Changes in Bilabial Contact Pressure as a Function of Vocal Loudness in Individuals with Parkinson’s Disease

Abstract

This study evaluated the impact of vocal loudness on bilabial contact pressure (BCP) during the production of bilabial English consonants in adults with Parkinson’s disease (PD). Twelve adults with PD produced sentences with the phonemes /b, p, m/ initiating a linguistically meaningful word within the sentence, while BCP was sensed with a miniature pressure transducer positioned at the midline between the upper and lower lips. Stimuli were produced at two loudness levels: Habitual and twice as loud as habitual loudness (Loud). A linear mixed model (LMM) indicated a statistically significant main effect of Condition (F (1, 714) = 16.210, p < 0.001) with Loud having greater BCP than Habitual (mean difference of 0.593 kPa). The main effect of Phoneme was also significant (F (1, 714) = 31.905, p < 0.001), with post hoc tests revealing that BCP was significantly higher for /p/ compared to /m/ (p = 0.007), and for /b/ compared to /m/ (p = 0.002). An additional LMM of the magnitude of the percent change in BCP in the Loud condition relative to the Habitual condition had a significant main effect of Phoneme (F (2, 22.3) = 5.871, p = 0.006). The percent change in BCP was the greatest for /p/ (47.7%), followed by /b/ (35.7%) and /m/ (27.4%), with statistically significant differences for both /p/ and /b/ compared to /m/ in post hoc tests. The results indicated that changes in vocal loudness cause changes in BCP in individuals with PD. A louder voice was associated with higher BCP for all three phonemes, although the increase was the greatest on bilabial stops compared to nasal stops. These results provide initial insights regarding the mechanism by which therapeutic interventions focused on increasing loudness in people with PD alter oral articulatory behaviors. Future work that details potential aerodynamic (e.g., oral air pressure build-up) and articulatory acoustics (e.g., burst intensity) is needed to better explain the mechanistic actions of increased loudness that can explain why loud-focused speech treatments for people with PD may improve speech intelligibility.

1. Introduction

Parkinson’s disease (PD) is a progressive neurodegenerative disorder characterized by motor, sensory, and cognitive symptoms that can impact voice, speech, and language function [1]. Disruption to basal ganglia pathways is implicated as a primary cause of symptoms, although widespread network disruptions beyond the basal ganglia contribute to the full set of motor and nonmotor impairments [2]. In terms of speech, individuals with PD often have hypokinetic dysarthria, which is characterized by a combination of features such as reduced loudness, voice quality changes, and articulatory imprecision, among others [3,4]. Of the estimated 6 million people globally who have PD [5,6], up to 90% have alterations to their speech sufficient to reduce their speech intelligibility and reduce their quality of life [7,8,9].

Changes to articulation of consonants and vowels are relatively common in people with PD, with studies indicating that approximately 40–45% have such impairments [10,11]. Distortions to both vowels and consonants have been reported [3], but “imprecise consonants” have been found to be the most perceptually deviant articulation feature [12,13]. Consonant production requires constriction of the vocal, either partially (such as for “s” with the tongue tip raised toward the front of the hard palate) or completely (such as for “p”, which requires complete closure of the upper and lower lips for a brief period of time) [14]. These constrictions alter the oral aerodynamics, resulting in frication noise from turbulent airflow (for partial constrictions) or bursts from oral air pressure build-up and release (for complete constrictions). Of note, consonants requiring greater vocal tract constriction are reportedly misarticulated more often by people with PD than consonants that require less constriction [15].

A primary communication rehabilitation focus for people with PD is on increasing vocal loudness through treatment programs such as LSVT LOUD^® [16,17] and SPEAK OUT!^® [18,19]. Outcome studies of such programs indicate that the programs are effective at increasing vocal loudness [20,21,22]. Of relevance here is that loudness-focused interventions for people with PD have been reported to also have a positive impact on articulation [23]. For example, when speaking louder, increased distinctiveness of consonants [19] and improved speech intelligibility [24] in people with PD have been reported.

