The assessment of second language speaking ability has played a central role in the evolution of language testing theory and practice. Educational Testing Service (ETS) has been a primary innovator in all dimensions of speaking assessment since the 1970s, addressing critical challenges through the advent of new test designs, scoring practices, and measurement technologies to make especially large-scale, standardized testing of speaking ability a reality. This volume presents a sample of key ETS research and development efforts related to speaking assessment over the years. The contributors highlight diverse contributions to conceptualizing the speaking construct, designing speaking test tasks, scoring speech samples, marshalling technologies for test delivery and automated scoring, and developing end-to-end speaking assessment procedures. The first part emphasizes how some of the earliest large-scale speaking assessments were designed and put into practice. The second part features research and development related to speaking assessments in the TOEIC and TOEFL testing suites, emphasizing their validity and scoring. The third part introduces research-based innovations in testing new and more nuanced speaking constructs, and in using automated scoring of speech to address diverse assessment needs. The volume will appeal to language testing professionals and test score users in illuminating how ETS has influenced the development of speaking assessment as well as pointing to multiple directions for future research and practice.