Accent Marks denoting Stress in Greek, and Other Diacritics

This page is part of the author’s set of pages on the Greek language

The Greek alphabet page explained the Greek letters, and in a separate paragraph it mentioned that there are some vowel digraphs in Greek: vowels that are put together and pronounced either as a single vowel sound (such as αι, ει, οι, υι, and ου), or as a combination of vowel-consonant sounds (such as αυ, ευ, and ηυ). This knowledge is a prerequisite to understand how stress is marked (written) in Greek. I’ll use “stress” to refer to the raising of the voice in speech, and “accent mark(s)” to refer to the marks placed over letters in writing, which denote stress or other features of the spoken language.

Here is a general, diachronic, and absolute rule regarding stress in Greek:

Only the last three syllables of a word can be stressed in Greek
Any apparent violation of this rule is an optical illusion! (more on this below). In Ancient Greek this rule was not exactly about syllables, but morae, a concept that will be discussed in the Ancient Greek section.

Do I need to explain what a syllable is? It’s a small chunk of a word, almost always containing a vowel or diphthong. For example, in English, the word “photography” includes four syllables: pho·to·gra·phy; the word “straight” has only one syllable (so it is called “monosyllabic”), because it contains a single vowel sound, the diphthong [ei]. Similarly, in Greek, the word σκουληκομυρμηγκότρυπα (never mind what that means(1)) has eight syllables: σκου·λη·κο·μυρ·μη·γκό·τρυ·πα.

Okay, so now you know: if a Greek word has a million syllables, you only need to worry about the last three to figure out where to put the stress mark. But where to put it? On which of the last three syllables? Unfortunately, at this point you can grab your crystal ball and try to divine the answer, because, unlike French, which stresses almost always the last syllable (the “ultima”), and unlike Spanish, which stresses very often the second-last syllable (the “penult”), Greek stress is more-or-less unpredictable. You have to know the sound of the word to know how to stress it. But if you know the sound, this page explains how and when to place the stress mark on the word, in Modern Greek, and Ancient Greek. And if you don’t already know the sound (stress pattern) of words, this page will explain why you see the accent marks placed as they are.

Modern Greek

In Modern Greek, since 1982, there is a single mark that shows where the stress goes, and it looks like a tiny slanted line over the vowel, in most fonts. Here is a normal (well, I mean more normal than σκουληκομυρμηγκότρυπα) Greek word: Αθηνά (the ancient Greek goddess of wisdom, akin to Minerva of the Romans). You can see the accent mark over the last vowel of the trisyllabic Α·θη·νά. But if we place the accent mark over another vowel, we make another word: Αθήνα, the capital of Greece.

Prior to 1982, the rules for Ancient Greek were applied also to Modern Greek, and it was an absolute mess, because there were five different marks in Ancient Greek, and the rules for placing them, although meaningful in ancient times, were meaningless in later times, because the pronunciation of the language had changed and wasn’t indicative anymore of which mark to use. Children had to memorize the stress for each word, and, besides allocating precious memory resources to a worthless system, they were making plenty of mistakes in writing. Fortunately, the system was simplified greatly: now there is a single accent mark, as I said. But first, we must learn where to put the accent mark, before we learn when.

Where to put the accent mark

I said the accent mark goes only over vowels. But there are three categories of vowels that receive stress in Greek:
  • Plain vowels,
  • Vowel digraphs, and
  • Diphthongs

Plain vowels are just what you imagine. There are seven of them in the alphabet: α, ε, η, ι, ο, υ, ω, and their stressed versions are: ά, έ, ή, ί, ό, ύ, ώ. Not much to mention here, except that when words are written in CAPITALS, the stress is not marked. But there is a case where capital letters can get an accent mark in writing: it is when the capitalized first letter of a phrase which is otherwise in lowercase, is stressed. For example:

Άσε με να φύγω : Let me go
Όμηρος : Homer

The capital vowels with stress look like this: Ά, Έ, Ή, Ί, Ό, Ύ, Ώ. As you see, the accent mark is written on the upper-left side of the capital letter, and does not point to it, but has the same look and tilt as the mark for the lowercase vowels.

Vowel digraphs are another story. When they are stressed, the accent mark is placed over the second letter of the digraph. Examples:

φαίνεται : it appears
είδα : I saw
φακοί : lenses
καθεστηκυία : established, prevailing [fem.]
καρπούζι : watermelon
αύριο : tomorrow
εύχομαι : I wish
απηύδησα : I got fed up

The last three examples, which correspond to each of the three vowel–consonant digraphs, αυ, ευ, and ηυ, might appear strange: although in speech what is stressed is — of course — the vowel, i.e., the first of the two constituents of the digraph, strangely the accent mark is shown over the second constituent, which happens to be pronounced as a consonant ([v] in αύριο and απηύδησα, and [f] in εύχομαι). But this is not without an explanation. If the accent mark were placed over the first letter (the vowel), then we wouldn’t know that this is a vowel digraph, and would pronounce each of its constituents separately, stressing the first vowel, and pronouncing the second letter also as a vowel (at its “face value”), making a second syllable with it. Indeed, this is exactly what happens with the word άυλος (“immaterial” [masc.]), which is made of three syllables: ά·υ·λος (pronounced [á·i·los]); compare it with φαύλος (“lacking judgment; paltry, trivial”), a disyllabic word: φαύ·λος (pronounced [fáv·los]). If we were to write *φάυλος, that would be pronounced [fá·i·los], a non-word in Greek.

Now, you might wonder, isn’t there any case where one must write the two letters of these digraphs (αι, ει, etc.), and pronounce them separately, stressing the second vowel? This last point is important, because if the first vowel is stressed then we have a case like άυλος: no problem, there is no digraph there. But what if the second vowel must be stressed, and the vowels must be pronounced separately? How do we write this?

