A pronunciation diagnosis device includes: articulation attribute data having articulation attribute values corresponding to preferable pronunciation concerning one of the tongue state, the lip state, the glottis state, the palatine uvula state, the nasopalatine groove state, the teeth state, and the jaw state or a combination including at least one of the articulation organ states, enforcing of the articulation organ state and a combination of the exhalation states for each of the phonemes constituting each voice language system; means for extracting acoustic characteristics from a voice signal pronounced by a speaker such as a frequency characteristic amount, a voice volume, a duration time, their change amount, or a combination of the change pattern, and a combination of them; attribute value estimation means for estimating an attribute value concerning the articulation attribute according to the extracted acoustic characteristic; and judgment means for comparing the estimated attribute value to the preferable articulation attribute data so as to make judgment concerning the pronunciation of the speaker.