This condition definition computes the part of the sentence that needs to be 
re-parsed. Its two arguments are its "output" or "return values" - a standard 
way to write things in Prolog.

On line 4, it is the value of "NewReversed" which is being computed, not that 
of "PreviousWords" as might at first be expected. "PreviousWords" and 
"SentenceBegin" are both known, and the line
   PreviousWords results from appending NewReversed to SentenceBegin
is a typical Prolog/COGENT notation used to compute the difference between the 
two lists.

An example would be:
Previous Words:   [cat, the] (reversed list)
Sentence Begin:   [the]      (reversed list, obtained from Choice Points)

[cat, the] = NewReversed @ [the]
                [cat]
              (reversed)
NewContent = [cat] (normal ordering; here a special case of length one)

In addition, we need to take into account the case in which the parser 
backtracks several times in a row, especially that of "nested" backtracking, 
that is, the case in which the parser parses several words, notices it is on 
the wrong track, backtracks to an earlier point, and then *while re-analyzing* 
has to backtrack again.

In this case, when the parser backtracks for the second time, the buffer 
"ReProcess" will contain words that must be re-analyzed. If we did not match 
them and append them after NewContent, they would be irremediably lost in the 
process, because they have already been "read" and are therefore not available 
from the input (the display) anymore.