Inclusive language has become an increasingly important topic in linguistics, translation studies and Natural Language Processing (NLP). Although numerous studies have examined inclusive writing from sociolinguistic and political perspectives, relatively few have investigated its automatic processing and its implications for multilingual applications. Furthermore, although recommendations on inclusive writing exist in several languages, no universally accepted standard has been established, while in France its use is restricted in certain official and administrative contexts. This paper presents an ongoing study that investigates the automatic recognition, generation and translation of inclusive language forms in French and Greek.

 The study is based on a manually compiled dataset of representative inclusive forms collected from French and Greek linguistic resources and authentic texts. The selected forms represent the most common strategies of inclusive writing and were chosen because they illustrate different morphological structures and translation challenges. The dataset includes French forms such as enseignant·e, étudiant·e, chercheur·euse and les étudiantes et les étudiants, together with Greek forms including καθηγητής/καθηγήτρια, καθηγητής/-τρια, καθηγητής(τρια), φοιτητής/φοιτήτρια and coordinated expressions such as οι φοιτητές και οι φοιτήτριες.

 The same dataset is used in both Unitex and ChatGPT. In Unitex, morphological dictionaries, lexical entries and formal grammars are developed to recognize and model inclusive forms automatically. ChatGPT is prompted to identify, generate, explain and translate the same linguistic forms between French and Greek while preserving their inclusive meaning. The generated outputs are analyzed according to linguistic accuracy, morphological correctness, consistency and translation adequacy. The rule-based descriptions implemented in Unitex serve as a reference framework for interpreting and evaluating ChatGPT's responses.

 Rather than comparing Unitex and ChatGPT as competing approaches, this study investigates how rule-based linguistic resources and large language models can complement each other in the computational processing and translation of inclusive language. The findings are expected to contribute to multilingual NLP, machine translation and the development of computational resources for gender-inclusive language.