R has been translated into 20 languages but currently not many packages have translations. In a survey of CRAN done last December, of the 8274 packages on CRAN, only 50 had any installed translations. Of those, 28 had only a single translation. As the plot below shows, the number of translated packages is almost indistinguishable from zero.
The RL10N project by myself and the excellent Thomas Leeper has just been funded by the R Consortium in order to address this problem, and help R users who aren’t native English speakers. In short, we want to ASSIST R TO TAKE OVER THE WORLD (of data analysis). Cue evil laugh.
There are three strands to the project:
Firstly, there are tools in the
tools package for working with translations, but they are a bit fiddly to use. Thomas has a work in progress package called
msgtools that aims to make things easier. We’ll develop this package to be robust, well documented, and easy for novice package developers to use.
MTurkR package provides an interface between R and Amazon’s Mechanical Turk API, a marketplace for human intelligence tasks (HITs), including translation tasks. We’ll develop a package that wraps
MTurkR, with functionality for creating and managing translation HITs.
Thirdly, Christopher Lucas and Dustin Tingley’s
translateR package provides an interface to the Application Programming Interfaces (APIs) for the Google Translate and Microsoft Translator services for automated translation of text. We’ll create an R package that wraps
translateR, with functionality for integrating the automated translations into a package development workflow.