The classic theory of knowledge is a theory of cognition concerning itself with propositions and concepts conceived either psychologically or logically. A semantically interpreted theory of cognition is a theory dealing with sentences and other verbal expressions. The author compares these two approaches and asks the question: When is it legitimate to transform a classic theory of knowledge into a semantic theory of knowledge? Or, to put the same problem in a different language: When is it legitimate to abandon logically or psychologically interpreted utterances and replace them with sentences and other linguistically interpreted expressions? Such replacement is tempting because propositions and notions lead to numerous, mostly insoluble, controversies, whereas syntactic and semantic formulations are by and large uncontroversial. It must be conceded, however, that such a replacement is not advisable before it has been shown on what conditions it can be executed without losing legitimacy.