v", LR ‘ _. . A ~ . : “firfrjj‘ ' ‘ fi-‘VLM-K LIB R A H Y 31:31:23“ 3 a n s tfitfi University w‘ This is to certify that the thesis entitled DIFFERENTIAL VOICE PARAMETERS IN HYPOPITUITARY CHILDREN presented by Connie Supal has been accepted towards fulfillment of the requirements for M. A. Audiology and Ypeech Selence fiegree in —-f Major professor Oscar I. Tosi, Ph.D., D.Sc. Date 30 July 1981 0-7 639 OVERDUE FINES: 25¢ per day per item RETURNING LIBRARY MATERIALS: Place in book return to remove charge fran circulation records DIFFERENTIAL VOICE PARAMETERS IN HYPOPITUITARY CHILDREN BY Connie Supal A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Audiology and Speech Sciences 1981 ABSTRACT DIFFERENTIAL VOICE PARAMETERS IN HYPOPITUITARY CHILDREN By Connie Supal The purpose of this study was to analyze acoustically . the voice parameters of a group of hypopituitary children before and after receiving hormonal treatment 23 a control group. All subjects were divided into age groups of two- year intervals and ranged in age from 6 to 17 years. Recordings were obtained from these subjects and choral spectra produced through the use of a PDP 11/40 minicomputer. The spectra were grouped according to hierarchical and multiple scaling algorithms in order to determine quantitatively the relationship between the spectra of hypopituitary children vs the control group. In addition, the glottal frequency of these children was determined by using a Visipitch device and compared by employing the t-test of statistical analysis. With one exception, the results of the analysis of glottal frequency showed no significant difference between the hypopituitary group v§_the control group. The results of the analysis of choral spectra suggested that the acoustical characteristics of the hypopituitary group after treatment approached the acoustical characteristics of the control group. I wish to dedicate this thesis to my parents who have always given me their love and support throughout every step of my life. ii ACKNOWLEDGEMENT I sincerely wish to express my gratitude to my thesis director, Dr. Oscar I. Tosi, for his ideas, effort, support, and encouragement throughout this project. I also wish to thank the members of the thesis committee, Dr. Leo Deal and Dr. Jerry Higgins, for their review and corrections on the draft on this thesis. iii TABLE OF CONTENTS Page LIST OF TABLES . . . . . . . . . . . . . . . . . iv LIST OF FIGURES . . . . . . . . . . . . . . . . v LIST OF APPENDICES . . . . . . . . , . . . . . . vi CHAPTER I. INTRODUCTION . 1 Purpose of the Study . 2 Definition of the Terms . . . . . . . . 3 Organization of the Thesis . . . . . . . 4 II. REVIEW OF THE LITERATURE . 7 Classification . . . . . . . . . . 7 Physical Symptomatology. . . . . . . . . 8 Treatment . . . . . . . . . . 13 Psychological Factors . . . . . . . . . 14 III. SUBJECTS, EQUIPMENT, PROCEDURES . . . . . 15 Subjects . . . . . . . . . . . . . 15 Phonetic Materials . . . . . . . . . . . 15 Equipment and Procedures . . . . . . . . 16 IV. RESULTS AND DISCUSSION . . . . . . . . . . 26 Results of Hierarchical Grouping . . . . 26 Results of Sammon' 3 Output . . . . . . . 28 Results of Fundamental Frequency . . . . 31 Discussion of Hierarchical Grouping . . . 35 Discussion of Sammon' s Output . . . . . 36 Discussion of Fundamental Frequency . . 37 V. CONCLUSIONS AND RECOMMENDATIONS . . . . . 40 APPENDICES . . . . . . . . . . . . . . . . . . . 42 BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . 45 iv LIST OF TABLES Table Page - 1. Causes of dwarfism . . . . . . . . . . . . . 9 2. Percentage of hierachical clustering . . . . 27 3. Pre-treatment euclidian distances from Sammon's graph . . . . . . . . . . . . . 29 4. Post-treatment euclidian distances from Sammon's graph . . . . . . . . . . . . . 30 5. Summary of glottal frequency . . . . . . . . 32 6. T- test of fundamental frequency by age group . . . . . . . . . . . . 33 7. T- test of fundamental frequency by total subjects . . . . . . . . . 34 LIST OF FIGURES Figure Page 1. Procedure to obtain choral speech . . . . . . 5 2. Example of choral spectrum . . . . . . . . . l9 3. Dendrogram for comparing pre-treatment 12-14 year male group . . . . . . . . . . 21 4. Minimum.span tree of 25 spectra taken from T031 (1979) . . . . . . . . . . . . . 22 5. Sammon's graph output . . . . . . . . . . . . 24 vi LIST OF APPENDICES Appendix Page A. Phonetic Materials . . . . . . . . . . . . 42 B. Months of Treatment . . . . . . . . . . . 43 C. Computer File Names . . . . . . . . . . . 44 vii I. INTRODUCTION Hypopituitarism.is a disorder of the pituitary gland. It is an endocrinological disorder which is one of the ‘major causes of dwarfism.in children. This pituitary deficiency manifests itself gradually through the years from birth until about the age of 21, at which time this disorder can usually be diagnosed by the lack of sexual development of the individual. This disorder may lead to a deficiency in growth and other hormones associated with the pituitary gland, Much of the research in the area of hypopituitarism has only begun in recent years, Clinical diagnosis of this disorder is extremely difficult. One of the major difficulties in diagnosing and treating this disorder is its frequent manifestation concurrently with other hormonal deficiencies. Another problem in diagnosing this disorder is the absence of obvious symptoms until the beginning of adulthood. Many of the researchers in the area of hypopituitarism . have noticed, among other characteristics of this disorder, that the voice may appear high pitched (Horstmann, 1949; Gardiner-Hill, 1937; Martin and Wilkins, 1958). Up to this date, however, no research has been done regarding the acoustic aspects of the voices of hypopituitary individuals. 