Inter-Speaker Interaction in Speech Rhythm:<P> Some Durational 
Properties of Sentences and Intersentence Intervals in 
Paragraph Reading
<P>
Kuniko KAKITA
<P>
Department of Liberal Arts and Sciences, 
Faculty of Engineering, Toyama Prefectural University<P>
Kosugi-machi, Imizu-gun, Toyama 939-03, Japan<P>
e-mail: kakita@pu-toyama.ac.jp
<P>
	The aim of the present study was to see if one 
speaker's speech rhythm was affected by another speaker's 
speech rhythm, and if so, how. The speech style examined was
reading of paragraphs consisting of short sentences. The 
speech parameters observed were the sentence duration and 
the intersentence interval duration, the two most essential 
and direct determinants of speech rhythm in reading of 
paragraphs.

	Utterance texts consisting of five short sentences 
were read by five speakers in three different ways. In one 
instance, a series of five sentences was read through by a 
single speaker; in the other two instances, a speaker took 
over from the preceding speaker (a) after the first sentence
and (b) after the third sentence. The duration of the 
sentences and the intersentence intervals following the 
takeover were compared with those at corresponding locations
in single speaker readings.  

	The results showed that the speakers' speech was 
indeed affected by the preceding speaker's speech. That is, 
both the sentence duration and the interval duration 
deviated from one's 'preferred' duration when exposed to 
another speaker's speech. However, sentence duration and 
interval duration differed characteristically in the way 
they were affected by another speaker's speech. The amount 
of deviation from one's 'preferred' duration was small for 
sentence duration while it was great for interval duration 
for all subjects. The across-subject variation in the amount
of deviation was small for sentence duration while it was 
great for interval duration. This was interpreted to reflect
the fact that the production of sentences is physically 
(articulatorily) constrained and is restricted in 
variability whereas the production of intervals has no such 
constraints and is much less restricted in variability. A 
speaker's speech was affected by the preceding speaker's 
speech mostly in an assimilative manner, the duration of 
sentences and intervals becoming more like those of the 
preceding speaker after the takeover. However, in some 
instances, the speakers adjusted the duration of the 
sentences and the intervals after the takeover in an 
opposite, or dissimilative, direction.

	The present study has provided some novel findings 
about what happens to one's speech when it interacts with 
another speaker's speech, and has confirmed that this line 
of study is well worth pursuing. An examination of acoustic 
parameters other than the ones reported here, e.g. the 
fundamental frequency and amplitude, are currently under way
to complement the results of the present study, and 
ultimately to provide the basis for a formulation of the 
speech interaction phenomena.
<P>
Keywords: interaction, speech rhythm, speech rate, pause 
duration<P>