Consider two propositions. First, word processors have improved by leaps and bounds since their advent about ten years ago. Second, what we actually do with word processors has not changed very much in the last decade. Both these statements may be true, but they create a tension: if the technology has been so revolutionary why should things remain pretty much as they always were?
One resolution to this slight conundrum is that the revolution has not yet begun in earnest. So far we have been watching a dress rehearsal: word processing is going to change radically in the next decade, and will make a huge difference to the way we write and think about writing. The improvements to the technology of word processing witnessed so far are preparatory to the really significant language and computing in a new light.
These changes are unlikely to come about through the gradual accretion of even more features to the already excellent packages in the market. Do we seriously imagine that Word Perfect 12.0 will arrive on our desks pretty soon with a manual three feet thick? Our guess is that several software innovations which are already hovering on the edge of the mainstream will blossom over the next five or ten years. These innovations will create niches for newcomers and simultaneously force the major packages to focus on parts of the language software market which they are best at. We will recognize that there is much more to language processing software than the rather similar functions covered by the best contemporary word processing packages.
Speech processing. It seems highly likely that speech processing will move more into the mainstream of personal computing. As ever with innovation, it may not happen in quite the way we first imagined. When speech processing was first mooted five years ago, it was seen as a way of circumventing the keyboard. Some of us have trouble with basic typing, and nearly all of us have difficulties with rarely-used command sequences or macros. While we look forward to the day we can instruct our machines to "print three copies", significant applications for speech processing are likely to appear before we get reliable automatic dictation machines. We will incorporate speech messaging well before we crack the problem of unconstrained speech recognition.
Most of the documentation that moves around corporations and between businesses is remarkably standard. The average document comes in about five almost-identical versions: file copies, agenda, schedules, minutes, ‘blind’ copies sent to colleagues, action copies sent to subordinates and so on. The five copies generally differ only in their destination and minor detail. A problem with such documentation is that it tends to be regarded as dross by its recipients, even if it is in fact important. An ideal way of giving apparent dross a high "impact factor” is to attach a voice message to some email.
Incorporating sound bites in documents may seem a superficial addition to the basic word processing function. But it is in line with the tendency towards object-oriented programs and user interfaces, hypertexts and multi-media documents. And all these powerful trends in the technology will encourage us to think of word processing as much broader than the typing or typesetting functions predominant in today’s word processing and desktop publishing programs. While word processing will encompass voice and graphics, we can also count on an equally strong tendency towards a more abstract and artificial view of the document as 'structured program'.
The drive to uncover structure in documentation is already very strong in the defence industries. We have all heard the story about the documentation for a Boeing 737 weighing more than the aeroplane itself, or the American frigates which carry more tons of paper than they do tons of missiles. But this mass of paper is no joke for the industries which build these machines, and they are consequently taking the lead in developing powerful tools for the automatic processing and interpreting of documentation.
SGML. One of the keys is the development of Standard Generalised Markup Language (SGML) schemes: SGML allows for documentation manipulation which is independent of the particular way in which the document has been processed or represented — on paper, screen, or magnetic tape for example. SGML has a real pay-off when your are dealing with large amounts of text. In typesetting or desktop publishing systems there is no way of distinguishing the italic into which you cast the title of a book from the italic you use for a foreign word.
The Next workstation, with its voice messaging and CD-ROM-based Digital Library is a pointer to the future.
But in SGML systems these differences are marked. The basic idea is that if you can treat the structure of documentation as abstract and declarative rather than being procedural and dependent on the manner of its representation, it should be much easier to make large-scale comparisons of documents. SGML markup is not easily intelligible to the human eye — quite the reverse, as it tends to look like a jumble of brackets and codes.
But new software will use SGML while concealing it from the user, in much the same way as our word processors now use and exchange ASCII without our needing to notice it. SGML will come into its own when users are able to incorporate documents, standard form contracts or advertising brochures into their databases, agreement files or catalogues without needing to consciously translate the structure to the formats they prefer to use.
