Traditional approach
One Markdown file for each language. This is the most commonly used approach to Markdown translations. For example, site generator plugins like jekyll-multiple-languages-plugin and mkdocs-static-i18n-plugin use it.
Advantages
Contextualization: The translator sees the entire file in the translation process, which can provide a better understanding of what is translating.
Programming vagueness: The programmer don’t need to deal with complicated translations workflows. He/she can sort files by translation directories or whatever other simple layout. The use of plain files limits the responsibility of the programmer.
Disadvantages
Markup errors: Translators have to learn Markdown language before start working in the translations because they need to work with Markdown markup directly.
The problem of dealing with Markdown markup can be prevented using editors that support the syntax, but these editors are not the ones translators are used to and you, as a programmer can’t impose the usage of certain tools to people that even don’t speak your language. You better think that they are creating the translations in a notepad using Windows 95.
This frequently causes failures in markup characters which breaks styles, so in the long run they must be revised anyway (sometimes in strange languages) or markup errors go unnoticed for a long time.
Desynchronization: A change in one file (the source language frequently) requires a manual review of all other files (translations).
This is a tremendous effort for the translators and therefore an incredible expenditure of money for companies. Think about it, the translators have to manually review each Markdown file, a format they don’t have to be used to, looking for these parts that have been updated… It just doesn’t make sense.
Redundancy: There are parts of Markdown files that don’t need to be translated, like code blocks. You can tell to the translators that they must ignore content inside blocks of text wich starts and ends with triple backticks (
`
), but at the end of the day, you will have blocks of code translated. These blocks, as well as other parts of the source file, must be hidden to the translator.