Wiktionary talk:About Proto-Indo-Iranian

@Tropylium Created. All non-conforming entries have been moved to correct transcriptions. —Aryamanarora ^{(मुझसे बात करो)} 17:44, 9 April 2016 (UTC)

Transcription of PII and PIr

Can we please standardize the transcription of PII and PIr to <ǰ>, <č>, <š> as opposed to <ȷ́>, <ć>, <ś>? It's the system Skjærvø, Cheung and Rastorgueva all use and it's just silly that we don't. --Victar (talk) 18:27, 27 April 2017 (UTC)

I'm a bit confused. Would the ancestor of हिम (hima) be *ǰimas or *ĵʰimás? I agree though that our system is unconventional. —Aryamanarora ^{(मुझसे बात करो)} 19:07, 27 April 2017 (UTC)

I believe *ĵʰimás is the correct reconstruction. *ǰimas would point to Proto-Indo-European *gʰ- and not *ǵʰ-. --Victar (talk) 19:22, 27 April 2017 (UTC)

I am confused as well as what the distinction was in PIr. I'll have a look at the published works again. --Victar (talk) 19:49, 27 April 2017 (UTC)

I'd want to see a wider survey before declaring one version more standard. I have seen ‹ć ȷ́› (though not ‹ś ź›!) in use quite a bit as well, e.g. by Kobayashi, Historical Phonology of Old Indo-Aryan Consonants; Windfuhr's chapter on early dialectology in The Iranian Languages by Routledge; Skjærvø's chapter on Old Iranian in the same (interestingly even despite him still using ‹k̂ ĝ ĝ› for PIE); or Mayrhofer's chapter on the prehistory of Iranian in Compendium Linguarum Iranicarum.

Phonetically I see the distinction for PII usually phonetically analyzed by Iranicists / Indologists as *ć/*ĉ *ȷ́/ĵ = [tsʲ dzʲ] (palatalized dental), versus *č *ǰ = [tɕ dʑ] (fully palatal). For Iranian (if you mean that by PIr) pre-sibilant chain shift, older works (up to at least ca. 1950) seemed to favor ‹ś ź› ([sʲ zʲ]?) for the first series, more recent ones instead have [ts dz] (since this is the stage that Nuristani and some loanwords in Uralic have).

If there's editor consensus in favor of the circumflex notation though, I don't personally really mind either way. --Tropylium (talk) 22:30, 28 April 2017 (UTC)

@Victar, Tropylium: I'm fine with the circumflex as well. Should I begin moving the entries? —Aryamanarora ^{(मुझसे बात करो)} 13:23, 1 May 2017 (UTC)

Obviously no objections from me. =) --Victar (talk) 06:56, 2 May 2017 (UTC)

Could we add a table showing the relationship of other transcription systems to ours to that there is no confusion, please? —JohnC5 17:58, 2 May 2017 (UTC)

I would appreciate that, since I just made a mistaken edit, assuming that the circumflex and caron were the same thing. — Eru·tuon 19:55, 2 May 2017 (UTC)

@Aryamanarora: Did you do this backwards? You've been moving all the descendants of PIE *ḱ to *č instead of *ĉ. Did I miss something? —JohnC5 20:30, 2 May 2017 (UTC)

@JohnC5: Crap, I knew I messed something up! Thanks for cleaning up after me. I confused myself. —Aryamanarora ^{(मुझसे बात करो)} 23:31, 2 May 2017 (UTC)

The use of epenthetic ᵢ in IIr

@Victar, CodeCat, Aryamanarora, माधवपंडित, hello all! I'd like to start a discussion about the use of ᵢ in IIr, Ir, IAr. Victar recently proposed moving *pHᵢtā́ and *HráwdHᵢti over but I'm not sure I agree. I think that if it truly is phonemic, it should be i and if it is merely phonetic it should not be shown. I'd also like to discuss the literature around this change. I'm not sure where to look for the history of this sound change or a careful accounting of the environment. Thanks! Also thanks for reverting the changes, Victar! —JohnC5 03:11, 12 August 2017 (UTC)

