Page 2 of 21

Re: Which Sciences Are the Most Useful?

Posted: Sun Jan 20, 2019 5:12 am
by teo123
That's a misconception, much like Ken Ham's belief that evolution is "historical science" rather than experimental science.
I was simply quoting what's written in both my sociology textbook and my psychology textbook. That's used as an example of the difference between observation and controlled experiments. That is, that while science should use controlled experiments where applicable (they can also be used to show causation, while observation can only show correlation), relying on systematic observation isn't necessarily a sign of pseudoscience.
There's no reason to believe those toponyms exist.
If "issa" meant "health-giving spring", then we indeed shouldn't expect those toponyms to exist (or, if they do, they should be confined to places where ancient thermae were built upon a spring, as in Dakovacka Breznica or near the Hadrian's thermae on Brijuni).
But if it meant, for example, "sea island", then we would expect those toponyms to exist and be confined to sea islands. Is that what we see when we look at the map? No, the element *issa~iasa appears once on an island known for the remainings of large ancient baths and two times in continental Croatia, both names of the places with large ancienct thermae. In fact, the element *malit~melit appears to be have been a word meaning "sea island", since it appears multiple times in Croatia (and once relatively near Croatia, but outside of it) and is confined to sea islands.
If "issa" meant "stream", then we would expect that element to appear all over Croatia, especially in hydronyms. Is that what we see when we look at the map? No, in fact, the word for "stream" appears to have been *ser.
We can try to postulate all sorts of meanings to "issa" (a place near water, town...), but then what we would expect to see doesn't agree with what we see on the map.
Also, I am not sure if that's relevant here, but what we see on the map is not the only kind of argument you can make here. A postulated word is more likely to be real if it has a plausible etymology. For instance, the well-known Indo-European root *yes (to boil, to foam) had a noun-form *yos (water source), and the Illyrian toponymic element *issa~iasa almost certainly comes from it.
So, I don't quite understand what you mean with "There's no reason to believe those toponyms exist.".
or show that they came from the same sources
Claiming that similarly sounding toponyms relatively near each other don't share the same root just because that wouldn't fit your theory is an obvious ad-hoc hypothesis, right? Especially if they appear to have the same meaning.
The only reason that might not be fallacious is if we have some relevant data from modern languages. For instance, we know that in most modern Indo-European languages, there is a word similar to *sal that means "salt". However, when we look at the toponyms, we see that the element *sal generally appears in the names of rivers, especially ones with waterfalls. Some people think they come from a different language family, but I think it's more likely related to the fact that in some old Indo-European languages, such as Latin, the words for "to jump" and for "salt" were near-homonymous (such as Latin "sal" and "salire"), and that languages don't like homonyms so the more rare word was displaced. Nevertheless, it's only because of that that it's reasonable to claim that the toponyms such as "Salapia" and "Sala" (the ancient name for many rivers, including Jadro and Zala) aren't related. Otherwise, it would have been an ad-hoc hypothesis.
You can always get incredibly lucky and find a Rosetta stone, but banking on dumb luck like that for some unlikely artifact to settle a dispute is not how science works.
Finding or just reconstructing a few new ancient toponyms wouldn't be quite comparable to finding a Rosetta stone. If some modern Croatian toponym ended in *-jes, but not in *-l(i)jes or *-r(i)jes (because of the metathesis of the liquids in Old Croatian), you could safely say that it comes from from an Illyrian word ending in *-issa with short 'i', because we see that the Illyrian sound Roman scribers transcribed as short 'i' gets borrowed as a 'yat' into Old Croatian, and the only way a modern Croatian toponym could end in *-jes but not in *-l(i)jes or *-r(i)jes is if it ended in yat+s+back-yer in Old Croatian and has gone through a Croatian dialect in which Old Croatian yat turned into /ie/.
That said, even finding a Rosetta Stone for deciphering Illyrian can't be compared to a supernatural event. There are probably hundreds of Messapian inscriptions surviving to this day waiting to be transcribed (and Messapian is generally believed to be a dialect of Illyrian), how do you know there aren't any multilingual inscriptions? And it doesn't even have to be a multilingual inscription, there were a few languages translated recently because somebody accidentally found an inscription that's incredibly easy to guess the meaning of and to translate. You know, the Ventris'es "Poloi polei" tablet and the Hrozny's "Nindaan ezzateni waatarma ekutteni" tablets?
A supernatural event would be incoherent with what we know about the universe, finding a tablet that's especially helpful in deciphering Illyrian is not.
the divinity of Jesus is falsifiable too because Jesus or God could come down to Earth and confirm or deny it.
I am not sure I understand what you are saying. How would Jesus coming down to Earth falsify Christian theology? That would, whatever he said, just confirm Christian theology. OK, maybe he could say something that would disprove some minor conclusion of the mainstream Christian theology (as if there was a consensus on anything), but the basic tenets of Christian theology (or any other theology, for that matter) are still unfalsifiable even in theory.
If you're talking fringe stuff like string theory, sort of, but those things are also models and not exactly hypotheses. That's another issue.
Can you elaborate on that, please?

Re: Which Sciences Are the Most Useful?

