FACTORING OUT WRITING FROM OTHER DURABLE VISUAL REPRESENTATION

A PICTURE, by conventional wisdom, is worth a thousand words. Given the choice of photographing or describing in words a Hawaiian sunset, few of us would disagree on the choice. We would also probably concur that, while both the photograph and the verbal description are representations of the event, the first is not language, while the second is. Moreover, if I should listen to your verbal description and then go home and write it down, my writing would be a second-order linguistic representation of the event, and a first-order (immediate) linguistic representation of your speech (see chapter 2).

How do we decide where to draw the line between durable visual representation which is linguistic and that which is not? How do we know whether a particular representation is a representation of an event or a representation of language? Are pictures and language historically related—ever? sometimes? necessarily?

To understand and answer these questions, let us consider samples of durable visual representation and compare their design, their content, and, to the extent we can determine it, their function. Our goal at this point is to familiarize ourselves with the sorts of visual representation whose linguistic status is problematic. (As we shall see in the section on the uses of written language, the structurally clear-cut cases of writing are interesting for a different set of reasons.)

Most of us have seen photographs of the cave paintings at Lascaux in Southern France and at Alta Mira in Spain. Less well known is that rock art has been found in widely diverse areas of the globe, including Australia (e.g., McCarthy, 1958; Berndt, 1964), North America (e.g., Smith and Turner, 1975; C. Grant, 1967), India (Brooks and Wakankar, 1976), and Africa (e.g., Willcox, 1963; Brentjes, 1969).¹ What is more, despite many differences in detail, there is a surprising degree of overlap in the subjects depicted and even (in some instances) in their manner of depiction.² Figures 6.1, 6.2, and 6.3 represent hunting scenes. The animals killed may differ from region to region, but the purpose in painting hunting scenes was probably similar: to induce success in subsequent hunts.

Questions of purpose are always difficult to resolve but especially so when some of our informants have been dead many thousand years.

Fig. 6.1 Horse Surrounded by Arrows, Lascaux Caves, France

Fig. 6.2 Bison Attacked by Hunters, Dharampuri, India (from Brooks and Wakankar, 1976:64)

Fig. 6.3 Man Attacking Buffalo, Concho County, Texas (from Jackson, 1938:281)

One of the most well reasoned analyses of the functions of rock art is that of Brooks and Wakankar (1976). Though written to examine the purposes of rock art in India, much of the theoretical framework is applicable to other regions as well. In their study, the authors identified twenty different “styles” of art, which they group into five historical periods. Styles are characterized by differences in subject matter, coloration, literalness of representation, and manner of drawing (e.g., heavy outline, x-ray style, solid figure). Based on this variation, Brooks and Wakankar hypothesize that alternative styles may serve different representational purposes, among which are religion, magic, secular records, symbolism (directional signals, taboo marks), and decoration. While many of these purposes may have been serious in character, it is also possible that these early artists laced their craft with humor. One mesolithic drawing depicts a nilgai antelope pregnant with an elephant (figure 6.4), which, according to the authors, “may illustrate a myth or reflect a sense of humor” (1976:84).

Fig. 6.4 Nilgai Antelope Pregnant with an Elephant, Hathi Tol Shelter, Raisen, India (from Brooks and Wakankar, 1976:84)

But what have rock paintings to do with writing? No one views these pictures as the equivalent of “written words.” In terms of the criteria of lexicon, syntax, and productivity identified in chapter 3 as necessary features of any language, we would have great difficulty identifying anything akin to rules of combination (syntax). Yet, with respect to lexicon and productivity, the case is not as clear. Consider the notion of a lexicon: Smith and Turner (1975), for example, present a “catalog” of petroglyphs from Southern California, in which they note similarities between glyphs, thereby hinting at the possibility of similar representational functions. Brooks and Wakankar go a step farther, illustrating striking similarities between the same motifs in rock painting and pottery (figure 6.5). Without some notion of syntax, it is hard to know how to speak of “productivity,” and yet it is clear that there was a great deal of combining of figures within a single vignette. Finally, there is the question of function. If Brooks and Wakankar are correct in the range of functions they attribute to rock art, then it appears that some of the purposes that art serves (e.g. secular records, religious ritual), language serves as well. How, then, do art and language differ in the ways in which they accomplish this task?

Fig. 6.5 Comparison of Motifs in Indian Shelter Painting and Pottery (from Brooks and Wakankar, 1976:32)

Let us pursue this question by comparing two types of visual records, both of which look like “art” to the untrained eye. Unlike the cave paintings, they both have an identifiable “order” to them (a crude syntax, if you will), and the motifs used are both more restricted in number and more stylized in presentation than the rock art; and yet, critical differences between the two types of records make us recognize one as art and the other as written language.

Fig. 6.6 Piero della Francesca, “Legend of the True Cross,” Church of San Francesco, Arezzo, Italy. Plan of the Fresco (from Venturi, 1954:56-57)

What are these differences? They are easier to talk about by first examining examples of both types of pictorial representation. In the first category we have, for instance, Piero della Francesca’s “Legend of the True Cross” (figures 6.6 and 6.7) and the Bayeux Tapestry (plate 1). The “Legend of the True Cross” is a fresco cycle, painted in the mid fifteenth century, in the choir of the church of San Francesco in Arezzo, Italy. The fresco panels, viewed in proper sequence (see figure 6.6), are based upon the legend of the cross found in Voragine’s Golden Legend, a late-thirteenth-century compendium of saints’ lives.

The fresco cycle begins with the death of Adam, over whose tomb his son Seth plants the branch of a tree given to him by the Archangel Michael. From the wood of this tree will be made the crucifix on which the Messiah is to die. King Solomon later has the tree cut down, but the Queen of Sheba recognizes the tree’s holy character. The scene then shifts to the conversion to Christianity of the Emperor Constantine. Legend has it that on the eve of battle against the tyrant Maxentius, an angel appeared to Constantine. The angel bore a sign of the cross on which was written In signo hoe confide et vinces (“In this sign believe and conquer”). And the story continues—Empress Helena, the mother of Constantine, sets about to find where Christ was crucified, and succeeds in unearthing the true cross. Chosroes, King of Persia, steals the cross, but the Emperor Heraclius succeeds in recovering it and returning it to Jerusalem.

Throughout the fresco cycle, the drawings themselves carry the story. The result is at once transparent and opaque. For the viewer already familiar with the legend, it is possible to “read” the chronicle of events without any written text. And yet, for the uninitiated who fails to recognize the Queen of Sheba or Empress Helena, the representational message is lost.

The Bayeux Tapestry is of the same genre as Piero’s fresco cycle in that it tells a story through a series of visual depictions of events. The “tapestry” (it is actually an elaborately embroidered roll of linen, not a woven tapestry) was made in the town of Bayeux, France, sometime during the twelfth century. “Reading” from left to right across more than 230 feet of “text”, we learn of the adventures of Harold, Earl of Wessex, and William, Duke of Normandy.

While Edward, King of England, is still on the throne, Harold crosses from England to France. The purpose of this trip is unclear. Harold may have gone to inform William that he, William, has been chosen as Edward’s successor; alternatively, Harold might be attempting to reclaim hostages being held in Normandy, or he might be on a pleasure trip. Harold eventually meets up with William and, after a series of adventures, returns to England. Edward dies, and Harold assumes the throne. On hearing of this, William has ships built and sails with his forces to England. William builds a castle at Hastings, provokes battle with the English, and defeats Harold.

