We have seen that if one concentrates on the similarities between LFG and ALEP formalisms,
a straightforward migration is partly possible. My goal was only to achieve
an output equivalence between the two grammars, just trying to keep the
grammatical descriptions in both formalisms as identical as possible.
This was possible, because I didn't have to migrate a lot of
metalinguistical descriptions or tools. The migration has mainly be done by just
porting the feature names (modifying them only if needed for
transparency) and transforming them into a type description.
The migration of the rules can
be very straightforward, but one has to think about the feature
descriptions and the type hierarchy, which are, in this explicit form,
not existent in LFG. Those remarks are also valid for the migration of
the lexicon. We also saw the need for a consequent use of macros in ALEP in order to keep the
grammar compact and readable.
The resulting ALEP grammar had to be modified later, since the DP
analysis was not the one adopted for the ALEP grammar, which avoids
empty categories. But the type system resulting from the migration
had to be just minimally modified, which is in the context of a
type-based formalism very important, since a significant time of the
grammar development
is concerned with the definition of the declaration component.