Iliad 3.420, πάσας δὲ Τρῳὰς λάθεν· ἦρχε δὲ δαίμων. -- why perfect?

[bcrowell]

Aphrodite drags a bitterly protesting Helen to Paris's love nest.

Ὣς ἔφατ ̓, ἔδεισεν δ ̓ Ἑλένη Διὸς ἐκγεγαυῖα,
βῆ δὲ κατασχομένη ἑανῷ ἀργῆτι φαεινῷ
σιγῇ, πάσας δὲ Τρῳὰς λάθεν· ἦρχε δὲ δαίμων.

Why use the perfect ἦρχε here rather than the aorist ἄρξε?

All I can think of is that the explanation is given after the statement that they were not seen, so it's already a retrospective.

[Hylander]

ἦρχε is imperfect, not perfect. The form of the imp. 3rd sing.. happens to be the same as the perf. 3rd. sing., however. The action is viewed as in progress. "She escaped the notice of all the Trojan women. The divinity was leading the way."

[bcrowell]

ἦρχε is imperfect, not perfect. The form of the imp. 3rd sing.. happens to be the same as the perf. 3rd. sing., however. The action is viewed as in progress. "She escaped the notice of all the Trojan women. The divinity was leading the way."

Thanks, it's interesting to know that it's also the imperfect form. Not really convinced by the semantics, though.

Maybe this is just a matter of the sequence of tenses:
https://en.wikipedia.org/wiki/Sequence_of_tenses#Greek
Since λάθεν is aorist, if the lack of detection is viewed as complete at the end (phew, we made it!), then we would use the aorist. If the lack of detection is viewed as co-occurring in time (still not being detected...), then we would use the imperfect. To me it seems like the semantics could be viewed either way here, but the use of the imperfect would imply the latter. I don't know if it matters that this doesn't really seem to have been constructed as a subordinate clause but more as a separate sentence.

Whoever tagged it in the Perseus treebank tagged it as perfect, but we have no way of knowing how much thought they put into that or what their thought process was.

[Paul Derouda]

The sequence of tenses isn't really relevant here, since ἦρχε is a different sentence. Bill is right about the semantics. The Greek present/imperfect verbs and aorist verbs are often used contrastively; this is can be quite tricky, as English doesn't have anything quite similar.

The perfect is not very frequent in Homer and would be definitely strange in this kind of context. Labeling this as perfect doesn't inspire confidence.

[Hylander]

Not really convinced by the semantics, though.

I think we need to derive our characterizations of Homer's grammatical/semantic categories from Homer's actual usage, and not vice versa.

Since λάθεν is aorist, if the lack of detection is viewed as complete at the end (phew, we made it!), then we would use the aorist. If the lack of detection is viewed as co-occurring in time (still not being detected...), then we would use the imperfect.

This is a slightly misleading way to think about the aorist. From the Cambridge Grammar of Classical Greek 33.28, p. 417:

The aorist indicative is used to present the occurrence of an action in the past, without reference to its duration or porcess, but presenting the action as a single, uninterruptable whole.. As such, the aorist is the default tense in narrative texts to record single, complete actions.

It's not that the aorist necessarily emphasizes the completed nature of an action: it's just that it simply states as a fact that the action occurred. Completion is generally implied under the circumstances, but that isn't the actual function of the aorist -- it simply states that the action occurred. The imperfect is in a sense "marked" in contrast to the aorist: the imperfect looks at an action as in progress.

this doesn't really seem to have been constructed as a subordinate clause but more as a separate sentence.

But this is very true as a general rule in Homer: he doesn't use subordination so much as parataxis. Here, ἦρχε δὲ δαίμων explains πάσας δὲ Τρῳὰς λάθεν. We might subordinate ἦρχε δὲ δαίμων, and introduce it by a causal conjunction, "because", but that isn't the way the Homeric language works.

[jeidsath]

Imo, it has a narrative purpose of bringing the reader into the immediacy of action. Notice how it gets followed by the οτε next line. When Paris was getting throttled a few lines earlier the same thing happened with αγχε immediately followed by a tension-heightening "και νυ κεν and he would've skragged him but for..." (Posting on my phone from memory, so please take this as approximate.)

[Hylander]

To Joel's point, another quotation from Cambridge Grammar, sec. 33.49, p. 427:

Aorist indicatives and imperfects are the main tenses of Greek narrative; both tenses locate an action in the past, but they differ aspectually. Imperfects, by suggesting that the actions they express are incomplete (imperfective aspect), typically do not 'push a story forward': rather they are used to 'set the stage' or to create a background/framework in which main events take place which do move the story forwards. These main events, in turn, appear in the aorist indicative.

[jeidsath]

Back at the computer. Lines 447-448:

Ἦ ῥα, καὶ ἄρχε λέχοσδε κιὼν, ἅμα δ᾿ εἵπετ' ἄκοιτις·
Τὼ μὲν ἄρ᾿ ἐν τρητοῖσι κατεύνασθεν λεχέεσσιν.

Here the use is narrative, not scene-setting: A rising imperfect tension, followed by the aorist narrative climax.

A scene-setting use, as described in Hylander's quote, use comes next, in 449

Ἀτρείδης δ᾿ ἀν᾿ ὅμιλον ἐφοίτα, θηρὶ ἐοικὼς,
Εἴ που ἐσαθρήσειεν Ἀλέξανδρον θεοειδέα.

And really the scene-settings continues until 455 when you get the aorist action of Agamemnon speaking. But then we're in the present with the speech, hearing it directly. So I'd take this "main events" versus "background" to be a very debatable sort of statement. Events are not necessarily more important than scene, especially in story-telling and poetry, where description can have the weight, with events being mere incident (as here in 449ff).