Reconstruction talk:Proto-Indo-Iranian/sarȷ́-

@माधवपंडित So, I traced the etymology back to PIE. —Aryamanarora ^{(मुझसे बात करो)} 00:23, 15 May 2017 (UTC)

@Aryamanarora: Wonderful etymology work man. Thanks a lot! माधवपंडित (talk) 01:54, 15 May 2017 (UTC)

Root entries

@AryamanA, JohnC5 I think this, and other root PII root entries should be deleted and moved to lemma entries. Like in PIE, there really is no such thing as a root word. --Victar (talk) 01:18, 5 December 2017 (UTC)

@Victar: I think it's mainly useful as an index to entries that otherwise don't link to each other. IMO they should be kept just for that reason. I don't really have a good enough understanding of PIE to say anything else; of course, I doubt the Proto-Indo-Europeans were actually aware of and actively used roots. —Aryaman^{A (मुझसे बात करें • योगदान)} 01:38, 5 December 2017 (UTC)

@AryamanA: That's what categories are for, i.e. {{PIE root|iir-pro|bʰer}}. If that's the only surviving reason, I think creating an artificial root entry is unnecessary. --Victar (talk) 01:43, 5 December 2017 (UTC)

@Victar: What about for roots without a clear PIE derivation? Honestly, this conversation is way beyond my level. JohnC5 will probably have more to say. —Aryaman^{A (मुझसे बात करें • योगदान)} 01:57, 5 December 2017 (UTC)

We also have the Related terms and Derived terms sections. --Victar (talk) 02:00, 5 December 2017 (UTC)

A moment I thought about how much there is a point in the claim that the Indo-European root entries around here are lying by stating hierarchies that never existed. But it looks to me like nobody has defined in an universally valid way what a root is. The concept is unlike such categories as noun or preposition because there is no direct observing of them imaginable. There are working definitions of the term root. And there has been quite some richness achieved with a certain working definition in the English Wiktionary. The reader can understand it too.

Categorization cannot replace caressed pages with the same amenity. And I point out that these reconstructions have a dedicated namespace that can have its own rules. The most undoubted use of reconstruction pages is being an index leading to actually attested forms. Palaestrator verborum (loquier) 04:29, 5 December 2017 (UTC)

@Victar: Hmmm, I very much appreciate your point that roots are a formal category implemented after the fact. In the case of Sanskrit, I will note that they were very much a category that people considered and which led to innovation and leveling. Where the IIr and Inc speaks perceived them that way is a different question; though the fact that Sanskrit possesses many nouns made of bare root words seems to indicate some familiarity with the root. To answer @AryamanA's question, there are definitely IIr roots that cannot be reconstructed back to PIE. It also seems a bit odd to have roots for PIE, none for IIr, Ir, and Inc, then suddenly have then again for Sanskrit and Avestan. Similarly, there are many Iranian verb forms that don't really come from any recognizable PIr conjugation but are just rebuilt wholesale from a root, and these would be hard to categorize in the absence of a root category. I do admit that is a strange to connect up these forms with a virtual root, but no more strange than our practice in Ancient Greek of choosing the 1st person present active singular then dumping a heap of semi-related conjugations under it. I think we should keep them given their utility and that academic literature also reconstructs them. I also think they do no harm. —*i̯óh₁nC^[5] 10:13, 5 December 2017 (UTC)

1. @JohnC5: "Sanskrit possesses many nouns made of bare root words seems to indicate some familiarity with the root." I would contest that. You could attempt to use the same logic for PIE, but from I have read, scholars all agree that roots, as we reconstruct them, did not exist in PIE. You can also find full-grade restorations in PGmc.
2. "Academic literature also reconstructs them [in root form]" As is often said, Wiktionary is not beholden to conform to other dictionaries. We don't do so for PGmc, we don't do so for PCelt, and we don't do so for Hittite. I've also never seen any literature reconstruct Avestan roots as done on this page, ex. harəz, with no trailing dash, like Sanskrit, but rather like harəz-, as found in Cheung.
3. "It also seems a bit odd to have roots for PIE, none for IIr, Ir, and Inc, then suddenly have then again for Sanskrit" Therein lies the conflict really -- we have a standard for Sanskrit with root and lemma forms that differ from all other PIE descendants. I don't think trying to force all other PII descendants into that box is the way to go. And just to mention, we have 482 PIE reconstructed roots, and 500 reconstructed lemmas, and the lemma forms are only going to grow from there. --Victar (talk) 00:12, 6 December 2017 (UTC)
4. To @Palaestrator verborum's point, I really don't see the advantage of having both *sarĵ- and *sr̥ĵáti. It just seems like a duplication of entries.

