Re: LUNATIC again

At 1:25 AM -0400 11/10/98, Logical Language Group wrote:
>David Durand:
>>That's fine. I think that the need for such proof is relatively small in
>>this forum. Most of us have had the experience of creating a system
>>_adequate_ to communication, if impoverished (in grammar or vocabulary, or
>>both).
>
>And the question I have essentially asked is: how can you know, unless you
>have
>used it in communication with others (however impoverished that communication
>might be)?  I guess I am a little weak on argument by faith.
I guess I trust the _incomplete_ knowledge that linguists have gathered by
studying thousands of natural languages.

I feel that If I can get respectable grammar coverage relative to a book
like Payne's describing Morphosyntax, or the Timothy Shopen volumes, then
I'm probably defining an adequate system. These are books about descriptve
linguistics, defining concepts and giving examples of constructions
typically found in ntural languages.

I guess I do have faith that I can cover the typical grammatical space of a
language, if I can cover all of the grammatical and semantic points that
professional linguists think of as belonging to a "checklist" of features
to look for in describing a natural language. If I'm making a
Euroclone-type language, that's easier, because I will naturally extend the
underspecified pieces of the grammar in compatible ways.

I've already conceded that scope of vocabulary is a problem for many
conlangs: one that I think is correctible (either by vocab creation, once
in use) or by the creation of a sufficiently powerful derivational
morphology.

I don't think that communicational adequacy is that hard, actually, because
_people_ will solve whatever problems they find within a system, and make
themselves understood, given even a little bit of ground to stand on.

Now your taste in vocabulary creation, and focus on that project, makes it
more important to your enterprise than it is to mine. I don't mind making
up words (both more and less precise than English) as I write something,
since I'm not expecting the kind of audience that you want. On the other
hand, I want the grammar to be as complete as possible, because that sets
the "rules" for the game that I am playing with myself -- to create
emotional tone through linguistic invention.

>>We _know_ that it's possible to create such a system.
>
>Umm, we know that it is possible that such a system can be created.  It is not
>entirely clear to me that anyone can create such a system by act of will. This
>question can be examined by looking at Esperanto, which you also cite:
Well, Z. created it by an act of will. The distinction actually seems funny
to me.

>>For that
>>matter, Esperanto is an existence proof in itself that language creation is
>>not impossible.
>
>Well, if we grant that Esperanto as defined by Zamenhof as of the Fundamento
>was at that point (its time of 'completion') adeqyate, then we have a case.
>I am reasonably sure this is so, but am not an Esperantist and have to take
>the word of others.  On the other hand, the core of Esperanto, the "16 rules",
>demonstrates how much Z. was assuming of the norms of Standard European.  He
>invented a language, but hardly started from scratch, and indeed had 95% of
>the grammar defined for him, if not more.
This is true, and is a significant factor in the brevity of the Esperanto
grammar. My copy of Matt's Tokana grammar is 142 pages (I don't have the
latest revision). He leaves a lot less grammar to the imagination.

>But all in all I am inclined to admit that Z "invented" the language, and that
>Esperanto is a "conlang" %^).  So no definitional quibbling is needed.
exactly.

>However the creation of E. may show the possibility of language creation,it
>does not show that any particular other creation is speakable, is "a system
>_adequate_ to communication".  Thus E. is NOT an argument that anyone else
>OTHER than Z. has "had the experience of creating a system
>_adequate_ to communication" until they have demonstrated it.
No, it's not an argument. It's an example of the creation of a grammar
sufficient to support full nmatural communication in a language. Enough
languages have been described that one can create an almost infinite number
of naturalistic grammars simply by reusing pieces already known to be
adequate for communication.

While it's not the same as an empirical _proof_ in usage, it's enough for me.

John Cowan's point about undebugged code is a much better one, but I think
it doesn't apply because human languages, unlike computer languages, can be
extended in use: If you're trying to say something, your words will be
intelligently understood _in context_ to mean whatever is _most sensible_
to the hearer in that context. Given that, I don't believe that grammatical
rules determine very much of a language's power in use.

In fact, if you read a lot of descriptive linguistics, you see so many
different variations of grammatical marking that you realize that very few
specific features are essential. Tenses aren't, singular/plural isn't,
distinction between second and third person isn't always, the variety of
argument-marking schemes is quite large, etc. Those things that do seem to
be essential are mostly the "compositional" features, embedded clauses
(relative, and non-relative), coordination of clauses, negation, etc.