For that case, the special mark called diaeresis (διαλυτικά: ¨ ) is employed: it is placed over the second vowel, thus “dismissing” the digraph. The diaeresis is still sometimes used in English, over the i of the word naïve, and for the same reason: showing the special pronunciation of the vowels of this word. In Greek, the diaeresis coexists with the stress mark over the second vowel:

παΐδι : rib
σεΐχης : sheik
προΐσταμαι : I act as superior [e.g., in a company]
ευφυΐα : wit, intelligence
προΰπαρξη : pre-existence
καταπραΰνω : I soothe
φορμαλδεΰδη : formaldehyde [a chemical compound, H2CO]

But don’t imagine that the diaeresis in Greek always goes with an accent mark! Quite the contrary: their co-occurrence is very rare. Much more often, the diaeresis is needed to “dismiss” an unstressed pair of letters that otherwise would look like a digraph:

μαϊδανός : parsley
θεϊκό : divine [adj. neut.]
κοροϊδεύω : I trick, I mock
μυϊκό : muscular [= of muscles]
προϋπάρχω : I pre-exist
αϋπνία : insomnia
ξεϋφαίνω : I un-weave

Diphthongs, finally, are very rare in Modern Greek. They do exist, but one has to search hard to find them. Caveat: if you read texts written by older Greek philologists, you might notice that what they call “diphthongs” in Greek are actually the digraphs, discussed above. This is so because the digraphs were actually true diphthongs in Ancient (e.g., Attic) Greek, and because the ancient grammarians were referring to them as “diphthongs”, this term prevailed and was in use in Modern Greek until fairly recently.

Now, what exactly is a diphthong? It’s a vowel with two constituent qualities, and it makes up the core of a single syllable. For example, in English, the “oy” in “boy” is a diphthong, pronounced [oi], making up a single syllable: boy. (It’s not bo·y.) Similarly, the “ow” in “tow” [toυ], and the single “o” in “go” [goυ], denote the same diphthong, [oυ], although native English speakers hardly recognize it as such, i.e., it is hard for the majority of English speakers to realize that the “o” in “go” has two constituents ([o] and [υ]), that it’s not a single, simple vowel. There are many other diphthongs in English. For readers who know Spanish, a diphthong is a more transparent notion because Spanish diphthongs are always written with two vowels, and always consistently, i.e., in one, specific way for each diphthong (e.g., lluvia, tiempo, dios, agua, puesto, monstruo, muy, estoy, aire, etc.). What about Greek? Modern Greek diphthongs are not as transparent as the Spanish ones, and they pass unnoticed by the native speakers. The latter can only sense there is something fishy in their language if you force them to face the “donkey conundrum”:

“Dear native speaker of Greek, what is the word for donkey in your wonderful language?”

“Why, γάιδαρος, of course.”

“And how many vowels do you count after the stressed ά, including it?”

“I count four: the stressed ά, an ι, another α, and an ο.”

“Exactly, four. So, if each vowel corresponds to a syllable, on which syllable does this word appear to be stressed?”


Thus the Greek speaker will remain speechless, having encountered a monster-word (in the form of a donkey), which seemingly defies the God-approved rule that no word can be stressed beyond the 3rd syllable from the end (see the blue box at the top of this page). Most likely, your Greek interlocutor will fall dumbstruck on the ground, paralyzed, unable to believe that his/her language violates the Divine Rule of Greek phonology. But you can help the fellow recover by pointing out that, actually, the Divine Rule is not violated, because the άι in γάιδαρος is indeed an example of a rare true Modern Greek diphthong. For this reason, the same fellow cannot break up this word between the stressed ά and the ι if the need arises to break it up across two lines in writing, but must respect the boundaries of its three syllables: γάι·δα·ρος. (Notice that a diaeresis is not needed over the iota in this word, thus not *γάϊδαρος, because the accent mark over the first alpha tells us that α and ι are distinct, they don’t form the digraph αι.) Here are some other examples of diphthongs in Greek, where I show the syllabification of the word explicitly:

νε·ράι·δα : fairy
αη·δό·νι : nightingale
πεϊ·νιρ·λί : [an oval-shaped snack with bread and cheese, of Turkish origin]
μποϊ·κο·τά·ρω : I boycott
φράου·λα : strawberry

Sometimes it’s not clear whether we have a diphthong or not: some Greeks might syllabify the last two examples as μπο·ϊ·κο·τά·ρω and φρά·ου·λα, respectively. (This becomes evident if we see words such as φράουλα in poems, where the metric convention might imply either the one, or the other syllabification.) Also, the first two examples show that the same diphthong can be spelled in different ways. But the accent-placing rules, which concern us here, are quite transparent:

  • If the diphthong is stressed, then its first constituent letter will receive the accent mark, and a diaeresis is not needed.
  • If the diphthong is not stressed, then a diaeresis is needed only when the written form of the diphthong resembles one of the digraphs.

In practice, you don’t need to know which vowel combinations are pronounced as diphthongs in Modern Greek. The writing will not reveal the information to you, but you don’t need it anyway. Pronouncing a diphthong as consisting of separate vowels (making up separate syllables) is not an error in Greek, and hardly anyone will notice (OK, with the exception of the donkey-monster, of course: you can’t pronounce γά·ϊ·δα·ρος and walk away as if nothing happend).

When to put the accent mark

The “when” in this section refers to which words show their stress explicitly with an accent mark in Modern Greek. There is an overarching rule that answers this question:
No accent mark is placed over monosyllabic (single-syllable) words in Greek.
The accent mark is shown only on the stressed syllable of words with two or more syllables.
It is obvious why the accent mark is not shown over monosyllabic words: because it would be redundant. But most rules gain in notoriety when they have lots of exceptions, right? Right: the above rule has lots of exceptions, in both of its sentences, and this is what the rest of this section explains.