--Victar (talk) 00:12, 6 December 2017 (UTC)

@Victar I don’t see it either, other than possibly one wants to link the root in one context and the verb in an other one. But better have such a entries than to remove all the roots, as this lack of advantages is not generally true. It could be a dangerous precedent. Palaestrator verborum (loquier) 00:21, 6 December 2017 (UTC)

@Palaestrator verborum: We're at the beginning, really, of adding PII entries, so we're only talking about 20 entries, most of which can be moved to a proper verb lemma entry. I don't see any danger in doing so. --Victar (talk) 00:33, 6 December 2017 (UTC)

@Victar:

1. You are certainly correct that there was no "root" per se, but the effects of a root-based system are still obvious in PIE and IIr. The formalism of the root itself may not have existed (though of course the earliest grammarians immediately identified the root as the governing part), but the continued expansion of the system with further derived conjugations, secondary roots, and secondary root nouns indicates that the system was still accessible in some form. Of course they didn't exist as a semantic category, but in the same way we have roots for Afro-Asiatic languages that used similar derivational system, so should we here.
2. We don't have them for Gmc, Celt, and Hitt because these languages no longer operated on a root-based system as Sanskrit and Avestan did. You say that we are not beholden to other dictionaries, but this is exactly the type of information that scholars expect to find. All three of my Indo-Iranian professors refer to IIr roots in class and papers. The fact that a user on Wiktionary doesn't like the idea does not mean we should just buck a whole history of scholarly tradition. As for a trailing dash in Avestan, it's hard to say when using the original alphabet. Kanga's dictionary does not provide trailing dashes in the native script, but I'm all for having them in the transliteration.
3. Again, we should only construct roots for languages and language families in which the root system is still active. Sanskrit and Avestan's verbal systems are not single paradigms but constellations of semi-predictable forms united only by their possession of a single shared part, the root. The systems in Germanic, Celtic, Latin, and to some extent AG have massively collapsed into unified paradigms. We would not give roots to these languages because these languages don't fit this system.

The main issue remains that there are strong scholarly traditions that use these as lexicographic tools for the easy categorization of descendants which do no harm to the project and which clearly relate reflexes that do not belong to any reconstructable conjugation. I appreciate that they may seem duplicative, but let's not throw the baby out with the bathwater in a quest for streamlining. Also, this still doesn't answer the question of categorizing related IIr forms with no PIE root. Using Related terms works to a certain extent (though it drives me crazy, sometimes), but having one place where you can find them all is crucial. To be clear, I'm all for the use of verb entries were reconstructable, but there is danger in removing an entry type that is typical and expected in the field.

I hope this clarifies my logic. I wonder what @Mahagaja, Rua, माधवपंडित, Barytonesis think of this. —*i̯óh₁nC^[5] 00:58, 6 December 2017 (UTC)

Thanks for the reply, @JohnC5.

1. Perhaps you can help explain this better to me. Theoretically, why would PII *bʰárati require a root to form a deverbal noun from it, like *bʰármā́, when you could just as easily drop the understood verb suffix and affix *-mā́ to it? Germanic has no trouble doing this, and even creating a secondary causative and making a whole new deverbal noun from that. Could you give an example of what I'm missing?
2. The same exact argument was made for keeping Celtic lemmas on Wiktionary caseless. Scholarly publications tend to prefer stems, as seen in the Leiden series, and on this project, we've intentionally chosen to break from that. So to paint me as some lone-gunman baby-killer is a bit inaccurate. =P I think when it comes to Avestan being a root-based system, it's being looked at with Sanskrit-tinted glasses, simply because it's an PII language. I'd like to see some examples of this as well, so that I might better understand the argument for it.