While articulatory changes have been described, the mechanistic reasons for improved articulation during loud speech remain unclear in the current literature. Increases in the amplitude and velocity of the tongue and jaw during loud speech in people with PD have been described [25]. However, features of the articulatory constriction for consonants have not been addressed, particularly as they relate to the articulation of bilabial consonants. A deeper understanding of how articulation changes when loudness is increased could help to further optimize therapy programs for people with PD and/or further serve to motivate people with PD to participate if not only loudness but also articulatory dynamics can be improved. The purpose of this study was to evaluate the impact of vocal loudness on the BCP of bilabial consonants. Two loudness conditions were evaluated: self-determined habitual loudness (Habitual) and twice as loud as habitual loudness (Loud). The hypothesis of the study is that increasing vocal intensity will result in an upscaling of articulatory movement that will be reflected in increased BCP. This hypothesis is based on the abovementioned positive changes to articulation in people with PD when speaking louder and studies on people without PD indicating that articulatory contact pressure is increased in loud speech [26].

2. Materials and Methods

2.1. Participants

Twelve adults (seven females and five males) ranging from 47 to 80 years old (mean = 63.7 years, standard deviation = 10.0 years) participated. Demographics, medical history, and PD-related clinical information are summarized in

Table 1. Demographics and Parkinson’s disease (PD) clinical information. S = subject, H&Y Score—Hoehn & Yahr Score, M = male, F = female.

S	Age	Sex	H&Y Score	Years Post PD Dx	Current PD Symptoms	PD Medication(s)
1	56	M	2	6	tremor, fatigue, depression	Sinemet, Mirapex, Selegiline
2	48	M	1	2	tremor, balance, shuffling gait, voice change	Sinemet, Rasagiline
3	67	M	4	14	bradykinesia, dysphagia, balance,	Parcopa, Amantadine
4	70	F	1	5	tremor, shuffling gait, fatigue	Sinemet, Ropinirole
5	67	M	1	7	tremor, bradykinesia	Sinemet, Parlodel, Mirapex
6	63	M	2	1	bradykinesia	Sinemet, Mirapex
7	65	M	1	2	tremor, bradykinesia, shuffling gait	Sinemet, Mirapex
8	77	M	1	6	tremor	Sinemet, Mirapex
9	80	F	3	2	tremor	Sinemet, Mirapex
10	60	F	2	2	tremor, rigidity	Sinemet, Requip
11	64	F	1	5	tremor, stooped posture	Sinemet, Mirapex
12	47	F	2	2	dystonia, tremor	Sinemet

. Although PD severity ranged from relatively mild to severe (Hoehn & Yahr Scale scores from 1 to 4), the majority fell toward the mild end of the range. Self-reported communication, swallowing, and cognitive issues are reported in

Table 2. Communication, swallowing, and cognitive issues. S = subject.

S	Speech and Voice	Speech Treatment	Smelling and/or Swallowing Issues	Cognitive Difficulties
1	imprecise articulation, slow rate	none	coughing, aspiration	concentration, problem solving
2	reduced loudness	LSVT	Coughing, aspiration, drooling, anosmia	short-term memory, concentration, problem solving
3	reduced loudness, imprecise articulation	LSVT	Coughing, aspiration	self-monitoring, insight
4	imprecise articulation	none	other GI problems	general deterioration
5	reduced loudness, lower pitch	group	Coughing, aspiration, pills stuck in throat	none
6	vocal fatigue	none	none	--
7	reduced loudness	group	none	general deterioration
8	reduced loudness	none	none	short-term memory
9	low pitch, rough voice	group	pills stick in throat, anosmia	none
10	no issues	none	anosmia	general deterioration
11	reduced loudness	none	anosmia	short-term memory, concentration, problem solving
12	no issues	none	none	short-term memory, general deterioration

. All patients were on pharmaceutical PD treatment. Patients who had undergone PD neurosurgery (such as deep-brain stimulation) were excluded, given reports of potential inducement of dysarthria from the procedure [27]. Most had not undergone speech therapy (n = 7), although two had completed LSVT and three others had completed group speech therapy. Participants were a convenience sample of patients followed through a movement disorders clinic who had a PD diagnosis confirmed by a neurologist. All spoke standard American English as their primary language.

