Reconstruction talk:Proto-Italic/fāðlā

RFD discussion: April 2023–June 2024

The following information has failed Wiktionary's deletion process (permalink).

It should not be re-entered without careful consideration.

---

Proto-Italic terms with only one descendant

@Sokkjo, Djkcel, Urszag, Kwékwlos, 沈澄心, Ioe bidome, Rua, Thadh.

Pretty much half of our Proto-Italic entries exist as ancestors of only a Latin descendant, like *fāðlā, *doukō, *erzāō, etc. etc. (and a few for Oscan, *āpā, *fuɣtēr, ...). To reconstruct Proto-Italic one should reconstruct from the bottom, i.e. with the internal evidence: the Italic languages. These reconstruction, all ripped straight from De Vaan (2008), are rather reconstructed from the top, i.e. with external evidence: other IE languages (or sometimes just because...). Moreover these entries contain no information that isn't already in the etymology of the Latin entries, and I hence propose their deletion. Catonif (talk) 12:59, 1 April 2023 (UTC)

Oppose. It's still one language. If we can reconstruct from the bottom and apply the appropriate sound changes to PIE to form Latin, then we should do it. Thadh (talk) 13:44, 1 April 2023 (UTC)

An entry should hold information. These entries only work as in-betweens for Latin and PIE, and the only useful things in them are:

1. the link to the Latin entry
2. the link to the PIE one.

It would be like having a museum, with the two only attractions being the entrance and the exit.

And "it's still a language", but of interest to who? Who would be interested in Latin words but their sound changes didn't get quite there? If one is, they can always read something about historical phonetics of Latin. We probably have some content on Wikipedia too. A dictionary is not the place for this: words are more than sound changes. Catonif (talk) 20:18, 2 April 2023 (UTC)

The entries show the vocabulary that can be reconstructed for Proto-Italic, which is pretty important information for historical anthropology. You don't have to reconstruct it yourself, but I don't see any problem with anyone else doing it, as long as it's based on the sound phonological system of Proto-Italic as deduced by means of comparing Oscan/Umbrian and Latin. Thadh (talk) 21:20, 2 April 2023 (UTC)

I find Catonif's argument convincing. You could have the Proto-Italic form in the Latin and/or PIE entry, achieving the same, but taking up less space. —Caoimhin ceallach (talk) 23:16, 4 April 2023 (UTC)

We’re an online dictionary - space is not a concern, and should never, ever be used to justify inclusion or exclusion of terms. I don’t have strong opinions on this topic, but I do have strong opinions on that kind of justification. Theknightwho (talk) 10:55, 13 May 2023 (UTC)

One thing is saving up bytes, the other is preventing visual and informational clutter, getting rid of redundancy, providing ease of mobility and making it harder for misinformation to leak in. Catonif (talk) 11:07, 13 May 2023 (UTC)

I'm in favour of deletion. Unless there is something 'special' about a given Proto-Italic entry, it should not exist if the only known descendant is Latin. On that basis, we would have potentially thousands of P-I entries showing nothing but putative sound-changes and crowding out the 'proper' reconstructions. Not to mention the risk of inappropriately projecting Latin innovations to the Proto-Italic level. Nicodene (talk) 10:21, 7 April 2023 (UTC)

Delete — Fenakhay ^{(حيطي · مساهماتي)} 15:39, 16 April 2023 (UTC)

Delete —Caoimhin ceallach (talk) 10:24, 18 April 2023 (UTC)

Also pinging @Fay Freak, @Mahagaja and @Vahagn Petrosyan who might have a say on this as they seem to have participated in similar discussions. Catonif (talk) 16:28, 6 April 2023 (UTC)

I don't like duplication and I don't trust most reconstructions. I would even prefer to ban Proto-Albanian, Proto-Armenian, Proto-Hellenic, Proto-Italic and Proto-Tocharian appendices altogether. Vahag (talk) 19:00, 6 April 2023 (UTC)

