The spell_quality.py script can be used to calculate basic spellchecker quality metrics.

It doesn’t have an installer, and depends on a working installation of Python and the pyenchant library. Simply download spell_quality.py and run it with your python interpreter. It has only been tested with Python 2.5, but Python 2.6 should work fine. Please report on your success with Python 3.


./spell_quality.py --correct correctcorpus.txt \
   --incorrect incorrectcorpus.txt --language zu_ZA

The corpus files should include one single word per line and should be in the UTF-8 encoding. It can be prepared based on any given corpus with commands similar to

tr -c 'A-Za-z' '\n' | sort  > corpus.txt

or if you only want to test unique words:

tr -c 'A-Za-z' '\n' | sort | uniq > corpus.txt