To think about archaeological decipherment is to leave behind many of the assumptions now current in the *humanities, such as: context determines meaning; form is an aesthetic category detached from reality; all interpretation is subjective; the text means whatever readers through history want it to mean; interpretation is a free play with no consequences. A document in an unknown script cannot be read for content or context: these may be guessed at (and in practice often are, repeatedly) but not assumed. The reading of an unidentified script can proceed in only one way: a formal methodology that captures statistical properties of recurrent elements of the corpus, leading to attempted correlations with the properties of known languages. Imagination, under these conditions, is usually a hindrance. The other side of this restricted mode of reading, confined to material features, is that interpretations can and must be tested against one another: there are better and worse answers to the question, what does this text say?, and in the end, perhaps, one and only one satisfactory answer. In decipherment it is not enough to say that the meaning of the text is what you make of it. The prospect of recovering lost archives and adding a new cultural province to the world’s stock of memories sets a goal for the activity.
Because a decipherment, unlike an opinion, can be deeply and demonstrably wrong, the annals are populated by cautionary examples. Consider Athanasius Kircher, *Jesuit, professor at the Collegio Romano and one of the seventeenth century’s most famous *polymaths. Primed by a philosophia perennis that treated earthly things as emblems and ciphers of celestial mysteries, Kircher saw in the hieroglyphic inscriptions covering the obelisks of Rome elaborate sermons in an allegorical mode: “Supramundane Osiris … flows down into the Osiris of the elemental world, Apis, beneficent Agathodemon, who distributes the power imparted by Osiris to all the members of the lower world,” translated Kircher from an inscription that present-day Egyptologists read as saying “Horus, strong bull, beloved of Maat … king of Upper and Lower Egypt,” and so forth. For Kircher, a hieroglyph was an image, and thereby a portent, an omen, a password, an initiation. Deciphering Egyptian writing required, first of all, putting oneself in the right frame of mind, and then calling up to memory the many esoteric Hellenistic texts that, for example, bestowed on the hippopotamus a moral character (bad) and saw promises of immortality in the outlines of the sun and moon. Knowledge of Coptic might smooth some edges. So inspired, one could ascend from the sensory images to a Platonic anamnesis of archetypes that brought the saving doctrines of Osiris once more to view.
A hundred and seventy years after Kircher’s Oedipus Aegyptiacus (Egyptian Oedipus) the efforts of Thomas Young and Jean-François Champollion had opened the way to reading the hieroglyphics effectively. They had the advantage of a key—the *Rosetta Stone, discovered in 1792, which bore one and the same inscription in hieroglyphic, demotic, and Greek—and of a method that treated meaning not as the precondition for interpretation but as its delayed reward. Rather than starting from analogies and assumed commonalities (the picture of the sun depicting the source of light, therefore standing for the origin of all beings, therefore denoting Osiris, the pharaoh’s celestial equivalent, and so forth), Young and Champollion sought to determine what value and function the graphs had in the different sentences where they occurred, allowing for the possibility that the writing system might use the same graph in several different capacities (as sign of an idea, as sound, as substitute for another sound, as silent classifier). Instead of beginning with the most exalted concepts of the Egyptian language (Osiris, the sun), they began with the transliteration of foreign names—a nearly nonconceptual use of the Egyptian characters. In the landscape of a language’s semantics, proper names are already at the very edges of meaning; foreign names even more so; foreign names transcribed are practically nonsense. But it was the non-Egyptian names “Ptolemaios” and “Kleopatra,” enclosed in cartouches to mark their pharaonic status, that opened the way to identifying the hieroglyphic signs that had been borrowed to write the sounds P, T, O, L, M, I, S, L, and A. From there it was only a short step to recognizing other names, including some of Egyptian origin: “Ramesis,” “Thutmosis.” From these beginnings, gradually the whole congeries of expedients that makes up the system of Egyptian writing came into view, with the range of “mimic,” “tropic,” and “phonetic” values that might variably be assigned to a character.
Admittedly, Egyptian hieroglyphic is a particularly complex writing system, which had a long history and a user base of professional scribes with no motive to reform it in the direction of simplicity. No one can say how long it would have taken to decode it without the Rosetta Stone’s parallel inscriptions in two forms of the unknown language and one in a known language, Greek. But the means of Champollion’s discovery are those common to all decipherments of unknown scripts. The underlying language, the principles of its writing system, the uniformity of the textual corpus—none of this could be assumed. Young and Champollion had to frame hypotheses and test them, discarding those that yielded nonsense (while allowing for the possibility that the standard for nonsense differs from one culture to another) and retaining those that made it possible to gain one more letter, one more name.
The methods are the same, that is, as those of cryptanalysts, with the circumstantial difference that the ancient scribes had no intention of disguising their meaning. Indeed many archaeological decipherers gained expertise in wartime cipher offices. Intentional *cryptography aims at leaving behind no trace of the regularities of the original language (the redundant formulas, the standard percentage of e’s or k’s, vestiges of syntax or accidence). The would-be *code breaker, in the words of a spy in the service of Louis XIV, “has no idea whether a certain sign stands for A or B or C or some other letter of the alphabet, whether it is a syllable or a word or perhaps a null sign; he hesitates everywhere, he doubts everything, and has nothing sure on which to fix his attention.” The very purpose of code clerks in foreign ministries, spy agencies, and armed forces is to ensure that their writings will be unreadable to the unauthorized—a series of genuinely arbitrary signs. But writing at any length with no meaning or pattern at all seems to be beyond human capabilities.