Exceptions where monosyllabic words do receive an accent mark

There are some “little words” in Greek that appear the same (i.e., spelled with the same letters), but are actually different words. In every one of such word-pairs, one of the two words in the pair is stressed in actual speech, when spoken within a sentence, whereas the other is unstressed. Although monosyllabic, we mark the stressed little word with an accent, to distinguish it from the other one, which we leave unmarked (as it should be). It’s that simple. Examples:
  • The που – πού pair:
    The first word (που) is one of those words that, grammatically, has a multiple personality disorder: it can be a pronoun, an adverb, or a conjunctive, but you’ll save all the grammatical confusion if you remember that in English it’s usually translated as “that”, as in “the car that passed”. Did you notice that (<— yes, that that) this word is not stressed in English, either? (“the cAr that pAssed”.) Same in Greek: this kind of που is not stressed in a sentence, and is not marked with an accent in writing. Examples:
    τον είδα που έφυγε : I saw him leaving [lit.: that he left]
    το αυτοκίνητο που πέρασε : the car that passed
    καλά που τους απέφυγα : good thing that I avoided them
    But there is this other πού, the interrogative adverb meaning “where”, which is always stressed, and it can be used either in questions, or in references to questions:
    πες μου, πού ήσουν; : tell me, where were you?
    πες μου πού ήσουν! : tell me where you were!
    δεν ξέρω πού ήσουν : I don’t know where you were

  • The πως – πώς pair:
    This one exactly parallels the previous pair. The conjunctive πως, meaning “that” (again, but in different cases than που), is not stressed, consequently not accented:
    σε προειδοποίησα πως καπνίζω : I warned you that I smoke
    πιστεύω πως θα τα καταφέρω : I believe that I’ll make it
    But the interrogative πώς, meaning “how”, is always stressed, both in direct questions and in indirect references to questions:
    πώς τα πας; : how are you doing?
    δεν ξέρω πώς να σ’ το πω : I don’t know how to say it to you

  • The η – ή pair:
    The first η, the unstressed one, is the feminine form of the article, in the nominative case, singular number:
    η Ελένη : Helen
    η φωτιά : the fire
    The second ή is the disjunctive particle meaning “either, or”:
    ή εγώ ή εσύ : either I or you
    μπούτι ή στήθος; : leg [lit.: thigh] or breast? [of chicken, OK? ]

  • The να – νά pair:
    The first να, the unstressed one, is one of the most common words in Greek (I think it’s either second or third in frequency). It is the particle that introduces the so-called subjunctive mood of verbs, which the English speaker usually perceives as the replacement of the absent infinitive in Greek. Examples:
    θέλω να φύγω : I want to leave
    μπορώ να πω κάτι; : may I say something?
    The other νά is quite rare, and it means “here!”, or “there!”. Examples:
    νά ο Γιάννης! : there is John!
    νά τα χρήματα που σου χρωστώ : here is the money I owe you
    The second νά is so rare that many (probably most) writers completely ignore it, leaving it unstressed.

  • The τι – τί pair:
    Here, the first (unstressed) τι is the very rare one, so rare that it can be said it’s encountered only in local dialects of Greek, or in some literary texts. It means “for”, in the sense of “because”:
    τι από δική τους χάθηκαν, οι κούφιοι, αμυαλωσύνη : for they perished through their own sheer folly
    τι δεν απέθανε στη γης ο μέγας Οδυσσέας : for the great Ulysses did not die on earth
    (Both of the above examples are excerpts from The Odyssey, Homer’s epic poem — above translation by Argýris Eftaliótis.)
    The other τί is the interrogative adverb or pronoun, meaning “what”:
    τί είπες; : what did you say?
    δεν σου είπα τί έπαθα! : I didn’t tell you what happened to me! [meaning: let me tell you what happened to me!]
    The first τι is so insignificant that most writers ignore its existence, and so they don’t put an accent over the interrogative τί — either because they don’t care, or because they think there is hardly any case of confusion. (However, in literary texts, such as the mentioned translation of the Odyssey, where the τι–for appears more often than the τί–what, the omission of the accent mark from the latter can become a real impediment in reading the text fluently.)

  • The για – γιά pair:
    This is more like the να – νά pair: the unstressed για is the most common one, meaning “for” in its regular use as a preposition, or “in order to” if followed by να:
    αυτό είναι για τη Μαρία : this is for Mary
    ήρθα για να φύγω : I came in order to leave [meaning: I came for a short time only]
    But the other γιά is often translated by “let’s” if followed by να, or by “here!” if not:
    γιά να δούμε! : let’s see!
    γιά πιάσε αυτό! : here, grab this!
    Once again, the stressed γιά is so rare that most writers will ignore it entirely. There are some other, localized uses of για, which don’t need to be mentioned.

Besides the above pairs, which consist of different words, there are the enclitic words: pairs made of a dative and a possessive case of the personal pronoun (“I”). The enclitic word pairs are a source of constant confusion, and the problem of how to distinguish among them has not been solved satisfactorily, in spite of Greek grammar books that will pretend the problem is solved. First, what are these “enclitic” word pairs?

They are pairs made of, e.g., “me” and “my”, or “you” and “your”, or “him” and “his”, “her” and “her” (notice the identity in English in the last pair), and so on. The first member of the pair is a personal pronoun in the dative case (yes, there is such a thing in English; the same forms constitute the accusative case), and the second is the same pronoun in its possessive case (usually called a “possessive adjective”, because grammarians enjoy confusing people). Examples:

  • μου δώσανε ένα δώρο : [they] gave me a gift (“me”, the indirect object of “gave”, is the dative form of the personal pronoun, “I”)
  • το βιβλίο μου : my book (“my” is the possessive form of the personal pronoun)
  • σου μυρίζει τίποτα; : does it smell something to you? (“you” is dative here, the indirect object of “smell”)
  • ήταν λάθος σου : it was your error (“your” is always possessive)
  • της έδωσα ένα φιλάκι : I gave her a little kiss (dative “her”)
  • μ’ αρέσει το φιλί της : I like her kiss (possessive “her”)

Now, what happens with these words is that normally they are not marked with an accent, just as you see them in the above examples. But there are some instances where genuine ambiguity can arise over which word is meant: the dative or the possessive, and the meaning is altered completely depending on which interpretation we take; for example:

η δασκάλα του πήρε το βιβλίο

The above can be interpreted in two ways:

  • either {η δασκάλα} {του πήρε το βιβλίο}, meaning “the teacher took the book from him” [dative],
  • or {η δασκάλα του} {πήρε το βιβλίο}, meaning “his teacher took the book” [possessive].

