Wiktionary talk:About Proto-Hellenic

Assumptions underlying Proto-Hellenic

@CodeCat, Angr, Erutuon Mostly directed to CodeCat. I notice that Proto-Hellenic is assumed to have (a) initial h- not yet lost due to Grassmann's Law (e.g. *honkʷʰā́ > ὀμφή (omphḗ, “voice, oracle”)); (b) neuter still in -m instead of -n (can't find any examples; I previously saw something like *poyyom but can't locate it now). What is the support for these? Benwing2 (talk) 21:30, 5 March 2017 (UTC)

Wikipedia says Grassmann's law is possibly post-Mycenaean. —CodeCat 21:49, 5 March 2017 (UTC)

@CodeCat "possibly" doesn't inspire much confidence. Wikipedia also says that the occurrence of Grassmann's law in both Sanskrit and Greek is "quite possibly due to areal influence from one language to the other", which would be impossible if it occurred post-Mycenaean. BTW it looks like Dbachmann added the "possibly post-Mycenaean" comment in August or September 2005. Do you have any actual evidence that Grassmann's Law is post-Mycenaean? If not, I think we should definitely not insert extra *h's, on the same logic you yourself use to justify not reconstructing *pekti and such in Proto-Slavic. Benwing2 (talk) 22:05, 5 March 2017 (UTC)

There's no evidence that I know of, whether it had occurred in Proto-Hellenic or not. Do you have positive evidence or are you concluding that it had occurred given the absence of evidence? —CodeCat 22:14, 5 March 2017 (UTC)

I think the alternation between ἔχω (ékhō) and ἕξω (héxō) probably indicates Grassmann's Law was still operative, because the loss of h in ἔχω (ékhō) can only have occurred after PIE initial s was debuccalized to h. But I am not sure exactly how the alternation would be analyzed, so maybe this reasoning is wrong. — Eru·tuon 22:22, 5 March 2017 (UTC)

*s > *h is also a likely early areal Indo-Greek innovation (found in Common Iranian and conditionally in Armenian), so this doesn't rule out a pre-PH date. Mycenaean also doesn't even indicate aspirates, so "possibly post-Mycenaean" seems mostly just like an ex silentio argument ("there's no explicit counterevidence").

When it comes to proto-languages such as this, where there are no authoritative dictionaries etc. to rely on, we should probably decide if we want to use a "minimalist" reconstruction (= by default do not reconstruct features that have been lost from all descendants) or a "maximalist" one (= by default reconstruct all features that aren't branch-defining innovations according to what we would expect in the parent proto-language). I would suggest the former, it's what we already do with a few cases such as Proto-Slavic where there are known chronological complications. Hence: better include things like Grassmann's or *-m > -n. --Tropylium (talk) 14:39, 6 March 2017 (UTC)

@Tropylium, Erutuon, CodeCat, Benwing2, Mahagaja First, if we assume that Grassmann's law was Proto-Hellenic, a number of Proto-Hellenic reconstructions in *seǵʰ- must be corrected.

Second, per De Decker 2015, Grassmann's law cannot be Proto-Hellenic, as it is actually definitively post-Mycenaean (and might even be post-Homeric). --Florian Blaschke (talk) 22:56, 17 December 2020 (UTC)

Difference between dz and ď

@Mahagaja, Erutuon: Is there a difference between dz and ď? I noticed that we use dz only in *dzeus and ď everywhere else, such as in *ďṓwō. --WikiTiki89 18:59, 12 April 2018 (UTC)

@Wikitiki89, Erutuon, JohnC5: If there is a difference, we should probably keep dz for words like ζυγόν (zugón) where the ζ comes from word-initial PIE *y-, but I'm not sure that source ever fails to merge with ζ < *dy/g(ʷ)y in any variety of Greek. Either way, *dzeus should probably be moved to *ďeus. —Mahāgaja (formerly Angr) · talk 14:30, 13 April 2018 (UTC)

This is the distinction between the two palatalizations in Greek: the first applies to primary (inherited from PIE) *Ty, the second to *Ky and secondary (analogically recreated) *Ty. The former gives what I've usually seen marked as *ts, *dz (> Attic-Ionic σ and ζ), the latter *ť *ď (> Attic-Ionic initial τ ζ, medial ττ ζ). See w:Proto-Greek#Palatalization. I do not know if a distinction between initial *dz- and *ď- is actually reconstructible though, or if this is simply inferred depending on the PIE origin. --Tropylium (talk) 15:40, 13 April 2018 (UTC)

