Data collection methods constitute a major area of concern in interlanguage pragmatics (ILP) research, since there is no easy way to collect the type of data that is relatively 'naturalistic' while at the same time allowing for researcher control. Further, current ILP research also suffers from the shortage of studies making use of self-report data to understand how L2 learners plan and execute their speech act behaviour. This article discusses the advantages of an innovative method devised for collecting naturalistic data on "institutional" criticisms and responses to criticisms in the L2 in a controlled manner. The article also addresses the potential of retrospective interviews in researching interlanguage (IL) pragmatic knowledge and issues involved in pragmatic decision-making, especially when used alongside the performance data. This article is grounded on data from 4 hours of recordings of 72 peer-feedback conversations by 36 EFL learners and 8 hours of recordings of 36 retrospective interviews with these learners.