Fig. 6.7 Panels from Top Left-hand Side of “Legend of the True Cross” (VIII, IX, XI)

VIII: The Finding of the Cross Empress Helena learns in Jerusalem that the only man to know the hiding place of the true cross is a man named Judas. When he refuses to reveal the place, Helena has Judas placed in a dry well. On the seventh day, Judas relents and leads Empress Helena to Mount Calvary, where three crosses have been buried.

IX: The Revealing of the True Cross Three crosses are unearthed, but it must be determined which is the True Cross. Each cross is lowered over the body of a youth who died that day. When the True Cross is brought down, the youth rises back to life.

XI: Heraclius Restoring the Cross to Jerusalem Chosroes, King of Persia, conquers Jerusalem and carries off the cross. The Roman Emperor Heraclius, incensed, defeats Chosroes in battle (panel X). Panel XI shows Heraclius reentering the city of Jerusalem, bearing the cross.

As we see in Plate 1, one striking difference between the “Legend of the True Cross” and the Bayeux Tapestry is that the tapestry adds Latin glosses to the pictorial story line. The viewer need not already know the sequence of events in order to interpret the tapestry. Nevertheless, we must not confuse the two forms of representation on the tapestry. The Latin language provides a written representation of the illustrations, which are a visual representation of the Norman Conquest. If the viewer does not read Latin, he must rely on prior knowledge of the events depicted in order to understand what is actually happening in the tapestry.

The “Legend of the True Cross” and the Bayeux Tapestry contrast with a second type of pictorial representation. This second type is exemplified by the eight extant pre-conquest manuscripts made in Southern Mexico. Some of the best known of these are the Codex Nuttall (housed in the British Museum’s Museum of Mankind in London), the Codex Seiden and the Codex Bodley (housed in the Bodleian Library in Oxford), and the Codices Becker I and II (now in Vienna’s Museum für Volkerkunde). These codices record Mixtec dynastic history from the late seventh century to the time of the Spanish conquest in the early sixteenth century. The manuscripts are made of treated animal hide that is painted and folded into sections. Mixtec ^writing” was one of four systems of durable visual representation to develop in pre-conquest Mesoamerica. Other systems are documented among the Maya, the Aztec, and the Zapotec. All four systems are historically linked, but as yet we know little about the actual relationships between them.

Plate 2 is taken from the Codex Nuttall.³ The codex, which was probably executed in the first half of the fourteenth century, tells the story of 8-Deer Tiger Claw (1011-1063 A.D.), the second ruler of the second dynasty of Tilantongo⁴ (I shall come back to the issue of his name below). Alfonso Caso has “translated” the original codex. As Caso explains, the manuscript is read by following the red guide lines from right to left.

What, then, is the difference between the pictorial-representations in the Piero fresco and the Bayeux Tapestry on the one hand, and the Mixtec codices on the other? A comparison of the Bayeux Tapestry and the Codex Nuttall makes the point clear.⁵

The Bayeux Tapestry and the Codex Nuttall bear a number of traits in common. Both are linear pictorial narratives of important events involving ruling families. Both order the events in such a way that the reader “knows” where to begin and where to end. Moreover, both employ some clearly identifiable pictorial conventions in the ways they represent their subject matter. In the Bayeux Tapestry, Englishmen have long hair and often wear mustaches, while the Normans are clean shaven. In the Codex Nuttall, stylized hills represent places, while arrows through the hills suggest that the places have been attacked (A.G. Miller. 1975:xv—see figure 6.8).

Fig. 6.8 Hill Pierced with Arrow Standing for Place Militarily Attacked. Codex Nuttall, p. 48 (from A.G. Miller, 1975:48)

But there are differences as well. The first that strikes the eye is that the Bayeux Tapestry depicts many people who are unnamed—foot-soldiers, cavalry, royal attendants who assist the major figures in their exploits. In the Codex Nuttall, only the key figures, who are all named, are depicted; there are “no genre figures, no spear holders, no bystanders” (M.E. Smith, 1973:21). The lack of unnamed auxiliary figures in the Codex Nuttall is an important indication that the codex is a linguistic document telling a story, rather than being an artistic rendering of an event, as is the Bayeux Tapestry. Moreover, since the names of the figures in the Codex Nuttall are embedded in the figures themselves (for example, g-Eagle is accompanied by her name sign, an eagle head with nine dots—see “Phoneticism and Pictorial Writing”), it is possible to “read” the document without going beyond the text. In the case of the Bayeux Tapestry, we must either know the story in advance or rely on the Latin glosses to make sense of the document. Without these, “we would know only that a group of clean-shaven people invaded by sea and defeated a group of people who wear moustaches” (M.E. Smith, 1973:173). The critical difference, though, is generally seen in the connection between the Codex Nuttall and the Mixtec language. Many of the pictures seem to stand for the things pictured (in Plate 2, the three pictures of children appearing under 5-Crocodile and 9-Eagle stand for the actual children the couple had), while others stand for ideas conjured up by pictures (in figure 6.8, the hill stands as a symbol for “place”). There is a third class of signs as well: those which unmistakably stand for sounds or words in the Mixtec language-so-called phoneticism. Caso offers the following example from Mixtec:

A particularly clear example of this is the place sign for the town of Coixtlahuaca as seen in figure 6.9. While Coixtlahuaca is a Nahuatl name, its Mixtec equivalent is Yodzo Coo (Serpent Plain).

I have intentionally begun this examination of written language somewhat informally in order to give a sense of the issues that arise in determining whether a particular durable visual representation is language. If we take cave painting, the Bayeux Tapestry, and the Codex Nuttall as representing a continuum from art towards language, then we might make the following observations about how written language and art differ:

Fig. 6.9 Place Sign of Coixtlahuaca (Mixtec Yadzo Coo). 1580 Map of Ixcatlan (from M.E. Smith, 1973:245)

The following discussion will examine the extent to which these observations dovetail with the criteria for writing posited in the linguistic literature.

CLASSIFICATION OF WRITTEN LANGUAGE

FORMAL CRITERIA

Treatises on writing have almost unanimously agreed that writing systems that represent the sound of spoken language, (i.e., alphabetic systems, syllabic systems, and “modified” hieroglyphic or ideographic systems like Chinese and Egyptian) historically derive from pictorial representations of experience rather than representations of language:

Plate 1 Bayeux Tapestry (Bayeux, France)
This portion of the tapestry shows Harold, then Earl of Wessex, leaving his manor house in Sussex and sailing for France (“Here Harold crossed the sea”).

Plate 2 Codex Nuttall (The British Museum)
Page 26 of this pre-conquest Mixtec manuscript tells part of the history of the ruler 8-Deer Tiger Claw. See page 158.

Whenever we are able to trace the origins of a phonetically based script, it seems that we ultimately find pictorial precursors. In many cases, as in the spread of the Phoenician script, the pictorial origins of the script are wholly camouflaged by the time the script is borrowed by another people.

