I given a brief sketch of the early visual display of "alphabetic" language that I think leads to three conclusions: First, in alphabetic cultures there has been a slow development towards an increasing visualization of written language. Second, as the way language is displayed changes, so does the way it must be processed. Third, because of the second conclusion, I do not think the study of reading can be divorced from the way language is written, which means that if a computer model is to be of significant help in understanding human processing, it will have to closely model the input that humans use.
1.2 Skoyles (1992) raises the issue of Hebrew, which is written in two forms today: with diacritical marks for vowels (pointed) and with just the consonants (unpointed). He writes that "all Israeli children start off reading pointed text, in spite of its being more complex (and different) visually, compared to unpointed text. Attempts to teach children without the indication of vowels by diacritics have not been successful." Hebrew in antiquity and now is written with breaks (spaces) between the words.
1.3 When the Greeks took over the Phoenician alphabet, however, they radically changed the display. Classical Greek was written in a continuous script with no breaks in an absolute sense: none for paragraphs, sentences, phrases, or words. In short, there was no punctuation whatsoever. Latin sometimes (and especially in inscriptions) separated words with "dots" or "points" (interpuncts), but in the second century A.D., in one of those great misbegotten regressions, it adopted the "scriptura continua" of the Greeks. In other words, the addition of vowels by the Greeks made words sufficiently distinctive to allow them to be run together and still be understood. If unpointed Hebrew were written all run together, it would be a nightmare to read. The diacritics, like the vowels, add information that makes the decoding easier, even if they are unnecessary or redundant to the expert. For this reason, classical Greek is now printed with diacritical marks.
1.4 Antiquity did not recognize the separation of the visual and aural aspects of text that enable us to treat print and speech differently. Seneca believed that separating words, as Latin sometimes did, denoted a stylistic difference between the Romans and the Greeks not just in writing, but also in speaking:
The pace of Quintus Haterius, a celebrated speaker of his day, is something I should have a sensible man keep well clear of: with him there was never a hesitation or a pause, only one start and only one stop.
But I also think that certain styles are suitable in a greater or lesser degree to different nationalities. In a Greek one will tolerate this lack of discipline, while we have acquired the habit of punctuating what we say, in writing as well as speech. (Seneca, Letters, 40.11-12)
The idea that writing in its physical layout should reflect the rhythm of speech foreshadows modern poets with their special printing effects. What is curious is that Seneca found the run-on style a strain to understand, saw a connection between writing and speech, and never thought of reforming the way Latin was written.
1.5 Antiquity knew the concept of the word as a unit of speech, but did not consider important the concept of the written word as a visual unit, with the result that children learned how to read solely by phonetics and not by the whole word. Consider Dionysius of Halicarnassus' description of how children learn to read:
When we are taught to read, first we learn by heart the names of the letters, then their shapes and their values, then, in the same way, the syllables and their effects, and finally words and their properties, by which I mean the ways they are lengthened, shortened, and scanned; and similar functions. And when we have acquired knowledge of these things, we begin to write and read, syllable by syllable and slowly at first. It is only when a considerable lapse of time has implanted firmly in our minds the forms of the words that we execute them with the utmost ease, and we read through any book that is given to us unfalteringly and with incredible confidence and speed.
Never in antiquity was the whole word method used to teach reading. Because all the letters were run together, the major problem every reader faced was figuring out what was a word. The only way to do that is syllable by syllable. This fact has a number of implications and not just for reading. For example, I think that the arranging of words in unbroken blocks impeded the development of our rich set of textual search aids, because most of these aids depend on the individual word. If you want to find something in a dictionary, thesaurus, encyclopedia, almanac, or most any reference, you look it up by the unit of the word.
1.6 Each element in our current repertory of punctuation marks has a separate introduction and each has generally taken a minimum of one hundred years to be fully adopted. Conversely, there are definite limits to the amount of punctuation and variety in arrangement of words that are useful in increasing our understanding of a written text.
