EGG – UNED Voice Lab

The electroglottograph or the electrolaryngograph is a device that generates a small high frequency electric current passing between two electrodes held externally around the thyroid notch by means of an adjustable band (see Figure 1). It measures, in a non-invasive way, the patterns of contacting of the vocal folds (Baken, 1992; Gilbert, Potter & Hoodin, 1984; Abberton, Howard & Fourcin, 1989). The vocal fold contacting is monitored through the measurement of differences in this current flow (Howard, 1995). This device was developed based on the fact that organic tissues tend to be good electric conductors (Baken, 1992).

**Figure 1.** Example of placement of electrodes that come with the Electrolaryngograph (Laryngograph^®, UK).

The interpretation of the electroglottographic waveforms gives some indirect insight into the complex three-dimensional vibrations of the vocal folds (Fourcin, 2000); it allows the understanding that the closing/opening motions of the inferior and superior segments of normal vocal folds are regular, correlated, quasi-periodic and independent from each other. Additionally, the vocal folds do not usually open or close at once in the horizontal plane, but show a zipper like motion (Baken & Orlikoff, 2000).

The electric current between the two electrodes as a function of time is converted and output as a voltage that varies in proportion to the vocal fold contact area (see Figure 2). It shows a positive change on the vertical axis when the electric current increases as a consequence of the touching of the vocal folds; thus, when the vocal folds are in contact, a larger signal is produced than when the vocal folds are apart (Abberton, Howard & Fourcin, 1989). The waveform is quasi periodic, since the vocal folds come together and move apart many times per second (Gilbert, Potter & Hoodin, 1984).

**Figure 2.** Schematic representation of two vocal fold vibratory cycles. The contact quotient (CQ) is intended to correspond to the fraction of the period time that the vocal folds are in contact; conversely, the open quotient (OQ) corresponds to the fraction of the period time that the vocal folds are not in contact. Many different definitions of CQ/OQ occur in the literature (adapted from Lã, 2012).

Electroglottographic waveforms can be characterized by several metrics, the one most known being the contact quotient. As illustrated in the Figure 2, the portion of one vibratory cycle that is spent in contact can be quantified as the contact quotient (CQ), and the portion without contact as the open quotient (OQ). Depending on the software, different contact and open quotient values may be obtained for the same pair of vocal folds vibrating in the same vibratory conditions. This is because the majority of software analysing electroglottographic waveforms apply different criteria to defined where contact starts (Herbst & Tersntröm, 2009; Lã & Sundberg, 2015; Herbst, 2020). The one presented in Figure 2 sets 3/7s from maximum peak as the onset of vocal fold contact (Lã, 2012).

More recently, both Wavegrams (by Christian T. Herbst) and FonaDyn (by Sten Ternström) compute the contact quotient without the need for setting a criteria for the onset of contact. In the first case (see Figure 3), after the normalization of consecutive electroglottographic waveforms, these are transformed into monochrome colour information and become represented as horizontal strips that then are rotated 90º anti-clock wise and graphically displayed to form the Wavegram. This type of display is particularly interesting when showing sudden changes in the vibratory patterns of the vocal folds, such as register breaks during registers transitions.

**Figure 3.** Example of a *Wavegram* representation showing a clear change in contact quotient (adapted from Herbst, Fitch, & Švec, 2010).

FonaDyn also does not apply a criterion to define the contact quotient, as the whole area beneath the waveform is considered (see Figure 4). FonaDyn allows the study of different electroglottographic waveshapes produced at different fundamental frequencies and sound levels, thus producing rich datasets of different waveforms plotted together over the entire vocal range (Ternström, Johansson & Selamtzis, 2018).

**Figure 4.** The quotient of contact by integration (Q_ci) is a definition of contact quotient that does not attempt to locate the instants of contact or separation of the vocal folds, but only the relative amount of contacting over the cycle (adapted from Ternström, JASA-EL July 2019).

It is a useful tool for research in voice, allowing the comparison within and between individual voices of several voice metrics, but also on the effects of a given approach to voice habilitation and rehabilitation (Lã & Ternström, 2020). You can learn more about FonaDyn and voice mapping for voice exploration and measurement under the tab “Voice maps”.

The real-time visual display of the electroglottographic waveform and different electroglottographic waveforms produced constitutes an excellent form of feedback to be used in voice lessons (Herbst, Howard & Schlömicher-Their, 2010; Lã, 2012).

Regardless of which tool is used to display electroglottographic waveforms, the benefit of electroglottography is that it is a non-invasive technique to assess indirectly the pattern of contacting of the vocal folds. Usually, direct methods require invasive techniques, such as the insertion of a rigid or flexible laryngoscope. Analysis of electroglottographic waveform allows the extraction of metrics related to vocal fold contact area, regularity of the pattern of vibration of the vocal folds and accurate measures of fundamental frequency (f_o) (Baken, 1992; Howard, 1995). The electroglottographic signal, which is an electromechanical signal rather than an acoustic one, thus provides indirect information on the different laryngeal excitations which determine the quality of the voice that is produced. When voice is “normal” or “modal” (Figure 5, lower panel), the waveform presents a rapid closing phase and a slower opening phase, usually associated with a good acoustic excitation of the vocal tract occurs for well-defined pitch. During “falsetto” voice (Figure 5, upper panel), the vocal folds are stretched and thin and they vibrate with higher fundamental frequencies. Only part of the upper edges of the vocal folds are contacting, so the amount of vibrating mass and the amplitude of vibration is smaller. The waveform shows that the open and closed phases are more similar, so that the waveform is simpler and sinusoidal (Fourcin, 2000).

