Abstract
Purpose:
Vocal fatigue (VF) is a primary vocal symptom experienced by professional voice users, such as teachers, whose voice is an occupational tool. The study determines the utilization of the Vocal Fatigue Index (VFI), a 19-item scale in identifying symptoms of vocal fatigue and its severity in teachers.
Methods:
Using responses of 695 teachers who completed the VFI, Mokken scaling was conducted on the items to identify the experiences of vocal fatigue and its associated hierarchical nature of vocal fatigue symptoms in teachers. Mokken scaling was completed on a total of four groups: (a) Total teachers group, (b) No VF group, (c) Low VF group, and (d) High VF group.
Results:
Results revealed differences in item hierarchies between total teachers and across the separate groups of VF severity. Item hierarchy for teachers highlighted items from physical discomfort at the mild end of the hierarchy to items from symptom improvement with rest at the severe end of the hierarchy. Items related to avoidance presented as a separate scale for teachers presenting with high VF.
Discussion:
Mokken scaling in teachers provides insight into the underlying complexity of the experience of vocal fatigue symptoms and reliance on differential behavioral strategies in its management, suggesting the heterogenous nature of latent trait for vocal fatigue in this specific population.
Keywords: Vocal Fatigue, VFI, Mokken, Teachers
1. Introduction:
Vocal fatigue is a common and debilitating symptom, affecting individuals with and without voice disorders1,2. Although vocal fatigue has been implicitly identified in the literature through attempts to identify the underlying mechanisms and treatment approaches, significant gaps exist in our understanding of this complex phenomenon1,3. Chief among these gaps in knowledge is the lack of a consensus operationalized definition of vocal fatigue across studies1,3,4. Definitions have varied including (a) considering vocal fatigue as a negative adaptation to a vocal task5, (b) vocal fatigue as a self-reported phenomenon6–8 and (c) vocal fatigue as a phenomenon of increased phonatory effort associated with a decrease in phonatory performance4. Such varied definitions have made it difficult to compare findings between studies.
Recently, a consensus paper sought to clarify the definition of vocal fatigue by differentiating it from other, related phenomena such as vocal demand, physiological response to demand, and perceptions of effort in response to that demand3. In this consensus, vocal fatigue has been defined as not only the sensation of fatigue, but also a quantifiable decline in function influencing vocal output. Despite this clarification, little is known about the subjective characterizations of vocal fatigue. This is likely due to the varied presentation of vocal fatigue in largely heterogeneous populations. Importantly, little is known about the behavioral characteristics of and group experiences related to vocal fatigue in various populations, especially professional voice users.
Vocal fatigue is a primary vocal symptom experienced by teachers who have both an increased occupational demand and are predominantly females9–11. Vocal fatigue in this population can result in chronic voice issues, increased health-care costs12, and decreased student engagement in the classroom13,14. Given the increased prevalence of vocal fatigue in teachers and its broader impact on health care and student engagement, it is imperative to identify behavioral characteristics of vocal fatigue in teachers to guide future prevention and treatment strategies to ameliorate this vocal symptom.
1.1. Vocal Fatigue Index
The Vocal Fatigue Index (VFI) was developed as a tool to reliably identify individuals with vocal fatigue and increase our understanding of its symptom profile and associated behavioral characteristics. The VFI is a global, retrospective, self-report of vocal fatigue which measures trait fatigue, defined as the perception of fatigue over an extended period of time3,15. The VFI was developed using a sample of individuals with a wide range of diagnosed voice disorders resulting in dysphonia1. Three factors were identified that characterized some of the complexity within the construct of vocal fatigue including: 1) Tiredness and Avoidance: 11 items; 2) Physical Discomfort: five items; and 3) Symptom Improvement (or lack thereof) with Rest: three items. Logistic regression analysis indicated that Factors 1 and 3 predicted the presence or absence of vocal fatigue; however, Factor 2 was not a predictor of vocal fatigue in this clinical population1. However, teachers sampled in other research have reported elevated scores on Factor 2 of the VFI, indicating potential hierarchical differences of symptom profiles in various populations16,17. Further, novel investigation of scaling hierarchy differences between populations is warranted to ensure efficient and targeted analysis of voice symptoms.
