How many times do you have to meet a word?

Stuart Webb wanted to know how many times you have to meet a word before it is yours, so he counted. In Webb (2007), 121 Japanese university students learning English met ten target words in one, three, seven or ten short sentences each, with a gloss for the meaning. The words were real English words dressed in invented spellings, so nobody could have known them already.

Then he tested ten different kinds of knowing: the spelling, the meaning, the grammar of the word, how it sits in a sentence, the words it keeps company with, and whether the students could recognize each of those or produce them. Knowledge went up with every step in repetitions, and the ten-encounter group was ahead of the rest on most measures. The tests were immediate, so this is about learning, not about remembering a month later.

The interesting part is that the kinds of knowledge arrived at different speeds. Recognizing the word and getting a rough meaning came after a few meetings. Using it, and knowing its collocations, the words it commonly goes with, took many.

Knowing a word is not one switch. It is several, and they flip one at a time.

Ten is not a law. Webb did not test stories against flashcards, and I am not going to pretend he did. What the study does say is that meetings in context are what build the whole word, not just its dictionary line.

A story gives you those meetings for free. Its vocabulary comes back because the story is about something, and each return puts the word in a new sentence with new neighbors, which is exactly the knowledge that took ten tries. A flashcard gives you the same meeting ten times.

So pick a book you can mostly follow, read it while you listen, and when a word comes back for the third time, notice what is around it. Aloglot is the version of this I use: it reads the book to you, lights each word, and explains any word you tap, so the tenth meeting is a tap away from the first.

Sources

• Webb (2007). The effects of repetition on vocabulary knowledge. Applied Linguistics, 28(1), 46–65.