Reconstruction talk:Proto-Indo-European/h₂ékʷeh₂

Persian

what about Persian ab?--71.111.229.19 23:17, 2 June 2010 (UTC)

It's from *h₂ep-. —Angr 10:18, 10 May 2013 (UTC)

Proto-Indo-European?

This word needs more than Germanic and Latin, preferably from an Asian IE language, in order to be securely reconstructible as a PIE root. EliasAlucard / Discussion 17:05, 10 May 2013 (UTC)

The Oxford Introduction to PIE lists it as a Northwest PIE word. — Ungoliant ^(Falai) 02:53, 11 May 2013 (UTC)

Yes, if it's only attested in Germanic and Latin, the most reasonable conclusion one can make is that it could have evolved in the European regional sphere, post-PIE. If there's a cognate in Anatolian, Indo-Iranian, Tocharian or Armenian, then it's a different matter and it can be said to be a PIE root. Keep also in mind that Germanic and Italic are closely related Centum languages. EliasAlucard / Discussion 06:14, 11 May 2013 (UTC)

I don’t know if it should be kept, but if it is, there should be a note. If you want to pursue this, you can add {{rfd}} to the page and state your case at WT:RFDO. — Ungoliant ^(Falai) 16:33, 11 May 2013 (UTC)

Why on earth would we want to delete it? There's no separate code for "Proto-Northwest-Indo-European", so there's no other place it could be, and it's the ancestor form of a hell of a lot of words in English and other languages. What difference does it make if it's only attested in two branches? Most Indo-European roots are attested in only a subset of the branches. I suppose we could add a note, but the fact that only Germanic and Latin are listed under Descendants already tells us that the word is attested only in Germanic and Latin, so the note would be kind of redundant. —Angr 16:58, 11 May 2013 (UTC)

“the fact that only Germanic and Latin are listed under Descendants already tells us that the word is attested only in Germanic and Latin” no, this just tells us that only Germanic and Latin descendants have been added to this wiki page. — Ungoliant ^(Falai) 17:11, 11 May 2013 (UTC)

We can call words used only in Australia "English", so why can't we call words used only in the Germanic and Italic proto-dialect "Proto-Indo-European"? —CodeCat 18:04, 11 May 2013 (UTC)

We tag words used only in Australia with {{Australia}}, so why not do the same for Northwestern PIE? — Ungoliant ^(Falai) 18:17, 11 May 2013 (UTC)

That may be a bit redundant because it already follows from the list of descendants. What I am more curious about is what criteria linguists have in distinguishing between post-PIE reconstructions and reconstructions that just happen to be attested rarely. —CodeCat 18:43, 11 May 2013 (UTC)

Readers won’t be able to guess whether the non-Latin and non-Germanic aren’t there because it’s NW PIE word, or because we haven’t added them yet. — Ungoliant ^(Falai) 18:47, 11 May 2013 (UTC)

Which is why I wonder how linguists themselves tell the difference. —CodeCat 18:53, 11 May 2013 (UTC)

Well, they can't. If a word is attested only in Latin and Germanic, there's no way to know whether it was present in PIE and lost everywhere else or whether it didn't come into being until the Northwest group had already separated off from the rest of the family. The second scenario is slightly more likely, though, just because there are so many other branches, some of them (like Sanskrit) with such huge corpora, that a word with such a basic meaning as ‘water’ would be bound to be attested somewhere somehow, even if just as an element in a rare, poetic compound, or in a placename from one of the IE languages we know next to nothing about, like Illyrian or Phrygian, if the word were really present in the oldest PIE. But even if the second scenario is the case, there's still no explanation for why the word is unattested in Celtic; there's certainly no grouping that includes Italic and Germanic but excludes Celtic. But Ungoliant's point is simpler: he just means that without a note readers might think we just haven't gotten around to adding the other languages yet rather than that there are no more languages to add. —Angr 20:56, 11 May 2013 (UTC)

Doesn't that apply to just about everything on a wiki? If we do leave a note, how accurate is that note, really? Just because we (as editors) aren't aware of any more descendants doesn't mean there aren't any. It doesn't seem right to put in a notice that says, effectively "no more descendants are allowed in this list". —CodeCat 20:59, 11 May 2013 (UTC)

I'm pretty sure that published sources have said this word is found only in Germanic and Latin. It's not just our guess. —Angr 21:05, 11 May 2013 (UTC)

