My article on counterfactuality and past is out now, and it’s free to access. In 2011, while I was working on the Daakaka grammar, I looked at the system of TAM markers and asked myself how hard it would be to figure out their meanings. Manfred Krifka suggested I start working with branching time to account for the realis/irrealis distinction, which I did. Except that the distinction we find in the Daakaka system isn’t binary, but tripartite.
Which was just as well because, as I then realised, this type of tripartite distinction is exactly what the branching-time framework affords. It still took me a while to figure out the main parameters of the Daakaka system: The mapping isn’t exactly one-to-one and I’m still learning new things about the TAM markers as I continue my empirical research. But the real struggle was to 1) realise that the modal semantics I was developing was fundamentally different from what people had been doing before, 2) figure out how to tell people that you could beautifully model English, German and any other language I know with the same approach.