The Impact of Communicative Strategies on ChatGPT Interaction and Text Quality: A Comparative Study

Abstract

ChatGPT is an effective tool for text rewriting, allowing for changes in structure and formulations, elimination of errors, and improvement of readability. The paper presents an attempt to compare the effectiveness of interaction and the ”quality” of texts after rewriting when formulating prompts to ChatGPT within three communication strategies: cooperative, informative, and persuasive. The effectiveness of such interaction was assessed using scales measuring the convenience of interaction with the chatbot and expert evaluation of text quality. The comparison of different communication strategies with ChatGPT revealed significant differences and trends when moving from the cooperative to the informative and further to the persuasive strategy. The upward trend reflected an increase in session duration, number of iterations, and text length, which was accompanied by growth in logical coherence, structural consistency, and detail of exposition, as well as increased accuracy and smoothness of the text. The downward trend indicated a decrease in indicators related to the convenience of using the chatbot and user support. The informative strategy occupied an intermediate position, providing structure, completeness, and textual closure in responses, while reducing flexibility and cohesion. The presence of specificity in interaction when using the informative strategy led to the identification of four latent variables determining the data distribution across the three communication strategies: quality of interaction and user support in dialogue, user satisfaction with text quality, session duration and dialogue volume, and content-related and semantic completeness of the text. The paper shows the positioning of the examined communication strategies with ChatGPT within the space of these latent variables.

Ähnliche Arbeiten