Document modeling

A backward glance

and

More on DTDs

C. M. Sperberg-McQueen, Black Mesa Technologies LLC

Rev. 18 September 2012

Nearby documents

Overview

Organizational notes

Bureaucracy, paperwork, and so on ...

Signups

Reminder about
  • minute-taking
  • vocabulary survey

Questions from last week?

Anything we need to clear up before proceeding?

Tour of vocabularies, cont'd

This week: TEI
  • Ryan Buller
  • Tynan Burke

Historical survey - Introduction

Where are we?

You are here

Overview
Representation of individual documents
  • introduction to SGML and XML
    • syntax (angle brackets)
    • model (trees)
    • DTDs
  • → historical survey
Schema languages
  • DTDs
  • XSD
  • Relax NG

But also here

Overview
Representation of individual documents
  • introduction to SGML and XML
    • syntax (angle brackets)
    • model (trees)
    • (→) DTDs
  • → historical survey
Schema languages
  • DTDs
  • XSD
  • Relax NG

What is document modeling?

Remember how we defined document:
  • (for physical documents) an artifact carrying writing; typically an artifact whose main purpose is to carry writing
  • (for electronic documents) an object managed by a computer system which resembles a non-electronic document in function

Some obvious questions

The basic technology: writing

In which we worry about foundations.

What is writing?

Coulmas* identifies three “fundamental characteristics” of writing:
  • artificial graphical marks on durable surface
  • intended to communicate something
  • has conventional relation to language
* Florian Coulmas, The writing systems of the world (Oxford: Blackwell, 1989), p. 17.

What is writing?

Sampson* is more general:
To ‘write’ might be defined, at a first approximation as: to communicate relatively specific ideas by means of permanent, visible marks.
* Geoffrey Sampson, Writing systems: A linguistic introduction (Stanford: Stanford University Press, 1985), p. 26.
Contrast:
  • sign language
  • works of art

A recursive loop?

Q. If writing is an attempt to represent some (ideational or linguistic) content, then ...

... doesn't that mean that writing is itself an attempt to model ideas and/or language?

So is document modeling just a second-level attempt to model something that is itself already a model?

A. Yes and yes, it certainly looks that way.

Benjamin's Kunstwerk

Cf. Walter Benjamin, Das Kunstwerk im Zeitalter seiner technischen Reproduzierbarkeit (1936).
A selective, tendentious summary: reproductions
  • may degrade / represent imperfectly;
  • may approach ubiquity;
  • not fixed in time or space;
  • lack the aura of the original.

Language and technical reproducibility

Q. What technology enables the technological reproducibility of language?
A. Writing.
Q. How does writing model language?

Kinds of writing

Sampson offers this typology*:
Writing:
  • semasiographic (?)
  • glottographic
    • logographic
      • based on polymorphemic unit (e.g. word) (?)
      • morphemic
    • phonographic
      • syllabic
      • segmental
      • based on phonological features
* “(?)” mark categories S sees as open to question.

Semasiography

Writing that represents ideational content directly.
  • Amerindian ‘picture writing’ and other remote examples (?)
  • language-independent* instruction manuals
  • international signage icons
  • mathematics
  • diagram languages (UML, chemical bond diagrams, ...)

Glottography

Writing that represents ideational content indirectly, by representing language.*
  • one word, one symbol? (unattested*)
  • one morpheme, one symbol (Han characters*)
  • one syllable (form), one symbol (katakana, hiragana, Linear B, Cherokee, ...)
  • one phonetic segment, one symbol (most alphabets)
  • one phonological feature, one symbol (Han-gul)
Note: most actual writing systems are mixtures, not pure.
* So document models model documents, which model language, which models ideas?

Some oppositions

  • iconic (motivated) vs arbitrary
  • complete vs incomplete (defective)
  • deep vs shallow
  • linear vs. non-linear*

All writing systems are incomplete

N.B. no writing system captures all aspects of language:
  • lexical identity (ambiguity)
  • suprasegmentals (intonation, volume, timbre)
  • vowels (?)
At one extreme, we may be unable to distinguish semasiography from very incomplete glottography.

The writing substrate

  • tablet
  • stone
  • sheet
  • scroll
  • codex

A simple document model

What document model does a Linear B inscription have?
  • A document
  • is a (relatively short) sequence
  • of syllables.

Scroll, codex, and linearity

Scrolls are sequential access devices.
Codices are random access or direct access devices.
Service books, missals.

