(1) Long, M. H. (1998). Focus on form in task-based language teaching. University of Hawaii Working Papers in ESL, 16(2), 35–39 (829 citations in Google Scholar)

I was torn between choosing this article or an even more frequently cited chapter (Long, Reference Long, De Bot, Ginsberg and Kramsch1991 – 3,719 citations) that covers similar ground. I opted for this one because I felt it provided the clearer account of focus-on-form and because it can be downloaded through Google Scholar.

Let me first clear up the difference between ‘focus-on-form’ and ‘form-focused instruction’. The latter refers to any instructional approach that involves overt or covert attention to phonological, lexical, grammatical or discoursal form while the former refers to a specific way of drawing learners’ attention to form. Let me also clear up another point. As Long (Reference Long, De Bot, Ginsberg and Kramsch1991) makes very clear, focus-on-form is not a method; it is a design feature that can be found in a variety of methods and approaches. But what exactly is ‘a design feature’? According to Long, it is a feature that is evident when we inspect transcriptions of lessons supposedly based on very different methods/ approaches – for example, how errors are corrected.

In his 1998 article, Long distinguishes three different ‘options’ for second language (L2) course design and, although he does not define what he means by ‘option’, it becomes clear that each option represents a different approach to language teaching. The first option is focus-on-forms, found in traditional teaching based on a linear set of discrete linguistic objects (i.e. a structural syllabus). The second option is focus on meaning where ‘the starting point is not the language but the learner and learning processes (p. 38) as in French immersion programs and Krashen and Terrell's (Reference Krashen and Terrell1983) Natural approach. The third option is focus-on-form, which also has a major focus on meaning but differs from the second option through the systematic provision of attention to language during communication. Here is Long's (Reference Long1998) seminal definition:

This definition rules out design features that do not briefly draw attention to linguistic elements, which do not arise incidentally, which do not occur in lessons where the main focus is on meaning or communication, and which do not arise in response to students’ online language problems.

There are problems, however, when it comes to identifying the specific procedures for delivering focus on form. Long draws on Doughty and Williams’ (Reference Doughty, Williams, Doughty and Williams1998) account of how focus on form differs in terms of the obtrusiveness of the instructional procedures. Some of these procedures – for example, input-flood and task-essential language – clearly involve focus on form as Long defines it, but others – especially those at the obtrusive end of the continuum – do not seem to. Consciousness-raising tasks (Ellis, Reference Ellis1993) and input-processing instruction (VanPatten, Reference Van den Branden1996) involve pre-selection of linguistic elements (and thus do not occur incidentally), they draw attention to the same form repeatedly rather than briefly, they occur in lessons where the main focus is on form rather than meaning, and they are not triggered by online language problems but by pre-emptive decisions about which linguistic problems to address. In his 2001 article, Long also refers to studies that involved the teaching of relative clause formation (e.g. Eckman et al., Reference Eckman, Bell and Nelson1988) as an example of how focus on form can speed up learning and lead to a higher level of ultimate attainment. But these studies involve focus-on-forms (i.e. there is pre-teaching of the target structure and mechanical practice exercises). I was left puzzled by what focus on form is and is not.

It might seem that I am being overly critical, especially as Michael Long is, sadly, no longer with us and so will not be able to respond to my comments. I have enormous respect for Long's work. He has had a huge influence on my own thinking in many ways. His exposition of the three instructional options in his 1988 article serves as the obvious starting point for teachers seeking to understand focus on form and it certainly shaped my own thinking about task-based language teaching. Reading the article, I found myself thinking ‘yes, yes’ but then also my questioning started. Of course, it is not surprising to find inconsistency in how Long conceptualized focus on form in its early stage. All constructs – especially those like focus on form that encompass both psycholinguistic and pedagogic perspectives – are subject to continuous rethinking, refinement, and redefinition. We should not expect watertight definitions of new constructs, and we should always be prepared to identify problems with them as a step towards greater clarity.

(2) Long, M. (2014). Second language acquisition and task-based language teaching (Chapter 12 – Methodological principles and pedagogic procedures). John Wiley & Sons (1,165 citations in Google Scholar)

Long continued to develop his ideas about focus on form in the following years (see Long, Reference Long, Ritchie and Bhatia1996, Reference Long, Lambert and Shohamy2000; Long & Robinson, Reference Long and Robinson1998) but brought them all together in this book, which I know was a long time in preparation and that I found myself waiting for. As the title suggests, the book explores the two areas that dominated Long's work – SLA and task-based language teaching. The book is a masterpiece. There are relatively few books that will survive the passing of time but I am sure this one will. The number of current citations in Google Scholar is amazing given it was published only a few years ago.

