kiyouh 19 1 031to057

TOMITA

Kaoru

Engl

i

sh

Phoneti

cs

0.Introduction

Phonological and phonetic features of speakers’ native languages affect their foreign language pronunciation. The effects last long so even for adult learners of advanced levels, making speech without foreign accents is not an easy task. Lenneberg (1967) proposes an important biological process of language learning and calls it "a critical period.” After the criticalperiod,it becomes difficult for people to acquire languages.

Learning foreign languages before or after the critical period is a controversial issue. Learners who start learning foreign languages after the critical period tend to show errors ofpronunciation on which their native language have a significant effect.

Foreign accented speech by Japanese advanced-level learners of English is observed in general societies, language classes and even academic meetings. It is caused by slight mistakes on both prosodic features, such as stress and intonation, and segmental features, such as consonants and vowels. Among these features, the vowel is focused on for the language experiment ofthis study.

English and Japanese have different vowel systems. English has eleven vowels and Japanese has five vowels. For example, English contrasts two vowels ([i] vs. [ɪ]) with vowel quality differences and Japanese contrasts them with short or long vowel ([i:] vs. [i]) but not with vowelquality differences.

Spoken languages are managed by people with supports ofwritten forms.They can be visualized with types of symbols, such as their corresponding letters or phonetic alphabets. On the basis of this multilayered process of spoken languages, visualization of vowel qualities is explored for linguistic and language learning studies.Language learners are able to produce spoken forms with reading these symbols orally. Reading the symbols orally, however,does not always lead learners to accurate pronunciation.

To remedy learners’ pronunciation, several tools are used. One of them is visualization of tongue shape with X-rays.This has been developed and has been made use offor a long time

in a field of speech science. As Trofimov and Daniel (1923, 25) points out, however, X ray pictures willnot ofitselfenable the students to pronounce the sound correctly.

Visualization of language sounds and its effect for learning pronunciation of foreign language has been discussed by limited number of researchers. Among them, one of the clearest results presented by Suemitsu et al. (2015, 6) concludes that short-term training with real-time articulatory visualfeedback approach improves the pronunciation ofJapanese learners in acquiring the non-native vowel[１].

Visualized vowels may not work for leaning language sounds ifthey are just figures with elusive scattering ofdots.They are,in a way,a picture ofa state that is sliced from a series of motion in three dimensions with axes of F1: openness of mouth, F2: tongue height, and time: motion of a vocal organ. As Cohn (2013, 3) explains, structured sequential sounds become spoken languages of the world, structured sequential body motions become sign languages,and structured sequentialimages literally become visuallanguages.

1.Methods

Recording of listed words and phonetic analyses of their vowels were conducted in a phonetic laboratory of Yamagata University. Formant frequencies were used for a measurement of vowel qualities. It is expected that vowels are not discriminated very well by non-native English speakers.

1.1 Participants

Three male speakers of American English (hereafter ME1, ME2 and ME3), three female speakers ofAmerican English (hereafter FE1,FE2 and FE3)participated in the experiment. They came from U.S.A.as exchange students with one year term.They were from 20 to 24 years old.

Two male Japanese students who majored in English (hereafter MJ1 and MJ2) and eight female Japanese students who majored in English (hereafter FJ1,FJ2,FJ3,FJ4,FJ5,FJ6,FJ7 and FJ8)took part in the experiment.They were from 20 to 22 years old.

1.2 Materials

were used for recording.A one-syllable word that begins with [h]and ends with [d],between which [u]was put in was not found,and so a two-syllable word,"hoodoo” was used instead.

1.3 Acoustic measurements

Files recorded in media (San Disk Extreme IV compact Flash UDMA) were put into a personal computer (Panasonic CF-W7CWU1JC). Six vowels, [i], [ɪ], [１], [a], [ʊ] or [u], were selected for measurements. Vowel analyses measured using Praat were conducted while focusing on formants,which were concentrations ofacoustic energy and the most dominant frequencies combined to produce the distinctive vowelqualities.F1,a reflection ofthe height of the tongue, and F2, a reflection of the location of the tongue that was the highest in production ofa vowel,were measured in Hertz.

1.4 Statisticalanalysis

2.Results

2.1 Phonemic contrasts

2.1.1 Minimalpairs produced by native speakers

Formant 1 and 2 values of six vowels by native speakers of American English are measured and listed in Table 1-3.

Table 1 Voweldiscrimination for[iː] and [ɪ] by native speakers

Comparison P

Z Mean N

hidS.D. heedS.D.

EM1

heed<hid <.000

-4.63 418

30 50964　

32860　 F1

hid<heed <.000

-4.26 2450

30 2292207

2608183 F2

EM2

heed<hid <.000

-4.78 343

30 41331　

27425　 F1

hid<heed <.000

-4.16 2530

30 2437111

2624133 F2

EM3

heed<hid <.000

-4.78 320

30 39933　

24119　 F1

hid<heed <.000

-4.78 2118

30 197975　

225853　 F2

EF1

heed<hid <.000

-4.37 412

30 48685　

33866　 F1

hid<heed <.000

-4.45 2407

30 2156177

2659376 F2

EF2

heed<hid <.000

-4.78 442

30 52520　

36024　 F1

hid<heed <.000

-4.78 2458

30 214374　

277334　 F2

EF3

heed<hid <.000

-4.78 499

30 58358　

41537　 F1

hid<heed <.000

-4.78 2590

30 227397　

290879　 F2

Table 2 Voweldiscrimination for[æ] and [ɑː] by native speakers

Comparison P

Z Mean N

hodS.D. hadS.D.

EM1

hod<had <.000

-3.49 731

30 71229　

75138　 F1

hod<had <.000

-4.78 1424

30 119145　

165865　 F2

EM2

hod<had <.000

-4.65 737

30 597136

87751　 F1

hod<had <.000

-4.78 1609

30 1046150

2173103 F2

EM3

had,hod NS

-0.73 650

30 65435　

64640　 F1

hod<had <.000

-4.78 1381

30 105385　

170875　 F2

EF1

hod<had <.000

-4.43 986

30 888102

1084100 F1

hod<had <.000

-4.35 1732

30 1553199

As is presented in Figure 1-3, vowel qualities by native-speakers are discriminated very well except two cases, which are produced by EF2 and EF3. In both cases, F2 values of "hoodoo” and "hood” do not show a significant difference statistically.