Why proto-tocharian? 19:57, 6 April 2023 (UTC) 2A02:3032:7:DA74:2D91:3A93:54C0:A408 19:57, 6 April 2023 (UTC)

Because it has only two descendants. To develop Catonif's museum analogy, the Proto-Tocharian museum has only an entrance, an exit and maybe additionally a cloakroom. Not an interesting place. Vahag (talk) 07:34, 7 April 2023 (UTC)

support deleting the proto-italic entries with only latin descendants but oppose deleting ones with only oscan or umbrian (or any other non-latin) descendants--Ioe bidome (talk) 13:51, 9 April 2023 (UTC)

I have mixed feelings about this. In principle, it should be possible to have meaningful reconstructions based on the intersection of a child and a parent. They aren't as reliable as ones created from multiple children, but they do tell us something about the overall patterns of the language at that stage. We may not know the exact form of the reflexes, but we know they exist. That said, we need to be clear about their limitations, and resist the temptation to fill in details using hunches and guesswork.

Unfortunately, our Proto-Italic entries are riddled with contributions by an extremely dubious IP editor, who has also edited as Inkbolt (talk • contribs • global account info • deleted contribs • nuke • abuse filter log • page moves • block • block log • active blocks) and Dim Blob (talk • contribs • global account info • deleted contribs • nuke • abuse filter log • page moves • block • block log • active blocks). They generally geolocate to Sarthe and the Pays de Loire in France, and used to be well known for their characteristic edit summaries: "Errors! Missing informations!" This is someone who has learned the basic rules for creating reconstructions, and likes to make believe that they know everything about the language that's being reconstructed. What they're really doing, though, is creating an elaborate conlang that superficially resembles the reconstructed language. I've reverted their addition of Gothic and Old English translations for words like television, and they also like to add IPA and inflection tables to reconstructions with as much detail as those in any attested modern language. The kind of details they've added to these entries are exacty the kind of thing we should be avoiding at all costs- unsupported at best, just plain made-up nonsense at worst.

So, basically, I think there's a place for a few limited entries of this sort- but we need to go through and clean out all the conlanged garbage and add explanations of how little we can be sure of in such entries. Chuck Entz (talk) 21:21, 9 April 2023 (UTC)

---

To my legitimate surprise this is at quite the distance from reaching consensus, so I'll now do my best to advocate my cause:

1. a lot of entries are said to be ancestors of words which do not in fact appear in any other IE language aside Latin and could be formations as late as Latin: fabula can simply be for + -bula, yet we have *fāðlā, irritus is in- + ratos (yet we have *enratos), audeo is pretty much avidus + -eo (yet we have *awidēō), agilis is ago + -ilis (yet we have *agelis), particeps is pars + -ceps (yet we have *partikaps), princeps is primus + -ceps (yet we have *priisemokaps), for subligaculum we frankly don't need *supoleigaklom (yet we have it), etc.
2. another lot of entries, even though they can't be direct formations in Latin from Latin words, still don't have any IE cognate and hence we cannot know whether they had a declension change/affixiation last-second, nor can we reconstruct proper senses without just copy-pasting the Latin ones. such cases include comis ⇐ *(s)mey- (but we have *komsmis), venustus ⇐ *wenh₁- (but we have *wenostos), stella ⇐ *h₂stḗr (but we have *stērlā), stupeo ⇐ *(s)tewp- (but we have *stupēō), lateo ⇐ *leh₂- (but we have *latēō), etc. How can we claim that the Proto-Italic people said any of this?

   For this copy-paste-from-Latin method, I'd also like to mention the funny example of *kūdō in the sense “to stamp, coin (money)”. Spoken proto-Italic can realistically be placed at around the beginning of the first millennium BC, with the first inscriptions appearing in the 7th century BC in Umbrian and Faliscan, while the earliest coins seem to be from 600 BC Lydia, with Roman coins appearing only in the 3rd century BC (although presumably having been in use already by a couple of centuries by the neighbouring Etruscans and Italiotes).