The book is of course about much more than focus on form. It is a full account of the theoretical bases for TBLT, the research that supports it, and the nuts and bolts of how to design, implement and evaluate a task-based programme. Focus on form receives full treatment in Chapter 2 where Long expands on his 1988 article again distinguishing focus-on-forms, focus on meaning, and focus on form. In this chapter, Long explains how focus on form brings together three major strands in his own research – the importance of negotiation of meaning (Long, Reference Long1983a) and, in particular recasts (Long, Reference Long2007) as a source of implicit learning, the need for form-focused instruction (Long, Reference Long1983b) as a source of explicit learning, and the role of maturational constraints on the acquisition of implicit learning (Long, Reference Long1990). These were all areas of SLA that figure in my own work where, like Long, I have grappled with how implicit and explicit teaching can be combined in ways that are compatible with how L2 acquisition takes place.

I agreed with Long that one way of achieving this combination was through focus on form. But I felt there was another way too. In Ellis (Reference Ellis1993), I put forward a proposal for ‘consciousness raising tasks. These were tasks that required learners to solve linguistic (usually grammatical problems) through reflection and collaborative talk. It seemed to me a good idea that learners could benefit from talk about language itself. The result would be explicit knowledge but I argued that learners could use this to fine tune their implicit knowledge gradually over time. In other words, whereas for Long the synergy of explicit and implicit knowledge needed to take place concurrently through focus on form, for me it could be delayed. (I also thought that such tasks could be considered ‘communicative’ through the talk they generated. Long was dismissive of consciousness-raising tasks but I feel he never really understood my rationale for them.

As in the early definition, focus on form is ‘brief’ (and now we are told that this can mean just ‘a few seconds’), it is embedded in meaning-based or communicative activities, and it is reactive (i.e. it is response to learner difficulty) rather than proactive. There are, however, some developments. It is now made clear that it can be explicit as well as implicit (i.e. it does not just attract but also directs attention to form), that it does not always arise because of a communication problem, and that it can (but need not) involve awareness although never of rules.

I found myself asking a number of questions:

1. 1. Why is attracting attention to form more prototypical of the construct than directing attention form?

2. 2. Why does focus on form always have to brief? Is there a case for taking time out from completing a task to provide a full and explicit explanation of a structural feature that is proving problematic? This is what the teacher in Samuda's (Reference Samuda, Bygate, Skehan and Swain2001) study did when the learners failed to make use of the structure targeted by the task. Long cites this study although it would seem it does not conform to his definition of focus on form.

3. 3. What else can trigger focus on form other than a communication problem? For example, do teachers sometimes direct attention to form even though they have understood what a learner said?

4. 4. Why does focus on form always have to be reactive? Isn't there a case for a teacher sometimes stepping in to prevent a problem arising? Do teachers do this when performing tasks with their students? Isn't the kind of task that Samuda used ‘proactive’ as it sought to focus attention on a predetermined grammatical structure? Focused tasks of this kind have in fact figured strongly in research on task-based language teaching (e.g. Doughty & Varela, Reference Doughty and Varela1998; Mackey, Reference Mackey1999).

5. 5. What are relative merits of explicit and implicit focus on form? Long clearly favours implicit focus on form (e.g. recasts) but he also recognizes a need for more explicit focus on form. But why should explicit focus on form not sometimes result in the learning of ‘rules’ especially if it involves metalinguistic comments?

I was also struck by the fact that ‘focus on form’ sometimes refers to a psycholinguistic condition (i.e. learners attending to a form) and sometimes refers to the pedagogic procedures that attract or directs attention to form.

I have always found reading Long's writing exciting because he shows how research can inform practice, he draws on an encyclopaedic knowledge of the field, he never holds back in expressing forthright views, and, especially, because he always makes me want to ask questions and to challenge. His book is important to me because it presents such a cogent and informed rationale for task-based language teaching, which, like Long, I am convinced is the solution to the problems with language teaching I witnessed when I lived and worked in Japan and, more recently, on my visits to Korea and China. Too many students were spending many years studying English without developing any true communicative ability. But it is also my experience of these countries that led me to think that Long's version of task-based teaching was too narrow and that there was a need for some explicit language teaching as well as focus on form. My thinking about language teaching has always been shaped by both research and my experience of practical exigencies.

(3) Ellis, R. (2016). Anniversary article. Focus on form: A critical review. Language Teaching Research, 20(3), 405–428 (266 citations in Google Scholar)

I was invited to write this anniversary article for Language Teaching Research and opted for focus on form as my topic. I accepted this invitation because it provided me with an opportunity to clarify my own (and hopefully other's) understanding of the focus on form.

