SentenceFlow

The Backstory

Every Korean English learner knows the pain: you memorize a sentence, but when you need it in real life, your mouth freezes. Reading comprehension is fine — but production fails. I built SentenceFlow to attack this specific gap using Progressive Chunking Recall, spaced repetition, and a daily unlock system that keeps learners coming back every morning.

Key Features

• 135 Curated Sentences — Organized across 5 CEFR-aligned levels (Foundation A1 to Mastering C2) and 8 real-life categories
• Daily 7AM Unlock — 7 new sentences unlock every morning at 7:00 AM KST, with countdown timer and notification badge
• 5 Levels — Foundation, Building, Expanding, Refining, Mastering — mapped to CEFR A1 through C2
• 8 Categories — Office survival, Travel, Daily life, Emotions, Job interview, Parenting, SNS/Trends, Nuance
• Progressive Chunking Recall — 4-step learning: full sentence exposure, chunk decomposition, fill-in-the-blank, complete recall from memory
• SM-2 Spaced Repetition — Sentences you struggle with appear more often; mastered ones fade to monthly reviews
• 3-Level Hint System — First letter, chunk meanings, full answer. No penalty for hints — just more practice
• Memory Strength Dashboard — Color-coded sentences: strong (green), shaky (yellow), at-risk (red), with level progress bars
• Level and Category Filters — Browse and filter all unlocked sentences by difficulty level and topic category

Technical Details

Single HTML file architecture (84KB) with React 18 via CDN and Tailwind CSS. The SM-2 spaced repetition algorithm manages review scheduling with ease factor adjustment. Daily unlock logic calculates KST timezone offset to release 7 new sentences at 7:00 AM each morning — all client-side with no backend required. All progress data persists in localStorage. Web Speech API provides native TTS for pronunciation. The learning engine implements adaptive difficulty based on performance.