Measurement of Pharyngeal Air Pressure During Phonation Using High-Resolution Manometry

Abstract

Purpose

The study of air pressure in the vocal tract is essential to understanding vocal function. Changes in vocal tract shape during different phonatory gestures are hypothesized to produce nonuniform air pressure across lower vocal tract locations. Current methods of air pressure measurement, however, are limited to a single location in the anterior oral cavity. The purposes of this study were (a) to assess the feasibility of a novel method of simultaneously measuring phonatory air pressure at multiple locations across the lower vocal tract using high-resolution pharyngeal manometry (HRM) and (b) to compare pressure across locations and among phonatory tasks.

Method

Two subjects underwent HRM while performing phonatory tasks. A catheter was passed transnasally and air pressure was measured simultaneously at five locations between the velopharyngeal port and the upper esophageal sphincter. Descriptive statistics were calculated for each location by task, and for each task averaged across locations.

Results

HRM was well tolerated, and air pressures from multiple locations in the lower vocal tract were able to be obtained simultaneously. During vocal tract semi-occlusion tasks, air pressures differed by location. Pressures averaged across locations demonstrated a pattern of increasing pressure with increasing semi-occlusion.

Conclusions

HRM is feasible for measuring air pressure simultaneously at multiple locations in the lower vocal tract during phonation with high spatial and temporal resolution, providing rich data to augment understanding of vocal function. The high spatial and temporal resolution yielded by this new method, paired with preliminary evidence that pressures change by location as a function of phonatory task, may be useful in future assays exploring differences in lower vocal tract air pressures between normal and disordered populations.

The study of air pressure in the vocal tract is central to the study of vocal function. Vocal tract air pressure is associated with phonation efficiency, and changes to air pressure impact the phonatory acoustic signal (Tanaka & Gould, 1985). As such, accurate air pressure measurement is important for research and clinical practice. Current methods of air pressure measurement in the vocal tract in clinical practice are limited to the anterior oral cavity and to a restricted number of phonatory gestures. This is problematic because air pressure at different locations in the vocal tract may vary. This feasibility study proposes a novel method of measuring air pressure simultaneously at multiple locations in the lower vocal tract using high-resolution pharyngeal manometry (HRM).

Air Pressure in Voice Production: Clinical Significance

Air pressure supplies energy for vocal fold vibration, and modulation of air pressure through the course of the vocal tract results in changes to the acoustic signals we perceive as sound (Yost, 2007). Source-filter theory, first described in the mid 20th century (Fant, 1960), describes the modification of an acoustic signal from a vibratory source (i.e., vocal folds) by a filter (i.e., vocal tract) to produce voice and speech sounds. More recently, nonlinear source-filter interaction theory has suggested that modifying the vocal tract shape affects vocal fold vibration (Kent & Read, 2002; Titze, 2008; Titze & Story, 1997). Such shape modification alters efficiency of voice production (Bartholomew, 1934; Laver, 1980; van Houtte et al., 2011; Yanagisawa et al., 1989, 1990), and is hypothesized to result in changes in air pressure at the location of the shape modification. Computer modeling can estimate impacts of varying vocal tract shapes on the acoustic profile of phonation (Titze & Worley, 2009), but direct measurement of changes in lower vocal tract air pressure at multiple locations simultaneously has not been possible. Thus, direct, simultaneous measurement of phonatory air pressure at multiple locations in the in the lower vocal tract in vivo would allow us to validate findings from computer simulations and would significantly advance our ability to quantify dynamic interactions of vocal function across speech tasks.

Understanding air pressure variation at different levels of the vocal tract is clinically relevant. There is considerable overlap between gestures produced by the aerodigestive pathway for phonation, swallowing, and airway protection. Some gestures, such as pharyngeal constriction and tongue base retraction, are necessary for swallowing, but when these same gestures are activated during phonation, they lead to reduced voicing efficiency and are associated with voice disorders including muscle tension dysphonia (Guzmán et al., 2013, 2016; van Houtte et al., 2011). Voicing efficiency refers to the efficiency of the conversion of aerodynamic energy to acoustic energy at the level of the glottis during phonation, described by van den Berg in 1956 (van den Berg, 1956). The sensation of increased voicing efficiency has been theorized to be felt throughout the thorax, neck, and head, whereas the sensation of decreased efficiency may be localized more to the neck (Titze, 2001). Thus, inefficient gestures could be considered “maladaptive” modifications to vocal tract shape.

