6
Formant Frequencies of Malay Vowels Produced by Malay Children Aged Between 7 and 12 Years *Hua-Nong Ting, *Alireza Zourmand, *See-Yan Chia, *Boon-Fei Yong, and Badrulzaman Abdul Hamid, *yKuala Lumpur, Malaysia Summary: The formant frequencies of Malaysian Malay children have not been well studied. This article investigates the first four formant frequencies of sustained vowels in 360 Malay children aged between 7 and 12 years using acous- tical analysis. Generally, Malay female children had higher formant frequencies than those of their male counterparts. However, no significant differences in all four formant frequencies were observed between the Malay male and female children in most of the vowels and age groups. Significant differences in all formant frequencies were found across the Malay vowels in both Malay male and female children for all age groups except for F4 in female children aged 12 years. Generally, the Malaysian Malay children showed a nonsystematic decrement in formant frequencies with age. Low levels of significant differences in formant frequencies were observed across the age groups in most of the vowels for F1, F3, and F4 in Malay male children and F1 and F4 in Malay female children. Key Words: Formant frequency–Malay children–Normal voice–Sustained vowels. INTRODUCTION Formant frequencies have been studied in children and adults across different languages such as American English by Peterson and Barney 1 and Hillenbrand et al, 2 Japanese by Homma, 3 Chinese by Andrianopoulos et al 4 and Chen, 5 Polish by Majewski and Hollien, 6 and Russian by Purcell. 7 Fundamental frequency (F 0 ) and perturbation measures of Malaysian Chinese and Malay young adults have been well studied by Ting et al, 8,9 but formant frequency data especially for Malay children in different age groups have not been well established so far. Thus, there is a need to look into the formant frequencies of Malay children. Peterson and Barney 1 studied the F 0 and the first three for- mant frequencies of 10 American English vowels in 33 men, 28 women, and 15 children. They found that children had the highest formant frequencies, whereas men had the lowest for- mant frequencies. The study was extended by Hillenbrand et al 2 in 45 men, 48 women, and 46 children to include the fourth formant frequency and two more American English vowels. They found numerous differences in terms of average frequencies of F1 and F2 compared with the study by Peterson and Barney. But in both of these studies, the age and sex of chil- dren were not specified. Peterson and Barney, 1 Hillenbrand et al, 2 and Bennett 10 re- ported that females had higher formant frequencies than males. These studies, however, did not determine the significant differ- ences in formant frequencies across the genders. Busby and Plant 11 found that there were gender differences in F2 for almost 11 Australian English vowels and in F1 for low vowels. In another study of Swedish children aged 11 years, White 12 re- ported that female children had significantly higher F1 and F2 than male children. However, the gender differences for formant frequencies among Malaysian Malay children are yet to be known. A few studies have investigated formant frequencies in mul- tiple vowels. 1–7,10 In general, they reported that low vowels had higher F1 than that of high vowels. However, most of these studies did not report the significant differences in formant frequencies across different vowels. White 12 reported that highly significant differences were observed across the vowels for F1 and F2 in Swedish children. On the other hand, some re- searchers suggested that F1 and F2 were sufficient to be used for vowel classification, so F 0 or upper formants such as F3 could add no additional advantage. 1–7,13 Potamianos and Narayanan 14 reported that because of the combination of a large acoustic parameter range and increased acoustic variability, the perfor- mance of vowel identification and classification seriously de- graded in children compared with that in adults. Eguchi and Hirsh 15 measured the F 0 and first two formant frequencies of vowels in 84 children aged between 3 and 13 years. Their results showed that the formant frequency values decreased with age. Busby and Plant 11 also reported that, in general, the F1, F2, and F3 values decreased with age. In another study of vowel development in English and Korean, Lee and Iverson 16 showed that the F1 and F2 of children aged 10 years were lower than those of children aged 5 years. In Italian language, a nonsystematic decrease in formants was also reported by Gerosa et al. 17 This trend of decrement in for- mant frequencies with age has not been reported. Besides that, the significant differences in formant frequencies across the age groups among the Malaysian Malay children are still unknown. The aim of this study was to determine the first four formant frequencies of sustained Malay vowels in Malaysian Malay children aged between 7 and 12 years. The study also investi- gates the effects of gender, vowel, and age on the formant fre- quencies among the Malaysian Malay children. METHODS Subjects Three hundred sixty normal Malaysian Malay children aged be- tween 7 and 12 years were selected from primary schools Accepted for publication August 22, 2011. From the *Department of Biomedical Engineering, Faculty of Engineering, University of Malaya, Kuala Lumpur, Malaysia; and the yDepartment of Audiology and Speech Sciences, Faculty of Allied Health Sciences, Universiti Kebangsaan Malaysia, Kuala Lumpur, Malaysia. Address correspondence and reprint requests to Hua-Nong Ting, Department of Biomedical Engineering, Faculty of Engineering, University of Malaya, 50603 Kuala Lumpur, Malaysia. E-mail: [email protected] Journal of Voice, Vol. 26, No. 5, pp. 664.e1-664.e6 0892-1997/$36.00 Ó 2012 The Voice Foundation doi:10.1016/j.jvoice.2011.08.008

