Output (Swain)

The traditional view of language teaching was that it involved learning the rules of grammar and individual words, then somehow putting it all together in output. This is what might be called a synthetic approach to language learning. But it doesn’t reflect reality because there’s simply too many bits to synthesize. For production to work, almost all of these bits need to be processed automatically.

Another view of language learning, promoted from the mid 1980s by Stephen Krashen, is that language is learned through input. But this view has also been critiqued.

Merrill Swain researched students in Canada who were studying at French-medium schools. In other words, the students were learning French by being immersed in the language through a content-based approach. She compared the results of 6th grade native and non-native speakers of French in grammar, discourse and sociolinguistic areas.

Wordle: Swain Output Hypothesis

What Swain found was that while these students were achieving near-native levels of comprehension in French when reading and listening, and were reaching grade-level expectations in curriculum subjects, they were not achieving as well when speaking and writing French. Their spoken and written language contained morphological and syntactical errors, lacked precise vocabulary and tended to use formal academic styles of writing at the expense of other types of writing.  In other words, their productive skills lagged well behind their receptive skills and they seemed to be able to understand input while only partly processing the language it contained.

Swain argued that language input wasn’t enough. What tended to happen in these content-based classrooms was teachers provided a lot of input, but students were not expected to produce much output. In whole class discussions, Swain found that 44% of student turns were limited to 1 or 2 words and only 14% of student turns were longer than a clause. The remaining 42% consisted of student turns of more than 2 words up to a clause.

Input also tended to be restricted: Swain found that of the verbs used by teachers, 75% of them were either in the present tense or the imperative, and only 15% in the past, 6% in the future and 3% in the conditional. This meant students weren’t being exposed to more advanced language.

Teachers also tended to accept grammatically incorrect but understandable responses – after all, these were content-based lessons, so the focus was on content not language. Only 19% of grammatical errors were corrected and Swain suggested that even these corrections seemed to determined as much as by ‘irritation’ than for pedagogical reasons. There seemed to be an assumption that the language would take care of itself. Children would simply ‘pick it up’.

There was explicit teaching of grammar, but in isolation from content. Swain found that grammar taught in these explicit lessons was rarely related to what was being taught in academic subjects and was instead focused on manipulating and categorizing language forms, not actually using them for communication.

Vocabulary instruction tended to be done in discrete reading sessions, when words were explained only in the context of the reading material, but the wider application of such words was rarely discussed.

Although they are able to communicate with some fluency in the second language, students often fall short of the high levels of linguistic accuracy that their years of schooling in the language might predict.

(Lightbown and Spada p.160)

an input-rich, communicatively oriented classroom does not provide all that is needed for the development of targetlike proficiency. It also makes clear that teaching grammar lessons out of context, as paradigms to be rehearsed and memorized, is also insufficient.

Swain, Focus on Form through conscious reflection,in (I think) Doughty and Williams 1998, page 65

Swain added a convincing argument that the reason why input alone was insufficient was because input only involved understanding meaning (semantic processing) while form accuracy in output involved syntactic processing.

Input alone is not sufficient for acquisition because when one hears language, one can often interpret the meaning without the use of syntax.

(Gass and Selinker, page 277)

In other words, you can understand what someone is saying or a reading passage without paying much attention to grammar. In fact, many learners simply have to ignore much of what is heard because, as is explained by input processing models, they can only process so much information – and their priority will always be meaning meaning and key vocabulary rather than grammar. An early learner who hears the sentence:

That black dog has just bitten the young girl on the leg.

might only process:


The learner can probably get the meaning from these content words, but this means a lot of the input can be ignored.

That black dog has just bitten the young girl on the leg.

The noticing hypothesis suggests that only those language items in the input that are noticed will become intake, and will be available for later use as output. Which means that when a learner is asked to speak or write a sentence like the above, what is likely to come out, influenced by other stuff in their interlanguage, is something like:

Dog bited girl in leg.

Input alone is therefore an ineffective way of learning function words and morphemes.

Many early learners can understand a lot of what they hear and read, but find it difficult to produce anything more than basic utterances. In linguistic terms, they aren’t carrying out a syntactic analysis of the language, only a semantic one.

Swain argued that students also needed to produce language, because only this would push the student to produce the correct grammatical structures and extend their use of vocabulary. It would force the learner to put words into order and ‘glue’ the meaning words together with function words – to ‘grammarize’ the sentence.

[Production] may force the learner to move from semantic processing to syntactic processing.

Swain 1985, page 249

Output sees the learner as being:

pushed toward the delivery of a message that is not only conveyed, but that is conveyed precisely, coherently and appropriately.

