Print version ISSN 1516-4446
Rev. Bras. Psiquiatr. vol.32 no.4 São Paulo Dec. 2010 Epub June 25, 2010
Análise vocal de personagens suicidas de filmes de cinema
Elaine Palinkas-SanchesI; Marsal SanchesI,II; Maria Cristina C. FerrariIII; Gisele OliveiraI,IV; Mara BehlauI,IV
ICentro de Estudos da Voz (CEV), São Paulo, SP, Brazi
IIDepartment of Psychiatry, Santa Casa de São Paulo School of Medicine, São Paulo, SP, Brazil
IIIFundação ABC Medical School, Santo André, SP, Brazil
IVUniversidade Federal de São Paulo (UNIFESP), São Paulo, SP, Brazil
OBJECTIVE: The aim of this study was to describe the auditory-perceptive evaluation and the psychodynamic aspects of voice samples among suicidal movie characters.
METHOD: Voice samples of 48 characters (27 male, 21 female), extracted from 36 movies produced between 1968 and 2006, were analyzed. The samples were evaluated through a specific protocol focusing on the auditory-perceptive evaluation (voice quality, resonance, pitch, loudness, modulation, pauses, articulation and rhythm) and the psychodynamic aspects of voice.
RESULTS: 85.5% of the samples exhibited abnormal findings in at least five parameters of the auditory-perceptive analysis, such as breathiness (n = 42; 87.5% of the samples), hoarseness (n = 39; 81.2%) and strain (n = 29; 60.4%), as well as laryngopharingeal resonance (n = 39; 81.2%), either high pitch (n = 14; 29.2%), or decreased loudness (n = 31; 64.6%). With respect to the psychodynamic aspects, dismay was detected in 50% (n = 24) of the samples, hopelessness in 47.9% (n = 23), resignation in 37.5% (n = 18), and sadness in 33.3% (n = 16).
CONCLUSION: Our findings suggest the existence of specific patterns used by actors during the interpretation of suicidal characters. The replication of these findings among real patients may contribute to improvement in the evaluation of potential suicidal patients, as well as the implementation of preventive measures.
Descriptors: Suicide; Voice; Motion pictures as topic; Intention to treat analysis; Evaluation of results of preventive actions.
OBJETIVO: O objetivo do presente estudo foi descrever a análise perceptivo-auditiva e de psicodinâmica vocal de amostras de fala de personagens suicidas em filmes de cinema.
MÉTODO: Foram analisadas amostras de fala de 48 personagens suicidas (27 homens, 21 mulheres), extraídas de 36 filmes produzidos no período de 1968 a 2006. As amostras foram analisadas utilizando-se um protocolo especificamente produzido para o registro das características da voz por meio da análise perceptivo-auditiva (qualidade vocal, ressonância, pitch, loudness, modulação, pausas, articulação e ritmo de fala) e da psicodinâmica vocal.
RESULTADOS: 85,5% das amostras (n = 41) apresentaram alterações em ao menos cinco parâmetros da análise perceptivo-auditiva, com destaque para soprosidade (n = 42; 87,5% das amostras), rugosidade (n = 39; 81,2%) e tensão (n = 29; 60,4%), bem como ressonância laringofaríngea (n = 39; 81,2%), pitch agudo (n = 14; 29,2%) e loudness fraca (n = 31; 64,6%). A análise da psicodinâmica vocal revelou desânimo em 50% (n = 24) das amostras, desesperança em 47,9% (n = 23), resignação em 37,5% (n = 18) e tristeza em 33,3% (n = 16).
CONCLUSÃO: Os achados do presente estudo apontam para padrões vocais característicos utilizados por atores na interpretação de personagens suicidas. Se replicados entre pacientes reais, estes achados podem contribuir para o aperfeiçoamento da avaliação do risco de suicídio e a instauração de medidas preventivas.
Descritores: Suicídio; Voz; Cinemas; Análise de intenção de tratamento; Avaliação de resultado de ações preventivas.
Vocal characteristics are highly valued by psychiatrists during suicide risk assessment.1,2 Nevertheless, the voice is routinely assessed by mental health professionals through a subjective approach, with no description or characterization of any vocal parameters.
