Why you still can’t understand native speakers, and what actually fixes it

The short version: native speech is hard because you are missing words, because the words you do know do not sound the way you learned them, and because nothing you have studied was ordered, so none of it compounds. Most learners attack only the first one, usually by doing more of whatever they were already doing, and stall for years.

Here is each problem, how to tell which ones are yours, and what actually moves each one.

Problem 1: you know far fewer words than the scene needs

This is the one people underestimate most, because vocabulary feels like the part they have been working on. The issue is the size of the gap.

To follow spoken material without help, research on English puts the requirement at roughly 6,000 to 7,000 word families for 98% coverage, which is the level at which you can infer the rest from context. Television specifically is friendlier: the most frequent 3,000 word families plus proper nouns covered about 95% of the dialogue across 88 programmes, with the figure ranging from 2,000 to 4,000 depending on genre.

Now compare that to where a finished app course leaves you, which is commonly in the 1,500 to 2,500 range. At that size you are not slightly short. You are meeting an unknown word every second or third sentence, and each one costs you the sentence it was in plus the start of the next one while you are still thinking about it.

Figures in this section come from two studies of English-language material: Webb & Rodgers, Vocabulary Demands of Television Programs (Language Learning, 2009), and Nation, How Large a Vocabulary Is Needed for Reading and Listening? (2006). They describe English, and the exact numbers for French or Spanish will differ, but the shape of the problem is the same.

The practical consequence is counterintuitive: 95% coverage does not feel like 95%. One word in twenty is roughly one per line of subtitle, which is enough to keep you permanently half a beat behind. We worked through what that means for a specific evening in front of Netflix in how many words you need to watch without subtitles.

Problem 2: the words you know do not sound like the words you learned

This is the one that makes people think their listening is broken. It usually is not. You learned words as isolated, fully pronounced units, and nobody says them that way.

In real speech, words at normal speed are run together, endings drop, vowels reduce to almost nothing, and the boundaries between words stop being audible. A Spanish learner who knows ¿qué es esto? perfectly well on paper hears quesesto and finds nothing to hold on to. A French learner who has drilled je ne sais pas hears chépa. The vocabulary was never the problem in those cases. The segmentation was: you cannot look up a word you did not manage to isolate.

The fix here is specific and it is not “listen more”. It is listening to speech while reading the same speech, so your ear gets told, line by line, where the boundaries actually fell. That is the entire mechanism behind dual subtitles, and it is why they help far more than subtitles in one language alone: one line shows you what was said, the other confirms what it meant, and you stop guessing at both simultaneously.

Problem 3: nothing you learned was ordered, so nothing compounds

The first two problems are well known. This one is the reason people stay stuck after fixing the first two.

If your input is an episode, the grammar in it arrives in the order the screenwriter needed, which is no order at all. A single scene can contain a subjunctive, a past conditional and a piece of slang, and you are meant to absorb all three from context while also following a plot. Beginners cannot, and the evidence that they cannot is that they watch for ten minutes, understand a third, and quietly switch the subtitles back.

Ordered material solves this, which is what a course is for. But a course on its own has the opposite failure: it teaches you a rule in a clean invented sentence, and clean invented sentences are not what anyone says. You end up able to conjugate on demand and unable to recognise the same conjugation when it goes past at speed in a contracted, mumbled, interrupted version.

So the two halves each fail on their own. We went through why the immersion half in particular fails at the beginner end in why immersion alone does not work at A1 and A2.

Why doing more of the same does not work

Each of the three problems has a different fix, and the usual instinct picks one tool and applies it to all three:

Read the right-hand column again. Every one of those is the thing a stalled learner does more of.

The loop that actually works

What resolves all three at once is not a tool, it is a sequence. Three steps, in this order:

The order matters more than any individual step. Step 2 without step 1 is immersion, which stalls beginners. Step 1 without step 2 is a textbook, which produces people who cannot follow a conversation. The loop only pays off closed.

How to run the loop without building it yourself

You can assemble this by hand: a structured grammar syllabus, a subtitle tool, a flashcard deck, and the discipline to keep all three pointing at the same notion in the same week. Plenty of people do, and it works.

It is also exactly what Streal is: a course of 32 chapters from A1 to C2, where each lesson ends on the real lines from the Netflix catalogue that contain the notion you just learned, with the timecode, so one click opens the scene at that second. While you then watch, it knows which notions you have covered and points those out as they go past, and holds back the ones you have not met yet rather than dropping them on you mid-scene. Chapter 1 is free in full if you want to see whether the loop does anything for you before paying for it.

How long this takes

Nobody can give you a date, and anyone who does is selling something. What can be said honestly is the shape of it: the vocabulary gap closes slowly and roughly in proportion to hours of real input, the segmentation problem improves noticeably within weeks once you start pairing audio with text, and the ordering problem is fixed the moment you start following a sequence rather than a mood.

If you are coming off an app and want a concrete next block of work rather than a principle, that is what to do after Duolingo, and the month-by-month version is in the A1 to B2 study plan.

FAQ

Is it normal to understand a teacher but not a TV show?

Yes, and it is the clearest sign that problem 2 is yours. Teachers and podcast hosts aimed at learners articulate fully and slow down at phrase boundaries. Actors do neither. The vocabulary is often the same; the delivery is not.

Should I watch with subtitles in my own language or the target language?

Neither alone works well below B1. Your own language means you read the plot and learn nothing; the target language alone means you decode text you cannot yet segment by ear. Both at once is what actually trains the boundary between the two.

Does listening without subtitles train your ear faster?

Only once you already know most of the words. Below that it trains tolerance for noise, not comprehension, because you cannot isolate an unknown word well enough to ever look it up.

How many hours of input do I need?

There is no honest single number, and it depends on how much of each hour you actually understand. Input you follow at around 90 to 95 percent does far more per hour than input you follow at 50 percent, which is the argument for building the foundation first rather than immersing early.