• Keine Ergebnisse gefunden

6.3. Age variation

6.3.1. Age variation across E ʋ e

One other interesting area examined in this study was speaker age and the use of the variables /e/, / ː/ and /ə/. The purpose was to determine the effect of age on the use of the three vowels across the entire E e. A statistical test was applied separately to the different age groups: Young age versus Old age across the two E e communities. First, the vowel spaces of each speaker group were compared using group mean F1 and F2 frequency values of each vowel. Figures 6.4a, b and c show both individual formant and group mean frequency values of the individual tokens spoken by the two age groups across E e.

109

(ai) Individual formant values of /e/ (aii) Mean formant values of /e/

(bi) Individual formant values of /ə/ (bii) Mean formant values of /ə/

(ci) Individual formant values of / ː/ (cii) Mean formant values of / ː/

Figure 6.4: Individual and mean formant frequency values of /e/, / ː/ and /ə/ measured at a steady state, plotted on S-transformed scales, F1/S(F1) on y- axis, F2-F1/S(F2-F1) on the x-axis spoken by speakers across Ewe

110

The red dots in all the figures indicate the vowel space for older age speakers, and the blue show that of the younger age speakers. From all the figures, we see differences in the formant frequencies within each vowel category. The DRESS in (a) occupies roughly three positions for each of the speaker groups while the NURSE in (b) occupies two different positions. The lettER, however, occupies four different positions on the vowel space as shown in (c), hence confirming the earlier results indicated in chapter five. Apart from the variations in each vowel category, there are variations also in the formants of the speaker groups for each vowel. The space of the DRESS for the younger age speakers is relatively lower and more fronted (i.e. higher mean F1-F2 values). The older age speakers’ is, however, relatively closer and more central (i.e. lower mean F1-F2 values). The younger speaker’s space for the NURSE is surprisingly more central than that of the older speakers’. The acoustic spaces of the lettER vowel for both speaker groups, although are not uniform, appear to be different for each group.

The differences in the mean frequencies within each vowel category even for each speaker group, suggest that the speakers are inconsistent in their realisation of the vowels. Figures 6.5a, b and c present the percentage of the variants used by each speaker group for each of the vowels.

111 (c) lettER

Figure 6.5: Percentage (%) distribution of the DRESS, NURSE and lettER vowels according to across E e

It obvious from the figures that the use of [ɪ] and [ ː] is infrequent among younger speakers, a group whose articulation of the DRESS and NURSE was mostly [e̠] and [eː]; about 633 tokens constituting 88% out of the total of 720 tokens for each. Older age speakers, on the other hand, showed a higher preference for [ɪ] and [ ː] than the younger speakers. This is an indication that younger speakers are more likely to retract the DRESS to [e̠], while fronting the NURSE. The older speakers are however more likely to favour significantly the raised variant [ɪ] of the DRESS, but centralizing the NURSE (i.e. closer and more back realisations). The margin of difference between the two speaker groups for the variants of the lettER, however, appears relatively small compare to those of the DRESS and the NURSE. A statistical test was performed to examine whether speaker age had significant effects on the variant choice. The observed and expected results are presented in Tables 6.4 and 6.5.

0 20 40 60 80 100

Old Young

[I] [o] [ə] [ ]

112

Table 6.4: Age Distribution of NURSE variants across Eʋe

Source: Field Data, 2017 **p<0.01 df = 94 (N = 96) Where Eta Square = 2, standard deviation = Std. Dev.

Variables Age

group N Mean Std. Dev. t-value Sig.

2

[eː] Lower

48 13.396 2.672

4.229** 0.000 0.150 Upper 48 9.479 6.021

[ ː] Lower

48 1.604 2.672

-4.119** 0.000 0.158

Upper 48 5.521 6.021

113

Table 6.5: Age Distribution of DRESS variants across Eʋe

Source: Field Data, 2017 **p<0.01 df = 94 (N = 96) Where Eta Square = 2, standard deviation = Std. Dev.

