Thoughts About Sight Reading, and Inner Listeners

by John Holbo on July 7, 2011

My performance/recording vs. oral/literary post has gotten lots of comments, so let’s see if I can drive you all off with a follow-up.

Two other books about the evolution of reading culture I read recently: Space Between Words: The Origins of Silent Reading, by Paul Saenger; and Silent Reading and the Birth of the Narrator, by Elspeth Jajdelska.

The Saenger book concerns the Rodney Dangerfield of punctuation marks: the space. (Why do you think that’s the largest key on your keyboard, hmmm?) Let me quote the publisher’s blurb in a way that makes Saenger’s point. Once upon a time, European scribes wrote like this:


That’s called scriptura continua or scriptio continua. It’s slow, isn’t it? But you can read it. How? Verify for yourself. You have to sound-out each word completely, inwardly, to figure out what word it is. All those little jack-rabbit-y saccades your eyes instinctively indulge in fail to garner whole word gestalts that afford speedy visual processing. Semi-contra Ong and McLuhan, this ‘visual’ format apparently affords a much more ear than eye-oriented reading experience, by our standards. When Augustine marveled at Ambrose’s silent reading of scriptura continua, really what he was admiring was the invention of snazzy, noise-reducing headphones for audiobooks.

Separated text was pioneered, in Europe, by Irish monks who, Saenger points out, starting using videre—to see—to mean to read. Continental readers of scriptura continua did not use this term. It is harder to find evidence that legere—to read—meant only to read out loud. (If it did not occur to people there was an alternative, they would not semantically encode the distinction.) But passages like the following, from Quintilian, suggest the range of ancient audio options was thought to exhaust the reading options:

Reading remains for consideration. In this connexion there is much that can only be taught in actual practice, as for instance when the boy should take breath, at what point he should introduce a pause into a line, where the sense ends or begins, when the voice should be raised or lowered, what modulation should be given to each phrase, and when he should increase or slacken speed, or speak with greater or less energy. In this portion of my work I will give but one golden rule: to do all these things, he must understand what he reads. But above all his reading must be manly, combining dignity and charm; it must be different from the reading of prose, for poetry is song and poets claim to be singers. But this fact does not justify degeneration into sing-song or the effeminate modulations now in vogue: there is an excellent saying on this point attributed to Gaius Caesar while he was still a boy: ‘If you are singing, you sing badly: if you are reading, you sing’” (Institutio Oratio, 1. 8.2).

Basically, Quintilian thinks of reading as rhetoric, as the art of speaking/performing a set piece. He would find it hard to understand, say, Sven Birkerts, rattling on about ‘the reading experience’ in The Gutenberg Elegies. That’s like asking someone about ‘the speaking experience’ or ‘the performance experience’. Maybe there is such a thing, or maybe not, but it’s not the point. (The audience should, of course, be having an experience. That might be the point. But they aren’t reading.) Note that he stipulates that the reader must understand. We would tend to think that is redundant. Reading needs understanding, by definition. If the student is just mouthing the words, she fails reading comprehension. Quintilian is thinking along different lines. Comprehension is the audience’s business, not the speaker’s. But if the speaker doesn’t understand, he will read in a flat monotone – will fail to be impressive and convincing. Try giving a good, audiobook-worthy reading of that passage of scriptura continua, above. You can’t, not first time through. You need to read it through five times, minimum, warming up your voice and committing it to semi-memory. It would still be helpful, for this Quintilianesque purpose, to have spacing and punctuation. That would allow you to ‘sight read’. Then you could do a decent job the first time. But only a decent job. If the goal is really good speaking, you still have to understand the whole passage beforehand, and rehearse it a few times. This in itself doesn’t explain why it took so long for word spacing to appear. But it helps explain. Consider ‘sight reading’ and musical scores. The ability to sight read, i.e. give a half-decent performance first time through, is very handy. But it is hardly the be-all and end-all of musicianship. Even if you thought ‘the musical experience’ named the thing; even if you identified that with the performer’s experience, rather than the audience’s, you would still hardly think the sight reading musical experience was the thing.

So long as reading is rhetoric, i.e. a genre of audible performance, it’s not like there is going to be this huge push to produce the most maximally sight-readable ‘musical’ notation. (You don’t see guitar players complaining that no one can sight read tab, after all. Why should that limitation be regarded as serious? It isn’t.) Add to this a high degree of haziness on the very concept of word. Plus you don’t have a box of punctuation symbols, just waiting to be deployed, so it might really take you quite a long time to invent all this, even though this apparatus seems so indispensable from our point of view today. (Yes, of course, the semitic language writing systems had spaces, going way back. They didn’t have vowels! They needed them! So the Irish monks would have been aware of this possibility, when they introduced spacing into their own texts. So: fine. They didn’t, strictly, invent this thing. Not invent-invent. Re-introduce. You get the point.)

Putting the point another way: adding space between words is a small step for graphic design, a huge step for mankind. In Thinking With Type, Ellen Lupton writes: “Although many books define the purpose of typography as enhancing the readability of the written word, one of design’s must humane functions is, in actuality, to help readers avoid reading.” Avoid reading in Quintilian’s sense. Instead, you read in Birkerts’ sense. Deeply. Or shallowly. You can skim and jump around and find what you need or want. You aren’t harnessed like an ox ploughing ahead in a line, unable to look around to see where else you might want to go. Lists. Indexes. Sections and subheadings and so forth. That’s the ‘reading experience’, for us. And a very visual affair it is (although not purely visual. No, of course not.)

Most of the things we read today, we sight read. That means at least two things: 1) we use our eyes a lot, to do things our inner ears couldn’t do, and can’t keep up with, and the page has been laid out to afford us a lot of help. 2) we read most stuff just once, ‘getting it’ more or less the first time. These are obviously two very different matters, although they tend to go together. Graphic design tends to go together mass production of printed material. The same technologies drive both. But the fact that you are a habitual visual reader, in sense 1, should be distinguished form the fact that you are a person who has read hundreds of thousands of documents once, and relatively few 10 times carefully, rehearsing them for oral performance.

So by what point had the shift in senses 1 and 2 occurred, pretty decisively? Ong (and McLuhan) think of visual literary culture as achieving dominance maybe earlier than it really did. But in this post I’m not going to quibble about what they claim, or don’t, about that. They are right about this much: they appreciate that there is an intermediate stage – between primary oral culture and modern reading culture – at which literacy sub-serves orality. They wouldn’t be surprised by that passage from Quintilian. They know there was a time when writing facilitated the study of classical rhetoric. At this point, reading was part of rhetoric. Quintilian would not have had any great trouble wrapping his head around the very notion of how pure oral poetry works. He wouldn’t have been gob-smacked by Milman Parry’s discoveries about Homer (which isn’t to say that he already knew it. Obviously not.) So put it this way: at what point are we far enough past Quintilian that we are gobsmacked by the Homer revelation, potentially? At what point did reading and writing become visual enough – in senses 1 and 2 – that we could really lose track of what it would be like to live in a primary oral culture?

This brings me to my second book: Silent Reading and the Birth of the Narrator, by Elspeth Jajdelska. (The Amazon link, above, will give you ‘look inside’ privileges, but I cringe at the price. Horribly overpriced academic books. What a world, what a world.)

In the late seventeenth and early eighteenth centuries, for the first time in England, a large enough group of children became sufficiently skilled in silent reading to constitute (in adulthood) an audience for a new style of writing. This style arose from the development of a new model of reading, as hearing rather than speaking (these terms are discussed in more detail below). The change in underlying reading models, and consequently in prose styles, created a need for what is now called ‘the narrator’ and was central to the development of all kinds of prose genres, including the novel. Changing reading models of this kind also interacted with the growing culture of politeness and new conceptions of the self.

The emphasis here on skilled silent reading should not, of course, be interpreted as a claims that there was no silent reading before 1700 and no reading aloud afterwards. Saenger has shown that during the Middle Ages a substantial body of monks read silently … [ok, we’ve been over the Saenger point already] … Between Augustine’s time [marveling at St. Ambrose and his silent reading] and the present day, then, there must have been a turning point in reading practices, when the balance of emphasis moved form reading aloud to reading silently. (4-5)

That’s about enough for one post, even though I’m only just getting into it. Jajdelska offers interesting stylistic analysis, which I will not summarize. Basically, she notes how changing norms for sentence formation and punctuation would naturally go with the shift she is discussing. I will say just a bit about what she means by ‘hearing’.

