Paleo-European languages

[Talskubilos]

Nobody in the field considers IE comparative linguistics a "hard science" on a par with physics, let alone mathematics. All reconstructions have to be taken with a grain of salt - this is precisely the reason why they are marked with asterisks. You are beating up a strawman here.

You're right. My complaint was more about the idiosyncratic use of the term "scientific method" by Ares Land for disqualifying my work.

[Ares Land]

I'm always open to discussion, but I'm rather reluctanct to authority arguments based on the established PIE theory is a "hard science" like e.g. physics or mathematics.

Nobody in the field considers IE comparative linguistics a "hard science" on a par with physics, let alone mathematics. All reconstructions have to be taken with a grain of salt - this is precisely the reason why they are marked with asterisks. You are beating up a strawman here.

I have a few reservations. Leaving mathematics aside (they're in a class of their own), you're either overestimating physics or underestimating linguistics.

It's still a science, and I don't see why it should be any less rigorous than the others.
Reconstructions are, of course, useful models. But so is a quark! There's actually a good bit of oversimplification going on in physics, and not everything can be experimented with or objectively measured.
And when it comes to objectively measurable criteria, while it's not linguistics isn't doing so bad, compared to other soft sciences.

That said, neither I nor anyone made unreasonably strict demands about the claims made here. Honestly I'd be satisfied with falsifiable hypotheses, which is really the bare minimum we could ask.

[Raphael]

I saw a claim elsewhere on the web years ago that there isn't really a single scientific method, but more like a collection of scientific methods, appropriate for different fields.

[WeepingElf]

Nobody in the field considers IE comparative linguistics a "hard science" on a par with physics, let alone mathematics. All reconstructions have to be taken with a grain of salt - this is precisely the reason why they are marked with asterisks. You are beating up a strawman here.

I have a few reservations. Leaving mathematics aside (they're in a class of their own), you're either overestimating physics or underestimating linguistics.

It's still a science, and I don't see why it should be any less rigorous than the others.
Reconstructions are, of course, useful models. But so is a quark! There's actually a good bit of oversimplification going on in physics, and not everything can be experimented with or objectively measured.
And when it comes to objectively measurable criteria, while it's not linguistics isn't doing so bad, compared to other soft sciences.

That said, neither I nor anyone made unreasonably strict demands about the claims made here. Honestly I'd be satisfied with falsifiable hypotheses, which is really the bare minimum we could ask.

You are of course right, Ares; and your reservations actually strengthen my point. If there is a difference between physics and linguistics, it is a gradual one. Nobody ever observed a quark, and, according to the theory now accepted among physicists, nobody ever will - quarks are just a formalism for explaining how particles such as baryons or mesons behave. Just as much as reconstructed PIE is a formalism for explaining why so many languages of Europe and South Asia are so similar to each other. Only that a language similar to reconstructed PIE probably was actually spoken somewhere at some time in the past, while isolated quarks never existed

That said, historical linguistics does have its rigorous methods, which must be applied with great care, lest one deceives oneself with false positives, as Talskubilos evidently has done.

[Talskubilos]

Only that a language similar to reconstructed PIE probably was actually spoken somewhere at some time in the past

I'm afraid it wasn't.

That said, historical linguistics does have its rigorous methods, which must be applied with great care, lest one deceives oneself with false positives, as Talskubilos evidently has done.

Ha, ha, ha.

[Nortaneous]

You are right. Hardly anything he posts is worth discussing (though we have a saying in German: Auch ein blindes Huhn findet mal ein Korn 'even a blind chicken sometimes finds a grain'), and our chance to convince him that his approach is flawed and yields far too many false positives is nil. He constantly dodges the question of regular sound correspondences, for instance, usually by pulling a new substratum language or loan pathway out of his hat that "just happens" to yield the sound correspondences he needs for the item in question. As I and others have said several times here and in various threads on the old ZBB, anything (and thereby nothing) can be shown that way!

I'm sure I'm rather untidy in my work, but I'm convinced consistent (a better term than "regular") correspondences are paramount.

For example, if I've linked Caucasian *ttsˀwǝ̄-nHē to IE *yoini is because I've also found other instances where a Caucasian affricate corresponds to IE *y. But this doesn't mean there couldn't be other correspondences as well. In fact, this is the main reason which have leaded me to think the classical monolithic PIE model is inaccurate.

