Part I of this post series about periphrasis ended with the observation about how translators handle two very similar passages in The Epistle of Barnabas:

14.2 Καὶ ἦν Μωϋσῆς νηστεύων ἐν ὄρει Σινᾶ “and Moses was fasting on Mt. Sinai”

4.7 Καὶ [ἦν Μωϋσῆς ἐν τῷ ὄρει] [νηστεύων ἡμέρας τεσσαράκοντα καὶ νύκτας τεσσαράκοντα] “And Moses was on the mountain, fasting 40 days and 40 nights”

The first is unanimously handled as periphrastic, the second as non-periphrastic. Outside of merely following the Greek word order regarding where the spatial prepositional phrase occurs, is there any compelling reason to adopt this analysis of the text?

Periphrasis as a sum of its parts

Stephen Levinsohn made an astute observation which has generally been assumed but, so far as I have been able to tell, never actually stated in description of periphrastic constructions. Periphrasis contains two distinct parts: a copula and a participle phrase.[1] Many different sentence elements can come between the copula and the participle in a periphrastic construction, but it is not a free for all. The elements which occur between the copula and the participle are either 1) the subject (which belongs to the copula), or 2) elements which syntactically belong to the participle. One does not find, for instance, an embedded subordinate clause between the copula and the participle in periphrastic constructions. Having tested Levinsohn’s observation on a large corpus of texts outside the NT, I am convinced it is fundamentally sound. This syntactic observation about how periphrastic constructions are ordered can serve as the basis for a helpful way to think about periphrastic constructions.

Think about periphrastics as a syntactic space where two centers of syntactic gravity come into close contact: the copula and the participle phrase. Each of these syntactic centers of gravity is capable of holding various syntactic elements in its own orbit. They can stand in varying degrees of relationship with one another, ranging from total semantic integration (the traditional notion of periphrasis), to syntactic subjugation (copula as main verb and participle as conjunct/predicate which, to be clear, is not periphrasis), and a variety of way points between those two. The closer the syntactic relationship, the more their gravitational force affects one another, eventually fundamentally changing how the two pieces relate to each other as they fuse into one.

To change the metaphor, a sentence with a copula and a participial phrase could be thought of as two “boxes” which can merge together into one. This ability is due to special properties of the verb εἰμί which allow it to become an auxiliary, mostly that it is aspectually wishy-washy and has little semantic might of its own.[2] At the non-periphrastic end of the continuum, they are two completely distinct entities. At the periphrastic end of the continuum, they have lost any separation between the two and function as one unit making one predication. Somewhere in between, they have a fairly ambiguous relationship with one another that can be analyzed as either/or.

What is the pay-off of such a model? It helps in asking focused syntactic and semantic questions of how periphrastic constructions are put together and how they function.

Syntactic limitations of εἰμί

In identifying periphrasis, we can focus attention on the role of εἰμί in the sentence, which is the easiest part to conceptually isolate.

What is so helpful about εἰμί is that it has a rather limited set of syntactic functions. The participle half of the equation has varying degrees of syntactic complexity, dependent on the participle. Incidentally, when the two fuse together, the syntax of the resulting periphrastic construction is driven by the participle, not the copula. On its own εἰμί has a short list of ways it functions to make a complete predication. And that is the key: if εἰμί is making a complete predication on its own, there is no periphrasis. By contrast, if there is no compelling reason for εἰμί to be in a clause on its own, periphrasis is underway.

εἰμί makes a complete predication in the following syntactic situations: 1) bare existential, 2) presentational, 3) subject + predicate nominative, 4) dative of possession, and 5) spatial or temporal adjunct. All these uses share in common that εἰμί makes a complete predication. It is a syntactic center of gravity that is complete, that has its own arguments and adjuncts, and is at peace with the world. In such instances, periphrasis is ruled out as εἰμί is not available to blend together with a participle phrase. It can have a conjunct/predicate/adverbial participle as a further modifier, but it won’t blend together with one.[3] Following is a brief description and example of each of these categories.

Bare Existential

On occasion, εἰμί functions to make a bare existential predication. That is, it can say “I am” in the sense of “I exist”. This is not at all common. An example can be found in Heb. 11.6: πιστεῦσαι γὰρ δεῖ τὸν προσερχόμενον τῷ θεῷ ὅτι ἔστιν “For the one coming to God must believe that he is/exists.” The emphasis here is on the existence versus non-existence of God as opposed to any attributes God might have or simply introducing God as a reference in the discourse.


εἰμί can be used to present a new entity (often a person) into the discourse (this usually goes by the cryptic name “thetic” in the literature). Rarely is the existence or non-existence of an entity an actual topic of interest in a discourse, outside of speculative philosophical discussion. When you say, “There is this guy that I know” in a conversation, you are usually not asserting the existence of someone as opposed to their non-existence, but rather introducing a new person into the discourse.  This usage is abundantly common, and often is paired in Greek with the indefinite pronoun τις. For example, Luke 16.9: Ἄνθρωπος δέ τις ἦν πλούσιος “now there was a certain rich man.” Here Jesus introduces the new character into his story.

