The Details of Modern Greek Phonetics and Phonology
What is it that distinguishes the pronunciation of a native speaker of Modern Greek from that of a learner of the language? As is the case with every other natural language, there are a few tips, tricks, and details that nearly no book on pronunciation mentions. Read the following if you intend to perfect your Modern Greek pronunciation, or if you are simply curious and interested to see what distinguishes Modern Greek phonetics and phonology from that of your own language. You can hear sounds of the example-words from a native speaker, the author of this text (click on the speaker-icon next to each word to hear its pronunciation).
I am assuming you are already familiar with the Greek alphabet, and hence with the pronunciation of each letter individually. This page lists the cases where the pronunciation of letters changes according to their surroundings. The list is meant to be exhaustive — there are no other cases of differing pronunciation in Greek depending on context, to the best of my knowledge. Modern Greek is supposed to be an almost what-you-read-is-what-you-speak language, not as “pure” as Spanish or Italian, but certainly much more predictable than English or French. Here is a list of topics discussed in this page:
[s] → [z] in front of voiced consonants
Note: when I put a letter in brackets, like this: [o], I refer to the sound of the letter; otherwise, the printed letter is shown without brackets. Actually, what I put in brackets is not letters, but IPA (International Phonetic Alphabet) symbols, all of which can be seen here (with Greek sounds highlighted).
When the letter σ (σίγμα, which stand-alone is pronounced [s]) is found before a voiced consonant except λ, (i.e., β, γ, δ, μ, ν, ρ), it is pronounced as the letter ζ (ζήτα), that is, [z]. Examples:
The same transformation takes place even if σ is the last letter of a word (thus, in writing it appears as “final sigma”: ς) and the next word starts with a voiced consonant. Examples:
When reading in slow speed, however, or speaking emphatically, it is possible to notice the cancellation of the between-words [s] → [z] transformation.
Notice that the letter λ, although voiced, does not always effect this transformation on σ. Some people, though, may perform this transformation for a final ς even before a λ. Examples:
In the page of the Greek alphabet it was explained that the letters μπ, when found together, are pronounced as [b]; the letters ντ, also together, are pronounced as [d]; and the letters γκ or γγ together are pronounced as [g]. In reality, the rule is not so simple.
What happens is that μπ is pronounced as [b], ντ as [d], and γκ as [g] only at the beginning of a word, or after a consonant (which is usually ρ); or sometimes when the word is a foreign loanword. Let’s call these the “dry” versions of μπ, ντ, and γκ. (Note that γγ never appears at the beginning of a word or after a consonant, so there is no “dry” version of it.) Examples:
It should be noted that although it is nearly acceptable to turn the nasalized version of these sounds to their corresponding “dry” ones (as in [ebr`os] for εμπρός), it is unacceptable (wrong pronunciation) to turn the “dry” sounds into nasal ones. For example, the pronunciation [mb`armbas] (for μπάρμπας, see example above) will never be uttered by a native speaker of Greek.(1)
Also note: the “always dry” pronunciation (as in [ebr`os] for εμπρός) can be heard in the locality of Athens by the younger generation ca. the beginning of the 21st century. This “peculiarity” seems to be specific to Athens. However, given the social significance of the capital of Greece, it is not hard to imagine that the “always dry” pronunciation might prevail over time and become “standard Greek”.
A phenomenon similar to the previous one occurs at word-boundaries: when a word ends in [n], then if the following word starts with [p] (and both words belong to the same part of speech, i.e., there is no pause due to a period, semicolon, comma, etc.) what happens is that the [n] changes to [m] and the [p] to [b]. Similarly, if the following word starts with [t], we have the transformation [n t]→[nd]; and if it starts with [k] we have [n k]→[ŋg]. Examples:
These transformations are very common due to the high frequency of occurrence of the definite article forms τον, την (singular, accusative case of masculine and feminine genders), and των (plural, genitive, all genders). Notice that it is only [n] that can cause these transformations, not [m], because in Greek there are no words ending in [m] (or, if there are any foreign loanwords, such as τραμ, they are usually nouns, causing a momentary pause in speech, which cancels this phonetic transformation).
Note also that this transformation is not universal among native speakers. Some native speakers (a small percentage, I believe) do not perform them, especially the second ([n t]→[nd]) and the third one ([n k]→[ŋg]), and especially when they want to speak exceptionally clearly (like anchormen/women). I myself do not do these transformations in my more phrases pages, precisely for this reason. But I am sure I perform them when I talk to other native Greek speakers.
As mentioned in the alphabet page, the typical case where the sound [ŋ] appears is in front of a velar consonant (sounds [k], [γ], [x], letters κ, γ, χ), and is denoted by the letter γ before the velar consonant. Examples:
In addition, a [ŋ] is produced also whenever a final [n] appears before an initial velar consonant (including [g]) at word-boundaries. Examples:
A third case where [ŋ] appears was already mentioned in a prior paragraph: it is when the digraph γκ or γγ appears after a vowel, and so is nasalized, pronounced as [ŋg]. For examples, see this paragraph, above.
