Introduction

Writing has generally been perceived as a solitary activity, completed by the writer working alone. Yet, over the years we have witnessed a growing interest among researchers and educators in Collaborative Writing, an activity that can be simply defined as the involvement of two or more writers in the production of a single text. This interest has been driven by two main factors. The first factor is the nature of workplace writing. Studies (e.g. Ede & Lunsford Reference Ede and Lunsford1990; Mirel & Spilka Reference Mirel and Spilka2002) have shown that in a number of workplaces, writing is often completed in teams rather than individually. The second factor is the advent of Web 2.0 applications such as blogs, wikis, and Google Docs, which have transformed literacy practices, making the creation and sharing of texts easier and more readily acceptable (Hyland Reference Hyland2016; Vandergriff Reference Vandergriff2016). In the field of second language (L2) learning, interest in collaborative writing was also spurred by early research conducted by Swain and her colleagues (e.g. Swain & Lapkin Reference Swain and Lapkin1995; Swain Reference Swain, Doughty and Williams1998; see also timeline for additional references) showing the language learning opportunities of communicative tasks which involve joint written output (e.g. Dictogloss).

Before outlining the research conducted on collaborative writing, let me begin by explaining what I mean by collaborative writing. Although some composition scholars (e.g. Bruffee Reference Bruffee1984) have suggested that all writing is collaborative to some extent (e.g. when a writer composes a text with a specific reader in mind), I view collaborative writing as an activity with very distinct traits (see Storch Reference Storch2013). I define collaborative writing as an activity that requires the co-authors to be involved in all stages of the writing process, sharing the responsibility for and the ownership of the entire text produced. These traits distinguish collaborative writing from cooperative writing. In a cooperative or team writing task, there can be a division of labour, with each team member completing one discrete section of the text or having the responsibility for completing one sub-task (e.g. collecting information, editing the final version). Although the distinction between collaboration and cooperation may be blurred at times, and admittedly the two terms have been used by researchers interchangeably, I have argued (see Storch Reference Storch2013, Reference Storch, Bitchener, Storch and Wette2017) that not all pair/small group writing tasks are necessarily collaborative and that it is collaborative writing that maximises the opportunities for language learning. Thus, in this timeline, I focus only on research on collaborative writing and not cooperative writing (e.g. reports on group assignments).

In collaborative L2 writing, the need to reach agreement on what and how to express ideas in the jointly produced text encourages learners to deliberate about language choice and grammatical accuracy. Swain used the term ‘collaborative dialogue’ (Reference Swain and Lantolf2000) and subsequently languaging (Reference Swain and Byrnes2006) to describe such deliberations, and the unit of analysis used to quantify and analyse instances of languaging is the language related episode (LRE) (Swain & Lapkin Reference Swain and Lapkin1995; Swain Reference Swain, Doughty and Williams1998). These terms are used frequently in empirical studies that investigate the nature of learners’ interaction during collaborative writing tasks and their potential outcomes for language learning. Researchers have also investigated a range of factors that may impact on the quantity and quality of these languaging episodes, including task type (e.g. jigsaw, dictogloss), the relative proficiency of the learners, task modality and learners’ perceptions of the tasks. The assumption underlying these investigations is that the more learners engage in languaging, the more conducive the activity is to language learning.

Given the language learning opportunities created by collaborative writing tasks, researchers have also sought evidence for whether engagement in such tasks results in language learning gains. Various approaches have been deployed in these investigations of outcomes (for a detailed discussion, see Storch Reference Storch2013), such as comparing individually and collaboratively produced texts or comparing learners’ writing before and after the implementation of collaborative writing (i.e. a pre- post-test research design).

Studies on collaborative writing have been informed primarily by sociocultural theory and activity theory (AT), both theories having their genesis in the work of Vygotsky (Reference Vygotsky1978, Reference Vygotsky and Wertsch1981). Key constructs from these two theories are often used by the researchers to analyse their data and explain their findings. These key constructs include the Zone of Proximal Development (ZPD) and scaffolding, language as a mediational tool, and learners’ actions as deliberate and goal driven. The ZPD distinguishes between a learner's current and potential levels of development and forms the basis for the argument that only instruction or assistance which is dynamically attuned to a learner's potential level of development leads to learning. This form of assistance is termed scaffolding. In collaborative writing, learners can pool their linguistic knowledge to co-construct the scaffold; a process that has been termed collective scaffolding (Donato Reference Donato, Lantolf and Appel1994; Storch 2002Footnote *).When co-constructing the scaffold, learners use language as a mediating tool. Language (whether it is the shared first language (L1) or L2) enables the learners to articulate their thinking and thus mediate their cognition. Research on collaborative writing has investigated the extent to which learners use their L1 and the functions that the L1 serves (e.g. Antón & DiCamilla 1998*; Swain & Lapkin 2000*; Storch & Wigglesworth 2003*).