My starting point was Long's definition of focus on form as originally stated in the 1988 article and elaborated in his 2014 book. It seemed to me that it was necessary to distinguish two different senses of the construct – as an instructional approach (Long uses the term ‘program’) and as a set or procedures for inducing attention to form. I suggested we should use ‘FonF’ for the former and ‘focus on form’ for the latter. FonF contrasts with FonFs but, as Long points out, focus on form has a place in FonFs as a methodological principle. I think that making this distinction is helpful because it enables us to consider the merits of FonF and FonFs as approaches separately from the merits of focus on form as a methodological principle.

In the course I teach on task-based language teaching to M.A. students, I use my article to present a rounded account by examining focus on form from three perspectives. The first is the pedagogic perspective. I argue that FonF and FonFs are not entirely distinct approaches but rather poles on a continuum with task-based language instruction a clear example of one pole and traditional structure-based instruction (as in the Audiolingual Method) at the other end. I point out that some structure-based approaches, such as present-practice-produce (PPP), also allow for focus on form as a methodological principle in the production stage of a lesson, which I think is also Long's view. Where I depart most clearly from Long is in my discussion of focus on form procedures. In particular, I argue that focus on form can be pre-emptive as well a reactive, need not always be brief, and can occur both within a task and outside a task (e.g. when learners engage in pre-task planning or when the teacher corrects errors after a task has been completed). I invite my students to consider to what extent they favour Long's narrow definition of focus on form and my broader one.

The second perspective I discuss is psycholinguistic. In this sense, focus on form is a mental phenomenon. I point to two key features – selective attention and cognitive comparison. Selective attention is required to focus on specific forms and the meanings they convey in context. Like Long, I consider that it can occur consciously as in Schmidt's (Reference Schmidt1993) Noticing Hypothesis or subconsciously when initial neural connections are strengthened and modified through exposure to input (N. Ellis, Reference Ellis2005). Cognitive comparison occurs when learners attend to the difference between their own output and input containing target language forms.

The third perspective is discoursal. Focus on form as an interactional phenomenon originated in the early work on the negotiation of meaning (Long, Reference Long1983a; Pica, Reference Pica, Kramsch and McConnel-Ginet1992). This has provided rich information about how the strategies such as recasts are employed to address communication problems that attract attention to form. Subsequently, researchers such as Lyster (Reference Lyster1998) have pointed out that similar strategies are used in classrooms even when no communication problem arises. Lyster coined the term ‘negotiation of form’ to refer to this phenomenon. Researchers have also shown that these strategies can be both pre-emptive and reactive.

Using one's own article with students is always risky because it can easily be seen a pre-empting debate by presenting a definitive position. I think, therefore, that is essential to use an article like this one side by side with other articles inviting students to identify areas of agreement and difference as a basis for constructing their own views. In his 2014 book, Long mentions a ‘long-standing affliction in applied linguistics’ – namely. ‘the tendency to dilute the meaning of new terms and the constructs behind them’ (p. 6) and specifically mentions ‘focus on form’ as an example. I do not think I have ‘diluted’ its meaning although I have clearly extended and hopefully clarified it in several ways. I have remained true to the central feature of Long's definition – namely, that focus on form is tied in some way to the performance of a communicative task. Once this is lost – as I think it is, for example, in Benati (Reference Benati2021) where the term is used to refer to non-task found in processing instruction – then the construct does indeed become vacuous.

(4) Pica, T. (1992). The textual outcomes of native speaker-non-native speaker negotiation: What do they reveal about second language learning? In C. Kramsch & S. McConnel-Ginet (Eds). Text and context: Cross-disciplinary perspective on language study (pp. 198–237). Heath and Company (324 citations in Google Scholar)

I met Tere Pica several times while I was working at Temple University Japan in Philadelphia in the 1990s. She was a professor at the University of Pennsylvania. I have always had huge respect for Tere and her work, which, like mine, aims to connect SLA with language pedagogy. My respect goes beyond her professional work; she did not have an easy life but bore it uncomplainingly.