…I looked over {{R:Sihler 1995}} and he does not seem to say anything about an initial distinction between *dz- versus *ď-; he only mentions *ď > δδ "in some dialects" which obviously can only be medial. At p. 194 he also has the following:

the development of G ζ- < PIE *y- (…) points to an early coalescence of *dy-, *gy-, and plain *y-.

which I think suggests that maybe we should be reconstructing Proto-Hellenic *dz- for all of these. --Tropylium (talk) 16:03, 13 April 2018 (UTC)

Or *ď for all of them? Surely there's no reason to write *dz initially and *ď medially. —Mahāgaja (formerly Angr) · talk 07:13, 14 April 2018 (UTC)

We would only seem to need *ď at all if given more info about this alleged *ď > δδ. For that matter, *-zd- (in *hízdō) also seems superfluous from the Greek-internal perspective. If these were all collapsed, simple *z might be sufficient (though with a note at this project page that this was unlikely to be just [z]). --Tropylium (talk) 08:50, 14 April 2018 (UTC)

Buck (Introduction to the Study of the Greek Dialects pp. 66–67) has several examples of dialectal medial -δδ- and initial δ- corresponding to Attic/Ionic/Koine ζ from *dy, *g(ʷ)y, and even initial *y: Boeotian γραμματίδδω (grammatíddō), ψαφίδδω (psaphíddō), τρέπεδδα (trépedda), δώω (dṓō), Δεύς (Deús); Cretan δικάδδω (dikáddō), ἐργάδδομαι (ergáddomai), δυγόν (dugón), Δῆνα (Dêna), Laconian γυμνάδδομαι (gumnáddomai), and so on. He doesn't give any examples of -δδ- from *-sd-, but on p. 76 he does mention that in Cretan, -ς δ- assimilates to -δ δ- across a word boundary, so it's conceivable that ἵζω (hízō) and ὄζος (ózos) (< *osdos) were *ἵδδω and *ὄδδος in these dialects, but no one seems to come out and say so directly. Sappho has ὔσδος (úsdos) for ὄζος (ózos), but Lesbian doesn't seem to be one of the dialects that has ď(ď) > δ(δ) anyway, so that isn't evidence against sd > δδ in the dialects that do. —Mahāgaja (formerly Angr) · talk 09:42, 14 April 2018 (UTC)

Source of Attic σ / Doric τ

@Rua, Erutuon: Which Proto-Hellenic consonant is the intermediate between PIE *t and Attic σ / Doric τ before ι, e.g. in τίθησι (títhēsi) (Doric τίθητι (títhēti)) or φέρουσι (phérousi) (Doric φέροντι (phéronti)? Would we reconstruct the Proto-Hellenic form as *titʰēti or *titʰētsi or *titʰēťi or what? —Mahāgaja · talk 11:54, 12 March 2019 (UTC)

It's just t. The assibilation didn't happen in all dialects. —Rua (mew) 11:57, 12 March 2019 (UTC)

@Rua: No, but it could have been undone in the dialects without it. So how do we explain why *titʰēti didn't become *σίθησι in Attic? Is it nonphonological/analogical? —Mahāgaja · talk 13:06, 12 March 2019 (UTC)

Perhaps the assibilation was conditional? —Rua (mew) 13:24, 12 March 2019 (UTC)

According to Sihler 1995, the following is known:

• Assibilation does not occur word-initially.
• It also does not occur before the suffixes -id-, -ikos, -izō.
• It remains unchanged in some Doric forms, including those you mentioned.
• *kʷi becomes *ti and never *si.

