臺灣唇顎裂已修補學童中文母音的聲學分析

全文

(1)國⽴立⾼高雄⼤大學西洋語⽂文學系碩⼠士論⽂文. 臺灣唇顎裂已修補學童中⽂文母⾳音的聲學分析 Acoustic Analysis of Mandarin Vowels of Children in Taiwan with Repaired Cleft Lip and Palate. 研究⽣生：許巧瑛撰指導教授：賴怡秀博⼠士. 中華民國⼀一○七年六⽉月.

(2)

(3) ACKNOWLEDGEMMENTS . The completion of this dissertation would not be possible without the help of. many people.. I am extremely grateful to those who have assisted and supported me. throughout this study. First of all, I would like to express my sincerest gratitude to my advisor, Dr. Yi-hsiu Lai, who spent plenty of her time guiding me with her professional advice and valuable suggestions to make this research successful. Dr. Lai, having expertise in phonology, phonetics, psycholinguistics, syntax, as well as SLA, has been my mentor for three years.. Her inspiring instruction has not only enlightened but also cultivated. me in experimental phonetics.. Her scholarly guidance and constant encouragement. during my study years at National University of Kaohsiung are most essential to the accomplishment of this thesis. The very deep gratitude I feel toward the committee members of my proposal hearing and oral defense, Dr. Raung-fu Chung and Dr. Song-lan Wang. Dr. Chung encouraged me to consider the issue from multiple perspectives. His profound knowledge in phonetics helped me construct a more organized analysis. Dr. Wang gave me valuable and constructive suggestions.. I am especially grateful to his. careful guidance on organization of this thesis. In addition, many thanks are given to all the participants in this study.. Without. their recordings and cooperation, the completion of this research would not have been made possible.. My heart-felt gratitude also extends to Jim and his wife, Donna.. Both of them gave strong support in retrieving the sounds and analyzing the data. I would also like to thank Lydia, who is the assistance in Dr. Chung’s language lab. She assisted me in coping with the technical problems in the computer. Last but not the least, my everlasting appreciation is devoted to my family and my dearest husband.. Without their love, patience, and wholehearted support, I. would not have had great courage to pursue the master degree. appreciatively and affectionately dedicated to them.. i . This thesis is.

(4) 臺灣唇顎裂已修補學童中⽂文母⾳音的聲學分析指導教授：賴怡秀博⼠士國⽴立⾼高雄⼤大學西洋語⽂文學系. 學⽣生：許巧瑛國⽴立⾼高雄⼤大學西洋語⽂文學系碩⼠士班. 摘要. 本論⽂文研究旨在探討臺灣已接受過唇顎裂修補⼿手術的學童與無唇顎裂的學童在中⽂文母⾳音發⾳音表現上之異同。︒｡結果顯⽰示上述兩群學童於中⽂文母⾳音的發⾳音位置有所差異，然⽽而在母⾳音空間之統計分析未達顯著差異。︒｡總計有六位已接受過唇顎裂修補⼿手術的學童、︑､與⼋八位無唇顎裂的學童作為本論⽂文的研究對象。︒｡在與⼗十四位受試者逐⼀一完成相同的中⽂文母⾳音語料採集程序後，以 Praat 語⾳音分析軟體擷取受試者母⾳音的 F1 值與 F2 值，並以統計分析檢視兩群體的平均值表現。︒｡研究結果顯⽰示，⾸首先已接受過唇顎裂修補⼿手術的學童所產⽣生的母⾳音 F1 平均值明顯⾼高於無唇顎裂的學童，顯⽰示已接受過唇顎裂修補⼿手術的學童發⾳音⾆舌位較無唇顎裂的學童為低；其次，已接受過唇顎裂修補⼿手術的學童所產⽣生的母⾳音 F2 平均值也與無唇顎裂的學童有顯著性差異，統計分析顯⽰示已接受過唇顎裂修補⼿手術的學童的母⾳音發⾳音前後位置較無唇顎裂的學童為前。︒｡最後，以兩組 F1 值與 F2 值所組成的母⾳音空間來做⽐比較，則未達到顯著性差異⽔水準。︒｡. 關鍵字：唇顎裂、︑､中⽂文母⾳音、︑､F1、︑､F2、︑､母⾳音空間. . ii .

(5) Acoustic Analysis of Mandarin Vowels of Children in Taiwan with Repaired Cleft Lip and Palate Advisor: Dr. Lai, Yi-hsiu Department of Western Languages and Literature National University of Kaohsiung. Student: Hsu, Chiao-ying Department of Western Languages and Literature National University of Kaohsiung . ABSTRACT. This study investigates the Mandarin vowel production of Taiwanese children with and without cleft lip and palate.. The production of vowels of. children with repaired cleft lip and palate differs from those of children without cleft lip and palate; however, the vowel space of these children is similar to that of the children without cleft lip and palate. Taiwan were recruited for this study.. Fourteen children from southern. These children were divided into two. groups both of which took part in the recording session. The first group was comprised of six children with repaired cleft lip and palate while the second group was comprised of eight children without cleft lip and palate.. Statistical analysis. of the recording data revealed major findings relating to the three research questions.. Firstly, children with repaired cleft lip and palate have higher F1. values, which means their tongue is positioned lower in the mouth compared to those of the children without cleft lip and palate.. The second finding is that the. F2 values of the children with repaired cleft lip and palate are significantly different from those of the children without cleft lip and palate. iii . They tend to.

(6) pronounce Mandarin vowels with their tongues further forward in the mouth. The third finding is that there is no significant difference in the vowel space of the two groups. Key words: cleft lip and palate, Mandarin vowels, F1, F2, vowel space . iv .

(7) TABLE OF CONTENTS ACKNOWLEDGEMMENTS ......................................................i CHINESE ABSTRACT ............................................................. ii ENGLISH ABSTRACT ............................................................ iii TABLE OF CONTENTS ............................................................ v LIST OF TABLES .....................................................................ix LIST OF FIGURES ..................................................................... x CHAPTER ONE. INTRODUCTION ....................................... 1. 1.1 Background of the Study ................................................................................. 1 1.2 Motivation and Significance of the Study ....................................................... 2 1.3 Research Questions of the Study ..................................................................... 4 1.4 Definition of Terms ......................................................................................... 5. CHAPTER TWO. LITERATURE REVIEW ........................... 7. 2.1 Current Research on Cleft Lip and Palate ....................................................... 7 2.2 The International Protocol Treatment and the Speech Therapy of Children with Cleft Lip and Palate ............................................................................... 12 2.2.1 Current International Protocol Treatment of Children with Cleft Lip and Palate........................................................................................... 12 2.2.2 The Speech Therapy of children with Cleft Lip and Palate ............... 13 2.3 Studies of Mandarin Vowels ......................................................................... 13 . v .

(8) 2.3.1 Articulation and Classification of Vowels ......................................... 13 2.3.2 The First Formant Frequency and the Second Formant Frequency of the Mandarin Vowels ........................................................................ 14 2.3.3 Vowel Space ....................................................................................... 17 2.4 Summary........................................................................................................ 17. CHAPTER THREE. METHOD .............................................. 19. 3.0 Introduction ................................................................................................... 19 3.1 Participants .................................................................................................... 20 3.1.1 GWC ................................................................................................... 20 3.1.2 GWOC ................................................................................................ 21 3.2 Instrument ...................................................................................................... 21 3.2.1 Instruments for Data Collection ......................................................... 22 3.2.2 Instruments for Data Analysis ............................................................ 25 3.3 Procedure ....................................................................................................... 26 3.3.1 Before Recording................................................................................ 27 3.3.2 During Recording ............................................................................... 27 3.3.3 After Recording .................................................................................. 28 3.4 Data Analysis ................................................................................................. 28. CHAPTER FOUR. FINDINGS AND DISCUSSIONS .......... 29. 4.0 Introduction ................................................................................................... 29 4.1 Vowel Positions in Group of the Children with Repaired Cleft Lip and Palate vi .