3. the same goes for unexpected phonological changes, as we cannot know if they occurred as late as Latin or as early as Proto-Italic, in cases like lutra ⇐ *údreh₂ (yet we have *utrā), sol ⇐ *sóh₂wl̥ (yet we have *swōl), etc.
4. as Chuck Entz illustrated, they're hotbeds for daydreamers (or "conlangers", as he said), as also evidence by the usage examples that have been added to some entries, such as in the above-mentioned *fāðlā, and in *əngʷnis, *wīs, *wīnom, *wai which Nicodene recently sensibly removed (mind you, these of course were straight up Latin sayings without taking into consideration comparitive Italic grammar). Another thing that shows the daydreaming of editors is for example the alternative form at *priisemos being qualified as "dialectal", or qualifying as "rare" one of the tables of *magjōs. All this not to mention the pronunciations, which Surjection thankfully recently got rid of (BP).
5. we even have non-lemma forms like *wenostiōs, *wenostisemos, *breɣʷjōs and *breɣʷisemos (and on what grounds are we saying *-isemos and not *-issəmos?).
6. so after we've crossed all those entries out, we're left with totally uncontroversial entries which must have existed in proto-Italic for sure, such as *mari, *doukō, *ezos, *swekrus, *neptis, *lakus, etc. This category of entries, unlike all the ones above, don't actually give any misinformation, but allow me to ask to the editors who would like to keep these just what have they learned after visiting such entries? What new information have you acquired? Being unreasonably lengthy is something we can definitely afford, but we must remember that every byte we host costs time for editors to maintain and for readers to read.

I beg opposers to reconsider, and passers-by to express themselves. Excuse the long post. Catonif (talk) 13:03, 15 April 2023 (UTC)

I think you're trying to solve multiple different issues at once. If there are any entries reconstructed as PIt. and/or PIE without any cognates at all, then of course, they should be created.

If there are entries that may have a cognate but where this cognacy is disputed and the formation might be parallel, then we should re-examine these cases and solve this on a case-by-case basis.

But when there is a clear cognacy to other IE terms in other branches (for instance piscis, regō etc.), and no attested other Italic cognates, but no indication of a borrowing and no controversy among experts as to whether this is an inheritance from PIE, then why not reconstruct PIt.? Thadh (talk) 13:27, 15 April 2023 (UTC)

Those fall under point 6. What information would you put in the entry that doesn't make it just an extra click? If we really must reach a compromise, I can allow those as being left unlinked. Catonif (talk) 14:15, 15 April 2023 (UTC)

I have already explained why it's important to have these entries: They show the lexicon of Proto-Italic, which is important for linguistic paleontology. Not all PIE terms made it to PIt, and not all Latin terms were already present in PIt. Thadh (talk) 14:18, 15 April 2023 (UTC)

This objection is easily allayed if you consider that you can include the Proto-Italic intermediate form in the Latin entry.

Or is it a neat list of all Proto-Italic terms that you want? What exactly would that add to wikt:Category:Latin_terms_derived_from_Proto-Indo-European that it justifies thousands of entries which seen individually are perfectly informationless? Caoimhin ceallach (talk) 19:17, 15 April 2023 (UTC)

Yes, it's the neat list that I want. And no, "Latin terms derived from PIE" doesn't cover this, because it doesn't include the terms that were formed in PIE specifically. How is it more useless than any reconstruction whose descendants are regularly developed? Thadh (talk) 19:47, 15 April 2023 (UTC)

What do you mean by "terms that were formed in PIE specifically"? Why aren't they in the list?

Isn't a lemma whose descendants were all perfectly behaved a rather exceptional case? Usually there are irregular changes and it's the job of the reconstruction to carefully evaluate all the available evidence and figure out what happened. That's the kind of stuff that gives and entry purpose and that I want to see in an entry.

Let me turn it around: what's so special about Proto-Italic that you want to do linguistic palaeontology on it? Why not Proto-Latino-Faliscan or Proto-Latino-Sabellic? If we let go of the of the criterium "the entry needs to be independently informative", where would you recommend drawing the line? Caoimhin ceallach (talk) 22:03, 15 April 2023 (UTC)

Sorry, I meant "formed in PIt. specifically".

