Go to previous page Go to main page Go to next page

The Sound of Greek

In this chapter, I will attempt to describe the Greek phonemes without introducing the Greek letters (except for a handful used for notational purposes) and their pronunciation, which is the object of the following chapters. I believe that the decoupling of phonology and alphabet facilitates learning each in turn.


Greek exhibits a remarkable (and unappreciated) symmetry and completeness in its phonology, both in consonants and vowels. Here, "completeness" should not be construed as implying that Greek comprises all possible sounds (that is a task that even the humongous IPA is struggling with), but rather that it does not have any gaps in its phonology. More details on the phonological completeness of Greek will be provided at the end of the chapter, after the individual Greek sounds are introduced.


In order to present the sounds of the Greek language, I will use three different approaches, each having its own advantages and shortcomings:

My hope is that each reader will find at least one of the approaches helpful for understanding and reproducing each sound.


As will be evident from the chapter on the pronunciation of the Greek letters, there is not a one-to-one correspondence between Greek letters and Greek sounds, as is also the case for many languages. In order to represent the various sounds more accurately, one has to use some kind of phonetic notation. Even though I will also make use of the IPA, I will for the largest part use my own notation system, let's (…modestly :-) ) call it "Angelos' Latin-based Phonetic Alphabet for Greek" or ALPAGIn the old version of these pages, I used a slightly different notation system, which was more symmetric, but employed unconventional (e.g., *bh* for /v/ and *ph* for /f/) and two-character (e.g., *Xh* and *X'* for fricative and palatal versions of *X* respecively) representations., which is largely based on the Latin alphabet, but without the shortcomings of the IPA, notably the weird non-Latin (actually, non-anything) symbols, such as ŋ, ɲ, ɟ, ʎ, etc. I have tried to use the normal Latin characters with their most popular phonetic values as much as possible; I have filled the gaps of the Latin alphabet with a few Greek characters (θ, δ, χ, γ). I have "extrapolated" the remaining sounds using some notational conventions:
IPA a/ɐ/ɑ e/ɛ i o/ɔ u p b f v t d θ ð c ç ɟ ʝ k g x ɣ m n ɲ ŋ s z t͡s d͡z r l ʎ
ALPAG a e i o u p b f v t d θ δ k χ g γ k g χ γ m n n s z ts dz r l l


The Greek vowel sounds coincide with the five vowels of the Latin alphabet: a, e, i, o, u, which will be used for their ALPAG representation.

Values & Examples

Examples are provided in English and four popular languages, wherein the pronunciation of the letters in bold represents (one way of pronouncing) the corresponding Greek vowel.

ALPAG IPA English Examples Other-Language Examples
*a*a /a/, /ɐ/, /ɑ/ father, spa, butt banderas[SPA] pasta[ITA|Italian] ca[FRA|French] Angst[GER]
*e*e /e/, /ɛ/ bed, said España[SPA]terra[ITA], che[ITA] café[FRA], frère[FRA] spät[GER], Fest[GER]
*i*i /i/ seat, Greek, decent viva[SPA]pizza[ITA] oui[FRA] Wiener[GER]
*o*o /o/, /ɔ/ saw, thought, dog costa[SPA] cotto[ITA] eau[FRA] Oktoberfest[GER]
*u*u /u/ fool, would cubo[SPA] tiramisù[ITA] tour[FRA] kaputt[GER]

If you are a speaker of Spanish, you have it easy: Greek has the exact same five vowel sounds as Spanish. The same can be said about Italian (just forget any distinction between open and closed) and (I guess) most Slavic languages. German "unadorned" (non-Umlaut) vowels are also, for all practical purposes, identical to the Greek (note, however, that e and o are generally more open in Greek than in German). If your mother tongue is English, you have the most difficult task, as you must learn to pronounce vowels monophthongally; if you have learned one of the aforementioned languages, you will be better prepared; if not, I can but give you one advice: drop the glides.

