Skip to main content

When to Start Speaking Japanese: The Output Debate, Settled Practically

The question of when to start speaking Japanese often splits learners into two camps. One delays speech until it emerges from heavy listening. The other pushes learners to produce language from the start. You do not need to wait for a green light to begin, but what you say should stay within what you have already heard.

Overview​

Both camps trace back to mainstream theories of second-language acquisition, and both rest on real findings. The input-first position draws on Krashen's claim that language is acquired by understanding messages.1 The output-first position draws on Swain's finding that understanding alone leaves a gap in what learners can produce.2

This article states each case fairly, separates the observed research from the advice built on top of it, and gives a concrete, low-pressure way to begin. The short answer is that early, low-stakes output is safe and useful as long as its volume tracks your input.

The two camps, in plain terms​

The debate is usually framed as a fight, but the two sides agree on more than they dispute. Both treat comprehensible input, language you can understand, as essential. They disagree on whether learners should also produce language early, or wait for production to surface on its own.

The input-first / silent-period case​

Krashen's Input Hypothesis holds that people acquire language in one way: by understanding messages, that is, by receiving comprehensible input containing structures slightly beyond their current level.1 In this model, speaking fluency is not taught directly. It emerges on its own once the learner has built competence through input, so production is a result of acquisition rather than a cause of it.1

Krashen also describes a silent period: a span during which a learner builds competence by listening before producing original speech.1 He documents it most clearly in children acquiring a second language naturally, and argues that forcing early production can raise the affective filter and impede acquisition.1

The affective filter is the part of this model that explains why pressure can backfire. Krashen's Affective Filter Hypothesis says anxiety, low self-confidence, and low motivation act as a filter. That filter reduces how much input is actually available for acquisition.1 One reason to allow a silent period is to keep the filter low by not demanding speech before the learner is ready.1

The strong-delay position is a community extension, not Krashen's rule

The popular advice to "delay all output for months until it emerges naturally" circulates in immersion-learning communities. It is a learner-community extension of Krashen's framework. Krashen describes the silent period and the affective filter as mechanisms. He does not prescribe months of enforced silence for adult self-learners in those terms.1

The output-first / early-speaking case​

Swain's Output Hypothesis came out of Canadian French immersion studies. Learners there received years of rich comprehensible input and reached strong comprehension. Yet they still fell short of native-like accuracy in production.2 Swain read this as evidence that comprehensible input is necessary but not sufficient, and that producing language plays its own role in acquisition.2

Swain identifies three functions of output. The noticing function makes learners see gaps between what they want to say and what they can say. The hypothesis-testing function lets them try a form and adjust based on feedback. The metalinguistic function strengthens knowledge as they use language to reflect on language.3

These functions operate under "pushed output," meaning output produced with pressure to be precise, coherent, and appropriate.3 Experimental work on pushed output reports gains in noticing and in subsequent learning of target forms relative to comprehension-only conditions.34

Long's Interaction Hypothesis adds a further argument for producing. Two-way interaction, meaning conversation where people negotiate meaning, modifies input to make it comprehensible and supplies the feedback that drives development.5 That mechanism requires the learner to produce, not only to receive.5

What the research actually says​

Stripped of ideology, the literature supports a narrower set of claims than either camp's loudest advocates assert. The silent period is real, but it is not a universal mandate. Input builds a base that output then converts into the ability to speak.

The silent period: real phenomenon, contested prescription​

The silent period is an observed phenomenon in some learners, most strongly documented in children in naturalistic immersion.678 Research disputes its status as a required, universal stage, especially as advice telling adult self-learners to stay silent.678

Saville-Troike's study of young second-language learners found that much of what looks like silence is actually active private speech: quiet self-directed rehearsal and language-learning strategy use.6 Silence does not necessarily mean the learner is idle, and she argues the silent period does not apply uniformly to all learners.6

Gibbons's examination of the silent period draws a sharper line. It distinguishes a genuine pre-production stage from prolonged silence caused by incomprehension or social withdrawal, rather than by a required processing stage.7 The caution is against treating extended silence as automatically beneficial.7

Introductory SLA (second-language acquisition) reference texts present the silent period the same way: as characteristic of an early stage for some learners, especially children, rather than as an obligatory phase every learner must pass through before speaking.89

Input is necessary but not automatically sufficient for output​

Comprehension of meaningful input is the foundation of acquisition. But considerable research and classroom experience challenge the stronger claim that input alone is enough for full acquisition, especially for accurate production. A focus on form and the chance to produce also matter.9

