An ambitiously titled, “From Monkey Alarm Calls to Human Language: How Simulations Can Fill the Gap,” by Jason Noble, J.P. de Ruiter, and Kate Arnold (available here) offers a good example of what these papers can achieve. Mainly it is richer in nutritious food for thought. Particularly intriguing is the discussion of the evolution of animal communication.
Dawkins and Krebs (1978) suggested that animal communications... happens when one animal, the actor, has been selected to produce a response in a second animal, the reactor, such that the reactor’s behavior (on average) changes to the advantage of the actor. Dawkins and Krebs acknowledge that such a definition moves a long way from our everyday understanding of the word “communication” and that there is no implication, in their view, that animals should be transmitting information to each other in order to qualify as communicating. They admit that they are “tempted to abandon the word communication altogether.” Dawkins and Krebs argued that animal communication serves not to inform but to persuade, and that advertising and propaganda are more apt metaphors than language for what goes on in the animal kingdom. 
Readers on this blog have sometimes wondered why I am so doubtful about the word communication. There is a nice summary of my concern. Animal communication and human communication serve two separate functions, obscured by a single word. The authors “suggest the general term ‘influence interaction’ to refer to an event where one animal acts in such a way as to influence the perceived states of the world”  and they provide a most valuable diagram of how an actor (Agent S) can influence a reactor (Agent R).
The diagram shows four ways to influence another:
- Accidental influence: An action by S triggers an irrelevant response by R. For example, a pig rooting for truffles makes noises that cause a mole to flee. The pig is not a predator, so the mole has no need to flee. The pig is not interested in the mole, so it reaps no benefit from its action.
- Exploitation: An action by S triggers a response that is useful to R but not to S. In this case, the reactor is taking advantage of something about S. For example, a cheetah stalks an antelope, but the antelope catches its scent and runs away.
- Manipulation: An action by S triggers a response useful to S but not to R. For example, the lure on an angler fish attracts a small fish which is eaten. The lure works properly for S but R’s response to it is, to say the least, counterproductive.
- Proper signaling: An action by S triggers a [mutually] beneficial response in R. For example, a bee dance sends other bees to a place for foraging. S and R both benefit from the increased food. In a sense, proper signaling is both exploitative (e.g., the reactive bees respond usefully to S’s behavior) and manipulative (e.g., S’s behavior triggers useful behavior in R).
Noticeably absent from the diagram is the speech triangle in which speakers switch roles as S and R while discussing a neutral topic. Also a bit of a puzzle is something like an alarm signal in which S (who already senses an approaching threat) makes a call that alerts others to the danger. It is useful to R, but what does S get out of it? Many observers say the group as a whole benefits from the activity, but the authors seem to believe the selfish gene is the whole story and they make no room for multi-level selection.
Having presented a typology of animal communications (or influences) they then propose a three step process for the evolution of “proper signaling.”
- Presence of accidental influence: The authors state as an absolute law that “all signaling systems must begin as cases of accidental influence.” The influence can be reflexive or learned, but it is a chance phenomenon available for evolutionary adaptation.
- Development toward manipulation or exploitation: In exploitation the R agents turn an accidental triggering event to their advantage. In manipulation the S agents turn an accidental reaction to their advantage.
- Emergence of proper signaling: The manipulative or exploitative influence of step 2 comes to be in the interests of both parties.
The authors conclude this section of their paper by saying:
The task for simulation modelers interested in a behavior such as monkey alarm calls or human language is to demonstrate an evolutionary trajectory along these lines. 
There is always a certain boldness to a priori statements like this one that take my breath away, and since their typology does not include the speech triangle I’m not inclined to take them entirely seriously. Nonetheless, they provide an excellent, quick survey of the theory of evolving animal communications and their diagram in particular is worth keeping in mind.