(9) and Group of the Children without Cleft Lip and Palate .............................. 30 4.1.1 Vowel positions of GWC and GWOC in Formant Setting 4 ............. 30 4.1.2 Vowel positions of GWC and GWOC in Formant Setting 5 ............. 38 4.1.3 Vowel Positions of GWC and GWOC in Formant Setting Judged by the Researcher ................................................................................... 45 4.2 Vowel Space in Group of the Children with Repaired Cleft Lip and Palate and Group of the Children without Cleft Lip and Palate .............................. 52 4.2.1 Vowel Space of the Two Groups in Formant Setting 4...................... 53 4.2.2 Vowel Space of the Two Groups in Formant Setting 5...................... 54 4.2.3 Vowel Space of the Two Groups in Formant Setting Judged by the Researcher ......................................................................................... 56 4.3 Discussion of Vowel Positions and Vowel Space in Group of Children with Repaired Cleft Lip and Palate and Group of Children without Cleft Lip and Palate ............................................................................................................. 59. CHAPTER FIVE. CONCLUSION ......................................... 62. 5.1 Summary of Main Findings ........................................................................... 62 5.2 Implications ................................................................................................... 64 5.3 Suggestions for Further Study ....................................................................... 64. REFERENCES .......................................................................... 66 APPENDIX A ........................................................................... 71 APPENDIX B ........................................................................... 75 vii .

(10) APPENDIX C ........................................................................... 78 APPENDIX D ........................................................................... 79 APPENDIX E ............................................................................ 82 APPENDIX F ............................................................................ 83 APPENDIX G ........................................................................... 84. . viii .

(11) LIST OF TABLES. Table 2.1 The average value of F1 and F2 in different reports (鄭靜宜 2004; Huang, 1996, Jeng, 2000; Hsiao, 2013) ................................................................ 16 Table 4.1 Means and standard deviations of F1 and F2 of the five vowels of GWOC and GWC in formant setting 4 .................................................................. 31 Table 4.2. ANOVA analysis of F1 of GWC and GWOC in formant setting 4 ......... 36. Table 4.3. ANOVA analysis of F2 of GWC and GWOC in formant setting 4 ......... 37. Table 4.4 Means and standard deviations of F1 and F2 of the five vowels of GWC and GWOC in formant setting 5 ............................................................... 38 Table 4.5. ANOVA analysis of F1 of GWC and GWOC in formant setting 5 ......... 43. Table 4.6. ANOVA analysis of F2 of GWC and GWOC in formant setting 5 ......... 44. Table 4.7 Means and standard deviations of F1 and F2 of the five vowels of GWC and GWOC in formant setting judged by the researcher.......................... 46 Table 4.8 ANOVA analysis of F1 of GWOC and GWC in formant setting judged by the researcher ............................................................................................ 51 Table 4.9 ANOVA analysis of F2 of GWOC and GWC in formant setting judged by the researcher ............................................................................................ 51 Table 4.10 ANOVA analysis of the vowel space of GWOC and GWC in formant setting 4..................................................................................................... 54 Table 4.11 ANOVA analysis of the vowel space of GWOC and GWC in formant setting 5..................................................................................................... 56 Table 4.12 ANOVA analysis of the vowel space of GWOC and GWC in formant setting judged by the researcher ............................................................... 58. . ix .

(12) LIST OF FIGURES Figure 3.1 The recording shelter ................................................................................. 24 Figure 4.1 Vowel /a/ distribution in formant setting 4 ................................................ 33 Figure 4.2 Vowel /e/ distribution in formant setting 4 ................................................ 33 Figure 4.3 Vowel /i/ distribution in formant setting 4 ................................................. 34 Figure 4.4 Vowel /o/ distribution in formant setting 4 ................................................ 34 Figure 4.5 Vowel /u/ distribution in formant setting 4 ................................................ 35 Figure 4.6 Vowel /a/ distribution in formant setting 5 ................................................ 40 Figure 4.7 Vowel /e/ distribution in formant setting 5 ................................................ 41 Figure 4.8 Vowel /i/ distribution in formant setting 5 ................................................. 41 Figure 4.9 Vowel /o/ distribution in formant setting 5 ................................................ 42 Figure 4.10 Vowel /u/ distribution in formant setting 5 .............................................. 42 Figure 4.11 Vowel /a/ distribution in formant setting judged by the researcher ......... 48 Figure 4.14 Vowel /o/ distribution in formant setting judged by the researcher ........ 49 Figure 4.15 Vowel /u/ distribution in formant setting judged by the researcher ......... 50 Figure 4.16 Means of the data fetched in formant setting 4 in vowel space and vowel positions of GWOC and GWC ............................................................... 53 Figure 4.17 Means of the data fetched in formant setting 5 in vowel space of GWOC and GWC ................................................................................................ 55 Figure 4.18 Means of the data fetched in formant setting judged by the researcher in vowel space and vowel positions of GWOC and GWC......................... 57. . x .

(13) CHAPTER ONE. INTRODUCTION. This study is an acoustic examination of Mandarin vowel production by children with and without repaired cleft lip and palate.. The first section of this chapter. introduces the theoretical background, followed by the motivation and significance of the study.. The three research questions are introduced in the third section.. The. final section provides the definition of the terms that are used in the discussion and analysis of this study.. 1.1 Background of the Study. The world is filled with the sound of voices. Walking down a busy street, sitting in a restaurant or waiting to see a doctor, the many and varied sounds of people’s voices fill the air.. Operating independently of the sense of sight, human. beings can perceive, process and understand these myriad sounds. For those with normal hearing ability it is easy to know if a voice belongs to an adult or a child, a woman or a man, or if it is expressing sorrow or joy.. The sound of a person’s. voice can even reveal their country of origin and their mother tongue.. But how is. this possible? How can a voice communicate details such as age, gender,. 1 .

(14) emotion, and native language? How can we know if the speaker is a man, a woman or a child? Vowels are the sounds that allow the human ear to distinguish these many variations. A vowel is produced through the opening of the vocal tract. Like musical instruments, the length of vocal tract influences the production of vowels. There are two characteristic vocal tract pitches, called “formants,” which differentiate vowel sounds (Ladefoged, 2011, p.23).. “The lower pitch” is referred to as “the first. formant and the higher one the second formant” (Ladefoged, 2011, p. 23, italics in original).. Research focusing on vowel production has identified a relationship. between the length of the vocal tract and vowel production.. For example, 鄭靜宜. (2011) claimed that because female speakers have shorter vocal tracts than males, female vowel production results in higher formant frequencies.. Consequently,. acoustic analysis of vowels in the first formant (F1) and the second formant (F2) has been the mainstream in this field.. 1.2 Motivation and Significance of the Study. According to the Noordhoff Craniofacial Foundation (https://www.nncf.org/encyclopedia), about one in six hundred babies in Taiwan are born with a cleft lip and palate.. Thanks to advanced medical technology,. these patients are able to receive corrective surgery as infants, in addition to receiving excellent post-surgery care. As a result, children in Taiwan born with. 2 .

(15) cleft lip and palate will show little outward sign of having been born with this birth defect. Some of them, however, may still struggle with pronunciation. Research examining pronunciation of children who have undergone cleft lip and palate repair has primarily focused on the production of consonants, such as palatal and nasal sounds.. The most distinctive features of pronunciation by. children with cleft lip and palate are nasal emission (Baghban et al., 2015) and middorsum palatal stops (Zajac, Cevidanes, Shah, & Hale, 2012).. While medical. research has been mainly concerned with the comparison of cleft lip and palate before and after surgery, little research has been conducted exploring vowel production of speakers with repaired cleft lip and palate in Taiwan.. As a result,. this study focuses on the production of Mandarin vowels by young speakers with repaired lip and cleft in Taiwan.. Acoustic analysis of vowel production serves as. the focus of this research. The first formant and the second formant play important roles in terms of the acoustic analysis of vowel performance.. Research has shown that formant. settings may influence the results (鄭靜宜, 2011; Hsiao, 2014).. Therefore, this. study will focus on particular on the vowel performance in three formant settings: formant setting 4, formant setting 5, and judged by the researcher. In sum, the main purpose of this study is to compare vowel production by children with and without repaired cleft lip and palate through acoustic analysis of the vowels.. . This study provides another perspective toward understanding speech. 3 .

