/ A Response to Charles Sanft

From the point of view of an historian of the ancient Mediterranean world, it would be hard not to be largely in sympathy with the approach that Charles Sanft takes in this immensely stimulating paper. For many of the key points he makes are well established in the literature on the subject of literacy, and the use of writing more broadly, in the Greco-Roman and Egyptian spheres.[1] The similiarities between much of the work of the last three decades in the West and some aspects of Sanft’s observations about China are thus striking. But so too are some of the differences. Both seem to me potentially illuminating, and both in some ways might lead us to nuance some of Sanft’s analysis and even to look at the Chinese evidence a bit differently than he does.

A large part of Sanft’s discussion is devoted to separating reading and writing as activities, particularly to insisting that they have different learning processes and different social roles, even if few people in the end may really have learned only one or the other skill. He is aware that this point has been made with respect to the Near Eastern world, and he cites Michael Macdonald’s work in this connection. Raffaella Cribiore has in fact shown from the hundreds of surviving textual artifacts from literate education in Greco-Roman Egypt that writing was taught first, reading second.[2] It is true, as Sanft points out, that writing tends to produce vastly more evidence than reading, and we should be cautious about assuming that we can detect the beginnings of learning to read. Nonetheless, for learning Greek we should decouple writing and reading as elements of literacy, with writing primary and reading secondary. That is not to say that reading was hard(er), but it was apparently not the first thing learned. That means that the approach taken by Sanft, which supposes more readers than writers, might not work for the classical side. Such a difference between Chinese and Greek is entirely plausible; it was surely easier to learn to write a 24-character alphabet than to master even a shorter list of signs than the five thousand required of the Chinese professional scribe.[3] But Sanft’s description of the separate processes of learning agrees with what we see in the Greco-Roman world, as does a stress on memorization, which was surely central to all education before the most recent times.

In any event, Greek terms in papyrus documents such as “unlettered” (agrammatos) and “because he does not know letters” consistently refer to writing ability in almost all cases. It is the inability to sign at the bottom of an agreement that is at stake. That the next step up the ladder is “writing slowly,” being barely able to write one’s name and a verb of subscription, makes that clear. That writing and reading were not decoupled, however, is shown by the fact that there is some evidence for people having contracts read aloud to them before someone else signs for them. But we do not know that such people could not read at all, nor do we know that they were telling the truth in saying they could not write at all. Could they have scrawled their name as a graffito, or read someone else’s graffito? Did they just not write well enough to want to expose their skills before others?

It is entirely possible that the processes of learning in China were more like those of traditional Egyptian scribes, with the hundreds of hieroglyphs to be mastered, than like those of people learning to write Greek or Latin. Even with the cursive form of the hieroglyphs that we find in Demotic, the number of signs was large. Phonetic value and semantic meaning were both expressed. It is perhaps not coincidental that Egyptian has a much stronger conception of professional scribes, as a separate social group, than does the Greco-Roman world, where even using the term “scribe” strikes me as misleading and prone to conceal a wide range of skill levels and degrees of professionalization. Few of those writers of Greek and Latin commonly referred to in modern literature as such would have had anything like the level of skill of Chinese scribes.

A second core part of Sanft’s argument has to do with the roles of speaking and listening in constituting literate communities. Much has been made of orality in classical studies for more than a half century, starting from the limited roles of writing in classical Greece and including more recently important studies on the place of oral recitation of literature under the Roman empire. The literature on the subject is enormous. William Johnson’s important book on reading culture under the Roman empire has brought the evidence for this phenomenon into the discussion.[4] But his subtitle is important: “a study of elite communities.” It was mostly the highly educated elite that engaged in reading literature aloud. As with high literature in China, aural reception of Greek and Roman literature required significant knowledge for a full understanding of many texts. It is by no means to be excluded that a Greek wheelwright would have felt much the same as the man in the story from Master Zhuang.