Vowel Chart

Greek Vowel ChartIt is not possible to pinpoint the Greek vowel sounds with sufficient accuracy in the IPA vowel chart. The reason is that the symbols used by the IPA correspond to a very fine quantisation of the vowel space, which is meaningless for a speaker of Greek. For example, in the particular case of *e*, there is no... competition from neighbouring phonemes (as between "é" and "è" in French) and a Greek speaker does not feel compelled to be very precise when pronouncing it; the produced sound may pend anywhere between (or even beyond) /e/ and /ɛ/ depending on the mood (lazy or diligent), context (surrounding phonemes) or intention (imitation of other voices). Furthermore, even if it can be argued that the vowels pronounced by Greek mouths are, more or less, identical (and represent a point in the vowel chart), few Greek ears can detect small divergences (e.g., between /e/ and /ɛ/) in the sound of a particular vowel, at least in a Greek context. I have, therefore, represented the Greek vowels as regions, instead of points, on the vowel chart, the most common value of the vowel being approximately represented by the center of the region. In the chart, the vowels on the right are "rounded" and those on the left are "unrounded" (more on "rounding" in the next section).


Vowels are the sounds that are produced with a relatively little-constricted vocal tract. The least constricted sound, the archetypal vowel, is *a*. It is the sound one makes with the mouth wide open, like when visiting the dentist or the otorhinolaryngologist. It is the sound of pain, awe, admiration, surprise. When pronounced, the top surface of the tongue is the furthest apart|one does not need to strain too much to pronounce it, though; just open the mouth and say AH! :-)|300 from all points in the mouth cavity. Thus, *a* is known as the open vowel (maximum cross-section of the vocal channel). It is also named central vowel, because it is neither back nor front (to be defined later).

The other vowels involve a certain constriction of the vocal tract by the tongue and (optionally) the lips. Tongue constriction occurs along two axes: a front vowel axis from *a* towards the hard palate and a back vowel axis from *a* towards the soft palate or velum. The least distance|without being considered (fricative) consonants between tongue and mouth walls (actually, roof of the mouth) along the two axes corresponds to the close vowels (minimum cross-section of the vocal channel) *i* and *u* respectively. In the case of *u*, a secondary constriction of the lips (indicated by the "phantom arrows" in the figure) complements that of the tongue.The linguists refer to this complementary constriction with the strange term lip rounding, as if it could also be achieved by... "lip squaring". :-) Personally, I find the complementary constriction by the lips necessary to produce a clear (*u*) sound, to compensate for the lack of flexibility of the mid-back of the tongue, which cannot produce the same constriction at the velum as the mid-front of the tongue produces at the hard palate for the pronunciation of *i*; but, of course, this impression may be due to my lack of sufficient exposure to unrounded back vowels. Due to the location of the narrowest part(s) of the vocal tract in each case, *i* and *u* may be considered as palatal vowel and (labio-)velar vowel respectively: further constriction along the left axis results in the palatal fricative consonant *γ* and along the right axis or the lips results in the velar fricative consonant *γ* or the labial fricative consonant *v*.The constriction has to occur only at one of the two places (velum or lips), otherwise one of the two articulations (*v* or *γ*) would mask the other. However, the linguists use different terms to describe them: front vowel and back vowel respectively, based on the location of the part of the tongue that is closest to the roof of the mouth.

About midway along the two axes are the other two Greek vowels, *e* and *o* respectively. Those involve a more open articulation towards *a* and are, therefore, called mid vowels. Other languages distinguish more vowels along each axis, but Greek has only one in each. As with *u*, *o* also has a compensatory lip constriction, which is analogous to the tongue('s velar) constriction (greater than that for *a* and smaller than that of *u*).


Values & Examples

In the provided examples, the pronunciation of the letters in bold represents the corresponding sound. Examples between braces {.} refer to approximate pronunciation or to one of many alternative pronunciations. When the English example may not be sufficient for describing the corresponding sound, examples in other languages are provided. Please, also pay attention to the remarks provided as footnotes or tooltips.

