Korean is particularly interesting because here it is even clearer than in Thai that
the differences cannot be explained just by differences in laryngeal-oral
timing.
The three voicing categories can be described as follows:
Voiceless aspirated.
Voiceless unaspirated, or weakly aspirated in initial position,
often voiced intervocalically; frequently referred to as "lenis".
Voiceless aspirated; "fortis" (also referred to as "forced"),
particularly characteristic of these stops is the abrupt onset of phonation
after closure release ("stiff voice", Ladefoged & Maddieson, 2008, p. 56),
i.e. it can be assumed that there is much higher tension in the vocal folds
at voice onset in this category compared to the lenis category.
Examples from IPA
Please note that p*, t*, etc. stand for the unaspirated fortis plosives
(Ladefoged & Maddieson 2008, p. 56), which in standard IPA are symbolized simply
as p, t, etc.
pʰal ("arm")
tʰal ("riding")
tʃʰal ("kicking")
kʰal ("knife")
pal ("foot")
tal ("moon")
tʃal ("well")
kal ("going")
p*al ("sucking")
t*al ("daughter")
tʃ*al ("squeezing")
k*al ("spreading")
Examples from our own data at IPS
For this speaker, there is actually a fairly consistent difference in VOT between
the lenis and the fortis stops (longer VOT for the lenis stops). However, the more
abrupt onset of the higher harmonics in the fortis category illustrates well the
difference in voice quality at vowel onset following these two sounds. In addition, a
consistent difference in fundamental frequency at voicing onset is also found (higher
for the fortis category). Findings of this kind have also been reported by Dart (1987).
pʰ
p
p*
Additional Analyses of the Korean data from IPS
Measurements of VOT for a large number of tokens show that the lenis
stop indeed lies on average between the fortis and aspirated stops,
but that individual tokens can be very close to both these categories. Click here
to see a plot of the VOT values. VOT is on the y-axis, the x-axis simply
arranges the tokens in the order they occurred in during the experiment.
For this speaker, we also measured intra-oral air-pressure. Despite
its name, there were no clear differences between the fortis stop
and the other two categories in terms of peak air-pressure during
consonantal closure. It can be seen in the sonagrams, however, that
the duration of closure is clearly longer for the fortis stop than
for the lenis stop.
A combination of voice quality and fundamental frequency (measured
at the onset of phonation) also distinguishes the three categories
well. Thus even when VOT is ambiguous, adequate differentiation
is possible. Click here
to see a plot of fundamental frequency (y-axis) plotted against
the closed quotient (x-axis). The color coding is the same as in
the VOT plot above. The closed quotient, which was measured from
the electroglottographic waveform, expresses the duration of glottal
closure as a proportion (in percent) of the duration of a complete vibratory
cycle during phonation. Lower values indicate more forceful glottal
closure and should be accompanied by stronger upper harmonics in
the specturm (refer back to the sonagrams).
Conclusion:
The lenis stop has lower fundamental frequency than the other two categories.
The fortis stop has higher closed quotient values than the other two categories.
Audio Source:
Illustrations of the IPA.
Institute of Phonetics and Speech Processing, LMU Munich.
References:
Ladefoged, P. / Maddieson, I. (2008): The Sounds of the World's Languages. Malden, MA: Blackwell.