The phenomenon of palatalization (a.k.a. palatization) is the single most important phonetic phenomenon of the Modern Greek language, and the one that largely distinguishes the speech of a native speaker from that of a second-language learner. Most native speakers of Greek are not even aware of this phenomenon. (Naturally, because it is a phonetic, not a phonological aspect of the language.) However, it is interesting that when Greeks want to mockingly imitate the non-native Greek pronunciation of foreigners, the main trick they do is that they avoid palatalization (because non-native speakers almost always omit it), but without being aware of doing so! That is, the average Greek is a proficient master of the phenomenon, and can even reproduce its omission from the language, but is not consciously aware of it — does not have access to the concept as a single idea, in cognitive terms. The following paragraphs will describe this concept. If you are a non-native speaker interested in the language, you will learn (consciously) something almost all Greeks know only subconsciously. If you are a native speaker of Greek, chances are you’ll be surprised by what you’ll read. So, let’s start. What is palatalization in Greek?
1. Regular Palatalization
→ The notion of palatalization refers to the change of four consonants, [k], [γ], [x], and [g], from the velar column of this table to the preceding palatal one, if the vowel that follows is either [i] or [e].
→ Also, two consonants of the alveolar column, [n], and [l] switch to their corresponding palatal ones if the vowel that follows is [i] (but not [e]).
Let us see examples of palatalization of the velar consonants:
The fact that many of the above words are commonplace gives a first indication of the frequency with which palatalization occurs in the language. Still, there is a lot more to follow. Let us see now some examples of palatalization of the alveolar consonants:
Note, however, that the palatalization of the alveolar consonants [n] and [l] followed by [i] varies in degree among regions of Greece. Athenians palatalize them a little bit, whereas people in some villages of Peloponnesus palatalize them a lot (to the point that Athenians — who are silly enough to feel “people of the capital” — make fun of them).
I mentioned earlier that the alveolar [n] and [l] do not become palatalized before [e]. That is, the letter-sequences νε and ναι are pronounced as: [ne]; similarly, λε and λαι are pronouced: [le]. Does this mean that the sounds [ɲe] and [ʎe] do not exist in Greek? Not at all — quite the contrary! They are very common. To write these last two sounds, an [i] (usually iota: ι) is inserted between the alveolar consonant and [e]. Examples, for all cases (palatalized and regular):
Now, the previous observation about [e] (i.e., that alveolar consonants can be palatalized before [e], except that in writing this is denoted with the insertion of an [i]), can be generalized to all vowels (except [i], of course). That is, to write [ca] (i.e., the palatalized [k] followed by [a]) an iota is inserted between κ and α. Thus, κια is pronounced [ca]. Similarly, [co] corresponds in writing to κιο (or κιω), and [cu] corresponds to κιου. The same remark applies to all other consonants we have examined so far. Examples:
It is important to understand that, in all the above examples, what you see written as an iota (ι) is not pronounced as [i]; it merely serves to denote the palatalization of the previous consonant. The entire [consonant + ι + vowel] is one syllable. It’s not even a diphthong, as in Spanish nieve (= snow); it is just a single palatalized consonant followed by a single vowel ([a], [o], [u], or even [e]). This is what distinguishes the speech of learners from that of native speakers: learners typically insert an [i] (even a very faint and short one) between the consonant (which they fail to palatalize) and the vowel. For example, suppose there were in Greek a word like νιέβε (a transliteration of Spanish nieve). Here is how a native speaker of Greek would pronounce this hypothetical word: [ ɲ`eve] (as if the Spanish word were ñeve). Compare that with the way a native speaker of Spanish(2) pronounces the actual Spanish word nieve: [n`(ie)βe] (two syllables again, but with [(ie)] forming a diphthong, and [n] being the regular alveolar sound).
The truth is that, even for a native speaker of Greek, it is impossible to pronounce one of these palatalized consonants and then go to the vowel without having the tongue pass from the position in the mouth-cavity where a very short [i] must be produced. This may be the reason why Greeks “feel” they pronounce an [i], and show this in their writing. That is, phonologically (in the Greek native speaker’s mind) there is an unpalatalized consonant, an [i], and a vowel; but phonetically (in actual sounds, as recorded and shown in a spectrogram) there is a palatalized consonant, the faintest idea of an [i], and a vowel.
Finally, let’s note that an earlier observation, in which I said that Athenians palatalize “a little bit” the alveolars [n] and [l] if followed by [i] (i.e., νι and λι), is not true if a second vowel follows after [i]. That is, νια, νιε, νιο, νιου, and λια, λιε, λιο, λιου are always palatalized by everybody, whether peasant in the countryside, or President of the Republic.