Thanks, JohnC5. I'll have to pull together some sources, particularly the source that uses *Hᵢ, but essentially what I've read is PIE H becomes i-colored interconsonantally in PII. --Victar (talk) 03:50, 12 August 2017 (UTC)

@Victar: I'll also readily admit that I was misinterpreting ᵢ as an epenthetic vowel rather than a coloration diacritic on the *H. I already like this much more and realize that I was foolish not to see that earlier. I would however like to discuss it more before implementing this across the board. —JohnC5 04:00, 12 August 2017 (UTC)

You're absolutely right, I should have started a discussion before moving those two entries. Incidentally, can you move back *HráwdHᵢti? I'm glad to see we understand each other's thinking now. --Victar (talk) 04:30, 12 August 2017 (UTC)

Done. —JohnC5 04:34, 12 August 2017 (UTC)

@JohnC5: Can you move back *pHᵢtā́ as well? --Victar (talk) 16:59, 12 August 2017 (UTC)

Old east Iranian dialects is very detailed paper on the matter, but actually argues for a syllabic *H̥ in PII.

A Grammar of Gatha-Avestan

The development of laryngeals in Indo-Iranian

The development of ∂/interconsonantal laryngeal in Iranian

--Victar (talk) 17:34, 12 August 2017 (UTC)

Sanskrit पितृ (pitṛ) & रोदिति (roditi) reflect this coloring. Also explains why these consonants aren't aspirated... ɱɑɗɦɑѵ (talk) 07:07, 12 August 2017 (UTC)

At first I was against it but now I see why it's useful. I'd like to see which source uses the _i though. BTW isn't *H between a consonant and consonant cluster realized as *i? E.g. *pitray, dative singular of *pHtā́. —Aryaman ^{(मुझसे बात करो)} 12:29, 12 August 2017 (UTC)

To play devil's advocate against myself, the alternative is to say laryngeals only aspirate the presiding consonant before a vowel and the insertion of -i- is a natural byproduct of IIR consonant cluster breaking. If that's the case, what are the rules of consonant cluster breaking and do we have examples of i-insertion without the aid of laryngeals? --Victar (talk) 16:05, 12 August 2017 (UTC)

To look at roots like *tep-, the normal result of cluster breaking in IIr is *a. —JohnC5 17:29, 12 August 2017 (UTC)

PII *bʰadʰHras -> Skt. बधिर (badhira) would be an additional example. But there's no interconsonantal laryngeal in PII *Hrudʰrás and yet we have Skt. रुधिर (rudhira). -- ɱɑɗɦɑѵ (talk) 06:41, 13 August 2017 (UTC)

The -i- or the aspiration of -d- in बधिर (badhira) could be influenced by another inflection of the word. If you have a look at the example in Old east Iranian dialects (actually Kuiper 1942), daughter is reconstructed as *dhugH̥₂tḗ (nom.), *dhugH̥₂térṃ (acc.), *dhugH₂tréi (dat.). --Victar (talk) 08:01, 13 August 2017 (UTC)

From the Indo-Iranian perspective, we are quite clearly dealing with a vowel and not a consonant. A decent option would be *ə or *ɨ (since this still cannot be identified with *i). A "syllabic laryngeal" *H̥ or *H_i is IMO overly abstract symbol-algebra.

As a sidenote, I also find the practice of retaining all laryngeals as *H into PII and even PIr. superfluous. AIUI it's done because some people argue that Eastern Iranian *iH > *i and *uH > *u never went through long vowels; but this does not mean that laryngeals would have to have survived in all positions. To reconstruct a PIr. *H as a distinct phoneme alongside real *h from PII *s (and also *x from PII *kC) is kind of absurd anyway. (Though I would like to read Mayrhofer's book on this before I propose adjustments.) --Tropylium (talk) 13:56, 13 August 2017 (UTC)

