This study investigates gender differences in the linguistic construction of character dialogue within role-playing video games, focusing on psychosocial factors. A corpus of game dialogue was analyzed using both a lexicon-based dictionary and a transformer-based emotion classification model. Basic linguistic analysis revealed that player characters spoke significantly more lines, sentences, and words than non-player characters but used simpler sentence structures. However, no significant gender differences were observed in these linguistic metrics. Word frequency analysis showed that female characters used significantly more language associated with cognitive reflection and social connection, whereas male characters exhibited higher rates of swear word usage. Emotion classification revealed that female non-player characters expressed a broader range of emotions, including more disgust, sadness, surprise, joy, and fear, while male characters showed a stronger association with anger. Principal component analysis identified an affective positivity–negativity axis that significantly predicted gender, with male characters clustering toward greater negativity. Analysis of neutral characters indicated that their psychosocial profiles most closely resembled male characters, supporting the notion of masculine defaults. These findings highlight how gendered psychosocial patterns are embedded in video game dialogue and reveal both persistent stereotypes and emerging shifts toward more balanced gender representation in interactive narratives.