(16) production of children with repaired cleft lip and palate.. While it has been clearly. demonstrated that these children have difficulties uttering certain consonants, there remains uncertainty surrounding the extent to which they are capable of producing vowels in the same way as children without cleft lip and palate.. 1.3 Research Questions of the Study. The purpose of this study is to examine if there is any difference in Mandarin vowel production of children with repaired cleft lip and palate and those without. Three central questions are investigated and discussed in this research: (1) To what extent does Mandarin vowel production by children with repaired cleft lip and palate differ from that of children without cleft lip and palate in terms of mean F1 values? (2) To what extent does Mandarin vowel production by children with repaired cleft lip and palate differ from that of children without cleft lip and palate in terms of mean F2 values? (3) To what extent does Mandarin vowel production by children with repaired cleft lip and palate differ from that by children without cleft lip and palate in terms of vowel space?. . 4 .

(17) 1.4 Definition of Terms. This section presents brief definitions of the terms in this study.. They. include cleft lip and palate, formant, and vowel space. (1) Cleft lip and palate According to the National Health Service (NHS) in the UK (https://www.nhs.uk/conditions/cleft-lip-and-palate/), a cleft is a split in the upper lip and/ or the palate, which is present from birth. This type of birth defect causes significant problems with speech production (Harding & Grunwell, 1996). The primary speech impacts of a cleft lip and palate are: lack of target stabilization, compensatory articulations, active and passive strategies, and systematic sound preference. (2) Formant A formant is a term from acoustic phonetics used in the classification of vowels and vowel sounds describing a concentration of acoustic energy and reflecting the way air from the lungs vibrates in the vocal tract (Crystal, 2008). The first, second and third formants are the standard formants used in the analysis of vowel quality. formant (F2).. This study focuses on the first formant (F1) and the second. The first formant is the lowest frequency describing the height of. tongue position, while the second formant is a higher frequency and is used to describe the forward placement of the tongue. the higher F1 will be. . The lower the tongue position is,. For example, /a/ has higher F1 values, while /i/ has lower 5 .

(18) F1 values.. The more forward the tongue is positioned, the higher F2 will be.. For instance, /i/ has the highest F2 values among all the vowels because the tongue is in the most forward position; the back vowel /u/, has the lowest F2 values. (3) Vowel space Acoustic vowel space is formed by several corner or cardinal vowels and is measured by the values of the first and the second formants (鄭靜宜, 2011).. In. Mandarin, there are five cardinal vowels including /a/, /e/, /i/, /o/ and /u/ (Chung, 2016).. Research has found a correlation between the area of vowel space and speech. intelligibility; the larger the vowel space, the higher the intelligibility (鄭靜宜, 2011; Bradlow, Torretta, & Pisoni, 1996; Liu, 2005; Turner, Tjaden, & Weismer, 1995).. . 6 .

(19) CHAPTER TWO. LITERATURE REVIEW. The purpose of this chapter is to offer an overview of the literature on cleft lip and palate.. The current research on cleft lip and palate is reviewed in the first part.. The second focuses on international protocol regarding treatment and speech therapy for children with cleft lip and palate.. The following section focuses on the vowels.. A brief summary is provided at the end of this chapter.. 2.1 Current Research on Cleft Lip and Palate. The oral production of the children with cleft lip and palate is quite different from that of the children without cleft lip and palate.. Harding and Grunwell (1996). regarded cleft palate speech as an articulatory disorder, identifying nasal resonance, nasal emission, and nasal turbulence as characteristics of cleft palate speech. The list below illustrates the ten cleft-type developmental processes in early speech. . 7 .

(20) Cleft-type processes occurring in early speech (Harding & Grunwell, 1996) Differences in voiced/voiceless plosive production Lack of target stabilization Phoneme mismatch Cleft-type strategies; passive and active compensatory strategies and adaptive strategies Backing pattern Systematic sound preferences Consonant range reduction Phonetic variability Persisting immaturities Developmental immaturities. Children with unrepaired cleft palate have also been shown to produce hyper nasal resonance and have difficulty producing consonants (Chapman & Willadsen, 2011).. Consonants require intraoral air pressure such as stops, fricatives and. affricates but children with cleft palate are unable to make these sounds due to the incomplete development of the palatal cleft (Chapman & Willadsen, 2011).. For. some children with velopharyngeal inadequacy, they might use passive or active strategy to produce sounds (Chapman & Willadsen, 2011).. Use of a passive. strategy causes nasalization of vowels and voiced consonants, nasal emission, turbulence and/or reduced intraoral air pressure on obstruent targets because they . 8 .

(21) are not able to lessen the unavoidable consequences of oral-nasal coupling (Chapman & Willadsen, 2011).. On the contrary, those who use an active. strategy generally change the place and/or manner of articulation, especially when they produce obstruents as they try to prevent the effects of oral–nasal coupling (Chapman & Willadsen, 2011).. Trost (1981) observed compensatory articulations in American English-speaking children with cleft palate.. He described back patterns of. articulation such as glottal stops, pharyngeal fricatives, and pharyngeal affricates as the children with cleft palate attempted to compensate for velopharyngeal inadequacy.. Trost’s (1981) findings further support observations by Chapman. and Willadsen (2011).. Hyper nasality has also been identified as one of the main. characteristics in children with cleft palate (Akafi, Vali, Moradi, & Baghban, 2013).. In Ha and Kuehn’s (2006) investigation of the temporal characteristics of. nasalization in children and adult speakers of American English and Korean they found that children with cleft palate exhibit hyper nasality showing longer durations of nasalization than children without cleft palate.. Similarly, Baghban. et al. (2015) conducted research measuring and comparing temporal patterns of nasalization in Persian children with and without cleft palate.. They concluded. that compared to children without cleft palate, children with cleft palate have longer nasalization durations (Baghban et al., 2015).. . 9 . They also found that total.

(22) nasalization duration in high vowel contexts was longer than low vowel contexts (Baghban et al., 2015).. In addition to hyper nasality, children with cleft lip and palate have difficulty producing palatal sounds (Zajac et al., 2012).. Research investigating the. maxillary arch dimensions and spectral characteristics of children with cleft lip and palate focused on the children’s production of middorsum palatal stops.. The. study included 19 children, with 5 adults participating as listeners in a perceptual identification.. The results showed that children with middorsum palatal stops. were found to have restricted anterior relative to posterior maxillary arch width. The researchers also found that middorsum palatal stops might occur in some children with repaired cleft lip and palate due to deviant palate morphology (Zajac et al., 2012).. Howard’s (2013) research aimed to investigate the phonetic and phonological parameters of speech production associated with cleft lip and palate in single words and in sentence repetition.. Two boys with persistent speech impairments. caused by unilateral cleft lip and palate were observed in the study.. Both of them. showed atypical phonetic features in their connected speech production, such as issues with alveolar-velar double articulation.. In addition, one of the participants. frequently omitted segments and syllables, which effected on his intelligibility (Howard, 2013).. Johnson (2004) observed that segmental and syllabic. reductions are strictly constrained in typical adult speech production and certain . 10 .

(23) key segments cannot be omitted if word recognition is to remain possible for the listener.. Klein and Flint (2006) also suggested that structural reductions are. more damaging to intelligibility. In Ahn, Kim, Kim, & Hong’s (2015) study investigating vowel sound changes before and after orthoganthic surgery included sixteen adult male patients with a mean age of 21.9. Eight of the participants exhibited mandibular prognathism while had normal occlusion.. Unlike previous research mentioned,. none of the participants in this study had cleft lip and palate.. Before the surgery,. the formant graphs showed that the eight patients with mandibular prognathism used a wide area in the production of each vowel; however, the area used by these patients had declined six months post-surgery (Ahn et al., 2015). The research above focused mainly on the production of consonants than vowels.. Even though Ahn et al.’s (2015) study centered on vowel change before. and after the surgery, the participants were not the patients with cleft lip and palate. Therefore, there are still significant knowledge gaps in terms of vowel production by children with cleft lip and palate.. . 11 .

