RE : [tied] Re: North of the Somme

Apologies if this comes over multiple times. Yahoo is refusing me the sent-message confirmation.

> > > > You're not paying attention. Let's go over this again - there
> > > > are two Baltic-Finnic words here, and you seem to be confused
> > > > as to what applies to what:
> > > > *piki "pitch", a trivial loan from Germanic (which might
> > > > itself be a substrate loan, but that's not relevant for BF)
> > > > *pihka "resin", regularly cognate with Khanty *peG@...; this
> > > > link means it cannot be a West European substrate loan, and
> > > > confirms that the *h is from former *S.
> > >
> > > No, *you* are not paying attention. I said 'substrate loan',
> > > not 'West European substrate loan'. A substrate common to
> > > FU/Uralic and northern IE.
> >
> > How far do you think this stretches exactly?
>
> Oh well. I'll repeat myself. The language I'm talking about is Kuhn's ur-/ar- language and Schrijver's language of geminates and language of bird names, all three of which I think is one and the same language.

Well, I've yet to spot any wider a/u alternation within Uralic substrate loans, and "having geminates" or "having names for birds" are not sufficient grounds for identifying a language.

I took a look at the message you link'd, and that discusses a Samic word, as well as the idea of a substrate carrying Uralic *maxi westward. Nothing seems to require said substrate to be in contact with Ugric.

> > Khanty is spoken a few thousands of kilometers away from the
> > Germanic hartlands. I don't buy the idea that pre-IE/Uralic
> > northern Europe/Siberia spoke only one language.
>
> Because?

Principle of uniformitarianism. Like Brian was saying in somewhat more detail.

Of course, your answer makes it fine again:
"Individual languages may hide behind this. THey are not to be taken more literally as a single language than Krahe's Old European."

If we allow for a family, we don't have to do with only trivial correspondences however. See later for a proposal.

> At least read Kuhn's articles.

Getting there…

> > > > > Anyway, you were saying
> > > > > you shouldn't have taken the easy option
> > > > > of choosing Germanic in the first place.
> > But I'm positing Germanic origin only for Finnic *piki, and this
> > link works without problems AFAICT (and as you say, it is also
> > the easiest explanation).
>
> Except now you have a BF *pis^k- different from Gmc. substrate *pikk-,

There are two different BF words anyway, *piki and *pihka (which survive intact into modern Finnish). No proposal is going to turn them into one and the same. If we are to pick one of them as an immediate cognate to the Germanic, considering all of semantics, existing loaning patterns, and external Uralic cognates, it should be *piki.

> after which you declare they're probably related but you don't know how. Occam no like.

I don't think you're beyond "working hypothesis" yet either.

> You do it by the book, I've gathered that much.

Yes, I tend to trust the standard view on a matter unless given good reasons not to. "It looks kinda similar to something else" is not a good enuff reason to abandon the idea that Finnic and Khanty descend from *piSka. If you find some more words where BF has *hk or *ht corresponding to an IE plain dental or velar, I'll be interested.

> > And to repeat the part you're failing to grasp with *piSka:
> > Khanty *L (voiceless lateral fricativ) can in no way come from
> > "preaspiration" or anything of the sort.
>
> How would you know that?

Well OK, by no kno'n soundlaws.

> > Taken together with Finnic, this unequivocally points to *-Sk-.
>
> Yes, taken together with Finnic and nothing else. But the languages around them swarm with lookalikes which must be ultimately related,

Germanic is not quite "around" Ob-Ugric. And looking at your other list of lookalikes, they're semantically quite some ways apart, having to do with milk rather than resin.

I keep seeing this "anything with superficial resemblance MUST be somehow related" attitude, but there is nothing that prohibits similarities from occurring by coincidence.

Anyway:
*piji "to milk": Finnic is proposedly a Baltic or Iranian loan. Hungarian /e/ from *iji is not acceptable. The comparision to Mordvinic, which would estabilish *peD'a, seems a little better, but still, we would expect Hungarian /J/ (palatal stop).

*pekta "to churn": irregular consonant correspondence of *-tt- vs. *-xt-; we would expect *-tt- vs. *-t- (from *tt), or *-ht- vs. *-xt- (from *kt). Probably from some substrate. May be related to *peksa- "to beat" (also quite irregular).

