LL-L "Grammar" 2005.12.12 (01) [E]

[Lowlands-L]

======================================================================
L O W L A N D S - L * ISSN 189-5582 * LCSN 96-4226
http://www.lowlands-l.net * lowlands-l at lowlands-l.net
Rules & Guidelines: http://www.lowlands-l.net/index.php?page=rules
Posting: lowlands-l at listserv.linguistlist.org or lowlands-l at lowlands-l.net
Commands ("signoff lowlands-l" etc.): listserv at listserv.net
Server Manual: http://www.lsoft.com/manuals/1.8c/userindex.html
Archives: http://listserv.linguistlist.org/archives/lowlands-l.html
Encoding: Unicode (UTF-8) [Please switch your view mode to it.]
=======================================================================
You have received this because you have been subscribed upon request.
To unsubscribe, please send the command "signoff lowlands-l" as message
text from the same account to listserv at listserv.linguistlist.org or
sign off at http://linguistlist.org/subscribing/sub-lowlands-l.html.
=======================================================================
A=Afrikaans Ap=Appalachian B=Brabantish D=Dutch E=English F=Frisian
L=Limburgish LS=Lowlands Saxon (Low German) N=Northumbrian
S=Scots Sh=Shetlandic V=(West) Flemish Z=Zeelandic (Zeeuws)
=======================================================================

   L O W L A N D S - L * 12 December 2005 * Volume 01
=======================================================================

From: Global Moose Translations <globalmoose at t-online.de>
Subject: LL-L "Grammar" 2005.12.11 (04 [E]

Ian Pollock wrote:
> PS Thanks to Sandy Fleming for her heroic summing-up of the discussion.

One click says it all:
http://www.lowlands-l.net/anniversary/index.php?page=sandy

Gabriele Kahn

----------

From: Paul Finlow-Bates <wolf_thunder51 at yahoo.co.uk>
Subject: LL-L "Grammar" 2005.12.10 (11) [E]


Thanks for the summary and explanation Sandy.  All explained, I might point 
out, in perfect "conventional" grammatical English!

Paul

----------

From: heather rendall <HeatherRendall at compuserve.com>
Subject: LL-L "Grammar" 2005.12.10 (06) [E]

Message text written by INTERNET:lowlands-l at LOWLANDS-L.NET
>What am I missing?<

Sorry! you're right...... tho' it sets one thinking that perhaps the brain
can distinguish ( needs to distinguish) between what can or cannot eb
inserted: hence function markers -

The quickly car     the quick car

The fast car ?????????

The lovely car ??????????

Isn't this what makes English so difficult for those looking/expecting
function markers?

Heather

----------

From: Jan Strunk <strunkjan at hotmail.com>
Subject: LL-L "Grammar" 2005.12.10 (11) [E]

Dear formal syntax bashers ;-)

Sandy Fleming wrote:
> I think at this point I should, while admiring the cleverness of some
> aspects of syntactic theories, explain why I don't think they're as good
> a description of language as they seem. While this may seem to be
> getting more into pure linguistics than Lowlands languages, I do think
> that it has a bearing on the relationship between language and dialect
> in the sense that grammarians seem to talk about an imaginary "standard"
> language and have more difficulty accepting the validity of dialects,
> even small dialectical variations, because they don't fit in with their
> theories.
Well, some grammarians do like to concentrate on one "major" (usually
European)
language. But there are many exceptions and linguistics as a whole is
incorporating
more and more data from "exotic" languages and dialectal varieties, etc.
You can even write a formal analysis of possessive constructions in Low
Saxon and
get an MA for that ;-)

