GG and change

[Robert Whiting]

----------------------------Original message----------------------------
Change is inevitable.  This is a by-product of a universe with
increasing entropy over time.  Therefore change is a "universal
universal" and has to do with the nature of the universe, not
with the nature of language as such.
 
It is my impression that almost all linguistic change is brought
about by sociological factors.  This impression is based on the
following observations:
 
1) Linguistic change is unpredictable.  By this I mean that the
   onset of a particular change at any given time can not be
   anticipated.  Once a change has started or been completed, it
   can be analyzed and classified and (probably) parallels for
   such a change can be found in other languages.  But the
   trigger for the change can not be shown to be the linguistic
   situation because one can also (probably) find parallel
   linguistic situations in other languages in which the change
   did not occur.  Therefore the trigger for the change must come
   from outside the language (or outside language in general).
   Analogical changes to restore morphological features levelled
   by phonological change might appear to be an exception (as
   might analogical change in general, since a model form must
   already exist in the language for analogy to be productive),
   but in this case it is simply a reaction to a change
   (phonological) that did not have its origin in the linguistic
   situation (on this seeming contradiction, see below, point 5).
 
2) Any part of language can change.  Changes can be observed in
   phonology, morphology, syntax, and lexicon.  I think (someone
   will correct me if I am wrong) that there is no part of a
   language that can be considered immutable.  Even features
   that were long considered inherent to a language, like word
   order or intonation, can and do change.
 
3) Linguistic change is irregular.  With the exception of sound
   changes (which the neogrammarians tell us are without
   exception and once a sound change is initiated, it will affect
   every instance of that sound in the language [but the
   unpredictability of point 1 still applies:  it cannot be
   predicted what will change or what it will change into]), a
   change in a form or construction may or may not affect similar
   forms or constructions.  Or part of a system may change and
   leave the rest of the system unchanged.
 
4) Linguistic change is not unidirectional.  A change (including
   phonetic changes) that goes in one direction in one language
   may go in the opposite direction in another language.  One can
   count up the number of instances for the change in each
   direction and say which direction is statistically more likely
   for the change, but in essence, there is no change that is
   impossible.
 
5) Linguistic change can cause conflict in the language.  This is
   a result of an inherent conflict in language between phonology
   and morphology.  Since language expresses meaning through
   phonological form, there is constantly a conflict between
   phonological simplicity (ease and speed of articulation) and
   morphological complexity (more overt morphological marking to
   disambiguate meaning).  What we have, then, is an "engineering
   trade-off" where changes for the better on one side will
   usually introduce changes for the worse on the other.  Thus a
   phonological change that reduces overt morphological
   distinctions will frequently be countered by an analogical
   change that restores some (if not all) of the lost
   morphological marking.  Thus while it may seem that analogical
   change is brought about by the linguistic situation, it is
   rather a response to a natural conflict between two competing
   systems, and, while it may be systematic, it is still
   unpredictable and irregular.  But even the phonological change
   that causes the response is not necessarily caused by an
   attempt a phonological simplification, since many phonological
   changes result in more complex phonology, so even a rule that
   phonological changes result in simplified phonology is not
   predictable.
 
Given these observations, it is obvious that the only thing about
linguistic change that is universal is change itself, and that
has nothing to do with language per se.  Furthermore, with the
exception of analogical restoration of morphological marking, a
linguistic situation is seldom seen to be the trigger for a
linguistic change, and even when it is, the nature and extent of
the change is not predictable.  Since linguistic change cannot be
seen to originate within language, it must be imposed by its
users, human beings, and therefore is sociological in nature
since language use is a socio-cultural phenomenon.
 
The sociological factors that affect (or effect) linguistic
change would seem to have to do with such things as intergroup
relationships and intragroup or cultural bonding. (I am not a
sociologist, so this terminology may not be current; I remember
sociology as the course where it didn't do any good to have last
year's exam -- they always asked the same questions - only the
answers changed :).)
 
A high prestige language or dialect is likely to trigger changes
in languages or dialects in contact with it by imitation.  But it
is not just high prestige languages that cause changes.  Thus
historical linguistics recognizes superstratum languages (higher
prestige, e.g., conquerors), adstratum languages (more or less
equal prestige, i.e, neighbors or ethno-linguistic mixtures
sharing the same territory), and substratum languages (lower
prestige, e.g., conquered populations).  While the nature of the
relationship between the languages may tend to influence the
types of changes that may flow between them (based on statistical
probability), again, any kind of influence of one language on
another is possible.  So all that is really needed for one
language to influence another is contact.  Even direct contact is
not needed, because, through writing, even long dead or unused
languages can cause changes (English has many Greek and Latin
neologisms).  Thus languages in almost any kind of contact can
cause changes in the lexicon (loan words and loan translations)
and grammar (areal features in phonology, morphology and syntax)
of one another.
 