1.2. Mokken Scaling
Mokken scaling assumes that to some degree all scale items reflect the underlying trait of interest (in this case, vocal fatigue), and that each item can be placed in a hierarchy, whereby some items reflect greater aspects of this trait than other item18. The hierarchy of items provides insight into the degree to which an individual presents on the underlying trait of a scale18. For example, if item A on the VFI reflected a mild degree of vocal fatigue and item B reflected a moderate degree of vocal fatigue, Mokken scaling would infer that all those who keyed B positively would have also keyed A positively. Mokken scaling predicts the likelihood that someone who keyed item B positively would possess a greater amount of vocal fatigue based on the likelihood that items A and B are hierarchically related. This analysis allows for investigations into unfamiliar territory when a scale is used to measure a trait within an unknown population. In this approach, scale development is used not just to measure the presence or absence of a trait (the binary approach), but also to investigate relational aspects of that underlying trait in different populations. Prior work using Mokken scaling analyzed responses from individuals with voice disorders to determine a tailored hierarchical structure of the VFI19. Within this group of individuals, Mokken scaling resulted in a single scale, with items related to tiredness at the easy end of the hierarchy, indicating mild vocal fatigue, followed by items related to avoidance, and finally, items related to physical discomfort. Items corresponding to vocal rest formed a separate scale. Mokken Scaling of VFI responses obtained from teachers can provide initial data on the potentially unique experiences of professional voice users, and may lead to a re-structuring of the original factors.
The aim of this study is to use Mokken scaling to identify the teachers’ experience of vocal fatigue and the hierarchical nature of vocal fatigue symptoms in teachers in comparison to those with voice disorders and vocally healthy controls. Towards this end, Mokken scaling can inform us how the construct of vocal fatigue may differ in a population of teachers compared to those with diagnosed voice disorders. Understanding the hierarchical nature of the VFI items and how they may differ in various populations provides a glimpse into the nature of vocal fatigue and its presentation in individuals with increased vocal demands. We hypothesize that in teachers, the hierarchical structure of the VFI will differ and provide insight into their behavioral strategies and experiences of vocal fatigue.
2. Methods:
2.1. Participants
Data for this study came from a sample of 695 full-time teachers (males −124, females- 571) described in previous work17,20,21, where teachers throughout the US completed the VFI as part of an online survey. The proportion of male to female teacher responses per grade level was consistent with the national average per the census of the year prior to data collection (2012). Grade level did not have a significant effect on reported levels of vocal fatigue in this sample22.
For analysis, based on the norms from the original VFI1, teacher responses were classified according to the VFI Factor 1 score into 4 groups: 1) Total Teachers, 2) No Vocal Fatigue (VF), 3) Low VF, and 4) High VF. The number of teachers per group and reported VFI cut-off values for each group are outlined in Table 1. Hierarchical item variations of vocal fatigue in those with low or high VF and without the presence of VF would indicate different signatures of vocal fatigue, reflecting group specific VF experiences and responses.
Table 1:
Classification of the groups based on Factor 1 scores of the VFI and participant demographics
| Vocal Fatigue (VF) Group | Number of Teachers (males, females) | Age (years) | Factor 1 VFI Score |
|---|---|---|---|
| Total Teachers | 695 (124, 571) | 42.7 | All Factor 1 Scores |
| No VF | 167 (30,137) | 43.6 | ≤ 11 |
| Low VF | 397 (55, 342) | 44.9 | between 11 and 24 |
| High VF | 131 (39, 92) | 43.7 | ≥ 24 |
2.2. Procedures- Mokken Scaling on Vocal Fatigue Index (VFI)
Similar to Rasch modeling in Item Response Theory, Mokken scaling suspends the strict assumption of these models based on binary responses (yes/no) and extends hierarchical analysis to items that have more than two responses. Furthermore, Mokken analysis allows for an analysis of multiple responses without specifically knowing if each item’s response has the same value distribution across all other items. For e.g., is the difference between a response of 2 and a response of 3 for item 6 the same as the distance between response 2 and 3 for all other items? Because the VFI allows for five possible responses per item, Mokken analysis is a more appropriate method of analysis. Mokken scaling was run on the Factor 1 scores of the VFI data from teachers. Figure 1 outlines the detailed steps involved in Mokken analysis.
Figure 1.
Mokken Scaling steps after an automated item selection procedure. See Appendix A for a detailed explanation of each of the steps and the norms.
2.3. Data analysis
Data obtained from teacher survey responses were analyzed in Rstudio software (Version 3.3.2) on the PC version (R Foundation for Statistical Computing, Vienna, Austria). For each of the four groups, Mokken scaling divided data into scales based on the automated item selection procedure. The number of scales, their reliability (Rho), and the invariant item ordering (HI) were recorded. Further analysis of monotone homogeneity and double monotonicity were completed to identify any potential outliers, repeating items, or signs of interactions/intersections between items.
3. Results
3.1. Total Teachers
Mokken scaling on all 695 teachers grouped all three factors under one scale. Three items were removed from the analysis; two items pertaining to avoidance behaviors and one item related to improvement with rest, (#1, #7, and #18), resulting in a single scale of 16 items. See Table 2 for a full report of items their means, H’, and the VFI Factor from which they belong. In general, the hierarchy ranged from items related to the Tiredness and Avoidance Factor, to items of the Physical Discomfort factor and ended in items related to the Symptom Improvement rest. Interestingly, results differed considerably when teachers’ groups were split representing low, high, and no vocal fatigue based on VFI scores.