I can appreciate your hard-core empiricism, but I'm not at all convinced
that it's the right level of proof to require. I believe that for skeptical
linguists, it probably is, but in this community, we've all had experiences
that show us that this "hard thing" (making a system for language) just
isn't _that_ hard.

Making a good one, is.

>So that:
>>That's fine. I think that the need for such proof is relatively small in
>>this forum.
>I accept that this is true, but am arguing that it may perhaps be too true.
>For artlangs, especially those that are presumed to be spoken by speakers
>quite unlike 20th century humans, there is arguably no need to prove
>effectiveness.  In a fictional world, a claim of adequacy can hold,
>because fictional
>elements can hold if plausible, and many conlangs are *plausibly*
>communicative even if not demonstarably so. I guess I can and will stop
>arguing completely insofar as the topic includes fictional langauges, because
>plausibility is even harder to argue about than the definition of language
>$%).
But that's the only point that interests me. It's your advocacy of a
definition of langauge that excludes what I find to be _obviously_
languages (if not necessarily langauges with a complete vocabulary). If we
agree to ignore that part of the discussion, and apply language to
sufficiently powerful linguistc framwork, then my work is done!

I don't think that fictional plausibility is the point: I think that Tokana
would be a perfectly functional language in everyday use, with the addition
only of more vocabulary. I doubt we will _prove_ this, but I also think
that it's not sensible to wait for such proof. I think we know, based on
studies of natural langauges in use, that the Tokana grammar is adequate to
that.

I'm pretty sure Teonaht also meets that criterion, but I've not had a
chance to read the whole grammar.

Most of my own languages do not, which is a shame. I only recently feel
that I've learned enough syntax to finish some of my projects adequately.

>>>We all can play Humpty Dumpty. %^)  Language is luckily not very
>>>proprietary.
>>>But most people rely either on dictionaries or experts for their definitions
>>>and categorizations of phenomena.
>>
>>Actually, most people rely on usage, not on either of those two things.
>
>John Cowan's quote answered this.  Most people, whether relying on usage or
>not, will change their belief when confronted by an expert saying something
>else.  Exceptions tend to be confined to arenas of life where other
>epistemologies can challenge science, i.e. anywhere where a person's religion
>dares to tread.
>
>>My dictionary (AHD college) has 10 meanings (or usages) for the word
>>language, including verbal and written utterances, animal comminication,
>>mathematical notations, programming languages, etc...
>>
>>The term is a broad one. Your stubborn insistence on a particular
>>(imperfect) definition of the term is what I'm arguing against.
>
>To the extent that this is NOT a linguisitcs forum, I will condcede this.
>And Sally among others seem to be strongly arguing that this is not a
>linguistics forum.  Fine.  But does this then mean that the rest of the
>plethora
>of linguistic hjargon terms that are regularly thrown around in this forum
>have Humpty Dumpty meanings as well?  How am I (or others) to knwo when one is
>to take the popular definition of a term as distinct from the linguistics
>definition, when so much of the discussion is dominated by linguistics
>information expressed in quite sophisticated linguistic jargon?
But we've already shown that your vaunted "technical definition" is not in
fact universally accepted by linguists, and I've taken considerable toruble
to explain why I don't even think it's at the core of your problems with
the representatives of the discipline that you've had to deal with,
whatever they may say.

I don't believe that language is a technical term in the way that Ergative
is. One way you can tell that is that ergative is not in common use outside
the discipline. Another way you can tell is that "language" is not in fact
used in a technical way within the discipline: the meaning is almost
entirely implicit, and based on the internal research agenda of the field.
That's why you look so alien to them (because you present things not on the
agenda) and why people suddenly start trying to use a definition to shut
you out.

>At worst (and probably true), I am guilty of taking a term with a jargon
>definition and presuming that this definition is intended by peopl who use it
>along with related jargon on this forum.  Where are the context clues that
>tell me NOT to use the jargon meaning of the word?  Some would think this
>is obvious, but I guess I am obtuse at understanding context then ( people
>have
>indeed accused me of having all the sensitivity to context of a ... well maybe
>I won't get into that %^).
I don't believe that language is a jargon term, and that's why I don't see
this problem. My exposure to linguistics and the academy come from being an
faculty brat, college/grad student for more than a decade, a Univerity
employee for some years (and freelancer within a university for some
others), a few linguistics courses (until the inferior quality of the
formal methods applied in traditional transformational/generative grammar
disgusted me), doing several semesters of graduate-level AI coursework in
syntactic parsing, and reading linguistics books from structuralism to GPSG
to current functionalist texts.

