Re: Word Order in typology

On Wednesday, October 13, 2004, at 09:02 , Chris Bates wrote:
[snip]
> I suppose you could argue that subject does exist as a notion in some
> ergative languages, even though it isn't exhibited in the morphology.
> For example,
>
> He passed the man and stopped
...except that I would not classify English as an ergative language. By
_ergative_ I mean, as do many other people, a language like Basque. I'm
afraid I accept the term 'ergative' only in the first two meanings listed
by Trask. I note that he begin definition #3 with "A label sometimes given.
...." & #4 with "A name sometimes given..."; the "sometimes" does not
include the times I use the term ergative. (His 5th definition refers to a
use in Government Binding theory)

Calling _stopped_ above ergative seems to be the 4th definition. In other
words, I do not think it is ergative.

> Would generally be interpreted as he stopped,
Is Always interpreted that way,

> not the man stopped, so
> the two verbs automatically share the same subject even though that
> subject is ergative in one and absolutive in another.
I don't think this is helpful. There was a time when the complain was that
Latin grammar was applied to languages for which it was not appropriate.
IMO this is an example of Basque grammar being applied to a language for
which it is not appropriate.

For one brought up on ancient Greek, the _stopped_ could be construed as
an example of the 'middle voice' for which the Romance languages use
reflexive forms (and Classical Latin simply used the passive). More simply
"stop" can be used in English either as a transitive verb [with quite a
wide range of meanings] or an intransitive verb meaning 'to come to an end,
  to come to a standstill'.

> I'm not convinced
> that trigger languages exhibit the notion of subject to any large degree
> though,
I am not familiar enough with such languages to be able to comment.

> and I've seen no proof that this is true of all ergative
> languages either. :)
As far I know, true ergative languages like Basque do not have grammatical
verbal arguments corresponding to the subject & direct object of
accusative languages, but use arguments known as absolutive & ergative. (I
realize split-ergativity as in Georgian makes things less straightforward)


>> [snip]
>>
>>> are in practice useless. I don't mean to disparage anyone who's a
>>> professional linguist, but I find a lot of linguistics to be a load of
>>> complete rubbish, or at least something strictly intuitive pretending to
>>> be rigorous. If you can't define something properly you should be
>>> honest, rather than pretending that your art is a science.
>>
>>
>> I'm not a professional linguist but, with respect, I find this a rather
>> short-sighted view. For centuries the physical explanation of fire was
>> not
>> known or was misunderstood. At one time physicists had a theory about
>> 'phlogiston'. We now know this is wrong. It doesn't mean that until
>> modern
>> science appeared all previous attempts at science was merely 'art'.
>>
> I guess not. :) The problem with a lot of science, and I guess
> linguistics, is that people try to distort the facts or the established
> theories slightly so that strong evidence that the current theories are
> very wrong fits.
Yes, but if others conduct scientific investigation with proper rigor this
will eventually be shown up.

> You just have to look at electromagnetism and the magic
> "ether" that appeared along with the assertion that the laws only hold
> in the Absolute frame, because the physicists didn't dare consider that
> the Galilean transformation would have to be scrapped.
Yes, yes - that's history. That the very point that some of us have been
trying to make. It is misleading to talk about "laws" in any science. All
we can do is experiment, examine facts and try to make sense of them; if
we can we come up with a _theory_ which in its turn can be tested,
verified, changed or scrapped as knowledge advances.

[snip]
>> Hang on - _argument_ roles may be defined _either_ grammatically
>> (subject,
>>  object etc) _or_ semantically (agent, patient etc). If I understand your
>> argument you are saying that grammatical arguments are fairly
>> meaningless.
>>  But how then are we to explain the grammar of a particular language?
>>
> I'm not arguing that GRs are useless... only that they are useless as a
> grounding for the study of language, because they are not language
> independent. :)
But you have to start with languages. It is only through the analysis of
individual languages & language groups that we ever hope to arrive at a
more general concepts.

> It seems strange to me to base the study of language on
> terms which do not have the same properties for each object under
> study... take group theory for instance.
It happens in real life however, in the study of, say, geology, biology
etc.

[snip]
> current group theory would just collapse. I have no object with using
> GRs in linguistic theory, but they shouldn't appear at all in the basis
> of linguistics,
Who said they were the _basis_? Linguistics surely covers the whole
complex of linguistic phenomena: phonology, grammar (both morphology &
syntax) and semantics.

> since they seem to me to be a language dependent
Just like phonologies.

> structure built on top of other language independent structures.
How do we know those structures? The semantic roles such as Agent, Patient,
  Recipient etc are abstractions that have been derived from our study of
individual languages (and of course are expressed through the medium of
language). We may assume they have universal significance, but it is just
that: an assumption - a theory.

