Kyung-ho Loken-Kim, Yoshihiro Kitagawa, Yoko Ohta
Linguistic Analysis of Speech Disfluency in
the ATR Spoken Language Database
Abstract:This is the first of two technical reports on speech disfluencies found in the ATR spoken
language database. This report presents the statistical analysis of the orthographic transcriptions,
and provides the following: 1) probability of fluent speech at various sentence lengths, 2) the
structure of reparandums (see section 2 in this report), 3) disfluency patterns and occurrences, and
4) word fragments and their significance in speech disfluency. The results of the acoustical
analysis will be reported in a subsequent report (TR-IT-0108).