**Figure 5.** Examples of electroglottographic waveforms displaying two types of vocal fold register: “normal” or “modal” (lower panel) and “falsetto” or “head voice” (upper panel).

During “vocal fry”, the vocal folds are thick and slack and there is a low air-flow rate. The waveform shows a rapid close and an extremely slow opening of the vocal folds. There are no regular cycles as there are alternations in the duration and amplitude of the cycles (Abberton & Fourcin, 1984; Howard, 1998). A “harsh” voice shows a smaller open phase and a less dynamic contact of the surfaces of the vocal folds (Howard, 1998). A “whispered” voice involves a narrowing of the glottis but there is no vocal vibration, so no waveform output will be observed. Only the restricted medial anterior vocal fold surface vibrates during “pressed” voice (Scherer, 1995). A “pressed” voice (Figure 6, left panel) presents a shorter waveform when compared with “flow” voice (Figure 6, middle panel). In the latter case, there is a maximum flow with complete glottal closure (Sundberg, 1987; Gauffin & Sundberg, 1989), indicating a higher dynamic contact of the surfaces of the vocal folds. A “breathy” voice (Figure 6, right panel) shows high peaks on the waveform that correspond to higher vocal fold contact area; however, since the glottis is open for a longer period when compared with “normal” voice, the baseline length of the Lx waveform is higher (Baken, 1987). Oedemas, nodules, polyps and other vocal aberrations are also distinguished through the analysis of the electroglottographic output (Altuzarra & Martin, 1996). For example, polyps or nodules will cause a longer flat segment of the rising edge, since these abnormal structures can increase or maintain the closed phase of the vibratory cycle for a longer period of time (Childers et al., 1986).

**Figure 6.** Electroglottographic waveforms for “pressed” (left), “flow” (middle” and “breathy” phonation types (adapted from Lã, 2012).

Despite the fact that electroglottography is useful for studying normal and pathological voice, it has some limitations. For example, the maximum contact area should not be considered as maximum glottal closure, since glottal closure might not be achieved when there is contact of the superior margins of the vocal folds (Baken, 1992). Thus, one should not consider vocal fold contact to be a direct complement of glottal area. The flow is measured by means of inverse filtering, and the waveform display refers to the amount of flow passing through the vocal folds when they are opened. When the glottis is completely closed, no air escapes from the glottis and the relative time in which this occurs per glottal cycle is referred to as closed quotient. Thus, contact quotient and closed quotient, although related, are different metrics. As illustrated in Figure 7, transglottal airflow starts before vocal fold contact ceases, as confirmed by the derivative of the electroglottographic waveform.

**Figure 7.** Comparison between vocal fold contact electroglottographic waveform (EGG) (middle panel) and its derivative (dEGG) (low panel), with the flow glottogram (higher panel) (adapted from, Lã & Sundberg, 2015).

Furthermore, the electroglottographic waveform amplitude cannot be directly correlated with vocal intensity, and there are some specific features of vocal fold vibrations about which firm conclusions cannot be made (Baken, 1992). Additionally, the waveform does not allow interpretations about the glottal width or motion of the vocal folds during the open phase of the glottal cycle. Rather, the EGG complements endoscopy, by giving information on what is going on while the glottis viewed from above appears to be closed. It should also be noted that a legible electroglottographic waveform is difficult to obtain for small larynges and for large fatty necks, since fatty tissues are poor electric conductors (Howard, 1988; Baken, 1987; Laver, Hiller & Beck, 1992). The electric current is spread not only between the electrodes; it also follows various paths around the vocal folds, so actually only a small part of the total current flow is affected by changes in the vocal fold contacting (Baken, 1987). The neck acts as a volume conductor, so the small high-frequency electric current does not flow straight from one electrode to the other, but spreads in several directions. Thus, the electrolaryngograph measures the impedance not only of the vocal folds but also of all neck structures close to the electrodes. A change in any of these structures will be also reflected in the EGG signal (Altuzarra & Martin, 1996; Baken, 1992). Furthermore, the position of the electrodes can change during phonation due to laryngeal movements. If the relative location of the vocal folds in the electric path is changed, variations in the electrolaryngograph output can occur, and these are not related to vocal fold motion (Altazurra & Martin, 1996; Baken, 1987; Baken, 1992). It is not uncommon to see the heartbeat in the EGG signal, since varying blood pressure can also affect the conductivity of the tissues. All in all, the variations in conductance that are caused by the changing contact between the vocal folds represent only 1% to 2% of the total variation observed in the EGG signal – yet they can be isolated, thanks to the frequency range of phonation, which is much higher than that of other physiological variations. Finally, laryngeal mucous can lead to inaccurate waveform interpretations. If a strand of mucous exists, the current can pass through that portion, and the electroglottographic waveform is calculated as being vocal fold contacting, hence causing a misreading of the truth vocal fold contacting (Childers et al., 1986; Baken, 1987).

Despite the above limitations, the electroglottograph/ electrolaryngograph: (i) allows the study of vocal fold contacting without obscuring supraglottal structures, thus without distorting the voice; (ii) it is non-invasive, so it is ideal for studying the voice during singing; (iii) it allows the study of large amounts of data, thus it is ideal to study phonation during the interpretation of repertoire; (iv) and it measures vocal fold contact area, so physiological changes in the vocal folds might be detected.