The purpose of the current paper is to present an on- tological analysis to the identification of a particular type of prepositional natural language phrases called figures of speech [1] via the identification of inconsis- tencies in ontological concepts. Prepositional noun phrases are used widely in a multiplicity of domains to describe real world events and activities. However, one aspect that makes a prepositional noun phrase poetical is that the latter suggests a semantic relationship between concepts that does not exist in the real world. The current paper discusses how a set of rules based on Wordnet classes and an ontology repre- senting human behavior and properties, can be used to identify figures of speech. It also addresses the problem of inconsistency resulting from the assertion of figures of speech at various levels identifying the problems involved in their representation. Finally, it discusses how a contextualized approach might help to resolve this problem.