The extent to which tasks could perpetuate authentic language production has been one of the major concerns in task based language teaching (TBLT) method. Although TBLT offers positive amendments to audio-lingual-structural approach, there have been notable issues in view of how meaning and form can be sufficiently dealt with in TBLT. Hence, this paper seeks to contribute to the long-standing debate over form-meaning issue in TBLT. It draws on findings from a case study involving third-year University students enrolled in a speaking subject. The main procedure entails the use of two speaking tasks which are similar but different in the way prompts were incorporated in order to regulate control and freedom. The data were collected through audio-taped as well as video recording. The findings show that the first task, which is less prompted, tends to result in a much less participation and language output than the second task which comes with more prompts. The second task allows students to demonstrate a higher level of participation and language output regardless of its accuracy. The findings attest to the need to design speaking tasks along the continuum of fully-prompted to zero/minimum-prompted tasks to scaffold learners’ confidence and creativity in using the language.