(24) 2.2 The International Protocol Treatment and the Speech Therapy of Children with Cleft Lip and Palate. The purpose of this section is to introduce the current international treatment and speech therapy protocol.. The first section is the introduction of the protocol. treatment described by Bennun, Tainijoki, Ylikontiola, Sandor, & Casadio (2016). The second section introduces speech therapy protocol as described by Casadio (2016).. 2.2.1 Current International Protocol Treatment of Children with Cleft Lip and Palate. Early intervention is important for children with cleft lip and palate and should begin as early as the perinatal stage (Bennun et al., 2016).. The complexity of the. medical, surgical, and psychosocial factors involved in the treatment of cleft lip and palate means that coordinated care is essential.. It is also essential to implement. protocol treatment before 6 years old to so that children with cleft lip and palate are not restricted to special educational programs but can attend regular classes. protocol treatment suggested by Bennun et al. (2016) is in Appendix A.. . 12 . The.

(25) 2.2.2 The Speech Therapy of children with Cleft Lip and Palate. In addition to medical treatment, speech therapy plays a significant role in helping children with cleft lip and palate and should start before the lip and palate is repaired (Casadio, 2016). Speech therapy must be tailored to the different stages of child development and the therapist must consider each case individually.. Along. with medical care, families should also commit to follow up treatment to help their children continue to develop.. The speech therapy is part of a multidisciplinary team. made up of a maxillofacial surgeon, pediatric dentist, pediatrician and an ear, nose and throat specialist (Casadio, 2016).. 2.3 Studies of Mandarin Vowels. This section aims to review studies regarding Mandarin vowels.. The first part. of this section focuses on discussing the articulation and classification of vowels. The second part is concerned with the first formant and the second formant of Mandarin vowels.. The third part describes the relation between vowel space and. speech intelligibility.. 2.3.1 Articulation and Classification of Vowels. A vowel is defined as “speech that is formed without a significant constriction of the oral and pharyngeal cavities, and serves as a syllable nucleus.” (Shriberg & Kent, 1982, p. 35). . Unlike consonants whose airstream through the 13 .

(26) vocal tract must be obstructed in some way, while producing vowels, the passage of the airstream is rather unobstructed (Ladefoged, 2011).. As a result, vowels. are characterized as sonorous and syllabic (O’Grady, 2016). produced by moving the placement of the body of the tongue.. Vowels are If the highest. point of the tongue is in the front of the mouth while uttering the vowels, then these vowels are classified as front vowels.. If the tongue is close to the back of. the mouth, then the vowels are viewed as back vowels (Ladefoged, 2011; O’Grady, 2016). There are seven major vowels in Mandarin, which are all regarded tense vowels (Tung, 1994). These include /i(ㄧ), u(ㄨ), y(ㄩ), a(ㄚ), o(ㄛ), ɤ(ㄜ) , e(ㄝ)/.. Lai (2006) categorized the seven vowels as follow:. (a) Front vowels: /i, e, y/ (b) Back vowels: /ɤ , u, o/ (c) High vowels: /i, y, u/ (d) Low vowel: /a/. 2.3.2 The First Formant Frequency and the Second Formant Frequency of the Mandarin Vowels. Research has shown that the most significant characteristic of vowels is their formant frequency.. For instance, the distance between the first formant. frequency and the second formant frequency of the front vowel /i/ is wider than . 14 .

(27) that of the back vowels such as /u/ or /o/.. Vowels are classified according to the. shape and constriction of the vocal tract (鄭靜宜, 2011; Honda, 2015).. Pickett. (1999) described the relationship between the constriction of the vocal tract and the formant frequency while producing vowels.. The length of the vocal tract, the. constriction of the oral and pharyngeal structures, the placement of the tongue as well as the shape of the lips cause different formant frequencies.. For example,. compared to the female speakers, the average of formant frequencies of male speakers is lower because male speakers usually have longer vocal tracts than female speakers do. Children have the highest formant frequencies because they have the shortest vocal tract of the three (鄭靜宜, 2011). Generally speaking, the first formant frequency is influenced by the height of tongue.. The lower the tongue position is, the higher the first formant frequency. will be.. On the other hand, the second formant frequency is influenced by the. advancement of tongue.. The closer to the front of the mouth the tongue is. positioned, the higher the second formant frequency will become (鄭靜宜, 2011). In terms of the first and the second formant frequencies of Mandarin vowels, researchers have found that regardless of the vowel, female Mandarin speakers have higher F1 and F2 than the male Mandarin speakers do (鄭靜宜, 2004, 2005; Huang, 1996; Jeng, 2000).. Hsiao’s (2013) study of 110 students from four elementary. schools in the southern Taiwan showed that the values of F1 of /a/, /e/, /i/, /o/, and /u/. . 15 .

(28) were 943, 628, 421, 559, and 485 while the values of F2 of the five vowels were 1682, 2334, 2886, 1267, and 1340 respectively.. Hsiao’s (2013) study revealed that the. values of F1 and F2 of the children are higher than those of the males.. Compared to. the values of F1 of the females, /i/ and /u/ of the children are higher than those of the females while /a/ is lower than that of the females.. As far as F2 is concerned, only. /o/ and /u/ of the children are higher than those of the females.. Table 2.1 is the. average values of F1 and F2 in different reports.. Table 2.1 The average value of F1 and F2 in different reports (鄭靜宜 2004; ⿈黃國佑 1996, Hsiao, 2013; Jeng, 2000). /e/. /i/ /o/. /u/. . Female. Children. Formant. Vowel /a/. Male 鄭靜宜. ⿈黃國佑. Jeng. 鄭靜宜. ⿈黃國佑. Jeng. Hsiao. (2004). (1996). (2000). (2004). (1996). (2000). (2013). F1. 743. 836. 767. 1013. 1053. 1090. 943. F2. 1240. 1358. 1329. 1641. 1658. 1686. 1682. F1. 520. 561. 510. 679. 631. 608. 628. F2. 1923. 1898. 1851. 2508. 2324. 2463. 2334. F1. 298. 293. 274. 396. 347. 352. 421. F2. 2243. 2274. 2123. 2782. 2852. 2908. 2886. F1. 445. 579. 468. 552. 672. 581. 559. F2. 814. 1037. 830. 994. 1194. 999. 1267. F1. 326. 326. 326. 397. 415. 400. 485. F2. 718. 877. 796. 838. 950. 806. 1340. 16 .

(29) 2.3.3 Vowel Space. Vowel space is used to show the auditory qualities of vowels since it is not easy to exactly point out the positions of the vocal organs in the vowels without x-ray or MRI (Ladefoged, 2011).. In Mandarin, the vowel space created by the. four vowels /a/, /i/, /u/, and /e/ is in the form of a quadrilateral (鄭靜宜, 2011). Linguists consider /i/, /u/, /a/, /e/, and /o/ cardinal vowels (Chung, 2016). Research has shown that because of the different length of vocal tracts, the vowel space differs between genders and ages.. In general, male speakers have a. smaller vowel space than female speakers do, while children have the largest vowel space due to the fact that their vocal tracts are shorter than those of adults (鄭靜宜, 2011).. Researchers have also established a relationship between vowel space and. intelligibility; the larger the vowel space, the more intelligible the speech (Bradlow et al., 1996; Turner et al., 1995). Liu (2005) also described a connection between vowel space area and both vowel and word intelligibility.. 2.4 Summary. This chapter begins with a review of current research on children with cleft lip and palate, followed by the international treatment and speech therapy protocol for children with cleft lip and palate. This is followed by a review of studies looking at. . 17 .

(30) the verbal production of vowels, acoustic analysis on vowels and the relation between vowel space and intelligibility.. A significant body of research has revealed speech. differences of children with cleft lip and palate from those without.. However, few. studies have investigated vowel production of children with cleft lip and palate, which is an important part of speech intelligibility. academic research in this area.. . 18 . This study addressed the lack of.

(31) CHAPTER THREE. METHOD. 3.0 Introduction. The purpose of this study is to measure and compare vowel performance in terms of F1, F2 and vowel space between children with cleft lip and palate and children with lip and palate free from any abnormality.. This chapter is composed. of four parts and describes how the study was designed. The participants in this study are introduced in the first section of this chapter. describes the instruments used to collect the data.. The second section. The third section covers the. procedure for obtaining the recordings of the participants.. The fourth section. covers data analysis. In order to protect the rights and welfare of research participants, the researcher attended a three-hour training course held by National Cheng Kung University Human Research Ethics Committee (Appendix C).. This study was supervised and. approved by this committee and has been verified as being in conformity with the ethical guidelines.. . Approval of this study can be found in Appendix D.. 19 .

