Intricacia gramatical em textos gerados pelo ChatGPT em língua adicional: uma perspectiva sistêmico-funcional [Grammatical intricacy in texts generated by ChatGPT in an additional language: A systemic-functional perspective]

Grammatical intricacy refers to the complexity of the syntactic organization of complex sentences in a text. This study analyzes this intricacy in a corpus of six Portuguese as an additional language tasks produced by Chatgpt. Sentence complexes were examined, considering the interdependence between sentences—parataxis and hypotaxis—and the logical-semantic system of expansion and projection. Data analysis included statistical treatment using Python and spss, as well as a qualitative evaluation of a sample of texts. The results indicated that the correlation between grammatical intricacy and lexical density generally follows what Halliday (1985) proposed in terms of proportionality. However, the correlation between grammatical intricacy and proficiency levels was not confirmed in the texts generated by artificial intelligence, indicating that Chatgpt does not consistently follow instructions regarding proficiency level. As for sentence complexity, hypotactic relations prevailed, and textual development occurred mainly through extension (+) and intensification (x) expansions. The contributions of this study highlight the importance of teachers who frequently use artificial intelligence to create teaching materials paying attention to the complexity and logical-semantic organization of texts, which also helps in identifying productions that do not follow the principles of academic integrity.