Guide · Problem

Is the Uzbek in language apps correct? A checklist for spotting bad Uzbek

Quick answer

Sometimes, and you can check it yourself. Reviews of large language apps often complain that smaller languages are machine translated or contain wrong grammar, and Uzbek is small enough to be at risk. A beginner can still spot careless Uzbek: Cyrillic mixed into Latin text, oʻ and gʻ missing or confused with plain o and g, Turkish-looking forms, suffixes that ignore sound changes such as -ga becoming -ka after k, verbs that are not at the end of the sentence, and audio with an English accent. One slip proves little. Several are a reason to change source.

Uzbekcha, launching soon, is built to show the root and every suffix of each word, so you see how a sentence is made instead of memorising it whole.

Launching on iOS and Android. One email when it's ready, and a founding-member offer. No spam.

People who learn smaller languages in large apps often report the same thing in reviews: the content reads as machine translated, or teaches incorrect grammar and wrong vocabulary. Those reviews are mostly about other languages, and we are not saying it of any particular Uzbek course. The risk is real for any small language, though, and Uzbek has features that make careless text easy to spot, even for a beginner.

We make Uzbekcha, which is not available yet. It is mentioned once, at the end.

Key points

  • You do not need to know Uzbek to test Uzbek. Script, spelling, suffix changes and word order can all be checked by rule.
  • The quickest checks: no Cyrillic in a Latin course, and kept apart from o and g, and the verb at the end of the sentence.
  • One slip proves little, and careful products make them too. A pattern of slips means the text was not checked by a native speaker.
  • When in doubt, compare with a second source or ask a person.

Why a small language is at risk

An app that offers dozens of languages usually writes its course once and then produces it in every language. For a large language there are many editors and many users to report mistakes. For a smaller one the translation may be automatic or done quickly, and errors can stay for years because few learners know enough to notice them. Uzbek adds two difficulties of its own: it changed script within living memory, and most of its grammar sits in suffixes, which automatic translation handles badly.

The checklist

1. One script only

Latin is the official script. A course that teaches Latin should contain no Cyrillic, except as a labelled reference. Cyrillic letters inside Latin words, or whole lines in Cyrillic with no explanation, mean the text was assembled from mixed sources.

2. The letters oʻ and gʻ

and are separate letters, not decorated versions of o and g. Ot is horse and oʻt is fire or grass. The language is oʻzbek, not “ozbek”. If a course drops the mark, it is teaching different words.

Officially the mark is a modifier letter (ʻ), and the tutuq belgisi in words such as san’at (art) and ma’no (meaning) is a right single quotation mark (’). In everyday typing both come out as a straight apostrophe (o', g'), and everyone copes, so a straight apostrophe alone is not an error. What to look for in a teaching product is consistency. Three different marks on one screen, or marks that come and go, point to text that nobody edited.

3. Turkish or Kazakh forms

Uzbek is related to Turkish and Kazakh, and material for those languages is easier to find, so their forms sometimes leak in. Two signs:

  • Suffixes that change their vowel. Standard Uzbek has largely dropped vowel harmony, so each suffix has a single form. The plural is always -lar: kitoblar (books). A plural that switches between two forms, the way Turkish -lar and -ler do, is not standard Uzbek.
  • Turkish vowels in common words. Turkish a is often Uzbek o: Turkish baş (head) is Uzbek bosh. The letters ö, ü, ş, ç and ı are Turkish. Uzbek writes , u, sh and ch.

4. Suffix sound changes

Correct Uzbek adjusts some forms to what comes before them. Two that are easy to check:

  • -ga (to) becomes -ka after k and -qa after q: Samarqandga, but eshikka (to the door) and qishloqqa (to the village). People will understand you if you get this wrong in speech, but written teaching material should get it right.
  • After a number the noun stays singular and the number takes -ta: ikkita non (two breads). Ikki nonlar is wrong.

A course that glues suffixes on without these changes was probably generated, not written.

5. Word order

The verb comes last. “I drink tea” is Men choy ichaman, literally “I tea drink”. Question words stay in place with the verb or predicate still at the end: Metro bekati qayerda? (where is the metro station?). Uzbek sentences that follow English order word for word are a sign of word-for-word translation.

6. Suffix order

On a noun the order is fixed: root, plural, possessive, case. Kitob-lar-im-da is “in my books”. Any other order is wrong.

7. The audio

An English-style accent, stress on the first syllable (RAH-mat for rah-MAT) or x read as “ks” means a text-to-speech voice built for another language. There is a full test in why some Uzbek audio sounds wrong.

One mistake is not a verdict

Every source slips sometimes, this one included. A single odd spelling on a page tells you very little; a pattern across several lessons tells you a lot. Judge a course on the pattern, and check anything important against a second source.

What to do when something looks wrong

  1. Check the item against a second source: a textbook, a tutor or a native speaker.
  2. Report it if the app has a way to do that.
  3. If slips are frequent, stop taking spelling and grammar from that source.

Where Uzbekcha fits

Uzbekcha is an AI tutor for Uzbek that we are building in Tashkent. It is not available yet, and there is only a waitlist. Large language models are shaky at Uzbek morphology on their own, so generated Uzbek in Uzbekcha passes through a morphological validator that checks suffix order, sound changes at suffix boundaries and Latin-script consistency, and it is grounded in verified grammar and dictionary sources.

Questions

Is the Uzbek in language apps machine translated?

It can be. Machine translation of smaller languages is a recurring complaint in reviews of large multi-language apps, though that does not tell you anything about one particular app's Uzbek. Read what the product says about who wrote and recorded its content, then run the checks in this guide on a few lessons.

How can I tell if Uzbek text is wrong when I am a beginner?

Look for things that need no vocabulary: Cyrillic letters in a Latin-script course, oʻ and gʻ written as plain o and g, plural nouns after numbers, -ga left unchanged after k or q, and sentences where the verb is not last. Each check takes seconds.

Is it wrong to write Uzbek with a normal apostrophe?

In everyday typing, no. The letters oʻ and gʻ officially use a modifier letter (ʻ) and the tutuq belgisi uses a right single quotation mark (’), but on a phone keyboard both come out as a straight apostrophe and everyone copes. In teaching material, inconsistent marks suggest unedited text, and a missing mark changes the word.

Does Uzbek have vowel harmony like Turkish?

Standard Uzbek has largely dropped it, so each suffix has a single form. If a course shows Uzbek suffixes changing their vowel to match the word, the way Turkish endings do, it is not showing you standard Uzbek.

What should I do if I find mistakes in my app?

Check the item against a second source, such as a textbook, a tutor or a native speaker. Report it to the developer if the app allows that. If mistakes are frequent, keep the app for listening practice at most and take your grammar and spelling from a source you trust.

Part of How to learn Uzbek: the complete guide, which maps every page on this site to the stage of learning it belongs to.

Early access

Learn Uzbek in your own language.

Uzbekcha generates every lesson for you: your language, your reason for being here, your level, with audio on every word. Join the waitlist and hear first.