> Firstly, consider the nature of scientific theories in general, by going
> back to our example of Newtonian vs Einsteinian theories in physics.
>
> When Kepler discovered that Mars had an elliptical orbit, it was a bit
> of a nail in the coffin of the then-favoured theory of epicycles. On the
> other hand, when Newton figured out a simple explanation of why the
> planets' orbits were elliptical, physicists were jumping for joy not
> only at how simple his theory was, but also how they were now able to
> calculate everything about the planets' orbits correctly, not only the
> shapes, but also such subtleties as the rate of precession.
>
> The only problem was that Mercury's orbit seemed to be precessing at the
> wrong rate. They tried to find all sorts of explanations and failed.
>
> Physicists actually stuck by Newton's theory and hoped that one day
> they'd find something to explain the anomaly in the rate of precession
> of Mercury's orbit. In the end they discovered that Newton's theory was
> only an approximation to reality and that General Relativity gives a
> better approximation. But imagine if they'd taken another tack. Imagine
> if they'd said, "Precession of Mercury's orbit, no problem, just add a
> constant and call it Precessional Augmentation so we still sound like
> scientists."
>
> Then they discover that light only bends by half as much in a
> gravitational field as Newton's theory predicts, but still, no problem,
> call it Photic Reduction and nobody's any the wiser.
>
> This is what you might call a patched theory. You can add as many
> patches as you like and call them as scientific names as you like but if
> it's a symptom of a flaw in the original theory, you'll never be done.
> Eventually people start building particle accelerators and in their own
> little world it begins to seem as if Newton's theory doesn't even come
> near to working, so the staunch Newtonians say, "No! Accelerating stuff
> to such high velocities is wrong!" and start an educational programme to
> make sure children never grow up with such disgusting behaviours as
> accelerating things.
>
> Of course, physicists don't behave like this. They can see all too
> plainly that they're up against hard reality and it's not to be argued
> with. But it's not so obvious to some that languages are hard reality
> and if you really want to understand them then trying to control them
> isn't the way to go about it.
Hoho, this is the old linguistics isn't a science, physics is, debate. I
agree with
your feeling that many strands of modern syntactic theory have a bad habit
of patching there theories, but this does not apply to the whole enterprise
of
linguistics. Just as the whole of modern physics isn't necessarily an
empirical science,
just look at all the speculative work on black holes, strings, etc.

> If you look at modern transformational grammar, I'd say that this
> started off from promising beginnings and then quickly grew into a prime
> example of a patched theory.
>
> One root of the problem might have been the importance of languages like
> Latin and Greek earlier in the educational system. I order to work with
> the morphology of these languages, people got used to thinking in terms
> of parts of speech and their paradigms. There was a feeling that words
> could somehow be wholly characterised in terms of whether they were
> nouns or verbs and suchlike. These ideas were inherited by modern
> grammarians who eventually managed to explain the whole of the grammar
> of a language in terms of a few simple structures - or in fact a single
> simple structure, the XP rule.
As I have said before, I agree with you that especially in transformational
grammar
many authors make up new principles and patch their theory for every new
language
or structural pattern they look at. I repeat though that there are many
formal syntacticians
that neither assume empty elements, nor transformation, nor assume that all
constructions
in all languages should adhere to one simple X-Bar-Schema. At least, I
don't. Moreover, most
linguists today realize that categories can be more or less fuzzy. But still
that doesn't remove their
psychological plausibility.
Last but not least, all people that feel dissatisfied with the way current
formal syntax is done are of course
invited to attack these theories, come up with better ones and test them on
empirical data. In fact, there is
currently a small revolution going on in Linguistics that questions the
validity of intuitive judgements and calls
for controlled experiments and corpus data to be used more extensively. So
better times might come
again for linguistics ;-)

> Impressive, except that as soon as you pull back from the simple
> examples (or "lab samples") that the theory works with, you run into the
> fact that it doesn't work with many real sentences.
>
> For example:
>
> "The girl will eat the cake."
>
> breaks down nicely according to the XP rule but as soon as we ask:
>
> "Will the girl eat the cake?
> What happened to our beautiful the XP rule? The idea in a
> transformational grammar is now to say that the XP rule describes the
> "deep structure" and to explain the fact that "will" shifted to the
> beginning of the sentence we introduce the idea of "surface
> transformations". It all sounds very scientific, but it's just a way of
> making a patched theory sound organised.
Well, I think one should not seriously expect one structural rule to cover
all
constructions, patterns, templates, call them what you will in all
languages?
Of course, you are allowed to amend your theory. I don't like the way that
is done in transformational syntax, either, but you don't have to through
away the
whole theory of endocentric (i.e. headed) structure just because some
constructions
work differently. Or would you do this in physics?

