Sequential word repetitions in the VMS

[MarcoP]

This are figures comparing exactly repeating words in Currier A & B “languages”. All data are based on Takahashi's transcription.

Currier A
words: 9025
repetitions: 96
different repeating words: 45

Currier B
words: 20063
repetitions: 181
different repeating words: 74

Since the B corpus I considered is about twice as large than the A corpus, the overall numbers are quite consistent. Of course, when one looks at individual words, things differ.

These diagrams are sorted by the number of repetitions of each word. Green bars are the “expected” number of repetitions based on word frequencies. The red bars are the count of actual exact repetitions for each word (daiin.daiin, chedy.chedy, etc).
   
I this data make clear that word frequency doesn't explain the number of repetitions.
For instance, in language A, the number of occurrences of chol is half the number of occurrences of daiin, yet chol has many more repetitions.
The preference for the repetition of q- words is clearer in language B, which has common words with that prefix. But language A also has several q- words that repeat (even if only once).
Language B seems to have a larger repertoire of often-repeating words. Both languages have long “tails” of words whose repetition only occurs once.


These are the same data ordered by decreasing word frequency.
   


I want to point out the comparison between daiin and aiin in language B. The two words have similar frequencies, but aiin (which is slightly more frequent) never repeats while there are three occurrences of daiin.daiin.
In language A the frequencies of daiin and aiin are quite different, but again daiin repeats and aiin doesn't.

These are counts on the whole manuscript based on two different transcriptions.

              Takahashi  Zandbergen

daiin            750       697
daiin.daiin       16        11

aiin             419       296
aiin.aiin          0         0