Path: utzoo!mnetor!uunet!husc6!hao!gatech!ufcsv!codas!killer!fmsrl7!eecae!crlt!michael
From: michael@crlt.UUCP (Michael McClary)
Newsgroups: comp.ai
Subject: Re: words order in English and Japanese
Message-ID: <1427@crlt.UUCP>
Date: 20 Jan 88 14:21:21 GMT
References: <1671@russell.STANFORD.EDU> <2617@calmasd.GE.COM> <1729@russell.STANFORD.EDU>
Organization: CRLT , Ann Arbor, MI
Lines: 56
Summary: Words order - how about a null hypothesis?

In article <1729@russell.STANFORD.EDU>, nakashim@russell.STANFORD.EDU (Hideyuki Nakashima) writes:
> I've got many (negative :-) responces to my postings.  Thank you. []
> 
> Words ordering in a syntax is not chosen at random.  There must be
> some explanation to it.  If you know a better one, please let me know.

Here comes another one.  (Am I delaying enough to be polite?  B-) )

Why must the choice of words ordering in the syntax rule set for a natural
language be non-random?  We see from the currently-used languages that
several orderings work about equally well.  Seems to me the null hypothesis
fits very well.

Any language at all confers a significant advantage over none.  Once a
language is begun, any >change< in something as basic as words ordering
rules increases confusion, reducing the utility of the language until
many people have learned the revised rules.  A change would have to confer
a very strong advantage to be a net gain during the transition, and
alternative words ordering rules do not do so.  Thus a randomly-selected
rule that works will be very strongly conserved.

(We see a a similar phenomenon with our keyboards.  Several layouts give
significant speed and accuracy improvements over QWERTY.  {Indeed, one
rumor claims QWERTY was developed specifically to >slow down< typists,
so early mechanical machines wouldn't jam.}  New machines can handle the
speed and don't jam, but the changeover to one of the improved layouts
would cost so much, in capital and confusion, that it doesn't happen.)

I would expect changes in words ordering to occur only in situations where
communications >barriers< were an advantage.  The only examples I can think
of are "oppressed classes":  Slaves, inhabitants of conquered provinces,
alternate-lifestyle subcultures, minority religions, secret societies,
traveling entertainers, teenagers.  All have business to conduct that must
be concealed from power-wielding members of the majority culture, and all
find or develop ways to hide their communication.

Thus I expect words ordering rule differences will prove to be a fossil
record of either separate development of languages from scratch (extremely
unlikely) or the resolution of political conflicts, not a parameter that
was tuned to maximize information transfer.

(And speaking of "oppressed classes" and words ordering rules: wasn't
there some work done recently {with descendants of about three groups
of slaves, in different parts of the world, with strongly differing
languages [both owners' and ancestors']} that indicated a "natural"
set of words ordering rules that occur in languages developed by
language-education-deprived people?)

===========================================================================
  "I've got code in my node."	| UUCP:  umix.cc.umich.edu!node!michael
				| AUDIO: (313) 973-8787
	Michael McClary		| SNAIL: 2091 Chalmers, Ann Arbor MI 48104
---------------------------------------------------------------------------
Above opinions are the official position of McClary Associates.  Customers
may have opinions of their own, which are given all the attention paid for.
===========================================================================