Skip to content
Dec 15 / Steve Runge

Exceptional thinking-2

This is the continuation of a post that started here, focusing on exception clauses in language. In the last installment (which you really should read before continuing), I discussed how some grammarians have viewed  ἀλλά and εἰ μή as nearly interchangable. I provided an example of a clause with ἀλλά, and now for the εἰ μή counterpart. The following passage from Mark 6:4-5 has two of them. The excepted element serves to replace an element that precedes.

The negated statements in vv. 4 and 5 are not entirely true; exceptions to each statement are added in the apodosis. There actually is a context where a prophet is without honor: in his hometown among his relatives and his household. Jesus actually was able to perform miracles in Nazareth: he laid hands on a few sick people, healing them. The statements are essentially incomplete until one reads the apodosis; it is only after this that the statement is completely accurate. The incomplete proposition of the negated clause sets the stage to highlight the restricted element that follows.

Note that the same propositional information could have been communicated using only positive statements, thereby avoiding the complexity of the negation and restriction.

a) A prophet is without honor in his hometown.

b) Jesus was able to lay hands on a few sick people and heal them.

Both of these statements communicate the same information as the negated statements in Mark 6:4-5, but without the same rhetorical punch. The pragmatic effect of using the negation + restriction is to highlight the restricted element because of its significance to the discourse. The focus here is not on the fact that Jesus was dishonored, but on the location where it takes place. Of all the people who know him and should accept him, it is the hometown crowd. Think of how many small towns tout some native son’s achievements as their claim to fame. In the case of Jesus, they reject him. As a result, he was apparently limited in the works he was able to perform there. Using only the positive statements would not have focused the reader’s (or hearer’s, for cwc) attention on these significant elements in nearly the same way as in the biblical text.

Okay, here is the problem I ran into (forward-pointing reference): If both ἀλλά and εἰ μή introduce something that replaces what precedes, what is the difference between them? What is the distinct role that each plays? It vexed me for days. And then, it hit me… Can you figure it out? Leave a comment. Next installment out Wednesday.


  1. Mike / Dec 15 2008

    I’m going to go out on a limb here…

    Is it that in αλλα point/counterpoint sets, the point anticipates and presupposes the counterpoint, whereas with ει μη the point could hypothetically standby itself without the counterpoint?

    Perhaps I could have checked a few more examples myself before making this guess…

  2. Steve Runge / Dec 15 2008

    Let me be more specific in the question: what is the semantic or processing constraint that is unique to each connective? If they completely overlapped, one of them would have dropped out. They each play a unique role. Keep thinking.

  3. Mike / Dec 15 2008

    Well I know the difference in syntax. αλλα is still functioning as a coordinating conjunction, whereas the ει of ει μη is a modal complementizer. αλλα must always introduce an identical phrase type and ει μη can only introduce a clause, though in the example you’ve give for αλλα, its still two clauses that are coordinated with the second one partially elided.

    For a moment, I thought it might be related to the ellipsis, but Mark 4:22 contains an ellipsis with ει μη, so it must be something else.

    I know there must be a semantic difference, but I’m at a loss (απορω).

  4. Seumas / Dec 15 2008

    I won’t phrase this right, but from your examples:
    αλλα takes the content of one phrase and replaces it completely with another. It’s saying something like, “take out X from the statement, and insert Y”
    ει μη is performing a different kind of substitution, at least in the examples you’ve provided, saying something like “statement X is true, only when statement Y which is a restricted negation of X is added”.

    I don’t think I’m being clear, but I’m feeling like there is a difference. No doubt you’ll enlighten us on your solution shortly

  5. Steve Runge / Dec 15 2008

    Give Seumas a prize! What exceptional thinking! The distinction has to do with the membership of the replacement with the original set. Was it a member of the original (theoretical) set or not? There is a difference in the constraints for each particle. Way to go! Where were you three months ago when I was undergoing hair loss?

  6. Mike / Dec 15 2008

    Ahh! That makes perfect sense, especially in light of the modal nature of ει.

    But it still seems incredibly subtle for Mark 4:22.

    οὐ γάρ ἐστιν τι κρυπτὸν, ἀλλʼ ἵνα ἔλθῃ εἰς φανερόν.

    My question is this: Would a second language Greek speaker such as Mark have recognized this sort of subtly?

  7. Steve Runge / Dec 15 2008


    Yes, I think he would have picked up on it. There is an example in the grammar from Mark where the synoptic parallel uses ALLA, but Mark uses EI MH (I think it is the one that Robertson cites). If you look at the context, Mark’s version changes the members of the set that is negated, the synoptic parallels do not. The post Wednesday (or whenever I get it finished) has a diagram to describe this, but it needs some tweaking. I am hoping for a Priscilla and Aquila that can pull me aside and straighten me out.

  8. Mike / Dec 15 2008

    I’ll take a look at the grammar and will look forward to the next post. Thanks, Steve.

Comments are closed.