He says that it "has the appearance of a sound law whose conditions have been disturbed beyond recovery by analogical leveling". —Rua (mew) 13:31, 12 March 2019 (UTC)

Hmm, if assibilation before i never happens word-initially, it must be separate from assibilation before u, because of σύ (sú) < *túh₂. I always sort of assumed they were the same. The τίθητι/φέροντι dialects also have τύ for σύ. —Mahāgaja · talk 14:59, 12 March 2019 (UTC)

Sihler doesn't even make any mention of assibilation before u, only i. Regarding the pronoun specifically, he says *tw- regularly becomes s- in the oblique forms, and that the s- in the nominative is by analogy with those. —Rua (mew) 16:52, 12 March 2019 (UTC)

Oh, right, it's tw → s, not tu → su. Forgot about that. —Mahāgaja · talk 09:51, 13 March 2019 (UTC)

Grassmann's Law

Several entries for Proto-Hellenic are written as if Grassmann's law predates Mycenaean Greek, but there is some evidence from Mycenaean Greek that suggests that Grassmann's law post-dates Mycenaean Greek in spite of the phonetic limitations of Linear-B script. I suggest checking the journal article at the following hyperlink before the link rots. https://www.academia.edu/4423181/2013_Another_attempt_at_a_chronology_for_Grassmann_s_Law_in_GreekRigognos Molinarios (talk) 22:14, 11 April 2021 (UTC)

Bibliography

I was wondering: what are the sources for reconstruction of forms and pronunciation for Category:Proto-Hellenic language? Thank you. ‑‑Sarri.greek ^♫ | 19:23, 7 July 2021 (UTC)

@Sarri.greek, Erutuon. This is an important question. There are many works discussing Greek historical phonology, but I'm not aware of any establishing a specific phoneme inventory of Proto-Hellenic like this page does. Without a source no claim about Proto-Hellenic is verifiable. —Caoimhin ceallach (talk) 12:28, 31 January 2024 (UTC)

Kiparsky Hypothesis, Codas, and Osthoff's Law

Before a retraction took action in eastern Greek, the third person plural of secondary active tenses (ἔλυον, ἔγνωσαν) was prosodically restricted to the last two syllables only, just as in words ending in -ξ and -ψ, because they originally ended in -ντ before the coda was lost. This is reflected in comments by grammarians, ancient and modern, about Doric Greek, and in the action of Osthoff's law in Doric plurals of the same (ἐλύον, ἔγνον), as in athematic participles, which were not leveled to long vowels, even in Aeolic present. If verbs are to be accented, this needs to be in place.

The circumflex accent was real and etymological, but limited in scope and not often due to sotera law (which didn't happen in Doric) and was already present in the oblique endings in first and second declension. -óèy to ῶι óìsù to οῖσι and the like, and likely in all accented diphthongs. At some point, third declension Nominatives and Accusatives were all leveled to be not-circumflex, but this leveling ceased to continue in effect before some nouns endings were made to contract. This is shown by the acute diphthongs as in Ζεύς and βασιλεύς from *Ζέϝ̀ς and *γ͜ϝὰτὶλήϝ̀ς but Ζεῦ and βασιλεῦ from *Ζέϝ̀ and *γ͜ϝὰτὶλήϝ̀ and in older contractions such as Σαπφώ N and Σαπφώ A instead of Σαπφώ N and †Σαπφῶ A from Σαπφώ N and *Σὰπφόϳν̀ A This happened before the contraction of un-like vowels with *σ → ͱ between, as is shown by ἀληθής next to ἀληθῆ from

• ν̀λᾱ̀θής next to *ν̀λᾱ̀θέσν̀ into *ὰλᾱ̀θέͱὰ to ἀληθέὰ to ἀληθῆ

The Kiparsky Effect in verbs In Rok Kutner's "K rekonstrukciji pragrške nominale morfologie" (Ljubljana, 2020) the idea is presented, and should be discussed here, that under a strict set of rules, (and then under analogical forces in the subjunctives), after an unaccented short non-nasalized vowel, a dental plus short ι are melted and metathetize. This results in the primary actives in -εις, -ει to come from

• -εις, 2S *-ειτ 3S from
• -εσι, 2S *-ετι 3S

rather than

• -εͱϊ ς 2S *-ει 3S from
• -εσι 2S and *-ει 3S.

Some further signs of this include that the -ει ending is not a short diphthong, but long, like the optative in -οι, which is said to be long because of the old coda *-οιτ, thus it not shortening from ει to ι like in the athematic datives. Another sign is the Ionic enclitic second singular present εἰς (=Attic εἶ, Epic ἐσσι) by the same mechanism when enclitic. A possible related happening to this is in the proclitic ποι- from ποτι- (≈προς) found in Peloponnesian inscriptions. 97.67.246.85 16:16, 8 October 2024 (UTC)