←back to thread

Learning to Learn

(kevin.the.li)
320 points jklm | 1 comments | 22 Oct 24 00:01 UTC | HN request time: 0.209s | source
Show context
dinobones ◴[22 Oct 24 03:39 UTC] No.41910980[source]▶
I've been wanting to try this approach for learning a language.

In English for example, learning the 800 most common words, you can understand 75% of the language: https://www.bbc.com/news/world-44569277.

I'd love to start fresh on a new language, take 800 new words, try to learn 10 a day, and see where I get after 3 months. Can I really understand 75% of text if I have perfect recall of those 800 words?

replies(9): >>41911047 #>>41911098 #>>41911173 #>>41911321 #>>41911390 #>>41912168 #>>41912979 #>>41913128 #>>41924943 #
mc3301 ◴[22 Oct 24 04:07 UTC] No.41911098[source]▶
Give learning Japanese a try. It's a meta-learning adventure! There are 3 distinct classes of characters (two syllabaries that each have a perfect matching pair with the other, 46 each plus some compounds) and the third are (mostly) chinese Kanji characters. Fun stuff!
replies(1): >>41911121 #
supriyo-biswas ◴[22 Oct 24 04:14 UTC] No.41911121[source]▶
I wonder if there is a similar "Pareto priciple"-esque approach that one could use to learn Japanese.
replies(2): >>41911168 #>>41911340 #
1. joshdavham ◴[22 Oct 24 05:03 UTC] No.41911340[source]▶
Kinda!

The frequency of words in every human language follows the Zipf distribution, which is a power law, like the pareto distribution.

Some learners create what are called frequency lists, which are lists of the most common words, and learn those words first. In general, you get (disproportionately) more bang for your buck from learning the most common words than the rarer words when it comes to understanding.

However, due to the very long tail of word frequency distributions, you eventually need to just start learning words as they come and stop trying to over-optimize with a frequency list.