7:22 AM Ward Powers has written a fine introduction to New Testament Greek. He introduces the middle voice on p. 71 by writing:

Commonly, the middle voice is used for intransitive verbs, or where the action of the verb does not carry over to an object but solely affects the subject. Often the subject of the middle voice verb is acting for himself or in his own interests.

In addition:

Closely related to the first use, the middle voice can have a reflexive sense — the action was done to or for the subject.

Powers emphasizes that these two usages can “shade into each other,” though many examples can be given where two distinct usages are obvious, as with:

Jesus washed (active voice) the disciples’ feet.

Pilate washed (middle voice) his hands.

They put on (active voice) him his own clothes.

Do not put on (middle voice) two tunics.

You yourself keep (active voice) the law.

They are to keep themselves (middle voice) from idol food.

I find it interesting that the example Powers focuses on in his ensuing discussion is the verb phulasso. As he puts it, “Thus Matthew and Luke use the active voice of phulasso in a passage where it is appropriate for the middle to be used, whereas in this passage Mark does use the middle:

Matthew 19:20  All these things I have kept (active of phulasso)

Mark 10:20         All these things I have kept (middle of phulasso)

Luke 18:21         All these things I have kept (active of phulasso)

This was precisely the evidence I used when I sought to respond to Robert Stein’s argument for Markan Priority based on this very construction. Stein, however, chose to take the evidence in a different direction. He argued that Mark used an incorrect form of the verb phulasso (the middle), and that this incorrect form was corrected by both Matthew and Luke into the active voice. I argued, in response, that I did not think there was anything inherently incorrect about Mark’s use of the middle here, especially when you take into consideration the use of the middle voice of phulasso in the LXX in contexts dealing with keeping the commandments of God. (See Some Dissenting Notes on R. Stein’s The Synoptic Problem and Markan “Errors.”) This is merely a small slice of the relevance of linguistic study to the interpretation of New Testament texts.

Lately it’s become clear to me that the question concerning correctness and incorrectness in language is not so much a linguistic one but a sociolinguistic one. In other words, it is people who determine what is correct and incorrect in language, not textbooks. In a sense, then, if everybody says “It’s me,” then this construction is correct. (One “should” say “It is I.”) I have the happiest memories of debating this issue with my fellow students while in Basel, as even then I had begun to question the Markan Priority Hypothesis and especially the so-called linguistic arguments for the posteriority of Matthew and Luke. I determined then and there that one day I would study the matter in greater detail, since my own teachers at Talbot had all espoused the consensus view regarding Synoptic origins. It was fascinating contrasting the arguments put forward by Stein and others with the statements of the church fathers. What struck me most, however, was the need to rethink the primary linguistic data in the texts themselves. I must also mention the work of William Farmer, who once invited me to spend a week in Dallas with his working group on the Synoptic Problem. It is a remarkable privilege, this process of thinking and rethinking one’s views about this and that. My task as a Greek teacher now is not to try and convince my students that I am right and someone else is wrong but rather to equip them with a tool (Greek) that will help them all become better Berean Christians.

(From Dave Black Online. Used by permission.