Error resolution strategies during speech interaction

This paper describes a study on error resolution strategies chosen when speech recognition errors occur in a command and control environment. Thirty-eight subjects participated in the study. The results indicate that strategies differ depending on the length of the error sequence. Users chose the following resolution strategies: switch to another input device, accept result, repeat command, work around using another set of commands to reach the same goal, or change goal. After incorrect recognition, users select another resolution strategy (e.g., the switch strategy). Users gave up on speech input after a maximum affive attempts to achieve the desired result.