The evidence question
Should You Look Up a Word While Reading — or Check It Later?
Looking up every unknown word can break your reading flow, but waiting can cost useful learning. Here is what research says about when to check a word.
A question worth asking.
08 sources, open to exploreIn this article 14 sections
You are halfway through a good English article when one word stops you.
You could tap it now. The definition is one second away.
Or you could keep reading, hold the rough meaning in your head, and check the word when you reach the end of the paragraph.
Both choices sound reasonable. They also protect different things.
One protects precision now.
The other protects the thread of the text.
So when is the interruption worth it?
Look up a word now when you need it to understand what you are reading. Wait when the word is not blocking comprehension and stopping would cost more than the answer is worth.
Research gives strong support to accessible meaning help during reading. Glosses and dictionary use can improve vocabulary learning. But the evidence does not support turning every unfamiliar word into an immediate lookup. Studies of real reading behavior show that excessive, poorly targeted dictionary use can hurt task performance. Post-reading vocabulary work can also improve retention. Yanagisawa, Webb & Uchihara, 2020 Prichard & Atkins, 2021 Peters, 2009
There is no well-established rule that says “always wait until the end of the paragraph” or “always check immediately.”
A better rule is based on the job the word is doing.
Timing matters because reading and word learning are not the same task
When you read a story, report, or essay, your main job is usually to build a coherent idea of what the text means.
A lookup creates a second job.
You leave the sentence, inspect a definition, choose the relevant sense, connect it to the sentence, and then return to the larger message.
That can be useful. It is still a switch in attention.
Digital dictionaries make the switch cheap enough that it is easy to stop without asking whether you need to. In a 2021 eye-tracking study, L2 readers often opened dictionary links even for words that were irrelevant to the task or could be inferred from context. Looking up many definitions nonselectively and spending little time processing them were associated with poorer task performance. Prichard & Atkins, 2021
The problem was not that dictionaries are bad.
It was that availability had become a habit instead of a decision.
Immediate meaning support is genuinely useful for vocabulary learning
There is a good reason learners want definitions close at hand.
Yanagisawa, Webb, and Uchihara combined 359 effect sizes from 42 studies involving 3,802 learners. Readers with glosses learned more target vocabulary than readers without glosses. Average learning scores were 45.3% versus 26.6% on immediate tests and 33.4% versus 19.8% on delayed tests. Results varied with gloss design, learner characteristics, comprehension, and the kind of vocabulary test. Yanagisawa, Webb & Uchihara, 2020
A 2020 eye-tracking experiment gives a closer view of what happens during reading. Forty-five high-intermediate learners read English text with or without L1 glosses. The gloss group performed better on meaning recall and recognition. Time spent processing target words and glosses was also related to later vocabulary scores. Ouyang, Huang & Jiang, 2020
So an immediate definition is not a shortcut that automatically weakens learning.
It can focus attention on the form–meaning connection at the moment the word matters.
That is especially valuable when misunderstanding the word would damage your understanding of the sentence.
One unknown word can have three different jobs
Imagine you are reading:
The committee rejected the proposal because its cost estimates were untenable.
If untenable carries the reason for the rejection and you cannot infer it, continuing may leave the whole sentence unresolved.
Check it now.
Now imagine:
She crossed the courtyard beneath an old colonnade and entered the library.
You understand the action. The exact architectural term may be interesting, but the story still works.
Keep reading. Mark it if you care.
Then imagine:
The new rule is provisional and will be reviewed in June.
You think provisional probably means temporary, but the distinction matters to the argument.
A quick check may be worth the interruption even though you can keep reading.
The timing decision is therefore not just about whether a word is unknown.
It is about how much uncertainty the word creates.
Waiting has one clear advantage: you get more context before you decide
Readers sometimes reach for a definition too early.
Prichard and Atkins found that participants often opened dictionary links before they had read contextual clues that could help them understand the word. Prichard & Atkins, 2021
Reading another sentence can change the problem.
The next clause may define the word. An example may make it obvious. A pronoun or contrast may narrow the meaning. Or you may discover that the word was not important at all.
This does not mean you must guess every word before checking it. That question is covered separately in Should You Guess an English Word from Context — or Look It Up?.
Here, the point is narrower:
A short delay can give the text a chance to answer first.
But “check it later” can quietly become “never check it”
Delayed lookup has a practical weakness.
You may forget the word.
You may remember that something confused you but not the sentence that made it meaningful. Or you may finish the chapter and decide you do not want to reopen six dictionary tabs.
Research on vocabulary learning from reading consistently shows that extra attention to target words can improve what survives after reading.
In a study of 137 college learners reading L2 German, Peters compared several ways of increasing attention to unfamiliar vocabulary. A post-reading vocabulary task improved later word retention. The broader pattern favored what Peters called input plus: reading combined with activities that make learners notice, look up, and process useful words again. Peters, 2009
This gives delayed lookup a condition:
If you postpone a useful word, you need a reliable way to return to it.
Otherwise delay is not a learning strategy. It is just omission.
42 studies show why “never interrupt reading” is too strong
The 2020 glossing meta-analysis included 42 studies and 3,802 learners.