Still, some research findings show evidence of specific voice patterns among patients at risk of suicide.3-5 These patterns are mainly described through the analysis of voice samples obtained from recorded suicidal messages, from suicide attempters, and from calls made to the police or suicide hotlines. Virtually all studies on this topic were carried out by physicians and engineers, using automated voice analysis protocols. No studies developed by speech-language pathologists, focusing on the auditory-perceptual evaluation of voice or on its psychodynamic aspects, were identified.
Recently, the study of suicidal movie characters has been considered as a feasible approach for a better understanding of suicide issues. These characters seem to show great similarity to real suicidal patients with respect to their clinical and demographic characteristics.6 Since different vocal strategies are routinely used by actors to characterize patients with mental disorders, the speech analysis of suicidal patients seems to be a feasible approach in the study of suicidal voice patterns.
The identification of such patterns has several potential implications. Among others, it may be useful in the training of health professionals or workers involved in telephone support services, contributing to improving their assessment of suicide risk. The present study has focused on characterization of the speech of suicidal characters from Hollywood movies, giving special emphasis to auditory-perceptual evaluation and the psychodynamic aspects of voice.
Speech samples of suicidal characters from films produced in Hollywood between 1968 and 2006 - were analyzed. All movies were, at the time of the study, on commercial release. Only movies that showed complete suicides were included in the study, and documentaries were excluded. The study was approved by the relevant Institutional Review Board (number 1115/07-CEV IRB).
The movies were analyzed by a speech-language pathologist and a board-certified psychiatrist. After watching the movies, the psychiatrist formulated the probable psychiatric diagnosis of the suicidal character, based on the movie plot and the behavior of the character in the movie. Only the main psychiatric diagnosis was considered. Simultaneously, demographic data on the characters were estimated. The speech-language pathologist, on the other hand, selected voice samples for each character, which represented the speeches that immediately preceded the relevant suicidal acts.
Initially, a pilot study was carried out by three independent speech--language pathologists, who analyzed part of the voice samples and described the variations identified. These professionals did not participate in the previous phase of the study. Based on this preliminary study, the parameters of the analysis were selected, and a specific vocal assessment protocol was created, in order to standardize the auditory-perceptual evaluation and the psychodynamic aspects of voice.
The protocol created (Appendix 1 available online at www.scielo.br/rbp) consisted of eight auditory-perceptual parameters and a question on the psychodynamic aspects of voice. The parameters of the auditory-perceptual evaluation were: voice quality, resonance, pitch, loudness, modulation, pauses, articulation and speech rhythm of speech.7 The evaluator was instructed to fill out the protocol according to the most outstanding findings in each sample, selecting variations within each parameter from a list included in the protocol.
The protocol further included a list of 17 descriptors related to psychodynamic aspects of voice,8 to be chosen by the evaluator. The options were: happiness, affliction, goodness, calm, dismay, discomfort, disarray, hopelessness, indifference, impatience, fear, disgust, resignation, anger, sympathy, surprise and sadness.
A brief description of the parameters for the auditory-perceptual evaluation and the psychodynamic aspects assessment is given in Table 1. In addition, Table 2 describes the main types of voice used in the characterization of voice quality.
Finally, all samples were analyzed by another speech language therapist, with expertise in vocal analysis, who listened to the voice samples from an AVI audio file. The duration of the samples ranged from six to 15 seconds. The evaluator was not provided with the movie scene corresponding to the voice sample. The samples were played as many times as the evaluator considered necessary to complete the analysis, filling out the protocol immediately after concluding the evaluation. The data provided by this evaluator were considered as being the final results of the analysis.
All voice samples were analyzed in their original language (English). All the evaluators who participated in the study were fluent in English.
1. Sample characterization
Forty-eight voice samples, extracted from 36 movies, were analyzed. Nine movies (25.0%) showed more than one suicidal character. With respect to the genres, the films were divided into three categories: drama/war drama (n = 26; 72.2%), thriller/horror (n = 8; 22.2%) and action/adventure/fiction/crime (n = 2; 5.6%).
A slight predomination of males (n = 27; 56.3%) was observed among the suicidal characters. Thirty-one characters were adults (64.6%), 15 were adolescents (31.3%) and two were elderly (4.2%). As far as their marital status was concerned, 31 characters were single (64.6%) and 13 (27.1%) were married. Thirty-three characters (68.8%) lived with someone else at the time of the suicide, and 32 (66.7%) were professionally active. There was great variability in the professions of the characters, with a predominance of students (n = 10; 20.8%), military personnel (n = 5; 10.4%) and housewives (n = 4; 8.3%).
