With the widespread use of Generative AI in education, effectively utilizing and integrating it into teaching have become key focal points and challenges in education. Different subjects and target audiences require varied norms and strategies for implementing Generative AI, such as ChatGPT. These differences directly impact the educational integration of Generative AI in various educational contexts. To address these disparities and establish common ground, we propose the concept of ChatGPT literacy to bridge research gaps. In this study, we tailor the concept of ChatGPT literacy specifically for language teachers, aiming to delineate the essential competencies needed to proficiently and ethically use ChatGPT as a language learning and teaching tool. We propose a theoretical framework encompassing six fundamental constructs: benefits, limitations, prompts, evaluation (of ChatGPT responses), assessment (assisted by ChatGPT), and ethics, to comprehensively conceptualise and evaluate ChatGPT literacy. Drawing on both quantitative and qualitative survey data from 492 language teachers across 41 countries, we validate the proposed ChatGPT literacy framework by examining teachers' practices and challenges associated with ChatGPT usage. Our analysis of Likert-scale data, utilizing item and confirmatory techniques, confirms the effectiveness of the six-construct framework in defining ChatGPT literacy. In addition, we collected qualitative data through open questions and conducted thematic analysis, demonstrating that ChatGPT has been integrated throughout the instructional cycle, from material preparation to formative and summative assessment phases. These quantitative and qualitative findings have significant implications for a range of stakeholders, including language educators, learners, AI technology developers, and policymakers, providing valuable insights to inform decisions regarding ChatGPT integration in language education. Ultimately, our study equips relevant stakeholders with the necessary competencies to responsibly exploiting ChatGPT's potential in language and other subject areas.