Re: Additonal features for CALS

On Wed, Jun 11, 2008 at 5:52 AM, Herman Miller <hmiller@...> wrote:

> JR wrote:
>
>> First ... congratulations and thank you Kaleissin, this site is really
>> great! Some thoughts/comments/suggestions:
>>
>
>  Would it be totally ridiculous to map langs based on the location of the
>> creator? I doubt we'd find any areal tendencies beyond the presence or
>> absence of (participating) conlangers themselves in different areas, but
>> you
>> never know. Well, it's just a thought ... I guess I really just miss the
>> maps, and even if those conlangs that are supposed to inhabit the Real
>> World
>> are mapped, most will still be excluded.
>>
>
> It's an interesting idea, but I think it makes more sense to do this page
> as if it's a collection of data on actual languages, with the place on the
> map based on where it is supposed to be spoken. A star map could be
> included, but languages like Ebisédian still would be hard to place (maybe
> the middles of oceans and the gaps between stars could be claimed by those
> with languages not spoken in our universe).
>
I agree it's preferable to map according to location of the language, but
wouldn't most langs be impossible to place, star map or no? Would people be
interested in assigning their lang to some empty area, even if that's not
really where it's supposed to be located? What would it mean? My main
objection to my idea is that I don't think it would be reflective of
anything interesting, unlike the WALS maps, which show areal trends and
demonstrate visually the locations of the langs in question (which is
particularly helpful for the more obscure ones). This would just be for
kicks. And actually there is already a site somewhere that puts conlangers
on a map, though without any reference to their langs' features.

 As for additional features, besides a priori/a posteriori, non-human
>> phonemes, based on, and length of active development (or rather, year
>> begun,
>> or to be more general, decade begun), some other possibilities are:
>>
>> 1) approximate lexicon size: not sure what values would be most useful,
>> but
>> something like 1-500/501-1000/1001-2500/2501-5000/over 5000, for example
>>
>
> For most conlangs this is really an attribute of the quality of
> documentation, not a property of the language. On the other hand, some
> conlangs have deliberately limited vocabulary, so this could be a binary
> feature (limited / unlimited).

 I agree that it's not a true linguistic feature in the way that most of the
others are, but neither are age of the conlang, or writing system type,
which is on WALS. They're all interesting though, and relevant enough IMHO.

 2) vowel harmony type: absent, backness, height, roundedness, ATR,
>> multiple,
>> other)
>> 3) vowel harmony pervasiveness: getting the values right would be tricky,
>> but something like absent/weak/moderate/strong/absolute, taking into
>> account
>> presence in both affixes (inflectional and derivational) and roots, and
>> whether loanwords undergo adaptation.
>> 4) allophone to phoneme ratio: not at all sure what values would be most
>> useful
>>
>
> This is probably one of the least constant features in my recent languages;
> I've gone through too many phonology revisions.
>
>  5) absence of common parts of speech:
>> nouns/verbs/adjectives/adverbs/multiple absent/all present (I suspect that
>> such a value set might obscure more tendencies than it would reveal, by
>> lumping too many langs into 'multiple', but I'm not sure which
>> combinations
>> would be best to single out. On second though, perhaps this could be split
>> into two features, one for nouns and verbs, and one for adjs and advs.)
>>
>
> Adjectives distinct from adverbs? yes/no
> Adjectives distinct from nouns?
> Adjectives distinct from verbs?
> Nouns distinct from verbs?
> Pre-/postpositions distinct from nouns/verbs?

Forgot about those ambipositions. You're asking quite different questions
here than I was, though, that seem to assume that if a class is missing,
it's because it's identifiably conflated with another. Sometimes this is the
case, as in, say, a lang that has a class of verbs called "adjectival" that
behave in a unique but still verbal way. But often, I think, the functions
of a certain class in one lang may in another be scattered rather messily
among multiple classes (not to mention among the inflectional or
derivational affixes/processes of other classes). In such cases I'd rather
not say that that class is "indistinct" from all the latter, but just that
the class is absent, period.

Also, how does one respond when one of the classes in a question is absent,
but not because of any conflation with the other class mentioned? For
example, in a lang with adjectival verbs, how do you answer if adjectives
are distinct from nouns? The subject of the question does not exist in the
first place, and either "yes" or "no" would be misleading. You could avoid
this by making only one feature, like this:

Adjectives: distinct class/conflated with verbs/c. w. nouns/mixed/other
(actually this would be a good place to stick in a "defective class" value
too, as some natlangs and at least one conlang have a class of adjectives
that is very small and closed)

Still though, if a class is absent I prefer to just say it's absent, rather
than indistinct from something else.

6) person distinctions (made anywhere in the language; mere syncretism in
>
>> certain instances is irrelevant): none/1,other/1,2,3/1,2,3,4/more than 4
>> persons distinguished/other. I'm not sure how relevant this is for most of
>> us, but I have one lang each to put into the second and penultimate
>> categories. Also, 4th person should not be taken to mean here some kind of
>> indefinite subject, as it sometimes is, but a category beyond 3rd that any
>> NP can belong to.
>>
>
> Could be interesting; Minza has 4th person pronouns.

Ah, that's good to see!


> 7) unusual number distinctions (made anywhere in the lang, but in a fairly
>
>> productive way): dual/trial/paucal/other/various combinations/none
>>
>> What say ye? I regret I probably wouldn't be able to do much actual
>> write-up, as my computer has just kicked the bucket and I don't expect to
>> be
>> able to replace it in the foreseeable future ... of course I'm on someone
>> else's at the moment.
>>
>
> There are all sorts of unusual features in conlangs, but some (e.g. a
> stack-based grammar, lack of voiced sounds) are confined to one or two
> examples. Self-segregating morphology? That could be a relevant feature for
> engelangs.
>
Don't even know what that is....

Josh

Replies