Firearm suicides were the most frequent among the characters (n = 15; 31.3%), followed by poisoning (n = 11; 22.9%), and hanging (n = 7; 14.6%). Other common suicide methods identified in the sample were suffocation (n = 5; 10.4%), defenestration (n = 4; 8.3%), and cutting (n = 3; 6.3%).
1) Psychiatric aspects of the characters
According to the respective movie plots, 27 characters (56.3%) exhibited evidence of planning preceding the suicidal act. Probable psychiatric diagnoses were formulated based on the movie plot and the behavior of the character. Thus, 81.2% (n = 39) of the characters showed findings suggestive of a psychiatric diagnosis. To systematize the diagnosis, DSM-IV-TR9 categories were adopted. Major depressive disorder was the most prevalent diagnosis identified among the characters (n = 17; 35.4%), followed by adjustment disorders and posttraumatic stress disorder (n = 11; 22.3%), personality disorders (n = 5; 10.4%) and schizophrenia (n = 2; 4.2%). Only one character met the diagnostic criteria for substance-related mental disorder alone, although seven characters (14.6%) showed apparent comorbidity with such disorders.
Most of the suicides were apparently mental disorder-related (n = 32; 66.7%). Some deaths were classified as altruistic or patriotic/political suicides (n = 6; 12.5%). Other possible motivations for suicide were found in 20.8% of the characters (n = 10).
2. Statistical analysis
The statistical analysis was performed using the software Statistical Package for Social Sciences (SPSS) version 10.0. Differences between characters with and without psychiatric diagnosis were searched with regard to voice quality and the parameters of the auditory-perceptual evaluation. In addition, the voice samples were successively divided into two groups, based on each of the most frequently found psychodynamic findings. Afterwards, these groups were compared as to the parameters of the auditory-perceptual evaluation. In the case of categorical variables, the chi-square test was used, whereas the Mann-Whitney test was utilized if the variables were ordinal. The significance level was established at 5%.
The auditory-perceptual evaluation showed that findings involving multiple parameters were frequent. Most of the samples showed a minimum of five abnormal parameters (n = 41; 85.5%). Eighteen samples (37.5%) showed abnormalities in seven parameters, and 14.6% (n = 7) had all parameters classified as altered
The proportions of abnormalities on each of the parameters can be seen in Table 3. The most frequent findings were related to voice quality, but most of the characters also showed abnormalities on loudness (n = 44; 91.7%), resonance (n = 42; 87.5%) and pauses (n = 38; 79.2%).
With respect to the voice quality (Table 4), all samples exhibited some abnormality, either occasionally or continuously. The most commonly identified findings were breathiness (n = 42; 87.5%), hoarseness (n = 39; 81.2%), and strain in the voice (n = 29; 60.4%).
Loudness was the second most common parameter identified as abnormal. Thirty-one (64,6%) of the characters showed weak loudness and 11 (22.9%) showed variable loudness, that is, loudness variations (weak versus strong) within the same voice sample. Resonance evaluation revealed that most of the characters utilized a laryngopharingeal adjustment pattern (n = 39; 81.2%) and only two samples showed hypernasal resonance.
Restricted modulation and reduced rhythm of speech were also identified, respectively in 64.6% (n = 31) and 62.5% (n = 30) of the samples. The articulation was imprecise in 66.7% (n = 32) of the characters.
Prolonged pauses were also frequent (n = 31; 64.6%) in the voice samples analyzed, sometimes associated with noisy breathing (n = 10; 20.8%). Finally, most of the voice samples had an adequate pitch. However, where there were abnormalities, the most common was acute pitch (n = 14; 29.2%).
The statistical analysis did not identify any significant differences between characters with and without psychiatric diagnosis with respect to any of the parameters cited above.
1. Psychodynamic aspects of voice
The evaluation of psychodynamic aspects of voice (Table 5) revealed that in 50% (n = 24) of the samples, the voice of the character suggested dismay. Hopelessness was detected in 47.9% of the samples (n = 23), whereas resignation was identified in 37.5% (n = 18) of the samples and sadness in 33.3% (n = 16). Similarly to the findings on the auditory-perceptual evaluation, no statistically significant differences were found between characters with and without a psychiatric diagnosis as to the parameters of the vocal psychodynamic evaluation.
