<![CDATA[The widespread use of ChatGPT as a conversational artificial intelligence has created new challenges in pragmatic communication between humans and machines, particularly when user prompts contain ambiguous deictic references. Unlike face-to-face interaction, ChatGPT does not possess physical embodiment, direct awareness of real-time situations, or interactional history beyond the available conversational context. These limitations make deictic expressions particularly vulnerable to ambiguity. This study aims to describe the forms of deictic ambiguity found in ChatGPT prompts and examine the conversational implicatures arising from such ambiguity based on Grice’s Cooperative Principle. This study employed a qualitative descriptive approach. The data consisted of fifteen prompt-response pairs produced by five active university students who regularly use ChatGPT, covering three categories of deixis: person, temporal, and spatial. Data were collected through non-participant observation and note-taking techniques and analyzed using the interactive model of Miles, Huberman, and Saldaña, involving data reduction, data display, and conclusion drawing. The findings indicate that deictic ambiguity occurred across all three categories, particularly through unspecified personal references, relative expressions of time, and locative demonstratives without clear referents. Such ambiguity led ChatGPT to provide clarification-seeking or overly general responses rather than direct answers, indicating the importance of contextual information in interpreting user intentions. Analysis based on Grice’s four maxims revealed that the maxim of quantity was violated most frequently (40%), followed by quality and relation (33.3% each), while manner was violated least (13.3%). These findings highlight the need for explicit, context-rich prompts to support cooperative communication with ChatGPT, particularly in academic settings.]]>
