Peter T. Daniels wrote:

> > Finally, neither "女" nor "馬" can be further subdivided
> > in graphic elements conveying either a certain meaning or
> > a certain sound.
> >
> [...]
>
> How does that approach clarify the use of "grapheme" for any other
> script?

Well, also Latin letters such as "a"-"z" are atomic elements of the script,
as opposed to digraphs such as "ch", which are decomposable in a string of
elements such as "c" + "h".

Notice that the decomposition of "ch" in "c" + "h" is *not* based on the
visual aspect (as it would be, e.g., to decompose "B" in a vertical stem "I"
+ in a double arc "3"), but rather on functional characteristics: the "c"
and "h" part have autonomous functional properties in the writing system.

E.g., the uppercase/lowercase distinction is a good example of a fact which
is better described as a property of the atomic elements, and not of the
composed elements.

You can say that uppercase letter "A" maps to lowercase letter "a", but you
cannot say which "uppercase digraph" maps to digraph "ch": it can be either
"Ch" or "CH", i.e., the elements such as "c" and "h" have autonomous
uppercase mappings, regardless that they participate in a digraph or not.

The fact that certain properties apply only to non-analyzable elements,
seems a possible rationale for having a cover term for that kind of
elements. Not that I am defending the term "grapheme" per se: it can be a
different term, or even just a descriptive phrase like "non-analyzable
graphical element", but I think that we do have a concept to apply the term
to.

> Why aren't the seven basic brushstrokes the "graphemes" of Chinese?
> Aren't they much more the "atoms" of Chinese writing?

Yes, but that doesn't make sense from a functional point of view.

Analyzing a Chinese character down to strokes(*) is like analyzing letter
"B" in "I" + "3": it makes sense only if you are interested in the graphical
appearance of the characters, i.e. if you are dealing with calligraphy or
type design.

But if you are dealing with the "grammar" of the writing system, that
doesn't make sense. The "女" and "馬" elements seen in "媽" do have
recognizable functional roles in the Chinese writing system, while the 13
strokes in it do not.

Similarly, a phon*e* [b] is analyzable in a series of "traits"(*) such as
"plosive", "bilabial", "voiced", but this analysis makes no sense if you are
dealing with phon*eme* /b/.

(* BTW, Italian "tratto" translates both "stroke" and "trait": am I being
influenced by my mother tongue's lexicon?)

> This is one of my basic examples for why not to use the term.
>
> If you can't tell me what you want "graphemes" to do, I see no use for
> the term.

Let's say I just wish a term for the "terminal symbols" in the "grammar" of
a writing system, i.e. a name for things which are not splittable anymore.

_ Marco