Preaspiration in the Nordic Languages: Synchronic and Diachronic ...

Preaspiration in the Nordic Languages: Synchronic and Diachronic ... Preaspiration in the Nordic Languages: Synchronic and Diachronic ...

www2.lingfil.uu.se
from www2.lingfil.uu.se More from this publisher
26.07.2013 Views

(i.e., short mVOffT and long mVOT). For unaspirated and preaspirated stops, there is often a considerable time lag between modal voice offset and stop closure, but not between stop release and modal voice onset (i.e., long mVOffT and short mVOT). Such a difference in timing relations cannot be attributed to articulatory reduction, and there is no reason to assume that preaspirated stops are derived from postaspirated stops by some type of articulatory or aerodynamic mechanism. At the same time, this does not mean that articulatory and/or aerodynamic factors are irrelevant in an explanation of why some CSw speakers exploit preaspiration (see Ch. 5). Other aspects of variation in fortis stops can be attributed to hypospeech. For example, instances in which the fortes are fricated and do not have a full stop closure can be seen as articulatory reduction. Also, the within-speaker variation in the amount of postaspiration or preaspiration produced may be a function of the hyper-hypo continuum. Lastly, it may be suggested that postaspiration in CSw fortis stops functions as a boundary marker (i.e., a “Grenzensignal,” cf. Trubetzkoy 1962:255), marking the onset of a word or a stressed syllable. Likewise, the absence of postaspiration, and the concomitant presence of preaspiration for some speakers, signals the absence of a word boundary (“negatives Grenzensignal,” ibid.) and indicates that the following vowel is unstressed. In CSw lenis stops, there does not seem to be such a functional division between word-initial and word-medial positions. 4.3 Tórshavn Faroese In this section, the stop production of four Tórshavn Faroese speakers is analysed, both from a quantitative and a qualitative point of view. First, in order to make it easier to refer to stops in different contexts, they will be designated to specific word types (section 4.3.2.1). Then, some problems regarding the phonological distribution of preaspirated stop variants in Faroese are adressed. These problems become evident when the literature on Faroese preaspiration is reviewed (cf section 3.3). The approach adopted here is to analyse the stop duration data without making a priori assumptions as to their phonological status as fortis or lenis stops (sections 4.3.2.3 and 4.3.2.4). Section 4.3.2.4 also – 146 –

contains an analysis of data on the proportional duration of preaspiration (i.e. the duration of preaspiration in relation to, e.g., the preceding vowel) for the four Faroese subjects. This allows us to compare the durational aspects of preaspiration production in Faroese with that of, for example, Central Standard Swedish. Lastly, section 4.3.3 contains a discussion of the variation in the production of preaspirated stops in Faroese. 4.3.1 Method and subjects The Faroese data come from four subjects, two male (SG and EI) and two female (MS and BJ), all speakers of the Tórshavn dialect. SG, EI and BJ are in their thirties and MS in her forties. Spontaneous speech interaction between the subjects was elicited using map tasks. The subjects were recorded in pairs, with EI and MS as one pair and SG and BJ as another. The material was transliterated by a native speaker of Tórshavn Faroese. Since Faroese orthography is not a good predictor of phonemic representation, the transliterations were converted from orthographic to phonemic representations using a grapheme-to-phoneme conversion program. 1 The total recording time was 20–35 minutes per subject. The amount of material analysed for each subject was for EI, approximately 9.5 minutes of uninterrupted speech 2 , comprising 2344 words: for MS 9.5 minutes, 2179 words; for SG 8.4 minutes, 1961 words; and for BJ 13.2 minutes, 2950 words. A rough estimation of speaking rate indicated that, EI uttered 6.37 syllables per second, MS 5.65, SG 5.82 and BJ 5.65 (cf. section 4.2.1 on how this estimate was obtained). Further details on the recording procedure and segmentation criteria were given in section 4.1 above. 4.3.2 Word-medial and word-final postvocalic stops 4.3.2.1 Stop and context classification In this section, Faroese stops and the contexts in which they occur are categorised on the basis of three variables (see Table 4–11). This will 1 This program was developed in connection with a project on developing a Faroese text-to-speech synthesis system (cf. Helgason & Gullbein, 2002). 2 Cf. section 4.2.1. – 147 –

conta<strong>in</strong>s an analysis of data on <strong>the</strong> proportional duration of preaspiration<br />