Table 2.
Mokken Scaling for all Teachers.
| Scale 1 | Rho = 0.945 | HT = 0.464 | Qualified |
|---|---|---|---|
| Question | Mean | HT | Original VFI factor |
| 2. My voice feels tired when I talk more | 3.052 | 0.58 | 1 |
| 4. My voice gets hoarse with voice use | 2.790 | 0.61 | 1 |
| 3. I experience increased sense of effort with talking | 2.675 | 0.58 | 1 |
| 11. My voice feels weak after a period of voice use | 2.549 | 0.67 | 1 |
| 14. My voice feels sore when I talk more | 2.537 | 0.66 | 2 |
| 6. I tend to generally limit my talking after a period of voice use | 2.476 | 0.55 | 1 |
| 5. It feels like work to use my voice | 2.456 | .64 | 1 |
| 10. I find it difficult to project my voice with voice use | 2.442 | 0.60 | 1 |
| 9. It is effortful to produce my voice after a period of voice use | 2.374 | 0.66 | 1 |
| 13. I experience throat pain at the end of the day with voice use | 2.304 | 0.63 | 2 |
| 15. My throat aches with voice use | 2.302 | 0.65 | 2 |
| 8. I feel I cannot talk to my family after a work day | 1.965 | 0.55 | 1 |
| 12. I experience pain in the neck at the end of the day with voice use | 1.863 | 0.57 | 2 |
| 16. I experience discomfort in my neck with voice use | 1.858 | 0.59 | 2 |
| 17. My voice feels better after I have rested | 1.468 | 0.41 | 3 |
| 19. The hoarseness of my voice gets better with rest | 1.449 | 0.36 | 3 |
| 2. My voice feels tired when I talk more | 3.052 | 0.58 | 1 |
| 4. My voice gets hoarse with voice use | 2.790 | 0.61 | 1 |
| 3. I experience increased sense of effort with talking | 2.675 | 0.58 | 1 |
| 11. My voice feels weak after a period of voice use | 2.549 | 0.67 | 1 |
| Items disqualified from scaling | |||
| Question | Original VFI factor | ||
| 1. I don’t feel like talking after a period of voice use | 1 | ||
| 7. I avoid social situations when I know I have to talk more | 1 | ||
| 18. The effort to produce my voice decreases with rest | 3 |
3.2. No VF group
Mokken scaling on the 167 teachers, who scored lower than 11 on the VFI, resulted in four separate scales. See Table 3 for a full report of items, their means, H’, and the VFI Factor from which they belong. Scale 1 included seven items and the hierarchy ranged from items from the Physical Discomfort factor to the Symptoms of Improvement with Rest. Interestingly, no items from the Tiredness and Avoidance factor appeared to reflect a hierarchical structure in this group. Scale 2, 3, and 4 did not qualify for further Mokken scaling.
Table 3.
Teachers without complaint of vocal fatigue.
| Scale 1 | Rho = 0.916 | HT = 0.517 | Qualified |
|---|---|---|---|
| Question | Mean | H T | Original VFI factor |
| 14. My voice feels sore when I talk more | 1.515 | 0.69 | 2 |
| 15. My throat aches with voice use | 1.323 | 0.73 | 2 |
| 13. I experience throat pain at the end of the day with voice use | 1.300 | 0.74 | 2 |
| 12. I experience pain in the neck at the end of the day with voice use | 0.970 | 0.65 | 2 |
| 16. I experience discomfort in my neck with voice use | 0.892 | 0.70 | 2 |
| 17. My voice feels better after I have rested | 0.545 | 0.43 | 3 |
| 19. The hoarseness of my voice gets better with rest | 0.515 | 0.42 | 3 |
| Scale 2 | Rho = 0.711 | HT = 0.277 | Did Not Qualify |
| Question | Original VFI factor | ||
| 2. My voice feels tired when I talk more | 1.024 | 0.34 | 1 |
| 4. My voice gets hoarse with voice use | 0.856 | 0.39 | 1 |
| 3. I experience increased sense of effort with talking | 0.671 | 0.33 | 1 |
| 11. My voice feels weak after a period of voice use | 0.455 | 0.37 | 1 |
| 10. I find it difficult to project my voice with voice use | 0.443 | 0.34 | 1 |
| 5. It feels like work to use my voice | 0.407 | 0.38 | 1 |
| Scale 3 & 4 | Did Not Qualify due to low number of items | ||
| All Items disqualified | |||
| Question | Original VFI factor | ||
| 1. I don’t feel like talking after a period of voice use | 1 | ||
| 6. I tend to generally limit my talking after a period of voice use | 1 | ||
| 7. I avoid social situations when I have to talk more | 1 | ||
| 8. I feel I cannot talk to my family after a work day | 1 | ||
| 9. It is effortful to produce my voice after a period of voice use | 1 | ||
| 18. The effort to produce my voice decreases with rest | 3 | ||
3.3. Low VF group
Mokken scaling completed on the 397 teachers, who scored between 11 and 24, resulted in three separate scales. See Table 4 for a full report of items, their means, H’, and the VFI Factor from which they belong. Scale 1 revealed a strong hierarchical structure and within this scale, the hierarchy of items of vocal fatigue was similar to the No VF group progressing from items on the Physical Discomfort factor to items of the Symptom Improvement with Rest factor. However, in this group, item ordering from the Physical Discomfort factor differed in hierarchy compared to the No VF group, indicating that the physical experience of vocal fatigue differs between these two groups. Scales 2 and 3 did not qualify for further hierarchical analysis. However, in scale 3, the items related to the Tiredness and Avoidance Factor were strongly related to each other and may explain aspects of avoidance when tired may be experienced by a subset of individuals. This finding might be reflective of the transitional nature of this group.