There are also cases such as that of the Ogham script which appear to follow a logic divorced from anything pictorial. Ogham is a script found on stone inscriptions in the British Isles (especially Ireland) and mostly used to represent the old Celtic language. Monuments bearing the script date from the fourth century A.D., although the script probably dates back much farther. Scholars debate its use; most conclude it was a secret script used by Druids, while others contend it served more practical purposes. More to the point is the way in which the script’s elements are formed:

These twenty possibilities are shown in figure 6.10. Figure 6.11 illustrates how the marks were actually used:

Fig. 6.10 Symbols for Vowels and Consonants in the Ogham Script (from Jensen, 1970:579)

Fig. 6.11 Example of Ogham Inscription (ibid., p. 580)

No one knows where this highly stylized phonetically matched system of markings came from. As Jensen points out (ibid., pp.580-582), Ogham has been traced to Latin by some and to Germanic Runes by others. (Still others would derive Runes from Ogham.) Unlike most alphabets, however, the signs in Ogham clearly exhibit a simplicity of organization that wholly belies any earlier pictorial source.

Furthermore, not every durable visual scheme of representation that has a phonetic base is necessarily language. A case in point is the West African system of oroko (sometimes called aroko), which makes use of homonyms for the names of numbers to get a message across. Jensen cites the following case from Yoruba:

An example of oroko is shown in figure 6.12. Use of the so-called rebus principle has, historically, been a central means by which pictorial representations of things are replaced by linguistic representations of words. We must, however, be careful not to assume that the presence of rebus implies an incipient written language, or, as we clearly see in the case of oroko, that rebus can only derive from a representation of things. Six cowrie shells represent the number “six”—not cowrie shells.

Returning to the issue of an artistic-linguistic continuum: At which point do we attach the label “writing” (what some authors have called “true” or “real” writing) to this durable representation? How much of the representation must be phonetically grounded? Does Mixtec, which seems to have a restricted amount of phoneticism, qualify? In fact, must there be phoneticism at all?

We can formulate the problem in terms of the models of representation and linguistic representation presented in chapters 2 and 3. If a form of representation is to be deemed a language, it must contain discrete elements which are productively combinable by an identifiable set of rules. This is one criterion of written language.

A second criterion is that writing is a representation of speech rather than of experience. Using the terminology developed in chapter 2, speech would be a first-order representation of experience, while writing, being a representation of speech, would be a second-order representation. I have not, at this point, specified the type of representation of speech I have in mind (e.g., lexical, phonological). A third criterion requires that the second-order representation be based on sound; it must represent speech on the phonological rather that the lexical level.

Which of these criteria are necessary for distinguishing writing from art and which are arbitrarily imposed (much as the restriction of linguistic theory to the study of messages transmitted through the auditory-vocal channel is arbitrary)? Does the imposition of the second criterion or of the second and third (n.b. the third implies the second) place artificial limitations on what we consider to be written language (limitations that we do not apply to spoken language)? To answer this question, let us see how these criteria have been used in the literature on writing.

Fig. 6.12 West African Symbol Messages (Oroko). Photograph by V.P. Narracott; from the Pitt Rivers Museum, Oxford.

The literature contains proponents of all three positions. David Diringer, in distinguishing between “embryo-writing” and “writing proper,” argues that embryo writing is isolated, arbitrary, and unsystematic

Diringer seems to be distinguishing between unordered, nonconventionalized drawings on the one hand (e.g., in petroglyphs) and sequential picture-writing on the other (e.g., North American Indian pictographs):

Diringer thus adheres to the first criterion I enunciated but allows his definition of writing to include direct, first-order representations of experience.

Diringer is, however, clearly in the minority in admitting picture-writing to be “true” writing. A somewhat stricter criterion (the second criterion) is adopted by George Trager, who insists that writing must at least represent speech:

(Chao, 1968:101 assumes a similar position.) Under this definition, writing systems may represent either sounds (as with alphabets and syllabaries) or words. However, most linguists and historians of writing insist upon adding the third criterion: that writing must have a phonetic base:

Other proponents of the phonetic position include Gelb (1952), Bloomfield (1933:283), and Goody and Watt (1968:35).

Why this insistence upon a phonetic base for “true” writing? Is the criterion wholly arbitrary, or does it imply unspoken presuppositions about thr uses to which speakers (and writers) put their language?

STRUCTURAL CONSIDERATIONS

What Do the Symbols Represent?

In principle, there are five different types of entities that a durable visual symbol may represent:

In practice, though, syllables and sounds are more easily discerned than the rest. Many of us are familiar with syllabaries in which each symbol represents a syllable in the language (like Korean, or in older times, Sumerian and Middle Egyptian). Similarly, little needs to be said about alphabets like Roman, Cyrillic, or Greek, in which each symbol represents a single sound. Not unexpectedly, there are minor complications with such straightforward characterizations—alphabetic systems have been known to use one letter to represent two sounds (the x in lox representing /ks/), or two letters for one sound (the ph for /f/ in phone). But the principle is clear enough.

Somewhat hazier are the distinctions between the representations of things, ideas, and words. Chapter 2 discussed the problem of whether the signifié of a linguistic sign was a thing in the objective world or an idea in a person’s head. Saussure chose the latter, given our inherent difficulties in finding tangible referents for the sign unicorn, upside-down, or perhaps.

Traditionally, the attempt to distinguish between signs referring to things and those referring to ideas takes the form of a distinction between logograms and ideograms. Logograms are said to represent tangible entities such as objects, numerals, or proper names, while ideograms represent abstract notions associated with the item visually depicted. Thus, a picture of a leg would, logographically, stand for a bent leg, while the same symbol, ideographically, would depict the activity of running. The term ideographic might mean—but traditionally does not—the use of a symbol to represent a concept (in the Saussurian sense of signifié). Under such a definition, the symbol of a bent leg might, ideographically, stand for either the concept of a bent leg or the concept of someone running.

The one kind of representation not yet discussed is the use of a symbol to stand in place of a word. Our difficulty is in figuring out when the symbol is standing for either an object or an idea on the one hand, or a word on the other. About the only time we can be sure that a word (rather than an idea or a thing) is the referent is when the pictorial representation is being used because of the homonymie value of the word which names it (recall the example in figure 6.9 of the use of feathers to conjure up the homonym for plain in Mixtec). The problem, of course, is that once we understand that the object of representation is a word rather than a thing or an idea, then, necessarily, the real object of representation becomes the sound of that word rather than the word itself.

How Much of the System Uses Which Kind of Representation?

All of these issues aside, Is it necessary for the symbols to represent sounds (as with syllabaries or alphabets) in order to consider the system a language? There are three possible answers: either no phonological representation is necessary, some must be present, or the representations must entirely represent sounds. What are the empirical data we must account for, and how do we evaluate them?

From the outset we can eliminate wholly syllabic and alphabetic systems; they clearly fill the bill. The first problematic cases that come to mind are Chinese and Japanese. Japanese is a mixed type, combining the use of kanji (word signs borrowed from the Chinese) and kana (which are syllabic). This, of course, leads to the question of how to explain the writing of Chinese, which is far too complex to discuss here. Suffice it to say that Chinese writing is neither wholly logographic nor wholly ideographic in the traditional senses of the terms; indeed, it has a rather marked degree of phoneticism.⁶

What about intermediate cases such as Mixtec and Mayan, in which there is limited evidence of phoneticism in the visual representation? Is this degree of phoneticism “enough” to warrant our consideration of these durable schemes as written language?