1 Lack of growth hormone has been noted to affect all the body's structures. In addition, there may be an imbalance of thyroid and sex hormones which are two of the major determinants of voice characteristics (Tanner and Marshall, 1970; Tosi, 1976). Consequently, it is logically hypothesized that hypopituitary individuals may present differential voice qualities. Since the spectrum is the physical correlate of voice quality, this study included a spectral analysis of a group of hypopituitary and "normal" children's voices. Also, glottal frequency was analyzed since pitch was the only voice characteristic mentioned by researchers in this field,as previously noted. An early diagnosis of hypopituitarism is crucial to start proper medical treatment or to avoid unnecessary hormonal injections and painful testing in these children. An acoustical analysis may serve as a complementary diagnostic tool; therefore, research of this possibility could be of great importance to bring more knowledge concerning this hormonal disorder. Purpose of the Study The purpose of the present study was to analyze acoustically the voice parameters of hypopituitary subjects before and after receiving various lengths of hormonal treatment, and to compare these characteristics to a group of control subjects. The following questions were formulated in order to define the scope of this research: 1. Is there a difference in the choral spectra between hypopituitary and control subjects' voices using the TOSI computer program of hierarchical and multiple scaling (Tosi, 1979)? 2. Is there a difference in the choral spectra between hypopituitary subjects' voices before and after receiving hormonal treatment? 3, Is there a difference between the age groups of hypopituitary subjects' choral spectra? 4. Is there a difference between the fundamental frequency of the voices of hypopituitary and control subjects? 5. Is there a difference between the fundamental frequency of the voices of hypopituitary subjects before and after receiving hormonal treatment? DefinitiOn of Terms Definitions of the major terms employed in this study are as follows: Hyp0pituitarism. In this study, hypopituitarism.is defined as a disfunction in the pituitary gland as diagnosed by endocrinologists at the Children‘s Hospital of Michigan Detroit Medical Center. ChOral spectra. Choral spectra as defined by Tosi (1979) are the long-term Fourier transforms of temporal choral speech, as introduced by Tarnoczy (1958). Choral speech is the temporal rearrangement of an individual's speech (Tosi, 1979). This rearrangement consists of dividing a speech.sammle into'n_segments of equal duration and then mixing thesen_segments into one. The resulting "choral" speech segment has the same duration as each of its 3 single components. This output would convey to a listener a perception of 3 persons talking simultaneously (as a "choir") . but, of course, the source is only one person (Figure 1). Hierarchical grouping of spectra. Hierarchical grouping of spectra consists of a computer procedure to rank the the similarities or dissimilarities of spectra as defined by Johnson (1967) and Sammon (1969). 'GlOttal frequency. It is the fundamental frequency of vibration of the vocal folds. In the present study, glottal frequency of subjects' voices uttering the sound /a/ in isolation was measured by using a Visipitch device (Model 6087, Kay Elemetrics Corporation). Organizationgof the Thesis Chapter I has introduced some general concepts on hypopituitarism and, in particular, the lack of research on the acoustical analysis of the voices of these children. Also mentioned was the great need for research in helping to diagnose this disorder in young children. Chapter II consists of a review of the literature including classification, physical symptomatology, treatment and psychological factors associated with this disorder. 3.. 08.1. 2.58% .111.» .33.... scene 3.5.3.. .33... 9°96» .Ammmav Hw0B_sonm :oxmu noomem Hmuono secure an mauvmoonm 1.3.510 00— a “mu 3 u .38. .333 .8 3.9-: .3...— = .3: :8 3:55: 2.2.»:— 593 u u Hoes.“ .239 .- utlooou m Oucm .H shaman . 2.428 us.» 463% .2226 T o: a lull T o: n III: a? e I I; 3 3:: 238.53.. 2. 32...»: 3 3.... - OOOHOA OMOA .3: :29... .333 m 2.2:: a. 3-. 252-2...» Egg-.3 uonn 1|.I1 fie_e_H___eL ~. ow an on km 0 v . too-o. o.-‘ m. coo-o. coo o"-‘h°°~b ‘00“‘a ‘-.°0“ ..m P. . m... .323: 6 Chapter II consists of a review of the literature including classification, physical symptomatology, treatment and psychological factors associated with this disorder. Chapter III discusses the selection of the subjects, phonetic materials, the equipment used in the study, and the procedures used to obtain the data. Chapter IV contains a presentation of the results obtained from hierarchical grouping, Sammon's graph, and glottal frequency analysis including the appropriate charts and tables. A discussion of these results organized in the same manner follows. Chapter V presents a summary of this study and conclusions which can be drawn from the results. In addition, it contains recommendations for future research. II. REVIEW OF THE LITERATURE The phenomenon of dwarfism has been noted throughout history, but it was not until the latter part of the 19th century that the medical community became deeply interested in the problem. Around 1900, Hutchinson and Benda were the first researchers to discover the significance of the pituitary , gland in dwarfism” After this time, there appeared more and more reports of dwarfism as a result of pituitary lesions. Erdheim.and Levi were-among the first to associate this disorder with the pituitary katammn,l949). It wasn‘t until the last two decades, however, that