Hallo conlangers!

On Monday 05 December 2011 16:04:24 Christophe Grandsire-Koevoets wrote:

> On 4 December 2011 23:16, Michael Everson <[log in to unmask]> wrote:
> > > I can think of plenty. But the main reason is: it's a conlang, and the
> > 
> > Indo-European classification is meant to use for natlangs exclusively.
> > 
> > Says who?
> 
> Every linguist except you.

Yes.  We have adequate ways of expressing that a particular
conlang is built on Indo-European material without just saying
"Indo-European language" as if it was a natlang.  Every linguist
except Michael Everson would say that

- Esperanto is an a posteriori artificial language based on
  Indo-European languages,
- Brithenig is a fictional Indo-European language.

The point is that Esperanto is *based* on Indo-European
languages and *not* an Indo-European language itself, as it did
not evolve from Proto-Indo-European in an unbroken continuity
of usage, and that in the case of Brithenig, that unbroken
continuity is an imagined one that never actually happened.
 
> > I'e been stuyding historical linguistics since I was 14

Then he ought to know better.  Someone mistakingly applying a
classification out of lack of knowledge errs not as badly as
someone who misapplies it against better knowledge.

> >     and I would
> > accept no such stricture.
>  
> Sorry, but I don't accept your authority on that subject. Too many people
> disagree with you.

Right.
 
> > Classification is a descriptive tool.
> 
> True. But using a tool for things it was not designed for can only result
> in chaos and unintended consequences.

At least, that is often the case, and the case here.
 
> > > Why? because it's based on diacronic linguistics:
> > Synchrony is also a tool that can be used in comparing languages.
> 
> Yes, but it results in different classifications entirely. The
> Indo-European family of languages is entirely and only based on diachrony.
> Typological comparison can lead to interesting classifications (like the
> Balkan Sprachbund, or SAE), but those are orthogonal to genetic
> affiliation. You're confusing different types of classifications, that
> exist on different levels and need to be kept separate.

I feel that Michael still confuses language families (such as
Indo-European) and language types (such as Standard Average
European).  An artificial language can be a member of the latter,
but not literally a member of the former.  It can be *based* on
material from a particular family, and if it has a fictional
history, it can be a *fictional* member of a particular family,
but in any case any historical linguist will insist that the
language is an artificial one and thus does not really belong
to any family.

> > > a language is Indo-European if and only if a set of regular sound
> > 
> > correspondences can be found between it and other Indo-European
> > languages, pointing out to a natural evolution from PIE via sound
> > changes.
> > 
> > A language can be Indo-European even if it does not participate in such
> > an evolution.
> 
> No it cannot. By definition. What you are trying to do is impose on us a
> different definition of "Indo-European" than the one that everybody else
> accepts. And we're disagreeing that it's a useful definition.

Michael is playing Humpty Dumpty here.  "Indo-European" has a
well-established definition, and that is a diachronic one.
A language is Indo-European by its ancestry, not by a set of
typological features.  Using the term in a different way only
leads to confusion.

Michael surely knows that if he has been studying historical
linguistics for years; that only makes things worse.
 
> > Does Volapük or Esperanto look Bantu, or Tai-Kadai? No, they do not.
> 
> No. But they hardly look Indo-European either. The resemblance is *very*
> superficial, and I say that as an Esperanto speaker.

Indeed.  All Indo-European languages I know well enough to say
something about them show traces of such Proto-Indo-European
patterns as ablaut, or suppletive (I mean, unrelated to the
singular case markers) plural case markers.  *Nothing* of this
is found in Esperanto.  All we have is a few thousand words
which resemble words from various Indo-European languages, but
do not correspond regularly to any of the source languages,
which in historical linguistics is taken as a sure sign of
borrowing, and borrowing is irrelevant to genealogical
classification.  Saying that Esperanto was an Indo-European
language is no way better than saying that English was a
Romance language.  It is the same thing.
 
> > > You can't do that with Interlingua, at least not enough to treat the
> > 
> > language as Indo-European.
> > 
> > If one were fool enough to try to use Interlingua as evidence for
> > linguistic reconstruction one would be making a mistake. To say that
> > Spanish is Indo-European but Interlingua is not is, well, to me quite
> > ridiculous.
> 
> To say that Interlingua is Indo-European *is* ridiculous.

It is at least misleading.  Interlingua is based on Romance
languages, but it is neither a Romance nor an Indo-European
language.
 