A change from reading aloud to skilled silent reading is important because it radically changes the underlying model of what writing and reading are. Reading aloud creates an identification between the writer and the reader. The reader is a speaker, the writer’s mouthpiece, with the writer’s words in his or her mouth. Silent reading creates a different relationship between writer and reader. Instead of identifying with the writer as a speaker of his or her words, the reader becomes an (internal) hearer of the writer’s words. So the move from reading aloud to reading silently involves a move from reading as speaking to reading as hearing, and from reading as declamation to reading as silent participation in an imaginary conversation between writer and reader. This is a radical change in the orientation of both writer and reader to the text. (6)

She notes, surely correctly, that this distinction between reading-as-speaking and reading-as-hearing crosscuts Ong’s oral/literate distinction in somewhat confusing ways. Literate culture is visual reading culture – but also hearing culture. When you speak, on the other hand, you don’t hear. Not exactly. Although it’s true that speaking while wearing noise-deadening earplugs is not conducive to comfortable, confident elocution, the experience of reading/speaking is not the experience of hearing what you sound like, while reading out loud. Most of us hate to hear the sound of our own recorded voices. It is unfamiliar and unwelcome. (No one reads out like singers sometimes sing: stick one finger in one ear so you can really get it right.) So there is something very wrong with associating orality with hearing. Or very half wrong. In oral culture, there is a reader/audience divide, and the experience is fundamentally different for the reader and for the audience. (When I read Daniel Pinkwater to my daughters, I am enjoying it and they are enjoying it. But I especially hate listening to John Holbo read things, whereas they especially like it. So there isn’t really one reading experience here, leaving aside the differences between adult and child appreciations.)

In silent reading it’s like your eye has an inner mouth with which it is sight-reading innerly out-loud to your inner (not outer!) ear, which is passively enjoying the performance. So you get to be both reader and listener. Pleasantly active and pleasantly passive, like an inner child on the knee of an inner adult. But, as Jajdelska notes, this is a fairly late development on the reading scene. It seems doubtful that St. Ambrose, with his invention of headphones for scriptura continua-format audiobooks, was quite this far along. At any rate, he wouldn’t have had the experience of inner sight reading, i.e. sitting and hearing your inner voice read you something for the first time, in a fairly competent manner. Because scriptura continua is just too damn hard to read. You would only really ‘hear’ it, inwardly, after you already understood it, i.e. had read it several times already and were now good enough to perform it, even inwardly.



Vance Maverick 07.07.11 at 5:26 am

Hm, I question whether we can judge what it’s like to read without spaces simply by trying it once. You might as well spell a paragraph of English backwards and claim you’ve proven Hebrew hard to read. Let’s spend 40 years with it, then get together again to discuss results.


Vance Maverick 07.07.11 at 5:34 am

Also I’m confused by the headphone analogy. Isn’t spoken text (“audiobooks”) continuous? (Thus the visual jokes along the lines of “mice elf”?)


nnyhav 07.07.11 at 5:37 am

In latest TLS Gail Trimble reviews Shane Butler, The Matter of the Page: Essays in search of ancient and medieval authors and William A. Johnson, Readers and Reading Culture in the High Roman Empire: A study of elite communities … alas not online, but fair use dictates:

Traditional accounts of ancient Greek and Roman reading have tended to stress two points. First, that the book as we know it, the codex of leaves bound along one edge, was not standard until the very end of Roman antiquity. Instead, an ancient book was a roll of papyrus paper which was most easily read in sequence […] The second traditional idea is that it was usual to read out loud, even when alone–perhaps because it was otherwise difficult to make sense of the continuous stream of letters, with no punctuation or even spaces between words, typical of ancient writing. [lf,cr] Shane Butler’s The Matter of the Page begins by complicating the first of these notions, William A. Johnson’s Readers and Reading Culture in the High Roman Empire the second. [… …] Johnson shows that ancient readers could read silently, and that doing so was not considered extraordinary. However, the evidence that reading aloud was normal in certain contexts should, he argues, make us ask more interesting questions about those contexts.

oops ran out of fair use …


Colin Danby 07.07.11 at 5:49 am

When did paragraphs come in to use?


nnyhav 07.07.11 at 6:06 am

still haven’t in blockquotes (thus lf;cr)


Brandon 07.07.11 at 6:41 am

To add to the point in nnyhav’s quotation, it’s worth pointing out that Augustine never actually expresses surprise at Ambrose’s reading silently; his point instead seems to be that he never felt that he could interrupt Ambrose with his problems, because Ambrose’s reading silently made it seem like he was very intent on his reading and wanted to be left alone. He does seem to have thought that it (or perhaps the fact that Ambrose almost always read silently) was unusual enough that it needed explanation — he suggests that perhaps Ambrose was afraid someone would overhear him and ask him a question, thus interrupting his alone time, or else that he was trying to preserve his voice — so at least he didn’t assume it as the default way to read, but he doesn’t ever seem to take it as astonishing in itself.

Like Vance, I’m not convinced that scriptura continua slows down reading for people who are practiced in it; one could just as easily argue that it’s the reverse: scriptura continua encourages taking in larger chunks of the line at a time, and separated text slows reading down by encouraging the reader to take things word-by-word. Instead of reading “francegermanyitalyandspain” you read “France,” “Germany,” “Italy,” “and,” “Spain.” (I don’t know which argument is better.) Likewise I think things are complicated by the fact that most of the manuscripts we are talking about were in Latin (or Greek) rather than English, and quick reading of Latin arguably requires taking in larger chunks of the line (as compared to a language like English) already, due to Latin syntax and grammar. Scriptura continua may well have made more sense for reading classical Latin in the first place, at least where people hadn’t already developed the habits for reading separated text. (I don’t know if this argument is sound, but it seems as plausible as the other.)

And isn’t Quintilian talking explicitly about reading speeches? Surely we can’t generalize from that to all reading. I suppose I’m not following parts of this argument very well.


John Holbo 07.07.11 at 7:08 am

Quick responses. Quintilian seems to be assuming that there is no such thing as primary silent reading. That is, he doesn’t say: now I’m telling you how to read speeches. He just assumes that anything worth reading is something worth reading out, on a suitable occasion. So the question of how to read is the question of how to read it out, properly, on that occasion. So that’s the art of reading.

Re: the difficulty of reading scriptura continua. You might think you’d get used to it. But you wouldn’t. This has been established experimentally. Sorry I don’t have the citation. I think it’s in Saenger. Divided text is just objectively better, if what you want to do is read quickly in the modern style.

Thanks for those references nnyhav. Don’t know those. I agree that it’s apocryphal to think Augustine regard Ambrose’s silent reading as miraculous. I think it would be better to compare it to playing an (electronic, presumably) keyboard without making sound. Obviously if you needed to practice piano without bothering someone, you could do that. Nothing remarkable about it. But it’s unusual. Someone who did it a lot would be noteworthy for that behavior. If you are in the habit of vocalizing everything you read, it will obviously take a bit of practice to suppress that, particularly if you are reading sciptura continua.

Maybe a better example would be: an actor practicing his lines silently, i.e. just reciting them inwardly. Not remarkable. But unusual.

Re: the audiobook crack. I’ve seen it said that Ambrose marks a turn to visual reading. Silent reading seems less an affair of the ear. People imagine that Ambrose, like us moderns, is zipping along, skimming and jumping around and doing all those usual modern reading things. But, since he was reading scriptura continua, he must have been doing exactly what we do when we read scriptura continua. That is, sounding each word out a bit laboriously. If you are like me, it takes a bit of trouble to keep my lips from moving just a little, reading that stuff. Because I have to puzzle it out. And, again, I think experimental evidence suggests humans just don’t get much better at reading that stuff. Do this little experiment yourself. How long does it take you to find all the letter e’s in this paragraph? How long does it take you to find all the spaces in this paragraph. It’s just objectively easier to count the latter than the former. You can skim along, noting them. You couldn’t train someone to recognize e’s just as quickly as you can recognize empty space between words. It’s not how the eye-brain works.


