You can finish a page in a foreign language, understand every word on it, and be wrong about what happened. That is not a vocabulary problem and no amount of looking words up will find it, because nothing about the experience felt like failure. This guide is about the check itself: why comprehension needs testing at all, what a question has to do to be worth answering, and how to build the check into ordinary reading without turning it into homework.
Why reading alone does not tell you
Reading gives you constant feedback about words and almost none about meaning. When you hit a word you do not know, you notice immediately: you stop, the sentence stalls, the gap is obvious. When you assemble a sentence into slightly the wrong situation, nothing stops. The text keeps going, your version keeps working, and you carry the error forward.
This is why re-reading feels so productive and proves so little. On the second pass the words are familiar and the page is smoother, so you read that smoothness as understanding. It is not evidence. It is recognition, which arrives whether or not you got the meaning right the first time.
The check has to come from outside the text. Close it and answer something you cannot answer by looking.
What a comprehension question has to do
Most questions attached to language-learning texts are vocabulary questions wearing a disguise. "What does mercado mean?" tells you whether you recognised a word. Useful, but it is not what you wanted to know.
A question is testing comprehension when it has these properties:
- The answer is not in any single sentence. If it is, you are testing whether the reader saw that line. Motive, sequence and consequence questions force the reader to have built the whole situation.
- The wrong options are plausible. Distractors assembled from words that really appeared in the text catch the reader who is pattern-matching. Distractors that are obviously silly test nothing.
- The text is closed. With the text open, a comprehension question becomes a search task. Searching is a real skill and an easier one, and it will happily hide the gap you are looking for.
- It survives translation. If the question stops making sense once you state it in your own language, it was about the wording rather than the meaning.
Four questions on a short story is usually enough. Beyond that you are mostly measuring stamina.
A wrong answer is the useful outcome
The instinct is to treat a wrong answer as a small failure and move on. It is the most informative thing that happens in the whole exercise, because it localises the problem. Somewhere in that text your model of events and the actual text diverged, and the question tells you roughly where.
So do not re-read the page. Find the sentence, and then work out which of these it was:
- A word you were confident about and had wrong. False friends live here, and they are invisible precisely because you did not stop.
- A grammatical signal you skimmed. A tense that placed the event elsewhere in time, a negation, a conditional.
- A referent you lost. Who "he" was, or in a pro-drop language like Spanish, Italian or Korean, who the unwritten subject was.
- A connector you ignored. The small words that carry cause, contrast and sequence, which are easy to treat as noise.
Over a few weeks you will notice you keep landing in the same category. That is worth more than any score, because it tells you what to work on.
Building the check into normal reading
The objection to all of this is that it turns reading into schoolwork. It does, if you do it on everything. Do not do it on everything.
A workable split: read extensively for volume, and test occasionally to calibrate. Most of your reading should be uninterrupted, at a level where you are not stopping, because that is what builds the fluency. Then once a week, take one short text and run it properly:
- Read it once, at speed, no dictionary.
- Answer the questions with the text closed.
- For anything wrong, find the sentence and name the category.
- Now look up whatever you want, and re-read once.
Twenty minutes. The point of the weekly check is not the score, it is finding out whether the ease you have been feeling all week is real.
Using the result to pick your level
Comprehension testing also answers the question people usually guess at: is this text the right level for me?
- Questions wrong and reading was hard. Too difficult. Drop a level; you are decoding, not reading.
- Questions right and reading was hard. Correct level. This is where the learning happens and it is supposed to feel like work.
- Questions right and reading was effortless. Too easy now. Move up, or the text stops teaching you anything.
- Questions wrong and reading felt easy. The important case, and the reason this article exists. You are skimming and your confidence is miscalibrated. Same level, slower, with the check every time until the two line up.
Where to try it
- The comprehension test. Read a short story, then four multiple-choice questions on it, in twelve languages. Free, no account, and it tells you immediately.
- The reader. The same stories with tap-for-meaning, for the lookup pass after you have answered.
- A CEFR level test. If you would rather start from a level than work it out from the table above.
- Free graded readers. Printable stories, one per CEFR level from A1 to C2, in twelve languages.
- Extensive and intensive reading. The wider version of the split described above.