Loading Grammar

   Loaded rule by rule.  Each rule's data structure has Source and
target production rules (left and right hand sides), source and target
(x and y side) alignments, and feature unification equation blocks
separated into parse, transfer, generation, and constraint.


Parsing

	Before parsing begins, the source text will or won't be all
upper-cased, depending on the value of the normalizecase variable.
This is to allow non-Western European languages to be processed
uncorrupted.  The source text is then tokenized into words.

    



Transfer



Constraint Checking