Laryngeal biomechanics of the singing voice JAMES A. KOUFMAN, Mb, TERESA A. RADOMSK(, Mil, GHAZI M. JOHARII, MD, GREGORY B, RUSSELL, MS, and DAVID C. PILLSBURY, MA, Winston-Salem, North lina By transnasal fiberoptic laryngoscopy, patients with functional voice disorders often dem- onstrate abnormal laryngeal biomechanics, commonly supraglottic contraction. Appro- Priately, such conditions are sometimes termed muscle tension dysphonias. Singers working at the limits of their voice may also transiently demonstrate comparable tension pattems, However, the biomechanics of normal singing, particularly for different singing styles, have not been previously well characterized. We used fransnasal fiberoptic laryngoscopy to study 100 healthy singers to assess patterns of laryngeal tension during normal singing and to determine whether factors such as sex, occupation, and style of singing influence laryngeal muscle tension. Thirty-nine male and 61 female singers were studied; 48 were professional singers, and 52 were amateurs. Examinations of study subjects performing standardized and nonstandardized singing tasks were recorded on a laser disk and subsequently analyzed in @ frame-by-frame fashion by a blinded otolaryngologist. Each vocal task was graded for muscle tension by previously established criteria, and objective muscle tension scores were computed. The muscle tension score was expressed as a percentage of frames for each task with one of the laryngeal muscle tension patterns shown. The lowest muscle tension scores were seen in female professional singers, and the highest muscle tension scores were seen in amateur female singers. Male singers (professional and amateur) had intermediate muscle tension scores. Classical singers had lower muscle tension scores than nonclassical singers, with the lowest muscle tension scores being seen in those singing choral music. (41%), art song (47%), and opera (57%), and the highest being seen in those singing jazz/pop (65%), musical theater (74%), bluegrass/country and western (86%), and rock/gospel (94%), Analyzed also were the influences of vocal nodules, prior vocal training, number of Performance and practice hours per week, warm-up before singing, race, smoking, and ‘alcohol consumption. (Otolaryngal Head Neck Surg 1996;116:527 37.) AXinos: everyone has seen ultra-sow-motion footage of runners, golfers, and other athletes, in which the positions of the major joints and muscles are sequen- tially analyzed in a frame-by-frame fashion, Biome: chanical analysis of athletic performance has become relatively familiar and is accomplished by linking a From the Contr For Voice Disorders of Wake Foren University (on. Koufman and Johari and Ms. Radomski: the Departments of Otolaryngology (Des. Koufinan and Jobst and Mr. Pillsbury) and Public Health Sciences (Mr. Russell), Bowman Gray School of Medicine, Wake Forest University Medial Center, and the De partment of Speech and Heating (Mr Pillsbury), Nonth Catolina Baptist Hospital, Presented at the Annual Meeting of the American Academy of tolaryngotozy-Head and Nock Surgery, New Orleans, La, Sept 17-20, 199 Reprint requests: James A, Kouta, MD. Center For Voice Disorders ‘of Wake Forest University. Department of Otolaryngology, Bow. ‘man Gray Sehool of Medicine, Medical Center Blvd, Winston x, NC 27157-1034, Copytight © 1996 by the American Academy of Otolaryngology Head and Neck Surgery Foundation, Ine 0194-5999/96/55.0040 24771913, videosystem to a computer system so that movement can be evaluated critically, With this technology, pat- tems can be identified that are at one extreme optimally efficient and at the other extreme maladaptive or even abusive Through the use of similar methods, laryngeal bio mechanics were studied at the Center for Voice Disor: dlers of Wake Forest U he purpose of these studies was determination of the normal laryngeal bio- mechanics of healthy singers and analysis some of the factors (¢.g., sex, vocal training, singing style) that may influence laryngeal biomechanics, METHODS AND MATERIAL collaboration between a laryngologist (1. A. K.) and @ vocal pedagogue (T. A. R.) was established for this, study, and a custom computer-controlled laser disk recording system was used so that videodata could be retrieved and analyzed with precision, ‘Through media advertising, volunteer singers were solicited for study. To be included, singers had to be both medically and vocally healthy (i., they could not be 527528 KOUFMAN et al. Otolaryngology ~ Head ond Neck surgery December 1996 Table 1. Laryngeal biomechanics of the singing voice: Breakdown of the study population by sex. professional/amateur status, and singing style Mote Female Singing style Pro ‘Amateur Pro Amateur Tot —— sw Opera a 3 7 6 19 Choral 3 6 3 6 18 Artsong 2 8 5 6 16 Musical theater 2 4 5 6 7 Popularfiazz 1 7 9 BluegrassiCaW 4 1 2 7 Rock/gospel 3 1 3 7 Barbershop = 4 3 7 Total 0 3 100 having any vocal problems at the time of enrollment in the study), Specifically excluded were subjects who had had any serious medical illness within 60 days of the study or who were known to have vocal cord lesions (€.g., nodules, polyps, granulomas). Subjects were en- rolled without regard to se; or age, although an attempt was made to solicit and include a balance of professional and amateur singers and singers of different styles of mus Before inclusion, each subject completed a detailed history form. This information formed the preexamina- tion database. Included was demographic information such as age, sex, race, occupation, professional status (full-time vocal professional, parttime vocal profes: sional, voice student, other amateur singer), number of yeuts of formal voice training, average number of prac- tice and performance hours per week, und style of sing ing. In addition, subjects were asked whether they were currently taking voice lessons, whether they warmed up before singing, whether they had musele tension when singing, and whether they exercised regulatly. The past medical history included the number of episodes of laryngitis during the past 5 years; any other history of vocal problems, specifically vocal cord le- sions, tobacco usage, alcohol usage, medications, and reflux symptoms; and any medical condition for which the subject saw a doctor regularly Demographic Description of the ‘Study Population AIL 100 subjects enrolled in the study were healthy, and none had any known voice problem. Studied were 39 male and 61 female singers, Forty-eight were pro- fessional singers, and 52 were amateur. Ninety were white, and 10 were black. Their mean age was 37 13.5 years (mean + SD), with a range of 18 to 69 years. The primary singing style of the 100 subjects, broken down by sex and professional status, is shown in Table 1 Instrumentation ‘The following equipment was used: a Pentax FNL- 13S, 3.7-mm external diameter fiberscope (Pentax Pre: cision Corp., Orangeburg, NY.), a Stor xenon light source (Storz Instrument Co., St. Louis, Mo,), a Pani sonic model AG7300 super VHS videorecorder (Pana sonic Co., Secaucus, N.J.), a Toshiba model 1K-C40 camera (Toshiba America, Inc., New York, N.Y.) & Pioneer model TQ-3031F optical disk recorder (Tech- nical Industries, Aanta, Ga.), and a Zenith microcom- puter (Zenith Electronics Co., Chicago, Ml.) ‘The optical disk-recording system was under foot- switch control by computer intesface for hands-off op- eration by the examiner. Frame addresses were stored and labeled in the computer program for subsequent retrieval and analysis. (The advantages of the optical disk videosystem for this study were its high resolution of the videoimage, precision random access, and sta- bility of freeze-frame fields.) Transnasal Fiberoptic Laryngoscopy Technique Transnasal fiberoptic laryngoscopy (TFL) was per- formed by one of us (J. A. K.) using the following technique: the procedure was explained, the subject's nose was topically anesthetized by spray with 2% co- caine and 1% ephedrine, and the subject was positioned in the sitting position with the neck in a modified “sniffing position,” with the arms on the lap, chin up, land shoulders relaxed. The fiherscope was introduced between the middle and inferior turbinates and posi- tioned in the hypopharynx so that it gave an_unob- structed view of the endolarynx. Vocal Tasks Each subject was asked to perform the following singing tasks: (1) the last line of the “Star-spangled Banner” (“O’er the land of the free, and the home of the brave") on key (provided by a keyboard) and (2) aOtolaryngology ~ Head and Neck Surgery Volume 115 Numbere KOUFMAN ef al, 529 Fig. 1. Features of MIP | are incroased tension and stitness of the vocal cords with an open glottie chink. selection of music in the subject’s own style. (Through: ‘out this manuscript the former will be referred to as the standard task, and the latter as the own-style task.) ‘The “Star-spangled Banner” line was selected as the standard singing task because itcan be sung in 6 seconds or less (and matches our 6-second recording window), it is familiar (0 every subject, it is relatively difficult 10 sing, and it includes the passaggio (the wansition from the middle to the high registers of the voice). For the own-style singing task, each subject was instructed to select something that he or she considered to be tech: nically difficult that also included the passaggio. Before TPL examination, each of the tasks was prac ticed by the subject under the supervision of the exam- um. We did this to evaluate the suitability of the sd by the subject for the own-style task and to encourage each subject to sing loudly, in full voice For actual videorecording (data collection), each task was performed at full voice und sufficiently quickly, so that the complete task was recorded within