18

Ancient Greek had many consonant clusters, like /pn/ in pneuma, /bd/ in bdellion, and /pt/ in pteron. But for some reason, /ks/ (ξ) and /ps/ (ψ) received special real estate in the 24-letter Greek alphabet. What was the logic behind this decision? Are there any theories by classicists on why writers of Greek found these two consonant clusters so important that they got dedicated letters?

Robert Columbia
  • 1,047
  • 1
  • 9
  • 18
Fomalhaut
  • 505
  • 2
  • 12
  • 7
    In most forms of Greek, they were the only clusters which could appear at the end of a syllable. They were quite frequent at the end of words. That’s at least one possible reason. It’s more problematic with ζ, since we don’t fully know when, where, if or to what extent ζ represented [zd], [dz] or even [dj]. It definitely represented some kind of cluster(s) originally, but not one(s) that could appear in coda position. – Janus Bahs Jacquet Sep 24 '23 at 10:09
  • 2
    @JanusBahsJacquet similarly, the Hebrew letter "צ" nowadays maps to the consonant cluster /ts/, both in Modern Hebrew and in most spoken/recited forms of ancient Hebrew, but may have had a different pronunciation in antiquity. – Robert Columbia Sep 24 '23 at 20:04
  • 5
    @RobertColumbia To be really technical, Modern Hebrew צ is an affricate, not a consonant cluster. – Fomalhaut Sep 25 '23 at 02:52
  • 1
    @RobertColumbia tsade has always been an affricate as far as we can tell (certainly since Proto-Semitic, and likely since Proto-Afroasiatic although it's harder to tell given the controversy around reconstruction there) – Tristan Sep 25 '23 at 08:30

1 Answers1

16

There are multiple possible reasons.

Synchronically, /ks/ & /ps/ are the only clusters that commonly occur word or syllable-finally and they also frequently occur as a result of inflection. Other clusters typically only occur word or syllable-initially or across syllable boundaries, and either as part of the root or due to derivational affixes.

It's possible this could have motivated their special treatment compared to most other clusters (other than zeta - /zd~dz/).

Diachronically we can look at the origin of the Greek alphabet in Phoenician.

Phoenician had a large inventory of sibilants: s, ṣ, š, and z corresponding to the letters samekh, tzade, shin, and zayin (using the Hebrew names of the letters). The exact phonetics of these is not entirely certain and varied over time and likely by location as well. At the earliest stages these were probably /ts/, /tsˤ~tsʼ/, /s/, and /dz/ respectively.

Meanwhile, by the time of the adoption of the Greek alphabet, the various Greek dialects each generally had just two sibilants: σ or Ϻ (depending on location) /s/ and ζ /zd~dz/ (with the realisation of ζ likely varying by dialect).

Unsurprisingly, there seems to have been significant confusion between the four Phoenician sibilants during the adoption of the alphabet, in particular at least three of the letters have shapes and positions corresponding to a different letter than the one they derive their name from.

Greek /dz~zd/ is represented by zeta, which has the shape and position of zayin /dz/, with its name either adjusted by anticipation of the following letter - eta, or continuing the name of tsade /tsˤ~tsʼ/.

The representation of Greek /s/ varied by location, either:

  • Sigma: which takes its shape and position from shin /s/, but derives its name from samekh /ts/
  • San: which takes its shape and position from tsade /tsˤ~tsʼ/, but derives its name either from shin /s/ or zayin /dz~zd/

That left a letter shape (and position in the alphabet) left over - that of samekh /ts/.

It has been observed by Jackson Crawford when discussing the Old Italic alphabets (and the origin of the runes) that when adopting an alphabet people tend to retain every letter of the original (and possibly add new letters if needed), but are usually reluctant to discard letters of the original even if they aren't needed to represent the phonology of the adopting language. This either leads to multiple letters representing the same sound, or letters being repurposed for entirely new purposes.

Cf Etruscan's dorsal stops, where they retained all four archaic greek letters for dorsal stops (kappa, gamma, chi, and qoppa ) to represent their two phonemic dorsal stops /k/ and /kʰ/, with the choice of whether to represent a /k/ with kappa, gamma, or qoppa depending on the following vowel (which gave rise to the modern use of <q> almost exclusively in the digraph <qu> for /kw/).

Following that principle, the Greeks put samekh to a new use for another cluster of a stop followed by /s/ - /ks/ and gave it a new acrophonic name xi.

That then left /ps/ as the only cluster of a stop followed by a sibilant (the others being /ks/ and /zd~dz/), and this asymmetry may have motivated them to create a new letter for it.

