Saturday, September 6, 2008

Williamsburg Paints on eBay!

For anyone who is interested, I got a bunch of Williamsburgs recently at a great price and have just put them on eBay. Click on the big fancy sidebar link!

Wednesday, September 3, 2008

Color in Old English (Part 2)

A Norman elite troop rides out to meet King Harold in battle. From the Bayeux Tapestry, c. 1077

It's interesting that until after the Norman invasion, the word "color" did not exist in English. Earl Anderson says, "Many languages do not have a taxonym for 'color': an example often cited is Zuni[.]" In English, the concept of color "was not fully lexicalized until the thirteenth century, when colur~colour was borrowed as a loanword from French.... [but it] was well established in English by the time of Chaucer[.]" Old English hiw, "hue" in modern English, "can mean 'color' but more often means 'appearance' or 'shape,' and bleo sometimes means 'color' but more often has the same meaning as Latin discolor [having many colors]." Along with fah, these were the English words for color, here with Anderson's attempts at differentiating them:
hiw - shape, appearance, as well as color
bleo - complexion, appearance, and form as well as color
fah - stained, guilty, criminal, shining, and variegated as well as colored
The poet of The Phoenix, "in a sequence of two verses, uses all three terms for color:
Is se fugel fæger__forweard hiwe,
bleobrydgum fag__ymb þa breost foran.
[The beautiful bird is foremost in color (or: appearance or shape), colored (or: adorned) with variegated colors in front, around his breast.]"
Of course, while there was no taxonym in the language for "color," there was always the underlying concept. This Anderson says was a "covert category" (a term first described by Benjamin Whorf in 1937) - for instance, to modern English speakers moths and butterflies belong in one category, bees and wasps and hornets to another. "English has no terms for these residual categories, yet they exist in the language as nonlexicalized, covert categories. Color is such a category in many languages around the world."

So, what about the actual color names themselves? After exhaustive reconstructions of the color taxonomies of Proto-Indo-European, Greek, Latin, and Germanic, Anderson concludes that Old English had (like Germanic) six basic color terms:
BLACK - sweart, with the innovation of blæc as a rival synonym
WHITE - hwit
RED - read
GREEN - grene
YELLOW - geolo
GRAY - græg
Such a limited color system may feel alien to us - note the lack of "blue" (compare traditional Japanese with its category covering green-and-blue, or "grue"). But we must remember that this doesn't simply mean an all-inclusive color spectrum chopped up into smaller segments. More from Anderson's discussion of "focality as a perceptual aspect of color":

It is a misconception widely held that cross-linguistic variation in color vocabularies results from differences in the ways 'color space' is divided from one language to another. The chromatic spectrum, according to this view, is a continuum without breaks, but languages use basic color words to divide up the spectrum in different ways. If this were so, we should expect speakers of a given language to agree as to the 'boundaries' between one color and another. This was not the case, however, in tests in which speakers of American English and Dani were asked to identify an assortment of color chips as exemplifying one color or another. Neither the Americans nor the Dani could agree among themselves as to the boundaries between yellow and green or green and blue [Eleanor Heider, "Universals in Color Naming and Memory," Journal of Experimental Psychology 93, 1972]. This leads to the conclusion that the boundaries between one color and another are idiolectic [different from person to person]. As Mead observed seventy years earlier, 'Most of us have a very limited color vocabulary, and we differ hopelessly in our terminology as soon as we move away from a few sharply defined colors[.]'

The mention of William Mead is interesting because of the role his paper ("Color in Old English Poetry," Publications of the Modern Language Association of America 14, 1899) undoubtedly had in perpetuating the idea of Old English color words denoting an aspect of surface shine or brightness. Mead correctly notes the "comparatively small number of genuine color-words," and the tendency toward light-dark contrasts - pointing out that, interestingly, read, which he calls the "strongest and most effective" color, is not found once in Beowulf. But he asserts that the "remarkable fact about a great number of the Old English ... color-words, is that they are so indefinite in their application as scarcely to permit us to decide whether a color-effect is intended or not." After brief examinations of them, Mead sometimes gives in to the notion that they mean no real color at all - or that, perhaps as in the case of read, their occurences are the result of poetic convention (I'll eventually have to get back to this brightness/shininess business later). But Anderson goes on,

What Mead was getting at [in saying "we differ hopelessly in our terminology"], putting his native speaker intuition to good use, was the primacy of color focalities rather than color boundaries. The notion of color vocabulary as the arbitrary division of color space is disconfirmed by our own experiences.