Formant Frequencies of Malay Vowels Produced by Malay Children Aged Between 7 and 12 Years

Embed Size (px)

Citation preview

Formant Frequencies of Malay Vowels Produced

by Malay Children Aged Between 7 and 12 Years

*Hua-Nong Ting, *Alireza Zourmand, *See-Yan Chia, *Boon-Fei Yong, and †Badrulzaman Abdul Hamid, *yKualaLumpur, Malaysia

Summary: The formant frequencies of Malaysian Malay children have not been well studied. This article investigates

AccepFrom

of MalaySciencesLumpur,Addre

BiomediLumpur,Journa0892-1� 201doi:10

the first four formant frequencies of sustained vowels in 360 Malay children aged between 7 and 12 years using acous-tical analysis. Generally, Malay female children had higher formant frequencies than those of their male counterparts.However, no significant differences in all four formant frequencies were observed between the Malay male and femalechildren in most of the vowels and age groups. Significant differences in all formant frequencies were found across theMalay vowels in both Malay male and female children for all age groups except for F4 in female children aged 12 years.Generally, the Malaysian Malay children showed a nonsystematic decrement in formant frequencies with age. Lowlevels of significant differences in formant frequencies were observed across the age groups in most of the vowelsfor F1, F3, and F4 in Malay male children and F1 and F4 in Malay female children.Key Words: Formant frequency–Malay children–Normal voice–Sustained vowels.

INTRODUCTION

Formant frequencies have been studied in children and adultsacross different languages such asAmericanEnglish byPetersonand Barney1 and Hillenbrand et al,2 Japanese by Homma,3

Chinese by Andrianopoulos et al4 and Chen,5 Polish byMajewski and Hollien,6 and Russian by Purcell.7 Fundamentalfrequency (F0) and perturbation measures of Malaysian Chineseand Malay young adults have been well studied by Ting et al,8,9

but formant frequency data especially for Malay children indifferent age groups have not been well established so far.Thus, there is a need to look into the formant frequencies ofMalay children.

Peterson and Barney1 studied the F0 and the first three for-mant frequencies of 10 American English vowels in 33 men,28 women, and 15 children. They found that children had thehighest formant frequencies, whereas men had the lowest for-mant frequencies. The study was extended by Hillenbrandet al2 in 45 men, 48 women, and 46 children to include thefourth formant frequency and two more American Englishvowels. They found numerous differences in terms of averagefrequencies of F1 and F2 compared with the study by Petersonand Barney. But in both of these studies, the age and sex of chil-dren were not specified.

Peterson and Barney,1 Hillenbrand et al,2 and Bennett10 re-ported that females had higher formant frequencies than males.These studies, however, did not determine the significant differ-ences in formant frequencies across the genders. Busby andPlant11 found that there were gender differences in F2 foralmost 11 Australian English vowels and in F1 for low vowels.In another study of Swedish children aged 11 years, White12 re-ported that female children had significantly higher F1 and F2

ted for publication August 22, 2011.the *Department of Biomedical Engineering, Faculty of Engineering, Universitya, Kuala Lumpur, Malaysia; and the yDepartment of Audiology and Speech, Faculty of Allied Health Sciences, Universiti Kebangsaan Malaysia, KualaMalaysia.ss correspondence and reprint requests to Hua-Nong Ting, Department ofcal Engineering, Faculty of Engineering, University of Malaya, 50603 KualaMalaysia. E-mail: [email protected] of Voice, Vol. 26, No. 5, pp. 664.e1-664.e6997/$36.002 The Voice Foundation.1016/j.jvoice.2011.08.008

