Abstract: The present invention is a new—linguographic—method of compiling the word dictionaries and lexicons, assigned for memories and other equipments of speech-recognition devices (SRD). The linguographic method is based on the simulation of the natural speech signals' accumulation and storage by the human language center (HLC). The linguographic method differs from the traditional lexicographic method of word dictionaries' composition by next features: A. It uses, as the primary element of word classification for the word dictionaries and lexicons, the stressed vowel of the speech signal; B. It uses a number of newly elaborated characteristics for the classification of speech signals in said dictionaries; C. It uses the Sound Alphabeth, invented by the inventor, for defining the sequence of signals' characteristics in the word dictionaries and lexicons.