Rating scales are a core tool connecting test-takers’ language performance and score interpretation in language assessment, and their psychometric properties directly affect test reliability and validity. However, existing reviews mostly adopt traditional narrative methods and lack systematic bibliometric analysis of international research on language skill rating scales. Using the Web of Science Core Collection, this study screened 99 core publications (2008–2026) and applied CiteSpace to analyze publication trends, institutional and national collaboration networks, keyword co-occurrence and clustering, and burst detection. Findings: publication trends show fluctuating growth with a 2023 peak followed by a sharp decline, reflecting a paradigm shift from classical scale validation to technological integration and cross-cultural adaptation; the institutional collaboration network is highly fragmented, with North America and Western Europe leading; the field forms six core clusters—language testing, generalizability theory, rating criteria, many-facet Rasch measurement, automated scoring, and localization; hotspot evolution has gone through three stages—scale development and reliability, psychometric validation and rating processes, and construct validity and comprehensibility. This study provides a knowledge map for rating scale development, validation, and cross-context adaptation.