It is proposed that a speech-dynamic analysis of oral fluency phenomena serves best to highlight the nondiscrete nature of the sound stream and to capture English syllable structure. Some current concepts and practices in oral testing are criticized, and previously neglected evidence concerning temporal variables in speech, automatic speech recognition, and recent experimental findings are presented that support a reexamination of the parameters of fluent speech based on quantitative and qualitative phenomena. The quantitative and qualitative speech parameters are outlined and discussed to underscore the range and complexity of features that nonnative speakers must master for fluency. Native and nonnative speech data on consonant attraction are examined, and the phenomenon of restoration in listening is also briefly discussed. It is concluded that with a topic as multidimensional as fluency, at least some of the features sensitive to speech flow and information flow in language have inevitably received insufficient attention. A more holistic understanding of these features can, along with innovative techniques, contribute to improved attainment of fluency. (MSE)