Table 4.
Teachers with low complaints of vocal fatigue.
| Scale 1 | Rho = 0.903 | HT= 0.567 | Qualified |
|---|---|---|---|
| Question | Mean | H T | Original VFI factor |
| 14. My voice feels sore when I talk more | 1.594 | 0.64 | 2 |
| 13. I experience throat pain at the end of the day with voice use | 1.372 | 0.63 | 2 |
| 15. My throat aches with voice use | 1.352 | 0.64 | 2 |
| 16. I experience discomfort in my neck with voice use | 0.884 | 0.60 | 2 |
| 12. I experience pain in the neck at the end of the day with voice use | 0.874 | 0.60 | 2 |
| 17. My voice feels better after I have rested | 0.456 | 0.41 | 3 |
| 19. The hoarseness of my voice gets better with rest | 0.446 | 0.36 | 3 |
| Scale 2 | Rho = 0.680 | HT= 0.124 | Did Not Qualify due to low HT |
| Question | Mean | H T | Original VFI factor |
| 4. My voice gets hoarse with voice use | 1.896 | 0.32 | 1 |
| 11. My voice feels weak after a period of voice use | 1.625 | 0.41 | 1 |
| 5. It feels like work to use my voice | 1.529 | 0.35 | 1 |
| 10. I find it difficult to project my voice with voice use | 1.479 | 0.31 | 1 |
| 9. It is effortful to produce my voice after a period of voice use | 1.433 | 0.40 | 1 |
| Scale 3 | Rho = 0.547 | HT = 0.540 | Did Not Qualify due to low Rho |
| Question | Mean | H T | Original VFI factor |
| 1.I don’t feel like talking after a period of voice use | 1.811 | 0.35 | 1 |
| 6. I tend to generally limit my talking after a period of voice use | 1.554 | 0.38 | 1 |
| 8. I feel I cannot talk to my family after a workday | 0.945 | 0.33 | 1 |
| Items disqualified from scaling | Original VFI factor | ||
| Question | |||
| 2. My voice feels tired when I talk more | 1 | ||
| 3. I experience increased sense of effort with talking | 1 | ||
| 7. I avoid social situations when I have to talk more | 1 | ||
| 18. The effort to produce my voice decreases with rest | 1 | ||
3.4. High VF group
Mokken scaling completed on 131 teachers, who scored about 24 on the VFI and resulted in three separate scales. See Table 5 for a full report of items, their means, H’, and the VFI Factor to which they belong. Scales 1 and 3 presented with a mild hierarchical structure. Within scale 1, items in the hierarchy were similar to the No VF and Low VF group progressing from items on the Physical Discomfort factor to the Symptom Improvement with Rest factor. In this group, one item from the Symptom Improvement with Rest factor reflected mild to moderate fatigue (#18: “The effort to produce my voice decreases with rest”; mean = 0.916), whereas the No VF and Low VF groups rated this factor as indicating severe fatigue. In this group of individuals who already possess high fatigue, the reordering of this item might simply reflect that improvement of rest is more common. Scale 2 did not qualify for further hierarchical analysis. However, unlike the Low VF group, where scale 3 did not qualify, the high VF group showed a consistent tendency to items related to avoidance of voice use from the Tiredness and Avoidance factor.
Table 5.
