This paper examines differences between Bakhtin's dialogic view and argumentation theories with respect to questioning and analyzes the significance of these differences for the theories of pedagogy. In argumentation theories, a question is thought to be shared among the parties in a discussion. In the fields of argumentation and education, in particular science education, not only is a question shared, but also an answer is integrated into one among the participants (Schwarz and Baker 2017). Bakhtin's view on questioning, advanced in his later writings, shows how new questions emerge continuously in answers to the previous questions so that a question is not shared by a questioner and an answerer. Using the Bakhtinian framework in the analysis of some Japanese pedagogical thoughts and classroom interactions, it is shown that each student can develop her/his own unique understanding of the topic -- not the shared, integrated understanding -- by finding out a new question in seemingly wrong answers, or by discovering different questions in the same problem. Finally, the reason why new questions emerge in question-and-answer exchange is investigated within a constructivist perspective from cognitive science.