> We can now have inversion, wh-movement, "do"-insertion, whatever doesn't
> work, we can make it work, as long as real language is sufficiently
> organised. Eventually, the attitude of our imaginary anti-accelaration
> physicists becomes reality in our non-imaginary grammarians when they
> run up against sentences like:
>
> "You do that, gonna leave you without a dime."
> "Don't make no sense."
> "Could be I ain't."
>
> Strangely, these American English sentences (possibly affected by
> Spanish?) are easy to make sense of while the following perfectly
> grammatical Scottish English sentences are more confusing (though not
> for me, as a Scots speaker!):
>
> "It was nice to meet her, I'm not saying that."
> "Don't worry about it, it's a matter."
> "Yes, will I." (OK, this one might need a "patch")
>
> This anomaly suggests to me that syntax is not central to language
> description, but that semantics might be.
Nobody in linguistics today questions the fact that there are more general
rules, templates,
constructions and more specific "idiomatic" templates. In fact, even the
"bad" transformational
grammarians often use idiomatic expressions as a type of evidence.

Now there might be some time for me to throw in my two cents with regard to
rules, templates, etc.
1. I believe that we indeed learn language using a more or less general
learning mechanism by observing
   patterns of language connected to patterns in the real world.
2. However, just as a simple computational neural network can learn precise
mathematical functions such
   as logical and, or, etc. Our brain is able to abstract more general
patterns or templates from several
   simpler patterns. And thus is able to form categories. These are not
necessarily 100 % discrete, but often
   fuzzy. It is mostly these more abstract patterns that linguists try to
describe while they often confine
   very specific patterns to "lexical knowledge".
3. Now, you should tell me, whether it makes a difference or not to call
these more abstract patterns
    templates, constructions, or indeed rules?
4. For me universal grammar is the collection of characteristics of human
language that arise from its use and
    historical forming, the way our brain works (e.g. we can process
recursive structures), and possibly
    (I would not totally do away with the possibility) also some inherent
properties of our neurological structures
    used for language processing (which would be nearest to Chomsky's
universal grammar or language organ).

> (An aside...
>
> Of course, too many grammarians, or even more so, people who have been
> educated with a certain grammatical bias, at some point depending on
> where they draw the line, will start declaring aspects of real language
> as "ungrammatical" and instead of attempting to understand it will use
> "grammar" as an excuse to disparage those who speak differently from
> themselves. This is _not_ in the spirit of scientific enquiry and can't be
> used as a
> basis for language study.
Yes, indeed. But from my experience it is rather the case that formal
grammarians sometimes disagree
about the grammaticality of examples because we humans are simply not able
to say for every example
whether it is grammatical or not. (This might have something to do with our
neural network ;-)
And the methodology of obtaining grammaticality judgements is often not very
scientific. But things are
changing, there is a small revolution going on in current linguistics about
what kinds of evidence one should
use for testing and building theories. Linguists often have problems to
communicate with "ordinary people" about
grammaticality because modern linguistics has totally shed prescriptivism
which is held dear especially by
other "language professionals" such as teacher, editors, etc. For example, I
often argued about language
questions with my mother who used to be a highschool teacher of German. She
would often say things
like "Meim Vatter sein Auto" is just plain wrong German, it's totally wrong,
etc. While I would say: No, its
a relatively new pattern that is emerging and has not been excepted in the
formal register, but it used in almost
all Germanic languages and languages all over the world, including e.g. many
Austronesian or Native American
languages.

> ...)
>
> However, to get back to the idea of semantics being central to language
> description, perhaps Paul hit on something nearer to the truth when he
> said that:
>
> "I can place an adverb between 'king' and 'runs' so: 'the king quickly
> runs' but I can't place an adverb between 'the' and 'king': 'the quickly
> king runs'. I conclude that I can't do this because 'the king' forms a
> discrete structure within the sentence, that the adverb 'quickly'
> modifies at a sentence level and that it therefore cannot break up the
> unity of the phrase, 'the king'."
>
> Here I think Paul is describing the phrase in terms of the semantics, he
> really only falls down when he labels words as "adverb" and so on. In
> fact the whole thing falls down on syntax, at least if we're talking
> about transformational grammar, because the XP rule is broken over and
> over again in the various placements of "quickly", and talking about one
> thing being at sentence level isn't part of transformational grammar
> theory.
Well, but in some languages you can place adverts where it wouldn't make
sense
semantically. Also, you can put things like relative clauses in places where
they are
not near to the "concept" they modify. (By the way, "modify" is also mainly
a syntactic
concept.)