Gelb epitomizes the traditional position that lack of “sufficient” phoneticism in a durable representation scheme is grounds for denying that scheme the status of language. It is on the phonetic issue that Gelb dismisses Central American systems such as Mixtec and Maya:

What is more,

Recall here the case of West African oroko (figure 6.12).

Gelb’s crowning argument belies a curious faith in the inevitability of progress in linguistic science. If the phenomena under consideration are indeed language, language scientists will recognize them as such:

A curious argument, to say the least. Even if we overlook the fact that the Maya languages we hear today have probably undergone considerable phonological change in the last 500 to 1,000 years since the documents in question were composed, we are still faced with problems. We would hardly be able to defend a thesis that decipherment of a phonetically based script is not only inevitable but inevitable in a predetermined period of time. Success in decipherment seems to entail a curious mix of genius and hard work; Gelb overlooks the former.

However, another factor that may be important is whether the people studying the system are open to finding phoneticism. Linda Scheie (1979) has argued persuasively that the failure of Maya scholars to find phoneticism may reflect, at least in part, their prior assumptions that it is not there to be found. The general disregard in which Bishop Landa’s Maya alphabet has been held (see Pagden, 1975), and the phonetic excesses (according to some) of Yuri Knorozovs work (e.g., 1958) has led many scholars to be highly sceptical of phonetic interpretations. Mayanists such as Tatiana Proskouriakoff (e.g., i960, 1973) and Heinrich Berlin (e.g., 1963) have developed

Crucial to the issue of decipherment is that

As Scheie demonstrates in her own work, once one drops one’s preconceptions about which factors are relevant in decipherment, one finds a great deal more phoneticism in Maya writing than had earlier been assumed.

And yet we should not attempt to press the issue of phoneticism in Maya writing too far. A great deal of Maya writing is pictographic (representing objects, ideas, or words—but not sounds). What is extremely important in deciphering Maya glyphs is understanding that depending upon the circumstances, the degree of phoneticism may be greater or smaller. The same glyph may on one occasion function as a logograph and on another as a phonetic sign.

What are the circumstances on which this “degree” of phoneticism depends? By the traditional hypothesis that signs become less pictorial and more phonetic as a representation scheme develops, we would tend to assume that the more phonetic signs in Maya appear in the historically later texts. Not so, says Scheie. Rather, she argues that the needs of the viewers of the text helped determine the type of visual representation used:

The situation is strongly reminiscent of the way in which the message of the Bible was made available in the European Middle Ages. Only priests had regular access to the written holy word. For the illiterate masses (the vast majority of the population), the stories of the Old and New Testaments were learned through sermons, morality plays, and, above all, through stained glass windows in churches.

The data from Maya cast serious doubts on the unexamined assumption that the higher the degree of phoneticism in a writing system, the more “linguistic” that scheme is in some absolute sense. But why have students of written language insisted on a phonetic base in the first place?

The primary assumption behind this insistence is that pictorially based representations are, in principle, unable to represent the full range of referents that can be represented in sounds. Speaking of pictographic means of communication, Goody and Watt state categorically that “however elaborately the system is developed, only a limited number of things can be said” (1968:35). The authors fail to mention, though, that there may be other considerations that make the superiority of phonetically based writing less certain.

Let us return to the aphorism at the beginning of this chapter: a picture is worth a thousand words. There are indeed cases in which a representation of an object or idea is far clearer in conveying a message than is a phonetic representation. Maps (as opposed to written directions) are good examples (at least for most people). The fact that do-it-yourself instructions almost invariably contain drawings—sometimes only drawings—suggests that sequences of events whose outcomes can be illustrated are sometimes more clearly represented by picturing rather than by describing events.

A second assumption is that pictorially based schemes that represent things, ideas, or words are necessarily iconic, and therefore inadequate to convey things, ideas, or words that are not easily pictured. Another alternative, of course, would be to have each symbol arbitrarily linked with its signifié. A good example of such a system is the arbitrary symbol languages used in teaching language to chimpanzees. Premack and Premack (e.g., 1972) had plastic forms arbitrarily paired with such abstract referents as if-then and different (see figure 6.13), and Duane Rumbaugh’s computer lexigrams are associated arbitrarily with referents like name-of, into, and eat (figure 6.14).

However, is this potential for limitless representation, which phoneticism provides, actually used? Or do many representations that qualify as “true” writing on the grounds of phoneticism actually accomplish no more communicative work than do systems that represent things or ideas? This question is reminiscent of the one posed in chapter 4 about the usefulness of the notion of an ideal speaker-listener. If our concern is with how people use language rather than with what they might do under idealized circumstances, then we want to consider what a functional classification of writing might look like.

Fig. 6.13 Examples of Signs Taught to Sarah the Chimpanzee. From “Teaching Language to an Ape” by Ann James Premack and David Premack, p.93. Copyright © 1972 by Scientific American, Inc. All rights reserved.

Fig. 6.14 Sample Lexigrams from Yerkish (from von Glasersfeld, Department of Psychology, University of Georgia, Athens, Georgia, 30601 and Yerkes Regional Primate Research Center of Emory University, Athens, Georgia.

FUNCTIONAL CONSIDERATIONS

Functional classifications of written language are barely in their infancy. Only one article (Basso, 1974) has appeared in the linguistic literature that directly broaches the question of which functions writing might serve. Basso’s classification is an overt attempt to mirror in writing the model of function as discursive competence that Hymes developed for spoken discourse (e.g., 1972). Basso suggests analyzing written language with respect to such categories as participants, form, topic, and function. If we were to consider the writing of letters, the participants would include at least one sender and one receiver; the form would take into consideration the quality of the stationery and of the grammar; the topic might range from “letter to the editor” to a letter to a friend; and the function could be anything from a letter of resignation to a thank-you note for a gift.

As we saw in chapter 4, though, the notion of “function” (and, in the present context, of “functional classification”) in which we are interested here takes us far beyond Basso’s initial distinctions. To get some idea of the range of issues a functional classification of writing might raise, let us briefly consider an example.

The Mnemonic Function of Writing

One of the reasons we write things down is so that we—or somebody else—will not forget what we have thought or observed or said. We write wills so that our descendants will not squabble over who gets the real estate and who gets the stocks. We keep records of how many seeds germinate in a botany experiment. We take notes in class lectures to help us prepare for final examinations.

If writing serves a mnemonic function (among others), then we can ask the following questions:

The answer to the first question is obviously yes. We tie strings around our fingers to remind us to pick up the dry cleaning; we leave out empty milk cartons to remind us to pick up more milk. Many societies rely on tally sticks to indicate how many items were processed in some way—they might make notches on a stick rather than writing in a record book “We baked seventeen loaves of bread.” Or consider the recurrent practice among high school students of assigning numbers to jokes, so that in the middle of a class someone will call out “Fourteen!” and the whole class will burst into laughter for reasons wholly unbeknownst to the teacher.

Even when the representation takes the form of phonetic writing, the amount of the message to be conjured up varies. This variance appears both when one looks at a single contemporary writing system (I might write bread or buy a loaf of Sunbeam white bread), as well as when one makes comparisons between systems or across time. Spanish orthography informs us at the beginning of a sentence—and then again at the end—that what we are reading is a question, while English only provides an interrogative marker at the end:

Spanish: @Cómo está?

English: How are you?

The Anglo-Saxon Chronicle regularly begins its entries with the word Her, meaning “in this place,” while today we would tend to use a phrase instead of the lexical abbreviation.