None of that experience made me think that my formal-systems definition of
language was inappropriate. It also mae me very sure that talking about
inventing languages would be regarded as weird, pointless, suspsicious, or
impossible. That's because of the socialogical factors, not the
intellectual content of the disciplines, in my opinion.

>>English is not entirely fixed, because people extend it all the time,
>>invent new narrative styles (Joyce did, certainly),
>With John Cowan the Joyce quoter in my camp, I should not argue this, but at
>times I wonder whether most people would consider Joycisms to truly be
>English (I enjoy them anyway %^).
I was thinking not of Finnigan's Wake, which is much at the borders
(intentionally) but of the "stream of consciousness" style of the Molly
Bloom section of Ulysses. It was very bizarre at the time, but has been
copied in a way that FW's has not. My favorite book in this kind of style
is Gabriel Garcia Marques' "The Autumn of the Patriarch."

>>>I would have to understand better what you mean by those aspects, since
>>>English aspect is so unlike Russian aspect which is how I have finally come
>>>to understand perfect/imperfect.
>>Assume it's just like russian. That's my point.
>  I don't see how your discussion of your"English-prime" has anything to
>do with
>Russian which has tense and number disticntions.  So I do not see your point,
>But i think that we've moved past whatever you were tying to show.
The point is that even with identical vocabulary, grammaticalizing a
different set of core notions will change the language fundamentally.
Word-for-word translation can become impossible even if the words are
_identical_, if the grammatical system is changed. That's true (as you
admit) for Lojban. There's a whole host of potential (but boring) conlangs
that involve taking the grammar of one language, and the lexicon of
another, and pairing them up. These will _boviously_ have the magical
"communicative power" property. If you start to also make changes in
specific grammatical rules (based on the range of variation observed in
human lanagues), you _might_ somehow lose that peoperty, but unless you run
into some human cognitive limitation, that's unlikely. The only verified
such properties are a very limited number of syntactic ones that don't seem
to be fully processable in speech situations. Center embedding and the
"respectively" construction are typical examples. They both occur (to very
limited extents) in human language, but they are never applied in full
generality. We might say "John and Jim ate and drank respectively", but
saying that "the cat, the rat, the dog, the car and the bird feces, chased,
skulked, bit me, went to town, and fell from the sky, repsectively" would
be unlikely to be understood in most situations.

So I think for naturalistic conlangers, there are good reasons to _believe_
in communicative adequacy. Particularly given the relevance of context, the
use of metaphor and so forth.

>>That E-prime is unconvincing to me because I think it's just paraphrasing
>>around one grammatical construction, and preserving the semantics whole.
>Yet it is "spoken" and adequate for communications, according to some
>proponents
>who claim to use it in everyday life.
Since the change is trivial and insignificant, this isn't surprising.

>  It is more complete than most conlangs,
>has speakers, has a grammatical difference from standard English (albeit
>essentially only one, it is a fairly big one).  Is it a "language"; is it
>a distinct language from English. Is it a conlang (it is invented and it is
>a language)?As you say, it largely preserves semantics whole- I would thus
>call
>it a code and not a language despite the grammatical difference.
You missed my criticism I think. I don't think the difference is a big one,
I think that it's a trivial one. If it were a big one, then I'd accept it
as an example.

But applying that same technique with meatier changes is what I'm getting
at. You seem to think that the vocabulary is the only repository of
semantics in the language. Many features of morpohology and syntax amount
to mandatory semantics. In English, you must decide whether something is
singular or plural to talk about it, even if that notion is irrelevant. In
other languages, you must also decide if plural objects are long and thin,
or small and round, or whether they are in bundles or in heaps. In English
you must decide if an action occurred int  the past present or future
(among other things), in other languages you must decide if an action is
just beginning, just ending, is an oscillating motion, or is a motion
repreated multiple times.

All of these semantic distinctions are priveleged by the rules of grammar
in some language. Changing these kinds of rules can make a new language,
with new semantics, just as can making up a new vocabulary.

  -- David
_________________________________________
David Durand              dgd@cs.bu.edu  \  david@dynamicDiagrams.com
Boston University Computer Science        \  Sr. Analyst
http://www.cs.bu.edu/students/grads/dgd/   \  Dynamic Diagrams
--------------------------------------------\  http://www.dynamicDiagrams.com/
MAPA: mapping for the WWW                    \__________________________