I think we're mostly on the same page. *H was certainly lost in some positions (i.e. Lubotsky's law) and it's no doubt more complicated than our current understanding. We don't truly know what form laryngeals look in PII, be it *h or a dorsal fricative of some kind. Nonetheless, I think we can agree that it's pretty self-evident that we're dealing with two forms in PII, a consonantal, and a vocalic. By what mechanism some laryngeals became vocalic, during what period this happened, and how/should we demarcate this distinction is the challenge at hand. --Victar (talk) 14:31, 13 August 2017 (UTC)

Kuiper (p. 148) actually claims that "*H̥ (…) became *i before the split" (of PII into separate branches), i.e. he seems to be using "Proto-Indo-Iranian" for the entire period from PIE to PII proper. On the other hand, he proposes a separate development *H > *i to have reoccurred in Indo-Aryan for some cases where Iranian has zero. This is probably not the current understanding. --Tropylium (talk) 21:15, 13 August 2017 (UTC)

@Tropylium: I don't we can say *H was fully realized into *i yet in PII, otherwise we would find -i- in Avestan 𐬞𐬙𐬁‏ (ptā‏). Yeah, it's a 41 year old rehashing of a 71 year old paper, so needs to be taken in that light. --Victar (talk) 22:49, 13 August 2017 (UTC)

We on the other hand regardless find -i- in Young Avestan (we do not seem to currently distinguish the Av. varieties, but we should). He proposes something like an alternation *ph₂tēr : *pH₂tre- > *pHtār : *pH̥tra- > *pʰtār : *pitra-, with the nom. sg. later levelled to *pitār outside of Avestan. --Tropylium (talk) 23:52, 13 August 2017 (UTC)

(We use dialect labels for the Avestan varieties, BTW, much like Vedic/Classical Sanskrit) —Aryaman ^{(मुझसे बात करो)} 14:21, 15 August 2017 (UTC)

Does anyone have access to this paper, Evidence of laryngeal coloring in Proto-Indo-Iranian? --Victar (talk) 20:30, 13 August 2017 (UTC)

@Tropylium also if you have a copy of Mayrhofer's Laryngalreflexe im Indo-Iranischen, I've yet to read it firsthand. --Victar (talk) 21:05, 13 August 2017 (UTC)

I've looked up the Ollett paper, but I'm not sure what it would contribute to this discussion: it just verifies that *Keh₂- gives *Kā- and not *Čā- (e.g. *kās- (“cough”) from *kʷeh₂s-), therefore a-coloring did happen in II as well.

There's a much more interesting paper in the 2016 volume, though: i-Epenthese im Altindoarischen: Laryngalvokalisierung oder „Verbindungsglied“?', from Aufderheide & Keydana. I'll read more into that one; the theory might not be reliable enough for us to work with, but we'll see.

As for Mayrhofer's book, it has a pdf version floating online if you look around, although it's kind of a tedious book that mainly lists various categories of Sanskrit/Avestan/Old Persian words for which a preform with a laryngeal has been supposed. There's no real attempt to reconstruct the common Proto-Indo-Iranian stage. --Tropylium (talk) 13:45, 15 August 2017 (UTC)

{{see desc}} in PII descendants trees

@AryamanA, JohnC5, Calak, माधवपंडित I'm wondering if we should use {{see desc}} in PII descendants trees beside PIA/Sanskrit and PIR entries by default. *dr̥Hgʰás is looking pretty busy. --Victar (talk) 15:38, 21 December 2017 (UTC)

I would do that in PII entries which have a lot of descendants from one branch (IA or Iranian) and only one descendant from the other and big entries like *dr̥Hgʰás. Normal sized PII entries can continue listing all descendants, IMO. -- माधवपंडित (talk) 16:17, 21 December 2017 (UTC)

@माधवपंडित: Well, that's the thing, if a Sanskrit entry is properly filled out, it's virtually always going to be long, so it's an eventually. In Balto-Slavic, we add {{see desc}} beside the Slavic entries, even if the Slavic entry only has 4-5 descendants listed. --Victar (talk) 17:06, 21 December 2017 (UTC)

I read the other day that 2/3 thirds of all the languages in IE were in the Indo-Iranian branch, so using {{see desc}} would seem prudent. —*i̯óh₁nC^[5] 00:16, 22 December 2017 (UTC)

