Natural language processing (NLP) plays a major role in knowledge engineering. However, NLP's usage is traditionally being seen as a means of extracting knowledge required for building knowledge resources, i.e. ontologies, knowledge bases. When it comes to the evaluation of these knowledge artefacts, then general trends are: expert reviewing, evaluating against existing ontologies and democratic ranking. We propose a new approach for evaluating domain coverage of application ontologies which is based on NLP techniques. The latter can be seen as one way of bridging the gap between terminologies and ontologies in order to create user-understandable expert systems.