As you see, the meanings are entirely different. Greek grammar books say that when a case of ambiguity can arise then we mark the dative case of the pronoun with an accent mark, whereas we leave the possessive unmarked. (There is a good reason for this choice: the dative case of the enclitics is stressed in speech, whereas the possessive one is not.) Thus:

  • η δασκάλα τού πήρε το βιβλίο : the teacher took the book from him [dative]
  • η δασκάλα του πήρε το βιβλίο : his teacher took the book [possessive]

The problem is (and grammar books never mention it), it’s not always clear whether there is an ambiguity or not. The example I selected above is one of a glaring ambiguity, a typical one that grammar books would present. But there are many cases where it’s not easy to spot the ambiguity (so the writer doesn’t mark the accent), and once you spot it you wonder, “Now did the author really mean the absence of accent (possessive case), or did s/he miss seeing the dative interpretation, just as I could have missed it, had I not paid attention? Or — as a third alternative — did the author actually mean the dative case, and neglected placing the accent mark on the enclitic, as it happens every so often with sloppy Greek writers?”

This causes a real headache to people who, like me, insist on understanding accurately what they read, and want to be understood accurately when they write. Notice that the enclitic words are extremely common: it’s hard to write a whole paragraph without a single one of them; so this is not a mere “how many angels can dance on the point of a pin” type of problem. For a writing system in Greek that avoids these problems, read this text [warning: link not functional yet].

Caveat: some words look polysyllabic, but are actually monosyllabic

Caution is needed with some Greek words that appear long, yet they’re made of a single syllable, and thus do not require an accent mark:

  • ποιος, ποια, ποιο : (interrogative) who [masc.], who [fem.], which one [neut.]
  • γεια : this word can be used in various ways, usually as “γεια σου!”, or “γεια σας!”, which can mean “hello to you!”, “bless you!”, “to your health!”, “bye-bye!”, “cheers!”, and probably more.
  • μια : one [fem.]
  • δυο : two

All such words are stressed on their last vowel. What appears as a digraph before the last vowel is not pronounced as a vowel, but serves to palatalize the previous consonant. If you wish to learn more about the phenomenon of palatalization, read this page.

Note that the first word (the interrogative pronoun “who” and “which”) includes not only the three genders, but also all cases and numbers of this pronoun, i.e., all of its forms: ποιου, ποιον, ποιας, ποιαν, ποιοι, ποιων, ποιους, and ποιες. None of these is written with an accent mark, they are all monosyllabic, and stressed on their last (and only) vowel. If you ever see the word ποιόν (stressed), that’s a completely unrelated noun meaning “quality”, is syllabified ποι·όν, and only coincidentally looks like the interrogative pronoun.

Note also that the last two words have alternative disyllabic forms: μία and δύο. When you see them written thus, they should be pronounced as consisting of two syllables, the first of which is stressed: μί·α and δύ·ο. There is no difference in meaning between the alternative forms, and both are equally likely in each case.

Exceptions where some words appear monosyllabic but they aren’t (and do receive an accent mark)

You might see the following written:
  • φέρ’ το : bring it
  • δώσ’ μου : give me
  • άσ’ το : leave it [let go]
  • είν’ ανάγκη : it is necessary
  • λίγ’ απ’ αυτά : a few of these
  • κάντ’ υπομονή : be [pl.] patient (lit.: “do patience”)

In each case, as in many similar ones, the first word is not monosyllabic but disyllabic, and its final vowel has been omitted because the following word allows the omission. (The words in full are: φέρε, δώσε, άσε, είναι, λίγο, and κάντε.) Thus, we don’t also omit the accent mark just because the word appears monosyllabic.

Exceptions where polysyllabic words don’t receive an accent mark

Yes, there are such cases, too. They arise from a different kind of vowel omission, this time one that happens at the beginning of a word:
  • να ’ρθετε : (may) you come
  • θα ’θελα : I would like
  • πού ’σουνα : where were you (Note: this is the interrogative πού, so it’s marked with an accent)
  • μου ’δειξαν : they showed me
  • σου ’φερα : I brought to you

Here, the “collision” of the two vowels (last vowel of the first word and first vowel of the second word) results in the omission of the second vowel, which is in the stressed syllable of the second word. (The words in full are: έρθετε, ήθελα, ήσουνα, έδειξαν, and έφερα.) In such cases we don’t place the accent mark on the first word, although it is stressed in speech. That is, if you pronounce, e.g., the first example, you’ll say something like [nárθete], where the small tick ( ´ ) shows which vowel is stressed; the second one would be [θáθela]; and so on. Note that the third example shows an accent over πού because this is the interrogative adverb, mentioned earlier.

Thus, because originally the stress was on the omitted vowel, after the vowel deletion the stress moves (in speech!) on the first little (monosyllabic) word, and the longer second word remains unstressed. But in writing we don’t mark the accent on the first monosyllabic word (don’t ask me why!), so the whole thing remains without an accent mark.

Exceptions to this exception!

What good would a rule be if it didn’t include exceptions to exceptions? Every self-respecting rule must have this feature, and so does this one. So:

Suppose I write the following:

  • θα ’ρθω : I will come

Where the apostrophe denotes the missing ε of the full form έρθω: the ε is elided because of the preceding α. But there are two ways to stress this in speech: [θárθo] and [θarθó]. (Notice that both ways are equivalent, i.e., no difference in meaning is effected, whichever way you choose to pronounce this.) If we don’t place an accent mark anywhere, we won’t know which pronunciation of the two is intended. To solve this problem, we do place an accent mark if [θárθo] is the intended pronunciation, whereas we don’t place any accent mark if the pronunciation is like [θarθó]. Thus:

  • θά ’ρθω [θárθo]
  • θα ’ρθω [θarθó]

A few more examples are possible:

  • νά ’ρθεις [nárθis] vs. να ’ρθεις [narθís] : to come [2nd person] (same if you replace να by θα)
  • νά ’ρθει [nárθi] vs. να ’ρθει [narθí] : to come [3rd person] (same if you replace να by θα)
  • νά βρω [návro] vs. να βρω [navró] : to find [1st person] (same if you replace να by θα)
  • νά βρεις [návris] vs. να βρεις [navrís] : to find [2nd person] (same if you replace να by θα)
  • νά βρει [návri] vs. να βρει [navrí] : to find [3rd person] (same if you replace να by θα)
  • νά μπω [nábo] vs. να μπω [nabó] : to enter [1st person] (θα is not common here)
  • νά βγω [návγo] vs. να βγω [navγó] : to exit [1st person] (θα is not common here)

(The careful reader may have noticed that I always put an apostrophe before ’ρθω, whereas I omitted it before the other verbs. This is because there is no Greek word that starts with ρθ-, so that ρθω somehow doesn’t look like a correct word; in contrast, βρ-, βγ-, and μπ-, are valid starters of Modern Greek words.)