Swain 1985, page 249

It is possible to comprehend input…without a syntactic analysis of that input. This could explain the phenomenon of individuals who can understand a language and yet can only produce limited utterances in it. They have just never gotten to a syntactic analysis of the language because there has been no demand on them to produce the language. The claim, then, is that producing the target language may be the trigger that forces the learner to pay attention to the means of expression needed in order to successfully convey his or her own intended meaning.

(Swain 1985, p.249)

Before Swain, output was generally regarded as a way of practising language. What Swain was arguing was that output was a creative  part of the language learning process.

Swain argued that learner output has three functions:

1. Noticing

Output raises learner’s awareness of a language item in three ways:

  • they notice the frequency and saliency of the language item
  • they notice a gap between their interlanguage and the target language
  • they notice a hole in their interlanguage: the learner can’t say what they want to say because they don’t have the words. And it’s while attempting to produce output that learners notice these holes. Talk For Writing is an effective way for these holes to be filled: children talk through their stories to other children (and sometimes the teacher or TA), but they will often need help to say what they want to say. This support can come from other children or the teacher or TA.

2. Formulating and Testing Hypotheses

Output gives learners opportunities to test their interlanguage to see what works and what doesn’t, although the effectiveness of this must depend on feedback.

3. Metatalk

Output gives learners opportunities to talk about language.This is often seen in dictogloss activities.

Ouput provides learners with the opportunity to produce language and gain feedback, which, through focusing learners’ attention on certain local aspects of their speech, may lead them to notice either: (a) a mismatch between their speech and that of an interlocutor, or (b) a deficiency in their output.

Swain 1985 page 290

Gass and Selinker describe four functions:

  1. hypothesis testing
  2. feedback
  3. automacity
  4. shift from meaning to grammar (semantic to syntactic processing)

processing language only at the level of meaning will not and cannot serve the purpose of understanding the syntax of the language, a level of knowledge that is essential to the production of language.

(Gass and Selinker page 290)

Semantic processing is fine if all you plan to do with a second language is to read the language. People that learn Russian so that they can read Pushkin in the original, for example, or students studying at universities in non-Anglophone countries who read academic texts in English but write essays in their mother tongue, or doctors who need to read medical journals and papers written in English. But  students who must create output in the second language need to be able to carry out syntactic processing so that their production is accurate in form as well as in meaning.

However, some studies have suggested that while output does lead to the acquisition of vocabulary, its impact on grammar is less clear. Therefore, output needs to be combined with noticing activities such as a focus on form and corrective feedback.


Swain suggested that output could be enhanced by:

  • focused input in problematic areas of grammar and vocabulary
  • more opportunities for production in meaningful contexts
  • systematic and consistent feedback

Learners need to do tasks that encourage output and also focus on language form as well as on meaning: “one task that we feel has been particularly effective in achieving these goals is dictogloss.” (Swain)

Swain carried out some research that used the dictogloss activity in the classroom and she found that learners were engaged in all three of the roles of output mentioned above while attempting to recreate the text. She suggests that the dictogloss is more effective when the task is initially modelled to the children, so that children are trained in how to accomplish the task.

Abstract of this study: Kowal and Swain (1994), Using collaborative language production tasks to promote students’ language awareness. Language Awareness 3(2): 73-93)

This paper presents data of 13 and 14 year old intermediate and advanced learners of French working collaboratively to complete a text reconstruction task. The task was designed to focus the students’ attention and discussion on the form of the message they were constructing. It was hypothesised that this kind of opportunity to produce language would promote their language learning by (1) making them aware of gaps in their existing knowledge which they would subsequently seek to fill; (2) raising their awareness of the links between the form, function and meaning of words as they worked to construct their intended message; and (3) obtaining feedback that they would receive from their peers and their teacher as they completed the task. The results support the hypothesis and also provide rich insights for teachers, researchers and curriculum planners into the language learning process in a collaborative setting; the students’ understandings of how language ‘works’; and the effects of certain grouping patterns on the ensuing student talk.

[My Note: this would also be a good way of ensuring the strategy was adopted in the classroom: the 2nd teacher could be the headteacher or the EAL Coordinator doing team teaching).

My Note: but all this depends on getting feedback on what they are producing.

Merrill Swain (1988), Manipulating and complementing content teaching to maximize second language learning, TESL Canada Journal 6(1):68-83

Merrill Swain (1995), Three functions of output in second language learning. In Cook, G. and Seidlhofer, B., Principle and Practice in Applied Linguistics. Oxford University Press

Merrill Swain, Integrating Language and Content in Immersion Classrooms: Research Perspectives (page 85 of document)


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s