I have chosen this book chapter because it provides the fullest account of Pica's work on the negotiation of meaning – the construct that initially informed focus on form. An alternative, which may be more accessible to readers, is Pica's (Reference Pica1994) article in Language Learning. I like the book chapter because in addition to a very clear analysis of the types of interaction found in negotiation of meaning sequences it also usefully summarizes a number of studies she completed with Doughty and Young (e.g. Pica et al., 1986, Reference Pica, Young and Doughty1987). These studies were motivated by Krashen's (Reference Krashen1985) claim about the importance of comprehensible input for acquisition and Long's (Reference Long1980) proposal regarding the special value of interactionally modified input. In Pica's studies, non-native speaker (NNS) – native speaker (NS) dyads performed different types of communicative tasks (jigsaw, information-gap, and opinion-gap). Their comprehension of the input that arose in the performance of the tasks was measured in terms of how successfully they achieved the communicative outcomes of the tasks. The main finding was that interactionally modified input very clearly facilitated comprehension. Interestingly, however, NNSs who observed negotiation were just as successful as those who actively engaged in it

The main part of Pica's chapter is entirely descriptive. It presents a framework for coding the negotiated nature of NS and NNS speech adjustments (see also Varonis and Gass (Reference VanPatten1985) for a similar framework). Pica's coding system distinguishes how speakers signal a need for clarification, confirmation, or repetition of another speaker's utterance, how this signal serves as a trigger for a negotiation sequence, and the variety of ways in which speakers respond to a signal. The appendix to the article provides the complete system along with examples of each category. In effect, this constitutes one of the most concrete accounts of how focus on form is accomplished interactionally. I have found that asking students to work through the examples is one of the best ways of developing their understanding negotiation of meaning.

The rationale for Pica's work on the negotiation of meaning is that ‘through negotiation learners are offered L2 structural and semantic information, feedback on their interlanguage, and opportunities to adjust, manipulate or modify their interlanguage semantically and structurally’ (p. 216) and that it is in these ways that negotiation is theorized to facilitate L2 acquisition. Pica is careful, however, to point out that the kind of cross-sectional, descriptive research she conducted does not provide actual evidence of acquisition and that longitudinal studies are needed for that. In the next section, we will look at experimental studies that have attempted to show a cause-and-effect relationship between negotiation and acquisition.

Pica's chapter influenced the research that I carried out in Japan with the help of some of my doctoral students in the 1990s. Pica's research was all laboratory-based (i.e. dyads were created for purposes of data collection). I was interested in investigating whether Pica's findings were applicable when tasks were carried out in a lockstep classroom context. Ellis et al. (Reference Ellis, Tanaka and Yamazaki1994) reported such a study where students negotiated their understanding of a listen-and-do task with a teacher. We were able to show that negotiation did occur in such a context, that it was not necessary for students to participate in negotiation to benefit from it, and that actual learning took place as a result of performing the task. It is important, I think to check whether the findings of laboratory-based research have ecological validity for the classroom.

In her later research, Pica continued to explore focus on form within classrooms. For example, her 2002 article reported a study of teacher-led discussions based on literary texts. Pica found that there were relatively few interactional opportunities that led to the kinds of input, feedback, and production likely to focus attention on form-meaning relationships and suggested ways of remedying this. Her 2004 article describes the kind of information-gap tasks that fit well into content-based language teaching and how to implement them in ways that promote focus on form.

Pica's research on negotiation like that of others at this time focused on young adult language learners. But there is also interesting work with young children by Rhonda Oliver (Reference Oliver1998, Reference Oliver2009), with whom I have had the pleasure of working alongside in the last few years.

(5) Lyster, R., & Ranta, L. (1997). Corrective feedback and learner uptake: Negotiation of form in communicative classrooms. Studies in Second Language Acquisition, 19(1), 37–66 (3,974 citations in Google Scholar)

I have had relatively little personal contact with Roy Lyster and even less with Leila Ranta but I do remember a very pleasant meal with Roy in Montreal shortly after he retired. I am an admirer of his work, which manifests the qualities I consider central to applied linguistic research – readability, insightfulness, a balanced perspective, and practical relevance. I chose this article because it is classroom-based, it dealt with one of the main ways of accomplishing focus on form (i.e. through corrective feedback), and it is very readable. I have found it always has an impact on students, making them rethink their ideas about what corrective feedback involves and what makes it effective.

Drawing also on discourse analysis, Lyster and Ranta described and quantified elements in the error treatment sequences extracted from their classroom data. This article is by no means the first descriptive corrective feedback article (see, e.g., Allwright (Reference Allwright, Burt and Dulay1975) and Chaudron (Reference Chaudron1977) but it is by far the most influential as it served as the basis for a host of experimental studies investigating the effect of corrective feedback on acquisition, with Lyster's own studies (e.g. Lyster, Reference Lyster1998, Reference Lyster2004, Reference Lyster2005; Lyster & Mori, Reference Lyster and Mori2006; Lyster & Izquierdo, Reference Lyster and Izquierdo2009) prominent.

