When constructing a translator, as a rule, it is necessary to apply the equivalent transformations of grammar of the implemented language, which convert the syntactic specification of the language into a form that allows for automatic or manual implementation of the source language, and solves the problem of constraint satisfaction of the chosen parsing technique. These problems arise both because of the variety of ways of defining the implemented languages, and because of the language ambiguity or nondeterminism of the recognizing automaton. Each transformation performed on translational CF-grammars can be defined as a likelihood ratio (homomorphism), determined on classes of grammars. Such relations preserve the semantic meaning of grammatical structures in equivalent syntax transformations, since the ultimate purpose of translation is to obtain a sequence of actions prescribed by the computing environment. The article presents a brief overview of various types of likelihood ratio over grammars used in cons