The present invention discloses a Tibetan character constituent analysis method, a Tibetan sorting method and corresponding devices, and relates to the field of natural language processing. The present invention is proposed to solve the problem that the existing Tibetan sorting methods have no universality or compatibility, which is inconvenient for the use of automatic computer Tibetan sorting. The technical solution provided by the present invention includes: S10, acquiring a Tibetan text to be analyzed; S20, using Tibetan characters in the Tibetan text as the input of a preset finite state automaton group; and S30, acquiring the constituents of the Tibetan characters according to a target finite state automaton, when the target finite state automaton in the finite state automaton group determines that the Tibetan characters in the Tibetan text are correctly spelled.
