Pushing Immersion Forward

Roy Lyster

Merrill Swain: Distinguished Immersion Scholar

In 2016, the Center for Advanced Research on Language Acquisition (CARLA) at the University of Minnesota conferred its first ever Distinguished Scholar Award at its Sixth International Conference on Immersion and Dual Language Education. The inaugural recipient was none other than Merrill Swain, in recognition of her role as a leader in the field of immersion education, having made significant contributions in the areas of both research and service.

Merrill Swain's seminal papers on immersion in the 1980s emphasized the importance of student production. Her output hypothesis triggered a turning point in the conceptualizations of immersion pedagogy and has been a source of inspiration for many researchers including myself. The output hypothesis can be seen as an important stage in her thinking, acting as a bridge between her earlier large-scale, evaluative research on immersion education, and moving toward a Vygotskian sociocultural approach to second language acquisition (SLA), emphasizing the interpersonal and collaborative co-construction of knowledge in the language learning process.

Merrill Swain began her long and productive career in 1973 at the Ontario Institute for Studies in Education at the University of Toronto, where she is currently professor emerita. During her career, she co-authored or co-edited 12 books or special issues related to immersion education, 95 book chapters and 135 papers in refereed journals. In addition, she co-supervised a remarkable total of 64 PhD students. Given the extent of her prolific career that was always moving forward, this chapter will focus on only a small part of it, namely, her early work conducted specifically in French immersion classrooms in Canada. Whereas Swain's earlier notions of functionally restricted input and the need for pushed output derived directly from her observations of immersion classrooms, her later research covered many contexts other than French immersion classrooms, ranging from adult university-level learners of French second language (L2) (Swain et al., 2009) to aging adults living in long-term care facilities (Motobayashi et al., 2014; Swain, 2013). Readers interested in her entire career will need to continue reading this timely volume dedicated to her wide-ranging contributions, aptly edited by Hossein Nassaji and Mari Haneda.

As for this chapter, it is devoted entirely to Swain's early pioneering work in French immersion contexts and its impact on immersion education. I suspect that many new scholars will be more familiar with Swain's later work, but I want to stress the importance of knowing her earlier work and its direct impact on moving French immersion forward at a time when its instructional practices still needed to be more clearly defined and based on research evidence. I will conclude that this body of seminal research is not always taken into account in the ever-increasing implementation of various types of content-based L2 programs around the globe. The purpose of this chapter is thus to highlight Swain's pioneering work in French immersion while rekindling the collective memory about its importance in program design and implementation.

Advocate for Change

Initial conceptualizations of immersion education and other forms of content-based language teaching underestimated the extent to which the target language needs to be attended to. It was initially believed that the L2 would develop naturally through exposure to comprehensible input in the form of content teaching and that students would pick up the language given sufficient time and input. Yet, as early as 1974, Merrill Swain (1974: 125) expressed concern about the L2 development of French immersion students, stating that 'some of the errors do not disappear' even after many years in immersion. To explain this, she argued that, first, teachers tend to ignore spoken errors so as not to disrupt the flow of communication and, second, communication with peers tends to 'reinforce their own classroom dialect of French' (Swain, 1974: 126).

Then, in a subsequent and well-known paper, Canale and Swain (1980) also drew from immersion findings (e.g. Harley & Swain, 1978: 11) to support their assertion that 'even with young children, grammatical accuracy in the oral mode does not improve much after a certain stage, perhaps when the learners have reached a level of grammatical accuracy adequate to serve their communicative needs'. They put forth that 'there seem to be no strong theoretical reasons for emphasizing getting one's meaning across over grammatical accuracy at the early stages of second language learning' and proposed instead 'some combination of emphasis on grammatical accuracy and emphasis on meaningful communication from the very start' (Harley & Swain, 1978: 14).

Thus, from the beginning, Merrill Swain was aware of some of the limitations of immersion education and subsequently became an advocate for change and a key player in proposing solutions. And, solutions were indeed proposed in her seminal papers that followed (Swain, 1985, 1988).

Pushed Output

In the first seminal paper, Swain (1985) proposed that comprehensible input alone is insufficient for successful L2 learning. She argued that exposure to input via subject-matter instruction engages comprehension strategies that enable students to process language semantically but not necessarily syntactically. Learners can draw on pragmatic and situational cues, vocabulary, real-world knowledge and inference to comprehend language sufficiently, without processing structural elements in the language. Bypassing language structure in this way, however, is harder to do when producing the language. Swain thus argued in favor of ample opportunities for student output and the provision of feedback that would push students to express themselves more precisely and appropriately. Having to produce the target language allows students to become more aware of language structure – if they are pushed to convey their meaning in precise and comprehensible ways. This is called pushed output.

