Obligation: πρέπει / έπρεπε

The frame πρέπει να and its past έπρεπε να are built on the impersonal modal page, which owns the machinery: πρέπει never conjugates, the person rides on the να-verb, and δεν πρέπει να prohibits. Assume all of that. This page is about two things that page leaves on the table, and they are the two that actually trip people up in conversation: πρέπει used to deduce rather than to oblige, and the way the aspect of the verb after να helps shape what kind of obligation you have just handed someone.

One word, two pressures

English speakers already have this split and never notice it. You must wash your hands is a duty. You must be exhausted is a deduction — nobody is ordering you to be exhausted. English uses the same word must for both, and so does Greek with πρέπει να. The technical names are deontic (duty, from the outside world) and epistemic or evidential (the speaker's inference, from the evidence in front of them).

What makes the Greek version worth a page of its own is that the two readings are separated by a signal English does not have: what you put after να — its tense, its aspect, and whether the state it names is one anybody can control. Get that right and the ambiguity mostly evaporates.

The evidential πρέπει: "he must be…"

Point πρέπει να at a state — being tired, knowing something, being at home — and it stops obliging and starts deducing. You are reading evidence and reporting your best conclusion.

Notice what these have in common: their contexts present conclusions from evidence, not instructions. Tiredness and knowing are not readily controllable; in the third example, the unanswered phone supports the inference that she is at work. A state can still be required — Το παιδί πρέπει να είναι κουρασμένο για να κοιμηθεί ("the child has to be tired in order to sleep") presents tiredness as a precondition, not a guess. Controllability and context decide, and a purpose or condition such as για να often signals that a requirement is meant.

A past tense after να tips it to deduction

Here is the cleanest signal in the system. Put a genuinely past verb after να — an aorist, an imperfect — and the sentence is heard as an inference: a requirement cannot reach back and be imposed on an event that is already over, so what is left is a guess.

The aspect keeps working inside the deduction: πρέπει να έφυγε (aorist) guesses at a bounded event, while πρέπει να κοιμόταν (imperfect) guesses at a state viewed from inside as in progress.

One important exception keeps this from being a blanket rule about "past-looking" complements. The perfect subjunctive (να έχει φύγει) has a present auxiliary, and it stays open to both readings — which is exactly how Greek expresses a deadline:

Negating a deduction

Negation is where the two readings stop being tidy, and it is worth being honest about it. Δεν πρέπει να… is read first as a prohibition ("mustn't"), which is the sense the impersonal page documents. In relaxed speech it also does duty as a negated guess ("can't be, surely isn't"), and only the situation tells you which.

If you want the deduction to be unmistakable, put the negation inside the να-clause with μην, or switch to δεν μπορεί να for a flat denial:

θα πρέπει να — the hedged version

Adding θα in front gives θα πρέπει να, which softens whichever reading is in play: a slightly less blunt obligation, or a slightly more tentative deduction. It is completely standard (neutral) and extremely common in news writing and careful speech.

For the other everyday way of guessing — bare θα in front of the verb, with no πρέπει at all (Θα κοιμάται, "he's probably asleep") — see θα for probability. Rough guide: θα is the quick, unelaborated guess; πρέπει να is the guess you have reasoned your way to.

Aspect shapes the kind of obligation

Now the second half of the page. Greek usually forces you to choose a stem after να — perfective or imperfective — and English "must" gives you no clue which. With πρέπει, a bounded instruction strongly favors perfective, while an open policy or habit strongly favors imperfective. This is a contextual default, not an event-count switch: a finite set of required acts can be packaged perfectively, and one required action can be presented imperfectively when its course is in view.

In this pair, the perfective packages one dose as bounded, while the imperfective presents an open repeated policy. The English translations therefore drift toward "do this now" and "you're supposed to / you should be doing this". The time words and situation reinforce those readings; the aspect forms alone do not mean now or habitually.

The same split runs through the prohibition:

Here the first presents an open practice and the second a bounded prohibition. The context selects those viewpoints; λες and πεις do not encode "for life" and "once" by themselves.

Έπρεπε: "had to" and "should have" in one form

The past έπρεπε να is genuinely two-faced, and learners are often taught only half of it. With an ordinary dependent verb after να, the sentence is open between an obligation that was carried out and one that was not:

Depending on what follows, that same Έπρεπε να φύγω στις έξι is either "I had to leave at six (and I did)" or "I should have left at six (and didn't)". Greek does not disambiguate it for you — the continuation does. When you want the regret to be unmistakable, put a pluperfect inside the clause:

The heavier, more layered forms of regret — θα έπρεπε να είχα…, μακάρι να μην…, ας μην είχα… — are gathered on counterfactual and regret.

Common Mistakes

1. Ignoring viewpoint in a standing rule. In this contrast, the open rule favors imperfective and the bounded immediate instruction favors perfective. These are strong contextual readings, not immutable event-count meanings.

2. Reaching for a past verb when you mean a future requirement. Both of these are good Greek, but the tense inside the clause changes the sentence from a deduction into a duty.

3. Negating a guess with δεν and getting a rule. Both are correct; they say different things, and the one you want for "he can't possibly know" is not the δεν πρέπει one.

4. Expecting Έπρεπε να to mean "should have" on its own. It carries both readings; if the point is that it did not happen, spell it out with the pluperfect.

Key Takeaways

• πρέπει να obliges and deduces: Πρέπει να πλύνεις τα χέρια σου is a duty, Πρέπει να είναι κουρασμένος is an inference — exactly the double life of English must.
• The deduction is the default reading when the να-verb names a state nobody can switch on at will (είναι, ξέρει, κοιμάται) — unless that state is a precondition for something else.
• A real past tense after να (έφυγε, κοιμόταν) steers the sentence to the inference reading; the perfect subjunctive (να έχεις τελειώσει) stays open, which is how Greek states a deadline.
• For an unmistakable negated guess, negate inside the clause (πρέπει να μην…) or use δεν μπορεί να…; bare δεν πρέπει να is heard first as a prohibition.
• Aspect shapes the obligation: perfective commonly packages a bounded instruction or delimited set, while imperfective commonly presents an internal or open policy or habit. Context, not event count alone, decides.
• Έπρεπε να is open between "had to (and did)" and "should have (and didn't)"; a pluperfect inside (Έπρεπε να είχα φύγει) makes it regret.