Such as in:
I will give you all the keys tomorrow that you need for your office.

"tomorrow" neither combines semantically with keys nor with the relative
clause. The relative clause
belongs more closely to keys than to "tomorrow". There are several
possibilites of modelling this.
But I don't think that you can explain linearization facts by semantics
alone.

> For example, we will say:
>
> "Quickly the king runs"
> "The king quickly runs"
> "The king runs quickly"
>
> But not:
>
> "The quickly king runs"
> "The runs king quickly"
>
> In transformational grammar this is done (I hesitate to say "explained")
> by categorising words as "nouns", "verbs", "adverbs" etc and in the
> "Deep Structure" using the clever XP rule to arrange it as:
>
> (The king)(quickly runs)
>
> And then apply various not-so-clever "Surface Transformations" or
> patches, to get the other allowed sentences. The problem here is that if
> we _had_ happened to say sentences like:
>
> "The quickly king runs"
>
> Then we could have covered this too by applying a new kind of Surface
> Transformation. Surface transformations do nothing for our understanding
> of language - you can allow what you like, you can disallow what you like.
>
> You'll notice that we can't generate the sentence:
>
> "Quickly the king runs"
>
> without breaking up the verb phrase. This is why we need patches.

> However, ignore syntax and bring in semantics the way Paul suggests (see
> how I cleverly make it _your_ idea in the hope that you'll agree with
> me? :) and things seem to work a lot better:
>
> "Quickly the king runs"
> "The king quickly runs"
> "The king runs quickly"
>
> A king is one thing we perceive in real life, running is another. "The
> king" is more qualified, but still one thing in real life, and the same
> can be said of "running quickly." "A running king" can also be
> visualised as a single concept, so why not apply the concept of
> "quickly" to that?
So why should "The king is quickly" not work?

> So now we can say:
>
> "Quickly the king runs"
> "The king quickly runs"
> "The king runs quickly"
> "*King the quickly runs"
> "*King the runs quickly"
>
> Now you'll notice that the adverbial phrase can break the XP rule while
> the noun phrase can't.
>
> I think we need look to something else to explain the many variations
> that occur in practice, not just a bunch of patches (which only
> generates language _we_ approve of, even though patches could be
> invented to generate anything).
Many grammarians that call themselves functionalists try to "explain" the
way
language is by appealing to historical forces, cognitive forces, economcy of
expression, etc. I think that this is a good endeavour, but I also think we
have to accept,
and this also follows from the templatic learning paradigm, that many facts
about
particular languages cannot be explained in principle, let alone by
semantics. One great
part of language simply consists of conventions which we have learned as
templates and
which we might sometimes use as productive rules and sometimes as fossilized
patterns.
I think this is also why sometimes the funny thing happens that students in
a language
class refuse to accept patterns in the language they want to learn and ask
"Why can you say
that, that's not logical, that's so weird!" And often language teachers make
efforts
to explain why something works like it does in a certain language while in
reality they can't
explain it because it's a simple convention and not necessarily logical.

> One idea is that instead of trying to analyse a sentence as a tree
> structure we try to look at it in terms of what we, as humans, can
> perceive as a single object or concept in real life. From:
>
> "The king quickly runs"
>
> we could isolate as concepts:
>
> "King"
> "The king"
> "running"
> "running quickly"
> "a king running"
>
> but not:
>
> "the quickly" &c.
So, how would semantics help you to rule out some orderings while allowing
others?
Note, that I don't say I want that using tranformations is right. In fact, I
totally despice
transformations. By the way, you should have a look at Cognitive Grammar (by
people like Langacker, Taylor, etc.). They explore language more in the way
you suggest...

> We might then be able to formulate a theory where sentence structure is
> driven by concepts rather than syntax, explaining why some phrases are
> syntactically more flexible than others. Ultimately the syntax rule (and
> perhaps there would only be one for each language, like the XP rule)
> comes into play, but beneath the concept level, not above it.
Well, how do you explain then that the same concept expressed in two
languages
can either have a very fixed structure or a very loose structure.
Consider for example Germanic noun phrases or "concepts referring to
entitities" (there is no
easy semantic characterization, I think) vs. Latin noun phrases.

