User talk:Victar/Reflexes/Proto-Indo-Iranian

Discussion
This is amazing... —Aryamanarora (मुझसे बात करो) 20:51, 1 April 2017 (UTC)
 * , oh, thanks. It still has a long ways to go. I'm just compiling data from various sources, with the goal of creating a guide for the proto form reconstructions. I'll add it to the about pages when I'm done. --Victar (talk) 03:49, 2 April 2017 (UTC)
 * I really like this and hope you don't mind my adding things. A question though: is there a reason you are adding accents to the Iranian descendants? As far as I am aware, we have no idea about potential accent placement in PI. — JohnC5 19:59, 27 April 2017 (UTC)
 * Yeah, sure thing. It's really just me going through Skjærvø's papers and putting it to a table. From what I've read, we know PIr and it's descendants had accents, so yes, I think that they should be included. I haven't created a row for it yet, but an accent before r in Avestan affects its voicing. Another note, I've been adding all the entries in sg. nom. even though many of the Avestan entries on Wiktionary are in the voc.--Victar (talk) 20:35, 27 April 2017 (UTC)
 * To be clear, Avestan and Sanskrit aren't using the vocative, they are using the "lemma-form" which is meant to show the declension (for instance in Snaskrit: "-a" is a-stem, "-as" is s-stem, "-ā" is feminine ā-stem, "-at" is nt-stem). The convention is probably borrowed from Sanskrit nouns into Avestan, it's actually something we should discuss for PII and PIr whether the entries should be under the nom.sg. or under the Sanskritic "lemma-form". We've been doing both until recently when Aryamanarora started converting them to the nom.sg. — JohnC5 20:51, 27 April 2017 (UTC)
 * Thanks for the clarification. I've never seen any papers reconstruct PII or PIr in noun-stem form (that is, without a trailing dash), so if we did, it would be unique to Wikipedia. I had already started a discussion on it not to long ago, but I can't recall where that is. --Victar (talk) 21:23, 27 April 2017 (UTC)
 * I'd love to be involved in that discussion if you can find it. I'm fine with using the nom.sg. for PII and PIr. I always found the Sanskrit system helpful and annoying at the same time. The rule seems to be to take the zero-grade output without desinence as the stem, then use that as the lemma. I believe these stems are also the "combining form" of the noun in many cases. That form is used both in Avestan and Sanskrit dictionaries pretty consistently. I don't think we should stray from that for either language, but I appreciate your use of the nom.sg. in these tables. — JohnC5 22:01, 27 April 2017 (UTC)
 * Thanks for all the edits, . Coming together. --Victar (talk) 04:26, 29 April 2017 (UTC)
 * Thanks to you too! I'm having a lot of fun. — JohnC5 05:50, 29 April 2017 (UTC)

I worry that our patterning for *sd- and *sdʰ- is being affected by the ruki rule and would otherwise not have retroflection. Is there a common IIR prefix ending in *-s- that I could try to find for examples besides ? — JohnC5 22:23, 29 April 2017 (UTC)
 * , I'll look into it. --Victar (talk) 03:15, 30 April 2017 (UTC)
 * I've found evidence to support my suspicion. — JohnC5 03:32, 30 April 2017 (UTC)
 * Great, thanks! --Victar (talk) 09:29, 30 April 2017 (UTC)

, I don't think we should detail Younger Avestan. They can have a look at the Avestan page for that. --Victar (talk) 03:15, 30 April 2017 (UTC)
 * Yeah, I only added that because Avestan was mislabeled. It would appead that our example is younger Avestan. — JohnC5 03:36, 30 April 2017 (UTC)
 * Yeah, my mistake on that. --Victar (talk) 09:29, 30 April 2017 (UTC)