For example, experiments have been done with the "visual chunking" of text to see whether such arrangements would increase comprehension.
They do not, except for poor readers (Crowder & Wagner 1992, 23-24)
1.7 Erasmus (1466?-1536), writing in a period when punctuation was still a novelty, puts it extremely well:
Just as the principle of symmetry, when properly applied, makes great improvement in the appearance of script, so you would hardly credit how much good punctuation contributes to the understanding of a passage; so much so that a certain scholar said rather wittily that punctuation was a kind of commentary of the text. It is useful therefore that a boy should quickly familiarize himself with this feature also, and that the habit should become second nature so that, even when his mind is on something else, he should, as he writes, attend to the punctuation marks as much as to the letters themselves. (Quoted in Saenger & Heinlein 1991, 255 n. 59)
I believe that the development of full punctuation, as we have it today, is directly related to the switch between reading texts out loud, as was the general practice in antiquity, and reading them silently, as became increasingly common through the Middle Ages. Saenger (1990, 54-55; 1982, 377-379) maintains that the adoption of word separation began first in the British Isles because readers there learned Latin as a second language and needed the extra help of the spaces to parse Latin texts into words. It may also be of interest to know that Greek often used the verb "akouein" ("to hear") in contexts where it must mean "to read" (Schenkeveld 1992).
1.8 Today the computer raises new issues concerning how much information is necessary to understand a particular text:
Cn u rd ths? F u cn, thn u cn str mor txt n ls spc.
In other words, text, as we are accustomed to seeing it, contains a lot of redundancy, which enables us to understand it far more quickly. The trade-off is between time to decode and the amount of storage space needed:
F u hv t tm, thn u cn rd stff lk ths.
If you don't, you prefer whole words with complete punctuation. Otherwise we would all be reading "zipped" files. Claude Shannon calculated that written English is about 50% redundant. The issue, however, is not whether you can read such a format, but whether you want to or should have to. Hence when Skoyles (1992) calls the "pointed" form of Hebrew more complex, he is considering it solely visually rather than from the point of view of information theory, in which it is simpler just because it is more elaborate in display.
2.1 I have given a very brief sketch of the early visual display of "alphabetic" language that I think leads to three conclusions: First, in alphabetic cultures there has been a slow development towards an increasing visualization of written language. Second, as the way language is displayed changes, so does the way it must be processed. Third, because of the second conclusion, I do not think the study of reading can be divorced from the way language is written, which means that if a computer model is to be of significant help in understanding human processing, it will have to closely model the input that humans use.
Crowder, R.G. and Wagner, R.K. (1992) The Psychology of Reading. An Introduction, 2nd Edition. Oxford University Press.
Dionysius of Halicarnassus. On Literary Composition, 25. Translation from the Loeb Classical Library.
Saenger, P. (1982) Silent Reading: Its Impact on Late Medieval Script and Society. Viator 13: 367-414.
Saenger, P (1990) The Separation of Words and the Order of Words: The Genesis of Medieval Reading. Scrittura e Civilta 14: 49-74.
Saenger, P. and Heinlein, M. (1991) Incunable Description and Its Implication for the Analysis of Fifteenth-Century Reading Habits. In: S. L. Hindman (ed.), Printing the Written Word. The Social History of Books, circa 1450-1520. Cornell University Press. Pp. 225-258.
Schenkeveld, D. M. (1992) Prose Usages of AKOYEIN "To Read." Classical Quarterly 42: 129-141.
Seneca. Letters from a Stoic (trans. R. Campbell) (Harmondsworth 1969).
Skoyles, J. R. (1991) Connectionism, Reading and the Limits of Cognition. PSYCOLOQUY 2(8) reading.1.
Skoyles, John R. (1992) Not All Phonological Reading Need Use Accurate Letter-Sound Rules. PSYCOLOQUY 3(6) reading.7.