EF2

hod<had <.000

-4.78 775

30 74627　

81540　 F1

hod<had <.000

-4.78 1499

30 129180　

170669　 F2

EF3

hod<had <.000

-4.78 893

30 82145　

96567　 F1

hod<had <.000

-4.78 1400

30 117470　

162678　 F2

Table 3 Voweldiscrimination for[uː] and [ʊ] by native speakers

Comparison P

Z Mean N

hoodS.D. hoodooS.D.

EM1

hoodoo<hood <.000

-4.78 444

30 49026　

39628　 F1

hoodoo<hood <.002

-3.14 1362

30 1515123

1210390 F2

EM2

hoodoo<hood <.000

-4.47 410

30 47452　

34585　 F1

hoodoo<hood <.008

-2.66 1610

30 1684233

1535315 F2

EM3

hoodoo<hood <.002

-3.15 385

30 42521　

34591　 F1

hoodoo<hood <.000

-3.54 1533

30 1413212

1654254 F2

EF1

hoodoo<hood <.000

-4.78 515

30 61432　

41653　 F1

hoodoo<hood <.000

-4.06 1509

30 1664166

1355139 F2

EF2

hoodoo<hood <.000

-4.78 474

30 54919　

40021　 F1

hoodoo,hood NS

-0.751 1717

30 173262　

1703164 F2

EF3

hoodoo<hood <.000

-4.78 506

30 61632　

39636　 F1

hoodoo,hood NS

-0.73 1600

30 1621210

2.1.2 Minimalpairs produced by non-native speakers

Formant 1 and 2 values of six vowels by Japanese learners of English are measured and listed in Table 4-6.

Table 4 Voweldiscrimination for[iː] and [ɪ] by non-native speakers

Comparison P

Z Mean N

hidS.D. heedS.D.

JM1

heed<hid <.000

-4.17 372

30 39124　

35718　 F1

hid,heed NS

-1.53 2190

30 2177155

2203116 F2

JM2

heed<hid <.000

-4.75 305

30 32724　

28213　 F1

hid<heed <.002

-3.15 2062

30 202262　

2102160 F2

JF1

heed<hid <.000

-3.89 385

30 41547　

35534　 F1

hid,heed NS

-1.01 2762

30 277272　

2751654 F2

JF2

hid,heed NS

-1.57 425

30 41171　

43940　 F1

hid,heed NS

-0.71 2852

30 2791449

2913365 F2

JF3

heed<hid <.001

-3.39 361

30 37927　

34334　 F1

hid<heed <.010

-2.4 2932

30 291562　

294856　 F2

JF4

heed,hid NS

-1.28 387

30 39226　

38140　 F1

hid<heed <.000

-4.78 2895

30 280472　

298656　 F2

JF5

heed<hid <.001

-3.27 497

30 51238　

48210　 F1

hid<heed <.000

-3.54 2837

30 280276　

2873165 F2

JF6

heed<hid <.004

-2.84 486

30 50762　

46533　 F1

hid<heed <.000

-4.42 2620

30 2483248

2757186 F2

JF7

heed<hid <.000

-4.57 413

30 44534　

38122　 F1

hid<heed <.000

-4.41 2870

30 281977　

292133　 F2

JF8

heed<hid <.006

-2.72 419

30 43126　

40735　 F1

hid<heed <.000

-4.72 2821

30 274573　

Table 5 Voweldiscrimination for[æ] and [ɑː] by non-native speakers

Comparison P

Z Mean N

hodS.D. hadS.D.

JM1

hod,had NS

-1.9 674

30 66245　

68638　 F1

hod<had <.000

-4.78 1348

30 1172104

152370　 F2

JM2

had<hod <.000

-4.48 460

30 　 49242　 42721　

F1

had<hod <.000

-4.78 1359

30 113342　

158495　 F2

JF1

hod<had <.000

-4.49 819

30 72379　

91481　 F1

hod,had NS

-1.53 2190

30 2177115

2203116 F2

JF2

hod<had <.000

-4.78 801

30 75538　

84729　 F1

hod<had <.000

-4.7 1357

30 127548　

143871　 F2

JF3

hod<had <.000

-3.94 774

30 72955　

81953　 F1

hod<had <.000

-4.78 1393

30 118672　

159986　 F2

JF4

hod<had NS

-1.56 748

30 722141

77311　 F1

hod<had <.000

-4.55 1523

30 1382150

166485　 F2

JF5

hod<had <.000

-4.27 795

30 74359　

84875　 F1

hod<had <.000

-4.22 1423

30 　 131956　 1528180

F2

JF6

had<hod <.000

-3.6 763

30 79047　

73769　 F1

hod<had <.000

-4.78 1520

30 120399　

1838245 F2

JF7

hod<had <.000

-4.76 902

30 831125

97462　 F1

hod<had <.015

-2.45 1122

30 108294　

1162149 F2

JF8

hod<had <.000

-4.63 744

30 66944　

81870　 F1

hod<had <.000

-4.78 1514

30 123692　

As is presented in Table 4-6,vowelqualities by nonnative-speakers are not discriminated very well. Among 60 cases, 12 cases do not show clear discrimination between minimal -paired vowels. As for "heed” and "hid”, three cases of F1 and one case of F2 do not show a significant difference between minimal-paired vowels statistically. As for "hod” and "had”, one case of F1 and one case of F2 do not show a significant difference between minimal -paired vowels statistically.As for "hoodoo” and "hood”,two cases ofF1 and three cases ofF2 do not show a significant difference between minimal-paired vowels statistically.

Table 6 Voweldiscrimination for[uː] and [ʊ] by non-native speakers

Comparison P

Z Mean N

hoodS.D. hoodooS.D.

