say what u will but i think kortlandt has a point with indo-uralic
I hear conflicting things about Indo-Uralic. My general impression is that it has promise, but I really know very little.
@possessivesuffix, do you have thoughts on Indo-Uralic?
Sure, maybe it's time I reiterate thoughts on this. Indo-Uralic definitely "has promise": it's got 100+ years of research and everyone on it agrees on a couple handfuls of basic data, the likes of PIE *wĆ³drĢ„ ~ PU *wetÉ 'water'; no one has given coherent alternate explanations, "they're loans" doesn't work in several ways; and "they're just accidental" could be right for any individual comparandum, but versions of this generally fail to consider the entire base of data (already because it's scatterd in literature, see below).
If some individual has a point is much more variable. Most people who assert a more detailed theory seem to be able to find a bunch more data for it, but often this can be easily criticized by others pointing out that they're e.g. confused about some well-known loanwords, or have been over-applying one family's morphology or morphophonology to the other (usually Indo-European stuff for Uralic), or have been using unreliable data from one side or the other. For example no, we probably should not think that Uralic *uwa 'flow' segments as *u-wa and contains the zero grade of a root *we- from which 'water' has been derived (or was it that *uwa- is a full grade and *we- is some kind of a zero grade from it?), as long as we don't have reason to think that anything at all is "in zero grade" in Uralic; or that there exist nominal suffixes *-d-, *-tÉ or *-wa.
Since this has not been a topic particularly in the mainstream since uhh the 1920s, the research is also fairly scattered and most people since then don't seem to know each other's work in good detail (and note that this includes a requirement to trawl thru work on Nostratic for relevant stuff). If I was working on this, I'd start from the side of bibliography to see what "already is" known. Kortlandt has put together some decent arguments that focus on the long-known material, but I'm unsure if this either achieves much further accumulation or later results yet, or how many of the promising-looking bits are his own contribution. And then there's the issue where most work seems to have been from Indo-Europeanists who don't know comparative Uralic very well (K is definitely in this camp) or occasionally also, from Uralicists who don't know comparative Indo-European very well.
There remain in any case also many open questions because (P)IE and (P)U have a large bunch of typological divergences despite being neighboring families. Though it seems to me that this is shrinking the more we learn about Anatolian, which e.g. has basically no prefixes and does have possessive suffixes; or for another intriguing point, Anatolian was recently argued to have evidence for *mel- 'to think', which now compares better with Uralic *mƤlÉ 'mind' than classical PIE *men- does. I wouldn't know if there's a connection between these two PIE roots (is *men- maybe from pre-PIE *menl- which is by n-infixation from *mel-?)
I will refrain from starting a mini-bibliography of recent things happening around the hypothesis, but the JIES 43 special issue from 2015 is a good place to start reading for anyone who wants some handle on details.
There is a lot to reply to. I see lots of questionable assumptions.
"Anatolian, which e.g. has basically no prefixes " But the Indo-Anatolian proto-language it derived from did have adpositions as prefixes. Like we see e.g. in Hittite pa-izzi 'to pass', pe-hĢ®ar(k)zi 'to hold ready', pe-iezzi 'to send', pe-dai 'to take somewhere', u-ŔŔiezi 'to draw open', Ćŗ-dai 'to bring' etc.
"Anatolian was recently argued to have evidence for *mel- 'to think', which now compares better with Uralic *mƤlÉ 'mind'" With 'recently', do you mean 2008 in Kloekhorst's Etymological Dictionary (page 545-546)? There is also Greek mĆ©lÅ 'to care about'.
"over-applying one family's morphology or morphophonology to the other (usually Indo-European stuff for Uralic)" Why would the parent language of Indo-European and Uralic have none of IE's weird morphophonology? Indo-European has a very early attestation, millennia earlier than Uralic. All things being equal it should better reflect the older state of affairs. And leveling out this morphophonology is the natural development. There is no modern IE language where ablaut, nasal presents, s-mobile or IE style reduplication is still productive. And with examples like PU *lƤmpi 'warmth' ~ PIE lehāp 'to glow', PU *wƤntV 'to look' ~ PIE *weid 'to see', PU *jƤÅÉ£i 'ice' ~ PIE *hāiĢÆeg 'to freeze'. PU *ÅaÅka 'to sting, to stick' ~ PIE *stegŹ° 'to stick' it is clear to me that assumption is probably invalid.
