Language and Translation Issues

This article is a WIP that aims to document issues the game has with language and translation.

Gender and Sex

The game's "gender support" for units merely considers sex rather than grammatical gender; languages that have more than two genders or don't mostly match it with sex are screwed.

We need to somehow implement per-language gender support that would let translators define what gender get associated with each unit type and variant and what genders exist in the language. A language can have different numbers and names for genders, for example, Old English, German, and Latin have masculine, feminine, and neuter, Spanish has masculine and feminine, and Polish has masculine (which is split into animate, inanimate, personal, and non-personal), feminine, and neuter.

And, of course, a different thing can have different genders across languages, and this can be particularly problematic for the undead; for example, in Spanish and Old English, a shadow is feminine but masculine in German. Also, while Portuguese piranha and its Spanish cognate piraña are feminine, piranha and piraña are masculine and invariant in Italian.

There also needs to be a way to let content somehow make use of this support, probably in a way similar to gettext plurals.

Finally, conscious beings that have no biological sex anymore but still have gender identity still need female variants. For example, in Spanish, a lich with a female identity would still say "Estoy cansada" ('I'm tired') rather than "Estoy cansado".

Grammatical Number

Gettext plurals support isn't exposed to WML and Lua, thus preventing content creators from using this feature.

Right now, the lack of support for it in Lua is problematic since the turn counter feature in the objectives dialog requests an arbitrary number; right now, it pretty much just caters to English and most other "Standard Average European" languages that just use the plural with everything above 1. In Scottish Gaelic, for example, to say that there are 12 turns left, the form "chuairt" (not a number word) has to be used, if 20 are left, "cuairt" has to be used, if 19 are left, "cuairtean" has to be used, and so on.

Date, Time, and Collation: Locales

We should stop depending on the operating system's locale for representing time and date. Language-wise, the operating system is the least flexible of the computer's components and most users don't know how to change it. Also, this is a problem for multilingual users or families sharing a computer.

Wesnoth seems to use short date format only, so it would be a matter of translating 7 weekday names and 12 month names for western languages.

We need to research if other cultural areas need a different date format.

Variable Subtitution

Variable subtituion is often (ab)used in translatable strings, and this can be problematic with languages that ... <insert more content>

Egocentric Directions

...

Vertical Text

...

Yes / No

Not all languages treat yes/no the same way, so individual translations each time these words occur are best.

Some languages have a yes/no system, other languages have an aye/nay system, and then there are languages that use the positive/negative form of the verb that was used in the question.

An example of the latter point from Portuguese:

Falas português? - Do you speak Portuguese? - Falo! - I do!

Gostas de Wesnoth? - Do you like Wesnoth? - Gosto! - I do!

Portuguese can circumvent this problem by using the emphatic "Sim!", which people have grown accustomed to in computerspeak. However, other languages like Scottish Gaelic can't.