JM1

hoodoo<hood <.000

-4.61 410

30 43723　

38216　 F1

hood<hoodoo <.000

-3.73 1852

30 1601222

1902180 F2

JM2

hoodoo<hood <.000

-4.78 341

30 36016　

32215　 F1

hoodoo<hood <.000

-3.87 269

30 130074　

123842　 F2

JF1

hoodoo<hood <.014

-2.45 428

30 41633　

44029　 F1

hoodoo,hood NS

-0.483 1201

30 1231201

1171185 F2

JF2

hoodoo<hood <.000

-3.84 510

30 53041　

49027　 F1

hood<hoodoo <.017

-2.38 1675

30 165484　

170576　 F2

JF3

hoodoo<hood <.000

-3.67 368

30 38230　

35420　 F1

hoodoo,hood NS

-1.37 1587

30 1656256

1519271 F2

JF4

hoodoo<hood <.000

-4.78 391

30 43833　

34324　 F1

hoodoo<hood <.000

-4.64 1162

30 1298137

1026126 F2

JF5

hood,hoodoo NS

-1.92 519

30 51134　

52826　 F1

hood<hoodoo <.000

-4.65 1758

30 165080　

1867133 F2

JF6

hoodoo<hood <.000

-3.53 449

30 47254　

4279　　 F1

hood,hoodoo NS

-0.638 1678

30 169190　

166676　 F2

JF7

hoodoo,hood <.000

-0.292 366

30 36630　

36625　 F1

hood<hoodoo <.040

-2.05 1214

30 117730　

1251167 F2

JF8

hoodoo<hood <.000

-4.52 436

30 45822　

41516　 F1

hoodoo<hood <.002

-3.07 1735

30 1828269

2.1.3 Visualization ofvowelqualities

Vowel discrimination for vowels by native-and nonnative-speakers is visualized with vowel spaces. There are six dots whose location is calculated with using perceptual adjusting (Story, et al. 2017, 460). Figure 1-6 presents vowel spaces by native speakers in which the scores offormant values are normalized with Bark factor equation.

These figures present various patterns ofvowelspaces.They are forming vowelspaces in which all vowels scatters very well, and that means these six vowels are produced with discrimination that also work for perception.

Figure 7-16 presents vowelspaces by nonnative speakers in which the scores offormant values are normalized with Bark factor equation.

Figure 5 Vowelspace by English female speaker2

These figures present severalpatterns ofvowelspaces,most ofwhose two-paired vowels are not separated enough. These tendencies are quite different from native English speakers’ones.

Among 30 minimal-paired vowels,21 minimalpairs are not discriminated very well.As for [i] – [ɪ] pairs, all 10 pairs are not discriminated. As for [１] – [ɑ] pairs, three pairs are not discriminated.As for [ʊ]– [u]pairs,eight pairs are not discriminated.

2.2 Phonetic contrasts

2.2.1 Differences on contexts produced by native speakers

Differences on contexts by native-speakers are presented in Table 7-9. Against the author's expectation,phonetic contrasts with F1 or F2 on contexts are not observed clearly even for utterances by native speakers.

Table 7 presents two cases, which discriminate F1 and F2 of "hid” on all three types of contexts:"said” in "What did you say?”,"did” in "Did you say …?”,and "sAid” in "Did you say …?”.Among 24 cases,13 cases show hyper-articulation ofvowels on one type ofcontext;six

Figure 15 Vowelspace by Japanese female speaker7

for interrogatives,four for repetitive statements,three for first statements.

Table continued

Table 7 Vowelqualities of[iː] and [ɪ] by native English speakers in three types ofcontext

Mean N

sAidS.D. didS.D.

saidS.D. EM1 　316 10 　2488 　40636 　2946 heed F1

2608 10

249674 2841101

248746 heed F2

　509 10

　53138 　43317

　56337 hid F1

2292 10

245244 201257

241248 hid F2

EM2 　273 10 　28333 　25913 　27920 heed F1

2625 10

2600128 2722103

2552114 heed F2

　413 10

　40011 　40115

　43842 hid F1

2437 10

2536102 241738

235899 hid F2

EM3 　241 10 　23612 　23317 　25522 heed F1

2258 10

226352 223355

227847 heed F2

　399 10

　38749 　40624

　40514 hid F1

1979 10

1990114 198353

196445 hid F2

EF1 　338 10 　39381 　29612 　32644 heed F1

2658 10

2768411 2552303

2656411 heed F2

　486 10

　42414 　59943

　43512 hid F1

2156 10

203674 238079

205477 hid F2

EF2 　360 10 37532 35315 　35316 heed F1

2773 10

278729 275035

278128 heed F2

　525 10

　50921 　54013

　52510 hid F1

2143 10

220378 207843

214932 hid F2

EF3 　415 10 　43438 　41233 　40034 heed F1

2909 10

296458 286592

289751 heed F2

　583 10

　55649 　57470

　61835 hid F1

2274 10

234377 228977

218972 hid F2

Comparison P P P Z Z Z

said-sAid did-sAid

said-did said-sAid

did-sAid said-did

sAid,said < did NS

<.005 <.005

-.153 -2.8

-2.8

said,sAid < did NS

<.005 <.005

-.153 -2.8

-2.8

did < sAid <said <.047

<.005 <.005

-1.98 -2.8

-2.8

did <said <sAid <.007

<.005 <.005

-2.7 -2.8

-2.8

did<said,sAid NS

<.036 <.019

-.459 -2.1

-2.34

said,sAid<did NS

NS <.008

-.968 -1.78

-2.66

sAid,did<said <.017

NS <.047

-2.39 -.178

-1.98

said,did<sAid <.007

<.022 NS

-2.7 -2.29

Table 8 presents one case, which discriminate F1 of "hod” on all three types of contexts. Among 24 cases, eight cases show hyper-articulation of vowels on one type of contexts; six for interrogatives,one for repetitive statements,and one for first statements.