Across those studies, gloss-supported reading produced higher vocabulary learning than reading without gloss support on both immediate and delayed tests. Yanagisawa, Webb & Uchihara, 2020
That does not settle the exact best second to open a definition.
It does show why a blanket rule such as “never look anything up while reading” throws away a useful learning tool.
Myth: stopping always destroys reading flow
Myth: A serious reader should never interrupt a text for vocabulary. If you stop, you are no longer reading naturally.
Reality: Some interruptions solve the exact problem that is preventing comprehension. A short, relevant lookup can let you return to the text with less uncertainty.
The real risk is not one lookup.
It is a chain of low-value lookups that turns a page into a list of disconnected definitions.
A 2025 study of digital dictionary behavior reinforces this distinction. Using eye tracking and lookup logs with 73 L2 readers, the researchers found that higher-performing readers were better at selecting context-appropriate meanings and used more strategic lookup behavior. The useful skill was not simply “use the dictionary more” or “use it less.” It was using it well. Zhang et al., 2025
Three moments to check a word
| When you check | Best fit | Main advantage | Main risk |
|---|---|---|---|
| Immediately | The word blocks the sentence or precision matters | Resolves uncertainty while context is active | Too many stops can fragment reading |
| After the paragraph or section | You have a workable meaning and want to protect flow | Gives context more time to help | You may lose the exact question |
| After the reading session | The word is useful for learning but not needed for comprehension | Separates reading from deliberate vocabulary study | Marked words may never get reviewed |
This table is a practical decision model, not a claim that research has identified three universally optimal lookup windows.
Direct experiments on exact lookup timing are much thinner than the broader evidence on glosses, dictionary use, and post-reading vocabulary work.
Be careful with claims about “delayed lookup”
Some studies compare immediate and delayed gloss presentation, but that is not identical to choosing to finish a paragraph before opening your own dictionary.
For example, one computer-mediated study with 100 engineering students compared immediate glosses with glosses shown after a short delay. Both immediate and delayed gloss conditions supported vocabulary learning and comprehension, with some differences by gloss language and outcome. That experiment is useful evidence that timing can matter, but it does not establish a universal delay interval for normal reading. Faghfouri, Aliasin & Mohammadi, 2022
This distinction matters.
A laboratory delay of seconds, a learner postponing a lookup until the next sentence, and a reader reviewing vocabulary after a chapter are three different behaviors.
We should not pretend one study answers all three.
The purpose of your reading should change the threshold
If you are reading for pleasure, protecting momentum may matter more.
If you are reading a contract, academic paper, technical manual, or exam question, one misunderstood word can change the whole claim. Precision deserves more interruptions.
If your main goal is vocabulary study, stopping is less costly because learning the words is part of the task.
If your main goal is extensive reading, stopping for every unfamiliar item can defeat the point of sustained exposure.
The same learner can therefore make different lookup decisions in different texts without being inconsistent.
The goal changes. So should the threshold.
A simple lookup-timing rule
- Read to the end of the sentence first, unless the word makes that impossible.
- Ask whether the word blocks the main idea. If yes, check it now.
- If the sentence still works, keep going briefly. Let the next sentence or paragraph add context.
- Mark words that seem useful, repeated, or important. Do not trust yourself to remember them later.
- Review marked words after the section or session. Reopen the original sentence when you check them.
- Stop marking everything. If unknown words are so frequent that nearly every line needs help, the text may be poorly matched to your current reading goal.
This rule is a practical synthesis of the evidence. It is not a tested six-step protocol.
For the separate question of whether you need to know every word in a text, see Do You Need to Understand Every Word When Reading English?.
Look up the word when the answer is worth the interruption
The research does not give us a stopwatch.
It gives us a better decision.
Meaning support during reading can improve vocabulary learning. Selective dictionary use can resolve important uncertainty. But automatic lookup of every unfamiliar word can pull attention away from the text, and useful post-reading processing can strengthen retention.
So do not ask:
Should I always look up words now or always wait?
Ask:
Does this word need an answer now?
If it blocks comprehension, check it.
If it does not, give the text a little more room. Mark the word if it matters. Then return to it on purpose.
The best lookup is not the fastest one.
It is the one that helps you understand the text without turning the text into a dictionary exercise.
Sources and further reading
- Yanagisawa, Webb & Uchihara (2020) — How do different forms of glossing contribute to L2 vocabulary learning from reading?
- Prichard & Atkins (2021) — Evaluating the Vocabulary Coping Strategies of L2 Readers: An Eye Tracking Study
- Peters (2009) — Learning L2 German Vocabulary Through Reading: The Effect of Three Enhancement Techniques Compared
- Ouyang, Huang & Jiang (2020) — The effects of glossing on incidental vocabulary learning during second language reading
- Hulstijn, Hollander & Greidanus (1996) — Incidental L2 lexical acquisition in reading: marginal glosses, dictionary use, and reoccurrence
- Zhang et al. (2025) — Strategic use of digital dictionaries in L2 reading
- Faghfouri, Aliasin & Mohammadi (2022) — Computer-Mediated Immediate and Delayed L1 and L2 Glosses
- Ng Lay Shi (2026) — The effects of glosses on English L2 incidental vocabulary learning through reading: a meta-analysis