Wiktionary:Requests for deletion/Reconstruction

__NEWSECTIONLINK__ '''This is a combined verification/deletion page for any reconstructed entries, i.e. those in the Reconstruction: namespace. This includes reconstructed entries in languages for which some attestation exists, such as Latin and Old English.'''

See the following table for other entries:

= 2019 =

Reconstruction:Proto-Germanic/erþaburgz
I have no problem with the word or whenever it existed or not. My issue is in regards to whenever the word should be reconstructed as 🇨🇬 or. This example is one of many PGmc where the first noun of the reconstructed compound ends with "ō" but the reconstructed compound has medial "a". I would normally check the descendent to see if I can deduce more information, however, most have no medial compound vowel e.g. 🇨🇬, 🇨🇬, 🇨🇬. So now, I'm left wondering what form it should be. 𐌷𐌻𐌿𐌳𐌰𐍅𐌹𐌲𐍃 𐌰𐌻𐌰𐍂𐌴𐌹𐌺𐌹𐌲𐌲𐍃 (talk) 03:19, 2 November 2019 (UTC)
 * The default medial vowel in pre-Germanic had become -o- for the thematic classes, as in Celtic and Latin. PGmc medial*-ō- would presumably have left some trace in OHG. Burgundaz (talk) 08:54, 28 April 2020 (UTC)
 * Kept. The discussion both here and on the talk page really seems to be whether the page should be moved to a different title, not deleted outright. —Mahāgaja · talk 14:00, 21 August 2023 (UTC)

= May 2020 =

Reconstruction:Proto-Germanic/gallô
Since all of the descendants from this have been moved over to, I think this can be deleted. DJ K-Çel (talk) 02:34, 30 May 2020 (UTC)
 * No. is the ancestor of the OE form, and  the rest. --  02:49, 30 May 2020 (UTC)
 * Well, at this discussion had said: "I've moved  to, since the West Germanic descendants are weak. I've also added the descendants of  to . I think we can delete ."
 * But it looks like 🇨🇬 and its ancestors were deleted about a week ago from . DJ K-Çel (talk) 03:01, 30 May 2020 (UTC)
 * It all depends on whether we want to keep the page solely for the lone 🇨🇬. Or we could consider the OE term a gender change from 🇨🇬 from 🇨🇬 and place it there. Leasnam (talk) 04:05, 22 July 2020 (UTC)
 * keep - this is a different word to . I've made updates, and removed the tag. Leasnam (talk) 19:50, 3 August 2023 (UTC)

RFD-kept —Caoimhin ceallach (talk) 14:26, 16 October 2023 (UTC)

Reconstruction:Proto-West Germanic/būti
Only has reflexes in one descendant, and an uncertain borrowing. This could easily have been formed within the separate history of Dutch. —Rua (mew) 10:52, 10 March 2020 (UTC)
 * Keep: The etymology is so widely circulated that even if it is wrong, which is hard to say (though I do prefer a direct Gaulish etymology for the Latin), it should just have an entry anyhow. -- 19:58, 10 March 2020 (UTC)
 * But should that be a Proto-West Germanic entry? The term is literally has only one descendant, that's not enough evidence to claim it's of PWG date. —Rua (mew) 20:15, 10 March 2020 (UTC)
 * Well, it's mostly reconstructed as PG, so PWG is even safer, no? -- 20:46, 10 March 2020 (UTC)
 * A term with only a Dutch descendant (which is doubtful, as Etymologiebank says the term is Low German in origin) can't even be reconstructed for PWG, let alone PG. —Rua (mew) 10:38, 11 March 2020 (UTC)
 * And others the opposite, and others still both inherited. -- 20:10, 14 March 2020 (UTC)
 * Related to this a Frankish label could be handy for PWG with only Dutch and Latin descendants. -- 20:58, 10 March 2020 (UTC)

After 4 years with no discussion, I've been bold and moved it to RC:Old Dutch/buti, which of course does not exclude the possibility that the term was present earlier. I've kept the redirect from PWG in case others expect to find it labeled as such. —Mahāgaja · talk 11:15, 7 May 2024 (UTC)
 * , but VL would have borrowed it from Frankish, not Old Dutch. -- Sokkjō 04:13, 8 May 2024 (UTC)

Reconstruction:Proto-West Germanic/fellō
Same as above. —Rua (mew) 12:37, 10 March 2020 (UTC)
 * Keep: OHG added. -- 19:52, 10 March 2020 (UTC)
 * You're aware that this is RFV, right? There's no keep/delete votes. —Rua (mew) 20:16, 10 March 2020 (UTC)
 * Keep: -- 20:25, 10 March 2020 (UTC)
 * ...kay. —Rua (mew) 10:38, 11 March 2020 (UTC)
 * @Rua: So does that resolve this? -- 20:08, 14 March 2020 (UTC)
 * I'm not entirely sure if having only an OHG descendant is enough either. But I'll leave that to third parties to decide. —Rua (mew) 20:18, 14 March 2020 (UTC)
 * If there really aren't any other West Germanic reflexes, then I'd be inclined to delete and just say the Latin is a loanword from OHG. It's not clear where OHG comes from, though, since  doesn't have an OHG reflex. —Mahāgaja · talk 21:14, 14 March 2020 (UTC)
 * Based on the context of the Latin attestations, it looks to have originated from Frankish, not OHG, and if we were to say it didn't exist in PWG, we have to somehow explain how it was novelly constructed in OHG. -- 08:00, 1 April 2020 (UTC)
 * Could it possibly be from 🇨🇬, a derivative of, or perhaps a derivative of / ? Leasnam (talk) 02:20, 24 August 2020 (UTC)
 * @Mahagaja @Victar I apologise for reviving a 4-year old discussion, but from what I can see, the Old High German term only exists in so far as Medieval Latin *might* come from OHG, thus constituting what Köbler calls a "lateinisch-althochdeutsch" mix. In the Althochdeutsches Wörterbuch, there is no 'fello'.
 * One option is to rework the page to *fillō or *fillijō, which several sources construct as ancestor for the Latin term (but that might also come from ).
 * (and then there's the whole issue with, which might simply be a ghost reconstruction, but I'd like to know if recent sources have something to say about those Old English compounds) Exarchus (talk) 15:53, 17 July 2024 (UTC)
 * I think it makes most sense to move the entry to, from , a derivative of . Leasnam (talk) 22:27, 28 January 2023 (UTC)
 * PWG was plenty productive, so  is also possible. --Sokkjō (talk) 09:44, 29 January 2023 (UTC)
 * I've moved the entry to RC:Proto-West Germanic/falljō. if you believe this addresses the RFD, please resolve. -- Sokkjō 19:44, 16 October 2023 (UTC)

The redirect from to  has been deleted as per 's deletion request. Kutchkutch (talk) 19:56, 26 December 2023 (UTC)

RFD-kept. This has long been resolved. Moved to *falljō. —Caoimhin ceallach (talk) 18:00, 1 June 2024 (UTC)

Reconstruction:Proto-Germanic/daliþō
If there are no non-North Germanic cognates, this should be moved to an Old Norse entry. -- 23:20, 28 March 2020 (UTC)


 * Why? — Knyȝt 09:10, 29 March 2020 (UTC)
 * Because it can be formed by, making it's existence in PG questionable with no other cognates. -- 17:19, 29 March 2020 (UTC)
 * That would render a **dald, which cannot be the ancestor of the descendants listed. The PG -i- is needed for the umlaut. — Knyȝt 19:42, 29 March 2020 (UTC)
 * Fair point, so an unattested ON, from + , which actually fits better semantically. --  20:12, 29 March 2020 (UTC)
 * Could you please create some kind of parent entry for these forms so I can delete this without losing info? Thadh (talk) 15:43, 29 December 2022 (UTC)

= April 2020 =

Reconstruction:Proto-Slavic/vьśь
Inaccurate reconstruction and meaning. -- Gnosandes (talk) 07:50, 6 April 2020 (UTC)
 * What makes you say it is inaccurate? 70.175.192.217 01:12, 26 September 2021 (UTC)
 * Is there any reason to doubt this reconstruction and meaning? This, that and the other (talk) 06:20, 23 April 2022 (UTC)
 * Per me, it should be something more like *vьsъ. I doubt that there is any need of <ś> and there definitely has to be a final /*ъ/ in order for the word to correspond properly to the Lithuanian and the BPSl ending. It's very rare for a 🇨🇬 to give 🇨🇬. An example is 🇨🇬 (whence 🇨🇬, 🇨🇬) which gave 🇨🇬. There is no need of <ś> in the reconstructed too, as this would have given a Lithuanian *višas, but it's 🇨🇬 instead. Both this and the Balto-Slavic reconstructions are wrong in this regard, esp. the former. Ентусиастъ (talk) 09:19, 23 April 2022 (UTC)
 * West Slavic data clearly points towards *vьśь (Polish, , Czech , Slovak ) and I don’t see how they could derive from **vьsъ, OCS (cf. gorazd), Russian  too points at least towards the final soft yer. It comes from older *vix- by progressive palatalization. This unpalatalized *x is actually attested in Old Novgorod forms like вхоу. Derksen in Etymological Dictionary of the Slavic Inherited Lexicon also reconstructs *vьśь and explains the -śь as originally locative plural ending (*-xъ in Slavic, generalized from PBSl *-šu < PIE *-su in ruKi contexts) and Lithuanian lack of š by levelling from forms to which ruKi did not apply:
 * "The origin of this etymon may be a Lpl. *uiṣu. In Lithuanian, the š < *ṣ may have been replaced with s when the variant -su of the Lpl. was generalized (F. Kortlandt, p.c.). Slavic generalized the ending -xъ < *-ṣu in the Lpl., which is why the pronoun has *ś < *x as a result of the progressive palatalization. In North Russian, we still find forms with x (cf. Vermeer 2000: passim)."// Silmeth @talk 11:26, 23 April 2022 (UTC)
 * @Silmethule I know about the West Slavic analogues. Also cope:-) Ентусиастъ (talk) 20:01, 23 April 2022 (UTC)
 * To be fair, how does a hard stem explain them? Asking honestly. Vininn126 (talk) 20:11, 23 April 2022 (UTC)
 * Lemma should be renamed to *vьxъ with note (in descendants) about West Slavic *vьšъ and East and South Slavic *vьsь (third palatalization). Sławobóg (talk) 12:32, 20 September 2022 (UTC)
 * ✅ Sławobóg (talk) 17:38, 20 January 2023 (UTC)
 * Are you going to move to *pěnęgь,  to *kъnęgъ,  to *otьkъ, etc. to make Wiktionary consistent with this change? It makes no sense to keep them if we allow vьxъ without the progressive palatization… I’d rather revert this move. // Silmeth @talk 18:23, 20 January 2023 (UTC)
 * No, for these we simply just add "from earlier X", like I did with . Sławobóg (talk) 18:27, 20 January 2023 (UTC)
 * Then why not add “from earlier *vьxъ” to, in order to keep consistency? If your argument is “because it’s not the earliest reconstructible form because Old Novgorodian doesn’t have the palatalization in this word”, then we should also move and other o-stems, because Old Novgorodian did not have the final yer, its  never had the *-as > -ъ change like the rest of Slavic. So… should we also move  to *xlaibas or something? Current Wiktionary Proto-Slavic reconstructions are not the earliest forms (and perhaps a bit anachronistic), but at least somewhat consistent in the features they show. // Silmeth @talk 18:37, 20 January 2023 (UTC)
 * Because ś didn't exist, it was made up by some linguists, meanwhile most dictionaries reconstruct this word as (care to see references?). vьśь was not helpful in any way. Comparing it to  is false analogy. Sources don't even mention Novgorodian (besides Derksen), it is not final argument (just supporting one), main arguments are sound changes and Baltic cognates. Third palatalization is pretty late. If we don't like -x- for some reason, we need to make separate lemma for East/South and West Slavic, and that is nonsense. Just google "vьxъ" and "vьśь" and see the results. Sławobóg (talk) 19:02, 20 January 2023 (UTC)
 * Care to share the recordings of native late Proto-Slavic speakers proving that “ś didn't exist”? Something like which then in different branches merges either with  (west) or  (south, east) seems like a reasonable model of the palatalization to me (in which *ś  would be a real phoneme at some point). Also, we do use the *ś notation for 2nd palatalization too, eg. in inflection of  (loc. *duśě, nom.pl. *duśi) or  (*muśě) which have different reflexes in west and rest of Slavic too.
 * References like Boryś, WSJP, Vasmer, or Melnychuk are right to list two forms, earlier *vьxъ and later *vьs/šь as they’re etymological dictionaries of specific modern languages, where the word went through those stages. But Wiktionary isn’t for specific Slavic language/branch, and generally has been treating *ś as a separate phoneme thus far, *vьxъ as the main form is inconsistent with this.
 * Of course we could change all -śi, -śě, etc. resulting from 2nd regr. palatalization to -xi, -xě, etc. too – but that’d be a bigger change and, I guess, a longer discussion. // Silmeth @talk 20:34, 20 January 2023 (UTC)
 * Then change it back. But before that, can you explain how PBS -as gave PS -ь? Sławobóg (talk) 20:48, 20 January 2023 (UTC)
 * You have the development in Derksen’s: *wiš- (PBS *š due to ruKi) > *vix- > *vьxъ > *vьśь (prog. palatalization changing the consonant which in turn influences the vowel). Same thing as with *-ingaz > *-ingas > *-in/ęgъ > *-ędzь in *kъnędzь, *pěnędzь.
 * As I understand it, Lithuanian having s suggests the original form might have been loc. pl. *wišu (not *wišas) and that Baltic replaced *-šu with *-su (variant of the ending outside of ruKi contexts, which was generalized in Baltic, compare Slavic *vьlcěxъ < *-šu with Lithuanian vilkuose – though I don’t know what the story exactly is here, I don’t know much about Baltic) // Silmeth @talk 21:15, 20 January 2023 (UTC)
 * pinging y’all cause you have been more active in Proto-Slavic than I have, maybe you have better input (or maybe I’m arguing under a non-issue and *vьxъ for main Wiktionary lemma is OK). // Silmeth @talk 21:25, 20 January 2023 (UTC)

Reconstruction:Proto-Germanic/wirtiz
This is said to be a neuter i-stem, but such nouns have a lemma in, while is reserved for non-neuters. Either the gender or the inflection is wrong. —Rua (mew) 12:36, 17 April 2020 (UTC)


 * And none of the alleged Germanic descendants is in Wiktionary! The Finnic loan is present, though. RichardW57 (talk) 13:31, 17 April 2020 (UTC)
 * Kluge reconstructs a z-stem as the ancestor to the OHG and ON. -- 22:21, 17 April 2020 (UTC)
 * I think this can be deleted in favour of more recent reconstructions. Just let me make sure we don't lose and descendants or break any links first. Leasnam (talk) 18:19, 18 April 2022 (UTC)

= July 2020 =

Reconstruction:Proto-Indo-European/bʰeh₂ǵos, Reconstruction:Proto-Slavic/bazъ
“Transformed” Pokorny stuff, ominously sourced by the Leiden school.
 * The beech isn’t in the range (!) of the Proto-Indo-European homeland.
 * The Slavic page is properly *bъzъ. The Serbo-Croatian descendant does not count for because Proto-Slavic *ъ generally gives a in Serbo-Croatian, the Russian and Ukrainian given are obscure dialectal forms, as well as the Bulgarian, which are unstressed while Bulgarian has suffered  and 🇨🇬 and  are very close; ominously one gives an Old Church Slavonic only for . The current Slovak form which I added, apart from being anomalous as a feminine, can also be from ъ, this can be seen  →  and the variation for . Against the evidence from all Slavic languages one cannot posit such a byform, more easily   is an etymologist’s fabrication to shoehorn all into an Indo-European-etymology. Which does not work anyhow because the Slavic words mean elder, not beech. These plants are not confusable.
 * The page is in ESSJa, ’tis true, but apart from the entry’s age as I have noticed often, they do not take a stand for every entry in their Proto-Slavic dictionary, which is but hypothetical. They apparently create some index files, here motivated by Pokorny, and look what they can find to support the form, then they publish all anyway if the result is negative. See the RFD already filed for the adjective in WT:RFDO, Useigor did not understand this and created bare objectionable entries this way.
 * 🇨🇬 means “book” but there is yet no proof the Germanic peoples used beechbark writing or anyone else as opposed to . And how can, different paradigms, be from the same Proto-Indo-European form? There is something unaccounted. The existence of that word also conflicts with giving , as the consonant outcome differs and because “the slot is filled” i.e. the alleged word for a tree is too similar to a word for the arm for both having existed.
 * Also the sound correspondence is not necessariy, I see that PIE eh₂ can give (with and without acute) ō, there is →,  → , but it depends on ablaut and more typical is the result ā or a, e.g.  → ,  → . Also considering the actual sound value in PIE [a] given at . This additionally decreases the likelihood of PIE origin.
 * 🇨🇬 is very tentative and random as always.
 * 🇨🇬 has been thrown out of the equation meticulously after the creation of the PIE, much reasoned at its entry.
 * Where is the Gaulish word attested? Probably fishy if it is claimed to be only Gaulish but not retained in other Celtic languages. What do the other Celtic languages have? With such things I am accustomed to have the suspicion that it is somehow conjectured from unfathomable placenames.
 * The Latin word may be an early borrowing from, like even ; as  and the beech is found in Italy only at some places and not right at Rome, only somewhat outwards. Whereas the beech is very frequent in the  area. In Latin likely a foreign word. I say this also from general impressions about substratum origins of Latin plant names, after having dealt with many Latin plant names and their origins.
 * This is well a loanword after Proto-Indo-European when Germans, /Romans and Greeks took new settlements judging by analogy. Remarkably the Slavic words and  are Germanic borrowings for some reason, apparently because the Slavs settled right at the Northeast of the distribution of the beech, of course also 🇨🇬 is loaned. So if not even the Slavs before expansion (3rd century CE) had a word for the beech, the Proto-Indo-Europeans hadn’t either; if the Slavs borrowed this word, the Germans and Greeks and Romans did it likewise earlier. The correct etymologies for the German and Greek words are “borrowed from an unknown source common to [Greek|Proto-Germanic]”. Fay Freak (talk) 15:37, 27 July 2020 (UTC)
 * agreed. This has always been a dubious reconstruction, made worse by shoehorning more descendants to it, and further comical by reconstructing it with *-eh₂-. Also see 86 -- 18:28, 30 July 2020 (UTC)
 * I have read it. So I have found it is actually a debunked canard since half a century ago, called beech argument. It might have went past the Soviet theorists. In it is expounded how the Gaulish name is derived by reconstruction, from placenames. It is to be added that the literature finds it problematic that the Greek word means an oak and not a beech. Fay Freak (talk) 20:44, 31 July 2020 (UTC)
 * I disagree. Where else would "book" come from
 * It says where it possibly comes from. Often explained as in Germanic from the word for beech, which last is a word borrowed from somewhere. I do not need to have an explanation for or know everything to disprove an etymology. Your argument is none. Otherwise aliens built the pyramids because “how else”. Fay Freak (talk) 23:32, 16 December 2020 (UTC)
 * Several invalid arguments here.
 * 1) “Transformed” Pokorny stuff, ominously sourced by the Leiden school." -- This is frankly just rude, not reasoned. Kroonen's dictionary is extremely respectable (even if one disagrees with it) and tested by peer-review, unlike this nomination for deletion. The reconstruction is cited by philologists in other "schools" than Leiden. Check out e.g. Ringe (Pennsylvania/Oxford). Wiktionary should be reflecting the general scholarly consensus, not novel, non-peer-reviewed proposals of independent-minded contributors.
 * 2) "beech not in homeland" -- irrelevant, as many words change in meaning over time, and with different environments in different geographical locations
 * 3) "yet no proof the Germanic peoples used beechbark writing". No, but Germanic peoples' first contact with "books" would probably be Roman writing-tablets, which were often made of beech wood. (See e.g. https://en.wikipedia.org/wiki/Vindolanda_tablets)
 * 4) "Latin word may be an early borrowing from Northwest Greek φᾱγός" -- Even if that is true, that still leaves Germanic (NW Indo-European) and Greek (S Central Indo-European) as cognates, which is generally regarded as sufficient to support the hypothesis that it is ancestral to both of those branches. But what on earth is "an unknown source common to [Greek|Proto-Germanic]" other than the common ancestor of the European side of PIE?
 * Signed: an anonymous academic peer-reviewer, who is a tenured Professor in a Philology Faculty (no, not Leiden). But the decision about whether to delete the page or not should be taken on the merits of the arguments alone. 82.132.228.243 11:19, 9 December 2021 (UTC)
 * Furthermore, the argument that “the slot is filled” i.e. the alleged word for a tree is too similar to a word for the arm for both having existed" is unreasonable, because homophony and doublets are actually perfectly common phenomena cross-linguistically. 94.196.220.242 14:57, 9 December 2021 (UTC)
 * You are rude and not reasoned, okay?
 * Bare editors are stricter about meaning differences than I am, e.g. I presume mildly not amused about this lumping beeches and elders and what not. To reconstruct we need to pin down a more or less vague meaning, which these equations do not meet, and formally it is a scarecrow greater than many reconstruction pages we decided to delete, just not on first glance, but after a review of the possibilities (possibilities are hard to assess for the casual observer by magnitude, hence all those antivaxxers; our judgement needs specific training for the assessment of specific possibilities, so even if you are a professor in one area you may stay without insight in a closely related area and ignore its possibilities even though these should influence the decision).
 * We all have read very odd things that are peer-reviewed, as some academics have built parallel universes to make a living. And the beech argument is one of it, not a respected theory any more (if I understood respect correctly as being more than being constantly repeated out of courtesy and the university habit of citing everything that is available) but a fringe view, certainly not adding, in the traditional meaning of, to our knowledge, but you are right that the decision about whether to delete the page or not should be taken on the merits of the arguments alone, since you yourself know your colleagues enough to distrust them.
 * It is symptomatic though that a tenured professor in philology fails to consider the presence of unknown language groups before Indo-European; that’s how one regularly comes up with reconstructions that should never be made: if all you have is a hammer, everything looks like a nail, if you only know Indo-European, “everything” is from Indo-European. It is partially not your fault, given that the other language groups of Europe are scatteredly or not at all attested, and partially it is your fault in so far as you never deal with language areas where well-documented languages have stood in manifest contact (the usual case with the bulk of scholarship about Greek and Germanic and Germanic languages, you know those who live in a Germanic country and study Germanic do kind of a cheap thing, and if a European classical language is added this only opens the horizon a little).
 * Different semantic fields of a language have various propensities to contain borrowed terms, plant-names are especially notorious in it. If I was only an Indo-Europeanist I would hardly know but in the Semitic, Turkic, Iranian languages half of them (I exaggerate but little) are certainly loans from each other or other less-known or completely unknown language groups; e.g. another Wiktionary professor saw that is surely borrowed and likely Hurrian but for the Greek  the mainstreamers fail to do anything but speculate Indo-European (native or “pre-Greek”) origins although with the Near East data they should have classified it as wanderwort. However about every second time I open a Greek etymology Beekes claims a Greek word to be pre-Greek: while the intrinsic value of this label is close to zero due to the multifariousness of the frequent Pre-Greek claim, the idea of unknown sources of borrowings has been defended very well and is being concretized while we edit Wiktionary etymologies more and more.
 * So we pray you, Professor, to register and solve words occasionally, and especially if to disprove people as rude and uneducated as me. The more you learn of this dictionary business the more you realize that there is a thin line between daring comparisons—adding to our knowledge by maverickism—and academic dishonesty. And IPs are evil. After all you already do not rely on the majority of the comparisons on which that PIE “reconstruction” is made, if Germanic and Greek are left: In our experience the farer away a reconstructed historical language the more descendants one needs, and for PIE two are regularly (without very good reasons) not enough, while for Proto-Slavic not rarely one is enough—if a term must have been formed in Proto-Slavic, e.g., while for PIE there are too many millennia in between of what could have happened and we do not know that *bʰeh₂ǵos must have been internally derived in PIE (usually between Arabic, Iranian and Turkic and often inside their language groups themselves we know where a term was formed and hence whence borrowed by our understanding the internal morphologies of the languages: all things you do not know for this term).
 * This is all to say that, in comparison to more certain etymologies, here you know absolutely nothing. But you should somehow be confident about a reconstruction rather than many mismatches and coincidences and alternative scenarios (and I have engaged in shaky reconstructions out of excitement, but this is so shaky that it crumbles apart the more you think about it—if it were better I would come to maintain this PIE term: obviously I come correct in thinking about reconstructions, you will hardly deny this experience in having a consistent and carefully weighed approach about reconstruction entries). Fay Freak (talk) 17:12, 9 December 2021 (UTC)
 * It is symptomatic though that a tenured professor in philology fails to consider the presence of unknown language groups before Indo-European; that’s how one regularly comes up with reconstructions that should never be made: if all you have is a hammer, everything looks like a nail, if you only know Indo-European, “everything” is from Indo-European. It is partially not your fault, given that the other language groups of Europe are scatteredly or not at all attested, and partially it is your fault in so far as you never deal with language areas where well-documented languages have stood in manifest contact (the usual case with the bulk of scholarship about Greek and Germanic and Germanic languages, you know those who live in a Germanic country and study Germanic do kind of a cheap thing, and if a European classical language is added this only opens the horizon a little).
 * Different semantic fields of a language have various propensities to contain borrowed terms, plant-names are especially notorious in it. If I was only an Indo-Europeanist I would hardly know but in the Semitic, Turkic, Iranian languages half of them (I exaggerate but little) are certainly loans from each other or other less-known or completely unknown language groups; e.g. another Wiktionary professor saw that is surely borrowed and likely Hurrian but for the Greek  the mainstreamers fail to do anything but speculate Indo-European (native or “pre-Greek”) origins although with the Near East data they should have classified it as wanderwort. However about every second time I open a Greek etymology Beekes claims a Greek word to be pre-Greek: while the intrinsic value of this label is close to zero due to the multifariousness of the frequent Pre-Greek claim, the idea of unknown sources of borrowings has been defended very well and is being concretized while we edit Wiktionary etymologies more and more.
 * So we pray you, Professor, to register and solve words occasionally, and especially if to disprove people as rude and uneducated as me. The more you learn of this dictionary business the more you realize that there is a thin line between daring comparisons—adding to our knowledge by maverickism—and academic dishonesty. And IPs are evil. After all you already do not rely on the majority of the comparisons on which that PIE “reconstruction” is made, if Germanic and Greek are left: In our experience the farer away a reconstructed historical language the more descendants one needs, and for PIE two are regularly (without very good reasons) not enough, while for Proto-Slavic not rarely one is enough—if a term must have been formed in Proto-Slavic, e.g., while for PIE there are too many millennia in between of what could have happened and we do not know that *bʰeh₂ǵos must have been internally derived in PIE (usually between Arabic, Iranian and Turkic and often inside their language groups themselves we know where a term was formed and hence whence borrowed by our understanding the internal morphologies of the languages: all things you do not know for this term).
 * This is all to say that, in comparison to more certain etymologies, here you know absolutely nothing. But you should somehow be confident about a reconstruction rather than many mismatches and coincidences and alternative scenarios (and I have engaged in shaky reconstructions out of excitement, but this is so shaky that it crumbles apart the more you think about it—if it were better I would come to maintain this PIE term: obviously I come correct in thinking about reconstructions, you will hardly deny this experience in having a consistent and carefully weighed approach about reconstruction entries). Fay Freak (talk) 17:12, 9 December 2021 (UTC)


 * According to Буковый аргумент, the modern range of w:Fagus orientalis falls inside the Proto-Indo-European homeland. And its range several millenia ago might have been broader than today. --Ghirlandajo (talk) 18:38, 26 June 2022 (UTC)
 * I'm surprised there haven't been any updates to this request for deletion thread since last year. Is anyone willing to contribute journal articles discussing the presence of this constructed word for beech outside of the late Northwest Proto-Indo-European dialect continuum? In the meantime, I will provide citations from the EIEC, Calvert Watkins: "The American Heritage Dictionary of Indo-European Roots", and Beekes Etymological Dictionary of Greek. Calvert Watkins was not part of the Leiden school as far as I'm aware; nor is Douglas Q. Adams, who was one of the main editors of EIEC.Rigognos Molinarios (talk) 12:23, 11 February 2023 (UTC)


 * This entire question is a bit of a mess, especially as some of the associated reconstructions are dubious, but I believe that the original posting user has made several critical errors, although the associated page should be updated.
 * The range of the beech tree not being within the hypothetical homeland of the Proto-Indo-Europeans, as mentioned earler, is irrelevant to whether the word occurs in PIE. Since the descendent languages cannot seem to agree to what tree it refers, which is not that unusual for tree or animal words that are not highly general, it could also originally have referred to a tree other than a beech, but seemingly one which is usable as a food source (per the different meanings for hypothesised descendents being 'elder', 'beech', and 'oak').
 * The Proto-Slavic word being properly *bъzъ does nothing to detract from the reconstruction, as Proto-Slavic *bъ- as a cluster regularly comes from PIE *bʰe-. The real question is whether *bъzъ can result from bʰeh₂ǵ-. *pazъ from *peh₂ǵ- suggests no, however, *bogъ being derived from *bʰeh₂g- suggests that the vowel a derivative of the word *bʰeh₂ǵos would reflect in Proto-Slavic is uncertain (admittedly the following consonant here is velar and not a palato-velar, but point is still relevant), so this is still a possible derivative. If another aspirate-eh₂-palatovelar cluster derivative could be found in Proto-Slavic anywhere, that would be helpful.
 * The Germanic reconstruction is an absolute ludicrous mess, agreed, and the attempted connection between 'beech' and 'book' is completely silly, plus one would expect a Proto-Germanic *bōkaz to come from *bʰeh₂ǵos and not *bōkō. However, the Proto-Germanic word for beech is apparently more appropriately reconstructible to *bōkijā, or more credibly neuter *bōkiją which is easily derivative of *bʰeh₂ǵos. Also *bʰeh₂- appears to regularly become *bō- in Proto-Germanic from a quick check of Proto-Germanic lemmas, so the additional attempted phonology points from the original poster are also spurious.
 * Agreed, the Gaulish word is not credible as it is derived purely from onomastics. However the Albanian word is only listed as possibly related, as it may possibly be, and it means 'oak' identically to Greek φᾱγός, so it is fair as a tentative or hypothetical relation given the difficulties inherent in Albanian reconstructions. Dismissing it out-of-hand is childish.
 * If the Latin word fāgus were an EARLY Greek loan from the more conservative Doric dialects, one would expect pāgus or phāgus in Latin and NOT fāgus as the Greek aspirate did NOT sound like a fricative in Latin even into the reign of Augustus, therefore it is near-impossible that the Latin word is a loan from Greek, and it is more reasonable to conclude genuine common descent. The beech not being immediately in the area of Rome but in the mountains a few dozen miles away is also completely spurious as an argument as this implies Latin speakers never left home, nor had need of a word for things not immediately in their environment but which were plentiful close by. Rome is also not on the sea coast, yet Latin also magically has more than one indigenous word to describe seas and seaborne fish! Plus, the Greek φᾱγός means 'oak', so magically the Latins would also have to inherit a Greek loanword with a completely different meaning to the Greek word for no discernable reason.


 * So to conclude, a credible word reconstruction in at least Latin and Greek, which is more than enough to posit a PIE origination, and additionally with plausible Germanic and Slavic derivatives, although the Slavic is on shaky ground. I'm inclined towards allowing the page for *bʰeh₂ǵos to stand, but the *bazъ page is spurious and should be deleted, and the Proto-Germanic 'beech' and 'book' derivations need serious attention by someone more capable than me at Proto-Germanic research, as they are in a completely silly state currently. I somewhat doubt any Indo-Iranian derivatives will be found given that beech and oak trees are not exactly plentiful on the Iranian plateau or in northern India, but otherwise at least four major families are covered. Someone could additionally look into Armenian as its word for beech, hačari, seems at face value to be similar-looking, but I am also completely unskilled at Armenian phonology and it would have to involve some manner of special cases as 'h' normally results from PIE *p in Armenian as I understand. This is still less sketchy than the still-existent page for PIE *h₂éyos, which is almost certainly an accidental fabrication, but nonetheless mistaken. Andecombogios (talk) 04:47, 22 February 2023 (UTC)
 * Mallory and Adams also accept the existence of a proto-indo-european term for beech, in their Encyclopedia of indo-european culture, (its on pages 58-60), so this isn't even limited to leiden school Ioe bidome (talk) 21:02, 7 August 2023 (UTC)


 * I am of two minds. I do not believe in the beechstave etymology and have to doubt the reconstruction. My opinion has not changed since I read this thread a year ago, noting that Fay Freak and Victar make compelling arguments. It is bogus, literally. Nevertheless, usually reliable sources are claimed to support this term.
 * 1. Kroonen is expressly "skipping well-established Proto-Germanic lexems" and is focused on "more rigorous" implementations of sound laws and regional grouping (, Preface). This does of course not take away from its value. Note that he discusses *bʰh₂ǵʰ- > on a different page, admittedly controversial, and he notes that /ǵʰ/ and /ǵ g/ do not match when he mentions  under  in order to reject "any direct connection". He does not mark his reconstruction is IE but EUR.
 * 2. Ringe 2006 and Ringe & Taylor 2014 do not mention it (though the latter mentions "books" shy of a PIE root).
 * 3. Adams & Mallory 2006 note the North-Western or Central distribution, noting the irregularity of Ru. buz.
 * More citations in the entry:
 * 4. De Vaan agrees with Kroonen. He notes especially that Albanian and Slav. *buzt "elder" are unrelated.
 * 5. Trubachev has pre-laryngealist *bhuǵ-, *bhāǵ-, *bha(u)ǵ-. Kroonen discusses the Germanic evidence of rounding away as internal borrowings after vowel changes.

