Open funderburkjim opened 3 years ago
This correction facility is implemented by the second step of redo.sh .
published declension from some Sanskrit grammar is more likely correct than the algorithmic declension.
We take it for granted, right.
This could be done by changing the underlying algorithm; but often it is hard to decide where an algorithm needs to be tweaked to effect the desired change.
French go the manual way, so why should we not?
Generally, rather complex algorithms are used to derive Cologne declension tables. It will sometimes be desirable to change a given Cologne declension table from its algorithmic value.
For example, we may believe that a published declension from some Sanskrit grammar is more likely correct than the algorithmic declension.
This could be done by changing the underlying algorithm; but often it is hard to decide where an algorithm needs to be tweaked to effect the desired change.
Thus, a rather simple correction facility has been implemented. We may effect a correction by manually entering a declension table for a given declension model and headword. Then, all such corrections will be used to override the algorithmic declensions.