Andrew Fisher 07.07.11 at 7:40 am

I think there are a whole bunch of issues skated over too lightly here, for instance:
– the technology of writing (and consequently the cost). The high medieval stuff I was (once) familiar with is very obviously shaped by the need to minimise writing costs;
– administrative reading and writing. Not only is this a major use of literacy, but there is a strong overlap between literary and administrative elites in both ancient and mediaeval European cultures (and no doubt in others too).
Without reading Jajdelska ($28 seems quite reasonable to me, but I’m not sure I can spare the time), it’s difficult to see why the emergence of a particular form of literature in the eighteenth century shouldn’t reflect increasing prosperity and decreasing costs of book production/distribution, rather than any fundamental shift in attitudes to reading; and the attempt to sketch out a grand narrative arc in this post seems unconvincing.


Phil 07.07.11 at 7:46 am

like singers sometimes sing: stick one finger in one ear so you can really get it right.

Does anyone, really? What (some) folksingers do is cup one hand behind an ear, shutting out everything behind the singer & hence focusing attention on the voice. (We tend to discount the rear-facing parts of the auditory field, but it actually accounts for an awful lot of the everyday sound-world. Next time you’re at the theatre, try listening in to the people in front of you.)


John Holbo 07.07.11 at 8:31 am

“$28 seems quite reasonable to me, but I’m not sure I can spare the time”

Ah, that’s a good point. Used price not so bad. $66 kind of steep for a new hardback on the other hand. More substantively, I agree that the post skates over a bunch of issues rather lightly. (But I did mention the technological component! although I didn’t discuss it.)

“Does anyone, really?”

Well, don’t stick the finger in too far, would be my advice. But, yes, it’s easier to sing in tune with one finger in one ear. So people do. I have myself (but I don’t sing very often.)


Tom Richards 07.07.11 at 9:08 am

I wonder if certain books or passages *should* be read in scripta continua by a modern reader, to force you into pronouncing them internally (I also had problems not moving my lips when reading the example), to process them more slowly (e.g. meditative religious texts), or to make returning to a reference more difficult in order to force people to read around in the text. It would be easy to produce examples by removing anything except characters from a text file and then making everything lowercase – I might just try it with poems for memorisation.

On the subject of the finger in the ear, if there’s a whole pub full of people singing different harmonies around me I find it difficult to concentrate on my own voice, and blocking out external sounds from one ear helps me to hear my singing more clearly. For me it’s a crutch that I’d discard as soon as I could, though!


Kadin 07.07.11 at 10:37 am

It’s not that difficult to read that scripta continua, and it doesn’t seem particularly slow to me. I tripped up in a couple of places and had to look closely to figure out what the words were, and I doubt I’d be able to effectively speed-read it, but I don’t think reading it is particularly slower than reading spaced text.


rm 07.07.11 at 12:03 pm

Andrew Fisher @8: it’s difficult to see why the emergence of a particular form of literature in the eighteenth century shouldn’t reflect increasing prosperity and decreasing costs of book production/distribution, rather than any fundamental shift in attitudes to reading

It can’t be both???

There’s a lot of scholarship establishing the truth of your point, but I don’t see why that precludes the internal-reader idea either. It was not inevitable that a literate middle class with lots of affordable reading material would be reading a particular type of literature, and the explanation of how “narrative voice” was a new thing sounds very convincing to me. Presumably Jajdelska has some interesting research into reading pedagogy.

There’s an old debate in literacy studies between cultural and technological determinism, and of course the lesson is that both culture and technology influence what we do.


Andrew Fisher 07.07.11 at 12:39 pm

@rm no.13

It could be both. I suppose its the mass of ‘inner’ devotional literature I know of from the late middle ages and early modern periods that makes me sceptical that ‘skill in silent reading’ is the main issue circa 1700.

Not having read Jajdelska’s work, I can’t really presume to criticise it. What I am therefore most sceptical about is John’s attempt to turn a couple of data points (however strong they may be individually) into some grand narrative arc about the ‘evolution of reading culture’ over many centuries.


Gaspard 07.07.11 at 1:04 pm

Can one make this argument work for logographic writing like Chinese, or would they have started silent reading sooner/ almost immediately?


Shelley 07.07.11 at 1:17 pm

Interesting. As a writer, this makes me wonder if all the texting/tweeting stuff is a weird way of writing without having that inner reader (on either side, really–what’s “inner” about it?)


Doctor Science 07.07.11 at 1:51 pm

This discussion doesn’t seem to be talking about the nature of writing or reading in general *at all*, only about how it (maybe) has worked for a particular set of scripts.

Chinese, for instance, has (a) an extremely long literary tradition, (b) no need for spaces between the words because they’re built in, and (c) is traditionally punctuation-free.

Compare/contrast with Sanskrit, which (a) has an extremely long and complex oral tradition, (b) not tied to any specific script, (c) in which punctuation and spaces are absent or afterthoughts.

For instance, in discussing Pāṇini’s very dense Sanskrit grammar Wikipedia says:

It is not certain whether Pāṇini used writing for the composition of his work, though it is generally agreed that he knew of a form of writing, based on references to words such as “script” and “scribe” in his Ashtadhyayi. … It is believed by some that a work of such complexity would have been very difficult to compile without written notes, though others have argued that he might have composed it with the help of a group of students whose memories served him as ‘notepads’ (as is typical in Vedic learning).

Compare/contrast again with Hebrew, which for centuries or more did without not merely punctuation or spaces, but *vowels*. Traditional Hebrew learning is very oral: studying the Talmud or Torah is done in pairs or larger groups, reading aloud back and forth, wrestling with the text and each other.

I don’t know how oral traditional Chinese literary culture was, or whether written texts were normally read aloud or silently — but ISTM you have to think about these cross-cultural issues to understand the general principles.


roac 07.07.11 at 2:03 pm

Agree with Holbo that the continuous format is inherently suboptimal in terms of reading speed. After going through the cited passage a couple of times, I see two major reasons: (1) a word break at the end of a line leads to a full stop, every time; (2) word breaks remove ambiguities which otherwise occur frequently, e.g. “wordsand.”
(Though “wordsand” is potentially a useful neologism, akin to “word salad.” Everybody has seen how words can halt social machinery the way sand in the gears destroys physical machinery; and the Sandman can certainly put you to sleep with words.)

No. 1 raises the question of whether and how much hyphenation retards comprehension, in comparison to non-right-justified text without word breaks. A question amenable to experiment, which has no doubt been answered. Anybody know what the answer is?


Martin 07.07.11 at 2:25 pm

For those interested in this stuff let me mention a book on a related topic that I found fascinating: The Printing Press as an Agent of Change, by Elizabeth Eisenstein which came out in the late 1970s. I don’t know if much of it has been superseded by later research, but, to repeat myself, I found it to be extraordinarily interesting.


john b 07.07.11 at 2:38 pm

This is a really interesting counterpart to McLuhan (whose point is that human narratives are traditionally oral, and the way Western Scholars see things has been inherently shifted by the move to books and the primary way knowledge is transferred being reading, and that the TV Age shifts that perspective) – in line with Augustine, this isn’t how Roman culture worked and isn’t how Latin literature should be read. But it is probably, 1000 years after Ambrose’s skills were considered amazing, how literature was read in the 1400s. As Vance and Brandon imply, once one dude’s had the idea, it gets picked up on quickly – mastering reading paragraphs like your citedoneabove is easier than learning French, never mind Arabic or Katakana, even before you move on to learning pictogram-based alphabets.

Will look up Eisenstein. Most study about the printing press is about its impact on mass consumption literature, rather than the way in which the elites who already had access to *whatever* on vellum were changed by it. The latter would be really interesting, although ethnography might be a bit of a struggle.