Just in case some Greek reader complains “This is not what we do! We write θα ’ρθω for [θárθo], and θα ’ρθώ for [θarθó]”, please note that I copied the above rules from Manolis Triantafyllidis, “Neoelliniki Grammatiki”, p. 25, ΟΕΔΒ edition (your well-known high-school textbook). If actual practice is not as Triantafyllidis prescribes, learners of Greek take note! (Thanks to Greg Brush for pointing this out.)

As far as I know, the above cases are all that there is; i.e., the 2nd & 3rd persons of μπω and βγω are almost always stressed on the ultima: να μπεις [nabís], να βγεις [navγís] (2nd p.), and να μπει [nabí], να βγει [navγí] (3rd p.).

It’s a mess, right? I know. Relax, no, there is no exception to the exception to the exception.

Exceptions where polysyllabic words receive two accent marks

Up to this point, if you pay attention to the previous rules, you’ll notice that the accentuation system of Modern Greek worked with the following idea in mind: “We have to reduce the ancient five accent marks into a single one, so let’s see how we can reduce them the most.” Everything else (such as the above exceptions) was postulated in order to disambiguate ambiguities. Thus, this accentuation system is not so much geared toward showing you where the stress goes up and down in speech, but toward simplifying the old multiple-marks system (the so-called “polytonic” one, see Ancient Greek). But the exception I’ll describe now is geared more toward showing you where the stress goes in speech. It occurs when a word is stressed on the antepenult (3rd syllable from the end), and an enclitic word follows it (see here for enclitic words). Examples with nouns:
  • το πουκάμισό μου : my shirt
  • ο έρωτάς σου : your [sg.] love (affair)
  • η προσπάθειά του : his effort
  • το χτένισμά της : her hairdo
  • το βράσιμό του : its boiling
  • η κυβέρνησή μας : our government
  • το δωμάτιό σας : your [pl.] room
  • η θάλασσά τους : their sea

Some verbs (all in the imperative):

  • μάθετέ μας : learn (about) us
  • δάγκωσέ το : bite it
  • φίλησέ με : kiss me
  • διάβασέ τα : read them [neut.]
  • αγκάλιασέ τις : hug them [fem.]
  • αντιμετώπισέ τους : confront them [masc.]

There is a good reason why the combination of such words is stressed in this way. Remember the God-given Rule of Greek stress? The one saying that no stress can go beyond the antepenult, the 3rd syllable from the end? (If you already forgot it, take another look at it in the blue box at the top of this page.) Well, combinations like the above make up a single unit of four syllables, at least the way they are pronounced. There already is a stress on the antepenult (e.g. μάθετε), so when the enclitic is adjoined (e.g., -μας), the whole unit makes up four syllables, the first of which is stressed (e.g., μάθετε-μας). And the Holy Rule says you drop dead if you do that kind of Greek sacrilege. So what you do to avoid Zeus’s thunderbolt is to put another stress on the penult of the entire unit (e.g., μάθετέ μας). Note that this additional stress is not secondary, i.e., weaker (not as in “dictionary”), but is produced with the same power as the first stress.

The same phenomenon (the Divine Rule in action) causes some rare cases where verbs in the imperative mood are followed by two enclitics:

  • πάρε τό της : take [sg.] it from her
  • φέρε μού τους : bring [sg.] them [masc.] to me
  • δώστε μάς το : give [pl.] it to us

As you see, although here we don’t have a word with two accent marks, we have the same phenomenon: the whole unit (word + enclitic + enclitic) has a stress on the fourth syllable from the end, and — to eschew the wrath of Zeus— an additional stress is placed on the second syllable from the end (the penult), which is also shown in writing with an accent mark.

A suggested revision: toward a more intelligent accentuation system in Modern Greek

Does the presented Modern Greek accentuation system appear weird to you? It does, to me. In this page [warning: link not functional yet] I discuss its problems in more detail, and propose a more reasonable system to other native speakers (or rather, writers) of Greek, like me. Will anybody listen? No. Do I care? Neither. But I employ my more intuitive system when I write for myself, and hope that, one day, other fellow Greeks will see the light.

Ancient Greek

As mentioned earlier in the Modern Greek section, there are three marks that denote not just stress, but the ups and downs of voice in Ancient Greek speech, as well as other marks that help the reader pronounce correctly. But why so much complexity? For a couple of reasons:
  • First, the ancient phonology was inherently more complex than the modern one. There were many more vowels than the present five ones (I refer to sounds, not to letters of the alphabet), and true diphthongs (those that turned to digraphs in the modern language). Also, vowels had different durations: there were long and short vowels, and the pitch of the voice could move up, or down, or up-and-down while pronouncing long vowels or diphthongs.
  • The second reason was that the ancient accentuation system was designed not for native speakers, but for second-language learners of the language. Native speakers had no reason to be told how the voice goes up or down: they knew how to talk. But learners of the language, who, after the conquests of Alexander the Great suddenly became too many, had no clue. For some centuries, well into the rise of the Roman empire, Greek played the role of the lingua franca of the Mediterranean world. (That’s why the original texts of the New Testament were written in Greek.) The lowercase letters were invented a couple of centuries before Christ by the Greek grammarians of Alexandria (Egypt), who also invented the complex accentuation system to help non-native speakers read Greek.

