The Centre for Speech Technology Research, The university of Edinburgh
EUSTACE
HOME DESIGN EXAMPLES SENTENCES PROCEDURE
RESULTS DISSERTATION LICENCE DOWNLOAD GLOSSARY


Procedure: Additional syllable labels


The additional syllables in disyllabic and trisyllabic keywords were measured for Keyword Series C and Keyword Series E (Series B and D feature only monosyllabic keywords - see EXAMPLES page for discussion of the keyword series). The labelling procedure was the same as for the test syllable, with labelling done keyword-by-keyword, and the order of speech files and speakers randomised. For the unstressed syllables, however, only full syllable durations were measured.

The additional syllables are labelled FIRST and SECOND. For the left-headed keywords, FIRST immediately follows the primary stressed syllable in the disyllable and trisyllable, and SECOND follows FIRST in the trisyllable: for example, in cap.tain.cy, ``tain'' is FIRST and ``cy'' is SECOND. For the right-headed keywords, FIRST immediately precedes the primary stressed syllable in the disyllable and trisyllable, and SECOND precedes FIRST in the trisyllable: for example, in mis.re.port, ``re'' is FIRST and ``mis'' is SECOND.

[Note: in English speech timing: a domain and locus approach (see DISSERTATION page), the syllables FIRST and SECOND are referred to as ``syllable-2'' and ``syllable-3'' respectively. This usage is not adopted here, for consistency with the original labelling scheme as illustrated in the .uns label files.]

The placement of labels for unstressed syllables is illustrated in Table 1 and Table 2 (these tables are adapted from Tables E.3 and E.4 in Appendix E of English speech timing: a domain and locus approach - see DISSERTATION page). The criteria used for segmentation are described in Appendix A of the dissertation.

Table 1: Placement of additional syllable labels for the left-headed keywords. The start of SECOND is taken from the end of FIRST.

The test-syllable-final stops in spec.ta.cle and cap.tain.cy are not always released, thus the closure durations of the FIRST onsets are not reliably measurable and so, as shown in Table 1, the syllables are measured from the FIRST stop release. The SECOND final stops are not always released in dog.ma.tist, part.ner.ship and cens.or.ship, so an additional label, SECOND.mid, is placed at the point of stop closure.

Table 2: Placement of additional syllable labels for the right-headed keywords. The start of FIRST is taken as the end of SECOND.

For the right-headed keywords, the onset of syllable-initial stop closure could not be measured utterance-initially. Thus, for the disyllables commend, dispose, produce, compose and descend, the stop release in the onset of FIRST is also labelled, as FIRST.mid. Similarly, for the trisyllables presuppose, decompose and condescend, the stop release in the onset of SECOND is also labelled, as SECOND.mid.

The FIRST-initial /h/ of the keyword humane is realised quite differently in the two contexts develop humane and inhumane. An additional label, FIRST.mid, is used to indicate the onset of voicing in the /h/, which coincides with the onset of /h/ itself in inhumane, but is later than the stop release in develop humane. The durations of /h/ by either measure may not be comparable in these two contexts and this comparison was excluded from the analyses in the original research.

Additional syllable labelling is discussed further in Section 4.3 of the dissertation.



Up to PROCEDURE page