Teachers with high complaints of vocal fatigue.
| Scale 1 | Rho = 0.907 | HT = 0.381 | Qualified |
|---|---|---|---|
| Question | Mean | H T | Original VFI factor |
| 14. My voice feels sore when I talk more | 1.389 | 0.65 | 2 |
| 15. My throat aches with voice use | 1.122 | 0.67 | 2 |
| 13. I experience throat pain at the end of the day with voice use | 1.100 | 0.67 | 2 |
| 18. The effort to produce my voice decreases with rest | 0.916 | 0.59 | 3 |
| 16. I experience discomfort in my neck with voice use | 0.733 | 0.59 | 2 |
| 12. I experience pain in the neck at the end of the day with voice use | 0.695 | 0.58 | 2 |
| 17. My voice feels better after I have rested | 0.405 | 0.51 | 3 |
| 19. The hoarseness of my voice gets better with rest | 0.374 | 0.44 | 3 |
| Scale 2 | Rho = .784 | HT = 0.087 | Did Not Qualify due to low HT |
| Question | Mean | H T | Original VFI factor |
| 4. My voice gets hoarse with voice use | 2.901 | 0.46 | 1 |
| 11. My voice feels weak after a period of voice use | 2.710 | 0.51 | 1 |
| 10. I find it difficult to project my voice with voice use | 2.603 | 0.51 | 1 |
| 9. It is effortful to produce my voice after a period of voice use | 2.580 | 0.51 | 1 |
| 5. It feels like work to use my voice | 2.573 | 0.40 | 1 |
| Scale 3 | Rho = 0.907 | HT = 0.328 | Qualified |
| Question | Mean | H T | Original VFI factor |
| 6. I tend to generally limit my talking after a period of voice use | 2.565 | 0.38 | 1 |
| 8. I feel I cannot talk to my family after a workday | 2.031 | 0.41 | 1 |
| 7. I avoid social situations when I know I have to talk more | 1.962 | 0.45 | 1 |
| Items disqualified from scaling | Original VFI factor | ||
| Question | |||
| 1. I don’t feel like talking after a period of voice use | 1 | ||
| 2. My voice feels tired when I talk more | 1 | ||
| 3. I experience increased sense of effort with talking | 1 | ||
4. Discussion
The purpose of the current study was to identify the hierarchy of items on the VFI in professional voice users, specifically teachers, and assess differences in the latent trait of vocal fatigue in this population. Mokken scaling provided insight into the differential experience of vocal fatigue between individuals with voice disorders and professional voice users, specifically, teachers19. Mokken scaling was first applied to responses from all teachers (Total Teachers group), then to teachers categorized as having No Vocal Fatigue (No VF), Low Vocal Fatigue (Low VF), and a High Vocal Fatigue (High VF). Overall, Mokken scaling item hierarchy differed between all teacher responses and those groups based on VF severity. Additionally, differences were captured for the Low and High VF groups showing that items from the Tiredness and Avoidance factor #6, and #8 are salient enough to form an additional scale and may reflect a unique aspect of the experience of fatigue that employs behaviors of intuitively reducing voice use and opportunities to talk.
4.1. Total Teachers Group
Although, Mokken scaling for this group was similar to individuals with voice disorders this was true only for the low end of the hierarchy (mild fatigue). Items related to avoidance in the Tiredness and Avoidance factor did not seem pertinent in the hierarchy of experience of vocal fatigue in teachers19. Additionally, in the total cohort of teachers, the items from the Symptom Improvement with Rest factor comprised the high end of the hierarchy (severe fatigue). This pattern of response may indicate the need for intermittent periods of rest throughout a teaching day for those with VF to maintain efficient voice use.
Given that items sensitive to the same underlying concept of a latent trait are collected in a single scale23, and because responses in the total teachers group were different than those with voice disorders, these results suggest a different experience of fatigue in teachers. Specifically, items in the Symptom Improvement with Rest in teachers are keyed positively only when the items in the low end of the hierarchy (mild fatigue) are keyed positively. This hierarchy suggests avoidance and rest behaviors may be present, but only when fatigue is high. Notably, the scaling of rest items was not evident in individuals with voice disorders19 and may reflect one prominent factor that predisposes one to voice disorder: rest, or lack thereof.
4.2. Separated Groups
Mokken scaling results were markedly different when teachers were categorized into different groups based on their reported VFI scores related to Tiredness and Avoidance. All three groups had one qualifying scale presenting with items from the Physical Discomfort factor at the low end of the hierarchy (mild fatigue) to items from Symptom Improvement with Rest at the high end of the hierarchy (severe fatigue). An exception was observed in responses from the High VF group. The item “The effort to produce my voice decreases with rest” was an additional item that qualified in the low end of the hierarchy (mild fatigue), which was not included in the scaling for No and Low VF group. It could be that, in those with High VF, rest is necessary in the earlier stages of vocal fatigue and that it is not needed or employed in those with No or Low VF.
Additionally, for the No VF group, only seven items in the VFI qualified for further scale analysis, suggesting that the remaining items were not adequately capturing their experience of fatigue. It could be that those without vocal fatigue complaints simply do not experience fatigue, presumably because their vocal systems are more robust, or their vocal behaviors adapt to increased activity. Another possibility is that items on the VFI do not capture milder or transient experiences of fatigue that might be present in this group. Considering additional items that might capture milder symptoms of fatigue will be important in the development of a teacher-specific scale of vocal fatigue. In fact, results from the group of teachers with Low and High VF that incorporated items relating to Tiredness and Avoidance of voice use, might be a starting point. These results suggest that for teachers with reports of fatigue (Low or High), feelings of tiredness or avoidance behaviors may be related to their overall experience of vocal fatigue.
