Usefulness of categorizing words based on occurances in different parts of the MS?

[joben]

I was wondering if this method is feasable:

The MS contains 6 different "parts", Herbal, Astronomical, Biological, Cosmological, Pharmaceutical and the Recipe part.

If I went through the Herbal part and wrote down the numbers of occurances for each word, and also the number of occurances per "part", I could hopefully draw some conclusions what type of words this is.

For instance, let's say I find a word in the herbal that exists in about 50% of the herbal pages, 15% of the pharmaecutical pages, 10% of the recipe pages and 5% of the biological pages, it means the word could mean something like "water, nutrition, stem, seed". This is not awfully specific, but what if this scenario happened:

A word is found in the herbal that only occurs a total of 3 times in the entire MS. The other two occurances are in the astronomical and the recipe part.
If this word is nearby a zodiac illustration, it means that the plant could theoretically share the name with the zodiac. If we want to identify the plant name, this could be a good approach. If a word is so uncommon that it only occurs 3 times in this massive MS and in different parts, it could be an indication that this is a name rather than a word.

Has this been done already or is this approach bad? Thanks for taking your time.