• Nie Znaleziono Wyników

5.2 Stressed vowel duration

5.2.6 Within-phrase relations

All of the above figures compare mean absolute durations of vowels in different types of contexts. They give a general insight into absolute vowel durations with respect to their intrinsic properties and context. This section is devoted to the length of vowels in stressed syllables within the same phrase in order to provide a better illustration of the rhythmic differences between the groups. Table 5.13 shows mean vowel length proportions in several head-to-nucleus and nucleus-to-tail relations.

Table 5.12.The effect of coda on vowel duration Vowel preceding_

Group _nasal _voiced _unvoiced _voiced:_unvoiced _nasal:_unvoiced

PL1 149 125 115 1.09 1.30

PL2 135 122 103 1.18 1.31

EN 150 127 108 1.18 1.39

None of the EN head:nucleus vowel ratios is higher than the corresponding PL proportion and four of the seven examples show relatively longer duration of the N vowel. N:T ratios show a similar tendency, where the main phrasal accent makes the vowel relatively longer in native English pronunciation.

The opposite relation in enough:gowns can easily be explained by the final lengthening effect overriding the power of accentual lengthening in English speakers. Final lengthening is attenuated by the unstressed final syllables in Cinders and (god)mother.

The figures displayed in Table 5.13 are still based on mean values without accounting for individual speakers’ speech rhythm characteristics. Table 5.14 presents data calculated from duration proportions in individual speech samples, indicating the mean proportions, individual proportion ranges for both groups and the number of Polish learners whose vowel length proportions were out of native English ranges.

This method of presentation suggests even more strongly a tendency in native English speakers to highlight the nuclear accent by increasing the relative length of the relevant vowel.2 It is possible that the most conspicuous differences have been enhanced by other factors, such as syntactic complexity, motivating a stronger verb-object boundary in Polish learners in longer phrases (thought:divine, buy:gowns) in comparison to simpler, shorter structures (bad:mood, find:hat), where PL ratios resembled native proportions

5.2 Stressed vowel duration 81

Table 5.13. Within-phrase stressed vowel quantity proportions (duration means in ms in parentheses)

Group

Tested units PL1 PL2 EN

VH:VN

honour:son (SH:SN) 0.9 (123:133) 1.1 (120:108) 0.8 (96:127) thought:divine (LH:DN) 0.6 (142:220) 0.5 (102:208) 0.3 (91:271) bad:mood (AH:LN) 0.9 (140:153) 0.9 (143:165) 0.9 (178:206) buy:gowns (DH:DN) 0.6 (178:303) 0.55 (143:262) 0.45 (131:292) started:shouting 0.8 (113:134) 0.8 (102:123) 0.8 (105:137) find:hat (DH:AN) 0.8 (112:142) 0.8 (97:126) 0.8 (131:158) sisters:gone (SH:SN) 0.3 (56:166) 0.4 (58:148) 0.3 (51:154)

VN:VT

enough:gowns (SN:DT) 0.5 (115:252) 0.5 (108:228) 0.4 (103:263) shouting:Cinders (DN:ST) 2.3 (134:57) 2.1 (123:58) 2.4 (137:58) god:mother (SN:ST) 0.8 (93:118) 0.8 (90:111) 1 (89:87)

2 Accented vowel length will be discussed in relation to higher level units in the subsequent sections.

despite considerable discrepancies in mean durations of individual tested vowels.

5.2.7 Summary

The results concerning native and non-native English vowel duration indicate the complexity of interaction of the duration determinants already discussed in the theoretical part of this dissertation, and the need to verify many of the suggested tendencies in more detailed studies.

The conclusions based on the gathered data are formulated as follows:

1. Mean vowel length in the read speech of native English speakers and Polish learners is comparable (130—133ms), although more advanced learners show a tendency to accelerate (122ms) especially in shorter, syntactically simpler phrases. This, however, sometimes leads to less natural duration proportions than in the first recording.

2. Standard deviations referring to the means mentioned above and differences in mean durations of vowels classified according to their intrinsic length (see section 5.2.1) indicate a more clearly specified distinction between short and long vowels, and diphthongs in native English pronunciation.

3. The scale of variation is expressed in the observation that the Polish learners make their English long vowels approximately 1.5 times longer and diphthongs twice as long as the short vowels. The corresponding proportions in native speakers amount to 2.4 and 1.7, respectively, which yields 40%

more difference between particular classes.