@Victar: I don't see any need yet, but you're that if we begin to have descendants for the hundreds of Indo-Aryan languages it will be necessary. The descendants at *dr̥Hgʰás are actually wrong, I'll be cleaning them up. Most of them are borrowed. —Aryaman^{A (मुझसे बात करें • योगदान)} 17:16, 21 December 2017 (UTC)

@माधवपंडित: You don't see *dr̥Hgʰás as being busy? I certainly do; assuredly, at least enough to selectively use {{see desc}}. Compare PBSL *bardā́ˀ. The question isn't if we should be using {{see desc}}, it's should we be using it by default.--Victar (talk) 18:14, 21 December 2017 (UTC)

@Victar:, No, like I said, I'm fine with using the new system esp. as it may improve our coverage of PIA and PIr. It's especially annoying when long lists of Sanskrit borrowings are given at the PII entry, cluttering up the whole thing. But at the same time, in case of small entries with just sanskrit and avestan descendants (or maybe just a few more), creating additional PIA and PIr reconstructions and putting {{see desc}} will be redundant. I have seen such activities here (like Proto-Italic entries or Proto Hellenic entries being made for just one Latin/Greek descendant) but I feel that's wasteful and Indo-Iranian could do without this. Again, only for entries with <10 descendants or so. -- माधवपंडित (talk) 01:21, 22 December 2017 (UTC)

Butting in to say that I agree with माधवपंडित: creating proto-entries for the sake of creating proto-entries is wasteful; Proto-Italic and Proto-Hellenic are indeed the worst offenders in that respect. I think his idea of having a lower limit of descendants might make sense.

Also, I'd like to make a point about usability: our PIE entries are much more fragmented now compared to when I got here (2011/2012), when they offered a more synthetic view of the descendants, and were thus more enjoyable for the layman: you could find vaguely related Latin, Lithuanian, Sanskrit, Tocharian words, or words of "one's own language" ("look, my French word long is related to your Assamese word দীঘল (dighol), that's so cool!!") on a same page. I'm not complaining: this evolution is natural and welcome, since what we've lost in "enjoyability", we've made up for ten times in accuracy and thoroughness. But let's not add unnecessary steps and clicks when the number of descendants is small.

That said, I think using {{see desc}} on *dr̥Hgʰás would be totally fine. --Per utramque cavernam (talk) 01:55, 22 December 2017 (UTC)

*ā or *ah in PII

Which one should be used in reconstruction? I see both in academic literature. —Aryamanarora ^{(मुझसे बात करो)} 17:02, 23 April 2017 (UTC)

Both existed, as it depends on the PIE root, i.e. *ō > ā and *oH > *aH. But yeah, a lot of older paper reconstruct *aH as *ā. --Victar (talk) 22:38, 23 April 2017 (UTC)

I would say the distinction should be maintained between *aH and *ā when known. —JohnC5 01:56, 24 April 2017 (UTC)

Yep, absolutely. --Victar (talk) 02:00, 24 April 2017 (UTC)

How would the difference be reconstructed, if not for PIE? What evidence is there within PII to reconstruct the difference? —CodeCat 17:09, 26 April 2017 (UTC)

This discussion probably warrants mirroring at WT:AIIR, but the arguments I have seen are all indirect, based on the maintenance of laryngeals in various other environments:

1. *iH and *uH give plain /i u/ in parts of northeastern Iranian, and may never have developed into long vowels;
2. *N̥H gives *ā, so *aH > *ā is at least later than the split of II from other IE languages.
3. *aHa is prosodically distinguished from *ā in Avestan (and Vedic?)
4. *Ha gives /ha/ or /xa/ in parts of southwestern Iranian (at least when *H = *h₂), though nowhere with full regularity.
5. *Ha gives *ka in a few loans into Uralic (cf. *karhu).