The rules of reconstructing a proto-language, especially proto-Indo-European (but also proto-Afro-Asiatic in the case of Semitic as a necessary component), is that the word in question must be attested in both the European and Asian branches. If it only has descendants in the European branches, then it's possible it could be a word that was borrowed into the European languages early on, in the European sphere of influence. Of course, the word must follow regular sound laws and all that too, so it's possible to distinguish loans from real inherited words, but the problem isn't that the word is attested in only two daughter languages; the problem is that Germanic and Italic are both in the same continent, relatively close to each other geographically and linguistically, and did to some extent, interact with each other. Had it been attested in only Italic and say, Avestan, then yeah, this would undoubtedly be a PIE root. But as it is now, it's questionable. Moreover, the more daughter languages the word is attested in, the better understanding linguists will have of its etymology and pronunciation in recostructing the root word. EliasAlucard / Discussion 06:07, 12 May 2013 (UTC)

┌────────────────────────────────────────────────────────────────────────────────────────────────────┘ That would only hold if you're postulating a "Proto-Western Indo-European" language, that was separate from a "Proto-Eastern Indo-European" language, or something along those lines. Otherwise, it would be genuine PIE- regional, perhaps- but PIE, nonetheless. There have been plenty of attempts to subdivide Indo-European along one set of isoglosses or another, but none of them (with the exception of Anatolian vs. the rest of IE , of course) has been accepted by many linguists. The problem is that you can split things a multitude of ways, depending on which isoglosses you choose. In some ways, Italic has more in common with Hellenic and Indo-Aryan than it does with Germanic and Balto-Slavic. As for your citing DNA evidence: sometimes it correlates with linguistic evidence, sometimes it doesn't- and there's no way to be sure which is which, especially when you're talking about thousands of years before recorded history, with only educated guesses as to the geography involved. It's tempting to try to organize things into tidy little patterns, but reality is a bit more muddled than that. Chuck Entz (talk) 07:04, 12 May 2013 (UTC)

Very late comment: It all depends on how we define PIE. Is it (a) the last stage of linguistic development in which all Indo-European languages shared, or is it (b) everything before the emergence of clearly identifiable sub-branches. Our definition is the latter, for good reasons. But I think it would be worthwile to add tags to those words which cannot be reconstructed for PIE in the stricter sense. Kolmiel (talk) 14:41, 6 February 2017 (UTC)

This seems to me to be a non-problem, actually. There are many very basic lexemes that are quite solidly reconstructed to PIE yet are still missing any trace of any reflex in well-attested branches, against all expectations, and we have no real clue why. It's evidently just the way language change works: inherited lexemes sometimes disappear without trace and without evident rhyme and reason.

There are many parallels to this observation in Romance, for example. (In Romance, there is sometimes an appeal to "phonetic fullness", to explain the replacement of certain classical by vulgar words that are longer and less quick to erode phonetically; this factor could actually have played a role in this case, too.) So the fact that this lexeme is missing even in Sanskrit does not really prove anything.

In fact, the conservativity of the European branches and especially of Germanic is often underestimated, and the richness of the inherited lexicon preserved in it; therefore, I am deeply sceptical of the way some quickly declare a reconstructed lexeme as "European only" or "regional" without any shred of indication, except a limited distribution, that the word may have been borrowed or formed secondarily.

The authors of the LIV have recognised this problem: discarding a perfectly PIE-looking lexeme completely only because of limited distribution is unwise. Also, and especially, because additional evidence can always appear or be discovered. Even if a lexeme is only attested in a single branch, if there is nothing to really prevent us from projecting it back to PIE, it should be done at least tentatively, with an indication that the PIE status of this lexeme is uncertain.

Sure enough, if a lexeme is extremely tenously attested, for example in a single modern Germanic dialect, we have reason to be much more sceptical (but even then, it's not completely excluded that it is a survival, it's just much less probable; probability is a matter of degree, after all). But if a lexeme is attested throughout Germanic, even and especially the oldest languages, of all branches, and therefore absolutely solidly reconstructible to Proto-Germanic, and displays no suspiciously un-Indo-European-looking irregularities (or indications pointing to borrowing between branches, or secondary formation, and the like), and just looks potentially old by all appearances, it would be unwise to not at least consider the possibility seriously that it is a lexeme inherited from Proto-Indo-European that just happens to not be attested anywhere else. Cases like this one are known. And it's the task of historical linguists to isolate these cases and distinguish them from patently secondary formations.