Remarkably, Derksen does not have it. Other sources follow IEW.
 * 6. Fortson 2004 includes Russian, remarks on the distribution (as does de Vaan in more detail). It is a rare example of a feminine o-stem.
 * 7. with *ah₂ (?), without Slavic. Laryngeal and palatal are not secure. *bʰāg- may be lengthened grade of **bʰag-'', viz. ? Ha!
 * 8. Kölligan (Handbook of Comparative and Historical Indo-European Linguistics, page 2236-2237) sets side by sidy *bheh2go-, *bherHg̑-, *perku̯us ‘oak’, *peu̯k̑-. Gaitzsch & Tischler (op.cit., page 86) plainly claim the word meant "beech", that the beech-argument which places the urheimat in the west is disproved (not cited though), and they include the Slavic word and a Kurdish būz "elm" too.
 * Gaulish is sometimes included based on toponyms but not discussed much. In sum, there is wide agreement in the literature. The only disagreement is about the random weights in the network or the nodes in the date tree, depending on how primitive you want your models to be. DurdyWendy (talk) 19:49, 12 March 2024 (UTC)
 * For Kurdish, you see its more pertinent connections in Chyet s.v. bizî and (nobody explained the Arabic barring my guess and Arabic etymological dictionaries don’t exist). Henning’s 1963 paper on The Kurdish Elm also touching the beech-argument is in.
 * This is not what agreement is. Again I stress that in this science one cites previous attempts without taking position; if one expands upon it the way you have done, it will conclude with a remark about a “highly uncertain reconstruction” or something in that direction; you see that academics write too polite to speak of bogus, but on the internet there is a higher necessity to reproach people that they have lost the plot. Fay Freak (talk) 21:31, 12 March 2024 (UTC)
 * Incorrect on the second charge, but not completely wrong. Scholarly debates are full of polemics, tending to be specific and well versed in the language. A notable example of polemics is Pereltsvaig & Lewis, "The Indo-European Controversy", who discuss the Buchenargument: "German philologists in the first half of the nineteenth century were preoccupied with finding the PIE Urheimat through tree names, favoring – unsurprisingly – a more northern homeland." (page 185). I lol'd. They consternate that only five roots which appear in Anatolian can be reconstructed for PIE and only one of them appears in Tocharian (viz. *dóru-, but here as in Slavic the meaning is "wood"), and they conclude, "Consequently, the majority of the reconstructed tree names can only shed light on the homeland of one of the main branches descending from PIE, not of PIE itself." (page 191). They agree that this is inconclusive (page 192), as do Gaitzsch & Tischler in the end. If that's what you mean, I stand corrected. Unfortunately it only proves the point: the sources agree to the extent that we usually concern ourselves with (WT:CFI).
 * I want to stress that the argument of a Latin loan is a possibility because Doric φᾱγός is less well attested and obviously a better match than φηγός. DurdyWendy (talk) 15:35, 13 March 2024 (UTC)

This case is a rather of the problems of Wiktionary at the moment. You people really have to make up your minds on whether you want this wiki to be based on verifiability and reflection of the current scholarly consensus, like Wikipedia, or whether you want it to be a place where a couple of anonymous editors get to play at being scholars themselves and the whole world is treated to the fruits of their (non-certified) wisdom and competence, filtered only through the principle of 'truth by consensus' (Stephen Colbert's notion of 'wikiality' may not describe current Wikipedia fairly, but it may still turn out to be quite fitting for Wiktionary). It's perfectly clear that multiple mainstream scholarly sources agree with the entry's reconstruction of the PIE term that 'beech' originates from (The Fay Freak's claim that they just mention it is wrong, as obvious for anyone who checks them), it's just that The Fay Freak and Victar don't agree with it, and the real question is whether Wiktionary is supposed to reflect the position of actual scholars or that of The Fay Freak and Victar. (AFAIC, most of the objections of the professor anon to The Fay Freak were compelling, and The Fay Freak's long-winded rant in response was both unacceptably aggressive and poor in actual counterarguments.) Ditto for the 'beechstaves' - this is the standard view, but DurdyWendy doesn't like it for whatever reason (this is actually a completely separate theory that The Fay Freak explicitly didn't dispute, but she apparently sees getting rid of the whole PIE reconstruction as a way to get at the etymology of 'book' that is her own personal bête noire). Again, the question is if Wiktionary is meant to be the place where DurdyWendy tells the world what she feels is true or whether it is the place that informs them about what the current consensus of actual credentialed scholars is. As far as I'm concerned, if you want to overturn the scholarly consensus 'with facts and logic', you should go publish your views in a peer-reviewed article and if you manage to convince actual scholars, then cite it on Wiktionary (not as the only truth, mind you, but just as one position). Instead of satisfying your ambitions to be a scholar here, be a real one. Just the fact that someone has a lot of spare time on their hands that allows them to spend time on Wiktionary doesn't give them the right to used it as a tool to push their pet theories and their version of the truth on the entire world while bypassing the standard 'screening' processes for such a thing. And in case some of the editors here do happen to be actual academics IRL, this doesn't change anything - they may have worked on something, but they clearly either haven't worked on the specific issue discussed or their contributions pertaining to it haven't managed to convince most other scholars and to overturn the prevalent views in the field - otherwise they would have been able to rely on published sources instead of embarking on imitations of scholarly debates here. Being that rare academic who happens to have a lot of time to spend on Wiktionary doesn't make your view on a specific subject more authoritative than that of the majority of your peers - it may still be highly deviant on that specific subject, and the question is what the predominant view among experts on a given issue is.--62.73.69.121 18:58, 14 March 2024 (UTC)
 * No idea what you are trying to say. See for yourself:
 * > -- Pierce, Marc. “The Book and the Beech Tree Revisited” Historische Sprachforschung / Historical Linguistics 119 (2006): 273–82.
 * Last but not least the Fay the Freak the luminary himself stands unchallenged --.
 * So much for consence. Anyhow, it hardly even matters for the topic at hand. IncrediblyBendy (talk) 20:30, 18 March 2024 (UTC)
 * Keep, clearly. Sources and arguments that question its existence may be noted on the entry under reconstruction notes or some such header. — Mnemosientje (t · c) 11:43, 19 March 2024 (UTC)
 * > -- Pierce, Marc. “The Book and the Beech Tree Revisited” Historische Sprachforschung / Historical Linguistics 119 (2006): 273–82.
 * Last but not least the Fay the Freak the luminary himself stands unchallenged --.
 * So much for consence. Anyhow, it hardly even matters for the topic at hand. IncrediblyBendy (talk) 20:30, 18 March 2024 (UTC)
 * Keep, clearly. Sources and arguments that question its existence may be noted on the entry under reconstruction notes or some such header. — Mnemosientje (t · c) 11:43, 19 March 2024 (UTC)


 * Keep Reconstruction:Proto-Slavic/bazъ at least since can't come from . I'm pretty sure strong ъ never becomes a in Lower Sorbian (though strong ь sometimes does). —Mahāgaja · talk 12:25, 24 March 2024 (UTC)
 * It can, arbitrarily—like the leading East-Slavic forms have, which can’t come from . You know well that language has arbitrariness as a principle. Reconstructing from the descendants, without prepossessed resolution to gather support for an even more remote reconstruction, this Proto-Slavic reconstruction would not have been created organically. Rather there is room for descendants of one form having some variation. Per Occam’s razor there is little reason to assume an alternative form in Proto-Slavic. Fay Freak (talk) 13:28, 24 March 2024 (UTC)
 * Keep. Until a consensus can be reached within and between all relevant fields (esp. Slavic, Germanic, Latin). Distribution issues have been mentioned. Thanks to the Fagus-Quercus sp.-Sambucus incongruity, an interdisciplinary approach is unlikely to solve the issue unless the linguists can agree on Sambucus as a secondary shift or an independent loan. Even if, as is likely to happen, a more detailed pollen analysis than those of Giesecke et al. 2007 and Magri et al. 2006 concludes Fagus was not present in the relevant region and time, or an admixture dating analysis is performed that concludes the ssp. orientalis introgression in ssp. sylvatica that formed Fagus × taurica did not extend even close enough to the time of formation and diversification of PIE, there is still the Quercus sp.>Fagus possibility, not to mention the biogeographical compatibility of Sambucus as the original meaning in PIE (should it be accepted as an inherited cognate). The phonetic and semantic issues associated with my own opinion notwithstanding, the lack of agreement between linguists justifies the entry's continued inclusion for the time being. Иованъ (talk) 15:57, 18 May 2024 (UTC)
 * Relevant for the history of the Germanic cognate is the recent confirmation of a longstanding hypothesis that the spread of Fagus sylvatica to Scandinavia was anthropogenic, with absolutely no presence in Northern Denmark until about 2500 kya, remaining an extreme rarity for several centuries thereafter. This matches the 3.2-2 kyr cal date range for the proliferation of F. sylvatica in Scandinavia given by Magri et al. 2006, including samples as far as southernmost tip of Sweden just before the beginning of that period; the same study however already detected the species in the macrofossil record of Zealand in the 5.7-4.5 kyr cal period ("predating pollen records by 1000 years"). So at the very least, in Germanic can hardly descend from the component that spread with "East Scandinavian Corded Ware", decreasing the robustness of a semantic argument for "Fagus" being the original PIE meaning as it increases the plausibility of a "southern" origin for the term within Germanic and therefore of it being a loan acquired by a descendant of PIE rather than PIE itself.
 * And since the current evidence favours there having been no Fagus on the steppe between the Western and Doric attestations at the time PIE spread but there is a closer semantic relationship between the Western and Doric cognates than the contentious Slavic cognate, separate loans from one or more non-IE families ought to be the preferred explanation (or a semantic shift from an otherwise unattested PIE term, but this is less likely where phytonymy is concerned). Иованъ (talk) 15:57, 18 May 2024 (UTC)

Reconstruction:Proto-Indo-European/h₁óytos
Germanic is often cited at being an early Celtic borrowing. Regardless, given how it's disputed, a PIE entry isn't warranted. -- 16:39, 8 November 2020 (UTC)
 * Delete; I remember that one has argued not without reason that this is a Celtic borrowing, and the likelihood of a Germanic-Celtic isogloss in comparison to a borrowing heavily speaks against this reconstruction. Fay Freak (talk) 01:17, 24 December 2020 (UTC)


 * Merge to, which in any case ought to mention at least Celtic . A loan is a likely possibility, and these details would be better discussed somewhere else such as on the PG and PC pages (and the latter does not even exist yet!). Note though also Greek as another suggested cognate. --Tropylium (talk) 18:41, 3 February 2021 (UTC)
 * In view of Beekes' own preferred etymology from, merging to seems no less speculative. I suggest that we add the Greek as well as , which is another perfect match, and so move the page to , where the two etymologies can be elaborated. In my opinion Beekes is right that the Greek evidence strongly points to  as the root, and the semantics fit better (“*that which is given” > “lot, fate”, “punishment”, “oath”; it is not even clear what a nominal in *-tó- from the semantically intransitive “to go” should mean). Perhaps other authors avoid connecting the Germanic and Celtic with it because this root has not otherwise survived in those branches or is somewhat obscure. — 69.121.86.13 16:11, 30 January 2023 (UTC)


 * Delete. Two words don't warrant a PIE reconstruction. Highly speculative. Ghirlandajo (talk) 21:17, 7 June 2022 (UTC)
 * Delete as probably nonexistent in PIE. Even Kroonen said that "it is unlikely that the formation goes back to Proto-Indo-European, only to surface in two neighboring branches at the far end of the IE-speaking area. It is more probable that the word somehow arose in a shared cultural zone with similar legal traditions." I have just merged whatever salvageable from this page to the root page, so now we can delete this. — Ceso femmuin mbolgaig mbung, mellohi! (投稿) 01:36, 19 September 2022 (UTC)
 * Delete . I have argued for high likelihood of fictivity below regarding a Proto-Semitic gloss “oath” at . Fay Freak (talk) 20:18, 22 May 2023 (UTC)
 * You already voted above. Chuck Entz (talk) 05:16, 23 May 2023 (UTC)
 * Keep: I do not understand why this request has not been closed after 3 years, 4 unanimous close votes and 1 suggestion to merge with the root. So, I vote contrarian to confirm the status quo. The reconstruction can be cited from reliable sources, who admit alternatives. The question of borrowing is undecidable according to (1998). PIE *oitos, as they put it, is formally acceptable for both; *h₁óytos is laryngealist cosmetics (that the online edition does usually add laryngeal notation when it is secure is not decisive). Adams & Mallory 2006 clearly state *h₁óytos and add Tocharian B  to the comparison. Sluis/Jørgensen/Kroonen 2023 discuss the Celto-Germanic isoglosses and do not improve on the status quo but confirm that *h₁óy-to- is a possibility, to conclude: "The Bell Beaker maritime network is not the only archaeological context for which the diffusion of linguistic features between these early IE groups can be hypothesized." (The Indo-European Puzzle Revisited 2023: 207). On the other hand, the evidence of Oscan eítuns is highly unreliable and that is just not good enough to reject Greek αἶνος, ἀναίνομαιor (Peters) and Hittite ḫai- (Puhvel) [apud EWAhd]. Adams and Mallory posit *h₂⸝₃ehₓ-, compare,  and note that "some would also include the Celtic [...] and Germanic [...] words for ‘oath’" (A&M 2006: 323). In sum, the entry is warranted but sufficient hedging will be needed. If you disagree entirely, you should be able to offer a better solution. I am looking at different options but this is not exactly an improvement, and so I cannot disagree. DurdyWendy (talk) 20:53, 8 March 2024 (UTC)

Reconstruction:Proto-Indo-European/dʰews-
Schwebeablaut isn't a root "variant", it's an environmental change. -- 00:31, 11 December 2020 (UTC)

Reconstruction:Proto-Indo-European/gl̥h₁éys
Not enough evidence, most likely did not exist; one used instead for all kinds of mice. The nominated word, somewhat varying by form in literature, is also claimed on the basis of some Iranian forms, which however have two much more likely explanations I mentioned at 🇨🇬 (an Iranian stem related to biting and stinging etc. (semantically, if it is not clear from an English horizon) and a Semitic borrowing from a widespread Semitic stem related to gnawing). Looking at the edit history has removed an alleged Indo-Iranian reconstruction allegedly descending from this Proto-Indo-European, likely without explanation because it is obviously baseless, before someone re-added the Sanskrit words  and  again, which apart from a new formation from the word for a hill may also be a borrowing like other words (from which we have  etc., “from a lost Vedic substrate language”). 🇨🇬 forms ɡoɖo ~ guɖu are not farther than the alleged Indo-European cognates. g + either r/l are too common consonants. Then, how has the paradigm given for the Indo-European page any relation to the paradigm of the Sanskrit word or the paradigm of the Latin word, which has the stem glīs- glīr-. means a mustelid, which is an animal not that similar or not all confusable or confoundable. I realize that  contains the word for mouse, but it does so because the weasel eats mice, but the weasel is no mouse by utmost historical phantasy, they had and have to be kept apart. So the glossing of the Indo-European page “mouse, dormouse; weasel” is impossible – no language can use one and the same word for mice and weasels. The 🇨🇬 is not similar at all and needs other explanations, which it has, mentioned in the linked article Studies in Thracian vocabulary I–VII. Fay Freak (talk) 23:03, 16 December 2020 (UTC)
 * Re the claim that a weasel is too different from a mouse, so the two have to be kept apart - the Norwegian word for a weasel is snømus, literally 'snow-mouse'. De Vaan, as cited in the Wiktionary entry, interprets mūstēla as an original diminutive *mūstr-elā 'a small mouse-like animal', not as 'mouse-eater'. Perhaps a taboo replacement (cf. the conflation of worm and snake, arguably an even more important difference)? --79.100.144.23 22:51, 10 July 2022 (UTC)
 * You're right that it might be a conflation of substrate words. But suppose it didn't exist in PIE―then why are the supposed cognates so similar? Sheer coincidence doesn't seem to be an option, so they would have come from related substrate/adstrate terms that had diverged semantically in their own distant past. It simply pushes back the question further into an unreconstructible era. Just because two animals don't seem "confusable" doesn't invalidate an etymological connection between words for them. Far stranger shifts have happened in far shorter time than the whole history of Indo-European. Also, the original term might have been an adjective or some other unspecific label that would have been applicable to various animals (cf., the supposed root of both and ; similarly semantically divergent substantizations are par for Indo-European linguistics). So your semantic arguments don't really hold up. If there were no formal problems with the phonetics, the cognate set would hardly be questionable. Likewise, some questionable items being formerly part of an etymology have no bearing on the present state of the etymology, nor do tentative, less-supported connections to other language families like Dravidian or Semitic, unless of course the evidence for those etymologies is stronger (in this case, it clearly isn't). The differing paradigms are an important question, but nothing unresolvable; most can be regarded as different ablaut variants of an i-stem, and the question of the exact form of the original noun (or other nominal) is at that point almost irrelevant to the validity of the etymology. Thus the only really worthwhile criticism here is on the phonetic grounds. ― 69.120.66.131 01:54, 16 January 2022 (UTC)
 * “Clearly” the evidence for Semitic and Munda origin of the Indo-Aryan terms are stronger—what does “supported” even mean? No, but coincidence is a stronger option, due to the ubiquity of the consonants in the game, and on the Arabic page I have offered a better derivation for the Iranian part. Anyway you admit that a noun is formally not reconstructible nor even any root of any defined meaning. The original might have been this, might have been that, or: there were multiple unrelated originals. You have a Greek and a Latin and Sanskrit word of bare different meaning. The Thracian is a meme speculation.
 * Now I also see the Greek and Latin derived from, on , underline how arbitrary the assumptions underlying the Indo-European reconstructions of these words are. Fay Freak (talk) 20:42, 8 April 2022 (UTC)
 * This reconstruction looks highly dubitable and is not supported by mainstream linguists. Delete. --Ghirlandajo (talk) 21:25, 7 June 2022 (UTC)
 * Delete: I removed the Proto-Indo-Iranian because the Sanskrit is a ghost word, only attested in dictionaries, and to quote Schrijver, "as it stands, the comparison [between and ] is neither formally nor semantically compelling." --  08:38, 23 May 2023 (UTC)
 * Delete -- 𝘗𝘶𝘭𝘪𝘮𝘢𝘪𝘺𝘪(𝘵𝘢𝘭𝘬) 07:01, 6 June 2023 (UTC)
 * RFD-deleted -- 𝘗𝘶𝘭𝘪𝘮𝘢𝘪𝘺𝘪(𝘵𝘢𝘭𝘬) 12:41, 20 May 2024 (UTC)

= January 2021 =

Reconstruction:Proto-Halmahera-Cenderawasih/wakaʀ
No descendants listed. Page creator is no longer active at Wiktionary. —Mahāgaja · talk 19:30, 30 January 2021 (UTC)
 * The reference gives three reflexes of for "SHWNG", which is another name for Halmahera-Cenderawasih:, Windesi "war" (Wikipedia treats Windesi as a dialect of the Wamesa language with the language code wad, while we treat wad as the language code for Wandamen, which Wikipedia says is the name of another dialect of Wamesa) and . Of course, we don't have entries for any of those and Blust is known to have regularized orthographies elsewhere in the same work to make comparison easier, so I wouldn't use it as a source for the terms themselves.
 * I would note that the Proto-South Halmahera-West New Guinea index lists wakaR, and a suffixed form *wakaR-i (with four more reflexes). Of course, it has similar indexes for proto-languages that probably don't exist, such as Proto-Western Malayo-Polynesian, so it would be nice to have some source to back this up. Chuck Entz (talk) 23:40, 30 January 2021 (UTC)

= February 2021 =

Reconstruction:Latin/consutura
Appears to actually be attested, though in New Latin, not Vulgar Latin. Hazarasp (parlement · werkis) 02:54, 21 February 2021 (UTC)
 * Indeed:, , . Should we interpret the New Latin use as "inherited" from Vulgar Latin? Otherwise, there are basically two homographic lemmas: a non-attested Vulgar Latin one, reconstructed from its descendants, and a re-invented one in New Latin. --Lambiam 10:13, 22 February 2021 (UTC)
 * The word isn't attested before the 14th century or so (DMLBS). We can't very well say that the Medieval Romance forms derive from New Latin, where the word is in fact reborrowed/calqued from Romance. Therefore these are separate, and the Proto-Romance (Vulgar Latin) entry has to stay. Brutal Russian (talk) 23:57, 3 April 2021 (UTC)


 * Hmm, I'm inclined to disagree; I would expect just one entry, which could explain that the word is inferred from Romance terms to have existed in Vulgar Latin but is not attested until later recoined in New Latin. By comparison, AFAIK we don't and shouldn't have separate reconstruction entries for attested terms in other languages (say, English or Ojibwe) where they can be inferred to have existed at some period before their actual attestation; we don't even have separate etymology sections for English words which existed in early modern English (Latvian, etc) and were later independently recoined in modern times. (OTOH, we still have a redundant reconstruction entry on lausa.) What do we do in comparable situations in Chinese or Hebrew? AFAICT we don't have any reconstructed Hebrew or Chinese entries, which suggests we may be handling "late (re)attested" words in those languages in mainspace only, as I would expect. - -sche (discuss) 06:58, 21 June 2021 (UTC)


 * Just noting that this has been moved to Reconstruction:Latin/cosutura (without n) and changed to derive from *cōsō, an unattested variant of cōnsuō. On one hand, this could all be handled in the mainspace by just mentioning the hypothetical n-less forms in etymology sections without spinning off whole entries, and I think such things are handled that way in situations were an n-form is attested earlier than an n-less variant; OTOH, it's now at least less stupid than having both *consutura and consutura... - -sche (discuss) 03:31, 22 March 2022 (UTC)

= March 2021 =

Reconstruction:Proto-Indo-European/ǵeh₂r-
Ditto. -- 22:22, 31 March 2021 (UTC)
 * Should be saved, but moved to *ǵar- —caoimhinoc (talk) 19:36, 18 August 2021 (UTC)

No consensus. No further action. —Caoimhin ceallach (talk) 15:29, 1 June 2024 (UTC)
 * Improper closing of an RDF. -- Sokkjō 00:52, 3 June 2024 (UTC)
 * Why is it an improper closing of an RFD? "A request can be closed once a month has passed after the nomination was posted." It's been years and there's no consensus. Same with Requests_for_deletion/Reconstruction. —Caoimhin ceallach (talk) 10:31, 3 June 2024 (UTC)

= April 2021 =

Reconstruction:Proto-Germanic/aganą
The conjugation table does not agree with the title, however I am unsure what it should be changed to. Pinging Mårtensås (talk) 15:17, 23 April 2021 (UTC)


 * Since the only direct descendant is, I suspect Reconstruction:Proto-Germanic/aganą should be moved to Reconstruction:Proto-Germanic/ōganą. —Mahāgaja · talk 15:41, 23 April 2021 (UTC)
 * And looking through the page history, I see it was at Reconstruction:Proto-Germanic/ōganą until Rua moved it to aganą in January 2019. —Mahāgaja · talk 15:47, 23 April 2021 (UTC)
 * The Proto-Germanic page title is fine. The only thing wrong with the conjugation is that the present singulars should be *ōg, *ōht and *ōg, respectively. The infinitive, not being inflected for number or person, is properly *aganą which follows the stem of the non-singular present, just like the present participle *agandz does, as we see in the fossilized Gothic . The Gothic form of the verb as *ōgan, with *ōg spread throughout the paradigm, including the secondary present participle, shows high degrees of regularization as is pretty typical for Gothic, which supports the theory that it was a superstratum language with many non-native speakers.


 * Likewise for aorist-presents, the infinitive stem follows the non-singular present stem of the verb, as seen in, although here too the present singulars should be *wīhō, *wīhizi, *wīhidi respectively; this is proved by the fact that otherwise -h- would have been unrecoverable to later Germanic speakers, and Gothic could not have regularized the -h- throughout... Which again in Gothic, we see regularization as the singular stem of the verb *wīh-, is spread to the other present formations: instead of original *wigam, and again in a derived form of the present participle, the original short vowel is kept, but the velar is regularized to -h- instead of original -g-:  instead of *andwigandō. Burgundaz (talk) 22:05, 18 May 2021 (UTC)


 * Seconding that the title agrees perfectly well with the IPA and conjugation table, even if perhaps the vocalisation of the present singulars is incorrect as Burgundaz says. Tristanjlroberts (talk) 23:48, 7 February 2022 (UTC)
 * Keep at current entry title. Burgundaz is correct. — Mnemosientje (t · c) 12:37, 19 March 2024 (UTC)

= May 2021 =

Reconstruction:Proto-West Germanic/krūci
This entry invents a completely new phoneme, that I have not seen in any sources, in order to reconstruct an entry. I know that Wiktionary gives some leeway for its own research, but a whole phoneme goes too far IMO. This is definitely something that needs to be sourced. —Rua (mew) 10:25, 6 May 2021 (UTC)
 * The Old Saxon, Old High German, and Old Dutch forms could maybe point *krūtsi, if I'm remembering the relevant historical developments correctly. The Frisian form is apparently a borrowing from Low German; if it was native, we'd expect *crēce. The OE form should maybe be excluded; unlike the other Gmc. forms; the consonantism may point to a loan from a dialect where Latin before front vowels gave, not . If  is removed and  is relegated to borrowing status, a case could be made for the page to be kept as . This would be a sensible adaptation of pre-Old French  into the late common WGmc. phonological and morphological framework. Hazarasp (parlement · werkis) 11:53, 6 May 2021 (UTC
 * Could it be that Proto-Germanic and Proto-West Germanic had phonemes only occurring in loanwords? Like the voiced postalveolar fricative occurs in German only in loanwords but in most familiar items like and . No doubt either the Proto-West Germans were able to pronounce . You have not seen it because of restricted use then, and by reason that Indo-Europeanists like to deal with inherited terminology rather than to sully themselves with language contact. Fay Freak (talk) 12:53, 6 May 2021 (UTC)
 * However, on second thought, the logic I employed further up is somewhat shaky; Dutch/Low German -s- isn't necessarily reconcilable with Low/High German -ts-. Additionally, the pre-OF form would be ; the note at about the vocalism in  (> later ) pointing to a late date of borrowing is spot-on. If the borrowing was late, it's not necessary to posit a common WGmc. source; separate borrowings in each WGmc. language could've easily resulted in similar phonological forms. In short, the WGmc. "cross" words seem to be separate, but interlinked borrowings from after the common WGmc. period (though this isn't watertight). However, the idea that some of the Germanic forms are borrowings from others is worth considering (Old High German → Old Saxon?). Hazarasp (parlement · werkis) 12:59, 6 May 2021 (UTC)
 * WG would have most certainly come into contact with Latin speakers that began to palatalize velars. The question in my mind isn't *if*, it's *how* we deal with these words. Using *c as a stand-in for an otherwise foreign sound is probably the easiest way to go about it. Compare also, which which has some interesting Old English variants. To quote what I wrote on the talk page, "1) even if the term entered one branch and quickly spread throughout WG, it's impossible to pinpoint the source, and 2) we're calling Frankish PWG so even if the word was adopted into Frankish and spread from there, that's still PWG yielding the word in every branch." -- 16:17, 6 May 2021 (UTC)
 * That doesn't really address the issues that I raised (I never said anything about the palatalisation of velars being a problem!), which mainly concern the vocalism (which could be seen as indicative of a later loan) and the discrepancy in the consonantism. Positing a PWG *c to cover for the discrepancy left me a bit skeptical with only one example, but now that you've found another, I'm a bit more open to the idea. You'll need to find one or two more to really convince me, though. Hazarasp (parlement · werkis)
 * By the way, shouldn't be ōn-stem ? Hazarasp (parlement · werkis) 18:12, 6 May 2021 (UTC)
 * Here's a fun one: . As for the ō-stem, I was going by Köbler. -- 01:45, 7 May 2021 (UTC)
 * OHG and Old Frisian both have ō-stems, but Middle Dutch and OE both have a ōn-stem. This kind of stuff makes me suspicious that they're seperate borrowings. Hazarasp (parlement · werkis) 02:39, 7 May 2021 (UTC)
 * Meh, that's WG for you. -- 05:13, 7 May 2021 (UTC)
 * Little details like these are important if the basis for assuming a common source is tenuous in the first place. Hazarasp (parlement · werkis) 09:14, 7 May 2021 (UTC)
 * ō- and ōn-stems were effectively merged by late Old Dutch, so you can't base any conclusions on that. This merger likely affected other dialects in the area, as we can see that modern German has it too. —Rua (mew) 08:30, 7 May 2021 (UTC)
 * Good to know, though I probably should've checked that; I'm not too well-versed in what happened to the Gmc. continental breakfast. That still leaves the OE form difficult to explain, though. Of course, such a problem can be sidestepped if we see them as seperate borrowings (as we probably should; at the very least, the OE form is not easily connected) Hazarasp (parlement · werkis) 09:14, 7 May 2021 (UTC)
 * . -- 05:52, 25 May 2021 (UTC)


 * In case people missed it, there is a ongoing discussion on the take page: Reconstruction talk:Proto-West Germanic/krūci. -- 16:17, 6 May 2021 (UTC)
 * There is also a related entry, derived from this noun. Leasnam (talk) 16:22, 6 May 2021 (UTC)
 * Given that, I note that the assumption that the word is “clearly Christian vocabulary” is not without doubt. Primarily this relates to a kind of punishment, applied to slaves, and one damn well imagines that Romans made Germans acquainted with it even in the 1st century, so just for and  I point out that it could be 🇨🇬 no less than postdating Proto-West Germanic. The real etymon of this verb is , from , in Medieval Latin written  (which is also in the TLL from some gloss, together with ). Formally, the common haplology points to proto-date, as also the parallel with 🇨🇬, 🇨🇬, 🇨🇬 from which the North-Germanic forms like 🇨🇬, 🇨🇬, 🇨🇬, 🇨🇬 are derived, as obviously the Christians vexed man with preaching from the very beginning, “very far before the religion itself started to become a thing”. Fay Freak (talk) 12:56, 7 May 2021 (UTC)


 * Whatever it's worth, the EWN calls it an "old Germanic borrowing" (so likely before Old Dutch) and the NEW vaguely calls it "a word of the conversion", which could mean a quite early date in relation to the Franks. ←₰-→ Lingo Bingo Dingo (talk)  18:13, 6 May 2021 (UTC)

I found this paper https://scholarlypublications.universiteitleiden.nl/access/item%3A2971774/view which deals with describing and dating various sound changes of the Latin dialect that West Germanic speakers would have been in contact with. It uses Germanic borrowings as part of its data as well. This seems very useful for figuring out how old they are, based on what Latin sound changes they have carried over. In 3.23 and 3.24, three changes are discussed that would have produced an affricate /ts/: Since PWG is considered to have ended around 400, this places it before the palatalisation and therefore forms like the one being discussed here are an anachronism. It is of course possible for Vulgar Latin /tj/ to end up in PWG as an affricate, but I find it unlikely that speakers would treat this as its own phoneme, since to their ears it would have sounded like a sequence /ts/ (compare how western European speakers nowadays hear Slavic c). So if we do want to denote this sound, I think a sequence *ts should be used and not *c. —Rua (mew) 08:21, 7 May 2021 (UTC)
 * From Vulgar Latin /tj/: 2nd century. Reflected also in Gothic, e.g..
 * From Vulgar Latin /kj/: not found in early Germanic loanwords, late enough for the velar to feed into the West Germanic gemination, e.g. (< ). Velars from Germanic feed this in Old French, e.g.  >.
 * From Vulgar Latin /k/ before front vowel: not found in early Germanic loanwords either, e.g., . Again, Germanic loans into Old French feed this, e.g. . Attested in inscriptions from the 5th century onwards.
 * Ending PWG at the 5th century excludes Frankish, and since we merged Frankish into PWG, that date needs to be pushed forward. -- 00:43, 8 May 2021 (UTC)
 * I'm not in favour of Wiktionary following a different standard from what is linguistically agreed upon. Wiktionary should be a linguistic source. —Rua (mew) 13:15, 8 May 2021 (UTC)
 * Academics don't even agree that there was a single Proto-West Germanic, let alone on a date when all its descendants diverged. I feel like we keep having the same conversation about finite PWG vs. a WG continuum. If a dialect absorbs a word and it spreads through the dialects, that's still the language absorbing the word. -- 19:02, 8 May 2021 (UTC)

Reconstruction:Proto-Germanic/ainijaz
With an unknown etymology and only Old Norse as a descendant, this entry should be deleted. -- 16:05, 9 May 2021 (UTC)
 * Delete. The Latin term is a borrowing, though from another Italic language, so you can’t well reconstruct from it but other things than inheritance become more likely for the other languages as well. The Old Norse may be back-formed from a compound with, Icelandic , borrowed from the Latin jūniperus in the form we have as the Vulgar Latin form, which is already suggested by Schiller-Lübben 1875. Fay Freak (talk) 08:41, 2 February 2022 (UTC)
 * Also note that in westrobothnian, all dialects show an initial j- (jên) which is not the expected outcome of *einir (it ought to be êin). For some reason whoever is keeping the westrobothnian wiktionary wrote it with a g but it doesn't change anything. If this word was truly native and occcured natively in old norse or proto germanic, it must have become êin. But no dialect shows this to be true. I am of the belief that it is a loanword aswell, and not even one old enough to have been nativized the same across scandinavia clearly.