> So any
> study of GRs should be done by considering how they related to other
> notions that aren't language dependent, and how languages group those
> concepts together to form GRs.
I agree this is a useful thing to do, if only to test the validity of the
theory of semantic roles. But unless one believes in some Platonic
absolutes, we must remember any notion of language independent concepts is
notional.

[snip]
> I am not arguing at all that we shouldn't study surface phenomena, but
> the study of surface phenomena must be grounded in the study of deep
> phenomena as much as possible.
The main problem - and it is a very real problem - is knowing what the
deep phenomena are or, indeed, whether there are deep phenomena at all. At
least we know the 'surface phenomena' are there.

[snip]
>> Ah, now that is hinting at the 'deep case' idea - a controversial and not
>> universally held idea.
>>
> It seems to me to be the most sensible idea. :) Not because of inbuilt
> case (I do not believe in the large scale language specific structures
> in the brain which Chomsky claims exist....
Good - we are agreed on something  :)

> I believe that language is
> shaped by shared enviroment and experience, and that the structures that
> are inbuilt from the beginning for language as very general: probably an
> area of the brain very good at parsing trees etc),
I basically agree with this also.

> but because humans do
> think about the same things because our experience shapes us that way:
Shades of Sapir-Whorf  ;)

> its natural to divide the world into those that do things (Actors),
> those that have things done to them (Patients), things that are used in
> doing an action (Instruments... tools are the basis of our civilization
> after all, along with language), etc.
I have reservations here. This seems to me intuitive thinking.

[snip - not that I think the points you made are irrelevant]

> If  the main word order is SOV then...
>
> The condition talks of a surface realization which may not be exhibited
> in a given language, so the universal is not applicable to all
> languages, and thus flawed,
I do not think anyone on this list would except any of the "If a language
has SOV then....." as universals applicable to all languages. At best the
so-called universals are tendencies - and some are very weak tendencies if,
  indeed, they are tendencies at all. No one AFAIK except Greenberg has
claimed they are applicable to all languages.

> since the whole point is to try to find
> phenoma true of all (or at least 99.9%) of languages.
Personally, I do not think our knowledge of language is sufficiently
advance to allow that to be done; but I am well aware that others will
disagree.

>> Grammar does exist IMO or do you think all so-called grammatical features
>> are explainable in terms of semantics?
>> (That is a genuine question - I hope it does not sound aggressive; it is
>> not intended to be.)
>>
>> Ray
>
>
> No, grammar does exist. :)
Good - something else we agree on  :)
================================
On Wednesday, October 13, 2004, at 12:47 , Chris Bates wrote:
[snip]
> .. so in a
> similar way its unsound to build general linguistic theory which is
> supposed to hold for all languages on notions like subject which haven't
> been shown to be applicable to all languages.
I agree. I have already said I don't think 'subject' is applicable to a
language like Basque, for example.

In any case, the "If a language has SVO..." can be definition be applied
only to languages where the concept of subject and (direct) object are
applicable. The rule cannot be universal and "if" surely, in any case,
means at best the rule (even if valid) is conditional.
=====================================

On Wednesday, October 13, 2004, at 12:54 , John Cowan wrote:

> Chris Bates scripsit:
[snip]
> Chinese is neither syntactically accusative nor syntactically ergative,
> interpreting such sentences as containing semantic ellipsis rather than
> a syntactic gap, and filling the ellipsis with whatever makes sense:
>
> 	George dropped the watermelon and broke
> 	George dropped the watermelon and was embarrassed
>
> are both legitimate and sensible sentences.
  Great language, Chinese      :-)

[snip]
>> I am not arguing at all that we shouldn't study surface phenomena, but
>> the study of surface phenomena must be grounded in the study of deep
>> phenomena as much as possible. My main argument was that the universals
>> regarding word order should be specifying that order using surface
>> phenomina which aren't exhibited by all languages.
>
> The difficulty is that surface phenomena are on the surface, available
> to all, whereas deep phenomena (which is something of a contradiction
> in terms: something that's "deep" precisely does not "appear")
Good point! But I am too lazy to go back & change what I've written above.
  They cannot, as you say, be phenomena; they are noumena.

> are theory-laden and doubtful.
Exactly!

They are theory-laden, and theory have to be tested. It is unsound to
build more theories about untested & doubtful theories. It seems to me
even very unsound to try and build universally applicable rules on
theory-laden and doubtful noumena.

[snip]
> Most actual language universals are conditional.
Are there in fact any which are unconditional?

Ray
===============================================
http://home.freeuk.com/ray.brown
ray.brown@freeuk.com
===============================================
Anything is possible in the fabulous Celtic twilight,
which is not so much a twilight of the gods
as of the reason."      [JRRT, "English and Welsh" ]

Reply