Semi-occluded vocal tract exercises (SOVTE) are therapeutic techniques commonly used in voice training and treatment of voice disorders (Angadi et al., 2019; Sundberg, 1974; Verdolini-Marston et al., 1995). SOVTE are often implemented in response to maladaptive vocal tract postures, to provide increased kinesthetic awareness of optimal vocal tract shape for most efficient voice production (Story & Titze, 1995; Titze, 2006). The acoustic and aerodynamic rationale underlying SOVTE is based on nonlinear source-filter interaction (Titze, 2006). In order to create a vocal tract shape that facilitates impedance matching of the glottis to the rest of the vocal tract, an individual phonates through semi-occlusions such as straws, lip trills, tongue trills, or bilabial fricatives. The semi-occlusion is theorized to result in kinesthetic feedback associated with improved impedance matching (Titze, 2006, 2020). The individual then uses that kinesthetic sensation as a target to be reproduced through epilaryngeal narrowing or relative pharyngeal widening in the absence of semi-occlusion, with associated changes to air pressure differentials throughout the lower vocal tract. Objective, in vivo physiologic evidence to support clinical use of SOVTE and identify optimal vocal tract configurations has been limited to pressure measurement in the anterior vocal tract (Maxfield et al., 2015).

Current Methods of Air Pressure Measurement: Limitations and Proposed Novel Methods

In most research studies and in clinical practice, intraoral pressure is measured from a single point in the anterior vocal tract during SOVTE to make estimates of oral, tracheal, subglottal, and transglottal pressures at certain moments in speech production (Guzmán et al., 2013, 2016; Hertegard et al., 1995; Maxfield et al., 2015; Patel et al., 2018; Smitheran & Hixon, 1981; Wistbacka et al., 2016). However, hypopharyngeal, oropharyngeal, and nasopharyngeal air pressures have not been simultaneously measured in these regions, despite the fact that air pressure is likely to differ across locations in the vocal tract during different phonatory gesture. To date, simultaneous pharyngeal air pressure measurements from multiple locations in the vocal tract have not been reported. Here, we propose the novel use of a validated clinical instrument, HRM, to overcome logistical limitations and quantify pharyngeal air pressure at multiple points in the vocal tract during speech/phonation tasks.

Pharyngeal HRM is a clinical procedure performed by speech-language pathologists, gastroenterologists, and otolaryngologists that objectively measures pressures from the velum to the proximal esophagus during swallowing. HRM uses circumferential pressure sensors embedded in a flexible catheter that is inserted transnasally and passed to the proximal esophagus (Omari et al., 2020). In addition to measuring contact pressure of corporeal structures during swallowing, HRM catheters are calibrated in an air pressure chamber and can accurately measure changes in intraluminal pharyngeal air pressure during behaviors such as expiratory muscle strength training (Hutcheson et al., 2017). Pressure data are collected from multiple, closely spaced sensors along the length of the pharynx, allowing for simultaneous measurement of pressure differentials at multiple levels of the lower vocal tract. This contrasts with single measurements of intra oral pressure in the anterior vocal tract, which have historically been used. Furthermore, HRM can be performed with almost any phonatory task, in contrast to clinical measurements that require very specific voicing tasks (Patel et al., 2018).

In this feasibility study, two subjects underwent HRM during various phonatory and nonphonatory tasks with different degrees of vocal tract semi-occlusion. The objectives of this study were to (a) assess the feasibility of HRM for measuring air pressure at multiple locations in the lower vocal tract simultaneously and (b) to compare air pressure across locations and among phonatory tasks. We hypothesized that (a) HRM would be feasible for measuring air pressure at multiple levels of the lower vocal tract and (b) air pressure would be increased near the velum and near the hypopharynx during SOVTE.

