A study on research trends in speech studies of Japanese learners with text mining

In the AI era, it is important to extract some keywords of a particular field in order to grasp trends of research and classify them in order to lead to a new research theme.<BR> In this study, we use text mining to analyze speech studies of Japanese learners.<BR> We search 669 previous researches published from the 1980s to the 2010s on Google Scholar with the keywords, ‘Japanese’, ‘Learners’, ‘Speech’, and ‘Pronunciation.’<BR> As a result of the search, we observed how its trend has changed.<BR> For text mining we use KH Coder, a free software.<BR> 1) As a result of conducting analysis of frequent words, we found a lot of more research regarding formation than perceptual parts of pronunciation.<BR> 2) As a result of analyzing the co-occurrence network of frequently occurring words, it was possible to visualize the frequency of research themes.<BR> 3) We also analyzed how research trends have been changing since the 1980s to the 2010s. The number of researches published has been increased, and themes have also become further diversified since the 1990s. During the 2000s, research in relation to speech recognition became outstanding with some useful techniques of it. Since the 2010s there has been a lot of research in regard to speech education and accent education.