Reconstruction:Old English/ynne
No reason to assume this term survived into OE outside of the compound which is inherited from WG. -- 07:27, 23 May 2021 (UTC)
 * There is one attestation as, the genitive plural of . I've added it to . It's listed as one of the Alternative forms of . Perhaps we simply need to move it to a non-reconstructed entry (?) Leasnam (talk) 11:02, 23 May 2021 (UTC)
 * Why is it and not  as it is in  ? I would expect  (ancestor of, ) and  (ancestor of ). *unnijā appears un-etymological, as the latin does not have a double n, but double n in GMW-PRO results later from gemination caused by j  Leasnam (talk) 11:07, 23 May 2021 (UTC)
 * Works for me. Please move and delete. And thanks for catching my typo. -- 17:11, 23 May 2021 (UTC)
 * Can you please take care of this? Thanks! -- Sokkjō 03:33, 27 February 2024 (UTC)

Reconstruction:Proto-West Germanic/maginōn
This is just a variant of that might not even go back to PWG. The descendants should just both live on there. -- 21:33, 24 May 2021 (UTC)
 * I'm not sure why you would want to create one page when both verbs and also the nouns that they derive from, have distinct descendants (either having or lacking the umlaut). I'm also not sure which one of them is the 'original' form, what makes you think that only *maganōn would be a valid reconstruction? If you want to keep / this creates an inconsistency in the fact that we have a PGM page for the noun, so we should change this to and list both variants there also. --DerRudymeister (talk) 17:26, 25 May 2021 (UTC)
 * I agree with the above. Descendants of *maginōn are distinct from those of *maganōn, and I prefer to keep such variations separate as well. Leasnam (talk) 07:14, 10 March 2023 (UTC)

Reconstruction:Proto-Indo-European/bʰruHg-
Iffy semantics, janky root shape. -- 05:36, 31 May 2021 (UTC)
 * Keep. I don't see anything wrong the semantics. The root shape is rare, but compare . De Vaan says "The restriction to Gm. and It., and the pervading zero grade, may cast doubts on a PIE origin; yet there is no decisive argument against it"; we can certainly add a note along those lines too, but I see no reason whatever to delete. —Mahāgaja · talk 07:25, 31 May 2021 (UTC)
 * All I see here is a Latin verb with several deverbal nouns, and a total left field guess of Germanic  being related to it. The connection is worthy of mention in an etymology, at best, not a full PIE root. --  01:16, 1 June 2021 (UTC)
 * : Our entry for is lacking, but its meaning is actually quite close to that of . L&S even glosses it this way: "to have the use and enjoyment of a thing, to have the usufruct". 2A02:2788:A4:205:4172:C92D:3A2E:3CF 13:33, 28 June 2021 (UTC)
 * If and  are from this root, then it is attested in two more families. --Vahag (talk) 09:43, 22 August 2021 (UTC)
 * I added these roots, but I could have done it wrongly. Please check. I'm minded to close this as keep unless comes back and says something. This, that and the other (talk) 07:14, 25 June 2022 (UTC)
 * The similiarity of Germanic and Latin words does not constitute a valid PIE lemma. Their connection with Greek and Armenian words is highly fanciful. Delete. --Ghirlandajo (talk) 21:09, 23 July 2022 (UTC)
 * Kölligin 2019 adduces <  +  (IF 2019: 115-136). He has also argued "From ‘throat’ to ‘self’?" (MSS 2022: 41-59), which is a bit much, but it nicely parallels  <,  and persona (IMHO). Point being, he does not insist on the prefix per se, discussing ,  separately, cf. "While semantically convincing, the phonological side of the equation is problematic" (IF 2019: 126; emphasis mine). The required sound changes are speculative. How is that "no decisive argument against it" (de Vaan, etc.)? Anyway, I'm saying that  is probably not too far away. IncrediblyBendy (talk) 14:35, 23 March 2024 (UTC)

= June 2021 =

Reconstruction:Proto-West Germanic/kauli
The variety of Old English forms shows that it was borrowed after the language no longer had a diphthong -au-, and also after umlaut. Otherwise it would have had the regular -īe-. In the other languages, the occurrence of stem-final -i must also point to a post-PWG borrowing, because a PWG -i would have been lost after a long syllable. This means that it cannot have been borrowed into the common ancestor. —Rua (mew) 08:37, 8 June 2021 (UTC)
 * I absolutely agree that this is a funky burrowing, but I think it's the result of two things: 1. borrowings from Latin sometimes resulting in diphthong braking, compare ~, so  seems a highly plausible vacillation 2. as with many plant words in WG,  probably played a role in its descendants. --  06:22, 9 June 2021 (UTC)

Reconstruction:Proto-West Germanic/buterā
The High German forms also have -t-, whereas from this PWG form would be expected, giving modern. Compare which does have the expected development. This means that the word must have been borrowed after the change t > z was productive in OHG, and therefore cannot be of PWG date. —Rua (mew) 14:47, 20 June 2021 (UTC)
 * I agree. Besides, Old High German had at least 5 other native words meaning "butter", hinting at the likelihood that was fairly new and had not yet ousted out the other terms. Leasnam (talk) 20:26, 30 October 2021 (UTC)

Reconstructed Latin
The following "Reconstructed Latin" entries are not actually reconstructed. Both are attested in Latin.

1) scepticus 2) absedium
 * If they're attested, please cite them on their entries. -- 08:15, 21 June 2021 (UTC)


 * ,, turn up both capitalized and uncapitalized citations from the 16-, 17-, 18-. 19- and 2000s, although many suggest a definition more like "sceptic" (n. or adj.) than the entry's current definition "the sect of skeptics". For absedium, I find only a very few modern examples, some of which call it out as an error, like this and
 * 1916, Archivio per la storia ecclesiastica dell' Umbria, volume 3, page 424:
 * Parmesius Michus ad absedium (3) Saxarię mansit, / Et magnum bellum a latere dextero dedit, / Et per vim terram Saxariam indixit (4).
 * PS, we're missing declension information and an inflection table at obsidium. - -sche (discuss) 18:16, 21 June 2021 (UTC)
 * This concerns an issue that comes up repeatedly: #1, #2. absedium is Medieval Latin, which is a language that postdates the emergence of Medieval Romance languages. If it also seems to underlie some Romance forms, then the reconstructed entry must stay unless the Romance formes are clearly borrowed from Medieval Latin. The problem is that no Romance descendants are given at, only two borrowings. The entry might be entirely in error and in that case should be moved and converted into mainspace Medieval Latin. Compare , which leads to which appears to be borrowed from Medieval Latin. In that case absedium would be a Medieval Latin borrowing from Medieval Romance such as that Spanish form. Du Cange has absedium as well as obsedium.


 * L&S gives Scepticī as an entry, but this is in fact erroneous since the word seems to be attested in the Greek script in the PHI corpus. It wouldn't be in TLL in any case because it's a proper name and these currently go up to the letter D. It doesn't seem to be attested in Medieval Latin either, so it should be moved to mainspace as a New Latin word. Brutal Russian (talk) 22:35, 25 June 2021 (UTC)


 * "If it also seems to underlie some Romance forms, then the reconstructed entry must stay unless the Romance formes are clearly borrowed from Medieval Latin": I disagree, for reasons I express in the RFD of Reconstruction:Latin/consutura, above. (We don't have separate entries or etymology sections for every time a nonce is re-coined, either, or for early modern English use of foobar as a survival from Middle English vs modern revival / borrowing of it from Middle English, etc.) The one entry can explain when it was attested. For etymology sections to be able to link to the term with an asterisk, one could either use piping or make scepticus a redirect, but I see no reason to duplicate the content. - -sche (discuss) 17:26, 26 June 2021 (UTC)


 * Anyway, I put citations at Citations:scepticus (and there are plenty more to be found using the searches linked above), so if someone would like to fix the definition (and add the right temporal label), that entry can be moved to mainspace. - -sche (discuss) 17:53, 26 June 2021 (UTC)


 * I moved Reconstruction:Latin/scepticus to mainspace given that it is amply attested. - -sche (discuss) 22:11, 12 July 2021 (UTC)

= August 2021 =

Reconstruction:Proto-Germanic/hagastaldaz
Duplicate of. -- 19:36, 24 August 2021 (UTC)


 * Thoughts on this? &mdash; Fytcha〈 T | L | C 〉 03:54, 26 January 2022 (UTC)
 * I personally think hagastaldaz is probably garbage and that any terms that appear to require such a reconstruction are later re-formations of hagustaldaz, but I'm not a Germanic languages expert. Benwing2 (talk) 05:01, 26 January 2022 (UTC)
 * Keep, at least for now, I don't see any reason for deletion. --Astova (talk) 21:32, 10 March 2022 (UTC)
 * Delete, seems clear-cut that *hagustaldaz was the original form. — Mnemosientje (t · c) 14:55, 24 January 2023 (UTC)

Reconstruction:Proto-Semitic/šawmān-
A homegrown reconstruction by Fay Freak, which attempts to tell a story that can unite all the Semitic words for "dove". The unfortunate fact is that they are probably not actually related; an irregular Akkadian reflex, then an intentional misuse of the (not widely accepted) sound law proposed in Al-Jallad (2015), then a strange shift in Arabic from /h/ to /ħ/ because the latter is more "lovely", then an even stranger shift in Northwest Semitic to /j/ because that phoneme is "popular"... Even the maximalists in the Semitic reconstruction game don't let their imagination run away with them as wildly as this. As I have mentioned before, protolanguage reconstructions should be a serious attempt at documenting the common ancestor of attested languages, not a playground for our hypotheses in historical linguistics that no Semitist has endorsed, despite extensive attention to this group of words. —Μετάknowledge discuss/deeds 06:40, 26 August 2021 (UTC)
 * And he qualifies of unlikely a reconstruction, *yawn-, which would perfectly work for Northwest Semitic and was posited by Kogan and Militarev in a chapter specifically on animal names. Is there a taboo about doves in Semitic cultures which would account for irregularities from a common etymon? Malku H₂n̥rés (talk) 15:42, 26 August 2021 (UTC)
 * This is a good question. I, for my part, found the forms more likely related than not, influenced by the acquaintance with taboos about animal names. Otherwise various terms for the dove are atomized without etymology, which is itself doubtful. Fay Freak (talk) 16:01, 26 August 2021 (UTC)


 * The Akkadian correspondence is not irregular. The most obvious examples for the correspondence are given, but it does not look like you even cared to read.
 * You should elaborate why the sound law of /ʃ/ to /ʔ/ is misused or not widely accepted. It is absurd to claim that the sound change is not widely accepted—maybe the exact formulation isn’t, but this subtracts nothing of its applicability here. /h/ to /ħ/ occurs in Arabic sometimes, as with examples referenced, this is not “strange”. Compare also, where is “strange”, yet true. I stress that one often has to search for sources of contamination for Semitic etymologies. For the sound change of leading /ʔ/ to /j/ there are not few examples, compare 🇨🇬 from , or 🇨🇬 borrowed from a form which began with glottal stop. As /w/ became /j/, the very distinction of the whole group of weak consonants in initial position became less relevant in Northwest Semitic at various stages.
 * It was good that imagination was running wild a bit. I do my reps wild like an animal too to get a beast body. One has think like a Biblical patriarch a bit, and sometimes one has to sit down and tell a story 🔦. Not everything that can be seen is obvious, and not all that cannot be immediately seen cannot be seen at all. But all I told has analogues, confirmation of regular occurrence! I say, you get flustered by a great array of information too fast. So I do not see any actual argument, only insults for a great work that but combines insights endorsed by Semitists. Where is the “attention” anyway? Do you call Militarev and Kogan’s proto-stage variation ~  attentive? You see I took great care. An unusual extent of, which was needed. Fay Freak (talk) 16:01, 26 August 2021 (UTC)
 * When I say "irregular", I mean it as a term of art. Not that it is unpredecented — although these are rare! — but that it does not follow the regular sound correspondences established by the comparative method. As for Al-Jallad's sound law, you need only read the paper to see that it doesn't apply to this word! (I said it's not widely accepted because the thought that it's morphologically conditioned still seems to be the mainstream idea; in that case, your application would still be a misuse, of course.) When you invoke multiple irregular sound changes to unite disparate forms, you are working against Occam's razor and it increasingly appears that you are justifying these sound changes by the fact that the semantics match, rather than because the sound changes are especially plausible. You say above that it's "doubtful" that the terms could be "atomized without etymology" — as I have told you before, our null hypothesis should be that there is no relation between any two given terms, and we should force ourselves to demonstrate beyond reasonable doubt that there is indeed one. You can work out however you like, and you can post your "great work" on a personal page, but we are trying to present solid reconstructions of Proto-Semitic. —Μετάknowledge discuss/deeds 16:20, 26 August 2021 (UTC)
 * Pretty solid reconstruction here. You are setting up a false dichotomy of regular and irregular sound changes here, an outdated of “sound laws”. “Irregular” and “regular” aren’t terms of the art, but subjective categories. The English and German strong verbs are called irregular, but to those who reconstruct Germanic they are regular categories.
 * Occam’s razor is also no law in humans, they like to do defy rules and make things complicated. I thought I disentangled it?
 * Obviously I read Al-Jallad’s and did not “see” (whatever that metaphor means) that it might not be applicable here. Nor do I know what “morphologically conditioned” is—sounds pretty esoteric. It’s just a sound shift that is known to happen bar certain constraints, boom.
 * I didn’t force anyone. It is still left to the readers to doubt the form. But it is also to their convenience. This is the most orderly attempt of connection of the terms, without which manaman is left confused, cognates here and there with sound changes explained here and there. I would need anyway to mention examples of sound changes and it would be really messy to do that at the individual languages, so at some point the Proto-Semitic unification developed naturally. I chose the most likely form. More likely than a null-relation too. Not biased in the beginning towards a null-relation—. Fay Freak (talk) 17:49, 26 August 2021 (UTC)
 * Again, you are intentionally missing the crux of the issue here. You say it is "left to the readers to doubt the form", but that's not how the dictionary works! We ask them to trust that all our mainspace entries are correct, rather than ask them to doubt (and though we have mistakes, we aim to eliminate them). A reconstruction is not truth, and we have a warning template to that effect, but we still ask readers to trust that we have presented them with something that is, beyond reasonable doubt, what we say it is. This may be "the most orderly attempt of connection of the terms", but it is still incredibly unparsimonious, and that is probably why, for example, Militarev & Kogan didn't try to unite them. You know full well they weren't ignorant of these terms, but they did not consider them to be related. I find this intriguing enough to merit discussion in an etymology section (probably under an autocollapsed box to save space), but we simply can't go around claiming these tenuous connexions as indisputable Proto-Semitic. —Μετάknowledge discuss/deeds 18:06, 26 August 2021 (UTC)
 * Militarev & Kogan didn't try to combine them because they did not have the Amorite form, they were ignorant of this form—which comes right between the other Northwest Semitic and the Arabic form, plus this was already sixteen years ago before various treatises on certain sound changes like this š → h → ʔ one. This was what made me think. Then I found Rescher connecting the Akkadian with the Arabic, knowing even less forms. Militarev and Kogan in all seriousness postulate the form as one of some “complex protoforms accounting for both type of reflexes” (SED vol. 2 page L), they did unite them! For all them even less substance was sufficient to see a family. Where was the reasonable doubt? The way people copy starred forms they saw somewhere it makes little a difference whether we just mention a starred form in a collapsible side-note in the individual language or in new page, reason looks bleak anyway. But aesthetics look served better the way it is. Of course I think whether a Proto-Semitic page exists here is merely a detail. (After people voted for Proto-Albanian pages, in which you did not weigh in by your vote … you recognize that you idealized reason a bit?) Fay Freak (talk) 18:32, 26 August 2021 (UTC)
 * They were indeed ignorant of the Amorite form, but it is a bit dangerous — we can only assume its meaning and its status as a genuine Amorite word, given the cultural mobility of names. It adds another line of support, but not enough to reduce the tenuous nature of the reconstruction as a whole. Militarev in particular is given to motivated reasoning, so my point was that this connexion was too much of a stretch to even make it into the SED — that is a measure of how unparsimonious it is. I didn't vote regarding Proto-Albanian because I don't know anything about it; I do know about Proto-Semitic, and that's why I am careful to ensure a consistent level of security in our reconstructions. —Μετάknowledge discuss/deeds 19:02, 26 August 2021 (UTC)

= April 2022 =

Reconstruction:Old Portuguese/africão
Reconstructed entry with no derived terms. - Sarilho1 (talk) 12:29, 5 April 2022 (UTC)
 * ‘no derived terms’— I think you meant descendants. But unfortunately, seeing as the term used in Modern Portuguese is, a cultismo, the unattested Old Portuguese word has no descendants. The entry looks legit, tho’ a reference would have been better. Keep for now. ·~   dictátor · mundꟾ  13:30, 5 April 2022 (UTC)
 * Question: If aflicão is attested (it has a quotation), why does it claim to be an alternative form of a merely reconstructed word? 70.172.194.25 00:46, 9 April 2022 (UTC)
 * Theoretically, a rarer form of a word assumed to be oftener in another but unattested form can be attested and against direct corpus evidence assumed the alternative form of an unattested form. But the same question has occurred to me and it is still left: the actual attestation situation against this speculation. Fay Freak (talk) 15:17, 9 April 2022 (UTC)
 * The issue here is that the word is only attested as a glaring hypercorrection. — Ungoliant (falai) 23:28, 11 April 2022 (UTC)
 * I do agree that is a cultism, but it's not clear to me why *africão is the expected native form. Wouldn't *afrigão be a more obvious reconstruction? - Sarilho1 (talk) 14:10, 19 August 2022 (UTC)
 * Delete, I reformatted so this entry is no longer needed. This, that and the other (talk) 02:32, 27 February 2023 (UTC)
 * The intervocalic /k/ is still problematic for an inheritance. Nicodene (talk) 03:55, 27 February 2023 (UTC)
 * It gets weirder, though. For some reason I haven't understood yet, the source normalizes the word as "afriçãos" with the "r" and a "ç" that has corresponding pronunciation of (implying a Latin form of *afritianus?). - Sarilho1 (talk) 00:08, 10 March 2023 (UTC)
 * Delete, this entry seems unnecessary if it is not attested in this form. I don't see why it matters that aflicão is a "glaring hypercorrection". And shouldn't /f/ in this context be subject to voicing (compare ábrego), making the actual hypothetical inherited form *avrigão?--Urszag (talk) 04:17, 27 February 2023 (UTC)

= May 2022 =

Reconstruction:Proto-Indo-European/Hrugʰís

 * after other references concludes that this has been borrowed separately into the subgroups of Balto-Slavic we reconstruct as (before Slavic palatalizations obviously) and thence into 🇨🇬 and 🇨🇬 from the 🇨🇬 word *wrugya transcribed in Greek as, if not a related word of the Trümmersprachen nearby.
 * Wikipedia about the cultivation of rye: “Domesticated rye […] is absent from the archaeological record until the Bronze Age of central Europe, c. 1800–1500 BCE.” Fay Freak (talk) 20:57, 8 May 2022 (UTC)

Reconstruction:Proto-Germanic/pīpaną
Proto-Germanic. It is probable that the Germanic word for "pipe" doesn't go back to Proto-Germanic; unlike the page says, is in fact attested, but it is late and scarce. This would seem to point to a borrowing from Medieval Latin and/or Middle Low German. (It also may be worth noting that when I created, Proto-West Germanic didn't exist, so the only option I had was to create the =gem-pro entry.) Hazarasp (parlement · werkis) 05:11, 18 May 2022 (UTC)
 * The edit history should merged into the PWG entry. --Sokkjo (talk) 15:46, 13 February 2023 (UTC)

= June 2022 =

Reconstruction:Old English/giestranmorgen
Old English. Like, this is most likely fictitious; is first attested as late  (in Caxton's 1481 The history of Reynard the Fox), and there's no cogent reason to think that it existed any earlier. The adduced  doesn't prove anything, as it could've easily been independently formed; c.f. modern  ). Hazarasp (parlement · werkis) 09:32, 16 June 2022 (UTC)


 * Delete. ·~   dictátor · mundꟾ  14:21, 13 July 2022 (UTC)

= September 2022 =

Reconstruction:Proto-Slavic/židъ, Reconstruction:Proto-Slavic/Rimъ
It is hard to find any good sources claiming these existed in Proto-Slavic, it is usually believed these were early borrowings in South Slavic and borrowed to other languages durning christianization. Etymologies explained in, ,. Sławobóg (talk) 20:01, 17 September 2022 (UTC)
 * Incidentally, there should not be a stage like "Romance *Rọ̄ma", because that is not an reconstructed form at all, and considering the time that the borrowing would have occurred, most likely we are dealing with Late Latin, 'Vulgar Latin', or however one prefers to describe it. Nicodene (talk) 22:22, 17 September 2022 (UTC)
 * Source must have been some Romance language and it had to sound *Rọ̄ma to give Slavic *Rymъ. Sławobóg (talk) 09:48, 18 September 2022 (UTC)
 * 1) That is exactly how the word sounded in Classical Latin. 2) Approximately when do you suppose the borrowing occured? Nicodene (talk) 12:33, 18 September 2022 (UTC)
 * Boryś suggests 6th-7th century. Most sources I saw mention some Romance etymon reconstructed as *something and we should stick to that. Sławobóg (talk) 13:58, 18 September 2022 (UTC)
 * "*Rọ̄ma" is simply wrong, as it implies that the form is unattested. Needless to say, Roma [ˈroːma] is one of the most thoroughly attested Latin words in history.
 * For the sixth or seventh century CE, it is far too early to speak of distinct Romance languages. I would suggest mentioning 'borrowed in late antiquity' if that is necessary. Nicodene (talk) 22:06, 18 September 2022 (UTC)
 * I think the best solution would be saying it came from "via"  or . I suppose one could add "or an early Eastern Romance language", but that does seem early. My impression is that the transition from Latin to the Romance languages was more like a number of changes running more or less in parallel at different rates and beginning and ending at different times than a single process. That would make it very hard to say exactly when one became the other. It's probably better to be vague and equivocal. Chuck Entz (talk) 22:52, 18 September 2022 (UTC)
 * Several sources say what they say. And this topic is about Proto-Slavic lemmas, not exact etymology of the modern words. My point is these lemmas didn't exist. Possibly all christian should be removed (,, possibly but this requires additional discussion). Sławobóg (talk) 10:37, 19 September 2022 (UTC)

Reconstructed Latin noun and verb forms
Nearly all of these are entirely predictable from the lemma form and, in fact, automatically listed there. The only exceptions are, , and - which can be kept. I would delete the other 180 entries in Category:Latin reconstructed noun forms and Category:Latin reconstructed verb forms. Nicodene (talk) 07:37, 19 September 2022 (UTC)
 * They're there for etymology reasons. Delete, and substitute (eg. 1st conjugation) every  with   to avoid the redlinks. On the Latin pages, links to declined forms (including also supine and past perfect) should not link to anything (maybe by making an argument like (eg.)   supported by ). Catonif (talk) 10:25, 19 September 2022 (UTC)

Update: Given that this RFD faced no opposition in 5 months, I've disabled links from the headwords of reconstructed Latin terms. All of the discussed entries are therefore orphans (excluding links from Romance etymologies, which are on their way to be removed). The only thing missing now is just actually deleting all the forms. Catonif (talk) 19:26, 8 February 2023 (UTC)

Reconstruction:Proto-Indo-European/h₂ewlo-
Strewn together entry of semantically impossible words. --Sokkjō (talk) 04:45, 30 September 2022 (UTC)
 * How are these words semantically impossible? Seems like a pretty short semantic leap from a word meaning "tube" or "hollow" to go on, over thousands of years, to mean things like beehive, stomach, passage, flute, etc, not to mention of course that there's academic sources for each of them.
 * And I wouldn't say they're "strewn together", every word I listed as a descendant either has its own page in the academic works cited for it or is listed as a cognate in others. GabeMoore (talk) 04:54, 30 September 2022 (UTC)
 * You think boat, beehive, hollow tube are solid cognates (**h₂ewlo- is also not a root BTW) ? I'm seeing a lot of those really questionable entries you're adding, like Reconstruction:Proto-Indo-European/pulu- which has descendants which can't possibly be there and no PIE formations to show your work on any. --Sokkjō (talk) 05:04, 30 September 2022 (UTC)
 * I do actually, and so do the people who research this type of thing, whose work I'm citing. Very little of what I (or most people, I believe) do on Wiktionary is original research or speculation. Beehives are found in hollow spaces, i.e. tree hollows. Tubes are hollow. Boats are notably hollow to have buoyancy, although I even listed that is only a possible derivative, as its etymology is not entirely clear. Still, Adams (2013) attributes it to . If all these somewhat phonetically-similar words have the common theme of having a "hollow" something, whether they're tubes, beehives, passageways, stomachs, or flutes, it makes sense to reconstruct a common term for them, which is what numerous academics have done.
 * Also worth noting that means visceral internal organs, the hull/hold of a ship, a hollow or cavity, and beehive, so I'd say that the semantic derivation there is already shown in Latin. Also compare 🇨🇬 >  . GabeMoore (talk) 05:17, 30 September 2022 (UTC)
 * The basis of the entry is Pokorny who should always be taken with a *HUGE* grain of salt, and trying to connect words together on an abstract concept like hollow is extremely tenuous, especially without any corresponding verbs. Latin, has also been connected to , which itself might be from . Many of these words could be from different roots, if not substrate borrowings. --Sokkjō (talk) 18:51, 30 September 2022 (UTC)
 * Okay, but this is still your personal speculation. While Pokorny is pretty outdated nowadays, the fact remains that leading scholars in their respective fields such as Adams and Beekes have connected these words as a direct part of their work, which have become the go-to publication in these fields - whether or not Pokorny connected them first is irrelevant. You're saying that the entry should be deleted because their conclusions don't make sense to you, with no other support. The fact that several authors independently reconstructed this root is more than enough to earn it an entry. GabeMoore (talk) 05:57, 1 October 2022 (UTC)
 * Firstly, one really has to pay attention to the publisher. Although Leiden is very highly looked upon, they're also a circle-jerk and you have to keep that in mind. Secondly, we're not beholden to any books or publications. In fact, as a project, we have our own reconstruction styles in PIE and other languages, and can decide if a reconstruction is worth an entry, or simply a mention in an etymology. --Sokkjō (talk) 06:30, 1 October 2022 (UTC)
 * See Votes/2013-10/Reconstructions need references. --Dan Polansky (talk) 09:19, 1 October 2022 (UTC)
 * Right, doesn't this support what I did? My editing on Wiktionary goes off of the academic work for a subject, Wiktionary "isn't a place for novel ideas", as someone in that discussion said - Wikimedia sites are not primary sources. If academic authors are unanimous in the derivation for a word, one dissenting Wiktionary editor is not grounds to delete a page, that's contrary to the whole idea behind the citation-culture that defines Wiktionary and Wikimedia in general. GabeMoore (talk) 10:34, 1 October 2022 (UTC)
 * We aren't "beholden" to any particular publications, but we have to use academic sources to cite our work, which is what I've been doing. Your issue with this is your own speculation, and you haven't provided any actual sources for your stance. (Although I'm fairly sure that the guidelines are that pages should be created even if there's an academic dispute, and this dispute should be marked; irrelevant either way because there isn't any academic dispute here.) The terms I listed are pretty much unanimously attributed - and not just in Leiden - to . GabeMoore (talk) 12:23, 1 October 2022 (UTC)
 * "I'm fairly sure that the guidelines are that pages should be created even if there's an academic dispute": absolutely not the case. comes to mind. --Sokkjō (talk) 02:05, 2 October 2022 (UTC)
 * creating IE-proto pages is a very difficult work. You should not do it until you are much more experienced. That said, the relationship of these words as inheritance from a PIE word meaning "hollow tube-like object" is universally accepted and is not just Pokorny and Leiden. Because as Skiulinamo says HeRR is an invalid root shape, people assume and (for Hittite) . --Vahag (talk) 11:54, 1 October 2022 (UTC)
 * I'm aware that it's quite difficult, and I usually spend nearly an hour on each PIE page, reading sources and finding cognates. I'm not just "stringing together" words I think sound similar - every single term I enter in any of my PIE pages has been attributed to its root by at least one author. This is the first time (at least since when I first started editing Wiktionary in high school) when anyone's raised an issue with my PIE pages; generally I think they've been a good addition to Wiktionary. GabeMoore (talk) 12:15, 1 October 2022 (UTC)
 * When you're inexperienced, time does not equal quality. I'll sit on a PIE entry for weeks before publishing it sometimes and I consider myself pretty knowledgeable. Your entries make it very clear you don't understand the instructions at WT:AINE. --Sokkjō (talk) 19:01, 1 October 2022 (UTC)
 * I've done quite a few PIE entries in the past few months and no one else has called me out for not understanding the basic instructions on how to make them. I may have created under the wrong lemma here, and I make errors here and there, but I really don't see anything on that page that I routinely ignore. GabeMoore (talk) 21:09, 1 October 2022 (UTC)
 * We don't have a team of people to check PIE entries, which is why we can't have inexperienced users creating entries. Reviewing your PIE entries, and I don't intend to be mean, but you clearly don't know what you're doing when you can't distinguish between a lemmatized noun and a root. You should try your hand at reconstructing Proto-Tocharian before embarking on PIE entries. --Sokkjō (talk) 00:25, 2 October 2022 (UTC)
 * Assuming we keep this entry, ~  might be a better reconstruction. I'm still very skeptical of the inclusion of Armenian  and especially Hittite, which Kloekhorst only guesses the meaning in an attempt to connect it here, while Puhvel speculates “milt, spleen”. --Sokkjō (talk) 19:01, 1 October 2022 (UTC)
 * I found another proposed descendant of : . I think we can securely reconstruct . The formation of this word and its relationship to the etymon of, and the Hittite should be investigated further. Vahag (talk) 17:51, 2 October 2022 (UTC)
 * Viredaz has connected with, , and thus with the root  which others reconstruct as . I think  can be explained from that root as "hollow object into which something can be fitted". Vahag (talk) 18:20, 2 October 2022 (UTC)
 * "to put on (clothes)" => "hollow object into which something can be fitted" seems rather improbable. Maybe it's related to PIE, whence Latin . Perhaps even a chiming root with , whence Latin . 🤷 Anyway, I'll try and clean up the entry. --Sokkjō (talk) 07:46, 4 October 2022 (UTC)
 * do you think could be borrowed from  or vice-versa? Alternatively, both could be from, if I'm not mistaken. It seems more likely that they share a common root than their meanings being innovated twice. --Sokkjō (talk) 22:49, 7 October 2022 (UTC)
 * a borrowing between those two geographically far-apart groups is excluded. The relationship should be explained within the framework of inheritance from PIE. PS. would give  Vahag (talk) 10:14, 8 October 2022 (UTC)
 * and what would a leveled render in Old Armenian? --Sokkjō (talk) 21:33, 8 October 2022 (UTC)
 * it would give . Vahag (talk) 07:05, 9 October 2022 (UTC)
 * seems to support at least some cases of PIE -yeh₂- > Armenian -i. How would you derive ? --Sokkjō (talk) 07:26, 9 October 2022 (UTC)
 * PIE *y becomes Armenian ǰ when preceded by *l, so is not comparable. The formation of  is unexplained. The  is not a problem, it can be an inner-Armenian addition. The ł instead of l is a problem: people assume a formation containing PIE *n, possibly paralleling, but this is uncertain. Vahag (talk) 07:53, 9 October 2022 (UTC)
 * Forgive me, as I'm no expert in Armenian, but what I was referring to with my example was a case of, as Martirosyan refers to it, "metathesis or y-epenthesis", i.e. *h₂el-yo- > *aly > ayl ~ ayɫ. --Sokkjō (talk) 04:51, 10 October 2022 (UTC)


 * OK, I've rescued the entry and moved it to Reconstruction:Proto-Indo-European/h₂ewlós. --Sokkjō (talk) 23:05, 12 October 2022 (UTC)
 * Thanks for the great work. By the way, Pokorny's *awlós probably should not be presented as an alternative reconstruction. It is a pre-laryngeal notation of the same thing as . Vahag (talk) 07:25, 13 October 2022 (UTC)
 * Hah, thanks. I beat you to *awlós. The school of "*a in PIE doesn't exist" is probably bunk and onomatopoeic roots with *a is probably a thing. --Sokkjō (talk) 07:45, 14 October 2022 (UTC)

= October 2022 =

Reconstruction:Proto-Indo-European/énu
The Tocharian just requires any phoneme between the nasal and dental and the Indo-Iranian could be secondary. Both entries at Reconstruction:Proto-Indo-European/h₂en-. --Sokkjō (talk) 23:40, 3 October 2022 (UTC)


 * Sidenote:, it's not recommended to even use R:ine:LIPP as a source; his work is that problematic. --Sokkjō (talk) 00:28, 4 October 2022 (UTC)


 * Keep. I disagree that LIPP is problematic. I agree though that Tocharian is a poor match, better off at with, , etc. —Caoimhin ceallach (talk) 17:37, 13 December 2023 (UTC)
 * No surprise there. Regardless, presuming an entry at RC:Proto-Indo-European/h₂en- is enough supposition. -- Sokkjō 18:50, 13 December 2023 (UTC)

No consensus. No further action. —Caoimhin ceallach (talk) 15:47, 1 June 2024 (UTC)
 * Improper closing of an RDF. -- Sokkjō 00:53, 3 June 2024 (UTC)

Reconstruction:Proto-Germanic/feþlą
Proto-Germanic. This shows only one descendant ; however the OHG word has an alternative form which is which would mean the PGmc term would have an intial p- not f-. Leasnam (talk) 06:35, 13 October 2022 (UTC)