I wonder if there's a better way to show ¯dh. Perhaps ːdh or V̄dh? What do you think? What do they use in other articles? --Victar (talk) 01:12, 1 May 2017 (UTC)
 * It's an interesting question since the compensatory lengthening is not completely simple. While i > ī and u > ū, az > ai > ē and až > au > ō? That's weird. If you'd like to take stroll through this article, it might elucidate things. I don't have time to read it at the moment because I just noticed the vowel table has a bunch of errors that I need to fix. — JohnC5 02:54, 1 May 2017 (UTC)
 * Just when you thought you were close to being finished. Sorry. — JohnC5 03:48, 1 May 2017 (UTC)
 * Haha, naw, I knew the vowel table was far from complete. =) Are we sure that VHV > VV happened before pre-PII? That seems like someone is going off the outdated theory of laryngeals being lost in PII. --Victar (talk) 04:01, 1 May 2017 (UTC)
 * Oooops! I didn't even notice. Yeah, please fix that to your liking. Also, the account of Avestan diphthongs is not reeeeally accurate. I'll update that soon. — JohnC5 04:10, 1 May 2017 (UTC)
 * Both *eHi, *ēy, *oHi, and *ōy word finally become *ā(y), so *eHi, *ēy, *oHi, and *ōy will have to be subdivided into 8 parts to be fully accurate. — JohnC5 04:40, 1 May 2017 (UTC)
 * So, my main issue with your proposed layout is that it makes Avestan's closed diphthong distinctions look like they are connected to one particular PIE diphthong. In truth, PIE *ey and *oy are indistinguishable after PII, and this holds for several other diphthong sets. — JohnC5 06:40, 2 May 2017 (UTC)
 * I don't think so. *ey and *oy were distinct in pre/early-PII, and as such, should have their own respective cells in the table. The problem with the previous format was it looked as though *ey and *oy were interchangeable in pre-PII, which they were not. --Victar (talk) 07:07, 2 May 2017 (UTC)
 * True, but they were entirely interchangeable from PII forward, and this layout makes it look like the distinctions further to the right in the table correspond to distinctions on the left, which certainly is not the case. — JohnC5 07:11, 2 May 2017 (UTC)
 * Perhaps, but I think it better to have an accurate table than pandering to the off chance that people may mistakenly misread it. Or we can go back to the original format, though I felt it a bit overkill. --Victar (talk) 07:27, 2 May 2017 (UTC)
 * FYI, we also have such split-merger-splits in the other two tables. --Victar (talk) 07:34, 2 May 2017 (UTC)
 * Yeah, we can use your solution. — JohnC5 15:40, 2 May 2017 (UTC)

Where did you find ? — JohnC5 17:48, 2 May 2017 (UTC)
 * -ə̄ is actually the OAv. sg.nom. ending and -ō is the YAv. --Victar (talk) 18:00, 2 May 2017 (UTC)
 * I get that but you are editing the wrong text. You need to be editing the alt text, but you are editing the link. — JohnC5 18:01, 2 May 2017 (UTC)
 * I looks as though you're mistaken. Please don't revert. --Victar (talk) 18:04, 2 May 2017 (UTC)
 * I'm sorry if this is frustrating to you, but you've broken all of the links. Take for example your text: . It should link to, but it currently links to and displays , both of which are incorrect. This is because you got the link and alt text backwards. — JohnC5 18:10, 2 May 2017 (UTC)
 * I'm not frustrated by the matter, but I am annoyed that you would revert my edits without discussing them first. I understand what you're saying, but if you compare the, you'll see that I only updated 3 and tr, not 2. What am I missing here? --Victar (talk) 18:23, 2 May 2017 (UTC)
 * Because Avestan is right to left, it does this annoying thing where 2 and 3 are flipped in the edit window. You also can't tell because we've overridden all of the automatic transliteration. I can fix this later, but it's gonna be annoying for precisely the same reason. — JohnC5 18:28, 2 May 2017 (UTC)
 * Gotcha. Will fix myself now. --Victar (talk) 18:32, 2 May 2017 (UTC)
 * Again sorry. I hate how r2l languages are handled. — JohnC5 18:46, 2 May 2017 (UTC)
 * , NP, thanks for bringing it to my attention. Yeah, super janky. Another reason to convert 3 to a/alt. --Victar (talk) 19:32, 2 May 2017 (UTC)
 * Also, what are you using to tell whether a word is Young or Old at the moment? — JohnC5 19:01, 2 May 2017 (UTC)
 * Either from what the source says or from its spelling, ex. β vs. b. --Victar (talk) 19:32, 2 May 2017 (UTC)
 * Fair enough. I've also come to realize that Kanga errs on the side of Younger Avestan. I wish there were a slightly more up-to-date dictionary. — JohnC5 20:07, 2 May 2017 (UTC)
 * Are you aware of this one? It's very abridged but useful. --Victar (talk) 23:16, 2 May 2017 (UTC)
 * No, thank you for this! — JohnC5 04:17, 3 May 2017 (UTC)