2.2. Speech Stimuli

The stimuli sampled in this study were three bilabial phonemes (/b, p, m/) initiating a bi-syllabic real English word (Bobby, Papa, and Mama, respectively) embedded in a meaningful sentence (“Buy ___ a puppy”). These three phonemes provided the opportunity to sample BCP across bilabial phonemes that varied in terms of their voicing feature (/p/ vs. /b/) and manner of production (/b/ vs. /m/; oral plosive vs. nasal continuant). The structure of the three words maintained the experimental phoneme of interest in the initiating position of the word, followed by a central vowel. The word initiating the sentence (“Buy”) ended in a vowel; thus, the experimental phoneme of interest was bounded by vowels, allowing for easy identification of the peak ACP of the target phoneme because vowels are produced with the lips parted (i.e., no central lip tissue contact).

The sentences were produced 10 times each in two loudness conditions: Habitual and Loud. The stimuli were presented on a computer screen with instructions to say the phrase in either their conversational loudness level (i.e., Habitual) or twice as loud as conversational (i.e., Loud). For the Loud condition, a direct magnitude estimation (DME) procedure previously described in a study of tongue–palate contact pressures in adults without speech deficits was used [26]. The instruction was to consider habitual loudness to have a value of 100, followed by the prompt to double their loudness so that the sentence was read at a loudness equivalent to a value of 200 (Figure 1). The target loudness level for a given production was indicated by altering the displayed text to show the intended loudness condition. Participants were not required to precisely double their loudness. Rather, the intent was to encourage louder speech, such as in LSVT LOUD^® and SPEAK OUT!^® treatments.

Each sentence was produced 10 times in the two loudness conditions (Habitual and Loud) for a total of 60 productions per participant, resulting in 720 samples for the 12 participants. All stimuli were fully randomized for each participant. Participants were allowed to take sips of water at any time during the sentence recordings, and the researcher controlled the pace of stimulus presentation to allow for a breath in between each production.

2.3. Instrumentation

The instrumentation has been described previously [28]. Briefly, an Entran EPI-BO transducer (Entran, Fairfield, NJ, USA) was used to measure BCP. This is a piezoresistive pressure sensor with a pressure range from 0 to 68 kPa and a high-frequency response, making it suitable for dynamic pressure measurements such as articulation contact pressure. The small sensing surface (2.58 mm), slim profile (0.45 mm thick), and light weight (0.085 g) with a temperature compensation module allow it to be placed between the lips without altering articulation. This transducer was mounted on a 0.5 mm acrylic strip held by an adjustable clamp so that the transducer was positioned on the lower lip of the participant. The transducer was placed in midline with the sensing surface against the lower lip (Figure 2). Positioning was such that the full sensing surface of the transducer was covered by the lips during closure for bilabial phoneme production. The transducer signal was amplified and routed to a PowerLab 8 digital system (ADInstruments, Colorado Springs, CO, USA) for signal display and recording using LabChart 8.1.16 software (20 kHz digitization, 16-bit precision; ADInstruments). The signal was low-pass filtered at 50 Hz to remove any voicing signal that might be present in the pressure curve, and a peak-picking routine in LabChart marked the maximum BCP signal excursion corresponding to the phoneme of interest. An audio signal was recorded in the second channel of LabChart using an AKG C410 headset microphone. With concurrent display of the BCP and audio signals, researchers could isolate the experimental phoneme of interest and play back the audio to confirm the production for measurement.

2.4. Procedures

Participants completed all data collection in one visit lasting approximately 30 min while seated in a quiet research lab. They were positioned with the computer monitor for displaying stimuli at a comfortable height for reading. Data recording proceeded with the researcher advancing the stimulus display (3 s interstimulus pause between sentences, or longer if a patient needed a sip of water). If a participant misread a sentence, the trial was repeated. During the data collection, participants could sit back from the recording instrumentation when needed (for example, to swallow saliva or to take a brief break). They were then repositioned with the pressure sensor resting in the same midline lower lip position. A small ink dot was placed on the lower lip at this location so that the transducer could be placed in the same position each time.