Posted: Sun Jan 20, 2019 1:51 pm
by brimstoneSalad
teo123 wrote: Sun Jan 20, 2019 5:12 amThat is, that while science should use controlled experiments where applicable (they can also be used to show causation, while observation can only show correlation), relying on systematic observation isn't necessarily a sign of pseudoscience.
It's not about observation vs. experiment (which is kind of a false dichotomy).

You really need to study more astronomy.
Evidence of the Big Bang is a good place to start: https://www.space.com/40370-why-should- ... -bang.html
Falsifiable theoretical models create strong predictions, and those predictions can be verified by observation in the same way as any experiment. This is as distinct from soft sciences where the predictions are weak at best and don't change things in themselves based on observation.
teo123 wrote: Sun Jan 20, 2019 5:12 amIf "issa" meant "health-giving spring", then we indeed shouldn't expect those toponyms to exist (or, if they do, they should be confined to places where ancient thermae were built upon a spring, as in Dakovacka Breznica or near the Hadrian's thermae on Brijuni).
No, they should not, because both people and words can easily move.
The lack of presence of these words wouldn't prove anything, because you'd have a cascade of very plausible explanations to ad-hoc explain either one.

Even if that observation increased or decreased the probability of a theory being true, because you may have already known of the lack of presence of these words, it creates a formulation bias and doesn't increase the probability of YOUR theory being true.

You need to study more of how statistics work in science, P-hacking, etc.
Probability can be very counter-intuitive. It's much like the Monty Hall problem. ONLY observations made after the formulation of the theory affect its probability of being true (and that's only if those observations very strongly reflect on the probability of the theory being true, IOW if they would actually prove it false and not be trivial to explain away like the lack of presence of "issa").

If we did not discover background radiation, then the Big Bang theory would have been false, period.
And crucially, we did not know about background radiation before the Big Bang theory was formulated, and the Big Bang theory predicted it. That's evidence, and that's what makes it a theory.

teo123 wrote: Sun Jan 20, 2019 5:12 amClaiming that similarly sounding toponyms relatively near each other don't share the same root just because that wouldn't fit your theory is an obvious ad-hoc hypothesis, right? Especially if they appear to have the same meaning.
And yet it is plausible, which makes all of this a pretty soft science.
The point is that the presence or absence of any of this is not a smoking gun, so does little to influence the probability of a theory being right, and only does anything at all if these observations were only made *after* the theory was proposed.

Forming a model to fit the evidence, like after the fact changes, isn't valid. Your model needs to be predictive of something that you don't already know, and then you need to figure out the probability of negative observations invalidating your model (which is a very difficult endeavor when it's anything less than effectively 100%, I can explain briefly how to do this if you're interested), and then you need to observe that thing to increase or decrease the probability of your model (now a theory) being true.

I'm guessing that you didn't follow this procedure. And if you didn't follow this procedure and your publication was accepted in a peer reviewed journal then that's evidence that the field as a whole is a pudding-soft science, if it's a science at.
teo123 wrote: Sun Jan 20, 2019 5:12 amSome people think they come from a different language family, but I think it's more likely related to the fact that...
That means nothing at all if you have not done the hard work of empirically establishing that probability.
That's one thing that makes it very difficult to harden a soft science.
teo123 wrote: Sun Jan 20, 2019 5:12 amFinding or just reconstructing a few new ancient toponyms wouldn't be quite comparable to finding a Rosetta stone.
I mean a literal ancient document which intentionally confirms these things in an undeniable way.
No, what you're doing isn't anything like a Rosetta stone.

teo123 wrote: Sun Jan 20, 2019 5:12 amIf some modern Croatian toponym ended in *-jes, but not in *-l(i)jes or *-r(i)jes (because of the metathesis of the liquids in Old Croatian), you could safely say that it comes from from an Illyrian word ending in *-issa with short 'i', because we see that the Illyrian sound Roman scribers transcribed as short 'i' gets borrowed as a 'yat' into Old Croatian, and the only way a modern Croatian toponym could end in *-jes but not in *-l(i)jes or *-r(i)jes is if it ended in yat+s+back-yer in Old Croatian and has gone through a Croatian dialect in which Old Croatian yat turned into /ie/.
Safely? Again, with what probability?
There are exceptions to that kind of thing everywhere. You'd need to tally every example of this being true and every counter-example to get a sense of the probability, but you need to do it with very good methodology. That's something that deserves to be a peer reviewed paper, not guessing at what is safe to say or not.
teo123 wrote: Sun Jan 20, 2019 5:12 amThat said, even finding a Rosetta Stone for deciphering Illyrian can't be compared to a supernatural event.
If it's true, then it's part of nature.
Either way you're relying on an outside force (god, or an ancient linguist who did your work for you) to prove something.
teo123 wrote: Sun Jan 20, 2019 5:12 amhow do you know there aren't any multilingual inscriptions?
There could be, but anything that's not copied has a tendency of degrading and being lost.
And again, you're relying on an outside force to do your work for you.
teo123 wrote: Sun Jan 20, 2019 5:12 amA supernatural event would be incoherent with what we know about the universe, finding a tablet that's especially helpful in deciphering Illyrian is not.
Plenty of ways to ad-hoc explain "god" and force compatibility with physics (god of the gaps).
teo123 wrote: Sun Jan 20, 2019 5:12 amI am not sure I understand what you are saying. How would Jesus coming down to Earth falsify Christian theology?
"Oh hai guyz, I'm just a hooman, I was abducted by aliens. Sorry about all the confusion."
teo123 wrote: Sun Jan 20, 2019 5:12 am Can you elaborate on that, please?
Things like string theory take existing information and form a model from it; kind of a way of visualizing and understanding the extant data. It's actually a very controversial area of physics, since there are some serious criticisms of its usefulness.

