I still think
that this would be an useful way of formally specifying a
dialect. For example, British English would have ISO639_2 code "en" and
ISO3166_1 code "uk" while Australian English would have ISO639_2 code
"en"
but ISO3166_1 code "au".
Even the ISO-639 codes in the table are there to connect what we are
doing in the Wikipedias and other projects. As it is a standard I added
it but in the database the ISO 639 fields are not compulsory, the "WMF
key" is. If we "need" these ISO639_2 codes, then we would adhere to the
principle that a language is a dialect with an army. Have a look at
http://nl.wiktionary.org/wiki/WikiWoordenboek:Lijst_van_messages#Schrijfwij…
and you will see how we do some of the uk and au stuff for you. This is
however not a great example because it is a mix of different spelling
but also vocabulary and scripts. As I was not content with this I came
up with the current ERD.
Nikola, if you'll look up information on English dialects, you'll find
that the division between British, Australian, American, etc. is all
very exaggerated. The British, Australian, and American standard
languages are all based on the same dialect.
For example, "to starve" is the same in British, Australian, and
American standard English, while in Yorkshire dialect it's "to clem".
Similarly, "mouth", which is the same in British, Australian, and
American English (as far as the standard languages go), is "flep" in
Yorkshire. Incidentally, "flep" also refers to the lips.
This could perhaps be compared to Serbo-Croatian: Serbian, Croatian,
and Bosnian are all based on Stovakian, and there's not much variance
between them; the real variance in Southwest Slavic is between
Stovakian, Cakavian, and Kajkavian. In this example, Australian,
British, and American correspond to Serbian, Croatian, and Bosnian,
while true dialects such as those of Yorkshire, Northumbria, or
Liverpool correspond to Stovakian, Cakavian, and Kajkavian.
Mark