In order to mediate the debate on whether common nouns are better interpreted as types or as predicates in type-theoretical semantic frameworks, the present paper shows that type theories whose models in category theory are toposes have access to a property drawing a one-to-one correspondence between first-order predicates and base types, thus enabling more flexibility for common noun interpretation. Using this flexibility and linguistic arguments based on negative predications, a subsequent proposal is made to interpret nouns as predicates with refined argument types.