Here is the basic prerequisite knowledge to understand the ancient system:

The duration of vowels

There are vowels of short, and of long duration in Ancient Greek. This is true also in English: for example, “bit” has a short [ı], but “beet” has a longer [i:], a vowel different not only in duration but also in quality. Exactly the same is true in Ancient Greek:
  • There is a short [o], denoted by the letter omicron (ο), and a longer [o:], denoted by the letter omega (ω). Omicron is pronounced as in British English “got”, whereas omega is pronounced as in “awe”, with the mouth a bit more open than in omicron.

  • There is a short [e], denoted by the letter epsilon (ε), and a longer [ε:], denoted by the letter eta (η). Epsilon is pronounced as in “bet”, whereas eta is pronounced as the first constituent of the glide i] in “make”, but longer. That is, take the glide [εi], the only vowel of “make”; break it up into its two constituents, and keep the first one, the [ε]; then lengthen that, but without gliding into [i] (if you can); that’s ancient eta. This is a bit hard for most native English speakers to do, because after producing the [ε], if they try to lengthen it, their jaws “automatically” (subconsciously) close somewhat and their tongue moves a bit forward, which results in gliding toward [i], but with a bit of practice it’s possible to do it.

  • Besides the previous four vowels, which have a one-to-one correspondence with the letters ο, ω, ε, and η, there are another three vowels, [a], [i], and [y], which can be pronounced as either short or long, without modifying their quality. Thus, the letter alpha (α) represents either the short [a] as in “father”, or a longer [a:] (not appearing in English in this longer duration); the letter iota (ι) represents either a short [i] (not appearing as short as that in English), or a longer [i:], as in “beet”; finally, the letter upsilon (υ) represents either a short [y], which is a rounded [i] as in French “une”, or a longer [y:]. Thus, here we have no different quality in vowels, just three sounds that can be heard as either short or long. Unfortunately, the writing of ancient Greek gives us no clue about whether the short or the long version is intended — that is, we’ll always see the same three letters, α, ι, or υ — unless they are marked with a circumflex, which means they are long (see why, below).

  • Diphthongs are always long, with a single exception: when the diphthongs αι or οι appear at the very end of declined words (nouns, adjectives, verbs), they are short. Exception to the exception: verbs in the optative mood still have their final αι or οι long.

Okay. Now, with the above knowledge regarding the duration of vowels, we can move on to understand the ancient accent marks.

Voice can go either up, or down, or both up-then-down during a stressed vowel

If what you have to pronounce is a short vowel, you have only one “tempo”, or mora as it is called in Latin (pl. morae), so the pitch of your voice can be either raised, or lowered, but not both (because there is not enough time to do both). But if you have to produce a long vowel, all possibilities are available: the pitch of your voice can go up-then-down during the pronunciation of the long vowel (up in the first mora, and down in the second one; the possibility of going first down and then up during a single long vowel does not appear in Ancient Greek); or the pitch can start low and keep rising during the two morae; or, finally, the pitch can start high and keep lowering during the two morae. Here is how the accentuation system denotes these possibilities in writing:
  • If the pitch of the voice starts low and keeps rising (whether in one or two morae), the acute accent mark ( ΄ ) is placed over the vowel. The shape of the acute accent ( / ) is indicative of its function.
  • If the pitch of the voice starts high and keeps dropping (whether in one or two morae), the grave accent mark ( ` ) is placed over the vowel. Again, the shape of the grave accent ( \ ) is indicative of its function.
  • Finally, if the pitch of the voice starts low and is raised during the first mora, and is lowered during the second mora, the circumflex mark ( ^ ) is placed over the vowel. Once more, the shape of the circumflex ( /\ ) is indicative of its function.
The circumflex mark lost its pointy tip in handwriting, becoming a mere curved line. Further, with the advent of typography, and long after the raisings and lowerings of the pitch in voice had been dropped from the language, it acquired an extra hook at its end, thus turning into a tilde ( ~ ). This final form of it is the one used in Greece in modern times when the polytonic system was still official (until 1982), and today when the need arises to print ancient, older Greek, or ecclesiastical (2) texts. The tilde-like form is the one we’ll use in this text, too, conforming to Greek typography; but the reader should have in mind what the shape of the original circumflex stood for.
From the above it follows that since ε and ο are always short vowels, you will never-ever-ever see them marked with a circumflex. If there are any absolute rules with no exceptions in Ancient Greek, this is one of them.

The Divine Rule of Three now concerns three morae, not three syllables

First, let me remind you of the terms used in this text for the last three syllables of a word:
  1. Ultima (λήγουσα) : the last syllable of a word. E.g., in English, the stressed syllable of the word e·clipse.
  2. Penult (παραλήγουσα) : the second-last syllable. E.g., the stressed syllable of the word ec·cen·tric
  3. Antepenult (προπαραλήγουσα) : the third syllable from the end. E.g., the stressed syllable of the word ex·pli·cit·ly.

Remember the Divine Rule of stress in Modern Greek? It stated that no word can be stressed earlier than the 3rd syllable from the end. In Ancient Greek, this rule can be better understood if modified somewhat, so that it refers to morae, not just whole syllables. We’ll see that the rule is not as strict as in its modern version, but still, it will help us understand some stress patterns. Here it is, rewritten for the ancient case:

Ancient Greek prefers keeping the stress of a word within the last three morae
What does that mean? Let’s see some examples.

The word means “race, descent”, and also “offspring”. On the first line it appears in the nominative case, in which it is stressed on the antepenult, over the first έ. It has three short syllables (ε – ε – ο), therefore, three morae. The third-to-last mora is stressed. So far, nothing very interesting. The second line, however, shows the same noun in the genitive case, which ends in the diphthong ου, a long vowel (recall that every diphthong except the final αι and οι is long). So now, if the stress were still on the first ε (i.e., *γένεθλου), we would count four morae starting from the stressed one: έ(1) + ε(1) + ου(2) = 4. Because Ancient Greek generally avoids such stress patterns, the stress descends to the penult: γενέθλου. This situation appears very often in nouns and verbs.

