Learning vocabulary from what you already read, with NoaLingua

Reading is the underused half

A page of ordinary prose holds several hundred words. An hour of reading exposes you to more vocabulary, and more varied vocabulary, than an hour of television, because written language is denser and less repetitive than speech.

It is also where the least is captured. Almost everyone has the same experience: look a word up, understand the sentence, keep reading, and have no memory of the word by the end of the chapter. The exposure happened. The learning did not.

Why looking things up mid-paragraph fails

Two reasons, and they compound. The first is that the lookup is a recognition event, not a retrieval one — the same failure described in why you forget vocabulary. You saw the answer; you did not produce it.

The second is that stopping destroys the read. Every interruption costs you the thread, so you either stop looking things up (and understand less) or stop reading properly (and enjoy it less). Most people oscillate between the two and conclude that reading in a foreign language is unpleasant.

Separate the reading from the studying

The workflow that solves this is old and works on paper as well as on screen: do the two jobs at different times.

1. Read for meaning. Do not stop. Where a word blocks you, mark it and move on — a pencil line, a highlight, a click, whatever costs you the least attention.
2. Guess as you go. Guessing from context is itself a retrieval act, and being wrong is fine; you will find out shortly.
3. When the chapter or article is finished, go back through your marks in one pass. This is now studying, and it is not competing with reading.
4. Keep only the words that are worth keeping: ones in the top few thousand by frequency, or ones you have now met twice. A rare word met once is not worth a card.
5. Save each with its sentence. The sentence is what makes it retrievable later.

Doing this on a screen

On paper the marking is a pencil. On a screen it can be cheaper still, which matters because the cost of the interruption is the whole problem.

NoaLingua exists partly for this: click a word on any web page or inside a PDF and the card opens with the meaning that word has in that sentence, with saving as one more click and no navigation away from what you were reading. The sentence, the page and the source come with it automatically.

It is not the only way and it is not necessary. A notes file with pasted sentences and a free flashcard program does the same job for nothing. What matters is that the capture is cheap enough that you actually do it, and that the sentence comes along.

What to read

Apply the same test as anywhere else: around 4 to 10 unavoidable lookups per 200 words, per the level test. Below that you are having a nice time; above it you are decoding.

Journalism is the most reliable starting point in most languages — short pieces, plain register, current vocabulary, and a natural stopping point every few hundred words. Novels are the harder and better long-term option, because the vocabulary repeats within a single book and repetition inside a context you care about is the most efficient learning there is.