Peter Constable wrote:
> It matters little to me whether someone considers these types or
> sub-types. My point (back in Nov. 2001) was to identify the structural
> criteria I considered relevant for defining the classes:
>
> [1] - in some scripts, the (prototypically atomic) symbols correspond to
> phones / phonemes, and there is no graphical structure
> corresponding to
> other units of phonological structure
>
> [2] - in some scripts, the symbols correspond to syllables; apart from
> modifications to the symbols (on the order of strokes), there are no
> other graphical structures corresponding to other units of
> phonological
> structure
>
> [3] - in some scripts, the symbols correspond (prototypically) to
> phonemes,
> but there are also graphical structures that correspond
> (prototypically)
> to phonological syllables
>
> [4] - in some scripts (notably Chinese), the graphical units correspond to
> units of meaning, or can be analyzed into two graphic components
> representing a semantic value and a syllabic value

(I numbered your list for ease of reference.)

I my *very* humble opinion, there is something extremely unsatisfactory in
this classification, as well in the many other I heard of: it is the fact
that sound and meaning are considered on the same ground, as if the fact of
indicating meaning or not had anything to do with which sound segment are at
the basis of how sounds are represented.

I mean: I would ideally put sound and meaning on two orthogonal axes,
completely independent of each other. In other terms, I would favor a
*bi-dimensional* classification.

The result of trying to mix up sound and meaning on the same axis is that
your category [4] is not general enough to describe all logographic writing
systems. E.g., Egyptian does have graphic components which represent
semantic value, just like Chinese, and it does have graphic components which
represent sound, just like Chinese. However, while Chinese phonetic signs
can well be called "syllabic", Egyptian phonetic signs cannot: Egyptian is
nearly a "logographic abjad" (excuse my oversimplification: the presence of
bi- and tri-litteral phonetic signs does not allow to talk about an abjad,
of course).

In a similar bi-dimensional framework, you can place Chinese and Cherokee on
the same "coordinate" on the sound axis (i.e.: they are both "syllabaries"):
what makes them different is that they have a different "degree of
ideographicity" on the meaning axis.

(OK, that's was just my layman's two pence. Now I can sit in my atomic
shelter and wait for Prof. Daniels' explosion. :-)

--
Marco