Materials and Method

Participants

To ensure participants could modify vocal tract configuration along the length of the vocal tract, we recruited two trained singers (Maxfield et al., 2015). Because of the exploratory nature of this research, we chose to study two subjects of relatively homogeneous size, age, and singing training. Two males (28 and 30 years old) without a history of voice, swallowing, respiratory, gastrointestinal, or neurological deficits participated in this study. The protocol was approved by the University of Wisconsin–Madison Institutional Review Board (IRB# 2013–1227-CP058). Informed consent was obtained from each subject prior to participation.

Equipment

A high-resolution manometer (ManoScan ESO, Medtronic) was used to measure pharyngeal pressures. The catheter has a diameter of 2.75 mm with 36 circumferential pressure sensors (each 2.5 mm in length) spaced 1 cm apart. Each sensor detects pressure via 12 circumferentially oriented sensors that are averaged to yield one data stream per sensor. The catheter is calibrated before each use in an air pressure chamber, and can record pressures between −20 and 600 mmHg with a fidelity of 2 mmHg. Pressure from each sensor was recorded at a sampling rate of 50 Hz.

To facilitate passage through the nares, less than 1 mL of topical 2% viscous lidocaine hydrochloride was applied to the catheter exterior and the nasal vestibule before insertion. After topical anesthetic application, the catheter was inserted transnasally until the distal tip was positioned in the proximal esophagus (see Figure 1). After catheter insertion, the participants acclimated to the catheter for 5 min.

Figure 1.

Left: fluoroscopic image of catheter placement during HRM. Arrow points to a pressure sensor. Right: spatiotemporal plot of pressure differentials during HRM. Warmer colors represent increasing pressure. Nasal quiescence is the region of low pressure in the nasal cavity. Pharyngeal area of interest demonstrates the location of the sensors on the y-axis from which data were analyzed.

Tasks

Phonatory and nonphonatory tasks were chosen to test a range of degrees of semi-occlusion and to facilitate comparison to data reported in the extant literature. Participants were asked to refrain from eating for 4 hr and drinking for 2 hr before the study. After the acclimatization period, the following tasks were performed by each participant: (a) sustained /a/ at a comfortable loudness for 5 s, three repetitions; (b) sustained /a/ at a soft loudness level for 5 s, three repetitions; (c) sustained /a/ at a high loudness level for 5 s, three repetitions; (d) resting nasal breathing; (e) resting oral breathing; (f) straw blowing without phonation for 5 s, three repetitions (straw inner diameter of 5 mm and length of 14.1 cm); (g) straw phonation for 5 s, three repetitions; (h) lip trill for 5 s; (i) bilabial fricative for 5 s; and (j) pinched straw blowing without phonation. All tasks were completed at a pitch that was perceived as comfortable to the participant. The number of repetitions for Tasks f–j was based on subject tolerance of the HRM procedure. Subject 2 additionally performed pinched straw phonation.

Data Collection and Analysis

Pressure data from the pharynx were analyzed with a custom-designed MATLAB program (MathWorks, Inc.). Data were assessed relative to atmospheric pressure to facilitate comparison to literature studying phonatory aerodynamics. The pharyngeal area of interest was defined as the five rostral-most sensors between the velopharynx and the upper esophageal sphincter (UES). Sensors were labeled sequentially from 1–5, with 1 being the most rostral. The velopharynx was identified as the region of pressure directly caudal to the area of continuous nasal quiescence, identified during comfortable /a/ (see Figure 1). The UES was identified as the region of tonically increased pressure caudal to the baseline low pressure zone of the pharynx and rostral to the low pressure zone of the esophagus (McCulloch et al., 2010). Data were analyzed for 3 s in the middle of each task, starting 1 s after task initiation. Pressures were analyzed from each of the five pharyngeal sensors, reported in mmHg. Thermal compensation was applied following catheter removal to account for possible sensor drift over the course of the study; this process involved the use of manufacturer-provided analysis software to record pressure at each sensor immediately after extubation, when no pressure was applied, but while the catheter was still at body temperature, then subtracting that measurement from the entire recording. In addition to absolute pressure measurements used for analysis, a delta measurement between nasal breathing prior to each task and reported absolute pressure was obtained to account for possible sensor drift over time (Lamvik et al., 2016). An intraclass correlation coefficient was calculated to compare the delta measurement to the absolute pressure.

