You have just met a Hebrew word and you want to know what it means. In English this is one action. In Hebrew it is two, and they fail in different ways: finding the entry, and then choosing between the entries you find. A dictionary solves the first problem; only you can solve the second.
The good news is that a failed lookup in Hebrew is almost never a missing word. It is nearly always an extraction failure — you looked for something the dictionary never listed.
Root-ordered or word-ordered? The tradition is split
There is a widespread belief that Hebrew dictionaries are "organised by root". Half of them are, and the half a learner would expect is often the wrong half.
Root-ordered works are the older tradition: Yonah Ibn Janah's eleventh-century כתאב אלאצול, "the book of roots", the Biblical lexicons that scholars still use (Brown–Driver–Briggs, Gesenius), and the Biblical lexicons scholars still use (Brown–Driver–Briggs, Gesenius). In these, every derived word sits under its three consonants: מִכְתָּב, כְּתוֹבֶת and הִתְכַּתְּבוּת are all under כ־ת־ב and nowhere near mem. Klein's etymological dictionary, free to read on Sefaria, is the hybrid worth knowing: the verbs are the root entries (ספר ᴵ, ספר ᴵᴵ), but each derived noun also gets its own alphabetical entry — מִכְתָּב sits under mem and הִתְכַּתְּבוּת under he, each pointing back at the root.
And crucially for a modern learner, מילון אבן־שושן — the standard Israeli desk dictionary, and one of the two used in the Hebrew matriculation exam — belongs to this camp for verbs. Its own principle, as Hebrew Wikipedia states it, is that you have to know the root of a verb in order to find it.
Word-ordered dictionaries list every word where it is spelled, and their founding figure is the one people assume was a root purist. Eliezer Ben-Yehuda deliberately broke with root order in מילון הלשון העברית הישנה והחדשה. The reason he gave is a good one: earlier dictionaries filed אֶגְרוֹף "fist" under ג־ר־ף, and he knew that very few of his readers could work that out. רב־מילים and its sister site Morfix follow the same principle, and add a morphological analyser on top.
| Reference | Ordered by | What that means for you |
|---|---|---|
| Klein, BDB, Ibn Janah | root | you must extract the root first |
| אבן־שושן | root, for verbs | לָמַד, לִימֵּד and הִתְלַמֵּד sit together under ל־מ־ד |
| בן־יהודה | alphabetically, by word | a deliberate break with root order |
| רב־מילים, מורפיקס | alphabetically, plus an analyser | type the form as you met it |
המילון הזה מסודר לפי שורשים, לא לפי אותיות.
hamilón hazé mesudár lefí shorashím, lo lefí otiyót
This dictionary is arranged by root, not alphabetically.
Verbs are where root knowledge is unavoidable
Nouns you can usually find as written. Verbs you cannot, because the form you met is almost never the form that is listed. Hebrew dictionaries cite a verb in the third person masculine singular past — כָּתַב, not "to write" — so a lookup has two steps: get back to that citation form, then get its root. The convention itself is set out on the citation form.
Three worked recoveries:
התחלנו. Strip the ־נוּ ending and you have הִתְחִיל. That is a hif'il, so the ה is the binyan's prefix and not a root letter, which leaves ת־ח־ל. Klein files it exactly there and marks the root post-biblical; the biblical verb for "begin" is הֵחֵל, from ח־ל־ל, so ת־ח־ל is a later root grown out of an older noun — Klein calls it a secondary base formed from תְּחִלָּה "beginning", itself the verbal noun of חלל ᴵᴵ. Even a "correct" root can turn out to be younger than the words it explains.
הבנתי. Again a hif'il ה, again not a radical. The citation form is הֵבִין, and the root is ב־י־ן — with a yod you will not see in most of the paradigm, because it is a hollow root. This is the recovery that defeats learners most often.
נפגשנו. The נ is the nif'al prefix. Citation form נִפְגַּשׁ, root פ־ג־שׁ.
התחלנו ללמוד עברית רק בספטמבר.
hitkhálnu lilmód ivrít rak beseptémber
We only started learning Hebrew in September.
לא הבנתי מה הוא רצה ממני.
lo hevánti ma hu ratsá miméni
I didn't understand what he wanted from me.
נפגשנו במקרה בתחנה המרכזית.
nifgáshnu bemikré batakhaná hamerkazít
We ran into each other at the central bus station.
Let the phone do the boring half
Be honest about this rather than romantic: Morfix, Pealim and רב־מילים all accept an inflected form exactly as you met it and hand back the base entry with a full morphological analysis. Rav-Milim's own documentation gives וּכְשֶׁתִּקְּחוּהוּ as a demonstration — an absurd pile of prefixes and suffixes — and the analyser returns the entry לקח with the parse spelled out.
Do not spend six months drilling a skill a phone performs instantly. Type the word in.
הקלדתי את המילה בטלפון וקיבלתי את השורש.
hikládti et hamilá batélefon vekibálti et hashóresh
I typed the word into my phone and got the root straight back.
Where the analyser stops and you start
The analyser's answer is a list, not an answer. It cannot read the sentence, so whenever a written form fits more than one root it offers you every candidate and steps back.
שכתבתי is the clean case. Two entirely different parses share that spelling:
- שֶׁכָּתַבְתִּי shekatávti — the relative ש plus כָּתַבְתִּי, root כ־ת־ב: "that I wrote".