(32) 3.1 Participants. There were fourteen children participating in this research including eight students with normal lip and palate and six students with repaired cleft lip and palate. All students spoke Mandarin as their first language and shared similar exposure to Mandarin.. All participants attended elementary school in southern Taiwan and. ranged in age from eight to eleven.. Research that focusing on vowel performance of. children doesn’t include gender in its comparisons. Vowel-specific formant patterns differ between genders and between adults and children due to differences in average vocal tract sizes (Maurer & Suter, 2015). While children’s vocal tracts are much shorter than adult’s (鄭靜宜, 2011), there is no significant difference in children’s vocal tracts in terms of gender.. As a result, the participants in this current study are. grouped and investigated according to their lip and palate rather than gender difference.. The following section provides detailed information about the two. groups.. 3.1.1 GWC. GWC consisted of six speakers with repaired cleft lip and palate ranging in age from eight to nine years old who were introduced by the Noordhoff Craniofacial Foundation.. All of them had undergone lip and palatal surgeries as infants and all. received care according to treatment protocol and speech therapy.. Aside from the. repaired cleft lip and palate, they did not have any hearing or perceptual impairments . 20 .

(33) that might interfere with pronunciation analysis.. The speakers in GWC were all. Mandarin speakers with basic reading ability; as a result, they were able to read the ten sentences consisting of the target five Mandarin vowels of the study without any difficulties.. 3.1.2 GWOC. GWOC was composed of eight children without cleft lip and palate between the ages of nine to eleven. perceptual impairments.. None of the eight participants suffered from hearing or Similarly to GWC, they were all their native Mandarin. speakers with a reading ability sufficient to complete the reading task without any guidance.. 3.2 Instrument. The following section is detailed information regarding the instruments employed for data collection and data analysis.. The reading lists, questionnaire,. recording system and recording equipment for data collection are described as well as the software used for analyzing the data.. . 21 .

(34) 3.2.1 Instruments for Data Collection. The instruments for data collection to fulfill the aims of this research include: a reading list, a questionnaire, Praat system ( Bocrsma & Weenink, 2009, version:6.0.16), a recording shelter, consent form, a laptop, a microphone. (1) A Reading List In order to measure the vowel space of the target participants, the fourteen children were given a list containing ten sentences with five Mandarin vowels designed by a professional linguist, Dr. Chung. were [u], [a], [i], [e], and [o].. The five target Mandarin vowels. These five vowels were added to the Mandarin. consonant [b] to form a CV and CVCV configuration.. The ten sentences were in. the form of two sentence patterns. Sentence 1: Wohuixie . Sentence 2: Wohuixie. (English: I can write .) de.. (English: I can write .). Sentence 1 is in the CV configuration while sentence 2 is the CVCV formation. The ten sentences were randomly distributed on the list.. This reading list is. attached to further illustrate the elicited utterances (Appendix E). (2) A Questionnaire In order to select the participants meeting the parameters of this study, a questionnaire (Appendix F) was designed to collect the participants’ basic personal information, including their gender, age, and the language that they mainly used at home. . 22 .

(35) (3) Praat Praat (version: 6.0.16) was used to collect the sound data.. Invented at the. University of Amsterdam in 2009, Praat is considered to be a functional and convenient software for recording, saving, and editing sounds, in addition to showing frequency values, which made it an appropriate choice based on the needs of this study.. The system was utilized to collect data in the form of spoken. sounds. (4) A Recording Shelter In order to reduce the interference of outside noise, a recording shelter designed by Hsiao (2013) was used during the recording session.. In his research, Hsiao (2013). found that acoustic interference during the recording process was compromising the quality of data being collected. To remedy this, he devised a recording shelter to improve the purity of the speech sounds.. The shelter was portable, enabling the. researcher to find a quiet place for recording.. When the participants sit in front of it,. they are sheltered from the outside environment, resulting in better quality sound recordings.. . Figure 3.1 is the picture of the recording shelter used in the study.. 23 .

(36) Figure 3.1 The recording shelter (5) Consent Since all the participants were under eighteen years old, consent was obtained with their guardians’ permission.. The consent form consisted of a description of. the research, including the motivation and the goals, the recording procedure, and information about how the privacy of the children would be protected.. The. purpose of the consent form was to inform the guardians that their children would be kept in a safe and secure environment throughout their participation.. In. addition, this research was not concerned with, nor would it investigate or reveal any personal information about the participants.. All of the data and personal. information will be stored in a safe place with keys and passwords for three years. After the three years, all the data and information will be destroyed. the consent form is in Appendix G.. . 24 . A copy of.

(37) (6) A laptop A portable laptop was one of the instruments utilized for collecting data. Praat, which was the primary software of this study, was downloaded onto the laptop for use during the analysis stage. A recording pen was not used because directly recording onto the laptop produced a higher sound quality.. Moreover,. the sounds that were recorded could be analyzed immediately with Praat. consequence, this laptop played a significant role in this research.. As a. The laptop. used was Apple MacBook Air. (7) A Microphone A headset microphone with a speaker made by Apple Company was used for this research.. Each of the children was asked to put on the headset microphone, which. was already connected to the laptop. They were then asked to read the ten sentences on the reading list to the speaker for recording.. 3.2.2 Instruments for Data Analysis. After finishing the entire recording, all the data gathered were analyzed with the following three instruments.. They were Praat, Microsoft Office Excel, and a. programmed keyboard, which are introduced respectively in the sections below. (1) Praat This software was employed during the data analysis stage.. Praat allows the. researcher to automatically fetch vowel frequencies while simultaneously. . 25 .

(38) displaying formants such as F1 and F2, which indicate tongue position when producing a particular sound. (2) Microsoft Office Excel 2010 Microsoft Office Excel 2010 was used for the statistical analysis of the values of F1 and F2 retrieved and evaluated by Praat.. This software allowed the. researcher to categorize the values of F1 and F2 respectively and display the results as vowel charts. (3) A Programmed Keyboard In order to save time and reduce some mistakes while analyzing the data, a technical keyboard programmed by Hsiao (2013) was used to transfer the digital values of F1 and F2 from Praat.. It also allowed the researcher to select, name, and. categorize the wave files effectively.. 3.3 Procedure. The sixteen participants were invited individually into a quiet room to read the five target vowels in their carrier sentences in Mandarin. The room also contained a laptop computer (Macbook Air), a high quality microphone (Mac), a shelter, the five target vowels in carrier sentences in Mandarin, the questionnaire and the consent form. The duration of recording was about ten to fifteen minutes each.. . 26 .

(39) 3.3.1 Before Recording. Before recording, the researcher briefly described the purpose and procedures of the study to the speakers’ guardians and politely asked them to sign the consent if they agreed to allow their children to participate in this study.. Prior to beginning in. the exercise, the researcher engaged in playing with the children in order to build rapport and ease any anxiety the children may have been feeling about working with an unknown adult. The researcher then gave each participant a detailed explanation about what they would be asked once they had settled into the room.. Each. participant was asked to fill in the questionnaire with the researcher’s assistance if needed.. After filling in the questionnaire, the speaker would be shown the five. target vowels in carrier sentences in Mandarin and asked to read them aloud several times to get familiar with them.. The last step of this session was to ensure that the. distance between the speaker’s mouth and the microphone was standardized at ten centimeters for the recording.. The speaker would also be asked to speak a sentence. with the microphone to test the quality of recording.. 3.3.2 During Recording. Once they were ready for recording, each participant was asked to read the ten sentences three times.. The main reason for asking the participants to read the. sentences three times to ensure that a quality recording had been achieved. . 27 . During.

(40) the recording, the participant was encouraged to read the ten sentences one by one in a spontaneous manner and at a consistent volume.. The researcher would not give. participants any hints or corrections when recording.. The recording session would. be monitored to avoid weak or overloaded signals. 3.3.3 After Recording. The recording session would last for about ten to fifteen minutes.. After. recording, each participant was given a small gift as appreciation. 3.4 Data Analysis. The entire recording would be transformed into wave files using the Praat sound recorder for further acoustic analysis.. In order to address the three research. questions, data analysis focused on three major dimensions: (a) the mean F1 values in the five target Mandarin vowels, (b) the mean F2 values in the five Mandarin vowels and (c) the vowel space of the Mandarin vowels.. To analyze the data, each target. vowel of every participant was isolated and given a mean value for the first formant and second formant.. All the figures were saved in an Excel file in order to calculate. the collected acoustic values and to plot the frequency charts and vowel charts.. . 28 .