And no, a lemma whose descendants are perfectly behaved are pretty regular for most small language groups and families. Proto-Permic *ku̇k, Proto-Finnic *neljä, Proto-Polynesian *ono, Proto-Slavic *ty. In fact, any historical linguistic model aims to explain as much as possible without irregularities, and the fact that there are a lot of irregularities often means that the reconstruction is inaccurate or that a sound law is not found yet.

"What's so special about Proto-Italic [...]?" If you really need to ask this question, I don't really understand why you'd bother to work on etymologies at all. Historical linguistics don't have such practical implications as mathematics do, sure, but if we do agree that it generally has value for this dictionary, then so does Proto-Italic. We're already reconstructing the language because the Italic languages are closer to each other than to other PIE branches, so we should be complete in our reconstructions and add also those forms that are intermediate before Latin. Thadh (talk) 22:43, 15 April 2023 (UTC)

How would you tell whether words formed in Proto-Italic and not in say Old Latin?

In the branches I'm most familiar with (Germanic, Celtic) straightforward reconstruction seem more the exception than the rule. Maybe they're unusual branches in that respect, but I doubt it. And it's not because we don't understand the sounds laws well enough. Rather it's morphological change, analogy, and borrowing that throw spanners in the works.

Actually since you mentioned numbers, they show quite a bit of irregular change in IE. Look at Latin vīgintī, Old Irish cóic, Proto-Indo-Iranian *šwáćš, Proto-Germanic *fedwōr. All of these changed in irregular ways and without cognates you'd be unwise to guess at specific ancestral stages. (Proto-Slavic *ty is problematic too if you zoom out a bit, because Baltic inherited a short-voweled version of PIE *túh₂.)

Yes, but why Proto-Italic and not say Proto-Latino-Faliscan or Proto-Latino-Sabellic? They were also real languages. My understanding is we reconstruct whatever we can by comparative means, because it's informative. But if there's nothing to compare, we don't, because it's not. Caoimhin ceallach (talk) 00:34, 16 April 2023 (UTC)

We can tell because of other Italic cognates.

Morphological irregularities have nothing to do with the phonological regularities of the lemma.

We're reconstructing the lemma only for PIt. If we were to reconstruct morphology, it would be even more informative, yes.

PS. *ty < PIE *tuH is completely regular and even if it weren't, it's a regular reflex of PBSl, so by your standards we would have to keep PBSl only?

I'm guessing because these are too close to PIt. to warrant a separate L2. But who knows, maybe we collectively forgot about them? They can't be much less informative than Proto-Eastern Polynesian, Proto-Nuclear Polynesian and others. Thadh (talk) 08:27, 16 April 2023 (UTC)

We've been talking about lemmas without Italic cognates the whole time.

Morphological irregularities have everything to do with reconstructability.

If you reconstruct Proto-Balto-Slavic *tūˀ you're ignoring the fact the Baltic did not inherit that form. We should reconstruct, but whatever we do, we need to deal with the irregularity.

Can you be more explicit? If we allow thousands of information-free Proto-Italic entries (ie without non-Latin cognates), why not Proto-Latino-Faliscan? Why is that stage not interesting enough for linguistic palaeontology?

Again, whether an entry for a proto-language is informative is determined by the presence of cognates, not by the closeness to its daughter languages. Caoimhin ceallach (talk) 10:01, 16 April 2023 (UTC)

I think you're misunderstanding. I'm calling for keeping terms without Italic cognates and with a solid IE etymology in PIt., because in combination with terms that do have Italic cognates, these terms give us insight in the linguistic and cultural reality of PIt. speakers. It's not about what language is more important, it's about being complete - if we're reconstructing a language, we have to be as complete as possible. There is no shred of doubt that these terms existed in PIt., so why on earth are you against the creation of entries for them? What you're proposing is equivalent to excluding attested terms that are only used in a particular dialect from the namespace and only giving them under the "synonyms" header of the standard word. Thadh (talk) 10:18, 16 April 2023 (UTC)

@Thadh I did not properly check the examples you gave before answering. Turns out that while rego does indeed fall under point 6, piscis actually falls under point 2, as we have no way of reconstructing it in the proto-Italic stage. How can we know it was an i-stem already when all other cognates reflect an o-stem paradigm? But that was just an example, so this is probably beside the point.

