> Anyway, I think that the “tríar manach do-rat” example is indeed a
> 'nominativus pendens’ and that the ‘nominativus pendens’ introduces topics,
> as MacCoisdealbha argues in his thesis.
> Topics can introduce information that is new to the discourse, as a way of
> framing what will follow. Imagine we are at the breakfast table: “this man I
> saw yesterday, he was clearly trying to break up with his girlfriend via
> text message! What do you think about that!?!” The topic comes up out of the
> blue and is not necessarily known to the hearer.
Yes, I hadn't thought of that use of topic. Interesting - a presentation topic. I'm not sure I've read much about that one. Anyway, it seems as though there's no real way to adjudicate between the nominativus pendens construction and the reduced cleft construction for the case of 'tríar manach do-rat'. One way to at least say that we have a reduced cleft here, possibly, would be if we could find any example at all where a presentational type meaning was encoded by a full cleft at the beginning of a discourse. If we could find that, we'd at least be able to say that both clefts and nominativus pendens constructions would be allowed to fulfil this presentational function.
Showing that nominativus pendens can start a story is fine (I do like your TBF example) but we would have to be absolutely certain that clefts can't. If we can show they can't then we'd be able to differentiate between clefts and nom. pendens in the following way:
Function 1 Function 2
1) nom. pendens: presentational focus old information (topic)
2) clefts: contrastive focus ?