VOWELS IN STANDARD AUSTRIAN GERMAN - Acoustics ...

67
Vowels in Standard Austrian German
the duration of the periods undergo changes in the first place. Therefore, it is not
surprising that absolutely no correlation can be observed between vowel duration and
F0, whereas a high correlation can be found between vowel duration and the number of
periods. This result holds for logatome reading task, the sentence reading task, and
spontaneous speech. Table 4.4 gives the correlation coefficient r for all speakers:
r Logatomes Reading Spontaneous Logatomes Reading Spontaneous
NoP/ms NoP/ms NoP/ms ms/F0 ms/F0 ms/F0
Sp012 0.96 0.87 0.94 0.16 0.02 0.07
Sp180 0.94 0.84 0.89 0.30 0.23 0.11
Sp082 - 0.89 0.83 - 0.22 0.11
Sp129 - 0.91 0.92 - 0.17 0.22
Sp126 - 0.92 0.96 - 0.15 0.12
Sp127 - 0.85 0.96 - 0.25 0.17
Table 4.5: Correlation coefficient r for NoP/ms and F0/ms for three speaking tasks, stressed
vowels, all speakers. Statistically significant results (p < 0.05) are in bold.
The number of periods (henceforth NoP) does not contain any information about F0 and
can therefore be taken as a pure durational measure 62 . Given the high correlation
between duration (in ms) and the number of periods, it can be concluded that the longer
a segment, the more periods it contains. Since fundamental frequency is hardly affected
by a truncation of periods, the number of periods renders more exact results than
traditional duration measurements.
The results on NoP display a high dependency on the speaking task. In the
logatome reading task, speaker sp180 discerns long and short vowels, i.e. vowels
termed as “tense” have a higher amount of NoPs than the vowels labeled “lax” (27 vs.
14, n = 313, t = 22.58, p = 0.00). The vowel /a/ is differentiated for NoP as well (27 vs.
14, n = 47, t = 10.74, p = 0.00). Duration, therefore, coincides with tenseness. Figure 4.6
shows the results of a cluster analysis for speaker sp180, which groups long/tense
vowels and short/lax vowels into two clusters:
62 In the chapters 3.2.2 and 3.2.3, NoPs could not be calculated, because no periodic signal
is available in the closure phase of unvoiced plosives.