This happened a few years ago, and gives some insight:
https://www.wired.com/2010/09/stringy-quantum/
Mathematician Peter Woit of Columbia University, author of the blog Not Even Wrong, thinks even claiming that the new paper is a test of quantum entanglement is going too far.

"Honestly, I think this is completely outrageous," he said. Even if the math is the same, he says, testing the quantum entangled system would only tell you how well you understand the math.

"The fact that the same mathematical structure appears in a quantum mechanical problem and some model of black holes isn't even slightly surprising," he said. "It doesn't mean that one is a test of the other."

Witten takes a more optimistic view of the theory's chances, pointing out that the mathematics of string theory have turned out to be coincidentally useful in other areas of physics before.

"In general, this kind of work shows that string theory is useful, and in fact by now it has been useful in many different ways," Witten said in an email to Wired.com.

"One might surmise that a physics theory that has proved to be useful in so many different areas of physics and math is probably on the right track," he added. "But that is another question."
In that snip, Witten's implication here verges on pseudoscience because he's suggesting a probability of being true without anything but his general feelings; you need to use objective methodology to demonstrate that.
That's the same kind of thing that's going on with linguistics here.

Re: Soft Sciences Vs. Hard Sciences

Posted: Mon Jan 21, 2019 4:28 am
by teo123
You really need to study more astronomy.
Ah, dude! I've put a lot of my time studying linguistics, I've put a lot of my time studying computer science, and I've put an insane amount of time studying various pseudosciences (Flat-Earth...). If that's not enough to have an educated discussion about philosophy of science, then nothing would be.
Evidence of the Big Bang is a good place to start
So, if I understand you correctly, that guy who is credited for the Big Bang theory proposed a way to test it, that couldn't be properly done using the equipment available back in the day. So, how is that different from what I was doing? You never know what kind of equipment will be available to archaeologists and people from other historical sciences ten years from now. The number of ancient manuscripts and inscriptions that can be read is increasing every day and will increase even faster in the future.
No, they should not, because both people and words can easily move.
Yes, but they don't move as randomly as you seem to assume. Have you heard of, for example, the Skok's Second Law of Toponymy? It states that modern names of the islands derive from the ancient names of the largest places on the islands. For instance, the ancient name for the island of Hvar was Pittousa, and the largest ancient place on Hvar, modern-day Starigrad, was called Pharos. Today, the name Pittousa disappeared, but the name Hvar is derived from the name Pharos. For the Croatian islands, that law is about 90% accurate (Dubravka Ivsic published a paper in 2013 in which she tested it using the data from some edition of the Anonymous Ravenian, sorry I can't find it using a quick search, but it's available for free online).
it creates a formulation bias
I think I understand what you are saying, but I think you are commiting a form of Ludic fallacy here. If you knew that a dice has been thrown 100 times and every time six was at the top, would you still assume it's a fair dice? You don't need to make a prediction and throw it again to be reasonably certain it isn't.
And yet it is plausible, which makes all of this a pretty soft science.
It's not plausible if you don't have any data that confirms that. See, how many words from a single language are expected to repeat themselves in toponyms. Water, flow, island, mountain, hill, waterfall, spring, abyss, coniferous forest... Let's be generous and say it's 20. So, how many possible toponymic elements are there? Let's be very generous and say all toponymic elements are in the form of consonant+vowel+consonant, and that the vowel is always ignored (which it isn't). Since there are about 20 consonants in a language, the probability of each element having each form is 1/(20*20)=1/400. So, the probability that some single element of the 20 elements that repeat themselves will be homonymius with some other element is 1-((400-1)/400)^20=4.8%. The chance that you've wrongly postulated a meaning to those elements is a lot higher than that.
Of course, the probability that any single of those 20 elements will be homonymous, under our simplified model here, is 1-(400 P 20)/(400^20)=38%, but for each of those 20 elements, by far simplest explanation is that it's not homonymous with any other. And that's even more true in a more realistic model.
And if you didn't follow this procedure and your publication was accepted in a peer reviewed journal then that's evidence that the field as a whole is a pudding-soft science, if it's a science at.
That's right, go insult people who have spent years studying that... because they don't do P-hacking to make their conclusions look more credible to people like you!
That means nothing at all if you have not done the hard work of empirically establishing that probability.
Probability of what? Of a particular term coming from an unattested language family? That's using the unknown to explain the unknown, and the probability that should be assigned to that is zero, right?
Safely? Again, with what probability? There are exceptions to that kind of thing everywhere.
Well, that question is very hard to answer. Sound laws are considered way more certain than any particular etymology. If some etymology doesn't agree with the sound laws, it's more likely a wrong etymology than an exception to a sound law. The exceptions are numerals and some very basic vocabulary, certainly not the toponyms.
You are probably using the regularity of the English spelling as a proxy to how regular sound changes are. That's wrong, most of the irregularities of English spelling come from the fact that the words were never actually pronounced the way they are spelt today. For instance, "come" was never pronounced with an "o" or an "aw" sound in it, it was actually pronounced more like "koom-uh". It used to be spelt with an "u", however, it was changed to "o" to avoid confusion in hand-written manuscripts ("um" is likely to be misread). The same goes for words like "love". Also, the spellings of the English words are often affected by flase or distant etymologies, such as in "isle", "island" or "doubt".
The exceptions we are talking about here are the results of dialectal diffusion which cause a word to undergo a sound-change two or more times when it should have undergone it only once. An example of that is the English word "bury" and the German words "tausend" and "Zwerg" (the "bury" has undergone both short 'u' to short 'a' change typical of English and the short 'a' to short 'e' change typical of English, when it should have undergone only one, and "tausend" has undergone both 'th'->'d' and 'd'->'t', and "Zwerg" has undergone both 'd'->'t' and 't'->'z'). Such cases are very rare.
I'd expect them to be even more rare in the toponyms (since a toponym is much less likely to be borrowed from another dialect to cause it to undergo the same sound-change twice), but I don't have an idea how to test it. Identification of ancient toponyms is partly based on the assumption that the sound laws in toponyms are exceptionless. For instance, the ancient hydronym "Oeneus" is not connected to the river Una only because "Oeneus", borrowed into Old Croatian, is expected to give something like "Inja" in modern Croatian. Similarly, the ancient hydronym "Murius" is not connected to the river "Mura" because, well, how come it didn't undergo any of the expected sound-changes?
"Oh hai guyz, I'm just a hooman, I was abducted by aliens. Sorry about all the confusion."
And why would it be reasonable to assume that alien is trustworthy?
Things like string theory take existing information and form a model from it; kind of a way of visualizing and understanding the extant data.
Really? I've always assumed it's some hypothesis about the composition of the sub-atomic particles that sounds smart to laymen, but sounds like a bunch of baseless and unfalsifiable assertions to the physicists.

