Unusual, evidence-based ways to get better at a new language • Hacker News Board

Re: Unusual, evidence-based ways to get better at a new language

#121

Post by Nadya » Wed, Aug 07, 2019, 4:41 PM UTC

JoshTriplett wrote: post #120

zzzcpan wrote: post #117

Earlier quoted context omitted.

Sure. I was watching tv shows with english subtitles and was looking up every word I didn't understand in a dictionary for an hour or more every day. Some days much more, if the show was interesting to me, so it was kind of motivating. At the beginning pausing video and switching between media player and dictionary proved to be a bit too exhausting, so I also wrote a tiny program that did that with a single hotkey be…

> At the beginning pausing video and switching between media player and dictionary proved to be a bit too exhausting, so I also wrote a tiny program that did that with a single hotkey before I really started doing all this. That sounds like a great idea for a service to help learn languages: subtitled videos with links from each word (or idiomatic phrase) in the subtitles to dictionary definitions.

I'll try to find the name of the program - but this already exists and made rounds a while back for Japanese learners where looking up words in a dictionary isn't always as straight forward as it is for latin-based languages.

(a few minutes of sitting on this post later...) And... found it! Or at least, a very similar idea.

https://github.com/juliango202/jijimaku

Re: Unusual, evidence-based ways to get better at a new language

#122

Post by mcguire » Wed, Aug 07, 2019, 4:48 PM UTC

f00zz wrote: post #53

notyourloops wrote: post #23

I learned how to turn a stream of nonsense (from my perspective) into words by listening to hundreds of hours of Spanish TV that I didn't understand. Spanish audio + Spanish subtitles (to figure out how to parse sound, and to map sound to writing), and then Spanish audio only (prevent reliance on reading). Surprisingly effective. Incomprehensible input appears to be quite valuable. At some point I'll try applying my…

Just bear in mind that there's a huge difference between being able to get the gist of something spoken in a foreign language and being able to convey your thoughts effectively in that language. Even after man-years of study I still sound like an idiot when trying to speak English in conference calls, even though I can understand nearly 100% of written and spoken English.

I sound like an idiot when trying to speak English and I'm a native speaker.

I don't know what my conclusion is; maybe speaking is difficult.

Re: Unusual, evidence-based ways to get better at a new language

#123

Post by kradroy » Wed, Aug 07, 2019, 4:52 PM UTC

zzzcpan wrote: post #117

abdusco wrote: post #110

Earlier quoted context omitted.

Can you expand on your "very different methodology"?

Sure. I was watching tv shows with english subtitles and was looking up every word I didn't understand in a dictionary for an hour or more every day. Some days much more, if the show was interesting to me, so it was kind of motivating. At the beginning pausing video and switching between media player and dictionary proved to be a bit too exhausting, so I also wrote a tiny program that did that with a single hotkey be…

I believe it's even better to watch dubbed shows that you already know very well. I greatly increased my spoken German comprehension by watching dubbed episodes of the Simpsons. I had seen most of them a dozen times in English and knew much of the dialogue by heart. It can also help with your pronunciation if you speak along with the dialogue, but you have to be careful you're not mimicking a "substandard" accent. You don't want to end up sounding like Cleetus when talking to business partners.

Re: Unusual, evidence-based ways to get better at a new language

#124

Post by mcguire » Wed, Aug 07, 2019, 5:03 PM UTC

commandlinefan wrote: post #19

mannykannot wrote: post #4

"Don't pay too much attention to grammar." Breaking that rule is one of the ways my high school failed to teach much German to most of its students.

Back in the ‘50s or ‘60s, the U.S. government Foreign Service Institute put together a series of language learning courses that covered most of the popular languages at the time. They’re public domain and if you look hard enough, you can find them on the internet - they’re dated, but they’re still accurate (there are some here http://www.fsi-language-courses.net/ but the hosting sucks). I’ve learned Spanish and Frenc…

Interesting. The text to go along with the audio includes a phonetic version.

    Good morning.  bwenozdiás [Some marks removed.]  Buenos días.

http://www.fsi-language-courses.net/languages/Spanish/Basic/..., Unit 1 - Useful phrases.

Re: Unusual, evidence-based ways to get better at a new language

#125

Post by tzs » Wed, Aug 07, 2019, 5:04 PM UTC

commandlinefan wrote: post #19

mannykannot wrote: post #4

"Don't pay too much attention to grammar." Breaking that rule is one of the ways my high school failed to teach much German to most of its students.

Back in the ‘50s or ‘60s, the U.S. government Foreign Service Institute put together a series of language learning courses that covered most of the popular languages at the time. They’re public domain and if you look hard enough, you can find them on the internet - they’re dated, but they’re still accurate (there are some here http://www.fsi-language-courses.net/ but the hosting sucks). I’ve learned Spanish and Frenc…

You can find much of the FSI material at the Live Lingua Project here [1].

