This study quantitatively measures the variation in language derived from a targeted set of digital game mechanics. Mechanics refer to the design elements of a game that make up the overall gameplay experience, determining player actions and the degree of language interaction. A corpus was compiled by extracting the language files from two popular commercial games, Fallout 4 and Skyrim, using modification software. The extracted language files were organized into three register categories following the register analysis framework detailed in Biber and Conrad (2019). The three categories include one spoken (dialogue trees) and two written registers (quest objectives and quest stages), which are common mechanics in many modern commercial games. Comparing results from three discriminant analyses, the findings indicate that statistical models cannot distinguish between the two games' linguistic environments at the level of the game; however, when considering the linguistic environments at the level of game mechanics, the model has high precision in accurately identifying the texts' game mechanic register categories. The results provide empirical evidence that digital game-based language learning (DGBLL) research designs could benefit from targeting specific design aspects and game mechanics rather than generalizing results at the level of genre or game title.