Lyster and Ranta's approach to focus on form is very different to Long's and Pica's. Their data base was immersion classrooms. Their theoretical base was the distinction between experiential and analytic teaching strategies (Stern, Reference Spada and Tomita1990), which can occur side-by-side in different degrees in different classrooms. They saw corrective feedback as an analytic teaching strategy that serves to introduce a focus on form into classrooms that are predominantly experiential. The key word here is ‘analytic’. Corrective feedback involves negotiation but negotiation of form rather than negotiation of meaning. That is, it is didactic rather than conversational in orientation as its purpose is to repair linguistic errors, not to remedy comprehension problems. In extending the sense of negotiation in this way, Lyster and Ranta included episodes where a teacher addresses learners’ language learning needs directly and sometimes very explicitly, a possibility not evident in Long's early account of focus on form but which he later incorporated into his extended account.

To my mind, the descriptive value of the study lies in its taxonomy of teacher feedback types and the coining of the term ‘uptake’ to describe how learners respond to feedback. The taxonomy distinguishes six types of feedback – explicit correction, recasts, clarification requests, metalinguistic feedback, elicitation, and repetition. There were two basic types of uptake – uptake-with-repair and uptake that needs repair – with subcategories of each. The appendix to the article provides examples of these feedback and uptake categories. The value of the article lies in this taxonomy that has informed multiple studies of corrective feedback, including one of my own (Ellis et al., Reference Ellis, Loewen and Erlam2006)

Lyster and Ranta also quantified the occurrence of the feedback and uptake categories. A major finding was that recasts predominated. A second major finding was that uptake-with-repair was much less likely to occur with recasts than after other types of feedback (in particular, elicitation and metalinguistic feedback). This led to them proposing that feedback will work better for acquisition if learners are prompted to self-repair rather than simply given the correct form as in recasts. In other words, Lyster and Ranta were effectively prioritizing the role of output over input in L2 acquisition, which, not surprisingly, given the importance attached to input by theorists such as Krashen (Reference Krashen1985), led to subsequent debate – see Goo and Mackey (Reference Goo and Mackey2013) and Lyster and Ranta (Reference Lyster and Ranta2013). I wonder, though, if this debate is of much relevance to teachers. My own view is that recasts and prompts can both facilitate acquisition in different ways. Also, teachers are unlikely to limit their feedback to just recasts or prompts and much more likely to employ a mixed bag of corrective strategies. This what I advise and, in fact, this is also the position that Lyster and Ranta (Reference Lyster and Ranta2013) espoused.

Descriptive studies such as Pica (Reference Pica, Kramsch and McConnel-Ginet1992) and Lyster and Ranta (Reference Lyster and Ranta1997) have provided us with rich detail about the two primary types of focus on form – the negotiation of meaning and the negotiation of form – but they rely on theory to make claims about their contribution to L2 acquisition. The next article also reports a study that is primarily descriptive in nature but it adopted an interesting research strategy to investigate whether focus on form leads to learning and that is the main reason I have chosen it.

(6) Loewen, S. (2005). Incidental focus on form in second language learning. Studies in Second Language Acquisition, 27(3), 361–386 (487 citations in Google Scholar)

Shawn Loewen was my student. He took classes with me at Temple University when he was completing his M.A. there and then, when I moved to New Zealand in 1998, he chose to register for a Ph.D. in the University of Auckland and took up a research assistant position in the Institute of Language Teaching and Learning where I was the director. He worked with me and Helen Basturkmen on a project investigating focus on form episodes in communicative classrooms (Ellis et al., Reference Ellis, Basturkmen and Loewen2001), which provided the basis for his subsequent doctoral research. After completing it, Shawn took up a position as lecturer in what became the department of Applied Language Studies and Linguistics. Shawn stayed for several years, working on a project investigating the testing and teaching of implicit and explicit grammatical knowledge (Ellis et al., Reference Ellis, Loewen, Erlam, Philp and Reinders2009) before taking up a position in Michigan State University. He has played a major part in my professional life and I have benefited enormously from his expertise, especially statistically. He is now a leading figure in instructed SLA.

Loewen's idea for moving research on focus on form forward was to replicate the descriptive approach of Ellis et al. (Reference Ellis, Basturkmen and Loewen2001) and then to use tailor made tests to investigate whether learners could recall the forms that had figured in the focus-on-form episodes (FFEs) that they had participated in during communicative lessons. The coding of the FFEs distinguished their linguistic focus (i.e. vocabulary, grammar, and pronunciation), negotiation of meaning and negotiation of form, whether uptake took place and, if it did, whether it involved repair of the linguistic problem. The tests were administered one day or two weeks later on the individual learners who had participated in specific FFEs. Loewen reported an overall recall level of 60% after one day and 50% after two weeks. Learners were more likely to recall lexical and grammatical forms if an FFE included uptake with repair and more likely to remember the correct pronunciation of an item if the FFE involved the negotiation of meaning rather than the negotiation of form. His study provides evidence that incidental focus on form can assist learning and thus was a step further than purely descriptive research. However, both Loewen and I were well-aware of the limitations of the study. Because it was impossible to administer pre-tests (i.e. there was no way of knowing whether FFEs would occur or what their focus would be), we could not compare pre- and post-test scores to see if learning was taking place. However, Loewen pointed out that a learner's participating in an FFE could be seen as indicating that they were experiencing linguistic problems either because they did not know the item or because their knowledge of it was incomplete.

