Slices of language

The ladybird browser project has an interesting methodology they say they are using to keep motivation up: instead of building every api from the bottom up, they are working in slices. Something like “make youtube work”, from the low level features to the high level features needed for that. The idea is that it’s more fun and fulfilling because you end up with an actual factual real modern website that works at the end of it, rather than implementing something like a camera api which isn’t necessary. And lots of other sites will benefit from the support across the stack that youtube needed, so just focussing on that one site means a lot of other sites work better too.

I wonder how applicable the same idea is to language learning. In particular, i think that slices could successfully replace topic oriented vocabulary lists, which i find quite confusing. I find that most topics cause some confusion with their similar words or similar meanings, depending on the topic and language.

So instead of learning all the colours, or a bunch of animals in one go, perhaps it would be better to have a series of stories, little vignettes, each of which uses a collection of words frequently. An adjective, a few nouns and verbs, that sort of thing. Knowing this sort of combination is a lot more immediately productive than a list of rooms in the house or soft drinks, so dialogues can more immediately be explored. Then over time with more slices, the domain gets bigger and bigger.

Using chinese text analyser, i’ve noticed that most books follow the same pattern: about a third of the words are hsk one or two; about a third of the words are hsk three to six; and about a third of the words don’t appear in hsk. Even more crucially, the words from hsk five and six, which themselves make up two thirds of all hsk words, tend to make up less than ten percent of the words in each book. What does this mean? Even if you learn all the hsk words, you’ll still have to look up three words for every ten that you read. That’s enough that reading wouldn’t be comfortable.

That suggests that this slicing approach might be better. Once you know the really common words from hsk one through four, don’t bother memorising the words in five and six unless you really need to take the exam; instead, start reading a book, and learn the words in the book. Learning them in the context of a book is beneficial, and because you’re reading a book that contains them you can be sure that you will see them.