2. Forced Palatalization
Everything in the previous sub-paragraph, under the heading Regular Palatalization, was about the change of six consonants: [k], [γ], [x], [g], [n], and [l]. In this sub-paragraph we’ll see that all the rest of the Greek consonants can be “palatalized”. But how does one palatalize a consonant such as [v] (letter β), for example, which is a labiodental fricative sound, when its corresponding palatal fricative sound [ ʝ ] has already been “taken” for the palatalization of [γ]?
Simple: [v] is “palatalized” by inserting a [ ʝ ] between it and the following vowel. So, when in Greek we write βια (and the stress is not over ι), we pronounce [vʝa], as if we had written βγια. We’ll soon see that there are exceptions, but this is true in the overwhelming majority of cases. I call this phenomenon “forced palatalization” because it does not result in a change in quality of [v], but in the insertion of a true palatal. Just like [v], every voiced consonant (other than the “regulars” [γ], [g], [n], and [l]), when palatalized, results in the insertion of the voiced palatal [ ʝ ] between itself and the following vowel. Similarly, every unvoiced consonant (other than the “regulars” [k] and [x]), when palatalized, results in the insertion of the unvoiced palatal [ç]. So, for example, φια is pronounced [fça], as if we had written φχια. The only exception among the voiced consonants is [m], a nasal, which (predictably enough) is palatalized by the insertion of the only nasal palatal sound: [ ɲ ]. So, μια is pronounced [mɲa], as if we had written μνια. Let us see as many examples as are necessary to make sense of these rules.
Let it be noted that an additional column, titled “+ [i]”, could be appended to the above table, because there are some cases where we have two successive [i]’s in writing, but the first one forces the palatalization of the previous consonant. Examples: ίδιοι [`iðʝi] (= same ones (masc.)), καινούριοι [cen`urʝi] (= new ones (masc.)). However, such examples are rather rare.
Also note that when we write the word μια (= one, feminine gender) without stress, it is understood that this is one syllable (that’s why the stress doesn’t need to be shown over alpha), hence, pronounced: [mɲ`a]. There is an alternative (and equally common) pronunciation: [m`ia] (same word, same meaning); in this case we write the stress over iota: μία, since we pronounce the two vowels as two syllables. A similar remark applies to δυο: [ðʝ`o], versus δύο: [ð`io]. In each case, although the meaning is the same, the stress mark shows in writing which of the two pronunciations is intended. (See rules for placing the accent marks to denote stress in writing, for more details.)
(A note on my notation: when I write two consonants in parentheses, such as (ts), (ks), and (ps), I mean that those two consonants must be pronounced one after the other as rapidly as possible, making nearly a single unit. I mention this because non-native speakers tend to insert a faint puff of air between the two consonants, which should be avoided in Greek.)
The previous note about an additional column titled “+ [i]” could be repeated for the unvoiced consonants, too. Examples: ποιοι [pç`i] (= which ones (masc.)), πιει [pç`i] (=s/he/it will drink) (note that there is no stress mark on the previous words, because they consist of a single syllable), ίσιοι [`isçi] (= straight (pl., masc.)). As before, such examples are rather rare.
Also, the earlier note about the pronunciation of μια vs. μία generalizes to this rule: whenever there is a stress over [i] (usually the iota), there is no forced palatalization, and the [i] and the following vowel make up two syllables. In the examples below I show the separation of syllables in pronunciation:
As another, triple-example, consider this:
As can be inferred from the previous examples on voiced and unvoiced consonants, the phenomenon of palatalization (whether regular or forced) in Greek is very common because several morphological changes of nouns denoting plural, genitive case, or nominative case in the feminine gender, are such that they imply palatalization. This is part of the reason that explains the predominance of palatalization in the language.
3. Exceptions in Forced Palatalization
For native Greek speakers, to avoid regular palatalization is impossible: the sounds [ke], [xe], [ni], etc., do not exist in the repertoire of Greek phonetics. (The corresponding “correct” ones are [ce], [çe], [ɲi], etc.) Forced palatalization, however, includes several exceptions.
Of course, all these exceptions make it hard, if not downright impossible, to know how to pronounce a word given its written form. Native speakers of Greek probably think this is a problem for learners of the language, only. They’d think that once one knows Greek natively, one knows how to read properly (i.e., given the written form of a word, one knows how to pronounce it). They couldn’t be more wrong! There are a few cases where the pronunciation is strictly ambiguous, even for native speakers, who I suspect will be surprised by the following examples (I give them together with some context, else it is not clear what the words refer to):
When the above words shown in red are encountered out of context there is no way to know how to pronounce them, or which of the two meanings is intended.