(i.e. <strong>the</strong> duration of preaspiration <strong>in</strong> relation to, e.g., <strong>the</strong> preced<strong>in</strong>g vowel)<br />

for <strong>the</strong> four Faroese subjects. This allows us to compare <strong>the</strong> durational<br />

aspects of preaspiration production <strong>in</strong> Faroese with that of, for example,<br />

Central St<strong>and</strong>ard Swedish. Lastly, section 4.3.3 conta<strong>in</strong>s a discussion of<br />

<strong>the</strong> variation <strong>in</strong> <strong>the</strong> production of preaspirated stops <strong>in</strong> Faroese.<br />

4.3.1 Method <strong>and</strong> subjects<br />

The Faroese data come from four subjects, two male (SG <strong>and</strong> EI) <strong>and</strong><br />

two female (MS <strong>and</strong> BJ), all speakers of <strong>the</strong> Tórshavn dialect. SG, EI <strong>and</strong><br />

BJ are <strong>in</strong> <strong>the</strong>ir thirties <strong>and</strong> MS <strong>in</strong> her forties. Spontaneous speech <strong>in</strong>teraction<br />

between <strong>the</strong> subjects was elicited us<strong>in</strong>g map tasks. The subjects<br />

were recorded <strong>in</strong> pairs, with EI <strong>and</strong> MS as one pair <strong>and</strong> SG <strong>and</strong> BJ as ano<strong>the</strong>r.<br />

The material was transliterated by a native speaker of Tórshavn<br />

Faroese. S<strong>in</strong>ce Faroese orthography is not a good predictor of phonemic<br />

representation, <strong>the</strong> transliterations were converted from orthographic to<br />

phonemic representations us<strong>in</strong>g a grapheme-to-phoneme conversion<br />

program. 1<br />

The total record<strong>in</strong>g time was 20–35 m<strong>in</strong>utes per subject. The amount<br />

of material analysed for each subject was for EI, approximately 9.5<br />

m<strong>in</strong>utes of un<strong>in</strong>terrupted speech 2 , compris<strong>in</strong>g 2344 words: for MS 9.5<br />

m<strong>in</strong>utes, 2179 words; for SG 8.4 m<strong>in</strong>utes, 1961 words; <strong>and</strong> for BJ 13.2<br />

m<strong>in</strong>utes, 2950 words. A rough estimation of speak<strong>in</strong>g rate <strong>in</strong>dicated that,<br />

EI uttered 6.37 syllables per second, MS 5.65, SG 5.82 <strong>and</strong> BJ 5.65 (cf.<br />

section 4.2.1 on how this estimate was obta<strong>in</strong>ed). Fur<strong>the</strong>r details on <strong>the</strong><br />

record<strong>in</strong>g procedure <strong>and</strong> segmentation criteria were given <strong>in</strong> section 4.1<br />

above.<br />

4.3.2 Word-medial <strong>and</strong> word-f<strong>in</strong>al postvocalic stops<br />

4.3.2.1 Stop <strong>and</strong> context classification<br />

In this section, Faroese stops <strong>and</strong> <strong>the</strong> contexts <strong>in</strong> which <strong>the</strong>y occur<br />

are categorised on <strong>the</strong> basis of three variables (see Table 4–11). This will<br />

1 This program was developed <strong>in</strong> connection with a project on develop<strong>in</strong>g a Faroese<br />

text-to-speech syn<strong>the</strong>sis system (cf. Helgason & Gullbe<strong>in</strong>, 2002).<br />

2 Cf. section 4.2.1.<br />

– 147 –

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!