Bigrams across uncertain spaces

[RobGea]

Bigrams across uncertain spaces.
All errors are mine and there probably are some.

Extract text from ZL_ivtff_2a using IVTT: from Windows cmdline "ivtt -x7 -s0 -h0 ZL_ivtff_2a.txt ZL2adot-comma.txt"
text file contains : pure text, keep commas and dots, also keeps '?', '*' and @123; notation

Find all the commas denoting uncertain spaces, get EVA-character from either side of the comma, concatenate to form bigram  i.e ('X,Y' = 'XY')
Find all the dots denoting certain spaces, get EVA-character from either side of the comma, concatenate to form bigram  i.e ('X.Y' = 'XY')

Count them, Rank them, Simple stat them:
abs(Percentage divide) =
percentage occurrence of Comma-bigram, percentage occurrence of Dot-bigram, take which ever number is higher and divide by the lower

abs(Rank Comma - Rank Dot) =
Absolute value of the Comma bigram rank subtracted from the Dot bigram rank

Did the 'percent divide' and 'rank subtract' because it just makes it easier to spot the differences and provides a simple way to compare them.
Top12 shown here:

Code:

Commas(X,Y) 2737 Total          Dots(X.Y) 30890 Total    abs(Percentage divide)  abs(Rank Comma - Rank Dot) Rank  count bigram  %          Rank  count bigram  % ('R1', 285, 'ra', 10.413)      ('R14', 730, 'ra', 2.363)    4.407      13 ('R2', 147, 'lc', 5.371)      ('R10', 1146, 'lc', 3.71)    1.448    8 ('R2', 147, 'lk', 5.371)      ('R33', 201, 'lk', 0.651)    8.25      31      --lk ('R4', 125, 'ls', 4.567)      ('R15', 672, 'ls', 2.175)    2.1    11 ('R5', 119, 'sa', 4.348)      ('R28', 245, 'sa', 0.793)    5.483    23 ('R6', 101, 'yk', 3.69)      ('R18', 443, 'yk', 1.434)    2.573      12 ('R7', 93, 'ol', 3.398)      ('R39', 118, 'ol', 0.382)    8.895      32      --ol ('R8', 90, 'ld', 3.288)      ('R17', 569, 'ld', 1.842)    1.785    9 ('R9', 83, 'ro', 3.033)      ('R6', 1355, 'ro', 4.387)    1.446    3 ('R10', 78, 'lo', 2.85)      ('R11', 996, 'lo', 3.224)    1.131    1 ('R11', 73, 'yd', 2.667)      ('R7', 1275, 'yd', 4.128)    1.548      4 ('R12', 68, 'yt', 2.484)      ('R23', 312, 'yt', 1.01)    2.459    11 ('R12', 68, 'ok', 2.484)      ('R52', 65, 'ok', 0.21)    11.829    40        --ok

Observations:
the 'o<character>' family turn up a lot with high abs(Rank subtract) scores ; ol, ok, or,oa,ot

The bigram 'lg' has the highest  abs(Rank subtract) score:
       comma                            dot
('R58', 5, 'lg', 0.183)      ('R206', 1, 'lg', 0.003)    61.0      148        --lg
One conclusion is that at least some of those 'l,g' bigrams are real bigrams and any apparent space is a scribal artifact.

Questions:
What does it mean when the Dot-bigram occurrence percentage is higher than the Comma-bigram occurrence percentage? e.g
('R38', 15, 'yo', 0.548)      ('R2', 2687, 'yo', 8.699)    15.874      36

Data attached:
  bigrams_un-certain_spaces.txt (Size: 10.34 KB / Downloads: 32)