92 Books Later: How To Learn A Language By Reading

I’m going to list three things and ask you to guess what they all have in common:

  • El Extranjero
  • Padre Rico, Padre Pobre
  • El Código Da Vinci

…apart from making a list of three random Spanish titles? Any ideas?

This is the list of books I´ve read in Spanish in June. I love learning languages by reading, and I love reading in other languages. Reading is an interesting, effective, and, above all, fun way to learn a language. Especially when you compare it with taking Spanish courses from any university in the world.

And, oh, wait! My year-long stock of Spanish literature cost me at least ten times less than would do a full-year Advanced Spanish course.

Forget about language classes. Investing in books is always a better deal when it comes to language learning. Here’s why this is so and how to get there.

We learn languages by reading

The good old statistics tells us that the vocabulary size of a typical college student comes down to approximately 20 000 words. We tend to expand it with the rate of 1000 words per year, and most of these gains originate from reading.

Reading is the primary and probably the fastest way to increase vocabulary, no matter in what language. And if you don’t remember memorizing a list of 20 000 most frequent English words as a child, it’s because you didn’t do any of that.

We don’t have to know all the words in a sentence to understand the point. We can derive the meaning of unknown tokens from the context, even if it gives us only a vague idea of what the word may signify.

Over time, as we encounter the same token over and over again, we add extra information to our “mental dictionary” until we construct a pretty good description of this word’s meaning. It adds +1 to our internal vocabulary. This process is called incidental learning.

Could the same trick work with foreign languages?

Learning a foreign language through reading

For many people (especially those with a degree in Education), rules of the normal world cease apply, when it comes to second language learning. Reading turns into a game of dictionary look-up. And suddenly, it becomes a highly challenging activity “for fluent speakers only”.

It’s true that you do already have to know some vocabulary before you can read in another language.

As Paul Nation, a leading vocabulary acquisition researcher from Victoria University of Wellington, explains, a learner should know 98% of vocabulary in a book in order to read it with ease and learn the remaining 2% incidentally.

No matter how insurmountable it sounds, incidental learning is still precisely what you want to count on when reading in a foreign language.

The very same mechanism that allowed you to learn 1000 English words per year would let you guess the meaning of un mot inconnu from the context, without consulting your little Larousse every two minutes. All you need to do is to read at the right level.

The more you read, the more times you encounter your target words, the more chances you have of finally learning them.

What is really crucial here is frequency because repetition is essential for vocabulary acquisition. However, there is no set “optimal” number of repetitions necessary for learning a word. Recommendations vary from study to study, but Dr. Paul Nation assures that 12 repetitions would be a safe bet in most cases.

What is clear is that if you want to learn a language by reading, you have to read a lot. But how much is this a lot?

How many books you have to read to learn another language by reading?

I love reading research papers on language acquistion, but only handful of them ultimately change my approach to learning.

This one was life-changing.

I refer to Paul Nation’s 2014 corpus-based study, in which he calculated the amount of input necessary to learn the most frequent 9000 words.1 Here are his findings:

Target level Target input
in tokens
Number of novels
(120 000 tokens per novel)
2000-level ~170 000 2
3000-level ~300 000 3
4000-level ~530 000 6
5000-level ~1 060 000 9
6000-level ~1 450 000 13
7000-level ~2 035 000 16
8000-level ~2 400 000 20
9000-level ~3 000 000 25

Let’s imagine that you already know the first 2000 most frequent words. In this case, you would have to read just about 300 000 words (or 3 books) to pick up another thousand words at the 3000-level.2

Here’s a caveat, however: reading three novels won’t guarantee a 100% coverage for all target words at this level. Out of one thousand tokens, roughly 800 would be met 12 times and about 80 one or two times. There’s nothing you can do about it, rather than read more. Due to the nature of Zipf’s law, some words always show up more often than others.

Covering the missing 1000-level

You might have noticed that 1000-level is missing. There are two reasons for that.

The first one is that Zipf’s law would simply ruin the calculations. 1000-level includes 1000 most frequent words and covers most “outsiders” such as function words (the, of, me) and general concepts (people, to like, park). These tokens are so frequent that the first 135 words alone account for 50% of any text.

The second reason is that you can’t read without knowing any words at all. For me, having a basic 1000 words vocabulary is a prerequisite for reading anything in another language. I know that without the basics I can’t access stuff I really want to read and the idea of reading children short-stories rarely excites me. So I typically spend a month just on learning these 1000 words with spaced repetition.

How much time should you devote to reading to learn another language from scratch?

Are you already making a list of 25 Italian books that would help you to learn 9000 words?

Don’t get into this trap. The table above is a build-up list. It means that if you know 0 words, you won’t turn into a fluent speaker after reading 25 books. These 25 books will help you to cover 9000-level words if you already know 8000 most frequent words by this point.

So how much time would it take to learn a language by reading from zero?

It depends on how you approach the task.

Method 1: Slow and steady

In his article, Paul Nation suggests learning vocabulary at the rate of 1000 words per year: similar to vocabulary acquisition rate for L1. In this case, you would need to finish just 2 books in the first year, 3 in the next, 6 in the third and so on.

