Saito & Tsuzuki (1989) investigated transcription process from KANA to KANJI compound-word. In their experiment, many pseudo compound words were produced by undergraduate students. To investigate an evaluation process for these pseudo compound words, compound-wordlikeness evaluation task for these pseudo compound words was conducted. Subjects' task was to judge whether a stimulus is legitimate compound word which appears in a dictionary. In a series of 4 experiments, subjects regarded some pseudo compounds as real existing compounds, and subjects' compound-wordlikeness judgement was affected by their knowledge on compound words. The results suggested that we acquire common meta knowledge for compound-acceptability (whether real or pseudo) through individual learning of single Kanji characters and real compounds, and through this meta knowledge, individual knowledge on real compound words affect the judgement of compound-wordlikeness of pseudo compound words.