> > If we want to devise a deeper connection, I would rather compare
> > this with *pec^a.
>
> If you leave out one of the supposedly reconstructed FU words your connection won't get very deep.

What word am I "leaving out"?

> > > > > Not if the loans was later.
> > > >
> > > > Too hypothetical for my taste, I like the contamination
> > > > explanation
> > > > better (thanks for bringing the 2nd word to my attention BTW).
> > >
> > > Well, suit yourself. But the presence of two similar-sounding
> > > reconstructions for similar-meaning sets of cognates should
> > > have alerted you to suspect loan.
> >
> > To clarify: I don't like positing "substrate alternation" or
> > other non-explanations
>
> ??? What do you understand by 'substrate alternation'?

*n vs. no *n in this case. We can say "OK, it's because of a substrate does that too" but that still leaves the distribution and the causes of the alternation ultimately unexplained.

> > (strikes me as akin to sweeping dirt under a
> > carpet)
>
> The substrate alternations is the dirt people most often sweep under the carpet (cf. Pokorny *si- "tröpfeln" etc with plenty extensions).

Do two wrongs make a right?

> > for irregularities that can be done solidly away in some
> > other fashion.
>
> But people don't, because it's not possible.

In general, no. I think in this case, the aberrant Mari -n- is easiest explained by influence from the other "conifer" word *pEnV- (which you may notice does exist in Mari too). This is an explanation that leaves nothing hanging viz. *pec^a, therefore preferrable.

> > Now sure, there might still be a relation between the two roots
> > we have remaining after this. That's a whole different topic.
>
> You wish.

OK, exaggerrating a bit. But we shouldn't make things harder than they really are.

> > But I think the evidence points to at least *pec^a being an
> > original Finno-Permic root, not something that was independantly
> > loaned to each branch.
>
> Okay, so
> pec^ä ~ penc^ä 'Kiefer, Föhre; Pinus sylvestris' FP
> is an original Finno-Permic root, and
> pihka FU "resin" (I persevere)
>
> are not related, at least not in a way you'd care to explain?

No. I'm saying they're original to the FP/FU/Uralic level, and any connections between them would lie deeper. For example - at some pre-Uralic stage, we have a root *pic^a and a derivativ *pic^ka (*-ka is a common Uralic nominalizer); then *i > *e in an open syllable before *a (such roots are rare), and *c^ > *S before *k.

We might also imagine that this original *pic^a is cognate to the IE words by palatalization.

(Disclaimer: these sound changes are made up on the spot and only serve to illustrate the idea.)

> > > What's insecure about *s´ala-, *kansa- and *sal3- ?
> >
> > *kansa "people": Outside of Fennoscandia, the only posited
> > cognates are Udmurt kuz, Komi goz, which do not correspond even
> > to _one another_ (viz. the initial stop voicing; otherwise
> > possible from *kansa). Also, they mean "pair". Germanic > Samic
> > contacts are kno'n to exist so that doesn't pose a problem.
>
> Please explain how Germanic-Lapp contacts solve the problem.

Like Brian says: by making Finnic & Samic Germanic loans. (It doesn't even need to be pre-Grimm, only older than Gmc *x > *h initially, and the Samic changes *a > *ó and *ns > *ss.)

> > Question remains what should we make of the Permic words, but
> > it's clear this cannot be an inherited Uralic word.
>
> How about ascribing it to a substrate?

That's fine for me. Do we need one behind Germanic *hansa too?

> > And I see no "influence" taking place here at any step - it's
> > quite straightforward.
>
> Erh, what? Did I say 'influence'?

"I see that so many times: Uralic or FU word gets influenced in Finnish by some Gmc/IE word which happens to sound like it and mean something similar."

I was specifially wondering here what you meant by "Uralic words getting influenced".

> > *s'ala "elm": Finnic *sala- and Mari *Sol are back-vocalic,
> > Mordvinic *s'äl'ej and Hungarian /sil/ front-vocalic. So one
> > possible explanation might be:
> >
> > 1) original Uralic root *s'äla- "to cut"
> > 2) a tree name "crack willow" is derived from this
> "cut" <-> "break" ?? I don't get it.

The Uralic verb has a variety of meanings - "to cut", "to breik", "to splinter" etc, the last specifically found in Finnic.