There are (at least!) three sets of regular ("consistent") sound correspondences between English and French, in three different lexical strata:
- the vocabulary inherited from PIE in both English and French
- the vocabulary that was loaned from Latin into English and inherited from Latin in French
- the vocabulary that was loaned from French into English

What, if anything, does this imply about the classical model of Germanic?

[Moose-tache]

I have found a consistent correspondence between people named Steve and people who are jerks (i.e. I have observed it more than once). Clearly this means that the standard model in which names do not determine personality is false. Psychology is not an exact science, after all, which means that all propositions are equally and simultaneously true. If you take umbrage with my "all Steves are jerks" theory, I will happily vindicate my findings by pointing out errors in 19th century psychoanalysis and phrenology. If this still does not convince you, I am willing to move on to my "all Beckies are allergic to figs" theory, also supported by entire ones of data points!

[bradrn]

I have found a consistent correspondence between people named Steve and people who are jerks (i.e. I have observed it more than once). Clearly this means that the standard model in which names do not determine personality is false. Psychology is not an exact science, after all, which means that all propositions are equally and simultaneously true. If you take umbrage with my "all Steves are jerks" theory, I will happily vindicate my findings by pointing out errors in 19th century psychoanalysis and phrenology. If this still does not convince you, I am willing to move on to my "all Beckies are allergic to figs" theory, also supported by entire ones of data points!

You may be on to something…

But more seriously: as an observer of this thread (the conversation can be entertaining!), it is clear that Talskubilos isn’t convincing anyone of his ideas. It is equally clear that no-one has yet convinced Talskubilos of their ideas. Now, I can’t say I see much chance of Talskubilos being convinced he’s wrong — but on the other hand, if Talskubilos does want to convince us of his ideas, Moose-tache does have a valid point here, for all that he presents it flippantly. Talskubilos: if you want to convince us that you’re right here, please give us more than one example at a time! It’s a lot easier to see if you’re right when we know you’re not just overgeneralising from a sample size of n=1. I suspect that won’t fully restore sanity to this conversation, but it would certainly help. (I think I will scream if I see even one more repetition of ‘I have an idea!’ ‘Where’s the evidence?’ ‘Look: ‹one farfetched sound correspondence›’ ‘That’s just one word!’)

[Richard W]

Nobody in the field considers IE comparative linguistics a "hard science" on a par with physics, let alone mathematics. All reconstructions have to be taken with a grain of salt - this is precisely the reason why they are marked with asterisks. You are beating up a strawman here.

I have a few reservations. Leaving mathematics aside (they're in a class of their own), you're either overestimating physics or underestimating linguistics.

It's still a science, and I don't see why it should be any less rigorous than the others.

The relevant comparisons should be with palaeontology or astronomy.

I'm still waiting for the true but falsifiable prediction derived from Grimm's Law.

[Ares Land]

I'm still waiting for the true but falsifiable prediction derived from Grimm's Law.

Uh? We've been through this already.
Like I said, Grimm's Law is in itself a true yet falsifiable prediction.
Though how about "there should be a number of words with similar meanings in Germanic languages and Latin, with Germanic f corresponding to Latin p, and that number should be significantly higher than expected through random chance"?


Comparisons with astrophysics or cosmology are indeed apt. As with paleontology or archeology, though coming up with objective criteria seems easier with linguistics.

There are difficulties, of course, that don't arise with physics. Semantic shift, in particular, is a particularly difficult point. That doesn't make basic scientific criteria unapplicable! Especially since our problem in this thread is chance resemblance; semantic shifts increase the odds of false positive, which is why I insist on regular patterns.

I saw a claim elsewhere on the web years ago that there isn't really a single scientific method, but more like a collection of scientific methods, appropriate for different fields.

There are indeed as many branches of epistemology as there are fields. Some criteria, though, cut across all fields.
Most notably, falsifiability or refutability, as introduced by Karl Popper, is applicable anywhere.

A good litmus test is asking the question: 'What would prove that theory wrong?'. If you can't answer that question, you can safely dismiss the theory.

[Richard W]

I'm still waiting for the true but falsifiable prediction derived from Grimm's Law.

Uh? We've been through this already.

Like I said, Grimm's Law is in itself a true yet falsifiable prediction.
The first prediction, though it seemed good, was actually falsified. However, I may have weakened it by looking for Germanic words that have survived into Modern English.

