Preaspiration in the Nordic Languages: Synchronic and Diachronic ...
Preaspiration in the Nordic Languages: Synchronic and Diachronic ... Preaspiration in the Nordic Languages: Synchronic and Diachronic ...
very large, since postaspiration was approximately only 7–9 ms shorter on average in till type (i.e., function) words than in tennis type (i.e., content) words. 140 120 100 80 60 40 20 0 GT CK MP FS Figure 4–4. The range of postaspiration durations for word-initial fortis stops for the four CSw subjects. 1 The mean postaspiration duration of initial stops in taverna type words was slightly shorter than in tennis type words for 3 of the 4 subjects, but t-tests do not indicate that the differences are significant. The difference was even smaller than that observed between the tennis type and the till type words (i.e., content vs. function words). It should be noted that there are far fewer tokens for the taverna type words than for the other categories, which affects the significance in the t-tests. There were also differences in mean postaspiration duration between subjects. For tennis type words, subject FS had the highest mean, 58 ms. GT and MP had similar mean durations, 46 and 44 ms respectively. CK 1 The notched box shows the median (central line) and the lower and upper quartiles (lower and upper box boundaries). The box thus comprises 50% of the observed data. The notched area shows the confidence interval around the median. The dotted-line (i.e. whisker) connects the nearest observations within 1.5 IQRs (inter-quartile ranges) of the lower and upper quartiles. The vertical line to the left of the box shows the 90% percentile range. The diamond in that line shows the mean and the 95% confidence interval around the mean. Crosses (+) and circles () indicate outliers, observations more than 1.5 IQRs (near outliers) and 3.0 IQRs (far outliers) from the quartiles. – 114 –
had an intermediate mean of 52 ms. For till type words, this pattern is repeated. FS had the highest mean, 50 ms, GT and MP had the lowest, 38 ms and 37 ms respectively, and CK had an intermediate mean of 45 ms. Table 4–2. Mean postaspiration duration (Po), in ms, for word-initial fortis stops in three different word groups (see text) and for all four CSw subjects. Standard deviations are given within parentheses and the number of tokens in italics. GT CK MP FS Po n Po n Po n Po n tennis 44 (18) 17 52 (11) 43 46 (14) 67 58 (14) 66 till 38 (13) 23 45 (13) 38 37 (14) 71 50 (19) 50 taverna 58 (20) 4 40 (6) 5 44 (13) 16 55 (11) 11 An analysis of the relationship of stop closure duration and postaspiration duration did not reveal any correlation between the two. This is perhaps not surprising, since stop closure duration varies greatly in the present data. One source for this variability is the fact that the context preceding the stops may contain any type of segment, a vowel, a sonorant, a stop or a fricative. Also, since the data are from unscripted (spontaneous) speech, factors such as prominence and speech rate are not controlled for. As a result, closure durations range from being practically absent (which can occur when a word-final nasal precedes a word-initial stop) to extending to several hundred ms. The data do not indicate that there are systematic differences in mean postaspiration duration between the different places of articulation (labial, dental and velar). Again, this is perhaps to be expected, since the material is from spontaneous speech and thus, for example, the number of tokens for each stop category, their stress conditions etc. are not controlled for. Analysis of the stop types with a further breakdown of the material (e.g., into content and function words) does not yield statistically valid results, since each category has such a small number of tokens. These data are therefore not very useful in examining the possible differences in the duration of closure and aspiration between different places of articulation. – 115 –
- Page 76 and 77: grams), with the occasional aid of
- Page 78 and 79: quantity system instead of the ON f
- Page 80 and 81: ON flytja). There are no examples o
- Page 82 and 83: Table 3-7. Reflexes of ON stops in
- Page 84 and 85: whether she thinks that leaving out
- Page 86 and 87: (Hede, Linsäll and Sveg) as well a
- Page 88 and 89: (VC vs. VC). For two pairs (/tak/ v
- Page 90 and 91: stop sequences that remain to be di
- Page 92 and 93: do occur systematic linguistic diff
- Page 94 and 95: Gräsö Forsmark Valö S W E D E N
- Page 96 and 97: There are very few examples of clus
- Page 98 and 99: Kökar, in the Åboland archipelago
- Page 100 and 101: one hand, speakers have a tendency
- Page 102 and 103: ecent studies that corroborate thes
- Page 104 and 105: example, in a particular dialect, t
- Page 106 and 107: does not give her segmentation crit
- Page 108 and 109: Scotland and Ireland 1 [hphthk] 2 [
- Page 110 and 111: approximately 140-150 ms in VC syll
- Page 112 and 113: 1. South Saami (Southern & Ume Saam
- Page 114 and 115: 1987a), Estonian (Engstrand & Krull
- Page 116 and 117: Table 3-13. Reflexes of ON postvoca
- Page 118 and 119: and Tórshavn Faroese data were eli
- Page 120 and 121: ducing breathiness into the voice q
- Page 122 and 123: Figure 4-3. An example of a sequenc
- Page 124 and 125: 4.2 Central Standard Swedish (CSw)
- Page 128 and 129: Several spectrographic examples of
- Page 130 and 131: The data indicate that there is lit
- Page 132 and 133: produced and in 9 cases there were
- Page 134 and 135: To examine effects of syllable stru
- Page 136 and 137: preaspiration for lexically stresse
- Page 138 and 139: durations exceeding 120 ms, preaspi
- Page 140 and 141: mately 70 ms. Just before the stop
- Page 142 and 143: noise becomes audibly postalveolar
- Page 144 and 145: As regards nasal + fortis stop sequ
- Page 146 and 147: most of the remaining instances he
- Page 148 and 149: (comparable to our taverna type) ha
- Page 150 and 151: The CSw data have shown that word-i
- Page 152 and 153: golf ‘golf,’ in which the utter
- Page 154 and 155: o Figure 4-35. Subj. GT: […tv]
- Page 156 and 157: word-initial lenis stop preceded by
- Page 158 and 159: (i.e., short mVOffT and long mVOT).
