To provide a language information apparatus which enables even one person to effectively practice daily conversations.

Roles are displayed at the heads of respective model sentences by a role instruction display 31 so that the roles are specified by each of the respective model sentences. The display of the model sentences of the roles uttered by the apparatus is masked and voices are outputted. The display (distance display 32) to announce to what degree the pronunciation uttered by a user is approximate to the pronunciation of the stored model sentences is made to enable the decision as to whether his own pronunciation is comprehensible by the other party or not. Further, the conversion progresses according to the distances.