Descriptive statistics calculated for each subject included mean, standard deviation, and confidence interval for the 3-s time interval of each repetition of the task (averaged across up to 9 s when three repetitions were obtained). Additionally, a composite mean value was calculated by averaging across all five pharyngeal sensors.

Results

Feasibility

The procedure was tolerated by both participants with minimal discomfort and was completed within a 15-min time frame. Acclimation time prior to task initiation facilitated tolerance of the procedure. Additional periods of quiescence were required between some tasks in order to re-acclimate participants to presence of the catheter and avoid gagging. Mean differences between (a) absolute pressure during tasks and (b) delta scores measured between pretask nasal breathing and absolute pressure were −0.4mmHG (SD = 0.71); the intraclass correlation coefficient was .985 (95% CI 0.977, 0.991], p < .01), indicating excellent reliability and suggesting minimal effect of sensor drift over time.

Pressure Differences Across Tasks

Table 1 contains means, standard deviations, and 95% confidence intervals for each subject and sensor across all tasks, as well as composite means averaged across all sensors. Composite means were lower for resting breathing and vowel production than for tasks that involved semi-occlusion for both subjects (see Figure 2). Mean pressures for each task were similar between subjects, and both subjects demonstrated generally increased mean pressures with increasing occlusion (see Figure 2). Pressures were similar but not identical between repeated productions of individual tasks within subjects (see Figure 3). Subatmospheric pressures were occasionally observed during tasks that involved very little semi-occlusion, but values do not fall out of the fidelity range from atmospheric pressure (±2 mmHg). These observations are not unexpected given the dynamic nature of pressure differentials in response to changing airflow direction (i.e., during biphasic breathing).

Table 1.

Mean pressures by subject and by sensor rank.

Task	Subject	Sensor rank	Mean (mmHg)	SD	95% CI
Comfortable /a/
	Subject 1	1	−0.621	0.906	[−1.647, 0.404]
		2	−0.851	0.005	[−0.857, 8.45]
		3	−0.932	0.025	[−.960, −0.903]
		4	−2.318	0.233	[−2.581, −2.055]
		5	−1.220	0.295	[−1.554, −0.887]
		Composite	−1.188	0.720	[−1.553, −0.824]
	Subject 2	1	0.337	0.060	[0.269, 0.405]
		2	1.718	0.501	[1.151, 2.286]
		3	−0.388	0.615	[−1.084, 0.308]
		4	0.396	0.182	[0.189, 0.602]
		5	0.513	0.238	[0.244, 0.782]
		Composite	0.515	0.773	[0.124, 0.906]
Straw phonation
	Subject 1	1	6.892	1.606	[5.075, 8.709]
		2	0.200	0.105	[0.080, 0.319]
		3	0.290	0.163	[0.105, 0.474]
		4	0.073	0.791	[−0.822, 0.967]
		5	−0.054	0.165	[−0.240, 0.133]
		Composite	1.480	2.886	[0.020, 2.940]
	Subject 2	1	0.252	0.045	[0.201, 0.303]
		2	4.789	0.693	[4.005, 5.573]
		3	0.308	0.052	[0.249, 0.367]
		4	−0.054	0.050	[−0.111, 0.002]
		5	0.666	0.064	[0.594, 0.738]
		Composite	1.192	1.895	[0.233, 2.151]
Lip trill
	Subject 1	1	6.096
		2	1.557
		3	1.699
		4	1.828
		5	1.446
		Composite	2.525	2.001	[0.771, 4.280]
	Subject 2	1	3.931	0.132	[3.781, 4.080]
		2	8.518	1.160	[7.206, 9.831]
		3	3.963	0.096	[3.855, 4.071]
		4	3.464	0.075	[3.379, 3.548]
		5	4.128	0.160	[3.946, 4.309]
		Composite	4.801	1.989	[3.794, 5.807]
Bilabial fricative
	Subject 1	1	8.297
		2	2.243
		3	2.466
		4	2.501
		5	1.758
		Composite	3.453	2.724	[1.065, 5.841]
	Subject 2	1	6.877	0.366	[6.462, 7.292]
		2	8.490	0.093	[8.385, 8.596]
		3	8.322	0.843	[7.368, 9.277]
		4	6.313	0.327	[5.943, 6.683]
		5	6.998	0.318	[6.638, 7.358]
		Composite	7.400	0.967	[6.911, 7.889]
Pinched straw phonation
	Subject 2	1	6.166	0.267	[5.864, 6.468]
		2	11.951	1.020	[10.796, 13.105]
		3	5.680	0.205	[5.448, 5.913]
		4	5.502	0.285	[5.179, 5.824]
		5	6.470	0.290	[6.142, 6.798]
		Composite	7.154	2.545	[5.866, 8.442]