... [F]or any given language, native speakers agree on the names and focal points of canonical colors. The hypothesis of focal colors was tested by Heider ... who, in the series of experiments alluded to earlier consulted twenty speakers of American English, and twenty-one speakers of Dani, a language that has only two basic color terms, mili for 'brightness' and for 'warm' colors and mola for 'darkness' and for 'cold colors' ... In one experiment, English and Dani informants were presented with color chips that exemplified both saturated or 'pure' colors and unsaturated colors, and were asked to recall the colors. In another, the Dani were asked to learn names for both saturated and unsaturated colors. In these tests, ... both Dani and English speakers named the saturated colors more rapidly, and gave these colors shorter names, compared with nonsaturated colors. Select saturated colors, therefore, are marked as 'focal,' and their focality influences cognition, for Dani speakers as well as for speakers of English, even though Dani has no chromatic color terms.

Anderson turns to a little puzzle from ancient Greek (italicized Greek words that follow are my transliterations): "Let us state the problem of color space versus focality another way. A basic color term is not just a larger 'color space' in whose territory are located specialized color terms with smaller semantic spaces. In the logic of the spatial metaphor, we would want to represent the relationships of the ancient Greek words erythros 'red,' phoinikos 'purple, Phoenician red, crimson,' and rhodos 'rose, rosy'" as in the following diagram:
.___________________________________________.
(...........................................)
(...............ERYTHROS "RED"..............)
(..._________......__________......______...)
(../phoinikos\..../porphyrios\..../rhodos\..)
(..\"purple" /....\"porphyry"/....\"rose"/..)
(___________________________________________)

"It is natural to think of erythros, the basic term for 'red,' as occupying a large semantic space in which phoinikos and rhodos are located as smaller spaces. We would probably want to think of the relationships among modern English terms for 'red' in the same way ...":
.___________________________________________.
(............._______.....________..........)
(.._________.( gules )...( russet ).______..)
( ( scarlet )......................( rust ) )
(..____________.......RED...........______..)
( ( vermillion ) ______..._________( roan ) )
(...............( lake ).( crimson )........)
(___________________________________________)

"Most native speakers of English would agree on the essential correctness of [the second diagram] as a way of representing the variety of terms for 'red.' By analogy we infer that [the first diagram] represents the relationship among ancient Greek terms for 'red,' but a diagnostic text demonstrates the inadequacy of the 'semantic space' metaphor as a way of representing the meanings of color terms." He adds that our first diagram does not hold up to the analysis, in Aristotle's Rhetoric (4th-century B.C.), of the phrase

'rosy-fingered dawn,' based on the verse formula rhodo-daktylos eos in the Iliad.... In a discussion of diction as an aspect of style, Aristotle notes his disagreement with the sophist Bryson, who says that there is no such thing as bad word choice because one synonym will do as well as another in cases where words have the same referent. Aristotle argues the contrary ... In the case of metaphor, well chosen words are

beautiful to the ear, to the understanding, to the eye or some other physical sense. It is better to say, for instance, 'rosy-fingered morn' (rhodo-daktylos eos) than 'crimson-fingered' (phoiniko-daktylos), or, worse still, 'red-fingered morn' (erythro-daktylos).

One could argue Aristotle was only thinking of the sound of the words [that is, that rhodo-, already part of a Homeric verse formula, was better for having only two syllables and not three like phoiniko- or erythro-] ... But this does not explain why Aristotle regards erythro- as the ugliest of all possible substitutions that he can think of, and this should give us pause, for Aristotle is a reliable witness, and erythros is the basic color term for 'red.' Come to think of it, whenever Aristotle, in Meteorologica, has occasion to refer to redness in the sky, he avoids erythros, using specialized terms such as phoinikos or porphyrios ... and he avoids metaphorically-derived terms such as aimatode chromata (blood-(red) colors) ... For Aristotle, erythros was not a color term that one used in a meteorological context, and for this reason, 'red-fingered dawn' was not available as a possible substitution for 'rosy-fingered dawn.' ... In their analyses of rainbow colors [which is a whole other discussion], Aristotle and Xenophon use phoinikos but not erythros, although a century after Aristotle (c. 260 B.C.), Poseidonios uses both phoinikos and erythros. Perhaps erythros in earlier Greek was focalized on a mineral color, like that of red bricks or iron ore or ocher. Whatever the precise solution to this problem might be, Aristotle's discussion of 'rosy-fingered dawn' illustrates that basic color terms must be thought of in terms of focalities; 'red' is not a genus that incorporates all supposed hyponyms.