than male children. However, the gender differences forformant frequencies among Malaysian Malay children are yetto be known.A few studies have investigated formant frequencies in mul-

tiple vowels.1–7,10 In general, they reported that low vowels hadhigher F1 than that of high vowels. However, most of thesestudies did not report the significant differences in formantfrequencies across different vowels. White12 reported thathighly significant differences were observed across the vowelsfor F1 and F2 in Swedish children. On the other hand, some re-searchers suggested that F1 and F2were sufficient to be used forvowel classification, so F0 or upper formants such as F3 couldadd no additional advantage.1–7,13 Potamianos and Narayanan14

reported that because of the combination of a large acousticparameter range and increased acoustic variability, the perfor-mance of vowel identification and classification seriously de-graded in children compared with that in adults.Eguchi and Hirsh15 measured the F0 and first two formant

frequencies of vowels in 84 children aged between 3 and13 years. Their results showed that the formant frequencyvalues decreased with age. Busby and Plant11 also reportedthat, in general, the F1, F2, and F3 values decreased with age.In another study of vowel development in English and Korean,Lee and Iverson16 showed that the F1 and F2 of children aged10 years were lower than those of children aged 5 years. InItalian language, a nonsystematic decrease in formants wasalso reported by Gerosa et al.17 This trend of decrement in for-mant frequencies with age has not been reported. Besides that,the significant differences in formant frequencies across the agegroups among the Malaysian Malay children are still unknown.The aim of this study was to determine the first four formant

frequencies of sustained Malay vowels in Malaysian Malaychildren aged between 7 and 12 years. The study also investi-gates the effects of gender, vowel, and age on the formant fre-quencies among the Malaysian Malay children.

METHODS

Subjects

Three hundred sixty normal MalaysianMalay children aged be-tween 7 and 12 years were selected from primary schools

Hua-Nong Ting, et al Formant Frequencies of Malay Vowels Produced by Malay 664.e2

around Petaling Jaya and Kuala Lumpur, Malaysia, for thestudy. Each age group consisted of 30 males and 30 females.None of the subjects selected had vocal pathology or voice dis-orders, symptoms of cold and flu, allergies, history of smoking,neurological disease, and respiratory dysfunctions.

Equipment and procedure

The subjects were shown the list of the vowels, and they wereasked to pronounce sustained vowels of /a/, /e/, /ə/, /i/, /o/,and /u/ for 5 seconds each, at a comfortable pitch and loudnesslevel. The recording was done using Shure SM58 microphone(Shure Brothers Inc., Agua Prieta, Mexico) in a normal roomenvironment. The mouth-to-microphone distance was fixed

TABLE 1.