To my mind the strength of this study is its non-interventional nature. It examined FFEs as they naturally occur in communicative classrooms. It confirmed that FFEs can occur very frequently – Loewen recorded 491 in 32 hours of teacher-led classroom interaction – pointing to the important role they can play in learning. Another study (Basturkmen et al., Reference Basturkmen, Loewen and Ellis2004) found that although teachers do focus on form a lot, they are doubtful about its legitimacy in fluency-oriented lessons and think they should avoid it. When they are shown that they do inadvertently do it, they are surprised but gratified to learn that it need not detract from the communicativeness of a lesson and, most importantly, can benefit learning. I have found that Loewen's study is more likely to convince teachers about the need for focus on form than the experimental studies that I consider in the next section.

(7) Mackey, A. (1999). Input, interaction, and second language development: An empirical study of question formation in ESL. Studies in Second Language Acquisition, 21(4), 557–587 (1,227 citations in Google Scholar)

I first met Alison Mackey when I was working for Temple University. She was on a visit from Australia where she was completing her Ph.D. at the University of Sydney. She was not yet a ‘name’ in SLA studies but I noted her confidence and thought that she might soon be. I was right, Alison rapidly became a leading figure in cognitive-interactionist SLA and has carried out a number of influential experimental studies. The article I have chosen was based on her Ph.D. It is an early article but arguably the most seminal.

In order to carry out her study, Mackey needed to operationalize the independent variable (i.e. interaction) and the dependent variable (i.e. acquisition). She formed four experimental groups that differed in how they experienced interactional modifications – there were two groups of interactors who participated actively in performing tasks (one group being developmentally ready to acquire the target structure and the other group developmentally unready), a group of observers who just listened to others performing the tasks, and a group of ‘scripteds’ who were exposed to specially modified input but had no opportunity to interact.

One of my reasons for choosing this article was that it involved the use of what I have called ‘focused tasks’. Such tasks are similar to the kinds of information and opinion gap tasks used by Pica in her research but differ from them in that they have been specially designed to provide occasions for the use of a specific target structure – English question forms in the case of Mackey's study. I have long felt that such tasks, which are required in an experimental study, also have a place in task-based language teaching. This is yet another way in which my ideas differ from those of Long.

For the dependent variable (i.e. development), Mackey drew on Pienemann and Johnston's (Reference Pienemann, Johnston and Nunan1987) account of the stages that learners pass through in the process of acquiring English questions. Development was operationalized in terms of whether there was any evidence of a shift from one developmental stage to another over time. We see here, then, how research on tasks was closely linked to work in SLA. There was a pre-test, three treatment sessions, and three post-tests. There was also a control group that just completed the tests. The results provided clear evidence that participating in negotiated interaction led to developmental gains in the target structure for both developmentally ready and unready learners and that these gains were durable (i.e. they actually increased over time). The interactors not only outperformed the control group but also the observers and the scripteds.

Mackey's study is not a model for the design of similar studies. Like most studies, it has its limitations, of which Mackey was well aware. For example, the experimental groups were very small and Mackey only investigated a single target structure. But the study has some real strengths. In particular, I liked how development was operationalized in terms of an attested sequence of acquisition. Most other studies have measured acquisition purely in terms of gains in target-like accuracy, which ignores the fact that acquisition can occur even if learners fail to produce a feature correctly. Another strength of Mackey's study is that she also illustrated how participation in specific interactions seemed to push the interactors’ development along. What I really liked about the study was that it was innovative. I sometimes feel that top journals attach too much importance to the technical aspects of a study at the expense of originality and innovativeness.

Mackey's study was framed by Long's Interaction Hypothesis (Long, Reference Long, Ritchie and Bhatia1996) and the preceding descriptive research (e.g. Pica, Reference Pica, Kramsch and McConnel-Ginet1992). Much of the other experimental research, however, focused on one aspect of interaction – corrective feedback – drawing on Lyster and Ranta's (Reference Lyster and Ranta1997) seminal article. This research has been very varied investigating a number of different dimensions of corrective feedback such as prompts versus recasts (e.g. Lyster, Reference Lyster2004) and implicit versus explicit feedback (e.g. Ellis et al., Reference Ellis, Loewen and Erlam2006). I have found it difficult to identify a single article that stands out as making a major contribution and so have decided to include a meta-analysis of experimental corrective feedback studies as my next selection. I also felt that it would be amiss not to include an example of meta-analysis in my essential bookshelf given its increasing importance in applied linguistics in general.