Note. Mean pressure, standard deviation (SD), and 95% confidence interval (CI) for each sensor for each subject for phonatory task. Composite indicates mean pressure of the five sensors for that task and subject. Subject 1 completed a single repetition of lip trill and bilabial fricative.

Figure 2.

Mean pressure averaged across all five sensors for each subject. Tasks performed only by Subject 2 are not included in this figure. Error bars indicate standard deviation.

Figure 3.

Mean pressure for each sensor for each task. Error bars represent one standard deviation. Absence of error bars indicates that the task was performed only once.

Pressure Differences Across Sensors

Pressures differences were observed between rostral and caudal sensors during tasks involving semi-occlusion. For both subjects, a single rostral sensor (Sensor 1 for Subject 1, Sensor 2 for Subject 2) was consistently higher than caudal sensors, typically by > 3 mmHg. This pattern was present during tasks that involved semi-occlusion, regardless of whether the task involved phonation (see Figures 3 and 4; see Table 1). There were minimal pressure differences across sensors in tasks that did not involve semi-occlusion. In phonatory tasks, a compression–rarefaction pattern was observed in individual tracings (see Figure 4), reflecting more periodic pressure changes during sustained pitches. The frequency of these periodic pressure changes is a function of the fundamental frequency of the subject, limited by the 50 Hz sampling frequency of the manometry system (data not shown). Likewise, the peak-to-trough amplitude of these cycles is increased during loud /a/, and decreased during soft /a/. This pattern was absent during nonphonatory tasks.

Figure 4.

Pressure tracings for each sensor for the three middle seconds of each phonatory task.

Discussion

The primary purpose of this study was to assess the feasibility of HRM for measuring phonatory air pressure at multiple locations in the lower vocal tract simultaneously. As hypothesized, we were able to use HRM to obtain these measurements and were able to do so with a high degree of spatial and temporal resolution, and with good tolerance of the procedure by both subjects. In addition, we observed increases in lower vocal tract air pressure associated with increases in vocal tract semi-occlusion. This finding speaks to the construct validity of interventions that use SOVTE to modulate vocal tract air pressures to treat voice disorders (Mills et al., 2017; Titze, 2006).

Lending concurrent validity to the use of HRM to measure air pressure during phonatory tasks, mean pressures averaged across sensors located only in the lower vocal tract during semi-occlusion in the current study were similar to those reported in studies that have assessed air pressure at only a single point in the anterior vocal tract during the same tasks (Maxfield et al., 2015). Additionally, our results demonstrate a similar degree of variability between productions of a single task and between subjects, as well as a relative lack of variability during straw phonation versus other SOVTE, and similar ranking of pressure based on degree of semi-occlusion as that reported by Maxfield et al. (2015).

HRM is unique as a method for measuring pharyngeal air pressure during phonation in that pressures can be measured at multiple points simultaneously. Compared to standard clinical methods of air pressure measurement, this results in a degree of spatial resolution that has not previously been possible. This is clinically relevant in that the increased spatial resolution possible with HRM could provide more nuanced insight into phonatory aerodynamic changes that have been reported following interventions for voice disorders of multiple etiologies (Adams et al., 1996; Giovanni et al., 1999; Rammage et al., 1992; Zheng et al., 2012). A novel finding of this study is that pressure differs by location in the lower vocal tract during semi-occlusion. For both subjects, a single rostral sensor demonstrated consistently increased pressure compared to caudal sensors during tasks that involved semi-occlusion, and the size of this increase varied by task. A possible explanation for this finding is that, as air pressure increases, velar and pharyngeal musculature near the velopharyngeal port stiffens to increase constriction to avoid air leak. This stiffness could contribute to nonuniform intraluminal pressures between the rostral and caudal vocal tract. Additional explanations are that changes in rostral air pressure may be associated with changes in airflow direction at a bend in the tube that is the vocal tract, or that relative widening of the hypopharynx locally reduced intraluminal pressure. While it is possible that velopharyngeal closure shifts the catheter to contact the posterior pharyngeal wall with resultant pressure artifact, this is unlikely given the absence of this observation during oral breathing and production of /a/, both of which require velopharyngeal closure. Future studies using either fluoroscopic imaging with simultaneous HRM, or repetition of the current study with a catheter that allows for circumferential discrimination of pressures would provide further insight into the mechanisms that result in the rostral–caudal pressure differences, and could account for possible artifacts of pharyngeal contact pressures.