Researchers investigating collaborative writing tasks have also observed that simply assigning learners to work in pairs or small groups does not necessarily mean that they will engage in languaging and scaffold each other's performance. As the seminal study by Storch (2002)* has shown, learners may form different relationships when engaging in collaborative tasks, some of which are not conducive to language learning. Storch's (2002) model of dyadic interactions distinguishes between four types of relationships according to the level of learners’ contributions to the task and engagement with each other's contributions. This model has been adopted in many other studies investigating collaborative writing in face-to-face (e.g. Watanabe & Swain 2007*) and computer mediated environments (e.g. Li & Zhu 2013*).

One reason why learners may not contribute to a collaborative writing activity is because they may prefer to write individually. This preference has been reported by several studies which have elicited learners’ perspectives about the collaborative writing activity, whether completed face-to-face (e.g. Storch 2005*) or using Web 2.0 tools (e.g. Kessler 2009*). A number of researchers have sought to explain learners’ behaviour during collaborative writing more systematically, including the relationships they form, by using AT as their theoretical framework. AT views learners as intentional agents whose actions can be explained by reference to their goals, orientation to the activity, and emotions (e.g. Storch 2004*; Li & Zhu 2017*).

The aim of this article is to provide a timeline of empirical studies on collaborative L2 writing that documents what has been the focus of investigation and the shift in this focus over the past 25 or so years (from the mid-1990s). I have selected seminal studies as well as studies conducted by emerging researchers in a range of educational contexts, in the teaching of English as well as other languages (e.g. French, Spanish, Korean) in second, foreign and immersion classes and to young and adult learners at different levels of L2 proficiency, in order to showcase the interest that this topic has stimulated across the world.

Two periods can be identified in this timeline. In the first period (approximately 15 years) most of the studies report on collaborative writing implemented in classrooms, with students working in pairs and interacting face-to-face. These studies have tended to focus on the nature of languaging (e.g. quantity, focus, quality and resolution of LREs) and the factors that influence the generation of LREs. A few also investigated whether engagement in collaborative writing resulted in language learning gains. The second period corresponds to developments in technology and thus includes studies which report on computer mediated (CM) collaborative writing using a range of digital tools such as wikis and Google Docs. Alongside this focus on CM collaborative writing, we have ongoing research on important aspects of face-to-face collaborative writing. In a sense, this second period reflects the reality of L2 classrooms, where both forms of collaborative writing complement L2 instruction. Studies investigating CM collaborative writing have focused predominantly on the nature of learners’ contributions to the jointly produced text: that is, whether learners add, delete or change the content of the evolving text, including correcting the text for linguistic accuracy. However, there are different approaches adopted in this analysis of learners’ contributions. Some researchers quantify the different types of contributions (e.g. Kessler 2009*), others employ a qualitative approach, describing and exemplifying the type of contributions learners make (e.g. Li & Zhu 2013*). Furthermore, in CM interactions, learners’ attention to language can take two forms. Because all the participants in a CM collaborative writing activity have access to the joint text (synchronously or asynchronously), they can simply edit it without any interaction. Alternatively, they can engage in text-based interaction, similar to the languaging episodes (i.e. LREs) in face-to-face interactions. On the whole, contributions which edit the text for errors in language use in CM collaborative writing are treated as equivalent to languaging episodes, regardless of whether there is some interaction between peers about the correction made, because the act of editing draws learners’ attention to form. Some researchers use LREs (or slightly different labels) when analysing learners’ language related corrections of the joint text (e.g. Kessler 2009*).

The timeline has categorised these studies according to the following themes:

1. A. Learner interaction during collaborative writing: the nature of languaging

   1. 1. Quantity and quality of languaging

   2. 2. Use of L1 during languaging

   3. 3. Collective scaffolding during languaging

2. B. Factors influencing the nature of languaging during collaborative writing

   1. 1. Task type (e.g. jigsaw, dictogloss)

   2. 2. Task modality (oral vs written)

   3. 3. Learners’ L2 proficiency

   4. 4. Mode of interaction

3. C. CM collaborative writing: the nature of learners’ contribution to the text

4. D. Relationships learners form

   1. 1. In face-to-face collaborative writing

   2. 2. In CM collaborative writing

5. E. The outcomes of collaborative writing

   1. 1. The impact on the quality of the text produced

   2. 2. Evidence of language learning

6. F. Learners’ perceptions of and reflections about collaborative writing

7. G. Explaining learner behaviour during collaborative writing activities using AT