The category CAT:Latin terms inherited from Proto-Indo-European should (though admittedly probably needs cleanup) contain all words that were surely formed as early as PIE and inheritedly ended up in Latin.

And I disagree with prioritising categories over mainspace. “Hey, you guys just made me read a full entry and I learned nothing. — Yeah, but we need it for the neat categories which nobody looks at, for linguistic paleontology.” (Hey what if we did From {{der|la|itc-pro||*term}}{{cln|itc-pro|lemmas}}? Haha tech people here would kill me.)

“I think you're trying to solve multiple different issues at once, […] we should solve this on a case-by-case basis.” I have a barrel full of mud and you're asking me to clean it with a sponge. I can't. I have to empty the barrel first. Catonif (talk) 21:42, 15 April 2023 (UTC)

As I already said before, if you don't want to clean the barrel, don't. It's up to you. But I don't think we should throw the barrel out just because you think it's too difficult to clean it. Thadh (talk) 21:56, 15 April 2023 (UTC)

Delete per proponent. PUC – 15:24, 16 April 2023 (UTC)

A good part of the uncontroversial examples you listed are sourced. I would say a good compromise would be to allow reconstructed terms with a single descendant if it is properly referenced. - Sarilho1 (talk) 10:15, 19 April 2023 (UTC)

Comment for now ... if we delete these pages, we wouldnt have anywhere to put the inflection tables, so we should at least make sure readers have a place to go to figure out what the inflected forms would be. That said, I'm abstaining for now because when I asked a similar question last year about why we have so many redlinked PIE terms in the proto-Germanic reconstruction space I was told that we don't make pages for IE roots that are based on a single descendant family. Is this policy also applicable to this situation, or is it different when we have both a descendant and a parent? —Soap— 16:05, 19 April 2023 (UTC)

Well a month has passed, and I'm seeing a 7-3 for deletion. Per RFD practices, I'd close this as RFD-deleted. I'll deal with entries individually by orphaning them first and then setting them with {{d}}. Catonif (talk) 12:05, 12 May 2023 (UTC)

(Pings:@Thadh, Caoimhin ceallach, Nicodene, Vahagn Petrosyan, Ioe bidome, Chuck Entz, PUC, Sarilho1, Soap)

I've had to revert @Catonif because they seem to believe that this vote forbids single-descendant Proto-Italic reconstructions on PIE entries and was deleting them. --{{victar|talk}} 21:36, 12 May 2023 (UTC)

I have already expressed my opinion on single-descendant reconstructions. Thadh (talk) 21:39, 12 May 2023 (UTC)

@Thadh: Right you did, but this was a vote for deleting Proto-Italic entries, not a policy vote to disallow reconstructions in the descendants sections on PIE entries. --{{victar|talk}} 21:46, 12 May 2023 (UTC)

@Victar I don't understand why you want to link to deleted entries. PIE entries are not a magical place where what is deleted elsewhere is allowed to stay. Catonif (talk) 22:09, 12 May 2023 (UTC)

What I want or don't want is irrelevant -- you simply went beyond the scope of this vote. --{{victar|talk}} 22:13, 12 May 2023 (UTC)

I stand by what I wrote earlier .... though more sure of myself now, because nobody answered my question .... but it looks like we made our decision and the proto-Italic terms are now deleted. I would have liked them to stay, because they contained useful inflection tables. I have no opinion on the newer question which I'd even say should be taken as an entirely separate policy debate, in a separate thread or at least a clearly marked subheader of this thread. It took me awhile to figure out what was going. Though I dont work with reconstructions that much, so perhaps its just me. —Soap— 22:16, 12 May 2023 (UTC)

this should be deleted too, right? just making sure I understand whats going on. Thanks, —Soap— 22:18, 12 May 2023 (UTC)

