198 Adrian Simpson
8000Hz1700ms7500Hz7000Hz
6500Hz
6000Hz
5500Hz5000Hz
4500Hz
4000Hz
3500Hz3000Hz
2500Hz2000Hz
1500Hz
1000Hz
500Hz
0Hz–40dBHz–50
–60
–70
–80
–90
–100
–110
–120 0 1000 2000 3000 4000 5000 6000 7000 8000–40dBHz–50
–60
–70
–80
–90
–100
–110
–120 0 1000 2000 3000 4000 5000 6000 7000 80001750ms1800ms1850ms1900ms1950ms8000Hz1400ms7500Hz7000Hz
6500Hz
6000Hz
5500Hz5000Hz
4500Hz
4000Hz
3500Hz3000Hz
2500Hz2000Hz
1500Hz
1000Hz
500Hz
0Hz 1450ms1500ms1550ms1600ms1650msFigure 4. Spectrograms and DFT sections of (a) ejective release of word-final plosive in
weht (ein) ‘blows’ and (b) pulmonically released word-final and pre-pausal plosive in mit
‘with’. Vertical arrows indicate location of plosive release (from Simpson 2007).The first account is that there is active movement of the larynx involved. However,
this account is problematic as it implies that such ejectives are not epiphenom-
enal, but actively produced, and the question of motivation arises, i.e. why would
a speaker choose to actively raise the larynx in this context? An enhancement of
the place of the articulation of the plosive in a burst which might otherwise be
masked by glottal closure would be a possible motivation. Nevertheless, a sec-
ond account seems more plausible since, on the one hand it explains the relative
intensity of some release bursts, and at the same time the explanation remains
epiphenomenal. This account assumes that air is still flowing through the glottis
after the supraglottal closure has been made, but prior to the point of oral stop
release the glottis is closed or configured for creak. The suggested mechanism is
shown in Figure 5 using Catford’s (1977) method of schematisation. Following
supraglottal stop closure, intraoral air pressure increases due to pulmonic airflow
through the glottis. At some point prior to the release of supraglottal closure, the
glottis closes or is configured for creak. When the oral closure is finally released,
the burst is auditorily and acoustically an ejective, although the pressure build-up