In the second paper, Swain (1988) illustrated how subject-matter teaching does not on its own provide adequate language teaching; language used to convey subject matter, she argued, needs to be highlighted in ways that make certain features more salient for L2 learners. She reported that content instruction did not invite much student production, was restricted in the range of language functions it generated, did not necessarily engage students in form-function analyses and provided students with feedback only very inconsistently.

Swain (1988) argued that typical content teaching tends to involve more teacher talk than student talk. She reported findings from an immersion observation study in which only about 14% of the turns produced by sixth-grade students were considered to be sustained (i.e. more than a clause in length). Swain concluded accordingly that typical content teaching does not provide opportunities for sustained student production. In light of (a) the input-based instructional approach associated with content teaching, (b) observations of minimal production by students and (c) their lower-than-expected levels of grammatical competence, Swain (1993: 159) proposed the output hypothesis: 'through producing language, either spoken or written, language acquisition/learning may occur'.

In 1993, Swain was an invited contributor to a special issue of The Canadian Modern Language Review to commemorate its 50th year. In her contribution, she expounded upon the output hypothesis by proposing four ways in which output may play a role in L2 learning (Swain, 1993). First, output provides learners with opportunities for meaningful practice of their linguistic resources and thus for developing automaticity in using these resources. Second, output pushes learners to move from semantic processing to syntactic processing and, as a result, to notice what they do not know or know only partially. When learners notice a gap between what they need to say and what they know how to say, they can respond in one of three ways: (a) ignore the gap; (b) identify the gap and pay attention to relevant input; or (c) search their own linguistic knowledge for information that might help close the gap by generating new knowledge or consolidating existing knowledge. Third, output has a metalinguistic function that enables learners to use language in order to reflect on language. Fourth, as learners stretch their interlanguage to meet communicative needs, they use output as a way of testing hypotheses about new language forms and structures, which in turn generates responses from interlocutors about the comprehensibility or well-formedness of their utterances.

Empowering language learners to engage actively in increasing the accuracy of their L2 production is thus at the core of the output hypothesis. When learners seek input to bridge gaps in their knowledge, they demonstrate agency over their learning as they discover aspects of the L2 of which they were previously unaware. Moreover, by incorporating hypothesis testing into the output hypothesis, Swain adds an element of conscious play and experimentation to the process of language learning.

I thought I was pushing my students in their output but I wasn't

Studying at the Ontario Institute for Studies in Education in the 1980s with Merrill Swain, Birgit Harley, David Stern, Sharon Lapkin and Jim Cummins was indeed an exciting time for me. I was a French immersion teacher from 1982 to 1988, during which time I completed my MEd and then began my doctorate in 1988, right on the heels of Swain's two seminal papers. Yet, perhaps as a typical doctoral student, I was skeptical about the idea that all that immersion students needed was to produce more output – or at least that's how I interpreted the output hypothesis at the time. In a professional journal, Swain and Lapkin (1986: 8) wrote that there 'seems to be very little real exchange of information which goes on in immersion classes' and that there is 'little working through of ideas interactively'. In a graduate paper I wrote, which became my first publication (Lyster, 1987), I reacted to their article as follows:

I have always tried to maintain a highly interactive classroom in which students are free to communicate amongst themselves as long as they do so in French, and in which group discussions and the expression of opinions form a significant part of the learning process. This approach seems to help students to develop fluency, but not accuracy. (Lyster, 1987: 714)

I have subsequently had the opportunity to analyze my 'highly interactive classroom' and have come to better understand a key component of pushed output that was missing. Here is the story (from Lyster, 2016).

In 1988, my eighth-grade students and I participated in a video produced by the Ontario Institute for Studies in Education in Toronto. In addition to the final product (a professional video about 75 minutes long used for immersion teacher education), I was given a video tape with hours of unused footage. From this source, I was able to transcribe various exchanges between me and my students for the purpose of analyzing my questioning techniques in terms of their effectiveness. The following exchange was extracted from a discussion during our study of the novel Max by Monique Corriveau (1966). Focusing on the plight of the eponymous character in Chapter 2, I asked students to put themselves in his shoes and to imagine how he must have felt as he took refuge overnight in a small projection room in the Aquarium du Québec. In doing so, I thought I was creating opportunities for sustained student output.