(41) CHAPTER FOUR. FINDINGS AND DISCUSSIONS. 4.0 Introduction. This chapter presents the results of the production of the five Mandarin vowels /a/, /i/, /u/, /e/, and /o/.. The main purpose of this research is to investigate the. acoustic properties of vowel performance of the fourteen participants grouped according to whether or not they had repaired cleft lip and palate. composed of three parts.. This chapter is. The first section compares the vowel positions of the two. groups in terms of formant setting 4, 5 and formant setting judged by the researcher to investigate the extent to which the two groups shared similar vowel positions.. In the. second part, vowel space is examined in order to determine similarities and differences.. The final part of this chapter briefly summarizes the statistical results. and offers general discussion about the vowel performance of the two groups.. . . 29 .

(42) 4.1 Vowel Positions in Group of the Children with Repaired Cleft Lip and Palate and Group of the Children without Cleft Lip and Palate. The aim of this section is to investigate the positions of the five target vowels of the children with repaired cleft lip and palate (GWC) and the children without cleft lip and palate (GWOC). This part of the analysis addresses the first two research questions (1) To what extent does Mandarin vowel production by the children with repaired cleft lip and palate differ from that by children without repaired cleft lip and palate in terms of mean F1 values in vowels? (2) To what extent does Mandarin vowel production by the children with repaired cleft lip and palate differ from that by children without repaired cleft lip and palate in terms of mean F2 values in vowels? The first to third sections explore F1 and F2 in formant settings 4, 5, and judged by the researcher.. The final section is a short summary of the findings .. 4.1.1 Vowel positions of GWC and GWOC in Formant Setting 4. This sub-section aims to investigate vowel positions of GWC and GWOC in formant setting 4.. Vowel positions of the two groups in formant setting 4 are. compared in terms of mean values of F1 and F2 and the distribution of each vowel.. . 30 .

(43) Table 4.1 Means and standard deviations of F1 and F2 of the five vowels of GWOC and GWC in formant setting 4. Group. GWOC. GWC p. Vowel. F1. F2. M. SD. M. SD. /a/. 970. 114. 1,007. 86. 0.090. /e/. 594. 103. 702. 58. 0.000. /i/. 369. 71. 446. 40. 0.000. /o/. 544. 78. 621. 97. 0.000. /u/. 445. 84. 542. 90. 0.000. /a/. 1,619. 183. 1,749. 122. 0.000. /e/. 2,409. 125. 2,534. 231. 0.005. /i/. 3,028. 163. 3,157. 225. 0.005. /o/. 1,013. 339. 1,168. 310. 0.032. /u/. 847. 156. 1,109. 459. 0.002. Note: M: mean; SD: standard deviations. Table 4.1 offers detailed descriptive statistics of vowel positions in formant setting 4.. The mean value of F1 of the /a/ of GWOC is 970Hz while that of. GWC is 1007Hz.. The data suggests that the position of the /a/ of GWC is. slightly lower than that of GWOC. . However, the result of the t-test shows there 31 .

(44) is no significant difference of the /a/ sounds between the two groups in terms of height.. In contrast, the forward positioning of the /a/ sounds of the two groups. are significantly different.. The F2 of GWC is 1749Hz and of GWOC is 1619Hz. meaning that GWC has higher F2 values of /a/ and GWOC has lower F2 values of /a/.. The statistical result (p<.05) shows that the /a/ sound of GWC occurs. significantly further forward than that of GWOC.. In brief, compared to the /a/ of. GWOC, the /a/ sound uttered by GWC is in a lower and more advanced position. The other four vowels of the two groups are significantly different.. GWC. has significantly higher F1 value than GWOC does, with the F1 of /e/ of GWC measuring 702Hz and GWOC measuring 594Hz.. GWC also has significantly. higher F2 values than GWOC does, with the F2 values of each group measuring 2,534Hz and 2,409Hz respectively.. The data analysis reveals that the /e/ of the. two groups does not share the same position.. On the contrary, the /e/ uttered by. GWC is significantly lower and more advanced than it uttered by GWOC.. The. other three vowels including /i/, /o/, and /u/ all share the same findings as /e/.. In. short, compared to GWOC, GWC tends to place the body of the tongue in a lower and more forward position while producing vowels. The following five figures illustrate the vowel distributions of GWC and GWOC in formant setting 4. Detailed descriptions of these five figures as well as statistical analysis follow.. . 32 .

(45) F2 (Hz) . F1 (Hz) . . Figure 4.1 Vowel /a/ distribution in formant setting 4 (GWC (blue) and GWOC (red)). F2 (Hz) F1 (Hz) . . Figure 4.2 Vowel /e/ distribution in formant setting 4 (GWC (blue) and GWOC (red)). . 33 . .

(46) F2 (Hz) . F1 (Hz) . . Figure 4.3 Vowel /i/ distribution in formant setting 4 (GWC (blue) and GWOC (red)). F2 (Hz) . F1 (Hz) . Figure 4.4 Vowel /o/ distribution in formant setting 4 (GWC (blue) and GWOC (red)). . 34 .

(47) F2 (Hz) . F1 (Hz) . . Figure 4.5 Vowel /u/ distribution in formant setting 4 (GWC (blue) and GWOC (red)). Figures 4.1 to 4.5 show the vowel distributions uttered by these 14 participants in formant setting 4.. The blue letters represent the production by. GWC and the red one stands for the production by GWOC.. These figures also. support the findings discussed above that the GWC participants tend to pronounce the five vowels in lower and more forward positions compared to the GWOC group. Table 4.2 and Table 4.3 are the ANOVA analysis of F1 and F2 of GWC and GWOC in formant setting 4.. Detailed descriptions of these two tables follow.. . . 35 .

(48) Table 4.2 ANOVA analysis of F1 of GWC and GWOC in formant setting 4 Source. SS. dF. MS. F. Group. 15,207. 1. 15,207. Vowel. 396,621. 4. 99,155. error. 1,369. 4. 342. Total. 413,197. 9. 44.433**. p-Value 0.002. Note: **p< .01. Table 4.2 shows the F1 in formant setting 4 for GWC and GWOC.. The result. of the ANOVA analysis shows that the two groups have significant differences in terms of the F1.. In other words, the height of the five vowel positions of GWC and. GWOC are not similar to each other. GWOC does.. GWC has significantly higher F1 values than. Therefore, according to the statistical analysis, it has been clearly. demonstrated that the positions of the target five vowels of the two groups are not the same.. . 36 .

(49) Table 4.3 ANOVA analysis of F2 of GWC and GWOC in formant setting 4 Source. SS. dF. MS. F. Group. 64,170. 1. 64,170. Vowel. 6,582,463. 4. 1,645,616. error. 6,435. 4. 1,609. Total. 6,653,067. 9. p-Value. 39.887**. 0.003. Note: **p< .01. Table 4.3 illustrates the values of F2 of the two groups in formant setting 4. The ANOVA analysis shows that the advancement of the five vowel positions is also significantly different. than those of GWOC.. The values of F2 of GWC are significantly higher. In short, the two groups have significantly different vowel. positioning in terms of the front-back dimension. The acoustic data analysis in formant setting 4 reveals that the five vowel positions of GWC are significantly different from those of GWOC. are positioned lower and further forward than those of GWOC.. The five vowels. These outcomes. address the first two research questions, revealing that the children with repaired cleft lip and palate do not have similar vowel positions as the children without cleft lip and palate, but tend to utter the five vowels in lower and more advanced positions.. . 37 .

(50) 4.1.2 Vowel positions of GWC and GWOC in Formant Setting 5. This sub-section focuses on the vowel positions of GWC and GWOC.. Formant. setting 5 in F1 and F2 and the distribution of each vowel are discussed in this sub-section. Table 4.4 Means and standard deviations of F1 and F2 of the five vowels of GWC and GWOC in formant setting 5 Group. GWOC. GWC p. Vowel. F1. F2. M. SD. M. /a/. 936. 149. 963. 140. 0.396. /e/. 603. 87. 672. 77. 0.000. /i/. 391. 66. 448. 49. 0.000. /o/. 532. 59. 589. 62. 0.000. /u/. 445. 69. 521. 66. 0.000. /a/. 1,535. 200. 1,591. 205. 0.214. /e/. 2,102. 410. 1,723. 392. 0.000. /i/. 2,801. 383. 2,554. 653. 0.048. /o/. 913. 113. 1,007. 133. 0.001. /u/. 852. 124. 946. 116. 0.000. Note: M: mean; SD: standard deviations . SD. 38 .

