The morph. search is reporting a single instance of a middle verb in a particular verse twice. I saw this behavior at 8:9, 11:16, 26:45, 27:34, 27:52 and 28:13 in the report. (This issue might already be known, but I wanted to make sure.) For example:

In terms of the exclusion of the lemma the query is evaluated as:

... @VL[^A] OR @VF[MU] OR (@VA[MU] ANDNOT lemma:ἀποκρίνομαι@VAP)

because AND'd terms are executed before OR'd terms. So the query only excludes verses in which @VA[MU] occurs with the lemma! If you wanted to exclude occurrences of the lemma with ALL the previous terms you need

(... @VL[^A] OR @VF[MU] OR @VA[MU]) ANDNOT lemma:ἀποκρίνομαι@VAP

They are (slightly) different issues. The previous post is a bug with counting the hits of a lemma@morph search. The double hits from your morph search above arise because two morphologies are assigned to the one word. I would have been content with that until I tried a couple of queries:-

RI GNTRI = ESV, LEB, etc GNT = LGNTI, NA27

@VR[^A] 2 2 for a word with alternative morpholgies

lemma:μίγνυμι@VR[^A] 1 3 in Matt 27:34 where lemma has alternative morpholgies

lemma:γράφω@VR[^A] 1 2 in Matt 27:37 where lemma has one morpholgy (NOTE UBS4, Byz & Scrivener 1881 give 1 result)

Clearly Logos aren't sure how these results should be counted!

Clearly Logos aren't sure how these results should be counted!

On reflection, the results for @VR[^A] are correct because it gives the number of occurrences of that term!

The lemma@morph results should always be 1 because the question is HOW MANY lemmas satisfy that morphology. If you want to dig deeper then Analysis will provide the details.