the 6-second window. In some instances, patients repeated tasks t0 censure that they were completely recorded Statistical Methods and Analysis Jata analysis the study population was broken down into sex. status, fessional status, and singing style ing styles were combined because the number of subjects in the pop, jazz, bluegrass, country and western (C&W), rock, and gospel groups was small These were combined into three paired groups— 4 popfjazz group, a bluegrasvC&W group, and a rock/gospel group. This gave a total of eight sing. ing-style groups for the purpose of statistical analy sis. (We believe that it is appropriate to pair these professional sex and pro Some of the sing groupings for that purpose because the paired styles demonstrated similar laryngeal biomechan. ies; that is, they had similar muscle tension [MT] seores,) Frequencies and descriptive statistics were generated for demographic data. Pearson's product-moment cor relations and analysis of variance were used 1 assess univariate relationships between MT scores and inde pendent variables. Multiple linear reg ward climination was used 10 examine multivariate530 KOUFMAN et al. Otolaryngology - Head and Neck Surgery December 1996 Fig. 2. The feature of MIP Ils partial or complete approximation of the false vocal cords (ie side-to-side compression, relations. Fisher's exact test was used to test for differ= ences in categorie variables. Both a univariate data analysis and & multivariate model for the outcomes of interest were analyzed in an effort to assess the impact of each variable individually, as well as in conjunction with the other variables observed. (The multivariate model takes every variable {nto account; however, it should be recognized that it is ‘not unusual for univariately significant variables in such models not to be significant in a multivariate model,) Laryngeal MT Patterns The operation of normal and abnormal laryngeal biomechanics as assessed by TPL examination of nor mal subjects and patients with voice disorders has been reported.'* Koufman and Blalock" have previously characterized four discrete patterns of laryngeal MT in patients with voice disorders, The basic patterns are termed muscle tension patterns (MTPs): type Lis glottal and types I-IV are supraglottal MIP I, MTP I (Fig. 1) was first described by Morrison et al, in 1986 as “an open glottic chink. (cristic finding of MTP is a gap between the vocal fold free edges during phonation, with a conspicuous pos- terior gap between the vocal processes. MTP [ has also been called a laryngeal isometric pattern because it presumably is caused by the simultaneous and inappro- priate overcontraction of the posterior cricoarytenoid muscles in opposition to the lateral cricoarytenoid muscles during phonation.’ (MTP 1 is common in pa- tients who have vocal nodules.) MTP II. MTP II (Fig. 2) is characterized by compres sion (medialward pressure) of the false vocal cords, In its mildest form, only the anterior portions of the false vocal cords are compressed, almost approximating or actually touching together. (When the false vocal cords, come into contact and are used for phonation, a severe, pitch-locked dysphonia is produced, False-cord voice, sometimes termed plica ventricularis, is a relatively The charac common functional voice disorder.)Qtolaryngology — Hood and Neck surgery Volume 115” Number 6 KOUFMAN et al, 831 Ure Of MIP Ills partial anteroposterior contaction of the larynx with arylenolds approaching the petiole, MIP It is the most common MIP MIP iil. MTP III (Fig. 3) is characterized by partial anteroposterior contraction of the larynx during phona: tion. Typically, the arytenoid are pulled forward towant the petiole ofthe epigtottis, obscuring the posterior one half to two thirds ofthe vocal folds. (MTP TIL is scen in pat so-called Bogari-Bacall syndvome.! Also itis routinely seen when singers sing the lowest note of the vocal range.) MIP IV. Extreme anteroposterior contraction (ie ccomplcte sphincterlike closure of the larynx), in which ints who speak with a very low-pitched voice, the the arytenoids actually contact and squeeze against the petiole, is characteristic of MTP IV (Fig, 4). (This pattern is uncommon. but it is seen in severe functional voice disorders and in patients with spasmodic dys- phonia.) Grading and Calculation of the MT Scores An otolaryngolo, ist (G. M. 1.) blinded to the demo: aphic and clinical information, reviewed and graded each vocal task frame-by-frame, Frames in which the vocal cords were not in contact (¢.g., during inspiration, before initiation of the task, and after completion of the task) were not scored. In addition, any obscured frames, were not scored culated as follow Thus the percent MT’ score was eal: Number 0 MIP x 100 Total aumber of scored frames For the task Specific criteria were established for the acwal classification and geading of each