Resources
--Victar (talk) 19:56, 3 May 2017 (UTC)
 * These are my IIr resources, most of which you can find online, and others I've OCR'd myself. Let me know if you need access to any, or if I'm missing any myself.
 * Indo.Iranian-Cheung-Etymological.Dictionary.of.the.Iranian Verb
 * Indo.Iranian-Doyama-Indo.Iranian.mans.dha
 * Indo.Iranian-Gharib.Sogdian.Dictionary
 * Indo.Iranian-Hock-On.the.Indo.Iranian.Accusative.Plural.of.Consonant.Stems
 * Indo.Iranian-Kanga-English.Avesta.Dictionary
 * Indo.Iranian-Kent-Old.Persian.Grammar.Texts.Lexicon
 * Indo.Iranian-Kim-Two.problems.of.Ossetic.nominal.morphology
 * Indo.Iranian-Kobayashi-Historical.Phonology.of.Old.Indo.Aryan.Consonants
 * Indo.Iranian-Lubotsky-The.Indo.Iranian.substratum
 * Indo.Iranian-Kümmel-The.development.of.laryngeals.in.Indo.Iranian
 * Indo.Iranian-Martinez.de.Vaan-Introduction.to.Avestan
 * Indo.Iranian-Mayrhofer-Etymologisches.Wörterbuch.des.Altindoarischen-v01
 * Indo.Iranian-Mayrhofer-Etymologisches.Wörterbuch.des.Altindoarischen-v02
 * Indo.Iranian-Monier.Williams-A.Sanskrit.English.Dictionary
 * Indo.Iranian-Phonology.of.Iranian
 * Indo.Iranian-Rastorgueva.Edelman-Etymological.Dictionary.of.the.Iranian.Languages
 * Indo.Iranian-Rau-The.Origin.of.Indic.and.Iranian.Feminines.in.ani.D
 * Indo.Iranian-Sandell-Compensatory.Lengthening.in.Vedic.and.the.Outcomes.of.Proto.Indo.Iranian.az.and.až
 * Indo.Iranian-Sastri-A.Grammatical.Dictionary.of.Sanskrit.Vedic
 * Indo.Iranian-Skjærvø-An.Introduction.to.Old.Persian
 * Indo.Iranian-Skjærvø-Avestan.and.Old.Persian.Morphology
 * Indo.Iranian-Skjærvø-Old.Avestan.Glossary
 * Indo.Iranian-Thackston-Sorani.Kurdish
 * A great resource for original Old Iranian texts is TITUS texts. —Aryamanarora (मुझसे बात करो) 22:16, 3 May 2017 (UTC)
 * Thanks for this! I was able to find or had the majority of these except Rastorgueva & Edelman and Rau. I also have:
 * (Allen 1953) Phonetics in Ancient India
 * (Beekes 1997) A Grammar of Gatha-Avestan
 * (Durkin-Meisterernst 2004) Dictionary of Manichaean Middle Persian and Parthian
 * (Hoffmann, Forssman 1996) Avestiche Laut- und Flexionslehre
 * (Kuz’mina 2007) The Origin of the Indo-Iranians
 * (MacKenzie 1971) A Concise Pahlavi Dictionary
 * (West 2011) Old Avestan Syntax and Stylistics
 * (Zwicky 1965) Topics in Sanskrit Phonology
 * — JohnC5 05:00, 4 May 2017 (UTC)
 * Cool. Yeah, Rastorgueva is really outdated for a modern publication, but if you just apply 50 years of progress to his reconstructions, they work. =P Do any of the ones you mention have etymologies or cognates listed? --Victar (talk) 14:08, 4 May 2017 (UTC)
 * Beekes and Hoffmann & Forssman both give historical/etymological accounts of Avestan phonology and morphology with examples from Sanskrit. Zwicky gives a pretty in depth, if outdated, description of Sanskrit historical phonology. The remaining texts are mainly just good for finding lemmata and citations in a given language. — JohnC5 14:41, 4 May 2017 (UTC)
 * I started downloading some of your resources, and I have to say, they're pretty cool. I got some others, mostly for Indo-Aryan stuff:
 * (Edgerton 1953) Buddhist Hybrid Sanskrit Grammar and Dictionary
 * (Pali Text Society 1921-25) Pali-English Dictionary
 * (Turner 1962-85) A Comparative Dictionary of the Indo-Aryan Languages
 * (Gray 1902) Indo-Iranian Phonology [note: really outdated but pretty comprehensive, and in English!]
 * (Ara 2008) Eschatology in the Indo-Iranian Traditions [note: mostly religious practices and history, not much lexicography; interesting nevertheless]
 * A Dictionary of Gandhari [note: one of the few comprehensive Prakrit resources available online]
 * —Aryamanarora (मुझसे बात करो) 20:00, 4 May 2017 (UTC)
 * I recommend checking out Compendium Linguarum Iranicarum (1989). --Tropylium (talk) 10:38, 31 August 2017 (UTC)

