Speech as process: A case study