The melody presentation can facilitate recall of sentences in a first language. This study was planned to investigate whether the same effect would be observed in second language learning. A 2×4 factorial design was used in the experiment. The first independent variable was type of auditory presentation in which the learners of Japanese have to listen to a Japanese song constructed from 3 verses either with melody or without melody. The second independent variable was periods of the free-recall test of the sentences: after the first-trial, second-trial, fifth-trial in the listening session (three immediate tests), and after 18 minutes from the fifth-trial (one delayed test). Three aspects of the sentence memory were adopted as dependent variables: sentence memory on both the literal level and the semantic level, and memory of the word order in each sentence. The main results were as follows: in three aspects of the sentence memory, (a) recall performance of learning Japanese sentences both with and without melody presentation were increased from the first trial to fifth trial, and were not decreased from the fifth trial to delayed test, and (b) learning of Japanese sentences with melody presentation had not better recall than that without melody presentation. It was suggested that the melody might not play the role as retrieval cues in second language learning. These findings were discussed from viewpoint of automaticity of language processing.