This study uses multimodal conversation analysis to investigate longitudinally tracked learning objects with explicit retrospective orientations in collaborative digital EFL writing tasks. Pupils' retrospections (e.g. "what did we say?") suggest a problem orientation to a previously topicalised language object, but also provide a second locus for language learning.
Collaborative writing provides rich pickings for exploring learning processes and learning achievements, since pupils routinely display their current epistemic status (Heritage 2012), particularly when they encounter some kind of linguistic 'trouble'. The emergent trouble thus becomes a trackable 'learning object' (Markee 2008) or 'learnable' (Majlesi 2014). What then happens when the need arises to reuse the same linguistic item on a later occasion? One option is to refer retrospectively to that occasion (e.g. "what did we say? återkommande [recurring] …?"). When would such a retrospective orientation be salient (cf. Jakonen 2018, Can Daskin & Hatipoglu 2019)? To address this issue, the study juxtaposes longitudinally tracked learning objects with and without retrospective orientations within the 'triadic ecology' (pupil-pupil-computer; Musk 2021) of collaborative EFL writing tasks written in Word and GoogleDocs. More specifically, how does the design of learning trajectories compare in the case of retrospective and non-retrospective orientations? Also, how do the respective orientations compare in terms of their interactional work? The collection of learning objects comes from 37 hours of video-recorded data from collaborative computer-assisted writing tasks in EFL classrooms of 5 Swedish upper secondary schools. The collection has been examined using multimodal conversation analysis and comprises topicalised vocabulary, grammar and spelling objects. The retrospective orientations corroborate Jakonen's finding that they occur "in situations where problems related to learning achievements surface" (2018: 770). Furthermore, these retrospections allow pupils to hold each other "morally accountable for contributing to the quality of the product-in-the-making" (Musk & Cekaite 2017: 168), insofar as they expect each other to know and remember aspects of language through a "shared referential history" (ibid.: 153). Thus, digital collaborative writing not only opens up learning spaces for resolving initial linguistic trouble, but also provides a second locus for learning when a previous learning object resurfaces as problematic.
