An article in today's Nature details the ear region of a 370-million-year-old fish called Panderichthys - and I'm an author! If you're not familiar with this fish, it is (for now) the most tetrapod-like lobe-finned fish known from complete body remains. It fits in between classic sarcopterygians like Eusthenopteron and the earliest known tetrapods such as Acanthostega and Ichthyostega.
What we have shown here is that the dramatic modification of the skull in the ear region began while the ancestors of modern tetrapods were still water-borne fishes. That is, they did not have all the requisite adaptations for life on the land. Consequently, we proposed that these modifications were retlated to some type of ventilation function (either passing air or water or both).
You'd probably think I'd say more about this, but you have no idea what a fill I've had of writing about this. Writing paper's for Nature is no easy task. They have to be short and to-the-point. Most importantly, you have to pack all the information you would put into a considerably longer paper into about 6 double-spaced pages - max! I re-wrote the manuscript nearly a dozen times to get it down to something snappy but still clear.
To that end, Nature has written a plain-language summary that should be accessible to all. The media likes to emphasize our functional model, which is by far the most speculative part of the whole paper. The main issue is that we can outline what sorts of changes are happening in the skull that 'set the stage' for later evolution that would eventually lead to a middle ear.