Let’s summarize all 10 active distinguisher paradigms we’ve seen so far (this will probably only layout properly if your browser is wide):

As we’ve already noted, some cells have identical distinguishers (for example, the ῶ of PA-2, PA-3, PA-4 and PA-5). More on that shortly.

But first note something about PA-6—it subsumes the next three paradigms and, in fact, in the case of 2SG subsumes every paradigm except PA-10. In otherwords, a word form from another paradigm technically matches PA-6 too. If you go back to part 10, you’ll see that our exemplar for PA-6 was δεικνύναι, δείκνυμι, and so on. The only reason PA-6 doesn’t have a vowel like PA-7, PA-8, PA-9 is that the vowel is always υ and hence it was dropped out of the distinguisher analysis. But we have no reason at this stage not to supposed the upsilon is an important part of the PA-6 paradigm (it just doesn’t distinguish cells within the paradigm). So I’m going to tentatively put it back for the purposes of comparing across paradigms. I’ll call this modified distinguisher paradigm PA-6a.

In repeating the paradigm of paradigms with this small modification:

Now let’s capture the common elements in the rows:

The INF, although coming in two variants, has the property that it gives us enough information to know every form of the word in the present indicative active.

No other slots in our paradigms do that.

• The 1SG can’t distinguish within the set {PA-2, PA-3, PA-4, PA-5} or within the set {PA-7, PA-9}
• The 2SG can’t distinguish within the set {PA-7, PA-9}
• The 3SG can’t distinguish within the set {PA-7, PA-9}
• The 1PL can’t distinguish within the set {PA-2, PA-3}, the set {PA-1, PA-8}, or the set {PA-4, PA-5}
• The 2PL can’t distinguish within the set {PA-1, PA-7}
• The 3PL can’t distinguish within the set {PA-2, PA-3} or within the set {PA-4, PA-5}

Among other things, this is why the 1SG isn’t a great choice of lemma (or headword, or citation form) for a lexeme. It’s the reason so many dictionaries and lemmatizations show the circumflex verbs uncontracted (e.g. ποιέω for ποιῶ) even though in many dialects, including the Koine, that’s a nonsense word. Even then, most dictionaries don’t distinguish PA-7 from PA-9 (τίθημι vs ἵστημι) although admittedly that’s not as important as they aren’t productive.

In almost all respects, the present active infinitive is the perfect lemma for the present active forms of a verb. Some have argued against the infinitive as lemma as it doesn’t form a clause by itself (although nor do verbs with obligatory complements). A close candidate is the 3SG, the benefit of which is how common it is. The main downside is just it doesn’t distinguish PA-7 and PA-9. But one could hardly go wrong focusing on the INF and 3SG as the forms to most associate with each present active verb.

It should be noted that even though the 1SG is the worst predictively, it’s completely predictable from any other forms. Also, despite some ambiguity in the 1PL and 3PL, they can be predicted from one another. Similarly, all the singulars in the PA-7 and PA-9 words predict each other.

Another way of thinking about this is to group our paradigm classes by their shared properties:

There are the other cross-cutting categories:

If one ignores accentuation, one could conceivably also come up with cross-cutting categories such as PA-{1,2} which shares the ει in the INF, 2SG, and 3SG. Or PA-{4, 9} which both have ατε in 2PL. Or PA-{1, 2, 3} which all have ουσι(ν) in 3PL.

Next we’ll look at the middles.