Mean Value of Formant Frequencies and Standard Deviation (H

Age (y) Vowel

Form

F1 F2

Male Female Male Fem

7 /a/ 815 ± 153 818 ± 193 1473 ± 127 1576 ±

/e/ 613 ± 152 611 ± 138 2343 ± 459 2192 ±

/ə/ 559 ± 115 626 ± 93 1612 ± 185 1659 ±

/i/ 358 ± 80 404 ± 62 2517 ± 628 2543 ±

/o/ 529 ± 95 582 ± 118 1044 ± 178 1141 ±

/u/ 419 ± 79 463 ± 51 987 ± 387 986 ±

8 /a/ 788 ± 150 825 ± 222 1486 ± 116 1535 ±

/e/ 548 ± 96 589 ± 114 2363 ± 569 2388 ±

/ə/ 528 ± 77 559 ± 97 1654 ± 141 1682 ±

/i/ 365 ± 51 359 ± 54 2669 ± 480 2296 ±

/o/ 503 ± 90 529 ± 81 1018 ± 142 1078 ±

/u/ 430 ± 52 466 ± 69 981 ± 326 1078 ±

9 /a/ 770 ± 161 774 ± 151 1471 ± 106 1523 ±

/e/ 566 ± 101 559 ± 104 2311 ± 575 2346 ±

/ə/ 529 ± 74 536 ± 69 1659 ± 151 1606 ±

/i/ 362 ± 54 361 ± 62 2631 ± 529 2479 ±

/o/ 499 ± 78 505 ± 84 1079 ± 167 1041 ±

/u/ 415 ± 61 438 ± 47 1021 ± 323 894 ±

10 /a/ 746 ± 138 820 ± 183 1450 ± 93 1567 ±

/e/ 527 ± 92 590 ± 110 2452 ± 354 2365 ±

/ə/ 520 ± 81 549 ± 75 1594 ± 163 1635 ±

/i/ 354 ± 58 363 ± 60 2684 ± 352 2488 ±

/o/ 472 ± 71 516 ± 82 999 ± 128 1086 ±

/u/ 404 ± 55 431 ± 62 925 ± 333 983 ±

11 /a/ 727 ± 163 746 ± 177 1482 ± 143 1499 ±

/e/ 537 ± 106 526 ± 95 2306 ± 357 2494 ±

/ə/ 502 ± 87 532 ± 69 1603 ± 180 1645 ±

/i/ 344 ± 69 335 ± 46 2597 ± 358 2369 ±

/o/ 465 ± 68 483 ± 77 1003 ± 150 1048 ±

/u/ 391 ± 67 403 ± 36 894 ± 231 1008 ±

12 /a/ 648 ± 113 739 ± 141 1358 ± 97 1482 ±

/e/ 492 ± 82 520 ± 79 2279 ± 326 2357 ±

/ə/ 458 ± 50 514 ± 58 1518 ± 156 1703 ±

/i/ 333 ± 45 344 ± 52 2626 ± 245 2396 ±

/o/ 448 ± 51 459 ± 73 944 ± 115 1019 ±

/u/ 387 ± 45 417 ± 40 895 ± 298 939 ±

at 2–3 cm. The speech sounds were digitally recorded byGoldWave (St. John’s, Newfoundland, Canada)18 digital audioeditor at a sampling rate of 20 kHz with 16-bit resolution.

A discrimination test was carried out to check the pronunci-ation of the vowels by the children. The listeners were asked toidentify the vowels based on the six Malay vowels. The pronun-ciation of the vowels was considered correct if seven of the10 listeners identified them correctly. The Praat (Boersmaand Weenink, University of Amsterdam, The Netherlands)19

software was used to determine the average values of the firstfour formant frequencies of the sustained vowels. Standardformant settings were used: 5500 Hz for maximum formantfrequency, five numbers of formants, 25 milliseconds for the

z) of Malay Children in Different Age Groups

ant Frequency (Hz)