(currently studying for an MPhil and hoping to study for a PhD in how new media affect the way people read literary narratives, so this is Of Great Interest. I wonder if John Q has anything to add here, as a scholar of That Kind Of Thing…)


candle 07.07.11 at 2:44 pm

I don’t buy the use of Quintilian here: sure, he isn’t explicitly talking about reading a speech aloud in this passage, but this is in the context of a treatise on oratory, and what he is concerned with is not primarily understanding but rather training-for-performance. All of his advice is aimed at improving the student’s skill at performing a text: understanding is not the aim of this but the prerequisite, and Quintilian makes no comment on how the understanding itself is to be achieved. I suspect that there was very little need among the Romans for treatises on how to read in its everyday sense – because, as you say, it was taken for granted by everyone – as opposed to how to become an orator.

And I’m suspicious of the Ambrose story too. Apart from the wholly valid points made by Brandon regarding Augustine’s purpose in telling this story, I suspect that it is also in part intended to mark a stage in his own development from his former job as a professional rhetor (and thus follower of Quintilian) towards his new position as a theologian and meditator on scripture. I agree that the point is not (or at least, not necessarily) that Ambrose was skimming through a passage that was new to him; but for me the important thing is his absorption in the text for its own sake, without regard for an audience. In other words, for all that Ambrose was a spellbinding preacher (says Augustine), when he performed the scriptures he did so in order to promote understanding, whereas Augustine had been trained (on the basis of Quintilian) to understand for the sake of performing.

Remember that this is Augustine considering whether or not to convert to Christianity and thus abandon his professional life. And his actual conversion scene (“Tolle, lege”) gives a primary role to reading for the sake of understanding: that is, the passage he reads he understands as having a specific message for himself alone. IIRC it isn’t stated whether or not he read the passage aloud. But whether he did or not, he will surely have imagined his reading of it as being more in line with Ambrose’s example than with Quintilian’s advice.

(I don’t know the Johnson book, as cited in the TLS review, but I can highly recommend Brian Stock’s Augustine the Reader. There is a long and interesting review by Mark Vessey at BMCR which may enlarge the point beyond where you want to go.)


Chris Crawford 07.07.11 at 3:30 pm

I can add some points from a different angle. First, I’d recommend the book “Reading in the Brain”, by Stanislaus Dehaene. It’s poorly written, but offers a great deal of information on exactly how the brain executes the reading process. I found material fascinating enough to justify slogging through the swill of the text. The book doesn’t directly address scriptura continua, but its discussion of the way that letters and words are understood makes it clear that word separation is crucial to efficient operation.

Another point: the comprehensibility of scriptura continua depends on the language. Latin and Greek spelling are closely phonetic, so it’s not difficult to make the jump from lexemes to phonemes. However, some languages, English especially, have poor correspondence between spelling and phonetics. Consider, for example, the many ways in which the letter “o” is pronounced depending upon its neighbors: ought, women, probe, boy, cot, coat… English pronunciation is highly dependent upon lexical context. The same problem occurs semantically in newspaper headlines such as “Deaf mute gets new hearing in killing”. (There are entire books of these, and they’re great fun!)

This language dependence upon the phonetic fidelity of spelling has all sorts of implications. A number of morphemic and syllabic writing systems rely heavily upon rebus to function. Egyptian hieroglyphics used rebus quite a bit, and the Chinese writing system is really complicated because it combines a bunch of elements: partly morphemic, but with tone and pronunciation cues for disambiguating morphemes.

This gets really bizarre when we get to languages with few consonants, such as the Polynesian languages. These require lots of syllables for each word. For example, suppose that I apply scriptura continua to a string of Hawaiian place names: maunkeahonolulumaunaloamauipepeekeonapililahana — I think you can get the idea.

Lastly, there’s been a lot of thought about the relationship between literacy and orality, how writing and reading have changed the way we think. The leading work on this is Orality and Literacy, by Walter J. Ong, but I think that the best starter book is The Alphabet Effect, by Robert K. Logan.


Brandon 07.07.11 at 3:30 pm

You might think you’d get used to it. But you wouldn’t. This has been established experimentally.

Established experimentally with people who haven’t yet developed reading habits based on separated text? That’s what it would take for the experiment to be relevant to the question of the reading speed of people who never saw a separated text in their lives. These things aren’t always symmetric: having A habits may easily allow picking up B habits, whereas having B habits may severely impede picking up A habits. Being able to do complex arithmetic with pen and paper (as an example) could very well allow for easy use of calculator; but at the very least it doesn’t follow from this that everyone handy with a calculator can easily pick up the knack required to do well the same problems with pen and paper.

Quintilian in the very next section (1.8.3) is clearly talking about reading speeches, because his comment is that young orators shouldn’t be taught to read speeches as a comic actor would. And according to the prologue, the whole section in which this passage is located is devoted to the question of cultivating good delivery. As candle notes, the point of the work as a whole is to lay down the education required to make an excellent orator.


Phil 07.07.11 at 3:46 pm

it’s easier to sing in tune with one finger in one ear. So people do.

Never seen it done in earnest, only as a caricature of folkies (who don’t do it). Live and learn.


David 07.07.11 at 4:53 pm

Back when the 911 final report came out there was a site that had a PDF of the report with new typography spacing for reading. The new spacing was great for the mind eye cognitive experience of reading the report.

In fact I could liken it to line being on a runaway horse as the experience of reading it captured the cognitive experience and carried (locked) the process of vacuuming the text and understanding (groking) the meaning. Words can not simply describe the experience.

This was the best reading experience in the full meaning of acquiring material from the page I have ever experienced for both understanding, speed, and not having text or layout hinder the experience.

Does anyone else remember or know the company or a way to find it?


Gene O'Grady 07.07.11 at 5:55 pm

For those who underestimate the difficulties of reading scriptura continua, if you look at the apparatus criticus in a decent edition of Homer (just to cite the author I’ve most recently been engaged with), you will see that there are lots of textual issues derived from word division. Many of them are comparatively trivial, although some of them are not, but the overall effect is to reinforce that it was definitely not that easy to read in that format.

If anyone is interested, some of Roger Bagnall’s recent work on literacy in Roman Egypt and other parts of the middle east have some well-founded revisionist ideas about ancient literacy and the physical shape of ancient writing, which used many other formats than the codex or roll. Getting a little farther astray (and less relevant), apparently the newly interpreted evidence from Herculaneum indicates that some rolls of prose texts were much longer and more capacious than previously assumed — up to something like 24 meters.


Bill Benzon 07.07.11 at 6:22 pm

As luck would have it, I just happen to have a very well-known poem in two versions that are relevant. In one version I’ve stripped out all spaces, punctuation, and capitalization. In the other version I’ve restored spaces between words, but no punctuation, much less line endings.


* * * * *

in xanadu did kubla khan a stately pleasure-dome decree where alph the sacred river ran through caverns measureless to man down to a sunless sea so twice five miles of fertile ground with walls and towers were girdled round and there were gardens bright with sinuous rills where blossomed many an incense-bearing tree and here were forests ancient as the hills enfolding sunny spots of greenery but oh that deep romantic chasm which slanted down the green hill athwart a cedarn cover a savage place as holy and enchanted as e’er beneath a waning moon was haunted by woman wailing for her demon lover and from this chasm with ceaseless turmoil seething as if this earth in fast thick pants were breathing a mighty fountain momently was forced amid whose swift half-intermitted burst huge fragments vaulted like rebounding hail of chaffy grain beneath the thresher’s flail and ‘mid these dancing rocks at once and ever it flung up momently the sacred river five miles meandering with a mazy motion through wood and dale the sacred river ran then reached the taverns endless to man and sank in tumult to a lifeless ocean and ‘mid this tumult kubla heard from far ancestral voices prophesying war the shadow of the dome of pleasure floated midway on the waves where was heard the mingled measure from the fountain and the caves it was a miracle of rare device a sunny pleasure-dome with caves of ice a damsel with a dulcimer in a vision once i saw it was an abyssinian maid and on her dulcimer she played singing of mount abora could i revive within me her symphony and song to such a deep delight ‘twould win me that with music loud and long i would build that dome in air that sunny dome those caves of ice and all who heard should seem them there and all should cry beware beware his flashing eyes his floating hair weave a circle round him thrice and close your eyes with holy dread and drunk the milk of paradise

* * * * * *

BTW, if you look at a sound spectrogram of continuous speech you won’t see any breaks in the sound that correspond to word divisions. Those ‘breaks’ are supplied by the inner ear, as it were, and aren’t in the sound stream.


roac 07.07.11 at 6:44 pm

Brandon @22: But is there an way to learn to read without starting with single lexical units in isolation? I can’t imagine one.


etv13 07.07.11 at 8:40 pm

I am flummoxed by the claim that the widespread adoption of silent reading created a need for a narrator. The Canterbury Tales has a narrator (in fact it has a narrator telling you how other narrators narrated their tales). Robinson Crusoe, which is at the early end of the period in which this development allegedly took place, has a narrator. It seems to me what might be a new development is the way in which the narrator disappears into the text. Austen’s and Trollope’s narrators sometimes directly address the reader, but they aren’t characterized in the same way that the narrator of The Canterbury Tales is (he doesn’t give one of his little thumbnail sketches of himself, but we do know he’s a guy going on a pilgrimage in April, and when he comes to tell his own tale, he’s an incompetent storyteller).