- שִׁכְתַּבְתִּי shikhtávti — a four-letter root שׁ־כ־ת־ב "to rewrite", built out of ש + כתב and now a verb in its own right: "I rewrote".
המכתב שכתבתי לך חזר אליי.
hamikhtáv shekatávti lekhá khazár eláy
The letter I wrote you came back to me.
שכתבתי את כל הפרק מהתחלה.
shikhtávti et kol hapérek mehatkhalá
I rewrote the whole chapter from scratch.
Nothing on the page distinguishes them. The first has a definite noun in front of it that needs a relative clause; the second is the main verb of its sentence and takes את. The syntax decides, and only a reader has the syntax.
The same happens inside a single root. הִתְעָרַבְתִּי sits under ע־ר־ב, but the dictionary lists ע־ר־ב as several numbered roots, and two of them produce this exact form: ערב ᴵᴵ "to pledge, go surety" gives "I bet", ערב ᴵᴵᴵ "to mix" gives "I got mixed up in it".
התערבתי איתו על מאה שקל שהוא לא יגיע.
hit'arávti itó al me'á shékel shehú lo yagía
I bet him a hundred shekels he wouldn't turn up.
לא התערבתי בעניין הזה בכלל.
lo hit'arávti ba'inyán hazé bikhlál
I didn't get involved in that business at all.
Common Mistakes
1. Counting a binyan prefix as a root letter. The נ of a nif'al and the ה of a hif'il belong to the pattern.
❌ שורש נ־פ־ג
Incorrect for נפגשנו — the נ is the nif'al prefix. Strip it and the ־נו ending and the root is פ־ג־שׁ.
✅ נפגשנו אתמול אחרי העבודה.
nifgáshnu etmól akharéi ha'avodá
We met up yesterday after work.
2. Taking the letters you can see and stopping there. A hollow root hides its middle letter in most forms.
❌ שורש ה־ב־נ
Incorrect for הבנתי — the ה is a hif'il prefix and the root is ב־י־ן, whose yod is invisible here. Under ה you will find nothing.
✅ עכשיו הבנתי למה הוא כעס.
akhsháv hevánti láma hu ka'ás
Now I get why he was annoyed.
3. Accepting the analyser's first candidate. When two roots share a form, the tool lists both and the sentence chooses.
❌ המפעל מחזר
the factory is courting
Incorrect — that reading is מְחַזֵּר from ח־ז־ר. With a factory as the subject it is מִחְזֵר, from the four-letter root מ־ח־ז־ר, 'recycled'.
✅ המפעל מחזר את כל הפלסטיק.
hamif'ál mikhzér et kol haplástik
The factory recycled all the plastic.
4. Reading the first numbered entry and moving on. ס־פ־ר is two roots, and the barber is not in the first one.
❌ הלכתי להסתפר
I went to tell a story
Incorrect — telling is ספר ᴵ, but הסתפר belongs to ספר ᴵᴵ, 'to cut'. It means 'to get a haircut'.
✅ הלכתי להסתפר לפני החתונה.
halákhti lehistapér lifnéi hakhatuná
I went for a haircut before the wedding.
Key Takeaways
- A Hebrew lookup is two problems: finding the entry and choosing between entries. Tools solve the first; only the sentence solves the second.
- The reference tradition is split, and not the way people assume: Klein, BDB and אבן־שושן file verbs by root, while בן־יהודה went alphabetical on purpose and רב־מילים and מורפיקס follow him.
- Verbs are cited in the 3rd person masculine singular past, so you must recover that form before you can recover the root: התחלנו → הִתְחִיל → ת־ח־ל; הבנתי → הֵבִין → ב־י־ן; נפגשנו → נִפְגַּשׁ → פ־ג־שׁ.
- Pattern letters are never radicals: the מ, נ, ת, ה and א that patterns add, and the plene ו and י that only mark vowels.
- Use the analyser. Morfix, Pealim and רב־מילים take an inflected form as typed. Then read its list critically — שכתבתי is both שֶׁכָּתַבְתִּי and שִׁכְתַּבְתִּי.
- Numbered entries (ספר ᴵ, ספר ᴵᴵ) mark separate roots, not shades of one meaning.
- A failed lookup is nearly always an extraction failure, not a missing word.
Now practice Hebrew
Reading grammar gets you part of the way. The exercises are where it sticks — free, no signup needed.
Start learning Hebrew →Related Topics
- Triliteral Roots and How to Extract OneA2 — Almost every Hebrew word stands on three consonants — but not the first three you see. Learn the eleven servile letters, strip the mould off the word, and test your answer against the rest of the family.
- Reading Unpointed Hebrew: Supplying the Vowels YourselfA2 — Unpointed Hebrew is not pointed Hebrew with the dots rubbed out — it is read by a different method, in which the reader supplies the vowels from the word, the pattern and the sentence. Here is that method, step by step.
- The Dictionary Form Is the Past-Tense הואA2 — Hebrew names a verb by its third-person masculine singular past — כתב, literally 'he wrote' — because that form carries no prefix and no suffix. Here is why, which references follow the convention and which do not, and the step-by-step procedure for stripping a form you met back to the form you can look up.
- Same Letters, Different Root: ס־פ־ר and FriendsC1 — Some Hebrew roots are not one root at all. ס־פ־ר is two, ע־ר־ב is five, and the reasons are ordinary history — a borrowing from Aramaic in one case, a sound change that let Proto-Semitic consonants collapse together in the other — not a hidden thread of meaning waiting to be uncovered.