Across Finnic and Samic, one of the more characteristic adjective endings is *-əta ~ *-ətä; yielding e.g. Finnish -ea ~ -eä, Estonian -e, Northern Sami -at. The Permic cognate *-i̮t is also at least relatively common. Because Of Reasons I have gone for a hunt for reflexes in Mordvinic, where no productive reflex survives. More specifically I’ve gone over Paasonen’s Mordwinisches Wörterbuch (a few more could be probably found in other sources). The scoop is as follows.
First, some cases well-known in the comparative literature. (Noticably often these have exact equivalents in Finnic, or indeed specifically Finnish).
- *kalgədə ‘hard’ (> Er. kalgodo, Mk. kalgəda) < WU *ka/ëlkəta > Fi. kalkea [1]
(MWB unwarrantedly lists this as a derivative of *kalgə ‘sheaf, etc.’, which is rather < WU *këlkə ‘haulm’) - *śejəďə ‘thick’ (> Er. śejeďe, Mk. śiďä) < WU *śikətä > Fi. sikeä ‘sound (of sleep)’
(the Moksha form miscited in Uralonet as śäjiďä — a real form, but rather from some Erzya dialect that has *e > ä) - *taŋgədə ‘firm, stiff’ (> Mk. taŋgəda) < WU *taŋkəta > Fi. tankea ‘id.’
- *valdə ‘light’ (> Er. valdo, Mk. valda) < WU *wëləta > Fi. vaalea ‘id.’
(in UEW / Uralonet, Mordvinic incorrectly under the longer variant *wëlkəta) - *vijəďə ‘straight’ (> Er. vijeďe, Mk. viďä) < WU *wojkəta > Fi. oikea, NS vuoigat etc. ‘right’
We see here reflexes as *-ədə / *-əďə after a consonant cluster, syncopated *-də after a PU sonorant (but apparently not after single *k). Moksha śiďä, viďä are probably due to secondary post-Proto-Mordvinic syncope (unclear to me if with fusion *jď > ď or, as might be suggested with *ej > i in the former, with vocalization of the glide). Not many other cases follow this exactly, though. I find only one other clear example + one possible example in *-ədə:
- ? *ľifčədə ‘loose’ > Mk. ľifčəda; from a stem common with e.g. *ľifčańa ‘pliable’, Mk. ľifčəm- ‘to relax’. Attested as both an ə-stem ľifčədə- and an a-stem ľifčəda- though, hard to tell which might be primary.
- *vačədə ‘hungry’ > Er. vačodo, Mk. vačəda; from *vačə ‘hunger; hungry’
For *-də after CVR-, I find two more examples, and also two nouns that might derive from former adjectives:
- Er. boďo ‘obese’. Perhaps distorted from *vojdo, and thus a derivative from *vaj ‘butter, fat’ (which in Erzya develops as > *voj > oj)? Still would have expected *-ďə, but there’s no possible soundlawful origin for an Erzya word ending in -ďo anyway…
- *naŕďə ‘firm, tough’ > Er. naŕďe, Mk. naŕďä (no base root that I can identify)
(update: or maybe from PU *ńërə ‘cartilage’??) - *śardə ‘elk, reindeer, deer’ > Er. śardo, Mk. śarda. Has clear cognates at least in Mari (*šårδə) and Khanty (*sūrtāj; Northern Mansi surti probably a loan from this), with the PU form usually reconstructed as *śarta. However I suspect this was originally rather an adjective *śarwəta ‘horned’ ← *śarwə ‘horn’. Loss of *-w- in clusters may have been early enough in Mordvinic and Mari to allow common syncope from *śarəda to *śarda. [2]
- Mk. šoľďä ‘crazy person, crybaby’. Could this be from a common root with Finnic *hullu ‘crazy’ (both pointing to earlier *šul-)? The morphology of the Finnic word remains obscure though, and the palatalization in Moksha would be unexpected; maybe suggests something like *šuljəta. Alternately, maybe ‘crybaby’ is more original, and the Mk. word is instead from a common root with Erzya čoľeďe- ‘to chirp, trill’? Either way this would probably have been an original adjective.
There are however several adjectives ending in *-adə, derived mostly from stems already ending in *-a-. This contrasts with the suffix’s behavior in Finnic and Samic, where it always carries a 2nd-syllable *ə even when attaching to *a-stems (e.g. Fi. lauha ~ lauhea ‘mild (of weather)’, notka ~ notkea ‘pliable’). I suppose the widepread Proto-Mordvinic reduction of 2nd-syllable vocalism led to a reanalysis of *-ədə as just *-də, and then later on the rise of new cases attaching to different stems.
- *kaladə ‘broken’ > Er. kalado, Mk. kalada; from a stem *kala- common with e.g. *kaladə- ‘to break (intr.)’, *kalaftə- ‘to break (tr.)’