(51) Table 4.4 exhibits all the mean values of F1 and F2 in formant setting 5. According to this table, the vowels /i/ and /e/ in formant setting 5 are in positioned further back in the mouth.. Similar to the findings in formant setting 4, the /a/. sounds of the two groups are not significantly different.. Although both F1 and. F2 of /a/ of GWC are slightly larger than those of GWOC, the results of t-test shows that the /a/ sound of the two groups is not significantly different, indicating that the GWC production of /a/ is closer in position of GWOC.. Aside from this. similarity, the data shown in Table 4.4 reveals significant differences with the other four vowels.. As far as the vowels /i/ and /e/ are concerned, the. performance of these two sounds by GWC is significantly different than that of GWOC.. The F1 values of /i/ and /e/ of GWC are significantly higher, indicating. the two vowel positions of GWC are significantly lower than those of GWOC. This finding is identical to that of the formant setting 4. However, in terms of F2 values, the mean value of F2 of /e/ for GWC is 1,723Hz and 2,102Hz for GWOC; F2 of /i/ of GWC is 2,554Hz and 2801Hz for GWOC.. This indicates that the performance of /e/ and /i/ of the two group is. significantly different.. The location of these two sounds uttered by GWC occurs. further back in the mouth than those uttered by GWOC.. Vowels /o/ and /u/ of. GWC are significantly lower and further forward in the mouth, making them identical to the findings in formant setting 4.. The t-test shown in Table 4.4. reveals that the performance of these two back rounded vowels differ significantly between the two groups. . The statistical findings provide strong evidence that /o/ 39 .

(52) and /u/ of GWC are significantly lower and further forward, compared to those of GWOC.. In summary, as the data in formant setting 5 illustrates above, the GWC. participants produce a similar quality /a/ as the GWOC group; however, their performance of the other four vowels shows measurable differences to those of GWOC. The following five figures illustrate the vowel distributions of GWC and GWOC in formant setting 5. Detailed descriptions of these five figures as well as statistical analysis follow.. F2 (Hz) . F1 (Hz) . Figure 4.6 Vowel /a/ distribution in formant setting 5 (GWC (blue) and GWOC (red)). . 40 .

(53) F2 (Hz) . F1 (Hz) . Figure 4.7 Vowel /e/ distribution in formant setting 5 (GWC (blue) and GWOC (red)). F2 (Hz) . F1 (Hz) . Figure 4.8 Vowel /i/ distribution in formant setting 5 (GWC (blue) and GWOC (red)). . 41 .

(54) F2 (Hz) . F1 (Hz) . Figure 4.9 Vowel /o/ distribution in formant setting 5 (GWC (blue) and GWOC (red)). F2 (Hz) . F1 (Hz) . Figure 4.10 Vowel /u/ distribution in formant setting 5 (GWC (blue) and GWOC (red)). . 42 .

(55) Figures 4.6 to 4.10 illustrate the distribution of the five vowels produced by each participant of the two groups.. The blue letters represent the production by. GWC, and the red letters represent for the production by GWOC.. The. distribution of the /a/ sounds for the two groups is in accordance with the previous finding showing little difference in the position of the tongue for /a/ between the two groups.. Unlike the findings in formant setting 4, the distribution of /e/ and. /i/ of GWC is lower but further back than GWOC.. As for the two back rounded. vowels, Figures 4.9 and 4.10 reveal the same trend as in formant setting 4.. The. production of /o/ and /u/ of GWC is comparatively lower and further forward in the mouth. Table 4.5 and Table 4.6 show the results of the ANOVA analysis of F1 and F2 for GWC and GWOC in formant setting 5.. A detailed description of these two. tables follows.. Table 4.5 ANOVA analysis of F1 of GWC and GWOC in formant setting 5 Source. SS. dF. MS. F. Group. 7,892. 1. 7,892. Vowel. 340,075. 4. 85,019. error. 634. 4. 158. Total. 348,601. 9. Note: **p< .01 . 43 . 49.813**. p-Value 0.002.

(56) Table 4.5 presents the F1 in formant setting 5 for GWC and GWOC.. The result. of the ANOVA analysis shows that the two groups have significant differences in terms of F1, which means the height of the tongue positions of the two groups is significantly different. GWOC group.. The GWC group has significantly higher F1 values than the. This means the GWC group tends to produce the vowels in. significantly lower positions than GWOC.. In brief, according to the statistical. analysis, the positions of the target five vowels of the two groups show significant differences.. This finding is in accordance with the outcomes discussed in formant. setting 4.. Table 4.6 ANOVA analysis of F2 of GWC and GWOC in formant setting 5 Source. SS. dF. MS. Group. 12,740. 1. 12,740. Vowel. 4,259,785. 4. 1,064,946. error. 92,263. 4. 23,066. Total. 4,364,789. 9. F. p-Value 0.552. 0.498. Table 4.6 shows F2 values in formant setting 5 for GWC and GWOC.. In. formant setting 5, F2 value of the two groups is not significantly different.. This. finding indicates that the two groups have similar tongue positions in terms of forward positioning of the tongue. . This does not necessary imply that the tongue 44 .

(57) positioning of the two groups have similar forward positioning because in formant setting 5, there is the possibility that the values of F2 may have been miscalculated by the computer program.. Research has shown there is a risk of calculating the wrong. frequency found that when using formant tracking (鄭靜宜, 2011).. For example, the. value of F3 might be incorrectly measured as F2 or F2 may be calculated as F1.. The. possibility of miscalculating the value of frequencies may compromise the reliability of these findings. 4.1.3 Vowel Positions of GWC and GWOC in Formant Setting Judged by the Researcher. This sub-section mainly exhibits the vowel positions of GWC and GWOC.. In. this sub-section, the formant setting judged by the researcher is compared in terms of mean values of F1 and F2 and the distribution of each vowel.. . 45 .

(58) Table 4.7 Means and standard deviations of F1 and F2 of the five vowels of GWC and GWOC in formant setting judged by the researcher. Group Vowel. F1. F2. GWOC. GWC. p. M. SD. M. SD. /a/. 978. 106. 968. 139. 0.713. /e/. 603. 90. 701. 59. 0.000. /i/. 385. 68. 440. 46. 0.000. /o/. 529. 60. 587. 63. 0.000. /u/. 442. 72. 520. 66. 0.000. /a/. 1,636. 164. 1,686. 163. 0.165. /e/. 2,362. 195. 2,486. 324. 0.049. /i/. 2,954. 238. 3,039. 547. 0.386. /o/. 913. 113. 1,016. 135. 0.0005. /u/. 847. 122. 947. 117. 0.000. Note: M: mean; SD: standard deviations. Table 4.7 shows all the mean values of F1 and F2 in formant setting judged by the researcher.. Despite the fact that F1 value of /a/ sound for GWC is slightly. lower than that of GWOC, which is in contrast to F1 values of /a/ in formant settings 4 and 5, the t-test still reveals that the difference between the two /a/ sounds uttered by GWC and GWOC is not significant. results of formant settings 4 and 5.. This finding echoes the. The F2 values of the /a/ sounds of the two. groups are not significantly different either which means the tongue position for . 46 .