I'm sorry for not answering you earlier, @Soap. The thing is, I don't think inflection tables are that valuable. If one is interested in common Italic inflection, Wikipedia is probably a good place to house that information. Inflection tables are essential for attested languages, to understand what each form found stands for, while for unattested languages, having that information scattered in each entry is firstly somewhat redundant (would only serve people who are trying to "speak Proto-Italic") and secondly somewhat risky, that is, it looks something very prone to have a Latin bias. See for example *fūmos that has as its nom.pl. "fūmōs, fūmoi" the first form being the original PIE one preserved in Osco-Umbrian and the second one being a Latin innovation. I also find it somewhat silly to list -osio as the gen.sg. ending in every single o-stem PIt noun after it being found only once (and note, although we have -i in Celtic as well, Osco-Umbrian actually agrees on an -eis for gen.sg., so reconstructing a PIt ending is a bit of a wobbly matter). This is especially true for verbs, where O.-U. data is more scattered and less clear. It's probably worth noting how an O.-U.-only PIt verb would list -esi as the infinitive ending, which is completely unknown to O.-U. (as those languages all agree on -om) and to pretty much any other IE language, after excluding the similar, yet not identical, Greek -ειν. Now I'm not proposing to delete all PIt inflection tables, but I'd like to make clear that they're in my opinion not something that we should let prevent us from deleting an entry.

And yes, *əngʷnis would be deleted. Is that a problem for the synonym thing? I think the priority is coherency within an entry. I wouldn't keep entire pages for "structural need", we're not Wikidata. But if you want we can make that a separate issue and discuss if that should be an exception, not letting one or more entries get in the way of dealing with the other hundred. Catonif (talk) 11:07, 13 May 2023 (UTC)

OK thanks. I dont know if Wikipedia as a whole would be willing to house all of the inflection tables we'd need, since I assume scholars disagree with each other here. We have them for PIE, but I guess PIE is a lot more popular in the sense of attracting scholarly attention. That, and I'd hope that if we had the tables here we'd have some way to feed a verb into it to see the inflected forms even if we don't have a page for that individual verb.

I think -osyo makes sense, for reasons I wont go into here. I guess that just proves the point, though, that what I think of as scholarly disagreement could also be called original research if we adopt a pick-and-choose approach to it. Thanks, —Soap— 10:34, 14 May 2023 (UTC)

Don't orphan the entries, just delete them directly --Ioe bidome (talk) 21:54, 12 May 2023 (UTC)

@Victar: a redlink is an implicit request to create an entry. If you want to display the reconstruction as plain text without linking to it that might be okay- but redlinking is a bad idea. Chuck Entz (talk) 22:41, 12 May 2023 (UTC)

If someone wants to start a policy vote to disallow Proto-Italic redlinks (or delete PItc reconconstructions, as they were doing), they can do that, but that was not the scope of this vote. --{{victar|talk}} 22:51, 12 May 2023 (UTC)

@Victar Please read carefully what Chuck and I have been telling you. It is like if we decided to delete all Esperanto entries and someone would keep adding them as redlinks in {{desc|eo}}s in their etymon's descendants because "the vote was about deleting entries, not a policy vote to disallow them in descendants sections". Catonif (talk) 11:07, 13 May 2023 (UTC)

Big nope, buddy, that's a false equivalence, the key being we're not "deleting all" Proto-Italic entries, only specificly ones with a single descendant. That vote would have looked like Wiktionary:Votes/2019-01/Banning Altaic. --{{victar|talk}} 19:55, 13 May 2023 (UTC)

