Persian stress sits on the last syllable of the word, almost always, which is one of the friendliest rules in this pillar. The exceptions are what matter, because they are not scattered: verbal prefixes take the stress off the ending and onto themselves, so the negative and the imperfective, which between them cover a very large share of all the verbs you will ever say, are stressed at the front.
A learner who has learned the rule and not the exception is putting the stress in the wrong place on most of their verbs, in every sentence, and being understood the whole time.
Six things to put in your mouth
- six vowels, three short and three long in origin, and nothing like the eleven of English. This makes Persian pronunciation genuinely easy for an English speaker.
- gh and q are one sound in modern Tehrani speech, made at the back of the throat, although they are written with two different letters.
- kh is the ch in Bach, and it appears in extremely common words including khub, good, and khoda, God.
- the stress is usually on the last syllable of a word, with the notable exception of verb prefixes and a few particles, where it moves forward.
- there is no w sound. The letter that looks like an Arabic waw is a v, so varzesh begins with a v.
- long and short are no longer distinguished by duration in modern speech, though the vowels remain distinct in quality, which is why older grammars describe a system that no longer matches what you hear.
One rule, and the prefixes that override it
On nouns and adjectives the rule holds with very few exceptions and can be trusted. On verbs it holds only for the bare forms. Add the prefix that marks a habitual or continuous action and the stress moves to that prefix. Add the negative and the stress moves there instead, and the negative wins over the other prefix when both are present.
That is a small closed system and it is learnable in an evening, and its effect is not small: a verb is the last thing in a Persian clause, so putting the stress in the wrong place there is putting it in the wrong place at the point of the sentence a listener is most attentive to. It also happens to be the piece a text to speech voice most often gets wrong, which makes machine audio an unreliable guide on exactly the words this page is about.
The other thing the spelling hides is the ezafe, the short e that binds a noun to its describer, and it is not stressed. So a noun phrase has one stressed syllable at the end and a run of unstressed joins before it, and hearing that shape is what teaches the eye to put the joins back on the page.
Applying the rule faithfully and being wrong on every verb
Here is the stall, and it is caused by the rule being good. Because final stress is nearly always right, a learner applies it everywhere with confidence and never revisits it. The verbs come out with the stress on the ending, which is a real Persian pattern belonging to a different form, so the result is not noise: it is a word that sounds like something else and is repaired by context.
Nothing corrects it. The exit is a small deliberate list rather than a general improvement: take twenty verbs, write each in the plain, the imperfective and the negative, mark the stressed syllable in each, and say them aloud in sets. An evening, and the pattern is installed. Getting somebody to catch the ones that still come out wrong is what a paid hour is for, in what a teacher is worth, and what you can do alone.
One verb, three forms
Where the stress lands as the prefixes go on
Each frame is the same verb. Nothing about the ending changed, and the stressed syllable moved twice.
- 01The bare past formStress on the last syllable, exactly as the rule says
- 02With the imperfective prefixStress moves forward onto the prefix
- 03With the negativeStress moves again, onto the negative, which outranks the other prefix
- 04In a full clauseAnd the verb is last, so the wrong stress lands where a listener is most attentive
The fourth frame is why this matters more than a stress rule usually would. Persian puts the verb at the end of the clause, so an error here is an error at the point of the sentence that carries the most weight.
Two ways to check where the stress landed
A Persian text to speech voice is the quick check and it has to be used with its limitation in view. It reads a sentence aloud and places the stress correctly on ordinary vocabulary, and it is unreliable on exactly the prefixed verb forms this page is about. It replaced needing a speaker present to model a sentence you invented rather than a word somebody recorded. Free in the operating system and the browser. Use it on nouns and phrases, and do not trust it on a negative verb.
A verb list with the stress marked on the prefixed forms is the thing to build, because almost no material marks it. It replaced discovering the pattern accidentally after a year, which is the usual route. Free, and twenty verbs written out in three forms each is the whole of it: the point is not the twenty verbs but the pattern they install, which then applies to every verb you meet afterwards.
The mistake: reading the ezafe as a syllable of its own
Told that an unwritten short vowel binds the words of a noun phrase, a learner gives it weight, and the phrase comes out as a row of separate words each with its own stress. That is the opposite of what the ezafe does: it is unstressed by definition, and its job is to make several words behave as one phrase with a single prominence at the end.
Hearing it is the fix. In speech the joins are perfectly audible and perfectly light, and a month of listening puts them back on the page without any conscious work. Which is why listening early is not optional in Persian in the way it can be elsewhere, and why the reading problem in thirty two letters, joined, with the vowels left for you to supply is solved by the ear rather than by the eye.
What the sounds are attached to
The prefixes that move the stress are the ones marking negation and aspect, so the stress rule and the verb system are the same subject seen twice, in no gender, no cases, and one of the tidiest verb systems anywhere. Several of those forms contract further in ordinary speech, systematically, which is two registers, and why the book does not sound like the street. The words carrying the stress come from two layers with different behaviour, in the Arabic layer, and the words English quietly took. And hearing all of it at conversational speed, with the vowels reduced, is Tehrani speech, and the vowels it swallows.