This study uses SFL to explore interpersonal meanings in the texts, comparing evaluative resources, verb processes, and personal pronouns across the datasets. Our preliminary analysis revealed significant differences in lexicogrammatical features between hotel managers and AI-generated responses. Key findings include that human managers are more likely to acknowledge complaints as “sincere” and “constructive advice,” whereas GAI often describes them as “valuable but unacceptable complaints.” In public forums, human managers tend to generalize complex complaints, encouraging guests to reach out privately for further discussion. In contrast, GAI quickly and accurately identifies core issues and provides concise paraphrases.
Human managers use more sensory verbs such as “note,” “look,” “hear,” and “read,” while GAI prioritizes apology and problem-solving strategies over gathering details for investigatory purposes, employing stronger expressions of regret like “apologize,” “lack,” “ensure,” and “assure.” Additionally, human managers incorporate more temporal adjuncts (e.g., “further,” “back,” “forward”) to establish action timelines in their responses, often utilizing intensifiers (e.g., “again,” “very,” “near,” “utmost”) to emphasize their commitment to addressing complaints. While Generative AI responses rarely set a timeframe, they sharpen their focus by using phrases like “regarding.” Furthermore, human managers use more personal pronouns like “I,” “my,” and “yours” to forge personal connections, while GAI responses may lack individuality and instead employ plural pronouns like “we” and “our” to create a collective voice.