It is clear from the statistical report that the two age groups differed highly significant statistically [p≤.000] for both the raised [ɪ] and the mid [e̠] realisations of the DRESS, with younger speakers showing an extreme preference for [e̠], more than the older group. There was however no statistically significant difference between the groups for the central realisation [ ː]. Table 6.5 equally shows that both speaker groups differed extremely significant statistically [p≤.000] for the fronted [eː] and the central realisations [ ː] of the NURSE. That is, when the NURSE vowel and speaker age were compared, the younger speakers showed an extremely higher preference for the fronted variant than the older speakers. Interestingly, the older speakers appeared to have shown an equal preference for both variants. Age however has no significant effect for the lettER vowel as illustrated in Table 6.6.

Variables Age

group N Mean

Std.

Dev. t-value Sig.

2

[ɪ] Lower

48 0.187 1.024

-4.329** 0.000 0.166 Upper

48 3.771 5.642 [e̠] Lower

48 13.104 1.519 5.032**

0.000 0.212

Upper 48 8.750 5.799

[ ː] Lower

48 1.708 1.352

-1.808 0.075 Upper 48 2.479 2.625

114

Table 6.6: Age Distribution of the lettER variants across Eʋe Variables

Age group N Mean Std. Dev. t-value Sig.

[ɪ] Lower

48 0.354 0.635

1.759 0.083

Upper 48 0.167 0.376

[ə] Lower

48 6.375 3.577

-0.244 0.808

Upper 48 6.562 3.946

[o] Lower

48 1.250 0.565

-1.302 0.197

Upper 48 1.437 0.823

[ ] Lower

48 7.021 3.348

0.265 0.791

Upper 48 6.833 3.569

Source: Field Data, 2017 **p<0.05 df = 94 (N = 96) Where Eta Square = 2, standard deviation = Std. Dev.

The margin of differences of ( 2 = 0.166), and ( 2 = 0.212) between the two age groups for [ɪ] and [e̠], according to Cohen’s (cited in Cohen et al. 2007) guidelines of interpreting eta square are large. Meaning that the difference of about 18% between the groups are large enough to attribute the result to the age of the speakers. The margin of differences of ( 2 = 0.150/158), that is, about 15 percent each for the NURSE variants [eː] and [ ː], is in the same way huge, and hence confirm the age effects on the DRESS and NURSE vowels among the E e of Ghana. The hierarchical graphs in Figure 6.6 reiterate the convergence and divergence between the two age groups for the three vowels.

115

(a) DRESS (b) NURSE

(c) lettER

Figure 6.6: Hierarchical graphs showing group mean frequencies for the tokens of the DRESS, NURSE and lettER vowels spoken by both age groups

The high mean scores for speakers below age 30 for [e̠] and [eː] in general clearly shows their preference for these variants. As seen earlier, the mean scores were very high for the younger age group for these variables, but declined for older speaker age group. The variant [ɪ] can be said to be restricted to the older age speakers since it was virtually absent from the sample of the younger speakers. It appears that the older speakers are gradually moving away from the mid retraction of /e/ while going for the raised retraction, and from / ː/ fronting to its central form. For, in each of the cases, the index scores for [e̠] and [eː] decreased, but increased for [ɪ] and also for [ ː] for the older age speakers. The assumption then is that the younger the speaker, the more likely

116

they will use [e̠] and [eː], and the older the speaker, the more likely they will use [ɪ] and [ ː]. But the two groups are identical in their use of the lettER vowel. Interestingly, the age effect on these variables seems to be strictly geography specific, that is, the age effect is different for each dialect group. To explore this further, a statistical test was applied separately to each variant and age in each dialect group: A lɔ lower age versus A lɔ upper age, and E eme lower age versus E eme upper age. That is, comparisons were made between two age groups: young and old, within each dialect region for age related variation.