Lastly, possible associations between the psychodynamic aspects of voice and the auditory-perceptual parameters were searched (Table 6). Dismay, which was the most frequently identified psychodynamic aspect, was statistically significantly associated with creaky voice (p = 0.02), breathiness (p = 0.03), and strained voice (p = 0.03), as well as with abnormal modulation (p = 0.002). Despair, on the other hand, was associated with a trembling voice quality (p = 0.001) and abnormal articulation (p = 0.02).
The present study focused on the auditory-perceptual evaluation and the psychodynamic aspects of voice among suicidal characters in movies. The approach adopted allowed us to obtain some original data regarding the voice quality of suicidal characters.
Our findings point to marked abnormalities in the voice quality of suicidal characters, as well as abnormalities in resonance, loudness and pauses. Regarding the psychodynamic vocal analysis, dismay and hopelessness were the emotions most frequently identified in the voice samples. Interestingly, the epidemiological characteristics of the characters were highly similar to the ones of real suicidal patients: single, depressed male characters were the most frequent, and the method of choice was a firearm. These findings are in agreement with data from the literature.6,10
As a public health problem, the suicide has been the object of particular attention.11-13 Even though various preventive measures are often empirically implemented, there is a great deal of controversy regarding the real effectiveness of these measures.14,15 On the other hand, there is in medical literature a consensus on the fact that patients who commit suicide often have some kind of contact with health services in the weeks that precede the suicidal act.16 Therefore, the identification of objective parameters to facilitate the identification of patients at risk represents an area of great interest for psychiatry.
Few previous studies have focused on this issue, perhaps because of the methodological and ethical difficulties of obtaining voice samples of patients who have committed suicide. These difficulties can be overcome through the utilization of voice samples of patients who have survived suicide attempts. However, conclusions obtained through this approach may not necessarily be similar to those for patients who have succeeded in committing suicide, since attempters seem to show different epidemiological and psychopathological issues when compared to successful suicides.10-12 Other studies have carried out vocal analysis of patients at high suicide risk, such as patients with certain diagnoses. These results are subject to the influence of confounding variables, since the vocal abnormalities detected may be related to the psychopathological characteristics of the sample rather than to the suicide risk.
The first study to examine this issue was authored by Ozdas et al., who analyzed voice samples of patients from heterogeneous backgrounds who were at potential risk of suicide.1 Subsequently, the same authors3 published results based on the voice analysis of 10 patients who survived serious, life-threatening suicidal attempts, compared with a control group. The samples were obtained from records of medical consultations held immediately before the attempts, or from recorded suicidal farewell messages. The vocal analysis basically focused on the search for disturbed short-term frequency (jitter), and the results showed a statistically significant higher frequency of jitter among the subjects who had made suicide attempts.
Further, the same group2 carried out a study that compared sample voices of patients diagnosed with unipolar depression (disthymia, single depressive episode or recurrent depressive episodes) compared to healthy controls. A total of 59 patients (22 of them at high suicide risk) and 34 controls were included in the study. Acoustical analysis of fundamental voice frequency (F0), amplitude modulation, formants and Power distributions was carried out. The results showed no F0 differences between groups. However, statistically significant differences were found in regard to the amplitude modulation and the distribution of formants.
Two additional studies by the same group4,5 compared the vocal patterns of 10 depressed patients at risk of suicide, 10 depressed non-suicidal patients and 10 controls. Voice samples were recorded during different daily activities. For the first group, the records preceded suicide attempts (by periods of time ranging from minutes to weeks). Statistically significant differences were found between the suicidal patients and the controls with respect to jitter, and across all groups regarding glottal flow spectrum.
The methodological differences found between the present study and the ones cited above make it difficult to compare those results with ours. All studies of the sample voices of real patients suggest the existence of particular voice patterns among patients at risk of suicide. Despite not evaluating real patients, our study reached similar findings. Recently, the utilization of movies for educational purposes has been the object of interest.17 The realism of some behavior patterns and life situations shown in movies has often allowed conclusions about movie characters to be extended to real patients.