ALPAG IPA English Examples Other-Language Examples
*b*baThe sound samples for the consonants are for their combinations with *a*, i.e., *ba*, *da*, etc. /b/ boat, bill
*d*da /d/ doll, day
*f*fa /f/ football, fish
*g*ga /g/ goal, gum
*k*ka/k/account, pink
*l*la /l/ left, love Greek *l* is never "dark" (except in dialects influenced by Turkish).
*m*ma /m/ mean, emission
*n*na /n/ noun, annoy
*p*pa /p/ appeal, please
*r*ra /r/ {river|particularly Scottish pronunciation} perro[SPA]
*s*sa /s/ stay, assign
*t*ta /t/ assist, mutant
*v*va /v/ vase, reveal
*z*za /z/ zoo, cousin
*θ*tha /θ/ thesis, thin
*δ*dha /ð/ this, brother An American colleague once told me that he could see no difference between the "th" of "this" and that of "thin". This confusion is, evidently, caused by the identical spelling and the fact that the existence of voice (which is the only difference between the two sounds of "th") is less perceptible for some (allegedly, the distinction between voiceless "Kate" and voiced "gate" is problematic for native Finnish speakers).
*χ*kha /x/ {loch|particularly Scottish pronunciation} Bach[GER], reloj|not to be confused with the /h/ pronunciation of some American-Spanish dialects[SPA]
*γ*gha /ɣ/ {woman} English "w", being essentially a semi-vocalic /u/, has two places of constriction: one at the back of the mouth (between the rear part of the tongue and the velum) and another at the lips (so-called "rounding"). Greek *γ* has only a (somewhat greater) constriction at the former location. When Greek *γ* is followed by *u*, the latter's "lip rounding" combines with the former to produce approximately the sound of "wo" in "woman". Before any other vowel (or consonant), the pronunciation is approximately that of "w" in "wait", "weep", etc. without the "lip rounding". {fuego[SPA]} Spanish "g" (before back vowels, such as /a/, /o/, /u/) is rather a so-called "approximant" than a fricative like Greek *γ*. One can say that it lies "between" English (hard) *g* and Greek (soft) *γ* in that it is "smoother" than the former, but less "perspicuous" than the latter. In some American Spanish, "g" tends to be pronounced like English *g*.
*k*kja /c/ keep, Kyoto équipe[FRA]
*g*gja /ɟ/ give, get The English examples of the palatals *k*, *g* may seem awkward to some, as one may wonder what the difference between, say, the "g" in "get" and "gum" is. In fact, Wikipedia describes this sound (*g*) as not existing in English and only "approximates" it with "g" of "get" and, strangely, "j" of "jump"! It is a common misconception that palatalisation of the velars produces not palatals, but palato-alveolars (cf. WILS05, p. 947, "change of pronunciation from keep ([kip]) to cheap ([tʃip])"), in essence like the development of "c" before front vowels from (assumed) /k/ in Latin ("centrum") to present-day /tʃ/ in Italian ("centro"). This could not be further from the truth, for even the English velars k and g cannot be easily pronounced as pure velars (/k/ and /g/) when followed by front (palatal) vowels (such as /e/ or /i/), as this requires that the (inflexible mid-back part of the) tongue be placed at two different points (velum for /k/, /g/ and palate for /e/, /i/) within the same syllable; one can easily notice that the location of "k" and "g" in "keep" and "give" is more forward than in "coat" and "goal", which is in line with the Greek practice. It seems that even professional linguists have trouble identifying that English velars before front vowels are palatalised, at least to some extent (for the two extremes, cf. "Transfer of Greek palatals in L2 English", which contends that "k"/"c" and "g" in English are pure velars even before front vowels, and WILS05, p. 948, "It is well-known that in many languages the velar stop consonants [k] and [g] are articulated further forward on the palate when they appear immediately before front vowels such as [i] and [e] than when they appear immediately before back vowels such as [ɑ]" including English and "'[t]he more front the vowel, the more front the velar' (p. 89) holds in all of the languages for which data were available" with reference to "Keating and Lahiri (1993)"). So, if you are still confused what *k* and *g* stand for, simply pronounce them as "k" and "g" are pronounced in the above-provided examples of English and in German. The main difference between English and Greek in that respect is that in the former the palatals are purely allophonic, whereas in the latter they may also form minimal pairs with normal velars, as in *káto*kjato (Κιάτο|a town in the Peloponnese) and *káto*kato (κάτω|down). So, the correct pronunciation of the combination "k"/"g"+*i*+vowel may be a challenge for English speakers. spaghetti[ITA], Lamborghini[ITA],guerre[FRA]
*χ*khja /ç/ {Houston, humid|not the /justən/, /juməd/ pronunciation of... George W} ich[GER] German "ch" (which is velar near a back vowel) has a palatal allophone near front vowels, similar to Greek *χ* and *χ*, with the difference that the place of articulation is determined by the preceding letter in German (cf. "Bach"bachsource LEO and "ich"ichsource LEO) and by the subsequent letter in Greek (cf. χέρι|hand, arm=*χéri*xeri and χάρη|grace, favour=*χári*xarh).
*γ*ghja /ʝ/ Yale, yes There is some confusion about the approximate vs fricative character of English (as well as Spanish, French, etc) consonantal "y". I guess there is also a lot of variation with many speakers pronouncing it as an approximant or "semivowel", but one would be too picky in identifying a significant difference between Greek *γ* and consonantal "y". Just to set the record straight though, the pronunciation of the word "φωτιά|fire"fwtia as [foˈtja], cited by Wikipedia as example of a Greek palatal approximant, is outright erroneous; the value of "ι" in that example is the voiceless palatal *χ* (/ç/). {ja[GER]}, {yo[SPA]}, Juventus[ITA]
*n*nja /ɲ/ {new}, {lasagna} España[SPA], cognac[FRA], bagno[ITA]
*l*lja /ʎ/ {million} "million" is a very crude approximation of the Greek palatal "l", but the closest one can come up with. If you do not understand (or cannot reproduce) the Italian "gl", then pronouncing it as "lli" of "million" is an acceptable alternative, since Greek palatal "l" is produced only when (normal) *l* is followed by (short) *i* and another vowel (in the same syllable). tagliatelle[ITA]
**nga /ŋ/ finger, sing
*ts*tsa /ts/ {tsunami} Blitzkrieg[GER], pizza[ITA]
*dz*dza /dz/ pods