As far as I know, inner-II evidence regardless provides no evidence for distinguishing *aH and *ā. It's also probably reasonable to expect that *iH and *uH vocalized later than *eH, *aH, *oH: the latter three give long vowels everywhere in IE, while the former can lead to breaking to *yā, *wā etc. or "shortening" to *i, *u. --Tropylium (talk) 07:15, 10 January 2018 (UTC)

Thanks, @Tropylium! Just to mention a few other environments laryngeals most likely survived in:

1. *R̥H, which is similar to *N̥H,
2. word-initially, PII and PIr compounds demonstrate that a) the thematic vowel of first element is lengthened by *#HC- in the second element, and b) a hiatus in Avestan is seen as a result of *#HV-,
3. and per Lipp (2009), PII likely underwent *H > *H̥ / (-)C_C´-, -C_(C), -C_sR-, which was probably lost in PIr in all positions except -C_(C).

One argument of note for the retention of *aH in PII (and PIA), per Lubotsky (1992), is that if the root contains a non-initial *H, including *aHC, in i- and u-stems, the accent is shifted to the stem. We don't see this same shift in i- and u-steams derived from āC. --Victar (talk) 23:04, 10 January 2018 (UTC)

Worth mentioning: much of this kind of stretches the definition of "laryngeal" and "surviving". They mainly show that laryngeals did not fully merge with the traditional non-laryngeal equivalent, but they very well could have already phonetically vocalized. *aH could have been [a̰ː] (a glottalized vowel) versus *ā being [aː] (a clear vowel) and *aHa being [a.a] (with merely a syllable break); *iH could have been [iə]; *H̥ could have been [ə]. Which is not a problem though as long as we acknowledge that "*H" is more of a graphical device than any one particular sound. (I will refrain from rehashing my argument from above that at least the notation "*H̥" seems misleading.)

What IS a problem is that when we don't have an IE etymology, we still have no means to tell if we should reconstruct *aH or *ā. Take for example, Sanskrit धारा (dhārā, “tip, edge”) (with also Iranian cognates). This is fairly old since some early version of this seems to get borrowed into Uralic as *terä (“blade”). Pre-PII *dʰor-eh₂ is out of the question at least, but we still have no idea if it comes from earlier *dʰēr- or *dʰeh₁r- (or, in theory, even *dʰērH-). So should the PII entry be *dʰāraH or *dʰaHraH (or even *dʰārā)? --Tropylium (talk) 00:51, 11 January 2018 (UTC)

See pp. 1881ff., and pp. 1883f. on intervocalic laryngeals specifically. However, see also Kümmel 2016: Not only the aspiration of voiceless stops, but also data from Middle and Modern Iranian do point to a Proto-Indo-Iranian phoneme /h ~ x/ rather than /ʔ/. The development of the laryngeals in Indo-Iranian is not completely understood: it does seem like they merged into a single phoneme on the way to Proto-Indo-Iranian, which was initially preserved and at least partially into more recent stages, even far more recent stages apparently, but the details remain controversial. --Florian Blaschke (talk) 21:33, 9 December 2019 (UTC)

Thorn clusters

What's up with weird reconstructions like *Hŕ̥ćšas? See pp. 10–12 on why these are really implausible. --Florian Blaschke (talk) 21:16, 9 December 2019 (UTC)

@Florian Blaschke: It's complicated. 😂 --{{victar|talk}} 22:07, 9 December 2019 (UTC)

According to you. According to Lipp and others, there was never an actual metathesis in Indo-Iranian, thorn never even existed as an allophone, and *TK > *Kθ > *Ks is a chimera. --Florian Blaschke (talk) 22:09, 9 December 2019 (UTC)

Thorn cluster metathesis will always be a point of debate. --{{victar|talk}} 22:13, 9 December 2019 (UTC)

"Will always"? Even if your crystal ball was right, that doesn't mean you can simply postulate whatever you like and expect to be taken seriously.

Even Lubotsky on p. 1885 disagrees with you. --Florian Blaschke (talk) 22:17, 9 December 2019 (UTC)

I say "will always" because either is plausible. If you read though the sources on the page I directed you to, {{R:inc:Kobayashi:2004|70}}, for one, supports thorn cluster metathesis, and the developmental steps are taken from that book. --{{victar|talk}} 22:29, 9 December 2019 (UTC)