RUKI rule
Does the apply to the vowel *i and *u as well? Not just the semivocalic *Vy and *Vw? —Aryamanarora (मुझसे बात करो) 15:34, 7 May 2017 (UTC)
 * I think so, yeah. — JohnC5 18:46, 7 May 2017 (UTC)

Should we actually do this?
, is it time to actually make WT:AIRA? — JohnC5 15:30, 22 June 2017 (UTC)
 * , sure why not. A work-in-progress is better than no progress. --Victar (talk) 15:40, 22 June 2017 (UTC)
 * As the person who made this, did you want to do it? :P — JohnC5 04:50, 23 June 2017 (UTC)
 * : I think this should probably be moved to WT:AIIR instead. What's the best way to merge entries? --Victar (talk) 18:00, 8 August 2017 (UTC)
 * I'll admit that I've been dragging my feet on this one. — JohnC5 04:52, 10 August 2017 (UTC)

*r/*n Sanskrit forms
What's your source for the Sanskrit case endings? An example like (See entries at ahan and ahar on this page). Both IAr. and Ira. expressed r ~ n alternation. — JohnC5 20:59, 30 August 2017 (UTC)
 * It's a *r-stem table, not an *r/n-stem. --Victar (talk) 21:31, 30 August 2017 (UTC)
 * Yes, I see that you changed that. That clears up my question. Thanks! — JohnC5 21:32, 30 August 2017 (UTC)
 * I didn't change it -- I was still in the middle of working on it. --Victar (talk) 21:40, 30 August 2017 (UTC)

I notice you have long accusative singulars in IIr. (*-ā́ram) but short in IAr. (*-áram). This believe the issue here is that PIE and Sanskrit had long *-ēr ( → ) and *-ōr ( → ). Unfortunately this means that your r-stems actually have (at least) 2 flavors for which we must account. — JohnC5 04:47, 31 August 2017 (UTC)
 * I think you're referring to tar-stems. --Victar (talk) 18:21, 31 August 2017 (UTC)
 * The division is not so clean cut. In Sanskrit for instance, it mostly splits between agent and kinship nouns, but not completely ( has the -ōr declension as expected, though it appears to have changed by analogy in Iranian if Skjærvø is correct about the inflection). I see that Skjærvø and to a lesser extent Martínez & De Vaan make that distinction, but it is not systematic by any means. Regardless, in both cases they are referred to as r-stems by traditional grammars. It is not sufficient to break them -ar versus -tar stems. — JohnC5 18:44, 31 August 2017 (UTC)
 * When is anything ever clean? Just another irregularity. --Victar (talk) 20:25, 31 August 2017 (UTC)
 * Yeah, sorry about that. I was just very confused when I saw IIr. (*-ā́ram) next to IAr. (*-áram), but then I figure out why. — JohnC5 21:09, 31 August 2017 (UTC)

PIA *š
Was this not retroflexed to *ṣ yet? Great job with the PIA btw. —Aryaman (मुझसे बात करो) 16:22, 25 September 2017 (UTC)
 * Already changed, and thanks! --Victar (talk) 01:07, 26 September 2017 (UTC)