2.5. Analysis

To address the study aim, a linear mixed model (LMM) was used to examine the fixed effects of the Condition (Habitual versus Loud) and Phoneme (/b/, /p/, /m/) on BCP, with Participant included as a random intercept to account for within-subject correlations. A model with an unstructured covariance matrix was chosen for the repeated measures factor to allow for unequal variances and covariances across repeated trials of the stimuli in each condition. The unstructured matrix gave the best model fit compared to other models tested (Akaike’s Information Criterion [AIC] values of 253 for the unstructured model compared to AIC values of 295 and 316 for compound symmetry and autoregressive structures, respectively). The Satterthwaite approximation was applied to estimate degrees of freedom. Additionally, descriptive statistics, including means, standard deviations, and ranges of BCP, were calculated for each speaking condition and phoneme, and box plots showing the median, interquartile ranges, and full ranges were constructed.

Percent change in BCP in the Loud condition relative to the Habitual condition was also of interest. Boxplots and descriptive statistics for these data were generated. Additionally, another LMM was conducted to examine whether the percent change from habitual to loud speech differed across the three phonemes. This model included Phoneme as a fixed effect and Participant as a random intercept to account for repeated measures. As above, an unstructured covariance structure was specified for the repeated effect (Phoneme) to allow for unequal variances and covariances between phonemes. The Satterthwaite approximation was used again to estimate the degrees of freedom for testing the fixed effects.

3. Results

Figure 3 displays box plots of the BCP group data for each phoneme and loudness condition. The LMM results from the Type III tests of fixed effects indicated a significant main effect of speaking condition, F (1, 714) = 16.210, p < 0.001, with higher BCP observed during the Loud condition (M = 2.307, SE = 0.666) compared to the Habitual condition (M = 1.714, SE = 0.494). Phoneme also had a significant main effect, F (2, 714) = 31.905, p < 0.001. However, the interaction between Condition and Phoneme was not significant, F (2, 714) = 0.971, p = 0.402.

For the fixed-effect of Condition, parameter estimates indicated that loud speech was associated with a 0.593 kPa increase in BCP compared to habitual speaking loudness (t = 16.21, p < 0.001).

Table 3. Means, standard deviations (SD), and 95% confidence intervals (95% CIs) for fixed-effect variables, and for differences between all levels within each fixed-effect (i.e., paired comparisons), which additionally include post hoc statistical outcomes—indicates no statistical test to report.

Fixed-Effect	Mean/Mean Difference	SD	95% CI	t	p
Condition
Habitual	1.714	0.482	1.442–3.156	--	--
Loud	2.307	0.556	1.993–4.300	--	--
Loud–Habitual	0.593	0.148	0.509–0.676	16.211	<0.001
Phoneme
/b/	2.086	0.730	1.673–3.759	--	--
/p/	2.387	0.663	2.013–4.400	--	--
/m/	1.558	0.615	1.210–2.769	--	--
/p/–/b/	0.301	0.835	−0.171–1.130	1.250	0.237
/b/–/m/	0.528	0.450	0.273–0.801	4.067	0.002
/p/–/m/	0.829	0.876	0.333–1.162	3.277	0.007

includes parameter estimates with 95% confidence intervals.

For the fixed-effect of Phoneme, post hoc tests revealed significantly higher BCP for /p/ compared to /m/ (t = 3.278, p = 0.007), and for /b/ compared to /m/ (t = 4.067, p = 0.002), but /p/ and /b/ did not differ (t = 1.250, p = 0.237). See Table 3 for descriptive statistics, including 95% CIs for the parameter estimates and the mean paired comparison differences.

Overall, the model explained 39% of the variance in contact pressure due to fixed effects (marginal R² = 0.39) and 56% when accounting for both fixed and random effects, i.e., individual subjects (conditional R² = 0.56).

