Perhaps, then, rather than “pragmatically unmarked” word order (which presumes that one pragmatic structure is unmarked relative to all others), it would be more fruitful to talk about “thetic word order” and “categorical word order” (and “predicate-focus word order”, etc. etc.).  

I thought, though, that the usual understanding of “pragmatically-unmarked” is “topic-comment” (i.e. “categorical”, as opposed to thetic). And I suspect this is the most common pragmatic structure in discourse (since thetic structures are only appropriate when there’s no established topic, and the various focus structures require a specific kind of presupposition).  


Sent with Mail Pilot  
On 19 Sep 2014 5:23:48 am, Basilius <[log in to unmask]> wrote:  

Jyri makes some valid points here.

However, unlike Jyri, I don't think this terminology ("pragmatically unmarked") is very apt, at any
rate, in the context of Scotto's original question which was about "languages such as Latin or
Russian", and of the subsequent discussion of word order in terms of formulas like "SVO" or "SOV".

The term "pragmatically unmarked" implies that there's a group of parameters that belong with
"pragmatics", and if you set all of them to their default ("unmarked") values, you get *uniformly*
the same WO in terms of "S", "V", and "O". That doesn't seem to be the case with many languages;
in particular, I'm quite certain that it's not the case with Russian.

Part of the problem is about defining "pragmatics"; for example, contrary to what is often said
about Russian WO, in a sentence answering a question like _What's going on?_ you'd normally use a
WO which is *not* pragmatically least marked, in colloquial Russian at any rate (in other words,
"thetic" doesn't warrant "pragmatically unmarked"). On the other hand, construing a
natural-sounding Russian sentence may seem to involve rules like "first set the temporal frame and
the spatial frame for the described event, in this order", which seem to overlap with what is
usually called pragmatics very much, but how would you avoid applying this rule if the time when
the event happened and/or the place where it takes place has to be specified in the sentence? And,
last but not least: there are no natlangs for which "pragmatics" would have been described in
sufficient detail to make accurate empirical generalizations about the "pragmatically unmarked" WO;
the term actually refers to a theoretical construct rather than an observable phenomenon (or even
points to a fictional entity - if you don't share any of the theoretical stances that need this

The other, and bigger, problem is that syntactic constraints working in "pragmatically unmarked"
sentences don't seem to operate directly on categories like formal "subject" or "object" in
Russian. There are restrictions on splitting certain types of constituents, some constructions
must be head-first, etc., but the positions of immediate dependents of main verb are not affected
by such constraints. Thus, e. g. a subject will be placed to a specific slot in the sentence not
because it's "subject", but e. g. because it sets the temporal frame (in a sentence like 'That day
was a rainy one'), or refers to "the most active participant in the event" (in 'Pete broke the
pencil") or to "the being whose attitudes are described" ('Mom likes roses') etc., and whenever
any of these happens to be expressed by something which isn't syntactically the subject, the
syntactic subject will go to another position. At a glance, the categories directly affecting the
ordering of "S" and "O" seem to be of semantic nature in Russian (although I won't insist on that,
and the categorization itself is probably language-specific). Sure, there is a strong correlation
between semantically motivated categories like "experiencer" and syntactic categories like
"subject", but it's just a statistical correlation. The term "nonconfigurationality" seems to
imply this: WO being primarily sensitive to factors that may correlate with syntactic roles but
are of different nature.

So, if you want an accurate terminology in describing a language which is indeed
nonconfigurational, I believe you are on the safer side when you say things like "the
statistically prevalent word order is SVO". Thus, I'd suggest *statistically prevalent* instead of
"pragmatically unmarked".

On Thu, 18 Sep 2014 13:49:44 +0300, Jyri Lehtinen <mailto:[log in to unmask]> wrote:

> 2014-09-17 14:04 GMT+03:00 Zach Wellstood <mailto:[log in to unmask]