I'm stuck at B1: move your listening up one notch, and speak as much as you listen

The problem

You passed B1 eight months ago. Since then you have listened to podcasts, done exercises, watched series with subtitles. And yet, on the last mock test, the same score, give or take a point. You understand the gist of what people say to you, but as soon as the topic leaves everyday life you lose the thread, and when you speak, your sentences are the same ones you used a year ago. You wonder whether you have hit your ceiling. You do not have a ceiling: you have a setting to change.

What the research found

Karen Lichtman and Bill VanPatten (Lichtman & VanPatten, 2021) went back, forty years on, to Stephen Krashen’s hypotheses about language acquisition. Their conclusion: Krashen’s formal framework has been set aside, but several of his ideas survive under other names in current research. The most robust is comprehensible input: you acquire a language first by understanding messages you hear and read, and that input should sit just above what you already know. Krashen’s famous “i + 1” was never measurable as such; the authors note that vocabulary research has given it a concrete form.

That form comes from Paul Nation (Nation, 2007). For a text or a recording to be useful for learning, 95 to 98% of its words should already be known, in other words two to five new words per hundred. Below that, you are decoding rather than understanding; above it, you no longer meet anything new. Nation also proposes a balance for a whole programme, which he calls the four strands: meaning-focused input (listening, reading), meaning-focused output (speaking, writing), language-focused learning (grammar, vocabulary, pronunciation) and fluency development (doing easy things, fast). He recommends giving roughly equal time to each.

The mechanism in plain words: a B1 plateau often comes from two imbalances. First, the input is badly calibrated. Easy content is reassuring but brings nothing new; “authentic” content at 85% known words exhausts you without feeding you. Second, output is almost absent. Many learners listen for hours and speak for a few minutes a week, whereas Nation puts the two on an equal footing. Lichtman and VanPatten make the same point: comprehension comes before production, but production emerges when you exercise it in exchanges, not when you wait for it.

Do this today

Take a short article or recording and count, in a hundred words, the ones you do not know. If there are two to five, that is your notch up. If there are more, look for something simpler; if there are none, look for something richer. Then, for every minute you listen, speak for a minute: summarise aloud what you have just heard, recording on your phone. Five minutes of listening, five minutes of speaking.

How our lessons use it

On the HumanFrench YouTube channel, the video « L’ascenseur » tells a short story in everyday words: at B1 you should know almost all of them, which makes it a good fluency exercise. Its shadowing version pauses so you can repeat each sentence: there, you produce as much as you listen. The reading activity « Je m’appelle Nadia » (A1) is below your level, but its last step, “Dis-le à ta façon”, is the move you need to multiply. Start with « L’ascenseur ».

The honest limit

Lichtman and VanPatten offer a theoretical reading, not a new experiment, and Nation’s 95 to 98% threshold comes from research on reading and vocabulary in English, not on the TCF or the TEF. The four strands are a teaching recommendation, not the result of a controlled trial. These markers help you tune your practice; they do not say how long it will take to reach B2.

Your questions