Since I'm rather puzzled by the situation myself, I will try to make it clear for others to judge. What I was doing is, after having closed the RFD, having gone and subsituted {{desc|itc-pro|*whatever page I just orphaned and tagged for speedy}} with Italic:, following what is already a widespread practice in PIE entries. I fail to see the illegitimacy of my edits, and would like to continue, but since, to my surprise, this was not well-seen by Victar, who put into question the validity for the vote to back up the changes (and note, it is not the deletion that is being opposed to, which Victar seems to accept, but the edits on the PIE entries), I ask here the involved editors, regardless of them supporting or opposing of the change itself, whether they as well believe that my changes went beyond the vote's purpose, and if not, if I have the green light to continue without being reverted. Catonif (talk) 12:50, 15 May 2023 (UTC)

I did assume that if we were to collectively disallow PIt. entries with only a Latin descendant from the reconstructed mainspace, that we would also do it in descendant sections. I can't think of an argument in support of keeping these other than ones that would support keeping the entries, as well. Thadh (talk) 15:11, 15 May 2023 (UTC)

I vote for "delete" (in both cases). I don't understand victar's point about "the key being we're not "deleting all" Proto-Italic entries, only specificly ones with a single descendant." I assume Catonif is not deleting all Proto-Italic forms in the descendants sections on PIE entries. What is being deleted are entries for Proto-Italic terms with only one descendant (not a ban of "Proto-Italic" as a whole so nothing like "Banning Altaic"), as well as links to such terms. I agree with Thadh that a vote for deleting the entries should be good for deleting the links as well. If there is actually some real loss of information from removing these links (and leaving the attested Latin forms that are the actual basis of the PIE reconstruction), I don't see what it is; I'd like to hear what the objection is aside from the purely procedural complaint.--Urszag (talk) 15:27, 15 May 2023 (UTC)

I've never questioned the legitimacy of this vote, and, to be honest, I support it. As Bhagadatta pointed out, we went through the same thing in Proto-Indo-Aryan, and Proto-Hellenic is in the same boat. What we didn't do in Proto-Indo-Aryan was delete the reconstructions in etymologies and descendants lists, and, to my knowledge, at no point did anyone assume that was going to happen.

If people are really concerned about single-descendant Proto-Italic redlinks, we can use |alt=, as suggested, but deleting sourced reconstructions is unprecedented and counterproductive to the project.

--{{victar|talk}} 19:07, 15 May 2023 (UTC)

I don't see how deleting things like Proto-Italic *akēō as an intermediary between Proto-Indo-European *h₂eḱ-éh₁-ye-ti and Latin aceō is harmful. Will anyone be confused by its omission, or aided by its presence? It seems trivial; for that reason, I suppose there isn't actually much doubt about the existence or form of the word at the common Italic stage, but it is also true that it was not reconstructed by comparison of multiple Italic descendants, so it seems arguable whether it falls under the definition of Proto-Italic as a theoretical construct. I'm not greatly concerned either way, but currently Catonif's edits seem appropriate to me.--Urszag (talk) 20:22, 15 May 2023 (UTC)

I think you know as well as I, not all Latin reconstructions are straight-forward and the Italic reconstruction is important for speculating an ancestral PIE reconstruction, as well as sourcing the chain of descent. Deleting them has no benefit and all the loss. --{{victar|talk}} 22:42, 15 May 2023 (UTC)

What I get is you're proposing to delete the entries, removing the forms from etymologies, but keeping them unlinked in descendants, or at least that seems to be the Proto-Indo-Aryan situation (except for the "unlinked" part).

About loss, like Urszag, I don't see the value or information in this, and I thought we discouraged mentioning unattested intermediary if the outcome is regular, e.g. we don't say Italian piede comes from Proto-Italo-Western *pęde, nor do we have that in the descendants of Latin pes, even though I'm fairly sure the reconstructed form could be easily found in sources. About benefit, getting rid of redundancy is always a benefit. Thinking pragmatically leaving them in PIE entries will also surely encourage laypeople to place the form in the Latin etymologies as well, considering them "missing", which I'd rather avoid.