In practice, it's likely both played a role.

Tristan
  • 8,262
  • 25
  • 46
  • 1
    The "confusion of the sibilants" theory is IMO doubtful, see https://latin.stackexchange.com/questions/9493/where-did-the-greek-sibilant-letters-come-from – TKR Sep 24 '23 at 13:15
  • 1
    @tkr what specific aspect? There absolutely has been some confusion, as evidenced by the names not matching the shapes and positions in the alphabets (even with the supposed native etymology of sigma, which seems most likely a folk etymology to me, san's pronunciation definitely doesn't match its origin in tzade). I can add extra information about the more likely "classical" phoenician pronunciation of these consonants (which is certainly significantly different from the earlier stage presented here) if that's the objection – Tristan Sep 24 '23 at 13:44
  • On the names see my comment to the selected answer in the link. – TKR Sep 24 '23 at 14:31
  • @TKR I don't see how that helps tbh. Why did this one letter receive a -ma? Why represent an original /n/ before an /m/ with a gamma? Why would clipping tzade give "san"? How do you get the nu? Either it should be "sa" or "sas" < "sad" which would be phonotactically disallowed. This involves far more special pleading than confusion of sibilants – Tristan Sep 24 '23 at 15:53
  • Thank you for this well-written answer. But before I give you the green checkmark, do you mind if you could explain for me the difference between "synchronically" and "diachronically"? – Fomalhaut Sep 24 '23 at 17:00
  • 3
    Fair enough, but deriving sigma from samekh or san from sin seems equally hard to me, and a priori a confusion scenario is unlikely because people learn letter names in order. – TKR Sep 24 '23 at 17:12
  • 3
    @HolyKnowing synchronic = "at the same time", diachronic = "through time". Basically synchronic linguistics looks at the present factors affecting speakers' choices, whilst diachronic linguistics looks at why speakers' choices are affected by historical factors – Tristan Sep 24 '23 at 17:41
  • "Sigma: which .. derives its name from samekh." Smegma? It is obviously relatrd to gamma, like digamma, which as W looks like shin (semidigamma?) but derives from waw, y, u, etc. lol. σ/ς bears no resemblence to shin tho – vectory Sep 27 '23 at 04:34
  • @vectory digamma was originally called wau. The name digamma was generally used to refer to it in dialects that had lost /w/ and abandoned the letter (so losing the name and instead describing its shape), which all dialects eventually did, hence that name became universal. Sigma and gamma also bear little resemblance - the only thing they have in common is being at-stems ending in -ma, and that's not exactly unusual (many loanwords get adapted as various t-stems and an at-stem -ma suffix is a common abstract-noun suffix) – Tristan Sep 27 '23 at 08:52
  • worth noting as well that the lower case forms are of ~byzantine era, the upper case forms are the older classical ones, but in general you should be looking at the shape in the archaic Greek alphabet, during which time sigma looked like a W on its side (so similar to a classical/modern Σ, but with angled top and bottom strokes, not parallel horizontal ones) – Tristan Sep 27 '23 at 08:55
  • What is the evidence of Greek calling it wau (cf. en.WP, "This form itself is not historically attested in Greek […]"). I thought this is a modern convention adopted from Hebrew ‹ו› waw, but modernity begins with Latin, "[…] from descriptions by contemporary Latin grammarians, who render it as vau." – vectory Sep 28 '23 at 13:10
  • @vectory, ϝαῦ /wau/ is not attested, but βαῦ /vaf/ & οὐαῦ /u.af/ are attested in later Greek after the letter had fallen out of general use except as a numeral. Together with Latin vau, and the knowledge of the Hebrew /vav/ and Syriac /wau/ names (which generally correspond with the Greek letters, suggesting they're cognate with the Phoenician names) it being the original name is pretty much certain. – Tristan Sep 28 '23 at 13:22
  • there's no rhebus principle here. It's the acrophonic principle. And the derivation of the Greek alphabet from Phoenician is not in any doubt. The vast majority of letters correspond in clear ways in sound, shape, name, and position in the alphabet. That cannot be explained by any other theory. It's not impossible there was an intermediate alphabet, but a derivation from Linear A or B, or Anatolian Hieroglyphs is simply not possible – Tristan Sep 28 '23 at 15:19
  • There aren't many scholars in a position to doubt the theory because reasonable doubt requires exceptional evidence. It is not enough to show that Χ and Ψ are not from Phoenician. Nevertheless, the fact that final consonants were frequently not spelled in Linear B (ἄναξ (ánax), wa-na-ka /wánaks/) is significant for this question. – vectory Sep 30 '23 at 14:32