(8) Li, S. (2010). The effectiveness of corrective feedback in SLA: A meta-analysis. Language Learning, 60(2), 309–365 (962 citations in Google Scholar)

There have been a surprising number of meta-analyses of corrective feedback studies. Plonsky and Brown (Reference Plonsky and Brown2015) identified a total of 18 although these also included studies of written corrective feedback. I have chosen Shaofeng Li's meta-analysis because it investigated the effectiveness of just oral corrective feedback and not just descriptive aspects as in Brown's (Reference Brown2016) meta-analysis. Also, it stands out as an excellent example of how to conduct a meta-analysis.

Shaofeng came to work in the Department of Applied Language Studies and Linguistics at the University of Auckland and was, in fact, the last of the appointees that I was personally involved in. I collaborated with him in a number of studies and we became friends. He left the University of Auckland not long after my own departure and has gone on to be a leading figure in SLA studies focusing on corrective feedback, individual learner differences and research methodology in applied linguistics.

To conduct a meta-analysis, it is necessary to define the domain to be investigated (i.e. oral corrective feedback) and then to conduct electronic and manual searches for primary studies, including both published and unpublished studies (e.g. Ph.D. theses). These studies then need to be sifted using explicitly stated inclusion and exclusion criteria (e.g. whether they used statistical analyses that would allow for effect sizes to calculated). Li ended up coding 33 studies for feedback type, outcome measures, and various moderator variables such as the timing of the post-tests, the task type, and the learners’ age. The results are reported in terms of effect sizes, which serve as practical measures of the effectiveness of corrective feedback. Overall corrective feedback had a medium effect size on L2 acquisition. Explicit feedback (e.g. explicit correction) had a stronger effect than implicit (e.g. recasts). The effects varied according to some of the moderator variables, notably research context where the effect was greater in foreign than in second language contexts and task type where the effect was greater in mechanical drills than in communicative activities. Age, however, did not have a moderating effect although another meta-analysis (Lyster & Saito, Reference Lyster and Saito2010), which included only classroom-based studies of corrective, did find age a factor, with younger learners more likely to benefit from recasts than older learners.

Li's meta-analysis like that of others (e.g. Brown, Reference Brown2016; Lyster & Saito, Reference Lyster and Saito2010; Spada & Tomita, Reference Shintani2010) provides clear evidence that corrective feedback has a positive effect on L2 acquisition and indicates which type is most effective and under what circumstances. Li went on to carry out a number of corrective feedback studies based on gaps in the research that his meta-analysis had revealed, including research that I was also involved in. For example, Li et al. (Reference Li, Zhu and Ellis2016) investigated the effect of corrective feedback on acquisition according to whether the feedback occurred during the performance of a task or after the task had been completed. We found no difference even when acquisition was measured by a test of learners’ procedural knowledge of the target structure.

One of the gaps that Li identified was the paucity of experimental studies of children. This gap is not only evident in corrective feedback research but also in other aspects of experimental focus on form research. The final study I have selected is unique because it looked at very young children who were complete beginners. It addressed a question of fundamental importance – whether FonF is more effective than FonFs, as Long and others (myself included) have claimed.

(9) Shintani, N. (2016). Input-based tasks in foreign language instruction for young learners. John Benjamins (86 citations in Google Scholar)

Natsuko Shintani was also one of my doctoral students at the University of Auckland. She completed her thesis in 2011 and went on to publish a series of well-cited articles based on it (Shintani, Reference Shintani2011, Reference Shintani2012a, Reference Shintani2012b, Reference Shintani2013, Reference Shintani2015). She eventually published a book, and this is what I have chosen for my essential bookshelf as it provides the most complete account of her research. Natsuko was a model Ph.D. student – hard-working, well-informed, creative, methodologically broad-minded, and statistically-savvy. These qualities are all manifest in her book. I thoroughly enjoyed working with her and, in fact, have gone on to do so in a number of projects in recent years.

A bit of background is needed to understand the importance of Shintani's research. Norris and Ortega's (Reference Norris and Ortega2000) meta-analysis of form-focused instruction studies included a comparison of the effectiveness of FonF and FonFs. They coded individual studies post-hoc as either FonF or FonFs. A true comparison of the two types of instruction, however, requires a single study involving the same type of students and the same instructor. Such studies are very rare and are often not well-designed. Shintani's was one of the first.