Swain's original argument is exactly this gap. Immersion learners with abundant input still under-performed in production. That is the empirical basis for treating output as a distinct contributor rather than a by-product of enough input.2

The same gap shows up in vocabulary as the receptive-productive asymmetry. A learner's receptive vocabulary, the words they can recognize, is consistently larger than their productive vocabulary, the words they can recall and use in speech or writing.1011 Words you can understand are not automatically words you can produce.10

Productive vocabulary knowledge typically follows and extends receptive knowledge, and develops more slowly.11 Recognition alone does not build the retrieval that production needs. Producing words is what tends to push receptive knowledge into productive control.11

When to start, in practice​

The practical answer falls out of the research above. Begin producing early. Expect production to lag comprehension, and keep what you say close to what you have heard. The rest of this section turns that into a routine.

Output from week one, with calibrated expectations​

The output-hypothesis mechanism works only when the learner actually produces. Noticing gaps, testing hypotheses, and getting feedback cannot happen during silence, which is the argument for beginning some output early rather than postponing it indefinitely.34

Calibrated expectations are the other half of the rule. Production often lags comprehension by a wide margin. That expected, well-documented pattern is the receptive-productive asymmetry, not a sign of failure.1011 A learner who understands far more than they can say is showing the normal shape of second-language development.10

Treat the comprehension-production gap as a feature, not a deficit

Plan for your speaking to trail your listening and reading for a long time. Measuring early output against your comprehension level guarantees disappointment; measure it against what you said last month instead.

Match what you say to what you have heard​

This is the central rule of the whole article. Keep early output inside structures you have heard enough times to feel familiar. Do not reach for forms you have only read about once.

The worry behind delaying output is fossilization, and it has a real kernel. Interlanguage, the learner's own developing rule system, can stabilize around a non-target form. Fossilized errors can persist even under continued exposure.12

Silence is not the fix for that risk. The output-hypothesis functions depend on output being pushed toward, and corrected toward, a target. What keeps output from rehearsing errors is noticing the gap and receiving feedback, not abstaining.35 Long's interaction work indicates that the feedback available in meaningful interaction is part of how production gets shaped toward the target.5

Keeping early output close to attested input, language you have actually heard or read, is what reconciles the two camps. It stays near structures you can have corrected, so it captures the output benefit while keeping fossilization risk low.

A low-pressure first-output ladder​

Early output does not have to mean conversation. The rungs below increase in social pressure. A learner can raise the stakes gradually rather than all at once, which fits the affective-filter logic of keeping anxiety low.113

Shadowing, listening while simultaneously reproducing speech, is the gentlest rung. The learner reproduces a model rather than generating original sentences. That exercises articulation without the gap-noticing pressure of free output. The deeper case for shadowing before conversation is that it builds the motor patterns speech depends on before any social stakes are involved.

A self-introduction (自己紹介, jikoshōkai) is the next step up. It is a conventional first-output anchor because it is highly formulaic. A standard sequence of はじめまして, your name, and よろしくお願いします forms a complete, socially appropriate introduction. You produce real, correct Japanese without inventing new structure.14

初はじめまして。よろしくお願おねがいします。14
"Nice to meet you. I look forward to working with you."

はじめまして belongs to first meetings only

はじめまして is used only when meeting someone for the first time, never as a general greeting. よろしくお願いします is a near-obligatory humble closing formula in an introduction. The polite ます-form shown here is the safe default for a beginner. The plainer よろしく on its own is casual.14

From the recorded monologue rung onward, you are generating original speech, first to yourself, then to a person who can give feedback. Each rung adds a little more interactional pressure than the one before it.

Good to know​

"Waiting until you're ready" usually means waiting forever​

Production trails comprehension as a structural feature of second-language development. A learner who waits to feel ready is waiting for confidence that the receptive-productive gap predicts will arrive late, if at all.1011 Producing closes the gap by building the retrieval that creates the feeling of readiness.3 Readiness is a result of early output, not a prerequisite for it.

Perfectionism and output anxiety are the real bottleneck​

Foreign-language anxiety is a distinct, measurable construct, not just generic shyness. Horwitz, Horwitz, and Cope defined foreign-language classroom anxiety as a specific mix of self-perceptions, beliefs, feelings, and behaviors. It is built on communication apprehension, test anxiety, and fear of negative evaluation, and is distinguishable from other anxieties.13

This anxiety carries a double cost. In Krashen's affective-filter terms, anxiety and low self-confidence raise the filter and reduce the input usable for acquisition. Output anxiety can therefore both block speaking and dampen the intake you get from listening and reading.1 The practical move is to treat each error as data about what to fix next, not as damage to undo.