Though how about "there should be a number of words with similar meanings in Germanic languages and Latin, with Germanic f corresponding to Latin p, and that number should be significantly higher than expected through random chance"?

As you noted, 'similar meanings' is difficult to evaluate, and 'random chance' is difficult to quantify. English fur and Latin pellis is one example, and Latin appareo and English form (as in 'Crystals form as the solution cools') is another.

[Ares Land]

The first prediction, though it seemed good, was actually falsified. However, I may have weakened it by looking for Germanic words that have survived into Modern English.

Um... Not sure I follow? Are you thinking of the exceptions you mentioned earlier?
If so, I don't think this invalidates Grimm's Law! I don't think anyone stated all words needed to be cognates.

Though how about "there should be a number of words with similar meanings in Germanic languages and Latin, with Germanic f corresponding to Latin p, and that number should be significantly higher than expected through random chance"?

As you noted, 'similar meanings' is difficult to evaluate, and 'random chance' is difficult to quantify. English fur and Latin pellis is one example, and Latin appareo and English form (as in 'Crystals form as the solution cools') is another.

[/quote]

Coincidental resemblances can be evaluated (I quoted Zompist's page on the subject earlier). It's just that we don't bother doing it. There are so many cognates between, say, Germanic and other Indo-European languages that we don't need a formal demonstration that it's not coincidence. On the contrary, it's pretty self-evident (well, it should be) that you can't build a genetic connection on the basis of a few words.

Semantics is indeed a delicate point. But AFAIK comparative linguists usually go through those words that haven't undergone dramatic shifts. Then, once correspondences have been established with some confidence, we can check the tricky ones and see if we can't find cognates. In the case of IE, it also helps that work was done on ancient languages. You get less shifts in meaning working with Sanskrit and Latin than French and Hindi.
(Which is incidentally why I'm bothered by proving links through words such as 'décombres'.)

Physics has similar fuzzy edges too. If you've ever done lab work in school, you know it can be hard work to get data that fit the neat graphs in the handbook; and we need elaborate set-ups to eliminate complex effects that are outside the study scope.

[Richard W]

The first prediction, though it seemed good, was actually falsified. However, I may have weakened it by looking for Germanic words that have survived into Modern English.

Um... Not sure I follow? Are you thinking of the exceptions you mentioned earlier?

I see no get-out clause in

You make the hypothesis that Germanic *f will match with *p in Sanskrit, Greek, or Latin.

The lack of a get-out clause is why I was surprised that it did so well.

If so, I don't think this invalidates Grimm's Law! I don't think anyone stated all words needed to be cognates.

The exceptions falsified the prediction. Obviously, the falsified prediction isn't an accurate deduction from Grimm's Law.

As you noted, 'similar meanings' is difficult to evaluate, and 'random chance' is difficult to quantify. English fur and Latin pellis is one example, and Latin appareo and English form (as in 'Crystals form as the solution cools') is another.

Coincidental resemblances can be evaluated (I quoted Zompist's page on the subject earlier). It's just that we don't bother doing it. There are so many cognates between, say, Germanic and other Indo-European languages that we don't need a formal demonstration that it's not coincidence.

I think the prediction is worth checking. It's possible that it might fail, though I think it won't. One technique often used is to tighten the conditions (and so weaken the prediction) by using Swadesh lists or similar to handle the issue of semantics. That does considerably reduce the number of words available. However, I recall a statement that Romanian has only inherited about 350 words (or non-grammatical morphemes?) from Latin.

My implicit point is that making accurate predictions is difficult. I'm not disputing the basic truth behind Grimm's Law. There seems rather to be an unstated level of interpretation, which quite possibly the practitioners can't articulate.

On the contrary, it's pretty self-evident (well, it should be) that you can't build a genetic connection on the basis of a few words.

Which of course raises the matter of the Afroasiatic hypothesis!

Physics has similar fuzzy edges too. If you've ever done lab work in school, you know it can be hard work to get data that fit the neat graphs in the handbook; and we need elaborate set-ups to eliminate complex effects that are outside the study scope.

Do you mean we didn't have dramatic variations in 'g' between the walls (timed drops) and the centre (pendula) of our physics labs? (For the latter, the problem is of course the conflict between 'small' and 'discernible' oscillations.)

