Currently the language selection boxes have:
1) Language, clarification/variant, namely:
"English, Old (ca.450-1100)"
"Frisian, Eastern"
"Frisian, Northern"
"Frisian, Western"
"German, Low"
"German, Swiss"
"Greek, Ancient"
"Khmer, Central"
"Ndebele, North"
"Ndebele, South"
"Sami, Inari"
"Sami, Lule"
"Sami, Northern"
"Sami, Skolt"
"Sami, Southern"
"Sorbian, Lower"
"Sorbian, Upper"
"Turkish, Ottoman"
+ several historic variants of German, French, Irish, etc., and 4 entries for "Creoles and Pidgins", all with zero-frequency.

2) Just language, namely:
"Chinese"
"Mandarin Chinese"
"Min Nan Chinese"
"Yue Chinese"
"Syriac"
"Classical Syriac"
(and possibly some more)

I understand that naming them like in "1)" is meant to ease finding historic or other variants, but from aesthetic point of view it looks a bit weird on release pages, etc.; and from i18n point of view, they are not very good to translate, if this "findability" is to be preserved – at least not without the end result looking even weirder than in English (e.g. in Estonian, names like "vanainglise" 'Old English' would become "Inglise: vanainglise", "Inglise, vana-" or such).

If the translators choose to translate them all like the latter group (so that they would look nice), then the finding advantage is gone. (The German l10n team has done this, and I intend to do this for Estonian.)

Possible solutions:
a) add a language search box, like there is for instruments, and rename "English, Old", etc. into "Old English, etc.
b) display "Old English", "Swiss German", "Yue Chinese", etc. with indent under "English", "German", "Chinese", etc.
c) both