I've used stemmers in the past but never looked at the exact mechanics behind them. Thank you for the information.

I notice that you evaluate the stemmers using the change in characters but how do the number of unique words change? I imagine this metric would play a more important role when using stemming within embeddings. Any thoughts?