It can also be argued that the more ambiguities possible in a system of writing, the more this system serves as a cue for formulaic phrases rather than as a representation of novel thoughts or words. In Eric Havelock’s words,

Havelock argues that while syllabic scripts “tend to partake of the formulaic” (ibid.), alphabets permit greater variety in individual expression. In fact, he attempts to establish a causal connection between the development of the Greek alphabet and the emergence of Greek democracy (1976, 1978).

Similarly, Chadwick suggests a causal relationship between formal considerations and the extent to which Linear B was a mnemonic device rather than a record of individual inventions. Commenting upon the apparent limitation of writing to administrative matters, he asks,

Chadwick’s observations seem to follow Havelock’s hypotheses about ambiguity and invention: Linear B, like the later prealphabetic Greek script, which was borrowed from the Semites centuries later, was also syllabic.

Thus we see that written languages might be classified with respect to the extent of their literal rendering of what was thought or said (as opposed to conjuring up a precomposed “formula” of words or phrases). In some cases, the mnemonic classification may correlate with structurally based notions of classification (e.g., syllabic vs. alphabetic). Our task, in such cases, is to explain the correlation rather than assume the redundancy or irrelevance of one or another set of features. This situation is precisely parallel to the case study in chapter 5, in which we explored the reasons behind structural and functional properties of trade jargons and pidgin languages.

Our discussion of the varying scale of mnemonic functions is but one example of a functional category by which to classify written languages, yet there are probably dozens of others. In order to get some sense of which functional ranges of written language are possible, we need to probe more deeply into the uses of actual written languages.

USES OF WRITTEN LANGUAGE

Written language, I have suggested, has an existence distinct from that of spoken language, but that, like speech, admits of variation. Nonetheless, speech and writing can, in principle, be used for referring to the same semantic domains. Poetry and legal statutes, genealogies and shopping lists, lectures and cooking directions can be expressed in speech, writing, or both. As we shall see in this section, speaking has typically served a wider range of semantic functions than has writing. Societies are recognized as becoming increasingly literate as they make the functional range of written language more closely approximate that of speech. In fact, the traditional insistence on recognizing as “true writing” only those systems of permanent visual representation that have a phonetic base can be formulated in functional terms: “true writing” (in the Bloomfieldian sense) is that which can be used to fill the same semantic functions as spoken language and can do so with the same lexical and syntactic devices. That is, “true writing” can be minimally defined as an approximate transcription of spoken language,⁷ without denying that writing may fill additional functions as well. This section will consider this minimal definition of writing, historically examining the extent to which writing has indeed assumed the range of functions commonly attributed to speech.

LINGUISTIC FUNCTION AND THE CONCEPT OF LITERACY

Written language may come to approximate the semantic range of spoken language in one of two ways: either the absolute number of uses to which writing is put can expand, or the number of senders and receivers of written messages can increase—or, of course, both. In principle, there is no reason why a restricted literate class could not increase the range of functions of written language without the size of the literate population increasing as well. Historically, however, the two factors often have gone hand in hand. Therefore, an examination of functional expansion of written language needs to be accompanied by a study of changing patterns of literacy among the populace. That is, we shall need to understand the access people within a society have to the written word, the context of that encounter, and the extent of their proficiency.

Havelock (1976:44) has argued that the Greeks invented literacy since they adapted the Phoenician alphabet in such a way that the Greek language could be easily and unambiguously transcribed and deciphered by any Greek citizen. Gudschinsky’s definition of literacy is consonant with Havelocks in that it considers transcription and decipherment as the goals of literacy:

Such a definition of literacy is deceptively simple. If we agree with Hymes (1973) and Bernstein (1971) in the “inequality of speakers”—that (spoken) linguistic abilities of different individuals within a language community are not necessarily comparable—then Havelock or Gudschinsky’s approach to literacy would force us to conclude that literacy is at once a univocal and a multivocal term.

An alternative approach to literacy is taken by authors such as Schofield, who argues that no single definition of literacy exists which is universally applicable:

Intimately tied up with the question of individual proficiency in reading and writing are the social and performative issues of whether reading is done silently or aloud and whether it is an individual or a group activity. Our twentieth-century perspective on the functions of reading emphasizes both silence and solitude. Yet these are recent developments: “silent reading as we know it was very rare until the advent of printing—in the ancient world books were used mainly for reading aloud” (Goody and Watt, 1968:42). Steiner observes that

The tradition of oral reading to a group makes written messages available to a segment of the population larger than that which is literate (Schofield, 1968:312-313). Thus, even an accurate count of the literate population (by whatever criterion) will not necessarily reflect the uses of writing within the population.

The “oral” and public character of reading did not begin to give way until printing made possible the private use of books. As McLuhan points out:

Similarly, the role of punctuation changed. Nineteenth-century editors of Shakespeare added grammatical punctuation, thinking

More generally, printing brings with it a shift from a verbal to a visual approach to the page:

SELECTIVE USES OF WRITING

Questions of literacy lead to more general questions about the use of writing. In addition to asking who in the population is involved with written language, we want to know what the writing is used for.

To begin with, we should be aware that knowledge of the possibility of written language does not, in itself, ensure that such writing will ever be used. Hawkes and Woolley, citing Diakonoff, admit that

Even in societies in which writing has been introduced, literacy may initially have highly circumscribed uses. Goody and Watt argue that while Sumerian, Egyptian, Hittite, and Chinese societies were in some sense literate,

Similarly, Havelock (1976:21) speaks of “craft literacy” in Greece for at least 150 years after the adoption of the Phoenician alphabet. Only for the sixth and fifth centuries B.C. does Havelock attribute first semiliteracy, then recitation literacy, and finally scriptorial literacy to the general citizenry.

One of the clearest cases of restricted literacy, with respect to both communicative function and size of the literate population, is Linear B. John Chadwick (1958, 1959) has argued that in Mycenaean Greece, writing was used fairly extensively for administrative purposes, but apparently had little or no use outside of bureaucratic circles.⁸

The recording of a particular kind of communication in writing does not, in itself, guarantee that the population will encounter it in written form. In the case of the Old Testament, for example, the presence of a written text initially had little effect on the people:

The spread of literacy may be hampered by pragmatic considerations such as complexity of script (Goody and Watt, 1968:35), social stratification (Sumerian and Akkadian kings were illiterate—ibid., p.37), or geographic mobility. In Somali, for example,

Equally important is the strength of the existing oral tradition. Again in Somali,

EXPANSION AND DECLINE IN LITERACY

We have, thus far, been looking at literacy largely in static terms, that is, at selected uses of written language at given periods of time. It is appropriate, however, to “dynamicize” the issue of usage and ask, Under what conditions do the uses of writing spread in a society, and under what conditions do some (or all) uses of writing fall into disuse?

The answer to the question of growth is fairly well documented in the rise of the European middle classes in the eighteenth and nineteenth centuries (cf. Defoe, 1726) and in plans for modernization in emerging countries around the world (e.g., Unesco, 1973). In both instances, the motivations for increased use of writing (and increased literacy) have been largely economic. The proliferation of grammar books in England aided lower class children in acquiring the necessary skills of reading, writing, and arithmetic for improving their station by entering business. In newly developing countries, limited literacy may help a farmer ensure that he is not cheated by traders; more sophisticated written skills may open up a new range of occupational possibilities.

