Came up recently in a 37signals podcast:
http://37signals.com/podcast#episode18
The straightforward, "brute-force" solution is to just use the first touch on a UI object as the hover.
It works. But are there other, more elegant solutions? For instance, could the hardware ever be made to support actual finger hovering?
It would mean a very different technology than what is currently used. I wouldn't try to force touch devices into screen device semantics. Touch devices are developing their own distinct set of user expectations. Touch to hover is perfectly acceptable. I've seen some nice examples of hinting to the user where "hoverable" areas are when the screen first loads. Highlighting hot spots in some way, then letting those highlights fade out after a second or two.
The point of hovering is to make some information available to the user about an element without triggering some primary action.
One approach might be to trigger the display of secondary information when dead space near an element is touched. That's not very discoverable though.
A better approach I think is basically the answer given above - touch and hold. But the first few time the user simply taps reveal the secondary information just before going off to the primary function, so that they know there is something to explore. It's not quite what 37 signals is saying though as I would not block the primary function, I would just reveal briefly it is there.
Instead of touch and hold you could also consider a swipe across something as seeking information, though again that's not easily discoverable unless you present an indication it is possible. I don't think there's anything wrong with the first few passes of an application presenting a few tutorial kinds of illustrations to let the user know something is possible, and many things like touch outside or swipe to reveal are very easy to remember even if they are hard to discover.
Read all answers and comments above and with all due respect to you and your solutions, I would simply suggest not to implement hover feature in touch devices. In touch devices users are not expecting something of that sort, So I would prefer not to integrate such a feature.
In mobile devices user and developers/UX engineer prefer lesser footprint as possible. So to reach a new view or to get the details user is not expecting and expected to tap twice on a button or link.
The best solution is what you have in your question - a single touch performs the hover effect. This works just fine on an iPhone and is completely intuitive. I tap a nav item, and it expands to show the secondary nav items. I tap it again, it takes me somewhere (if it is itself a link). I don't see the need to replace something that already works and makes sense to the user.
Some touchscreen technologies do support hovering. For example, with Microsoft Surface and some specialized tablet displays distinguish between hovering and pressing. It seems like it might be frustrating, though. I used ZBrush with a tablet, which allows you to literally hover, and accuracy was significantly reduced (the cursor was jumping all over the place). In a graphics program this is okay, since you get the feel of drawing (and the imperfections can make it look more natural), but for things where the user needs to be accurate, I'm not sure it would be a great option.
Don't expect devices like the iPad to support a cursor anytime soon. It's much wiser to focus on avoiding the hover and any other annoying mouse-only feature. I've recently written a detailed post on why we should stop putting anything mouse-over related to the web and what are the best ways to replace it.
If an object has the focus for a period of time with no activity, hover. It's probably an indication the user needs more information. Granted this is very difficult on single-touch action items like links and buttons (Easier to get the focus when in a keyboard environment where you can Tab to it.).
Why not simply write the name of the function on the button (unless it's tiny). That way the user doesn't have to hover over every button until they find the one they want. Having mystery buttons or links that would require hovering is known as Mystery Meat.
If you really want to have pictures instead of text, then I like the idea of one touch to 'hover' another touch to 'press' the button. However, it's not always clear that the UI supports this. Buttons look like you touch them to activate and a new user might be wary of touching a button in case they activate a function they didn't want.
An alternative would be to use little sprung switches.
This way it's clear that just touching them won't activate them. However, it doesn't let the user know that they could touch them for a tip.
