The use of block-based visual language in introductory programming is a popular method in education. However, there is little research which provides evidence showing advantages of block-based language. This paper presents the results of learning data analysis with fine grain logs recorded by students’ development environment where the students can select their language in block-based or Java. A total of 400+ students’ logs collected each of four years were analyzed. The results show that migration from Block to Java can be consistently seen each year, although the whole block-editing rate was influenced by the method of the instructor’s introduction. Though block-editing did not affect working time and Lines of Code (LOC), it could reduce the compile error correction time, whereas using Java requires approximately 20% of compile error correction time for students. We concluded that block-based language worked to encourage students to focus high-level algorithm creation, as well as it provides an advantage to understanding text-based language.