Once a sizable portion of a population becomes literate, written language can, in addition to its more distinctive purposes such as recording laws or serving as a cultural medium, begin to approximate some of the functions of spoken language. A lecturer can jot notes to himself before delivering a speech; friends may exchange letters just to “keep in touch” rather than waiting for a significant event to occur before writing; newspapers can provide the sensationalism that the previously nonliterate classes used to witness in public gatherings.

But the uses of written language may also diminish in number, either when an established writing system is adopted by a previously illiterate people or when the needs of an already literate society change. Just because written representation may historically progress from the potential encoding of a wider range of (phonological) messages to actually doing so, does not ensure that these functions will always be preserved. The data on literacy decline—either in numbers of people or in language function—are, by definition, more difficult to gather than data on the increase in literacy. Yet, disregarding instances in which the decline—or loss—of literacy was brought about by the fall of a civilization (as in the case of Linear B), we can point to at least one case of decline: that among members of cargo cults in coastal and island Melanesia.

The cargo cults, millenarian movements which have grown up in Melanesia at various times during the twentieth century, are Western adaptations of indigenous religions in which a primary purpose is the acquisition of “explicit socio-economic benefits for the practitioners” (Meggitt, 1968:301). Christian missionaries were, by comparison, wealthy and powerful and assured their audience of equality under Christianity. The native population was initially willing to embrace the new religion as a way of sharing in the missionaries’ wealth. Such religious participation included schooling:

From the beginning, however, literacy did not have the same function in the eyes of the Melanesians as it did in those of their Christian teachers:

Instead, the Melanesians looked upon written language as “one more of those inherently ambiguous models of communication with the supernatural with which they were already familiar” (ibid.).

Soon, however, the Melanesian faith in the missionaries—and in the benefits of literacy—was shattered, for despite their participation in mission rituals—and despite their sons’ learning to read and write—they remained materially poor. The result was a withdrawal from mission churches and schools and from uses of literacy that we would recognize. The native population did, however, retain a ritualized manipulation of writing as a way of understanding the supernatural (ibid., pp. 303-304). Meggitt concludes:

THE MINIMAL DEFINITION OF WRITING AS TRANSCRIPTION

We have been looking at the uses of written language, working under a minimal definition of writing as a transcription of spoken language in order to see variation in the degree to which so-called literate societies have utilized writing as an approximate permanent representation of the semantic domains covered in spoken language. However, as argued earlier, this representation can be, at best, approximate. Not only does writing underrepresent what is spoken (cf. Chao, 1968:110), but it also goes beyond its transcriptive function in that transmission of a message to an absent or unspecified interlocutor creates its own structural and stylistic constraints.

To what extent is this minimal definition of language empirically correct? Is writing ever “merely a way of recording language by visible marks,” or are there always features that structurally and stylistically set it apart from speech? The answer to this question varies historically: at one extreme lies classical Chinese writing, which has a lexicon, syntax, and diction quite distinct from that of even educated Mandarin speech (Brandt, 1944); at the other extreme lie the phonological transcription systems devised by linguists or missionaries for recording previously unwritten languages. At this same end of the spectrum we find the writings of newly “literate” people who have learned to write the alphabet and to sound out words but have not been trained in a “literary” tradition. Most written languages lie somewhere in between.

The history of written English is an interesting example of how the degree of correspondence between written and spoken language can be repeatedly redefined. If our records accurately reflect the history of writing in the English speaking world, we can identify an early period in which the written style (be it poetic, as in Beowulf; historical, as in the Anglo-Saxon Chronicles; or ecclesiastical, as in /Elfric’s Homilies) was clearly distinct from colloquial usage. Colloquial writing (writing approximating a transcription of everyday speech) begins to appear in Middle English (e.g., the Northern mystery plays, Chaucer) and, though not dominant, continues in Early Modern English (e.g., Shakespeare) and Modern English (e.g., Dickens—see Brook, 1970). But does the balance ever shift? Does colloquial speech ever dominate written style? This, I suggest, is precisely what has happened in contemporary American English.

THE FUNCTIONS OF WRITTEN LANGUAGE IN TWENTIETH-CENTURY AMERICA

It is always difficult to be objective when assessing the present—we must be suspicious of perennial Jeremiahs who would have us believe that our age is one of cultural or spiritual decay. Yet it is clear from the popular and academic literature of the past few years that many highly literate people are concerned with what they feel to be a growing illiteracy among school-age—and especially college-age—Americans (e.g., Sheils, 1975; G. Lyons, 1976). Some skeptics (e.g., Stewart, 1976) point out that complaints about declining literacy go back at least a century, leading us to question whether we are indeed falling under the sway of those “perennial Jeremiahs.” However, since so many universities and colleges are insisting on English composition or remedial reading courses, the problem merits our attention.

The role of the linguist in American higher education has always been distinct from that of the English composition teacher, partly because of the linguist’s preoccupation with speech, and partly because of the structuralist tradition of descriptivism. Yet, in the context of the present approach to writing, perhaps the time has come when the linguist and the pedagogue can join forces.

I have argued that the structures of a language are a function of the uses of that language, that societies differ in their linguistic needs, and that individuals differ in their linguistic abilities as well as needs. Furthermore, I have demonstrated that written language can be analyzed in this same framework. If we encounter a society that has a writing system that can, in principle, express a potentially infinite number of messages, but in which the potentially literate population encounters difficulty in using writing for many of these purposes, then we have two possible explanations. The first is that the functions of literacy themselves remain intact but either the quality of education is deficient or student motivation is flagging. The second is that the functional balance of literacy has shifted. Educators have generally chosen the first alternative. We shall, instead, look at the second.

Have we reason to believe that the functions of writing have changed —or have been changing—over the past century? Steiner, writing independently of the current pedagogical discussion, has argued that during the twentieth century the uses of written language have radically altered. In eighteenth- and nineteenth-century Europe, reading was a private encounter between the individual and the book:

Riesman describes roughly this same period in similar terms:

Riesman contrasts “the bookish education of these inner-directed men” (ibid.) with that of their other-directed successors who are “men molded as much by the mass media outside their education as by their schooling; men who are more public relations-minded than ambitious” (ibid., p.115). Similarly, Steiner speaks of the loss of privacy in encounters with the written word. Reading has become a public behavior to be practiced in a university library or in one’s office rather than at home. With the coming of the paperback, books have become quasi-disposable. Even highbrow paperbacks carry “no manifest sign of economic or cultural elitism. Mickey Spillane and Plato share the same book rack in the airport lounge or drug store” (Steiner, 1972:200). More important, print has lost much of its communicative value, as it is increasingly being replaced by tape deck or video:

To Steiner’s list of reasons for the retreating functions of print we can add several more. The first of these is the telephone; a vast amount of communication that previously was transmitted in writing can now be done in spoken language. A second factor is the growing degree of geographic mobility; the need for letter writing, if not obviated by a phone call, is likely to be met by a personal visit instead. A third (and admittedly more ephemeral) reason is a growing degree of informality. In a lecture several years ago, I proposed that the growth of a youth-dominated culture in post-World-War-II America was partially responsible for the decline in writing abilities. One of my colleagues countered that literacy was not declining—only the degree of formality was. He argued that it was now acceptable to allow our writing to mirror casual conversation—full of false starts, changes of direction, and so forth. If he is correct, then writing is indeed changing its function from being a reflective medium in which the sender must anticipate his receivers reaction in constructing his written message, to an idiosyncratic stream of consciousness in which the role of the receiver becomes all but irrelevant.