Printing

The second most important technology.

Woodblocks

7c China (for Buddhist scriptures); 15c Europe (block books).
Mixture of text and image.

Movable type

11c China (baked clay)
13c Korea (metal)
15c Europe (metal)
What characters to cast into type?

Industrial printing

Post-incunabular books develop secondary non-linear properties:
  • title page
  • front, back matter
  • running heads
  • catch words
  • page numbers
  • signature numbers

Envelope-pushing

Both before and after print, some books go beyond reproduction of language.
  • initials, miniatures
  • indexing (12c, first Biblical concordances)
  • image poetry

Facsimiles

A logical continuation of woodblock line: an analog representation.
Not: text and image, but text as image.

Codes

In which we speed through history.

Semaphores

Semaphore telegraph / Napoleonic semaphore:
  • line-of-sight communication
  • angled rods (or shuttered panels)
  • one position, one symbol (in French case, two arms × seven positions × one crossbar × 4 angles, 7 × 7 × 4 = 196 symbols; two-symbol codes using 92 positions, 8,464 meanings (92 × 92).
  • code books

Flag semaphores

Also 19th century (?)
  • A-Z
  • 1-9 (= A-I), 0 (= K)
  • rest / space
  • error / attention
  • invitation to transmit (= K)
  • acknowledge / correct (= C)
  • shift-to-numbers
  • shift-to-letters (= J)

Morse code

1836-1844, Morse et al.
  • A-Z
  • 0-9
  • gaps (1, 3, 7 dots long)
  • spelled-out punctuation, or punctuation conventions
    • full stop (AAA)
    • comma (MIM)
    • etc. (. , ? ' / ( ) : ; = + - " @)
  • prosigns (message control): wait, invitation to transmit, error, end of work, understood, starting signal

The document model of Morse code

For Morse code (and flag semaphore), a document is a sequence of characters and gaps.

Punched cards

In which we arrive on the threshold of electronic computing.

Hollerith cards

1890 census: Herman Hollerith.
Unit record equipment:
  • card punch
  • collator
  • sorter
  • tabulator
  • interpreter

Punched-card (‘Hollerith’) code

In standard design (IBM, 1928), a card has:
  • 80 columns
  • rectangular holes
  • ten rows for digit punch
  • two rows for zone punch
     ______________________________________________
    /&-0123456789ABCDEFGHIJKLMNOPQR/STUVWXYZ
 Y / x           xxxxxxxxx
 X|   x                   xxxxxxxxx
 0|    x                           xxxxxxxxx
 1|     x        x        x        x
 2|      x        x        x        x
 3|       x        x        x        x
 4|        x        x        x        x
 5|         x        x        x        x
 6|          x        x        x        x
 7|           x        x        x        x
 8|            x        x        x        x
 9|             x        x        x        x
  |________________________________________________

Punched-card model

In early uses, card divided into fields.
  • columns 1-20: name
  • column 21: sex
  • columns 23-24: age
  • ...
Cf. simple database records.

Horizontal text

One way to represent texts with cards:
  • columns 1-72: one line of text
  • columns 73-80: line ID (play, act, scene, line; book, chapter, verse; ...)

Further on DTDs

Some constructs you'll see.

Notation declarations

<!NOTATION
name
external identifier
>
Examples (in Oxygen)

Parameter entity declarations

<!ENTITY
%
name
external identifier
optional ‘internal subset’
>
Examples (in Oxygen)

What got left out of XML DTDs

Several SGML DTD constructs support features omitted from XML.
  • SHORTREF (<!SHORTREF foo "bar" baz>, <!USEMAP foo barracuda>)
  • RANK
  • and-connector (&)
  • case folding of element, attribute names

What got trimmed out of XML DTDs

Several SGML DTD constructs were trimmed down in XML.
  • default entities (entity name #DEFAULT)
  • tag omissibility (<!ELEMENT list-item - O (#PCDATA | %phrase;)*>
  • attribute types (NUMBER, NUMBERS, NUTOKEN, NUTOKENS)
  • inclusion and exclusion exceptions

Assignments

Due: Sunday morning 30 September 2012.
1 Do a TEI version of your document, valid against some schema or DTD built by the TEI's Roma software. (If you want to keep it simple, use TEI Lite.)
2 Questions about ISO 8879 Annex E.
3 Simple DTD syntax quiz.