(59) /a/ sounds of the two groups are not significantly different.. This finding is in. accordance with the finding in formant setting 5 but contradict the findings in formant setting 4. As far as the other four vowels are concerned, the (F1) values of F1 of GWC are all greater than those of GWOC. In addition, the statistical analysis shows significant differences between the F1 values of the two groups, which means the height of the tongue for the four vowels of GWC and GWOC reach a significant level.. In other words, the participants of GWC tend to utter /i/, /e/, /o/, and /u/ in. significantly lower positions than those of GWOC. The findings of F2 values reveal that the vowels produced by the GWC group are all positioned further forward in the mouth than those produced by GWOC. However, the statistical result shows that only /e/, /o/, and /u/ of the two groups are significantly different, while the /a/ and /i/ sounds are not.. In other words,. the pronunciation of the /e/, /o/, and /u/ of the two groups showed significant difference in terms of how far forward the tongue is positioned.. In brief, the. comparison of acoustic data from GWC and GWOC in formant setting judged by the researcher shows that the vowel production of /e/, /o/, and /u/ for the GWC group are positioned significantly further forward than for the GWOC groups, while the vowels /i/, /e/, /o/, and /u/ are in significantly lower positions, with the exception of /a/. The following five figures illustrate the vowel distributions of GWC and GWOC in formant setting judged by the researcher. Detailed descriptions of these figures as . 47 .

(60) well as statistical analysis follow. F2 (Hz) . F1 (Hz) . Figure 4.11 Vowel /a/ distribution in formant setting judged by the researcher (GWC (blue) and GWOC (red)) F2 (Hz) . F1 (Hz) . Figure 4.12 Vowel /e/ distribution in formant setting judged by the researcher (GWC (blue) and GWOC (red)). . 48 .

(61) F2 (Hz) . F1 (Hz) . Figure 4.13 Vowel /i/ distribution in formant setting judged by the researcher (GWC (blue) and GWOC (red)). F2 (Hz) . F1 (Hz) . Figure 4.14 Vowel /o/ distribution in formant setting judged by the researcher (GWC (blue) and GWOC (red)). . 49 .

(62) F2 (Hz) . F1 (Hz) . Figure 4.15 Vowel /u/ distribution in formant setting judged by the researcher (GWC (blue) and GWOC (red)). Figures 4.11 to 4.15 provide another perspective from which to interpret the vowel positions of GWC and GWOC.. The blue letters stand for the vowel. positions of GWC and the red letters stand for the vowel positions of GWOC. The distribution of the /a/ sound of the two groups falls close together, which is identical to the findings in the formant settings 4 and 5.. As for the front vowels. /e/ and /i/, the positions of GWC are lower and further forward, as are the back vowels /o/ and /u/.. As per the first two research questions, these findings reveal. that the two groups do not share the same quality of vowel positions in terms of F1 and F2.. . 50 .

(63) Tables 4.8 and 4.9 show the results of ANOVA analysis of F1 and F2 of GWC and GWOC in formant setting judged by the researcher.. Detailed descriptions of. these two tables follow. Table 4.8 ANOVA analysis of F1 of GWOC and GWC in formant setting judged by the researcher Source. SS. dF. MS. F. Group. 7,424. 1. 7,424. Vowel. 381,506. 4. 95,376. error). 3,194. 4. 798. Total. 392,124. 9. p-Value 9.298*. 0.038. Note: *p< .05 Table 4.9 ANOVA analysis of F2 of GWOC and GWC in formant setting judged by the researcher Source. SS. dF. MS. Group. 21,855. 1. 21,855. Vowel. 6,697,817. 4. 1,674,454. error. 1,521. 4. 380. Total. 6,721,193. 9. Note: **p< .01. . F. 51 . 57.476**. p-Value 0.001.

(64) Tables 4.8 and 4.9 present the values of F1 and F2 in formant setting judged by the researcher.. The findings reveal significant differences in vowel positions. in terms of tongue positioning of the 14 participants of the two groups.. In. general, GWC has significantly higher F1 and F2 values than GWOC. The aim of this section is to answer the first two research questions, regarding the extent to which Mandarin vowel production by the children with repaired cleft lip and palate differs from that by children without cleft lip and palate in terms of mean F1 and F2 values.. The findings shown above demonstrate that there are significant. differences in Mandarin vowel production by GWC and GWOC.. The analysis of the. distributions of the five vowels further illustrate that the vowels uttered by the GWC group are positioned lower and further forward compared to those by the GWOC group.. 4.2 Vowel Space in Group of the Children with Repaired Cleft Lip and Palate and Group of the Children without Cleft Lip and Palate. The aim of this section is to analyze the vowel space of the children with repaired cleft lip and palate (GWC) and the children without cleft lip and palate (GWOC).. Vowel space area is a two-dimensional area formed by lines connecting. F1 and F2 coordinates of vowels (Sandoval, Berisha, Utianski, Liss, & Spanias, 2013). According to 鄭靜宜 (2011) vowel space is calculated by adding the areas of two triangles, which form a quadrilateral vowel space area. . 52 . The vowel space of this.

(65) current study is calculated according to the formula suggested by 鄭靜宜 (2011). The vowel utterances of the two groups are analyzed according to these measures and the findings are presented in this section.. In order to ensure validity, three data. points related to vowel production are utilized including average values in formant setting, 4, 5, and the formant setting judged by the researcher.. 4.2.1 Vowel Space of the Two Groups in Formant Setting 4. The vowel space of GWC and GWOC in formant setting 4 is presented in this sub-section.. Figure 4.16 represents the mean values retrieved in formant setting 4.. Table 4.10 presents the detailed statistical analysis of vowel space of the two groups.. F2(Hz) . F1(Hz). Figure 4.16 Means of the data fetched in formant setting 4 in vowel space and vowel positions of GWOC (red) and GWC (blue). . 53 .

(66) Figure 4.16 illustrates the vowel space of the two groups in formant setting 4. The blue letters and lines stand for GWC and the red letters and lines represent GWOC.. According to the figure displayed above, it is apparent that the two groups. do not share the same vowel space.. The vowel space of the GWOC is 579,280 Hz. 2. while the GWC is 524803 Hz which means the vowel space of GWC is 54477 Hz 2,. 2. smaller than that of GWOC.. Table 4.10 ANOVA analysis of the vowel space of GWOC and GWC in formant setting 4 Source. SS. dF. MS. F. Between groups. 14,977,806,834. 1. 14,977,806,834 1.05. Within groups. 171,081,371,695. 12. 14,256,780,975. Total. 186,059,178,529. 13. p-Value 0.32. Table 4.10 presents the result of the ANOVA analysis of the vowel space of GWC and GWOC in formant setting 4.. According to this analysis, there is no. significant difference between the two groups in terms of vowel space.. 4.2.2 Vowel Space of the Two Groups in Formant Setting 5. This sub-section discusses the vowel space of GWC and GWOC in formant setting 5. . Figure 4.17 represents the mean values retrieved in formant setting 5 in 54 .

(67) vowel space and vowel positions of GWOC and GWC.. Table 4.11 presents the. detailed statistical analysis of vowel space of the two groups.. F2(Hz) . F1(Hz). Figure 4.17 Means of the data fetched in formant setting 5 in vowel space of GWOC (red) and GWC (blue) The vowel space of the two groups in formant setting is 5 is illustrated in figure 4.17.. This figure reveals that the GWC’s vowel space is smaller than GWOC’s as. well, with a vowel space of 281078Hz and 454798Hz respectively. 2. 2. The results are. compelling because in formant setting 5, which is considered to be the standard setting when retrieving sounds, the vowel space of GWC is almost half the size of that of GWOC.. This surprising finding infers that there may have been interference with. the acoustic data in formant setting 5 while calculating the values of F1 and F2. Miscalculating the value of frequencies may interference the reliability of findings (鄭. . 55 .

(68) 靜宜, 2011). Table 4.11 ANOVA analysis of the vowel space of GWOC and GWC in formant setting 5 Source Between groups. SS. dF. MS. 97,276,746,918. 1. 97,276,746,918. Within groups. 277,093,398,420. 12. 23,091,116,535. Total. 374,370,145,338. 13. F. p-Value. 4.21. 0.06. The result of the ANOVA analysis in Table 4.11 reveals another interesting finding that there is no significant difference between GWC and GWOC when it comes to vowel space.. In regards to the third research question, this seems to. indicate that there is no significant difference in terms of vowel space between children with cleft lip and palate and those without.. 4.2.3 Vowel Space of the Two Groups in Formant Setting Judged by the Researcher. The purpose of this sub-section is to examine the vowel space of GWC and GWOC in formant setting judged by the researcher. Figure 4.18 shows the mean values retrieved in formant setting judged by the researcher. detailed statistical analysis of vowel space for the two groups.. . 56 . Table 4.12 presents the.