Theories of Truth and Learnable Languages

Abstract

Compositionality (frege principle) says that the meaning of a complex expression depends functionally on the meanings of its parts. it is shown to amount to semantical context-independence, which fails in e. g., english, as exemplified by branching quantifiers, backwards-looking operators, "any", etc. davidson notwithstanding, compositionality is also unnecessary for learnablility. moreover, t-schema (tarski, davidson) fails (witness "'anyone can become a millionaire' is true if anybody can become a millionaire").