did,sAid<said NS

NS <.022

-1.78 -.653

-2.29

did,sAid,said NS

NS NS

-1.53 -1.07

-1.68

sAid,said,did NS

NS NS

-1.37 -1.58

-.153

did,said,sAid NS

NS NS

-.968 -.561

-.867

did < said,sAid NS

<.007 <.047

-.764 -2.7

-1.98

did,said <sAid <.005

<.005 NS

-2.7 -2.7

-.969

sAid,said < did NS

<.005 <.005

-1.78 -2.8

-2.8

sAid,said < did NS

<.005 <.005

-.764 -2.8

-2.8

said,did,sAid NS

NS NS

-1.88 -1.88

-.153

did<said,sAid NS

NS <.047

-.408 -1.88

-1.98

sAid,said<did NS

<.009 NS

-1.68 -2.59

-1.78

did<said,sAid NS

<.012 <.005

-1.78 -2.49

-2.8

said,did,sAid NS

NS NS

-1.47 -1.27

-.764

did,said<sAid <.007

<.028 NS

-2.7 -2.19

-.663

sAid,did<said <.008

NS NS

-2.65 -1.78

-1.32

said,did<sAid <.009

<.022 NS

-2.59 -2.29

-1.88

Table 8 Vowelqualities of[æ] and [ɑː] by native English speakers in three types ofcontext

Mean N

sAid S.D. didS.D.

saidS.D. EM1 　753 10 　76941 　72329 　76730 had F1

1658 10

169288 164251

164135 had F2

　711 10

　72137 　71324

　70022 hod F1

1191 10

118636 117439

121352 hod F2

EM2 　876 10 　88464 　87539 　87250 had F1

2173 10

2163112 222479

2133105 had F2

　597 10

　591164 　604159

　59784 hod F1

1046 10

1067162 1072136

　999155 hod F2

EM3 　646 10 　63651 　66435 　63930 had F1

1705 10

170565 168353

173797 had F2

　646 10

　64324 　65527

　63930 hod F1

1053 10

105034 1048135

106161 hod F2

EF1 1084 10 108651 1067156 109971 had F1

1912 10

199068 1774232

1973104 had F2

　888 10

　82555 　89870

　942132 hod F1

1553 10

1582254 1568177

Table continued

Table 9 presents one case,which discriminate F2 of"hood" on allthree types ofcontexts. Among 24 cases,eight cases show hyper-articulation ofvowels on one type ofcontexts;five for interrogatives,one for repetitive statements,and one for first statements.

EF2 　815 10 　82041 　84220 　78232 had F1

1703 10

175854 164520

171668 had F2

　746 10

　73818 　76929

　73017 hod F1

1291 10

124258 134682

128569 hod F2

EF3 　965 10 101074 　96452 　92141 had F1

1677 10

162968 166580

158370 had F2

　822 10

　80050 　85240

　81228 hod F1

1175 10

112655 121455

118472 hod F2

Comparison P P P Z Z Z

said-sAid did-sAid

said-did said-sAid

did-sAid said-did

did < sAid,said NS

<.021 <.013

-.045 -2.31

-2.49

said,did < sAid <.005

NS NS

-2.8 -1.58

-.051

said,did,sAid NS

NS NS

-1.47 -.561

-1.17

did,sAid,said NS

NS NS

-1.47 -1.07

-1.73

said,did,sAid NS

NS NS

-.663 -.663

-.051

said,sAid,did NS

NS NS

-.561 -1.58

-1.531

sAid,said,did NS

NS NS

-.764 -.459

-.663

said,sAid,did NS

NS NS

-1.37 -.051

-.968

sAid,said,did NS

NS NS

-.255 -1.07

-1.22

did,sAid,said NS

NS NS

-.968 -1.17

-1.47

said,sAid<did NS

<.050 <.050

-.765 -1.95

-1.95

did,sAid,said NS

NS NS

-.408 -.969

-.051

did,sAid,said NS

NS NS

-.357 -.255

-.153

did < said,sAid NS

<.009 <.009

-.059 -2.59

-2.59

sAid < did,said <.007

<.007 NS

-2.7 -2.7

-.051

said,did,sAid NS

NS NS

-1.07 -.153

-1.17

sAid,did<said NS

NS <.009

-1.07 -1.88

-2.6

sAid,said<did NS

<.005 <.007

-1.17 -2.8

-2.7

sAid<did<said NS

<.038 <.017

-1.07 -2.07

-2.39

sAid,said<did NS

<.037 NS

-1.17 -2.09

-1.88

said,did,sAid NS

NS NS

-1.88 -1.73

-1.27

said,sAid<did NS

NS <.025

-1.07 -.561

-2.24

sAid,said<did NS

<.021 <.028

-.357 -2.31

-2.19

sAid,said<did NS

<.013 NS

-1.83 -2.49

Table continued

Table 9 Vowelqualities of[uː] and [ʊ] by native English speakers in three types ofcontext

Mean N

sAidS.D. didS.D.

