Sorry, I haven’t been following properly.
Ναυσικάαν ἐύπεπλον· ἄφαρ δ ̓ ἀπεθαύμασ ̓ ὄνειρον,
First off, to rehearse what you all probably recognize, the verse has a 3rd-foot “feminine” caesura, just like its immediate neighbors (and also 51 and 52, as it happens). In Homer this is a bit more common than the 3rd-foot “masculine” caesura following the long, and more to the point, it appears to be the more traditional form. Note that the syntax also breaks at this point.
I don’t know if I’ve ever read the cited article of Witte’s (and I don’t currently have access to JSTOR), but “Hermann’s bridge” refers to the late-18th-cent. observation that if the second half (the “biceps” in Maas’s terminology) of the 4th foot is disyllabic (consisting of two shorts), there is very rarely word division between the two shorts. ἄφαρ δ ̓ ἀπεθαύμασ ̓ ὄνειρον| is in conformity with this rule, as uncompounded *ἐθαύμασ ̓ ὄνειρον| would not be.
*θαύμασ ̓ ὄνειρον| (uncompounded and unaugmented), for its part, would be bucolic diaeresis (word-break before 5th foot, resulting in the closing cadence found in other verse forms too). This is a common pattern in Homer (here in 48, 51, 53, …), so while ἀπεθαύμασ ̓ ὄνειρον| does “prevent” bucolic diaeresis, it does so only incidentally. That can’t be its purpose or motivate the use of the compound.
Witte was a pathbreaking scholar in Hermann’s wake who significantly anticipated Milman Parry in his recognition of the part played by formulae in Homeric composition. Aetos explains his argument, which in itself is very plausible. But to my mind Witte’s point as applied to ἀπεθαύμασ ̓ ὄνειρον| would be more cogent if θαύμασ(ε) were attested in formulaic phrases and/or at bucolic diaeresis. Which it’s not. That doesn’t invalidate Witte’s position (no significant semantic difference) but it does weaken it with respect to this particular verse.
Stanford for his part was a good reader of Homer, but I’m not much attracted by his interpretation of the compound here (a Homeric hapax, as Sean points out). It seems quite tenable in principle, despite Hainsworth, but over-ingenious, and it’s not how the word was understood in antiquity.
Sorry not to be more helpful.