How did σσ differ from σ?

Question

Varro mentioned in this answer:

I think it's highly likely that originally Greek σσ had a distinct sound from σ which made it a closer match to a foreign [ʃ] than σ would have been, which is why it turns up in μεσσίας (and perhaps "Elissa"), but I have no reason to think the same applies to Latin.

I'd never heard of this before, but it would make sense with the transcriptions of messías and Elissa. On the other hand, we see Iēsous with a single sigma.

What is the consensus on this? Is there other evidence for σσ having a distinct sound? And if so, what might it have been?

Answer 1

My guess is that this refers to the origin of some cases of σσ in palatalization (various examples noted in Wikipedia: the outcome of the processes is dialect-dependent). In principle, [ʃ] is a plausible outcome of a process of palatalization and may have been involved in the development of eventual [sː].

However, I'm not sure it's correct that the σσ in μεσσίας has anything to do with this.

I believe we have metrical evidence that σσ is a long consonant, as would be expected from its spelling with two letters.

"On the pronunciation of Ancient Greek Zeta", by Sven-Tage Teodorsson (1978) says the following about this sound's development:

The history of the phonetic correspondant of ⟨Ζ⟩ is parallel to that of Ionic-Attic ⟨ΣΣ⟩/⟨ΤΤ⟩, which stand for the reflex of *[t(h)-j], *[k(h)-j], *-[tw]-, but this reflex never fused with [st]. The result of the palatalization was probably at the affricate stage *[tʃ] when the alphabet was introduced. The use of one sign ⟨Ͳ⟩ (beside ⟨ΣΣ⟩ in Ionic dialects, and ⟨[unable to reproduce character*]⟩ in Pamphylian, indicates the monophonemic interpretation. The chronology of the following Ionic development *[tʃ] > [ts] > [sː] > [s] (Koine) is not established.

(page 327-328)
*the Pamphylian variant of Sampi doesn't seem to be in Unicode yet

The Ionic(-Attic) development may perhaps have been *[tʃ] > [ʃː] (> [sː]) and *[ʒdʒ] > ʒː] (> [zː]) instead (cf. Witton 1898).

(page 329)

Answer 2

This seems extremely implausible to me for a couple of reasons.

Odysseus is spelled as both Ὀδυσσεύς and Ὀδυσῆος (Iliad 1.138) as needed in order to fit the meter in Homer. It seems like a bit of a stretch to do this if the sound of σσ was actually the qualitatively different [ʃ]. He's an epic figure who was respected. Christians don't arbitrarily say Shaint Shteven, and in the US we don't say Thomash Jeffershon whenever we feel like it. It would sound like some kind of deliberately disrespectful joke.

It would also be extremely odd that (1) σσ was the only digraph in the Greek writing system, but no ancient writer ever remarked on it; (2) σσ stood for the only non-geminate consonant with the power to affect poetic meter in the same way as a double or geminate consonant; (3) juxtapositions like αὐτὸς σὺ=ΑΥΤΟΣΣΥ were just coincidentally treated exactly the same for metrical purposes as words spelled with σσ, even though the sounds were qualitatively different; (4) just like λλ and νν, [ʃ] is never word-initial or word-final, and never occurs in larger consonant clusters.

Answer 3

Coming back to this several years later, some scholars do believe σσ was used to represent a different quality of sound than σ—at least in Anatolia. From Obrador-Cursach's The Phrygian Language page 58, in the context of an alternation between Ανγδισι, Ανδισσι, and Ανδξι in Greek transcriptions:

Indeed, we cannot ignore that ⟨σσ⟩ is used in other Anatolian areas to represent a sound other than /s/. This is the case of Carian [personal names] arliš and uśoλ, found in Greek inscriptions as Αρλισσις and Υσσωλλος respectively.

Based on this, he cites Gusmani for a claim that the Phrygian name could have had a palatalized [ʃ], especially given the surrounding high front vowels.

Adiego's The Carian Language discusses these personal names in only slightly more depth, but does mention that the Carian š and ś are transcribed medially as σσ, while s is transcribed as σ.

However, Melchert (in the Cambridge Encyclopedia of the World's Ancient Languages) points out that Carian λ reflects a historical geminate *ll, and is always transcribed into Greek with two letters (variously λλ or λδ). Similarly, at least some instances of ś go back to historical *ss. He warns that "The nature of the contrast with […] s remains to be defined", but it seems entirely possible to me that Carian λ, ś, and š were actually longer than l and s.

In this case, counter to Obrador-Cursach, σσ could simply indicate a quantity difference instead of a quality difference. Carian phonology is still not very well understood and I'd be hesitant to pin too much of the theory on it.