Reconstruction:Proto-Indo-European/ḱlinéh₂ti
11:38, 19 October 2022 (UTC) None of the formations in the descendants actually match the PIE form, so why does this exist? There was a Latin form listed before, but per De Vaan, it doesn't belong there. —CodeCat 14:04, 31 July 2017 (UTC)
 * Beekes clearly cites *klin-je/o- as the root of the the Greek form, and that itself comes from an older nasal present. Kroonen cites the older original form as either *ḱli-neh₂- or *ḱli-neu-. That is what was reconstructed on this page. de Vann concurs and explains the long -ī- as being "introduced from the root aorist *klei- i *kli- (cf. cliēns)." --Victar (talk) 14:12, 31 July 2017 (UTC)
 * This doesn't address the main point at all, but merely confirms it. Why do we reconstruct this if there are zero forms which actually descend from it? —CodeCat 14:15, 31 July 2017 (UTC)
 * The Latin form is cited in sources as an example of this original form. Source add. --Victar (talk) 14:24, 31 July 2017 (UTC)
 * Perhaps, but the entry is now in conflict with the (also sourced) etymology given at and . We clearly cannot follow the sources here, as they contradict each other. This is an issue with a lot of your editing, Victar. You blindly go with sources which often posit very bold hypotheses that don't have widespread acceptance, you don't critically examine them. You also do not seek out consensus; whenever someone reverts your questionable additions, you ignore the fact that reverting an addition means no consensus, and repeatedly reinstate. When discussions finally start, you also refuse to wait for a consensus, but reinstate your edits as soon as you think you have proven your point, as on Reconstruction talk:Proto-Indo-European/ḱlitós. You need to stop hiding behind sources and start listening to editors. —CodeCat 14:27, 31 July 2017 (UTC)
 * If we're going to bring this to personal attacks simply because you disagree with the sources, let me say that your actions are very unbecoming of an editor. I attempt to start a dialog with you, as exampled here, and your solution every time is to simply assert you are correct and delete the entry. You were stripped of your adminship because of your uncooperative behavior which you continue to this day. --Victar (talk) 14:34, 31 July 2017 (UTC)
 * You attempted to start a dialog, then completely ignored it and put the content back anyway. Multiple times. Without consensus. And you still haven't given a good argument for keeping this entry, and you still continue to add content without consensus. —CodeCat 15:02, 31 July 2017 (UTC)
 * Ignored it? I put forth my thoughts, and your response was you're wrong, I'm deleting it again. That is not a dialog on your part. You are not one to talk at all about consensus. For example, there was a clear consensus that laryngeals existed in PII and many of its descendants, and despite that, you systematically deleted them, at which point had to insist that you stop. This project is not a dictatorship for your ruling. --Victar (talk) 15:45, 31 July 2017 (UTC)
 * No person is correct in all decisions. CodeCat has made mistakes, as have I. Recently, Victar, you're been running roughshod across these entries, frequently ignoring language specific-considerations in favor of utopian reconstructionism. I've tried to be fairly conservative in the way I've edited on here, and CodeCat is very useful in reining in my reconstructive excesses. Similarly, sometimes she has personal opinions which need curtailing, but this is done in discussion with other editors. If there is no consensus, it is better to not add it all then to add highly speculative material. Again, I've made these mistakes many times, and indeed at one point CodeCat told me to go read a bunch of literature since I was adding so many bad reconstructions.
 * In this matter, I would agree with CodeCat that there seem to be several competing explanations or maybe several competing forms. This entry is at this point too speculative to merit its own entry. — JohnC5 17:57, 31 July 2017 (UTC)
 * I am completely open to being wrong and absolutely crave discussions so that I might learn from them. I do ping you and CodeCat on entries that I hope either of you can look over. I probably would have done the same once I finished working on the entries, like I did here.
 * My objection is in CodeCat's method of simply reverting someone with no explanation other than "you're wrong", if that much, which I find in bad form. We should be encouraging editors to go through the sources and add this information to the project, and, in turn, correcting mistakes, not disregarding them with reverts and throwing the baby out with the bath water.
 * You both know, I'm a clean and methodical sourcer (I added passage to many reference templates to improve them further), and I want to get these entries done by the books as much as anyone else. Three sources, Kroonen, de Van, and Beekes all suggest a nasal verb root, but if that's too speculative, lets have a conversation and come to some less speculative alternatives. I'm happy to see that CodeCat has now added some to the root entry, but this was only after my objections to the deletion of the entry. --Victar (talk) 18:27, 31 July 2017 (UTC)
 * I usually err on the side of caution, and I wouldn't even have created a PIE page in this instance. The descendants are too different to really pin them down onto a single form. Kroonen etc. may be right about the original -neh₂- suffix, but this is very speculative considering that it doesn't appear in that form in any language. It appears to me like they pulled it out of thin air. The nasal itself is plentifully attested, but its exact nature is too unclear. What I find especially telling is that Greek added an additional -ye- present suffix to it, as if it wasn't "present enough" in its old form. The Germanic form, with a stative formed, seemingly, from a characterised present, is even more puzzling. So my preferred option would be to just say "We don't know" and only list the forms without trying to pin a particular underlying formation on them, as Kroonen and De Vaan do. Note that Kroonen and De Vaan disagree on what they think the original form was, and I find neither of their proposals particularly compelling. De Vaan's proposal might work if the root had an additional final laryngeal, but that's ad hoc and only solves the puzzle for Latin, it makes it harder for all the others. —CodeCat 18:37, 31 July 2017 (UTC)

Haha, "present enough" made me laugh. I can certainly see where you're coming from and Kroonen definitely phones it in at times. I think we can agree though that three independent innovations of seems highly unlikely. I've moved over the sources for to, so please feel free to delete. Also, please copy your comment regarding Germanic weak class 3 over, which I think is helpful. Thanks. --Victar (talk) 03:06, 1 August 2017 (UTC)
 * Re CodeCat and "...personal attacks simply because you disagree with the sources...": +1 on the Uther-meter. Chuck Entz (talk) 02:59, 1 August 2017 (UTC)
 * That is not a helpful comment for dialog. --Victar (talk) 03:06, 1 August 2017 (UTC)

Hoping to get some more input on this at the proper venue. This, that and the other (talk) 11:38, 19 October 2022 (UTC)
 * Keep - again, if it's properly sourced to credible material, we keep it. if there are conflicting opinions in sourced material, we include information about the conflict. we shouldn't be making the call on which is "right", that is original research (though debating/arguing about it is fun! ^__^). our job on wiktionary, as with wikipedia, is to provide a compilation of the existing information. it's not our job (on here, in the article-space) to express our own ideas/opinions about the etymologies (or definitions, etc.) Lx 121 (talk) 13:18, 29 April 2019 (UTC)


 * Delete. I think that it doesn't make complete sense to keep it as it's own entry, but rather to move it to, and then show it as forms descending from the derived form *ḱlinéh₂ti, Qwed117 (talk) 23:31, 9 November 2022 (UTC)

Reconstruction:Latin/werra
Extra note: this should preferably be history-merged to guerra per the discussion below. —Al-Muqanna المقنع (talk) 02:44, 29 October 2022 (UTC)

This has some similarities to Reconstruction:Latin/circa above, which I don't have strong feelings about, but it strikes me that the "Keep" argument is much less tenable in this case. First, / is one of the most frequently used novelties in Medieval Latin, attested copiously across a wide area down to the early modern era (the citations in DMLBS go up to 1545) and as a term of art in medieval law, werra ending up as the standard spelling in English Latin, guerra on the Continent. So unfortunately this series of edits by labelling it non-standard, sporadic, and early medieval was incorrect. I've largely reverted them (and added some quotations).

But, maybe more importantly, we have the following observation by Laury Sarti, Perceiving War and the Military in Early Christian Gaul (ca. 400–700 AD) : "[The Frankish Germanic idiom's] vocabulary, as far as it can be reconstructed, however does not contain a synonym for the Roman term bellum to refer to warfare. The word that comes closest is werra. This term only gained this specific meaning in the course of the Merovingian Age, at the earliest." (Note that Frankish does not mean "war".)

This means that the term's acquisition of its dominant Latin meaning is virtually coincident with its appearance in the sources (9th century at the latest per Niermeyer). I'm generally more conversant/interested in historical work with Medieval Latin documents than the diachronic linguistics, but it seems to me, at least, to make little sense then to distinguish a reconstructed Vulgar Latin entry from the well-attested Medieval Latin one. —Al-Muqanna المقنع (talk) 12:42, 27 October 2022 (UTC)
 * Hi . I am in favour of removing this 'reconstructed' entry and moving its contents to the attested entry. 'Ninth century at the latest' is certainly early enough for a first attestation.
 * I'm not convinced that guerra or werra can really be considered standard, since the literary Latin term, the one regarded as most 'correct', remained bellum throughout the Medieval Ages, to the best of my knowledge. That is, it did not stop being used, only to be revived later. But if a citation can be found that does directly support the notion of guerra/werra becoming the standard word for 'war' in literary Latin, at least temporarily, I'd have no further objection. Nicodene (talk) 12:56, 27 October 2022 (UTC)
 * Thanks . Bellum was in use as well, of course, but of the two sources I've added, one is a monastic chronicle and the other is a formal treaty document. We also have other learned sources like, one of the most prominent late medieval jurists, ad Digest 14.2.2 " qui tempore guerrae propter defensionem vadunt " (cited here, but also just search guerra in that book). That seems fairly standard to me. —Al-Muqanna المقنع (talk) 13:07, 27 October 2022 (UTC)
 * @Al-Muqanna I see- that is a higher degree of acceptance than I'd expected. Perhaps we can simply add a note such as 'coexisted during the Middle Ages with the Classical equivalent bellum before being eliminated during the Renaissance'. Nicodene (talk) 15:22, 27 October 2022 (UTC)
 * I don't mind that usage note, though it might be redundant? Given the humanist reformation of Latin-writing during the Renaissance I don't think there should be any expectation that novel terms from Medieval Latin will be carried over into New Latin; at least I would (and do) treat it as "opt-in" and mark it as "Medieval Latin, New Latin" if it is. —Al-Muqanna المقنع (talk) 15:55, 27 October 2022 (UTC)
 * Yeah, true. Nicodene (talk) 23:29, 27 October 2022 (UTC)


 * If this page is going to be moved to the mainspace, it needs to be moved, not . --Sokkjō (talk) 01:54, 29 October 2022 (UTC)
 * Since a page already exists at the target it cannot be moved (, main entry: ). Content-wise, thanks to Nicodene's work, everything has already been moved except the PItWRom pronunciation (which is potentially problematic given the timeframe of semantic evolution and the loan to Latin). If the problem is preserving attributions the relevant procedure would be a history merge, though there doesn't seem to be a process for those on WT and a talk page note should also be sufficient. —Al-Muqanna المقنع (talk) 02:20, 29 October 2022 (UTC)
 * There's no process for history merges, but we certainly do them here. It's been a while since I've done one, but I know how to do them. Chuck Entz (talk) 02:33, 29 October 2022 (UTC)
 * Nice, will add a note to the top about it in that case. —Al-Muqanna المقنع (talk) 02:44, 29 October 2022 (UTC)
 * just going ahead and deleting an RFD page without giving people time to comment first is really bad form. Do you also plan on me about it? --Sokkjō (talk) 07:27, 30 October 2022 (UTC)
 * No, I didn't 'delete the page' (?), I moved its now-redundant content and left a link. And no, you can't 'move' the reconstruction page when the mainspace already has the corresponding entry, which you should have seen, considering that I both linked it and referred to it in the RFD comment box.
 * The 'reconstructed' page is inevitably going to be deleted, considering that the term is extensively attested (and early on), even in that exact spelling. If you want to stubbornly edit-war away my edit to the page, and stalk through my profile, have at it. Nicodene (talk) 12:45, 30 October 2022 (UTC)
 * You neglected to read the above. We have the ability to merge revision histories. --Sokkjō (talk) 23:44, 30 October 2022 (UTC)
 * @Sokkjō You neglected to re-read your own comment where you demanded that the page be moved. And falsely claimed I deleted it. Nicodene (talk) 23:51, 30 October 2022 (UTC)
 * Conclude this. The Latin word is first attested 858 CE (borrowed from Romance obviously) but the descendants are Italo-Western. So, keep the reconstruction page and add a Medieval Latin descendant. Kwékwlos (talk) 11:13, 7 March 2023 (UTC)
 * It's not obvious at all, and all the sources I can find treat its use as continuous, not a re-borrowing: the FEW treats it as a Merovingian borrowing, Strecker's Introduction to Medieval Latin lists werra as a typical example of Medieval Latin originating in Merovingian usage (p. 31), the more recent Medieval Latin: An Introduction and Bibliographical Guide likewise lists werra as a Latin term borrowed in the Merovingian era. Seems far more likely to me that the Carolingian example from 858 is continuous with earlier Merovingian usage and not a novel re-borrowing. —Al-Muqanna المقنع (talk) 11:51, 7 March 2023 (UTC)
 * The ninth century is quite early in any case and a written can be taken simply as the orthographic representation of the popular term, rather than a borrowing from Romance into Latin (which had only just barely begun to be distinguished). Nicodene (talk) 12:56, 7 March 2023 (UTC)

= November 2022 =

Reconstruction:Proto-Semitic/ʔabay-
Proto-Semitic. — Fenakhay ( حيطي · مساهماتي ) 19:01, 2 November 2022 (UTC)
 * This reconstruction is based on non-standard meanings of, which are copiously shown in Landsberg and Nöldeke to which I have updated the Arabic entry, however the latter author prefers  as their origin, and on my first glance the Akkadian is also related rather to this, as also our Assyriologist thinks, and with this alone the Indo-European relations fall away. For obscure  I point to theories at , of which root it can be assimilated. The present Proto-Semitic page should be moved to . Fay Freak (talk) 21:13, 2 November 2022 (UTC)
 * What do people think of the proposed PIE etymology? It does have a source, but the meanings in PIE and Proto-Semitic are pretty different, even if you can posit "love" as an intermediate meaning or something. Also, it's one of only two entries in Category:Proto-Semitic terms derived from Proto-Indo-European. 98.170.164.88 22:39, 2 November 2022 (UTC)
 * Chronologically and formally, it would make most sense (I rate the chances higher) to assume Semitic having borrowed later, perhaps from an an unattested Old Iranian descendant of the Proto-Indo European into the oldest West Semitic languages, there being other basic borrowings like, maybe dying it out in the Iranian branch of Indo-European unlike the Indian one owing to its vulgarity. In any case it remains that it is better to move the entry to Proto-West Semitic, as I seem to have not explicitly enough suggested, , as its presence in there is yet somewhat better evidenced for it than for Proto-Semitic, even if we decide that it is not enough either; the entry as Proto-West Semitic seems to be quite bearable for me though now, giving an etymology to various Semitic words including Ethiosemitic ones and a borrowing in the . Fay Freak (talk) 22:51, 4 March 2023 (UTC)

Reconstruction:Proto-West Germanic/biggō
Proto-West Germanic. Old Dutch has, , and Middle Dutch , , but forms with b- do not appear till almost Modern times, and may be due to conflation of the aforementioned and 🇨🇬. If this is the case, perhaps the reconstruction can be moved to the Alternative form listed on the entry; but that doesn't explain the Low German and Frisian terms, they may be simply borrowings from the Dutch. 06:36, 29 October 2022 (UTC) Leasnam (talk) 03:51, 11 November 2022 (UTC)

= December 2022 =

Reconstruction:Proto-Indo-European/ḱóymos
This should be either deleted or moved to (see ). Likely the only term that belongs to this root is Greek. . --Sokkjō (talk) 00:08, 3 December 2022 (UTC)


 * Can you offer some evidence for this, or even the existence of a root ? --Lambiam 18:11, 4 December 2022 (UTC)


 * For what it's worth, Kroonen and Derksen support the derivation from, and likewise Matasović, while not as explicit, reconstructs an original anlaut *ḱ- rather than *tḱ-, so he doesn't seem to think of , either. --Florian Blaschke (talk) 00:29, 5 December 2022 (UTC)

Reconstruction:Proto-Germanic/in walą fallaną
Proto-Germanic + the 2 presumed descendants (Old English and Old Norse). SoP. Also, very likely calqued between the two languages, or formed on similar model. Leasnam (talk) 18:13, 18 December 2022 (UTC)
 * Also, the PGmc translates as "fall into the corpse/battlefield" (which doesn't make much sense). The Old Norse is "fall into the dead/slain", the Old English as "fall into the slain", which do make sense Leasnam (talk) 18:25, 18 December 2022 (UTC)


 * Delete all three. --Sokkjō (talk) 04:08, 19 December 2022 (UTC)
 * @Leasnam. Delete . —Caoimhin ceallach (talk) 14:11, 20 October 2023 (UTC)
 * Keep all three. I've changed my view. I don't see why this is necessarily calqued, compare and the like. —Caoimhin ceallach (talk) 16:01, 1 June 2024 (UTC)


 * Delete the Proto-Germanic, but keep the Old English and Old Norse phrases, as they are idiomatic. —Mahāgaja · talk 14:23, 20 October 2023 (UTC)
 * I think you're right. I was under the impression that simply meant battle and val could be dative. Do a-stems always have -i in the dative in Old Norse (unlike Icelandic)? —Caoimhin ceallach (talk) 17:54, 20 October 2023 (UTC)

= January 2023 =

Reconstruction:Proto-Brythonic/llew
Proto-Brythonic.

A UtherPendrogn creation. has questioned this on the grounds that it is a modern Welsh spelling and that Proto-Brythonic would have been too early for this Roman borrowing to have entered the language at that stage. I don't deal with Celtic languages myself, so I'm bringing it here. Also pinging. Chuck Entz (talk) 18:44, 13 January 2023 (UTC)
 * Schrijver, given as a citation, has the Brythonic as *lewū . —Al-Muqanna المقنع (talk) 18:47, 13 January 2023 (UTC)
 * It looks fine to me. Considering the enormous number of Latin loanwords in Proto-Brythonic, I have no idea why anyone would think it was too early. As for the spelling, Schrijver follows different conventions than we do. He shows the final syllables as they were in Proto-Celtic, while we show the latest Proto-Brythonic form after final syllables were lost. We also show the fortition of l and r in initial position, rendering them as ll and rr (see WT:ACEL-BRY). This spelling is entirely in keeping with our conventions for Proto-Brythonic. —Mahāgaja · talk 19:07, 13 January 2023 (UTC)
 * You (and Wiktionary as a whole) are using the wrong terminology. Proto-Brittonic would be the ancestor of Brittonic - the ancient Celtic language of Britain attested from the 4th century BC through the mid-6th century AD, after which it transitioned into Neo-Brittonic. The time period of Proto-Brittonic would be somewhere between the break-up of Proto-Celtic (in the early Iron Age) and the 4th century BC. This is way before Rome conquered Britain and British speakers borrowed this word for lion, so it makes no sense to speak of "Proto-Brittonic" *lewu:. As the word (along with most other Latin loans) was borrowed during the Roman era in Britain (1st century through early 5th century AD) t would properly be Brittonic, not Proto-Brittonic. M.Aurelius.Viator (talk) 04:34, 14 January 2023 (UTC)
 * > We also show the fortition of l and r in initial position, rendering them as ll and rr.
 * This is completely wrong for Brittonic (no less Proto-Brittonic)! This sound change only occurred in medieval Welsh. M.Aurelius.Viator (talk) 04:36, 14 January 2023 (UTC)
 * It was discussed here: Wiktionary_talk:About_Proto-Brythonic. Note that the spelling "ll" in the context of Wiktionary's Proto-Brythonic transcriptions isn't supposed to be the sound of modern Welsh  [ɬ]: rather, it's supposed to represent a "fortis" [l͈] as opposed to extra-short [l̆]; likewise, [r͈] and [r̆] are represented as  and . The argument given, which seems sound, is that reconstructing the loss of final vowels to Proto-Brythonic requires the liquids to have already developed some kind of contrast, since mutation is based on the historical presence of final vowels. Unless you argue that Proto-Brythonic did still have final vowels, or that the mutation of these consonants in Welsh is an innovation introduced only by analogy to other consonant alternations, the chronology seems to require this.--Urszag (talk) 04:48, 14 January 2023 (UTC)
 * Sorry, it's just wrong and not a single professional Celticist uses such spellings! And, once again, you guys are using completely incorrect terminology! When you say "Proto-Brythonic" you actually mean to say "Neo-Brittonic" - i.e., the new stage of the British Celtic languages that starts around the middle of the 6th century AD, when old Brittonic final syllables were now completely gone via apocope and the internal compound vowel was lost or greatly weakened via syncope. Proto-Brittonic - which is not a term that is typically used by Celticists - would be the immediate ancestor of Brittonic (1st millennium BC through the mid-6th century AD) and daughter language of Proto-Celtic (2nd millennium BC). How could so many of you here get this basic terminology so wrong?? It's so infuriating! M.Aurelius.Viator (talk) 17:53, 14 January 2023 (UTC)
 * "The mutation of these consonants in Welsh is an innovation introduced only by analogy to other consonant alternations" may indeed be the case; David Stifter posits this exact scenario for Old Irish's "un-"lenited l, r, and n. — Ceso femmuin mbolgaig mbung, mellohi! (投稿) 17:21, 15 January 2023 (UTC)
 * (/Brythonic, like other proto-languages, is the common ancestor of the, i.e. the phase up to the ~6th century AD; Wiktionary's terminology is perfectly accurate. It does not mean "the ancestor of the Brythonic language", singular, any more than Proto-Indo-European is the ancestor of some other language called Indo-European. —Al-Muqanna المقنع (talk) 04:56, 14 January 2023 (UTC))
 * You don't know what you are talking about - this is not the terminology used by any professional Celticist today! The language tree for the British Celtic languages is as follows:
 * Proto-Indo-European > Western Indo-European > Italo-Celtic > Proto-Celtic > Common Insular Celtic > Proto-Brittonic (which not a commonly used term!) > Brittonic > Neo-Brittonic (also called Common Archaic Neo-Brittonic) > Old Welsh/Cornish/Breton/Cumbric > Middle Welsh/Cornish/Breton > Modern Welsh/Cornish/Breton. M.Aurelius.Viator (talk) 18:01, 14 January 2023 (UTC)
 * It's pretty easy to find in recent literature actually. —Al-Muqanna المقنع (talk) 00:59, 15 January 2023 (UTC)
 * I have been a student of Celtic historical linguists going on 40 years now - numerous well known scholars in the field have consulted with me on various linguistic problems and cited/thanked me in their papers. M.Aurelius.Viator (talk) 21:52, 15 January 2023 (UTC)
 * are absolutely correct on all points. This entry looks fine to me. --Sokkjō (talk) 20:32, 14 January 2023 (UTC)
 * No mainstream Indo-Europeanists or Celticists believe in "Western Indo-European" or "Italo-Celtic" as actual nodes in a family tree; they're just polyphyletic regional groupings. There is no point whatsoever in reconstructing "Proto-Brittonic", "Brittonic", and "Neo-Brittonic" as separate protolanguages since they all have the same attested descendants. The latest stage of the language ancestral to the Brythonic languages is just barely attested, which is why we do have a few Proto-Brythonic lemmas that are not reconstructions. Other scholars might prefer to use a different name for an attested language, but we have chosen not to. (Likewise for Proto-Norse, which also has a few attested terms.) —Mahāgaja · talk 20:47, 14 January 2023 (UTC)
 * I don’t know one example of anyone using the term “Proto-Brythonic”, common terminology should be used as is stated above, Proto-Celtic > Proto-insular-Celtic > Brittonic > Neo-Brittonic. Adopting this approach would be less confusing to people researching the subject and bring Wiktionary closer with other works on the subject. The arguement of not using separate terminology because they have the same attested descendents is complete nonsense, by that logic Old, Middle and Modern Welsh should not have separate terminologies. The sheer number of phonological, morphological and grammatical changes during the time between the 1st and 6th centuries would have rendered the language unrecognisable to speakers at either end and therefore I think its completely justified using separate terminologies for the respective eras.
 * My opinion on transcribing Neo-Brittonic is to use spelling conventions from earliest attested Medieval Brittonic languages look at the most common shared spellings and the rest decide as a collective which is best to use, we could obviously have an IPA transcription next to the chosen orthography. Silurhys (talk) 18:49, 15 January 2023 (UTC)
 * What the heck are you talking about? Google Scholar Google Books Back to the subreddit from whence you came! -- Sokkjō (talk) 21:23, 15 January 2023 (UTC)
 * The term isn't commonly used - but if you even bothered to read the links you shared (or even understood what they are saying!) you would realize that they are using the term to refer to the ancestor of Brittonic (i.e., a language that pre-dates the earliest attestation of Brittonic in the 4th century BC), not Wiktionary's idiosyncratic (and idiotic) use of the term to refer to the early medieval, immediate ancestor of Welsh/Cornish/Breton - a term that should properly be Neo-Brittonic (or, if you insist on being Welsh-centric), Neo-Brythonic. This is the standard among all Celticists today. M.Aurelius.Viator (talk) 21:58, 15 January 2023 (UTC)
 * The only two people here who have a clue what they are talking about is @M.Aurelius.Viator and @Mellohi!, you should certainly take guidance from them. They are clearly knowledgeable in up to date terminology and obviously have a much better understanding of Celtic Historcal Linguistics in general than the rest of you here. At the end of the day, Wiktionary isn’t about your ego, it’s about presenting up to date, academic information in the best way possible. It seems clear to me that there is a major lack of care, ignorance and refusal to learn here. Take the Uindorix entry for example, it has been put under Proto-Celtic when it clearly does not belong there and the answer to trying to fix the issue is essentially “stop being pedantic”. It’s not pedantic, there is nothing wrong with a high level of detail, it increases the quality of the entries and will be much better for other people who come to Wiktionary for information in the subject. It seems to me, Celtic is a complete mess on Wiktionary at the moment. Silurhys (talk) 22:32, 16 January 2023 (UTC)


 * Mahagaja's point was strictly about reconstructing them as proto-languages. A proto-language is a reconstruction based on the comparative method. The stages of Welsh are all attested- we don't need to reconstruct them.
 * If you only have one set of data, how can you reconstruct multiple stages? It doesn't matter how different they had to have been, you have no way to even guess at how many there were, or their relative chronology. Chuck Entz (talk) 21:27, 15 January 2023 (UTC)
 * We do have attestation from several periods though? We have an abundance of names attested (place, personal, tribal etc.) from different periods, we have inscriptions on gravestones, boundary markers, Christian monuments and the bath defixiones (which granted is possible it’s Gaulish but there is no evidence either way). Of course we have an idea of the scale of sound change and a relative chronology, Jackson (Language and History in Early Britain) particularly has done loads of work on chronology (and sound change) and McCone (A a relative Chronology of Ancient and Medieval Celtic Sound Change), Schrijver (Studies in British Celtic Historical Phonology) has done massive amounts on sound change, as have countless others. Silurhys (talk) 22:23, 15 January 2023 (UTC)
 * Once again, the issue is that Wiktionary is using completely wrong terminology. Proto-Brittonic (or Proto-Brythonic, if you insist on using a Cambro-centric term) would be a daughter branch of Proto-Celtic and the immediate ancestor of Brittonic - the ancient Celtic language of Britain which is attested from the 3rd century BC and lasted until the mid-6th century AD, after which Celticists refer to its new form (post-apocope, post-syncope, et al.) as Neo-Brittonic. Wiktionary is using Proto-Brythonic where they should actually be using the term Neo-Brittonic, because all the "Proto-Brythonic" reconstructed forms they are using show apocope - something that didn't occur in Brittonic languages until at least the late Roman era and was not complete until the mid-6th century (using Kenneth Jackson's dating, which is still the standard). As Silurhys states, we have plenty of onomastic evidence and even some limited inscriptional evidence for Brittonic and for Archaic Neo-Brittonic (mid 6th through 8th century AD). The next stage in the British Celtic languages is the break up into separate daughter branches, Old Welsh, Old Cornish, Old Breton, and Cumbric, by the early 9th century AD. M.Aurelius.Viator (talk) 20:17, 16 January 2023 (UTC)


 * By the way, I see a lot of vague without any details- basically name dropping without the names. The entry in question has references- do you? Chuck Entz (talk) 21:39, 15 January 2023 (UTC)
 * @Mahāgaja Do you have any background in Celtic historical linguistics?? Because you are saying absolutely ridiculous things! Italo-Celtic is absolutely recognized as a node in a family tree and your insistence that Wiktionary stick to the completely inaccurate term "Proto-Brythonic" for what is properly (Archaic) Neo-Brittonic/Brythonic (the preferred term among all Celticists today for early medieval, immediate ancestor of Welsh, Cornish, and Breton) is just bizarre. I don't even understand the motivation behind it! The term is wrong and is used by no one outside of Wiktionary to refer to Neo-Brittonic. M.Aurelius.Viator (talk) 22:06, 15 January 2023 (UTC)

Reconstruction:Proto-West Germanic/pinti
Seeing as #p is exceedingly rare in Germanic, most likely pointing to borrowing, I added. doesn't quite agree, so I'm creating an rfv to get other editor's thoughts. ? -- Sokkjō (talk) 06:09, 30 January 2023 (UTC)
 * I fail to see the point of this. Regardless of the etymology, the verification of the entry is based on the descendants, not the etymology. Perhaps you wanted to start a discussion at Etymology Scriptorium instead (?) Leasnam (talk) 14:09, 30 January 2023 (UTC)
 * The point is to have a conversation instead of an editwar. --Sokkjō (talk) 23:39, 30 January 2023 (UTC)
 * My point is that this and any conversation we may have needs to be at the Etymology Scriptorium, not here. Leasnam (talk) 03:42, 31 January 2023 (UTC)

Reconstruction:Proto-West Germanic/pendul
Proto-West Germanic. Request verification of the reconstruction. Multiple issues: phonetically, it doesn't make sense. Alternative forms appear fabricated as adhoc guesswork. The OHG looks like a derivative of an early borrowing of Latin. The OE is clearly a diminutive of, a word well attested in other early Germanic languages with the same meaning, "penis" (see ). I will concede the possibility that this group of words (which includes )  may  be borrowed from Latin, but it's not certain. However, this doesn't explain other apparently related terms in Germanic showing ablaut (,, etc. - granted, these are not sure reconstructions in themselves either). I think "Uncertain" is adequate for all. Leasnam (talk) 18:06, 30 January 2023 (UTC)
 * pfendil* agrees with it being in formation borrowed from ML., however the meaning was influenced by . See the .The existence of OHG makes OE  a cognate. -- Sokkjō (talk) 21:22, 30 January 2023 (UTC)
 * How does come from   ? I would expect . Leasnam (talk) 21:45, 30 January 2023 (UTC)
 * Granted, the -t- is unexpected, but Latin borrowings with -Vl- were almost regularly altered to -il- (compare ) triggering i-umlaut. That, and Latin often vacillated e ~ i. -- Sokkjō (talk) 23:39, 30 January 2023 (UTC)
 * It's not unexpected. It's incorrect: pintel does not come from pendulum Leasnam (talk) 01:56, 31 January 2023 (UTC)
 * 1. Unexplained #p pointing to a borrowing, 2. nearly identical Latin term in both form and meaning, 3. OHG cognate backed up by sources. All you're bringing to the table is an unsourced, outdated, and dubious #b PIE root, and you're hinging its superiority on a -t- instead of a -d-? My guy, you're gonna really hate Old English ~  ~ . --Sokkjō (talk) 03:27, 31 January 2023 (UTC)
 * The OHG is not a cognate. I would refer you to remember what a cognate is (you should know this). And no, I know about fled, it's already listed at . Leasnam (talk) 03:40, 31 January 2023 (UTC)
 * You've addressed none of my points. --Sokkjō (talk) 03:42, 31 January 2023 (UTC)
 * I know :) Leasnam (talk) 03:43, 31 January 2023 (UTC)

Reconstruction:Proto-Indo-European/gʰer-
Dumping ground of an entry, full of unsourced reconstructions. -- Sokkjō (talk) 07:30, 19 January 2023 (UTC)
 * what do you mean by "unsourced"? there are sources on the page --Ioe bidome (talk) 22:02, 17 February 2023 (UTC)
 * Keep. Source or not, most of the etymologies are demonstrable with the comparative method. Saph668 (talk) 16:27, 23 November 2023 (UTC)

Reconstruction:Proto-Balto-Slavic/maréitei, Reconstruction:Proto-Indo-European/moréyeti
Forged PBS word and PIE word made up on it. There and no Baltic descendants and PBS word is based on false etymology of PS. Sławobóg (talk) 13:25, 19 January 2023 (UTC)


 * Keep. Not sure what's wrong with the etymology. Checks out at the PIE and PBS levels, though I'm not too familiar with Slavic sound laws. Claiming that it doesn't exist in Proto-Indo-European is also disregarding the Indo-Iranian descendants. - saph 🍏 13:51, 8 January 2024 (UTC)

Reconstruction:Proto-Germanic/pundijaną
, I'm having trouble seeing the semantic connection between ON and PWG. Perhaps instead the ON is related to PWG, , itself borrowed from Latin , (maybe a vulgar form of ?) -- Sokkjō (talk) 08:03, 28 January 2023 (UTC)

And, ON  is borrowed from OE ? -- Sokkjō (talk) 09:07, 28 January 2023 (UTC)

, to me it seem that the ON verb has a sense of 'forcing' or 'compelling' which it shares with west-germanic, although the modern icelandic exclusively means 'to torture'. I'm okay with removing the proto-germanic entry. But we should state that the west-germanic entry has a possible cognate in Old Norse pynda. --DerRudymeister (talk) 10:57, 28 January 2023 (UTC)


 * is a loanword from . Leasnam (talk) 17:26, 28 January 2023 (UTC)

Reconstruction:Proto-West Germanic/pinn
Duplicate entry of Reconstruction:Proto-West Germanic/pinnā. Etymology also unsourced and highly dubious, deriving from a *#b PIE root. --Sokkjo (talk) 23:48, 30 January 2023 (UTC)