All the cited studies used acoustical analysis of voice samples, a time-demanding procedure that requires highly specific equipment.5 Our approach, on the other hand, allows the rapid identification of voice patterns, enabling the evaluator to take immediate preventive measures regarding the suicide risk.
In our study, the high number of altered parameters in the auditory-perceptual evaluation may reflect the importance given by actors to their vocal performance when they try to personify a suicidal person. Most of the abnormal findings in the auditory-perceptual evaluation were represented by abnormalities in the voice quality. Since these findings were unrelated to the presence or the absence of possible psychiatric diagnosis, we may hypothesize that actors, when playing a suicidal character, tend to valorize only the imminent suicide rather than other psychopathological elements.
Similarly, the high frequency of prolonged pauses in the voice samples may be related to acting techniques used by actors to build up the characters, and may not necessarily be typical of real suicidal patients. The same can be stated with respect to the reduced rhythm of speech, since possible depressive disorders, which could justify that finding due to psychomotor retardation, were found in only 35.4% (n = 17) of the samples.
On the other hand, the influence of mood and psychological issues on voice and speech patterns has been previously described and was superbly discussed in the classic text "The voice of neurosis".18 Our findings point to the predominance of dismay and hopelessness among suicidal characters. Curiously, sadness was detected in only 33.3% (n = 16) of the samples. These findings are consonant with clinical data obtained from real suicidal patients, which indicate that despair represents one of the most important elements in the mental process that results in suicidal behavior.15
The relationship between vocal psychodynamic characteristics and the auditory-perceptual evaluation shows the role of objective voice elements in the emotions translated through the voice. In some cases, the voice quality per se seems to be the key element of the relevant emotions, whereas in others voice quality is strongly associated with other parameters.
When considering two of the most frequent findings in the psychodynamic analysis (dismay and hopelessness), we can reach certain conclusions. There was a statistically significant association between dismay and creaky voice, breathiness and strained voice, as well as with abnormal modulation. In the specialized literature, these voice characteristics are associated with fear, affliction, weakness, impotence and anxiety.7 In our study, hopelessness was associated with a trembling voice and abnormal articulation. These findings, in literature, are associated with excessive sensitivity, fragility, decision-making difficulties, fear and restricted emotional expression.7
Our study has some methodological limitations, which are worth discussing. First, the similarity between suicidal movie characters and real patients may be questioned, since movies tend to glamorize suicidal behaviors.6 However, the clinical and epidemiological characteristics of suicidal characters were in great part similar to the ones classically found in medical literature regarding real suicidal patients.10-12,19,20 Depression was the most frequently identified diagnosis, and the principal method of suicide adopted was the gunshot. Almost 20% of suicidal characters did not meet the criteria for any psychiatric diagnosis, slightly higher than is found in the literature. The low prevalence of substance-related disorders is not in agreement with most literature data, and, further, the psychodynamic analysis is, to some extent, a subjective approach, and some degree of inter-rater variability may be expected when the present protocol is administered by independent evaluators. Still, that analysis was considered highly relevant for our study, given the nature of the data it provides and the theme of the present research.8 Finally, given the strong association between mental disorders and disruptive behavior usually portrayed in Hollywood movies, it was not feasible to obtain a control group (composed of voice samples of non-suicidal characters with psychiatric disorders) that could be matched to the study group according to age, gender and psychiatric diagnosis.
Overall, our results indicate that movie actors use marked changes in voice quality as an acting resource when playing suicidal characters. Breathiness, hoarseness, strained voice quality and decreased loudness laryngopharingeal resonance were the most frequently identified changes in the voice samples. The evaluation of the psychodynamic aspects of voice identified mainly dismay, hopelessness, resignation and sadness in the samples analyzed. The replication of similar findings in real patients may contribute to the development of strategies targeting the identification of patients at high risk of imminent suicide and the implementation of preventive measures.