At times, it has been argued that the last nine sounds (*k* - *dz*) should not be considered as independent phonemes, but merely as allophones (or combinations) of other phonemes. Whatever the case may be, they are real and distinct sounds that exist in Greek and need to be described, in order to provide a complete overview of Greek phonology.


There are four categories of Greek consonants: basic consonants (16), nasals (4), sibilants (4), and liquids (3).

Basic Consonants

There are three properties that distinguish the basic consonants from each other: location (or place of articulation), voicing and sustainability. I consider the following detailed discussion of these three properties essential for describing some Greek sounds that do not exist in other languages; for example, I find it insufficient to describe the sound of Greek "γ" merely as "soft German g" and I cannot otherwise explain to a Spaniard how "χέρι" (*χéri*kheri) differs from, say, "jerbo" (/'xerbo/jerbosource LEO)

Four different groups of consonants can be identified according to their place of articulation, as illustrated in the adjacent drawing:

Voicing is an on/off feature of consonants, which refers to the contemporaneous participation (vibration) of the vocal cords or lack thereof. Consonants pronounced with voicing are called voiced, while those that lack it are termed unvoiced or voiceless. The difference between voiceless and voiced consonants of the (otherwise) same type is exemplified by the difference between "pill" and "bill", "feel" and "veal", "tame" and "dame", "Kate" and "gate", "seal" and "zeal".

Finally, sustainability refers to the inherent ability of a consonant to be (uniformly) pronounced for a time that exceeds an "instant". Some consonants, like English "p", "b" and "t", involve a complete obstruction of the vocal tract and its violent release by the accumulated air pressure, at which instant (and only then) the sound of the corresponding consonant is produced. These consonants are, thus, named plosives (due to the… air dynamics) or stops (due to their duration). If the obstruction is moderately relaxed or slightly modified, a narrow passage is created, through which the air can be pass to produce some friction; a typical example is English "v", wherein the lower lip rests not on the upper lip to produce the closure of "b", but on the upper teeth to define an imperfect closure due to the interstices of the teeth, through which air can freely flow to produce a sound that can be protracted as long as one desired. These consonants are named fricatives or spirants (air dynamics) or continuants (duration).