F3 F4

ale Male Female Male Female

164 3303 ± 340 3252 ± 456 4492 ± 291 4436 ± 320

609 3333 ± 307 3423 ± 273 4408 ± 377 4434 ± 351

256 3184 ± 507 3244 ± 499 4555 ± 424 4588 ± 350

651 3618 ± 273 3683 ± 280 4475 ± 337 4540 ± 255

183 3179 ± 400 2933 ± 632 4343 ± 299 4372 ± 349

244 2998 ± 316 2969 ± 514 4286 ± 357 4211 ± 378

118 3389 ± 179 3171 ± 486 4519 ± 215 4454 ± 297

569 3383 ± 289 3418 ± 307 4435 ± 289 4391 ± 352

200 3354 ± 184 3223 ± 412 4669 ± 239 4550 ± 370

711 3691 ± 241 3563 ± 319 4551 ± 267 4447 ± 294

110 3211 ± 353 3118 ± 434 4385 ± 336 4351 ± 327

383 3105 ± 250 2980 ± 316 4246 ± 280 4201 ± 335

159 3270 ± 307 3277 ± 389 4427 ± 354 4461 ± 234

469 3288 ± 290 3408 ± 223 4307 ± 385 4423 ± 287

213 3260 ± 210 3216 ± 471 4537 ± 242 4588 ± 334

626 3617 ± 263 3679 ± 253 4526 ± 277 4504 ± 280

139 3233 ± 285 3218 ± 371 4387 ± 280 4398 ± 314

204 3060 ± 240 2970 ± 289 4372 ± 295 4192 ± 322

138 3269 ± 231 3365 ± 304 4365 ± 247 4538 ± 247

447 3309 ± 211 3344 ± 217 4380 ± 316 4386 ± 326

208 3213 ± 309 3175 ± 296 4461 ± 321 4562 ± 308

618 3593 ± 257 3613 ± 214 4496 ± 262 4549 ± 271

119 3137 ± 270 3263 ± 346 4274 ± 321 4445 ± 279

310 2952 ± 229 3117 ± 415 4194 ± 281 4379 ± 346

105 3136 ± 353 3194 ± 304 4371 ± 298 4329 ± 290

341 3282 ± 182 3327 ± 220 4421 ± 241 4464 ± 307

143 3182 ± 255 3219 ± 206 4475 ± 312 4554 ± 304

679 3553 ± 204 3503 ± 277 4460 ± 280 4443 ± 332

156 3157 ± 266 3209 ± 234 4324 ± 345 4361 ± 316

294 3034 ± 277 3001 ± 258 4247 ± 304 4237 ± 316

119 2979 ± 365 3222 ± 211 4138 ± 300 4260 ± 250

423 3054 ± 231 3201 ± 180 4214 ± 270 4311 ± 307

206 3043 ± 229 3101 ± 205 4360 ± 273 4410 ± 257

644 3364 ± 260 3463 ± 270 4299 ± 278 4363 ± 304

148 2980 ± 243 3116 ± 200 4114 ± 278 4245 ± 302

310 2921 ± 193 2947 ± 241 4150 ± 288 4203 ± 351

TABLE 2.

The Mean and Standard Deviation (Hz) of Malay Children’s Formant Frequencies by Vowels and Age Groups

Grouping

Formant Frequency (Hz)

F1 F2 F3 F4

Male Female Male Female Male Female Male Female

Overall 510 ± 88 536 ± 92 1665 ± 269 1670 ± 317 3240 ± 272 3254 ± 320 4380 ± 299 4405 ± 310

Vowel

/a/ 749 ± 146 787 ± 178 1454 ± 114 1530 ± 134 3225 ± 296 3247 ± 359 4385 ± 284 4413 ± 273

/e/ 547 ± 105 566 ± 107 2342 ± 440 2357 ± 476 3275 ± 252 3354 ± 237 4361 ± 313 4401 ± 322

/ə/ 516 ± 81 553 ± 77 1607 ± 163 1655 ± 204 3206 ± 283 3197 ± 348 4509 ± 302 4542 ± 320

/i/ 353 ± 59 361 ± 56 2621 ± 432 2428 ± 655 3573 ± 250 3584 ± 269 4468 ± 283 4474 ± 289

/o/ 489 ± 76 512 ± 86 1015 ± 147 1069 ± 142 3150 ± 303 3143 ± 369 4305 ± 310 4362 ± 314

/u/ 408 ± 60 436 ± 51 950 ± 316 981 ± 291 3012 ± 250 2997 ± 339 4249 ± 301 4237 ± 341

Age (y)

7 549 ± 112 584 ± 109 1663 ± 327 1683 ± 351 3269 ± 357 4427 ± 347 4427 ± 347 4430 ± 334

8 527 ± 86 555 ± 106 1695 ± 296 1676 ± 348 3356 ± 249 4468 ± 271 4468 ± 271 4399 ± 329

9 524 ± 88 529 ± 86 1695 ± 309 1648 ± 302 3288 ± 266 4426 ± 305 4426 ± 305 4428 ± 295

10 504 ± 82 545 ± 95 1684 ± 237 1687 ± 307 3245 ± 251 4362 ± 291 4362 ± 291 4476 ± 296

11 494 ± 93 504 ± 83 1648 ± 236 1677 ± 286 3224 ± 256 4383 ± 297 4383 ± 297 4398 ± 311

12 461 ± 64 499 ± 74 1603 ± 206 1649 ± 308 3057 ± 254 4212 ± 281 4212 ± 281 4299 ± 295

3500

4000

4500

5000

z)

Male F1

Female F1

Male F2

Female F2

Male F3Female F3

Male F4

Journal of Voice, Vol. 26, No. 5, 2012664.e3

window length, and a dynamic range of 30 dB. The averagevalue of all the first three formant frequencies could be obtainedin every vowel. However, in some vowels, the fourth formantfrequency was undefined at certain segments of time. Neverthe-less, the average fourth formant frequency value was still calcu-lated based on the available fourth formant frequency values inthe vowel.

Statistical analysis was carried out using SPSS Statistics 17.0(SPSS, Inc., Chicago, IL)20 to determine the significant differ-ences in formant frequencies across the vowels, genders, andage groups. Independent sample t tests were used to determinethe group mean differences in formant frequencies betweengenders for every individual vowel. One-way analysis of vari-ance (ANOVA) with Tukey post hoc analysis was used to deter-mine the groups’ mean differences in formant frequency acrossthe vowels and age groups. The significant level for indepen-dent sample t test, ANOVA, and Tukey post hoc test was setat a¼ .05.

