The paper contributes to the methodological discussion of what evidence the philosophers of language (e.g. Devitt, Machery, Stich, etc.) use in support or disconfirmation of semantic theories. Three classical theories of empirical confirmation are discussed. We suggest that in the methodological literature on semantics one of them – namely H-D conception – prevails. Further, we provide a basic framework for testing modal semantic statements. Finally, we discuss the role of intuitions, language corpuses and other linguistic data in the methodological considerations.