In R.M.W. Dixon's book, "Ergativity," I read that ergative-absolutive marking is generally morphological. IIRC, that goes for split-S alignment and fluid-S alignment as well.
For those who came in late, split-S alignment occurs when the subjects of some intransitive verbs are marked with absolutive case & do not stand for agents (e.g. with verbs 'die,' 'fall,' etc) and the subjects of other intransitive verbs are marked with nominative case & do stand for agents (e.g. 'jump,' 'blink,' etc.). Fluid-S alignment occurs when the subject of an intransitive verb is marked with nominative case if it stands for an agent and absolutive if it doesn't. So one verb could mean 'drop to the ground' or 'fall' depending on the case of the subject.
My question is, do isolating natural languages ever exhibit non-nominative/accusative morpho-syntactic alignment. e.g. Could an isolating language have semantic marking (e.g. arguments occur in order of prototypical agency)?