Interaction Hypothesis

Michael Long's Interaction Hypothesis (1981, revised 1996) accepts Krashen's claim that comprehensible input is necessary but rejects the claim that bare input is enough. Acquisition is best supported when input becomes comprehensible through interaction, specifically through the negotiation of meaning that occurs when communication breaks down and interlocutors work to repair it.

The Original Formulation (1981)

Long observed that conversations between native speakers and non-native speakers contain systematic adjustments not found in native-native talk: confirmation checks, clarification requests, comprehension checks, repetitions, and reformulations. These interactional modifications make input comprehensible to this learner, at this moment, on this form. The hypothesis reframed input from a static corpus into a dynamically tailored stream.

The 1996 Revision

The updated version integrated insights from the Noticing Hypothesis and Output Hypothesis. Long now argued that interaction supports acquisition because it brings together three conditions in a single discourse moment:

Element	Role in acquisition
Modified input	Interlocutor adjusts speech to become comprehensible (slowing, paraphrasing, exemplifying)
Selective attention	Negotiation directs the learner's attention to the form that caused the breakdown (noticing the gap)
Output / pushed production	Repair turns require the learner to reformulate, often more accurately (pushed output)

Acquisition is most likely when these three converge: the learner notices a form, has it modeled comprehensibly, and produces a more target-like version, all within one conversational sequence.

Negotiation of Meaning

The mechanism at the heart of the hypothesis. A typical negotiation sequence:

1. Speaker A produces an utterance.
2. Speaker B signals incomprehension ("Sorry?", "What do you mean?").
3. Speaker A reformulates: slower, simpler, with synonyms or examples.
4. Comprehension is achieved; the conversation continues.

The reformulation provides modified input precisely calibrated to the learner's processing limit. If B is the learner, step 2 is also pushed output: B has to articulate what they didn't understand. Recasts (Speaker A repeating B's incorrect utterance correctly) deliver implicit corrective feedback without disrupting the flow of meaning.

Interaction vs Input vs Output

Hypothesis	Core claim	What it gives the learner
Input Hypothesis	Comprehensible input at i+1 is sufficient	Exposure
Output Hypothesis	Pushed production drives syntactic processing	Practice + noticing
Interaction Hypothesis	Negotiation supplies modified input and pushed output and attention to form simultaneously	All three, locked together by communicative pressure

Long's framing is integrative rather than competitive: interaction is the discourse environment in which input and output become acquisition-relevant.

Empirical Base

A large body of task-based research supports interaction effects. Studies (Mackey 1999; Mackey & Goo 2007 meta-analysis; Pica 1994) consistently find that learners exposed to interactional feedback acquire targeted forms more reliably than those receiving only premodified input. Recasts and clarification requests have particularly strong evidence for short-term gains, with effects on longer-term retention varying by feature type and learner readiness.

Classroom Implications

• Design tasks with information gaps that require negotiation to complete (jigsaws, opinion gaps, decision-making tasks).
• Pair learners at slightly different levels: asymmetry triggers genuine negotiation.
• Train teachers and stronger peers to recast rather than overtly correct, preserving communicative flow.
• Build in time for post-task reformulation so learners can revisit forms surfaced during negotiation.
• Avoid lockstep IRF (initiation-response-feedback) sequences that suppress learner-initiated repair.

Criticisms

• Recast ambiguity. Learners may interpret recasts as content responses rather than corrections, missing the corrective intent.
• Form-meaning trade-off. Heavy negotiation can disrupt fluency and reduce engagement.
• Cultural variation. Negotiation norms differ across cultures; what counts as a clarification request in one classroom may be face-threatening in another.
• Limited evidence on complex grammar. Interaction effects are clearest for vocabulary and discrete morphology; impact on complex syntax is less established.
• Sociocultural critique. Treats interaction primarily as a delivery mechanism for input/output rather than as constitutive of learning itself (sociocultural perspectives frame interaction as the site where cognition is co-constructed).

References

• Long, M. (1981). Input, interaction, and second language acquisition. Annals of the New York Academy of Sciences, 379, 259–278.
• Long, M. (1996). The role of the linguistic environment in second language acquisition. In W. Ritchie & T. Bhatia (Eds.), Handbook of Second Language Acquisition. Academic Press.
• Mackey, A. (1999). Input, interaction, and second language development. Studies in Second Language Acquisition, 21(4), 557–587.
• Mackey, A. & Goo, J. (2007). Interaction research in SLA: A meta-analysis and research synthesis. In A. Mackey (Ed.), Conversational Interaction in Second Language Acquisition. Oxford University Press.
• Pica, T. (1994). Research on negotiation: What does it reveal about second-language learning conditions, processes, and outcomes? Language Learning, 44(3), 493–527.