medical experiments began being performed following the discovery of producing pituitary extracts by Evans and Long in 1921 (Hortsmann, 1949). Classification Hypopituitarism is an endocrine disorder, specifically of the anterior pituitary gland. This endocrine disorder has been identified as one of the major causes of dwarfism since hypopituitary individuals are noticeably smaller than others of their sex and age (Wilkens, 1958). Some of the causes of dwarfism.are bone disease, nutritional and metabolic disorders, circulatory and respiratory disorders, constitutional delayed growth and adolescence, "primordial" and genetic, and other types. A complete table of the causes of dwarfism taken from Wilkens (1957) is transcribed in Table 1. Hypopituitarism may be considered to be either idiopathic or organic. Idiopathic usually refers to a biochemical abnormality present from birth, whereas the latter stems from tumors or other gross destructive lesions of the pituitary gland. Organic causes of this disorder are rare. Human growth hormone (HGH) is only one of the hormones of the pituitary. This hormone, however, has no specific target organ. Instead, almost every organ of the body responds to the action of HGH. Human growth hormone is a single polypetide of 190 amino acids (Roth, Glick, Yalow and Berson, 1963). This hormone is responsible for the transportation of amino acids into cells and allows for the incorporation of these amino acids into protein, an ingredient necessary to maintain all living tissues of the human body. In addition, HGH plays a role in DNA replication, RNA transcription and lipid and carbohydrate metabolism. Humans show physiological responses to growth hormone deriVed only from other humans or anthropoid sources. Physical Symptomatology It is unusual that hypopituitarism reveals itself in a family history of growth delay (Hortsmann, 1949). Children with pituitary deficiency usually have a normal birth size Table 1. Causes of Dwarfism II. III. IV. V. VI. VII. . Bone Diseases A. Chondrodystrophy, Hurler‘s syndrome, etc. B. Rickets, all types C. Osteogenesis imperfecta—«with multiple factures D. Diseases and anomalies of spine Nutritional or Metabolic Disorders Celiac disease and cystic fibrosis of pancreas . Chronic renal diseases with acidosis and/or rickets . Hepatic diseases . Nutritional defects and chronic infections . Electrolyte disturbances (hypokalemia, hypercalcemia, etc.) Circulatory or Respiratory Disorders A. Some congenital malformations of the heart B. Extensive chronic pulmonary disease—- with anoxemia Constitutional Delayed Growth and Adolescence Endocrine Disorders A. Hypothyroidism B. Hypopituitarism C. Sexual precocity with early epiphyseal fusion "Primordial" or Genetic Dwarfism A. Familial B. Spordic (intrauterine dwarfing?) C. Syndrome of gonadal aplasia and dwarfism D. Autosomal anomalies (mongolism and other trisomies) Other Types A. Progeria (Gilford—Hutchinson) B. Cockayne's syndrome C. Dwarfism with severe brain defects FIDO UH> Source: Wilkins (1957). 10 and experience a normal growth rate for the first 2-4 years of life. As Wilkins (1957) pointed out, this suggests that growth of very young children may not be dependent on the action of growth hormone; but, instead, growth of the individual occurs as a result of the intrinsic qualities of the tissues. Growth does not stop entirely in these children but tends to show a progressive deviation away from the normal growth pattern (Hortsmann, 1949; Wilkins, 1955; Martin and Wilkins, 1958). Physical examination of children with hypopituitarism typically shows that their body proportions are normal for their heflflu:but not for their chronological age. Therefore, body build serves of little value when trying to classify this disorder. In the literature there is usually a distinction made between the terms growth and development. According to Hortsmann (1949), physiologically speaking, growth is an increase in the size of an organ or a system of organs. Growth indicates a quantitative increase, whereas development refers to qualitative changes in tissues or organs towards greater differentiation. The distinction is made between growth and development because in certain pathological conditions the two do not go together. Hypopituitarism.is very difficult to diagnose. One reason for this difficulty is that this disorder gradually progresses from birth to the adult stages of development. It is indistinguishable from other types of dwarfism in 11 childhood, particularly constitutional growth delay and adolescence (Wilkins, 1957). As was previously mentioned, it is rare that medical experts find obvious signs for the cause of this disorder, such as a tumor or lesion in the pituitary region. In addition, involvement of . the pituitary gland can often mask other symptoms which may be suggestive of an organic lesion. Often times, when there are no obvious explanations for other types of dwarfism, there is a tendency to assign these to a pituitary deficiency (Hortsmann, 1949). Also, differences in the degree of the deficiency of the various pituitary hormones may lead to differences in the clinical picture of these individuals. It is absolutely necessary to diagnose this disorder in childhood in spite of the difficulties involved since it may cause abnormal growth or delayed puberty. Clearly, additional methods are needed to help in the differential diagnosis of this disorder at a time when it may be crucial for effective treatment. It has been suggested that one major hallmark for hypopituitarism is the retardation of skeletal age. This delay in the osseous development manifests itself by a marked proliferation of the cartilage in the epiphyseal ends of long bones (Hortsmann, 1949; Martin and Wilkins, 1958, Gardiner-Hill, 1937; Seckel, 1960). Often times, other pituitary functions, such as gonadotropic, adreno- tropic or thyrotropic deficiencies, may serve as a clue 12 that there is a pituitary disorder. On other occasions there are no other signs that suggest that this disorder is present, and it may not be until the patient grows older that the lack of gonadotropic activity becomes conspicuous with the failure of the individual to mature sexually. Primary and secondary sex characteristics may completely fail to appear. Some indicators of a pituitary disorder in males are small genital organs; lack of facial, pubic, and axillary hair; and a high pitched voice after the time of normal puberty. In women with this disorder, it may be underdeveloped breasts and the absence or scanty flow of menstruation (Hortsmann, 1949; Wilkins, 1957; Henneman, 1960; Gardiner-Hill, 1937). Also, because of the lack of muscular development which usually occurs in normal children in adolescence, hypopituitary individuals may shOW'muscular weakness and fatigability. During the pre-adolescent years, it may be impossible to distinguish hypopituitary disorders from.other causes of dwarfism” In addition, until about the age of 21, it is difficult to determine whether the deficient pituitary activity is temporary or permanent (Wilkins, 1957). If an individual remains sexually immature well into adult years, he is presumed to be deficient in both GH and gonadotropic hormones of the pituitary. After reaching the adult years, however, individuals with hypopituitarism can usually be divided into groups depending on their pattern of sexual development. 13 TREATMENT As was mentioned previously, only human and anthropoid sources of pituitary growth hormone are capable of stimulating growth. The treatment of this disorder varies from individual to individual, depending upon the degree of the deficiency. Growth hormone causes growth in the tissues and increases size without causing excessive osseous or other maturational development. It has been recommended by medical experts (Wilkins, 1957) that intramuscular doses be given from 1 to 5 mg per day. Other experts recommend injections on an interval schedule, such as three times a week instead of every day. The optimal dosage and schedule of treatment have not yet been determined. Often times, growth hormone is also administered with other hormones, depending on the exact nature of the problems Most physicians disapprove of unnecessary use of hormonal therapy at an early age but realize that delayed adolescence and sexual development may cause serious psychological maladjustments. If the onset of puberty in males has not occurred by the age of 14, it is recommended that intramuscular injections of chorionic gonadotropin, and often times small doses of testosterone, be administered. For females, the problem of hormonal therapy is more serious as follicle stimulating hormone (FSH) has been found to cause cystic changes in the ovaries. At the present time, this drug is only available for research purposes. 14 Due to the difficulties in diagnosing this disorder, there also seem to be similar difficulties in treating it with the proper drug therapy. Psychological Factors It seems obvious that hypopituitarism may cause some psychological problems due to the lack of sexual maturity and growth in these children, particularly in adolescence. The psychological handicaps that may develop depend on the individual's capacity to cope with people's reactions to their small size (Wilkins, 1957). It is not known whether psychological problems are a cause of growth failure or whether it is a consequence of this failure. Little research or reports on this aspect have been produced. Thus, from the review of the literature it appears that divergent opinions exist regarding the clinical diagnosis of hypopituitarism” The nature of this disorder'makes it difficult for medical experts to establish an exact definition of hypo- pituitarism and, consequently, it is crucial to produce a careful diagnosis in order to rule out other possible causes of abnormal growth and development. As Hortsmann (1957) said, "no other field of medicine demands such a precise diagnosis as hypopituitarism." III. SUBJECTS, EQUIPMENT, AND PROCEDURES Subjects A total of 26 subjects were utilized in this study. These subjects consisted of 11 male and 3 female children diagnosed by a staff physician at the Children's Hospital of Michigan, Detroit Medical Center, as being hypopituitary. The control group consisted of 12 "normal" children matched by age and sex. All subjects were divided into age groups of two-year intervals, ranging in age from 6-to 17 years. They were native to the United States, Speaking a midwestern _dia1ect of Standard American English. The hypopituitary subjects were patients at the above- mentioned medical center. All patients were diagnosed as having idiopathic hypopituitarism except one, who was diagnosed as having an organic etiology. They received hormonal injections as prescribed by their physician according to standard medical procedures, as discussed in "Treatment." The mean time of treatment for the hypopituitary subjects was 30 months. Phonetic Materials The phonetic materials used for all recordings consisted of the name and age of each subject, isolated vowels /a/, /e/, 15 16 /i/, /o/, /u/, counting from one to ten, and various pre- selected sentences. A complete sample of the phonetic materials used is contained in Appendix A. Equipment and Procedures The voices of the hypopituitary subjects before treatment were recorded by the staff at the Children's Hospital of Michigan using standard recording equipment in a normal