Table 5.14. Within-phrase stressed vowel proportions in individual speakers Group

honour:son (SH:SN) 0.9 (0.7—1.2) 3 1.2 (0.7—2.2) 8 0.8 (0.4—1.0) thought:divine (LH:DN) 0.7 (0.3—1.0) 10/12 0.5 (0.3—0.8) 5/12 0.3 (0.2—0.5) bad:mood (AH:LN) 0.9 (0.6—1.2) 0 0.9 (0.4—1.8) 1 0.9 (0.4—1.2) buy:gowns (DH:DN) 0.6 (0.4—1.0) 5 0.55 (0.4—0.7) 5 0.45 (0.3—0.6) started:shouting (LH:DN) 0.9 (0.6—1.7) 1 0.9 (0.4—1.4) 4 0.8 (0.6—1.2) find:hat (DH:AN) 0.8 (0.6—1.2) 0 0.8 (0.5—1.3) 1 0.9 (0.6—1.6) sisters:gone (SH:SN) 0.4 (0.2—0.6) 1 0.4 (0.2—0.6) 2 0.3 (0.2—0.5)

VN:VT

enough:gowns (SN:DT) 0.5 (0.3—0.7) 2 0.5 (0.3—0.7) 2 0.4 (0.3—0.6) shouting:Cinders (DN:ST) 2.4 (1.9—3.3) 0 2.3 (1.3—4.2) 0 2.6 (1.1—5.0) god:mother (SN:ST) 0.8 (0.5—1.2) 7/12 0.8 (0.6—1.1) 6 1 (0.8—1.2.0)

4. Vowels in nuclear syllables are generally longer in native English pronunciation not only in terms of timing relations but even in absolute values.

5. Polish learners do not increase their English vowel duration “flexibility”

after seven months of training.

The above conclusions are supported by Table 5.15, which presents the data contained in Table 5.2, rearranged in order to show a ranking of tested words according to PL2:EN mean vowel duration ratio. This ratio is preferable to PL1:EN because we believe that differences which remain after training indicate more serious problems than those signalled by some even larger initial discrepancies that disappear in the course of training.

Columns 7—8 give the number of responses out of the EN range (shorter=“–”, longer=“+”) specified in column 6.

5.2 Stressed vowel duration 83

Table 5.15.Ranked POL:ENG vowel duration proportions in individual tested words Vowel in Position PL1

divine fDN1+ 220 208 271 203—383 4– 5– 0.8 0.8

mood fLN1+ 153 165 206 130—287 4– 2– 0.7 0.8

hat fAN1– 142 126 158 94—205 1– 3– 0.9 0.8

bad mAH1+ 140 143 178 116—248 5– 2– 0.8 0.8

gowns fDN1+ 303 262 292 223—376 0 2– 1.0 0.9

gowns fDT1+ 252 228 263 205—348 1+, 3– 4– 1.0 0.9

shouting mDN3– 134 123 137 93—174 1– 2– 1.0 0.9

parties wLN2– 155 150 166 124—211 1– 2– 0.9 0.9

son fSN1+ 133 108 127 94—195 0 6– 1.1 0.9

gorgeous wLN2+ 173 164 169 128—207 1+, 1– 0 1.0 1.0

started mLH2– 113 102 105 64—186 0 1– 1.1 1.0

had to mAH2+ 79 75 74 44—95 1+ 2+, 1– 1.1 1.0

gone fSN1+ 166 148 154 105—206 2+, 1– 1+, 1– 1.1 1.0

godmother mSN1+ 93 90 89 66—115 1+, 1– 1– 1.0 1.0

Cinders wST2+ 57 58 58 27—84 0 1+ 1.0 1.0

buy mDH2– 178 143 131 92—154 8+ 7+ 1.4 1.1

thought mLH1– 142 102 91 56—123 6+ 2+ 1.6 1.1

Cinders wSN2+ 70 68 64 45—105 1+ 1+, 1– 1.1 1.1

enough mSN1– 115 108 103 74—146 1+ 1– 1.1 1.1

Cinders mSH2+ 62 59 55 32—90 0 0 1.1 1.1

sisters mSH3– 56 58 51 27—69 0 3+ 1.1 1.1

looking mSH4– 62 55 54 25—87 1+ 0 1.2 1.1

mother mSH2+ 120 107 86 59—99 11+ 9+ 1.4 1.2

honour mSH4+ 123 120 96 71—121 8+ 5+ 1.3 1.2

godmother wST2+ 118 111 86 64—118 6+ 4+ 1.4 1.3

The ranking proves that shorter mean PL durations never concern short vowels, whereas longer realisations always pertain to short vowels. The two exceptions of buy and thought, where we observe the final lengthening effect in Polish learners (especially PL1) due to phrasing reorganisation, have already been mentioned in section 5.2.6.

The influence of contextual factors, such as pre-fortis clipping, foot complexity, accentual lengthening and (especially) final lengthening has also been suggested but, owing to their interaction, it would be rather unwise to attempt to isolate any of these variables while controlling the others, given the amount of data gathered for the purpose of this study. Therefore no conclusions are formulated in this respect on account of insufficient evidence.