Arabic is declared three times, and tagged once
Three of the nine vendors that declare a site language declare Arabic. Two use the bare code and one names Egypt. Separately, the longest voice list in this register carries Arabic as a base entry plus thirteen named regional variants. As of 2026-09-22.
| Vendor | Code as declared | Form of the code | The set it sits in |
|---|---|---|---|
| CapCut | ar-EG | Region-tagged | 27 locales, every one but English region-tagged |
| Kling AI | ar | Bare language code | 20 bare codes |
| SceneMixer | ar | Bare language code | 15 codes, one of them script-tagged |
Inclusion rule. Vendors declaring this language as a same-host hreflang alternate. A vendor serving the language without declaring it cannot be seen by this method and is not listed. Order. Alphabetical by vendor.
1One declaration names a country, two do not
Two vendors declare ar and one declares ar-EG. Egyptian Arabic is the variety with the widest reach across the region through film and television, so naming it is a defensible choice and a specific one.
The bare declarations say only that the pages are in Arabic. Written modern standard Arabic is fairly uniform, so for a marketing site the bare code is less lossy here than it would be for Portuguese or Spanish, where the spoken norms diverge further.
2The variant question moves to the voice column
Where Arabic actually splits is in speech. One vendor here publishes a table in which Arabic appears as a base entry with thirteen named regional variants, from Bahrain to Yemen, which is nearly a tenth of that entire table.
That is the single clearest case in this register of one language behaving like fourteen. A buyer in the Gulf can use that list; a buyer reading a bare count of a hundred and something cannot tell whether their variety is one of the entries or none of them.
3What three declarations settle for a producer
They settle that three vendors expect Arabic-reading buyers and have paid to translate pages for them. Two of those three publish nothing about the language a finished video would speak, so the row stops there.
Arabic also raises a question no vendor in this register addresses: text direction. Nothing published says how a right-to-left script behaves in generated subtitles or in letters drawn inside a picture, and that is a rendering problem rather than a language list problem.
4Sources
Each row was read from the vendor's own markup on 2026-09-22 and is kept in the register; what counts as a declaration is on how counted. The variant question is set out under dialect.
Languages either side of this one in the index: Vietnamese and Chinese. Every language any vendor declares is listed at one language at a time.
- Site languagesEvery alternate but English is a language-region pair; Russian is declared only as ru-BY and Chinese only as zh-TWhow the codes are written
- Site languages20 languages: ar, de, en, es, fil, fr, hi, id, it, ja, ko, ms, nl, pl, pt, ru, th, tr, uk, videclared as same-domain alternates
- Voice languagesArabic appears as a base entry plus 13 named regional variants, from Bahrain to Yemen, inside the same 143-row tableone language's variants inside the list