Pronominal Clitics Attaching to Verbs with Focal Constituents

Specifically, here, we are talking about instances where the pronominal clitic attaches to the verb in spite of the fact that there is a fronted focal constituent which would provide an optimal host for the clitic. In a previous post, some tentative conclusions were drawn regarding this phenomenon on the basis of several contrastive examples. Because the construction is so incredibly rare in Hellenistic Greek, I have expanded my study beyond the New Testament to a wider selection of Hellenistic Greek literature. This phenomenon does not appear in Philo. There are three instances in the Apostolic Fathers, all of which are in the Epistle of Barnabas. There is a single occurrence in Josephus’ works and small handful in the Greek Pseudepigrapha. Not including the LXX (for reasons explained below), there are fewer than 20 instances of pronominal clitics attaching to verbs even when there is a fronted focal constituent in the clause.

In contrast to those texts, this structure is fairly common in the LXX, which I would suggest is caused by the translational nature of the text. Hebrew pronouns are not affected by prosody and pulled forward by fronted focal constituents the way Greek clitic pronouns are. The fact that the Greek pronouns stay in their position following the verb so regularly suggests a close adherence to the word order of the Hebrew text on the part of the translator. This is clearly seen in the following examples with the focal constituent and pronoun underlined in each sentence.

(1) Genesis 43:6 Τί ἐκακοποιήσατέ με ἀναγγείλαντες τῷ ἀνθρώπῳ εἰ ἔστιν ὑμῖν ἀδελφός;

לָמָה הֲרֵעֹתֶם לִי לְהַגִּיד לָאִישׁ הַעוֹד לָכֶם אָח׃

Other examples include: Gen 29:25f.; Ex 5:22; Num 14:11; Josh 9:22; Ruth 1:21; 1 King/1 Sam 19:17; 20:8; 28:12, 15; 2 King/2 Sam 19:44; 23:15; Psalm 21:2; 41:6, 12; 42:2, 5. There are numerous others. The results noted here were found by only searching Genesis through Psalms and only for clauses with content questions with a pronominal constituent. Interrogatives with fronted question words are prototypically associated with Focus. Were I to expand my search to include the rest of the LXX and a broader set of fronted focal constituents, the number of instances would likely multiply.

But even the eighteen examples provided here are substantial evidence that this ordering, where the Hebrew word order is rigorously follow, is a result of literal Greek translation. Considering the rarity of the ordering non-translation Greek—virtually non-existent—it can probably be safely said that the order is, at best, highly infelicitous and at worst, ungrammatical. In the discussion below, we will examine all instances occurring in Josephus, the Old Testament Pseudepigrapha, the Apostolic Fathers, Philo, and the New Testament of the order:

Focus, V=Clitic.Pro, (X)

The goal is to determine whether any patterns are recognizable that might point toward determining the grammatical acceptability of the ordering.

Clauses Quoted from the LXX

The ordering of the following three examples can be attributed to the same translational phenomena as example (1) and have little need for any more comment than that. These quotes are the only explicit places where we can determine the ordering is caused by a Hebrew source text because these are the only places where we have access to a Hebrew source text.

(2) Mark 15:34 ὁ θεός μου ὁ θεός μου, εἰς τί ἐγκατέλιπές με;

(3) Barn 6.1 Τίς ὁ κρινόμενός μοι; ἀντιστήτω μοι· ἢ τίς ὁ δικαιούμενός μοι;

(4) Barn 16:2 ποῖον οἶκον οἰκοδομήσετέ μοι

Clauses with Two Prosodic Phrases

Of the thirteen clauses where this pattern occurs, a number of them can be viewed as prosodic parallels to the word order:

[ɸ: Topic=Clitic.Pro] [ɸ: V (X)][1]

I previously argued that this ordering involved two prosodic phrases, rather than the more typical one. This is most clearly seen in elevated emotive clauses such as those in the following examples, though there are other causes as well.

(5) John 8:46 διὰ τί ὑμεῖς οὐ πιστεύετέ μοι;

Contextually, this little question falls in the second half of Jesus discourse in a dialogue with the Jews. In this section, Jesus makes a heated argument that his interlocutors are not God’s children. The answer to the question of word order here is rather straight forward. We have a combination of two phenomena already described: first of all, we have noted in the previous post that particularly salient prosodic phrases can create a larger than normal break in prosodic phrasing. Likewise, we have observed that verbs, because of their syntactic and pragmatic status tend to receive a major stress accent. For that reason, I propose that διὰ τί pulls ὑμεῖς forward into its more salient prosodic phrase, but at the same time, πιστεύετέ still receives a stress accent of its own, to which μοι then attaches. The result is a prosodic pattern quite similar to what I described for pronominal clitics attaching to Topics in the previous post, where we have two major phonological phrases.

