There is an emerging interest in how people coordinate multiple activities with multiple partners. One important aspect relates to how interruptions are managed. Interruptions often disrupt a task-related interaction between participants. Dealing with them requires suspending the interaction, resolving the problem, reinstating the interaction, and continuing the task. Suspending interaction is the process by which participants in an interaction decide to stop interacting for an indefinite period of time, without taking leave of each other. In many cases, they wait for another to finish the interrupting task (e.g., holding the line on the telephone). Suspending interaction is a recurrent process in the coordination of complex tasks.

The goal of the project was to test a model of the suspension process. Several interactional problems must be solved in a suspension, and any model must specify those problems, their sequence, and the factors that influence them. Important steps in the suspension process are (1) initiating the suspension, (2) excusing, justifying, or explaining the suspension, (3) dealing with the interruption, (4) initiating reinstatement, and (5) continuing the task. Two aspects are important. First, facework and politeness is often involved in suspensions, because asking one's partner to wait while one does something else is face-threatening. Two factors that to affect the extent of the politeness phenomena that occur are the degree of personal responsibility of the participant proposing the suspension and the length of the suspension. Second, continuing the task requires reconstructing a joint perspective on the task (e.g., "where were we?"). This process is called backtracking. It is affected by factors like the duration of the suspension, the person interrupted, the point at which the task is suspended or the available cues.

Four studies tested the model. Study 1 used existing corpus data to code naturally-occurring suspensions in telephone conversations. Results revealed that the duration of the interruption and the conversational role of the initiator affected both politeness during interaction and task reinstatement. Study 2 replicated findings of Study 1 in an experimental setting. Pairs of participants came to the laboratory and discussed a topic together. Their discussion was interrupted at several points. The duration of the interruption and the conversational role of the person interrupted (speaker vs. listener) was manipulated. Study 3 manipulated duration, the location in the task where the interruption takes place, and the target role (both persons interrupted or only one) in a structured task. Duration and location of interruptions affect the reinstatement of the task. When both participants are interrupted, task reinstatement is more effortful than when only one is. Study 4 extended this design by incorporating a third variable, the presence or absence of visual information between the participants.

The research is relevant to theoretical issues in the psychology of language as well as applied issues in computer-supported collaborative work and the management of interruptions.