 * Keep. We have consistently shown roots with distinct genders and inflections since the beginning (compare, , ). We do not customarily lump them all together on a single entry. is supported by 🇨🇬 and 🇨🇬, and likely by 🇨🇬. Leasnam (talk) 00:03, 31 January 2023 (UTC)
 * That statement is demonstrably false, with many examples, like, , etc. Further, in PWG entries we find the exact opposite due to the weakening of case endings causing widespread stem and gender vacillation in Old West Germanic, and eventually leading to full mergers in Middle West Germanic, ex. , , etc. In the specific example of WG , the etymology, both borrowing and inherited, supports an original ōn-stem. --Sokkjo (talk) 00:42, 31 January 2023 (UTC)
 * Ok, we haven't kept consistent with it - I erred in stating that. Thank you for the list ! It was a big help. But, at one point we had done that. Those "deal" pages were some of our first, and are a template to how we intended to create these. We also considered creating bare-stem-only forms (e.g. *pinn-) and use the Derived terms to group the descendants. But at any time if we already have a more precise breakout of forms, we should leave it - that is the best-case, most precise way to display the data. I had already created  first. Then you came up behind it and created, which I think is great. I never said a word to you about it. I didn't call for the deletion of *pinnā, but you added the st-M forms already present at *pinn to it, and that was not needed, they were already represented at *pinn. It is more likely that PGmc had at least 2 distinct forms for *pinn-: *pinn and *pinnā. We already show them in the most ideal way possible, so why do you want to remove one of them ? It is less likely that OE, OSX, and OHG independently converted a pre-existing feminine n-stem to a strong masc a-stem in concert (btw, OE  is definitely not a weak n-stem, it's either strong fem o-stem or more likely masc a-stem, or neuter [very slim chance there though]) like its cognates in OSX and OHG.
 * Several in your first "demonstrably false" list are multi-gendered at PGmc, like, so the descendants will show a mix - this is correct. In the others, like , the PGmc gender-decl usually aligns with the source it was taken from, not arbitrarily decided, - both Koebler and Kroonen reconstruct it as strong-o fem. In cases where one descendant is different to the others, I see no need to create a whole new entry - it would be a waste. But when I see that the majority of descendants have both a strong and a weak case side by side I think it's preferable to group them on separate pages. And I'll remind you that I created *pinn because it was a stronger case, there are ZERO attestations of *pinnā until the Middle stages of the languages. I'm not advocating for the removal of either of these, I think we should leave them both (fyi, I was on my way to creating *pinnā myself anyway...it was on my list, but you beat me to it !), but if one were to go, I think *pinnā should be the loser. It's attested too late. Leasnam (talk) 03:29, 31 January 2023 (UTC)
 * PWG is better aligned with 1. the majority of sources, which cite this as being borrowed from Latin,  2. the alternative etymology, which claims it's an vairient of PWG , 3. the gender of most of the descendants of this word. What sources even cite it as an a-stem? "I had already created *pinn first.", is a rather disingenuous argument as you only created it because you were minute-by-minute shadowing my edits in the area, but if you want to merge histories, sure.
 * I never said that the whole list was of such cases, just that many such cases can be found in the link I provided. "There are ZERO attestations of *pinnā until the Middle stages of the languages", not true as we have OHG . Combine that with Middle West Germanic feminine forms, it makes a stronger case for an original ōn-stem, especially since the weakening of case ending usually caused them to be rebuilt as a-stems, not the other way around.
 * To restate my argument for deletion, should deleted because its descendants are a duplication of those on . --Sokkjo (talk) 21:22, 19 February 2023 (UTC)
 * OHG pinna is not inherited from PWGmc, it is a borrowing from Latin. The Low German and Dutch words are probably also borrowings of the Latin, possibly through OHG. Your reconstruction is flawed. is a surer reconstruction. It may or may not be an earlier borrowing from Latin, that is beside the point.   should be deleted. Leasnam (talk) 22:01, 19 February 2023 (UTC)
 * ok, the entry page *pinn was not created first, nor was I aware at the time I created it that you had hours earlier created . I was mistakenly counting the edit I made to Old English here [], which postulated the Proto-West Germanic reconstruction (so in my mind I had planned at that moment to create it, but didn't actually get around to creating it till days later). This is NOT important to this discussion, but if it makes you feel more confident knock yourself out. Leasnam (talk) 22:14, 19 February 2023 (UTC)
 * OHG is a readjusted form, which is pretty common for WG borrowing from Latin. You keep on making all these claims, but backing none of them up with sources (and ) . --Sokkjo (talk) 22:15, 19 February 2023 (UTC)
 * Your claim is vapid. You can't simply make facts up in order to push a false narrative. Leasnam (talk) 22:17, 19 February 2023 (UTC)
 * I am a source. Leasnam (talk) 22:18, 19 February 2023 (UTC)
 * I'm done with discussing this. This is not getting us anywhere. Leasnam (talk) 22:19, 19 February 2023 (UTC)

= February 2023 =

Reconstruction:Proto-Slavic/mok-
Proto-Slavic. Tagged by 2A00:23C7:9C97:8201:A977:F253:F051:826F, but not listed. 70.172.194.25 06:47, 10 February 2023 (UTC)
 * Unusual approach for Proto-Slavic, but I see no flaw nor obvious alternative. Fay Freak (talk) 05:59, 31 December 2023 (UTC)

Reconstruction:Proto-West Germanic/pinnā
Proto-West Germanic. Several issues with this: 1). The Old High German does not show the expected shift from p- to pf-, it is likely a borrowing from the Latin. 2). Middle Dutch and Middle Low German forms do not appear till Middle stages of the languages, they are either borrowings of the Old High German or more recent parallel borrowings from Latin. 3). The non-feminine forms (,, and appear to be much older, likely (if also from Latin) borrowed much earlier. They are already represented correctly at and should be removed from. Leasnam (talk) 22:06, 19 February 2023 (UTC)
 * This RFD should be discussed above on . --Sokkjo (talk) 22:20, 19 February 2023 (UTC)
 * It's a separate RFD. No need to conflate the two discussions. Leasnam (talk) 22:23, 19 February 2023 (UTC)

Reconstruction:Illyrian/Drūn
More of these Illyrian entries User:Balltari added: --Sokkjo (talk) 01:50, 21 February 2023 (UTC)
 * Reconstruction:Illyrian/Bardillis
 * Reconstruction:Illyrian/Glaukias
 * Reconstruction:Illyrian/Blaedarus
 * Reconstruction:Illyrian/Licca
 * Reconstruction:Illyrian/Teutana

Some of these enries, including Drun, are valid. I can provide sources needed, and should not be deleted. —04:45, 10 September 2023 (UTC)

Reconstruction:Proto-West Germanic/þwahilā
Redundant entry of, I'm guessing to accommodate for the Old English? Overkill duplication and *-ilā doesn't seem to have been a productive PWG suffix. – Sokkjō · ᛏ 07:37, 23 February 2023 (UTC)
 * Again you're "guessing" incorrectly, / is also weak (and there is additionally a strong variant). You don't own the Reconstruction space, so please stop acting like you do. Leasnam (talk) 02:24, 24 February 2023 (UTC)
 * how am I "owning" the reconstruction space? I'm just putting in corrections. 1. you failed to address the issue of duplication again, 2. is a mostly unproductive plant/animal suffix, so at best, the etymology is wrong. The term is formed with the higly productive instrumental suffix, making  the principle reconstruction and any variation intra-WG.  --– Sokkjō 20:21, 26 February 2023 (UTC)
 * Obviously it's not only a plant/animal suffix, it is also an agent suffix, a feminine agent suffix. <<1. you failed to address the issue of duplication again>> okay, let me address it to you again - it's not duplication if the declensions are distinct. Easy enough ? Leasnam (talk) 01:12, 28 February 2023 (UTC)

= March 2023 =

Reconstruction:Proto-Slavic/Strybogъ, Reconstruction:Proto-Slavic/Xъrsъ
Both created by User:Sławobóg, who displays apparent tendencies to shove Polish etymologies (often augmented with his personal interpretations) on otherwise controversial terms: Безименен (talk) 08:44, 8 March 2023 (UTC)
 * 🇨🇬 is described (in The Tale of Prince Igor) to have association with the winds and is viewed in this way by plenty of Slavists. Phonetic evidence also points towards *-i- (including for the related verb 🇨🇬; it's reconstructed as by ).
 * 🇨🇬 has an alleged connection with the Sun. For example, in some regions of Bulgaria, it is associated with summer solstice Remark: Dubious sources.. It is viewed as a solar deity by a sufficiently large group of Slavists in order to disregard this hypothesis entirely. Also, there is not enough evidence to assert Slavic origin despite what Sławobóg claims. The theonym is attested so sporadically that there is always going to be a shadow of doubt.

''TL;DR: User:Sławobóg explained in (unrequested) depth why the view he supports is correct. And I still don't get how this justifies ignoring all points of disagreement with past theories.'' Безименен (talk) 01:37, 9 March 2023 (UTC)


 * In there is mention about that other (wind) etymology, but it is also explained why it is unlikely. I respect Melnychuk, but he is wrong on this one, it is definitely, Smoczyński reconstructs  and derives it from 🇨🇬 (y < ū cf. , , ). -t- between [s] and [r] in Slavic is normal (cf.  < ). Mixation of -i- and -y- is mentioned (per Trubachov, but there are others), and it is confirmed since 11th century, I'm glad you ignored that. Root *stryb- is widely attested hydronyms, especially rivers (also in Poland and Polabian), that makes derivation from  problematic on semantic ground. Another semantic problems face Ukrainian words with root stryb- that mean "to run (away)"; "to jump" - that has semantic shift cognates for example in 🇨🇬, . Is there any example for "to extend" > "to flow"/"to jump"? There is also dialectal 🇨🇬 which has almost exact semantic cognate in 🇨🇬 (“to flow (about electricity)”) from the same root. It should be mentioned, that the context in text also allows aquatic etymology (arrows are brought by the wind from the sea). TLDR: wind is correct on phonetic side, but the supposed root of  doesn't explain semantics of Ukrainian/Russian words and doesn't explain why that root is used in so many hydronyms. Plus, there is kinda-morphological problem -  in the theonym is attested for  only, but that theonym has very unique etymology, from interjection /.
 * This is simply a lie. Scholars are divided on this one, some connect Him to the sun, others to the moon. Researchers who support the solar interpretation have no evidence for it. The Tale of Igor's Campaign:
 * Prince Vseslav was a judge for his subjects, he distributed cities among princes, but by night he ran like a wolf, from Kiev he ran to Tmutarakan, before the cock crowed, as a wolf he ran along the road of the great Khors.
 * Borissoff's 🇨🇬 is completly wrong, PIE ǵʰ gives Slavic g, z or ž, never x/k, he also can't tell which sources are credible. Iranian etymology is also rejected on phonetic grounds. That etymology also assumes we had like 3 Sun Gods in Slavic mythology, but not Moon God, which was basic god for Indo-Europeans (and all humans actually). Etymology is pretty simple here: there are mixations of k : x in Slavic languages (x comes from earlier k/s), I gave examples for that. That makes earlier form *Kъrsъ completly correct. In theonymy/onomastics there is simple rule, that theonyms come from homophonous appellative, and we have that appellative: ; compare also 🇨🇬 from , or 🇨🇬. Bezimenem also forged some relation of Khors with Bulgarian solstice, there no such thing, Khors is known only from Old East Slavic texts and toponyms.
 * There is general problem with Iranian etymologies (loved so much by East and South Slavs). In the past some scholars believe there were dozens of Iranian words borrowed into Slavic, including theonyms, today it is just few. Scholars pushed Iranian etymology even more for words that had [x] in them, but for many years now we know thats native sound. All gods have native etymology (Simargl is only weird one). I work on Slavic mythology a lot, meanwhile Bezimenem believes that pagan Slavs knew concept of the sin, probably because he refuses to understand that concept of morality and concept of sin or two different concepts, even our dictionary explains it, just like most pagan peoples. He said that I have no basic knowledge and that I'm ignorant, but he reversed my correct (kinda) change, because he doesn't know about ę : ě alternation, even tho it is mentioned and explained in our both dictionaries of Proto-Slavic. I also explained some of theonyms stuff to him on my talk page. Sławobóg (talk) 12:05, 8 March 2023 (UTC)
 * You are wasting everyone's time with your explanations. The fundamental issue with your etymologies is not that they lack justification, but that they don't acknowledge alternative opinions made by experts. It's not your business to decide which etymology is the most accurate.
 * Also, where exactly have I claimed that sin and morality are the same? Stop twisting my words. I've claimed the two categories are related and that pagans had some comprehension of categories that are related to the concept of sin. You, on the other hand, have asserted that proto-Slavs had "no such thing like" a concept of sin. This is an absolute statement. It implies absolute lack of comprehension of anything related to sin, including morale and ethics. If you want me to stop criticizing you, stop talking in absolutes.
 * PS The account of Bulgarian folklore traditions related to Xъrsъ are attested by Анчо Калоянов and mentioned in Xърс (ref: СбНУ). Go argue with them what they have "forged", I'm not their advocate. Even without that, it's undeniable that the lunar theory is not universally excepted, hence, it's unjust to impose it as such. That's the whole point of my complaint! Безименен (talk) 16:50, 8 March 2023 (UTC)
 * So thank you for wasting my time on writting stuff you don't even care to read. As for that alleged Bulgarian thing, it his just his [weird] opinion, not a fact, there is no historical information about Khors' feast. You are completly ignorant on the topic. You can't even distinguish opinion from historical fact and you don't understand what you are reading lol. I even downloaded these books and I can't confirm Калоянов saying anything like that, there different topic on page 163 (first and second part) vol 2, vol 3 talks "Jewish" epithet used on some occassion. It doesn't matter, it is 19th century book, we have books from 21th century that understand that topic better. Sławobóg (talk) 17:51, 8 March 2023 (UTC)
 * Wow hahaha is it another one source which discuss the "Jewish Angel Khors" from the book "The discussion between the three priests" from the 14-15 centuries. You can find very weird mythology stuff in the related Russian apocryphal texts, like a Great Solar Chicken and the Anti-Satan, so i'm not sure if any quote from "The Discussion" can be used for the research on etymology of any word. Any apocryphal text of this kind has rough translated borrowings and anachronist conceptions. By the way, Kaloyanov doesn't mention it as a Russian text unrelated to Bulgarian geography, probably because the legend about a Bulgarian author.
 * PS Slawobog, do you have more sources on k/ch sound change in Slavic languages? Seems like a very fresh researches. Are you sure they are legit? Tollef Salemann (talk) 18:34, 8 March 2023 (UTC)
 * Well, religion is barmy. If you get too much into it you can . Not bad to make mention of the alternative versions you consider less likely. I don’t like the tendency of Bezimenen giving all versions in a randomly ordered list as if no theory had advantages over the other, but this does not mean that we can’t at least add the references of the theories you consider “pseudo-scientific” (science is a big word when it’s about assessment of probabilities by experience with conjectural material) and note nigh the reference (in a further reading or references header, depending on what you deem their purposes) why you don’t even let it into the main article text. Fay Freak (talk) 18:35, 8 March 2023 (UTC)
 * Haven't read the whole Kaloyanov yet, so im not saying he is a pseudo-science-dude making a mummy collection. I just know from my experience that the topic "Names of Slavic Pagan Gods" has tendency to get a lot of doubtful sources with shady agenda. Tollef Salemann (talk) 18:52, 8 March 2023 (UTC)
 * I wouldn't take Kaloyanov's views uncritically, however, that's not the point here. It seems he bases his observations on ethnological research made by God knows who and on the following sources: "Ходене на Богородица по мъките", "Беседа на тримата светители", "Слово на Христолюбеца" и "Слово на тълкователя". I apologise that I can't reflect more on the problem in hand. My goal is not to refute Slawobog's etymology, but to make him acknowledge the dissonance among the scientific community on the topic of "Pagan Gods". Безименен (talk) 19:23, 8 March 2023 (UTC)
 * It can be a Bulgarian text behind it, but the passage about "Jewish Angel Khors" may come from another not-Bulgarian source, but anyway is kinda weird to use it in etymology research. He is not the one who did it. I hope that the origins of this angel-passage gonna be revealed some day, but it seems like some Greek-Slavic translation localization, compare e.g. to translations of Jupiter in the old Norse Trojumanna saga as Krita-Thor, very similar to that one which was made with Khors in this old Russian texts. Im still not sure if it is anything to do with the etymology tho :) Tollef Salemann (talk) 19:34, 8 March 2023 (UTC)
 * Admittedly, it is not directly related. I just don't see why we should leave an option for this discussion to arise in the first place. One could simply address common past controversies under Wiktionary's etymology and to forget about them once and for all. If this was done beforehand, I wouldn't have dug these dubious/confusing/forged views up. Безименен (talk) 23:15, 8 March 2023 (UTC)
 * @Tollef Salemann It is not related at all. It is used by some supporters of fringe theories. "Jewish" thing is indeed very weird here. My book say it just meant "Pagan", just like "Greek", but it doesn't give further explanation or other examples of this use. Sławobóg (talk) 10:41, 9 March 2023 (UTC)
 * Yeah there is used the word "Ellin" as "Greek". It is usually used in the old Russian sources as a synonym of the pagan Greeks and their beliefs (except maybe one or two examples). Tollef Salemann (talk) 10:58, 9 March 2023 (UTC)
 * Again, where in your etymologies it is implied that 21th cent. theories have superseded old ones? If an etymon in question is so apparent as you make it appear, there wouldn't have been so many theories and fault opinions in the past.
 * I'm fine with trimming down some of the sources that I add, as long as it is apparent that given contested question bears uncertainty. I often struggle to represent faithfully the complexity of a given etymology without going overboard. Безименен (talk) 18:55, 8 March 2023 (UTC)
 * I'm fine with trimming down some of the sources that I add, as long as it is apparent that given contested question bears uncertainty. I often struggle to represent faithfully the complexity of a given etymology without going overboard. Безименен (talk) 18:55, 8 March 2023 (UTC)

Reconstruction:Proto-Slavic/regti
Proto-Slavic. There are no sources, none of the given descendants match the reconstructed verb, the proposed PIE root does not exactly match the semantics of the alleged Slavic verb. A related term 🇨🇬, which the creator advices us to see, is given by Slovene linguists as a derivative of 🇨🇬 < 🇨🇬.

? 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:)
 * I've had to revert because they seem to believe that this vote forbids single-descendant Proto-Italic reconstructions on PIE entries and was . --  21:36, 12 May 2023 (UTC)
 * I have already expressed my opinion on single-descendant reconstructions. Thadh (talk) 21:39, 12 May 2023 (UTC)
 * 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. -- 21:46, 12 May 2023 (UTC)
 * 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. -- 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, . 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)
 * : 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. -- 22:51, 12 May 2023 (UTC)
 * 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 eos 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, the key being we're not "deleting all" Proto-Italic entries, only specificly ones with a single descendant. That vote would have looked like Votes/2019-01/Banning Altaic. -- 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 itc-pro with, 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.
 * -- 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 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. -- 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)
 * 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. seems like a clear example of this. Benwing2 (talk) 19:36, 24 June 2023 (UTC)
 * 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)

Reconstruction:Proto-Indo-European/h₁erh₂-
Problematic entry. *h₁erh₃-o-s works fine for the Hittite, but the Latin demands either some irregular lengthened o-grade, or a laryngeal/liquid metathesis. The Latvian stands alone, requiring a separate o-grade form, but it may as well just have been borrowed from Latin, as it was into Old English. It's all a lot of acrobatics that is better left to an etymology section. -- Sokkjō 07:59, 11 April 2023 (UTC)
 * Delete (not that I find the theory that the Latvian term is a Latin borrowing plausible) -- 𝓑𝓱𝓪𝓰𝓪𝓭𝓪𝓽𝓽𝓪(𝓽𝓪𝓵𝓴) 02:52, 8 May 2023 (UTC)
 * RFD-deleted -- 𝘗𝘶𝘭𝘪𝘮𝘢𝘪𝘺𝘪(𝘵𝘢𝘭𝘬) 12:51, 20 May 2024 (UTC)

Reconstruction:Proto-Albanian/gritlā
Etymology is rather semantically dubious, >, and should be left to the etymology section of. Entry also created by block-evade user. Pinging -- Sokkjō 01:29, 18 April 2023 (UTC)

Reconstruction:Proto-Indo-European/plúnksneh₂
No cognates outside Baltic. — *Diwodh₃rós 13:10, 18 April 2023 (UTC)

merge with --Ioe bidome (talk) 17:38, 23 April 2023 (UTC)

RFD-delete. Descendants moved to PIE *plewk-. —Caoimhin ceallach (talk) 16:47, 1 June 2024 (UTC)

= May 2023 =

Reconstruction:Latin/Hispaniolus
Latin. See. Earlier, speakers of such languages would have called themselves ladino or cristiano when Old Occitan began to innovate the word. Kwékwlos (talk) 10:08, 3 May 2023 (UTC)


 * Delete. Nicodene (talk) 13:48, 2 May 2024 (UTC)
 * Delete, contradicts information already on Wiktionary. Fay Freak (talk) 14:37, 2 May 2024 (UTC)
 * @Fay Freak this is not a reason for deletion. If it were, one could equally make the same argument for deleting whichever pages contradict this page. You need to reference actual reliable sources in order to establish the reliability or non-reliability of this reconstruction. Brusquedandelion (talk) 00:13, 12 May 2024 (UTC)
 * It is: the nominated-for-deletion page contradicts the information, the facts on the other page which I assume as true due to detail and references, not simply the expression of it. Indeed I care whether pages contradict each other but don’t put this forward as a reason. You made this up in your head from my rhetorics. Fay Freak (talk) 00:22, 12 May 2024 (UTC)
 * Comment: @Romanophile, as the page creator, can you comment on this RfD? Thank you. Also, @Kwékwlos, can you provide reliable sources arguing in favor of your view here? Likewise for Romanophile. Brusquedandelion (talk) 00:11, 12 May 2024 (UTC)
 * https://www.etymonline.com/word/spaniel (((Romanophile))) ♞ (contributions) 23:37, 13 May 2024 (UTC)
 * That site provides no factual basis for such a reconstruction. Nicodene (talk) 01:33, 14 May 2024 (UTC)
 * Check The Barnhart Dictionary of Etymology. (((Romanophile))) ♞ (contributions) 01:51, 27 June 2024 (UTC)

Reconstruction:Proto-Germanic/jumjaz
Ymir of Norse mythology is obscure and its etymology is purely conjecture, not warranting a full entry. -- Gowanw (talk) 20:05, 8 May 2023 (UTC)


 * All modern specialists in Germanic and Norse language and mythology reconstruct *jumja- which is generally connected to Indo-European *yémHos, cf. Böldl (2013: 112), which can be reconciled, as Kroonen (2013) does, by reconstructing a zero-grade pre-Proto-Germanic form *i̯m̥H-i̯o-. Interestingly, as observed by Kroonen (2013), Old Irish emon can actually go back either to a zero-grade *imH-o- or a full grade *iemH-o- (the traditional reconstruction). This doesn't mean we have to reconstruct a zero-grade but it does leave open the option.
 * Positing a long stem vowel **Ýmir as per Huld apud Mallory & Adams (1997) on the other hand is entirely unwarranted and unnecessary. There doesn't appear to be any reason to outright reject the connection posited by the majority of experts in the field.
 * Böldl, Klaus (2013). Götter und Mythen des Nordens: Ein Handbuch [German: Nordic Gods and Mythologies: A Handbook]. Munich: C.H.Beck. Hörntje (talk) 13:38, 10 June 2024 (UTC)

Reconstruction:Proto-Italic/kruks
Proto-Italic.

Chuck Entz (talk) 06:01, 16 May 2023 (UTC)


 * Deletion seems appropriate based on the policy discussed above. In the case of this word, De Vaan even expresses uncertainty about the exact form of the Latin word's ancestor and whether it ultimately is a borrowing or inheritance.--Urszag (talk) 00:23, 20 May 2023 (UTC)
 * Delete. Fay Freak (talk) 14:39, 2 May 2024 (UTC)

Reconstruction:Proto-Dravidian/weṇney
Does the derived terms section at Reconstruction:Proto-Dravidian/weḷ make a separate entry for this unnecessary? It is listed as a separate entry on page 531, column 2 of R:dra:DL. Kutchkutch (talk) 23:51, 19 May 2023 (UTC)
 * No, it is not unnecessay, we do this all the time with Proto Indo-European entries (create a separate entry for a derived term when it has a lot of formal descendants and is well supported by the sources). -- 𝓑𝓱𝓪𝓰𝓪𝓭𝓪𝓽𝓽𝓪(𝓽𝓪𝓵𝓴) 01:34, 20 May 2023 (UTC)
 * Agreed - there are enough descendants for a separate page. We have other derived/compound terms with such pages. MSG17 (talk) 10:15, 2 June 2023 (UTC)

Kept - this also isn't really a valid deletion request, and the term should be listed as Proto-South Dravidian (a separate language). Theknightwho (talk) 20:56, 24 February 2024 (UTC)


 * @Theknightwho why should it be listed as Proto-SD? Which source argues that? Brusquedandelion (talk) 00:19, 12 May 2024 (UTC)

Reconstruction:Proto-Semitic/yamīn-
Proto-Semitic. Rfv-sense: oath. Tagged by an IP editor with the comment "Attested anywhere before Classical Arabic?" Vox Sciurorum (talk) 17:46, 22 May 2023 (UTC)
 * 1. The development of this sense is secondary even in historical times in so far as “the ancients” (of the Arabs) used the terms listed at . I gotta add them now, for fixes the poetry locations known by his time.
 * 2. The cognitive bias behind the meaning ascription is well-exemplified. People are inclined to project later dogmatization of the meaning of an “oath“ as “aught right” back: Like Rua in 2014 added this sense to according to careless Leydenite literature. There is some hereditary demagogy for this social understanding here to which I am immune after four centuries of inbreeding of the later   which believed in a prohibition of oaths the scriptural basis of which is still controverted in this century. No coincidence, 2020 I fixed the etymologies of on one hand  and on the other hand  with two PIE pages, and namely I knew back then already what most jurists, barely taking courses in Roman Law, don’t even know: That the base meaning of  is “court of law”, Gericht, precisely not . Find out what an in iure cessio is if you want to know how so. Due to Roman formalism in law (a transaction was not valid if not following a particular form like mancipatio, stipulatio etc.; circumstances in the actual historic Roman law that were barely understood either before the twentieth century), this means that  meant everything but “to take an oath”: Apparently it meant appearing in court and pronounce some formal statement there. (Roman procedural law is largely lost.) Hence the idea of an oath, that later became so relevant in the Middle Ages after fusing barbarian law understanding with Catholic rigidity, was only secondarily expressed as ius iurandum. The important oath for military had another name,.
 * 3. By the same fashion I understand that the Arabic oath was only a formal expression of allegience, a speech act of allegiance, perhaps to the, and see also , and the material understanding as right is also secondary.
 * 4. Therefore it is demonstrated that we have to remove the alleged meaning even if copiously found as reconstruction in references, as people choose their social beliefs according to group habit more than reasonable argument, in such a fashion that Christian sectarian practices trickle down in secular researchers who have not learned to check their biases appropriately.
 * 5. A minore ad maius, as I have exposed the development with the Romans and Arabs, we have to theoretize that oaths had no place with Indo-Europeans, and indeed, as an experienced editor with good judgment, above moves for deleting  to which  is falsely traced, with univocal agreement, by three other editors. Fay Freak (talk) 19:36, 22 May 2023 (UTC)
 * 5. A minore ad maius, as I have exposed the development with the Romans and Arabs, we have to theoretize that oaths had no place with Indo-Europeans, and indeed, as an experienced editor with good judgment, above moves for deleting  to which  is falsely traced, with univocal agreement, by three other editors. Fay Freak (talk) 19:36, 22 May 2023 (UTC)

Reconstruction:Proto-Indo-European/bʰardʰéh₂tos
Proto-Indo-European. Reconstruction by (which although straightforward, it must be noted it is unsourced). I don't see why we can't handle this in the derived terms of, as we usually do. Catonif (talk) 19:12, 23 May 2023 (UTC)
 * Delete: The PIE word for beard has always been contentious. No sense making it worse with derivatives that are clearly secondary. -- 23:52, 23 May 2023 (UTC)
 * Delete. Derivatives of this extension are already listed on . - saph 🍏 16:34, 7 December 2023 (UTC)

Reconstruction:Proto-Germanic/parrukaz
Not attested in Old Norse. History should be merged into. -- 00:29, 25 May 2023 (UTC)


 * What a twist.
 * Pinging @Leasnam as well, who added the mention of an Old Norse parrak. Do you recall where that may have been from? Nicodene (talk) 05:47, 25 May 2023 (UTC)
 * Koebler has it in Old Norse, as and  (B&T also lists parak and ), but Koebler postulates it as a loanword from Old English or Middle Low German. However, I want to say I had gotten it from somewhere else long before I started using his works, but I cannot remember. parrak wouldn't be right anyway, it would have to be  or something like that and the form  always bothered me. Leasnam (talk) 06:21, 25 May 2023 (UTC)
 * The circular reference in the etymologies of and  also might need to be looked at. Leasnam (talk) 06:50, 25 May 2023 (UTC)
 * Indeed, that's what prompted this investigation (Talk:parricus).
 * So did both Danish and Swedish inherit some form of this par(r)ak? The meaning 'fish-pond' does seem consistent with that. On the other hand, the meaning 'park' suggests a modern loan directly or indirectly from French . Nicodene (talk) 07:01, 25 May 2023 (UTC)
 * As far as I can tell, parrak in only attested in medieval Icelandic as a nickname of unknown meaning: “Eystein rejste fra Norge til Island; han var søn af Rane, der var søn af Hildir Parrak.” Hellquist is the one that cites  as a “laan fra ags. pearroc”, but the -ak is a more typical Saxon vocalisation, fwiw. --  07:22, 25 May 2023 (UTC)
 * Hellquist does not give the form "parraka", either in the first edition (which is on Runeberg) or in the third (which is in my lap). Instead he says "En gammal biform föreligger i ags. pearroc, inhägnad (eng. parrock) = fhty. pferrich, (får-)fålla o.d. (ty. pferch), osv., mlat. parricus." For the main form of the word (park, y. fsv. parker) he says "Väl ytterst från fra. (enl. somliga i sin tur av iberiskt urspr.)" 217.210.158.171 20:53, 9 July 2023 (UTC)
 * exists but is defined as . It conceivably derives in meaning from, but even so, it would be more than likely a Low German borrowing. I can't find any source that claims also means , though it wouldn't be too hard of a semantic shift. , it seems you added that back in 2012 (have we really been working on these entry for 11 years? 0_0).  --  08:01, 25 May 2023 (UTC)
 * I added the Scandinavian descendants in 2016 []. Rua (CodeCat) and I (and a few others) made many of the early ones...and we didn't yet distinguish a node at PWG. Yeah, it's been that long. Leasnam (talk) 18:56, 25 May 2023 (UTC)
 * Yeah, I created the PWG back in when we still called it Frankish. :p --  19:14, 25 May 2023 (UTC)
 * Yep, and at that time the two were still considered distinct :) Leasnam (talk) 15:01, 26 May 2023 (UTC)
 * From where is the claim that parak exists as a Swedish word? Wiktionary doesn't have it, it's not in SAOB, not in Hellquist, and Korp gives no relevant hit, only a spurious one arising from inserting a space into parakit. 217.210.158.171 21:01, 9 July 2023 (UTC)
 * This is the main form for parr[i~u]kaz. If we delete this, would we delete parrikaz as well? — Soap — 08:33, 15 March 2024 (UTC)

= June 2023 =

Reconstruction:Proto-Indo-European/ḱh₂d-
There has been an RFV (now auto-converted into an RFD) on this page since April 2019. Nicodene (talk) 22:12, 1 June 2023 (UTC)


 * I deemed this RFV passed a year and a half ago, but apparently forgot to remove the tag. I'm sorry. Unless anyone brings up further objections I can remove it. —Caoimhin ceallach (talk) 23:22, 1 June 2023 (UTC)
 * Can we move it do though? I see it was proposed in the past RFD as well, but never done. Catonif (talk) 13:33, 15 June 2023 (UTC)
 * Delete. Not everything starred somewhere needs a page. Fay Freak (talk) 14:41, 2 May 2024 (UTC)

Reconstruction:Proto-Indo-European/sh₂ómn̥
Created by User:Dyḗwsuh₃nus, who has a known track record of creating dubious reconstructions. It formally fits only Sanskrit as the Ancient Greek term could be from a related but distinct formation. This means the entry has been created on the basis of a single descendant. -- 𝘗𝘶𝘭𝘪𝘮𝘢𝘪𝘺𝘪(𝘵𝘢𝘭𝘬) 03:49, 6 June 2023 (UTC)
 * Should probably be moved to, beside , making it slightly less tenuous. -- 04:45, 6 June 2023 (UTC)
 * Delete. Not enough support by material to be credible. Fay Freak (talk) 14:42, 2 May 2024 (UTC)
 * Beekes doesn't even say with certainty that ῠ̔́μνος is from *sh₂em- and lists *sh₂em- as one of the many possible etymologies. Delete -- 𝘗𝘶𝘭𝘪𝘮𝘢𝘪𝘺𝘪(𝘵𝘢𝘭𝘬) 01:51, 23 May 2024 (UTC)
 * RFD-deleted. As for *sh₂em-, anyone taking the initiative can create a new entry. Moving this one to the root form and cleaning it up would've taken the same amount of effort. -- 𝘗𝘶𝘭𝘪𝘮𝘢𝘪𝘺𝘪(𝘵𝘢𝘭𝘬) 05:19, 3 June 2024 (UTC)

Reconstruction:Proto-Germanic/skaggiją
Duplicate entry of Reconstruction:Proto-Germanic/skagją. -- Sokkjō 04:04, 21 June 2023 (UTC)

Reconstruction:Proto-Indo-European/leyd-

 * For context: *leyd- § History, ludo § History and . I invite editors to check these links and the entries' revistions, and of course to check the sources firsthand, before stating an opinion.