4.3. Response to Sensations of Fatigue
Results from this analysis suggest that vocal fatigue may affect behaviors differently in various populations. Specifically, in this population of teachers, they experience symptoms of physical discomfort at the low end of the hierarchy (mild fatigue) and engage in vocal rest behaviors, observed at the high end of the hierarchy (severe fatigue).Teachers reporting high vocal fatigue rated vocal avoidance behaviors under a second scale of Mokken analysis. Those without vocal fatigue did not clearly engage in avoidance behaviors. This begs the question, are teachers without vocal fatigue not experiencing fatigue, and therefore not engaging in avoidance behaviors; or are those in this group not aware of physical discomfort to the same extent and therefore not behaviorally reacting to fatigue. In other words, are those without reports of fatigue more efficient vocalizers or simply not in tune with the physical sensations of laryngeal discomfort, and thus do not report symptoms? If they are more efficient vocalizers, would they eventually follow the same pattern of fatigue by engaging in avoidance behaviors if they did feel fatigue? Additionally, if they did feel fatigue, could it be that they might not engage in avoidance behaviors for some other reason such as prioritizing communication over feelings of fatigue? One study by Whitling et al. suggested that there might be a group of individuals who do not change vocal behaviors despite feelings of fatigue24. Thus, this group could be heterogenous in nature. These questions could be empirically tested by increasing their vocal activity, above and beyond their normal voice use, to see if they in time, respond similarly to the groups with fatigue. Such individual differences may shed light in to the complicated nature of vocal fatigue and the need to explore factors that modulate the psychological experience of vocal fatigue24. It also suggests that additional items on a teacher-specific measure of vocal fatigue might be necessary to disambiguate between awareness of fatigue or obliviousness to this symptom.
4.4. Sensitivity of the VFI for Teachers
One might expect to see all teachers responding similarly on the VFI, given that many teachers use their voices extensively, and often at the limits of their vocal capacities each day. However, there appear to be distinct subsets of this population. Understanding the factors and separating these subsets would be a major consideration in developing a teacher-specific measure of vocal fatigue. It appears that when the voice is pushed to the upper limits of ability, additional response strategies and characteristics of vocal fatigue might be introduced that are not observed in normal voice users or those with voice disorders. Because the VFI was originally developed on individuals with voice disorders and may not be sensitive in capturing symptoms of extensive voice users, additional validation of VFI may be warranted in professional voice users.
Consideration in the future development of the VFI for teachers must include a re-ordering of the VFI factors. Whereas Factor 1, Tiredness and Avoidance, was the primary factor that determined vocal fatigue in voice patients, it was Factor 2, Physical Discomfort, that determined vocal fatigue in teachers. Reorganizing the VFI to focus on physical symptoms of discomfort to reflect vocal fatigue appears necessary. In addition, a teacher specific VFI might require the inclusion of more items relating to physical discomfort to allow for sensitivity the degree of vocal fatigue at the milder end. Our results also show that items from Factor 3, Symptom Improvement with Rest, presented as a related experience of fatigue in teachers, particularly those with reports of high fatigue. Additional items relating to this construct might prove necessary in determining the presence and amount of fatigue in teachers.
4.5. Clinical Implications
There are a few clinical implications that can be gleaned from this study. First, this study highlights teachers’ differential responses to vocal fatigue, although teachers as a profession have a voice risk factor in common. This is not surprising given that the vocal demand (educational level and environment) and the vocal demand response (individual differences in physiology) of each individual may be different. Specifically, they present with a range of vocal fatigue symptoms from physical discomfort to vocal rest and avoidance. Second, items in our analysis clearly show that response/coping strategies may differ depending on reported symptoms of vocal fatigue. Understanding the nature of varying symptoms and the coping strategies, will help incorporate additional measures to investigate individual differences’ in employing one strategy over another and which voice conservation and treatment approaches are appropriate to delay the onset of vocal fatigue. Finally, the clinician needs to accommodate for the differences in VFI responses in teachers by potentially employing other analysis methods. However, exactly which methods to employ requires additional investigation.
5. Limitations and Future Directions
As in all studies, there are limitations and possible next-step options and implications. One confound to this study is that the teachers were grouped based on the Factor 1 scores on the VFI and responses to Factor 1 may have influenced how the participants responded to other items. Nonetheless, the hierarchy of each group did not reveal any specific hierarchy based on the Factor 1 responses. Another limitation of this study could be the low number of items in some of the Mokken scales, most likely because of the low number of total items in the VFI. Nonetheless, Mokken scaling has been performed on scales with fewer items and demonstrated that even though a scale may be shorter in item numbers, it can reliably possess a hierarchical structure25. This structure was evident across all Mokken scaling analyses reported in our study. This work also provides evidence that the VFI has the necessary items and independent factors to create reliable hierarchical scales for different populations. However, it also shows that different populations might require more specific items pertaining to their unique voice-use characteristics.
