Wayne Leman of the Better Bibles Blog has a recent post titled Paragraphing Ephesians 5 where he discusses some of the peculiarities of paragraph breaks.
Yep. Paragraph breaks. And paragraph breaks (and sentence breaks, and punctuation) are all important and sometimes overlooked elements of the original language editions of the New Testament. For example, the NA27 Greek text and the UBS4 Greek text differ in their paragraphs in Ephesians 5. They both have paragraph breaks between vv. 20 and 21. But UBS4 has vv. 21-33 as a single paragraph while NA27 has vv. 21-24 as a paragraph and 25-33 as a pargraph (... with a sub-paragraph break before v. 29!). Different folks think different things are going on here.
Another helpful source to consult is the OpenText.org clausal analysis. OpenText.org have analyzed the Greek New Testament for its syntactic structure. (Small plug: Logos Bible Software have developed and will release a version of the OpenText.org analysis; see the Logos Bible Software blog for some more details). The text they analyzed, as I understand it, had no punctuation -- so their clause breaks (and secondary clauses, and embedded clauses) reflect their own judgment as to the underlying syntactic structure of the text. They plan to use their clausal analysis as a basis for a higher-level analysis of paragraph structure, so their clausal analysis could be helpful for us to consult in this case as well.
OpenText.org have Eph 5.18b-21 as a single primary clause with a series of embedded clauses. Like below, where each paragraph represents a clause. The first clause (v. 18a) has what OpenText.org calls a secondary clause, it is indented. The second clause (vv. 18b-21) has a series of five embedded clauses (these would traditionally be labeled "participial clauses", likely), they are indented as well. I've forced wrapping on some of the embedded clauses; there are only five.
καὶ μὴ μεθύσκεσθε οἴνῳ,
ἐν ᾧ ἐστιν ἀσωτία,
ἀλλὰ πληροῦσθε ἐν πνεύματι,
λαλοῦντες ἑαυτοῖς ἐν ψαλμοῖς καὶ ὕμνοις
καὶ ᾠδαῖς πνευματικαῖς
καὶ ψάλλοντες τῇ καρδίᾳ ὑμῶν τῷ κυρίῳ
εὐχαριστοῦντες πάντοτε ὑπὲρ πάντων
ἐν ὀνόματι τοῦ κυρίου ἡμῶν Ἰησοῦ Χριστοῦ
τῷ θεῷ καὶ πατρί
ὑποτασσόμενοι ἀλλήλοις ἐν φόβῳ Χριστοῦ
You can see the OpenText.org representation online (click New Testament in the Texts sidebar menu, then click Ephesians, then check out the clause view of chapter 5).
This analysis is helpful for me because I can start to piece things together and think about what the text is really communicating. The second clause is talking about "being filled in the spirit". It then lists five ways that we can focus on to do this. The participle clauses "attach" to the primary verb (which, thanks to the OpenText.org analysis, we know is the predicator/verb πληροῦσθε). So we know that it is through speaking psalms, hymns and spiritual songs; through singing; through making music in our heart to the Lord; through always offering thanks for all things; and through submitting to one another out of reverence/fear of Christ.
OpenText.org list their next clause as vv. 22-23:
αἱ γυναῖκες τοῖς ἰδίοις ἀνδράσιν
ὡς τῷ κυρίῳ
ὅτι ἀνήρ ἐστιν κεφαλὴ τῆς γυναικὸς
ὡς καὶ ὁ Χριστὸς κεφαλὴ τῆς ἐκκλησίας
αὐτὸς σωτὴρ τοῦ σώματος
As Wayne notes in his blog post, v. 22 (in NA27 and UBS4) has an implied verb, likely ὑποτάσσω as some form of the word is in several MSS (Sinaiticus and several minuscules; Byz texts too). This would likely function as the predicator (primary verb). There are two immediately secondary clauses; an adverbial clause and a subordinate clause. The subordinate clause has some other stuff going on. The adverbial clause is a contrast, it helps explain the primary portion of the clause ("Wives, to your own husbands; as to the Lord ... ") with the subordinate clause providing more information ("... because the husband is head of the wife as Christ is head of the church; he [Christ] is Savior of the body").
As to Wayne's original post about whether there is a paragraph break between v. 21 and v. 22; I'm not too sure. He is, of course, correct that there is cohesion between the two clauses especially shown with the use of the assumed verb in v. 22 (NA/UBS editions). Either way, there is a consistent vibe working through the last half of Ephesians 5.
All of this (Wayne's post particularly, and my few thoughts above) underscores the importance of thinking about higher-level discourse issues.
Too often (and I'm one very guilty of this) we get stuck in the words, doing searches and looking at concordances to see how particular words are used; we forget that the primary thing that affects word meaning is immediate context. Words dissociated from context are difficult to deal with. But in context, we can work to quantify meaning. One method of evaluating context is to step up from the word level to the clause level and the paragraph level. Trace the flow of the larger section (paragraph, pericope, whatever). Evaluate where different editions have sentence breaks and paragraph breaks. Look at syntax; see where the phrase or clause appears in the larger discourse. What is the primary verb, and where does the currently evaluated text lie in relation to that verb? What is the subject (who is doing the acting)? What is being acted upon (or, what is/are the objects)? What other sorts of things modify the verb?
Starting with words is fine; but we also need to be diligent to examine how words relate to each other to form larger units (OpenText.org would call these "word groups"); and then how those units form into clauses; and how those units form into paragraphs. Then we've worked the text from a variety of angles and we can know the big picture (for vv 18b-21, it could mean being filled with the spirit and evidencing that in particular ways) and the smaller picture (practically, what "making music to the Lord" could mean). But all sorts of stuff — not just word meaning — affects how a verse or section is interpreted and then applied in the life of a Christian.
Update (2006-03-18): Note that the break between Eph 5.21 and 22 has been a subject of discussion on the OpenText.org forum.