Yes getting more traction for Wikisource would be useful, particularly for non-English texts. The ability to show multiple languages side by side is an excellent way of transcribing and translating texts, however one that is rarely used by anyone. I would be surprised if this attracted many new people who would stay on and become regular wikisourcerers.
Having 'been around' for quite a while, dabbled in Wikisource and lurked around its back passages, I still find it comparatively hard to understand. If this is to attract newcomers, then it would be nice to see this go hand-in-hand with improving both the guidelines on exactly how to proofread (there's a complex multi-stage process that could do with a simpler work-flow), the peculiarities of how text is marked-up there and the rather convoluted underpinning process for turning a document/book into a djvu file, loading it on Commons and then setting it up as a book on Wikisource (phew). I'm fairly wizardly but I found the "norms" hard to work out and arbitrary.
I agree with Charles' point about low-hanging fruit. With ancient text transcriptions falling into disrepair (as University IT departments cut back) there is significant educational value in publishing transcriptions of Latin and ancient Greek inscriptions, however hardly any are on Wikisource, as this is much harder than transcribing a page from a 19th century journal. Having talked to a couple of academics about this area, I personally would not recommend Wikisource to any historians over a custom solution at the moment, mainly due to its poor interface, lack of standards for transcriptions (e.g. how do you mark up "this letter is likely to be a delta" or "this word is missing from the original" apart from making generic custom pop up notes?) and general clunkiness, which is a great pity.
Fae