0

500

1000

1500

2000

2500

3000

7 8 9 10 11 12Age Group (Years old)

Fre

qu

en

cy (H Female F4

FIGURE 1. The mean formant frequencies by age groups for both

Malay male and female children.

RESULTS

Formant frequencies

Table 1 shows the mean formant frequencies of Malaysian Ma-lay children aged between 7 and 12 years in six Malay vowels.The overall means of F1–F4 were 510 ± 88, 1665 ± 269,3240 ± 272, and 4380 ± 299 Hz, respectively, for males and536 ± 92, 1670 ± 317, 3254 ± 320, and 4405 ± 310 Hz, respec-tively, for females. The highest F1 was observed for /a/,whereas the lowest F1 was observed for /i/. On the otherhand, /u/ had the lowest F2 in all the age groups and /i/ hadthe highest F2, except females aged 8 and 11 years. In F3, /i/had the highest value in males and females and /u/ had the low-est value except in females aged 7 years. Themean and standarddeviation of the formant frequencies by vowels and age groups

are shown in Table 2. The effect of the age group and vowel onthe formants for both Malay male and female children is illus-trated in Figures 1 and 2, respectively.

Significant differences across genders

In general, formant frequencies of Malaysian Malay femalechildren were higher than those in males except in certainvowels in some age groups. Table 3 shows the significant levelsin formant frequencies across the genders in different vowels

TABLE 3.

Levels of Significance for Formant Frequency Differences

Across Gender, by Age Group and Vowel

Age (y) Vowel

Significant Levels

F1 F2 F3 F4

7 /a/ 0.946 0.009 0.620 0.481

/e/ 0.962 0.283 0.233 0.788

/ə/ 0.017 0.414 0.645 0.746

/i/ 0.014 0.874 0.371 0.406

/o/ 0.061 0.042 0.077 0.732

/u/ 0.013 0.983 0.793 0.428

8 /a/ 0.449 0.113 0.025 0.333

/e/ 0.135 0.866 0.654 0.597

/ə/ 0.168 0.545 0.117 0.144

/i/ 0.644 0.020 0.086 0.156

/o/ 0.257 0.071 0.363 0.692

/u/ 0.028 0.296 0.093 0.576

9 /a/ 0.930 0.149 0.942 0.663

/e/ 0.813 0.795 0.077 0.189

/ə/ 0.726 0.269 0.648 0.101

/i/ 0.984 0.313 0.358 0.755

/o/ 0.770 0.345 0.865 0.892

/u/ 0.121 0.075 0.199 0.027

10 /a/ 0.080 0.000 0.172 0.009

/e/ 0.019 0.410 0.531 0.948

/ə/ 0.155 0.398 0.634 0.144

/i/ 0.550 0.138 0.735 0.445

/o/ 0.032 0.009 0.120 0.032

/u/ 0.074 0.489 0.062 0.027

11 /a/ 0.666 0.599 0.503 0.582

/e/ 0.673 0.041 0.387 0.549

/ə/ 0.149 0.322 0.545 0.324

/i/ 0.571 0.108 0.428 0.832

/o/ 0.322 0.259 0.419 0.659

/u/ 0.408 0.102 0.635 0.896

12 /a/ 0.008 0.000 0.003 0.091

/e/ 0.184 0.425 0.008 0.201

/ə/ 0.000 0.000 0.304 0.469

/i/ 0.399 0.073 0.154 0.392

/o/ 0.528 0.032 0.021 0.084

/u/ 0.007 0.574 0.640 0.527

The bold values indicate that the P values are less than 0.05 which indi-

cates the significant gender and age differences.

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

/a/ /e/ / / /i/ /o/ /u/

Vowel

Fre

qu

en

cy

(H

z)

Male F1

Male F2

Female F2

Female F1

Male F3

Female F3

Male F4

Female F4

FIGURE 2. Themean formant frequencies by vowels for bothMalay

male and female children.

Hua-Nong Ting, et al Formant Frequencies of Malay Vowels Produced by Malay 664.e4

and age groups using independent sample t test analysis. Gen-erally, no significant difference between the genders wereobserved in most of the vowels in all age groups except F1 inthe age groups of 7 and 12 years, F2 and F3 in the age groupof 12 years, and F4 in the age group of 10 years. Significant gen-der difference was observed in more vowels in Malay childrenaged 12 years than those in other age groups.

