Stress Shifts Inside a Verb Paradigm

Hebrew stress is מלרע — on the last syllable — for the overwhelming bulk of the vocabulary, and the general rules treat everything else as a set of small exceptions. The verb is the biggest of those pockets, and the only one where the stress moves within a single word's paradigm. Conjugate one verb through its past tense and the beat shifts twice.

This matters more than an accent usually does, because Hebrew stress is contrastive. בנו is bánu "in us" or banú "they built" — that pair is worked through on stress minimal pairs and is not repeated here. What this page does is walk the shifts you meet inside the paradigms themselves.

The past tense of a strong verb

Why it retracts at all

The split is not arbitrary, and knowing the reason makes it stick. The third-person endings ־ָה and ־וּ are old inflectional endings: they were part of the word when the stress was assigned, so the stress landed on them. The first- and second-person endings are historically pronoun fragments — ־תִּי, ־תָּ and ־נוּ are worn-down relatives of אני, אתה and אנחנו — and they were attached to a word whose beat was already fixed. Nothing came along afterwards to move it.

This is the same story as the segolate nouns, where the second vowel is a later insertion and the stress stayed where it was. In both cases the rule is: material added late does not attract the beat. The one first- or second-person ending that breaks it is ־תֶּם, which is stressed in the normative reading — and that is precisely the cell Israelis have since pulled back into line, which is the next section.

The one cell where the books and the street disagree

The normative stress on the second-person plural past is on the ending: כְּתַבְתֶּם ktavtém, הֲלַכְתֶּם halakhtém, אֲמַרְתֶּם amartém. Nearly every Israeli says katávtem, halákhtem, amártem — restoring the stem vowel (הָלַכְתֶּם, אָמַרְתֶּם) and pulling the beat back to match the rest of the first- and second-person forms.

This is not sloppiness and it is not new; it is the paradigm regularising itself, and it is what you will hear in ordinary speech at every level of education. The normative reading survives in careful reading aloud and in liturgy.

The present: one form pulls back

Every present form is מלרע except one — and the exception is a shape you already know from nouns.

The feminine singular in ־ֶת ends in a segolate shape, and segolates are stressed early — the same reason סֵפֶר is séfer. It holds across the binyanim: מְדַבֶּרֶת medabéret, מַסְבִּירָה is masbirá because it has no ־ֶת, but מִתְלַבֶּשֶׁת is mitlabéshet and יוֹשֶׁבֶת is yoshévet.

Hollow verbs: the beat carries the tense

This is the shift you will meet on your first day in Israel, and it is worth getting the direction right, because it inverts easily in the memory.

In verbs whose middle root letter is ו or י — בא, קם, גר, שר, טס — the feminine singular is spelled identically in the past and the present. The past is מלעיל and the present is מלרע.

The plurals behave the same way in the past — bá'u, kámu, gáru, sháru are all מלעיל — but there they cause no confusion, because the present plural is spelled differently (באו against באים).

Why the split? The present is the participle, and its feminine ־ָה is a stressed adjectival ending like the one in טוֹבָה tová and גְּדוֹלָה gdolá. The hollow past has only one stem vowel and keeps the beat on it, exactly as it did in Biblical Hebrew. So the two forms are not a coincidence of Israeli pronunciation; they were never the same word.

שרה is not a clean minimal pair

You will see שרה offered as a three-way stress puzzle. It is worth being exact about it, because two of the three readings differ in more than the beat.

Unpointed שרה has three readings: the name שָׂרָה, the past שָׁרָה shára, and the present שָׁרָה shará. Only the last two are a genuine stress minimal pair — same consonants, same vowels, different beat. The name opens with שׂ /s/, not שׁ /sh/, so it differs in a consonant; unpointed writing hides that, which makes שרה a homograph, not a minimal pair. And the name's own stress is unstable: normatively sará, colloquially Sára, so it does not even sit on one side of the contrast.

Almost everything else is מלרע

Once past the pockets above, the verb system is nearly uniform.

• Every infinitive is final-stressed, without exception: לִכְתּוֹב likhtóv, לָבוֹא lavó, לְדַבֵּר ledabér, לְהִסְתַּכֵּל lehistakél.
• A future or an imperative with no vowel ending is final-stressed: אֶכְתּוֹב ekhtóv, יִכְתּוֹב yikhtóv, נַסְבִּיר nasbír, יָבוֹא yavó, תָּקוּם takúm, כְּתוֹב ktov.
• Add the ־י or the ־וּ ending and the beat normally moves onto it: תִּכְתְּבִי tikhteví, יִכְתְּבוּ yikhtevú, יְדַבְּרוּ yedabrú, כִּתְבִי kitví, כִּתְבוּ kitvú.

"Normally" is carrying weight in that third line, and the exception is large enough to learn on its own. Where the stem itself holds a long i or u — that is, in hif'il and in the hollow verbs — the stem vowel keeps the beat and the ending never takes it. And it is not a fact about the future alone: the same long stem vowel holds the beat in the past too, so hif'il gives hitkhíla, hisbíra, hirgísha, hevína where pa'al gives katvá, and the hollow verbs give gára, káma, bá'u. The one place it lets go is the present participle, whose ־ָה is a stressed adjectival ending: matkhilá, masbirá, mazminá. So within hif'il the present and the future run opposite to each other.

That is why בּוֹאִי and קוּמִי sound nothing like כִּתְבִי, and why bo'í for bó'i gives a learner away instantly — בואי being one of the first words anyone will say to you.

One more מלעיל future form exists and can be filed away at once: the ־נָה feminine plural, תִּכְתּוֹבְנָה tikhtóvna, which is correct but rare in speech.

Common Mistakes

1. Pushing the first-person past to the end. The ־תי ending never takes the beat.

2. Inverting the hollow-verb pair. This one changes the tense, not the accent.

3. Stressing the feminine present on the ending. ־ֶת is a segolate ending and pulls the beat back.

4. Carrying the English beat into the infinitive. Infinitives are final-stressed without exception.

Key Takeaways

• In the past tense of the strong pa'al verb, the third-person endings ־ָה and ־וּ are stressed (katvá, katvú) while the first- and second-person endings are not (katávti, katávta, katávnu). Where the stem holds a long vowel the ending never takes the beat — hif'il hitkhíla, hisbíra, and the hollow verbs gára, káma, bá'u.
• The second-person plural is the one contested cell: normative ktavtém, everyday Israeli katávtem — and the same retraction on הלכתם, אמרתם.
• In the present, only the feminine singular in ־ֶת is מלעיל: kotévet, medabéret, mitlabéshet. The other three forms are מלרע.
• Hollow verbs put the tense in the beat: past bá'a, káma, gára, shára are מלעיל; present ba'á, kamá, gará, shará are מלרע. Get the direction right — the participle is the final-stressed one.
• שרה is a homograph, not a minimal pair: the name is written with שׂ and differs in a consonant, and its own stress varies between sará and Sára.
• Infinitives are מלרע without exception, and so are futures and imperatives with no vowel ending. Add ־י or ־וּ and the ending normally takes the beat (tikhteví, yikhtevú) — except in hif'il and the hollow verbs, where the stem vowel keeps it: tagídi, yagídu, bó'i, yavó'u.