saidS.D. EM1 　396 10 　38526 　41222 　39230 hoodoo F1

1210 10

1078182 1093221

1460551 hoodoo F2

　490 10

　49426 　47926

　49826 hood F1

1515 10

145559 1579161

1511103 hood F2

EM2 　373 10 　45862 　29433 　36692 hoodoo F1

1535 10

1342262 1871195

1392156 hoodoo F2

　474 10

　45862 　47041

　49450 hood F1

1684 10

1661166 1773361

161853 hood F2

EM3 　345 10 　351100 　380115 　30314 hoodoo F1

1655 10

1530264 1885104

1549198 hoodoo F2

　429 10

　42620 　42620

　43519 hood F1

1413 10

1291114 1547279

1400132 hood F2

EF1 　416 10 　41472 　42824 　40656 hoodoo F1

1355 10

1272114 1451157

134284 hoodoo F2

　615 10

　60930 　61843

　61625 hood F1

1664 10

1731114 1503147

1758106 hood F2

EF2 　400 10 　40810 　3744 　41910 hoodoo F1

1704 10

1719152 157687

1815155 hoodoo F2

　549 10

　54610 　55523

　54523 hood F1

1732 10

172229 174184

173464 hood F2

EF3 　393 10 　38553 　38615 　41921 hoodoo F1

1579 10

1516156 1520128

1701129 hoodoo F2

　616 10

　60132 　62326

　62436 hood F1

1621 10

148686 1804203

1574186 hood F2

Comparison P P P Z Z Z

said-sAid did-sAid

said-did said-sAid

did-sAid said-did

sAid,said < did NS

<.012 NS

-.612 -2.49

-1.47

sAid,did,said NS

NS NS

-.612 -.359

-1.47

did,sAid,said NS

NS NS

-.51 -1.42

-1.37

sAid < said,did <.028

<.036 NS

-2.19 -2.09

-.968

did,said<sAid NS

<.005 NS

-.357 -2.8

-1.78

sAid,said<did NS

<.007 <.005

-.357 -2.7

-2.8

sAid,did,said NS

NS NS

-1.68 -.561

-1.27

said,sAid,did NS

NS NS

-.866 -1.78

-1.88

said,sAid,did NS

NS NS

-1.02 -.153

-1.78

sAid,said<did NS

<.005 <.005

-.357 -2.8

-2.8

did,sAid,said NS

NS NS

-.663 -1.47

-.663

sAid<said,did <.022

NS NS

-2.29 -1.78

2.2.2 Differences on contexts produced by non-native speakers

Differences on contexts by nonnative-speakers are presented in Table 10-12. As is expected by the author,phonetic contrasts with F1 or F2 on contexts are not observed very clearly among utterances by nonnative speakers.

Table 10 presents three cases,which discriminate F1 and F2 of"heed” and F2 of"hid” on allthree types ofcontexts.Among 40 cases,nine cases show hyper-articulation ofvowels on one type of contexts, five for interrogatives, one for repetitive statements, three for first statements.

said,sAid,did NS

NS NS

-.255 -.714

-1.27

sAid (said)<did NS

<.005 NS

-1.27 -2.8

-1.58

sAid,did,said NS

NS NS

-1.22 -.459

-.255

did < sAid,said NS

<.007 <.007

-1.04 -2.7

-2.7

did<sAid<said <.036

<.005 <.005

-2.09 -2.8

-2.8

did<sAid,said NS

<.037 <.009

-1.88 -2.09

-2.59

said,sAid,did NS

NS NS

-.051 -.561

-.663

sAid,said,did NS

NS NS

-.866 -.255

.153

did<said,sAid NS

NS <.017

-1.68 -.051

-2.39

sAid,did<said NS

NS <.022

-1.88 -.153

-2.29

sAid,did,said NS

NS NS

-1.78 -1.68

-.051

sAid,said<did NS

<.009 <.022

-1.07 -2.59

-2.29

Table 10 Vowelqualities of[iː] and [ɪ] by non-native speakers in three types ofcontext

Mean N

sAidS.D. didS.D.

saidS.D. JM1

　357 10

　35317 　36224

　3559 heed F1

2203 10

217382 2279146

215776 heed F2

　392 10

　38124 　41123

　3819 hid F1

2167 10

217382 2216117

211149 hid F2

JM2

　282 10

　28313 　28813

　27611 heed F1

2102 10

　216152 1979231

216643 heed F2

　327 10

　34412 　2988

　33815 hid F1

2023 10

199054 206138

201471 hid F2

JF1

　349 10

　37732 　32418

　34628 heed F1

2751 10

273953 277350

274055 heed F2

　415 10

　43149 　39346

　42043 hid F1

2772 10

2801100 277061

274635 hid F2

JF2

　439 10

　45322 　40623

　45748 heed F1

2913 10

3087113 2693513

2959253 heed F2

　411 10

　40489 　40718

Table continued JF3 　343 10 　32621 　37824 　32627 heed F1

2948 10

294649 291424

298566 heed F2

　379 10

　37534 　36919

　39323 hid F1

2915 10

293546 286124

295068 hid F2

JF4 　381 10 　40235 　34312 　39837 heed F1

2976 10

299235 299752

296976 heed F2

　392 10

　39415 　38218

　40038 hid F1

2887 10

283077 299752

283539 hid F2

JF5 　482 10 　4837 　4849 　48113 heed F1

2873 10

292532 2777266

291727 heed F2

　512 10

　52941 　48713

　52142 hid F1

2801 10

274061 285950

280667 hid F2

JF6 　466 10 　47323 　46032 　46645 heed F1

2757 10

278678 2718301

2767110 heed F2

　507 10

　50351 　48481

　53444 hid F1

2483 10

2519171 2430388

2499119 hid F2

JF7 　381 10 　37623 　38024 　38620 heed F1

2921 10

292135 292144

292022 heed F2

　445 10

　47224 　42229

　44032 hid F1

2819 10

2795103 279626

286665 hid F2

JF8 　470 10 　39744 　42034 　40523 heed F1

2897 10

287545 296639

285044 heed F2

　431 10

　42835 　43713

　42927 hid F1

2745 10

275773 272986

275065 hid F2

Comparison P P P Z Z Z

said-sAid did-sAid

said-did said-sAid

did-sAid said-did

said,sAid,did NS

NS NS

-.561 -.237

-.357

said,sAid,did NS

NS NS

-.051 -1.59

-1.78

said,sAid<did NS

<.024 <.028

-.204 -2.25

-2.19

said,sAid,did NS

NS NS

-.051 -1.59

-1.78

said,sAid,did NS

NS NS

-1.27 -1.17

-1.78

did<sAid,said NS

<.047 <.005

-1.27 -1.98

-2.8

did<said,sAid NS

<.005 <.005

-.869 -2.8

-2.8

sAid,said<did NS

<.008 NS

-1.17 -2.65

-1.78

did<said,sAid NS

<.009 <.020

-.663 -2.6

-2.29

sAid,said<did NS

<.022 NS

-.296 -2.29

-1.68

did,said,sAid NS

NS NS

-1.12 -1.58

-1.58

said,did,sAid NS

NS NS

-1.88 -.764

Table 11 presents two cases,which discriminate F2 of"had” and that of"hod” on allthree types ofcontexts.Among 40 cases,five cases show hyper-articulation ofvowels on one type ofcontexts,and that is five for interrogatives.

