Post II of this series on periphrasis left off with the acknowledgment that even a good model of how the copula and participle interact with one another in periphrastic constructions is not always adequate for deciding what is and is not periphrasis. A further piece of the puzzle is necessary: constituent order. We need to be able to assess constituent order and make good arguments about what is likely to be periphrastic or not. For this we need a model of constituent order in Greek. While there is no perfect description of Greek constituent order, there has been good work advancing our understanding of it in the past several decades. I will follow and adapt the work of Stephen Levinsohn to move forward with discussing periphrasticity.[i]

Constituents, phrases, and words…oh my!

Before engaging periphrasis and constituent order, there is one brief terminological point of order: what is constituent order, or, more pointedly, what is a constituent? You may have come across discussions of Greek “word order.” Grammarians and commentators periodically make some comments in that direction. Often when people talk about word order what they are discussing is what linguists now call constituent order (but sometimes they are actually talking about word order). A constituent just means an element of the sentence which is filling a slot in a sentence—like a subject, object, prepositional phrase, and so forth. Constituents range in size from single words, through phrases, up to complex clauses in their own right. For example, compare the following two sentences:

  1. The dog is barking.
  2. The black and white dog with droopy ears, a lot of slobber, and a fear-inducing snarl which has on more than one occasion nearly given me a heart attack, is barking.

In both examples the italicized text is a constituent, namely, the subject of the clause. These sentences show that constituents fill a slot in the sentence and can vary widely in size and complexity. Indeed, in sentence (2) there are even constituents within the subject constituent.

A word order observation would be to note that the adjectives “black and white” stand before the noun they modify instead of after:

  • “black and white dog”


  • “dog black and white.”

A constituent order observation would be to note that the subject constituent precedes the verb in its entirety. You can’t say, for example,

  • The black and white dog is barking with droopy ears

and have it mean quite the same thing or sound quite right. A summary way of stating this constituent order observation is that in English constituents stay together. When you move “with droopy ears” it ceases to be part of the subject constituent. This is also the case in Greek as well. Greek has various possible word order variations within a phrase, but constituents stay together.[ii]

Levinsohn’s model of constituent order

Stephen Levinsohn has done a good deal of work on modeling Greek constituent order in the NT from a functional linguistics perspective. As it turns out, the same basic framework he built on was also being used by other scholars in the Classical Greek tradition to do the same thing.[iii] It is not until around 2009 that there is significant cross-citation in these streams of work, which leads me to believe they were developed in isolation from each other.[iv] This is heartening, considering that in all the essential details these models of word order agree with each other.

While all the various models of constituent order deal with periphrastic constructions to some degree (as periphrastic constructions are present in all the corpuses (or corpi, I suppose) which have been extensively mined for work on these models), only Levinsohn has done specific work on them. I think his model for periphrastic constructions is essentially correct, so I will use it here without any other comment (if you are terribly interested in more detailed engagement, just wait for my dissertation 😐). Levinsohn argues that there is a default periphrastic construction constituent order.[v] By default he means a usage which is not pragmatically communicating anything in particular beyond the semantics of the elements in the sentence. The default pattern is:

  • copula + (subject) + participle phrase

The subject is routinely absent in these constructions (a so-called null subject), resulting in a copula and participle adjacent to each other. The “participle phrase” section can contain just a participle, or a participle and any of its arguments and adjuncts (i.e., stuff that goes with it syntactically). This is modeled as follows:

  • copula + (subject) + participle + (object) + (adjuncts)

The main takeaway from this brief model is that the basic sentence schema gets used when there is no pragmatic motivation to convey anything beyond the bare semantics of the periphrastic construction. Translating that into more normal sounding English, whenever a writer puts one of these elements in a different position than the proposed model there is pragmatic motivation. The order communicates something about how the elements in the sentence relate to the discourse. Some of the orders denote different emphasis, some indicate how different parts of the clause/sentence relate to the context, and so on. It follows from this model that if we see an order like

  • subject + copula + participle + (arguments and adjuncts)


  • prepositional phrase + object + participle + copula

that the writer is conveying some sort of pragmatic goal through these orderings.

This constituent order model gives us the final piece of the puzzle needed to make robust arguments about whether a passage does or does not contain periphrasis. That will be taken up in the next post of this periphrasis series.

[i] Levinsohn has published extensively on constituent order and other aspects of discourse analysis in NT Greek. His broadest reaching work is Stephen H. Levinsohn, Discourse Features of New Testament Greek: A Coursebook on the Information Structure of New Testament Greek, 2nd ed. (Dallas: SIL International, 2000). Many of his conference papers and presentations, which further develop and present the ideas in that book, are available for free at the link cited above. As a word of caution, his work is not very accessible as it was mostly done with the goal of training Bible translators with a strong background in linguistics, making it tough slogging to get into. You have been warned.

[ii] This is a strong rule, but not iron-clad. Greek possess the ability to break up constituents, which is called hyperbaton in traditional grammars, also discussed as discontinuous syntax. Most constituents, though, stay together as a unit, even when they have the possibility of varying word order within the constituent. Of course, we ignore post-positives such as δέ and γάρ when considering whether a constituent is “together” or broken apart.

[iii] In case you are interested, both Levinsohn and Helma Dik, a classicist, developed a constituent order model based on the work of Simon C. Dik, The Theory of Functional Grammar: Part 1: The Structure of the Clause, ed. Kees Hengeveld, second, revised edition, Functional Grammar Series 20 (Berlin: Walter de Gruyter, 1997).

[iv] The earliest work I have come across which shows awareness of both streams, and influence from both of them, is that of Bailey, Nicholas Andrew. “Thetic Constructions in Koine Greek with Special Attention to Clauses with εἰμί  ‘be’, γίνομαι ‘occur’, ἔρχομαι  ‘come’, ἰδού/ἴδε  ‘behold’, and Complement Clauses of ὁράω ‘see.’” Academisch Proefschrift, Vrije Universiteit, 2009.

[v] Technically speaking, Levinsohn argues that there is a default constituent order for sentences which follow the εἰμί + participle pattern, which contains more than just periphrasis. For this post, the details of Levinsohn’s argument on this front, as well as on clauses of the εἰμί + participle pattern which present a new entity into the discourse (called ‘thetic’ or ‘presentational’) will be left aside.