3.4 Acoustic analysis of word stress in Assam Sora
3.4.2 Fundamental frequency in first and second syllable of Assam Sora disyllablesAssam Sora disyllables
tical test substantiates the fact that vowel duration is significantly longer in second syllable. Therefore, it is confirmed that vowel duration effectively contributes to the prominence of the second syllable in Assam Sora disyllables.
3.4.2 Fundamental frequency in first and second syllable of
This includes calculating the average pitch variation between vowel start and vowel end for every vowel in each syllable of the disyllables. However, considering that aver- age fundamental frequency may ignore an unique pitch change within a given range, maximum fundamental frequency or maximum pitch variation within vowel start and vowel end in first and second syllable is also calculated here. Thus, considering the average fundamental frequency first, Table 3.3 presents the region-wise averages of average fundamental frequency distinction between first and second syllable in Assam Sora disyllables with standard deviation in parenthesis.
Table 3.3: Average of average f0 in first and second syllable
Singrijhan Sessa Koilamari Lamabari
Syllable 1 172.4 (51.09) 142.17 (34.15) 167.79 (59.15) 162.42 (51.63) Syllable 2 181.61 (50.42) 153.27 (32.01) 181.27 (53.47) 161.94 (39.21) From Table 3.3 it observed that average fundamental frequency is always higher in the second syllable than in first syllable, except in case of Assam Sora speech data recorded in Lamabari region. This indicates that there is higher pitch variation in the second syllable than in the first syllable of Assam Sora disyllables. However, while this is true for Assam Sora data recorded from Singrijhan, Sessa, and Koilamari regions, the speech data recorded from Lamabari region does not show similar results.
Significantly, by considering the maximum fundamental frequency next, a similar pattern is again found from the comparison of maximum fundamental frequency in first and second syllable of Assam Sora disyllables. Table 3.4 presents the region-wise average of maximum fundamental frequency distinction in first and second syllable of Assam Sora disyllables.
Table 3.4: Average of maximum f0 in first and second syllable
Singrijhan Sessa Koilamari Lamabari
Syllable 1 182.85 (61.18) 152.28 (43.78) 180.48 (67.24) 174.24 (61.41) Syllable 2 198.43 (62.81) 164.68 (36.81) 196.4 (63.50) 173.58 ((48.47) From Table 3.4 it is observed that maximum fundamental frequency is also higher
in second syllable of Assam Sora disyllables, except in case of speech data recorded from Lamabari region. Thus, it is evident that both average and maximum funda- mental frequencies are higher in second syllable than in the first syllable, except in case of Assam Sora speech data recorded from Lamabari region. In this regard, it was assumed that there are speaker variations in the speech data of Lamabari region since fundamental frequency is sensitive to the physiological differences in vocal tract length of different speakers. Therefore, in order to reduce the speaker effect on fundamental frequency measurements, both average and maximum fundamental frequencies are normalized for all speakers across the four regions using thez-score equation. Hence, Figure 3.3 and 3.4 presents the graphical representations of speaker normalized aver- age and maximum fundamental frequency distinction in first and second syllable of Assam Sora disyllables.
Figure 3.3: Speaker normalized average f0 in first and second syllable
Significantly, Figure 3.3 and 3.4 reveals that both average and maximum funda- mental frequency are higher in the second syllable than in the first syllable of Assam Sora disyllables. Moreover, the normalized speech data shows that a similar pattern
Figure 3.4: Speaker normalized maximum f0 in first and second syllable
is also observed in Assam Sora data recorded from Lamabari region that was pre- viously not found while examining the non-normalized speech data. However, the distinction in Lamabari region is not as clearly visible as it is visible for the other three regions. On the contrary, a closer examination of the Lamabari speech data reveals that, the 10 Assam Sora speakers recorded in Lamabari region actually have an uniform fundamental frequency across first and second syllable of disyllables. This pattern can also be observed in the distribution of Lamabari data for average and maximum fundamental frequency measurements. Data distribution of average and maximum fundamental frequency measurements in first and second syllable of As- sam Sora disyllables is estimated separately through kernel density estimation. Thus, Figure 3.5 and 3.6 presents the kernel density distribution of speaker normalized aver- age and maximum fundamental frequency in first and second syllable of Assam Sora disyllables.
From Figure 3.5 and 3.6 it is evident that both speaker normalized average and
Figure 3.5: Distribution of speaker normalized averagef0 in first and second syllable
maximum fundamental frequency are clearly separated in first and second syllable of Assam Sora disyllables, except in case of Assam Sora data recorded from Lamabari region. Thus, the analysis of fundamental frequency reinforces the fact that in Assam Sora disyllables there is higher pitch variation in second the syllable than in first syllable, except in case of Assam Sora speech variety spoken in Lamabari region. In case of Lamabari region, a closer examination of the speech data reveals that the pitch in the vowel nuclei is completely uniform in first and second syllable of As- sam Sora disyllables. Additionally, a one-way ANOVA test conducted by considering normalized average f0 as dependent variable and syllable positions (first or second syllable) as factors also reveals the same variation. For all regions except Lamabari, averagef0 is found significantly higher in the second syllable than in the first syllable (Singrijhan,F(1, 4710) = 333.114,p <0.001; Sessa, F(1,1907) = 134.417, p <0.001;
Koilamari,F(1,3080) = 114.524,p <0.001; Lamabari F(1,2534) = 1.582, p = 0.209).
Figure 3.6: Distribution of speaker normalized maximumf0 in first and second syl- lable
Similarly, a one-way ANOVA test conducted by considering normalized maximumf0 as dependent variable and syllable position as factor also reveals the same pattern (Singrijhan,F(1,4710) = 298.997, p <0.001; Sessa,F(1,1907) = 109.652, p <0.001;
Koilamari,F(1,3080) = 78.969,p <0.001; Lamabari, F(1,2534) = 1.446,p = 0.229).
Hence, from the analysis of fundamental frequency it is confirmed that, except for Assam Sora speech data recorded from Lamabari region, higher pitch variation in second syllable contributes to the prominence of second syllable in Assam Sora disyl- lables. Unfortunately, the reason for finding an exception in Assam Sora speech data recorded from Lamabari region is not known yet.