23-09-2023, 10:07 PM
(23-09-2023, 06:00 AM)Mark Knowles wrote: You are not allowed to view links. Register or Login to view.I question whether "they form a representative subset for the word types". By selecting only those occurring 7 and 8 times you introduce some bias. I believe that words occurring 7 and 8 times in the Voynich text behave differently from words occurring at some other frequencies. I would be curious as to what your results say for words that occur 1 time in the Voynich text.
In which way do you believe they behave differently? Do you have any evidence for your hypothesis?
Words occurring only one time (hapax legomena) are a bad choice for selecting a limited sample. There are over 5500 hapax legomena in the voynich text, whereas the whole idea of a limited sample is to check only a limited number of words. However, to check the more than 5500 words only means that more work is necessary to reach the same result. If you are curious you can check it yourself. Just count the word tokens ending with EVA-m or starting with EVA-d for different positions within a line. Words ending in EVA-m occur more often than expected in line final position and words starting with EVA-d are more common than expected in line initial position (see Timm 2014, p. 85).
However it would be correct to argue that words in Currier B behave a little bit different than words in Currier A. For instance words ending in EVA-m occur more frequently in line final position in Currier B than in Currier A (see Timm 2014, p. 85).
Position of words ending with EVA-m within a line:
initial between final
Currier A 13 169 188
Currier B 4 106 401