Although several studies have developed models and type hierarchies for named entity annotation, no such resource is available for semantic relation annotation, despite its utility for various applications (e.g. question answering, information extraction). In this paper, we show that there are two issues in semantic relation description, one concerning knowledge engineering (what to annotate?) and the other concerning language engineering (how to deal with modality and modifiers?). We propose a new annotation scheme, making it possible to have both a precise and tractable annotation. A practical experiment shows that annotators using our scheme were able to quickly annotate a large number of sentences with very high inter-annotator agreement.