Implement system `bgnpcgn-ara-arab-latn-1956` (BGN/PCGN Romanization System -- Arabic (1956)

[ronaldtse]

This issue is to implement the transliteration system of bgnpcgn-ara-arab-latn-1956.

This system is referred in the GeoNames database as ara_Arab2Latn_BGN_1956, with the system title 'BGN/PCGN Romanization System -- Arabic (1956)'.

Tests should rely on the data extracted for the ara_Arab2Latn_BGN_1956 system in https://github.com/riboseinc/geonames-transliteration-data .

[ronaldtse]

@manuel489 Would you be able to help encode this map? Spec located here: Thanks!

https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/858000/ROMANIZATION_OF_ARABIC.pdf

Ping @nbh707

[manuelfuenmayor]

Ok @ronaldtse

[manuelfuenmayor]

Hi @ronaldtse,
Apparently, there is an inconsistency between the romanization character : th and its Unicode value: 0073+0304.

Using an unicode converter like: https://www.branah.com/unicode-converter

For \u0073\u0304, I get the character s̄ instead of th. (The unicode value for th is \u0074\u0068.)

In this case, should I use the unicode value (\u0073\u0304) for mapping or the romanization character (th)?

[ronaldtse]

I think it is th, and this is an error in the BGN/PCGN document.

In these columns, notice that th and dh both have ...+0304. However, kh and sh have ...+0068.

So I checked another transliteration document (https://transliteration.eki.ee/pdf/Arabic_2.2.pdf), and it seems that 062B is always transliterated as a t, th or a modified t.

This makes me think that it ...+0304 is an error (for both th and dh).

[ronaldtse]

Actually, I stand corrected. It is possible that instead of th they meant s̄ and instead of dh they meant z̄. Let me clarify with them.

[ronaldtse]

Under "Special Rules" at the end, it says:

When the definite article precedes a word beginning with one of the “sun letters” t, th, d, dh, r, z, s, sh, ş, ḑ, ţ, z,̧ l, or n

So this alludes to th and dh...

[manuelfuenmayor]

Yeah, in addition, if we look at the provided example:

We can see that the phrase only contains th (instead of s̄). Which makes me think that th is the correct character.

Oh, and talking about examples. In the example columns, there are "pointed" script and "unpointed" script in Arabic language:

Which one should I use for encoding the tests?

[ronaldtse]

@manuel489 we have to use the "Pointed Script". "Unpointed Script" omits short vowels so transliteration by mapping isn't possible.

[manuelfuenmayor]

@manuel489 we have to use the "Pointed Script". "Unpointed Script" omits short vowels so transliteration by mapping isn't possible.

Ok, understood.

Map added in #236

[nbh707]

th is correct. This is the sound like "th" in "the"

[nbh707]

In some dialects it is pronounced like "t" so that would account for the instances of "t"

[nbh707]

As for pointed text, there are a few entries in the BGN file that do have the vowels written, or some of the vowels written. But the majority do not. So we need to use the transliteration in the BGN file to determine what vowels are missing in the original Arabic.