Mathematical analysis of Scottish Stone Art points to lost language?
At New Scientist web site:
"Elaborate symbols and ornate depictions of animals carved in stone by an ancient Scottish people have given up their secret – to mathematics. Statistical analysis reveals that the shapes are a forgotten written language. The method could help interpret many other enigmatic scripts – and even analyse animal communication....."
I thought to begin with this was just to do with april 1st but maybe not. It says in the paper there are like 43 categories of symbol. Not much of a language with 43 words in it. And they admit some stones only have one picture on them. And if on a stone with lots on, there's a picture of a deer, could that not symbolise A Deer or maybe a group of people represented by a deer, is that really much of a 'language' either? Any more than any paintings can be said to be a language, with all the straightforward representations / symbolism / connotations of the images that they contain?
And another thing (And I know I don't understand the statistics of what they're talking about, I admit it) but isn't the data set for the pictish stones stupidly small, and yet they're comparing whatever it is they've statistically analysed with various languages with thousands of examples.
Ok so some of the symbols are Mysterious and clearly have some meaning we can't easily guess at. But doesn't that just make them symbolic? Like if you draw a crown it could be symbolic of the queen, although anyone who's never seen a crown wouldn't know that? Also yes you can construct some sort of narrative that your fellow countrymen can understand if you carve a load together on one stone. But isn't that obvious, do you need statistics to prove that? Otherwise it would just be a bizarre jumble of random symbols on a stone, and they clearly weren't doing it for carving practice. But that doesn't really make it a language like English is a language, surely. Isn't there a slope between representational imagery, symbolic imagery, things like hieroglyphics, and eventually alphabets? Can it not be a progression, why does it have to be shoehorned into the category of 'language'?
What am I ranting about. I'm not sure. I think it's against the need to squeeze statistics out of beautiful objects. I've got a bit of a science/art dilemma in my head as a matter of course you see.
I haven't read the paper in full yet - so hope it's not a 1st of April joke (paper was published online on 31st of March so if it is they are/were cheating!)
I come from an Electronics and Engineering background and at first sight the idea of using Shannon's law to identify the amount of information (ie non-randomness) in any data set makes sound sense. Indeed early cryptographic approaches made use of this approach for many years.
I'm going to dig out my old (very old) text books and try to understand the text and maths a bit more.....
OK - not gone back to the textbooks but read the paper and it all looks sensible.
In simple terms what they have done is:
1) Counted the number of unique symbols present (43 - but varies a bit by source/interpretation).
2) Identified the number of unique two symbol combinations and non-unique (ie happen more than once) two symbol combinations.
3) Taken a load of known languages and styles of writing, and random data and done the same with them to create value sets for each and then shown that their approach can spot the difference between different language types, different styles of wring and random data.
4) Shown that low data set values make it impossible to tell the difference between different styles of writing but can spot languages from random data and spot and types of language.
5) Run the data set for Pictish symbols through the same process and deduced that it looks like a language, but they can't be sure if the symbols depict words or syllables (50/50 split based on the different data sets/interpretations).
6) Based on the fact that many stones have only one symbol made the rational, but unproven assumption that a single symbol probably equates to a word and not a syllable.
Don't know enough stats to say if their statistical analysis is valid given the size of the data set but the claim 99.9% confidence levels which is normally more than OK for experimental results.....
"Isn't there a slope between representational imagery, symbolic imagery, things like hieroglyphics, and eventually alphabets? Can it not be a progression, why does it have to be shoehorned into the category of 'language'?"
Good point Rhiannon, though I'm not sure I'd call (written) language a slope or progression from imagery to alphabets. Pictographs can and do stand alone in conveying ideas (as in the case of Chinese characters). Each Chinese character also has a sound (or sounds). The difference between a Chinese character and a phonetically written word is that the former also carries an image. I can't do it here but if you imagine a square, that's the Chinese character for mouth (or entrance). The character also has an accompanying sound. The difference between the Chinese character for mouth and the phonetically written word for mouth however is that with the Chinese character you get a visual message first and then you get a sound. With the phonetically written word all you get is the sound (though some words like our written word 'look' do have a pictographic element to them). The closest we have in English to the pictographic way of conveying ideas is with (road) signs. But imagine a whole language written with the equivalent of road signs!
There's something else. From the reality of a mouth (assuming such a thing :-) we might have the following sequence. The actual mouth. Our concept of the mouth. The sound of the word mouth. The character or pictogram of the mouth. The phonetic spelling of the word mouth. That's four steps removed from an actual mouth. In other words, I suppose we could say that languages that skip the phonetic aspect of conveying the idea of a mouth are one step closer to reality.