This study aimed to examine the effect of two types of collaborative output tasks on Iranian EFL learners’ comparative adjectives with two or more syllables. Thirty Iranian EFL learners participated in this study which were then divided into two experimental and one control groups; one experimental group received dictogloss task in 4-pairs and the other experimental group was given text reconstruction editing task in 6-pairs. Using pretest, posttests and delayed posttests, the data were collected through a grammaticality judgment test (GJT) and audio-recording of the learners’ interaction. Comparing the mean scores of three groups in GJT generally indicated that experimental groups gained more than the control group and text reconstruction editing group outperformed dictogloss group in noticing and learning the English comparative adjectives with two or more syllables. The transcripts of the students' verbal interactions indicated that text reconstruction editing group generated larger number of turns and language related episodes (LREs) in comparison to dictogloss group. Findings of this study also indicated that the majority of problems encountered in all the dyads were correctly solved in both dictogloss and text reconstruction editing groups, while the dictogloss group correctly solved more LREs in comparison to the text reconstruction editing group which showed that the dictogloss group was better regarding the quality of LREs. It can be concluded that the current study found evidence in support of Swain and her colleagues' claims (e.g. Kowal & Swain, 1994) that task implementation required the learners to produce output collaboratively which in turn leads to the internalization of grammatical features.