Re: Soft Sciences Vs. Hard Sciences

Posted: Tue Jan 22, 2019 3:18 am
by teo123
Here is the relevant Dubravka Ivsic's paper, Skok's Second Law is discussed beginning with the page 363.
By the way, in the second paper I published, I even entertained the notion that the patterns I see in the toponyms are actually not statistically significant. So, I tried to calculate the p-value of one of the most striking patterns I saw, the disproportionate number of rivers have their names starting in k(+vowel)+r. That is, there are five of them in Croatia: Krka, Karasica (two rivers with the same name), Krapina, Krbavica and Korana. So, the probability of a river name starting with a particular pair of consonants is 1/(20*20)=1/400. Let's say there are around 100 river names in Croatia. So, what's the probability of that pattern arrising by pure chance? Well, that question is a modified version of the birthday paradox. Since I didn't feel confident trying to derive a mathematical formula for it, I wrote a computer program estimating that probability. Here it goes:

Code: Select all

#include <stdio.h>
#include <stdlib.h>
#include <time.h>

int main() {
	srand(time(0));
	int number_of_repeats;
	scanf("%d",&number_of_repeats);
	const int number_of_rivers=100;
	float sum=0;
	int counter[400];
	for (int i=0; i<1000; i++)
	{
		for (int j=0; j<400; j++)
			counter[j]=0;
		for (int j=0; j<number_of_rivers; j++)
			counter[rand()%400]++;
		int max=0;
		for (int j=0; j<400; j++)
			max=(counter[j]>max)?(counter[j]):(max);
		if (max>=number_of_repeats) sum++;
	}
	printf("%f\n",sum/10);
}
So, you enter the number of repeats of the same two-consonant pair, and it outputs the approximate probability of that arrising by chance. If you enter 1, it outputs 100. If you enter 2, it outputs either 100.0 or 99.9. If you enter 3, it outputs a number close to 56%. If you enter 4, it outputs a number close to 5. So, that would already be statistically significant. If you enter 5, it outputs 0.0 no matter how many times you run it. So, the p-value of that pattern is very close to 0. So, I can claim that the Illyrian root meaning "river" (or something like that) is *kar~kur ('u' can be borrowed into Old Croatian as back yer and disappear due to the Havlik's law, 'a' either remains 'a' or turns into 'o' according to the law of the metathesis of the liquids) with close to 100% certainty, right? I mean, that program doesn't calculate the probability of that particular pattern appearing, it calculates the probability of any such pattern appearing in random data.
The P-value of that pattern with "Issa" is easily calculated. Let's assume there are 50 (there are more, but let's say that for the sake of argument) ancient Croatian toponyms known by now. There are five places in Croatia in which it's reasonable to believe there were springs believed to be health-giving: Vis, Brijuni, Daruvar, Varazdinske Toplice and Dakovacka Breznica. The element "Issa" repeats itself 3 times, in Vis, Daruvar and Varazdinske Toplice. So, the P-value of that pattern with "Issa" is (5 C 3)/(50 C 3)+(5 C 4)/(50 C 4)+(5 C 5)/(50 C 5)=5.32E-4.
So, do you think I've added anything new to the story with those calculations? I don't think so.

Also, if you insist social sciences are not real sciences, how can you possibly be against anarchism? If social sciences are not real sciences, anything lawmakers could possibly do is forcing their subjective opinions onto people, and that's authoritarian.