reverberant room. The recordings obtained prior to receiving hormonal treatment were labelled as "pre-recordings." The recordings of the hypopituitary subjects after receiving hormonal treatment and recordings of the control group were obtained by the author using a portable Sony Cassette tape recorder (Model #110) and a ferrum.oxide cassette tape. The recordings obtained after receiving hormonal treatment were labelled as "post—recordings." These recordings were also obtained in a normal reverberant room with no extraneous noises. Original cassette recordings were played back using the Sony Cassette tape recorder (Model #110) and dubbed onto 1/4" open reel magnetic tapes (Scotch, AV176, Low Noise) using the Ampex tape recorder (Model AG600). Further, these tapes were physically spliced to include only the subjects' voices, editing the voice of the staff person prompting the children during the original recording. In addition, lengthy pauses and extraneous noises which occurred in the pre- 17 recordings were extracted by segmentation. The final product consisted of 3 onedminute long samples from the voice of each subject/condition (pre, post, control). Each of the subject's sample was separated by a leader tape properly labelled. These tapes will be referred to as master tapes, These master tapes were played back through a Tanberg tape recorder (Model 3300K) and input to an analog-digital (A—D) computer peripheral device in order to be digitalized at a rate of 10,000 readings per second. This rate of sampling was used in order to keep the range of frequencies within 5K Hz (Nyquist, 1928; Tosi, 1979). The digital output from the A—D peripheral was fed into a PDP 11/40 minicomputer (Digital Corporation), and one choral spectra was obtained from.each subject's sample through the program labelled CHORAL (Tosi, 1979), A total of 78 spectra were obtained through this procedure. Each spectra received a file name according to the following protocol: ”MH" referring to the initials for the hospital, "B" or "A" referring to before or after treatment recordings, a subject number, and a spectra number. For example, the file name MHBlZ refers to Children's Hospital of Michigan, pre-treatment recording, subject number one, and spectra number two. A complete list of these file names is contained in Appendix B. These files were subsequently used for the hierarchical grouping completed in the program labelled TOSI3. 18 Each spectrum included an X-axis which portrayed frequencies from approximately 60 to 5K HZ‘With one ordinate every 2.4414 Hz (Dubes, 1979). The vertical axis included energy for each frequency from a range of 0 to 60 dB every 0.5 dB. In other words, each spectrum consisted of approximately 2048 bytes on the horizontal axis and 120 bytes on the vertical axis. Consequently, each spectrum carried a total information of 2.4576 x 105 bytes. Thus, each spectrum could be considered as a 2048+ dimensional vector for the purposes of further computer calculations (Figure 2 ) . The 78 spectra were stored in the RAM (random access memory) disk of the computer to be retrieved for hier- archical grouping and multidimensional scaling. These operations were accomplished by using the software package denominated T0813. This package of software starts with hierarchicalgrouping of spectra as defined by Johnson (1967). Hierarchicalgrouping consists of comparing all combinations of spectra taking two at a time, point by point, and establishing the average difference of the 2048 ordinates compared. Consequently, each combination of any two spectra yielded a spectra difference number. These numbers are arranged in a square matrix of n x n (n indicates the number of spectra to be compared) columns and rows. The computer searches for the smallest of these numbers in that matrix and collapses (groups) the two spectra for 19 \\ I t x x . xx x x xx x xx xx x x xx xx x x x xxx xxx x xxxxxxxx xxx x xxxxxxxxx xxxx xx xxxxxxxxx xxxx xx xxxxxxxxx xxxxxxx xxxxxxxxx xxxxxxxx xxxxxxxxxx xxxxxxxx xxxxxxxxxx xxxxxxxx xxxxxxxxxx xxxxxxxx xxxxxxxxxx xxxxxxxx xxxxxxxxxx xxxxxxxx xxxxxxxxxx ######### ########## ######### # ########## ’ ######### ############ . ######### ############# ## # # ######### ############# ## # # # ####################### ##### # ## # ######################## ##### ### ## # ######################## ##### #### ## # ######################## ##### ####### # ######################## ##### ####### # ######################### ##### ######## # # ########################## # ############## # # I########################## # ############## # ## ############################ ################ # # # ### ############################# ################ # # ##### ############################# ################# ### ## ##### ############################# ################# ### ## ##### ############################# ################## ##### ######## ############################## ################## # ##### ######### ############################## ##################### ##### ######### ############################## ########################### ######### ########################################################## ######### ########################################################## ######### ########################################################## ######### ########################################################## ######### ########################################################## ########## ########################################################## ########## ########################################################### ###################################################################### ###################################################################### ###################################################