Codex is a new language-learning tool created out of frustration with existing apps. In a landscape dominated by gamified web apps which rely on context-free snippets of text and audio, aspiring language learners find themselves progressing along the app's series of lessons without getting much closer to understanding and speaking a real, living language. Codex is based on real-world content from around the web; news articles, videos, blog posts, and more can be found in Codex's existing content library or imported with the click of a button.
Codex is built on the concept of comprehenbible input-- a theory of language acquisition developed by American linguist Stephen Krashen. Krashen describes the theory in more detail here, but the core principle of comprehensible input is that we acquire language by exposure to material which we can at least partially understand. As we become more familiar with the language, we gradually comprehend more and more as words and phrases become associated not with spellings or verb conjugations but with the actual idea meant to be communicated through language.
Reading and viewing is made easy by Codex's interactive text display. Codex keeps track of users' vocabulary and reading / watching activities in detail; and each word in the target language is highlighted in a certain color to reflect how well the user knows the word. Words that the user has never seen before appear in yellow, seen words appear in shades of blue. The color of a seen word reflects the difficulty rating that the user has assigned to that word.
Clicking on a word opens it in the details panel. For a new word, suggested translations are displayed which, when clicked, will change the word's status to 'seen' and assign it the selected translation. For a seen word, the details panel allows the user to add, edit, and delete translations, and to change the word's difficulty level.
TODO: describe how phrases work
The home screen displays curated content aimed at the user's skill level. Each bit of content has a difficulty rating based on the user's vocabulary data. Content links are accompanied by breif descriptions and lists of tags.
The Codex reader view is designed to resemble the content's original web page as closely as possible. The content page is nested within the reader window with text in the target language with interactive text elements loaded dynamically from the user's vocabulary data.
The video viewer consists of a video player and a synchronized set of captions displayed as interactive text below the frame. Users can also click the arrow button next to a caption and the video player will jump to the beginning of the caption.
So far, Codex is has been tested with X languages:
- English
- Spanish
- French
- German
- Russian
- Turkish
Next for testing:
- Hindi
- Portuguese
- Vietnamese
- Perisan (need script support)
- Indonesian (Bahasa Indonesia)