> Date: Sat, 7 Jun 2014 12:59:38 -0700
> From: [log in to unmask]
> Subject: Re: Definition of a-priori/a-posteriori
> To: [log in to unmask]

> On Jun 7, 2014, at 12:10 PM, R A Brown <[log in to unmask]> wrote:
> > Yes, indeed. But that doesn't conform to an important part
> > of Couturat and Leau's definition of _a priori_, i.e. that
> > the authors takes *no account* of natural languages.
> Most of the modern conlanging crew, I’d say, makes absolutely no reference to these original definitions.

Indeed. Prior to this discussion, I was not aware of these original definitions. If that makes me a hapless rube who doesn't know my conlanging history, so be it. However, I've been on this list for *quite* a while now, and my understanding of the terms "a priori" and "a posteriori" as they apply to conlanging (and my understanding is the more modern interpretation) came from *this list*. A quick noodle around the ether indicates this is now the prevalent sense. The horse is out of the barn, the genie is out of the bottle. If the narrower historical usages have been around for more than a century, that simply means that there are now two tiers of definition and one needs to be aware of the who, when, and why of what one is reading. What else is new? Taking the two original terms and applying them to language invention required its own modification of the definitions from the get-go (so actually, there are three tiers of definition), so I find it neither surprising nor jarring that the definitions have further modified as the nature of conlanging has itself evolved.

>The modern de facto definitions—that a priori and a posteriori deal with vocabulary sources, and, to an admittedly vague extant, >grammatical patterns—are much more useful, and apply to all conlang types. A conlang is either a priori, or a posteriori with the >language sources provided in parentheses (and the language source could be another conlang, as is the case with ámman îar).

Again, indeed. Calling Quenya "an a priori artlang with big nods to Finnish et al." plunks it right where it needs to be in my mind. If it were "an a posteriori artlang with big nods to Finnish et al.", I would anticipate something rather different from what, I surmise, it is (something far more Finnish-like, regardless of the recognizability or opacity of the Finnishesque vocabulary). And calling it an a priori *artlang* means it won't be standing next to Babm in the group photo.

> In short, the three features that have been brought up so far (or at least those I mention here) do not exclude each other—and, >indeed, they’re also more of a continuum than binary distinctions. For classification purposes, though, binary distinctions with an >added explanation are more useful than an essay describing exactly what a conlang is or isn’t with no summary.

> Date: Sat, 7 Jun 2014 20:10:30 +0100
> From: [log in to unmask]

> Yes - I have written before that lumping Quenya and
> Esperanto together as _a posteriori_ is IMO just as
> misguided as lumping Quenya and Babm together as _a priori_.
>   Indeed, it is my contention that while the _a priori_ ~ _a
> posteriori_ distinction is useful one for auxlangs, it
> doesn't make sense for artlangs which are artistic creations.

Well, "lumping" lends an air of supposed indiscriminate ridiculousness to it all.  "Lumping" a rhododendron and a pine tree together may seem misguided without the qualification that one is an evergreen *broadleaf* and the other is an evergreen *conifer*. So here, one has:

Wenedyk is an a posteriori artlang
Esperanto is an a posteriori auxlang
Quenya is an a priori artlang
Babm is an a priori auxlang

Jameld is a posteriori; Teonaht is a priori. Useful, understandable, easy (no lumping :) ).

"Well, then, where would you place ConlangX in that scheme?" One is always going to, and should, have that conversation when a language doesn't fit neatly into a binary classification system. Continua, they're a good thing. ;) And just because there may be some messy conversations like, "Well, LangY is predominantly a posteriori with some major a priori elements", or vice versa, doesn't mean we have to throw out the classification as uninstructive, useless, or meaningless.

> Yes, indeed. But that doesn't conform to an important part
> of Couturat and Leau's definition of _a priori_, i.e. that
> the authors takes *no account* of natural languages.

The point being that contemporary conlanging usage of the term no longer conforms to this narrow interpretation as the contemporary conlanging landscape has changed.

> The confusion arises IMO in that we are taking categories
> that were designed and have been used for more than a
> century for categorizing auxlangs and trying to apply them
> to a very different sort of conlang.

What trying? They *are* being applied usefully, understandably, and successfully to artlang and auxlang alike.  The most cursory websearch illustrates this.

To my mind, if, in writing for a contemporary conlanging audience, one employs "a priori" in its Couturat/Leau sense without any kind of qualification, explanation, or nod to Couturat/Leau, *that* is where the confusion will arise. Indeed, is that not where communication broke down in the Bālaybalan thread in the first place?

> > I think what we've got is two points on a continuum, but
> > divided differently into two parts depending on one's
> > point of view. For artlangers, the line is drawn closer
> > to the "a posteriori" end. The confusion results mainly
> > from not agreeing on where the line should be drawn.
> Or whether the line is applicable or meaningful for artlangs.

By Couturat and Leau's definition, probably not. But is that really a line anyway? More of a light switch, I think. On/Off. A language is either a priori or it isn't. Full stop. Not a continuum. Can't be "sorta pregnant". The modern de facto definitions are being actively employed, however, which says to me that this is a continuum the general conlanging community finds applicable and meaningful for both artlangs and auxlangs.  

> But I shall stick with the Gnoli triangle, unless I can be
> convincingly persuaded otherwise.

Personally, I don't see these as antithetical. The triangle is a wonderful monument to integrating messy continua. It even adds color to the mix so one can argue ad infinitum if LangZ is more cerulean or aquamarine or baby blue. And wasn't there talk a few years back about adding an extra dimension so that it became a pyramid, or even a cube? It seems particularly conscious of and adaptable to various criteria for classifying conlangs, and I would think another axis (a priori/a posteriori) would add to, not detract from, its depth.