Yep, the soma-soma synapse in that paper appears to be a chemical synapse between two cell bodies. There are no axons or dendrites in that preparation so the synapse must be between the somas. They are using two snail neurons that are far from each other (in different ganglia) in vivo, but which are known to synapse. This in vivo synapse is long-range and so must be mediated by an axon. In the dish, they can encourage the synapse to form even without encouraging axons and dendrites to grow.
Note that this is an artificial in vitro preparation and they are, in some sense, forcing a synapse to occur in an unusual way in a cell culture dish. They do this because it makes it easy to study certain aspects of the signalling mechanisms for generating synapses. In your paper, they appear to be using it as an easy preparation for a proof of principle for the interface between a biological neuron and a non-biological semiconductor substrate.
The paper you cited references this paper that outlines the soma-soma preparation with a pretty picture to make things clear: http://www.jneurosci.org/content/19/21/9306