did<sAid,said NS

<.005 <.040

-.4 -2.8

-1.98

did,said,sAid NS

NS NS

-1.68 -1.88

-.96

sAid,did,said NS

NS NS

-.153 -.051

-.652

said,did,sAid NS

NS NS

-.459 -.459

-.204

said,sAid<did NS

<.008 <.005

-1.32 -2.66

-2.8

did,sAid<said NS

NS <.032

-1.32 -1.68

-2.14

did,sAid<said NS

NS <.008

-1.59 -.816

-2.66

did<sAid,said NS

<.005 <.005

-.459 -2.8

-2.8

did<said<sAid <.005

<.005 <.007

-2.8 -2.8

-2.7

said,sAid,did NS

NS NS

-.663 -.357

-.663

did,sAid,said NS

NS NS

-.051 -1.88

-1.22

sAid,said<did NS

<.022 <.005

-.153 -2.29

-2.8

said,sAid,did NS

NS NS

-.255 -.119

-.561

did<said<sAid <.005

<.005 <.005

-2.8 -2.8

-2.8

did,said<sAid NS

<.028 NS

-1.83 -2.19

-1.83

sAid<said<did <.028

<.007 <.047

-2.19 -2.7

-1.98

did,said,sAid NS

NS NS

-.459 -1.12

-.77

did,said,sAid NS

NS NS

-.153 -.459

-1

did,sAid,sAid NS

NS NS

-1.47 -.561

-1.47

did,said,sAid NS

NS NS

-.051 .051

-.765

sAid,did,said NS NS NS .459 .051 .867

said,did,sAid NS NS NS .051 .306 .459

did,said<sAid NS <.009 NS 1.68 2.6 1.07

sAid,did<said NS NS <.028 1.88 .306 2.19

sAid,said,did NS

NS NS

-.357 -.714

-1.63

said,sAid<did NS

<.007 <.007

-1.68 -2.7

-2.7

sAid,said,did NS

NS NS

-.459 -.306

-.83

did,said,sAid NS

NS NS

-.102 -.866

-.255

Table 11 Vowelqualities of[æ] and [ɑ] by non-native speakers in three types ofcontext

Mean N

sAidS.D. didS.D.

saidS.D. JM1 　686 10 　66142 　70333 　69528 had F1

1523 10

152424 1533119

151230 had F2

　662 10

　68436 　61940

　68322 hod F1

1172 10

124078 106576

121258 hod F2

JM2 　427 10 　42519 　43026 　42717 had F1

1584 10

162162 149794

163560 had F2

　492 10

　50230 　46449

　51133 hod F1

1134 10

115832 109834

Table continued JF1 　915 10 　94394 　88970 　91276 had F1

1473 10

1496168 138686

1538100 had F2

　724 10

　74574 　69456

　732100 hod F1

1126 10

115474 109698

112454 hod F2

JF2 　847 10 　84927 　85529 　83732 had F1

1439 10

146457 　139140

146187 had F2

　755 10

　74234 　77935

　74536 hod F1

1525 10

127131 125458

130045 hod F2

JF3 　819 10 　78431 　87638 　79634 had F1

1599 10

1576135 160650

161646 had F2

　729 10

　72848 　69433

　76559 hod F1

1186 10

116143 117438

1224105 hod F2

JF4 　738 10 　800156 　685123 　730144 had F1

1664 10

163593 172855

162968 had F2

　725 10

　800156 　63564

　730144 hod F1

1382 10

1487123 121839

144299 hod F2

JF5 　848 10 　82368 　89256 　82883 had F1

1528 10

1526233 1515154

1544161 had F2

　743 10

　76472 　70433

　76150 hod F1

1319 10

132344 129766

133656 hod F2

JF6 　737 10 　68859 　79071 　73334 had F1

1838 10

1783189 1901340

1830186 had F2

　790 10

　75935 　82446

　78740 hod F1

1203 10

121399 1187101

1208107 hod F2

JF7 　974 10 100135 　89722 102525 had F1

1165 10

101033 115471

133174 had F2

　831 10

　90051 　698127

　89544 hod F1

1082 10

111146 　98369

115167 hod F2

JF8 　818 10 　85063 　74840 　85647 had F1

1689 10

1701150 166873

170997 had F2

　669 10

　67036 　69147

　64641 hod F1

1236 10

125082 1212113

124584 hod F2

Comparison P P P Z Z Z

said-sAid did-sAid

said-did said-sAid

did-sAid said-did

sAid,said<did NS

<.017 NS

-1.88 -2.39

-.561

said,sAid,did NS

NS NS

-1.58 -1.22

-1.58

did<said,sAid NS

<.032 <.017

-.306 -2.14

-2.39

did<said,sAid NS

<.007 <.005

-.866 -2.7

Table 12 presents no case which discriminate F1 or F2 on all three types of contexts. Among 40 cases, eight cases show hyper-articulation of vowels on one type of contexts, six for interrogatives,one for repetitive statements,one for first statements.

