NoaLingua for Chrome
Add to Chrome — Free Add to Chrome

Language

Learning Serbian: what actually works

Serbian is one of the 28 languages NoaLingua offers, and 7 of its ten capability axes are verified from the extension's own registries. This page lists what exists, what is documented as absent, and the specific things about Serbian that do not apply to any other language.

Српски

BCP-47 sr · cyrillic script

Capability by capability

Support for Serbian, capability by capability
Capability State Detail
Chrome on-device translation per pair No en–sr has no Chrome on-device model at all — unavailable, not merely undownloaded
YouTube caption translation Yes Explicitly named as covered by YouTube's translation
Cloud translation (MyMemory) per pair Not documented Opt-in, quota-limited, and the only source that sends sentence text off your device.
Offline word dictionary per pair Not documented FreeDict/WikDict coverage is per pair and not enumerated
Grammar tables (UniMorph) Yes 31 MB · adjectives (424,520), verbs (221,222) and nouns (150,844)
Definitions & IPA (Wiktionary / kaikki) No No kaikki dictionary exists for this language — probed and returned 404
Example sentences (Tatoeba) per pair Yes Real bilingual sentences, searched locally once downloaded.
Speech recognition (Chrome pack) Not documented Chrome owns this pack; availability is Chrome's, not ours
Uncommon-word highlighting No No bundled list, so uncommon-word underlining does not run
Punctuation for captions without it Yes One 56 MB model covering 47 languages, downloaded once per profile.

Specific to Serbian

Things that are only true here

  • There is no Chrome on-device translation model for English–Serbian. It is unavailable, not merely undownloaded — which is precisely why the extension keeps YouTube’s own caption translation as a fallback.
  • Serbian is one of two languages the handbook names as covered by that YouTube fallback, so translation works on YouTube — but not offline.
  • There is no kaikki/Wiktionary dictionary for Serbian at all. It was probed and returns 404, so definitions and IPA come from the online sources or not at all.
  • Serbian grammar comes from UniMorph’s `hbs` Serbo-Croatian macrolanguage dataset — 840,799 rows, unusually adjective-heavy (424,520). It does not distinguish Serbian, Croatian and Bosnian spelling, which is a limitation of the upstream data.
  • Serbian is covered by the punctuation model.
Setting up for Serbian Choose Serbian as the language you are learning in the extension popup or Settings → Languages. Packs begin downloading immediately and in the background — leave the browser open for 15 to 30 minutes and check the corner of the screen. When it goes quiet, everything is local. The full first-run guide →

Common questions

Does NoaLingua work for Serbian?

Serbian is one of the 28 languages in the learning and translation selectors, and 7 of its ten capability axes are verified. The table on this page shows exactly which ones — including the ones that are documented as absent.

Can I get grammar tables for Serbian?

Yes. 31 MB · adjectives (424,520), verbs (221,222) and nouns (150,844) Conjugation and declension tables open at the exact form you clicked.

Does Serbian work offline?

Partly. Example sentences work offline once downloaded, but there is no offline dictionary for Serbian — definitions come from the online sources or not at all.

What does “not documented” mean in the table?

That we have not verified it ourselves, so we will not claim it either way. It is not the same as “no”. Several of those columns depend on data Chrome owns and does not publish per language.