However, the above rule concerning morae should not be taken as dogmatically in the ancient language as it must be taken in the modern one. There are plenty of examples where, if we count morae including the stressed one, we find out that there are four of them:

The first example means “the person”, the second “Homer”, and the third “I was leaving”. (Ignore for now the small hooks over initial vowels, or next to them, i.e., the aspiration marks, which will be explained in the next section.) In the first example, since ω is long and ο short, we already count three morae before the stressed one, so if the initial α is short (it is), then the total morae including the stressed one are four. Similarly, in the second example, we count Ό(1) + η(2) + ο(1) = 4, and in the third example, again, έ(1) + ευ(2) + ο(1) = 4. So it looks like this “stress at most the last three morae” does not hold much water, does it?

And yet, if we don’t take it too literally, we’ll see that we can explain several phenomena with its help. For example, the above two nouns in the genitive case have their stress descend to the penult:

So we see the same “reluctance” to hold the stress on the antepenult in the genitive, in view of the fact that with the final long ου we’d have ά(1) + ω(2) + ου(2) = 5 morae in the first example, and Ό(1) + η(2) + ου(2) = 5 also in the second example. If 4 doesn’t sound good, 5 is definitely prohibitive. This descending of the stress over the penult is sometimes expressed by the rule:

When the ultima is long, the antepenult cannot be stressed

Thus, if the antepenult cannot be stressed, the stress “descends” to the penult.

But there are several more observations we can make with the help of the “relaxed” Divine Rule employing morae. First, the Modern Greek rule is also valid in the ancient language, in an absolutist way:
Only the last three syllables of a word can be stressed in Ancient Greek
There is no exception to the above rule. (That’s why it was mentioned at the top of this page as holding diachronically unchanged, in both the ancient and modern languages.) But in Ancient Greek we can make one further elaboration of it, concerning the stress marks:
Only an acute accent mark is possible on the antepenult
Right, because a circumflex on the antepenult would mean the following: the first mora of the antepenult has the voice raised (count 1); then the second mora of the antepenult (there must be a second one, i.e., the vowel must be long, otherwise we couldn’t have a circumflex) has the voice lowered (count 2); and then there is the penult (count at least one more mora, i.e., 3), and the ultima (count at least 4). Four is a bad idea. The above is an absolute rule, too — no exceptions. (Unless I mention it explicitly, the rules I list are absolute ones.)

Then there are two rules concerning the kind of accent mark on the penult:

A stressed long penult receives an acute accent mark if the ultima is long
Or: “Long before long gets an acute”, which again complies with the “last three morae” rule of thumb: if the long penult received a circumflex, then we’d count morae as follows: the first mora of the penult has the voice raised (count 1); the second mora of the penult has the voice lowered (count 2); and then there are two more morae in the long ultima (count 4), which is not good. Examples:

The first word is the plural of “person” in the dative case (we saw the nominative in the singular of this noun in a prior example), and the second is the name of the goddess of wisdom (Roman Minerva), the modern version of which we encountered in the beginning of the Modern Greek section. In both cases, we have a long penult before a long ultima. The third example, meaning “the dolphin” in the accusative singular, is trickier: it tells us that the ultima must be short. Notice that the letters ι and α do not reveal their long or short duration. The penult, we have no doubt what it is: we see a circumflex over it, so it must be long. But the ultima? Well, if the ultima were also long, the penult would receive an acute mark by the above rule; so the ultima must be short. In this third example we cannot figure out what mark to place beforehand, but we can figure out the quality of its vowels after seeing how the word is stressed.

And what happens when the ultima is short? Here:

A stressed long penult receives a circumflex if the ultima is short
Or: “Long before short gets a circumflex”, once more complying with our rule of thumb: the voice is raised on the first mora of the penult (count 1), lowered on the second mora of the penult (hence the circumflex; count 2), and there is one more mora in the short ultima (count 3). Three is good.

(You might think, “two is also good, isn’t it?”, meaning that an acute accent on the penult wouldn’t hurt. Yes, but the stress pattern of Ancient Greek does not allow this possibility.)


The first word means “who” (the interrogative pronoun), in the masculine gender, singular, nominative. We see a long diphthong (οι) in the penult, and a short vowel in the ultima, hence the penult will get a circumflex. In the second example, meaning “soft, gentle, meek” (masculine, singular, nominative), we conclude that the α of the penult must be long after seeing how it is stressed. We don’t really need the above rule in this case (if the α were short, the only possibility for it would be an acute accent), but we see an application of it: given that the α is long, it must receive a circumflex, by virtue of the previous rule.

The previous rules tell us what accent marks to use in many cases over the antepenult and penult. What about the ultima?

Unfortunately, any of the three marks can be placed over the ultima: the circumflex, the acute, or the grave accent. Hmm... is this the first time we mention the grave? Yes, right? But there is a reason for that. Here it is:

The grave accent can appear only on the ultima

All right. The good thing about the grave is that it is more or less predictable. Here is one reason:

The grave is only the “alter ego” of the acute (only over the ultima, of course).
Simply, if there is no period, comma, or other punctuation mark that interrupts
the flow of speech after the stressed word, instead of an acute we put a grave.
This doesn’t mean we cannot have a circumflex over the ultima! But if the circumflex is ruled out, and the decision is between the acute and the grave, then the way to decide between the two is given by the above simple rule of thumb, which says that if a comma, period, colon, semicolon, question mark, or exclamation mark follows the word, we put an acute; otherwise we put a grave accent. (Reality is a bit more complex than the simple image I paint here, but let us stay with the above for now.) Example:

This fragment of a sentence says, “and so the beautiful/fair Calypso” [in the accusative; Calypso was a nymph in The Odyssey]. As you see, as long as the voice keeps uninterrupted we keep placing grave accent marks — assuming words are stressed on the ultima and that no circumflex is required. When the flow of voice is interrupted, we place an acute, signifying the raising of the voice at the end of this part of speech (i.e., the noun phrase; in reality, a noun phrase ends at its final noun without needing to be followed by a punctuation mark, and that is the “complexity” I refer to in the previous paragraph).