Group or cultural bonding can have the opposite effect of causing
a language to deliberately be altered to make it more unlike its
neighboring languages.  Loanwords may be systematically purged
from the language to make it more specific to its culture or
group.
 
If linguistic change is sociological in nature, then like other
sociological changes (changes in government, religious, and
economic systems) it originates with the few, not the many.  The
many just follow along once the change is set in motion.
 
The point to generative grammar is that grammar has to have a
synchronic form that is independent of its history.  The speakers
of a language learn its grammar, very seldom the history of its
grammar.  But since the grammar (and lexicon) of a language is
constantly changing, a synchronic (generative) grammar is at best
a "snapshot" of the grammar of the language at a particular point
in time.  A synchronic grammar of the English of a century ago
would be different from a synchronic grammar of today's English
which would be different again from the synchronic grammar of
English a century from now.  What historical linguistics can do
is link the different "snapshots" and explain the mechanisms of
the changes.  But the history of these changes is irrelevant to
generative grammar which has to be able to explain the current
grammar of the language as used by its speakers on its own terms.
 
Now if universal grammar is a grammar that all languages must
obey and is supposed to encompass immutable rules of language,
then if everything in language can change in an unpredictable,
irregular, non-unidirectional way that can lead to conflict in
the language or if cross-linguistically every possible way of
accomplishing a linguistic result occurs, then universal grammar
either is a chimera or it is at best a restatement of Murphy's
Law:  Anything that can happen, will happen.
 
 
Bob Whiting
whiting at cc.helsinki.fi
 
 
On Thu, 30 Jul 1998 Isidore Dyen wrote:
>
> ---------------------------Original message---------------------------
>
> How about contemplating whether language change is inevitable. If
> it is not, theren should be some stable languages somewhere, If
> it is, then it must be inherent in all languages and thus a
> universal.
>
>On Thu, 23 Jul 1998, bwald wrote:
>
>> --------------------------Original message---------------------------
 Before I forget, I have some comments on Robert Ratcliff's last
message. He states:
 
>> >... if one takes seriously the generative claim that the
>> >goal of formal linguistic analysis is the discovery of an innate,
>> >biologically determined language faculty, then you sever the link
>> >between historical and formal linguistics.
>>
 
>> I would like to offer a different perspective.  It is not about
>> "severing the link", but about distinguishing between what is
>> innate and thus presumably immutable, unchangeable, universal etc
>> etc, VS. everything else in language.  The "everything else" is
>> what is relevant to historical linguistics, because it is what
>> varies and changes within and across particular languages from
>> one time to another.  Therefore, the search to isolate what is
>> innate or invariant in all languages also serves historical
>> linguistics by revealing those aspects of language, or of any
>> particular language, which are subject to change.  The two
>> programs complement each other, and work together.
 
>> Stated differently, GG, and no doubt any serious synchronic
>> framework for analysis which claims to be applicable to all
>> observable (and "possible") human languages, seeks to provide the
>> invariant parameters of language within which variation and
>> change are possible -- and to which variation and change are
>> *limited*.  This is quite different from severing the link
>> between historical and formal (i.e., "universal") linguistics.
>> So, despite the difference in emphasis, synchronic linguistics
>> continues its historic mission to provide a grounding for the
>> study of linguistic change.  One need not be misled by what some
>> GGists claim they are trying to do (not to mention what they
>> claim is "important").  It is no different from what historical
>> linguists are trying to do when they compare two changes and say
>> they reflect the SAME process of change.
>>
>> Having said that, then, it turns out that virtually every
>> substantive proposal that GG has made for something invariant in
>> language turns out to be too concrete, and the exceptions in some
>> language or other show that those features of language are indeed
>> subject to change.  And so the search goes on, as proposals for
>> concrete universals retreat into greater abstraction as the data
>> from more and more languages accumulate.  Each failed universal
>> is an opportunity for the historical linguist to contemplate and
>> try to determine how it is that languages can evolve in one way
>> or the other.