In ancient Attic Greek, how (un)stable were "ΝΣ"/"ΝΖ" and preceding vowels?

Question

In Latin, it is thought (as far as I know) that within a single word, /ns/ and /nf/ were always preceded by a long vowel.

This is a somewhat complicated result of a hypothesized sound change in words containing /ns/ or /nf/. The original sound change is supposed to have turned Vns and Vnf sequences to [Vːs] and [Vːf] sequences (or possibly [Ṽːs] and [Ṽːf] sequences; the exact realization and occurrence of nasality as a contrastive element for Latin vowels is somewhat debatable). According to Allen 1978, consonantal [n] was later re-inserted before the fricative in the pronunciation of Classical Latin while the vowel remained long, resulting in [Vːns] sequences (Vox Latina 2nd edition, p. 29-30). In Classical Latin, long vowels occurred even in prefixed words that started with in- or con- followed by s or f.

In Greek, the cluster /ns/ apparently was similarly unstable, but it doesn't seem to have become strictly eliminated by sound changes, and as far as I can tell Ancient Attic Greek differs from Latin in having a fair number of words where this cluster comes after a short vowel. My basis for the preceding statement is the spelling in words with epsilon-nu-sigma; I'm asking this question to learn whether the spelling is misleading, and whether we know the length of alpha, iota and upsilon in this context.

For example, the prefix ἐν- seems to occur fairly freely before σ or ζ. As far as I can tell, this prefix doesn't have an alternative spelling like εἰ- in this context, even though it does have alternative spellings ἐγ, ἐμ to represent place assimilation to following velar or bilabial consonants and ἐλ, ἐρ to represent full assimilation to a following liquid consonant. I would think that this indicates that Attic Greek words starting with ἐν-σ- or ἐν-ζ- were in fact pronounced with a short vowel [e] followed by a nasal consonant [n].

But I'm not certain, because Ancient Greek orthography seems to have indicated assimilations or sound changes only sometimes, not always. For one word spelled with ΝΣ—the proper noun Τίρυνς—Wiktionary suggests that the accentuation implies that "the standard Classical Attic pronunciation" was actually "Τίρῡς".

What is the general distribution of ΝΣ in Attic Greek, and what do we know about its pronunciation and the pronunciation of preceding vowels?

Answer 1

History and Distribution

I found a blog post that discusses some of the history, although not the pronunciation, of words spelled with ΝΣ: "Greek etymology: hádrynsis", by Bas Clercx. Published 10 July 2018. I also found a online handout by Roberto Batisti that summarizes some of the relevant sound changes: Compensatory Lengthening in Ancient Greek. Phonetics, Phonology, Dialectology.

It seems [ns] was lost multiple times in Greek, with different results. Words with original *ns (with s from Proto-European *s) ended up retaining the nasal and losing the fricative. As with intervocalic and word-initial *s, this seems to have progressed by debuccalization of [s] to [h] (which would produce *[nh]); the [h] in this cluster was lost in all dialects, but with somewhat different ultimate outcomes. In Attic, the vowel before the *n was subject to "compensatory lengthening".

In contrast, the secondary *ns resulting from palatalization of *t to *s regularly lost the nasal and retained the fricative. Compensatory lengthening of the preceding vowel occurred in these words also in Attic (in other dialects, the preceding vowel was diphthongized instead).

It seems that Carl Darling Buck wrote about the history of these clusters in works published in 1933 and 1955, but I haven't read what he says yet.

The word-final context (nominative singular forms)

Nominative singular forms ending in ΝΣ, such as Τίρυνς, are exceptional and extremely uncommon. The only others that I know of are the following:

• ἕλμινς, ἕλμινθος. The LSJ entry says "also nom. ἕλμις".

• πείρινς, πείρινθος, although I'm not sure how many examples there are of the nominative singular in ancient sources. The LSJ entry says "πείρινθος cited as nom. by Hsch., EM668.21".

(Perseus's copy of the LSJ gives ἀέρσι-πονς, but that seems to be a scanno for ἀέρσι-πους.)

Apparently, nouns with stems ending in -nt- or -n- usually have nominative singular forms ending in n or s depending on certain conditions discussed in the following paper: "Synchronic and Diachronic Derivation of Greek n- and nt-stem Nominative Singular Formations", by Ian Hollenbaugh.

The rarity of words ending in ΝΣ is a specific example of a more general rarity of resonants before Σ, and particularly before word-final Σ. The only Ancient Attic Greek word ending in ΛΣ is apparently ἅλς.

The pronunciation is still unclear to me

I found a Wiktionary page "Ancient Greek transliteration" that says that the clusters νσ~νς and νζ are "Not viable in Classical Attic, where the spelling is rare, and the ν is silent and audibly lengthens the preceding vowel instead", but I don't know how to tell whether this is accurate as a general rule, or whether it is an overly absolute description that is only based on the historical sound changes (and the single example of Τίρυνς, where there seems to be specific evidence for a pronunciation with a long vowel) without taking into account the possibility of exceptions to the sound changes.

I'm kind of skeptical that the usual distinction in Attic orthography between short ε and long ει would be lost in this context, since I don't know of any other situations in Attic where ε was regularly used to represent a long vowel.