Does extensive reading work for learning a language?

Extensive reading means reading a large amount of text that is easy for you, quickly, for the story or the information, without stopping to study the language. The Extensive Reading Foundation describes it as reading “very easy, enjoyable books to build their reading speed and reading fluency”, ideally a book a week or more at the reader’s level. For most learners the books are graded readers, written or retold with a limited vocabulary for each level. In Japanese it is called tadoku (多読), literally “reading a lot”.

Richard Day and Julian Bamford’s ten principles of extensive reading (2002), which the 2025 meta-analysis drew on, start with four that define it: the reading material is easy; a variety of material on a wide range of topics is available; learners choose what they want to read; and learners read as much as possible. The 2025 meta-analysis below used a similar definition: sustained, independent reading of longer texts at the learner’s level, chosen by the learner, with a focus on enjoyment.

Intensive reading is the opposite approach: a short, harder text read slowly and closely, with a dictionary and grammar notes, to learn the language in it. Extensive reading uses easy books read fast to practise what you already half know. The Extensive Reading Foundation puts the relation in one line: intensive reading “introduces new language items”, while extensive reading helps learners “practice and get a deeper knowledge of them”.

An effect of d = 0.41, the overall figure in the 2025 meta-analysis, means the average learner who read extensively ended up ahead of about two in three learners in the comparison group. That is a real gain for something learners can do on their own, but it is not a shortcut that replaces everything else. The effects for vocabulary and general proficiency were a little larger, and the one for reading comprehension a little smaller.

Two limits are worth knowing. Most of the evidence comes from classrooms: in the 2025 review, 56% of the comparisons (46 of 82) were with university students, and they ran from four weeks to a full academic year. And much of the evidence is about learners of English: Jeon and Day’s meta-analysis covered only English as a second or foreign language. Results for readers of Spanish, French or German are an extension, not a measurement.

The evidence points that way, though almost all of it comes from classes. In Jeon and Day’s meta-analysis, adults gained more than children and teenagers: d = 0.70 against 0.52 and 0.35. The 2025 review found no significant difference by educational level; it could not test age, because too few studies reported it. Neither studied self-taught learners as a group, so for someone reading alone the best guide is what the classroom studies share: easy books, a lot of them, read regularly.

One finding matters more for self-learners than for classes. Effects were larger when the choice of books was limited, for example to a level, and when learners had some form of accountability. A class provides both; a self-learner has to build them in, for example by choosing from one level of a graded reader series and keeping a log of books finished.

It is one way to get it. Comprehensible input is Stephen Krashen’s term: his input hypothesis (1982) holds that we acquire a language by understanding messages in it, input that contains language slightly beyond our current level, which he wrote as i + 1. Extensive reading is the cheapest source of that input in quantity: a graded reader at your level is, almost by definition, text you understand.

• One book teaches few words for good. After reading one graded reader with 25 invented words, learners could translate on average 4.6 of them right away and 0.9 three months later.
• Structure helps. In the 2025 review, programmes that limited the choice of books and added accountability did better than pure free reading.

• A book a week at your level is what the Extensive Reading Foundation recommends.
• About 200,000 words a year. In a study of 76 Japanese university students, the 14 whose reading speed rose most, from 97 to 130 words per minute, read on average 208,607 words in the year, counted in standard-length words. The authors tentatively suggest 200,000 a year as a minimum for similar learners.
• From 200,000 to 3 million words a year to meet English words often enough to learn them, by Paul Nation’s count. He needed just under 200,000 words of text to meet the second thousand most frequent word families at least 12 times each on average, and 3 million for the ninth thousand. At 150 words per minute, five days a week for 40 weeks, a million words a year is about 33 minutes a day.

Yes, but give extensive reading part of the time. In Namhee Suk’s study (2017), four classes of Korean university students had 100 minutes of reading lessons a week for a 15-week semester. Two classes (88 students) spent all of it on intensive reading. Two classes (83 students) spent 70 minutes on intensive reading and 30 on extensive reading. The second group did better on reading comprehension, reading speed and vocabulary.