Subject and Predicate Nominative

εἰμί can work with a subject and predicate nominative to make a predication. For example, John 1.49: σὺ εἶ ὁ υἱὸς τοῦ θεοῦ “Are you the Son of God?”

Possessive Dative

εἰμί can also pair with a possessive dative to make a complete predication: “It belongs to me.” This usage appears in Acts 8.21: οὐκ ἔστιν σοι μερὶς οὐδὲ κλῆρος ἐν τῷ λόγῳ τούτῳ “there is neither part nor inheritance to you in this word” which we would more normally say in English, “You don’t have a part or inheritance in this word.” The copula can easily be elided in such situations, resulting in a passage like  John 1.6: ὄνομα αὐτῷ Ἰωάννης “(there is) a name to him, John”, or more normally, “his name is John.”

Spatial or Temporal Adjunct

Lastly, εἰμί can pair with a spatial or temporal adjunct to make a complete predication: “he is there”, “it is morning”.  In John 11.10 we find an example of this use: ὅτι τὸ φῶς οὐκ ἔστιν ἐν αὐτῷ “because the light is not in him.”

The Pair of Periphrastics, reconsidered

So where does this discussion of εἰμί get us in assessing the two texts from Epistle of Barnabas? Well, it gives us the strong start to a partial answer. We now have in place the conceptual architecture needed to assess these texts: look for whether εἰμί is making a complete predication.

14.2 Καὶ ἦν Μωϋσῆς νηστεύων ἐν ὄρει Σινᾶ

4.7 Καὶ [ἦν Μωϋσῆς ἐν τῷ ὄρει] [νηστεύων ἡμέρας τεσσαράκοντα καὶ νύκτας τεσσαράκοντα]

A close examination of these two passages, however, leaves us with further, frustrating questions. In the first text, the prepositional phrase ἐν ὄρει Σινᾶ is read by our translators as a spatial adjunct modifying the whole predication “he was fasting.” In our box model terminology, ἐν ὄρει Σινᾶ is part of the participial phrase on this reading. In the next instance, the strikingly similar ἐν τῷ ὄρει is read by the translators as syntactically belonging to the copula, making a complete predication there and closing off the possibility of periphrasis. But couldn’t one argue that this spatial adjunct belongs to the participial clause here, as well? After all, a fronted spatial adjunct is certainly possible and quite common in clauses. Also, there is no rule saying that a temporal adjunct and spatial adjunct can’t both be attached to one verb (in fact, it is quite common). So, what is going on?

At this point, we must introduce one further piece to the periphrastic puzzle. A piece which historically has been largely overlooked (though not entirely) and which I have come to believe is quite consequential for assessing and interacting with periphrastics: constituent order. That will be the topic of Part III. Stay tuned.

[1] He makes this point in Stephen H. Levinsohn, “Functions of Copula-Participle Combinations (‘Periphrastics’),” in The Greek Verb Revisited: A Fresh Approach for Biblical Exegesis, ed. Steven E. Runge and Christopher J. Fresch (Bellingham (WA): Lexham Press, 2016), 307–27. See also the important prior work (published later), “Constituent Order in and Usages of εἰμί: Participle Combinations in the Synoptics and Acts,” in From Ancient Manuscripts to Modern Dictionaries: Select Studies in Aramaic, Hebrew and Greek, ed. Tarsee Li and Keith Dyer, Perspectives on Linguistics and Ancient Languages 9 (Piscataway (NJ): Gorgias Press, 2017), 423–41. While he has his own theoretical reasons for this distinction between copula and participial phrase which I do not fully understand, so I can’t say for certain I fully agree with, the basic insight is helpful for conceptualizing periphrasis.

[2] I am non-committal on the exact aspectual properties of εἰμί. At the very least, εἰμί does not have any aspectually contrastive forms (that is, there is no aorist or perfect of εἰμί). I think Stanley Porter’s name for it, “aspectually vague,” is a useful way to conceive of this particular verb, Stanley Porter, Verbal Aspect in the Greek of the New Testament, with Reference to Tense and Mood, Studies in Biblical Greek (New York: Peter Lang, 1989), chap. 10. There are auxiliaries which are not aspectually wishy-washy, like ἔχω “I have”, but it is easy to see how εἰμί can come to function as an auxiliary. Incidentally, forms of “to be” and “to have” are exceedingly common in usage as auxiliaries in multi-verb constructions across languages. Consider English “I am running” or “I have seen the truth”.

[3] The one major modification to this is the common instances in which εἰμί has a subject and is predicating a variety of adjectival modifiers to that subject, e.g., “the dog is brown and lazy.” A participial phrase can also occur in this sort of context, e.g., “the dog is brown, lazy, and drooling on the floor.” I would argue, contrary to many traditional representations of periphrasis, that the Greek equivalent of these instances is actually a case of periphrasis. Klaas Bentein has argued this most vociferously and effectively in Verbal Periphrasis in Ancient Greek: Have- and Be- Constructions (Oxford: Oxford University Press, 2016). It is not periphrasis of the same degree as the “traditional periphrasis”, but it is periphrastic nonetheless.