sAid,said,did NS

NS NS

-.255 -.225

-.408

did<sAid,said NS

<.007 <.005

-.652 -2.7

-2.8

did,sAid,said NS

NS NS

-1.42 -1.27

-1.68

did<said,sAid NS

<.007 <.047

-.866 -2.7

-1.98

did,said,sAid NS

NS NS

-1.27 -1.88

-.764

did<sAid,said NS

NS <.022

-.459 -1.68

-2.29

did,said,sAid NS

NS NS

-.663 -1.68

-.561

did,said,sAid NS

NS NS

-.968 -1.22

-.968

said,sAid,did NS

NS NS

-.969 -.255

-.765

did<said,sAid NS

<.022 NS

-.153 -2.29

-1.78

sAid,said,did NS

NS NS

-.357 -1.83

-1.88

did,sAid,said NS

NS NS

-1.37 -.459

-1.78

sAid,said<did NS

<.005 <.005

-.663 -2.8

-2.8

sAid,did,said NS

NS NS

-.764 -.561

-.357

did<sAid,said NS

NS <.022

-1.78 -.969

-2.29

sAid,did,said NS

NS NS

-1.68 -.561

-1.17

did,said,sAid NS

<.028 NS

-.153 -2.19

-1.07

said<sAid,did NS

NS <.022

-.408 -.408

-2.29

did<said,sAid NS

<.028 NS

-1.17 -2.19

-1.42

did<said,sAid NS

<.005 <.005

-1.27 -2.8

-2.8

sAid,said<did NS

<.037 NS

-.051 -2.09

-1.47

did,said,sAid NS

NS NS

-.459 -.357

-.357

did<said,sAid NS

<.022 <.012

-.255 -2.29

-2.49

did,sAid,said NS

NS NS

-.204 -.714

-1.58

sAid,said<did NS

<.028 <.022

-1.68 -2.19

-2.29

sAid,said,did NS

NS NS

-1.17 -.764

-.459

sAid<said,did <.028

<.022 NS

-2.19 -2.29

-1.17

did,said,sAid NS

NS NS

-.051 -.051

-.255

did<sAid,said NS

<.005 <.005

-1.68 -2.8

-2.8

sAid<did<said <.005

<.005 <.007

-2.8 -2.8

-2.7

did<said,sAid NS

<.005 <.005

-.357 -2.8

-2.8

said<did<sAid <.013

<.005 <.005

-2.49 -2.7

-2.8

did<sAid,said NS

<.013 <.005

-.357 -2.49

-2.8

did,sAid,said NS

NS NS

-.347 -.866

-.663

said,sAid<did <.037

NS NS

-2.09 -.764

-1.68

did,said,sAid NS

NS NS

-.153 -.561

Table 12 Vowelqualities of[uː] and [ʊ] by non-native speakers in three types ofcontext

Mean N

sAidS.D. didS.D.

