Word orders in comparative constructions

I've been reading John A. Hawkins' _Word Order Universals_,
and have come across his discussion of adjective-marker-standard
word orders.   It reminded me of something I thought of when
I read _Describing Morphosyntax_ and the relevant passage
in Payne (who uses the more general term "quality-marker-standard",
perhaps to more suitably apply it to languages with stative verbs
or quality-nouns instead of adjectives per se): it may be the
reason I haven't mastered gjâ-zym-byn's comparative construction after
all this time, while becoming relatively fluent with most aspects
of its grammar, is that it violates a universal.

Payne and Hawkins give only examples of languages with
the order quality-marker-standard and standard-marker-quality.
Neither mentions any languages with any of the other four
theoretically possible permutations, e.g. QSM, MQS, MSQ
or SQM.  Joseph Greenberg in his article on universals
says conservatively that they are the two common orders,
and neither Payne nor Hawkins asserts the nonexistence
of languages with one of the other four orders, they just
fail to mention any.

<http://books.google.com/books?id=_B2BVl2JpT0C&pg=PA53&lpg=PA53&dq=adjective+marker+standard&source=web&ots=bfd1_p2nug&sig=YMp1h5rzEJo1jGS4RTDqRI78V_8&hl=en&sa=X&oi=book_result&resnum=1&ct=result>

Anyway, gzb does not use one of the four alternate orders,
but it does do something else a bit odd, which I haven't
seen in the languages Payne gives examples from or in
other languages I've studied: it puts the comparee between
the quality (usually surfacing as a stative verb) and the
standard.  The other languages where I know something
about the comparative constructions, or have seen
examples in Payne, put the standard and its marker
right next to the quality-word, either to the left
or to the right.

gzb comparatives have until now worked like this:

žu-sô-sra-van tam-ram θe karl-ram.
care-tending-COMP-V.STATE Tom-NAME than Karl-NAME
Tom is more careful than Karl.

That is, "Tom than Karl" is the topic of the sentence;
{θe} is an intraphrasal conjunction like {pe} (and).
The sentence has the same structure as

žu-sô-van tam-ram pe karl-ram.
care-tending-V.STATE Tom-NAME and Karl-NAME
Tom and Karl are careful.

For some reason, this comparative structure has been hard
for me to learn to use fluently.  It's probably partly
because I have occasion to use explicit comparison
less often than many other parts of the grammar,
so I get less practice with it, but it may also be because
it's unnatural to put the standard of comparison into
the topic phrase with a conjunction like this.

In addition, according to Hawkins, gzb's using
QMS (or Q...MS) is rare among postpositional
languages, -- nonexistent among the languages of
his sample, though he has data on QMS/SMQ order
in relatively few languages and doesn't assert
that such languages are impossible.  (He does
assert that such an order can't occur in postpositional
languages where the adverb precedes the verb
(they follow the verb in gzb), on the basis of the
same data; I'm not sure why but it has to do
with his theory of why various word order
correlations exist, which I'm still trying to understand.)

It's recently occurred to me that I might drop {θe}
and mark the standard of comparison with a derived
postposition: probably {dîfu-i}, from the process root
{dîfu} "to compare, look for differences between"
(from Unix 'diff'').

karl-ram dîfu-i žu-sô-sra-van tam-ram.
Karl-NAME compare-at care-tending-COMP-V.STATE Tom-NAME

Here the standard (karl-ram), marker (dîfu-i) and quality
(žu-sô-sra-van) are all in a row with nothing interrupting
them.  I like this and will probably stick with it, but it will
be a while before I find out if it's really easier to learn
than the {θe}-construction.

Questions:

1. Does anyone know of any natlangs or conlangs with
a word order other than QMS or SMQ?

2. Does anyone know of a language where other
sentence constituents can or routinely do come between
the elements of the Q/M/S sequence?

3. How does your conlang form explicit comparatives?

If these are universals, even statistical universals, of
human language, they might be good things to violate
in an exolang.

--
Jim Henry
http://www.pobox.com/~jimhenry/

Replies