I'm not trying to be stubbornly anti-root entries. As expressed above, perhaps I'm missing some key points. What I am against though is creating root entries by default, with only one derivation listed, when they could instead be made into lemma entries, which reflect its descendants. --Victar (talk) 17:06, 6 December 2017 (UTC)

@Victar: Props for the phrase "lone-gunman baby-killer"—I'm gonna' use that in meatspace. I suppose to your first point, sometimes when creating a noun like *bʰármā, it's a matter of source. why do the *-mā stems come from the present as opposed to a root aorist? Choosing one over the other could get very dicey, unlike say Latin from which the secondary deverbative formations effectively could be built only to the present stem or the perfect stem. And this is not to say that words were not formed from the conjugations; the aorist active participle of a causative definitely is not built wholesale from a root. But if you have a full-grade root noun, does it come from a full-grade present, a full-grade aorist, or an unreduplicated perfect? Of course, in the actual system, they probably had a system of analogy with other similar constellations of forms that helped them choose how to create new forms, but sometimes it's really hard to tell what would be the parent.

Yeah, the stem versus nominative singular thing still haunts me. The reason why people use stems, I have always believed, is that they are easier: you don't have to figure out which word-final sound laws you're going to show, you don't have to reconstruct the desinences, and it frequently avoids ambiguity in the nominative singular. Fortunately, we have Rua to decide all those things for us (;P)! In truth, I think we chose the nominative singular because in most languages this is the preferred lemma (though not in Sanskrit and Avestan, of course). I know this countervails my preference for roots somewhat, but then again Hebrew and Arabic both have these roots as well and they serve as very helpful tools. When I'm looking at Arabic, I'm always pleased to jump to the root page and see the variety of fun formations.

Again, my main concern is that they are very useful and familiar lexicographic tools that I would sincerely miss. Of course they don't exist, but when your system is still as whacky as Sanskrit and Indo-European, where the present conjugation will tell you nothing about what type of aorist, perfect, or participle you can expect, it makes sense to have a unified place to look. —*i̯óh₁nC^[5] 05:15, 7 December 2017 (UTC)

Hmm, we rarely have entries for verbs and roots in PII; we mostly do nouns and adjectives. I'm wondering where an entry like *bʰiš- would go if we decided to get rid of roots. Like @JohnC5: said, since both Avestan and Sanskrit operate on verb roots, and since we also have entries for verb roots in Sanskrit, I think we can keep these PII root entries -- माधवपंडित (talk) 01:08, 6 December 2017 (UTC)

I don't know a whole lot about PII, but since we do have root entries for both PIE and Sanskrit, and since I consider root entries in those two languages useful (even in Sanskrit there may be more than one present stem, and therefore more than one Verb entry, for a root), I'm not opposed to root entries for PII. —Mahāgaja (formerly Angr) · talk 10:10, 6 December 2017 (UTC)

More thought into this. Sanskrit (and maybe Avestan) roots reflect the same PIE --> Skt sound tranformations that full, meaningful lemmas do. If not always, then many a time. I've always wondered how this can be, if we are to believe that Sanskrit roots are a later creation of grammarians. This is why in my early days with Wiktionary I used to believe that PIE roots were reconstructed by applying (in reverse) sound laws to Sanskrit verb roots. But that is not so. This may mean PII speakers were aware of the concept called verb roots, which they inherited from their PIE days and passed on to their descendants. Correct me if I'm wrong... -- माधवपंडित (talk) 13:07, 6 December 2017 (UTC)

Oh Panini et al. were definitely aware of certain sound changes like Grassmann's Law even though they did not have the luxury of comparative linguistics. I think roots were contrived by grammarians as a way to analyze sound changes when prefixes and suffixes were added, and also to document ablaut. Note that obviously made-up letters like ॠ (ṝ) were only used in roots. —Aryaman^{A (मुझसे बात करें • योगदान)} 22:01, 6 December 2017 (UTC)