An effect that signals syntactic involvement in sluicing (They knew someone was approaching, but they didn’t know who), and an underlying full structure, is a choice to omit prepositions in languages that allow preposition-stranding under wh-movement, as in (1).

(1)	a.	Peter was talking with someone, but I don’t know (with) who.
	b.	Who was he talking with?

The choice is not available in non-preposition-stranding languages. This observation, due to Merchant (2004, 2006, 2007), has not yet been falsified, though various alternatives have been posited that argue for no more structure than is visible in sluicing (cf. Ginzburg and Sag 2000, Culicover and Jackendoff 2005, Stainton 2006).

Based on diachronic and synchronic data, I argue that Merchant’s claim has little support. Notice that Old English (OE) and Middle English (ME) provide a good testing ground because wh-interrogatives cannot strand prepositions until the 13c (cf. Grimshaw 1975, Allen 1980, Van Kemenade 1987, Fischer 1992, Bergh and Seppaenen 2000). This means that a difference in the use of prepositions could accompany the transition. My OE and ME data shows no such difference, however; what it shows is that ME prepositions are consistently retained where they could be omitted. These results and a closer look at non-P-stranding languages suggest that: (1) omission of prepositions in sluicing is not sensitive to the availability of P-stranding, especially when both merging and sprouting are brought into focus; (2) the final form of a sluice may be, cross-linguistically, a function of how easy it is to integrate it into the proposition expressed by its antecedent.