I've got a new idea of how to find "and" (if not in the form of a clitic). Given that "and" would be expected to be amongst most frequent words, one could check the list of top-N most frequent vords and see which of them do never occur:
a) as a label
b) at the end of paragraphs.
Vords satisfying both conditions at the same time would be "and"-suspects.
This implies that there is no shuffling, of course.
a) as a label
b) at the end of paragraphs.
Vords satisfying both conditions at the same time would be "and"-suspects.
This implies that there is no shuffling, of course.