Bilinguismo jurídico chinês-português em Macau: análise e alinhamento de corpus com IA || Chinese-Portuguese legal bilingualism in Macao: AI-powered corpus analysis and alignment

This article explores legal bilingualism in Macao, a Special Administrative Region of China that adopts the “one country, two systems” principle. The analysis focuses on the coexistence of Chinese and Portuguese in local legislation, highlighting linguistic hierarchy and legal translation challenges. Based on a robust methodology, the study compiles and annotates parallel corpora of legal texts in Portuguese and Chinese, employing advanced artificial intelligence tools such as Bertalign for automated alignment and spaCy for linguistic annotations. Due to the distinct writing systems between the languages, alignment required AI-based solutions to overcome limitations of conventional aligners developed for Romance languages. The analysis highlights the significant influence of Portuguese on Chinese legal terminology, with a high prevalence of linguistic calques. The study also addresses the challenges of Chinese language tokenization and machine translation, proposing practical solutions and comparing the performance of different translation tools, such as LLMs (language models) and NMTs (neural machine translation systems). The detailed exploration of the corpora, including n-gram analyses and syntactic patterns, offers valuable insights for corpus linguistics and legal translation. The computational tools of Sketch Engine enable, through practical examples extracted from this aligned corpus, the study of specific aspects of Macanese legal translation. The article contributes significantly to the field of corpus linguistics, especially in multilingual legal contexts, and offers valuable methodological and analytical resources for researchers and legal translation professionals. Keywords: legal bilingualism; Macao; legal translation; corpus linguistics; linguistic calques.