Transitivity isn’t a binary thing. You can scale it across usage. This is clear in things like lexical semantics. Consider the middle instances of φοβέω, for example.
ἐφοβήθησαν δὲ ἀκούσαντες ὅτι Ῥωμαῖοί εἰσιν,
they grew afraid when they heard they were Roman citizens (Acts 16:38).
ἐφοβήθησαν φόβον μέγαν·
There was greatly terrified (they feared [with a] great fear )(Luke 2:9)
ἐφοβήθησαν τοὺς ὄχλους
they feared the crowd (Mark 12:12)
Intransitive change of state –> Semi-transitive experiencer-stimulus –> Syntactic transitive experiencer-stimulus
Now from the perspective of the prototypical transitive clause (volitional animate participant acting on an inanimate non-volitional object), all three of these clauses are divergences. And the most transitive version of this verb necessarily involve the active voice:
μηδὲν αὐτοὺς ταραχῶδες ἐφόβει
Nothing terrifying was causing them fear (was frightening them (Wisdom of Solomon 17:9)
What’s the point of all of this? Well, transitivity is interest to me. That’s mostly it. On the other hand, another point might be that when it comes to describing the semantics of the Greek middle, subject affectedness may certainly be a necessary condition for middle morphology, but it still isn’t the whole story.
Anyway, I’m enjoying my wife’s research on the topic. Her larger thesis project deals with this and many other questions (this is from her data). And her forthcoming presentation/paper at the Cambridge Greek Verb conference is going to examine the impact of metaphor and analogy in language change in the middle voice with particular reference to -θη forms in the Koine period.
Expensive Books that Deal with these Questions: