On Wikisource, it's required to specify the language of a source. For example, this Balinese source (part of work on a current WMF grant) is specified as ban-bali, which means Balinese language in Balinese script.
However, as far as ProofreadPage is concerned, the language field is just one of various index fields that it does not try to interpret. This means that when processing pages in the Page namespace corresponding to an Index page, ProofreadPage does not pass along the language, and UniversalLanguageSelector cannot use it. For example, on this page (from the same source as the above Index page), the text is not marked as ban-bali and so cannot get the webfont specified in UniversalLanguageSelector (font should be Vimala). This is a problem for my work because most users do not have adequate Balinese fonts installed.
Can ProofreadPage be modified so that it recognizes one of the index fields as containing a language code, and passes it on as the language of corresponding pages in the Page namespace? It looks like we could just say that a field in the index data config with "language" in its data field will be treated this way.