Can we say the VMS is a "free word order"?

[Emma May Smith]

On the original topic: some researchers have sought to analyse words by their relationship to one another and sort them into "classes". This was achieved by stating that classes are less likely to occur adjacently (or some other similar rule) and then sorting of words into classes which minimised class adjacency. The nature of such classes is unknown, but provide us an interesting starting point.

Languages can be thought of as head-initial or head-final. Basically, that means whether the key word in a multi-word phrase comes at the start of end of that phrase. Some languages are quite strict, others less strict. But this characteristic of a language is linked with word order, thus knowing the head directionality of a language you can make a guess (a preference) about word order.

Word classes could be used to model head directionality, as given enough text certain word class sequences should predominate. But it would require an input as to the identity of those word classes. For example, we might be able to say, "if [daiin] is a noun then the language is head-final and maybe has an SOV word order." Or, more practically, "if labels are nouns then ... and ... ."

I think this would be a very shaky approach and a null outcome wouldn't prove the negative. Though I can't see how we would otherwise try to understand word order at our current state of knowledge.