This paper examines the effectiveness of a mobile assisted blended learning scenario for pronunciation in Korean language. In particular, we analyze how asynchronous oral communication between learners of Korean and native speakers via "kakaotalk" (an open source mobile phone application) may be beneficial to the learner in terms of pronunciation. Our methodological approach is based on task resolution (Ellis, 2003) in peer-to-peer collaborative settings and the spaced repetition concept (Ebbinghaus, 1885, cited by Dempster, 1988). The outcomes of our study show that the learners appreciated the possibility to interact with native speakers but most of them preferred synchronous communication for training their pronunciation skills.