Why this is not a "speak from day one" absolutist pitch either​

Output's benefits, noticing, hypothesis-testing, and feedback, all require something to produce. That material comes from input. A learner with near-zero input has near-zero attested structure to output. Output volume should therefore track input volume rather than run ahead of it.23

Comprehensible input remains the foundation in every model discussed here, including Swain's. The output hypothesis was offered as a complement to input, not a replacement. An input-free "just talk" approach has no support in this literature.29

The fossilization worry, right-sized​

The worry as commonly stated is that speaking before you are ready locks in bad habits, so you should stay silent. The flaw is the assumption that not producing prevents fossilization. Fossilization is about a system stabilizing without corrective pressure. Silence removes that corrective pressure rather than supplying it.12

The literature's mechanism for moving interlanguage toward the target is noticing gaps and getting feedback through interaction. Both require producing and correcting output.35 Manage the risk by keeping early output close to structures you have heard and getting it corrected, not by abstaining from speech.1235

See also​

References​

Footnotes​

  1. Krashen, Stephen D. Principles and Practice in Second Language Acquisition. Pergamon Press, 1982. Full text: https://www.sdkrashen.com/content/books/principles_and_practice.pdf ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8 ↩9 ↩10

  2. Swain, Merrill. "Communicative Competence: Some Roles of Comprehensible Input and Comprehensible Output in Its Development." In Gass, S. and Madden, C. (eds.), Input in Second Language Acquisition. Newbury House, 1985, pp. 235–253. ↩ ↩2 ↩3 ↩4 ↩5 ↩6

  3. Swain, Merrill. "Three Functions of Output in Second Language Learning." In Cook, G. and Seidlhofer, B. (eds.), Principle and Practice in Applied Linguistics. Oxford University Press, 1995, pp. 125–144. ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8 ↩9

  4. Russell, Victoria. "A Closer Look at the Output Hypothesis: The Effect of Pushed Output on Noticing and Inductive Learning of the Spanish Future Tense." Foreign Language Annals, vol. 47, no. 1, 2014, pp. 25–47. https://doi.org/10.1111/flan.12077 ↩ ↩2

  5. Long, Michael H. "Native Speaker/Non-Native Speaker Conversation and the Negotiation of Comprehensible Input." Applied Linguistics, vol. 4, no. 2, 1983, pp. 126–141. Oxford University Press. ↩ ↩2 ↩3 ↩4 ↩5 ↩6

  6. Saville-Troike, Muriel. "Private Speech: Evidence for Second Language Learning Strategies during the 'Silent' Period." Journal of Child Language, vol. 15, no. 3, 1988, pp. 567–590. Cambridge University Press. ↩ ↩2 ↩3 ↩4

  7. Gibbons, John. "The Silent Period: An Examination." Language Learning, vol. 35, no. 2, 1985, pp. 255–267. https://doi.org/10.1111/j.1467-1770.1985.tb01027.x ↩ ↩2 ↩3 ↩4

  8. Saville-Troike, Muriel. Introducing Second Language Acquisition, 2nd ed. Cambridge University Press, 2012. ↩ ↩2 ↩3

  9. Lightbown, Patsy M. and Spada, Nina. How Languages Are Learned, 4th ed. Oxford University Press, 2013. ↩ ↩2 ↩3

  10. Webb, Stuart. "Receptive and Productive Vocabulary Sizes of L2 Learners." Studies in Second Language Acquisition, vol. 30, no. 1, 2008, pp. 79–95. Cambridge University Press. ↩ ↩2 ↩3 ↩4 ↩5

  11. Nation, I. S. P. Learning Vocabulary in Another Language, 2nd ed. Cambridge University Press, 2013. ↩ ↩2 ↩3 ↩4 ↩5

  12. Selinker, Larry. "Interlanguage." International Review of Applied Linguistics in Language Teaching (IRAL), vol. 10, no. 1–4, 1972, pp. 209–231. ↩ ↩2 ↩3

  13. Horwitz, Elaine K., Horwitz, Michael B., and Cope, Joann. "Foreign Language Classroom Anxiety." The Modern Language Journal, vol. 70, no. 2, 1986, pp. 125–132. https://doi.org/10.1111/j.1540-4781.1986.tb05256.x ↩ ↩2

  14. Cotoacademy. "Jikoshoukai: How to Introduce Yourself in Japanese." https://cotoacademy.com/jikoshoukai-self-introduction-in-japanese/ ↩ ↩2 ↩3