Harold 07.07.11 at 9:02 pm

Petrarch complained that the cramped Gothic style and abundant use of abbreviations in Medieval books made them seem “designed for anything but reading”. After his death, but even before the invention of the printing press, there was a reform of style, with a change from Gothic to Italic (Carolingian) script, and with spare, classically inspired ornament instead of the former extravagant illumination, in order to forefront the text and facilitate reading. These Renaissance manuscripts appear more “modern.”


Bill Benzon 07.07.11 at 9:46 pm

@roac, 26: very good question.


John Holbo 07.07.11 at 11:29 pm

“I don’t buy the use of Quintilian here: sure, he isn’t explicitly talking about reading a speech aloud in this passage, but this is in the context of a treatise on oratory, and what he is concerned with is not primarily understanding but rather training-for-performance.”

First, I take credit or blame for the Quintilian case. I didn’t borrow it from Saenger or anyone else, so if it’s bad evidence, it counts against no one’s case but mine. Second, I don’t think it’s a grand slam proof, but I think it’s telling. There isn’t anything in the passage like: ‘reading for speaking is different from reading for yourself, or reading privately’. There’s just ‘here’s how to read’. Now compare this to, say, Birkerts, in “The Gutenberg Elegies”. He says a lot about what reading involves. None of it fits the case of reading in Quintilian’s case. Now, Birkerts is obviously aware that it is possible to read stuff out loud, as a kind of performance or oratory. But he just doesn’t think about that. My point is that Quintilian thinks he is talking about ‘how to read’ and Birkerts is talking about ‘how to read’ and their venn diagram circles aren’t intersecting, never mind not perfectly matching. Now in fact if we mapped their actual spheres of awareness there would be intersection, if not perfect matching. But it’s significant that they are talking past each other in this way. It indicates a real shift in attitudes towards reading. That’s all I’m trying to maintain.

“Established experimentally with people who haven’t yet developed reading habits based on separated text?”

If I recall – and it’s a bit hazy – some people tried to train themselves to do this, to see how good they could get. And the prognosis was: we’re not getting very far, and it seems doubtful we’re going to. Conclusion: you are always going to be slower, reading this way.

More later …


John Holbo 07.08.11 at 12:37 am

roac, I’ve been wondering this myself and I haven’t seen any relevant texts. How do you teach basic reading in continuous text? I’m guessing that you would use what Saenger calls semi-aerated (? sorry I’m forgetting his word for it) script. There is an intermediate stage between continuous text and modern word separation at which the text is broken up by spaces, so you get these semi-arbitrary chunks. Basically, the eye leaps from chunk to chunk and there aren’t any hard and fixed rules about how big they should be.

I think you could teach reading by teaching first the letters (obviously). Then teaching


That’s not much harder for a kid than HOP ON POP, actually. You have to laboriously sound it out. So, in a way, teaching via short bursts of scriptura continua would not be that much harder than teaching with word separation. Maybe. But I’ve not seen any scholarly discussion of this question, although I’m sure it’s out there somewhere.

It’s important that there aren’t any dictionaries until, like, the 1600’s. I’m sure there were translation manuals. It’s so hard to imagine studying a foreign language without having a proper concept of ‘word’ …


waw 07.08.11 at 2:45 am

There are narrators in/for several of the Platonic dialogues, so is the claim really that the use of narrators had await the rise of silent reading? Or is there something more to that point that I am not getting?


John Holbo 07.08.11 at 3:04 am

Sorry, I just got around to rescuing several comments from moderation. I see that Doctor Science did a double-post, so I’ll now go back in and delete the extras.


John Holbo 07.08.11 at 3:22 am

“There are narrators in/for several of the Platonic dialogues, so is the claim really that the use of narrators had await the rise of silent reading?”

No, that would clearly be crazy. The thesis is more modest than that. (I agree that the title is a bit misleadingly flamboyant that way. The book is not.)


Vance Maverick 07.08.11 at 3:27 am

“the taverns endless to man?” Bill, you’re not being very careful there — on the other hand, taverns measureless sound pretty good to this man.


Glen Tomkins 07.08.11 at 5:06 am

Surely the introduction of spaces between words is like the earlier introduction of punctuation in Greek, and the vowel points in Hebrew — done largely for the benefit of readers for whom these were not their native languages. Sort of like having differently colored shapes on the different keys of a reduced keyboard when you’re trying to teach a toddler piano. Not something people who know the language need, but a handy learning aid for those who don’t. Medieval copysists started using spaces to give themselves the advantage of a visual analysis into separate words, a crutch that native speakers didn’t need.

Of course, that doesn’t explain a point you make that I would not disagree with, that for a native English-speaker such as myself, scripta continua even in my native tongue is much harder to get through quickly than my native tongue with the spaces.

But I don’t think that means that scripta continua wasn’t at least as efficient, if not more so, for the ancients. I have trouble accepting the idea that they would not have hit on spacing earlier — not have left this innovation to Irish monks — had it also been more efficient for them with their native language, as it is for me with mine.

Perhaps it’s a difference in the “efficiency” sought, what they valued in the process of reading. I have to admit that not infrequently the chief advantage I derive from struggling through a pasage in the original Greek, is that it takes so long that it’s harder to misread and overtranslate, and not really because the original otherwise adds anything to the understanding, as compared to a competent translation. Festina lente, and so forth. Had the Library of Alexandria survived, and with it an unedited superfluity of everyday ancient prose, including all sorts of stuff that they did not prize for its literary value, perhaps we would see that they did put in spaces for their equivalent of the NYT, or the minutes of faculty meetings — stuff no one values for itself, and just wants to get through as quickly as possible. Perhaps they reserved scripta continua just for literature, but since all that survived is their literature, their NYT equivalent is mercifully lost, all we see from this period is scripta continua.

The other possibility is that they really just didn’t think in terms of separate words. That’s clearly not categorically true, in that they did distinguish parts of speech, and other grammatical concepts that require separate words, much as modern physics requires fundamental particles. But I wonder if that wasn’t a level of abstraction that had as little impact on how they experienced language as quantum mechanics has on how we experience the physical world. We grow up today taught grammatic and orthographic fundamentalism from the cradle, as if the rules of these were prescriptive and constitutive of language, rather than merely descriptive of language, and largely quite incidental and not at all fundamental to how language works.


Harold 07.08.11 at 8:06 am

“It takes so long it’s harder to misread” — Slow, intensive Medieval reading. Eloquently and movingly described by Anthony Grafton (in his December 23, 2010, essay, “Jumping through the Computer Screen”, a NYRB review of McNeely & Wolverton’s Re-inventing Knowledge: From Alexandria to the Internet):

“Benedictine monasteries did emphasize a distinctive approach to texts, one that might be called “slow writing and reading”—and that contrasts as sharply with contemporary practices in reading and writing as Slow Food does with McDonald’s. The Benedictine rule allowed each monk to borrow one book a year from his monastery’s collection. This he was to read and meditate on, slowly and with concentration, in his few free hours. Public readings from the Bible and other central Christian texts, held at mealtimes, reinforced the instruction drawn from the carefully chosen Christian classics in individual cells. So, even more powerfully, did a central Benedictine task: that of copying the canon of sacred texts and their Christian commentators, precisely and accurately, on sheets of skin that would last for centuries, when bound into codices, and serve generations of Benedictine readers in their turn.