Some Iranian comments
If the plan is to on make this into an official reference sheet, I have some possible corrections: --Tropylium (talk) 16:29, 26 September 2017 (UTC)
 * As far as I know, *ĵ > d is the default development in Old Persian (just like *ĉ > θ), *ĵ > z only occurs in the cluster *ĵw and in loanwords from other Iranian languages. Where do you get the claim from that *ĵ > d would be particular to positions before front vowels?
 * *kH > *kʰ gives PIr. *x (or *kʰ, see below), not /k/.
 * You have the output of *ḱt and *ǵʰt the wrong way around for Av. / OP: it is the former that has /st zd/, the latter /št žd/.
 * Your reconstruction of a PIr. *ŝ for PIE *sḱ is interesting. Is this from any particular source?
 * *l > *r is not yet PIr.; some languages like Ossetian still sometimes retain /l/ (the usual example seems to be ).
 * There are a few other changes that might be, per some authors, non-PIr. despite being found in both Av. and OP. One is the spirantization of initial *pʰ *tʰ *kʰ (from *PH and in clusters like *Pr) to *f *θ *x, which is not evidenced in Balochi, Parachi, or Sakan (Wakhi, Khotanese, Tumshuqese).
 * Thanks, . I fixed the obvious errors. I'll have to look into the others. The development of *sḱ is from Kobayashi (2014). There are some alternatives hypotheses that I'll try and include in the table as well. --Victar (talk) 19:39, 26 September 2017 (UTC)

PIE *sw
Hi. PIE. *sw > PIr. *hw- > OP. ʰuv, Av. xᵛ (foe example PIE. *swé > PIr. *hwá > OP. ʰuva, Av. xᵛa). Please correct PIr. and OP. forms. Thanks.--Calak (talk) 08:00, 6 October 2017 (UTC)

tt > tˢt law name
, is there a law name for the insertion of an epenthetic sibilants between dentals, i.e. tt > tˢt / dd > dᶻd? --Victar (talk) 05:28, 17 November 2017 (UTC)
 * Not really. It's normally just considered the phonetic outcome of *-DD- in PIE. I looked around and can't find any such rule. — JohnC5 05:56, 17 November 2017 (UTC)
 * I've seen "double dental rule" used. --Tropylium (talk) 08:48, 17 November 2017 (UTC)
 * Oh yeah, I've definitely seen that as well. — JohnC5 09:34, 17 November 2017 (UTC)
 * Cool, thanks both. --Victar (talk) 17:11, 17 November 2017 (UTC)

Bartholomae's law
Hey,, I have another law question. Do dt > tt and ds > ts also fall under Bartholomae's law, or is that a different phenomenon? Thanks. --Victar (talk) 05:30, 18 November 2017 (UTC)
 * No, they fall under normal voicing assimilation. The general principle is *-DT- in PIE becomes *-TT-. — JohnC5 08:13, 18 November 2017 (UTC)
 * Thanks again, . So a follow up, is there a reason we don't reconstruct PIE lemma on the project with this and the rule above, if they applied to PIE? Case in point, > *wéyttōr > . --Victar (talk) 21:33, 18 November 2017 (UTC)
 * This question has been asked before. So this is a phonetic change, but it is effectively unconditional, so all instances of /*-tt-/ become [*-tst-]. We stand at an awkward intersection between phonemic and phonetic representation. We don't, say, show all laryngeal vowel coloring, Siever's/Lindeman's, or the Saussure Effect, but we do show the effects of Stang's, Szemerényi's, Pinault's, etc. in PIE. You might ask about this, as she was instrumental in the formulation of this. To my knowledge, however, it is uncommon to show the double dental rule in PIE reconstructions because it makes the words more etymologically opaque (but so do several other rules, so I don't know). We very well could write . The superscript ˢ irks me a bit because, phonetically, there was a full *s. But then again, we also write  and not . I dunno; we do some things but not others. — JohnC5 21:58, 18 November 2017 (UTC)
 * Thanks, . I figured it was something exactly along those lines. --Victar (talk) 23:45, 18 November 2017 (UTC)