> > 3a) a secondary (tertiary?) meaning of "elm" develops
> > 3b) Finnic and Mari, under IE influence, revert to front-vocalism
>
> Ok, so they are influenced by the IE *sal-ik- which happens to sound and mean similar?

Which is entirely reasonable if IE and Uralic turn to be genetically related. Or if one is an ancient loan from one into the another (possibly via a substrate). Etc.

> > Alternate scenario:
> > - Finnic *salaka is an older IE loan (the substitution *s > *h in
> > *halaka is typical of Germanic loans only)
>
> ?? I thought s > h was a BF development independent of the source of the word (which makes halaka an old, salaka a young substrate loan)?

No, *s (alveolar) remains; *S (postalveolar) > h.

The loan substitution *s > BF *h is found only in old Germanic loans, not in Baltic, Slavic, Iranian nor Swedish. PGermanic, not having an *s <> *S contrast, apparently had a freer realization for *s, which was identified as closer to BF *S than *s. (It also seems there was a dental allophone, since Gmc *st- > BF *s-.)

(And for completeness, later *s > *h development occurs conditionally in some cases, but initially only in unstress'd function words, eg. 3PS pronoun *sän > hän)

> > - The "elm" words are unrelated, and since the vocalic
> > correspondences are irregular, likely from a substrate of some
> > sort.
>
> So they borrowed a word for "elm" which was similar to their own word for "willow"?

No, in this alternate scneario I'm suggesting there to have been no original Uralic tree word of this shape (seems good also since there exists a separate Uralic word for "willow", *paja) and that Finnic is from some other origin.

Actually, *salik- doesn't seem to exist in Balto-Slavic or Iranian, so it looks like the Finnic s- variant would have to be from some substrate, if not inherited.

> > > And all UEW has to say is 'cf.'
> >
> > Well, I've said it before: the UEW works mostly as a repository
> > of data. (Aluckily a more up-to-date Uralic etymological
> > dictionary is in the works.)
>
> I wasn't criticizing UEW's reconstruction. I was criticizing its nonchalant attitude to the mysterious fact that the *sal-ik- root is distributed over all of the FU and IE fanilies.

Yes, it does that too.

> > > (BTW Danish 'savl' "saliva", guessed, with '?', to be from 'a
> > side form' *sak- of *sag- "humidity", seems to suggest a
> > substrate word *saG-l- instead, which could > *sal- or
> > (metathesis) *sal-G-,
> >
> > What does saliva have to do with willows anyway?
>
> Slush. Mud. Unhealthy fluid.

Saliva has limited antibacterial qualities actually, and "liquid" hardly cuts it.

OTOH we might bring up the fact that willow bark contains salicylic acid. "Antiseptic" and "painkiller" are still not the same, but I think closer than "saliva" and "mud".

> > > Remember also that the fact that some Baltic Finnic dialect was >
> > > spoken in eastern Poland
> >
> > No. This is very much not a fact, but your own hypothesis.
> No. It's documented by
> 1) Hydronyms
> 2) Roman authors.

*walga is not a BF hydronym. If you want to find a specifically BF substrate, look for eg. *joki "river". You've probably taken a look at Janne Saarikivi's "Substrata Uralica"?

Roman authors don't provide detail'd information on linguistic affiliations, I'm afraid. Nothing requires the "Aesti" to have been ancestors of the Estonians - I would rather suggest that the Finnic tribes who settled a former area of their took also their name from them (especially since by the Roman times, there likely was no Finns/Estonians/Veps/etc. distinction yet.)

So there are several possible interpretations here. Maybe you do have sufficient evidence to even call it a theory, I couldn't tell yet; however, in no way is this a _fact_.

(Or if you have records from Poland written in a Baltic-Finnic language, do tell!)

> It is not advisable to keep the question of loans to and between FU and IE languages separate from the question of where they were in contact with each other and with a substrate language.

Agreed.

> > Derive *pik- from *pei, fine; also derive *piT- from *pei, maybe.
> > However, I don't see *piT being derivable from *pik (or vice
> > versa).
>
> But both are from *pit-ik- etc.

How so? I see no *-t- in the *-k forms.

> > One thing I think makes your thought process hard to follow is
> > that you list huge amounts of data but do not explain how exactly
> > do you think it's all related.
>
> Don't hesitate to ask.
>
>
> Torsten

Don't hesitate to just explain outright. ;)

John Vertical