"confused about some well-known loanwords" The problem with loan etymologies is that they have been overused by Uralicists and there is a lot of junk out there. Some even say that the likes of nimi 'name', weti 'water, etc. are Indo-European loans. And then fault anyone calling it Indo-Uralic inheritance for 'being confused about well-known loanwords'. While we are at it, let's call the pronouns loans too, like Vovin tried to do for Altaic. The only way to go about it is to establish solid sound correspondences between Indo-European and Uralic. And then anything that behaves according to those sound correspondences and is reasonably regular in both IE and Uralic is potential Indo-Uralic inheritance. Even if someone once called it a loanword.
There is no modern IE language where ablaut, nasal presents, s-mobile or IE style reduplication is still productive.
Ablaut and nasal presents are very much alive in Lithuanian, even if notĀ āproductiveā in the senseĀ āapplied to recent loansā.
On the other hand, I doubt that there is a modern IE language which has no fossilized trace of ablaut. E.g., Modern Irish has nom. sg. bean āwomanā, gen. sg. mnĆ”, gen. pl. ban - and this in a language which has gone through several periods of severe phonological and morphological change.
Indo-European has a very early attestation, millennia earlier than Uralic.
Proto-Uralic reconstruction is not perfect (neither is PIE) but it is sufficiently reliable to tell that it did not have traces of IE-like morphophonology. Proto-Uralic is not much younger than PIE, and certainly older that earliest attested IE languages.
"Proto-Uralic reconstruction is not perfect (neither is PIE) but it is sufficiently reliable to tell that it did not have traces of IE-like morphophonology."
That is just not true. The reconstruction does not tell that. A more truthful way of stating it is that there is no need to assume IE like morphophonology. For example, there are a lot of unexplained vowel alternations in Uralic. If you a priori exclude ablaut as a possible explanation, that is a choice, an assumption you make beforehand. But it certainly is not a result.
No, even if by āa lot of unexplained vowel alternations in Uralicā you mean genuine alternations (within one language), and not just unexplained sound correspondences (between several languages). There is indeed a lot of such alternations, e.g., in Udmurt, and in some other Uralic languages. But what we need to advance Proto-Uralic ablaut as a possible explanation are alternations reconstructable for Proto-Uralic. The best example I know is *pala-Ā āto burn (intr.)ā ~ *poltta-Ā āto burn (tr.)ā, but the causative here is limited to West Uralic and the basic verb did not even meanĀ āburnā in Proto-Uralic, it meant āto devourā and was transitive.
As a general historiographical note, after the work of SetƤlƤ thru Steinitz, we're kind of instinctively wary in Uralic studies about reconstructing proto-alternations from not very substantial evidenceā¦ not because it's impossible for evidence to be worn thin, but because it's by now clear that this approach allows too many different possibilities, no way to constrain choices between them, and sidelines attempts to find conditional factors for alternations.
From this angle, I even think the notion of Indo-European "ablaut" is itself probably underspecified. If a language only has two proper vowels *e *o, asserting "*e ~ *o ~ zero" will refer to literally any possible vowel alternation that may happen for any reason in any environment. This is not a definition that can be consistently applied to any other language variety, and indeed everyone recognizes that secondarily arising umlaut or metaphony processes in IE languages are not ablaut. Within IE, we can continue to posit some lexical instances of vowel alternation as "ablaut" if and only if we can etymologize them as continuing PIE ablaut. This holds even if they're not synchronically distinguishable from other vowel alternations. English /ÉŖ/ ~ /oŹ/ "is ablaut" in ridden ~ rode (was ablaut already in PG *ridanaz ~ *raid), but "is umlaut" in gild ~ gold (is secondary from PG *guldijan- ~ gulĪøan).
Looking outside IE, I think we'd also need a similar level of evidence to claim something as "being ablaut" historically and not just descriptively. If we had distinct Uralic cognates for both nom.sg. *wod-rĢ„ and gen.sg. *wed-nĢ„-s, then sure, that would seem to be ablaut; but if you we find some vowel alternation, for some unknown reason, in entirely different lexemes ā or even in partly same lexemes but in completely different parts of the paradigm or derivation tree ā that's not going to be sound grounds to infer a connection.
