The paper discusses grammatical construction of speech as a self-contained mechanism independent of meaning and semantics. The thesis about its autonomy is argued by the identification of a hidden non-semantic component connected to the word meanings in speech. This component is composed of bits of non-semantic information which form quantitative and elective (carrying directed substitution) relations. The structures thus formed determine the classes of words in the sentence and at the same time the way they are syntactically connected. The paper briefly touches on such issues as the relationship between sign and meaning in verbal communication, the difference between meaning and linguistic information.