Since Champollion, translation from so-called dead languages has claimed a series of successes: Babylonian, through names and royal titles preserved in the otherwise unrelated language of modern Persian; Akkadian, working through Old Persian and relying on the verbal tics of self-important monarchs; Hittite, using Akkadian letter equivalents and a grammar recovered through Indo-European analogs; Brahmī and Tokharian through bilingual inscriptions and analogies with known scripts. After holding out for centuries, Mayan glyphs were finally decoded, proceeding through the usual first steps: numbers, dates, regnal names, and astronomical data.
What David Kahn calls “the most elegant, the most coolly rational, the most satisfying, and withal the most surprising” of archaeological decipherments, that of Linear B script, took place in the years 1945–52 after many false starts. The first tablets containing Linear B inscriptions were brought to light by Arthur Evans in 1900. Evans’s apparent reluctance to publish and two world wars delayed the appearance of significant numbers of Linear B texts from Knossos and Pylos until the early 1950s. No one could identify the language of the inscriptions, much less the structure of the writing. With little to go on, archaeologists experimented with solutions that involved analogies to Cypriot, Etruscan, even Basque. In 1952, building on work by Alice Kober, Michael Ventris published his solution, which has been received as definitive. Kober had recognized in the unknown graphs certain regularities that recalled the shape of words having a constant stem and varying case endings—a common word structure in Indo-European languages like Greek and Latin. Ventris, an architect, seems to have had the orderly mind and tolerance for detail required to document the behavior of each of eighty-seven signs, tabulating their co-occurrences and mutual exclusions. Phonetic equivalences were not yet in play; up until the publication of his solution Ventris had been skeptical of the suggestion by Kober and others that Linear B represented an earlier form of Greek. In discerning the organizing principles of the writing system, comparisons were useful, but only indirectly so. The large number of signs suggested a syllabary rather than an alphabet. Some pictograms and numerals had already been distinguished by E. L. Bennett. In addition to the sequences that stood in relation to one another recalling those of the inflected forms of a noun, there were apparent classifiers, marks that seemed most likely to preface a noun with a category such as “boy,” “girl,” “grain.” But the classifiers appeared also in the middle of words, as one would not expect from mere labels. Reasoning from the apparent catalog style of the tablets (a *genre well represented in the Bronze Age eastern Mediterranean), Ventris took the hypothetical step of supposing that some of the inflected modifiers that headed lists of commodities might be place names, and he experimented with inserting a few nearby Greek localities. A-mi-ni-so, representing later Greek Amnisos, Ko-no-so for Knossos, and Tu-li-so for Tulissos were at least plausible readings for those sequences and suggested substituting equivalent sounds wherever the same graphs occurred in the corpus. Ventris soon produced six or eight recognizable Greek words from the marks on the clay, and that was the turning of the corner. Phonetic equivalents for nearly all the remaining graphs followed, and by June 1952 it could be declared that Linear B was Greek and legible.
The elegance of Ventris’s solution is partly a product of its unusual history. Until the very end Ventris was not persuaded that the language of Linear B was Greek, and so his initial hypotheses were not jiggered to favor Greek (consciously or subconsciously). Analogies with other scripts were called in only rarely and with reservations. Ventris circulated his working notes among the ten or twenty researchers who had attempted to crack Linear B, whose assumptions varied greatly. Well before attempting to read words, regularities of a minimally semantic kind (numbers, evidence of inflection, categories appearing from layout) were amassed, until finally it was possible to begin ascending the “chimney” (as rock climbers might call it) of the place names and thence to attempt reconstructing nouns and verbs.
A bit of linguistics jargon has found its way into anthropology—the difference between “etic” and “emic” description. An “etic” description of behavior, say the organization of an annual temple feast in Taiwan, supplies an account of what goes on from the perspective of an outsider, perhaps a Dutch missionary with an inherited set of definitions (what is a ritual, what is devotion, what are favors, why do people attend festivals, and so on). An “emic” description, by contrast, attempts to clarify and explain what is going on in the terms available to the participants. (The linguistic analogy is to the difference between phonetics, which situates the sounds of a language on a putatively universal grid like the International Phonetic Alphabet, and phonemics, where every sound is charted in its relations of analogy and opposition to other sounds within the same language.) The project of decipherers like Ventris can be said to be quite literally “emic”: one must draw up the inventory of units first and then catalog their combinations and other ways of behaving. Supposing no language had been found that corresponded to Linear B, and thus no sound or meaning equivalents been assigned to its “words,” a complete description of the distribution of the signs in the spirit of Zellig Harris would nonetheless have been possible; and with some scripts it may be that that is as far as knowledge can go. The leap from forms to meanings may always be desired—in most eyes it is the definition of successful decipherment—but it is not always to be had.
Upon decipherment, mute signs begin to speak. We can overhear ancient scribes committing to writing the matters that they or their rulers thought were important: lists of commodities, tax rolls, tribute, kingly titles, and eventually dates, proclamations, laws, and treaties. The general message is, as Shelley put it while the Egyptian writings were still unreadable: “My name is Ozymandias, King of Kings! / Look on my works, ye Mighty, and despair!” Not every civilization writes down its myths, poems, epics, tales, fables, philosophies, and jokes. The great majority of recovered ancient writings consists of receipts, inventories, law contracts, and business letters; the Epic of Gilgamesh is the rare exception, preserved by a school curriculum that maintained it in the *canon of three successive languages. For recovering the daily life of ancient societies these dry records nonetheless have their poetry.
Haun Saussy
See also accounting; archivists; art of memory; bureaucracy; diplomats/spies; encrypting/decrypting; inscriptions; landscapes and cities; translating