Read our latest articles and guides.

Say "I want a coffee" out loud. Under a second, one unit, no assembly. Miller, Wray, Ellis and Pawley all point the same way: the phrase is the unit of speech, not the word.

Five years is roughly 400 hours of exposure and fewer than 20 of actual speaking, against the thousands a Knowing Brain needs. It would be astonishing if you could speak.

Every language product asks the same thing first: open the app. That single requirement is where most language learning quietly dies.

Involuntary musical imagery is well documented and almost never discussed in language learning. When a song loops in your head, motor regions fire as though you were singing it.

You've told them to stick at difficult things. They've heard you say it and never watched you do it — and a language is one of the few things you can be seen to learn.

Sixty minutes, thirty pounds, and about fifteen minutes of actual production — most of it the safe sentences you already knew. A tutor is worth it later, not first.

Four core chunks, taken verse by verse, multiplying into thirty or more usable phrases by the time the song ends. Here is the whole thing taken apart.

Flip. Red. Wrong. By the tenth wrong answer you're not learning any more, you're defending. A song has no failure state — you sing what you can and catch more tomorrow.

Five stages from zero to fluent speaker, what your brain is doing at each, and how long each takes. Most readers of this are stuck in Stage 2, and more input won't move them.

Apple said the best camera is the one you have with you. Two to three hours a day where your ears are free — and almost no language method can use a minute of it.