I am not planning to reply any more here, I hope this was enough to make you think.

Re: Soft Sciences Vs. Hard Sciences

Posted: Wed Jan 23, 2019 9:05 am
by teo123
I’d be interested in reading what other people on this forum educated in social sciences, such as @EquALLity or @Jebus, have to say about this issue.
Brimstone, I see you are otherwise a smart guy, but your comments here seem arrogant, uninformed and even somewhat insulting. They make about as much sense as when the Flat-Earthers deny that light (in this case languages) obeys certain laws.
If you think you have something sensible to say in defense of your position (that the field I was studying and published three peer-reviewed papers about, that is the study of the Croatian toponyms, is not a real science), go ahead, maybe you indeed have! But what you’ve done in the last few posts is losing much of your credibility in my eyes, and waste a lot of my time I would use to prepare myself for the electrotechnics test I failed.

Re: Soft Sciences Vs. Hard Sciences

Posted: Wed Jan 23, 2019 1:02 pm
by Jebus
teo123 wrote: Wed Jan 23, 2019 9:05 am I’d be interested in reading what other people on this forum educated in social sciences, such as @EquALLity or @Jebus, have to say about this issue.
I haven't followed the thread but if you can break it down to a couple of specific questions, I will let you know what I think.

Re: Soft Sciences Vs. Hard Sciences

Posted: Thu Jan 24, 2019 3:17 am
by brimstoneSalad
teo123 wrote: Mon Jan 21, 2019 4:28 am
You really need to study more astronomy.
Ah, dude! I've put a lot of my time studying linguistics, I've put a lot of my time studying computer science, and I've put an insane amount of time studying various pseudosciences (Flat-Earth...). If that's not enough to have an educated discussion about philosophy of science, then nothing would be.
Linguistics, as I explained, is a soft science.
Computer science is more of a field of engineering; it doesn't typically apply scientific methodology. "science" being in the name is misleading.
Studying pseudoscience is a good start, but you need some background in a real hard science.
teo123 wrote: Mon Jan 21, 2019 4:28 am So, if I understand you correctly, that guy who is credited for the Big Bang theory proposed a way to test it, that couldn't be properly done using the equipment available back in the day.
At the time, then, there was no reason to believe it over anything else since it couldn't be currently tested. There are a few things like that in physics.
teo123 wrote: Mon Jan 21, 2019 4:28 amSo, how is that different from what I was doing? You never know what kind of equipment will be available to archaeologists and people from other historical sciences ten years from now. The number of ancient manuscripts and inscriptions that can be read is increasing every day and will increase even faster in the future.
What's distinct for something like "we need a higher energy particle accelerator" is that it's something we can build. Linguistics have to rely on luck, and it may very well be that the information you need doesn't and never will exist. You might just need a time machine.
teo123 wrote: Mon Jan 21, 2019 4:28 amYes, but they don't move as randomly as you seem to assume. Have you heard of, for example, the Skok's Second Law of Toponymy? It states that modern names of the islands derive from the ancient names of the largest places on the islands.
That's cool, but it's meaningless without a P value, and P value is meaningless post hoc.

Read this:
https://xkcd.com/882/

It quite cleverly explains why P value is only meaningful in a context where you don't get to select from a pool. What you're doing in linguistics is selecting a correlation, and it doesn't matter if your after the fact calculations of the probability of it being a coincidence are 10% or 0.1%.
You have a large pool to choose from, and THAT fact means you're not practicing good science.

You have to develop a theory, delineate what would falsify it in a statistically meaningful way, and *then* collect the data to do so.
Now, to be fair, a LOT of scientists in the public view are also not practicing good science. Look at Brian Wansink. P hacking is how you get all the research dollars, because it gets you results fast and most people don't understand it well enough to see how those results are worthless.
teo123 wrote: Mon Jan 21, 2019 4:28 amI think I understand what you are saying, but I think you are commiting a form of Ludic fallacy here. If you knew that a dice has been thrown 100 times and every time six was at the top, would you still assume it's a fair dice? You don't need to make a prediction and throw it again to be reasonably certain it isn't.
If the hundred times were hand picked from a pool of billions of billions of dice being each thrown a hundred times, yes.

It's inevitable that SOME of those dice will roll six a hundred times in a row without being "unfair". It would be unusual if they did not.

Again, see that XKCD comic for an illustration.

If you test enough things, some of them will come out appearing to have good P values, but that doesn't make them true.
What you're doing is applying a P value ad hoc to something that already appeared to have a good P value, but there's no good reason to believe that wasn't by chance (that it wasn't the green jellybean).

Again, this is basically textbook P hacking. Some researchers do this, look in a bunch of data for correlations and then ad hoc apply P values. It's a scandal. Bad science.
teo123 wrote: Mon Jan 21, 2019 4:28 amIt's not plausible if you don't have any data that confirms that.
No, that's not how science works. You don't get to assume you're right until somebody else shows another theory is just as likely to be right.
The alternatives are plausible, period.

