Greek Syntax: Article Introducting Prepositional Phrase

Awhile back over on the Logos Newsgroup for Greek, someone asked a question:

Someone has commented that there are 484 occurrences of the definite article occurring without a noun introducing a prepositional phrase, such as, "τα επι τοις ουρανοις." I wonder if someone would teach me how to search my GNT (N/A27) to confirm this statement?

The example is (I believe) from Eph 1.10:

εἰς οἰκονομίαν τοῦ πληρώματος τῶν καιρῶν, ἀνακεφαλαιώσασθαι τὰ πάντα ἐν τῷ Χριστῷ, τὰ ἐπὶ τοῖς οὐρανοῖς καὶ τὰ ἐπὶ τῆς γῆς ἐν αὐτῷ. (Eph 1:10, NA27)

as a plan for the fullness of time, to unite all things in him, things in heaven and things on earth. (Eph 1:10, ESV)

Note that the same structure is used in "things on earth" in the same verse.

Anyway, the best way to find stuff like this — where you’re really searching for a relationship between words and/or phrases even though it looks like proximity will get you close enough — is a syntax search. In this example, the relationship is between the article and the prepositional phrase. It is more than proximity (occurring close to each other or in sequence); it is functionally that the prepositional phrase in some way further modifies/qualifies/distinguishes the article (which, in cases like these, tends to function like a relative pronoun).

The Syntactically Analyzed Greek New Testament makes this relatively easy to find. Let’s look at this portion of Eph 1.10 first to see how it is analyzed:

Here the word group contains a head term; the head term contains a word (τα) and the structure that modifies it. Here the structure is a relator. A relator is basically a prepositional phrase that functions adjectivally, modifying a substantive (instead of functioning adverbially, modifying the primary verb of the clause). So all we need to do is find where a relator modifies a word that that is an article.

There are two basic cases to consider. The first is like Eph 1.10, where the word is the root word of the head term, and the relator modifies it. The second case is where the word is a modifier itself, like in Mt 5.16:

Here note that τον is a definer, and the relator (adjectival prepositional phrase) modifies the definer.

These are the two cases to consider. A syntax search that looks like the following should account for both of them:

You’ll notice I’ve used an unordered group to contain the word+modifier portion of the query. Why did I do this? Because I really want to find where a word and a modifier are siblings (occur at the same ‘level’ in the annotation) because this implies they are in relationship with each other. The containing structure(s) (here the head term or modifier at the root of the query) constrain the elements to already being in the same unit. The unordered group allows for this, letting you specify the elements you care about (here a word and a modifier), and it will run the permutations, including optional elements occurring between them, while it searches. It makes query specification a whole lot easier.

When the search is run, 298 occurrences are located. Here’s a snapshot of the results dialog:

The different colors in the results come in because of the "OR" in the query. In this way you can tell that some results come from one half of the "OR". Here the greenish color represents the top half of the "OR" (word is a direct child of head term); the brown represents the bottom half (word is a direct child of modifier).

So, to answer the question posed on the Greek newsgroup; I’d respond that according to the Syntactically Analyzed Greek New Testament, there are 298 instances of the definite article occurring without a noun introducing a prepositional phrase.


  1. It might be possible that the other 200 or so occurances are found in the LXX, but we wont know until we get some kind of Syntactically Analyzed Greek Old Testament.

  2. I understand the search request a bit differently than you, Rick. I think the person is looking for articular prepositional phrases functioning as substantives only and not as adjectives. I also think there is an easier way to do the search, but I’m still trying to figure this stuff out, so I’m not sure. Since I wanted to include pics of my search dialog, and I couldn’t figure out how to include them here, I posted a followup on my blog. I’ll check back here on your blog for comments/corrections.

  3. Rick,
    Wouldn’t it be nice to be able to take the proximity search and have Libronix do a comparison and provide a listing of differences between the two search results? We can compare reference lists, but not verse lists. We can import searches into verse lists but not reference lists.
    Perhaps a comparison could be implemented as it would prove very helpful in situations such as this.

  4. The plot thickens! Mike over at ?? ????? identified himself as the source of the 484 number. It looks like you understood him correctly, Rick, in that he was looking for either substantive or adjectival instances. I played around with something different search terms, and I come up with either 295 or 306 hits, but I am pretty sure both of those are wrong. Partly it is just an issue of handling duplicates or multiple hits within a single verse. E.g., your first hit, Matt 2:16 actually has two hits: “the ones IN Bethlehem and IN all it regions…”
    So, the 484 number looks high to me, and I think we can confidently say that there are somewhere around 300 instances!

  5. Actually, the “someone” was me, and interestingly, I got my results (484) from my own syntax search. I’ve revised the count and now have 526 and I’ve written a post explaining why – something that I consider a problem with Opentext’s tagging.
    The Greek Article & Preposition

  6. Dave Hooton says

    This has turned into quite a marathon looking at the blogs of Mike and mgvh …So here are my comments!
    A. The unordered group produces spurious matches:
    1. 1x extra hit in Acts 6:9 where
    art rl2
    is correct (rl = relator with prep), but the spurious rl1 -> art is an error of proximity and order.
    2. 2x extra hits in Rom 4:16 where
    art1 rl1 +
    art2 rl2
    are correct, but the spurious art1-> rl2 + rl1 -> art2 are also errors of proximity and order!
    B. There are 2x hits where the relator does not contain a preposition (Act 11:12, Rev 11:2).
    As there are relators that only contain a preposition, I can’t assume that a Specifier is necessary for an alternate query!?
    I’m tempted to simply accept Mike’s modified query with 526 results! But a Greek Morph Search saves a lot of time …
    NB. There is a total of 10 relators that do not contain a prep and there are 2 relators that contain only a prep! Are these tagging errors?

  7. By Gideon Goldenberg on September 11, 2009 1:22 PM
    The question has been discussed in standard grammars, and it will be good practice to consult the available literature before anything else. To search GNT e.g. in “Accordance”, you may search the sequence Article + Preposition. The question, however, is more general, and should also refer to Article + any nominal in genitive or another oblique case. Matt. 22:21-Lk 20:25 has “ta Kaisaros” Art in the accusative + N in the genitive, also “ta tou theou” Art accus + [def-N] genitive. A prepositional phrase in syntactically an exponent of oblique case, Matt. 6:9 being the classical locus. One approach regarded such syntagms as implying an ellipsis of the N to which the Art refers; otherwise such Art was considered a “substantivizer” of the following attribute-expression; others thought of the Art as retaining its original demonstrative force thus embodying the head noun. A closer idea may regard such Article as implying a zero-marked substantival head rather than representing it. If we go through the existing rich literature about Ancient Greek we find scattered there most of the ideas recently suggested.
    Needless to say, all these constructions are common Greek and show nothing special to Koiné or to Biblical Greek.