- *komadə ‘turned over’ > Er. komado, Mk. komada; from *koma- ‘to turn over (< PU *kuma-)
- *naksadə ‘rotten’ > Er. naksado, Mk. naksada; from a stem *naksa- common with e.g. *naksaftə- ‘to let rot’, *naksalgadə- ‘to begin to rot’
- *ozadə ‘sitting’ > Er. ozado, Mk. ozada; from *oza- ‘to sit’
- *panžadə ‘opened’ > Er. panžado, Mk. panžada; from *panžə- (!) ‘to open’ (< PU *panča-)
- *śťadə ‘straight, standing’ > Er. śťado, Mk. śťada; from *śťa- ‘to stand’
- *štadə ‘naked’ > Er. štado, Mk. štada; from *šta- ‘to be exposed, cold’
- *tajadə ‘stupid, grumpy’ > Er. tajado; from a stem *taja- common with e.g. *tajardə- ‘to be timid, dejected’, *tajaskadə- ‘to become grumpy’
Itkonen (1963, CIFU 1) has proposed to consider a chunk of these to be instead primarily adverbs, formed with the homophonic ablative suffix *-də, but I’m not sure if this is a good analysis: Mordvinic infinitives and participles are generally marked, not formed by appending case endings to a bare verb stem. Also, I would analyze *kala, *naksa, *taja to be primarily noun roots ‘brokenness’, ‘rottenness’, ‘unsatisfiedness’.
Still more interestingly, I can also find adjectives where the final vowel looks to have escaped vowel reduction.
- Mk. aluda ‘underlying; under’. Another adverb/adjective, seemingly pleonastic from an unattested *aləŋ > *alu ‘underlying, undery’ (maybe ousted by the homophonic lative adverb: Er. alov, aloŋ, Mk. alu ‘(to) under’).
- Er. čando, čonda ‘pricey; price’. Probably not a cognate of Fi. hinta ‘price’ as traditionally compared. MWB hesitantly but I think more likely correctly suggests a connection with Er. Mk. čana ‘price’, which is ← Ru. цена.
- *pärda > Er. ala-berda, Mk. ala-pärda ‘missshapen’ (“under-pärda“). Probably still an independent word in PMo., given how Erzya and Moksha differ in if they adopt compound-medial stop voicing (“rendaku”, we might call it).
- *säŕďa ‘fragile of old age’ > Er. seŕďa, Mk. śäŕďä; evidently from a common root with *säŕəďə- ‘to hurt, be sick’ (? < PU *särä-, though intriguing resemblance also with Finnic *särke- ‘to hurt’).
- *šopəda ‘dark; darkness’ > Er. čopoda, čobda, Mk. šobda, šovda; from *šop ‘in a day, for a day’
- *topəda ‘dark (of color), maroon’ > Er. topoda, Mk. tobda; from *topə ‘full’, the meaning apparently thru expressions like *topəda_seń ‘full blue’ = ‘dark blue’.
- #ťožda ‘light’ > Er. čožda ~ Mk. ťožďä (no base root that I can identify). Reconstruction difficult due to several irregularities. Is Er. č- maybe by contamination with čova ‘thin, fine’?
In three of these, we find a similar environment to where PU 2nd-syllable *a survives: after a 1st-syllable *o < PU *u. Maybe the same would have originally allowed even retention of a 3rd syllable *a? — By contrast the disharmonic *pärda, *säŕďa pretty much have to be Mordvinic-internal formations. Could an adjective suffix *-da have been generalized / extracted just from cases like *topəda?
No further answers today; just a look at what other etymological candidates we might have in Mordvinic for residues of this ending.
[1] Close to a ghost word, though; kalkee ‘poor, low-quality’ is only known from one Finnish dialect. This can only really link to ‘hard’ thru kalki ‘poor, unlucky’ (“having a hard time”) from one early dictionary. The reported “dialect variant” kalkkea ‘loud, talkative, lively’ seems likely to be unrelated and instead from the verb kalkkaa ‘to ring (bell), make loud noise’ (many similar derivatives from this, also e.g. kalkas ‘lively’, kalkatti ‘blabbermouth’). — Estonian kalk, kalge ‘hard, brittle’ is a more reliable cognate in any case at least.
[2] In Mo. loss of *w probably postdates medial voicing though: by a few examples, *-tw- *-sw- seem to yield PMo. *-t- *-s-, not **-d- **-z- (at least *latə ‘shelter, roof’ ~ Finnic *latva ‘canopy’, *kas- ‘to grow’ ~ Finnic *kasva-).
