Component reanalysis is a general process in the development of the script. The use of basic stroke types in modern printing styles is one aspect of this, so, you don't have to look very hard to find examples of the general type. Any Hanzi whose development can be related to pre-square-script forms will do. The oldest inscriptional materials, bone and shell, imposed different constraints on the kinds of basic script elements which were conveniently written. Likewise for bronze, pottery, silk and paper. For various reasons, the set of basic script elements can change, can be refined and adjusted. The cursive Seal shapes in Han times became regularized further in later calligraphic styles and then in block printing, and finally in computerized typography.

Assemblages of basic stroke types can naturally be analyzed into components. Just *how* this analysis is to be done is quite subjective sometimes, and so particular authorities must be referred to for specific component analyses. Even for common characters, e.g. [U+72ac] 'dog', there may be various analyses of earlier forms. The current writing of [U+72ac] 'dog' as [U+5927] 'big' + [U+4e36] 'dot' does not probably mean that anyone thought of a dog as a 'big man with a dot on his shoulder', though some big men with dots (chips?) on their shoulders may be rather dog-like.

Similarly, for animal characters such as [U+99ac] 'horse' and [U+9ce5] 'bird', the four dots of the feet are written as [U+706c], a combining form of [U+706b] 'fire'. This does not necessarily mean that anyone thought of the feet of various animals as fire-like, though perhaps some clawed animal feet are highly motile and rather rapacious, like fire. In the seal forms these four dots are lines, clearly legs/feet of various types (depending on the animal).

There may in some cases be philosophical musings at work behind the scenes in the "original" writing of a form in any given period, that is, in terms of the selection of graphical elements for combination based upon some understanding of the natural or scribal form trying to be conveyed. We cannot always recover what these might have been.

The class of "xiangxing" characters, sometimes called "pictographs" or "zoographs" (stylized pictographs with conventionalized pronunciation) are representations of natural objects, written using a set of basic script elements. Even at the earliest stage, this requires translation or reanalysis, in that the set of shapes identified in the natural object get converted to a set of stylized conventional graphical objects. Sometimes this conversion goes more smoothly than others. In the case of the Eastern Han text Shuo Wen, since it analyses all of the forms that it collects, we can compare these with earlier writings, equated on some basis or another, though usually not explicitly analyzed in the source (but since they are conventional stylizations, the analysis is often implicit).

Take as an interesting example [U+9f0e] 'cauldron' (= [U+2a502] = [U+9f11] = [U+25140] = [U+230ca] = [U+230d2] = [U+230b0] = [U+230a8] = [U+3ac0] = [U+251f4] = [U+2fcd]). The numerous encoded variants I've listed here illustrate how different people have reinterpreted the forms (old and new) in different ways.

In the Seal forms of DUAN Yucai's Shuo Wen (1815 edition of a text said to originate in 121AD) the character is explained as composed of [U+8c9e] over ([U+723f] + [U+7247]), (i.e. [U+24570], 'split [U+6728] wood'), though [U+8c9e] has been abbreviated to simply [U+76ee] 'eye' in the Seal writing. The character [U+8c9e] in Shuo Wen is related to 'divination', and perhaps specifically, to "divination by shells", but has an alternate and rather circular explanation in Shuo Wen, in which [U+9f0e] 'cauldron' abbreviated is *its* phonetic component. This doesn't correspond to its use in abbreviated form in [U+9f0e] 'cauldron', unless perhaps we understand that the divination by shells was done by heating and cracking them in fire, and that fire was built upon 'split wood', and that large seashells might have served as a primitive cooking vessel (cauldron). The results of divination are (you guessed it) "insight (into the future)", and so the writing with 'eye' rather than 'fire' or vessel can be further explained in terms of Han or Pre-Han philosophizing and reanalysis of the forms. When the abbreviation/reanalysis apparent in Shuo Wen's forms took place is unclear to me; it does however seem traditional, and is tied to the _Book of Changes_ philosophy said to be rooted in earliest times. In the _Book of Changes_, [U+9f0e] 'caudron' is written as [U+4DF1], which has [U+2632] 'fire (~eye)' over [U+2634] 'wood (~wind)'.

So, this is an example (perhaps not the clearest) of a character that (some might say) obviously derives from a picture of an ancient Chinese bronze cauldron: the 'eye' shape at the top is really the mouth of the vessel, and the 'split wood' shape at the bottom is the feet of the vessel. The manner in which the picture is realized in the script at any stage is however not completely obvious, and neither is the interpretation of its graphic elements. In fact, the familiar shapes of "eye" and "split wood" are perhaps false friends, used in this context
purely for their shape, and without any deeper significance, though in the Rorschach test of Eastern Han philosophy anyone can find meaningful components. Particular associations might drive the character to have one appearance rather than another, as we saw in the list of ding3 'cauldron' variants.

Another example of reanalysis over long stretches of time is U+8fb0, a very ancient chronometric sign of the class called dizhi "earthly branches". I wrote a whole book about this one several years ago (which is unfortunately out of print). In brief, the Eastern Han (Shuo Wen) explanation is that it is a picture of 'a man lying dead at the grassy base of a cliff' (presumably, he fell over the edge and plummeted to his death while contemplating the stars). The older pictures, however, are highly stylized scorpions, indicative of the parts of the constellation Scorpio with its prominent breast-star Antares. Stylization of the scorpion anatomy had been taken to such extremes over time that the original associations were all but completely lost by Han times.

Richard

On Monday, Jan 26, 2004, at 08:08 US/Pacific, ECOLING@aol.com wrote:
> A question on historical reanalysis of components of Han characters.
>
> Could some of you who are familiar with the history of Han CJK characters please give me good examples of characters for which (in the Unicode Han tables) we can clearly recognize standard components of the characters, but for which earlier forms of those same characters had a single unanalyzable glyph, for which the following description would be appropriate, or else (other example), for which earlier forms of those same characters had a *different* set of components whose visual result was somewhat similar.
>
> "an earlier historical unity was reanalyzed by later scribes; the later writing is an approximation of the earlier one, expressed using familiar elements"
>
> In other words, instances in which an unanalyzable character = glyph went out of use, replaced by a combination of components
> which persisted into later use?
>
> I am looking for a pedagogically useful illustration of this phenomenon.