Trust in automation, and particularly maintaining an ade- quate level of trust in automation is now recognized as a major perfor- mance factor in supervisory control. Leveraging man-machine interaction is seen as a promising approach to influence the level of trust of an operator. Two problems need to be addressed in order to reach this goal: (1) measuring the level of trust; (2) acting on the level of trust to reach a more appropriate level. In this paper, we tackle problem 1, and propose to use a computational dialogue modelling approach to evaluate trust dynamically. We describe our model on two examples and give some perspectives.