Re: Tree writing [Was: Non-linear / full-2d writing systems?]

For a typological "ordinary" language -- say, SOV/Po/AN/GN, or SVO/Pr/NA/NG, 
or VSO/Pr/NA/NG -- translating the tree-writing into a linear sequence of 
phonemes is not especially difficult. Take the first of these. The reader 
speaks each child glyph before its parent, starting with the left child. 
This reader has it easy; one traversal order -- postorder -- handles pretty 
much every construction in the language. VSO has it equally easy. Speak each 
parent glyph before its children, reading the children from left to right. 
VOS, just as easy -- parent before children, children right to left.

(By habit I picture subject children left of object children. So I hadn't 
really been picturing *unordered* children. The order of glyphs on the page 
is fixed, the order of reading what is written depends on the language.)

<verb meaning="go">
<noun meaning="dog">
<adj meaning="blue"/>
</noun>
<adposition meaning="to">
<noun meaning="store">
<noun meaning="Bob">
</noun>
</adposition>
</verb>

Postorder: Blue dog Bob's store to go.
Preorder: Go dog blue to store of Bob.
Preorder, Reversed: Go to store of Bob dog blue.

Things are slightly more complicated for speakers of, say, English. Inorder 
traversal, in general, but postorder traversal for adjective-noun sequences. 
Still. not so bad. Amharic? Trickier... Welsh AuxSVO? Trickier still... But 
luckily we get to make up the languages, too. A big help to speakers in 
general would be some little indication on or near each glyph for its part 
of speech or the sort of phrase it heads. Say, one radical of the glyph 
indicates part of speech.

In general, though, learning how to traverse a tree in a specified order is 
something that's pretty easy to learn. I'm a logic teacher by trade, and the 
idea that a formula is just a certain traversal of a tree is something 
students can learn almost immediately, probably because on some level they 
*already* know this.

As for learning to write... well, drawing trees is cake compared to learning 
multiple thousands of glyphs.

If we don't wish to order the child nodes in some a priori manner -- say, 
subject left, object next, then adpositional phrases -- we could indicate 
semantic or grammatical role by some decoration upon the connections between 
them and their parent. Agent, straight line, Patient, wavy, Experiencer, 
crooked, etc. Or variations on the Lindisfarne Order if that's one's thing. 
This allows for participant languages that have a freer word order but 
inflect dependents. 

On 5/7/05, Henrik Theiling <theiling@...> wrote:
> 
> Hi!
> 
> Patrick Littell <puchitao@...> writes:
> >...
> > figure two languages with a slightly different noun-number-classifier 
> order
> > wouldn't put up too many barriers, but SOV and VSO languages would have 
> a
> > tough time "collaborating" in such a way.
> >
> > But a *tree* (or a more general directed acyclic graph) is independent 
> of
> > the order in which one traverses it.
> >...
> 
> Very interesting!
> 
> I myself have been struggling in the design of S2 mainly with the question
> of how to represent the underlying tree a) on paper, b) as a stream of
> phonemes. The design started with a tree, without a word order yet fixed.
> 
> My question is, how you can stop the design of a conlang at the tree
> level when in fact people want to talk and read/write? You excluded
> talking from your experiment, so I restrict my question to this: how
> to represent an unordered set of children of a node in the tree on a
> sheet of paper without ordering/serialising them? Or do you want
> to use something like tree depth for ordering?
> 
> I'm very curious about this, because, as I said, main work of S2 was
> about serialising trees.
> 
> **Henrik
> 
-- 
Patrick Littell
PHIL205: MWF 2:00-3:00, M 6:00-9:00
Voice Mail: ext 744
Spring 05 Office Hours: M 3:00-6:00