***the statistics must vary with every single ms.***
No, I disagree strongly there, I can show that the numbers from many different ms show pinpointed statistics.
That is exactly what I am saying, for..I think 3 years now. (I will not discuss entropy again, that is an entirely different discussion.)
The main differences between texts using my stats are:
- the time period (1300 is different from 1400) but still you can see the used language clearly
- the exact language
- there is often 1 most frequent word in the top ten, that is the subject in the book. For example a name.
This is nothing special, there are (commercial) tools, that can tell for many languages what the % chance is, that the text is in that language.
For example it will say: 5% Modern English, 55% Middle English, 40% Vulgar Latin.
My suggestion for the VMS, is that the polyphonic characters in combination with some nulling and/or a start/and/or ending formula is used. The final ciphered polyphonic set will consist of 10 sets.
If that is true, that is enough to confuse any statistical tool.
The advantage with this MS text is that it is *very* long, which favours the use of statistics.
You don't even know if you have a coherent set of data. That is one of these unproven assumptions about the ms. You only have to take the trouble of looking at any modern catalogue of mss. to realise that the ms. is not a normal book
@ David
I am talking about mss. whoses abbr. are not dissiolved, that is the case in the VMs
This is getting a bit off the original topic, but the first and perhaps the biggest problem we have is that we don't know how to read the writing. Different people have made different assumptions, and they are all essentially certain to have some mistaken assumptions in them.
All statistics that one can generate based on the transcribed versions of the text depend on these assumptions.
In fact, one good way of proceeding is to 'play with' these assumptions, i.e. to tune them or experiment with them, and to see what happens.
A typical example is the word length distribution. Different interpretations of what are word spaces leads to different results. Different definitions of what are individual characters also lead to different results.
On the other side, the expected word length distribution of an abbreviated text will be different from one without abbreviations.
On the homogeneity of the text, there is little doubt that the text is not homogeneous.
Different statistics, based on all different available transcriptions confirm each other.
It is possible to identify blocks of homogeneous text.
On the level of difference between these blocks, people have different opinions.
The work of Prescott Currier on this topic is known by most people, and often quoted, but has to be 'enjoyed with care' (to use a Germanism). It is a mix of objective and subjective statements.
Yes sorry that it's becoming off topic. To reply on Rene's text "...biggest problem we have is that we don't know how to read the writing"
I had prepared a rather long reply, but here the short version:
Rene, you talk about assumptions;
I have worked on result sets that are hard and true for (on average) 90%,
When people write on this forum that accepting an error margin of 10% (or less) is nonsense to them,
all discussion stops there, and there is no point in sharing anything anymore on those results, because the results are nonsense too in the eyes of the beholder.
I have sets that are for 100% true, but you will have to accept the fact that there the scribe made some errors.
If that is impossible to do, everything based on that observation will be invalid.
I can also show that those errors are errors based specific methods, but of course, without a final decrypted text I can not never proof it,
However, based on that single argument we can get nowhere.
I talk about observation and not assumption.
The main difference is that an observation should be made objectively; of course when I see a tree, there are always people that say: no that is a bush. I will quickly give up then, because I can spend my time better than convincing those people.
An assumption would imply that you start somewhere, without showing exactly why you start there or have no underlying metrics.
I agree that an assumption is most of the time a scientific mistake, sometimes it leads to new paths however.
google says:
Assumption - the act of taking for granted, or supposing a thing without proof;Observation - the act or the faculty of observing or taking notice; the act of seeing, or of fixing the mind upon, anything. ....18 okt. 2011
Keep up the good work!