If you want to show it's less plausible than YOUR theory, then YOU need to do the leg work and disprove them. And you need to do so with every plausible alternative (and there could be dozens, even hundreds, thousands).
The vast array of variables and plausible alternatives is one of the things that keeps soft sciences soft; it's very very hard to boil it down (as done with physics) to a small handful of options.
teo123 wrote: Mon Jan 21, 2019 4:28 amSee, how many words from a single language are expected to repeat themselves in toponyms. Water, flow, island, mountain, hill, waterfall, spring, abyss, coniferous forest... Let's be generous and say it's 20. So, how many possible toponymic elements are there? Let's be very generous and say all toponymic elements are in the form of consonant+vowel+consonant, and that the vowel is always ignored (which it isn't). Since there are about 20 consonants in a language, the probability of each element having each form is 1/(20*20)=1/400. So, the probability that some single element of the 20 elements that repeat themselves will be homonymius with some other element is 1-((400-1)/400)^20=4.8%. The chance that you've wrongly postulated a meaning to those elements is a lot higher than that.
Of course, the probability that any single of those 20 elements will be homonymous, under our simplified model here, is 1-(400 P 20)/(400^20)=38%, but for each of those 20 elements, by far simplest explanation is that it's not homonymous with any other. And that's even more true in a more realistic model.
All of that, I just see the flat-Earther teo is back.
You don't want to accept that you've invested a lot of time in a soft science. That doesn't mean it's not useful or a fulfilling profession, and that doesn't mean you can't be more rigorous and more scientific about it than others, but it's very hard to turn something like this into a hard science. Any of those numbers, and more, are likely very easy to manipulate to P hack.

I'm sure @Jebus knows psychology is a softer science, but that doesn't mean it's not useful. It's also a harder science than linguistics because there's something concrete and testable that actually comes out of clinical psychology: either you help people get better, or you don't. There's something concrete there, and you can use good methodology to evaluate interventions. It's also applicable in testable ways to advertising, propaganda, etc.

What is the parallel there in linguistics? So what if you get the root of a word wrong? Is there a clinical application? Can we do some actual test to confirm it's right or wrong on the basis of people understanding or misunderstanding something? No.
The best it can do is help us understand writings in dead languages, but usually all we can do is say if the results are intelligible or not. Doesn't mean it's the correct translation. Very VERY rarely writing might lead us to a prediction about some archaeological discovery. You translate something and it tells you there's a pyramid filled with cat mummies at the mouth of some river, and then you find exactly that... well, good job. That's pretty convincing. NOW you can argue that you translated it right if it made a concrete prediction. But you're relying on an astronomically rare occurrence (which in a lifetime working in linguistics you'll probably never see) and it's accomplished what exactly? Stocked a museum with cat mummies or something.
Meanwhile a working psychologist can make and confirm concrete predictions very reliably with a few years of study.

Linguistics is a soft science. And it's among the softest of the soft. Again, doesn't mean it's useless, doesn't mean you can't do better. But raging against the fact that it's a soft science isn't going to help your case. It's making you look delusional and unhinged.
teo123 wrote: Mon Jan 21, 2019 4:28 amThat's right, go insult people who have spent years studying that... because they don't do P-hacking to make their conclusions look more credible to people like you!
Honest linguists will admit linguistics is a soft science. If they're really honest they'll also admit that it doesn't have that much of material value to offer humanity outside of contributing to our understanding of ancient human history... which, unfortunately, is a mostly academic interest... at least outside of biblical/quranic stuff (which has real implications because people believe in that stuff).

There's also, at best, some efforts at documenting and preserving contemporary human languages which are going extinct. But why? What material value do these languages hold? You basically have to appeal to a very exaggerated notion of how linguistics affect human cognition, and claim that with these languages whole ways of thinking will go extinct, and we may lose the ability to solve some real human problems in other sciences because we couldn't think in a certain unique way. That's not very plausible.

People do what makes them happy, and sometimes that's exploring history and culture rather than working in a hard science and discovering cures for diseases. Maybe it's very writing fictional novels, making movies, games, whatever.

I'm not here to hate on people who study soft sciences, or who do other work in the humanities and entertainment. Doing something other than curing cancer doesn't make you a bad person.

I'm just telling you it's not a hard science, and your experience in it doesn't really qualify you to understand what hard science is about.
You need to study a hard science to understand that well.
teo123 wrote: Mon Jan 21, 2019 4:28 am
That means nothing at all if you have not done the hard work of empirically establishing that probability.
Probability of what? Of a particular term coming from an unattested language family? That's using the unknown to explain the unknown, and the probability that should be assigned to that is zero, right?
You just can't. Thus soft science.
teo123 wrote: Mon Jan 21, 2019 4:28 am
Safely? Again, with what probability? There are exceptions to that kind of thing everywhere.
Well, that question is very hard to answer.
Exactly, soft science.
teo123 wrote: Mon Jan 21, 2019 4:28 amSound laws are considered way more certain than any particular etymology. If some etymology doesn't agree with the sound laws, it's more likely a wrong etymology than an exception to a sound law. The exceptions are numerals and some very basic vocabulary, certainly not the toponyms.
How certain? What probability?
teo123 wrote: Mon Jan 21, 2019 4:28 amYou are probably using the regularity of the English spelling as a proxy to how regular sound changes are.
I'm not using anything for anything. I'm saying if you don't know the actual probability of these things then you're practicing a soft science.
If that's what you want to study that's cool, just stop pretending it's a hard science.
teo123 wrote: Mon Jan 21, 2019 4:28 am
"Oh hai guyz, I'm just a hooman, I was abducted by aliens. Sorry about all the confusion."
And why would it be reasonable to assume that alien is trustworthy?
Why would you assume some scroll is a trustworthy example of a language? What if it was an ancient linguist playing a joke on the future?
My point exactly. You're asking for outside and (once) conscious forces to sort it out for you.
teo123 wrote: Mon Jan 21, 2019 4:28 amReally? I've always assumed it's some hypothesis about the composition of the sub-atomic particles that sounds smart to laymen, but sounds like a bunch of baseless and unfalsifiable assertions to the physicists.
Hypotheses are testable. Models are different, any model that is truly descriptive is basically true as far as most people are concerned (and may in fact be, metaphysically).
Say we have a black box, and we know when you put a 1 in it, it gives you a 2. That's all it does, because all we can put into it are ones. The opening is too small for anything else, it would break the box.
So one model might be "it has an adding machine in it, and it adds 1 to whatever you feed into it"
Another model might be "it's a multiplier, and doubles whatever number it gets"