Starting with the labials, which is the consonant series that exists in most languages|One notable exception is Spanish, where there is no distinction between B and V in the same form as in Greek and can thus serve as the model for the other series of basic consonants, the two stops *b* and *p* are produced when air pressure builds behind the tightly closed lips and is released in an audible "plosion". They are both thus "plosives" or "stops", as their sound cannot be prolonged for more than the instant of the "plosion". The difference between the two is voicing, which results in a "fuller" sound in the case of *b*. In order to produce protracted sounds or "continuants" out of *b* and *p*, the obstruction of the vocal tract has to be relaxed, so that passage of a continuous airstream be possible. One way to achieve this would be to partially open the closed lips, e.g., in the middle; this would produce a narrow opening similar to the position of lips while whistling. The (bilabial) continuant counterparts of *b* and *p* produced in such a manner are indicated in IPA with the Greek letters /β/ and /ɸ/, even though these sounds do not exist in Greek. In fact, the Greek labial continuants are identical with the English ones, namely "v" and "f" (hence the notation *v* and *f*), where the lower lip rests against the bottom of the upper teeth providing a passage through the interstices of the teeth.For this reason, linguists refer to them as "labiodental". Personally, I do not consider the bilabial continuants anything more than approximants. The more stable and distinct sound produced by the cooperation of the lower lip and upper teeth is probably the reason why the labiodental continuants are so much more widespread in the world languages than the bilabial continuants.

Similarly, the dental stops *d* (voiced) and *t* (voiceless) are produced when the airstream is trapped by the perimeter of the tongue sitting on the tops of the upper teeth or gum.Whether it is the tip or blade of the tongue on one hand and the teeth, gum or alveolar ridge on the other it does not make a big difference, for a Greek ear would perceive all the (similar) sounds produced in these manners as dental stops. The dental continuants *δ* and *θ* are not produced by merely distancing the tongue from the teeth/gum (which would produce blurred z/s-like sounds), but by placing its perimeter along the bottom ridge of the teeth, to again take advantage of the well-defined interstices of the teeth.

Skipping the tricky palatals for the time being, the velar stops *g* (voiced) and *k* are formed when the rear part of the tongue is raised towards the soft palate to block the airstream at the rear part of the mouth. As there is no well-defined gap inside the mouth, the only way to provide a narrow passage is to distance the tongue from the soft palate barely enough to let the airstream through.Whether the most proximate point of the tongue to the palate is more forward, more backward or at the same height as the point of contact for *g* and *k* I cannot tell (although the former seems more likely). Thus are formed the velar continuants *γ* (voiced) and *χ* (voiceless).

Coming now to the palatals, it has to first be noted that in Greek and several other languages these are the result of the combination of a velar with a front (palatal) vowel, such as *e* and (chiefly) *i*. Thus, when one compares the "g" of, say, "got" and "get" and the "k"-sound of, say, "cup" and "keep", it becomes evident that the latter are pronounced "further forward on the palate" (WILS05, p. 948), with the middle (instead of back) of the tongue touching the hard (instead of soft) palate. So, for all practical purposes, the Greek palatal stops *g* (voiced) and *k* (voiceless) are not to be distinguished from English "g|But only when it is a, so-called, hard g" and "k" as pronounced before front vowels. The loosening of the closure of the palatal stops to let the airstream through (in a manner analogous to the generation of *γ* and *χ*) produces the palatal continuants *γ* and *χ*. The former corresponds to more or less English consonantal "y", as in "yell", "yawn", etc, while the latter can be approximated with the consonant of "hew|Although George W and other Americans make it sound like the y of you".

This concludes the presentation of the basic consonants, each of which can thus be described as a combination of three properties, namely location ("L"=labial, "D"=dental, "P"=palatal, "V"=velar), voicing ("+"=voiced, "-"=voiceless) and duration ("."=stop, "_"=continuant); for example, *v* can be characterised by the triplet "L+_". These three properties can be combined to 4x2x2=16 different triplets, which is exactly the number of basic consonants in Greek. Thus, it is possible to arrange the basic consonants in a table with each dimension defined by one property; as this would require a 3D table, in the above illustration I have "flattened" it into four sub-tables corresponding to cross sections at each place of articulation. Appended at each table is the corresponding nasal for that place of articulation, which will be described next.


The four Greek nasal consonants are produced in a way similar to its stops, namely the mouth cavity is blocked at the same locations as for the stops. Thus, *m* involves a closure of the lips (like *p* and *b*), *n* involves the perimeter of the tongue tightly resting on the roots of the teeth or gum (like *t* and *d*), *n* involves the middle of the tongue pressing against the hard palate (like *k* and *g*) and ** involves the back of the tongue pressing against the soft palate (like *k* and *g*). The only difference is that the airstream is allowed to escape through the nasal cavity, the entrance of which is usually blocked by the velum in all other (non-nasal) sounds.