The Renaissance Benedictine abbot Johannes Trithemius argued, in a famous screed against the printing press, that this Benedictine way of engaging with texts was uniquely valuable because it eliminated all the slippages that can come between books and readers. Those who adopted this regime would be transformed by what they read:

‘He who copies accepted and holy texts will not be burdened by vain and pernicious thoughts, will speak no idle words, and is not bothered by wild rumors…. And as he is copying the approved texts he is gradually initiated into the divine mysteries and miraculously enlightened. Every word we write is imprinted more forcefully on our minds since we have to take our time while writing and reading. The repeated reading of Scripture will inflame the mind of the writer and carry him happily to total surrender to God.'”


candle 07.08.11 at 11:53 am

First, I take credit or blame for the Quintilian case.

Yes, no problem – I realise that. I still think it’s a misleading example, though.

I think it’s telling. There isn’t anything in the passage like: ‘reading for speaking is different from reading for yourself, or reading privately’. There’s just ‘here’s how to read’.

Well, that’s true, but it rests on the surprising assumption – well, I find it surprising – that Quintilian would have any reason to specify that they were different. As you say, Birkerts doesn’t feel the need in the reverse case. Sure, their Venn diagrams don’t interact, but I can’t see how this means that one attitude is ancient and the other modern, when all of the difference can much more straightforwardly be accounted for by the explicit purpose for which each writer is writing. You would find precisely the opposite if you took Andrew Fisher’s medieval devotional literature and contrasted it with a modern guide on how to give a Best Man speech.

This is not to say that there isn’t a difference between ancient and modern reading. But to think about it needs much more sensitivity to context.


candle 07.08.11 at 12:07 pm

It is worth pointing out that we can read some actual ancient Latin writing for ourselves, such as the famous letter from Octavius to Candidus from the Vindolanda texts.

This is obviously difficult to read even if you have Latin (and a transcription), but the bottom line of the text can be made out to read nisi mittis mi aliquit with what look like an attempt at word spacing. This is fairly low-level informal writing: it seems that when scribes or the highly educated got involved (as on tablet 291, the birthday invitation, which for some reason I can’t link to) the spaces were ignored. And of course this is true of the epigraphic writing we are used to seeing on monuments, and was probably true on the parchment scrolls kept in upper-class homes and in libraries.

Latin/Greek makes a difference too, obviously.


John Holbo 07.08.11 at 1:22 pm

“but I can’t see how this means that one attitude is ancient and the other modern, when all of the difference can much more straightforwardly be accounted for by the explicit purpose for which each writer is writing. You would find precisely the opposite if you took Andrew Fisher’s medieval devotional literature and contrasted it with a modern guide on how to give a Best Man speech.”

But Quintilian is much earlier. A thousand years. Aren’t you thinking about post 10th century medieval stuff? In which case this would potentially confirm Saenger’s point? Also, I doubt that a modern guide to giving a Best Man speech actually would use ‘read’ as a synonym for ‘speak’. If you found a modern guide to reading a prepared speech, it would probably begin by saying ‘we all know how to read, but not many people are good at reading a text out load, effectively, in front of an audience.’ Which is precisely not what Quintilian would say. So my point about a contrast would be established. (Obviously we would need to check and see what any such guide really did say.)

I agree with your general point that it’s a bad idea to make broad claims about what people could ‘conceive of’ when, really, it’s probably a matter of what they were interested in. That’s sufficient, I think, to support my ‘thinks about it quite differently’ claim.


John Holbo 07.08.11 at 1:23 pm

“This is obviously difficult to read even if you have Latin (and a transcription), but the bottom line of the text can be made out to read nisi mittis mi aliquit with what look like an attempt at word spacing.”

That is interesting.


roac 07.08.11 at 1:38 pm

Just as a random data point, the manuscripts I have looked at (in facsimile), which are in Olde English and Old Norse from the 12th c. or thereabouts, are not only word-divided but eminently legible; once you learn a few archaic characters and conventional abbreviations, it is no harder to read Beowulf in manuscript than from a modern printed text. No doubt this fits somewhere into the narrative of the Saenger book, which I am going to have to get hold of despite the price.

Incidentally, I believe that there was an unbroken tradition of reading aloud as family entertainment in Iceland from the Saga Age onward — the country was notably literate even when dirt-poor.


Random Lurker 07.08.11 at 1:44 pm

@roac 28

I don’t know if this can be of interest, however from the point of view of a non english speaker, the difference between written words and pronunciation of those same words in english is much bigger than in italian: in italian, I think that also a not very competent speaker (such an immigrant, for example) can guess the correct spelling of a word very easily when he hears it the first time; in english, on the opposite, this seems very hard to me.
The wikipedia page about “Old English” ( states that in early middle ages english presented a much closer relationship between spoken and written.
IMHO it is possible to teach reading in scriptura continua, if the difference between spoken and written is very small.


Tim Silverman 07.08.11 at 2:32 pm

Dr Science @17: on the subject of technical Sanskrit, even more disconcerting is the technical mathematical and astronomical tradition, which was also basically oral, with results including (at the extreme) cases of trig tables set to verse in order to make it possible to memorise them and pass them on orally, and, more generally, highly technical textbooks in the form of laconic notes designed to aid memory and learning within a basically oral context.

Also perhaps related to this discussion is the very late appearance of fully developed mathematical symbols (basically at the end of the 16th century) after millenia in which mathematical problems, proofs and theorems were in the form of word problems (sometimes with accompanying geometrical diagrams). The rapid development of algebra and calculus thereafter is presumably connected with this.


Tim Silverman 07.08.11 at 2:35 pm

Glen Tomkins @38: there are a lot of surviving Greek papyri. You can see images of some of them on them web, and scriptura continua seems to be the norm. They aren’t all “literature” (on the other hand, there’s no equivalent to the NYT or many other modern forms of writing).


Tim Silverman 07.08.11 at 2:39 pm

JH @33:

there aren’t any dictionaries until, like, the 1600’s

You mean lists of definitions? There are parallel word lists for translation purposes from ancient times, e.g. Akkadian/Sumerian cuneiform dictionaries from the Old Babylonian period.


Gene O'Grady 07.08.11 at 2:42 pm

I may be misreading (how appropriate!) Glen Tompkins, but it’s not true that only literature survives from Classical antiquity — tons of technical and scientific stuff, even excluding Christian authors. So much Galen (some of which is literary and most of which apparently is rhetorical) that it fills 24 or so volumes in the old edition Cambridge is reprinting.

Also, lots of people wrote Latin and Greek who were using them as second or third languages — could have had Aramaic or Oscan or some Celtic language as their native tongue. Plus at least one major Latin poet (Claudian) apparently was a native Greek speaker.

Finally, the world of the Vindolanda material (on leather as I recall), the world of graffiti from Pompeii and now sites in Syria, the world of ostraca, and the worlds of private and bureaucratic papyri are all quite different from each other and from the higher culture writing we are familiar with. To a much greater degree than language and script differences in our world. A semiliterate person might be much more likely to put in spaces for a semiliterate audience.


roac 07.08.11 at 2:44 pm

RL @ 20:

Believe me, pretty much all native English speakers are aware of the perversity of English spelling. (Very few of them understand the various and complicated historical reasons for it. I understand a little.) We have never lacked for advocates of a radical simplification to bring spelling in line with modern pronunciation. The problem is that on from the date you start teaching the new system in school, everything printed according to the current system starts inexorably to turn into a foreign language text. And Americans at least are horrible at foreign languages, as you no doubt know.

As for your suggestion that reading without word breaks is easier in a phonetically-spelled language, I have no doubt of it.