[Talskubilos]

Coincidental resemblances can be evaluated (I quoted Zompist's page on the subject earlier). It's just that we don't bother doing it. There are so many cognates between, say, Germanic and other Indo-European languages that we don't need a formal demonstration that it's not coincidence. On the contrary, it's pretty self-evident (well, it should be) that you can't build a genetic connection on the basis of a few words.

But the thing is you can't also prove all these connections go back to a single proto-language.

Semantics is indeed a delicate point. But AFAIK comparative linguists usually go through those words that haven't undergone dramatic shifts. Then, once correspondences have been established with some confidence, we can check the tricky ones and see if we can't find cognates. In the case of IE, it also helps that work was done on ancient languages. You get less shifts in meaning working with Sanskrit and Latin than French and Hindi. (Which is incidentally why I'm bothered by proving links through words such as 'décombres'.)

Generally speaking, the likelihood of semantic shifts tends to be proportional to time, less so in the case of Wanderwörter.

[Talskubilos]

But more seriously: as an observer of this thread (the conversation can be entertaining!), it is clear that Talskubilos isn’t convincing anyone of his ideas. It is equally clear that no-one has yet convinced Talskubilos of their ideas. Now, I can’t say I see much chance of Talskubilos being convinced he’s wrong

In the case I were so, of course.

B— but on the other hand, if Talskubilos does want to convince us of his ideas, Moose-tache does have a valid point here, for all that he presents it flippantly. Talskubilos: if you want to convince us that you’re right here, please give us more than one example at a time! It’s a lot easier to see if you’re right when we know you’re not just overgeneralising from a sample size of n=1. I suspect that won’t fully restore sanity to this conversation, but it would certainly help. (I think I will scream if I see even one more repetition of ‘I have an idea!’ ‘Where’s the evidence?’ ‘Look: ‹one farfetched sound correspondence›’ ‘That’s just one word!’)

I've already listed a few correspondences between reconstructed Gaulish words with no convincing Celtic etymology and Baltic, as well as a couple of putative "satem" substrate loanwords which could correspond to some extinct Indo-Iranian language(s),as suggested by Richard.

[Ares Land]

If so, I don't think this invalidates Grimm's Law! I don't think anyone stated all words needed to be cognates.

The exceptions falsified the prediction. Obviously, the falsified prediction isn't an accurate deduction from Grimm's Law.

I'm not sure I get your point. My initial phrasing left that implicit, but I later corrected to mean, that of course, the prediction does not require 100% accuracy.

One technique often used is to tighten the conditions (and so weaken the prediction) by using Swadesh lists or similar to handle the issue of semantics. That does considerably reduce the number of words available. However, I recall a statement that Romanian has only inherited about 350 words (or non-grammatical morphemes?) from Latin.

I don't really know much about Romanian, but that statement seems surprising. (As I recall there are more words than that shared across all Romance languages)

My implicit point is that making accurate predictions is difficult. I'm not disputing the basic truth behind Grimm's Law. There seems rather to be an unstated level of interpretation, which quite possibly the practitioners can't articulate.

I think historically, and especially with IE, the need to be really explicit about the level of interpretation wasn't really pressing.
Now of course, IE historical linguistics benefits from a wealth of data that's positively decadent and some formal model could be useful...
It'd be interesting to apply some kind of statistical analysis to that supposed Greek-Italic substrate we were talking about earlier. (While the paper is interesting and the hypothesis not unlikely, I still have a hard time chasing off the feeling the authors are trying a little too hard...)

On the contrary, it's pretty self-evident (well, it should be) that you can't build a genetic connection on the basis of a few words.

Which of course raises the matter of the Afroasiatic hypothesis!

I'm really not up to date on Afroasiatic, but I think the morphological similarities are difficult to explain otherwise.
Though for controversies based on insufficient cognates and probably chance resemblances, we only need to look on Greenberg's work on American languages...

Do you mean we didn't have dramatic variations in 'g' between the walls (timed drops) and the centre (pendula) of our physics labs? (For the latter, the problem is of course the conflict between 'small' and 'discernible' oscillations.)

Ah, yeah, in much the same way, I'm sure macroscopic quantum effects in electronics class.

But the thing is you can't also prove all these connections go back to a single proto-language.

That's right: evidence on PIE indeed point at multiple stages, and while the IE family can still be modeled as a tree the 'branches' are really more akin to intersecting subgroups.

In the case I were so, of course.