MTP: MTP I, the n the cords must be Hront to back, and create an angle at the : MTP IT, compression by the false cords must be 50% (oF g gap betwe it must open anterior eater) compared with the base- line measurement between the mid false vocal cords MTP IIL, anteroposterior contraction must be at least one third of the baseline measurement be- tween the arytenoids and the petiole; and MTP IV, fone or both arytenoids must actually touch the pe tiote, and no part of the true vocal cords should be Visible RESULTS For all subjects, the mean MT score for the standard task (the last fine of the “Star-spangled Bunner"”) was 40.41% + 45.9% (mean + SD), The mean MT score for the own-style task was 60.9% 4 41.9%,532 KOUFMAN et al Otolaryngology ~ Head ond Neck Surgery December 1996 Fig. 4. Features of MIP IV are complete anteroposterior contraction with the arytenoid being pulled forward against he petiole and foreshortening and thickening of the false vocal cords {and the aryepigiottic folds, with no patt of the true vocal cords being visualized Table 2. Mean muscle tension scores by singing style (In rank order from least to most) Singing syle No. of subjects Mr score +50 Vai Choral 8 ante fet song, 6 46984141 5742 Opera 9 577242 0.282 Barbersho 7 B13437t 0.2630 Populatliaz 8 6514429 o.1a72 Musical theater 7 354986 1924 Bluegrass/CAW 7 8564377 O.o19t Rock/gospel 7 9402 146 0.0040 Calculated by comparison of the MT Influence of Singing Style (MT Scores for the ‘Own-style Task) The mean MT scores for the different singing styles 41.1% 4 44.8% 46.9% + 41,19; opera, 57.7% + 42.7%; bar bershop. 61.3% + 37.19; popular/jazz, 65.1% + 42.9% musical 73.5% + 38.6%; bluegrasIC&W, 85.6% + 37.7%: and rock/gospel, 94.09% + 14.6%, The differences among some of the groups were statistically significant (Table 2) (in rank order) were as follows: chor theater 2ch group with that of the choca group, which had the lowest MT sco Influence of Sex and Professional Status For male subjects, the mean MT score for the standard task was 40.1% 45.1%, and for female subjects the mean was 40.6% 46.7% (p= 0.8379), The mean MT scores for male and female subjects for the own-style task were also. virtually identical (61.9% + 39.6% and 60.3% + 43.7%, respectively) (p= 0.9999) For professional subjects, the mean MT seore for the standard task was 64.4% + 42.6%Otolaryngology ~ Head and Neck Surgery Volume 115" Number e KOUFMAN etal, 533 Table 3. Mean muscle tension scores for the standard singing task and the own-style task “Sendard task MT ‘Own-siyle task MT Group/subset score # SD (%) Vou sc010 25D 0) p Volue Professional 2ageai2 oozatt 466 4422 0.0220 worn ‘Amateur men 3182423 sag rata Professional men 497 £477 7A7 #342 ‘Amateur women 5634470 735#415 Allmen 40.1245. 08379 194396 9999 ‘ll women, 4062487 603437 All professionals 6447426 1108 3392450 o2iat All amateurs 46.4 £463 srarata Tota! group 4044459, eogratg “Denote the aiference among te groups. {Denotes stastical sgnifeance (p< 0.05, Table 4. Surnmary of variables correlating (or not correlating) with the MT scores for singing tasks (univariate statistical analysis) Variables that strongly correlated with MT scores (p< 0.05) Classical singing Classical singers had lower MT scores than non Classical singers, (The classical designation i cluded art song, opera, and choral styles; all of the ether siyies were considered nonetassical ) Vocal nodules Singers who Rad vocal nadules had higher MT ‘scores than those who did not, Race \White singers had lower MIT scores than black singers, Variables that weakly correlated with MT scores = 0.05 to 0.15) Vocal taining Singers who had had vocal training had lower IMT scores than those who had not Professional status Professional singers had lower MT scores than, farmateur singers. warm-up Singers who warmed up before singing had lower IMT scores than those who dic not Practice hours MT seoees of singers increased as the number of practice hours per wook increased Variables that did not conelate with MT scores (p> 0.15) sex Sex did not influence MT sect. Tobacco use Tobacco use aid nat influence MT score ‘Alcohol use Alcofol uso did not influence MT score Prior vocal problems Singers with history of vocal problems did not have highor MT scores than those who had not hhad such problems, Performance hours For bath professional and amatour singers. the umber of performance hours per week did Not influence MT scoro. Tension when singing Singers who reported that they fot tension in the eck when they sang did not have higher MIT Scores than those who did not have such tension, Voice lessons Singers who wore taking voice fessone atthe time ‘inclusion in the study cid not have lower MT scores than those who were not Regular exercise Singers who exercised regularly did not have lower IMT scares than those wha did nol 46.4% + 46.3%; however, the difference was not statis- professionals. had