נפעל or התפעל?

Hebrew has two binyanim whose subject is not doing anything to anybody else, and a learner meeting a new verb has to know which one is in front of them. This page is the procedure: the question to ask, the pairs to practise it on, and — more usefully — the two places where the question gives you the wrong answer. The inventory of roots that fill both slots, and the productivity evidence behind guessing, are on where nif'al and hitpa'el overlap. The prefix reshuffling that hides a hitpa'el under a shape you would not look for (הִשְׁתַּמֵּר, הִצְטַלֵּם) is on metathesis, and the full range of ways Hebrew says himself is on reflexive strategies.

The question

Did an outside agent do it, or did it come from the subject?

• Outside agent — somebody or something else did it, and the subject is on the receiving end → nif'al. It is the binyan that pairs with pa'al: כָּתַב katáv → נִכְתַּב nikhtáv.
• From the subject — it did it to itself, they did it to each other, or it happened from within → hitpa'el. It is the binyan that pairs with pi'el: גִּילֵּחַ giléakh → הִתְגַּלֵּחַ hitgaléakh.

Same root, same three letters, and the two sentences are about different things entirely: in the first, a person wrote and the letter was written at; in the second, nothing comes in from outside — the two of them are doing it to each other.

Four live pairs

These are the pairs to practise the question on, each one checked in Milog before it was printed here. Note that the hitpa'el gloss is never a mechanical do-it-to-yourself version of the nif'al gloss.

Row two is the one to look at hardest. התפתח does not mean "opened itself." Milog's definition is עבר תהליך הדרגתי של גדילה — went through a gradual process of growth — with a fourth sense, הקצין; החריף, for a situation getting worse. Nothing in "the subject did it to itself" would have produced that from open. Row four is worse still, and the rest of this page is about why.

Where the question gives the wrong answer

Two large classes break it, and knowing their names is worth more than the question itself.

A great many nif'al verbs are not passives at all. נִמְצָא, נִשְׁאַר, נִכְנַס, נִרְדַּם, נִשְׁבַּר describe a state or a change of state with nobody doing anything. Ask "who did it to him?" of נשארתי and there is no answer, because the question does not apply. This is the middle voice, and it has its own page: נשבר, נפתח, נסגר.

A great many hitpa'el verbs are not reflexives. התעייף, התבגר, התאהב, התייאש, התרגש are all things that happen to you, and none of them involves doing anything to yourself. This is the inchoative class, and it is the one hitpa'el has taken over: rule of thumb, an unfamiliar "become-ish" verb is far likelier to be hitpa'el than nif'al, because hitpa'el is the productive intransitive binyan while pa'al and nif'al are essentially closed (Bat-El 2013, Brill EHLL 704–707; Bolozky 1978, 2007).

So the honest scope of the question is this: it sorts a pair you are already looking at; it does not tell you which binyan a meaning lives in. For production, look the verb up.

One pair, measured: נרגש against התרגש

ר־ג־שׁ is the pair where the semantic question fails completely — both members mean got excited, was moved, and neither has an outside agent. What separates them turns out not to be meaning at all but which cells of the paradigm each one occupies. Counting Hebrew Wikipedia's article namespace with insource:/…/ on 16 August 2026, each figure run twice with the same result:

That zero is only worth printing because the controls were run. First-person nif'al pasts are perfectly at home in this corpus — נכנסתי returns 258 and נשארתי 173 — and the nif'al past of this root is not extinct either: נרגשו returns 31 and נרגשה 3. It is the first person that is empty. Talking about your own past excitement, Hebrew reaches for the hitpa'el; describing a crowd's present state, it reaches about equally for either, with the nif'al slightly ahead.

Common Mistakes

1. Expecting an agent behind every nif'al. English was broken invites by whom, so learners bolt על ידי onto verbs that have no agent slot. Both sentences below are correct — the point is how differently they behave. In the same corpus, על ידי follows the true passives in bulk (נכתב על ידי and נבנה על ידי are five-figure and four-figure floors respectively) and follows the intransitives essentially never: נכנס, נשאר, נרדם and נעלם return seven hits between the four of them.

2. Picking the wrong member of a live pair. Both members exist, so nothing looks wrong; the sentence simply says something else.

3. Giving a nif'al a direct object. English enter the room, leave the building, pass the exam are transitive; the Hebrew verbs that translate them are not.

4. Assuming the other half of the pair exists. For most roots only one of the two binyanim is filled, and the gap is the normal case rather than something to be repaired. Neither sentence below has a partner in the other binyan, and neither needs one.

Key Takeaways

• The question: outside agent → nif'al; from the subject, to itself, or to each other → hitpa'el. נכתב against התכתב is the clean case.
• The hitpa'el of a pair is never a mechanical reflexive of the nif'al. התפתח is develop, not open itself; התכתב is correspond, and in criticism engage with.
• Two classes break the question. Nif'al verbs like נמצא, נשאר, נכנס, נרדם are intransitive statives, not passives; hitpa'el verbs like התעייף, התבגר, התאהב are inchoatives, not reflexives.
• The question sorts a pair in front of you; it does not generate verbs. To produce, look the verb up — and if you must guess, guess hitpa'el, which is the productive intransitive binyan while pa'al and nif'al are essentially closed.
• נרגש and התרגש gloss identically, and what separates them is the paradigm: נרגשתי returns 0 against 83 for התרגשתי, while the plural present splits 147 to 113. Ask which forms are used, not which one means what.
• Do not bolt על ידי onto an intransitive nif'al. It follows the real passives in bulk and the intransitives essentially never.