when we see a sound change diagrammed as *σπενδσω > *σπενσω > σπεισω, do you take that to mean that at some point in time people actually said "σπενδσω" and then "σπενσω" and finally (in the historical period) "σπεισω"?
I have no way of knowing this, and maybe no one else does either. But I suspect that the assibilation or assimilation of -δσ- to -σ- (in effect, the disappearance of δ) occurred when sigma became the marker of futurity. That seems to me like a plausible and "natural" (whatever that means) sound change -- something like that happens in casually spoken English -- although in general the motivations for sound changes are poorly understood. The change from -νσ- to -σ- with compensatory lengthening would have to have occurred after the disappearance of δ, which brought -ν- into contact with -σ-.
All of this happened before Greek began to be written, so it's really speculative to ask how the actual pronunciation changed. The absence of writing meant that there was no stabilizing factor that would impede changes in pronunciation.
Incidentally, Smyth sec. 532 says that the sigmatic future arose out of the sigmatic aorist (first aorist) subjunctive, but of course some verbs have second aorists, so the process must have involved a generalization of the sigmatic aorist with ε/ο vocalism as the future marker. I haven't chased this down, though.
Smyth also asserts (sec. 100):
ντ, νδ, νθ before ς form νσς (98), then νς, finally ν is dropped and the preceding vowel is lengthened (37).
It seems unlikely to me that -νσς was ever part of the sequence of changes. The only evidence I can detect is the Homeric (and post-Homeric poetic) form ποσσι instead of Attic ποσι, but that can probably be explained by archaic formulas that originally included the word
podsi, where
aoidoi lengthened the first syllable to ποσσι when the sound change occurred so that the old formula would continue to fit the meter. Having a word shaped _υ alongside a word shaped υυ would have offered metrical flexibility for aoidoi, and subsequent poets no doubt took advantage of that following the Homeric model. But, hey, what do I know?
is there a post-Smyth source on the evolution of the ancient Greek language, from a linguistic perspective, you might recommend?
The rules of "euphony" in Smyth are really not bad, as long as you realize that "euphony" is something of a misnomer. These are descriptions of low-level, relatively recent (in relation to the classical language), phonological changes. I think a discussion of these rules is also found in the
Cambridge Grammar of Ancient Greek.
To go back further, you are getting into pre-Greek and before that into Proto-Indo-European. In English, the most complete exposition is in Sihler,
A New Comparative Grammar of Latin and Greek, but this work is dense amd difficult and frustrating to use, and on top of that very expensive. I generally use two French books,
Lejeune, Phonétique historique du mycénien et du grec ancien for historical phonology, and Chantraine,
Morphologie historique du grec, when I want to educate myself on specifics in the history and pre-history of ancient Greek. These date from mid-century, and are probably less up-to-date than Sihler (though apparently Sihler too is showing its age), but I find them -- by orders of magnitude -- much better organized and easier to use than Sihler. There's also a German work by Rix on Greek historical phonology,
Historische Grammatik des Griechischen, but I find this too algebraic.
I should emphasize that I'm no scholar myself, just a reader of ancient Greek with some interest in the history of the language.