lower MT scores (33.9% + tically significant (p=0.1108), For professional sub- 45.0%) than amateurs (57.2% + 41.4%), but again, Jects, the results for the own-style task were oppo- the difference was not statistically _sigoi site from those observed for the standard task— — (p =0.2131)534 KOUFMAN of al. Interaction between Sex and Professional Status When sex and professional status were treated as independent variables (as in the preceding section), neither appeared to correlate significantly with MT score: however, when the study population was divided into four discrete cohorts by sex and professional status, the MT score for each group was significantly different from that for the other groups. For the standard task professional female subjects had the lowest MT scores (24,396 41.2%), followed by amateur male subjects (31.8% + 42.3%), professional male subjects (49.7% + 47.7%), and amateur female Subjects (56.3% 4: 47.0%). The differences among these groups were statistically significant (p= 0.0281), For the own-style task, professional female subjects still had the lowest scores (46.6% + 42.2%). followed by ama- teur male subjects (50.9% + 41.29), amateur female subjects (73.5% +: 41.5%), and professional male sub- jects (74.7% + 34.2%). The differences among these roUups were statistically significant (p= 0.0220). These data are summarized in Table 3 Variables That Strongly Correlated with MT Scores Classical styles of singing (choral, art song, and opera) had lower MTT scores than nonclassical singing styles. Subjects who were found t have vocal nodules, her MT scores than those who did not have thern, and black subjects had higher MT scores than white data are described below and summa: subjects, The rized in Table 4. Classical singing styles. When the classical styles (choral, art song, and opera) were combined into one group (n= 53) and compared with a group combining the other five singing style groups (n= 47), the MT scores for the classical group were lower for both tasks. For the standard task the mean MT score for the classical group was 28.1% #41.2%, compared with 54.2% + 47.4% for the nonclassical group (p = 0.0040). For the own-style task the mean MT score for the classical group was 45.9% + 42.6%, compared with 74.9% + 36.8% for the nonclassical group (p = 0.0013), Vocal nodules. In the course of performing this, study, 15 subjects were found to have asymptomatic vocal nodules on laryngeal examination. Seven subjects, had mature (organized/keratinizing) nodules, and eight had immature mucosal swelling-type nodules. Both MT scores were higher for subjects with nodules. For the standard task the mean M ule group was 57.9% + 49.3%, compared with 37.3% +44.9% for the group nodules (p = 0.0626). For the own-style task, the mean MT score of the nodule group was 93.3% + 15.1%, compared P score of the nod Otolaryngology - Head and Neck Surgery December 1996 with 55.2% + 42.6% for the group without nodules (p = 0.0082), Race. The singing styles (own style) of the 10 black subjects were opera (n=4), gospel (n= 2), popliazz (n= 2), art song (n= 1), and bluegrass (n= 1). Com: pared with the white subjects, the black subjects had significantly higher MT scores. For the standard task, black subjects had a mean MT 88,0% + 31.6%, compared with 35.2% + 44.2% for white subjects (p = 0.0004). For the own: style task, the black subjects had a mean MY score of 99.1% 42.8%, compared with 56.3% + 41.9% for the white subjects (p= 0.0018). Variables with Borderline Correlation with MT Scores Lower MT scores we training and in those who warmed up before singing Compared with the amateurs, professional singers had lower MT scores for the own-style task but higher MT scores for the standard task, In addition, there was & correlation between the number of practice hours per week and the MT score—as the number of practice hours increased so did the MT score. These data are described below and summarized in Table 4. Vocal training. Seventy-three subjects had at least | year of vocal training (range, 1 to 23 years). Vocal training was treated as a continuous variable, and it appeared to correlate with lower MT (p= 0.0838), Professional status. Horty-cight subjects were pro: fessional singers, and 52 were amateur. For the own- style task, the professional singers had lower MT scores (33.9% + 45.0%) than amateur singers (57.2% 4 41.4%) (p= 0.2131). For the standard task, the MT scores were higher for the professional subjects (64.4% + 42.6%) than for the subjects (46.4% + 46.3%) (p = 0.1108). Warm-up before singing. Seventy-four of the 100 subjects reported that they warmed up before singing. For the standard task those ned up had a mean MT score of 39.6% + 46.1%, com- pared with 48.5% #47.2% for those who did not (p= 0.1392). For the own-style task, those who warmed up had a mean MT score of 57.8% + 41.9%, com pared with 73.3% + 