The current books’ scope is within the Natural Language Understanding sub-field of Natural Language Processing. From the multiple possible tasks from this domain, we stopped at Discourse Analysis. We analyzed the main approaches existent in this field and identified the flaws of each of the presented approaches. Starting from them, we proposed an adaptation of an existing framework (the Polyphonic framework) using ideas derived from the theory of a known linguist (Tannen) regarding the importance of repetitions in discourse. After presenting our adaptation, we showed how it would solve most of the indicated problems with the other approaches.