the tones typical of American English intonation. Gobl (1988) described a stronger excitation of a word in focal position than in non-focal position and found the excitation to be stronger for vowels than for the surrounding consonants. Stevens and Hanson (1995) for American English, Sluijter et al. (1995) for American English and Dutch, and Jessen et al. (1995) as well as Claßen et al. (1997) for German show the dependence of voice quality on stress (details are given in section 17).
Voice quality is also used in phrase boundary marking (Fant & Kruckenberg, 1989 for Swedish).