Bengalese finches are the domesticated strain of the wild white-rumped munias. Bengalese finches had been domesticated for over 250 years from the wild strain white-rumped munias and during this period the courtship song became phonologically and syntactically complex. The purpose of this study is to understand proximate and ultimate causes for song complexity in Bengalese finches. Field observation of white-rumped munias in Taiwan suggests that populations of munias show a gradient of song syntactical complexity: when the population has more sympatric species, the population showed less syntactical complexity, suggesting that syntactical complexity does not develop under the pressure for species recognition. Laboratory study of cross-fostering between the two strains revealed that white-rumped munias are more specialized in accurately learning own-strain phonology while Bengalese finches learned equally but less accurately learned phonology of both strains suggesting that Bengalese finches lost species-specific bias to accurately learn own phonology. By a nest-building assay, we found that females work more when stimulated with complex songs but not with simple songs. Taken these evidences together, we suggest that phonological and syntactical complexity in Bengalese finch songs evolved first because domestication freed them from pressure for species recognition based on song characteristics and then sexual selection advanced the complexity. This is enabled by longer song learning period in Bengalese finches. Neural and molecular studies also support the notion that Bengalese finches keep more song plasticity as adult. In conclusion, song complexity in Bengalese finches provides a unique opportunity for integrative study of animal communication.