There are known mechanisms to succinctly describe regular languages, such as nondeterministic finite automata, boolean automata, and statecharts. The MERLin project is an investigation into and comparison of different description mechanisms for the regular languages. In particular, we are concerned with descriptions which, for a specific application domain, often achieve succinctness. To this end we implemented a Modelling Environment for Regular Languages (MERLin). This paper describes the application of the MERLin system to analyze the behaviour of selective nondeterministic finite automata.