There is no reason to suppose that this practice of oral recitation extended uniformly throughout daily life and the practical uses of writing. It is simply not true that the ancients did not routinely engage in silent reading.[5] But that does not mean that reading aloud was unimportant in such contexts. There is, for example, evidence that family letters were routinely read aloud and even translated from one language into another in oral presentation for the convenience of those who did not read. “You who are reading the letter, whoever you are, make an effort and translate to the women what is written in this letter, and tell them” says the writer of one letter.[6] In posting government orders in public, the Roman administration did not anticipate that everyone could read them, but it probably did suppose that there was always someone who could read such notices aloud to others within a “circle of trust.”[7] As Sanft says, focusing on percentages of individuals who can read and write is not productive as a means of understanding how reading and writing worked in ancient societies.

Another distinction stressed by Sanft has to do with utilitarian vs. non-utilitarian motives in learning letters. This seems problematic to me. Greek education of the higher type certainly formed people for social roles; is this more or less utilitarian than having fun? Perhaps more, for homo politicus or homo economicus, but the line does not seem clear to me; it might be seen as on the same spectrum as the nomadic self-expression and entertainment for which Sanft cites Macdonald’s studies of graffiti. (This, it should be noted, has been challenged in Karen Stern’s recent book on graffiti written by Jews, which thinks that some of Macdonald’s ludic graffiti were in fact religious in character, expressions of devotion.[8]) One may certainly see prestige as a motivation for learning to read without denying a utilitarian character. The lower forms of writing competence were surely for the most part more crassly utilitarian, in the economic sense, than the higher. But the motivation is as appropriate to the class of individual doing the learning as prestige is for the member of the governing stratum.

This sort of class distinction, and gender distinction as well, is also relevant when one thinks about the use of dictation. For those who could not write or who wrote only haltingly because they came from unprivileged backgrounds, dictating was a necessary means of participation in the culture of writing.[9] Sanft says, “It might seem self-evident to modern literates that those who can write do so.” But this is in no way self-evident. Many people in antiquity who could write perfectly well nonetheless dictated, by wish rather than necessity, in the same way that modern executives do, or at least used to. Dictating in this case is a sign of privilege, of being too busy or (self-)important to do the actual writing oneself. Even if writing did not require great effort, a provincial governor, for example, would at most subscribe a greeting below a letter actually penned by a secretary. At the same time, however, there is plenty of evidence for a high level of actual writing by members of the propertied elite in the course of managing their estates.

Such flexibility in the use of dictation is just one piece of an overall picture of a literate community. Rather than seeing a necessary choice between analyzing literacy at the individual or group level, I would prefer to see these as inextricably joined. Of course Sanft is aware, as he says, that communities are made up of individuals. Writing is a complex social system, in which many people who cannot write nonetheless use documents all the time, through a network of trusted helpers. There has been much study of this on the classical side. An interesting example that has figured in my own work recently is the funeral workers, or nekrotaphoi, of Egypt’s Great Oasis, from whom we have a mass of family papers dating to the third and fourth centuries CE. Not a single one of them was able to write, as far as we can tell from their documents. But they had no trouble in supplying themselves with people who could write, who produce their documents, sign them on their behalf, witness to them, and even travel to Alexandria to register them in official archives.[10] Some of these were members of the local elite. And the undertakers kept papers about status and property for decades, knowing their potential use in litigation. They were full members of literate society, even if they could not write at all. The community, or society, was in fact a unit not only of transmission and consumption of writing, but of its production.

I have used the terms community and society interchangeably, rather against Sanft’s proposed distinction. This terminological difference does not seem to me very useful. An individual is not a member of one or the other, but of both. For analytic purposes, both smaller communities and larger societies are useful and necessary, and both played a role in making literacy a phenomenon that transcended personal capabilities. The concept of a textual community is another matter. In Brian Stock’s usage, which Sanft cites, it has some useful meaning, but the idea has been much more widely used in recent decades, and it runs a risk of reifying what are only transient constellations of people as if they were essential bodies. Early Christian studies have been particularly culpable of assuming that almost any text must reflect a community, something for which I can see no evidence. In any case, individual, community, and society are all relevant descriptions of particular situations and must be seen in connection with one another.

