---

This study was intended to compare processing instruction (PI), an input-based approach to L2 grammar instruction developed by VanPatten (1996), with an output-oriented type of instruction (OI) to assess their relative effects on learners' ability to interpret and produce English causatives. A pretest and posttest (immediate and delayed effects) design was used. 151 university students from four intact classes were randomly assigned to three treatment groups of PI, OI, and EI (Explicit-information-only) and one uninstructed control group (C). Students were assessed on interpretation and controlled written production tasks at the sentence level. Within-group comparisons indicated that the three instructional options, as compared to the control group, resulted in some kind of knowledge gain in both interpretation and production tasks, but the gains were not equal. The results of between-group comparisons contradicted VanPatten's claims about the superiority of PI over OI. While PI and OI were equally better than EI on interpretation tasks, OI group outperformed both PI and EI on production tasks. No significant difference was found between PI and EI on production tasks. The same results were obtained after a one-month interval, reflecting the durability of the instructional effects on the interpretation and production of the target structure.