Re: Emacs i18n

[Yuri Khan <yurivkhan@gmail.com>]

> It sounds like Russian has various declensions, like Latin.
> Do I understand right?

Yes, Russian has several declension types for nouns, and several
conjugation types for verbs, and knowing the declension type helps one
deduce the ending for each grammatical case and number, most of the
time.

> Are there standard names or codes for the declensions?

At school, they teach us there are three declensions. First declension
for feminine nouns and a few masculine nouns ending in -а and -я;
second declension for masculine nouns ending in a consonant and neuter
nouns ending in -о or -е; and third declension for feminine nouns
ending in -ь.

This is not the complete picture. There are also non-declined nouns,
nouns that decline as if they were adjectives or pronouns, and
exceptions.

> In Latin, each
> declension has a number.  I think those numbers are standard in
> teaching of Latin, so everyone who can read Latin knows those numbers,
> and translators would find it natural to use a construct that
> specifies the proper declension by number.

The difference is that any translation to Latin would be done by a
linguist, while translation to Russian would be done by ordinary
Russian-speaking people. Most of whom do not keep the scientific
details of declensions in their heads any longer than necessary for
their current study and work.

> We would not have to implement all of them -- only those that
> are useful enough to be worth implementing and documenting.

Right, and pretty much the only part of grammar that is useful for
software UI localization is the grammatical number, because that is
the only thing that changes by circumstances outside the control of
the code. It just does not happen that the same message would put a
noun in different cases, or a verb in different tenses, depending on
the value of some expression.