Chris Crawford 07.08.11 at 2:57 pm

Obviously, I failed to make clear my point in #22. There is no need to debate the fundamental question of whether scriptura continua can be read as quickly as word-delimited text. That matter has been all but settled by developments in the study of the neurophysiology of reading. Although an experiment directly assessing the difference has not, to the best of my knowledge, been carried out, a great quantity of experiments carried out using magnetoencephalography (real-time high resolution imaging of the brain) during reading has taught us much about that process. Obviously, reading is such a complex process that we’ll never nail it down, but some points have emerged clearly:

1. Letters are recognized as combinations of basic elements: lines, curves, and dots, joined together in three basic forms: corners, perpendicular joins, and convergences (Ls, Ts, and Ys).

2. Bigrams — pairs of letters — are recognized and identified, and the brain carries out a fitting process using most likely pairings. For example we can understand “bdage” as “badge” with very little delay.

3. If spacing between letters of a word is increased, the brain can still comprehend until the space length equals about two characters, at which point normal processing collapses and the reader must proceed letter by letter.

4. The crucial area of letter recognition is in the left occipito-temporal region.

5. Once basic shape recognition has been completed (about 200 msecs), the comprehension process spreads all over the left occipto-temporal region, integrating knowledge of spelling conventions (e.g., “q is almost always followed by u”) with syllable probabilities (e.g., “tion” is common, but “toin” is rare) and then up to word comparisons. But it doesn’t stop there; the fit of the word to the context must also be worked out. If the fit is good (e.g., “necessary and proper” versus “necessary and paper”) then recognition is faster.

Anyway, all of this has led to the development of a model of reading that’s too complicated for me to present here, but that model clearly indicates that scriptura continua should take longer for the brain to process than space-delimited text. If you’d like to delve into it, again I recommend “Reading in the Brain” by Stanislas Dehaene.


michael e sullivan 07.08.11 at 4:26 pm

“(You don’t see guitar players complaining that no one can sight read tab, after all. Why should that limitation be regarded as serious? It isn’t.)”

As a guitar player, I’m not sure your point is made here.

I don’t see anything inherent in tab that makes it harder to sight read. If anything, it would seem to be easier as one does not have to translate a note to a string location, and one does not have to decide where to play a note that has multiple options on the fretboard.

I cannot, in fact, sight read tab well at all, but I’m pretty sure that’s because I have very little practice with it compared to with traditional musical notation (which I can sight read fairly well, as do most good classical guitar players). I’ve always assumed that pro/near-pro rock players (who see a lot more tab than we do) could sight read tab as well as I sight read regular notation.


candle 07.08.11 at 4:47 pm

But Quintilian is much earlier. A thousand years. Aren’t you thinking about post 10th century medieval stuff? In which case this would potentially confirm Saenger’s point?

Well, I’m thinking about the high/late Roman empire, but I couldn’t off the top of my head come up with an uncontroversial example of something obviously written for a context which excludes declamation. That is, I was trying to think of the opposite of Quintilian, which isn’t easy to do. Still, you’re right that that does help Saenger’s point. Naturally I prefer to think it shows how hard it is to find uncontroversial examples.

My main point was really that Quintilian is a poor test-case: a teacher of rhetoric is going to stand out even from his time in chiefly being interested in reading aloud. Although of course it’s true that Roman literature did contain a lot of writing which either derived from an oral context or pretended to: not just plays and poems, but the speeches of Cicero and philosophical and legal writings were all things which at least presented themselves as to be spoken aloud.

So that’s significant in itself. But what we really need – or at least, what I would like – is some sense of how the Romans talked about reading the Latin novels or the technical treatises or other kinds of writing which didn’t present themselves as part of the world of forensic rhetoric. It may well be that the kind of reading imagined for those texts was also talked about in terms of declamation, which would be very interesting indeed.

A semiliterate person might be much more likely to put in spaces for a semiliterate audience.

I’m a bit uncomfortable with making this divide between literate and semiliterate, especially if the Vindolanda letters are included in the latter. The difference may just as easily be between formal and informal, or professional and everyday writing (and maybe reading). It might be parallel to the difference between drawing up a contract and writing a letter; or the difference in presentation between a church missal and an exam script; or between typesetting your textbook in LaTex vs. dashing off a blog post.

It could be the case that spaces were the ‘literate’ norm and that scribes and other professionals were taking them out for the sake of a hyperliterate audience. In that case the difficulty would be part of the charm: writing and reading scriptura continua would be a way of advertising your education or your copious spare time, and would be very different from what an ordinary literate person would use in a letter to his/her friends.


candle 07.08.11 at 4:55 pm

I recommend “Reading in the Brain” by Stanislas Dehaene

Irrelevantly, I would just like to recommend to everyone Georges Perec’s Experimental demonstration of the tomatotopic organization in the Soprano (Cantatrix sopranica L.), with its citation to Donen, S. & Kelly, G. Singing in the brain. Los Angeles, M.G.M. Inc. Press. 1956.

I’m going to have to stop this long-post-followed-by-short-follow-up habit, for which I shall blame reading too much Holbo.


Brandon 07.08.11 at 5:45 pm

Anyway, all of this has led to the development of a model of reading that’s too complicated for me to present here, but that model clearly indicates that scriptura continua should take longer for the brain to process than space-delimited text.

Since people don’t actually read at the same rates, whatever the speed of brain processing is, what we actually would need to know here is the degree of typical overlap in speed between reading continuous text (by someone primarily familiar with it rather than separated text) and reading separated text, and the extent of the difference. If, for instance, the difference in brain processing speed for separated text allows for greater reading speed, but only for optimal readers (those with unusually small time gaps between word recognition and integrated understanding of the text as a whole), or if the difference in speed at the same level of comprehension amounted to (just for example) a second per hundred words, this would make the difference useless for the type of argument considered here — as far as the effects considered here are concerned, such a difference would for practical purposes be as good as no difference. The difference would have to be (1) discernible in a large body of typical rather than optimal readers and (2) signficant enough to have the effects in question. (An interesting result, for instance, would be if the difference made very little difference for optimal readers but a very big difference for typical readers.) Admittedly, (1) is tricky for historical reasons: the population of typical readers in the first century could very well have a higher proportion of optimal readers than it would have in the thirteenth century; but things like this would, of course, be relevant to the viability of the argument.

One of the reasons I keep hitting this point is that it seems on the evidence very, very likely that the primary impetus for separated text are due not to advantages for reading but to advantages for copying. Copying requires nothing but word recognition (well, in a strict sense it requires nothing but recognition of letter-shapes, but obviously it would be highly facilitated by word recognition), whereas reading requires in addition integrated comprehension. In this sort of context, then, we really need to know whether separated text’s value for things like word recognition has a major effect on typical reading speeds, or whether it is small enough to be swamped out by other things.

For instance, I had no problem reading Holbo’s continuous text example — certainly it went slower than separated text, and certainly I had to backtrack a few times due to initial misreading that would have been unlikely with separated text, but it wasn’t a huge difference. So I lose, at conservative estimate, a sizable fraction of a minute per paragraph, and that without having gotten used to it. I’m very much closer to the optimal reader side of the reading speed spectrum than the opposite, so that might make a difference, and it may very well be that I would never read continuous text as quickly as I can read separated text, but the difference here is slight enough that one can’t obviously attribute much to it; it’s difficult to build a causal explanation in which being able to read a few words per minute faster would plausibly have such a massive effect (if it did we should be able to see similar effects, at least partially, in differences between slower and faster reading populations today, so that might be another place to look to strengthen the argument).


Gene O'Grady 07.08.11 at 8:18 pm

Candle is right to object to my use of semi-literate and literate, particularly in regard to Vindolanda, about which I know very little. And I should also observe that the level of literacy, and what literacy means, is a very disputed area in classical studies right now.

The things I used to know a bit about (Pompeiian and Roman graffiti) and some of the papyri and Ostraca from the Roman East and Egypt do show both a literate-semi-literate- barely literate spectrum and a formal-informal spectrum. I also think that the use of the term hyperliterate is useful, given that so-called classical Latin represents a massive effort on the part of people like Cicero and Virgil (and maybe Gaius Gracchus and Gallus and Calvus, for all we know) to create an elevated and flexible literary language, with Greek culture as some kind of model or rival in the background.

