Let's see how applications will work in this one device.
Calendar/Alarm/Scheduling/Travel - I suspect that scheduling will be voice driven where you schedule events by voice rather than by keyboard. This interaction makes the device feel like it is your personal assistant. The feedback from the application is what will drive the experience of a personal assistant. Apple is kicking ass with Siri in this area.
Maps - Voice and touch are big in this area. Just imagine the idea of telling your device where you want to go and using touch to explore an area. Imagine asking your device to show me "where are all the restaurants in the current area" or "where are all the tech companies who are currently hiring" that's just a another small part of the personal assistance market. Siri has this in its arsenal.
Note-Taking/Math/Graphics - This comes from my own experience. I love the way I can scroll with my finger and write with my pen. The experience from these two input methods is amazing in that I no longer need a notebook for keep my notes. This is especially useful for students, doctors, scientists, and whoever else like taking notes. Microsoft Onenote is a great piece of software for this task.
Documents - The keyboard plays an integral role in letting people express their thoughts, but seeing as how this is already been done, I can't see how this can be extended any further. I can't say if voice will be big driver in this area since a lot of document writing takes place while being quiet. Collaboration seems to be the wave of the future for this particular aspect of computing. Maybe some type of brain wave recognizer will take the place of a keyboard in this area. Microsoft Word and Google Docs do very well in this area.
Text Messaging/Email - I think voice will be the biggest input method in this one device because it doesn't take much thought to really respond. Obviously, this is just like documents, but it is a much shorter stint and it doesn't take much to write an email or text message. A cool thing would be for the device to summarize what you just said into a text-message since texts are limited. Yet again, Siri is on the ball game with this.
Translation - Voice is a big driver in this area and rightfully so. The idea is to allow people to continue speaking their language, but to break down language barriers by providing a 'middle man' to facilitate the same interactions regardless of language. Microsoft Research has been doing some great things in this area.
Other areas to think about are Identity and Payments.
What will it look like?
I would say a Microsoft Surface is ideally what it should be structured like. All the inputs are already put onto this device.
It may as well be a stretchable phone with a hologram interface and a projector.