Then, the system takes out the headword take, the top candidate translation word J2, and the entered translation word J11 (step 722), and registers a set of these three pieces of information in a form of the learning data as shown in FIG. 18 described above (step 723) with the translation word addition mark in ON state.