In this method, two subjects read a prepared text in synchrony with one another, using a methodology described further in [2]. At the start of each reading, subjects were given a countdown by the experimenter, and were asked to commence reading when the count reached zero. Dysfluencies were caught and the sentence restarted with a new countdown. In the final corpus, any such dysfluencies have been omitted, leaving only the fluent readings.
Note: because dysfluencies have been omitted, the pauses in the resulting corpus are not to be taken as actual pauses for the purposes of research.
http://chains.ucd.ie/corpus.php
