This is my first curation since agreeing to curate. I am soooo excited! The link is to a piece that Omer has on syntactic atoms. I won't be giving much away if I say that he thinks that it is not entirely clear what these are, though whatever it is, it is not what most people take them to be. I won't say what his argument is, because you should read it. But I will say that the main point he makes has been, in part, made by others.
Chomsky has long argued that whatever "words" are they do not have the referential properties that semanticists take them to have. And in this they contrast with animal calls, which, Chomsky points out, fit the referential/denotational paradigm quite tightly. See (here) for some discussion and references.
Similarly, more recently, Paul Pietroski has argued that syntactic atoms do not denote concepts or extensions but something more abstract; something akin to instructions for fetching concepts. As he points out the key desideratum for lexical meanings is that they compose (here Paul follows Fodor who made a very good career pointing out that most of the things that philosophers proposed as vehicles for meaning failed to compose even a little bit). If this is so, then the idea that linguistic atoms are concepts cannot be correct and the question of how our syntactic atoms emerged to mediate our way to concepts becomes an interesting question. Combine this with Chomsky's observations and one has a real research question to hand.
Omer presents another take on this general view; the idea that our standard conceptions of syntactic atoms are scientifically problematic. In fact, given that we have learned something about syntax and the basic operations it might involve over the last 60 years, just what to make of the atoms (of which we have, IMO, learned little) might be even more urgent.
Here is the link to Omer's piece.