I did consider this but the effort of doing a full dictionary pass is a lot to ask for marginal improvement and I don't know anyone quite as obsessed as me who would do it. Paying someone would be possible, but finding the right combination of "willing to do it for pay" and "I trust their judgement" is hard.
In practice, the way I approach this is by reacting to complaints from players who either don't know words I included or were disappointed I didn't include a particular word.
How about doing a pass where you categorize as in for sure, out for sure, and idunno. There may not be too many of those idunno‘s, so may be possible to enlist help.
That said your approach seems to have worked well. Kudos!
Yup, that's a really good point. I kind of wish I had marked ambiguous words on a first pass, and then taken a bit of a different approach for a second pass of just the difficult ones.
I don't, unfortunately. I'm trying to avoid having a dedicated backend for this so there are Google Analytics but they don't allow that granular of a metric.
That definitely could be something interesting, but I'd probably need a decently larger player base to get enough data, considering how many words are possible.
In practice, the way I approach this is by reacting to complaints from players who either don't know words I included or were disappointed I didn't include a particular word.