Inflection tables
, if you don't mind. I'm going to work on those myself. Could you move your editions to your own sandbox? Thanks. --Victar (talk) 08:21, 24 November 2017 (UTC)
 * I apologize. I had grown a little frustrated with the declensions being incorrectly (sorry) conflated. As mentioned before, the Iranian accusatives in -āram belong to the *-or declension and those of -aram to *-er (of which is an example). Skjærvø's declensions demonstrate this. You may of course move them somewhere else, but I don't really think that reverting it back to the old table would be advisable, since it was wrong. I did not mean to step on your toes. — JohnC5 08:33, 24 November 2017 (UTC)
 * No worries. I (and Kümmel, from what I understand) disagree on your thoughts on r-stems, but yep, I'll be moving my ongoing work on it to my other user page. --Victar (talk) 08:40, 24 November 2017 (UTC)
 * Which is the Kümmel material in question? The main evidence here is the fact that the Avestan and Old Persian examples in Skjærvø table 46 perfectly align with this distribution and agree with Sanskrit. I'm not really sure how there can be disagreement without any counterexamples? — JohnC5 08:46, 24 November 2017 (UTC)
 * I meant to ping you. — JohnC5 08:49, 24 November 2017 (UTC)
 * Kümmel (2017) Einfuhrung ins Altiranische --Victar (talk) 08:56, 24 November 2017 (UTC)
 * Thanks for this! Very cool. Also §4.2.2.8 agrees exactly with what a said. I don't mean to be a pest, but this really isn't crazy what I'm saying, it is very much the orthodoxy. — JohnC5 09:02, 24 November 2017 (UTC)
 * Actually, I hate to say this, but many of the tables in Kümmel §4.2.2.X represent multiple related declensions (the *u-stems have at least 3 separate declensions in IIr!). Indo-Iranian is very good at preserving the variety of inflection types of Indo-European and provides much of the evidence we use to reconstruct. If you're going to create a IIr. declension module, it will have to show that diversity. I'm not trying to stomp on your work by any means—I really appreciate that you work in this language family—, but the inflections are often more complex than your current work shows. I'm just trying to make sure all of the detail gets captured. — JohnC5 10:25, 24 November 2017 (UTC)
 * John, I'm sharing this document with you, not the other way around and I'm not daft, I know how to read a table. Yes, I very much understand that there is variety. I'm not disagreeing with that. But for one, you can see in 4.2.2.8 r-Stämme that *-r̥ is one of those variants, so that table shouldn't have been deleted, like your did. Secondly, I'm disagreeing there being clean-cut forms true to the etymology of each word, as opposed to free variation. Will there be lemma that use case endings true to their etymology? Sure, but we also see varied use to case endings within a single lemma. I think you're trying to put forth the idea of some perfect pristine form of PIr that simply did not exist, at least not to any level you're suggesting. --Victar (talk) 15:18, 24 November 2017 (UTC)
 * Also, I feel again that I need to stress that these tables on are still on my user pages, meaning they are works-in-progress, and are still missing many forms, ex. n-stems and s-stems. To judge them as complete works would be premature. --Victar (talk) 15:42, 24 November 2017 (UTC)
 * Sorry for the delay; I was traveling. To answer your question: yes and no. You're certainly correct that no inflectional paradigms are perfect, that syncretism and analogy occur all the time, and that frequently anomalous forms from other declensions pop up. That is par for the course, but the overwhelming evidence from the earliest attested terms in the earliest IIr languages cluster into predictable declension that are related and match other IE branches. To argue that we shouldn't followed the reconstructable trends of declension that are widely agreed to exist just because the evidence may contain some anomalies would certainly be throwing the baby out with the bath water. I appreciate that this is a work in progress, but I've always admired the input of my peers at any stage of development. I don't find it too odd that someone would like to point out an error in your ongoing word, but I appreciate that you may feel otherwise. I'd also gladly see the evidence for the proposal you are putting forward, that IIr and Ir did not distinguish within stem classes, because you clearly believe it quite strongly. You are of such great conviction, and I would love if you would show me why. You are under no obligation to do so, of course, but unless one of us has a change of heart, this problem will rear its head down the line. But if you don't want to discuss this while it's still in progress, that's fine. In the mean time, I hope you will be amenable to commentary on other matters when they appear. Also, if you seen any other spooky ghost versions of my account show up, please tell me ;) — JohnC5 07:01, 25 November 2017 (UTC)