Despite the complete closure of the mouth cavity, all nasals are to be considered as continuants, since their sound can be sustained by the airstream escaping through the nose (the mouth cavity acting as a resonator). It is generally considered that they are also voiced: all assimilatory phenomena in Greek agree that *m* is voiced; most of them (with the possible exception of non-voicing of *s* before *n*) indicate that *n* is also voiced, a conclusion that we can fairly safely extend to *n* and ** (which, being themselves results of assimilation, do not participate in further assimilatory phenomena that would reveal their voicing).


Sibilants, in particular *s*, are sometimes described as "whistling" or "hissing" sounds. *s* (voiceless) and *z* (voiced) are produced when the airstream is guided through the interstices of and small opening between slightly parted teeth; the tongue is loose and its tip is placed behind the opening between the upper and lower teeth at a relatively small distance (but still greater than the distance between the tongue and the palate in the palatal or velar continuants; note that there can be no comparison with the others, since in the dental continuants the tongue touches the teeth and in the labial continuants it does not participate). There is no doubt that *s* and *z* are continuants. They are also generally considered "dental" (or rather "alveolar", but this term seems unnecessarily specific and even flawed, at least for the Greek sibilant continuants, since the tongue neither touches the alveolus nor does it come closer to it that to the teeth).

Clearly dental are the other two Greek sibilants, *ts* (voiceless) and *dz* (voiced). They start exactly like *t* and *d*, but instead of being released to a fairly relaxed mouth, as in those sounds, they are released to the shape of *s*/*z*; that is, after release (of *t* or *d*), the tongue assumes the position behind the parted teeth, which it has when pronouncing *s* or *z*. Since it appears that their sound is composite, *ts* and *dz* are often defined as "affricates", a term which alludes to their resembling the "fricatives" (the various sounds which I refer to above as "continuants") or having a fricative component. However, it is evident that *ts* and *dz* are not continuants, since they are only produced at the instant of release, while any attempt to prolong their sound results in *s* or *z* being heard, but not *ts* or *dz* (SIAM88, §2,240, p.379).Another issue is whether *ts* is one sound or two (similarly for *dz*). The fact that it is represented by a single character in many languages, e.g., c|Slovene, z|German, ц|Russian, צ|Hebrew, ţ|Romanian, hints towards a single sound. On the other hand, the fact that the digraph "ts" is considered enough to describe the sound suggests that these single letters may merely be shortcuts, similar to "x" for *ks* (cf. Alexander vs. Алекса́ндр|Russian). However, it seems that, while *ks* ("x") is a true double consonant ("Mac sale" and "Max ale" sound the same), *ts* is not necessarily equivalent to *ts*, as is evident when *t* and *s* are on opposite sides of word boundary (cf. "it seals" and "its eels"). It appears that the reason for this is the difference in the place of articulation of the two constituents of "x" (*k* being a velar and *s* being a dental), while the two constituents of *ts* (namely *t* and *s*) are both dentals and can provide a combined effect in a single sound.

The Greek sibilants can, therefore, be arranged in a 2x2 matrix similar to those of the basic consonants in the above illustration:

D\V - +
_ *s* *z*
. *ts* *dz*

It is worth noting that Greek lacks the entire series of so-called "post-alveolar" sibilants ("pressure", "measure", "rapture", "injure") except for idiomatic or dialectal speech, for instance former Prime Minister Konstantinos Karamanlis who was thus pronouncing all Greek *s* (much like Sir Sean Connery's esses) or Cretan which uses it in lieu of the palatal series (much like Italian, although in Cretan they are rather "alveolo-palatal"). Any "post-alveolar" sibilants contained in loanwords are converted into normal dental sibilants , e.g., "σουτ"=*sút*sout for "shoot" and "Τσώρτσιλ"=*tsórtsil*tswrtsil for "Churchill".


The consonants of this last category are "peculiar", in that their manner of articulation is unlike anything discussed before and that they are rather isolated (they have neither voiceless nor plosive counterparts|and it is doubtful whether such counterparts exist). These include one trill (*r*) and two laterals (*l* and *l*).