And reading 3 Italian books in a year, 10 minutes a day, doesn’t sound like a bad idea at all. But… This way it would take you 9 years to “learn” a language.

Slow and steady wins the race? Well, probably not in this case. Nine years is a huge period of time, and your goals may change 180 degrees south – and multiple times (sometimes to the point where you don’t even feel like learning this language anymore).

It happened to me at least trice. And if I didn’t manage to reach working proficiency before this change of the course, the time was practically wasted.

Method 2: Fast and furious

That’s why I like to take everything to the extreme and reach a high proficiency level as fast as possible before I changed my mind again.

Back to the table, if we add all the numbers from the second column, we get approximately 11 000 000. That’s the number of tokens you have to process to learn the most frequent 9000 words from scratch.

This equals to something like 92 books. With the reading speed of 200 words per minutes (which is comfortable enough: I tend to read English at 295 wpm, and English is far from being my native language), it would take you 150 minutes a day (2.5 hours) to process all this beauty in a year. 

And as a person who actually does aim at that, I can testify that this number is insane.

However, there are ways to make your life easier but still learn a language through reading in record time.

How to learn foreign language vocabulary in record time?

Language learning is an ongoing process. You’ll never be “done”, even after you hit this 9000-level milestone. However, more often than not, you can consider yourself fluent way before you learn all these words.

That’s why there is a number of things you want to rethink about language learning, reading, fluency, and vocabulary size:

1. You don’t have to get all the input from reading

Reading in a foreign language is one way to get the essential comprehensible input. But it’s not the only one way.

Any activity you do in your target language will contribute to that cherished number of 11 million tokens you need to upload to your brain. You may listen to podcasts, watch movies and YouTube, talk to strangers, read magazines, play games… When it comes to vocabulary acquisition, anything goes, seriously.

Diversify your input, and you won’t ever feel bored. But keep in mind that spoken input is often way slower than written. Read a book for two hours and you will process approximately 24 000 words. Spend the same two hours watching a movie and you will get just about 10 000 words3With a typical speech rate for movies of 83 words per minute.4.

2. You don’t have to learn all 9000 words in a single year

If you never managed to finish 92 books in a year, don’t expect to finish 92 foreign language books in the same time.

You would need to have a well-established reading habit (as well as a ton of free time and motivation flying high 365 days in a row) in order to read this much. So save this idea for your retirement.

Instead, aim to cover, say, 1000-5000 levels in the first year, 6000-7000 in the next, and gradually work your way through 8000-9000 levels during the following 18 months. This way, you’ll consistently read 20-30 foreign language books a year, which is more than manageable.

3. You don’t even have to know all 9000 words

Fluency is not measured in vocabulary size.

According to Paul Nation, you need to know 7000 to 8000 words in order to read unsimplified texts and 6000 words to understand movies. But two Canadian linguists, Stuart Webb and Michael Rodgers, would argue that 3000 words are more than enough for watching foreign language TV.

You may feel more than confident with a vocabulary of 3000 words. To understand what I’m talking about, check out some words from English 4000-level: judicial, urged, clearing. Now tell me, how often you use them.

So aim to reach the 5000-level through natural input in the first year. Then spend the next two years solidifying your knowledge.

The good news is that the more you read at the higher levels the better you end up knowing the more frequent vocabulary. Thus, if you read a novel at the 6000-level, you automatically refresh and consolidate your understanding of 5000 most frequent words.

What books to read to learn a language?

4-year old kids don’t start their reading adventures with War and Peace by Leo Tolstoy. Because it’s too hard and kills motivation, right?

There is an assumption that beginner language learners shouldn’t do anything similar either. Nobody thinks, however, about the motivation of adults dying of boredom while reading Little Red Riding Hood in Portuguese just because it was “their level”.

Always prefer challenge to boredom. We linguists pray at comprehensible input, but it doesn’t mean that you’re doomed to read short stories until you gained 5000 words vocabulary size. Input can be made comprehensible with a dozen reading strategies or simple one-click dictionary lookups on devices like Kindle Paperwhite or Kobo Aura.

The rule of the thumb here is to read as much as possible. This way, book after book, you’ll expand your vocabulary, enhance your understanding of grammar and increase your proficiency level. Try to learn a language through reading and you’ll never want to do it any another way.

Image Credits: Photo by Ahmad Ossayli on Unsplash

Author Details
Polyglot, Author and Founder of Linguapath
Hey! I\'m Alina Kuimova, and my long-lasting obsession with learning languages led to the creation of this site. Apart from being a grammar enthusiast, I enjoy reading smart books in any language available, finding easier ways for the brain to learn things and buffing productivity stats by 180%.
  1. Nation, I. S. P. (2006). How much input do you need to learn the most frequent 9,000 words? Reading in a Foreign Language, 26, 1–16.
  2. Paul Nation based his study on English corpora. So his calculations will still be within the right range with most European languages. I wouldn’t apply it to languages that are radically different from English, such as Mandarin, Japanese and Arabic.

Read these next:

Get language learning hacks based on proven linguistic research.