38.9% for those who did not (p =0.1250), Number of practice hours per week. The number of practice hours per week was (reated as a continuous variable, and it appeared to correlate with the MT scores for the own-style task—as the number of pructice hours. increased, so did the own-style MT score (p = 0.0838). For the standard task, however, there was no correlation 0.8985). seen in those with vocal amateurOtolaryngology — Head ana Neck Surgery Volume 115" Number e A number of other variables were investigated, but hone appeaved {0 correlate with the MT scores (Table 4). Correlation Between MT Scores for the Standard Task and for the Own-style Task On cursory examination of the data, it may appear that the MT scores on the standard and own-style tasks fe independent. However, when the MT scores of the two tasks are compared, the correlation is highly sta- fistically significant (p= 0.0001), Muscle Tension Pattern Data {In addition to collection of the total MT score data for ‘each vocal task, each of the MTP patterns was scored, and the data were analyzed. MTP I was by far the most common MT pattern, For the standard task, for all subjects the mean MTP values were as follows: MTP 1, 3.9% £18.19: MTP. TI, 6.0% £22.2%; MTP II, 0.4% 442,656; and MTP IV, 2.89% 4 15.0%, For the own-style task the mean MTP values for all subjects were as Follows: MTP 1, 3.4% 415.6%: MTP IL 10.8% + 28,24; MTPHU, 46.5% + 41.49%; and MTP E 4.0% £ 14.2%, ‘The individual MTP patterns were analyzed to detect differences among professionals and amateurs. men and ‘women. None of the differences among the groups was statistically significant, except MITP IT was more com- mon in male singers (10.1% + 27.3%) than in female singers (3.4% + 18.05) (p = 0.0365) and MTP UL was more commion in amateur singers (38.54% + 45.5%) than in professional singers (21.656 + 37.8%) (p = 0.0261), Mutivariate Mode! ‘A multivariate model used to investigate the interac tion among the variables showed a significant correla tion between MTP score and sex, race, classical singing style, professional status, practice hours per week, al cohol use, an vocal nodules, These data are shown in ‘Table 5. (The implications and interpretation of this model are addressed in the Discussion section.) DISCUSSION ‘The normal biomechanical configuration for effort- less phonation is that the vocal cords approximate “lke two hands clapping on a hinge”; vocal cord closure is achieved along the length ofthe vocal cords, without the participation of supraglottic structures. (There is neither front-to-back foreshortening nor side-to-side compres sion of the larynx, and the aryepiglotic folds remain thin and rounded.) Effortless phonation, however, is not always what singers exhibit when they are singing. It must be emphasized tha the data presented here do not suggest that singing styles associated with high MTT cores should be avoided, nor do they suggest that a high KOUFMAN ef al, 535: MT score is in any way abnormal or pathologic. We do, however, make one assumption that underlies this ‘work (and that is supported by clinical experience): high MT scores imply high levels of laryngeal work, whercas low MT scores imply relative vocallaryngeal eff- iency. It would be premature to speculate about the specific implications of MT scores except to say that a very high MT score represents the laryngeal mechanism working al maximal effort, In some singers this may be necessity and acceptable to achieve a desired vocal result, and in others it may be a precursor to a voice disorder. In other words, in a patient with a voice disorder, observed MT patterns appear to have pathologie significance: how- ever, in an asymptomatic singer, «high MT seore may simply indicate an elevated level of laryngeal work for a specific vocal task and nothing more eis on the basis of our experience with patients hhaving vocal nodules that we can extrapolate the pos sible significance of prolonged and excessive high lev- els of laryngeal tension/work. In this series 1 patients hhad axymptomatic voeal nodules Gncidentally found at the time of examination), Although this group (by Univariate statistical analysis) had higher MT scores than subjects without nodutes, its important to note that by the multivariate model, nodules were not associated with increased MT seores for those subjects’ own-style tasks (Table 5), A possible explanation for this finding is exemplilied by one of the study subjects, a professional elub singer. Her MT score on the standard task was very high, but when she sang her own style, her MT score was 0.0%—she sang “Misty” using a soft, breathy glottal attack, In professional singers, with few exceptions we found 4 matching of the singer with his or her own singing style. In the multivariate model, professionals had rela- tively high MT scores for the standard task, but not for the own-style task. Vocal training