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


Design of the sentence materials



Key to symbols used in sentence design tables



Keywords

The recorded sentences are designed around sixteen different test syllables, each spoken in three different lexical contexts: monosyllable, disyllable and trisyllable. The word containing the test syllable is referred to as the ``keyword''. The three keywords for each test syllable are referred to as ``keyword triads''. There are eight keyword triads in which the test syllable is always word-initial, the ``left-headed'' keywords, and eight in which the test syllable is always word-final, the ``right-headed'' keywords. The full set of right-headed keywords is shown in Table 1 and the full set of left-headed keywords is shown in Table 2.

The syllables immediately adjacent to the test syllable are unstressed, to control stress-adjacent lengthening, and are identical in the disyllables and trisyllables in each keyword triad, to minimise any effect of local phonetic variation on test syllable duration. Stress and phonetic environment are also controlled in the carrier sentence.

Table 1: Right-headed keywords used in the recordings
Note: Transcriptions are given for Standard Southern British English. Unlike American English, juice and the final syllable of produce and reproduce are homophonous, and pronounced as shown.


Table 2: Left-headed keywords used in the recordings


Carrier sentences

For the recordings, the keywords are placed in sentences designed to be read as meaningful utterances, with a full normal prosodic realisation. At the same time, the length of the utterances and the immediate environment of the test syllables is controlled. The full design of the experimental materials is shown schematically in Table 2 for right-headed keywords. The design of the materials for left-headed keywords is the mirror-image of the syllable configurations shown.

The design shown in Table 3 systematically varies word length and utterance length. Where the test syllable is utterance-medial, these factors are varied both independently and concurrently. Where the test syllable is close to the utterance edge, utterance length covaries with word length and varies whilst word length is fixed.

Table 3: Schematic representation of the keywords and carrier sentences

Example sentences illustrate the five series of test syllable environments resulting from this design (see EXAMPLES page).



Up to EUSTACE HOME page Forward to EXAMPLES page