- Page 160 and 161: make it easier to refer to stops in
- Page 162 and 163: [mh] ‘food’ is used, and for fu
- Page 164 and 165: clear stop correlate, 26 (7%) had a
- Page 166 and 167: All subjects considered, for lenis
- Page 168 and 169: Figure 4-45(a) plots preaspiration
- Page 170 and 171: cluster and 2 instances with a kt c
- Page 172 and 173: 4.3.2.5 Postaspiration duration Ins
- Page 174 and 175: closure gesture. As the escape chan
very large, s<strong>in</strong>ce postaspiration was approximately only 7–9 ms shorter<br />
on average <strong>in</strong> till type (i.e., function) words than <strong>in</strong> tennis type (i.e.,<br />
content) words.<br />
140<br />
120<br />
100<br />
80<br />
60<br />
40<br />
20<br />
0<br />
GT CK MP FS<br />
Figure 4–4. The range of postaspiration durations for word-<strong>in</strong>itial fortis stops<br />
for <strong>the</strong> four CSw subjects. 1<br />
The mean postaspiration duration of <strong>in</strong>itial stops <strong>in</strong> taverna type<br />
words was slightly shorter than <strong>in</strong> tennis type words for 3 of <strong>the</strong> 4<br />
subjects, but t-tests do not <strong>in</strong>dicate that <strong>the</strong> differences are significant.<br />
The difference was even smaller than that observed between <strong>the</strong> tennis<br />
type <strong>and</strong> <strong>the</strong> till type words (i.e., content vs. function words). It should be<br />
noted that <strong>the</strong>re are far fewer tokens for <strong>the</strong> taverna type words than for<br />
<strong>the</strong> o<strong>the</strong>r categories, which affects <strong>the</strong> significance <strong>in</strong> <strong>the</strong> t-tests.<br />
There were also differences <strong>in</strong> mean postaspiration duration between<br />
subjects. For tennis type words, subject FS had <strong>the</strong> highest mean, 58 ms.<br />
GT <strong>and</strong> MP had similar mean durations, 46 <strong>and</strong> 44 ms respectively. CK<br />
1 The notched box shows <strong>the</strong> median (central l<strong>in</strong>e) <strong>and</strong> <strong>the</strong> lower <strong>and</strong> upper quartiles<br />
(lower <strong>and</strong> upper box boundaries). The box thus comprises 50% of <strong>the</strong> observed data.<br />
The notched area shows <strong>the</strong> confidence <strong>in</strong>terval around <strong>the</strong> median. The dotted-l<strong>in</strong>e<br />
(i.e. whisker) connects <strong>the</strong> nearest observations with<strong>in</strong> 1.5 IQRs (<strong>in</strong>ter-quartile ranges)<br />
of <strong>the</strong> lower <strong>and</strong> upper quartiles. The vertical l<strong>in</strong>e to <strong>the</strong> left of <strong>the</strong> box shows <strong>the</strong> 90%<br />
percentile range. The diamond <strong>in</strong> that l<strong>in</strong>e shows <strong>the</strong> mean <strong>and</strong> <strong>the</strong> 95% confidence<br />
<strong>in</strong>terval around <strong>the</strong> mean. Crosses (+) <strong>and</strong> circles () <strong>in</strong>dicate outliers, observations<br />
more than 1.5 IQRs (near outliers) <strong>and</strong> 3.0 IQRs (far outliers) from <strong>the</strong> quartiles.<br />
– 114 –