################### ###################################################################### ###################################################################### ###################################################################### ####################################################################### ####################################################################### ####################################################################### #######################################################t############### ####################################################################### ####################################################################### ####################################################################### ####################################################################### #IOQQQQOOOOOIOIOOII#**#*####****#II*###*#####*###*##############**###**#**## Figure 2. Example of choral spectrum. Horizontal axis is reduced to 76 points in order to accommodate the computer printer and obtain the hard copy. 20 which the difference is that smallest number. Therefore, a new (n - 1) x (n - 1) matrix is obtained; and the process continues until the matrix is reduced to 1 x 1. These groupings are portrayed in a graph called a dendrogram. (Figure 3). Examination of these dendrograms reveals the closeness or similarities/dissimilarities of a group of spectra. Since in the present case it was known which spectra belonged to which individual, examination of the dendrograms revealed the homogeneity or heterogeneity the voices of each group analyzed. Dendrograms include quantitative information concerning the promptness of grouping of spectra pairs. These quantitative data are labelled "birth" and "lifetime" (Johnson, 1967). Next in the software package TOSIB is the multiple scaling and modified.Sammon‘s graph (Sammon, 1969; Tosi, 1979). In the present case, multiple scaling consists of reducing the dimensions of the spectrum vector from 2048 to 2 dimensions. This reduction is essentially calculated by a minimum span tree as introduced by Kruskal (1964) and seen in Figure 4 taken from Tosi (1979). This minimum span tree consists of taking one spectrum and plotting all others by calculating the minimum distances or "stresses" between any two pair of spectra. This plotting is redone several times until the best fitting minimum distance among all spectra is reached. In this process, each spectrum is finally determined by two coordinates, .' “- ‘ i r .L '..' -. .. . .‘ ... .. . L, , ...... toe nu .‘ u o. .\ f" n ._, . ~. .. s ' u’ U .1. ...... u. _- - . . , '..‘ I l .. .i. a... ‘...‘ . .1. ....... .9 ...' u. n‘ . ... . .. s 9 .e'. '...‘ .. A4. . .1 ‘ 4 .. s x.) (u x.‘ .‘ ... . .-'\ ...‘ . , _ ' \« V An I" at. I \n e'. 0 I "' ‘ III. ‘ -. , .-. ..'.. ._ l 'm‘ 'u-e’ 'm' T bu" ' + ... ...’. . 3.3 ,2. 3.. ...' v -...‘ - . ‘m‘ an 'n.‘ ....... ._ . nu - I; no. 1.... .1. \oo' .. . a l -‘ s I - "J 1.) l... ' ‘ .I’l .\l u x I .. . '..) ....... _. i ‘7" . "l f": a t O. 0“ v‘ . u, ..., .\.'. ' ' " R. l s .‘ I ." ..\ ..., . ... .. . ... , . ‘a‘. .- ...... -' we. put. one .. ‘ T ." ... I. do u ... ... \ ... 5.. .. .. a}. m a. u u.’ s" a .. ...' r -. 3‘ u‘ i. x "T * \ ".. "a. _' \ . ,. . I I I ‘0‘ I.- .b- I ' in 4’ . v ‘ . ... ... I .A. A .' Figure 3. 21 I ) Is I.) I) I“; 0 (fl If) (‘I 1’: f l 1‘: {‘1 (:1 a j (‘0 .‘1 (f! l ‘1 .“. 'fi. 1‘. ‘ .“. ,v I l .l .g 4 . ’..-‘ . ' a: U a; U U U .l. l l l .-. l I. .. , . 1 no" .o‘. l 'a._ ‘ ' (:3 \ l .1. x ‘1" k ‘ " (:3 Ix.’ 1 \ J x.) ;‘ l ' n l l l r l I I l n z i u z t I ' I a i 0.. 0 a 0 . . l , | I l I I 1 l I ' I 1 60- D... u 0-. I 0 O... I DI- ! C v I - o I l I l 5 e on no. .0.- u 5.0 u- . I ' a I r t 1 ' v D O 9 0.. 00 I v I t 4 Y 0.. O. I“ m. D... O. 0.. O l l 9 . - . 1 l I I. a... It. ‘O‘ l.— Ii“ 0... I... O... I... .0. ”I. -. .000 A“. I e ' l l .' I I~ I... 00.. a... I... I... too. Cool OI‘ D... O” O... I... O... 0.0 0-. O... ' Dendrogram.for comparin prevtreatment 23 control 12-14 year male group. ISpectra 01—12 correspond to the four hypopituitary subjects, and spectra 13-21 correspond to the three control subjects.) 22 .w was x .cowmamsHv can can SH abuuooom woo mucomouaou unwoa 50mm .A¢mmav HmoH_EOHMu .mx< om muoonnsa owwum>m unmaumouu omnuo>w unmaummuu muoonn3m maouu ow< o..H aouucoo umom o.m umom om mum on mum om .uwmommm ........ H.__»%ofionwmum.Hmuuoam_mo humaanm .m oHan 33 OO OOH O HOH OH OOO O OOO .OO OHO Houuaoo O.O O. O.O O.O O.O OO OOH O OOH OO OOO OO OOO OO OOO ON OOOO OO OOH OO HOH OH OOO O OOO OO OHO Honucoo O.O O. O.O O.O O.O; O OOH OO OOO OO OOO, OO OOO O OOO OM mum OO OOH OO OOH OO OOO OO OOO OH OOO “OOO O.O OOO“ .O.Of wO.O O.O O OOH OO OOO OO OOO OO OOO O OOO ON OOO m 0.0 x O 0.0 x O 0.0 x O 0.0 x O 0.0 O 83258 | A III II- II III OH 2 «HmmIII-Iumqmaum OHO: ”OHO: .OOO OH-OH .OOO OH-OH .OHO OH-O .OHO HH-O .OHO O-O O Ago-Ow emu-hob >283on Hangs—Hm mo 33de ”Emu-H. . o wand“. 34 Aao>oa oufimowmaawOmv Nuaawnmnoumum nowumw>oc purchumu.o.m CNQEHM muoohnam mo Honeucuz ,WII- om.o om OOO OH om OHO OH Houucoo mm OOOO O0.0 OO OOO OH OO OOO OH Houucoo mm OOO OO;O OO OHO OH OO OOO OH OOOO OM OOO m -a-m N. z- 4a.m N . .z .qu, x z aowuwvcoo HOHUCOU Ufiwfimmhul “mom USQEHGMHUI GHQ A3833 Hmuou OOO—V >083on H3553 mo OHO-Damn ummuum. .N oHan 35 The standard deviation of the mean glottal frequencies from age groups was computed to observe its size and any trend correlated with age and maturation as described by Kent (1976). However, this type of variation was not encountered in the present study. Sizes of standard deviation were large in all groups, as predicted by Murray and Singh (1980). Discussion of Hierarchical Groupings Hierarchical grouping is correlated with homogeneity or heterogeneity of subjects' voices. In most sex/age group/ conditions a 100 percent grouping was obtained for the hypopituitary and control subjects. One exception was for the 12-14 year male group. In this group, the control subjects clustered only an average of 84 percent, whereas in other age groups, the control subjects clustered 100 percent. This could possibly be explained by considering the instability of the voices at puberty (Tosi, 1976; Tanner, 1970). For the male hypopituitary subjects in the 12-14 year group, it was observed that pre—treatment subjects clustered 100 percent, whereas the same patients after treatment clustered only 92 percent, thereby approaching the data yielded by the control subjects. This result suggests that the treatment has a possible effect on pubertal development as it relates to voice characteristics. 