urdunorm · rule review
We are building a text normalizer for Urdu speech recognition. It decides which spellings of the same word should count as the same word. Every rule below changes how accuracy is measured, so a wrong rule quietly makes the results wrong without breaking anything.
None of this has been checked by a native speaker yet. That is what we are asking you for.
یہ صفحہ اردو املا کے قواعد کی جانچ کے لیے ہے۔ ہر اندراج کو درست، غلط، یا غیر یقینی کے طور پر نشان زد کریں۔ آخر میں اپنے جوابات محفوظ کر کے ہمیں بھیج دیں۔
If two spellings look identical on your screen, open Show characters on that card. Urdu has several pairs that look the same in most fonts and are different letters, and that is exactly what we are trying to get right.
Each rule says: wherever we see the spelling on the left, treat it as the spelling on the right. Two questions for each one. Do people really write the left-hand spelling? And is the right-hand spelling the correct modern one? If the two mean different things in any sentence, the rule is wrong.
Used to write digits out in words. Assume every one is correct unless you say otherwise, so just click the ones that are wrong and type the correction. We are least confident about 50 to 99. Two we would especially like ruled on: is 6 written چھ or چھے, and is 26 چھبیس?
The multipliers. The thing to confirm is that the system is the South Asian one, so lakh is one hundred thousand and crore is ten million.
These are switched off. Each one is either a spelling that collides with a real unrelated word, or a pair where we could not tell which spelling is standard. Mark a rule Correct only if you are confident it is safe to turn on.
Press Save my answers and email us the file. Nothing is uploaded from this page and nothing is sent anywhere; the file goes to your own Downloads folder and stays there until you send it.