Significant differences across vowels

Significant differences in all formant frequencies were ob-served across the Malay vowels in both Malay males andfemales for all age groups except F4 in female children aged12 years. F1 and F2 had stronger significant differences acrossthe vowels than F3 and F4. The post hoc test analysis showedthat significant differences in formant frequencies were ob-served between the vowel pairs, mostly for F1 and F2 comparedwith F3 and F4, for both genders in each age group.

Significant differences across age groups

Table 4 shows the F number and significant level in formant fre-quencies across the age groups for six Malay vowels. Signifi-cant differences in formant frequencies across the age groupswere observed in most of the vowels for F1, F3, and F4 inMalaymale children and for F1 and F4 in Malay female children.However, the significance was not strong because the F num-

bers were less than 10. The post hoc test analysis showedthat significant differences in formant frequencies across theage groups were found in some vowel pairs in certain agegroups only.

DISCUSSIONS

Malaysian Malay female children had higher formant frequen-cies than male children, in general. This result is in agreementwith the studies of Peterson and Barney,1 Hillenbrand et al,2

Busby and Plant,11 and White.12 Based on the independentsample t test, there were no significant differences in all four

TABLE4.

FNumbers

andPValuesin

Form

ants

AcrosstheAgeGroupsforSix

MalayVowels

Vowel

F1

F2

F3

F4

Male

Female

Male

Female

Male

Female

Male

Female

FNumber

PValue

FNumber

PValue

FNumber

PValue

FNumber

PValue

FNumber

PValue

FNumber

PValue

FNumber

PValue

FNumber

PValue

/a/

4.728

0.000

1.423

0.218

5.319

0.000

2.210

0.055

6.845

0.000

1.050

0.390

6.755

0.000

4.006

0.002

/e/

4.293

0.001

3.571

0.004

0.550

0.738

1.200

0.311

5.988

0.000

3.724

0.003

2.162

0.060

0.809

0.545

/ə/

5.018

0.000

7.467

0.000

2.951

0.014

0.828

0.531

3.453

0.005

0.596

0.703

3.589

0.004

1.802

0.115

/i/

1.171

0.325

5.306

0.000

0.527

0.756

0.581

0.714

5.944

0.000

3.377

0.006

2.955

0.014

1.761

0.123

/o/

4.429

0.001

6.977

0.000

2.819

0.018

2.648

0.025

2.571

0.028

2.685

0.023

3.263

0.008

1.331

0.253

/u/

2.313

0.046

6.880

0.000

0.825

0.533

1.324

0.256

2.204

0.056

0.903

0.480

1.930

0.092

1.296

0.268

Thebold

valuesindicate

thatthePvaluesare

lessthan0.05whichindicatesthesignificantgenderandagedifferences.

FIGURE 3. Ratio of female-to-male formant frequencies (K-factors,

in percent) across the Malay vowels.

Journal of Voice, Vol. 26, No. 5, 2012664.e5

formant frequencies between males and females. But some sig-nificant gender differences in certain vowels and formants wereobserved in each age group. This observation was also similarto the previous studies.Statistical analyses showed that there were significant differ-

ences in formant frequencies across the Malay vowels in mostof the Malaysian Malay children. The significant differenceswere stronger in F1 and F2 compared with those in F3 andF4. White12 also reported that there were strong significant dif-ferences in the first two formants across the vowels among theSwedish children. The study also suggested that the use of F1and F2 was enough for vowel identification or classification.Generally, the Malaysian Malay children experienced a dec-

rement in all formant frequencies with age. The result was inagreement with most of the studies of Busby and Plant,11

Eguchi and Hirsh,15 and Gerosa et al.17 However, the decrementin formant frequencies was not systematic, with some of theformant frequencies being higher between the ages of 7 and12 years.Fant21 used the formant scale factors of K-factors to describe

the relationship between the formant frequency values of malesand females. Figures 3 and 4 show the formant scale factors

FIGURE 4. Ratio of female-to-male formant frequencies (K-factors,

in percent) across the age groups.

Hua-Nong Ting, et al Formant Frequencies of Malay Vowels Produced by Malay 664.e6

across the Malay vowels and the age groups, respectively.Generally, the four K-factors for each vowel and each agegroup were less than 9% for Malaysian Malay children. Thereported percentages were much lower than those in thestudies of Bennett,10 White,12 and Eguchi and Hirsh.15 The pre-vious studies reported all positive values for the K-factors; how-ever, some of the Malay male children had higher formantfrequencies than females, which resulted in negative valuesfor some K-factors. This was very obvious in /i/ for F2.