I think the above is good as a collection of rules that cover the basics of stress in Ancient Greek. There is more, but it’s beyond the scope of this page to mention everything known about ancient stress. We move now to the other two marks seen over vowels (and sometimes even over consonants).

Aspiration is marked with a special symbol; so is its absence!

The sound [h] (the “aspiration”) in English (as in “horse”) has a letter that usually represents it, the H. Ancient Greek had this sound, too, at the beginning of some words. In Classic times (ca. 5th C. BCE), the same letter, H, was used to write it. At that time, there were only capital letters, no lowercase ones. Thus, the word ΙΠΠΟΣ (“horse”) was written ΗΙΠΠΟΣ. (And when such words passed into Latin, they retained their initial H, thus hippopotamus, Hippolytus.) But this use of the H was confusing, because there was the use of the same letter, eta, to denote the long vowel [ε:], and there were words that started with the vowel eta, followed by another vowel, as in ΗΕΡΙΟΣ (“in the air”). Now what did that initial H represent? An aspiration (English [h]), or the long [ε:]? (The latter was the case in ΗΕΡΙΟΣ. Compare that with ΗΕΛΙΞ, “spiral, helix”, where the H stood for aspiration.) Independently of such — admittedly rare — problems, the use of H to denote the aspiration was abandoned after the Classic times. The sound of aspiration (the [h]), however, still existed in the language: as an initial sound in words starting with a vowel or with the consonant rho (ρ, [hr]), and within the so-called aspirated consonants θ [th], φ [ph], and χ [kh]. When the lowercase letters were invented in Hellenistic times (last three centuries BCE), Greek grammarians needed a way to show to non-native speakers (learners) of Greek when an aspiration was to be pronounced, and when not. So, what they did is that they broke up the letter H into two halves, one like this: |- and one like that: -| and used the first symbol to denote the aspiration, and the second for the absence of it. But those symbols were unwieldy. Remember, there was no typography at that time to standardize anything — all written language was handwritten. Handwriting often results in the distortion of symbols (one such example was mentioned in an earlier paragraph regarding the circumflex). So the two symbols soon changed form: the |- of the aspiration became like a right-facing hook:

That was the rough breathing mark, placed over the lowercase initial vowel or rho. The absence of it in words starting with a vowel was marked with the symmetric symbol, the transformed -| :

That was the smooth breathing mark. But I exaggerated the size of those symbols; see them in actual size in the following words:

Did you notice that I didn’t provide any example of the smooth breathing mark over a upsilon or rho? There is a reason for that:

The letters υ and ρ are always marked with the rough breathing mark, word-initially
This is so because obviously ancient Greeks were pronouncing an obligatory aspiration before starting a word with either of those two sounds. Such obligatory phonological phenomena are quite common in languages. In English, for example, native speakers “cannot” start a word with [ps] or [ks], and so, “psyche” is pronounced [sáiki:], and “xenon” is pronounced [zí:non]. Spanish speakers cannot start a word with [sp], [st], or [sk], and so, the word for “spasm” is “espasmo” (σπασμός), the word for “stadium” is “estadio” (στάδιον), and the word for “scorpion” is “escorpión” (σκορπιός). Similarly, it wasn’t possible for ancient Greeks to start without an aspiration before υ and ρ, word-initially. (Note that in modern Greek there is no aspiration at all, so ρ starts words just as in languages like Spanish, Italian, Russian, etc., which also have the ρ-sound and lack aspiration.)

Now, fine, the rough breathing mark denoted the aspiration, [h]. But why did its absence need to be marked with another symbol, the smooth breathing mark? Take this as an idiosyncratic feature of the ancient Greek writing system. There is no evidence showing that the smooth breathing mark had any effect in pronunciation. Simply, if there is an initial vowel and you don’t have to put a rough breathing mark over it, then you put a smooth one.

Another question the learner might have is if there is any way to predict which mark to use. Other than the absolute rule for υ and ρ discussed above, unfortunately there is no way. At least not for us, learners of Ancient Greek, who, unlike the ancient native speakers, receive no clue from pronunciation. There are a few partial rules (in the sense that they apply to a few types of words only), such as that the smooth breathing mark is way more common than the rough one, or that the so-called “augment” of the past tenses (the Greek “aorist” and “imperfect”), i.e., an initial ε- or η-, always takes the smooth breathing mark. But such rules do not save the day. The bottom line is, the learner has to learn by heart which mark goes with which word. That’s what I had to do as a child, when the same ancient polytonic system was still applied to Modern Greek (I finished my high school years in 1980, just two years before the polytonic system was abandoned, and the monotonic one — described in the Modern Greek section — was officially established).

A few more things to note regarding the breathing marks: first, if the initial vowel is a diphthong, the breathing mark is placed over the second letter of the diphthong, just as it happens with the accent marks:

By the way, here is a another rule of thumb:

Almost all words that start with the diphthongs αυ, ευ, or ηυ, have a smooth breathing mark over the υ
The only major exception is the verb ευρίσκω (“I find” listed above) and its derivatives.

And second, when the breathing and accent marks coincide over a vowel, there are rules for how to place them: the breathing mark goes before the acute or grave, and under the circumflex:

Is that all?

The above by no means exhaust the topic of accentuation in Ancient Greek, but should be construed as a sufficient starting point for beginners.

Footnotes (clicking on the footnote-number, on the left, brings back to the text)

(1). And yet, some readers will be left with the pressing, I-want-to-know-now need to learn the answer: what does this eight-syllable-long Greek word mean? To let such readers sleep peacefully at night I’ll reveal its meaning immediately: it has no real meaning. It’s a kiddy-word, meaning something like “worm-ant hole” (σκουλήκι = worm, μυρμήγκι = ant, and τρύπα = hole). I suppose Greek kids see those holes in the ground and figure out (wrongly) that they might be made either by worms or by ants (they’re made only by ants), so they coined this word. And because they looooove mysterious and long words, they made this word part of the modern Greek language.

(2). The Greek Orthodox Church still uses the polytonic system officially.

Back to the “main” Greek language page