And that's really the key to the matter... what would prove you wrong?
Or, in other words, taking for instance the Baltic connection you're suggesting, how do we test if it's correct or not?

[WeepingElf]

If so, I don't think this invalidates Grimm's Law! I don't think anyone stated all words needed to be cognates.

The exceptions falsified the prediction. Obviously, the falsified prediction isn't an accurate deduction from Grimm's Law.

I'm not sure I get your point. My initial phrasing left that implicit, but I later corrected to mean, that of course, the prediction does not require 100% accuracy.

Indeed not. I think that what the Grimm's Law example is about is that it was soon observed that there is no shortage of Germanic words which show the "wrong" kind of consonants - voiced obstruents where one would expect voiceless spirants - and another scholar, Karl Verner, showed that these "wrong" consonants occur in specific positions, governed by the regular sound change now called, in his honour, "Verner's Law". (Some other seeming irregularities turned out to be due to a sound change in Greek and Old Indic, known now as "Grassmann's Law".) This is how science progresses: a hypothesis is made, observations confirm it but also throw up exceptions, and a more refined hypothesis is made that accounts for the exceptions.

[...]

On the contrary, it's pretty self-evident (well, it should be) that you can't build a genetic connection on the basis of a few words.

Which of course raises the matter of the Afroasiatic hypothesis!

I'm really not up to date on Afroasiatic, but I think the morphological similarities are difficult to explain otherwise.
Though for controversies based on insufficient cognates and probably chance resemblances, we only need to look on Greenberg's work on American languages...

Afroasiatic is an interesting case. There are not many lexical cognates, yet the morphologies of those six families are so similar that other explanations seem unlikely. (It is the same IMHO with Mitian, though in that case the morphological resemblances are a good deal less perfect, so the argument is weaker, but the difference seems mostly to be a gradual one.)

Amerind, in contrast, is a clear example of mistaking the question for the answer. What Greenberg's lists say is not "These languages are related to each other!" but "Are these languages related to each other?" Of course, asking the right question is the first step towards finding the answer - but only the first step. Amerind did seem to make sense as long as the "Clovis first" model was not disproven by archaeological finds that showed that humans had been in the Americas earlier, but this only shows how hazardous it is to build a language family hypothesis on extralinguistic evidence.

[...]

But the thing is you can't also prove all these connections go back to a single proto-language.

That's right: evidence on PIE indeed point at multiple stages, and while the IE family can still be modeled as a tree the 'branches' are really more akin to intersecting subgroups.

Exactly. Talskubilos is just beating up a strawman when he attacks the "monolythic" (sic!) PIE. The family tree model has its limits, and the wave model is IMHO more accurate. Just look at any dialect continuum, it doesn't really matter which ones. What you will see are intersecting isoglosses - innovations that have spread through parts of the continuum and often overlap with each other, such that it is hard to draw a family tree. And guess what IE was 4,000 years ago? Right - a dialect continuum! And indeed, we do see intersecting isoglosses in IE. For instance, Germanic shares some innovations with Italic and Celtic, and others with Balto-Slavic, which in turn shares some other innovations with Indo-Iranian. Etc. Loans between dialects are also a thing, as are blurry isoglosses. (Ever heard of the Rhenish fan, where the line between "with High German sound shift" and "without High German sound shift" is different almost for every single word?)

Yet, Talskubilos's ideas are jumpy and adventurous, and the evidence he adduces in most cases insufficient. That doesn't necessarily mean that his ideas are wrong, but the burden of proof is on him, and as long he doesn't bear it, they are not worth much.

[Richard W]

If so, I don't think this invalidates Grimm's Law! I don't think anyone stated all words needed to be cognates.

The exceptions falsified the prediction. Obviously, the falsified prediction isn't an accurate deduction from Grimm's Law.

I'm not sure I get your point. My initial phrasing left that implicit, but I later corrected to mean, that of course, the prediction does not require 100% accuracy.

Is this what you mean by correcting:

As for Grimm and Werner's law, yes, the rule worked well enough to get the whole field of comparative linguistics started

Don't bother too much with the exceptions. We really specify that the correspondences mostly work; exceptions are to be expected.

You also mentioned Verner's law, but that's not relevant to the prediction.

I thought that for a test using the 'scientific method', a prediction had to be right. I don't think 'X or not X' is falsifiable, except possibly when X is meaningless.

On the contrary, it's pretty self-evident (well, it should be) that you can't build a genetic connection on the basis of a few words.

