What is the evidence for the arbitrariness of the sign?
Continuing this question, what English words can not be motivated and should be considered arbitrary?
I think only the natural meanings would remain and that all words that are compounds of those natural sounds. But what does the linguistic community think? Which words are 100% arbitrary and please say why?
Which words are 100% arbitrary and please say why?
In some ways, none of them. In other ways, all of them.
In the question you linked, I used = /kæt/ as an example of an arbitrary word. But in many ways, it's not arbitrary at all. I call a /kæt/ because my parents called it a /kæt/, and my teachers called it a /kæt/, and their parents called it a /kæt/, and so on. Go far enough back and they called it a /katte/, and even farther back a */kattuz/, and so on, and we can track the specific sound changes that eventually led to this becoming the modern English word that we all know.
That is, I didn't just one day decide on a whim that this small furry animal would be a /kæt/. I call it a /kæt/ based on centuries upon centuries of oral transmission from generation to generation.
Now, why did my distant Proto-Germanic ancestors (I'm sure I had some somewhere in the family tree…) decide to call a */kattuz/? Unknown. Some linguists think it was borrowed from an Afro-Asiatic language, or a Uralic language, or some unknown substrate that's long since died out, or that it goes back to an obscure PIE root. There must have been some reason, but there's just not enough surviving evidence to say. This is the same reason we can't reliably talk about anything older than Proto-Indo-European: we can hypothesize and speculate, but there's not enough evidence to make falsifiable theories. And that's the standard that all modern science is weighed against.
The standard assumption is that there's nothing inherently -like about /kæt/, which is why has different names all over the world (in Kenya it's a /paka/, in Japan it's a /neko/, and so on). That's what people mean when they talk about "the arbitrariness of the sign". But there's no origin of language we can point to to say "this, this is where the arbitrariness comes in". We can just say "if language weren't arbitrary, we'd expect to see all languages across the world having similar words for similar things, and we don't see that".
It sounds like what you're interested in is—is there a built-in mechanism in our brain that somehow associates with /kæt/? In other words, is there an intrinsic meaning to the phonemes /k/, /æ/, /t/ that connects them to ? And as you can see from the answers to my other question, the answer is pretty solidly no. This can be tested and falsified: take a large sample of people who don't know anything about language X (or its relatives), then show them the language-X words for and , and see if they can identify which is which. It's been found that people with no outside knowledge of language X or its relatives usually do no better than random guessing.
(For an example you can try yourself: /toːtoːtɬ/ is an attested word for a certain type of animal that I'm sure you're familiar with. Can you tell me what animal that is? How about /tokatɬ/? Or /mit͡ʃin/?)
However…
While people can't reliably tell and apart, there are some interesting results showing that humans, regardless of language, do associate certain sounds with certain qualities. Several different experiments have shown that people tend to associate words like kiki, takete, and Kate with sharp, pointy shapes, and words like bouba, maluma, and Molly with soft, rounded shapes.
This implies that there is some sort of intrinsic association between sound and quality embedded in the human brain. It's not anywhere near as strong as you've suggested (which is why people can't reliably distinguish and in an unknown language), but it does exist.
Could such an association have been involved in the origins of human language itself, millennia and millennia ago?
Sure! Absolutely! That's entirely possible. But—and this is an important "but"—that's so far in the past that no current linguistic models can make any sort of falsifiable claims about it. And as long as the claims aren't falsifiable, they're not considered science.
If you want to seriously pursue this idea, and take it in a more scientific direction, I'd recommend changing your approach. If you think that a certain phoneme is associated with a certain meaning, turn that into a falsifiable hypothesis: "humans, regardless of language, will instinctively associate phoneme X with meaning Y". Then set up an experiment that could falsify that hypothesis: gather a bunch of test subjects who speak English and a bunch who speak Mandarin, have them categorize pictures based on nonsense-word labels, and see if the accuracy is the same regardless of native language. This might support your theory, or might disprove it, but either way, you'll have a falsifiable theory, and hard evidence one way or another. And that's what modern science demands—nothing more, nothing less.
