[HN Gopher] What makes comprehensible input comprehensible?
___________________________________________________________________
What makes comprehensible input comprehensible?
Author : surprisetalk
Score : 31 points
Date : 2025-06-22 18:51 UTC (4 days ago)
(HTM) web link (cij-analysis.streamlit.app)
(TXT) w3m dump (cij-analysis.streamlit.app)
| joshdavham wrote:
| Oh wow! I'm surprised to see someone post my analysis haha
|
| Happy to answer any questions here. I kept my analysis really
| high level for a general audience but since this is HN, we can
| get a bit nerdy :D
| flippyhead wrote:
| I love this. I made a totally free, just for fun, tool based
| around learning Japanese via Youtube using the CI approach.
| https://seikai.tv The trick is finding content that is at the
| right level but that you also find interesting. Great article,
| thank you!
| joshdavham wrote:
| Thanks for the kind words!
| ragazzina wrote:
| > Word length - At least in English and French (the languages I
| know best), longer words are generally considered harder.
|
| I think in a language with a lot of similar sounds or even
| homophones, longer words are easier. For a beginner Chinese
| speaker that knows both words, hearing "che" will probably be
| ambiguous, but "chuzuche" will be parsed immediately.
| joshdavham wrote:
| That's a good point.
|
| I don't think the 'longer equals harder' pattern holds for
| every language. I actually reached out to the head teacher at
| CIJ when I first made this analysis and she said the same.
| EdiX wrote:
| I don't think this captures the whole situation. Much of what
| makes comprehensible input comprehensible, at lower levels, is
| presence of visual hints.
| joshdavham wrote:
| That's exactly right.
|
| Much of the beginner videos make use of visual hints like you
| say (images, props, etc), and none of these were taken into
| account in my analysis.
|
| I do think it could be cool to do a 'visual' analysis of CI in
| the future where you attempt to measure how much context is
| present (or not) in each video and see what insights you could
| draw from that.
| joshdavham wrote:
| Here's the source code for this analysis to those interested:
| https://github.com/joshdavham/cij-analysis
|
| I will note that the transcripts (and parsing scripts) are not
| included in the repo. The transcripts are not my intellectual
| property so I can't share it (and the parsing scripts are a bit
| of a dumpster fire).
___________________________________________________________________
(page generated 2025-06-26 23:02 UTC)