Summary of comparative studies on mutual intelligibility, to get a sense of the consistency of our current language fallback settings and find any obviously missing fallback.
Written mutual intelligibility of slavic languages per Lindsay (percentage for written language, in italics otherwise; asterisk denotes formal intelligibility study, bold denotes a likely candidate for language fallback):
- sh -> mk (77 %), sl (60 %)* done sh -> various, gerrit:332178
- sl -> sh (94 %)*, mk (70 %)
- bg -> mk (82 %), uk (80 %)
- mk -> sh (90 %), cs (79 %), bg (75 %), sl (73 %), sk (67 %)
- pl -> be, uk (67 %), ru (60 %)
- hsb <-> dsb (86 %) (gerrit:325693)
- cs -> hsb (100 %), sk (97 %)*, ru (75 %) (done with gerrit:304799)
- sk -> cs (98 %)*, szl (70 %) (done with gerrit:304799)
- ru -> pl (88 %), be (85 %), uk (80 %), bg (74 %), sl (71 %), sk (63 %)
- be -> ru (? %), uk (80 %), pl (67 %)
- uk -> be (75 %), sk (72 %), ru (60 %), bg (60 %)
- sr -> mk (58 %)
The best candidates for a new fallback seem to be cs <-> sk and sl <-> sh; likely, but not proven yet (or not for written language), are hsb <-> dsb, ru <-> be, ru -> uk, uk -> be, sh -> mk. be, sk and sl seem to be the languages which could benefit most from a fallback, given they have millions of speakers but few translations.
The source however needs to be verified... assuming we have interest in confirming any of the above.
- gl->es (MediaWiki messages are fully translated)
- mt->it? (but English is very common)
- fr/ca done: gerrit:329391, gerrit:332544
- rup->ro (gerrit:329506)
- ido->eo (gerrit:324711)
- eu->es (only for geographic reasons)
The relationships between those languages are very non-obvious.
- ki -> sw ???