What is comprehensible input (and how to actually get it)
“Comprehensible input” is the idea — popularized by linguist Stephen Krashen — that you acquire a language mainly by understanding messages slightly above your current level. Not by memorizing rules, but by getting lots of real language you can almost follow.
The phrase often used is “i+1”: input (i) plus one small step beyond what you already know. Enough to stretch you, not so much that you’re lost.
Why it works
When you understand a message, your brain quietly absorbs the patterns that carried it — word order, collocations, how sounds blend. You pick up grammar implicitly, the way you did your first language, instead of consciously applying rules mid-conversation.
That’s why learners who consume hours of understandable content often develop a better ear and more natural phrasing than those who only drill exercises.
The hard part: making real input comprehensible
Authentic native content is the best input — but at full speed it’s often not comprehensible, which is the whole problem. The trick is to add just enough support to push a clip from “lost” to “i+1”: a transcript, instant definitions, and the ability to replay a phrase until it lands.
You can build that support yourself around any video, or use a tool that has it ready. Sygmatic, a sister app from the same maker, does exactly this with real YouTube clips — a synced transcript, tap-to-define and a short quiz — and is free once you sign in.
How to use comprehensible input every day
Pick content slightly above your level. Watch it once for the gist, then again with the transcript, tapping unknown words. Re-listen to the parts you missed. Keep sessions short and frequent — consistency beats marathon study.
Finish with a quick check (a quiz or a quick recap in your own words) so the new vocabulary moves from “recognized” to “remembered.”
Where output comes in
Input builds the store of language you can draw on; speaking is where you find out what you can actually retrieve under time pressure. The two feel very different. Plenty of learners who understand a great deal still stall when they have to answer, because they have never practised pulling the words out fast.
That is the half banter is built for. You hold a short spoken conversation with an AI character in an everyday scene — ordering dinner, checking into a hotel, making weekend plans — and at the end your own lines are quoted back with corrections. It works best on top of input, not instead of it: the phrases you have heard most are the ones that come out when you need them.