CONCLUSIONS

The study investigated the first four formant frequencies of sus-tained vowels in normal Malaysian Malay children aged be-tween 7 and 12 years. The results showed that, generally,Malay female children had higher formant frequencies thantheir male counterparts. However, no significant differencesin all four formant frequencies were observed between theMalay male and female children in most of the vowels andage groups. Significant differences in all formant frequencieswere found across the Malay vowels in both Malay male andfemale children for all age groups except for F4 in female chil-dren aged 12 years. Generally, the Malaysian Malay childrenshowed a nonsystematic decrement in formant frequencieswith age. Low levels of significant differences in formant fre-quencies were observed across the age groups in most of thevowels for F1, F3, and F4 in Malay male children and for F1and F4 in Malay female children.

Acknowledgments

The authors thank the Ministry of Science, Technology andInnovation, Malaysia, for supporting this research under theScience Fund project (06-01-03-SF0516).

REFERENCES1. Peterson GE, Barney HL. Control methods used in a study of the vowels.

J Acoust Soc Am. 1952;24:175–184.

2. Hillenbrand J, Getty LA, Clark MJ, Wheeler K. Acoustic characteristics of

American English vowels. J Acoust Soc Am. 1995;97:3099–3111.

3. Homma Y. An acoustic study of Japanese vowels: their quality, pitch, am-

plitude, and duration. Stud Sounds. 1973;16:347–368.

4. Andrianopoulos MV, Darrow K, Chen J. Multimodal standardization of

voice among four multicultural populations’ formant structures. J Voice.

2001;15:61–77.

5. Chen YY. The acoustic realization of vowels of Shanghai Chinese. J Phon.

2008;36:629–648.

6. Majewski W, Hollien H. Formant frequency regions of Polish vowels.

J Acoust Soc Am. 1967;42:1031–1037.

7. Purcell ET. Formant frequency patterns in Russian VCV utterances.

J Acoust Soc Am. 1979;66:1691–1702.

8. Ting HN, Chia SY, Abdul Hamid B, Sarah Mukari SZ. Acoustic character-

istics of vowels by normal Malaysian Malay young adults. J Voice. 2011;

25:e305–e309.

9. Ting HN, Chia SY, Kim KS, Sim SL, Abdul Hamid B. Vocal fundamental

frequency and perturbation measurements of vowels by normal Malaysian

Chinese adults. 2010. J Voice. In press.

10. Bennett S. Vowel formant frequency characteristics of preadolescent males

and females. J Acoust Soc Am. 1981;69:231–238.

11. Busby PA, Plant GL. Formant frequency values of vowels produced by pre-

adolescent boys and grils. J Acoust Soc Am. 1995;97:2603–2606.

12. White P. Formant frequency analysis of children’s spoken and sung vowels

using sweeping fundamental frequency production. J Voice. 1999;13:

570–582.

13. Sakayori S, Kitama T, Chimoto S, Qin L, Sato Y. Critical spectral regions

for vowel identification. Neurosci Res. 2002;43:155–162.

14. Potamianos A, Narayanan S. A review of the acoustic and linguistic prop-

erties of children’s speech. Multimedia Signal Processing, 2007. MMSP

2007. IEEE. 2007;22–25.

15. Eguchi S, Hirsh U. Development of speech sounds in children. Acta Otolar-

yngol Suppl. 1969;257:1–51.

16. Lee S, Iverson GK. Vowel development in English and Korean: similarities

and differences in linguistic and non-linguistic factors. Speech Comm.

2009;51:648–694.

17. Gerosa M, Giuliani D, Brugnara F. Acoustic variability and automatic rec-

ognition of children’s speech. Speech Comm. 2007;49:847–860.

18. GoldWave Inc [CD-ROM]. Release v5.55; 2009.

19. Boersma P, Weenink D. Praat: Doing Phonetics by Computer [computer

software and manual]. Amsterdam, The Netherlands: University of Am-

sterdam; 2009. Available at: http://www.fon.hum.uva.nl/praat/. Accessed

May 11, 2009.

20. SPSS Inc [CD-ROM]. Release 17.0.0; 2008.

21. Fant G. Non-Uniform Vowel Normalization. Speech Transmission Labora-

tory Quarterly Progress Status Report. Stockholm, Sweden: Royal Institute

of Technology; 1975. pp.1–19.