Since we can only ever give it one, those models are equivalent and unfalsifiable ways of understanding the mechanism. They're both essentially true, for all we know, and the difference between them is a very annoying philosophical issue.

String theory is basically that. At least it is right now. Wait until it makes a testable prediction, and maybe it can be something more. But that's why people are annoyed at the time being put into it.

Re: Soft Sciences Vs. Hard Sciences

Posted: Thu Jan 24, 2019 3:45 am
by brimstoneSalad
teo123 wrote: Tue Jan 22, 2019 3:18 am Also, if you insist social sciences are not real sciences, how can you possibly be against anarchism? If social sciences are not real sciences, anything lawmakers could possibly do is forcing their subjective opinions onto people, and that's authoritarian.
Anarchism is a position that makes claims, you're doing the same thing as any politician.

A lot of lawmakers DO that, which is why I'm not a big fan of politics.
However, there are testable components to government and economics (these are much harder sciences than linguistics, actually).
For example, when people run pilot programs and collect actual data on the effects of a policy before mindlessly promoting it. This is rare in politics, but it's happening more today than it used to. We're moving in the right direction.

The bottom line is that you don't make changes to how things run until you TEST those changes and you find that they're effective.

So we need to TEST the idea of changing the minimum wage. We need to TEST basic income. TEST changes to welfare, etc.
And yeah, if you put a LOT of things the government does to the test, what you find out is that quite a few of them are just wasting money and sometimes making things worse. But sometimes they aren't, and sometimes there's a better policy. You can't assume government is bad and anarchy is great until you test that.

Anarchy is just another untested dogma.
teo123 wrote: Tue Jan 22, 2019 3:18 amI am not planning to reply any more here, I hope this was enough to make you think.
I got somebody complaining about how a soft science isn't a hard science, and trying very hard to yet again prove the Earth is flat.
You really didn't learn anything from the Flat-Earth thread, did you?

It's like Red said.
teo123 wrote: Wed Jan 23, 2019 9:05 am Brimstone, I see you are otherwise a smart guy, but your comments here seem arrogant, uninformed and even somewhat insulting.
And that's what it comes down to. I hurt your feelings by denying that you're qualified to evaluate hard sciences because you published a few peer reviewed papers in linguistics. You're also not qualified to evaluate quilting technique, does that insult you too?
teo123 wrote: Wed Jan 23, 2019 9:05 amThey make about as much sense as when the Flat-Earthers deny that light (in this case languages) obeys certain laws.
I'm not denying anything established in linguistics, I'm saying it's a soft science, and it is. These things are weakly established.
Basic laws sounds follow and how words work and move around and change I have no doubt are generally correct, but it's the unquantified "generally" part that's the problem, and moreover the large pool you're sampling from and then doing some ad hoc p hacking.
Light just not just "generally" behave as such. It does. If it ever does not that's not just a quirk or exception to the rule, that would mean that basically ALL physics is wrong. Soft sciences tolerate exceptions like that.
teo123 wrote: Wed Jan 23, 2019 9:05 amwaste a lot of my time I would use to prepare myself for the electrotechnics test I failed.
Like in the flat-Earth thread, you're wasting your own time by not putting in the effort (and not opening your mind) to understand what I'm explaining. I'm being exceedingly patient with you yet again, and explaining in detail what a soft science is and why, but it's all going over your head.
You can't rationalize your way out of this. Linguistics being soft science is not an insult, it's just a fact. Linguistics just doesn't lend itself to being easy to objectively confirm and evaluate, if it did then your peers would not have been able to so easily reject your findings. You're just going to become increasingly frustrated if you continue to expect linguistics to behave like a hard science.

Re: Soft Sciences Vs. Hard Sciences

Posted: Thu Jan 24, 2019 4:17 am
by teo123
Jebus wrote: Wed Jan 23, 2019 1:02 pm
teo123 wrote: Wed Jan 23, 2019 9:05 am I’d be interested in reading what other people on this forum educated in social sciences, such as @EquALLity or @Jebus, have to say about this issue.
I haven't followed the thread but if you can break it down to a couple of specific questions, I will let you know what I think.
You can read the details of my theory on the long web-page I've made about it (the three peer-reviewed papers I published about it aren't available on-line and they are in Croatian). I'll summarize it here.