Which of course raises the matter of the Afroasiatic hypothesis!

I'm really not up to date on Afroasiatic, but I think the morphological similarities are difficult to explain otherwise.

Checking back, Robert R. Ratcliffe reckoned that Ehret and Orel & Stolbova agreed on 70 items of vocabulary, so perhaps we have enough vocabulary. I'll bet Omotic is painfully short of common vocabulary, though. Curiously, Aroid (a.k.a South Omotic) has plausible morphological agreements with both common AfroAsiatic and Nilo-Saharan.

Though for controversies based on insufficient cognates and probably chance resemblances, we only need to look on Greenberg's work on American languages...

Some of what I see of the rejection of some relationships is disconcerting - 'Pan-Americanism', 'No, also shared by...'. We need better filters, but all I can think of is 4- or 5-way comparison. Unlike RF filters, there's a limit to the data we can gather.

[Richard W]

Amerind did seem to make sense as long as the "Clovis first" model was not disproven by archaeological finds that showed that humans had been in the Americas earlier, but this only shows how hazardous it is to build a language family hypothesis on extralinguistic evidence.

But is this earlier settlement linguistically relevant?

[bradrn]

But more seriously: as an observer of this thread (the conversation can be entertaining!), it is clear that Talskubilos isn’t convincing anyone of his ideas. It is equally clear that no-one has yet convinced Talskubilos of their ideas. Now, I can’t say I see much chance of Talskubilos being convinced he’s wrong

In the case I were so, of course.

Well, yes, of course. I’m talking in hypotheticals here: my point was that you aren’t going to be convinced that you’re wrong — and neither is everyone else going to be convinced that they’re wrong. (It would be convenient if English had a productive subjunctive!)

B— but on the other hand, if Talskubilos does want to convince us of his ideas, Moose-tache does have a valid point here, for all that he presents it flippantly. Talskubilos: if you want to convince us that you’re right here, please give us more than one example at a time! It’s a lot easier to see if you’re right when we know you’re not just overgeneralising from a sample size of n=1. I suspect that won’t fully restore sanity to this conversation, but it would certainly help. (I think I will scream if I see even one more repetition of ‘I have an idea!’ ‘Where’s the evidence?’ ‘Look: ‹one farfetched sound correspondence›’ ‘That’s just one word!’)

I've already listed a few correspondences between reconstructed Gaulish words with no convincing Celtic etymology and Baltic, as well as a couple of putative "satem" substrate loanwords which could correspond to some extinct Indo-Iranian language(s),as suggested by Richard.

You’re talking about these ones, right?

1) Gaulish *borwā 'sludge' ~ Baltic *purwa- 'dirt, marsh'
purported Celtic etymology: *borwā 'hot spring' (Ablauting variant of *berwā 'brew, cook'). Phonetics is good but semantics is bad.

2) Gaulish *komboro- 'heap, accumulation' ~ Baltic *kumb(u)r- 'soil elevation, hill'
purported Celtic etymology: *kom-bero- 'confluence (of rivers)'. Phonetics and semantics are both problematic.

3) Gaulish *sant-ikā 'ladle, milking vessel' ~ Lithuanian sámti- 'ladle, wooden spoon'* ~
purported Celtic etymology: *sfanyā (feminine variant of *sfenyo- 'teat, pap'). A complete disaster, due to improper reconstruction of the Gaulish form.

On the other hand, there're some traces of "satem" loanwords elsewhere, seemingly from a non-Baltic source:
- Cisalpine Gaulish karnitu '(he) erected'**, Lepontic karite- '(he) made' ~ Sanskrit karóti, kr̥ɳóti 'to do, to make'
- Basque (Biscayan) sarda 'fish school' ~ Sanskrit śárdha-, śardhas- 'host, troop'

*Also Etruscan śanti 'a kind of vessel'
**There's a purported Celtic etymology from *karno- 'heap of stones'.

The problem, in this case, is that having only three correspondences is nearly as useless as having only one: as zompist points out, this is easily within the bounds of chance resemblances. In fact, you can find more correspondences between Mandarin and Quechua than you have between Gaulish and Baltic!

Really, here is what we want to see from you: a list of sound correspondences, backed up with significant data (at least 20 words or so, if not more), preferably without making up unattested words. If you can do that, I suspect that you’ll start seeing a lot fewer complaints.