To further capture the magnitude of the pressure change as a function of loudness, the percent change in BCP in the Loud condition relative to an individual participant’s Habitual condition was calculated for each phoneme. Group data for these percent differences are presented as box plots in Figure 4. The LMM analysis revealed a significant main effect of Phoneme on percent change in BCP, F (2, 22.3) = 5.871, p = 0.006. Estimated marginal means showed that the percent increase was greatest for /p/ (M = 47.7%, SE = 13.783), followed by /b/ (M = 35.7%, SE = 10.317), and least for /m/ (M = 27.4%, SE = 7.913). Post hoc pairwise comparisons with Bonferroni adjustment indicated that the percent change for /p/ was significantly greater than the change for /m/ (p = 0.016). There was not a significant difference in the percent change for /p/ compared to /b/ (p =0.060) or for /b/ compared to /m/ (p = 0.190). See

Table 4. Descriptive statistics for the percent change data for each phoneme. SD = standard deviation, 95% CI = 95% confidence interval.

Phoneme	Mean (%)	SD	Minimum	Maximum	95% CI
/b/	35.738	14.289	17.994	55.520	27.7–63.4
/p/	47.744	21.646	18.430	90.755	35.5–83.2
/m/	27.413	14.802	13.704	68.711	19.0–46.5

for additional descriptive statistics.

4. Discussion

This study evaluated the effect of vocal loudness on bilabial contact pressure for the phonemes /b, p, m/ during speech from people with PD. Overall, the findings supported the hypothesis that BCP is increased with an increase in vocal loudness. While a few studies have reported improvements in articulation and speech intelligibility when people with PD increase vocal loudness through approaches such as LSVT^® and SPEAK OUT! [19,23,24], there is limited investigation regarding how this occurs. Studies have demonstrated that the jaw, lip, and tongue range of motion and velocity are reduced when people with PD are speaking at their comfortable loudness [29,30,31]. Conversely, Kearney et al. [25], utilizing a three-dimensional electromagnetic tracking system, demonstrated that articulatory working space, range of movement, and average speed of movements of the jaw and tongue increased when PD speakers increased loudness. Dromey [32] specifically assessed lip velocity and displacement during bilabial phoneme production in people with PD and found that both increased when speakers increased their sound pressure level. Based on these prior articulatory kinematic studies, it seems likely that the Loud condition in the current study may have resulted in increased velocities and displacement of the lips with a corresponding increase in the collision force between the upper and lower lips, reflected in higher BCP. Future studies are needed that combine measurement approaches of the articulator movement (i.e., kinematics) and the resulting end product of those movements (i.e., articulatory contacts).

The percent change in BCP in the Loud relative to the Habitual condition varied across phonemes, with group means ranging from approximately 27% to 48% with /p/ being significantly greater than /m/ (and no other significant paired comparisons). Increases in tongue–palate contact pressure for the stop consonants /t/ and /d/ in the Loud compared to Habitual speaking condition were 38% and 36%, respectively, in adults without speech disorder and no neurological impairment [26]. Data for lip-to-lip contact pressure (BCP) during loud speech in adults without speech disorders are not available. Overall, the current results for people with PD suggest a relatively comparable upscaling of articulatory contact pressure in loud speech, although this occurred to a lesser extent on the phoneme /m/. The most likely explanation for the greater percent change in the Loud compared to Habitual condition on /p/ compared to /m/ relates to the aerodynamics and velopharyngeal closure status for each phoneme. That is, for the oral consonant /p/, the velopharynx is closed to allow air pressure build-up to produce the burst release. When increasing loudness, even greater oral air pressure build-up is anticipated, potentially necessitating more forceful bilabial closure to contain the mounting pressure. In contrast, the nasal bilabial consonant /m/ is produced with an open velopharyngeal port and without the need for high air pressure build-up (i.e., /m/ does not have a burst release like /p/). In this case, the articulatory system does not need to increase lip contact forces to the same extent.

