In this paper we are specifically interested in the relationship between natural language and the knowledge representation (KR) formalism referred to as ontology. By ontology we mean a taxonomic, hierarchical data structure. The reason we use natural language for terms in ontologies is so that we humans can understand the ontologies. Machines and humans who have to understand ontologies interpret like terms in a given natural language similarly, though their interpretations may not completely coincide. Human understanding of natural language terms involves premise smuggling, by which we mean the unconscious or implicit use of background knowledge and context in the process of interpretation. Machine understanding of ontological terms is explicit, and is based on their positions in an ontology. While two machine ontologies may contain similar terms, they may not be completely homologous. Ontological terms are interpreted beyond the control of the designers of ontologies, by both humans and machines. When natural language is used for ontological terms, the interpretation of these terms by humans is complex and idiosyncratic. In this paper we discuss some of the problems and issues involved in human and machine understanding of natural language ontological terms.