Following Emma's observations, I have computed a couple of slightly more accurate bigram diagrams. The main differences are:
Benches were changed into C and S and benched-gallows into K T P F;
* I computed bigram prefixes and suffixes considering the three disjoint sets of image-breaks, line-breaks and "ordinary" space-breaks;
* Here line-breaks do not include what appear to be paragraph-breaks (which appear to be another different set because of Grove words).
I still only considered data from pages that contain at least a line interrupted by illustrations.
The following diagram illustrates suffixes before the various breaks. Those before and image-break are comparable with suffixes before a line-break: the red and green bars are similar, while the blue bar often stands out. As already discussed, the only major difference is -am, which has a spike at line-end but not before an image-break.
Prefixes after the various breaks are much harder to compare. For instance, che- and she- (Ce- and Se- in the graph) behave differently in the three cases; yt- yk- are another, total different, example.
PS: in term of Emma's You are not allowed to view links. Register or Login to view., one can observe that words after an image break typically favour a weak initial (e.g. cho- ok- yk- yt-) symmetrically "disliking" a strong initial (qo- and the almost total absence of initial gallows). However, there are exceptions to this hypothetical pattern: in particular da-.
Benches were changed into C and S and benched-gallows into K T P F;
* I computed bigram prefixes and suffixes considering the three disjoint sets of image-breaks, line-breaks and "ordinary" space-breaks;
* Here line-breaks do not include what appear to be paragraph-breaks (which appear to be another different set because of Grove words).
I still only considered data from pages that contain at least a line interrupted by illustrations.
The following diagram illustrates suffixes before the various breaks. Those before and image-break are comparable with suffixes before a line-break: the red and green bars are similar, while the blue bar often stands out. As already discussed, the only major difference is -am, which has a spike at line-end but not before an image-break.
Prefixes after the various breaks are much harder to compare. For instance, che- and she- (Ce- and Se- in the graph) behave differently in the three cases; yt- yk- are another, total different, example.
PS: in term of Emma's You are not allowed to view links. Register or Login to view., one can observe that words after an image break typically favour a weak initial (e.g. cho- ok- yk- yt-) symmetrically "disliking" a strong initial (qo- and the almost total absence of initial gallows). However, there are exceptions to this hypothetical pattern: in particular da-.