In this case, the lexeme is extremely well attested within Germanic and solidly reconstructible to Proto-Germanic, clearly old in Latin too, and borrowing between Germanic and Latin is clearly implausible. For the standards of the LIV, this would actually suffice (to reconstruct a verb), while noting the limited distribution, but not even with a question mark. This Proto-Indo-European reconstruction, to be sure, isn't completely certain. But it is reasonably plausible.

Maybe the theoretical issue here is that people refuse to consider shades of grey. This may be because in the natural sciences, insights are often presented in a black-and-white way: either 100% certain, or 0% certain. But that's not realistic: even in the natural sciences, nothing is literally 100% certain. But usually, things we think of as reasonably certain in the natural sciences are at least 95% certain. In fact, usually much more certain: something like 99.99999% is not rare, even common; and for established theories, the certainty is far greater still.

In the natural sciences, margins of error are usually much lower than in the social sciences and especially humanities, including history. Therefore, we may treat insights as reasonably established that we are in reality only 90% certain about, or even less. The steppe hypothesis is highly probable, but maybe overall it is not that certain, after all; it's only that rival hypotheses are even less plausible (due to their own shortcomings and problems) and therefore probable. If the hypothesis may stand at 80% and no rival hypothesis comes close, that's still pretty good. 99.9% is excellent. In the natural sciences, this would be ridiculously uncertain for a hypothesis to be accepted as established fact. The margins of error would be far too large. At best it would be a promising hypothesis still due to further research ("further research needed" being the typical conclusion).

Therefore, we shouldn't apply the same standards to this kind of research than to the natural sciences. It's even more crucial here to think in degrees of probability. This reconstruction is certainly not highly certain – but there's a good chance that it is was really a lexeme present in the language, certainly good enough to make a page/entry about it here. --Florian Blaschke (talk) 20:55, 9 April 2019 (UTC)

I just thought of a good example in Germanic: German Hader (“dispute, quarrel”) seems to presuppose a Proto-Germanic formation *haþ(a)raz or the like (related to *haþuz (“battle, fight”); Kroonen reconstructs *haþarō, formally identical to the Slavic word, but pace him, Hader is masculine in German, and seems to have always been), which has parallels in Sanskrit and Slavic, but it is only attested in Middle and Modern High German (but in Middle High German, there is only a single attestation in the 14th century, and initially, the word is regionally limited to East Middle German). It's just not plausible to assume that hader was formed late in Middle High German, and it can't be a late borrowing from Slavic (since it reflects Grimm's law and o > a, and the gender doesn't fit either), so the assumption (evidently shared by Kroonen) that it was inherited from Proto-Germanic and is for some unknown reason missing everywhere else in Germanic makes the most sense and is virtually unescapable.

The statement "but a cognate of aqua would be expected to be attested somewhere else in Indo-European, at least in Sanskrit, if it was inherited from Proto-Indo-European!" is then equivalent the statement "but a cognate of Hader would be expected to be attested somewhere else in Germanic, at least in Gothic/Old Norse/Old English, if it was inherited from Proto-Germanic!" Well, the expectation is simply wrong. And the "regional limitation" problem exists in Germanic too: various evidently old formations are attested only in West Germanic (only sometimes in North Germanic too), but not in Gothic. The present of the word do has survived only in West Germanic. Why? We have no idea. But regional limitation does not automatically prevent us from reconstructing a Proto-Germanic form – in the past, scholars often followed a principle that a word had to be attested in East Germanic (Gothic or Crimean Gothic) to be accorded full Proto-Germanic status (ideally in West, North, and East Germanic), but this principle has evidently been abandoned (to be fair, the East Germanic evidence is somewhat limited, but on the other hand, it was accorded so much weight that scholars could reconstruct a word or form attested only in East Germanic to Proto-Germanic, and that's just not following the principle consistently that regional limitation makes a reconstruction dubious). Regional limitation to Europe without attestations in the Asian branches is a fully analogous situation – if we don't proceed that way in Germanic, we shouldn't proceed that way in Indo-European, either. Regional limitation, per se, is not an argument against Proto-Indo-European status. It just one of several possible factors forming an argument that makes it relatively less certain. And sometimes the alternative to "etymology unknown" is really "it could be inherited, after all, and lost everywhere else". --Florian Blaschke (talk) 12:43, 20 April 2019 (UTC)