A study was planned to generate a model for testing second language education that would incorporate task types to allow the display of fluency. It is argued that the display of oral fluency can be used as a selection criterion for inclusion of test items in multitask or portfolio assessment. The following three research questions were formulated: (1) how different speaking tasks varying in structure in terms of cuing and expected response affect the display of oral fluency; (2) the difference in oral discourse produced in the performance of the four research tasks; and (3) the task-specific nature of production of formulaic speech. Young adult students (N=32) from the highest English fluency levels of a school for English in Brazil would produce cued and uncued dialogues and picture-cued or uncued narratives (monologues). Testing sessions were to be evaluated by two examiners. It was expected that: task structuring involving manipulation of cued/uncued and dialogue/monologue conditions would affect the amount of formulaic speech produced, and the fluency ratings obtained on the tasks with cued tasks would generate more formulaic and more fluent speech. Data to be analyzed include: 128 audiotaped speech samples from the subjects; subjects' proficiency and fluency scores on tasks; subjects' written responses to a survey questionnaire; and subjects' responses to semi-structured interviews. The test conditions and documents to be used in the study are appended. A 74-item list of references is included. (SLD)