Finds of everyday writing are often not easy to interpret for the size of the reading community, because of the mediating role of nodal individuals with skills. But papyrology (and graffiti) support the view that the number of people with skills was not minuscule; and almost everybody was connected to and through these people. The Eastern Desert of Egypt furnishes a precise parallel to the border military posts mentioned by Sanft. The forts along the roads leading from the Nile valley to the Red Sea ports and the desert quarries and mines have yielded thousands of texts on potsherds (ostraka). Most of these are letters and documents, but along with them have come some (para)literary texts, ranging from apparently original compositions of an obscene nature to exercises suggesting that people were indeed taking the opportunity provided by free time and boredom, in conjunction with literate companions, as a spur to learning to write and read.[11] The thousands of personal letters written and received in these posts show that there were plenty of literate persons present, even if their spelling and grammar would win no prizes, just as Sanft points out was the case on the Han border regions. One can in fact parallel such small, remote outposts with nothing to do but “manning watchtowers and waiting for something to happen” in Egypt as far back as the Old Kingdom. If the papyrus texts that probably made up a large part of the writing in circulation in this milieu had survived, and not only the ostraka, we would no doubt have an even richer view of the place of writing in these tiny communities.


    1. Much of the literature on learning letters in Greek and Latin over the past few decades has been stimulated by William Harris’s landmark book Ancient Literacy (Cambridge, Mass.: Harvard University Press, 1989), which distilled for the ancient Mediterranean world many of the approaches against which Sanft is reacting.return to text

    2. R. Cribiore, Writing, Teachers, and Students in Graeco-Roman Egypt (American Studies in Papyrology 36; Atlanta: Scholars Press, 1996).return to text

    3. This is not entirely ignorant supposition, as I have watched my grandson learning to make Chinese characters.return to text

    4. W. A. Johnson, Readers and Reading Culture in the High Roman Empire: A Study of Elite Communities (New York: Oxford University Press, 2010).return to text

    5. See, e.g., A. K. Gavilov, “Techniques of Reading in Classical Antiquity,” Classical Quarterly n.s. 47 (1997): 56–73.return to text

    6. A. Bülow-Jacobsen, Papyri Graecae Haunienses, fasc. 2 (Bonn: Habelt, 1981), 12–17.return to text

    7. A. E. Hanson, “Ancient Illiteracy,” in Literacy in the Roman World (Ann Arbor: Journal of Roman Archaeology, 1991), 159–98, discusses this phenomenon and much else concerning the functioning of literate communities.return to text

    8. Karen B. Stern, Writing on the Wall: Graffiti and the Forgotten Jews of Antiquity (Princeton, N.J.: Princeton University Press, 2018), chapter 1.return to text

    9. It is by no means always easy to determine whether dictation was used in a particular case. See R. S. Bagnall and R. Cribiore, Women’s Letters from Ancient Egypt 300 BC–AD 800 (Ann Arbor: University of Michigan Press, 2006), 59–65.return to text

    10. See R. S. Bagnall, The Undertakers of the Great Oasis (P.Nekr.) (London: Egypt Exploration Society, 2017), 12.return to text

    11. See particularly J.-L. Fournet, “Langues, écritures et culture dans les praesidia,” in H. Cuvigny (ed.), La route de Myos Hormos. L’armée romaine dans le désert Oriental d’Égypte (Cairo: IFAO, 2003, 2006 [2nd ed.]), vol. 2: 447–67; R. S. Bagnall and R. Cribiore, “O.Flor.inv. 21: An Amorous Triangle,” Chronique d’Égypte 85 (2010): 213–23.return to text