Future studies must correlate items of the VFI to other measures of vocal fatigue such as changes in physical performance or perceptual experiences. For example, correlating the VFI in teachers with and without vocal fatigue to measures of phonatory output (e.g., acoustically, aerodynamically, or muscularly) may serve to clarify how these groups differ and refine our knowledge of what constitutes vocal fatigue. Likewise, future investigations into communicative motivation and drive might explain why some groups engage in avoidance behaviors while others do not, thereby clarifying the relationship between avoidance behaviors and vocal fatigue. Finally, future studies are warranted to look in to restructuring of the factors of the VFI and further validation of VFI in teachers.
Conclusion
Mokken scaling provided insight into hierarchical nature of vocal fatigue as captured by the VFI in teachers and the potential differences in VF reports between clinical and non-clinical populations. The Total Teachers group behaved very similar to the voice disorders population, however, differences in hierarchy were seen for teachers when they were categorized in to No VF, Low VF and High VF groups providing further information on the heterogeneous nature of vocal fatigue.
Acknowledgements
Research reported in this publication was partially supported by the National Institute on Deafness and Other Communication Disorders of the National Institutes of Health under award number R01DC012315. The primary author would like to thank her graduate student Kelli Morgan, BS for her assistance with the Mokken RStudio software.
Appendix A
Mokken scaling (Figure 1) initiates with an automated item selection procedure, where all items are analyzed and grouped appropriately to a single scale or multiple scales. Each identified scale is then subjected to individual Mokken Scaling to reflect a hierarchical structure of another observed trait25. For each Mokken scale, the initial approach is to identify if the scale qualifies for further analysis, established by Loevinger’s coefficient HT. Scores of HT = 0.3 to HT = 0.4 denote a valid, but weak Mokken Scale. HT values between 0.4 and 0.5 suggest a moderately strong Mokken Scale, and scores above 0.5 verify a strong Mokken Scale. After assessing the entire scale, Mokken scaling calculates Loevinger’s coefficient H’ for each item, which measures the correlation of each item with all the other items in the scale. Items are then subject to an invariant item ordering (iio) to identify the hierarchy of the items within the scale. The Mokken Scale is then assessed for it’s internal reliability, Rho, that creates groups of items, that reflect an aspect of the underlying trait based on the probabilities of respondents keying item pairs similarly. A value of Rho ≥ 0.7 indicates a reliable scale. The final step of Mokken scaling is to calculate the mean score of items within each Mokken scale, which reflects the number of people who key that item positively. Thus, higher mean scores reflect a basic level of that trait, in this case a lower level of fatigue (mild fatigue); low mean scores reflect a more advanced level of the trait, or higher levels of fatigue (severe fatigue).
If during this process, there are fewer than three items, they cannot be subjected to further analysis. Additionally, if the Loevinger’s coefficient HT is <.3, they do not qualify as a second Mokken Scale.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
6. References
- 1.Nanjundeswaran C, Jacobson BH, Gartner-Schmidt J, Verdolini Abbott K. Vocal Fatigue Index (VFI): Development and Validation. J Voice. Published online 2015. doi: 10.1016/j.jvoice.2014.09.012 [DOI] [PubMed] [Google Scholar]
- 2.Nanjundeswaran C, VanSwearingen J, Abbott KV. Metabolic Mechanisms of Vocal Fatigue. J Voice. Published online 2017. doi: 10.1016/j.jvoice.2016.09.014 [DOI] [PubMed] [Google Scholar]
- 3.Hunter EJ, Cantor-Cutiva LC, van Leer E, et al. Toward a consensus description of vocal effort, vocal load, vocal loading, and vocal fatigue. J Speech, Lang Hear Res Published online 2020. doi: 10.1044/2019_JSLHR-19-00057 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Solomon NP. Vocal fatigue and its relation to vocal hyperfunction. Int J Speech Lang Pathol. Published online 2008. doi: 10.1080/14417040701730990 [DOI] [PubMed] [Google Scholar]
- 5.Boucher VJ, Ayad T. Physiological Attributes of Vocal Fatigue and Their Acoustic Effects: A Synthesis of Findings for a Criterion-Based Prevention of Acquired Voice Disorders. J Voice. Published online 2010. doi: 10.1016/j.jvoice.2008.10.001 [DOI] [PubMed] [Google Scholar]
