The syntax of Mark 10:36 has been discussed at length of B-Greek a few months ago. (Mark 10:36 Τί θέλετέ [με] ποιήσω ὑμῖν;). Generally, speaking we all didn’t really know what to do with the construction or the particular reading in this verse that the editors of the NA27 had chosen as most likely being original. The clause in question, following the NA27 is:

τί θέλετέ [με] ποιήσω ὑμῖν;

The variation in the textual history of the clause is rather severe. The two main alternatives are:

τί θέλετε ποιήσω ὑμῖν;
τί θέλετέ με ποιησαι ὑμῖν;

Some describe the NA27’s choice as being grammatically impossible, a conflation of the other two main possibilities. For example, R. T. France states (2002), “The syntactically impossible reading of א B, τί θέλετέ με ποιήσω, must result from a conflation of the two constructions. The reading which best explains the variants is τί θέλετε ποιήσω (with ἵνα understood), the abruptness of which led to correcting the subjunctive to an infinitive, with the consequent addition of με.”

That the NA27 text is ungrammatical is generally agreed upon. It’s just not possible.

By chance, however, I spent part of my summer reading through Theodore Markopoulos’ The Future in Greek: From Ancient to Medieval (2009), a volume that focuses on the periphrastic constructions (e.g. μέλλω+inf, for example) that would eventually muscle out the synthetic future during the Byzantine and Medieval period. θέλω constructions play a prominent part in every chapter of the book, and his discussion of various formations that appear with the verb: V + INF, V + ἵνα + SUBJ, V + V_subj. There’s a lot of interesting data that Markopoulos provides that’s worth looking with reference to this supposedly “impossible” constructions.

Because I hold that understanding Greek in historical context is an absolute necessity for understanding the language in the 1st century, I think it is worthwhile to look at Markopoulos’ analysis of this construction in the Classical period, the Hellenistic-Roman period and also in the Medieval period.

Θέλω+V_subj in Classical Greek

Now no one questions the acceptability of the plain old θέλω+V_subj. Its grammatical status is not in question. But it is worth commenting on its development in the Classical period. Markopoulos (2009: 38-39) notes the following facts about the construction:

1. At least two other modal-like verbs are used in the same way: βούλομαι ‘want’ and κελεύω ‘urge.’
2. The construction is only used in dialogue and likely were a feature of the spoken language more so than the written.
3. All known examples from the Classical period involves questions.
4. Goodwin (1875) suggests that the construction originated as two paratactic sentences that, by the time of Ancient Greek, likely involved some kind of subordinate relationship. Whether this is an accurate origin cannot be determined because there are no examples in Homer.
5. There is however, a parallel from Early Latin with the V +  V_subj pattern.

There’s not much conclusive here. And at this point is time there is no instances of a shared argument between the two clauses like we seen in the NA27’s text of Mark 10:36 (τί θέλετέ [με] ποιήσω). With that said, Goodwin’s proposal about the origin of the construction parallels that of both Moulton and Howard (1929) and Robertson (1923) who suggest that the construction is *still* paratactic. That interpretation results in treating this as two questions: “What do you want of me? What should I do?” But if, according to Markopoulos, there was already a sense of subordination in the Classical period, then it is not likely that the language would have degrammaticalized away from subordination a few hundred years later.

Θέλω+V_subj in Hellensitic-Roman Greek

This is definitely where the construction becomes more interesting. Unfortunately, Markopoulos does not provide a particularly large number of examples. And, equally frustrating, he does not provide any direct parallels to our clause—though he does state that they do exist. With that said, provides the following two charts of statistical data (74; 76).

Table 3.11 Volitional (ἐ)θέλω in the H-R period (papyri)

Complement	3rd c. bc	1st c. bc -1st c.ad	1st c.ad	2nd c.ad	3rd c.ad	4th c.ad	Total
INF.	35	6	318	75	35	89	558 (92.7)
ἵνα+SUBJ (disj.)	—	—	9	—	—	2	11 (1.8)
ἵνα+SUBJ (co-ref.)	—	—	1	—	—	1	1 (0.2)
VS (co-ref.)	17	1	24	3	—	3	32 (5.3)
TOTAL	36	7	352	78	35	94	602

Table 3.12 Volitional (ἐ)θέλω in the H-R period (papyri)

Complement	3rd c. bc	1st c. bc -1st c.ad	1st c.ad	2nd c.ad	3rd c.ad	4th c.ad	Total
INF.	9	19	30	74	44	46	222 (93.4)
ἵνα+SUBJ (disj.)	—	—	1	1	—	—	2 (0.8)
VS (disj.)	1 (?)	—	—	5	5	1	12 (5.0)
VS (co-ref.)	—	—	—	1	—	1	2 (0.8)
TOTAL	10	19	31	81	49	48	238

Now, numbers aren’t particularly useful out of context, but I’ll do the best I can to make the data relevant for our purposes here. The disjunctive/co-reference parentheses refer to the subject of the subordinate clause being either the same as or different to that of the matrix clause. And it’s the disjunctive data that we’re primarily interested in here since that’s what we’re dealing with in Mark 10:36. And with that, its somewhat disappointing to see that Markopoulos does not provide numbers for disjunctive subjunctive complement clauses in the non-papyri texts. He doesn’t explicitly list his corpus anywhere, so I don’t know if he simply did not have any examples available or what. I do know that he used at least part of the NT because he lists Matthew 26:15 as one of his infinitive complement examples. So who knows there what he was thinking. If there are no other examples in non-papyri texts, then at the very least, it is clear that Mark has much more in common with the every day language of the people than with the literary writers—but we already knew that.