And then "ablaut", which has no clear conditioning and applies in all sorts of places, seems to be itself most likely similarly heterogeneous and to represent multiple processes. Especially since it's three grades and not two, so at minimum it has to have two origins: one process that yilded *e ~ *o and another that yilded *e ~ zero (and maybe, though not by absolute requirement, *o ~ zero)
Beyond ablaut, it's then of course impossible for Uralic to have a cognate process or even any kind of an especially convincing trace of s-mobile or root zero-grades, when Proto-Uralic (or even most branch protolanguages) does not even have any stem-initial consonant clusters. A similar argument goes I think for morphological mismatches: PU only seems to have had agglutinative TAM marking, so finding any cognate for PIE perfect reduplication or present tense nasal infixation looks very unlikely on those grounds. (There is clearly room to reconstruct PU verb inflection in more detail though.)
On a very broad level I'd suggest that familiarity especially with IE does not prepare people to understand well the evolutionary trends of morphophonology in languages with more agglutinative morphology. There are relicts of old processes all around for sure, but they're by and far relicts of minor juncture effects, where the originating sound changes will likely remain broadly identifiable; mostly not relicts of fully morphologized mutation systems that completely disappeared.
As a general historiographical note, after the work of SetƤlƤ thru Steinitz, we're kind of instinctively wary in Uralic studies about reconstructing proto-alternations from not very substantial evidenceā¦ not because it's impossible for evidence to be worn thin, but because it's by now clear that this approach allows too many different possibilities, no way to constrain choices between them, and sidelines attempts to find conditional factors for alternations.
As a first approach, we must assume that all vowel reflexes of a stem go back to one vowel in Proto-Uralic. But maintaining this too rigidly leads to overconditioning, ad-hoc sound changes and unmotivated intra-Uralic loan etymologies. And there are obvious cases where maintaining this assumption is not tenable. The word for 'two' in Uralic is an example of that.
We must always reconstruct what seems to be the state of the proto-language based on evidence. Sometimes things happened that make the proto-language hard to reliably reconstruct. Rejecting those things does not make the reconstruction more rigid, it makes the reconstruction wrong.
Indo-Uralic ablaut is still a hypothesis. Ultimately it needs to be clear what the 'ablaut grades' are, which vowels they result in under what conditions, and in which situations what 'ablaut grade' is used.
Especially since it's three grades and not two, so at minimum it has to have two origins: one process that yilded *e ~ *o and another that yielded *e ~ zero (and maybe, though not by absolute requirement, *o ~ zero)
Indo-Uralic probably had more grades than that. Even Indo-European proper had *Ä and maybe also *Å. But you are right, without a complete reconstruction of the Indo-Uralic vocalism it is hard to separate ablaut from other processes. And you can't just take the vowels of Uralic and reconstruct that as the Indo-Uralic vocalism. The reconstruction of the Indo-Uralic vocalism is anything but trivial.
Beyond ablaut, it's then of course impossible for Uralic to have a cognate process or even any kind of an especially convincing trace of s-mobile or root zero-grades, when Proto-Uralic (or even most branch protolanguages) does not even have any stem-initial consonant clusters.
Of course it is true that Uralic does not have initial *sC consonant clusters. But Indo-Uralic did have them and they became shibilants in Uralic. I don't know if Indo-Uralic had true zero-grades. Perhaps the best way to think about IE's zero grade is in terms of something like an Indo-Uralic schwa or something like the Slavic yers.
It would be ideal if IE's zero grade corresponds to one vowel in Uralic. But that is not how it seems to work. It looks as if it splits into different Uralic vowels based on nearby consonants (laryngeals, semivowels) and vowels. So you get one group of stems where the vocalism is mostly determined by the environment. And another group where the vocalism is mostly determined by derivation. There were probably also some umlauts leading to vowel harmony. And you end up with a situation that is even more complex than Tocharian.
On a very broad level I'd suggest that familiarity especially with IE does not prepare people to understand well the evolutionary trends of morphophonology in languages with more agglutinative morphology.
The typology within language family is not constant and similar typology often more reflective of a Sprachbund than true inheritance. Modern Uralic languages can be very fusional, sometimes even more so than what is found in Indo-European.
You wish to explain the failure to remain within the bounds of the 'agglutinative' assumption of proto-Uralic as ignorance. But there is clear evidence that the Indo-Uralic proto-language was transfixing just as Indo-European. And I think that the widely held but wrong assumption that Indo-Uralic was not transfixing is one of the reasons that Indo-Uralic has been stagnating so much.