and warming up before singing, forexample, may be useful in inereasing laryngeal efficiency; however, the multivariate model demonstrated that the MT score is result of the complex interactions of many variables, Some of the variables that influence the MTT score interact in ways that may be interpreted. There was no difference between the MT scores of male and female subjects, orbetween professionals and amateurs, and yet the “sex/professional status” cohorts were sigaificantly different from each other (Table 3). The explanation is simply that the variables that influence MT’ score inter act in complex ways, For example, « higher proportion of professional women sang classical styles, warmed up before singing, and had vocal training than professional men and amateur women.536 KOUFMAN ot al Otolaryngology - Head and Neck Sur ‘December 1996 Table 5. Results with a multivariate model: Variables correlating (or not correlating) with the muscle tension scores for the standard singing task and the own-siyle task Varibie Standard ‘own syle Sox a.0120" 0.1820 Race ‘.0006" ors Classical singing 0.0012" 0048 Professional status 0.0012" 0.4610 Perfomance hoursiweok 0.5057 0.4610 Prior vocal training 0.6035, o.7705 Practice hoursiweek 6224 0887 Prior vocal problems 9.9673 0.7502 Warm-up belore singing 281 0.1097 Tension when singing 08658 0.9923 Ongoing voice lessons 0.9837 0.7984 Regular exercise oss23 06066 Tobacco use 0.3260, 0.9686 Alcohol use 0405" 0.382 Vocal nadules 07st 0.6958 Shown are the p values of a backward stepwise regression; soe the Statistical Methods and Results sections for analysis and dation of forms “Denates statistical significance (o (105) {Denotes naar sigreance (p< 010). Variables That Favorably Influence Laryngeal Biomechanics Generally. some variables appear to have the effect of decreasing the MT score or protecting against exces laryngeal tension, and these include: (1) vocal training, (2) singing classical styles of music, and (3) warming up betore singing. Intuitively, each of these variables should exert a score-lowering influence. Vocal training. It is presumed, particularly by vocal pedagogues, that the purposes of vocal training are to teach good vocal technique (vocal hygiene), to teach an appropriate repertoire, and to increase the efficiency of the laryngeal mechanism. Although the definition of good vocal technique may vary from teacher to teacher, xz0od breath support, good posture, and relaxation of the vocal tract and its supporting musculature are among its professed elements. In other words, reduction of excess laryngeal tension is one of the primary goals of training. In particular, ‘many voice teachers focus on avoiding laryngeal cleva- tion during singing. This makes sense because excessive tension of the suprahyoid musculature is associated with supraglottic contraction (MTPs I through IV). In this study 79% of the subjects had some vocal training, and the MT data appear to affirm the proposition that vocal training does to some degree increase laryngeal/vocal efficiency (i., decrease MT scores), Classical versus nonclassical singing styles. ‘The complexity of the model makes it difficult to determine the relative contribution of every variable, and it appears. classical singing may be associated with other MT score-reducing variables. However, when a classical singer sang in a nonclassical style, such as musical theater or gospel, we observed that the MT scores increased dramatically. (As part of a related study, some subjects did sing more than one style, but herein we are jjust reporting the results of the subjects’ primary singing style.) There is no question, for example, that “belting” singing styles require more work than classical styles, and it appears that the ranking of singing styles by MT. score in Table 2 probably represents real differences in laryngeal tension/work. (We should point out that belting is sometimes used in the classical [operatic] repertoire; however, by comparison with musical theater and other nonclassical styles, itis infrequent.) Even when all variables are considered, certain sing: ing styles demand increased laryngeal work: increased MI scores appear to correlate with certain singing styles, One subject, an amateur bluegrass singer, stated, “If it don’t hurt, it ain’t bluegrass.” (He had a very high MT score for his own-style task.) Clearly, there may be social and cultural factors that affect how certain styles ‘of music are sung. To the trained ear, what you sce (on ‘TRL) is what you hear, Warm-up before singing. Most athletes warm up before performing, and vocal athletes appear to be no different. Its likely that vocal warm-up limbers up the laryngeal musculature much like stretching does for other muscles. This is evidenced by the data ofthe study. Variables That Unfavorably Influence Laryngeal Biomechanics Several variables appear to correlate with the MT score and have the unfavorable effect of increasing MT scores, or of generating increased laryngeal tensio ‘These include vocal nodules, race, and an increased ‘umber of practice hours per week.