9. Description of the EGG waveform.

This section contains a detailed description of the EGG signal especially with respect to the shape of the waveform and to the time domain characteristics of the physiological features.

As already mentioned, the EGG signal is regarded as a correlate of the glottal area or the glottal opening width. In an experiment by Gilbert, Potter and Hoodin (1984) an insulating strip was inserted between the vocal folds of an adult male during phonation to prevent electrical contact between them. There was no apparent effect on the production an acoustic wave, but after the removal of the insulator the amplitude of the EGG signal increased by ca. four times. Additionally, the results published by Scherer et al.(1988) enable the researcher to establish a linear relaltionship between the vocal fold contact area and the output of the electroglottograph. However, proper placement of the electrodes is of a special importance since a slight shift might cause spurious effects in the recorded signal.

The relationship between the gross features of the EGG waveform and the major phases of the glottal cycle are given by Baken (1992), Rothenberg and Mahshie (1988), Childers and Krishnamurthy (1985) and presented in Fig.16. Modal phonation with a full glottal closure has been assumed for this particular representation. There is no common agreement as to the direction in which the changes in vocal fold contact are represented. In this entire study however, the increased vocal fold contact is consistently plotted upwards on the y axis.

Figure 16. The model of the EGG waveform with annotated vocal folds movement phases: a) Representation of the vocal fold vibratory cycle (from: Childers & Krishnamurthy, 1985:137), b) Phases of the idealized EGG waveform related to the vibratory cycle of the folds. The six segments of the EGG waveform are denoted with the letters a,..,f, while instances of the fold movement are denoted with the numbers 1,..,9.

Phases of the vocal folds contact:

• When the vocal folds are open and it is ensured that there is no lateral contact between the vocal folds, the impedance is maximal and peak glottal flow occurs (segment (e) in Fig.16.). The waveform in this segment is flat, with small fluctuations.
• Then, the upper margins of the vocal folds make initial contact (segment (f) in Fig.16.).
• In the next phase of the movement (denoted as (a)) the lower margins come into contact and the vocal folds as a whole continue to close zipper-like. If the vocal folds close very rapidly and along their whole length, the phases (f) and (a) become indistinguishable and consequently the slope of the closure phase ((f)+(a)) becomes steep. The presence of this knee is typical for low to normal voice intensities and the slope of segment (f) is more gradual than the slope of (a).
• Glottal closure is achieved during phase (a). Childers and Krishnamurthy describe the vocal fold movement during this phase as follows: "Over large portion of the closing phase, the vocal folds adduct towards their medial position with little or no change in the length of contact along the midsagittal line. Just prior to closure, the vocal folds are almost parallel with a narrow opening along their entire length. Closure occurs almost simultaneously along the entire midsagittal line. Thus, while the glottal area does not reflect this fact, the glottal closure is an abrupt phenomenon. This type of closure is typically seen as the pitch is raised." (Childers & Krishnamurthy, 1985: 143).

• During the next phase (indicated as (b) in Fig.16) the vocal folds remain in contact and the airflow is blocked. Like in phase (e), limited fluctuations of the impedance are observed. However, the waveform is not flat, but rather forms a smooth hill (or hump). During this phase contact increases until the maximum is reached and then slowly decreases again. The maximum of the EGG amplitude usually occurs after the instant of glottal closure. Childers and Krishnamurthy (1985) explain this as the result of the elastic collision of tissue. This leads to mainly perpendicular vocal fold extension, which may cause the rounding of the Lx waveform, whose typical shape during the full contact phase is parabolic. If the contact area and its depth remain unchanged, the EGG is flat. For some voices, however, a small fluctuation in this phase can be observed (as in Fig.17). One can hypothesize that this is due to unstable contact between the folds.