Proto-Indo-European, in the sense "to play", added by. This revision shows the references which endorse a single root meaning "to let, allow; to let go, release". The semantic explaination can be found here. The current revision puzzlingly cites the same sources while claiming something wildly different, which is supported, as the PIE entry currently stands, by the two additional references of Pokorny and Adams (of course potentially many more works of the same and earlier time-period can be cited). Their works respectively are undeniably of enormous value, yet science is moving forward, and I've yet to find an authoritative source explicitely rejecting Rix's view in favour of the simpler traditional Baltic-less "to play" one. Catonif (talk) 20:29, 29 June 2023 (UTC)
 * To echo what I wrote earlier, most recent does not equal most right, and just because something is published, doesn't mean it shouldn't be scrutinized. Rix posits the theory that the Balto-Slavic terms meaning, which are traditionally found under  , should instead be reconstructed as from . He then goes one step further and suggests the root is one the same as the well-established root.
 * The semantics of this merge are dubious, which Rix himself claims tentatively, marking each and every reconstruction as uncertain. The safer bet is to continue to keep this as two separate roots, whilst noting Rix's theory on the entry. -- 00:03, 30 June 2023 (UTC)
 * So we're removing references now? The Leiden books are just citing LIV and we don't need to recite it every time it's mentioned, just the original source. Scholars citing each other is exactly the point, it means the idea is widespread and makes sense. I understand the concept of reference clutter whenever there's an undisputed claim, but here you're claiming that Rix is the only one that tries to merge these roots, and it seems like you are trying to make it seem like that's the case, while it undeniably isn't. [T]hey aren't studies on PIE. Bonkers, they're Indo-Europeanists! Do I need to wait for Lubotsky's project to be finished for the ref to be usable? [M]ost recent does not equal most right. Certainly can't be applied to every single case, but given your expertise in the field I don't think I need to be the one explaining to you that the vast majority of times [n]ewer works that are more up-to-date with modern scholarship have precedence over the old ones (from WT:REF). We can place your question marks wherever you will, but I don't think we should split the root in two for them. I personally find Rix's semantical explaination convincing, considering the same shift is observable notably in 🇨🇬 and relax : cf. 🇨🇬, as Latin lūd- was initially the opposite of work, for which compare also 🇨🇬 → school, with the Latin parallel . Catonif (talk) 20:05, 3 July 2023 (UTC)
 * No sources were removed, but attributing them on the entry as a primary source to the reconstruction is incorrect. If we cited every secondary source of a reconstruction, there would be dozens upon dozens. In fact, for what might be the oldest reconstruction for, theorizing the connection between the Balto-Slavic and the Albanian, but with no mention of the Latin and Greek words for . --  21:30, 3 July 2023 (UTC)

Reconstruction:Proto-Indo-European/amma
Two descendants of questionable relation. -- 07:36, 3 August 2023 (UTC)
 * Delete. Enough to speedy. Fay Freak (talk) 14:44, 2 May 2024 (UTC)

Reconstruction:Proto-Tungusic/lökkö
Proto-Tungusic. Created by TheSilverWolf98. And tagged by Ardahan Karabağ. -- Dubukimchi (talk) 16:36, 13 August 2023 (UTC)

Reconstruction:Proto-Slavic/xujovъ
Inconsistent reconstruction (per other Proto-Slavic reconstructions, *o would have transformed to *e after *j), and I'm uncertain that the Polish descendant is directly from Proto-Slavic, it could even be a very late borrowing from Russian. OosakaNoOusama (talk) 23:48, 31 August 2023 (UTC)


 * As a Russian speaker: Haahahaa. Aahaahah. Hah.
 * Delete. Nicodene (talk) 23:55, 31 August 2023 (UTC)
 * And likewise Reconstruction:Proto-Slavic/xujьňa.
 * Nicodene (talk) 23:56, 31 August 2023 (UTC)
 * can confirm that the Polish word is either a late borrowing or independently coined. it has only 4 occurrences in Korpus Języka Polskiego PWN and the earliest occurrence in NKJP is 1996 88.156.139.16 18:57, 5 September 2023 (UTC)
 * Delete. Catonif (talk) 10:09, 11 September 2023 (UTC)
 * Delete per all the above. PUC – 08:29, 18 October 2023 (UTC)
 * Delete. Fay Freak (talk) 14:43, 2 May 2024 (UTC)

= September 2023 =

Reconstruction:Proto-Japonic/niako
I made a mistake there. It's supposed to be *niakua, not *niako. At first, I just thought that if I made the word "niako", someone won't change the title. Now, I changed my mind. I want to delete this page because now I realized it's reconstructed as *niakua. Please delete the "*niako" page and use "*niakua" instead. 115.84.96.189 02:13, 1 September 2023 (UTC)
 * I made it a hard redirect instead of deleting it, in case someone looks it up under the *niako form. —Mahāgaja · talk 06:12, 1 September 2023 (UTC)
 * Okay. 115.84.95.49 13:31, 1 September 2023 (UTC)
 * The current niakua reconstruction appears to be a mistake -- see Reconstruction_talk:Proto-Japonic/kua. ‑‑ Eiríkr Útlendi │Tala við mig 04:06, 5 September 2023 (UTC)

Reconstruction:Proto-Germanic/marþraz
No entra-WG descendants. Edit history should be merged into PWG and deleted. -- Sokkjō 07:54, 1 September 2023 (UTC)
 * Delete. Catonif (talk) 10:09, 11 September 2023 (UTC)

Reconstruction:Latin/-tora
Latin. Tagged but not listed by Nicodene. Ultimateria (talk) 03:31, 4 September 2023 (UTC)

Reconstruction:Proto-Northeast Caucasian/x̱änɦï
* x̱änɦï is an alternate (Starostin 1994) reconstruction of (reconstruction from Schrijver 2021), I don't see why there is a separate Wiktionary page for, when the alternate reconstruction is already in the page  --Qmbhiseykwos (talk) 00:20, 8 September 2023 (UTC)
 * You're going about it backwards; should be deleted, and then  would be moved to Reconstruction:Proto-Northeast Caucasian/ɬ:ɔn, assuming that's what people want. You should start a discussion first on making your point on why this is a preferable reconstruction, because as is, all our PNC reconstructions so far have been based on Starostin and we don't want two competing systems. --  01:21, 8 September 2023 (UTC)
 * Here is the case I will make for why Schrijver's reconstructions are preferable reconstructions to Starostin's PNEC reconstructions. I quote Johanna Nichols ("The Nakh-Dagestanian Consonant Correspondences," 2003:208) at length
 * "Given this agreement, the obvious research priority would seem to be reconstructing Proto-Nakh-Daghestanian (PND). But, oddly enough, this is the one thing that has not been done. The groundwork for Nakh-Daghestanian (ND) comparative studies was laid by Trubetzkoy (1930), who established some notoriously non-transparent but quite regular correspondences; however, his cognate sets are for 'North Caucasian' and include Northwest Caucasian (NWC) as well as Nakh-Daghestanian words. Gigineishvili (1977) reconstructs Daghestanian only. Bokarev (1981, and the earlier 1961 work on which it is based) deals with ND, but unequally, comparing Daghestanian protoforms and cognate sets to Nakh translation equivalents without determining whether the Nakh words are actually cognate to their Daghestanian translations (often they are not). Diakonoff and Starostin 1986 assume that Hurro-Urartean is related to Nakh-Daghestanian, and assemble putative cognate sets so as to maximize similarities between the two families. Nikolayev and Starostin (1994) assume that Nakh-Daghestanian and Northwest Caucasian form a family and offer reconstructions for that putative family, again assembling putative cognate sets so as to maximize matches and similarities between ND and NWC. This method offers no guarantee that the ND subpart of an ND-NWC putative cognate set is a proper ND cognate set, and in approximately a third of the cases they are not (Smeets 1989, Nichols 1997). As the putative cognate sets determine the correspondences and hence the reconstructions, there is no certainty that a proper PND has actually been reconstructed. Neither Diakonoff and Starostin, nor Nikolayev and Starostin, take on the burden of proof and discuss whether the incidence of resemblances exceeds chance expectation, nor do they present examples of the kind of shared morphological paradigmaticity that would strongly support genetic relatedness. Accordingly, the possibility of external relations to NWC and/or Hurro-Urartean must be regarded as an opinion for which no support has been offered."
 * Nichols further explains that "Nikolayev & Starostin 1994 frequently split Nakh-Daghestanian cognate sets up among two or more putative North Caucasian cognate sets. Nichols 1997 found that over one-third of the then-known secure Nakh-Daghestanian cognate sets were split up and redistributed by Nikolayev and Starostin."
 * The reconstruction provided here is based on regular (Nichols 2003) consonant correspondences that Nichols identifies in the same paper, and regular vowel correspondences worked out in Schrijver (2021 "A history of the vowel systems of the Nakh languages (East Caucasian) with special reference to umlaut in Chechen and Ingush", 2018 "The Origin of Vowel Alteration in Avar-Andi-Dido (North-East Caucasian)")- it is limited because not every vowel phoneme of PNEC has been reconstructed in Schrijver 2021, nor has data from Khinalug, Dargwa, Lak, or Lezghian languages been incorporated into the reconstruction. However, I still think this is better than Starostin's reconstruction, because while Starostin's reconstruction can claim inspiration from Khinalug, Dargwa, Lak, or Lezghian languages in addition to Avar-Andi-Tsezic (or Avar-Andi-Dido) languages, it has the significant disadvantage of rather irregular correspondences everywhere. And those irregular correspondences are used to explain a set of data which contains 2/3 cognates, 1/3 false cognates. Nichols and Schrijver instead proceed from established cognate sets in Gigineishvili and build sound laws off of those, rather than building sound laws off of a wider data set that includes false cognates. This is why, although their reconstruction is less "complete" than Starostin's, their reconstruction is still significantly better methodologically than Starostin's, and unlike Starostin's reconstruction, seems to actually explain a good deal about ablaut and vocalism in Nakh and Avar-Andi-Tsezic languages.
 * Based on this case I think that this Schrijver (2021) based reconstruction should be used instead of Starostin's. Qmbhiseykwos (talk) 02:33, 8 September 2023 (UTC)

Reconstruction:Proto-Germanic/klainiz
Like above, duplicate of which has no certain etymology. -- Sokkjō 01:03, 10 September 2023 (UTC)


 * Delete. Catonif (talk) 10:00, 11 September 2023 (UTC)
 * BUMP -- Sokkjō 06:50, 22 September 2023 (UTC)
 * Deleted. Leasnam (talk) 02:12, 18 March 2024 (UTC)

Reconstruction:Proto-Germanic/winkilaz
Hey, you created Reconstruction:Proto-West Germanic/winkil, but really Reconstruction:Proto-Germanic/winkilaz should have been moved to there. Could you merge the history of PG into the PWG entry and delete the PG entry? -- Sokkjō 04:55, 18 September 2023 (UTC)
 * , How can I do this ? Leasnam (talk) 15:35, 26 September 2023 (UTC)
 * Erutuon laid out the steps . -- Sokkjō 06:45, 27 September 2023 (UTC)

Reconstruction:Proto-Germanic/bermô
Since Reconstruction:Proto-West Germanic/bermō is of uncertain origin, the PG should be deleted. -- Sokkjō 04:03, 20 September 2023 (UTC)
 * Delete, do it, no loss, given the PWG page. Fay Freak (talk) 14:46, 2 May 2024 (UTC)

Reconstruction:Proto-West Germanic/-mō
Delete: I see no evidence of this being productive in Proto-West Germanic. -- Sokkjō 04:26, 20 September 2023 (UTC)
 * I don't think there is enough evidence to even say this was productive in Proto-Germanic. -- Sokkjō 01:31, 21 September 2023 (UTC)
 * Does a suffix have to be productive to be considered a suffix and warrant an entry ? Can we not simply label it nonproductive ? Leasnam (talk) 22:17, 21 September 2023 (UTC)
 * I would say yes. Creating entries for unproductive elements breeds confusion and encourages false reconstructions. -- Sokkjō 06:32, 22 September 2023 (UTC)
 * We do it for other languages, even English. Consistency matters. Leasnam (talk) 15:28, 26 September 2023 (UTC)
 * No it doesn't, and even if everyone did it, that's an . -- Sokkjō 06:51, 27 September 2023 (UTC)
 * I see no reason whatsoever to restrict affix entries to productive affixes. —Mahāgaja · talk 09:21, 27 September 2023 (UTC)
 * You would be attributing meaning to them that the speakers didn't have. All they would see is a root ending in *-m and the suffix . -- Sokkjō 09:37, 27 September 2023 (UTC)
 * Keep I agree with Mahagaja. @Sokkjo, you're describing a situation in which (part of) a suffix has been reanalysed as something else. It would be fair then to no longer call it a suffix. But I don't think that is the case with . Take Modern German for instance. Even now I think synchronically people would segment as blu-me, based on a root blu- which also gives  and . This makes German -me an unproductive and very rare suffix (compare  and, which only differ because they're masculine). See Category:English_unproductive_suffixes. Other English examples, which are nevertheless suffixes, are , ,  (unproductiveness is often debatable of course). —Caoimhin ceallach (talk) 19:38, 8 October 2023 (UTC)

Reconstruction:Proto-Indo-European/létrom
Two descendants which are speculated to be borrowed from one another. -- 04:50, 20 September 2023 (UTC)
 * Delete. Fay Freak (talk) 05:06, 20 September 2023 (UTC)

Reconstruction:Proto-Indo-European/h₁sówr̥
Single Hittite descendant. -- 04:51, 20 September 2023 (UTC)
 * Delete. Fay Freak (talk) 05:06, 20 September 2023 (UTC)
 * Delete. Even Kloekhorst, whose book this comes from, is tentative in this suggestion and doesn't provide any possible cognates. It could easily be an innovation within Hittite.
 * —Mahāgaja · talk 06:52, 20 September 2023 (UTC)
 * RFD-deleted -- 𝘗𝘶𝘭𝘪𝘮𝘢𝘪𝘺𝘪(𝘵𝘢𝘭𝘬) 12:55, 20 May 2024 (UTC)

Reconstruction:Proto-Germanic/þaisimô
Reconstruction:Proto-West Germanic/þaisimō can be reconstructed a couple different ways and there are no Norse or Gothic cognates. -- Sokkjō 05:17, 20 September 2023 (UTC)


 * Agree. Delete. —Caoimhin ceallach (talk) 19:51, 8 October 2023 (UTC)

Reconstruction:Proto-Indo-European/mr̥k-
See etymology of Reconstruction:Proto-Germanic/murhǭ. -- 05:54, 20 September 2023 (UTC)
 * Delete. Fay Freak (talk) 09:17, 20 September 2023 (UTC)

Reconstruction:Proto-Indo-European/h₂rew-
Highly speculative, better suited for an etymology section. -- 06:03, 20 September 2023 (UTC)
 * Delete. Fay Freak (talk) 09:17, 20 September 2023 (UTC)
 * RFD-deleted -- 𝘗𝘶𝘭𝘪𝘮𝘢𝘪𝘺𝘪(𝘵𝘢𝘭𝘬) 12:56, 20 May 2024 (UTC)

Reconstruction:Uzbek/emoq
There is already an Uzbek entry at so this is already an attested term, not a reconstructed one? Unless this is a reconstruction for midevil/pre-modern Uzbek, in which case it should be in the Arabic script not Latin. سَمِیر | Sameer (مشارکت‌ها • کتی من گپ بزن) 17:11, 23 September 2023 (UTC)


 * Also, Uzbek is still written in the Arabic script in Afghanistan and in China. Rodrigo5260 (talk) 18:04, 23 September 2023 (UTC)
 * Keep and delete emoq instead. emoq is completely unattested (as far as I know) in Modern Uzbek and is only the hypothetical infinitive form of its conjugations. Note that the page emoq is of very low quality also, sporting an impossible to confirm pronunciation, an erroneous conjugation table (the verb is defective and takes a very limited number of conjugations) and a poor gloss. Newgrass 82 (talk) 04:18, 5 June 2024 (UTC)
 * Comment Also, Sameer, how did you decide which article was the correct one to delete? Why are you sticking your nose in languages you don't even remotely understand? Newgrass 82 (talk) 04:25, 5 June 2024 (UTC)

= October 2023 =

Multiple Proto-Basque reconstructions
The following Proto-Basque reconstructions lack sources, don't conform to Proto-Basque phonology or are otherwise dubious. Santi2222 (talk) 16:32, 10 October 2023 (UTC)
 * Reconstruction:Proto-Basque/arhe unsourced
 * Reconstruction:Proto-Basque/beh- unsourced, probably based on the hypothesis that and  are related (according to Trask "[this link] cannot be evaluated")
 * Reconstruction:Proto-Basque/caheca unsourced (it doesn't show up in the linked sources), Proto-Basque doesn't allow word-initial . According to the General Basque Dictionary it is onomatopoeic.
 * Reconstruction:Proto-Basque/caio same as above (unsourced, initial ). Mitxelena links it to French and Corominas to Spanish.
 * Reconstruction:Proto-Basque/cedaŕ same as above. Possibly a derivative of, whose conservative form is probably.
 * Reconstruction:Proto-Basque/concoŕ initial . Trask lists it as one of several kV(n)kV- expressive formations.
 * Reconstruction:Proto-Basque/eɫhauso Trask lists modern Basque (standard spelling ) as a variant of the straightforward compound  +
 * Reconstruction:Proto-Basque/gernu contains the unusual cluster, according to Trask "Probably native, with an unrecoverable phonological history".
 * Reconstruction:Proto-Basque/izarr- the only descendants are Romance borrowings. There's no indication that Romance borrowed specifically from Proto-Basque and not from some other substrate language.
 * Reconstruction:Proto-Basque/katena initial . While modern Basque is clearly a (later) Latin/Romance borrowing, Proto-Basque borrowed Latin  as  in general.
 * Reconstruction:Proto-Basque/kaka initial . Probably just a borrowing from Spanish.
 * Reconstruction:Proto-Basque/-kunbe no indication in the given source that the suffix (from, itself from Proto-Basque ) dates from Proto-Basque.
 * Reconstruction:Proto-Basque/sapar Trask lists modern Basque as one of many dialectal variants of.
 * Reconstruction:Proto-Basque/xainx́ unsourced, the palatal fricative and affricate (assuming that's what x and x́ stand for) were not phonemic in Proto-Basque.
 * Reconstruction:Proto-Basque/ziNaguRi unsourced, Proto-Basque nouns with more than two syllables are uncommon.
 * These all look like sound reasons for deletion to me, though I wont vote yet as I know little about the language. In fact I have a question ... do we know at what point the deletion of -n- occurred? Was this before or after the language began to comfortably loan words with initial voiceless stops?  Because katea, while a redlink, is so far as I know a modern Basque word for chain, and if /n/ was lost at a time when we say that the language still only allowed voiced stops in word-initial position, perhaps there were exceptions to this rule. Best wishes, — Soap — 17:07, 10 October 2023 (UTC)
 * A note: it seems the word for chain may have irregularly changed to kate. — Soap — 17:39, 10 October 2023 (UTC)
 * I don't know at what point Basque acquired initial voiceless stops, but according to Trask, the loss of intervocalic happened during the early medieval period and was complete by the early 11th century. Again according to Trask, all instances of Romance  are borrowed as  (with many exceptions in the initial position). Regarding the loss of final, modern Basque  is a reanalysis of  as . Actually,  is attested as an indefinite form, but nowadays it is the singular (definite) absolutive form of  (I don't know if any varieties still have the form with final ). Santi2222 (talk) 18:09, 10 October 2023 (UTC)
 * Also, while Im here, though Im probably not the first to notice, I want to point out that the Basque Etymology link we use on so many pages currently redirects to the website of a mountain resort in Bulgaria. Since the original link was also from Bulgaria, my guess is that it's the same company and that someone at what is now a resort had an interesting side hobby. Nonetheless the link isn't much use if we can't get to the original document, and I wonder if it might have been compiled from a different original source. Thoughts? — Soap — 17:56, 10 October 2023 (UTC)
 * The "Basque Etymology" link used to point to an etymological dictionary by John Bengtson. The main point of that dictionary seems to be to connect Basque with "Proto-Sino-Caucasian" so I seriously doubt we can find anything useful in there.Santi2222 (talk) 18:24, 10 October 2023 (UTC)

Reconstruction:Proto-Germanic/lapilaz
Likely a West-Germanic construction. See RC:Proto-West Germanic/lapil. -- Sokkjō 19:33, 14 October 2023 (UTC)

Reconstruction:Proto-Slavic/tvьrdogolvъ
Proto-Slavic. IP keeps adding what looks like unsourced BS. Vininn126 (talk) 21:48, 17 October 2023 (UTC)


 * @Vininn126 there seem to be reflexes in all three Slavic branches - what's a satisfying explanation for that? Chernorizets (talk) 02:50, 18 October 2023 (UTC)
 * @Chernorizets One problem is the chronologization - For example the Polish reflex is attested rather at the end of the Middle Polish period or beginning of the modern period, which suggests innovation or borrowing. Ideally there should be Old East Slavic and Old Church Slavonic attestations as well, etc. Any source I know that would have information on this word does not claim PS origin at all, making me believe it's common innovation. After all, there are other languages that all innovated such a construction, like hardheaded, which can point to just common innovation overall, suggesting that PS is not needed to explain this word. Vininn126 (talk) 08:18, 18 October 2023 (UTC)
 * Delete pervininn. PUC – 08:29, 18 October 2023 (UTC)
 * Delete, neighbours inspired each other. Fay Freak (talk) 14:47, 2 May 2024 (UTC)

RFD-deleted. Vininn126 (talk) 09:23, 3 May 2024 (UTC)

Reconstruction:Proto-Slavic/sъmyslъ života
Proto-Slavic. Made by IP that is adding what looks like unsourced BS. Vininn126 (talk) 21:55, 17 October 2023 (UTC)
 * Delete, BS reconstruction. PUC – 08:27, 18 October 2023 (UTC)
 * Delete, too philosophical to be credible for the ancient Slavs. Fay Freak (talk) 14:48, 2 May 2024 (UTC)

RFD-deleted. Vininn126 (talk) 09:22, 3 May 2024 (UTC)

Reconstruction:Proto-Slavic/tvьrdogolvostь
Proto-Slavic. More unsourced reconstructions by an IP. Vininn126 (talk) 22:13, 17 October 2023 (UTC)
 * Delete, BS. PUC – 08:28, 18 October 2023 (UTC)
 * Delete, as the adjective. Fay Freak (talk) 14:49, 2 May 2024 (UTC)

RFD-deleted. Vininn126 (talk) 09:21, 3 May 2024 (UTC)

Reconstruction:Proto-West Germanic/ilīk
There's only one descendant. In addition, meant "he" in Proto-Germanic, not "self, same". —Caoimhin ceallach (talk) 23:18, 20 October 2023 (UTC)
 * -- Sokkjō 00:38, 21 October 2023 (UTC)
 * Deleted. Leasnam (talk) 00:51, 21 October 2023 (UTC)
 * What's the etymology of ? I don't think it should have been deleted so quickly. -- Sokkjō 00:54, 21 October 2023 (UTC)
 * Restored, and moved to PWG. Not sure if the RFD applies to this as well. The Old English etymology is equivalent to ang. has a partial cognate in 🇨🇬, 🇨🇬. Not sure at the moment what the best way to handle this is. Leasnam (talk) 01:14, 21 October 2023 (UTC)
 * If it's from PWG, wouldn't the outcome in OE be (or ?) -- Sokkjō 02:47, 21 October 2023 (UTC)

Reconstruction:Proto-Germanic/skuldōną
The Nordic terms are borrowed from Middle Low German, according to R:no:NEO, which is usually the case with so few terms. -- Sokkjō 21:32, 22 October 2023 (UTC)

Reconstruction:Proto-Germanic/plakjō
Reconstruction:Proto-West Germanic/plakkju is dicey enough, no need to make it worse with a Proto-Germanic entry. Merge history into aforementioned PWG entry. -- Sokkjō 04:19, 23 October 2023 (UTC)

Reconstruction:Proto-Indo-European/dléwkus
Proto-Indo-European. A term with descendants only in Latin and Hellenic, of further unknown etymology, not without phonological problems for which some ad hoc solutions have been proposed, often marked with question marks even in relatively early works such as Pokorny's. Two modern sources, square brackets mine: The Armenian is occasionally mentioned by some sources, for which I ping  for input, although judging from the etymology we provide for it, I doubt it's what is going to save the PIE entry. Catonif (talk) 15:57, 24 October 2023 (UTC)
 * Even accepting this proposal though [referring to explainations to the phonological problems], there is little reason for proposing an IE form given that the distribution is limited to neighbors with a long history of contact.
 * Even if it [the explainations to the phonological problems] were correct, we would still not have an etymology for Graeco-Latin *dl(u)ku-. It is likely that we are dealing with a common borrowing from an unknown source.
 * Even if it [the explainations to the phonological problems] were correct, we would still not have an etymology for Graeco-Latin *dl(u)ku-. It is likely that we are dealing with a common borrowing from an unknown source.
 * Even if it [the explainations to the phonological problems] were correct, we would still not have an etymology for Graeco-Latin *dl(u)ku-. It is likely that we are dealing with a common borrowing from an unknown source.


 * The Armenian is super-irregular and is not helpful for your root. I have a more economical explanation for Armenian as my original research, which I have not put in our entry for yet. Vahag (talk) 17:39, 24 October 2023 (UTC)
 * Keep: It's used as a big case study for #dl in PIE and is well sourced because of it. I can think of dozens of PIE entries that should be deleted before this one. -- Sokkjō 16:01, 22 May 2024 (UTC)

== Reconstruction:Proto-West Germanic/kwellan Reconstruction:Proto-Germanic/kwellaną ==

I moved the entry content for RC:Proto-West Germanic/kwellan to RC:Proto-West Germanic/kwalljan, as it's less gymnastics explaining the germination with a strong j-present then a wonky secondary full-grade.

Thus RC:Proto-West Germanic/kwellan and RC:Proto-Germanic/kwellaną should me history merged/deleted into RC:Proto-West Germanic/kwalljan. -- Sokkjō 07:26, 31 October 2023 (UTC)


 * Delete I think its better if it is a j-present. I guessing the germinated entry has to done with the source giving the verb as . Without a non-west Germanic descent there is no telling whenever this verb is a j-present or formed through a germinate. 𐌷𐌻𐌿𐌳𐌰𐍅𐌹𐌲𐍃 𐌰𐌻𐌰𐍂𐌴𐌹𐌺𐌹𐌲𐌲𐍃 (talk) 20:06, 2 November 2023 (UTC)


 * I'm not opposed to remove the Proto-Germanic entry, but I'm not in favor of reconstructing a class 6 j-present as there is no evidence for that at all. All the descendants show a class 3b inflection. Not only Orel but also Seebold reconstructs this verb as such. None of them bother to explain the geminate, but if I were to do an educated guess where it originates from I would say the verb was influenced by semantically related . --DerRudymeister (talk) 20:07, 6 November 2023 (UTC)
 * That's a good point, though j-presents have a way of being rebuilt. Kroonen believes in full-grade iteratives, so I rewrote the etymology using that on RC:Proto-West Germanic/kwellan and reverted RC:Proto-West Germanic/kwalljan. RC:Proto-Germanic/kwellaną should still be deleted, given how unsure this verb is. -- Sokkjō 05:16, 11 November 2023 (UTC)

= November 2023 =

All entries in Category:Proto-Sunda-Sulawesi lemmas
Reconstruction:Proto-Sunda-Sulawesi/aba-h

Reconstruction:Proto-Sunda-Sulawesi/aduh

Reconstruction:Proto-Sunda-Sulawesi/aba

Reconstruction:Proto-Sunda-Sulawesi/azak

Reconstruction:Proto-Sunda-Sulawesi/bagus

Reconstruction:Proto-Sunda-Sulawesi/(h)abu(s)

Reconstruction:Proto-Sunda-Sulawesi/hi(m)pi

Reconstruction:Proto-Sunda-Sulawesi/wad

Reconstruction:Proto-Sunda-Sulawesi/wair

Reconstruction:Proto-Sunda-Sulawesi/wair mata

Reconstruction:Proto-Sunda-Sulawesi/waji

Reconstruction:Proto-Sunda-Sulawesi/waji-ŋ

Reconstruction:Proto-Sunda-Sulawesi/waji-ʔ

Reconstruction:Proto-Sunda-Sulawesi/(w)akar

The overwhelming consensus in discussions both here and at Wikipedia has been that the Sunda Sulawesi language family isn't a valid grouping, so there's no point in reconstructing a proto-language for it. We need to delete these so we can remove this language and its categories from Wiktionary Chuck Entz (talk) 08:23, 10 November 2023 (UTC)
 * Agree. The proto-language apparently was copied from a spurious Wikipedia construct based on a misreading of a source. So not just not valid, but never actually proposed by anyone in the first place. The Wiktionary reconstructions were produced I guess in good faith by an editor who apparently saw the clade in the tree and tried their luck based on cognate sets and wanderwörter in major regional languages of Indonesia that are comprised in that clade. But even if Sunda-Sulawesi was a valid clade, these reconstruction would be mostly wrong and too Malay-centered (e.g., the final *r and *h, where overall evidence from the languages included in the clade would require use to reconstruct *ʀ and *ʔ). –Austronesier (talk) 18:59, 27 May 2024 (UTC)

Reconstruction:Proto-West Germanic/losōn
Proto-West Germanic. I removed the Old English as that comes from a root meaning "loss" shared with Old Norse. The GOH is incorrect, means to "hear, listen"; and  goes back to. The Low German and Dutch look derived from the adjective. Leasnam (talk) 16:52, 10 November 2023 (UTC)

Reconstruction:Proto-West Germanic/narrjō
I am grappling with a couple of issues related to. If this existed in PWGmc, shouldn't we at least see i/j-mutation in German ? German is pretty reliably affected in this regard. Also, it is generally accepted that the Middle Dutch and Middle Low German words are borrowings, the latter especially so since it is first attested in a translation of Luther's Bible.