1. Ozdas A, Shiavi RG, Wilkes DM, Silverman MK, Silverman SE. Acoustical correlates of near-term suicidal risk. JASA. 1999;105(2):1248. [ Links ]
2. France DJ, Shiavi RG, Silverman S, Silverman M, Wilkes DM. Acoustical properties of speech as indicators of depression and suicidal risk. IEEE Trans Biomed Eng. 2000;47(7):829-37. [ Links ]
3. Ozdas A, Shiavi RG, Silverman ES, Silverman MK, Wilkes DM. Analysis of fundamental frequency for near term suicidal risk assessment. IEEE Trans Biomed Eng. 2000;3:1853-58. [ Links ]
4. Ozdas A, Shiavi RG, Silverman SE, Silverman MK, Wilkes DM. IEEE Trans Biomed Eng. 2004;51(9):1530-40. [ Links ]
5. Ozdas A, Shiavi RG, Wilkes DM, Silverman MK, Silverman SE. Methods Inf Med. 2004;43(1):36-8. [ Links ]
6. Ferrari MC, Sanches M, Uchida R. Descriptive analysis of suicidal characters in films. Presented at the 2nd Congress of the Latin America and Caribbean Suicidology Association; 2007 Jun 28-30; Belo Horizonte, Brazil.
7. Behlau M, Madazio G, Feijó D, Pontes P. Avaliação de voz. In: Behlau M, editor. Voz: o livro do especialista. Rio de Janeiro: Revinter; 2001. v.I, p.85-180. [ Links ]
8. Behlau M, Ziemer R. Psicodinâmica vocal. In: Ferreira LP, editor. Trabalhando a voz. São Paulo: Sumus; 1988. p.71-88. [ Links ]
9. American Psychiatric Association. Diagnostic and statistical manual of mental disorders. 4th ed. Revised Version. Washington DC: American Psychiatric Press; 2000. [ Links ]
10. Botega NJ, Rapeli CB. Tentativa de suicídio. In: Botega NJ, editor. Prática psiquiátrica no hospital geral. Porto Alegre: Artmed; 2001. p.365-77. [ Links ]
11. Daly R. Youth suicides on rise question is why? Psychiatr News. 2007;42(19):1-34. [ Links ]
12. Biddle L, Brock A, Brookes ST, Gunnell D. Suicide rates in Young men in England and Wales in the 21st century: time trend study. BMJ. 2008;336(7643):539-42. [ Links ]
13. Nock MK, Borges G, Bromet EJ, Cha CB, Kessler RC, Lee S. Suicide and suicidal behavior. Epidemiol Rev. 2008;30:133-54. [ Links ]
14. Bertolote JM, Fleischmann A, De Leo D, Bolhari J, Botega N, De Silva D, Tran Thi Thanh H, Phillips M, Schlebusch L, Värnik A, Vijayakumar L, Wasserman D. Suicide attempts, plans, and ideation in culturally diverse sites: the WHO SUPRE-MISS community survey. Psychol Med. 2005;35(10):1457-65. [ Links ]
15. Mann JJ, Apter A, Bertolote J, Beautrais A, Currier D, Haas A, Hegerl U, Lonnqvist J, Malone K, Marusic A, Mehlum L, Patton G, Phillips M, Rutz W, Rihmer Z, Schmidtke A, Shaffer D, Silverman M, Takahashi Y, Varnik A, Wasserman D, Yip P, Hendin H. Suicide prevention strategies: a systematic review. JAMA. 2005;294(16):2064-74. [ Links ]
16. Elfrink V, Schlachta-Fairchild L, Szczur M, Chang HF, Young-Weeden E, Rocca M. The Preventing Suicide Network: delivering online tailored resources to those who help others. Stud Health Technol Inform. 2006;122:1001. [ Links ]
17. Rosenstock J. Beyond a beautiful mind: film choices for teaching schizophrenia. Acad Psychiatry. 2003;27(2):117-22. [ Links ]
18. Moses PJ. The voice of neurosis. New York: Grune and Stratton Inc.; 1954. [ Links ]
19. Webster Rudmin F, Ferrada-Noli M, Skolbekken JA. Questions of culture, age and gender in the epidemiology of suicide. Scand J Psychol. 2003;44(4):373-81. [ Links ]
20. Mello-Santos C, Bertolote JM, Wang YP. Epidemiology of suicide in Brazil (1980-2000): characterization of age and gender rates of suicide. Rev Bras Psiquiatr. 2005;27(2):131-4. [ Links ]
Elaine Palinkas Sanches
5210 Weslayan st, B102
7005 Houston, Texas, US
Submitted: August 1, 2009
Accepted: March 10, 2010
This study was carried out at the Centro de Estudos da Voz (CEV), São Paulo, Brazil