Skip to main content

Because it is so common, the particle δή often escapes notice. Careful attention to its distribution, however, reveals that it is not just randomly sprinkled into texts to add a vaguely defined ‘emphasis’, but rather plays a key role in articulating transitions between units of narrative. Consider, for instance, Herodotus’ Thermopylae narrative, which opens thus: οὗτοι μὲν δὴ οὕτω διενένωντο ποιήσειν· οἱ δὲ ἐν Θερμοπύλῃσι Ἕλληνες… (7.207.1). The historian has just explained the Spartans’ reasoning in sending so few men to Thermopylae, so it is hardly unusual that he refers to them as οὗτοι here (with the anaphoric stem οὑτ- further echoed in the adverb οὕτω), then deploys μέν to set up the expectation that he will move on to another group, in turn introduced by δέ. Herodotus marks the end of the account with a similar transition: οἱ μὲν δὴ περὶ Θερμοπύλας Ἕλληνες οὕτως ἠγωνίσαντο, Ξέρξης δέ… (7.234.1): once more, a μὲν δὴ … δέ pair guides the reader from the previous subject to a new one, with the summarizing role of the first half of the sentence signaled with an anaphoric demonstrative, again οὕτως. But why does Herodotus choose to use δή in both sentences?

That there is no ready answer to this question is one sign that δή has not received the attention it deserves: in Sicking & Van Ophuijsen 1993, Sicking’s study of text articulation in Lysias relegates δή to an appendix (treating it primarily as an interactional particle), and, while Van Ophuijsen devotes nearly twelve pages to it, he aims largely to deny it any connective force, reducing it to meaning simply “clearly”; meanwhile, Wakker’s 1997 study of emphatic particles in tragedy helpfully distinguishes between the nature of emphasis conveyed by ἦ, δή, and μήν—but more from the perspective of μήν than that of δή. This is, of course, completely reasonable: insofar as μήν acts as the speaker’s personal guarantee of the truthfulness of a statement, it is a more robust particle, more restricted in use, than the blander, more multi-purpose δή, which is said either to indicate that the statement is clearly true (Sicking & Van Ophuijsen) or to call the addressee’s attention to the statement (Wakker).

These formulations, however, do not go far enough in explaining some of the distributional peculiarities of δή, as seen in microcosm in the Thermopylae narrative. First is its overwhelming propensity to occur with other particles. Of the 19 examples of δή in this episode, it is found eight times after μέν (42%), five times after τε (26%), and once each after δέ and καί (together, another 10%). Second is its marked fondness for οὑτ-: nine times in 7.207–33 (47%). These two contexts together account for all of the examples of δή in this passage.

That δή should collocate so readily with other particles is one indication that, whatever else it does, its role in articulating the text should not be dismissed lightly (cf. Revuelta Puigdollers 2009: 103). Such a function is further suggested by its tendency to occur with οὗτος. While δή is usually described as showing a preference for demonstratives, what is particularly remarkable is that it is not found a single time in Herodotus with either ἐκεῖνος or ὅδε. In other words, δή does not have a preference for demonstratives; it has a preference for οὗτος in particular. Since this is the anaphoric pronoun par excellence, and since μέν, the particle δή most often combines with, also frequently signals that the preceding material is coming to a close, this distribution lends support to Sicking & Van Ophuijsen’s view that δή marks statements that are already established as true. But since οὗτος is also strongly allied with second-person deixis, we can also see it as bringing this to the reader’s attention, in line with Wakker’s position—a position further bolstered by Herodotus’ disproportionate use of δή with other devices for establishing discourse cohesion.