Is it true that has a rule, a language where subject and object aren't explicately marked will always have SVO or OVS order? I've been thinking that it may be possible to get away with having no marking for subject/direct object in a verb-final language. Well, one that relies heavily on coverbs anyway.
Just to demonstrate my point, I was thinking of a 'primitive' conlang that had a lot of unusual features. One of them being complete lack of any sort of abstract 'grammatical' particles. Every verb in the language is a coverb that can only take ONE argument (typically a direct object). A 'subject' may stand alone at the start of a sentence, though some intransitives can take an agent as its only object (the language has ergative alignment).
Of course, my conlang isn't meant to be spoken by modern people. Its my own speculation on what an early proto-language may have looked like. I figured a language like this would be the next stage after an 'improvised' grammar that relied solely on an agentivity heirarchy (like a direct-inverse language) to differentiate subject from object. Simply a single hominid may realize that it could guess which role an object performed in an event based purely on what it is. It wouldn't be 100% reliable of course, but it would serve some benefit that individual to be able to decipher everyone else's babbling with some amount of certainty.
Anyway, I diverged a bit. But I was wondering if there was any known language that demonstrated zero-marking with either a verb final or verb initial order?