The trill *r* is the one commonly called "rolled r". It is supposed to be produced when the airstream causes the front part of the tongue to vibrate. How this is done I do not know, as I have never been able to produce it the same way the other Greeks do|so, it is not in the genes! :-). The best I can do is vibrate the middle part of the tongue, which appears to be a fairly good approximation that goes almost undetected. Since this sound is a problem for native speakers of English, French and German, my only recommendation is to try your best imitation of Spanish "r" ("pero") or rather "rr" ("perro").

The lateral *l* is similar to *t* and *d*, but it is not the periphery of the tongue that rests on the roots of the teeth or gum, but only its front part (tip), while the sides are kept at a small distance from the molars. This results in an imperfect closure of the mouth cavity and allows the airstream to escape from the small gap between the sides of the tongue and the molars; hence the name "lateral".

The last sound *l* is the palatalised variation of *l*, wherein the middle part of the tongue is raised towards and presses against the hard palate, the airstream still escaping over the sides of the tongue.

*r* and *l* are often referred to as "liquids", a term that can be traced back to the ancient Greek grammarians (BEKK16, p.632), obviously because they are both related to the (always wet?) tongue. Although they are related, modern linguists distinguish them into trill and lateral, the distinction I have used above. All three Greek liquids are continuants, since they involve an imperfect closure of the vocal tract. As to the question of voicing, the liquids are rather tricky; however, there are assimilatory phenomena that clearly indicate that they are voiced, but in a few cases there are indications to the contrary (but this is not doubtless).


Symmetry and Completeness

Now that all Greek phonemes have been introduced, it must be clear why Greek is phonologically symmetric and complete. Theoretically speaking, if each phoneme is considered as a combination of discrete values of a number of properties, say values A1, A2, A3 for property A and values B1, B2 for property B, then a phonologically complete language would comprise all possible combinations of the values of these properties, A1B1, A1B2, A2B1, A2B2, A3B1, A3B2; if one pair, say A3B2, is missing, then one of the corresponding property values should be absent altogether, i.e., either A3B1 or both A1B2 and A2B2 should also be missing. In the above presentation, the properties considered were location (labial, dental, palatal, velar), voicing (voiced, unvoiced), duration (stop, continuant) for the basic consonants, only location for the nasals, duration and voicing for the sibilants.The liquids are not considered, since not all of their variations exist (in any language). In other words, none of the above presented tables has an empty cell and every cell has a counterpart at the same position in all other tables.

Compared to the Greek model, the lack of completeness of some of the major European languages is evident when one observes that, although all of them comprise the labial series, they have significant gaps in the other consonant categories. For example, English lacks the velar continuants (*γ* and *χ*), German only the voiced velar continuant (*γ*), but also the dental continuants (*δ* and *θ*), and French and Italian all of the above. Spanish is pretty consistent in terms of basic consonants (due to its lack of distinction between voiced stops and voiced continuants; "b"="v"), but it has abnormal sibilant series, having only one dental sibilant (the voiceless continuant "s") and only one "post-alveolar" sibilant (the voiceless stop "ch").


In a couple of words, mainstream Greek knows nothing of semivowels or approximants. This general aphorism needs clarification, in particular to define what is a "semi-vowel". It appears that everyone understands this term differently and providing a precise definition is not an easy task; so let me draw you a picture. :-)

The voiced velars (*g* and *γ*) and palatals (*g* and *γ*) can be considered as the "projections" of the back and front vowel axis respectively. In particular, the place of articulation of the palatal consonants is at the end of the front vowel axis. In the adjacent figure, which illustrates the vowel triangle, the front axis (*a*-*e*-*i*) is extended beyond the last vowel (*i*) towards the hard palate; its point of intersection with the palate is the point of contact of the tongue when pronouncing *g* (also *k*, but since that is voiceless it better not be compared with the vowels). *γ* (also *χ*), being a continuant, is located at a short distance from the palate, barely enough to allow the air stream to break *g*'s "barricade". Somewhere in the narrow space between *γ* and *i* is the border between (palatal) consonants and vowels. But the distance between *i* and *γ* is not enough to define an unambiguous new phoneme. Sometimes, when *i* is followed by another vowel and is shortened to form a single syllable, its shortening results in it being less clearly pronounced, the tongue not attaining the necessary distance from the palate, but pending in the "gray zone" of the "consonant/vowel border". The resulting sound is neither a clear fricative (*γ*) nor a clear vowel (*i*) and thus termed a semivowel (I guess, it could also have been termed a "semiconsonant", but for some reason its vowel character prevailed). A similar situation arises when *u* is followed by another vowel in the same syllable; however, the labial component (a.k.a. "lip rounding") in *u* prevents it from being confused with *γ*; but the tongue location during articulation of semivocalic *u* (corresponding to English "w") is certainly between *u* and *γ*, somewhere around the back part of the "grey zone".Some linguists define a further related phoneme, the spirant approximant, but they must either be splitting hair or be confused by the orthography, since the space between the close vowels (*i* and *u*) and the voiced continuants (*γ* and *γ*) is too small to accommodate one clear sound (the semivowel), much less a second one (the "spirant approximant"). To summarise the above in a few words, I understand a "semivowel" to be nothing more than a blurred (imprecisely articulated) short vowel.

