6 ms·

Foreign Dispatch: Ideas from code editors applied to foreign language writing

67 points by ichverstehe 4y ago

geza 4y ago
Note you might be interested in looking at interactive machine translation systems and the associated research papers - for example Lilt [1, 2] or TransSmart [3, 4] - while those are designed for professional translators rather than for language learners, what they do is very similar to your long-term aims for this project (a specialized MT system that, given a partially written translation to the foreign language and some context written in in the user's native language, suggests the following word). [1] https://lilt.com/technology/translate https://lilt.com/technology/translate [2] https://lilt.com/research https://lilt.com/research [3] https://transmart.qq.com/ https://transmart.qq.com/ [4] https://arxiv.org/pdf/2105.13072.pdf https://arxiv.org/pdf/2105.13072.pdf
- ichverstehe 4y ago
  Very interesting, I will look into that. Thank you!
luminouslow 4y ago
Love the idea! Are you planning to keep the core tool stack OSS? Would love to contribute to growing the project in the long term. I imagine there could be avenues for monetization while keeping it OSS (e.g. Licence for business use)
- ichverstehe 4y ago
  No plans to monetize. Contributions very welcome! :)
zote 4y ago
Wonderful idea, Viel Glück.
turtleyacht 4y ago
Great idea. I wanted to use pmd to learn some subtle aspects of Java. Tool-assisted learning. It's either operate on existing code and learn the conventions and techniques, or write hundreds of programs in the language. From a maintenance perspective, it's a superpower.
- thaumasiotes 4y ago
  I thought the idea sounded interesting, but it appears to be very narrowly focused on the problem of "I'd like to write something in [German], but I don't know [German] all that well". I'm more interested in the problem of "I'd like to read something in [German]". The "right way" to solve the problem of knowing how to write something in German is, unfortunately, to read several dozen examples of people writing similar things. Dictionary entries are complex things. (Well, often they aren't, but they should be. The amount of information one dictionary entry needs to provide is quite large.) I don't see a lot of value in having the editor pop up abbreviated dictionary entries for words that I know I need to look up. If I know that, I can look up the word in a high-quality dictionary and benefit from a much fuller entry.
  - turtleyacht 4y ago
    That's a good approach: one should read many examples in German to learn writing it that way, whether fiction, or news, or mathematics. Foreign Dispatch looks to eventually be an ensemble of tools. Autocomplete being the first, with one-for-one analogs of programming: linter, in-line documentation, and language server. Going back to your writing concept, how cool would it be to pop up snippets of German literature for similar sentences in context? That would be neat.
  - visarga 4y ago
    > I'm more interested in the problem of "I'd like to read something in [German]" Something can be done here. Books with aligned/inlined/popup translations, tracking known/unknown vocabulary, spaced repetition for new words.
    - 4y ago
llanowarelves 4y ago
I had this idea (and way down the list on things i planned executing on) but I'm very glad you're doing it so I can enjoy it without having to
red_trumpet 4y ago
I think it should be "Ich ziehe Sauerteig vor." (or in past tense, "Ich zog Sauerteig vor"). "Zuge" is not a word[1]. So yes, a kind of spelling+grammar linter would be great I guess! [1] https://www.duden.de/suchen/dudenonline/zuge https://www.duden.de/suchen/dudenonline/zuge
- ichverstehe 4y ago
  I mixed it up with bevorzugen[1] I think. Will fix it, thanks! :) https://en.wiktionary.org/wiki/bevorzugen https://en.wiktionary.org/wiki/bevorzugen
remix2000 4y ago
This is very similar to Vim's built-in "thesaurus completions": https://vimhelp.org/insert.txt.html#compl-thesaurus https://vimhelp.org/insert.txt.html#compl-thesaurus

67 points by ichverstehe 4y ago

geza 4y ago

Note you might be interested in looking at interactive machine translation systems and the associated research papers - for example Lilt [1, 2] or TransSmart [3, 4] - while those are designed for professional translators rather than for language learners, what they do is very similar to your long-term aims for this project (a specialized MT system that, given a partially written translation to the foreign language and some context written in in the user's native language, suggests the following word). [1] https://lilt.com/technology/translate https://lilt.com/technology/translate [2] https://lilt.com/research https://lilt.com/research [3] https://transmart.qq.com/ https://transmart.qq.com/ [4] https://arxiv.org/pdf/2105.13072.pdf https://arxiv.org/pdf/2105.13072.pdf

ichverstehe 4y ago
Very interesting, I will look into that. Thank you!

luminouslow 4y ago

Love the idea! Are you planning to keep the core tool stack OSS? Would love to contribute to growing the project in the long term. I imagine there could be avenues for monetization while keeping it OSS (e.g. Licence for business use)

ichverstehe 4y ago
No plans to monetize. Contributions very welcome! :)

zote 4y ago

Wonderful idea, Viel Glück.

turtleyacht 4y ago

Great idea. I wanted to use pmd to learn some subtle aspects of Java. Tool-assisted learning. It's either operate on existing code and learn the conventions and techniques, or write hundreds of programs in the language. From a maintenance perspective, it's a superpower.

thaumasiotes 4y ago
I thought the idea sounded interesting, but it appears to be very narrowly focused on the problem of "I'd like to write something in [German], but I don't know [German] all that well". I'm more interested in the problem of "I'd like to read something in [German]". The "right way" to solve the problem of knowing how to write something in German is, unfortunately, to read several dozen examples of people writing similar things. Dictionary entries are complex things. (Well, often they aren't, but they should be. The amount of information one dictionary entry needs to provide is quite large.) I don't see a lot of value in having the editor pop up abbreviated dictionary entries for words that I know I need to look up. If I know that, I can look up the word in a high-quality dictionary and benefit from a much fuller entry.
- turtleyacht 4y ago
  That's a good approach: one should read many examples in German to learn writing it that way, whether fiction, or news, or mathematics. Foreign Dispatch looks to eventually be an ensemble of tools. Autocomplete being the first, with one-for-one analogs of programming: linter, in-line documentation, and language server. Going back to your writing concept, how cool would it be to pop up snippets of German literature for similar sentences in context? That would be neat.
- visarga 4y ago
  > I'm more interested in the problem of "I'd like to read something in [German]" Something can be done here. Books with aligned/inlined/popup translations, tracking known/unknown vocabulary, spaced repetition for new words.
  - 4y ago

llanowarelves 4y ago

I had this idea (and way down the list on things i planned executing on) but I'm very glad you're doing it so I can enjoy it without having to

red_trumpet 4y ago

I think it should be "Ich ziehe Sauerteig vor." (or in past tense, "Ich zog Sauerteig vor"). "Zuge" is not a word[1]. So yes, a kind of spelling+grammar linter would be great I guess! [1] https://www.duden.de/suchen/dudenonline/zuge https://www.duden.de/suchen/dudenonline/zuge

ichverstehe 4y ago
I mixed it up with bevorzugen[1] I think. Will fix it, thanks! :) https://en.wiktionary.org/wiki/bevorzugen https://en.wiktionary.org/wiki/bevorzugen

remix2000 4y ago

This is very similar to Vim's built-in "thesaurus completions": https://vimhelp.org/insert.txt.html#compl-thesaurus https://vimhelp.org/insert.txt.html#compl-thesaurus