If is native, and related to, , then the entry would still have the wrong form to give rise to 🇨🇬 and 🇨🇬. Leasnam (talk) 23:16, 20 November 2023 (UTC)
 * Moving the entry to with an unexplained germination is probably the way to go. The term could go back to a frequentative, i.e. . I don't see any strong reason to assume the MDut. and MLG forms are borrowings from MHG. It's first attested in MLG by Agricola, not in Bugenhagen's Bible translation. -- Sokkjō 05:31, 12 December 2023 (UTC)
 * I moved it to RC:Proto-West Germanic/narrō and added Lloyd as a source, who seems to be convinced in a native formation. They also note some Frisian forms. -- Sokkjō 02:07, 13 December 2023 (UTC)

Reconstruction:Old Norse/stríða
Identical to … Should be deleted. Saph668 (talk) 04:04, 21 November 2023 (UTC)
 * ✅ Leasnam (talk) 18:09, 21 November 2023 (UTC)
 * Actually, a reconstruction should be made for the strong verb. Leasnam (talk) 18:13, 21 November 2023 (UTC)
 * I've restored and updated . Please have a look. Leasnam (talk) 18:24, 21 November 2023 (UTC)

= December 2023 =

Reconstruction:Proto-Arawak/-nʊma
Proto-Arawak. The entry is ill-formatted. The Arawak term for "mouth" is apparently roko, which does not resemble this reconstruction - this is of course not fatal, but it is cause for suspicion. This, that and the other (talk) 03:52, 20 December 2023 (UTC)
 * I haven't really worked on Proto-Arawak entries in like 8 years, but I moved it to RC:Proto-Arawak/nama and added a better source and descendants. -- 10:29, 21 December 2023 (UTC)

Reconstruction:Proto-West Germanic/beustameluk
Proto-West Germanic. Tagged by but not listed. This, that and the other (talk) 06:59, 30 December 2023 (UTC)
 * RFD not created by me, but an IP. I just commented out the long rational and asked them to create a RFD discussion.
 * Keep: The IP's rational is that, "the compound is so obvious, it could even have developed independently in various languages." That can be said of countless Germanic terms, and the simplicity of a compound has never been grounds to reject a reconstruction. They also claim that it should also be deleted because it is limited to Continental West Germanic, but again, regionality is also not valid reason for disclusion, see CAT:Regional Proto-West Germanic. -- Sokkjō 10:27, 2 January 2024 (UTC)

Reconstruction:Proto-West Germanic/cibollā
Proto-West Germanic. Tagged by (Victar) but not listed. The edit summary was "not the place from a huge rational". This, that and the other (talk) 07:00, 30 December 2023 (UTC)
 * : look again. It was tagged by a (generally knowledgable) IP, who laid out their rationale- chapter and verse- in the template itself. Sokkyo merely commented out the verbiage with the explanation (if I may paraphrase) that it would be better to present it in an RFD discussion than on the page. Chuck Entz (talk) 07:44, 30 December 2023 (UTC)
 * @Chuck Entz Thanks Chuck, and apologies to Victar. I understand what was meant by that edit summary now. Here's the IP's rationale:
 * "There is no evidence that this old enough to qualify as 'Proto-West Germanic'. In fact there is strong evidence to the contrary: 1.) Not present in English; 2.) First attested in OHG in the 11th century, not attested in any other old language; 3.) The underlying form outside High German is clearly 'cipolla' as no b-forms exist (Middle Dutch 'chibole' is from French [see the source below]; MLG forms with 'zw-' are obvious High German loans); 4.) Even in High German it's probably 'cipolla', p-forms are attested in MHG. The -b- comes from 🇨🇬. If so, it lacks the first phase of the High German consonant shift, which would automatically exclude it. -- But either way, there's nothing: No English, no unity in form, no early attestation." This, that and the other (talk) 08:50, 30 December 2023 (UTC)
 * 1. There are plenty of valid PWG reconstructions without an Old English form, see CAT:Regional Proto-West Germanic for some examples. 2. Many PWG reconstructions are only attested in OHG simply because OHG is the most well attested. 3. What often happens is Middle English and Dutch are reenforced by Old French, so you'll find quite a few ME terms deriving both from OE and OF. For the discussion on ML /c/ > PWG, see 4. Not sure the point they're trying to make here, as there is no "p-form" in HG. -- Sokkjō 04:03, 31 December 2023 (UTC)
 * Keep. The IP arguments are kind of non sequitur. “-b- comes from 🇨🇬.” – could have been even more so in PWG? Dialectal variation /p/ ~ /b/ in early times for such a lovely smol plant can be expected. MLG onsets can also be pseudo-High-Germanisms, like when a word was borrowed from Romance but perceived as borrowed from High German, because the High-German-speaking Franconians ruled France. All arguments are much less pressing than IP thinks.
 * It is also not comfortable to assume that in the 5th century CE, after Germanic invasions in Rome but before the High German consonant shift, the only word for onions was . Wasn’t Albanian also borrowed before even /k/ in Latin was affricatized? Don’t know what the “uncertain” part is. At the same time, for comparison, Arabic  long was borrowed in addition to, such important plants are internationalisms in earliest possible antiquity. In Africa south of the desert when Arabs came they spread the onion. The Slavs I think only knew ramsons before borrowing  to name the onion, yes, in Proto-Germanic times European nutrition was depressing, in Late Antiquity less so. Fay Freak (talk) 04:34, 31 December 2023 (UTC)


 * As for the ultimate origin, the comparisons to etymologize Latin cēpa are of course garbo. With Welsh cennin going back to compared to 🇨🇬, the equation with its singulative is only motivated to claim PIE inheritance; the Albanian one is contradictory when at the same time it is in the descendants section. I might have chosen to leave it that, some contradictions cannot be resolved, but I ween that it is an Anatolian borrowing cognate to, hence the variation with  and  only used in , and weird indeclinable  in addition to ; anno urbis conditae hence the food with the Italics was also depressing. Fay Freak (talk) 04:54, 31 December 2023 (UTC)
 * If kept, rename to *tsibollā. Using a single letter implies that this is a distinct phoneme, which I previously expressed doubts about. It's more likely that native speakers treated it as a consonant cluster, albeit one that doesn't exist natively. Similar to how non-Slavic speakers will treat Slavic "c". —Rua (mew) 10:16, 9 February 2024 (UTC)

Reconstruction:Proto-West Germanic/pundu
Proto-West Germanic. A creation of Victar/Sokkjo, tagged by with the edit summary "one descendant with unknown gender". I have to say I tend to agree. This, that and the other (talk) 07:01, 30 December 2023 (UTC)
 * You're overlooking the derived term. -- Sokkjō 04:03, 31 December 2023 (UTC)

Reconstruction:Proto-Germanic/dulją
Proto-Germanic. Tagged by with the reason "PWG entry  already exists and is on unknown origin", but not listed. This, that and the other (talk) 07:09, 30 December 2023 (UTC)


 * Tagging the entry creator @Leasnam.
 * Not sure this can really be taken back to Proto-Germanic. If it helps, I can comment on the Romance series. The latter don't really suggest much of a time depth either, as their distribution mostly follows follows the zones of contact with Germanic. The exception is one 'prong' that has penetrated through Occitan and into Basque and Catalan, but that doesn't do much to change the overall picture. Overall we are well within the range of typical for Frankish borrowings (probably with some degree of lexical "piggy-backing" via French). Mind, I've not yet sorted out the details of the various vowel outcomes. Nicodene (talk) 14:25, 19 February 2024 (UTC)
 * I don't have a great deal of objection to removing this entry; however, there is also the 🇨🇬 which suggests a 🇨🇬, with 🇨🇬 being a diminutive/derivative of it. I tend to think that PWG terms with stem-final j-gemination likely occurred in PGmc and were inherited as fossilised irregular nouns rather than gemination being part of a productive paradigm in PWG. Is it fair to say this or am I missing something ? Leasnam (talk) 17:36, 19 February 2024 (UTC)

= January 2024 =

Reconstruction:Proto-Malayo-Polynesian/qaRutay
There are no listed descendants in the entry. Also, the thesis linked in "Further reading" itself does not support this reconstruction (p. 150). Ελίας (talk) 11:55, 1 January 2024 (UTC)

Reconstruction:Proto-Italic/swerjos
Disputed, phonetically problematic reconstruction with only one direct descendant. — 69.120.69.23 00:40, 11 January 2024 (UTC)


 * Delete. This does not deserve its own page as the etymology is already discussed more in depth on the Latin page and there are no other Italic cognates. - saph 🍏 14:01, 11 January 2024 (UTC)
 * should result in, surely. Delete. Nicodene (talk) 14:02, 19 February 2024 (UTC)

Reconstruction:Proto-Slavic/ablušьko
Proto-Slavic. One branch does not Proto-Slavic make, especially since Czech and (second) Upper Sorbian don't work, the header is broken, and four descendants (of which two are Sorbian) can easily innovate this later. Thadh (talk) 15:54, 16 January 2024 (UTC)


 * Keep. Fixed errors and added more material. One branch can make Proto-Slavic and that view is popular in Slavistics. Sławobóg (talk) 16:27, 16 January 2024 (UTC)

Reconstruction:Proto-Hellenic/pʰílos
1 descendant. Same policy from Italic should be applied to Hellenic. - saph 🍏 06:53, 17 January 2024 (UTC)


 * What constitutes a valid Proto-Hellenic entry, for the record? One where the Greek descendant is found in sufficiently divergent forms in multiple dialects? Nicodene (talk) 13:49, 19 February 2024 (UTC)
 * Presumably that or a Mycenaean cognate. - saph 🍏 14:05, 19 February 2024 (UTC)
 * Right.
 * Delete, in any case. Nicodene (talk) 21:34, 25 February 2024 (UTC)

Reconstruction:Proto-Japonic/wənə
In the Nihon Ryōiki, it's attested as 乎乃 according to JDB p. 837. Apparently the Nihon Ryōiki perserves ko otsu distinction of no. @Eirikr Chuterix (talk) 19:10, 18 January 2024 (UTC)


 * Considering that the ko-otsu vowel differences were merged in the 800s, and that the  probably wasn't completed until 824, and that all manuscripts are apparently incomplete with the oldest dating to 904, this is all potentially late enough that I'd want to be certain that the spellings there actually reflect ko-otsu vowel values, and not post-merger spellings where the older distinctions were already lost.
 * Given also the possibility that wo might have only represented, for which Arisaka's law would necessitate that the following o would have to also be , we should make sure to confirm. ‑‑ Eiríkr Útlendi │Tala við mig 21:05, 18 January 2024 (UTC)

Reconstruction:Proto-Balto-Slavic/(s)prénˀstei
Proto-Balto-Slavic. No sources, added by IP who has (re)added unsourced material that did not survive RFD previously. Vininn126 (talk) 20:14, 20 January 2024 (UTC)
 * I've added sources for this one at least. -- 02:47, 21 January 2024 (UTC)

Reconstruction:Proto-Indo-European/kʷeh₁y-
I don't think we should start creating entries for alternative reconstructions to PIE roots. That information can be just kept on the main entry, like seen on RC:Proto-Indo-European/péysks and RC:Proto-Indo-European/h₃merǵ-. -- 00:14, 24 January 2024 (UTC)


 * @Victar not disagreeing, but is there a reason why you say this? Specifically, what I'm thinking is:
 * This is how attested languages handle things, albeit that "alternative reconstruction" is different from "alternative forms"
 * Different dictionaries have different reconstructions, which means that either a user may try to look up an alternative reconstruction and see page-not-found or words in the daughter languages might link to a page that wouldn't exist (e.g. has  but  has ). It's helpful if these links are active.
 * I would be in favor of a template similar to like  that simply changes the word used to make it more correct. Dragonoid76 (talk) 05:39, 24 January 2024 (UTC)
 * I concur with Dragonoid here, alternative-form-style soft redirect entries are useful for reconstructions given how variable they can be between authors at times. — Ceso femmuin mbolgaig mbung, mellohi! (投稿) 15:31, 24 January 2024 (UTC)
 * I think such a template would be a good idea. I agree with @Victar that all information should be kept in one entry. —Caoimhin ceallach (talk) 15:33, 24 January 2024 (UTC)
 * I also agree with Victar that there should not be separate entries for alternative reconstructions, but hard redirects instead. The difference between reconstructions and attested entries is that mainspace entries cover multiple languages, while Reconstruction entries are language-specific. It makes much more sense to keep our discussion of a controversial reconstruction on a single page rather than spread across pages. —Mahāgaja · talk 15:51, 24 January 2024 (UTC)
 * Any redirect is fine by me, just not straight deletion. — Ceso femmuin mbolgaig mbung, mellohi! (投稿) 16:41, 25 January 2024 (UTC)
 * To come down clearly, my comment above is to be interpreted as a vote to change to a hard redirect. —Mahāgaja · talk 19:49, 25 January 2024 (UTC)
 * Keep. Useful for looking up reconstructions from papers which do not agree with the 'mainstream' reconstruction; though, flat-out redirects would have the same effect with less effort to create and without losing that ability to easily look up alt reconstructions, so in the future we should probably tend towards redirects. - saph 🍏 19:15, 25 January 2024 (UTC)

Reconstruction:Proto-Indo-European/lúkʷos
Same as above, made presumably by the same user. -- 03:07, 24 January 2024 (UTC)


 * Keep. Etymology section on here probably would not fit in on the main page. It's also not really an alternative reconstruction, more like an  alternative form. - saph 🍏 19:19, 25 January 2024 (UTC)


 * Convert to a hard redirect. —Mahāgaja · talk 19:50, 25 January 2024 (UTC)

= February 2024 =

Reconstruction:Proto-Samic/čājē
Proto-Samic. Entries with no information other than derived terms. This, that and the other (talk) 00:17, 5 February 2024 (UTC)
 * Delete. Should be kept in etymology sections. - saph 🍏 18:21, 7 February 2024 (UTC)
 * Keep: We've never deleted reconstructions just because they only have derived terms. -- 21:20, 7 February 2024 (UTC)
 * They don't even have meanings... what is the point of such an entry? This, that and the other (talk) 09:46, 8 February 2024 (UTC)
 * I've fixed the templating. It's a noun with two denominative verbs, but yes, it lacks a meaning. I could venture to guess, but I'll let the Sami editors, ex. User:Rua, figure that out. -- 17:47, 8 February 2024 (UTC)
 * Okay, that's better. I'll close this discussion as far as it concerns *čājē. This, that and the other (talk) 00:13, 10 February 2024 (UTC)
 * Keep - two derived terms with productive suffixes is usually enough to reconstruct a base noun at some stage of the language. Thadh (talk) 18:17, 8 February 2024 (UTC)
 * It seems to me that in the case of there is only one derived term, as the two terms listed are alternative reconstructions of the same term. This, that and the other (talk) 00:11, 10 February 2024 (UTC)
 * In that case we need to fix the entries for the derived term and move everything there. Thadh (talk) 00:53, 10 February 2024 (UTC)
 * Keep, so long as the derived terms are valid. Nicodene (talk) 11:26, 9 February 2024 (UTC)

Reconstruction:Proto-Hellenic/tʰéwō
Proto-Hellenic. One descendant, same policy should be applied to Hellenic as to Italic. Bundling this in with all the other one-descendent Proto-Hellenic pages. - saph 🍏 18:55, 6 February 2024 (UTC)


 * See also Requests for deletion/Reconstruction. - saph 🍏 18:57, 6 February 2024 (UTC)

Reconstruction:Proto-Slavic/směati
Proto-Slavic. Tagged for speedy (!) deletion in September 2023 with the reason "misspelling of ". This, that and the other (talk) 09:45, 8 February 2024 (UTC)


 * Not sure what the issue is (or how this constitutes a 'misspelling'?). From what I can tell, OCS reflects this form. - saph 🍏 16:42, 8 February 2024 (UTC)

Reconstruction:Proto-Italic/wergō
Proto-Italic. 1 Latin descendant. - saph 🍏 21:57, 17 February 2024 (UTC)
 * Not to mention created by an IP editor notorious for making stuff up, unreferenced, and unedited since by humans except for minor formatting tweaks. As it stands now, Delete. Chuck Entz (talk) 23:10, 17 February 2024 (UTC)
 * Delete. Nicodene (talk) 13:43, 19 February 2024 (UTC)
 * RFD-deleted -- 𝘗𝘶𝘭𝘪𝘮𝘢𝘪𝘺𝘪(𝘵𝘢𝘭𝘬) 03:36, 21 May 2024 (UTC)

Reconstruction:Gaulish/abalo-
Gaulish reconstruction with no descendants or derived terms. —Mahāgaja · talk 21:28, 18 February 2024 (UTC)


 * Not a reconstruction, this is attested as avallo in Endlicher's glossary. Keep and move out of reconstruction namespace. - saph 🍏 13:11, 19 February 2024 (UTC)
 * I have added the quote. User  moved this page and formerly claimed an alternative spelling while admitting the fruit is attested. Against the cognitive dissonance I point out a recent discussion about lemmatization of Trümmersprachen attested in transcriptive mentions.  has to move back; avallo or avallon are likely options for me, but inconsistent with the tree name kept at, which should perhaps be moved across the reconstruction namespace to . On the other hand one can normalize at b with the argument that the u in the manuscript stands for /β/ already, depending a bit on what Mahagaja thinks the Gaulish sound. Fay Freak (talk) 15:13, 19 February 2024 (UTC)
 * If it's attested as, then obviously that's what the entry needs to be named. The gloss (which seems to be plural, unless Medieval Latin has made a feminine noun out of it) isn't sufficient to know whether the fruit or the tree is being referred to. —Mahāgaja · talk 16:51, 19 February 2024 (UTC)
 * I mean hyphenation has not developed in the 5th century CE to designate a stem or root to be expanded by endings. It is possible that the author of the glossary abstracted the endings away but we should add something for it to be comprehensible to modern readers; we do so in the inflection table anyway, but if it is the tree then the table is unfounded. Actually I tend to assume that it is the tree, due to the topographic content of the glossary overall, so move to avallos (where s is the inflectional ending that we don’t separate in titles by hyphens, and add the URL which I forgot when formatting the quote)! Otherwise we don’t know which is to be had in the mainspace and which a reconstruction! Fay Freak (talk) 17:57, 19 February 2024 (UTC)
 * Whatever this is, it's unlikely to be a masculine ending in -os, since the Proto-Celtic word for the fruit is and the word for the tree is . The -ll- (which can come from *-ln-) makes the tree more likely, while the -o makes the fruit more likely. But I see no reason at all to put the entry at anything other than the attested spelling . —Mahāgaja · talk 18:05, 19 February 2024 (UTC)
 * What about ? Would be artificial though, nothing compellingly reasonable I see. -ll- is a weak argument for the tree, and even in spite of the topographic content of the glossary a Roman might have heard the name of the fruit and assumed it to equal the name of the tree, as in Latin usually only the ending differs for fruit and tree, just like I did it right now. And given the Proto-Celtic tree word both reconstruction pages are unreasonable and should be deleted. Due to the Romance meanings of poma I assumed an apple fruit first so we have it attested and the tree would be different. Fay Freak (talk) 19:04, 19 February 2024 (UTC)
 * Dear all, So in French studies of the Gaulish language, there is a distinction between :
 * the attested form of Late Gaulish avallo which is translated as poma in Endlicher's wordlist. avallo is a late form with spirantisation of b into a v, and the loss of the final -s at the nominative singular seen in Late Gaulish.
 * *abalo-, for apple : is the reconstructed based on the above term and many descendents in Latin and onomastic studies of Celtic base toponyms -> acerabulus (leading to the French érable), the many Alon, Avalon, Valuéjols and Valeuil in France or known toponyms like Αβουλα (given by Ptolemee in Spain), Abulobrica or personal names like Abulus, Abalanis, Abbula, etc. Most probably neuter.
 * *aballo-, for apple tree : most likely one of the few -o- words being feminine in Gaulish. Like in Brittonic, there was likely a distinction aballo- / aballo- for apple / apple tree in Gaulish, probably lost in the late form of the language. Attested in current placenames like Ollon, Avalleur, etc. and in former proper names like Abullius, Abelus, Abullus, etc.
 * The final "-" is part of the reconstruction of the Gaulish form as it indicates that we don’t know for sure whether it was -on and -os.
 * Kind regards, Treehill (talk) 20:04, 19 February 2024 (UTC)
 * I hardly think the compiler of Endlicher's glossary listed a noun stem in his word list rather than a whole word. It's probably avallom with the final m in the process of being absorbed into the preceding vowel (as happened in Vulgar Latin) on the way to the final syllable being lost (as happened throughout Insular Celtic). (There was no &#8209;s because the word was neuter, not masculine.) The fortis ll is phonologically regular in, but not (AFAIK) in Gaulish. Perhaps there was influence from the word for the tree *aballā on the word for the fruit *abalom, making the latter pick up an unetymological fortis/geminate ll. —Mahāgaja · talk 20:10, 19 February 2024 (UTC)
 * So according to Treehill there was actually an aballo in Late Gaulish writen avallo by a Roman due to Romance spirantization but the for earlier stages we can reconstruct two more pages? Thanks both for not shirking to add complications. 😣 Fay Freak (talk) 20:48, 19 February 2024 (UTC)
 * Dear User:Mahagaja, I think you misunderstood me, avallo, is not a noun stem. It is the Gaulish word as attested by the compiler in a very late form of Gaulish. The "o" is the final letter. The two others are Classical Gaulish forms reconstructed from an earlier form of Gaulish whose ending is uncertain *abalo- (was likely an -on word) or *aballo- (likely ended in -os). So the three entries as they exist now are the only one attested. Writing avallon (or worse, avallom) would be an incorrect anachronism putting a late spirantised letter with an earlier form of the nominative singular.
 * Same with aballo User:Fay Freak as it would do the opposite -> a pre-spirantisation "b" of earlier forms of Gaulish, with a late nominative singular.
 * The three entries should remain as they are, as stated in the sources indicated : avallo, Reconstruction:Gaulish/abalo- an Reconstruction:Gaulish/aballo- are the only correct forms. Treehill (talk) 21:52, 19 February 2024 (UTC)
 * My comment about the noun stem was in response to Fay Freak's suggestion of using with a hyphen as the page name. What is the evidence that Gaulish had an o-stem noun aballos? RC:Gaulish/aballo- also lists no descendants or derived terms and should therefore also be deleted. —Mahāgaja · talk 08:21, 20 February 2024 (UTC)
 * Descendants were added as per your request. Treehill (talk) 10:14, 20 February 2024 (UTC)
 * Same as below. —Caoimhin ceallach (talk) 13:19, 20 February 2024 (UTC)

David Stifter wrote an article (in German) about the Celtic word for apple: | An apple a day.... The Gaulish forms are discussed in the first section. About Endlicher's glossary he says:
 * "de"

- Zu verschiedenen Anlässen habe ich argumentiert, dass Endlichers Glossar kein Zeugnis eines noch gesprochenen Spätgallischen ist, sondern dass es sich um eine gelehrte Wörtersammlung nach dem Sprachtod des Gallischen handelt, die zum größeren Teil auf Exzerpten aus spätantikem lateinischen Schrifttum beruht. Sein Zweck war die antiquarische Erklärung gallischer Ortsnamen bzw. Ortsnamenelemente (so auch Blom 2011: 177–181).

So while it's a valueble source and pertinent to the entry, I don't think we can use Endlicher's glossary as "attestations" in the strict sense. Therefore both the Gaulish words for apple and apple tree should be considered reconstructed. —Caoimhin ceallach (talk) 14:08, 20 February 2024 (UTC)
 * No, what follows is that the attestation is not an additional Late Gaulish in addition to two earlier Gaulish reconstructions. A mention excerpted from an earlier mention. Hence one of the pages has to be moved to the mainspace. Unless we assume the part was reconstructed from toponyms in the beginning, hence the ending at a morpheme boundary. Fay Freak (talk) 15:02, 20 February 2024 (UTC)
 * Delamarre writes in his Dictionnaire gaulois français :
 * "fr"

- *abalo-, *aballo- : pomme, pommier Mot qui apparait dans le glossaire de Vienne sous la forme avallo, glosé poma avec déjà spirantisation du b en v. [...]


 * He then goes on listing descendents and cognate in other IE languages.
 * Savignac writes : "fr"

- pomme nf, abalo- (neutre) Mot donné par le Glossaire de Vienne sous la forme avallo, glosé poma « pomme », avec spirantisation du -b- en -v-, et déduit du nom propre Abal(l)us, comparable au viel irlandais ubull (de *ablu-), gallois afal, breton aval « pomme ». Remonte à une forme *ablu-/*abol- « pomme », d'où viennent le lituanien óbuolas, vieux slave abluko, vieux haut allemand apful, allemand Apfel, anglais apple « pomme », cf la glose thrace dinupula (de *kun-abola « pomme à chien »)


 * Best Treehill (talk) 17:32, 20 February 2024 (UTC)
 * I read all of the above again and the Beer Parlour thread you linked to, but I'm still not following. Which page has to be moved to the mainspace and why? —Caoimhin ceallach (talk) 22:48, 20 February 2024 (UTC)

Reconstruction:Gaulish/aballo-
Adding this term as it also has no descendants or derived terms. —Mahāgaja · talk 08:23, 20 February 2024 (UTC)


 * As per your request, attested descendants and derived terms have been added. Treehill (talk) 10:15, 20 February 2024 (UTC)
 * I reordered them and added Wikipedia links for placenames. Is this a good way of doing that? —Caoimhin ceallach (talk) 11:50, 20 February 2024 (UTC)
 * Thanks, but since this form is attested in the form it should be moved there (main space, not Reconstruction space). —Mahāgaja · talk 13:28, 20 February 2024 (UTC)
 * Savignac writes : "fr"

- pommier nm, aballo- Mot déduit de NP Abellus, Abullius ..., du nom de lieu Abella, en Campanie, qualifiée de malifera « porte-pomme » par Virgile, d'où vient le terme aveline « grosse noisette », Aballo, devenu Avallon (Yonne) ..., comparable au vieil irlandais aball, gallois afall, breton avallenn « pommier ». Clairière-des-Pommiers *Aballo-ialon, Avaloiolum, Valuéjols (Cantal) et Valeuil (Dordogne, Eure), La Pommeraie Aballo, Avallon, Ollon (Drôme, Avalono, en 1 252), Marché-du-Pommier *Aballo-duron, Avalleur (Aube). Remonte à une forme celtique *aballo-, issue d'un plus ancien *abalnosl*abalna « pommier », dérivé en *-no- du nom de la pomme *ablu-/*abol-.).


 * Best Treehill (talk) 17:32, 20 February 2024 (UTC)

While the minds are focussed on Celtic apples, maybe anyone is interested in this: Requests_for_moves,_mergers_and_splits. —Caoimhin ceallach (talk) 00:37, 21 February 2024 (UTC)

Keep both. In the reconstruction namespace. —Caoimhin ceallach (talk) 02:12, 1 June 2024 (UTC)

Reconstruction:Dacian/aba
No descendants and we know little enough from the language that we can't conclusively reconstruct anything. Bogdan (talk) 12:28, 21 February 2024 (UTC)


 * Delete. - saph 🍏 13:22, 21 February 2024 (UTC)
 * Speedy delete. Fay Freak (talk) 19:44, 25 February 2024 (UTC)
 * Delete. Nicodene (talk) 21:32, 25 February 2024 (UTC)
 * Speedied. Thadh (talk) 22:35, 25 February 2024 (UTC)

Undeletion of Reconstruction:Proto-Dravidian/poẓal
Now that there are codes for CAT:Proto-South Dravidian language and CAT:Proto-South Dravidian I language, can this entry be undeleted or recreated? Kutchkutch (talk) 03:50, 24 February 2024 (UTC)


 * But it only has two cognates. Wait for now. Illustrious Lock (talk) 13:53, 24 February 2024 (UTC)

Reconstruction:Old English/ons
Redundant entry of, the expected outcome in OE as a result of the , inherited from Proto-North Sea Germanic. -- Sokkjō 03:44, 27 February 2024 (UTC)


 * Counterpoint: Onswini is attested on the Collingham Stone(s), by a runic inscription reading "æfter, onswini cu(ning)", showing that an older varient of the name Ōswine was used during the early OE period, thus allowing an older form of the term ōs to be fossilized. Æzelf89 (talk) 04:10, 27 February 2024 (UTC)
 * Archaic feature are sometimes retained in proper nouns, but regardless, it isn't evidence of it surviving in this term. -- Sokkjō 05:23, 27 February 2024 (UTC)
 * That's fair. However, I don't feel comfortable simply deleting the page all together. Same with . I think it would be better to note both of them as archaic, and state in both pages that both words would've only survived within given names and not in normal usage. Æzelf89 (talk) 15:05, 29 February 2024 (UTC)
 * You're also duplicating content, which, even if it was a valid alternative, is not how to deal with alternative forms, see . -- Sokkjō 19:57, 29 February 2024 (UTC)
 * Fair, fair. I'll take care of that and edit them accordingly. Æzelf89 (talk) 20:00, 29 February 2024 (UTC)


 * , I moved to to, as that is how it is attested. It also alludes to that possibility of this simply being an intentional archaism, as is quite common with late inscriptions, cf. the Old Persian Artaxerxes inscriptions.
 * Also, I want to point out again that given and place names are dubious sources for term attestations. Names like Ansigar and Aslac are likely to be foreign names of immigrants to England, see Migrants in Medieval England, c. 500-c. 1500. The moneyer Aslac, which you cite as a source for, was most certainly a Dane, as the coin which names him was minted under Danelaw for King Cnut. And the memorial coinage the moneyer Ansigar appears on was also likely a foreigner from continental Europe, see . -- Sokkjō 19:38, 1 March 2024 (UTC)
 * , pinging you again. -- Sokkjō 17:01, 5 March 2024 (UTC)
 * still hasn't responded., can you read the above and take action? -- Sokkjō 06:50, 31 May 2024 (UTC)

Reconstruction:Old English/Gydda
Only the names Giddinge and Gyddingas are attested, which might derive from a personal name, but it's not certain. Best left to an etymology section. -- Sokkjō 05:49, 27 February 2024 (UTC)
 * , pinging you. -- Sokkjō 06:52, 31 May 2024 (UTC)

Reconstruction:Old English/gydda
Like above, not strong enough evidence to reconstruct this term. -- Sokkjō 05:51, 27 February 2024 (UTC)

Reconstruction:Proto-Germanic/wulkaną
This should be merged into. -- Sokkjō 03:08, 28 February 2024 (UTC)

Reconstruction:Old English/ans
Same as Reconstruction:Old English/ons above. -- Sokkjō 19:39, 29 February 2024 (UTC)

Reconstruction:Old English/as
Again, per above. -- Sokkjō 22:19, 29 February 2024 (UTC)

= March 2024 =

Reconstruction:Proto-Ugric/kälä-
A Proto-Ugric word with a solid Proto-Uralic etymology with no morphological change - we don't have these, but only terms which are not inherited from Proto-Uralic. &mdash; S URJECTION / T / C / L / 08:49, 2 March 2024 (UTC)
 * "to wade" > "to rise" does not seem solid. Aikio can rely on Khanty kül- "arise, go to shore". Okay. The different meaning still justifies an entry, whether as separate definition, ety or lemma, still. 2A02:3033:206:9573:F921:E292:BEAC:54D1 23:12, 2 April 2024 (UTC)
 * It doesn't. The semantic shift can be recorded in the Proto-Uralic entry by giving glosses. &mdash; S URJECTION / T / C / L / 06:13, 3 April 2024 (UTC)

Reconstruction:Proto-Indo-European/(s)neg-
Dubious. See rational at Reconstruction:Proto-Germanic/sneganą. -- 09:07, 2 March 2024 (UTC)


 * And how does that invalidate the other proposed descendants...? - saph 🍏 00:49, 30 March 2024 (UTC)
 * Delete. Much too uncertain. is probably not related. —Caoimhin ceallach (talk) 11:00, 24 April 2024 (UTC)
 * On what basis do you deny its cognacy? - saph 🍏 11:01, 24 April 2024 (UTC)
 * See the etymology in the entry. —Caoimhin ceallach (talk) 11:48, 24 April 2024 (UTC)
 * Again, *sneganą is not the only proposed relation here. I don't see a reason why *(s)nogHō > *snakô "snake" and *(s)nogós > *nāgás "snake" is so unreasonable. - saph 🍏 13:57, 24 April 2024 (UTC)
 * It is not unreasonable in and of itself, but the other etymology is better because a) it accounts for the meaning "elephant", which is old, and b) it has a precise Lithuanian cognate. With that, the PIE basis for *sneg- is gone. —Caoimhin ceallach (talk) 14:37, 24 April 2024 (UTC)
 * "naked" > "snake, elephant" seems like a bit of a stretch. - saph 🍏 15:56, 24 April 2024 (UTC)
 * I wouldn't know why. Words meaning "naked" are often used to describe hairless animals. In English too, e.g. . —Caoimhin ceallach (talk) 08:45, 25 April 2024 (UTC)
 * As a descriptor. You wouldn't just say "naked," no? - saph 🍏 12:00, 25 April 2024 (UTC)
 * In English, adjectives aren't freely substantivised. But in other languages this is normal and common. Take for instance, , . —Caoimhin ceallach (talk) 12:36, 25 April 2024 (UTC)
 * Has this ever happened in Sanskrit? Why would it develop into two only vaguely similar concepts, anyway? - saph 🍏 15:59, 25 April 2024 (UTC)
 * Substantivization of adjectives is extremely common in Sanskrit, aided by the fact that adjectives and nouns aren't distinguished at all in inflection. I would even go as far as saying that most adjectives also have a substantival meaning. Eg., ,.
 * Anyway, originally meant "elephant". It was later used to mean "snake", but originally only in contexts where its mythological closeness to the elephant was to be emphasised. For example in the Mahābhārata the king of the snakes (nāga) is, the divine elephant. (See P. Thieme, Zeitschrift für vergleichende Sprachforschung auf dem Gebiete der Indogermanischen Sprachen, vol. 78, no. 3/4, p. 178, fn. 1) —Caoimhin ceallach (talk) 10:30, 26 April 2024 (UTC)
 * I wonder if the semantic shift 'elephant' > 'snake' really comes from them both being hairless, or if it has to do with the similarity of an elephant's trunk to a snake. —Mahāgaja · talk 10:53, 26 April 2024 (UTC)
 * That's Wackernagel's theory too, from *nāgahasta lit. "snake-arm" or so, like the Latin . But in Vedic nāga (almost?) always means "elephant", not "snake". —Caoimhin ceallach (talk) 11:19, 26 April 2024 (UTC)
 * Delete, then. - saph 🍏 12:50, 26 April 2024 (UTC)
 * @Caoimhin ceallach you can't just call for pages to be deleted based on your own personal speculation that it's "much too uncertain". This reconstruction is attested in reliable sources. At a minimum you have to provide other reliable sources that dispute this reconstruction, but even then that would not necessarily be enough to delete the reconstruction, only to problematize it. Brusquedandelion (talk) 00:25, 12 May 2024 (UTC)
 * I added a reliable reference at the entry for, contesting this reconstruction. Contrary to what you say, this article has no references. If you have references for , please add them. If you want more doubt that this reconstruction is correct, check , which also connects , , and with , which are all irreconsilable with a root meaning “crawl”. Some root meaning “pointed” or “stinging” is more likely. —Caoimhin ceallach (talk) 15:26, 12 May 2024 (UTC)

Reconstruction:Proto-West Germanic/talentā
I hate to have to do this, but I highly doubt this term was present in PWG the way we show in the entry. Several issues exist: the Old High German lacks the shift from PWG t- to z-; the OHG declension was strong rather than weak; the Middle Low German term was neuter and a known borrowing from the Latin talentum. Therefore each language borrowed the term independently. Leasnam (talk) 20:04, 4 March 2024 (UTC)
 * Happy to go over it. 1. There are many exceptions to OHG *t > z and *p > pf found in Latin borrowing. Compare OHG, from PWG , from , which underwent fricativization of the the *k but not the *t. I assume this is under the influence of the Latin term, but I'm open to other suggestions. 2. The vacillation of ō- and ōn-stems is seen in native terms too, so I'm not overly concerned by that, especially with a borrowing. 3. How is the MLG a known direct borrowing Latin? -- Sokkjō 20:47, 4 March 2024 (UTC)
 * Köbler states is a "Fremdwort". Leasnam (talk) 21:25, 4 March 2024 (UTC)
 * I mean, it's definitely a foreign word... I added the Dutch as well. -- Sokkjō 23:56, 4 March 2024 (UTC)
 * No. When he uses it it means it's been borrowed into GML from another language. Leasnam (talk) 00:44, 5 March 2024 (UTC)
 * Not sure why you added a reference on the page that doesn't support the reconstruction: J. de Vries (1971), Nederlands Etymologisch Woordenboek, Leiden talent znw. o., mnl. talent ‘geldswaarde; geestesgave’ <lat. talentum < gr. tálanton. De bet. overdracht naar ‘aangeboren geestesgaven’ berust op de parabel in het NT Matth. 25, 14 vlgg., Luk. 19, 12 vlgg. Leasnam (talk) 00:52, 5 March 2024 (UTC)
 * And yet Köbler gives no reason. The citation is on the etymology, not the term. -- Sokkjō 00:56, 5 March 2024 (UTC)
 * It's toast. Don't add it again with out a credible source that suggests this was a Proto-West or Proto-Germanic term. Leasnam (talk) 01:07, 5 March 2024 (UTC)
 * Another explanation for tunihha and the like is that the dialects that borrowed it had no intervocalic single [k] at the time, only [x]. —Caoimhin ceallach (talk) 09:49, 24 April 2024 (UTC)

Woof, just deleted unilaterally. So much for the RfD process. -- Sokkjō 01:07, 5 March 2024 (UTC)
 * Others have an opportunity to voice their opinions. If it's deemed justified it can easily be reinstated. Your edits are illogical and my confidence in the work is now shaken. Leasnam (talk) 01:11, 5 March 2024 (UTC)
 * Deleting the entry, you've not given others the chance to even look at its contents. Its deletion seems to me an emotional outburst borne of frustration, and not actually made in good faith and trust in the process. -- Sokkjō 01:21, 5 March 2024 (UTC)
 * Undeleted, as it was indeed deleted out of process and at least deserves its due discussion term of one month. That said, I also lean delete, the arguments for its reconstruction seem very weak. The source cited also does not support the reconstruction, as Leasnam has noted. — Mnemosientje (t · c) 12:09, 19 March 2024 (UTC)

Reconstruction:Proto-West Germanic/nunnā
Being an ecclesiastical term, I don't think this can go back to PWG. OE and OHG were independent borrowings. OHG nunna is not attested till the latter half of the 9th century (?) Leasnam (talk) 20:24, 4 March 2024 (UTC)
 * Ecclesiastical Latin still falls in the same timeframe as Proto-West Germanic. -- Sokkjō 20:51, 4 March 2024 (UTC)
 * Parallel to 🇨🇬 from 🇨🇬, the 🇨🇬 appears to be a borrowing of, the Old English itself being an early borrowing from the Latin. Both OHG terms and  appear Late 9th c. after the Anglo-Saxon missions to Germany :\ Leasnam (talk) 22:19, 4 March 2024 (UTC)
 * That's certainly possible. Hard to say morphologically. -- Sokkjō 00:00, 5 March 2024 (UTC)
 * This is why it's also important to know the history :] Leasnam (talk) 00:55, 5 March 2024 (UTC)
 * You could make the same argument for any Christian term, but that's certainly not the case. -- Sokkjō 00:59, 5 March 2024 (UTC)
 * Do you think this may be the reason why we have so few reconstructions for religious terms ? We didn't overlook these...they're not candidates. Leasnam (talk) 01:03, 5 March 2024 (UTC)
 * We? These entries weren't created previously because we did not have a Proto-West Germanic language on the project before. Why not just mention on the entry that the OHG attestation is late, and may instead be a borrowing from Old English? -- Sokkjō 01:27, 5 March 2024 (UTC)
 * The idea seems to be that reconstruction should be backed up by evidence. EWAhd says that o > u is a High German internal development from ecclesiastical Latin, from a Late Latin Lallwort of rather different meaning, as also noted on and . That's not a reliable basis for any kind of reconstruction. If it were, you would have to consider the ecclesiastical usage possibly borrowed from (pseudo) PWG. DurdyWendy (talk) 16:00, 10 March 2024 (UTC)
 * That change u > o is a general West Germanic a-umlaut feature, so likely it was borrowed from attested Medieval Latin variant . That, or the underlying WG form is actually, from + . -- Sokkjō 19:40, 14 March 2024 (UTC)
 * Comment: Christianity existed in the West-Germanic Sprachraum since Late Antiquity, particularly the areas bordering the historical Roman Empire. The existence of words pertaining to Christianity in PWG is certain (Category:gmw-pro:Christianity). Simply being an ecclesiastical term does not preclude its existence in PWG. — Mnemosientje (t · c) 12:09, 19 March 2024 (UTC)

Reconstruction:Proto-Indo-European/bak-
*b and *a are irregular reconstructions. Other problems: same west-central distribution as *bʰeh₂ǵos (in RfD), an Afro-Semitic comparison (not cited), three out of the six references do not support this root, two of the other three are agnostic about grouping, outdated in many parts, and the combination of meaning and sound sounds like you really serious with the comparative methadone. The Reconstruction note is a red-flag: "Various points suggest a post-Indo-European borrowing".

