--- title: "Proč se zasekáš na B1 (a 30denní plán, jak z toho ven)" description: "Kartičky tě dostanou na střední úroveň. K plynulosti ne. Tady je to, co opravdu funguje při přechodu B2/C1." canonical: https://talktodia.com/cs/blog/breaking-the-intermediate-plateau language: cs published: 2026-05-13 updated: 2026-06-09 author: Bhada Yun (Founder, TalkToDia) license: see https://talktodia.com/.well-known/ai-policy.txt --- # Proč se zasekáš na B1 (a 30denní plán, jak z toho ven) Kartičky tě dostanou na střední úroveň. K plynulosti ne. Tady je to, co opravdu funguje při přechodu B2/C1. You can survive a holiday in the country. You can read menus, follow maybe 60% of a Netflix show, and order food without pointing. But you can't hold a 30-minute conversation about anything that matters, and you've been recycling the same 800 words for a year. That's the intermediate plateau, and it's where most people quit — Richards (2008) describes it as the predictable attrition wall between B1 and B2. This article is the playbook we wish someone had handed us at B1: why the plateau happens, the five specific skills that get you out, and a 30-day plan you can run with any conversation partner — human or AI. ## Why do you plateau at B1? You plateau because the methods that got you to B1 are optimized for a problem you no longer have. Flashcard apps, beginner textbooks, and Duolingo teach high-frequency vocabulary and core grammar. Once you've absorbed those, the low-hanging fruit is gone — and your *receptive* knowledge has run far ahead of your *productive* knowledge. You understand roughly twice as much as you can say (Laufer 1998 measured the gap at about 2:1 for intermediate learners). The plateau also hurts in a way the beginner stage doesn't. Beginners are excused. Plateau-stage learners can *almost* do it, which is worse — you start suspecting you're not the language-learning type. You are. You're just at the point where your tools stopped paying. ## What skills actually move you from B2 toward C1? Five things, and none of them is "more grammar": 1. **Deeper vocabulary in narrower contexts.** Stop chasing the next 1,000 most-common words; chase the next 500 in *one* domain you genuinely care about — cooking, finance, gaming, parenting. Hu & Nation (2000) put the lexical-coverage threshold for unassisted reading at ~95% known words, which you reach faster by going deep in your domains than wide across everything. (If you're earlier in the journey, start with [the first 1,000 words](/en/blog/first-1000-words-cover-75-percent) instead.) 2. **Longer turns.** Plateau-stage learners speak in 5-second utterances. Breaking out means comfortably holding the floor for 30–60 seconds. That sixfold jump is where the discomfort lives, and there is no shortcut — only a daily ramp. 3. **Hedging, modality, and stance.** "I think", "it seems that", "rather than", "as far as I'm concerned" — the connective tissue of adult speech. Learners skip these and sound flat; native speakers use them constantly. 4. **Prefab chunks instead of word-by-word assembly.** Erman & Warren (2000) measured around 58.6% of conversational English as formulaic sequences. Fluency is partly memorized. Collect whole phrases, not words. 5. **Recovery skills.** When fluent speakers lack a word, they paraphrase around it — *circumlocution* (Dörnyei & Kormos 1998). Plateaued learners freeze. Paraphrase is a trainable skill, and it's the single highest-leverage one on this list. ## The 30-day plateau-breaker This is the plan we give TalkToDia users who arrive stuck. It works with any conversation partner — a tutor, a patient friend, or Dia. - **Days 1–10: load the domain.** Pick one domain. Read 10 articles in it. Mark every word you don't know; put the 100 most frequent into your flashcard system. (TalkToDia's word bank does this automatically from your conversations — every word you actually use gets tracked and recycled.) - **Days 11–20: speak the domain daily.** A 15-minute conversation in that domain, every day. The target is 30-second turns, not one-liners. If your partner is human, tell them to stop rescuing you. If it's Dia, this is the default: she keeps asking the follow-up question instead of filling your silence. - **Days 21–30: close the loop.** Record yourself talking about the domain for 5 minutes a day (or do a voice call and review it). Note three things you couldn't say cleanly. Look up how a native says them. Use all three in tomorrow's session. The pattern behind all three phases: input narrowed to one domain, output stretched past your comfort, and a feedback loop within 24 hours. ## Why doesn't more input fix the plateau? Because after B1 your bottleneck isn't understanding — it's retrieval under time pressure. Krashen's input hypothesis explains how you got to B1, but Swain's output work (1985 onward) and Long's interaction hypothesis (1996) explain the next jump: you have to produce, get a reaction, and repair. Watching another 200 hours of Netflix widens a reservoir you already can't pump from. The retrieval-and-articulation loop only gets faster by speaking under mild pressure until retrieval becomes automatic — what DeKeyser (2007) calls *proceduralization*. That's the design principle behind TalkToDia: enough scaffolding that you don't freeze, enough demand that you can't coast. It's also why we built [voice calls at native speed](/en/blog/why-native-speakers-sound-fast) rather than slowed-down listening drills. ## How do you know it's working? Skip "feeling fluent" — it lags months behind reality. Track these instead: - **Turn length.** Time your answers. 5 seconds at week 0 → 30+ seconds by week 4 is on pace. - **Recovery rate.** Count how often you abandon a sentence vs. paraphrase your way out. The ratio should flip. - **Same-topic re-runs.** Explain the same article twice, three days apart. The second take should be noticeably smoother — that's proceduralization you can hear. ## FAQ ### How long does the intermediate plateau last? As long as you keep using beginner methods — for many learners that is literally years. With deliberate output practice (daily conversation, 30-second turns, feedback within a day), most learners feel clear movement within 4–8 weeks. The plateau is a method problem, not a talent ceiling. ### Am I stuck because I am bad at languages? No. The plateau is structural: the methods that work for A1–B1 (flashcards, gamified apps, graded textbooks) train recognition, and B2 requires fast retrieval and production. Almost everyone who reaches B1 hits this wall. Changing the training fixes it; doubling down on the old method does not. ### What should I do after finishing Duolingo? Switch from recognition practice to production practice. Pick one domain you care about, read heavily in it, and have a daily 15-minute conversation where you hold 30-second turns. Any patient conversation partner works — TalkToDia is built for exactly this stage, with conversations that adapt to your level and a word bank that recycles the vocabulary you actually use. ### Is B2 fluent? B2 is the first level most people would call functionally fluent: you can argue, joke, and work in the language with occasional gaps. C1 adds precision and ease in specialized or abstract topics. For living in a country, B2 is usually the practical target worth aiming at first. ## Sources - [Richards (2008) — Moving beyond the plateau](https://www.cambridge.org/elt/blog/wp-content/uploads/2017/04/Moving-Beyond-the-Plateau.pdf) - [Common European Framework — B1/B2 descriptors](https://www.coe.int/en/web/common-european-framework-reference-languages) --- Cite as: Proč se zasekáš na B1 (a 30denní plán, jak z toho ven) — TalkToDia Blog, https://talktodia.com/cs/blog/breaking-the-intermediate-plateau