Based on the above definition, it can be safely said that Greek does not have any semivowels, since it has a clear preference for "decisive", well-spaced apart sounds and for unambiguous articulation. Although an *i*-semivowel can be observed in the pronunciation of some speakers (e.g., "γεια" being pronounced as **In ALPAG representation, a connected vowel digraph (*V1V2*) represents two vowels (*V1* and *V2*) pronounced as one syllable, namely a (true) diphthong, be it opening or closing.ia instead of *γá*ghja), this is either dialectal or idiomatic.


Excluding cases of "synizesis" in poetry, diphthongal pronunciation is reserved for three vowel combinations, *ai*, *oi*, *ei*, with *i* following another vowelA forth diphthong, *ui*, may also be added (cf. VILA57, p. ιδ´, with examples ακούει|he/she/it hears/listens=*akúi*, κρούει|he/she/it knocks/beats=*krúi*), but this is not very clear, as *u* (being rendered as a digraph in Greek) is already considered a "diphthong" and as *i* and *u* are both closed vowels.. When *i* precedes another vowelVilaras (VILA57, p. ιδ´) describes the combinations of *i* with each of the five vowels as "καταχρηστηκες δηφθογγες|improper diphthongs"., instead of diphthongisation Greek prefers consonantisation. Greek *u* never forms diphthongs (possibly due to the orthography as a digraph "ου").

Stress, Intonation and Quantity

Greek stress is dynamic, like in English and most (if not all?) European languages. That is a stressed syllable sounds louder than a non-stressed syllable; for instance, in the word μαμά|mommy=*mamá*mama the (only) difference between the two (otherwise identical) syllables is the loudness of the second one.

Greek intonation, that is the "raising" of the voice (e.g., when making questions), is identical to that of English and needs no further explanation.

Finally, there is no quantitative distinction of Greek vowels (i.e., no long and short vowels), except, of course, the vowel shortening during diphthongal pronunciation. It is often maintained that stressed vowels are "long"cf. BLAS82, p. 107, "der jetzige Grieche spricht betonte Vokale lang, unbetonte kurz|the Greek of today speaks stressed vowels long, unstressed short" or "longer"cf. CARA95, p. 177, "Greek, on the other hand, pronounces all syllables distinctly and isochronously with one of the syllables having a somewhat more dominant stress and hence being slightly longer than the others because of the percussion"; however, this must be either an illusion or a conclusion based on the phonology of other languages, for I know no Greek that tends to even slightly protract his/her stressed vowels (outside poetry and singing, of course, and possibly some "musical" dialects); thus, all (*a*) vowels in πατάτα|potato=*patáta*patata(≠*patáata*pataata) are of the same length, irrespective of whether they are stressed (i.e., louder) or unstressed (i.e., quieter).Time Plot of πατάταAn inspection of the time plot of the word πατάτα (see right side, top) is enough to convince anyone that there is no such thing as lengthening of Greek stressed vowels. All three syllables ("lobes" of the plot) are equally long (any impression that the middle one is "slightly longer" than the others is a mere illusion caused by the fact that the loudness of the middle syllable amplifies the smaller "peaks", which are thus more visible than in the other two syllables). Time Plot of πατάαταOn the other hand, the time plot of the (actually) long version, πατάατα (see right side, bottom), clearly shows what a "long" vowel should look like.


Go to previous page Go to main page Go to next page