To say that the functions of writing may be changing is not, however, to give up on making the population “literate.” Rather, it is to suggest that before we can hope to solve an educational problem, we should know its systemic causes; and linguistic theory may prove helpful in unearthing that etiology.

PHONETICISM AND PICTORIAL WRITING

We have been looking now in some detail at the range of uses to which written language can be put, and comparing them with the functional range of speech. What remains is to explore how this heightened sensitivity to function assists us in solving some of the definitional problems posed in the beginning of this chapter. In particular, we want to ask whether functional considerations help in deciding whether the requirement that a durable representation have a phonetic base is an arbitrary criterion to impose on written language. Our discussion will center on the representation of proper names in two graphic representational systems that have sometimes been denied the status of “writing”: preconquest Central Mexican texts, and North American Indian texts.

THE REPRESENTATION OF PROPER NAMES

The argument hinges on the distinction between proper and common names and how such a distinction is recorded in durable representation. All human languages (see Hockett, 1963) have ways of distinguishing between individuals (proper names) and classes (common names). For example, Albert belongs to the class of men, Providence to the class of cities. Phonetically based systems of writing (alphabets or syllabaries) can represent this distinction unambiguously. In pictorially based systems of representation, the matter is more complex. Yet the pictorial component is not the only—or even the most important —factor to consider.

Paradigmatically, when one represents a proper name nonphonetically, one needs a way of distinguishing between the individual and the class of which it is a member. If I wish to represent my grandfather, I need to be sure that you don’t interpret the symbol as “man.” Assuming I am a good artist, and assuming further that you, as my audience, know what my grandfather looks like, the problem should be obviated.However, since writing is, par excellence, a means of communicating with an audience not present at the formulation of my message, I have no way of guaranteeing that the second assumption will hold.

It has traditionally been assumed (e.g., Gelb, 1952:103-104; Goody and Watt, 1963:35) that this problem of distinguishing between class and individual representation has been one of the primary motivating forces behind the introduction of phoneticism into earlier pictographic systems of representation. A man on a throne could be any king, but either the alphabetic “William” or pictures of a will and a yam, interpreted as rebus elements, make it clear that the referent is a specific individual. This formulation of the problem presupposes, I suggest, that the “meaning” of proper names is not transparent (this theory, while appropriate for much of modern western Europe, is highly inappropriate for many other societies). Naming dictionaries may tell us that Charles means “man of the common people” or that John means “Yahweh is gracious,”⁹ but in common usage these meanings are not obvious. Even names that may be transparent on reflection (Rio Grande, Pennsylvania, Miller, Baker) we tend to treat as arbitrary labels.

The semantics of naming is, however, quite different in other societies, including the two whose writing will be considered here: North American and Central Mexican Indians. Most of this discussion will deal with personal names, although many of the same points can be made about place names (toponyms) as well. We need to look at what structure these names had, how they were represented in writing, and what happened when additional or foreign names had to be represented. Before beginning, a few comments on the writing systems themselves are in order.

The use of pictography among the American Indians is well documented (see Mallery, 1893, reprinted 1972b). Isolated symbols or chains of symbols were used to keep records, including religious rituals, histories of past events, and censuses of the members of a community. Typical of these mnemonic records were the “winter counts,” such as the so-called Lone Dog’s Winter Count (figure 6.15). Pictographs were also used for sending novel messages, and after the coming of the whites some of these were even sent through the United States mails (Mallery, 1972b, Vol. L363). Because of their restricted usage and lack of a phonetic base, American Indian pictographs have not generally been regarded as actual writing (e.g., Gelb, 1952:21).

Fig. 6.15 Lone Dog’s Winter Count (from Mallery, 1972b, vol. L266)

On the other hand, Gelb notwithstanding, it is now commonly acknowledged that the Indians of Mesoamerica developed writing systems before the coming of the Spanish (Benson, i973:v). The best-known (and most sophisticated) of these was the system of Maya glyphs used in what is now the Yucatan, El Salvador, British Honduras, and Guatemala. A different system (although probably connected at some point historically with Maya writing) developed in Central Mexico. Sometimes known as “Aztec,” other times more generally as “Mixteca-Puebla” (Nicholson, 1973:1), it was used on monumental stelae and in painted codices to record historical and genealogical information, and ritual. Most of the representations are pictorial, having either iconic or ideographic import (A.G. Miller, i975:xiv-xv). In addition, they contain some phoneticism through rebus.

NORTH AMERICAN INDIAN “PICTURE WRITING”

Keeping this brief background in mind, let us return to the issue of proper names. In native North America, personal names generally connote an object or attribute with which the person so named is identified. Many of the names refer to animals and attributes or positions of those animals (Spotted-Horse, Sitting-Bull), while others refer to exploits or attributes of the individual himself (Caught-the-Enemy, Spotted-Face, Johnny-Belches-When-He-Eats). The common feature of these names is that they are easily represented graphically; sometimes the figure named is represented with the distinguishing characteristic (Roman-Nose, figure 6.16; Spotted-Face, figure 6.17), while other times the identifying object or characteristic is drawn separately and connected with the human figure by a line (Stabber, figure 6.18; Spotted-Elk, figure 6.19). In some cases the identifying characteristic appears by itself without the generic human figure (Red-Shirt, figure 6.20; Takes-the-Gun, figure 6.21). In all cases though, it would appear that the name is graphically representable and would be recognizable to a member of the tribe “reading” a text.

MIXTEC

Naming in preconquest Central Mexico is more complex, although, in principle, equally amenable to graphic representation. Among the ruling classes, some of whose exploits are described in the surviving Central Mexican codices, each individual is given two names. The first of these, a calendrical name, indicates the day on which the person was born. The Mesoamerican calendar had twenty day signs (e.g., alligator, deer, rabbit, and reed—see figure 6.22) and thirteen number indicators.¹⁰ Thus, one ruler’s calendar name is “8-Deer” (figure 6.23) and another’s “10-Rabbit” (figure 6.24). In addition, each figure has a personal name or nickname. Although we don’t know how the personal name was ascribed to the individual (M.E. Smith, 1973: 27), we do know that at least many of the names (those for which we have records) were graphically representable. As with the representation of names in North American Indian texts, the personal name may be incorporated into a generic figure (e.g., 5-Rain Smoking Mountain, figure 6.25, in which a smoking mountain is worn as a helmet) or indicated separately but attached to the generic figure by a line (e.g., 5-Rain Smoking Mountain portrayed in a different manuscript, figure 6.26). From some of the postconquest census cadasters (e.g., Códice de Santa Maria Asuncion, Codex Vergara), we know that common people as well had personal names that are graphically represented.

Fig. 6.16 Roman-Nose. Red-Census Cloud’s Census (from Mallery,1972b, Vol. 1:450)

Fig. 6.17 Spotted-Face. Red-Cloud’s Census (ibid., p.451)

THE QUESTION OF NECESSITY

Graphic representation of proper names in North and Central America appears to be adequate as long as the name to be represented is depictable and the pictorial representation is likely to be understood by the audience. However, what happens when either of these conditions is violated (as is likely when the script must be used to refer to foreigners)?