Shintani's research is a model process-product study. She utilized six-year Japanese children who were enrolled in classes in her own private language school in Japan. They were all complete beginners. FonF was operationalized as input-based tasks. They involved focused tasks where the children listened to the teacher's directions and demonstrated their comprehension by selecting the correct picture card from an array of picture cards. The target language was a set of words naming animals, home appliances, and fruit and vegetables and two grammatical structures (plural-s and copula be). FonFs was operationalized as present-practice-produce (PPP). There was a pre-test and two post-tests that assessed comprehension and both constrained and free production of the target features. Shintani found that both types of instruction were equally effective in enabling the children to learn the target words. Interestingly, though, FonF resulted in better acquisition of the adjectives that arose incidentally in the input as the teacher worked to help the children comprehend her directions. Grammar acquisition only occurred in the FonF group but only for plural-s, not for copula be, which Shintani explained was because successful task completion made attention to plural-s but not to copula be essential.

Another reason for including Shintani's study in my essential bookshelf is that it involved a whole series of lessons in a real classroom and thus has greater ecological validity than the typical laboratory-based experimental study involving just one of two treatments. The same communicative tasks were repeated nine times over five weeks. All the lessons were recorded and transcribed, enabling Shintani to compare the process features of the FonF and FonFs. This is another strength of the study as it has become increasingly important in experimental research to document the process as well as the product of the treatment. Shintani reports results for two key process features – turn-taking and repair – with clear differences evident in both. For example, turn-taking sequences were longer in FonF lessons, students were more likely to initiate questions, and the turn-taking was in part managed by the students. In other words, the input-based tasks did not exclude contributions from the students who engaged actively, initially using their L1, but later English in their struggle to understand the teacher's directions. In contrast, FonFs constrained both the teacher's input and the students’ output.

It is always pleasing to see one's doctoral students progress in their careers. After finishing her thesis, Natusko first worked in the National Institute of Education in Singapore, then in the Faculty of Education and Social Work in the University of Auckland and is currently a full professor in Kansai University, Osaka. She has continued to be an active researcher and a leading advocate of TBLT in Japan.

(10) Bryfonski, L. (2021). From task-based training to task-based instruction: Novice language teachers’ experiences and perspectives. Language Teaching Research, 13621688211026570 (3 citations in Google Scholar)

Van den Branden (Reference Swain and Lapkin2015, Reference Van den Branden2016) has written informatively about the role of teachers in task-based language education and generally about teacher education for TBLT but I wanted a reading that reported an actual study of teachers’ professional development. One article I was familiar with was Vásquez and Harvey (Reference Varonis and Gass2010). They asked student-teachers to replicate Lyster and Ranta's (Reference Lyster and Ranta1997) descriptive study to see if doing so led to changes in their beliefs about corrective feedback, reporting that it did lead to a broader acceptance of the need to correct errors in task-based instruction. But I wanted an article that looked at the effect of training on teachers’ actual practice of focus on form. With the help of Google Scholar, I arrived at Bryfonski's interesting study based on her doctoral dissertation, noting that Alison Mackey had been her supervisor. I also liked the context of her study – English–Spanish bilingual schools in Honduras – as this was very different from the contexts of other studies in my selected readings.

The main reason for choosing this article rather than say Erlam (Reference Erlam2016), whose study examined the effect that training had had on teachers’ ability to design tasks, was that it focused on the implementation of tasks, including focus on form. Bryfonski based the training she provided on Long's methodological principles. She video-taped the teachers’ lessons after the training and also collected their stimulated recalls through post-lesson interviews, coding both the lesson observations and the teachers’ recalls in terms of the extent to which they demonstrated the principles. Interestingly, both ‘focus on form’ and ‘provide negative input’ were among the lowest scores for the observational data while ‘provide negative input’ was the very lowest for recalls. In other words, the training had had less effect on the teacher's use of and cognitions about focus on form than it had had on the other principles, many of which relate to very general aspects of good teaching (e.g. ‘promote a cooperative learning environment’). Bryfonski suggested that the results of her study point to areas that need special attention in a training programme. One of these is clearly focus on form.

I think one way of using Bryfonski's article in a TBLT course might be to ask teachers to design their own study investigating the effects of training on the implementation of tasks and on focus on form in particular. Rather than using Long's principles, which as Bryfonski admitted require high-inferencing coding, such a study could draw on Ellis et al.'s (Reference Ellis, Basturkmen and Loewen2001) taxonomy of focus-on-form episodes and investigate what effect training has on their occurrence in task-based lessons.