36 Another exception to the perfect clustering is found in the only female group studied, in which both pre- and post-treatment hypopituitary subjects yielded the same average figure, 78 percent. Further, by exploring clustering of the individual hypopituitary subjects in this group, it was disclosed that the female who received less treatment (15 months) clustered better (100 percent) than those who received treatment for a longer time (39 and 45 months). These subjects clustered only 67 percent. This may suggest an optimal duration of treatment for female hypopituitary subjects. Discussion of Sammon's Output Concerning the Sammon's graph outputs, average disper- sion of subjects' spectra is correlated to the homogeneity of the group considered. A comparison of Tables 3 and 4 reveal that the range of the average dispersion for pre-treatment hypopituitary subjects is larger than that of post-treatment hypopituitary subjects. The quotient of the ratios of the pre— and post-teatment hypopituitary subjects over their distance to the control subjects could be correlated in some way by the degree of improvement of the voice produced by treatment. In all sex and age groups, there was an improvement of the voice char- acteristics quantified from 23 to 97 percent. The smallest percentage was found in the youngest age groups studied 37 (6-8 and 9-11 year males; 39 and 23 percent). This result may suggest that treatment produces less effect on hypo- pituitary subjects before reaching the puberty years. In addition, the 15-17 year male group showed a smaller percentage (61 percent) than the 9-14 year male group, which may suggest an optimal time for treatment. Discussion of Fundamental Frequency Although no statistically significant differences were observed through the use of the t-test, contradicting observations of previous authors regarding the high-pitched voices of hypopituitary children, it should be noted that in most cases of this study there was a decrease in the average fundamental frequency after treatment. However, the decrease was not enough to grant a statistically significant difference. The following factors must be considered to properly interpret the lack of statistical significance: 1. Limitations of the number of subjects. 2. Poor pre-recordings. 3. The existence of only one voice sample per individual for the pre- and post-conditions. 4. No consistency in the manner of utterance of the vowel /a/ in isolation during the pre- and post-recordings. Failure for the standard deviation of the mean fundamental frequency to follow the decreasing rate with age and maturation suggested by Kent (1976) may also be due to the abovementioned factors. 38 The questions formulated on page 3 can now be answered as follows: 1. Is there a difference in the choral spectra between hypopituitary and control subjects' voices using the TOSI program of hierarchical and multiple scaling (Tosi, 1979)? In general, the spectra for the hypopituitary and control groups in this study were highly homogeneous, clustering closely. 2. Is there a difference in the choral spectra between hypopituitary subjects' voices before and after receiving hormonal treatment? In all age groups, the post-treatment hypopituitary spectra came closer to the control group spectra than the pre-treatment spectra. 3. Is there a difference between the age groups of hypopituitary subjects' choral spectra? Yes, the results between the various age groups differed as shown in Tables 3 and 4. The groups that showed the largest percentage increase were the 9-14 year females and the 12-14 year males (92 and 97 percent). The 15-17 year male group also showed an increase (61 percent); however, it was less than the 12-14 year male group and more than the 6-8 year male group. 4. Is there a difference between the fundamental frequency of the voices of hypopituitary and control subjects? The t-test of statistical analysis showed no significant difference except for the pre vs post condition, 12-14 year male group. 39 5. Is there a difference between the fundamental frequency of the voices of hypopituitary subjects before and after receiving hormonal treatment? Again, the t—test of statistical analysis showed no significant difference except for pre vs post, 12-14 years male group. V. CONCLUSIONS AND RECOMMENDATIONS All the acoustical measurements obtained in this study showed an effect of the treatment for these hypopituitary children. The effect was variable in degree, with much less effect concerning the fundamental frequency. The results of this study, therefore, suggest that acoustical analysis of hypopituitary voices can be a valuable and relevant concomitant tool for differential diagnosis. It would be desirable to perfect the acoustical analyses of hypopituitary children by standardizing the procedures for obtaining the samples including prescribed intervals of time for recordings during the medical