Whereas the linearization of words in a Latin nominal phrase is quite free
and words can even
appear spread out over the whole sentence, the same is not true at all for
German, for example.

ille adiutor Mercatorum clarus mihi notus non est.
adiutor ille Mercatorum clarus mihi notus non est.
ille clarus adiutor Mercatorum mihi notus non est.
etc. pp.

Dieser Helfer der Händler berühmte ist mir nicht bekannt.
? Helfer dieser der Händler berühmte ist mir nicht bekannt.
Dieser berühmte Helfer der Händler ist mir nicht bekannt.

Would you then want to say that the semantics of Latin and German is so
different, rather than the structural
syntactic organisation?

> I don't know the answer but I do feel that transformational and similar
> grammars are inherently unscientific and a break from trying to
> categorise words as nouns, verbs &c and a move over to semantics as
> central to language structure might give us a better understanding and a
> more sensible system of values for what's "right" and "wrong" in
> language production.
Well, you probably feel that (as I sometimes do, even though I am a
practising formal syntactician) because
you concentrate on certain varieties of transformational grammar that make
use of very abstract ideas and
try to force the data into their theory. I agree with you that these kinds
of theories never get assembled into
one coherent whole (every author adds his/her own ad-hoc principles) and
thus cannot really be tested empirically.
(By the way, things like empty elements sometimes get tested empirically
using psycholinguistic experiments...)
However, please widen your horizon and look at other formal theories, too.
Such as HPSG, LFG, cognitive
grammar or even formal construction grammar, which is somewhat closer to
your templates!
And moreover, it is one thing to say that you find the (or some) current
formal theories of language unscientific
and to conclude from that that languages are organized according to meaning
and concepts
and don't need a separate structural level. In order to support your claim,
you would have to look at the arguments
why you need syntax (such as e.g. my argument regarding German vs. Latin
above) and argue convincingly that
meaning and concepts is all you need.

To sum up my somewhat chaotic reply:
1. I think you do need syntax (and not only for computational linguistics)
2. I agree that it is not done in a very scientific way in some current
theories.
3. But there is no reason to dismiss the formal study of language as a
whole, because it has
    allowed linguists to ask question that no one would have come up with
had we only talked
   about language in traditional or philosophical ways. (Concepts and the
like...)
   Even the frameworks that I do not like at all because everyone makes up
new ad-hoc rules,
   presses language into as few rules as possible, assumes empty elements
galore, etc., have made
   significant contributions by asking questions noone had ever thought of
before.

Jan Strunk, defender of formal syntax ;-)
strunk at linguistics.rub.de

----------

From: Sandy Fleming <sandy at scotstext.org>
Subject: LL-L "Grammar" 2005.12.06 (05) [E]

> From: Brooks, Mark <mark.brooks at twc.state.tx.us>
> Subject: LL-L "Grammar" 2005.12.05 (06) [E]
>
> Hello Syntax Collectors (pun intended):
>
> I would like to provide an example of my 2-year-old grandson's language
> acquisition.  He talks quite a bit, and seems to understand more than
> he can
> say.
>
> One of his current phrases is:  "Pick you up."  From the context and
> timing
> it's easy to see that he means "pick me up."  But, he seems to be
> repeating
> what he hears his parents ask him, albeit pared down.  They say and ask

In BSL, probably most other sign languages too, pronouns are indicated
by "indexing" ie pointing with the index finger (or, in more adult
speech, often the eyegaze).

The strange thing is that Deaf children make exactly the same mistake
and sign, "Pick you up," pointing at the adult, to get themselves picked up.

Not sure what that says about language learning, but it does seem
interesting. Perhaps it gives an idea of how _very_ imitative and
unanalytical infant speech is? Or maybe it is analytical but they just
think the structure means, "You pick up"? I don't know!

Sandy Fleming
http://bsltext.org/
http://www.livejournal.com/users/cochlear_my_eye/ 

==============================END===================================
* Please submit postings to lowlands-l at listserv.linguistlist.org.
* Postings will be displayed unedited in digest form.
* Please display only the relevant parts of quotes in your replies.
* Commands for automated functions (including "signoff lowlands-l") are
  to be sent to listserv at listserv.linguistlist.org or at
  http://linguistlist.org/subscribing/sub-lowlands-l.html.
======================================================================