I stumbled upon a comment on Omer Preminger’s post on linguistic atoms:
I really don’t understand the methodological criterion of “there [is/isn’t] observable evidence for their existence.” This is theory construction. The evidence for their existence is the success of the theory that they support. About a century passed from the hypothesizing of electrons until electrons were first “observed” (depending on how you define observing an electron, exactly). I don’t think those are grounds on which to have been skeptical of electrons, and I don’t think those are grounds on which to be skeptical of syntactic atoms of the kind being discussed here.
Compare that to a quote from Pullum’s review of Nostalgic views from Building 20 that I blogged about before:
It does suffer, though, from two familiar maladies of linguists. The first is an epistemological affliction: difficulty in telling the difference between (i) data that are well accounted for if construct C is posited and (ii) evidence that C actually exists.
Kenstowicz’s useful review of reasons for favoring some constituent structure assignments over others when doing metrical analyses should not be confused with a demonstration of the actual existence of metrical constituents.
I’m wondering (without coming to any valuable conclusion) whether you can compare that easily theoretical constructs in fundamental physics and in linguistics. The objects that linguistics deals with are vastly more complex than those of fundamental physics. On the other hand, we also don’t want to dial up our physics envy and claim that as long as we can’t express something in the language of quantum field theory we haven’t understood anything.
But quantizing electric charge is one thing, explaining aspects of differential object marking with a functional sequence NameP > DefP > SpecP present in every nominal (read about it here) is another.