saidS.D. JM1 　382 10 　37918 　38311 　38421 hoodoo F1

1902 10

1940134 1891184

1876224 hoodoo F2

　447 10

　44723 　43018

　43328 hood F1

1559 10

158187 151897

1704348 hood F2

JM2 　322 10 　3329 　3046 　3317 hoodoo F1

1254 10

125234 121843

124345 hoodoo F2

　360 10

　36715 　3445

　37013 hood F1

1300 10

136357 124329

129375 hood F2

JF1 　440 10 　43633 　44727 　43729 hoodoo F1

1171 10

1218127 　981136

131599 hoodoo F2

　416 10

　39723 　43735

　41328 hood F1

1231 10

1120161 1393202

1180135 hood F2

JF2 　492 10 　49713 　46733 　50611 hoodoo F1

1705 10

172364 167771

171690 hoodoo F2

　530 10

　54638 　50626

　53848 hood F1

1655 10

165883 167079

163895 hood F2

JF3 　354 10 　34219 　36217 　35821 hoodoo F1

1562 10

172364 1739110

1525261 hoodoo F2

　381 10

　37027 　40415

　37034 hood F1

1552 10

1293216 170587

1657304 hood F2

JF4 　343 10 　34330 　34817 　33825 hoodoo F1

1026 10

1085173 　97076

102391 hoodoo F2

　438 10

　44636 　42435

　44526 hood F1

1219 10

1085173 1225125

1347145 hood F2

JF5 　528 10 　53316 　51125 　54028 hoodoo F1

1867 10

1919159 1806100

1876119 hoodoo F2

　511 10

　51345 　49815

　52335 hood F1

1650 10

1659115 167159

162051 hood F2

JF6 　427 10 　4307 　4198 　4318 hoodoo F1

1666 10

170680 160759

168552 hoodoo F2

　471 10

　43818 　53544

　44218 hood F1

1691 10

167179 1718108

168381 hood F2

JF7 　366 10 　37330 　36624 　36022 hoodoo F1

1251 10

1208141 136579

1180205 hoodoo F2

　366 10

　36718 　35619

　37544 hood F1

1177 10

1172140 125290

Table continued JF8 　416 10 　42415 　4125 　41123 hoodoo F1

1643 10

167838 1554137

169688 hoodoo F2

　458 10

　46723 　44415

　46223 hood F1

1828 10

1913352 1835172

1735249 hood F2

Comparison P P P Z Z Z

said-sAid did-sAid

said-did said-sAid

did-sAid said-did

sAid,did,said NS

NS NS

-.889 -.358

-1.53

said,did,sAid NS

NS NS

-.866 -.866

-.051

did<said,sAid NS

<.022 NS

-1.63 -2.29

-.204

did,sAid,said NS

NS NS

-.866 -1.37

-1.88

did<said,sAid NS

<.005 <.005

-.255 -2.8

-2.8

did,said,sAid NS

NS NS

-.866 -1.47

-.971

did<sAid,said NS

<.007 <.005

-.534 -2.7

-2.8

did<said,sAid NS

<.005 NS

-1.93 -2.8

-1.58

sAid,did,said NS

NS NS

-.102 -.562

-.866

did<sAid,said NS

<.013 <.005

-1.71 -2.49

-2.8

sAid,said<did NS

<.009 NS

-1.42 -2.6

-1.78

said,sAid<did NS

<.013 <.017

-.664 -2.49

-2.39

did<sAid,said NS

<.025 <.022

-1.42 -2.24

-2.29

did,said,sAid NS

NS NS

-.153 -1.58

-.765

did<said,sAid NS

<.041 NS

-1.22 -2.04

-1.02

said,sAid,did NS

NS NS

-.968 -1.78

-.866

sAid,said,did NS

NS NS

-1.58 -1.88

-.306

said,sAid,did NS

NS <.022

-.153 -1.58

-2.29

said,sAid,did NS

<.021 <.047

0 -2.31

-1.98

sAid<did,said NS

<.007 NS

-.561 -2.7

-.153

said,sAid,did NS

NS NS

-.459 -.764

-1.17

did,said,sAid NS

NS NS

-.968 -1.88

-1.27

did<said,sAid NS

<.012 NS

-.357 -2.5

-1.63

sAid,did,said NS

NS NS

-1.27 -1.88

-1.83

did,sAid,said NS

NS NS

-.764 -1.63

-1.68

did,said,sAid NS

NS NS

-.612 -1.88

-1.47

did,sAid<said NS

NS <.037

-1 -.612

-2.09

said,sAid<did NS

NS <.028

-.652 -.968

-2.19

did<sAid,said NS

<.037 <.007

-.102 -2.08

-2.7

did<said,sAid NS

<.008 <.005

-.764 -2.66

-2.8

sAid,said<did NS

<.005 <.005

-.51 -2.8

-2.8

sAid,said,did NS

NS NS

-.459 -1.27

-.764

said,did,sAid NS

NS NS

-1.12 -.764

-.663

said,sAid<did NS

<.005 <.037

-.764 -2.8

-2.09

did,sAid,said NS

NS NS

-.357 -1.22

-.866

said,sAid<did NS

NS <.022

-1 -1.17

-2.29

said,did<sAid NS

<.028 NS

-1.47 -2.19

-.153

did<sAid,said NS

<.028 <.037

-.764 -2.19

-2.09

did,said,sAid NS

NS NS

-.764 -1.88

-1.42

said,did,sAid NS

NS NS

-.764 -.357

3.Discussion and conclusions

This study focuses on vowel spaces produced by native English speakers and Japanese speakers. Results show that two-paired vowels produced by Japanese native speakers are not separated enough in vowel spaces. This tendency is quite different from the one by native English speakers. Different contexts do not always produce hyper-articulation for Japanese native speakers and this tendency is also observed for native English speakers.

Vowels dotted on a vowel space are to be used for grasping characteristics of learners’ pronunciation by themselves. It is expected that they sense them not only by ears but also with eyes. As Gregory (1970, 155) claims, we are able to read function from structure, and engineers can 'see’the functionalsignificance ofthe parts ofquite complicated systems.

One syllable words are used for oralreading except the one,which is a two-syllable word of 'hoodoo’. For a word list, a single word, not a phrase, is selected to equalize conditions of reading vowels.The author ofthis study,then,is recommended by other researchers to use a contraction of 'who would’, for the next experiment. This comment works very well as these words are treated separately in a written form but they do not act by themselves in a spoken form.The word of'who’d’is sure to present phonetic features ofa single word.

This study focuses on phonetic features of F1s and F2s. To discriminate vowel qualities, however, duration also should play an important role. For listeners, interaction of duration values and formant ones would decide which vowels speakers are uttering.

One thing that happens to be found against the author’s expectation for contrast ofvowel qualities on contexts is hyper-articulation for interrogatives. In dyad, one asks to the other, for example,"Did you say "hood”?” after she/he says "What did you say?”,and the other say, 'I said "who’d”.”.This "hood” is hyper-articulated very welland it is much more than the one repeated the second time as in "I said "who’d”.”. This study shows the repetition does not always cause hyper-articulation ofwords.The condition ofrecording in phonetic laboratory, ofcourse,might get rid ofnaturalsituations ofhyper-articulation for repeated words.

Acknowledgments

Much appreciation goes to colleagues, students, loved ones for their support of my language experiments, especially Dr. Ooigawa Tomohiko for his comment to normalization of formant values. This research is partially supported by a Project Grant-In Aid for Scientific Research by the Ministry of Education, Culture, Sports, Science and Technology (Basis C-26370655, "Applied study on ability of analyzing English sound with visualized vowelspaces”).

References

Cohn,N.(2013).The VisualLanguage ofComics.London:Bloomsbury. Gregory,R.L.(1970).The IntelligentEye.London:Weidenfield&Nicolson.

Lennerberg,E.(1967).BiologicalFoundationsofLanguage.New York:Wiley & Sons. Story, B. H. and Bunton, K. (2017). Vowel space density as an indecator of speech

performance.JournalofAcousticalSociety ofAmerica,141:5,458-464.

Suemitsu, A., Dang, J., Ito, T. and Tiede, M. (2015). A real-time articulatory visual feedback approach with target presentation for second language pronunciation learning.The Journalofthe AcousticalSociety ofAmerica ExpressLetters(Published Online),138(4). Trofimov,M.V.and Daniel,J.(1923).The Pronunciation ofRussian.Cambridge:Cambridge

TOMITA

Kaoru

Engl

i

sh

Phoneti

cs

Abstract

This study explores vowel qualities differentiated between native and non-native speakers.Words which contain six vowels respectively,[i],[ɪ],[１],[ɑ],[ʊ]or [u]are orally read in carrier sentences (e.g."What did you say? I said "heed”.Did you say "hid”? I said "heed””.). They include statements and interrogatives uttered between two speakers. Forman values oftarget vowels are measured with Praat,among which formant one and two are dealt with statistically.

Phonological and phonetic features of speakers’ native languages affect their foreign language pronunciation. The effects last long and so even for adult learners of advanced levels,making speech without foreign accents is not an easy task.Lenneberg (1967)proposes an important biologicalprocess oflanguage learning and calls it "a criticalperiod.” After the criticalperiod,it becomes difficult for people to acquire languages.

Effects of critical period for learning foreign languages are controversial issues. Learners who start learning foreign languages after the critical period tend to show errors of pronunciation for which their native language have a significant effect.

Foreign accented speech by Japanese advanced-level learners of English is observed in general societies, language classes and even academic meetings. It is caused by slight mistakes on both prosodic features, such as stress and intonation, and segmental features, such as consonants and vowels. Among these features, the vowel is focused on for the language experiment ofthis study.

English and Japanese have different vowel systems. English has eleven vowels and Japanese has five vowels. For example, English contrasts two vowels ([i] vs. [ɪ]) with vowel quality differences and Japanese contrasts them with short and long vowels ([i:] vs. [i]) without vowelquality differences.

Phonol

ogi

cal

and

Phoneti

c

Contrasts

on

Vowel