Presumably with the goal of encouraging more complex language, I was trying hard to get students to articulate the emotions that might be felt by someone hiding and being sought after. Although the initial questions had been planned with this in mind, I clearly had not planned any follow-up questions. I did not provide any corrective feedback and did not follow up with elaboration questions to push students beyond their useful but very short utterances: la peur (fear), confus (confused), fâché (angry). Instead, I was quick to affirm their responses with 'yes' or 'that's good'. Yet, as Wong and Waring (2009) have since noted, the use of approval markers such as these may inhibit learning opportunities insofar as they serve a 'finale' function that precludes further attempts by others to articulate their understanding or explore alternative answers. So, instead, I should have asked these students to explain why Max would be afraid, confused or angry. I also could have expected longer utterances with subject pronouns and verb phrases more consistent with those in the questions, that is, either first- or third-person subject pronouns and verbs in the conditional to identify probable yet uncertain feelings: J'aurais peur (I would be scared); Il serait confus (He would be confused); Il serait fâché (He would be angry). If it's worth the time, in the context of this novel study, to ask these questions about feelings, then it must be worth the time to expect more than fragmented answers of only one or two words.

In their seminal study of classroom discourse, Sinclair and Coulthard (1975) found that the most typical teaching exchange consists of three moves: an initiating (I) move by the teacher; a responding (R) move by the student; and an evaluation (E) move by the teacher. It is this IRE sequence that I was following in the above abstract.

The IRE sequence is seen as the quintessence of transmission models of teaching and typical of teacher-centered classrooms. It has been criticized for engaging students only minimally and for maintaining unequal power relationships between teachers and students. Nevertheless, the IRE sequence continues to permeate classroom discourse, probably because it helps teachers to monitor students' knowledge and understanding (Mercer, 1999). By assessing their students in an ongoing manner in the course of interaction, teachers are better equipped to plan and evaluate their teaching. Furthermore, IRE exchanges can develop into more equal dialogue if, in the third turn, the teacher avoids evaluation and instead requests justifications or counterarguments (Nassaji & Wells, 2000). In this regard, the evaluation move needs to be replaced by a follow-up move that aims to: (a) elaborate on the student's response or provide clarification; (b) request further elaboration, justification, explanation or exemplification; and (c) challenge students' views (Haneda, 2005). This kind of push helps students to deepen their understanding of ideas and concepts and provides opportunities for students to use language that is more complex than that found in the shorter answers observed by Swain and her colleagues. This would be pushed output.

Functionally Restricted Input

In addition to limited opportunities for student output, Swain (1988: 75) characterized immersion classroom input as 'functionally restricted' and illustrated this with two examples. The first was the finding that singular vous as a politeness marker was almost completely absent from classroom input, which helps to explain its absence from immersion students' sociolinguistic repertoire. The second was the finding that 75% of all verbs used by immersion teachers were restricted to the present tense or imperative forms, whereas only 15% were in the past tense, 6% were in the future tense and 3% were in the conditional mood, which helps to explain immersion students' limited use of conditional forms and their inaccurate use of past tense forms.

To illustrate functionally restricted input, Swain provided an example of a history lesson (translated from French to English) in which sixth-grade students are listening to their teacher initiate a discussion about life in late 18th-century Antilles:

How do you think these plantations ... are going ... to change ... life in the Antilles? [...] These people are going to sell their sugar, rum, molasses, brown sugar. They are going to make money. With the money, they are going to buy clothes, furniture, horses, carriages ... all they want and they are going to bring them back to the Antilles. (Swain, 1988: 71)

Even though this was a history lesson about events that took place almost 200 years earlier, the teacher used the immediate future tense to convey her message, which Swain (1988) qualified as 'superb from a content teaching point of view', but then continued:

Its use has brought the distant past into the lives of the children, got them involved, and undoubtedly helped them to understand the social and economic principle that this historical unit was intended to demonstrate. However, as a language lesson these examples illustrate several problems — problems which may arise in any instructional setting based on authentic communication; problems which arise at the interface of language and content teaching. (Swain, 1988: 72)

Swain (1988: 76) provided many useful pedagogical suggestions that are worth revisiting. Two that stand out for me are that teachers need to (a) be aware of their language use so that they can engineer contexts that demand specific and otherwise infrequent uses of language and (b) explore content sufficiently so that language in its full range emerges.