This Arabic root pattern extractor takes Arabic words and recovers their most likely triliteral root (جذر) by reversing the templates (أوزان) that Arabic uses to build words. Arabic morphology is famously systematic: a small root of usually three consonants is poured into a pattern to make nouns, verbs, and participles, so peeling those patterns away gets you back to the shared root that ties a word family together. That is invaluable for dictionary lookup, vocabulary study, and corpus work.
How it works
The tool runs a sequence of morphological rules on each word. First it strips diacritics (tashkeel)
and normalises the hamza-carrying alef forms أ إ آ to plain ا. Then it peels prefixes — the
definite article ال (and its fused forms وال, بال, كال, فال), the conjunctions و and ف, and
the prepositions/markers ب ك ل س — but only when removing them still leaves at least three letters.
Next it removes common suffixes: plural endings ون ين ات ان, the feminine ة, and attached pronouns
ه ها هم هن كم كن نا ي ك.
What remains is the stem. The extractor then deletes the letters that templates typically insert — the
long vowels and hamza ا و ي ء ة first, and if more than three letters still remain, the form-augment
consonants ت م ن س. If a four-letter result contains a doubled (geminate) letter, one copy is dropped.
The first three surviving consonants become the candidate root, displayed as ك-ت-ب style. A word that
lands on exactly three consonants is flagged triliteral (high confidence); anything else is flagged
uncertain.
Example and tips
Feed it a family like الكاتب يكتب كتابا المكتبة مكتوب and every word should collapse toward the same
root ك-ت-ب (“to write”), which is exactly the relationship a learner wants to see. Because the rules
are heuristic, sound roots like this resolve cleanly, while weak roots — say words from ق-و-ل or و-ج-د
where a و or ي is part of the root itself — may show an uncertain flag because the tool cannot tell a
root letter from a template vowel without a dictionary.
Use it to group vocabulary by root before memorising, to guess a dictionary headword, or to pre-process a
word list for further analysis. Treat uncertain rows as prompts to verify manually. Since everything runs
in your browser, your word lists never leave your device.