Figure 17. EGG waveform with fluctuations in the contact phase

• The opening and the open phase are described analogously. In the process of vocal fold separation the contact between the folds starts to diminish and subsequently the lower margins of the vocal folds begin to separate, initialising the opening. Lower margin separation proceeds gradually during phase (c) (Fig. 16). Then the upper margins also begin to separate, resulting in an acceleration in the growth of impedance (phase (d)) until the full opening is reached. The glottis grows in size during this phase. As the contact between the vocal folds is not maintained anymore, the EGG waveform does not reflect the glottal width or the glottal area. It also does not contain any information about the glottal flow.

The time derivative of the Lx waveform is usually used in the determination of the periodicity of the signal. It can also be used to identify the distinguishable changes in the slopes during the phases of increasing and decreasing impedance which correspond to those of the simplified model of the EGG (see Fig.18).

Figure 18. Several periods of the EGG signal and their time derivative (modal /a/). The time markers are as follows: t0 - start of the EGG period, maximum of the signal derivative, t1 - maximum of the amplitude, t2 - opening instant, minimum of the signal derivative, t4 - minimum of the amplitude, t5 - start of the EGG period.

The positive peak of the derivative serves as an indication the instant of the glottal closure (CGI). Almost all researchers regard it as a robust and reliable marker of the pitch period for various voice qualities and intensities (Childers & Krishanurthy, 1985; Baken, 1992; Fourcin, 1993; Colton & Conture, 1990; Vieira et al., 1996). The procedure is both simpler than measuring F0 in the acoustic signal (Hess, 1991) and allows a more precise measurement of it. Thus, if a relatively strong EGG is registered, it is recommendable to use it as a reference for the precise measurement of speech fundamental frequency (Askenfeldt et al., 1980; Hess & Indefrey, 1987). Problems occur if the signal-to-noise ratio is poor or if F0 is measured during voice offset and onset, when several EGG pulses can exhibit a modified (or distorted) shape. It should be noted that the distribution of the EGG-measured pitch is even used to evaluate the degree of voice pathology (Fourcin,1993; documentation of the Laryngograph Processor; Fourcin & Abberton, 1977) when the acoustic methods of F0 determination fail.

The negative peak of the EGG derivative is regarded as an indicator of glottal opening. During the opening phase the glottal area increases monotonically until it reaches its maximum. Baer et al. (1983), Childers and Krishnamurthy (1985) state (based on high-speed films of the vocal folds vibration) that the movements of the vocal folds that are reflected in the EGG have two distinct phases. First, the EGG decreases monotonically, reflecting the decreasing in lateral contact between the vocal folds. During this interval the EGG waveform is convex. Then, as the upper margin separates the waveform of the EGG changes to concave. It is however questionable whether this should be used as a reliable marker of the instant of opening (because of the impact of additional mechanisms that can mask or delay it in the EGG signal). There is evidence for the so-called mucus bridging effect (Childers & Krishnamurthy, 1985; Titze & Talkin, 1979), in which a strand of mucus can bridge the glottis i as the vocal folds initially open. When the mucus bridge breaks, the EGG waveform records a sharp fall even though the glottis is already open.

It is usually supposed that the EGG reflects the contact area between the vocal folds. This area can be viewed laterally and transversely. Assuming that the depth of contact does not change during vocal fold vibration, the lateral contact area depends on the length of contact area along the upper margins of the vocal folds. Childers, Smith and Moore (1984) found a strong correlation between the EGG and the length of the vocal folds contact. Scherer et al. (1988) show a linear dependence of the EGG signal on the contact area between the folds.

The description of the EGG waveform fails for voices with a continuously open glottis. In that case, the variation of the larynx impedance does not correspond to the glottal area. Particularly the amplitude of the Lx waveform fluctuation may, but need not, change in accordance with the reduced contact between the folds.

In conclusion it can be assumed that the EGG waveform provides an adequate representation of the approximation and the closure of the vocal folds. The open phase of the folds movement is better represented by other means, especially by inverse filtering and invasive visual inspection (stroboscopy).