Dou Dizhu For All: An Accessible Card Game for Visually Impaired Persons in China Powered by ChatGPT's Language Model

This paper presents Dou Dizhu For All, an adapted version of the traditional Chinese three-player card game Dou Dizhu designed for both blind and visually impaired players and sighted players. The authors argue that video games remain largely inaccessible to visually impaired persons because of their visual-centric design, and that game accessibility research has not yet been conducted in China. The project positions itself as an initial step toward addressing that gap, with the goal of producing interface design and accessibility features that can inform future accessible games in the Chinese context.

The design centers on three main accessibility features. A custom text-to-speech system provides game-specific audio feedback, such as identifying the previous player by name and role when an avatar is tapped. Audio cues convey card holdings and board status so players can follow game context. A voice control system uses ChatGPT's language model to interpret natural-language commands, such as requests to bid for the landlord role or to hear the rules, and executes them after player confirmation. Interface decisions include eliminating registration to shorten the path to starting a game and keeping button layouts consistent across screens to support touch-based navigation.

The evaluation plan is qualitative. The authors propose recruiting 8 to 10 blind and visually impaired adults over 16 through online gaming communities, with a range of experience levels, and conducting semi-structured interviews after gameplay. The interview guide draws on three components of the Immersion Experience Questionnaire, namely cognitive involvement, control, and challenge, and responses will be analyzed thematically. The paper does not report results, and the authors explicitly acknowledge that more rigorous research will be needed for a fuller evaluation of the game.