This is actually a really solid observation, Cao Thai — and honestly, something I’ve thought about a lot too.
The problem with how most of us learned English is that we studied the language but never studied speaking. You pass 12th class with 90+ in English. You write essays. You answer comprehension questions. But the moment someone asks you something in a meeting, your brain freezes — because nobody ever taught you the chunks.
And that’s exactly what native speakers use: chunks. Not individual words assembled in real-time. Pre-built phrases that come out automatically. “I was just about to…” / “Does that make sense?” / “Let me get back to you on that.” These are daily conversation sentences for spoken English — and they’re learned through repetition, not grammar rules.
Your idea of highlighting + saving to a doc + exporting to Anki is honestly the right workflow. Anki is underused for this. Most people use it for vocab — single words. But if you feed it full sentence chunks from real conversations, real articles, real dialogue — that’s when it becomes powerful. Your brain starts storing whole patterns, not just meanings.
Here’s what I’d add though: the source matters a lot. The sentences you highlight should come from natural spoken contexts — not textbook English. Think web series subtitles, podcast transcripts, Reddit threads. Even your 12th class English question answers have certain formal sentence patterns (“The author implies that…”, “This suggests that…”) — which are useful for written English, but a completely different register from what you’d actually say out loud.
So maybe two separate Anki decks: one for formal written patterns (useful for exams, emails, essays), one for spoken chunks (useful for actual conversations, interviews, daily use). Different muscle, different practice.
The feature request itself is simple and specific enough that it’s genuinely useful — not just a “nice to have.” Would love to see this built. +1 from me.