The Science of Language Learning: What Actually Works

Language learning is one of the most studied fields in cognitive science. Researchers have been investigating how humans acquire second languages for decades. Yet the most popular language learning products largely ignore what that research says.

This post is a summary of what the science actually shows — and why it should change how you approach learning a language.

The Two Types of Language Knowledge

First, a crucial distinction that most people miss: there's a difference between knowing about a language and knowing a language.

Explicit knowledge is declarative — the grammar rules you can articulate, the vocabulary you can recall when asked. This is what traditional study primarily builds.

Implicit knowledge is procedural — the intuitive feel for what sounds right, the ability to produce a sentence without consciously thinking about its construction. This is what native speakers have. This is what enables fluent speech.

The problem: explicit knowledge doesn't automatically become implicit knowledge. You can memorize all 14 Spanish verb tenses without being able to use them in real-time conversation. The conversion from explicit to implicit requires extensive practice in the form of actual communication.

Krashen's Input Hypothesis

Linguist Stephen Krashen proposed one of the most influential theories in second language acquisition: we acquire language primarily through comprehensible input — messages we understand that are slightly beyond our current level (i+1).

In this model, understanding slightly-beyond-current input is what causes acquisition. Explicit grammar study can raise awareness, but doesn't directly cause acquisition. Only comprehensible input does.

The practical implication: you need to hear and read a lot of the language at the right level. Not too easy (nothing to learn), not too hard (can't comprehend). The zone just beyond your current ability.

Swain's Output Hypothesis

Merrill Swain observed that even learners with extensive comprehensible input exposure often plateaued without reaching full fluency. Her research in French immersion programs showed that students could understand French well but struggled to produce it accurately.

Her conclusion: producing language is not just a way to demonstrate acquisition — it may itself be a mechanism of acquisition. When you try to speak and realize you can't express something, it draws your attention to gaps in your knowledge and drives you to fill them.

The implication: passive exposure (reading, listening) isn't enough. You need to produce the language in meaningful communication to develop full fluency.

Noticing and Corrective Feedback

Schmidt's "noticing hypothesis" proposes that acquisition requires conscious attention to input features. In practice, this means error correction can accelerate learning — but the type of correction matters enormously.

Research consistently shows:

• Explicit metalinguistic correction (stopping the conversation to explain the grammar rule) disrupts communication and often doesn't stick
• Recasting (incorporating the corrected form naturally in the tutor's response) is highly effective and maintains conversational flow
• Repetition with correction (repeating what the learner said correctly) is also effective

The lesson: good language tutors correct errors naturally, in the flow of conversation, not through grammar lectures.

Spaced Repetition

Hermann Ebbinghaus's research on the "forgetting curve" established that memory decays in a predictable pattern, and that reviewing material at increasing intervals — spaced repetition — dramatically improves long-term retention.

Spaced repetition is most commonly applied to vocabulary flashcards. It works there. But the most effective form may be using vocabulary in real conversation at optimal intervals — which is exactly what happens naturally in extended language practice. When you've used a word in conversation multiple times over weeks, it tends to stick in a way that flashcard review never quite achieves.

The Role of Motivation and Anxiety

Language learning research consistently identifies two psychological factors as major determinants of success:

Integrative motivation (wanting to connect with speakers of the language, genuinely caring about the culture) predicts better outcomes than instrumental motivation (learning for a specific practical goal like a job requirement).

Language anxiety — fear of making mistakes in front of others — is one of the most significant barriers to progress. Learners with high anxiety speak less, take fewer risks, and progress more slowly. Learners with low anxiety attempt more, make more productive errors, and learn faster.

The practical implication: conditions that reduce anxiety (low-stakes practice, patient feedback, safe environment for errors) accelerate learning significantly.

What This All Means for How You Should Practice

Putting the research together, the most effective language learning approach:

1. Maximizes comprehensible input at i+1 — lots of language just above your level
2. Requires meaningful output — real communication, not just exercises
3. Provides natural corrective feedback — recasting, not grammar lectures
4. Uses spaced repetition naturally — through conversation, not just flashcards
5. Reduces anxiety — creates conditions where you're willing to take risks
6. Maintains motivation — connects learning to real communication goals

This is what shaped Festivities. Real conversations. Adaptive difficulty. Natural error correction. Low-stakes practice. And always, the goal of actual communication rather than exercise completion.

See how Festivities applies this research →