Linguist

This is early days for this project, and it has been throwned together in short order as a proof of concept. Everything is subject to change.

linguist is a command-line tool to practise translating sentences between languages.

$ linguist --lang de 
Translate: Hi
> Hallo
✓

Translate: Run
> Lauf
✓

Translate: Wow
> Plotzinger
✘ Potzdonner
  Donnerwetter

Translate: Fire
> Feuer
✓

Translate: Help
> Hilf
✓

Installation

go get -u github.com/KyleBanks/linguist

How It Works

A translation dataset is loaded from the dataset directory, which contains tab-delimited translation pairs from a native language to the language being learned. For example, the most basic samples from the English -> German dataset look like so:

Hi.	Hallo!
Hi.	Grüß Gott!
Run!	Lauf!
Wow!	Potzdonner!
Wow!	Donnerwetter!
Fire!	Feuer!
Help!	Hilfe!
Help!	Zu Hülf!

A sample is picked from the dataset and the user is prompted to enter the translation. A simple Levenshtein Distance measurement is done on the sanitized input and possible translations, which is used to determine if the user entered the correct translation.

TODO

There's a lot of work to make this a useful tool, and contributions are greatly appreciated:

Difficulty scaling
Randomization
Non-binary results (when answer is close enough but not perfect, display the correct response)
Mode variations (Translate from Native, Translate to Native)
Additional language datasets (currently only supports learning German for native English speakers)
Improvements to the game UI (colors, etc.)
Improvements to validations (currently uses a simple/static levenshtein distance comparison. This should at the very least scale based on the sample difficulty)
General code/structure improvements

Acknowledgements

Translation dataset provided by https://tatoeba.org.