> > > You *can* do that with Brithenig, because that's how it was designed.
> > > In
> > 
> > that case, you can call it Romance and Indo-European, and I'd have no
> > problem with that (as long as you don't forget to mention it's
> > artificial). But you can't do that with Novial or Esperanto or
> > Interlingua. Call them "of Indo-European stock" if you want, but they
> > are not Indo-European languages.
> > 
> > An engineered/artistic/auxiliary/constructed language can most certainly
> > be described, based on its lexical and grammatical features, as
> > Indo-European if it shares lexical and grammatical features with
> > Indo-European languages.
> 
> No. It can be described as Indo-European if it can be proven through the
> comparative method (and only through it) that there is a set of regular
> sound changes between PIE and the language in question.

A well-crafted diachronic fictional language based on an Indo-
European antecedent may perhaps pass such a test; perhaps
Brithenig is done well enough to pass it.  Yet, the regular
sound changes Brithenig is based on never actually happened
in the real world.  Brithenig can thus be called a *fictional*
Indo-European language, the same way Hercule Poirot can be
called a fictional Belgian.  But a language such as Esperanto
won't pass the test - no way.
 
> Typological comparisons only lead to groupings of resemblance, not of
> affiliation. You can probably find someone somewhere in the world who has
> near identical features to mine, in terms of face, eye colour, skin
> complexion, posture, etc. That won't make this person related to me in any
> way unless DNA testing proves we're related.
> 
> Being in a language family is like a blood relationship: it exists only
> through a very strict set of criteria. Typological and lexical resemblance
> do *not* belong in those criteria. Those are orthogonal to the whole
> question.

Exactly.  Michael tries to redefine "Indo-European" as a set of
typological features, which it isn't.  Typological classification
certainly has its uses, but it is orthogonal to genealogical
classification.

> > Talossan is a marvellous Romance artlang. I have heard that it has been
> > criticized because it does not have regular sound changes from Latin or
> > Proto-Romance. Perhaps a narrow sort of expectation about what a Romance
> > artlang might be would lead someone to be disappointed with it. For my
> > part, I like the energy of Talossan. And of course it is Romance. And
> > that means that of course it is Indo-European.
> 
> Not if it doesn't have regular sound changes from Proto-Romance. It may be
> Romance-based, but it's not Romance.
> 
> And that doesn't change anything to its quality as a conlang. Why do you
> *have* to put things in places where they don't belong? Can't you just
> enjoy them for what they are, rather than for what you'd like them to be?

Well said.  Nobody denies that conlangs such as Esperanto or
Interlingua, or Brithenig, show *resemblances* to Indo-European
languages, yet, the resemblances are due to processes of a
different nature than those that have happened in Indo-European
natlangs.  All the languages mentioned are artificially created;
in the case of Brithenig, in a way that mimicks diachronic
processes of an ordinary kind (but it still only *mimicks* them),
in the case of Esperanto and Interlingua, not even that.
 
> [...]
> > It seems that the Euroclones are the result of language engineering done
> > on the basis of one or more languages that *already existed* and evolved
> > naturally from an earlier form.
> 
> The nature of the work is still quite different. What the translators did
> was the equivalent of sewing a new pocket to a pair of trousers. The
> trousers look a bit different, but they still are essentially the same
> trousers. Euroclone creators, on the other hand, take a look at different
> trousers (all similar in that they have two legs, a way to close the
> trousers at the waist, etc...), take fabric, cut it in ways similar to the
> trousers they've looked at, and sew the pieces into a new pair of trousers.
> This is quite a different thing to do. They didn't modify something that
> already existed, but made something new, inspired by something that already
> existed. Those are activities that cannot be compared (indeed, they require
> quite different skills).

An apt comparison.
 
> > It doesn't make "unu, do, tri, kvar, kvin, ses, sep, ok, naŭ, dek"
> > anything but obviously Indo-European in form. They don't have to be
> > "unsullied by an engineer" to be relatable to their Indo-European
> > heritage.
> 
> Eh... no. Looks can be deceptive. Some non-IE languages have borrowed the
> IE numbers wholesale. That didn't transform them suddenly into IE
> languages. The Esperanto numbers may look IE, but they work quite
> differently, in a way that is typologically more similar to how numbers in
> Chinese or Japanese work. Their form may have been inspired by IE
> languages, but that's it.

Exactly.  These numerals, like most if not all Esperanto words,
are of Indo-European origin, but that does not make Esperanto
an Indo-European language.

> > > The main stock was natural, and was only tampered with on a limited
> > 
> > scale.
> > 
> > From this I guess we may understand that you admire "natural"
> > development.
> 
> No I don't. What does it even have to *do* with what I've written?

Nothing.  There is nothing dismissing in saying that language
X cannot be a proper member of family Y because it is artificial.
It is just a statement of facts.