(6) a. [ɸ: διὰ τί ὑμεῖς] [ɸ: οὐ πιστεύετέ μοι;]
b. [ɸ: Why do you] [ɸ: not believe me?]

A similar explanation fits well with Mark 8:2 / Matt 15:32.

(7) Mark 8:2 / Matt15:32 σπλαγχνίζομαι ἐπὶ τὸν ὄχλον, ὅτι ἤδη ἡμέραι τρεῖς προσμένουσίν μοι

In terms of information structure, both the nominative and the verb are focal, both conveying new asserted information, though it is the nominative that is the more salient of the two. The position of the pronominal clitic appears to be motivated by the very same constraint as before: an impassioned statement/exclamation with two distinct stressed phrases.

(8) a. σπλαγχνίζομαι ἐπὶ τὸν ὄχλον, [ɸ: ὅτι ἤδη ἡμέραι τρεῖς] [ɸ: προσμένουσίν μοι]
b. I have compassion for these people, [ɸ: because already three days] [ɸ: they’ve stayed with me.]

A non-emotive example of multiple prosodic phrases can be seen in Acts 10:29.

(9) Acts 10:29 πυνθάνομαι οὖν τίνι λόγῳ μετεπέμψασθέ με;

Like many other examples, the clause at hand is a question, but unlike the previous examples, there is no clear elevated emotion driving it. Instead, we have another phenomena that occurs in several of our examples: a complex interrogative construction, where the semantic force of the question (reason: “Why?”) is created by the combination of τίνι and λόγῳ. This clause provides collaborative evidence of the specific placement of a pronominal clitic in a given prosodic phrase. As hypothesized previously, the pronominal clitic must attach as close to the phonological peak of the phrase as possible. In questions, that peak is filled by the question word: τίνι. And that is where we would expect με to attach, but in this example, that position is blocked by λόγῳ. This problem is resolved by 1) having two prosodic phrases

(10) a. πυνθάνομαι οὖν [ɸ: τίνι λόγῳ] [ɸ: μετεπέμψασθέ με;]
b. So I asked, [ɸ: Why] [ɸ: did you send for me]

A number of our examples can be explained in this manner: 3 Baruch 4:1; T. Zebulon 6:6; Apoc.Moses 24:2.

Clauses with Significant Textual Variation

One of the examples contains significant textual variation, suggesting that the Greek speaking scribes recognized the ungrammatical nature of this construction.

(11) Mark 10:36 ὁ δὲ εἶπεν αὐτοῖς· τί θέλετέ [με] ποιήσω ὑμῖν;

I cautiously accept the NA27’s text as correct simply because it can easily be viewed as thoroughly ungrammatical from a couple different perspectives. If the subjunctive is correct, then we should expect the pronominal clitic to be pulled forward to connect with τί, which make the clause felicitous. Or, if the με is actually syntactically the subject of the complement clause, then the subjunctive mood is a production error where there should be an infinitive. In that case, the με cannot be moved forward any more than it already is since that would move it out of its syntactic domain. All of this syntactic conflict is seen rather clearly in NA27’s apparatus.[2]

Residue Clauses

(12) a. John 13:37 διὰ τί οὐ δύναμαί σοι ἀκολουθῆσαι ἄρτι;
b. Acts 8:36 τί κωλύει με βαπτισθῆναι;

The situation of examples (12a-b) is difficult to explain. We might be inclined to say that because με and σοι function at a lower syntactic domain (a subordinate clauses), they cannot be pulled forward beyond the edge of their clause. Unfortunately, that is clearly not the case. Even a cursory look at Hellenistic texts shows this not to be the case. Counter examples are discovered quite quickly, shown in (13a-c):

(13) a. Matt 16:15/Mark 8:29/Luke 9:20 ὑμεῖς δὲ τίνα με λέγετε εἶναι; “Who do you say I am?”
b. John 7:19 τί με ζητεῖτε ἀποκτεῖναι; “Why are you trying to kill me?”
c.  Isaiah 27:4 τίς με θήσει φυλάσσειν καλάμην ἐν ἀγρῷ; “Who will appoint me to guard stalks in a field?”

