Re: My Three Assertions
|From:||Damian Yerrick <tepples@...>|
|Date:||Friday, February 25, 2005, 2:40|
Quoting Mike Ellis <nihilsum@...>:
> "constructed" ... "having been constructed" = "having been deliberately put
> together / shaped / engineered with the goal of making a language"; whether
> it's from scratch or from existing language(s) doesn't matter here.
> No, there aren't characteristics inherent to conlangs AFTER their
> construction that seperate them from natlangs.
So in other words, sometimes the difference between natlangs and some
conlangs can be expressed only diachronically.
> But I'd say there are plenty of a-priori and semi-a-priori langs
> that wear their constructedness visibly.
One way to make a conlang more naturalistic is to put it through an
iteration or two of sound change rules, which obscure the strict
regularity (cf. English umlaut for |foot|->|feet| and |mouse|->|mice|),
and possibly to toss in some occasional stem suppletion from
near-synonyms (|go|->|gade| and |wend|->|went| becoming |go|->|went|).
> However, having been deliberately constructed (usually by one person)
> is itself a very different way for a language to come about than
> having developed over generations of speech without deliberate
> creation. THAT is the unique characteristic.
The line you seek isn't as bright as a fellow might hope: Look at
prescriptivist reforms by authors of popular reference grammars.
A couple famous grammars of the eighteenth(?) century tried to fit
English into a Latin mold, deprecating features such as double
negatives, sentence-final prepositions, and split infinitives.
Some would claim that Sanskrit is a conlang as well, a
regularization of one of the prakrit dialects.
See also pidgins and creoles. Pidgins are consciously created
languages used for trade among groups with incompatible languages;
creoles are what you get when a pidgin becomes so pervasive that
it becomes a group's L1. John McWhorter in _The Power of Babel_
gives three examples of empirical tendencies of creoles, which a
Martian anthropologist could use to determine whether an arbitrary
human language is probably a creole:
1. Creoles tend toward isolating morphology.
2. Creoles tend toward lack of phonemic tone.
3. Creoles tend toward regular compounding, with few words that are
completely different from the sum of their parts (compare the
German compounds that end in |-loesung|).
> Natural languages are shaped by certain tendencies that can show us how
> the human mind works. When you create a language you can deliberately
> go against these, either subtly or by creating something that's
> impossible for a human to process without working it out on paper first.
When you first learned arithmetic, you probably had to work out multiple
digit addition on paper, carrying the 1. Nowadays, many of us can see a
pair of 3-digit numbers and add them mentally.
> The whole stack-syntax
> thing, for example, could result in sentences that require a speaker to
> remember the precise order of a larger 'stack' of words than most people can
> hold in their memory at one time.
But would somebody speaking a stack-based language as L1 be able to hold
a deeper stack than somebody speaking a stack-based language as L2?
Compare the language of most cultures to that of the Piraha~ to see how
language and culture influence one another.
However, some claim that the tool-use and -making centers of the brain
helped lead to the language centers, and the hierarchical construction
of an utterance resembles the hierarchical construction of a physical
tool. Languages that can be parsed into trees make sense in part
because the steps for making a tool can also be made into a tree.
> Yes, I'm aware that Fith is not a "human language", but remember with
> conlangs all that is required for it to be a human language is that its
> creator calls it one!
I'd suggest a different criterion: a human language is one that a
human being can learn to understand and speak in real time.