Of course, in a language like Brithenig, which purports to
represent a Romance language of a fictional Britain, it does
matter whether the imagined diachronic development is plausible
or not; but that has no bearing on the fact that it is not a
Romance language in the same way French or Italian are.  Its
membership in the Romance family remains qualified by the fact
that it is fictional.

> > That's fine, but it doesn't trump a description of Esperanto or
> > Interlingua as being Indo-European in nature.
> 
> No. The fact that they are diachronically unrelated to IE languages does.

Yes.
 
> [...]
> > >> Only diachronic conlangs based on natlangs, such as Brithenig, could
> > >> be
> > 
> > said to belong to the base language's family, but even that is not the
> > usual practice in linguistics; for instance, Wikipedia classifies
> > Brithenig as artificial, not as Indo-European. I'd say that the
> > Wikipedia is wrong here.
> > 
> > > I'd say it's right: Brithenig doesn't exist as a spoken language
> > > *here*.
> > 
> > So?
> 
> So it does not belong to the Indo-European language family here. It's a
> family of natlangs.

The fact that Brithenig now has a small community of real-world
users (but it is very small, and they don't use it in daily life)
has no bearing on the fact that it is artificial, and its
membership in the Romance language family is fictional.
 
> > > It doesn't belong in the *real* Indo-European tree.
> > 
> > That depends on the purpose of the classifier.
> 
> The purpose of the Indo-European family is to classify *natlangs* which are
> all descended from PIE and thus share a common ancestry. That is all there
> is to it. Typological and lexical resemblances don't enter into it. A
> language typologically unlike anything IE, but that could convincingly be
> related to PIE through regular sound changes would still be IE.

Yes; actually, there is no "Indo-European type" defined by
synchronic typological criteria.  Indo-European is not very
homogenous in this regard.  The typology of a language has no
bearing on its genealogical classification.
 
> [...]
> > I am compelled to point out that the fictional Ill Bethisad exists within
> > our universe.
> 
> No. *Descriptions* of Ill Bethisad exist in our universe. That's quite a
> different thing. A description of something is not the thing itself.

Exactly.  All that exists of Ill Bethisad in our universe is a
collection of texts, maps, grammar sketches etc. that *represent*
this fictional world, and of course the mental images of that
world those documents evoke in the minds of their readers.
 
> > > In the same way, my Narbonese is very much a Romance language, with
> > 
> > nothing out of the ordinary, but I would only put it in the IE tree if I
> > was talking of its fictional setting (also Ill Bethisad). *Here*, it's
> > primarily a conlang.
> > 
> > If its features are Indo-European, then it's a fictional Indo-European
> > language.
> 
> Yes, in that it's an IE language *in the fictional world*. That's the only
> thing "fictional IE language" means.

Just that.

> > > To me, that's like saying that my imaginary brother should be
> > > considered
> > 
> > a true member of my family, at the same level as my real life sister. It
> > makes no sense whatsoever.
> > 
> > No, it's like saying that your imaginary brother should be considered to
> > be an imaginary member of your family.
> 
> And what advantage do you get out of it?

To conclude:

All of us (probably even Michael Everson) agree that Brithenig
and Narbonosc are fictional languages.  We can say that they
are "fictional Romance languages", i. e. they are Romance
languages in a fictional setting.  In the real world, they are
artificial languages, just as Esperanto or Interlingua are.
It is certainly very useful to distinguish between entities
that exist in the real world and entities that exist in an
imaginary world.  Ignoring this distinction is delusionary.

Where we disagree is whether such artificial languages can be
rightly considered members of the Indo-European family.  All
of us except Michael Everson opine that they are not true
members of the Indo-European family - Brithenig and Narbonosc
are members only in a fictional world, while Esperanto and
Interlingua (which do not have a fictional background) are
not at all.  Period.

The genealogical classification of languages does not proceed
by filling in check-lists of typological features that are
thought to be "characteristics" of the family in question.
Nor does it operate by tallying how many words of a language
are from which language.  It operates by establish regular
sound correspondences between languages and reconstructing
the diachronic processes that shaped the member languages
of a family.

The membership of a language in a family is established by
demonstrating that a core of the lexicon and the morphology
of the language corresponds regularly with their counterparts
in other languages of the family.  Languages such as Esperanto
or Interlingua would fail such a test.  Brithenig or Narbonosc
would perhaps pass, but that would only mean that their authors
have done a good job imagining the fictional history of their
languages.  They would still be fictional languages, though
well-characterized ones.

--
... brought to you by the Weeping Elf
http://www.joerg-rhiemeier.de/Conlang/index.html
"Bêsel asa Êm, a Êm atha cvanthal a cvanth atha Êmel." - SiM 1:1