We have a different situation from examples (14a-b):

(14) a. Matthew 26:34 ἔφη αὐτῷ ὁ Ἰησοῦς· ἀμὴν λέγω σοι ὅτι ἐν ταύτῃ τῇ νυκτὶ πρὶν ἀλέκτορα φωνῆσαι τρὶς ἀπαρνήσῃ με.
b. Matthew 26:75 καὶ ἐμνήσθη ὁ Πέτρος τοῦ ῥήματος Ἰησοῦ εἰρηκότος ὅτι πρὶν ἀλέκτορα φωνῆσαι τρὶς ἀπαρνήσῃ με·

These two clauses could be viewed as having multiple prosodic phrases, though it is not entirely clear where such a reading is acceptable in the context. But that is the most likely scenario. Note also the significant textual variation in verse 34.[3]

(15) Mark 6:25 θέλω ἵνα ἐξαυτῆς δῷς μοι ἐπὶ πίνακι τὴν κεφαλὴν Ἰωάννου τοῦ βαπτιστοῦ.

It is not clear in Mark 6:25 whether the clitic can be pulled forward by this adverb. And there are so few instances in Greek literature and documentary texts, that no conclusions can be drawn. All instances in Perseus’ Greek texts are from the New Testament and Documentary Papyri. With that said, there are a number of other adverbs that can and do pull pronominal clitics forward.

(16) John 9:26 Τί ἐποίησέν σοι;

The clause in (16) fits within the narrative of John’s Gospel where Jesus has just healed a blind man the Jews are trying to determine what exactly happened. This particular clause is rather difficult to explain. The context does not necessarily suggest that this question would have been said with any special or emotive force that would cause two separate prosodic phrases. Yet even still, this might be the best explanation for the ordering here. The majority of manuscripts, including P66, א‎2, A, L, Θ, Ψ, 070, 0250, 33 and Families 1 & 13,  along with the Textus Receptus read πάλιν just before this clause.[4] If we accept this reading, it could be interpreted as reflecting some frustration on the part of the Jews who were questioning the healed man. However, this explanation must presently be limited to speculation, rather than being a confident conclusion, the perspective a textual critic would be valuable on this question.

(17) Josephus, Ant 6.226 ὁ δέ οἶδα γάρ, ἔφη, πάντα σε χαρίζεσθαί μοι καὶ παρέχειν ἐθέλοντα

The position of  μοι in this clause seems rather inexplicable. There is no reason for posit a complex prosodic structure with multiple phrases. In context with the fronted and focal πάντα, one would expect the phenomena of clitic doubling to appear: πάντα σέ μοι χαρίζεσθαι. Yet that is not the case here. Once again, we must turn to supposition. Contextually, this clause appears within Josephus’ narrative of David, Jonathan, and Saul. In this pericope, David and Jonathan devise the plan where David hides in the field instead of dining with the King in order that Jonathan might find out what Saul’s opinion of David is without him around.

Must of the narrative itself is quite close to the Old Testament account in 1 Samuel 20, but this particular speech from David is prefaced by an added clause with our unusual clitic placement. In this context, there is the possibility that πάντα σε χαρίζεσθαί μοι reflects a translation of the Hebrew text that we no longer have access to. That would make this unusual construction parallel to the LXX translational phenomena described previously.


Most of the data for this rather unusual structure can be explained with relative ease. Some are clearly caused by a translation’s source language. Others can be explained as a result of complex phonological/prosodic structures. This leaves us with seven residue examples and even with these, they might fit within the other categories are mentioned in the discussion. Whether this structure is as infelicitous as it appears to be could only be determined with access to a native speaker of Hellenistic Greek.[5]

With that said, it is highly significant considering how prevalent the structure is in the LXX that it does not appear in the Gospel texts more often. If, as some have claimed, the Gospels are originally documents translated from Hebrew or Aramaic, we would expect far more instances of this phenomenon. That we do not is evidence that all four Gospels are original Greek compositions, particularly since the few occurrences there are can  generally be attributed to other causes.

[1] ɸ = Prosodic/Phonological Phrase.


[2] 1 2 4 C Θ f 1.13 565. (1241). 1424 pc 1 4 a b i ❙ 4 D ❙ τι θελετε ποιησαι με (א2) A (L W 427*) m ❙ txt א1 B Ψ 2427c (NA27, 124).

[3] 2 3 1 A ❙ 1 3 2 א* 33 pc τρις απαρνησει με P53 B C Θ 565. 579. 892. 1424 al txt א2 D L W 067. 0160vid f 1.13 m (Ibid., 77).

[4] Ibid., 280.

[5] A Modern Greek speaker might be helpful on this question, but not necessarily. While the pronominal system went through significant change, there are still a few parallels that can be drawn.