The FSI is not the only US government created language program. There's also the Defense Languages Institute (DLI). Much of their material is also on Live Lingua [2].

[1] https://www.livelingua.com/fsi/

[2] https://www.livelingua.com/dli/

Re: Unusual, evidence-based ways to get better at a new language

#126

Post by overcast » Wed, Aug 07, 2019, 5:08 PM UTC

f00zz wrote: post #53

notyourloops wrote: post #23

I learned how to turn a stream of nonsense (from my perspective) into words by listening to hundreds of hours of Spanish TV that I didn't understand. Spanish audio + Spanish subtitles (to figure out how to parse sound, and to map sound to writing), and then Spanish audio only (prevent reliance on reading). Surprisingly effective. Incomprehensible input appears to be quite valuable. At some point I'll try applying my…

Just bear in mind that there's a huge difference between being able to get the gist of something spoken in a foreign language and being able to convey your thoughts effectively in that language. Even after man-years of study I still sound like an idiot when trying to speak English in conference calls, even though I can understand nearly 100% of written and spoken English.

Unless English is an easier language to understanding all broken up, nearly 100% of the time I can figure out what someone who claims to be "bad" at it is saying. English speakers, particularly in the US seem to be pretty tolerant of it. While other countries seem to be dismissive of anyone attempting their language, as if they can't POSSIBLY understand a single word you're saying.

tldr: I'm sure you sound fine, and we can all understand you.

Re: Unusual, evidence-based ways to get better at a new language

#127

Post by workthrowaway » Wed, Aug 07, 2019, 5:20 PM UTC

klmr wrote: post #9

hwj wrote: post #3

At least for Christians a good way to learn a foreign language is to read the Bible in that language, because (1) they are probably used to read a chapter or two on a daily basis (thus, the routine is already there), (2) and they probably remember some passages in their native tongue (thus, already having a translation of the text). This helped me. YMMV.

This is good advice, but > they are probably used to read a chapter or two on a daily basis This sounds mental. Reading the Bible on a daily basis? I’m pretty sure that this is not the norm for the vast majority of (devout) Christians (at least in Western/Central Europe).

first time i am using a throwaway account lol. but, hey, i do read the bible on a daily basis or at least try to. i don't consider myself the most "devout" tbh. but reading the bible daily is very easy!

i mean everyday readings comes in my email along with a bunch of other newsletters. i read all that stuff every morning, so it is definitely not hard.

Re: Unusual, evidence-based ways to get better at a new language

#128

Post by NoodleIncident » Wed, Aug 07, 2019, 5:26 PM UTC

mannykannot wrote: post #4

"Don't pay too much attention to grammar." Breaking that rule is one of the ways my high school failed to teach much German to most of its students.

I had a similar experience; my high school teacher and/or textbook seemed to only reluctantly introduce new vocabulary, so that they had some words to construct their grammar exercises out of.

Re: Unusual, evidence-based ways to get better at a new language

#129

Post by doubleunplussed » Wed, Aug 07, 2019, 5:32 PM UTC

zzzcpan wrote: post #84

T1glober wrote: post #76

Earlier quoted context omitted.

This makes sense. The act of recalling it is what cements your 'knowledge' of it, and repeating that recollection makes it easier to draw that connection each time.

It's not really how we learn new things. We see, hear, experience them in context many times, not force ourselves to recall them. Forced repetition is one of the worst most inefficient approaches to language learning.

I've found spaced repetition a great way to increase the size of my vocabulary whilst learning a language. You're right that it's unnatural, but a large unnaturally-learned vocabulary can then be polished by practice in more natural contents, and I feel like the net result is more efficient than using the natural contexts alone. This is especially true since I do not have easy access to native speakers for half an hour every day, but I can practice with the Anki app whilst on public transport during that time.

Last time I learned a language (German), a small vocabulary was my perpetual weakness. Now I'm learning a different language (Spanish) with a different approach that makes heavy use of spaced repetition, and vocabulary is no longer my weakest point. Instead, listening comprehension is now my weakness. But with a decent vocabulary, I am now working on improving the listening comprehension using podcasts. If my vocabulary were weaker, this would be much more difficult as I would need access to a native speaker who would considerately restrict themselves to mostly use vocabulary I know.

Re: Unusual, evidence-based ways to get better at a new language

#130

Post by evrydayhustling » Wed, Aug 07, 2019, 5:34 PM UTC

It's interesting that the first two - listen even without comprehension, and deprioritize grammar - correspond significantly to advances in NLP on the last few years. Unsupervised language models (that learn to predict text without any labeled task) turn out to be valuable through transfer learning in specific comprehension tasks. And old school NLP methods that focused on grammar tree construction have fallen behind attention and transformer based methods that learn softer, more abstract representations of structure.