Freelance reconstruction 
Er. alaberda, Mk. alapärda looks borrowed from the Turkic personal name Al(l)aberdi (< *Allah berdi 'Allah has given'). The name is known among the Volga Kypchak speakers and is sometimes adapted as Алаберда. It is a recurrent pattern in the region that names of specific individuals are used as pejorative terms for certain traits of the human body or personality. E.g., many Mari terms with this type of semantics ('sloven', 'fastidious', 'arrogant', 'naughty', 'bully', 'mischievous', etc.) are Turkic personal names in origin.
Maybe — could explain the sequence *-ärd- at least. Does Moksha otherwise get also medial -p- from foreign -b- though? Not in cases like ćebäŕ ‘pretty’, saban ‘plough’.
Butylov says Mord. p- is a default way to substitute b- word-initially (e.g., pakśa Mk. apaĺams ‘испортить, окорябать’, but — if unparalleled — this example is probably not quite reliable.
Sorry, my comment was clipped, this is the correct one:
Butylov says Mord. p- is a default way to substitute b- word-initially (e.g., pakśa Mk. apaĺams ‘испортить, окорябать’, but — if unparalleled — this example is probably not quite reliable.
Clipping problems: you’re probably trying to use < > without escaping them to < >.
Ok, thanks, one more attempt:
Butylov says Mord. p- is a default way to substitute b- word-initially (e.g., pakśa from Tat. baqča ‘field’). Perhaps the Mk. -p- reflects the fact the word is etymologically a compound, indeed. He also provides an example for the intervocalic position, Mk. apaĺams ‘испортить, окорябать’ from Tat. äbälä- ‘id.’, but — if unparalleled — this example is probably not quite reliable.
Devoiced substitution of all stops remains common enough in Moksha even in Russian loans, but yes intervocalic cases look very rare. One Moksha case from Paasonen is even päbak ‘bud, base of fruit’ ← Tat. / Bashk. bäbäk. Adopting the word knowing that it is a compound sounds plausible for this though (especially since it continues to kind of parse as that in Mordvinic).
Voiceless medial substitution could perhaps be still expected at some point in early Mordvinic, shortly after *-b- > -v- which leaves a Mari-like situation with arguably no phonemic /b/ ([mb] analyzable as /mv/, [db] as /dp/ etc.)… on the other hand, cases like common Mordvin ava ‘mother’ seem to point to -b- having rather been adopted as -v- at this stage. A theoretical alternative maybe would be that *-b- > -v- is post-Turkic contact entirely and these oldest loans went *aba > ava as a native development, but yeah I’d like a fair bit more evidence before setting up such an assumption.
My understanding is that Mord. ava ‘mother’ is a relatively recent loan (from Old Chuvash *äbä, ca. 14-15th centuries) as the intervocalic voicing -p- > -b- in Chuvash certainly postdates the Mongol invasion.
If ava was loaned as *aba, it would not even strictly need to be from a voiced original: Mordvinic medial -p- is half-long [pˑ] (as all voiceless obstruents), so also short lenis [b̥] could have been adopted as *b, when it still existed. But it does look like a correspondence that would benefit from some additional intermediary where late (*p >) *b > v or w is known.
Contamination could be also an option! Already PU likely had *apa for an older female relative of some sort (continued at least in Mansi ūp ‘younger maternal uncle’s wife’, Khanty E *ɔpi̮ ~ W *ăpə ‘older sister, cousin, aunt, etc.’, Samoyedic *apå ‘older sister’). This protoform would have given PMo. *avə or *ava, which could have been remapped to ‘mother’ by Turkic contact.
mordwinisches
According to Paasonen, Md. *komadǝ ‘turned over’ is an adverb rather than an adjective. As such it likely originated in a fossilized partitive case form identical to Saami *komōtē ‘turned over’ (North Saami gopmut, etc.). The same noun root *kumo is also attested in Finnic in various case forms (Finnish kumossa, kumoon, etc.), but not in the partitive form, I suppose.
As a side note, Md. *śejǝdǝ ‘dense’ cannot reflect a proto-form with *s- due to its initial palatalized sibilant. Instead, we must reconstruct PU *ć- (*ś- according to the traditional reconstruction), because PU *s- would have been preserved as s- in Erzya.
Fair, the parallels do give it extra support. Still no nominalizing suffix in sight though, but *kuma-w as suggested by Finnic and Samic could have also given just *koma in Mordvinic…
*sikətä was a simple typo, fixed.
The same correspondence can be actually found also in two other of Itkonen’s examples, i.e. Samic *pōncōtē ‘naked’, *čōnčōtē ‘standing’ (if cognate to *śťa-, as per e.g. UEW). This is again not really evidence for treating the Mo. counterparts as residual partitive adverbs from the bare verb stem, which would predict in Samic **pōncētē, **čōnčētē; but it could point to denominalized *panča-w-ta, *ćańća-w-ta. Interesting how the former is residual in Samic too (only South thru Lule which do not otherwise retain the partitive case). This in mind the other two clearly deverbal cases *ozadə, *štadə might be also better treated as similarly formed adverbs after all. But I am hesitant to extend this all the way to *kaladə, *naksadə, *tajadə without any evidence of a base verb.
