You can read a news article and follow the argument. Ask you to explain it out loud in the same language, and the sentences come out slower, simpler, full of pauses while you search for a word you definitely recognized on the page a minute ago. If that gap feels like proof you're not actually learning, it's worth saying plainly: it isn't. It's the normal, expected shape of how a second language develops, and the frustration usually comes from expecting understanding and production to move together when they don't.
Two different skills, not one skill at two speeds
Comprehension and production draw on genuinely different cognitive processes, even though they both get called "language ability" as if they were the same thing measured at different points. Comprehension is recognition: given a word or a structure, matching it to a meaning you already hold. Production is retrieval and assembly: starting from a meaning you want to express, searching your own memory for the right word, the right form, the right order, under real time pressure, without the word already sitting in front of you as a prompt.
Recognition is simply an easier cognitive task than retrieval. This is true even in your native language: most adult native speakers can understand vocabulary in a book review or a formal document that they would rarely, if ever, use unprompted in their own speech. The gap between a passive and an active vocabulary is a normal feature of language ability generally, not a defect specific to learners.
Why reading widens the gap faster than speaking closes it
Reading, especially reading real text regularly, is an unusually efficient way to build recognition. Every article you read exposes you to vocabulary and sentence patterns in context, reinforcing the recognition pathway, often without you consciously drilling anything. This is part of why a dedicated reading habit tends to produce comprehension gains that feel almost effortless compared to the deliberate, often uncomfortable practice that builds speaking fluency.
Speaking fluency, by contrast, requires retrieval practice specifically: producing language under time pressure, ideally with feedback, which is a different activity from reading and doesn't happen as a side effect of it. A learner who reads constantly but rarely speaks is, correctly, building one skill much faster than the other, not failing to build a skill evenly. The imbalance isn't evidence of an inefficient method. It's the direct, expected consequence of which activity you've actually been doing.
Why this gap is worth naming as progress, not a problem to fix immediately
There's a real temptation to treat the comprehension-production gap as something wrong that needs correcting right away, usually by feeling bad about not speaking enough. That framing undersells what's actually happened: strong reading comprehension is real, hard-won progress. You've built a large recognition vocabulary and a working sense of how the language's sentences are actually structured, both of which are genuine prerequisites for speaking well later, not wasted effort in a different direction.
The practical implication is that closing the gap, when and if you want to, is its own separate project, one that requires its own kind of practice, not a sign that the reading was the wrong investment. Someone with strong reading comprehension who starts speaking practice has a real advantage over someone starting from nothing: they already have the vocabulary and structural sense sitting in recognition memory, ready to be pulled into retrieval practice, which tends to make the transition faster than building both skills from scratch would.
A concrete version of the gap
Picture two learners who've both spent six months on the same language. The first spent most of that time in an app, drilling short dialogues and speaking exercises, with little exposure to longer real text. The second spent it reading real news daily, with almost no speaking practice. Given a real article, the second learner will very likely outperform the first at extracting accurate meaning, correctly parsing a complex sentence, and recognizing far more vocabulary. Given a live conversation, the first learner may well sound more fluent in the narrow range of things their drills covered, even though their actual vocabulary and structural knowledge is smaller.
Neither learner is doing it wrong. They've built different skills because they practiced different activities, and that's the whole explanation, not a sign that one method is broadly superior to the other. The reading-heavy learner has built a foundation, recognition of real vocabulary and real sentence patterns, that will make future speaking practice more efficient once they start it, because they won't be learning new words and structures and trying to produce them at the same time. They'll only need to practice production of things they already understand, which is a smaller task than learning and producing simultaneously.
What actually helps close it, if you want to
Retrieval practice specifically, not more comprehension input, is what narrows the gap. Reading a passage and then trying to summarize it out loud, in your own words, in the target language, forces exactly the retrieval process that reading alone doesn't exercise. This is a genuinely different task from reading itself, even though it uses material you've already comprehended, because now you're the one generating the sentence rather than recognizing one someone else wrote.
Speaking with real feedback, even briefly, accelerates this further, because retrieval attempts that go uncorrected can build fluency in producing incorrect forms just as easily as correct ones. The point isn't that reading was the wrong thing to prioritize. It's that comprehension and production are separate skills that respond to separate kinds of practice, and understanding that distinction is what turns "why can't I speak as well as I understand" from a source of frustration into an accurate description of where you actually are and what would move the other skill forward.