There was an expectation that BCP in the Habitual condition for the PD speakers would be lower than values reported in the existing literature. However, this was not the case. The group mean BCP for the PD subjects during the Habitual condition was similar to values reported for adults without a neurological or speech condition [28], i.e., approximately 1.5 to 2.4 kPa. Given that the amplitude and velocity of articulator movements are reduced in people with PD [25,33], the expectation was that during habitual speech, individuals with PD in the current study might generate lower BCP because of their motor impairment. There are several possible reasons why BCP was similar for the PD participants compared to non-PD participants in an earlier study [28]. One possible reason why BCP in the PD speakers was within the previously reported range for non-PD adults relates to PD severity. The majority of the PD speakers in the current study had relatively early-stage PD, as evidenced by Hoehn and Yahr Scale scores of 2 or less for 83% of the participants. Although each had some changes to their communication, these were fairly mild in severity, and it may be that the motor activity of the lips was not heavily impacted. A second possibility for explaining the similarity in BCP between PD and non-PD participants is that individuals in the pre-PD and early PD stages may adjust their articulation in a compensatory manner to try to maintain speech intelligibility [34]. If this were the case, perhaps BCP is maintained via compensation. Besides PD severity and the possibility that they were using compensatory articulation, a third possible reason for the similar BCP between PD and non-PD participants in the earlier study is that there were subtle differences in the speech stimuli between the current study and Searl [28]. It may be that such differences could influence BCP, although this seems unlikely because the target phonemes (/b/, /p/, /m/) were the same, as was position within the word (initiating a bisyllabic word) and surrounding context (vowels immediately preceding and following the target phoneme). Future replication studies with increased sample size are needed to confirm the current findings. Such studies would be strengthened by collecting BCP, PD severity, and speech intelligibility measures longitudinally to see what relationships emerge and how they evolve as a function of disease progression.

From a clinical perspective, training people with PD to increase vocal loudness appears to result in increased BCP. Higher oral air pressure build-up behind the point of articulatory constriction is anticipated in the Loud vocal condition, given the greater aerodynamic power and subglottal air pressure that occurs during loud voicing [35]. Increased oral pressure during a stop consonant is associated with a more intense burst release on the phoneme, which could positively impact speech in terms of clarity [26]. A more salient burst release could lead to listeners’ perception of better speech precision and intelligibility, as has been reported when loud-focused therapies have been applied to various patient populations. Attempts were made early in the current study to place an oral air pressure sensing tube between the lips and into the oral cavity, which would have allowed direct assessment of the relationship between the contact pressure and the aerodynamic pressure when the PD participants were speaking. Although this was possible in earlier studies in non-PD speakers, the presence of the oral pressure tube in the PD participants in which this was trialed resulted in a clear distortion of their speech. An additional problem with obtaining oral air pressure measurements in the PD participants is a greater occurrence of saliva pooling in the mouth (not uncommon in PD), which not only can create speech distortion, but it can also intermittently plug the opening of the air pressure sensing tube, resulting in invalid measurements.

This study demonstrated that BCP increased when people with PD increased their vocal loudness. The results add to the kinematic results in the literature regarding the loudness impact on articulator movement velocities and ranges, with a general upscaling of oral articulatory activity when vocal loudness is increased in people with PD. These results provide initial insights into the mechanistic action of vocal loudness on oral speech movements that may occur in clinical scenarios where a louder voice is trained, such as LSVT and SPEAK OUT! Although future studies would be needed, it may be possible to use BCP as visual biofeedback during loud speech training for people with PD. It may also be that there is an optimal BCP range that can be determined that balances improved speech clarity or intelligibility with the physical exertion required to increase loudness and articulatory precision.

Several limitations of the current study were pointed out above. This includes the need for increased sample size and multimodal assessment of speech production involving oral air pressure and perhaps kinematic measures. These would allow a more complete understanding of the changes in bilabial speech production. Additionally, longitudinal assessment of BCP with corresponding perceptual data from listeners judging speech intelligibility or clarity would be informative to understand how articulatory dynamics of lips change with disease progression. In such longitudinal studies, it may be possible to learn whether PD participants continue to demonstrate an increase in articulatory contact pressures even as they progress to a more severe motor disease state. Lastly, the current study mimicked therapeutic interventions such as LSVT and SPEAK OUT! in terms of the instruction to increase loudness. However, the PD participants did not undergo the actual therapeutic program, which extends over multiple weeks. As a person progresses through such a therapy program, articulatory dynamics may continue to evolve in ways that cannot be captured in a cross-sectional study such as this one. Assessing articulatory dynamics as an outcome measure following LSVT and/or SPEAK OUT! therapy has clinical value in helping explain reported improvements in speech intelligibility and clarity in some people with PD.