That actually reinforces the similarity to the IE *-tó- adjectives, of which most are participles, but some (e.g. bearded) are directly denominal…
What do you think about PGerm. *tanhu- ‘tough’ (Kroonen 2013: 509), which seems to me to be related to WU *taŋkəta? I assume no one presented this comparison so far.
Seems formally decent! — but if we start looking into IE, Baltic *tankja- ‘thick’ (> Lith. tankus, also loaned as Finnic *takja ‘dense’) would seem to make for a strictly better source for geographic reasons. Germanic loans in Mordvinic are not at all well-known… mostly just some possible Wanderwörter from/via Gothic.
Finnish has also variants tanea ‘id.’, tanakka ‘robust, stout’. If these are not back-derivatives, they would point to *taŋkəta segmenting as *tan-kəta, perhaps then still from a common IE root with *tankja- though.
(Another possible Uralic direction of comparison yet would be Hungarian dagad ‘to swell’, which could be reconstructed back to *taŋkV-nta- ‘to be/become *taŋkV’ and could be related to WU after some semantic shifts on either side.)
— I also do not understand why does Kroonen claim *tanhu- ‘tough’ to be from an IE root for ‘to bite’, but this may be neither here nor there for purposes of Finnic/WU comparison.
Have you listened to Aikio’s presentation in Greifswald? As far as I know, he dealt with the Proto-Germanic loans in Mordvinic, Mari and Permic. Surely he was in a negative position about these. What do you think for instance about runko that can be compared to Mordvinic mdE rungo and M ronga? PGerm. *skrunkaz- has been presented as a possible loan original. Is runko a Wanderwort form/via Gothic? I do not know whether PGerm. *skrunkaz- has been inherited to Gothic.
If Finnic tankëda and Mordvinic *taŋgədə are cognates each other, can Baltic *tankja- ‘thick’ be the common loan original? Baltic *tankja- has already been loaned as Finnic *takja-. Of course, Finnic tankëda and Mordvinic *taŋgədə could be older loans from the same loan original than *takja-.. Anyway, Baltic *tankja- is not as convincing as PGerm. tanhu- from a semantic point of view, in my opinion.
Indeed yes, I was even present on-site myself. For a decent while I have already agreed with the general gist that this is a far-fetched loanword layer. It will be good to see a review put together by him if any of this still holds up in any form. Many have by now already been given better alternate etymologies.
For ruŋgo even coincidence could be possible, cf. e.g. Latin truncus > Eng. trunk which has nothing to do with Germanic *skrunkaz but still looks like a near-perfect match with Finnic *runkV in the meaning ‘trunk’ (— and who knows, maybe it’s even the case that Finnish ‘trunk’ and wider Finnic ‘body’ are actually two separate etymologies, not original polysemy). In Mordvinic, there’s also an unclear mismatch here between Erzya /u/ ~ Moksha /o/ which might be trouble for any attempt to give this a pre-Mordvinic etymology but could be some support for a Wanderwort explanation. Not that a word for ‘body’ is what we most expect to see as one. So currently I’m fairly skeptical.
Finnic loans retaining Baltic (or Germanic) *-ja-stems as *-Cja- do seem to be a slightly newer layer, older loans show different patterns such as *-Tj- > *-ćć- (metsä) and this could be the case here.
Finnish tankea ‘stiff, rigid’ has an extremely limited dialect distribution; according to SSA it is only attested in some locations in South Ostrobothnian and Kainuu dialects. Such a distribution would be highly unusual for an old inherited word, although it’s not quite impossible, of course. But if I only considered the Finnic evidence, to me this would certainly look like a local dialectal contamination of tanakka ‘stout, sturdy, firm’ and kankea ‘stiff, rigid’; in any case some kind of associative connection between thw words kankea and tankea must exist, considering their identical meanings. But admittedly, this analysis would leave the phonologically and semantically regular correespondence to Moksha Mordvin taŋgǝda ‘strong, firm, tight’ as a weird coincidence.
Indeed, adjective tankea is marginal word in Finnish dialects, but common Finnish verb tankata ’to stutter; to harp’ is obviously formed of the same stem as tankea (or is derived directly of this word?). The semantic (and likely derivational, too) parallel is, for instance, änkeä (in North Tavastia) ’strong, stiff’ ~ änkätä (widely in Southwest dialects) ‘to stutter’.
If Moksha Mordvin taŋgǝda is coincidence, how about Livonian da’nktõ ‘strong, stiff’? Is it also a coincidence? I think it is more likely that kankea has supplanted tankea in Finnish dialect. And if this is right, there should be no problem in reconstructing tankea to West Uralic.