In classical Greek, the only consonant digraphs that have corresponding letters in the alphabet are ks(Ξ), ps(Ψ), and zd(Ζ, though this one's debated) -- why aren't other consonant digraphs (e.g. ts, kp, kth, sz) identified with distinctive letters? I thought it might had something to do with their frequency of use, but I am not certain. It'd be very helpful if there are some sort of frequency statistics of all consonant digraphs available.

  • 2
    Good question! It could well have to do with where the Greek alphabet comes from. Have you looked into that?
    – Joonas Ilmavirta
    Commented Jun 20, 2019 at 22:39
  • 2
    I've only just read several Wikipedia articles and Wiktionary appendices. I noticed there were more letters representing digraphs and clusters in archaic alphabets, like the Pamphylian Sampi. But what I've found so far doesn't seem to answer my question directly. Commented Jun 20, 2019 at 22:51
  • 3
    FWIW, the specific example clusters you cite (ts, kp, kth, sz) happen to be either impossible or rare in Greek. But of course one could cite others that are common (e.g. st, sk, or clusters with liquids). AFAIK there isn't a clear answer to this question.
    – TKR
    Commented Jun 23, 2019 at 19:59

2 Answers 2


The question really isn't so much why there weren't more signs representing consonant clusters as it is why the ones that do exist are there. The Phoenician alphabet, from which the Greek derives, had more signs for sibilants (𐤆 z, 𐤎 s, 𐤑 , 𐤔 š) than the Greek language had sibilant phonemes (at most two), so you could say the Greeks were incentivised to come up with creative new uses for them.

It's worth noting, though, that not every Greek alphabet had signs for [ks] and [ps]; if I may steal a map from Wikipedia:

enter image description here

Only the dark blue and red regions had signs for [ks] (a variant of Ξ for dark blue, Χ for red; it's the red alphabet that was adapted into the Etruscan and ultimately the Latin alphabet), and only the dark blue had a sign for [ps] (Ψ). The other alphabets, including the Old Attic alphabet (light blue), used digraphs; typically ΧΣ for [ks] and ΦΣ for [ps].

Eventually the Ionian alphabet, a dark blue alphabet, became the norm throughout the Greek world, and it did have ξ and ψ; why? Very likely because in Ionic, as in Attic, [ks] and [ps] are the only consonant clusters permitted at the end of a word (with exactly one exception: ἁλς). This, and the presence of the sibilant there just as in the excess Phoenician signs, must have made these clusters unusual enough in the minds of the Ionians that they decided to devote some dedicated signs to them. The single exception must not have been prominent enough to add another one for [ls].
Note that not all dialects share this restriction: most Doric dialects, for instance, permit -νς (prominently seen in the accusative plural -ᾰνς), and while some Doric-speaking regions do use a dark blue alphabet for undoubtedly interesting historical reasons, most use a green alphabet that doesn't have ξ or ψ, perhaps for exactly this reason. Conversely, Thessalian and Boeotian, Aeolic dialects which do share the restriction, don't have ξ and ψ, showing that their creation was ultimately a pretty arbitrary decision.

The matter of ζ is slightly trickier, because we aren't entirely certain where and when the Greek alphabet was initially created (if it did have a single origin), so we don't really know what the original phonetic value of the sign was. There are good reasons to believe it was [dz], however, which, unlike the later value of [zd], couldn't have been represented as a digraph and so would have needed its own sign. They could have used ζ just for the [z] sound, but—if this is right—they correctly noted that it only occurs as part of the monophonemic affricate /d͡z/ and decided to just use Ζ to represent the whole thing instead of rendering it as ΔΖ (though I believe that spelling is attested epigraphically at least once). Later, of course, the metathesis will have happened, and ζ came to stand for [zd] regardless of whether it came from /d͡z/ or /sd/.
The phonetic realisation of the Phoenician sibilants themselves is famously disputed, but it's not unlikely one or more of them were affricates, so this use of a single sign for an affricate wouldn't be an incredible innovation on the part of the Greeks—Krahmalkov even suggested /d͡z/ for Phoenician 𐤆, the ancestor of Greek Ζ, itself, which would make it even less of a stretch.
(See also this earlier discussion on zeta.)

  • 1
    I completely forgot about sampi while writing this, but it fits into this story as well.
    – Cairnarvon
    Commented Mar 21 at 19:25

My feeling is that the "doubleness" is a bit of an illusion, conflating letters which seem "double" for different reasons.

Ζ is a perfectly respectable single letter, a cognate of the Hebrew zayin. If the Greeks happen to find themselves dissimilating it in pronunciation, that is a secondary phenomenon.

Ξ is, similarly, a single letter, a cognate of samech, which is one of those Daffy Duck-style sibilant consonants which Semitic languages like and Greek doesn't. That a "splashy s" becomes a "ks" is not unparalleled. The letter "x", for instance, in parts of Iberia and also in Mexico (even in the word "Mexico" itself), is used for various kinds of "sh" sound. Again, this is a single letter for which Greek happened to find a double-consonant pronunciation.

The letters φ and χ (which you didn't mention) are clearly late additions because they come after the end of the alphabet. They had to come into existence because (a) aspirated π and aspirated κ mean something different from their unaspirated versions and so need to be distinguished in writing and (b) after quite a lot of vacillation Η turned out to be more useful as "long Ε" rather than as "aspirate". (That, again, is not unheard-of. In some accents in French the final vowel in oui has a definite "h" sound.)

Of the digraphs you mention, κπ, κθ, κφ are not true consonants, just the collision of a κ at the end of one syllable with another letter at the start of the next. But there are better examples. κτ, πτ and σθ, for instance, are genuine consonant clusters that can occur at the beginnings of words. The first two of these, at least, have a chance of being genuine "consonants-in-themselves", hybrids of their components and not just one sound after the other. Compare the "gb" in "Igbo" which is not two sounds but a single sound articulated in two places at once. If one were evolving a "rational Greek alphabet" those would be good choices for being made a single letter; but it didn't happen.

All this doesn't answer your question so much as sharpen it. Why, out of all the conceivable consonant clusters, did it become desirable to write ΦΣ as Ψ and to make no other change? I say Φ rather than Π because I think that inscriptions used to use ΦΣ rather than ΠΣ. The resemblance between Φ and Ψ also makes me think that this new letter followed a general rule for adding letters to an alphabet, which is not to add new and unprecedented shapes.

Was it, perhaps, something to do with the special character of the S sound, making it, like the H sound, feel more like a modifier of its preceding consonant rather than a consonant in its own right?

  • 4
    Some of the phrasing in this answer is puzzling to me. Of course Z and Ξ are single letters; the point is that they stand for double sounds. The Greeks didn't "dissimilate" Z. I don't know what "splashy" or "Daffy Duck-style" are supposed to mean, but samekh is thought by many to have been pronounced [s], just like Σ (though there's debate on this point). And the fact that Igbo has true labiovelar consonants seems irrelevant since this wasn't the case in Greek.
    – TKR
    Commented Jun 23, 2019 at 19:57
  • 2
    I'm not sure I understand your point fully, but two points might be helpful to address: (1) Greek treats ζ, ψ, and ξ as double consonants for the purposes of poetry, and (2) the σ in aorist and future inflections is combined by a regular rule with certain consonants, e.g. βλέπω -> βλέψομαι.
    – brianpck
    Commented Mar 20 at 21:42

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.