So, the proportion of modern Croatian toponyms that make sense to those who speak Croatian is insignificant. Yet, according to the mainstream history, the only language that was widely spoken in Croatia for the last more than 1000 years is Croatian. I consider that to be a contradiction. Mainstream linguistics claims that many of those toponyms indeed come from Croatian, but from some words that don't exist in modern Croatian. So, they try to explain the Croatian toponyms using the Slavic (since Croatian is a Slavic language) roots. However, I believe that to be fallacious, for about the same reason the Nostratic hypothesis is fallacious. You first need to establish using the statistical methods that the languages are related or came into contact before you claim particular words are related. So, by that logic, you should have to establish that a significant proportion of the Croatian toponyms is explicable using Croatian, before you claim a particular toponym comes from an obsolete (or even unattested but reconstructed from a Proto-Slavic root) Croatian word.
Many modern Croatian toponyms have a well-known Latin etymology. However, I argue that most of those etymologies can't be true because of the known sound laws that would have affected a Latin toponym borrowed into Old Croatian.
There were many toponyms attested in ancient sources that aren't explicable using Latin or Greek. They are usually attributed by mainstream linguistics to the "Illyrian languages", implying there were many and possibly unrelated "Illyrian languages". However, I think that Illyrian was more-or-less a single language (closely related languages at worst), because those toponyms appear to have elements that repeat themselves with certain meanings.
For example, the ancient name for Vis was "Issa", the ancient name for Daruvar was "Balissa", and the ancient name for Varazdinske Toplice was "Iasa". So, the element that repeats in them (and, as far as I am aware of, in no other ancient Croatian toponym) is *issa~iasa. What else is common to those places? Well, as everyone who has studied a bit of Croatian history knows, in all those three places there are remainings of large ancient thermae built upon a spring. There are perhaps only five such places in Croatia: Dakovacka Breznica, Varazdinske Toplice, Hadrian's thermae on Brijuni, Vis and Daruvar (places such as Solin don't count because the thermae there were built in Roman times and they got their water from a river). So, I think I can claim with reasonable certainty that *issa~iasa meant "health-giving spring" in Illyrian. Is there a plausible etymology for that term? Indeed there is, the well-known Indo-European root *yes (to foam, to boil) had a derivation *yos (spring).
This can also be applied to modern Croatian toponyms. For instance, a disproportionate number of Croatian river names start in k(+vowel)+r: Krka, Korana, Karasica (2 rivers with the same name), Krbavica and Krapina. So, I can claim with reasonable certainty that *kar~kur meant "river" (whatever be its etymology) in Illyrian. And a disproportionate number of the Croatian mountain names appear to start with k(+vowel)+l: Klek (2 times), Kalnik and Claudius (the ancient name for Papuk). So, *k(u)l meant "mountain" in Illyrian, and it almost certainly comes from the Indo-European root *kelH, meaning the same. Now, I've noticed that the word "Kalnik" must have been formed after the Third Slavic Palatalization ceased to operate in Croatian. Therefore, the Illyrian language must not have died out in ancient times, as the mainstream history claims, but it must have been alive (despie never being attested) until at least 12th century (when the Third Slavic Palatalization ceased to operate in Croatian, but the Havlik's Law still operated, for the short 'u', borrowed as a back yer, to change into 'a').
This was, I think, more than enough to get a general idea.

Now, Brimstone claims that's not a real science. His basic arguments are:
1. My claims are practically unfalsifiable without a time machine.
My response was that they simply aren't. You can postulate all sorts of meanings to "Issa" (island, town, place near water, stream...), but then what you should expect to see when you look at the map doesn't agree with what you actually see on the map.
Brimstone's response to that was that you shouldn't really expect anything because "words easily move".
My response was that it's not correct because toponyms don't change the places they refer to randomly, but according to certain laws. My example was the Skok's Second Law of Toponymy (that modern names of islands derive from ancient names for the largest inhabited place on that island), which is, according to the research published by Dubravka Ivsic (I linked to it in my previous post) more than 90% accurate for the Croatian islands.
2. That, because the sound laws are supposedly not 100% accurate here, I need to calculate P-values before making any statements relying on them.
I've made a few arguments against that in my last two posts:
a) P-values are here almost impossible to calculate because the identification of the ancient Croatian toponyms partly relies on sound laws being exceptionless (that's why the ancient hydronym "Murios", for example, isn't connected to the river Mura).
b) The main reason for the apparent exceptions to the sound-laws is dialectal diffusion, and we can assume it doesn't exist in the toponyms.
c) In the second peer-reviewed paper I published, I did attempt to calculate the P-values of some patterns I saw in toponyms, and they were extremely low.
d) P-values don't drastically increase the chance of being right here.

(I will reply to Brimstone's last two posts later if I find something useful in them and find the time to, we were apparently writing this at the same time.)

Re: Soft Sciences Vs. Hard Sciences

Posted: Thu Jan 24, 2019 4:23 am
by teo123
Something is apparently wrong with the first link in my last post (though you don't really need to look at it to understand what's going on), the correct address is:
http://flatassembler.000webhostapp.com/toponyms.html