‘Otolaryngology — Head and Neck Surgery ‘Volume 115" Number 8 Virtually by definition, vocal nodules are associated with aberrant laryngeal biomechanies, namely, higher than normal vocal fold tension/stfiness (i.e. increased laryngeal muscle tension). Among voice clinicians, itis presumed that nodules are the result of vocal abuse (too loud, too long), an excessively hard glottal attack, poor breath support, and/or excessive vocal cord ten- sion/stffness, In addition, nodules appear to be self-perpetuating. ‘The mass of the nodules themselves prevents complete closure of the cords along their free edges. As a con- sequence, in an attempt to achieve closure for certain vocal tasks, people with nodules will automatically increase the tension of the vocal cords and increase th subglottic pressure to drive the stiffened cords. Thus increased laryngeal work can ereate a vicious eycle in which additional vocal cord trauma and swelling are a consequence of compensatory biomechanical adjust- ‘ments for the nodules Ici important to emphasize that the study subjects with vocal nodules were asymptomatic (ie. they had no vocal symptoms). In addition, more often than not vocal nodules in singers result from problems with the speak- ing, and not the singing, voice. Consequently, voi clinicians working with nodule patients must establish the cause of nodules speaking voice, singing style, or both—if therapy is to be effective. When asymptomatic nodules are found, they should be addressed by a similar approach, namely, establi ment of why they are there by survey of laryngeal biomechanics during TFL with an assortment of vocal tasks. Observation of how MTP patterns change with different tasks is essential to the understanding of a- ryngeal biomechanics and to the determination of a person's vocal strengths and weaknesses from a bio- mechanical point of view. (This can be likened t0 a tennis coach examining each of his or her pupils” different strokes.) ‘This concept is basie—laryngeal biomechanics (MIPS) both reveal and forecast the vocal output, Some people with asymptomatic nodules are healthy, and others potentially have voice disorders. The difference between the two groups may be related to whether the nodules are “wanted.” In any case, vocal nodules ap- pear to be associated with inereased MT scores, In summary, we have attempted to characterize and quantify laryngeal biomechanies so that some of the variables that may affect it could be studied. In view of the obvious complexity of these interactions, we do not believe that the influence of each individual variable can KOUFMAN ef al, 537 be fully described. Nevertheless, the methods used for this study appear to be useful for analysis of laryngeal biomechanics, CONCLUSIONS 1, MT scores for male and female singers were not significantly different. 2. MT scores for professional and amateur singers were not significantly different. 3. Singers of classical styles of music, such as art song, opera, and choral music, appeared to have lower MT scores than singers of nonclassical styles. 4. The mean MT scores for eight singing styles (in increasing order of muscle tension) were as follows: choral music, 41%; art song, 47%; op- cra, 38%; barbershop, 61%; popularjjazz, 65%: musical theater, 74%; bluegrass/eountry and western, 87%; and rock/gospel, 94%, 5. The most common biomechanical alteration dut- ing singing was contraction of the supraglottis in an anteroposterior direction (ie., MTP IID. 6. Singers who had had formal vocal training ap- peared to have significantly lower MTT scores than those who had not. 7. Singers who warmed up before singing appeared to have significantly lower MT scores than those who did not. 8, Singers with vocal nodules. appeared to have Significantly higher MT scores than those who did not. 9. Black singers appeared to have significantly higher MT scores than white singers. 10. Alcohol use, tobacco use, prior vocal problems, and regular exercise did not appear to influence the MT scores. REFERENCES. 1. Koufouan JA, Blalock PD. Vocal fa professional voice user: Bogart-Baca 1988;98:493-8, 2. Koufman JA, Blalock PD. Functional voice disorders, Otolaryngol (Cin North Am 1991;24:1059-73, 3. Koufman JA. Gasroesophageal refux and yoice disorders. In Rubin JS, Sataloft RT, Korovin GS, Gould W), eds, Diagnosis reatment of voice disorders. New York: Izaku-Shoin, 1995:161 1 4. Morrison MD, Nichol H, Rammage LA. Disgnosic criteria in funetiona dysphonia. Laryngoscope 1986:94:1-8 5, Yanagisawa E, Estill, Mambrino Le al, Supraglttic cots tions to pitch using. Videoendoscopie study with spectroanaysis, ‘Aan Otol Rhino Laryngol 1991100: 19-20, an dysphonia nthe syndrome, Laryngoscope