The electronic document, like traditional kinds of documents, is based on writing. Writing is a graphical form of language, though its exact linguistic nature is subject to debate, even after 5,000 years of literate history. Depending upon the philosophical perspective of the analyst and on the way that writing is used in a given context, a written text can be considered either as a recording of speech, in which case it is a subsidiary level of linguistic representation, or as a direct expression of language, in which case it is an alternative to speech, but expressed visually rather than aurally.

The different modes of perception used by speaking and writing make for important differences between speech and text. The acoustic stream of speech is temporal and linear, but the graphical image of text is spatial and planar. Writing has always been durable, but speaking was ephemeral until the development of acoustic recording technology. Moreover, although reading is a sequential decoding of a text image, text elements can also be perceived simultaneously, whereas speech sounds are perceived sequentially.

An analysis of the text image reveals some aspects that represent various elements of speech and language, and other aspects that appear to be artifacts of the graphic medium, though sometimes such distinctions are subtle.

Electronic documents in their graphical expression use the form of writing called typography. A fundamental principle of typography is the dual nature of the image: the complementary relationship between figure and ground (positive and negative space) at each level of organization. At the level of the letter, there is the marked form, and the unmarked space inside it, traditionally called the counter. At the level of the word, there is the grouping of letters, and the spaces between them, or letter-spaces. At the level of the line, there is the word space. At the level of the column, there is the space between lines, or leading. At the level of the page, there are the spaces separating and surrounding the columns, or gutters and margins.

When a graphical and linguistic unit coincide, the ground space is significant; it marks some element of speech or language. For example, in modern English orthography, the word space differentiates two words. When the graphical and linguistic units do not coincide, the space still serves a purpose; it organizes the text in the plane of the page, for the reading process. For example, the spaces at the beginnings and endings of lines define a column as a coherent body of text, though they do not necessarily have any linguistic significance. Other uses of the ground may have significance at higher levels of text organization, such as paragraph indentations, or the line endings, indentations, and line spacings of poetry.

Beyond the page, the text of a document is organized by higher level linguistic groupings, such as chapters, by symmetry principles, such as repetition, pattern, homology, etc., and by conceptual structures, as seen in the indexed array of pages used in the codex form of books.

The structure of a writing system usually emphasizes a particular level of language, isolating certain elements for representation and ignoring others. Alphabetic writing represents phonemes. The Graeco-Roman alphabet represents both consonantal and vocalic phonemes, whereas the traditional Hebrew and Arabic alphabets represent consonants only (though there is some provision for vowels). Logographic writing represents words or morphemes. Chinese writing and Japanese Kanji (based on Chinese) are logographic. Syllabic systems and hybrid systems are also used.

The nature of the signs used to represent these linguistic elements can be iconic or symbolic, in the semiological terminology developed by C. S. Pierce, or, in the semiological terminology of De Saussure, motivated or unmotivated. Many writing systems began with iconic or motivated signs which resembled some object, and which later evolved into symbolic or unmotivated signs that were abstract shapes. The histories of cuneiform, hieroglyphics, the alphabet, and Chinese and Japanese writing exhibit this tendency, although to different degrees.

At the level of individual signs, the alphabetic system has become abstract and symbolic, but this has encouraged the evolution of an additional dimension of signification, that of typographic variation. Since the 15th century, the typographic alphabet has become successively richer and more varied through a process of amalgamation in which stylistically different alphabetic forms are joined together in typeface families. Formal oppositions like majuscules and minuscules (capitals and lower-case), or roman and italic, define dimensions of graphic contrast that are used to mark significant features of a text.

The antiquity of the amalgamation determines the strength of the bond between the variations, probably because the more ancient formal distinctions have had more time and opportunities to become incorporated into the standard semantics of the text. CAPITALS and lower-case were first conjoined early in the 15th century, and today it is almost obligatory to include both in a standard character set like the ASCII set or a typewriter font. Roman and italic types were first conjoined in the middle of the 16th century, and today italic is a necessary companion to roman in most literary texts. Normal and bold weights were first conjoined in the 19th century, and today boldface is a frequent supplement to italic in many typographic texts. Seriffed and sans-serif types began to be used together a few decades ago, and appear to be on the way to forming an extended family of typographic variation.

In English orthography, there are reasonably well-defined rules for the use of capitals and lower-case, though these change with time. Capitalization in an 18th century book is often different than it would be today. Most publishing houses and periodicals have manuals of style that set forth rules for the standard usage of italic and bold faces, though these rules may be less rigid than those for capitalization.

Semantic interpretation of graphical variation appears to be more effective with abstract, symbolic systems than with concrete, iconic systems. This can be seen in modern workstations as well as in historical scripts. In a workstation user-interface that supports both icons and text, the icons will generally be graphically invariant (other than by video-reverse), whereas the text will often support variations such as italic and bold face.

The graphical invariance of the icons in this context suggests that they are not the same kind of writing as alphabetic or logographic systems. This is borne out by the non-syntactic nature of most screen icons -- unlike letters or characters, they cannot be combined into larger and more complex expressions -- and by their lack of standardization; the forms of icons vary from system to system. It may be that screen icons are an inchoate writing system still in the process of formation, or that screen iconography will remain a ``pictographic'' system (to use an older terminology) or a ``semasiographic'' system, to use a more recent terminology for signs that represent objects rather than elements of language.

The different graphical aspects of the text image each deserve a thorough analysis. The basic principles of text organization are part of the passive graphic vocabulary of most literates, but an active, refined understanding of such principles was, in traditional literate societies, the province of the professional scribe, and today is the domain of the typographer and graphic designer. This understanding has in the main been intuitive, though in comparison to the other visual arts, typography is more susceptible to rational analysis.

The following paper, ``The Design of Lucida: an Integrated Family of Types for Electronic Literacy'', is an attempt to provide an analysis of the design decisions that determine the visual appearance of a typeface. As such, it is focused on a relatively low level of the typographic image, the letterforms. Analyses of the visual principles of the higher levels of text organization will be necessary if the design and formatting of electronic documents is to become part of computer science as well as a part of the graphic arts.

Copyright Ó 1986 by Charles Bigelow