Notice that the unpalatalized version (second row in each example) adds an extra syllable to the word. For example, the palatalized άδεια is formed by two syllables: [`a·ðʝa]; whereas the unpalatalized άδεια is formed by three syllables: [`a·ði·a].
4. Summary of palatalization rules
Summarizing this very long section, we can say the following:
Note: the phrasing “when [i] is inserted in writing” means that one of the six ways to write the sound [i] in Greek (ι, η, υ, ει, οι, and υι) is inserted before the vowel, and after the palatalized consonant. This is a mere writing convention, because in terms of sounds the [i] is hardly present.
Thanks to Greg Brush for coming up with the above summary.
This is not a rule of pronunciation that you have to learn or else you won’t be pronouncing right, but a phenomenon that occurs very frequently, so if you are aware of it you won’t be surprised by what you hear. The phenomenon occurs when an unstressed vowel is between two unvoiced consonants in a word, as in the word άσος (= ace), where the unstressed [o] is between two [s]’s, and [s] is an unvoiced consonant. Then the vowel is often not voiced. How can a vowel not be voiced? Well, you pronounce it as you would if you whispered it. Here are a few more examples: άνοστος (= tasteless), αλλόκοτος (= weird), λύσης (= of solution), μίσους (= of hatred). Two notes here: first, this is an optional phonetic transformation, which means that sometimes you’ll hear native speakers producing it, sometimes not. And second, native Greek speakers are not aware of it: they think they always pronounce the vowel normally. But you, as a non-native learner of the language, do not have the “tuned” ear of a native speaker, so you notice all the details. (There is nothing wrong with you, this is a well-known phenomenon.) So don’t ever start an argument with a native speaker about the way they really speak — they’re not going to agree with you, but you’ll know you are right.
Strictly speaking, what is described in this paragraph does not concern any phonetic phenomenon, but refers to an ambiguity that may be perceived by learners of the language in writing, when the three-letter combinations ντσ, and ντζ are encountered in some words, for example: βίντσι, μπρούντζος. Although the native speaker of Greek knows how to read these words (because their pronunciation is known), the learner may interpret the three consonants as either ντ+σ (and ντ+ζ), or ν+τσ (and ν+τζ). To clear out any ambiguity, let us note that the second interpretation is always correct, that is, these consonants are always expected to be read as ν+τσ and ν+τζ. Examples:
That double consonant letters are pronounced as a single consonantal sound has already been noted somewhere in the alphabet page. For example: κόκκινο [k`ocino] (= red), Σάββατο [s`avato] (= Saturday), αλλεπάλληλα [alep`aʎila] (= one after the other). But there are two cases that might confuse the learner: words that start with ευφ- and ευβ-. In the case of ευφ-, since ευ is in front of a voiceless consonant ([f]), it must be pronounced as [ef]. But then what happens with the following φ? Does it result in a second [f] sound after [ef]? No, the rule that says “pronounce double consonants as single” applies here, too. For example, the adjective εύφορο, -ρη, -ρος (= fertile) is pronounced [`eforo]; also: ευφράδεια [efr`aðia] (= eloquence, fluency), ευφυΐα(3) [efi`ia] (= intelligence), ευφωνία [efoɲ`ia] (= euphony). All these are words that start with the prefix ευ-, meaning “well-” in ancient Greek, followed by various stems starting with φ-. Similarly, there is a single word starting with ευβ-: it’s the proper name Εύβοια [`evia], the name of the second-largest island of Greece, but coming so close to the mainland (in fact, connected to it by a bridge) that it doesn’t look like an island at all on the map (see it here). Here too, the voiced β causes ευ- to be pronounced as [ev], but the resulting double [v] is pronounced as a single one.
Footnotes (clicking on the footnote number brings back to the text)
(1) The exception is Cypriot Greeks, who pronounce exactly like that: [mb`armbas] (not only this word; they nasalize all cases described here), and make some other interesting transformations. What I describe in this page is Athenian Greek, the de facto (or “received”) standard understood throughout the rest of Greece and Cyprus.
(2) Thanks to my friend, Irma Verόnica Alarcόn, for supplying the pronunciation of this Spanish word. Notice that Irma, unlike most other Spanish speakers, pronounces the letter “v” with the labiodental [v] instead of the bilabial [β]. This is a feature particular to my friend’s pronunciation, which I did not record in the transcription of “nieve”.
(3) Note the diaeresis and the stress over iota. That’s because υι (without diaeresis) is a digraph in Greek, pronounced as a single [i] (see the digraphs in the alphabet page). But the upsilon and iota in ευφυΐα denote two different [i]-sounds, so the diaeresis shows they are pronounced separately (as a prolonged [i] made of two parts, and stressed on its second part).
Back to the Greek Alphabet page
Back to the “main” page for the Greek language