Meilet takes -culum for un suffixe de nom d'instrument comme en grec, but this is not clear cut because it either is as in Greek or it is not as in Greek and so irrelevant to the comparison. The suffix is supposed to be -culum, which would leave ba to be explained, so has to be addressed. acclamo, acclaro, acclino, acclivis, accolo etc. are not clear counter examples. -ulum is a dumb diminutive. DurdyWendy (talk) 16:57, 14 March 2024 (UTC)

Reconstruction:Proto-Germanic/wrībaną
Should be history merged into RC:Proto-West Germanic/wrīban and deleted. -- Sokkjō 02:29, 16 March 2024 (UTC)
 * Delete. — Mnemosientje (t · c) 12:09, 19 March 2024 (UTC)

Reconstruction:Proto-Germanic/slutilaz
Intra-West Germanic construction. Should be history merged into RC:Proto-West Germanic/slutil and deleted. -- Sokkjō 06:26, 17 March 2024 (UTC)
 * Same for WT:RFDR above. -- Sokkjō 18:14, 17 March 2024 (UTC)
 * Delete. — Mnemosientje (t · c) 12:09, 19 March 2024 (UTC)

Reconstruction:Proto-Celtic/dawnā
Proto-Celtic. Tagged but not listed over four years ago by Victar with the comment “ don't create PC entries if it only has a single branch. This should be left in an etymology.”. —Mahāgaja · talk 20:21, 17 March 2024 (UTC)
 * Delete. — Mnemosientje (t · c) 12:09, 19 March 2024 (UTC)
 * Delete. —Caoimhin ceallach (talk) 01:36, 1 June 2024 (UTC)

Reconstruction:Proto-Germanic/bērilaz
Entry created for as a possible borrowed form to, but if the Gothic existed at all, it could just be an intra-Gothic formation, like  and. -- Sokkjō 20:29, 17 March 2024 (UTC)
 * Delete. — Mnemosientje (t · c) 12:09, 19 March 2024 (UTC)

Reconstruction:Proto-Germanic/brēþiz
Intra-West Germanic construction. Should be history merged into RC:Proto-West Germanic/brāþi and deleted. -- Sokkjō 03:22, 18 March 2024 (UTC)
 * Delete. — Mnemosientje (t · c) 12:09, 19 March 2024 (UTC)

Reconstruction:Proto-Germanic/slībaną
Uncertain etymology. Should be history merged into RC:Proto-West Germanic/slīban and deleted. -- Sokkjō 07:04, 18 March 2024 (UTC)
 * Delete. — Mnemosientje (t · c) 12:09, 19 March 2024 (UTC)

Reconstruction:Proto-Germanic/faką
The etymology for this term isn't super convincing and I think should be consolidated to RC:Proto-West Germanic/fak, with no Germanic cognate outside West Germanic. -- Sokkjō 04:32, 20 March 2024 (UTC)


 * Delete. —Caoimhin ceallach (talk) 11:43, 24 April 2024 (UTC)

Reconstruction:Proto-Germanic/finkaną
Unknown origin with no cognates outside West Germanic. History merge into RC:Proto-West Germanic/finkan and delete. Same for: -- Sokkjō 06:23, 21 March 2024 (UTC)
 * RC:Proto-Germanic/funkô → RC:Proto-West Germanic/funkō
 * RC:Proto-Germanic/fangijaną, RC:Proto-Germanic/fankijaną → RC:Proto-West Germanic/fangijan
 * delete unneeded alternative RC:Proto-West Germanic/fankijan

Reconstruction:Proto-Germanic/dēhǭ
Etymology of RC:Proto-West Germanic/dāhwā is uncertain. Should be history merged and deleted. -- Sokkjō 03:30, 23 March 2024 (UTC)

Reconstruction:Proto-Germanic/hagatusjō
Etymology of RC:Proto-West Germanic/hagatussjā is uncertain. Should be history merged and deleted. -- Sokkjō 06:09, 23 March 2024 (UTC)


 * From what I can tell this reconstruction is valid when it comes to the form, it's what would be expected. However, we have no idea how far back it goes, thus I can neither concur nor argue against you. Haimariks (talk) 09:00, 23 May 2024 (UTC)

Reconstruction:Proto-Germanic/skēkaz
Etymology of RC:Proto-West Germanic/skāk is uncertain. Should be history merged and deleted. -- Sokkjō 22:21, 23 March 2024 (UTC)

Reconstruction:Proto-Italic/flūō
If it's possible to history merge this with Latin fluo, I think that would be better than having a separate page. As described on the pages, there is not a consensus on the pre-Latin form of this verb. I also don't see any cognates within Italic. Furthermore, I am skeptical of the perfect *flūzai and past participle *flūssos. The redirect at Reconstruction:Proto-Italic/flowō should also be deleted. Urszag (talk) 23:48, 26 March 2024 (UTC)

Reconstruction:Proto-Italic/pawjō
Proto-Italic. One Latin descendant with a dubious (to say the least) etymology. - saph 🍏 22:29, 29 March 2024 (UTC)
 * Created by Dim Blob, notorious for making stuff up. Chuck Entz (talk) 22:56, 29 March 2024 (UTC)

Reconstruction:Proto-Indo-European/h₁op-
Reconstructed as an o-grade to already existing. -- 04:46, 5 April 2024 (UTC)
 * The *h₁op- entry could be folded into the *h₁ep- entry although Mallory and Adams only state that *h₁op- may derive from *h₁ep-. As well derived roots often have their own entries on Wiktionary. ElkandAcquerne (talk) 13:33, 5 April 2024 (UTC)
 * Before you attempt to create any other Proto-Indo-European entries, please read WT:AINE. You'll see there that we reconstruct entries in e-grade. -- Sokkjō 02:34, 6 April 2024 (UTC)
 * Delete and merge into . This isn't significant enough that we have to make an exception to policy. - saph 🍏 03:07, 13 April 2024 (UTC)

Reconstruction:Proto-Italic/haizēō
Proto-Italic. One Latin descendant. Made up on a dubious etymology of Pokorny's. - saph 🍏 12:04, 9 April 2024 (UTC)

Reconstruction:Proto-Italic/temalom
Proto-Italic. One Latin descendant, reconstruction forged on an unlikely etymology. - saph 🍏 03:02, 13 April 2024 (UTC)
 * See WT:Requests for cleanup: they created it as an IP, then edited it later as Dim Blob. The only other human who edited the entry was an IP who couldn't do anything to stop them. Chuck Entz (talk) 03:22, 13 April 2024 (UTC)
 * Delete. It seems to be back-formed through Latin, but I'll need the input of someone knowledgeable in Italic, like . -- 𝘗𝘶𝘭𝘪𝘮𝘢𝘪𝘺𝘪(𝘵𝘢𝘭𝘬) 03:57, 23 May 2024 (UTC)
 * RFD-deleted, striking. Catonif (talk) 11:21, 26 May 2024 (UTC)

Reconstruction:Old Dutch/fro
I don't see any evidence of this form surviving into Old Dutch. -- Sokkjō 17:41, 13 April 2024 (UTC)
 * I made this entry because its genitive plural (frōno) did survive (attested) as an adjective, and *frō or its variant *frōn were found back many times in toponyms. Therefore, my conclusion was that it must have been around in Old Dutch. Preupellor (talk) 17:57, 13 April 2024 (UTC)
 * The Old Dutch noun descends from PWG, itself, yes, built from the genitive plural of . -- Sokkjō 20:02, 13 April 2024 (UTC)
 * Well, the noun was never attested, but the adjective was attested in the Mfr. Reimb. A, and if that does in fact stem from PWG too, that still doesn't explain the various toponyms that use frōn- rather than frōno- ("fronland", "fronakre", "vronlo"). Well, unless *frōn was also directly inherited from PWG, rather than a form of . Preupellor (talk) 20:11, 13 April 2024 (UTC)
 * PWG simply had more productivity than, which was displaced by  in West Germanic, and why it didn't survive into Dutch. -- Sokkjō 20:40, 13 April 2024 (UTC)
 * Sorry for the late reply, but I think we should just let a moderator decide over this. Preupellor (talk) 16:23, 30 April 2024 (UTC)
 * There are no moderators, but if there is anyone you'd like to comment, you can tag them using ping. -- Sokkjō 17:04, 30 April 2024 (UTC)
 * Thank you. Honestly the lack of people who check other people's edit, and even though I don't completely agree with you in this case, I'm at least happy that you're at least trying to filter out the errors... Preupellor (talk) 17:39, 30 April 2024 (UTC)

Maybe I'm just bored and meddling, but this appears to have solid descendents only in West-Germanic. The hypothetical Gothic term with is unreliable as the etymology is undecided between Frankish, Gothic or Lombardic. The Gothic reconstruction (@Fakename 2016, @Villager 2020) is not sourced. Apease a Zulu (talk) 16:47, 19 April 2024 (UTC)
 * Agreed and that's probably the case for a lot of alleged Gothic borrowings. Looks like was the first to add them to the page. -- Sokkjō 16:39, 29 April 2024 (UTC)
 * Move, this is indeed a secondary formation with only West Germanic evidence. In Proto-Germanic this would be a root-ablauting n-stem (weak stem *skin-), whence also 🇨🇬 from the strong stem. — Ceso femmuin mbolgaig mbung, mellohi! (投稿) 06:00, 18 July 2024 (UTC)

Reconstruction:Latin/aetaticum
No basis for assuming a Proto-Gallo-Romance form considering the absence of a cognate in Occitan (or Catalan). In all likelihood e(d)age was formed within Old French from e(d)et + -age and borrowed/calqued into Old Franco-Provençal. Nicodene (talk) 22:35, 19 April 2024 (UTC)

Reconstruction:Proto-Indo-European/key-
Entry only contains Germanic terms and a possible dubious "extended" root. -- Sokkjō 02:14, 25 April 2024 (UTC)


 * @Sokkjo is securely attested. this source mentions  as the underlying root of the Germanic terms. In such situations, would we note on the page that  is not securely attested outside of Germanic rather than deleting the page outright? Dragonoid76 (talk) 07:49, 25 April 2024 (UTC)
 * PIE isn't in question. It's worth mentioning a possible connection in the etymology sections of the Germanic entries, and even on, but it isn't strong enough evidence for a  entry. -- Sokkjō 07:56, 25 April 2024 (UTC)
 * Delete and discuss hypothetical *key- elsewhere. Two of three references don't actually mention *key-. Also, and  don't belong together. The former was originally an s-stem. —Caoimhin ceallach (talk) 10:31, 7 May 2024 (UTC)
 * Delete -- 𝘗𝘶𝘭𝘪𝘮𝘢𝘪𝘺𝘪(𝘵𝘢𝘭𝘬) 06:42, 22 May 2024 (UTC)


 * I cleaned up and moved the entry to RC:Proto-Indo-European/keHy-, which is inline with Mallory. Any connection to RC:Proto-Indo-European/keyt- is folly though. Still unsure if it's worth keeping. Kroonen is adamant that the Baltic terms are borrowed from Germanic. -- 04:31, 29 June 2024 (UTC)

Reconstruction:Proto-Germanic/swagilaz
, should we delete RC:Proto-Germanic/swagilaz and RC:Proto-Germanic/sugilaz and just chaulk up the OE forms to being of PWG ? I'd also venture to guess the sw- forms are under the influence of,. -- Sokkjō 18:16, 12 May 2024 (UTC)
 * I've been patiently waiting to do so... :] Leasnam (talk) 18:44, 12 May 2024 (UTC)
 * ✅ Leasnam (talk) 20:21, 12 May 2024 (UTC)
 * 👍 I also created an entry for WG, if you want to use that for your etymology. -- Sokkjō 20:30, 12 May 2024 (UTC)

Reconstruction:Sanskrit/झापयति
Sanskrit. Tagged but not listed. Created with edit summary "meh, why not". Ultimateria (talk) 01:25, 23 May 2024 (UTC)


 * Should be moved to Ashokan Prakrit. The Indo-Aryan editors decided long ago that reconstructed terms given by Turner are to be treated as neither Sanskrit nor Proto-Indo-Aryan, but rather reconstructed Ashokan Prakrit. I'll start working on it right away. -- 𝘗𝘶𝘭𝘪𝘮𝘢𝘪𝘺𝘪(𝘵𝘢𝘭𝘬) 02:01, 23 May 2024 (UTC)
 * I have rescued the entry. -- 𝘗𝘶𝘭𝘪𝘮𝘢𝘪𝘺𝘪(𝘵𝘢𝘭𝘬) 02:57, 23 May 2024 (UTC)

Reconstruction:Latin/hebuit
Latin. Tagged but not listed. Ultimateria (talk) 17:46, 23 May 2024 (UTC)


 * Delete both. We do not need nonlemma forms of reconstructed terms except maybe in the case of suppletion. —Mahāgaja · talk 20:30, 23 May 2024 (UTC)
 * Agreed on all counts. Nicodene (talk) 05:02, 24 May 2024 (UTC)

Reconstruction:Proto-Ryukyuan/panas-
Proto-Ryukyuan. Rationale was "not attested until Early modern japanese". Ultimateria (talk) 20:47, 26 May 2024 (UTC)


 * I'm not sure what that's got to do with the price of tea in Okinawa, since (1) Japanese isn't among the descendants of Proto-Ryukyuan and (2) the whole point of reconstructed words is that they aren't attested. —Mahāgaja · talk 21:39, 26 May 2024 (UTC)
 * It's that hanas- in Japanese is not attested until early modern Japanese, which means all Ryukyuan terms containing this word were all borrowed from Japanese. Chuterix (talk) 00:39, 6 June 2024 (UTC)

Reconstruction:Proto-Germanic/dōljō
Duplicate Proto-Germanic entry. The form is incorrect, it should be, which already exists. Leasnam (talk) 01:57, 28 May 2024 (UTC)
 * Good catch, please delete. -- Sokkjō 06:41, 31 May 2024 (UTC)

Reconstruction:Proto-Indo-European/mitstos
Here we go with users going crazy with etymon creating incorrect proto reconstructions in areas they have no knowledge in. Delete RC:Proto-Germanic/missaz too while we're at it. -- Sokkjō 06:33, 31 May 2024 (UTC)


 * Delete. - saph 🍏 19:24, 1 June 2024 (UTC)
 * Delete. Proto-Germanic may exist, though the Dutch and Old Norse terms may be explained differently. —Caoimhin ceallach (talk) 17:14, 6 June 2024 (UTC)

Reconstruction:Proto-Indo-European/stéh₂tus
Another entry that shouldn't exist but was created because of etymon. -- 18:01, 3 June 2024 (UTC)


 * At Reconstruction:Proto-Indo-European/steh₂-, *stéh₂-tu-s ~ *sth₂-téw-s is one of the derivatives listed. Is this incorrect? Ioaxxere (talk) 18:22, 3 June 2024 (UTC)
 * At no point should lemmas have hard redirects to a root entry, but the larger issue, as pointed to above, is users with no knowledge in proto languages, like PIE, creating entries, simply because they're trying to use etymon. That needs to be discouraged. -- 18:38, 3 June 2024 (UTC)
 * So if *stéh₂tus is a legitimate lemma then it seems like the best course of action is to expand Reconstruction:Proto-Indo-European/stéh₂tus into a full entry rather than deleting. Ioaxxere (talk) 19:55, 3 June 2024 (UTC)
 * You cannot expect PIE editors to clean up junk entries just so some user can play with etymon. The entry should be deleted just as any other poorly created proto entry. -- Sokkjō 08:21, 4 June 2024 (UTC)
 * I see it as a good thing that such cases are now higher-profile. If the entry is wrong, so is the etymology that spawned it, and so a problem already existed beforehand. Nicodene (talk) 20:13, 3 June 2024 (UTC)
 * The threshold of viability for a PIE reconstruction is very high, and just two possible descendants on a productive suffix often doesn't cut it. -- Sokkjō 08:24, 4 June 2024 (UTC)
 * Indeed, “if the entry is wrong, so is the etymology that spawned it” is fallacious since neither the wrongness of an entry, nor that of an etymology, is simply its truth value. Whether or not something is mentioned, linked, and created, ontologically is distinct and not interdependent in its assessment of verisimilitude, which requires and suffices contextually adaptive space allotment. Now this epistemology is difficult to understand but is able to:
 * The overapplication of the fundamentals of science, the treatment of every respect of life in a kind of coordinate system by which man makes a stand for it, sticks around since Descartes in spite of being spelt out by Vico as actually damaging the informatory value of any publication project, for which we have to strike a balance between demonstrating and cuing: >Sola hodie critica celebratur; topica nedum non praemissa, sed omnino posthabita. Incommode iterum: nam ut argumentorum inventio prior natura est, quam de eorum veritate diiudicatio, ita topica prior critica debet esse doctrina. At enim eam nostri facessunt, et nullius usus putant: nam sat est, inquiunt, homines modo critici sint, rem doceri, ut quid in ea veri inest inveniant; et quae circumstant verisimilia, eadem ipsa veri regula, nullam topicam docti vident. Sed qui certi esse possunt vidisse omnia? Unde illa summa et rara orationis virtus existit, qua «plena» dicitur, quae nihil intactum, nihil non in medium adductum, nihil auditoribus desiderandum relinquit. Natura enim incerta est, et praecipuus, immo unus artium finis, ut nos certos reddat, recte fecisse: et critica est ars verae orationis, topica autem copiosae. etc.
 * Delete. Fay Freak (talk) 12:00, 4 June 2024 (UTC)
 * I see now that it's a matter of the reconstruction entry being unnecessary (not enough desc's to justify it) rather than incorrectly formed. At any rate my vote was/is to delete.
 * As for the etymon-chain, it can be fixed by pointing to  and . Nicodene (talk) 23:33, 4 June 2024 (UTC)
 * Keep - IMO we can have redirects if a full entry is not justifiable. -- 𝘗𝘶𝘭𝘪𝘮𝘢𝘪𝘺𝘪(𝘵𝘢𝘭𝘬) 13:50, 4 June 2024 (UTC)
 * Keep per User:Pulimaiyi. Although we wouldn't have a redirect in this situation for an attested language, there's no reason why PIE needs to play by the same rules. Ioaxxere (talk) 14:36, 4 June 2024 (UTC)
 * Delete. I trust PIE editors when they say that this entry shouldn't exist. As stated, Reconstruction entries are supposed to have a higher threshold before creation, and if we don't do redirects for attested languages, we should certainly not do them for reconstructed languages. Again, this is part of the issue I recently mentioned at Beer Parlour. AG202 (talk) 15:07, 4 June 2024 (UTC)


 * Delete as a redirect, or else create a full-fledged entry for it. I don't object to redirects within Reconstruction space when it comes to alternative reconstructions, but this isn't a case of an alternative reconstruction. —Mahāgaja · talk 15:23, 4 June 2024 (UTC)
 * Delete, there should not be a redirect as there should not be links to this in the first place. These kinds of reconstructions are left unlinked. Catonif (talk) 22:50, 4 June 2024 (UTC)
 * Delete. —Caoimhin ceallach (talk) 17:22, 6 June 2024 (UTC)

Reconstruction:Middle English/fnore
@Hazarasp: Is there any grounds for reconstructing a form that basically is a ghost? There's no process fn- > sn- anywhere I know of. Chuterix (talk) 00:36, 6 June 2024 (UTC)
 * : see the etymologies for sneeze, snore and snort. This isn't isolated at all. Chuck Entz (talk) 05:30, 6 June 2024 (UTC)
 * RFV passed. See the comment I agree with below. Chuterix (talk) 05:36, 6 June 2024 (UTC)


 * Delete, even if this variant is likely to have existed alongside the verb, there's no benefit in our having a Reconstruction entry for it when it has no descendants. —Mahāgaja · talk 06:04, 6 June 2024 (UTC)


 * Delete per above. —Caoimhin ceallach (talk) 17:27, 6 June 2024 (UTC)

Reconstruction:Pictish/ᚚᚔᚈ
Pictish. A creation of the now-banned eternal optimist, UtherPendrogn. There's a References section but no actual references, just assertions that it's found in two place names and that it was borrowed into "the Gaelic languages" (with no examples; is inherited from Proto-Celtic, and  is borrowed from French). —Mahāgaja · talk 21:17, 16 June 2024 (UTC)
 * Incidentally, I tagged the entry with rfv, not rfd. Are RFVs for Reconstructions automatically RFDs? —Mahāgaja · talk 21:19, 16 June 2024 (UTC)
 * Delete. -- Sokkjō 23:40, 16 June 2024 (UTC)
 * Delete. Nicodene (talk) 07:14, 17 June 2024 (UTC)

Reconstruction:Proto-Germanic/Rīnaz
I think there's a good argument to be made for being borrowed from. -- Sokkjō 19:42, 18 June 2024 (UTC)

Reconstruction:Proto-Indo-European/trep-
Poorly created entry. -- 01:12, 19 June 2024 (UTC)

Reconstruction:Proto-Indo-European/ǵembʰ-
Debatable root. Should be left to the etymology on RC:Proto-Indo-European/ǵómbʰos. -- 01:15, 19 June 2024 (UTC)

== Reconstruction:Proto-Indo-Iranian/Hámsas, Reconstruction:Proto-Indo-Iranian/sarpás ==

Proto-Indo-Iranian: has descendants only in Indo-Aryan. —Svārtava • 13:38, 19 June 2024 (UTC)
 * Keep. We have no policy that states PII roots with only Sanskrit descendants whould be deleted. -- 15:04, 19 June 2024 (UTC)
 * Keep, per Victar. There is a PIE ancestor and a descendant, so it's reconstructable, and we don't have community consensus to delete this type of entries for PII, only for Proto-Italic. Thadh (talk) 15:10, 19 June 2024 (UTC)
 * And unlike Proto-Italic, where Latin is lucky if it has an Italic cognate, the chances of a Iranian cognate to Sanskrit are very high. In fact, is cognate with /. --  15:27, 19 June 2024 (UTC)
 * Withdrawing RFD for on that point. But I'd still argue that  be deleted if no descendant of it is attested beside Sanskrit. —Svārtava · 15:48, 19 June 2024 (UTC)
 * I would support moving the entry verb form, which is more secure. That work for you? -- 20:08, 19 June 2024 (UTC)
 * I still think it is unnecessary to have the PII entry with only one branch of descendants as that can be comfortably shown on the PIE entry. —Svārtava · 17:46, 20 June 2024 (UTC)
 * PIE had a different linguistic reality than PII speakers, than PIA speakers, than Sanskrit speakers. Having a category with solidly reconstructed terms in PII is useful to understand about the speakers and their worldview. Thadh (talk) 18:25, 20 June 2024 (UTC)
 * @Thadh: I can undrestand that argument, but in this case the PIIr entry adds really nothing - it is at best shown on the PIE page's descendants and on the Sanskrit page's etymology. I also wouldn't want hundreds of PIIr entries created on this basis alone. —Svārtava · 02:50, 22 June 2024 (UTC)

Reconstruction:Proto-West Germanic/trottōn
Proto-West Germanic. The Old English (< trieddian ?) is a variant of  (if not a sole survival of ). The West Frisian and Dutch forms are ultimately from the French. The French is a borrowing of a High German term parallel to 🇨🇬. Leasnam (talk) 16:51, 21 June 2024 (UTC)

Reconstruction:Proto-West Germanic/trettōn
Proto-West Germanic. Similar to above, this should be moved to, or maybe. Leasnam (talk) 16:55, 21 June 2024 (UTC)

Reconstruction:Proto-Italic/kerweiks
Proto-Italic. I think this hypothesis should be presented at the entry for Latin rather than at a separate entry for two reasons: it falls under the already-established criterion for deleting Proto-Italic terms with only one descendant, and the given etymology is dubious per De Vaan.--Urszag (talk) 01:04, 26 June 2024 (UTC)

Reconstruction:Proto-Koreanic/mati and other Proto-Koreanic reconstructions created by User:Chuterix
additional pings: @Chom.kwoy, @Surjection, @Mahagaja, @Theknightwho Requesting a speedy deletion for all of them. Reconstruction:Proto-Koreanic/tàLá & Reconstruction:Proto-Koreanic/hyeng were deleted for this same reason. It's been brought up to this user directly (on Discord) since the start of this year that their reconstructions simply do not make sense in terms of Koreanic family history. The sources also do not support the exact constructions, except for, but that one is shaky as well. This user has openly stated that they do not have that much experience with Koreanic languages, "I don't know much about Korean linguistics" at, let alone to create their own Reconstruction. They've also since been told to stop working on Koreanic entries for this reason. Said user has also been blocked before for similar issues. The entries in question include: This has caused a severe problem with the quality of our entries, leading to more misinformation spread, as I've openly talked about beforehand. See: this Reddit thread for an example. This is precisely why I'm getting tired of cleaning up Koreanic entries, as stuff like this gets to stay up after creation even though they're blatantly incorrect, while we wait for the slow wheels to Wiktionary to get to them. I'm very annoyed to have to put this work in now, and this is why we have good editors leave. AG202 (talk) 03:31, 27 June 2024 (UTC)
 * Reconstruction:Proto-Koreanic/cenki: If anything it should be put at Reconstruction:Proto-Koreanic/ceNki, the current reconstruction doesn't make sense logically, and the source does not support it.
 * Reconstruction:Proto-Koreanic/matV: This reconstruction solely exists to posit a Japanese borrowing, completely ignoring how Koreanic forms &  would point to an earlier *matVk, per basic historical Koreanic linguistic principles. The source also does not support it. Edit while writing: Since me bringing this problem up, it's been changed from  to, but the same issues apply.
 * Reconstruction:Proto-Koreanic/kwòmá: Having tones in the reconstruction is the first blatant problem. We're not even sure if Old Korean had tones, per Lee Ki-Moon and S. Roberts Ramsey in A History of the Korean Language (2011), let alone Proto-Koreanic. Additionally, the source cited does not have any PK reconstruction listed, solely talking about the initial vowel likely being /o/ due to vowel raising in Baekje, with Western Old Japanese likely being a borrowing from it. There's nothing that shows that the second vowel is /a/.
 * Reconstruction:Proto-Koreanic/kopoLi: What is the  supposed to represent? Also, according to the source listed, Western Old Japanese borrowed from Baekje and not Proto-Koreanic. Vovin also states that the MK form comes from *kopor but even that is shaky because it's missing the explanation for the /h/.
 * Reconstruction:Proto-Koreanic/i-: There's no reason for this reconstruction to exist. Why is it a prefix or verb stem when it's neither of those in Koreanic languages? We can show a borrowing to Jurchen -> Manchu without a poor reconstruction; the reconstruction is also not found in the source cited.
 * Reconstruction:Proto-Koreanic/seki: This is the only one that may be passable since it's directly cited by Vovin, but as @Chom.kwoy has noted, "It does seem like one possibility of a internal reconstruction but theres not enough evidence to be sure other than the tentative loanword in Japanese, so same situation as . It could have been different suffixes added to the same root, like < *se-ih and  < *se-k. Compare [Middle Korean], ,  "three years old (animal)"


 * All Proto-Koreanic etymologies have been removed from their attested entries. :) Chuterix (talk) 03:48, 27 June 2024 (UTC)
 * p.s. Not just that I don't know (much) about Koreanic linguistics, but also that I don't know much Korean at all. :/ Chuterix (talk) 03:51, 27 June 2024 (UTC)
 * For the mistaken pK *matV(k), I was fooled by the mistaken assumption that -h < (*)-k was some sort of (place or locative) suffix (e.g. AFR's hypothesis that it is somehow cognate to OJ ko2 "place"). For the taLa that was deleted, it was based on some assumption that Vovin cited, that Old Korean did not have vowel harmony, although a look at Ramsey's History of Korean Language said: "For as we move back in time from Contemporary Korean to Early Modern Korean to Middle Korean, we find that the system of vowel harmony becomes more and more regular, with fewer and fewer exceptions to its rules."
 * The syllable  in *koporVk supposes it was either */r/ or */l/. Chuterix (talk) 03:57, 27 June 2024 (UTC)

Reconstruction:Proto-Germanic/hūdijaną
The etymology for this terms is contested. Should me history merged into RC:Proto-West Germanic/hūdijan. -- Sokkjō 06:26, 1 July 2024 (UTC)

Reconstruction:Proto-Austroasiatic/k.raw
Created by banned IP-user 24.108.18.81.

Cited sources:
 * Michel Ferlus, "Formation of Ethnonyms in Southeast Asia" pdf;
 * Result of my attempt at source-verification: Ferlus reconsctructs the "etymon *k.raːw", yet he does not classify it as Austroasiatic.
 * Sidwell, Paul and Felix Rau (2015). "Austroasiatic Comparative-Historical Reconstruction: An Overview." In Jenny, Mathias and Paul Sidwell, eds (2015). The Handbook of Austroasiatic Languages. Leiden: Brill.
 * Result of my attempt at source-verification: Sidwell & Rau (2015:340-365) reconstruct no such item as PAA *k.raːw "man"; relevant pages are 350 & 351

I found *k(ə)ra:w "human being" on page 646 of Frédéric Pain (2008). "An Introduction to Thai Ethnonymy: Examples from Shan and Northern Thai". Journal of the American Oriental Society. 128 (4): 641–662. Even, Pain (2008) is old & may not reflect current scholarly consensus in Austroasiatic linguistics. Erminwin (talk) 17:14, 3 July 2024 (UTC)

Reconstruction:Proto-Indo-European/-néh₂
I question this as a PIE suffix. "*o-neh₂" nouns are simply derived from r/n-stems + or, which are themselves late. -- 03:03, 7 July 2024 (UTC)

Reconstruction:Proto-Germanic/Gaizaharduz
PWG construction. -- 21:48, 16 July 2024 (UTC)


 * How to explain the Icelandic reflex? Is it plausible it wasn't listed in Old Norse due to an unlucky attestation accident? Or was it calqued instead? — Ceso femmuin mbolgaig mbung, mellohi! (投稿) 05:45, 18 July 2024 (UTC)

Reconstruction:Proto-Malayic/pe(rR)aŋ
Proto-Malayic.

This was created using templates and language codes for a Malay entry, which I fixed. The more serious problem is that it was obviously created based on the Austronesian Comparative Dictionary entry that it links to at the bottom, and that entry only cites the 1938 Dempwolff reconstruction to contradict it. The ACD instead explains the forms in the other languages as borrowed from the Malay one. Chuck Entz (talk) 02:19, 20 July 2024 (UTC)