Some More Data

What’s equally disappointing is that Markopoulos provides no examples of disjunctive subjunctive complements, so we do know know if there are other examples where we have the enclitic object pronouns appearing between the matrix verb and the complement verb. So we’re pretty much back where we were before in terms of whether or not, the appearance of the με between the two verbs is an acceptable construction or not. I’ve been digging through examples from the Duke Databank of Documentary Papyri, as well as other contemporary and literary Hellenistic Greek texts for other θέλω+Vs and could only find a few examples of pronouns appearing between the two verbs and in all cases case it was a dative pronoun. Here’s one example from Epictetus:

τί ἔτι ἀγωνιᾷς, μὴ οὐ δείξῃς ἡμῖν, τίς εἶ; θέλεις σοι εἴπω, τίνα ἡμῖν ἔδειξας;
Epictetus, Dissertationes ab Arriano Digestae 3.2.14

There are a few more, but they’re not particularly different.* But even though there isn’t the same issue of the accusative clitic =με that suggests the subjunctive of Mark 10:36 should rather be an infinitive, this is still a clear instance of the object of the subjunctive verb being pulled forward to the matrix verb. So there’s nothing, in principle, wrong with a pronoun appearing there.

*There are also some great false hits like this one: εἴ τι δέ ποτε θέλεις, σοὶ πέμψω (DDBDP PSI Document 825: IV/Vspc? [LINK])—don’t let it trick you.

The question before us then is this: if its perfectly acceptable to have the pronominal object of the subordinate clause between the two verbs, why is it then not acceptable to the the pronominal object of the main clause between the two verbs? And if Markopoulos is correct and that the V+Vs pattern goes beyond θέλω to other modal-like verbs, then what we have is a relatively rare, but consistent alternative construction to the standard infinitive and ἵνα+subjunctive constructions.

With that in mind, I’ve done some searching for the pattern with the two other verbs that Markopoulos mentions: βούλομαι ‘want’ and κελεύω ‘urge.’ We get some very nice examples:

βούλει φῶμεν, ὦ Κρίτων, ᾗ ἄλλους ἀγαθοὺς ποιήσομεν;
Plato, Euthydemus 292d
τί γὰρ βούλεσθʼ εἴπω;
Demonsthenes, Adversus Androtionem 70
ἀλλʼ ὅ τι βούλεσθʼ ἐξετάσωμεν.
Demosthenes, On Organization 28
βούλεσθε εἴπω καὶ συνεξέλωμεν;
Aelius Aristides, Orationes 34

You may have noticed that every single one of these examples appears in dialogue—just as we saw above Markopoulos noted. And what’s more, they continue to only be. I haven’t tried to search for examples with intervening pronouns. With literally millions of words of text, sifting through false hits becomes quickly becomes overwhelming as the complexity of the searches go up.

A Proposal

In terms of significance, I’d say that we cannot simply disregard the pattern, even though it definitely doesn’t translate directly/naturally into English without add an English infinitive (to+bare verb). Contrary to what France stated above, we cannot simply chalk up the NA27 text to someone viewing the V+Vs structure as ungrammatical and then changing the subjunctive to an infinitive and then someone else adding the με and then someone else correcting it back to the subjunctive. When you combine that with the fact that there’s no evidence that the ἵνα is “understood,” France isn’t really left with a leg to stand on. The V+Vs construction has a long and impressive pedigree, in spite of its rarity.

And in terms how we explain the construction, there’s a possibility that takes Moulton, Robertson, and Goodwin’s observations about paratactic origins seriously (i.e. deriving from the interrogative subjunctive), but recognizes that there is still some level of subordination in the construction. This type of complex clause is called Co-Subordination. I know, it’s not a great name. The idea is that it provides a way for conceptualizing constructions that aren’t quite subordination and aren’t quite coordination. This has been a real problem for a number of minority languages around the world, where complex sentences are more difficult to analyze in terms of the traditional terminology and since then, the concept of co-subordination has been extended to see how it works (with much success) for the more well-studied languages.

I’ve created a tree diagram below that I hope demonstrates the relationship. The formalism is Role and Reference Grammar (there’s a brief overview of RRG here). I’ve tried to label and explain anything that might possibly need explanation.†††

Now, its entirely clear to me whether or not the first tree or the second tree is more accurate. The difference would be determined on the basis of whether any clause level modifiers could appear between τί θέλετέ [με] and ποιήσω ὑμῖν;. And without more data, a final conclusion is unlikely. Still, since the construction is primarily limited to dialogue, its not as common as it could be in the surviving texts, but all of the examples of the construction I’ve thus see would suggest the second representation is the correct one and that clause level modifiers aren’t permitted between the two pieces of the construction.

Now then, this post has gone on for quite a while now and I’m really curious about any feedback about this proposal.

Works cited:

France, R. T. 2002. The Gospel of Mark. NIGTC. Grand Rapids, Mich.: Eerdmans.
Goodwin, William Watson. 1875. Syntax of the moods and tenses of the Greek verb. Boston: Ginn and Co.
Markopoulos, Theodore. 2009. The future in Greek: From Ancient to Medieval. Oxford: Oxford University Press.
Moulton, J. H. and Wilbert Howard. 1928. Grammar of New Testament Greek: Accidence and word formation. Edinburgh: T&T Clark.
Robertson, A. T. 1923. A grammar of the Greek New Testament in light of historical research. London: Hodder & Stoughton.