(March 19, 2023 at 6:47 pm)BrianSoddingBoru4 Wrote:

(March 19, 2023 at 6:11 pm)FlatAssembler Wrote: What do you think about this comic:

I think it’s poorly drawn, unfunny, and is attempting to reinforce a false distinction.

Boru

I thought the message of that comic is undeniable. Bullshitting gets caught earlier in natural sciences than in relatively-hard social sciences such as linguistics, where it is still caught much earlier than in soft sciences such as sociology, which are still better in this regard than humanities such as literature criticism. You don't think so?

And what do you think about p-values?
The p-values in particle physics are regularly on the orders of magnitude of 1/1'000'000. If you read a paper in particle physics, you can be almost certain it is not describing a coincidence.
In softer parts of linguistics that is etymology, it is usually not obvious how to even calculate the p-values. And when you can, you get very weak p-values. In my latest paper about Croatian names of places, I was only able to provide math that shows the p-value is somewhere between 1/300 and 1/17. And my informatics professor Franjo Jović thinks it is likely closer to 1/17.
As for harder parts of linguistics such as historical phonology, well, it's hard to tell. I don't know of anybody who tried to calculate the p-value of the Havlik's Law or something similar. If I had to guess, I'd say the p-value of the Havlik's Law is probably around 1/10'000. We have to keep in mind the Havlik's Law is not exceptionless. Notably, in the Chakavian dialect of Croatian, the schwa in the first syllable sometimes changes to 'a' even when the Havlik's Law predicts it should disappear.
I'd be interested in what @polymath257 thinks about this.

Maybe you should post your question to a forum where people actually give a shit about what's on your mind? Just a suggestion.

(March 20, 2023 at 10:02 am)Angrboda Wrote: Maybe you should post your question to a forum where people actually give a shit about what's on your mind?  Just a suggestion.

Which question do you have in mind? And which forum would you suggest me to post it to?

The moderator eutow from forum.hr has asked me to not post stuff about my linguistic theories again on that forum. Here is what he said to me in a private message:

Quote:Bok

Vidim da nema više nikakve smislene rasprave o tvojim teorijama. Zato od sada nemoj više objavljivati takve stvari na Jezikoslovlju, jer se ti i drugi korisnici samo krenete vrijeđati kao mala djeca.

LP

(March 20, 2023 at 10:58 am)FlatAssembler Wrote: The moderator eutow from forum.hr has asked me to not post stuff about my linguistic theories again on that forum. Here is what he said to me in a private message:

Quote:Bok

Vidim da nema više nikakve smislene rasprave o tvojim teorijama. Zato od sada nemoj više objavljivati takve stvari na Jezikoslovlju, jer se ti i drugi korisnici samo krenete vrijeđati kao mala djeca.

LP

What he said, but in English.  Stop it.

I find it unconscionable that a serious subject can use a p value of more than .005 as a guide. To do so leads to too many unnecessary errors.

Errors in human thinking? Say it isn't so.

(March 20, 2023 at 9:55 pm)polymath257 Wrote: I find it unconscionable that a serious subject can use a p value of more than .005 as a guide. To do so leads to too many unnecessary errors.

But my paper with math showing that the p-value is somewhere between 1/300 and 1/17 is still better than the vast majority of papers about Croatian names of places, which do not even attempt to calculate the p-value, right?

(March 20, 2023 at 10:19 pm)Tomato Wrote: Errors in human thinking? Say it isn't so.

Not errors in thinking. Errors as in writing a paper about a coincidence and presuming it's an actual phenomenon. Empirical errors.

(March 21, 2023 at 6:02 am)FlatAssembler Wrote:

(March 20, 2023 at 9:55 pm)polymath257 Wrote: I find it unconscionable that a serious subject can use a p value of more than .005 as a guide. To do so leads to too many unnecessary errors.

But my paper with math showing that the p-value is somewhere between 1/300 and 1/17 is still better than the vast majority of papers about Croatian names of places, which do not even attempt to calculate the p-value, right?

I don’t think you truly grasp what p-values mean. You seem to be mistaking statistical significance for scientific significance. It looks as though you’re trying to apply p-values to your hypothesis about Croatian place names. That’s not what p-values are intended to test.

Boru