Abstract
The method of continuous discourse tracking was evaluated in terms of the strategies used by the talker and the types of responses elicited from the listener. Talker utterances were classified into four categories:
1. 1. Complete repetition (of the initial utterance);
2. 2. Partial repetition;
3. 3. Repetition with change in emphasis; and
4. 4. A combined strategy using two or more correction strategies.
Listener responses were classified into three categories:
1. 1. Correct repetition of intended utterance;
2. 2. Partially correct repetition; and
3. 3. Totally incorrect repetition or no response.
The listeners showed small but statistically significant differences in their response patterns. Much larger differences were observed in the pattern of correction strategies used by the talkers. Differences in correction strategy were also observed between the early and later stages of a talker—listener exchange.