Speech Errors Management in Air Traffic Control Communications: A Detailed Study

Many studies have reported on some human factors influencing the communication process, especially in aeronautical framework (see Davison (2003) for example). When spoken, communication comprises three different components: production, perception and understanding. The communication is often disturbed by one or many errors that affect one or several of these components. Consequently, one way to make air traffic control (ATC) communications more efficient and robust is to have as much knowledge as possible on these problems and their usual management. This paper presents the interests brought by corpus-based studies to Air Traffic Control (ATC) applications, especially interactions/communication between controllers and pilots. The corpus recorded represent dialogues during exercises where air-traffic controllers being formed interact/converse with people simulating pilots in practice. We propose error and strategies typology in accordance with the phraseology Then, we describe the principles and the specification adopted both for the recording and the annotation of corpus. Then, we report first results obtained from corpus analyses on errors and correction strategies of the air-traffic controller, and comment them in regards with ATC oriented applications