This feasibility study has a number of limitations. These include the two-subject design, which precludes more-sophisticated comparison of pressures among sensors and tasks, as well as the fact that subjects were both male and homogenous in terms of voice experience, size, and medical history. Because results are likely to vary as these factors are altered, future research should include a diversity of demographic, medical, and historical factors across a larger sample size. An additional limitation is fidelity of pressure measurement of ±2 mmHg, which is larger than that recommended by Patel et al. (2018). For example, if a change of 1 mmHg were to occur between two different tasks, it is possible that the sensors would not detect this change; this may explain the absence of rostral pressure differences across loudness levels during production of /a/. Despite this limitation, internal validity of the current findings is maintained in that pressure differences observed between sensors were greater than 3 mmHg and differences observed between tasks were greater than 15 mmHg. It should also be acknowledged that fundamental frequency is likely to have had some influence on pressures. While the substantial differences in air pressure between SOVTE and non-SOVTE tasks would likely be preserved regardless of fundamental frequency, fundamental frequency will be particularly important to consider when assessing air pressure in the absence of vocal tract semi-occlusion, when vocal tract shape may be altered relative to fundamental frequency in order to facilitate appropriate impedance matching. Furthermore, differences in lung volume could contribute to pressure changes in the vocal tract (Maxfield et al., 2015; Stathopoulos & Sapienza, 1993), and should be accounted for in future work. Finally, studies that implement this methodology in the future would benefit from establishing external validity through comparison of simultaneously obtained measurements of pharyngeal air pressures and anterior oral air pressure.

Pharyngeal HRM is a procedure commonly performed by speech-language pathologists, gastroenterologists, and otolaryngologists to assess swallowing. The fact that HRM was well tolerated during phonatory activities indicates that it would be feasible for voice specialists including speech-language pathologists and laryngologists to perform HRM to assess some aspects of vocal function. A primary benefit of HRM over traditional aerodynamic assessment of voice is the high degree of spatial resolution across the lower vocal tract during phonatory air pressure measurement. This may be clinically relevant, as relative differences in air pressure among lower vocal tract locations may differ between individuals who exhibit greater and lesser degrees of voicing efficiency. For example, during efficient voice production (i.e., appropriate impedance matching), a decrease in pharyngeal air pressure due to pharyngeal widening might be observed (Vampola et al., 2011) compared to less-efficient voice production (i.e., suboptimal impedance matching). This may serve to produce a visual biofeedback target for individuals who have difficulty with kinesthetic feedback during voice therapy (Vaiano et al., 2019). Further understanding of air pressure differences across populations with voice disorders could also aid in differential diagnosis. Data from normal and disordered populations should be obtained in future studies to characterize relative phonatory air pressure differences among locations in the lower vocal tract to further explore this hypothesis.

Conclusions

It is feasible to use HRM to detect changes in phonatory air pressure at multiple levels in the lower vocal tract. The current study demonstrates inconsistencies in pressure at different locations in the lower vocal tract during SOVTE. This instrumental assessment may be useful in future assays exploring differences in air pressure in the lower vocal tract in normal and disordered populations.

Funding Statement

Jesse D. Hoffmeister and Christopher L. Ulmschneider received support from National Institutes of Health (NIH), National Institute on Deafness and Other Communication Disorders (NIDCD) Grant T32 DC009401. Corinne A. Jones received support from NIH Grant F31 DC015709. Michelle R. Ciucci's research is supported by NIH, NIDCD, Grant R01 DC014358.