This thesis aims to analyze Korean pronunciation errors of Japanese learner based on the spontaneous speech data obtained from each learner over a certain period. In most earlier studies, learners’ pronunciation errors tended to be analyzed based on manipulated data. In this thesis, however, the learners’ pronunciation errors are treated as part of the natural utterance. Japanese learners’ pronunciation errors in actual speech generally corroborate the conclusions of earlier research regarding certain phonological phenomena, such as the high vowel alternation, nasal consonant alternation, and alternation between aspirated and lax consonants. Some phenomena, however, require another perspective that analyzes the learners’ pronunciation errors. Open syllabification, which is the most typical type of error, does not occur frequently in the actual speech of Japanese learners, and final consonant deletion, which is not considered a representative type of error, is more common than open syllabification.