Some readers may be interested to know that we apparently possess a few letters written by Bar-Kochba, the leader of the Jewish revolt, and they are all in Greek.


candle 07.08.11 at 8:50 pm

The things I used to know a bit about (Pompeiian and Roman graffiti) and some of the papyri and Ostraca from the Roman East and Egypt do show both a literate-semi-literate- barely literate spectrum and a formal-informal spectrum.

Yes, that’s fair enough. I didn’t mean to imply that no one could be classed as semi-literate, but just that we shouldn’t take Cicero as normal and everyone without his education and training as struggling to make it through a shopping list. But yeah, literacy in the ancient world is mostly a matter of surmise and extrapolation. So much like everything else in the ancient world, then.

Now I think about it, it might be interesting to think of scriptura continua as the equivalent of typesetting, with all the room that leaves for ordinary reading and writing (ie handwriting, in the modern world). Claudia Severa’s use of a scribe in the Vindolanda birthday invitation might equate to using calligraphy on your wedding stationery.


maidhc 07.08.11 at 9:27 pm

Most lutenists can sightread lute tablature. As michael e sullivan points out, it indicates what string you play the note on, which regular music notation does not.


Doctor Science 07.08.11 at 11:10 pm

Chris Crawford @51:

Is the process different for Chinese? I wouldn’t be surprised if one-symbol-per-word scripts were faster to read than alphabets, PROVIDED you are a skilled & experienced reader. I believe both James Fallows and Language Log have discussed problems in China and Japan with literacy degradation after schooling ends, as all but the heavy readers gradually forget many symbols (kanji, in Japan).

I have also gathered that one-symbol-per-syllable scripts are the easiest to learn — a lot of the Cherokee population got up to basic literacy (at least, the ability to send letters) within a week or so after being introduced to Sequoia’s script. Many of the scripts missionaries devised for writing North American languages are syllabaries, because with them they could spend less time teaching reading and more time teaching the Bible.


Gene O'Grady 07.08.11 at 11:53 pm

I promise to be quiet, but one last perhaps relevant item on scriptura continua is that its use may have something to do with the cost and the scarcity of material to write on. The data on the price of papyrus is hard to interpret, the data on parchment is even worse, but the common practice of palimpsesting (is that a word) and reuse of the reverse of pieces of papyrus indicates that this may have been a consideration.


nnyhav 07.09.11 at 2:25 am


John Holbo 07.09.11 at 3:07 am

“As a guitar player, I’m not sure your point is made here.

I don’t see anything inherent in tab that makes it harder to sight read. ”

The analogy isn’t perfect, and my point was really a compound one – not every bit if which holds, on reflection – but I think it’s right that tab is not sight-readable in a certain sense. I doubt anyone can look at an unfamiliar piece of music, in tab notation, and have an immediate idea what it sounds like. That’s because the tab doesn’t denote the notes but the position on the fretboard. It’s also because tab doesn’t include any notation for rhythm.

Basically, standard musical notation represents pitch, and is semi-iconic of pitch: higher on the staff means higher note, so forth. Tab represents what your fingers should do, and is semi-iconic of finger-position. So it’s very hard to ‘see’ in tab what the musical outcome should sound like because, strictly, that isn’t even represented in this notation. I don’t deny that someone might be able to sight-read it effectively. But it would be hard, and I doubt anyone would ever have occasion to become a facile sight-reader of tab. Tab is not for sight-reading. It’s for figuring out where your fingers go, rather laboriously. So that then you can do it over and over many times, and eventually it sounds good.

There are obviously exceptions. If the piece is really simple. Just a power chord, chunk-chunk-chunk-chunk, then same thing up a string, then up a step, you know you’ve got some sort of 1-4-5 basic thing, and you know what that sounds like. But if what’s being tabbed is a significantly unfamiliar musical structure, I doubt anyone could sight-read it effectively.


Scott Martens 07.09.11 at 10:35 pm

I relied a lot on Spaces Between Words in the first part of doctoral thesis, but I came to a different conclusion about the origins of “wordhood”. While I can’t criticize it for its theory about why spaces were invented in the West, the notion that language could be broken up into word-like chunks is far, far older, and I proposed something quite different to explain it. The reason you can got without spaces in Latin, Greek and Sanskrit and get away with it is because most words have only a few possible suffixes. Readers notice those implicit barriers.


John Holbo 07.10.11 at 12:30 pm

That’s interesting Scott. I found myself puzzled by the same thing. I agree that it seems the concept of word has to go way back. I don’t think that Saenger really says different, but he does hint different, which is puzzling. I think it’s probably fair to say that lots of ordinary scribes in the pre-word separation period would have produced significantly different results if you’d held a gun to their heads – after explaining what a gun does, of course – and told them to introduce spaces between ‘the words’. Some would have separated every morpheme. Others would have run together whole phrases. In the absence of a clear decision procedure, you can see why there would need to be some special impetus. ‘Just do it’ wouldn’t quite cut it.

Your point about suffixes is probably right, at least that’s part of it.


Scott Martens 07.12.11 at 9:02 am

John, I’ve been looking high and low for research on just that. There are plenty of languages that use no explicit word markers in writing. Increasingly though, public education is imposing a notion of word on those people so they can use dictionaries. There *ought* to be a study on how Chinese people naturally cut up their texts while there are still a few left who never learned about words in school.

Lacking that… Latin was written with a form of word demarking punctuation until the 2nd century, and there is research on how consistently it was used. Let me save you the reading: A conjugation or declension suffix was the best predictor of a word boundary. Prepositions and pronouns were routinely left unmarked. Priscian, writing in the 5th century, was visibly unable to make distinctions between prepositions and prefixes when they were spelled and pronounced the same.

I was unable to find any instance of separating at every morpheme in any written language. Every syllable, yes. Early Brahmi texts – the Asokan pillar edicts in particular – separated every syllable in some cases. But they were clearly under the influence of Aramaic. The Semitic languages separated words systematically, although there was no canonical division until very late. But as early as 1100 BC, you see word boundaries in Phonecian and Aramaic. Hebrew manuscripts show a fairly high level of inter-scribe agreement in the Dead Sea texts.

But Semitic languages usually are not written with vowels. They needed markers to be readable. Itsonethingtoreadtextswrittenlikethis. Btlkths? Much easier to read that as “Bt lk ths” Disemvoweled and unspaced, no Semitic text longer than a short inscription is readable.

The conclusion I came to – which I think no one else has floated – is that adding vowel markers to Brahmi and Greek resulted in the loss of interword markers. I found some supporting evidence in early Greek inscriptions, and noted that Hebrew stopped being written with word boundaries about at the same time as the invention of niqqud vowel markers. I would have liked to have traced the same history in Ge’ez, but lacked the resources.

As for Latin, they dropped their interword punctuation because they imitated the Greeks – that’s well documented. I found some bitching in the Venerable Bede’s works about how young British scribes misread texts because they don’t know where the spaces are. The impetus for the British/Irish invention of spaces may have come from the cultural disconnect between young Britons in the 8th century and the Latin tradition they were supposed to be upholding. Saeger seems right on in seeing the space as becoming entrenched because it was *useful*, but it’s origins may simply have been as a pedagogic tool for getting young anglos (and Celts) over the language barrier. That’s the same reason Monier-Williams gave for adding spaces to Sanskrit in his 19th century textbooks.

So… It’s still more or less right to suggest we have spaces because the English can’t read too good.


John Holbo 07.12.11 at 9:47 am

Thanks Scott, that’s quite interesting and, potentially, actually rather useful re: some stuff I’m writing on. Maybe you could drop me a line and share some scholarly references: jholbo at mac dot com.


Ed 07.13.11 at 10:01 pm

I am disappointed to see that the parent thread to this one ( has been closed to new comments.

I took Chris’s point to heart and went and read the Wald, which is both a great book and an object lesson in the value of truth in advertising. As Mark Richardson says in the entry on ‘How the Beatles Destroyed Rock’n’Roll’ in the rather good Pitchfork round-up of music books ( “People I know who love this book lament the title.”

Comments on this entry are closed.