- 6.Kitch JA, Oates J. The perceptual features of vocal fatigue as self-reported by a group of actors and singers. J Voice. Published online 1994. doi: 10.1016/S0892-1997(05)80291-7 [DOI] [PubMed] [Google Scholar]
- 7.Kostyk BE, Rochet AP. Laryngeal airway resistance in teachers with vocal fatigue: A preliminary study. J Voice. Published online 1998. doi: 10.1016/S0892-1997(98)80019-2 [DOI] [PubMed] [Google Scholar]
- 8.Laukkanen AM, Ilomäki I, Leppänen K, Vilkman E. Acoustic Measures and Self-reports of Vocal Fatigue by Female Teachers. J Voice. Published online 2008. doi: 10.1016/j.jvoice.2006.10.001 [DOI] [PubMed] [Google Scholar]
- 9.Hunter EJ, Tanner K, Smith ME. Gender differences affecting vocal health of women in vocally demanding careers. Logop Phoniatr Vocology. Published online 2011. doi: 10.3109/14015439.2011.587447 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Simberg S, Laine A, Sala E, Rönnemaa AM. Prevalence of voice disorders among future teachers. J Voice. Published online 2000. doi: 10.1016/S0892-1997(00)80030-2 [DOI] [PubMed] [Google Scholar]
- 11.Smith E, Lemke J, Taylor M, Kirchner HL, Hoffman H. Frequency of voice problems among teachers and other occupations. J Voice. Published online 1998. doi: 10.1016/S0892-1997(98)80057-X [DOI] [PubMed] [Google Scholar]
- 12.Cantor Cutiva LC, Burdorf A. Medical Costs and Productivity Costs Related to Voice Symptoms in Colombian Teachers. J Voice. Published online 2015. doi: 10.1016/j.jvoice.2015.01.005 [DOI] [PubMed] [Google Scholar]
- 13.McKinnon SL, Hess CW, Landry RG. Reactions of college students to speech disorders. J Commun Disord. Published online 1986. doi: 10.1016/0021-9924(86)90005-5 [DOI] [PubMed] [Google Scholar]
- 14.Schmidt CP, Andrews ML, McCutcheon JW. An acoustical and perceptual analysis of the vocal behavior of classroom teachers. J Voice. Published online 1998. doi: 10.1016/S0892-1997(98)80052-0 [DOI] [PubMed] [Google Scholar]
- 15.Enoka RM, Duchateau J. Translating fatigue to human performance. Med Sci Sports Exerc. Published online 2016. doi: 10.1249/MSS.0000000000000929 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Gao Y, Dietrich M, Pfeiffer M, And Desouza GN. Classification of sEMG Signals for the Detection of Vocal Fatigue based on VFI Scores. In: Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS. ; 2018. doi: 10.1109/EMBC.2018.8513224 [DOI] [PubMed] [Google Scholar]
- 17.Hunter EJ, Banks RE. Gender Differences in the Reporting of Vocal Fatigue in Teachers as Quantified by the Vocal Fatigue Index. Ann Otol Rhinol Laryngol. Published online 2017. doi: 10.1177/0003489417738788 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Sijtsma K, Molenaar I. Introduction to Nonparametric Item Response Theory.; 2011. doi: 10.4135/9781412984676 [DOI] [Google Scholar]
- 19.Nanjundeswaran C, van Mersbergen M, Morgan K. Restructuring the Vocal Fatigue Index Using Mokken Scaling: Insights Into the Complex Nature of Vocal Fatigue. J Voice. Published online 2019. doi: 10.1016/j.jvoice.2017.09.008 [DOI] [PubMed] [Google Scholar]
- 20.Banks RE, Bottalico P, Hunter EJ. The Effect of Classroom Capacity on Vocal Fatigue as Quantified by the Vocal Fatigue Index. Folia Phoniatr Logop. Published online 2018. doi: 10.1159/000484558 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Cantor-Cutiva LC, Banks RE, Hunter EJ. The Effect of Upper Airway Ailments on Teachers’ Experience of Vocal Fatigue. J Voice. Published online 2020. doi: 10.1016/j.jvoice.2020.05.024 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Banks RE. Understanding Vocal Fatigue and Grade Level Demand as Quantified by the Vocal Fatigue Index (VFI). Published online 2015. https://d.lib.msu.edu/etd/2440
- 23.Van Abswoude AAH, Vermunt JK, Hemker BT, Van Der Ark LA. Mokken scale analysis using hierarchical clustering procedures. Appl Psychol Meas. Published online 2004. doi: 10.1177/0146621604265510 [DOI] [Google Scholar]
- 24.Whitling S, Rydell R, Lyberg Åhlander V. Design of a clinical vocal loading test with long-time measurement of voice. J Voice. Published online 2015. doi: 10.1016/j.jvoice.2014.07.012 [DOI] [PubMed] [Google Scholar]
- 25.Stochl J, Jones PB, Croudace TJ. Mokken scale analysis of mental health and well-being questionnaire item responses: A non-parametric IRT method in empirical research for applied health researchers. BMC Med Res Methodol. Published online 2012. doi: 10.1186/1471-2288-12-74 [DOI] [PMC free article] [PubMed] [Google Scholar]