treatments of these patients. Recommendations for future research should include the use of a larger number of subjects with a very well defined medical diagnosis and treatment, and good recordings pro- duced by competent people using the proper standards. A good research design prior to the performance of a formal experiment is highly desirable. However, because of the lack of consistency in the medical aspects of hypopituitarism and the fact that the subjects are patients, experimenters may encounter great difficulty in adhering to a strict research design. Since the computer algorithms utilized in this study seem to yield relevant information regarding the differences 40 41 in voice qualities of these hypopituitary and control children, it is recommended that future research be specifically aimed at standardizing the interpretations of the algorithms dis- cussed in this study. This could be achieved through large amounts of data. In addition, the construction of a perceptual scale to evaluate hypopituitary patients by qualified listeners (such as a group of experienced speech pathologists) could also be relevant as a screening method for early detection of voice disorders in hypopituitarism. APPENDICES HHH NHO \OGDNO‘UJ-‘UNH 42 APPENDIX A PHONETIC MATERIALS My name is Today's recording date is My birthdate is I mm years old. /a/, /e/, /1/. /o/. /u/ My sister has shiny new shoes. Put the puppy in my lap. I like ice cream. I like to chew chewing gum. This is the first of the month. Count from.one to ten. This is the end of the recording. 43 APPENDIX B MONTHS OF TREATMENT Sub :‘Lect Months treatment J. VUOMN>OWUNUEW wWWOOOWM>£EDEZ S. 20 23 8 34 3O 39 45 62 45 44 21 15 34 9 30 mos . l6 44 APPENDIX C COMPUTER FILE NAMES Hypopituitary Control File Name Subject Pre Post Subject File Name J.S. MHBll MHAll D.F. MHCll 12 12 12 u,; 13 13 13 a3 3.3 MHBZl. MHA21 T.L. MHC21 a: 22 22 22 $74 23 23 23 0 J.M. MHB31 MHA31 32 32 33 33 i“. D. 3 14111341 MHA41 H. Y. MHC-31 “a, 42 42 32 :13 43 43 33 u z: E.W. MHBSl MHASl B.W. MHC.41 °‘ 52 52 42 53 53 43 . D.A. MHB61 MHA61 A.R MHCSl g 62 62 52 :>~.cu 63 63 53 ‘77; P.P MHB71 MHA71. C.S MHC61 '7' g 72 72 62 oxk. 73 73 63 C.B. MHB81 MHA81 L.M. MHC71 82 82 72 83 83 73 C.B. MHB91 MHA91 G.F MHCBl 92 92 82 a; 93 93 83 1.; A.G MHBlOl MHAlOl J.S MHC 91 a) 102 102 92 73'; 103 103 93 (4,: E.C MHBlll MHAlll D.L. MHClOl H 112 112 102 113 113 103 E. C MHBlZl MHA121 122 122 123 123 45 COMPUTER FILE NAMES (CONT.) Hypopituitary Control File Name Subject Pre Post Subject File Name 3' 13.3. 14113131 311111131 T.M. ’MHClll ;. 132 132 112 hug 133 133 113 .H a: L.P. MHBl41 MHA141 M. S. MHC121 J.S 142 142 122 F4 143 143 123 MH=Children's Hospital of Michigan B=pre-treatment recording .A=post-treatment recording C=control subjects BIBLIOGRAPHY BIBLIOGRAPHY Gardiner-Hill, H. Abnormalities of growth hormone and development: The clinical and pathological aspects. British Medical Journal, 1937, 1, 1241-1245. Henneman, P.H., Forbes, A.P., Molaware, M., Dempsey, E.F., and Carroll, E.L. Effects of human growth hormone in man. Journal of Clinical Investigation, 1960, 39, 1233-1238. Horstmann, P. Dwarfism, a clinical investigation with specific reference to the significance of endocrine factors. Acta Endocrinology, 1949, SS, P- 1'23. 25-26, 38-40, 83-109. Johnson, 8. Hierarchical clustering schemes. Psychometrica, 1967, 32, 3, 241-253. Kent, R.D. Anatomical and neuromuscular maturation of the speech mechanism: Evidence from acoustic studies. Journal of Speech and Hearing Research, 1976, 19, 3, 421-4471 Kruskal, J.B. Multidimensional scaling by optimizing goodness of fit to a non-metric hypothesis. Psychometrica, 1964, 29, 1, 1-27. Martin, M.M., and Wilkins, L. Pituitary dwarfism: Diganosis and treatment. Journal Of Clinical Endocrinology, 1958, 18, 679-693. Murray, T., and Singh, S. Multidimensional analysis of male and female voices. JOurnal of the AcoustiCal Society, 1980, 68, 5, 1294-1300. Nyquist, H. Certain factors affecting telegraph speed. Bell system Technology, 1924, 3, 324-338. Raben, M.S, Action of growth hormone in man in Clinical Endocrinology, ed. by E.B. Astwood, (Grune and Stratton, New York, 1960), p. 15-18. Raben, M.S. Growth hormone: Clinical use of HGH. New England JOurnal of MediCine, 1962, 266, 31-35. 46 47 Roth, J., Glick, S.M., Yalow, R.S., and Berson, S.A. Secretion of human growth hormone: Physiologic and experimental modification. Metabolism, 1963 12, 577-583. Sammon, J. A nonlinear mapping for data structure analysis. Transactions on Computers, 1969, 18, 5, 401-409. Seckel, H.P.G. Concepts relating the pituitary growth hormone to somatic growth of the normal child. A.M.A. Journal of Disorders in Childhood, 1960, 99, 349-356. Shepard, T.H., II, Waxman, S., Berstein, N., and Ferrier, P. Human growth hormone, II. Further study of its effect on growth in dwarfism. Journal of Pediatrics, 1960, 57, 363-369. Talbot, N.B., and Sobel, E.H. Endocrine and other factors determining the growth of children in Advances in Pediatrics, (Interscience, New York, 1947), p. 238-250, 286. Tanner, J.M., and Marshall, W. Variations in patterns of pubertal changes in boys. Archs. Dis. Childhood, 1970, 45, 13-23. Tosi, 0., Postan, D., Bianculli, C. Longitudinal study of children's voices at puberty in Proceedings XVI Congress IALP, (Karger, Switzerland, 1976), p. 486-490. Tosi, 0., Voice Identification. Theory and Legal Applications, (University Park Press, Maryland, 1979), p. 92-98. Wilkins, L. Hormonal influences on skeletal growth. Ann. N.Y. Aca. Sci., 1955, 60, 763-766. Wilkins, L- The Diagnosis and Treatment of Endocrine Disorders in Childhood and Adolescence, (Charles C. Thomas, Illinois, 1958, 2nd ed.), p. 160-183. ”7111711717111111111117111111“