SwaPS: A Method for Efficiently Relearning

Abstract

Character amnesia is a recent phenomenon in which native Chinese or Japanese speakers forget how to write Chinese characters (kanji in Japanese), although they maintain the ability to read them. To solve this problem, we previously proposed a novel pronunciation-based input method called G-IM (Gestalt Imprinting Method), which is effective in preventing the loss of character shape memory. G-IM sometimes outputs GIM characters whose shapes are slightly incorrect, which forces users to pay close attention to character shapes and thus strengthens retention and recall. However, the task of finding and correcting such slight errors in writing documents is an originally high cognitive load act that is burdensome and discourages users from using G-IM. Therefore, in this work, we focus on the act of reading behaviors and propose a novel method for generating incorrect character shapes named SwaPS, which can effectively correct and strengthen the memory of character shapes by simply reading a document without increasing the user’s workload. SwaPS generates incorrect characters named PS characters by using a deformation method that swaps the position of the semantic radicals and phonetic radicals of the phonogram characters, which account for 80% of all Chinese characters. By reading a document that includes PS characters, the user’s attention is drawn to the character shapes, which is expected to correct and strengthen his or her memory of the character shapes. In the future, we aim to build an e-book reader that automatically generates and presents PS characters. In this paper, we conducted a basic investigation by printing manually created incorrect characters on paper and presenting them to users. The results of the user study confirmed that reading a document that includes PS characters significantly strengthens character shape memory compared to reading a document that contains only correct characters or slightly different incorrect characters used in the G-IM system. We also confirmed that reading a document that includes PS characters does not increase the user’s load compared to reading a normal document that contains only correct characters.