Following a previous study using locus equation (LE) and electropalatographic (EPG) data to examine coarticulation of voiced consonants and vowels in CV syllables [1], the present study examines voiceless stops and fricatives using the same analysis techniques. It is found that when LE data for stops is sampled at the stop burst, rather than at vowel onset, the correlation between LE data on coarticulation and EPG data on coarticulation is quite high. By contrast, results for the fricatives are quite poor. It is suggested that the LE is capable of capturing rather gross differences in coarticulatory resistance, such as that involving a tongue tip rather than a tongue body, but that it is not capable of capturing more subtle differences in coarticulation, such as those involving different coronal articulations. This explanation is supported by work in progress on Australian Aboriginal languages which have up to four coronal places of articulation [2].
Cite as: Tabain, M. (1998) Consistencies and inconsistencies between EPG and locus equation data on coarticulation. Proc. 5th International Conference on Spoken Language Processing (ICSLP 1998), paper 0668, doi: 10.21437/ICSLP.1998-366
@inproceedings{tabain98_icslp,
author={Marija Tabain},
title={{Consistencies and inconsistencies between EPG and locus equation data on coarticulation}},
year=1998,
booktitle={Proc. 5th International Conference on Spoken Language Processing (ICSLP 1998)},
pages={paper 0668},
doi={10.21437/ICSLP.1998-366}
}