The Science of Learning
Why most language apps fail, and the 5 peer-reviewed cognitive science pillars Clamo is built on.
“Five pillars of cognitive science, 21 personalized parameters, zero dark patterns.”
The science of memory and language learning has advanced dramatically since 2000. Yet most mainstream language learning apps are engineered around slot-machine engagement mechanics (hearts, streaks, locked levels) rather than long-term neurological retention. Clamo takes the opposite approach: every feature is anchored in peer-reviewed cognitive science.
The Five Pillars Clamo Rests On
1. FSRS Spaced Repetition
Stochastic shortest-path memory optimization
Unlike legacy 1980s SM-2 algorithms, the Free Spaced Repetition Scheduler (FSRS) uses 21 trainable parameters to model individual memory stability, difficulty, and retrievability curves. Research validates significant efficiency gains over fixed-interval schedulers.
2. The 85% Optimal Difficulty Rule
The mathematical sweet spot of learning
Research published in Nature Communications proved that learning rate is mathematically maximized when error rate is ~15% (85% accuracy). Training that is too easy creates an illusion of competence; training that is too hard triggers cognitive fatigue.
3. High Variability Phonetic Training (HVPT)
How varied speakers help learners hear tone categories
High Variability Phonetic Training (HVPT) research uses multiple speakers to help learners generalize tone categories beyond one voice. Clamo's complete multi-voice audio library is still being completed.
4. Retrieval Practice & The Testing Effect
Active generation over passive recognition
Testing memory produces 50–100% higher long-term retention compared to passive re-reading. Clamo prioritizes active production (type-hanzi, type-pinyin, tone synthesis) over low-effort multiple choice recognition.
5. Desirable Difficulties & Interleaving
Contextual interference for robust discrimination
Practicing confusable items together (such as 他 and 她, or tone minimal pairs) creates contextual interference. While it feels harder during practice, it dramatically reduces long-term confusion and cross-character interference.
Clamo vs Traditional Language Apps
| Dimension | Traditional Apps | Clamo |
|---|---|---|
| Scheduling | Opaque, engagement-driven streaks | Mathematically open FSRS spaced repetition |
| Tone Training | Incidental or single-speaker | Multi-voice tone audio coverage in progress |
| Difficulty | Fixed linear path or penalty gates | Dynamic 85% optimal accuracy governor |
| Character Learning | Isolated word flashcards | Radical decomposition, component trees & stroke order |
| Ethics | Hearts, countdown timers, dark patterns | Zero dark patterns, transparent methodology, open export |
Experience Evidence-Based Chinese Learning
Try our interactive tone drills and spaced repetition flashcards without creating an account.