Fig. 6.18 Stabber. Red-Cloud's Census (ibid., p.448)

Fig. 6.19 Spotted-Elk. Red-Cloud’s Census (ibid., p.456)

Fig. 6.20 Red-Shirt. Red-Cloud's Census (ibid., p.448)

Fig 6.21 Takes-the-Gun. Red-Cloud’s Census (ibid., p.454)

Fig. 6.22 The Twenty Day Signs in Mixtec and Valley of Mexico (from Picture Writing from Ancient Southern Mexico by Mary Elizabeth Smith, pp.24-25. Copyright 1973 by the University of Oklahoma Press, Publishing Division of The University. Composed and Printed at Norman, Oklahoma, U.S.A., by The University of Oklahoma Press)

Fig. 6.23 8-Deer. Codex Nuttall, p. 52 (from A.G. Miller, 1975: 52)

Fig. 6.24 10-Rabbit. Codex Nuttall, p. 48 (ibid., p.48)

In North America, there was a consistent attempt to make foreign names fit the graphic mold. European names were regularly “translated” into the local native American language, and could, in theory, be represented by the same graphic principles as indigenous names. William Penn was known as Onas, the word for “feather-quill” in Mohawk; the second French governor of Canada, de Montmagny, was known (in English translation) as “great mountain,” a title applied to all subsequent Canadian governors (Mallery, 1972b, vol. 1:443). The only exception that Mallery reports to this “translation” process is the representation of a General Maynadier in a text as “many deer,” that is, with a graphic representation of near homonyms (figure 6.27). Mallery explains, though, that

That is not to say that the rebus principle was too sophisticated to be used in native North America—rebus is found in a number of nonliterate societies (e.g., the West African oroko—see Jensen, 1970:31) whose discursive graphic representation is not as complex as that of the North American Indians—the North American Indians simply did not use rebus.

Fig. 6.25 5-Rain Smoking Mountain. Codex Nuttall, p.56b (from Picture Writing from Southern Mexico by Mary Elizabeth Smith, p.221. Copyright 1973 by the University of Oklahoma Press, Publishing Division of The University. Composed and printed at Norman, Oklahoma, U.S.A., by The University of Oklahoma Press)

The case was quite different in Central Mexico. Although the extent to which phoneticism was used to represent proper names (through rebus) is unclear (Nicholson, 1973)—particularly with regard to how much of the phoneticism was indigenous and how much reflects Spanish influence—considerably more phonetic representation was used than in North America.

Phonetic representation of names was used either by itself or in conjunction with graphic sumbols. For example, in the Codex Mendoza, the place name Tochpan is represented by a rabbit (tochtli in Nahuatl) and a stylized depiction of a banner (pamitl or pantli) (figure 6.28). In other cases, both graphic and phonetic elements are used redundantly for either part or all of the name. So, (again in the Codex Mendoza) the place sign for Tzompanhuacan (figure 6.29), meaning “place of those who possess a skull-rack,” is denoted by both a (graphic) skull-rack (tzompantli) and “phonetically, by a banner, pantli, decorated with hair, tzontli, thus Tzon (m)(tli)-pan (tli) ” (Nicholson, 1973:13).

Fig. 6.26 5-Rain Smoking Mountain. Codex Becker I, p.6 (ibid.)

Fig. 6.27 Many-Deer. American-Horse’s Winter Count (from Mallery, 1972b, vol. 11:596)

Fig. 6.28 Place sign: Tochpan. Codex Mendoza, fol. 521* (from Nicholson, 1973:10)

Fig. 6.29 Place Sign: Tzompanhuacan. Codex Mendoza, fol. 35r (from Nicholson, 1973:12)

The use of phoneticism for a proper name, either by itself or redundantly with graphic representation, varied widely from scribe to scribe.¹¹ Moreover, although the data have not been properly analyzed by Mesoamerican specialists, it is probable that some genres (e.g., a general census) lend themselves to phoneticism more than others (e.g., the history of the exploits of a few well-known figures). Generally, it seems that “phonetic use of graphemes was considered to be less necessary in the case of especially well-known places and persons” (Nicholson, 1973:35) and “in the case of foreign names (places or persons), their phonetic rendering may have been standard” (ibid., PP. 35-36).

But was the introduction of phoneticism necessary in Central Mexico in a way not necessary in North America? We would like to answer this question, but like most causal questions dealing with language change, it cannot be settled decisively. Given the systems of naming in both societies—systems that presumably predated any written representation —phonetic representation for proper names was less necessary in either society that it would be in, say, modern Europe. That is, were Europe now developing a written tradition, there would be a much greater need for a phonetic component than in pre-European America, despite the fact that the same lexical category (proper names) is being represented in both instances. Thus, in formulating a minimal definition of written language (e.g., must its combination rules include rules of phonetic collocation), we need to consider the social context in which the written language is used.

Fig. 6.30 Householder Andrés Oyohuatl. Códice de Santa Maria Asuncion, No. 316 (from Nicholson, 1973:30)

Fig. 6.31 Householder Antonio Oyohual. Códice de Santa Maria Asuncion, No. 159 (ibid.)

Fig. 6.32 Householder Antonio Oyohual. Códice de Santa Maria Asuncion, No. 82 (ibid.)

One type of evidence militates against a simple solution to the question of whether phoneticism is necessary: the same name might sometimes be represented phonetically, and sometimes not. In the Codice de Santa Maria Asuncion, for example, the name Oyohuatl or Oyohual is denoted in three different ways: nonphonetically, with a picture of a bell (the Nahuatl for “bell” is oyohualli)— figure 6.30; wholly phonetically, with a picture of a road (Nahuatl ohtli) and a picture of night (Nahuatl yohualli), thus 0 (htli)yohualli— figure 6.31; and both methods used concurrently—figure 6.32. The use of rebus, either by itself or in conjunction with a direct graphic representation of the name, may indicate that the name itself is no longer recognized as being associated with “bell” (cf. John or Pennsylvania), or that the name is not indigenous to the area, and therefore the bell representation, though logical, will be novel to the audience. However, we have no evidence at the moment for either of these hypotheses.

Even if we cannot argue for necessity for phoneticism in Central Mexico, several factors make phoneticism more likely to develop in Central Mexico than in North America. It is quite possible that the Indians of Central Mexico had encountered and needed to refer in their writings to a larger number of foreigners (either Indian or European) than did their northern counterparts. The point becomes particularly important if the name of the foreign person (or place) is not easily translatable into graphic representation. After the Spanish conquest, the Mexicans regularly had to cope with European names (sometimes assuming them themselves), and it is not obvious that these could be handled as easily as the North American Indians had dealt with Penn or de Montmagny. Another important factor was potential influence from Maya writing; although the amount of phoneticism in Maya writing is still strongly debated (e.g., Thompson, 1959; Knorozov, 1958; Kelley, 1976), it is clear even to Thompson (1972) that many of the proper names in Maya were represented phonetically.

In comparing two pictorial based systems of representation, North American Indian and Central Mexican, we have seen that the same problems that we found in establishing empirical bounds on the lexicon of spoken languages apply to other representational schemes as well. Only by looking at the actual uses to which a community puts its language can we determine whether phoneticism is a necessary component of a representation system that bears the title language.