It is also true that in much cases, like *akēō for example, we can't be sure already existed in PIt, as the stative suffix -éh₁yeti > -ēō > -eō is productive all throughout its history, so we theoretically can't exclude a post-PIt derivation from an adjective which then died out before being able to be attested. This reasoning would condemn the listed PIE *h₂eḱ-éh₁yeti as well, though the hyphens help much there, making it fairly clear that the proposed "Proto-Indo-European" etymon is nothing more than a morphological "surface analysis". A form like Proto-Italic *akēō on the other hand is confidently telling us "If you went to Proto-Italy^[sic], you would 100% hear people say this", which although probable is just speculation. For this I'd rather not list these forms at all, though I guess *ak-ēō is also a solution.

I wish it wasn't, but if all this is the only way to go forward without being reverted, I see myself as probably having to comply. By now I've also given up on the first and probably most important thing of the entire situation, which is that the Proto-Italic as a theoretical concept is whatever is reconstructible from comparison of the Italic languages, which I see only Urszag taking into consideration. Catonif (talk) 15:38, 17 May 2023 (UTC)

Delete. A similar policy was adopted for Proto-Indo-Aryan entries with only Sanskrit as the descendant. -- 𝓑𝓱𝓪𝓰𝓪𝓭𝓪𝓽𝓽𝓪^{(𝓽𝓪𝓵𝓴)} 02:48, 14 May 2023 (UTC)

Delete per Bhagadatta. I'm strongly against these useless and redundant entries. Svartava (talk) 16:45, 14 May 2023 (UTC)

@Catonif I am not expressing a strong opinion here but I should add that Ringe, who I respect a lot, allows Proto-Germanic reconstructions if there's only one descendant but a clear PIE antecedent. *fuɣtēr seems like a clear example of this. Benwing2 (talk) 19:36, 24 June 2023 (UTC)

@Benwing2 I'm sorry the contorted discussion doesn't make it very clear, but this was closed a while ago. I've already deleted many entries. As for Ringe, we probably don't need to make that analogy since de Vaan himself, as shown, is the one reconstructing most of these forms. Although it was just an example, it may be good mentioning it isn't as neat as it may look. The expected PIt. outcome of the nominative *dʰugh₂tḗr would be **fugatēr. There are formal problems with the proposed etymology (the unvocalised laryngeal (possible levelling?), and de Vaan mentions the absence of any attestation with [h] as striking) and overall no clear-cut evidence the word properly means "daughter" (as most ancient and poorly attested languages' words sadly do). Catonif (talk) 23:16, 24 June 2023 (UTC)

Post-closing comment I vehemently disagree with deleting one-descendant reconstructions as a general principle if scholars generally reconstruct it. But I can understand how Proto-Italic has an imbalance leading to such entries feeling more worthless than usual. Italic's attestation is extremely Latin-heavy, Latin has pretty productive affixes, only one source (De Vaan) ever posits many of these pre-forms, and the morphology is often trivially derivable from Latin's. — Ceso femmuin mbolgaig mbung, mellohi! (投稿) 15:04, 12 September 2023 (UTC)

Well, I think this particular discussion has only ever been about the specific case of Proto-Italic, even if some of the arguments involved could have applicability to other cases. Another reason I would favor deleting single-descendant reconstructions in this case is that I have seen some linguists dispute the validity of "Italic" as a clade of Indo-European (in the words of Benjamin W. Fortson: "The differences between Latino-Faliscan and Sabellic are not trivial and have led some to view Italic as a pseudo-branch or Sprachbund that arose through convergence of geographically contiguous but phylogenetically not closely connected dialects, rather than as a node on the Indo-European Stammbaum. This view goes back to the 1910s and 1920s and originated in Germany, though it came to be identified especially with Italian scholars"). Here on Wiktionary, we reconstruct Proto-Italic in accordance with the mainstream view, but I think including a bunch of forms based only on Latin makes it harder to use our entries to see how much evidence there is for the family itself.--Urszag (talk) 17:45, 12 September 2023 (UTC)

RFD-delete Proto-Italic entries with only a Latin descendant with a tally of 9–3 or so. I think this decision was taken long ago, so it's time to close this. —Caoimhin ceallach (talk) 16:38, 1 June 2024 (UTC)