Even when speaking conditions and listener responses are very tightly
controlled, some talkers are easier to understand than others. In a se
ries of intelligibility tests using both English words without context
and English sentences, both native and non-native listeners found one
of five talkers difficult to understand. Since all talkers read simil
ar materials and the task was the same for all listeners, the differen
ces in intelligibility must have resulted from particular phonetic cha
racteristics used by the talker. Spectrograms were made of all test wo
rds produced by the talkers and compared on selected acoustic-phonetic
properties. In comparison with the more intelligible talkers, the lea
st intelligible talkers produced test words at shorter durations; abbr
eviated vowel durations; used the least differentiated vowel space, as
defined by the first two formants; used minimal cues for consonantal
contrasts; and had the most varied amplitude of stressed vowels. These
characteristics are similar to those distinguishing deliberately clea
r speech. That non-native and native listeners found the same talker d
ifficult to understand suggests that the effect of clear speech, thoug
h different in degree, reflects the use of the same acoustic-phonetic
information by both groups of listeners. Further, clear speech has sim
ilar acoustic-phonetic characteristics whether deliberately or inadver
tently produced.