Are there any features present in body text that aren't present in labels?

[galen]

The lack of punctuation suggests some aspect of the text may function as a 'separator' between one thought and the next. 

Since labels don't need punctuation,
if there are any properties of body text that are absent from label text,
they'd make compelling candidates for playing the role of punctuation.

What has been noticed about the differences between body text and label text? 

[dashstofsk]

Labels are not representative of the paragraph text. Words starting ot and ok are more frequent in labels than in the paragraph text, and words starting qo or ch less frequent. Here are some lists of the top paragraph words and the top label words.

Also labels have more variability. 20% of the paragraph text is made up of just 18 words. For labels the equivalent number is 38 words.


   
   

[Bluetoes101]

I believe it is more "q" than "qo" missing, in that there are many labels that start "o" that can then be found in text "qo". Obviously outside the many that are unique. 
okeody > qokeody etc, though correct me if wrong. 

There's probably also something to be said about labels mostly existing in "B" language rather than "A", so some differences may be down to that rather than label/non-label.

[ReneZ]

Here are some lists of the top paragraph words and the top label words.

These labels also include the single characters in 'sequences', and are not near some drawing element. One may wonder if these should be included as labels.

[galen]

Yes, I should have been more specific. I was thinking of a label as a word or two accompanying some illustration. I know that's harder to run the numbers on -- I thought maybe this had already been investigated and was curious what the findings were. 



The number of duplicate words in the label data is interesting. Are there any instances of the same label being applied to two separate illustrations? 

Bluetoes101 -- you mentioned that many of the 'o' labels can be found in the text but with 'qo' instead. Are they usually on the same page or nearby pages, or are they just somewhere else in the whole manuscript?

[ReneZ]

I know that's harder to run the numbers on -- I thought maybe this had already been investigated and was curious what the findings were. 

It's not really more difficult. If you use any of the transliteration files in the IVTFF format, such as listed in the table here: You are not allowed to view links. Register or Login to view.

then you can identify labels that are not near a drawing element with the code "L0", just like this case:
<f57v.1,@L0>      dairol

The symbol preceding "L0" can be ignored.