Anderson's possible solution to this puzzle takes into account the material origin of basic color names. Also, "Languages maintain specialized subsets of color words that relate to particular domains within the material culture" - horses, cattle, and fabric being some of these subsets. Some languages of East Africa "have two dozen or more color terms that apply only to cattle. Colors applied only or mainly to horses are characteristic of the Altaic languages, viz. Kirgiz, Mongol, Uzbek, and Turkish.... Old French has thirty-nine words that apply only or mainly to animals, and twenty-four words that apply only to horses[.]" Sometimes a "specialized horse or animal color or a fabric color [can] become a basic color by sliding from a specialized semantic role upward to a Level I taxonomic slot." Old English purpur ("purple" in modern English) was originally used only of fabrics; Germanic *brunoz (English "brown" and French brun) was once used only of animals, and the "Germanic horse term *blanka- displaced Latin albus as the basic word for 'white' in Italian bianco, French blanc, and Spanish blanco."

These specialized terms are the crux of another puzzle, from the Old English Prognostics - "where the writer interprets the meaning of dreams about horses of different colors:
Gyf mon mete þæt he hwit hors hæbbe oððe on ride, þe byð weorðmynd.
Gyf him þince þæt he on blacum horse ride, þe byð his modes angnes.
Gyf him þince þæt he on readum horse ride, þe byð his goda wanigend.
Gyf him þince þæt he on fealawan horse ride, þæt byð god, oððe grægan, þæt byð god swefn.

[If someone dreams that he owns a white horse or rides on one, that means honor. If it seems to him that he rides on a black horse, that is anxiety of mind. If it seems to him that he rides on a red horse, that is waning of his property. If it seems to him that he rides on a fallow horse, that is good, or a gray one, that is a good dream.]"

Carole Biggam (in Grey in Old English, 1998), says Anderson, "first noticed the semantic puzzle of horse colors in this passage: it deviates from its Latin source, where the colors are albus (signifying a good event), niger (signifying anxiety), baius (signifying an expedition), castaneus ('dark red-brown' signifying a horrible business), and flavus (signifying a loss). Both the Latin and English versions have five horse colors, but the English version omits baius, a word for which Old English has no true counterpart,* and adds græg at the end" - perhaps because, to quote Biggam, "the translator appears not to believe that the list could have omitted a græg horse."

An attempt to match up the two groups of color words does not work:
Latin ......Old English
albus.......hwit
niger.......blacum
baius
castaneus...readum
flavus......fealawan
............grægan
The key to this puzzle, Anderson says, is the "role of Old English horse colors as classifiers" - that is, like red and white when speaking of wine, they don't function as general descriptors of color. The fact that some of these colors also happen to be basic-level color names is what is misleading. The mismatch between the Old English passage and its Latin source is "understood more readily when we realize that Old English has five canonical horse colors, used as classifiers": hwit (or also blanca, used as a substantive meaning "white horse"), blæc, read, fealu (or fealwe), and græg. "The Anglo-Saxon Prognosticator, in making his list of five horse colors, was not translating color descriptors from Latin; rather, given a list of Latin color terms that were more or less suggestive of the English horse-categories, he matched the Latin colors with English categories as best he could."

Which takes us back to the Bayeux Tapestry, now believed to have been constructed in England by English artists. In it Anderson sees "indirect confirmation" that the horse color words (tracing back to old horse words like *blanka- and *falwa-) functioned as classifiers, not just as descriptors:

The approximately 152 horses in the tapestry are all monochromatic, in five hues that range from light golden yellow to blackish brown, representing Old French sorel 'golden yellow,' fauve 'brownish yellow' (cognate of OE fealu), ferrant (from Latin ferron) 'iron gray' with a blue coloration, bai (from Latin badius) or baïart 'reddish brown,' and noiron 'blackish brown.' These colors are used, as well, as the proper names of horses: Sorel, Fauvel, Ferrant, Baiard, and Noiron. In several tapestry scenes, horses in groups of four are differentiated from each other by colors and in these cases the colors usually are sorel, fauve, bai, and noiron. A horse ferrant is seen less often, and then as part of a pair, partnered with either sorel or fauve. Color terms are used in Old French, then, as in Old English, to classify horses, not merely to describe them, and the semantic influence of French color-classifiers is so strong that it gives rise to the proper names of horses and it allows for the non-realistic coloring of horses in the Bayeux Tapestry.

__________
*[Anderson's note:] Alternatively, the Anglo-Saxon Prognosticator may have thought that OE readum covered the same semantic space of both baius and castaneus.