Full-Article Learning Mode

It's no longer just "learn languages from videos." LangKey's Full-Article Learning Mode lets you open any English article on BBC, NYT, Medium — enter learning state in one click: original text + AI translation, saved vocabulary auto-highlighted, switch any practice mode from the top.

Entry Point

On article-type pages (with article / main / long-text detection), LangKey auto-injects a "📖 Learn this article" floating button at the bottom-right. Click and:

1. Extract main content (Readability algorithm)
2. Smart cleaning (remove nav / ads / footnotes / citations)
3. Split into sentences (SubtitleItem)
4. Open the learning panel in browser Side Panel

Carrier is Chrome Side Panel (Chrome 114+), original and learning panel side-by-side, no tab switching.

Learning Panel Layout

┌────────────────────────────────────────────┐
│  LangKey · Article Learning                │
│  ─────────────────────────────             │
│  Source: BBC · 2026-08-10 · 3 min read      │
│                                             │
│  [ Read ] [ Input ] [ Pick ] [ Cloze ] [ Listen ] [ Shadow ]  │
│                                             │
│  ── Original ──────────── Translation ──     │
│  The pandemic has...   The pandemic has...   │
│  In a recent study...  In a recent study...  │
│  [vocab: pandemic highlighted]               │
│                                             │
│  ── Vocabulary (auto-synced) ──             │
│  pandemic (B2) | study (A2) | recent (B1)   │
└────────────────────────────────────────────┘

7 Practice Modes Reused

The 5 "practice modes" work identically on article content as on video:

• Input Mode: Word-by-word input of article sentences
• Option Mode: Sentence ordering
• Fill-in Mode: AI cloze
• Listening Mode: TTS reads the article
• Shadowing: Read after article sentences

New additions:

• Dictation Mode: AI generates dictation at your level (v2.4.0)
• AI Conversation: Scenario-based free conversation (v2.5 roadmap)

Vocabulary Auto-Highlight

• Words already in your VocabularyDB are auto-highlighted in the original (colored by CEFR level: A1 gray / A2 yellow / B1 orange / B2 green / C1 blue)
• Hover to see translation
• Click for full entry (CEFR level / etymology / synonyms / review progress)

Site Compatibility

Tested working:

• BBC News
• The New York Times
• The Guardian
• Medium
• Substack
• Personal blogs (WordPress / Hexo / Hugo)

Pages that can't be extracted (minimal SPA / strict anti-scraping) will show "Cannot identify article structure."

Data Storage

Extracted article content is stored in chrome.storage.local, auto-cleared after 24 hours. All wrong words / vocabulary persist in VocabularyDB (local IndexedDB), shared with video learning records.

Privacy

AI translation requests only send the paragraph currently being read (not the whole article). All other data is stored locally, not uploaded to any server.