TEI Lite: a 90% solution

Lou Burnard

TEI Lite: a view of the TEI

what do most people want, most of the time?

realistic for existing texts, e.g. OTA, UVA, HTI...

realistic for document production, e.g. TEI technical documentation

see http://www.tei-c.org/Lite/

Tutorial now available in French, Italian, Russian, Japanese, and Korean

Basic structure(s)

A TEI-conformant document comprises a header followed by a text

the header is essential for:

bibliographic control and identification

resource documentation and processing (see later)

Structure of a TEI text

A text may be unitary or composite

a unitary text contains

front matter

back matter

a body

in a composite text, the body is a group of texts (or nested groups)

Basic TEI structure...

A text usually has divisions

generic, hierarchic subdivisions

vanilla or numbered

type attribute

associated head and trailer elements from the divtop class

for example...

Use of global attributes

Applicable to all elements

id for unique identification

n for (non-unique) name or number

rend for rendition (appearance)

lang for language and hence writing-system

Extensible, like other classes

Character Encoding Recommendations

non normative

extend, using standard entity sets or transliteration

document transliteration scheme with formal Writing System Declaration

Text components in TEI Lite

What are divisions made of?

prose is mostly paragraphs (<p>)

verse is mostly lines (<l>), sometimes in hierarchic groups (<lg>)

drama is mostly speeches (<sp>) containing <p> or <l> and interspersed with stage directions (<stage>)

These may be mixed, and may also appear directly within undivided texts.

Verse: an example

Drama: an example

Texts are not just words...

… but probably only people know that

an encoding may claim to capture

just visual salience,

just its assumed causes

both

encoding makes explicit one (or more) sets of interpretations

For example...

And this Indenture further witnesseth that the said Walter Shandy, merchant, in consideration of the said intended marriage...

"And this Indenture further witnesseth"

And this Indenture further witnesseth that the said Walter Shandy, merchant, in consideration of the said intended marriage...

Who does the work?

TEI scheme allows for close reading -- and the reverse

can tag very detailed features of discourse function

can normalise or simplify (e.g. dates numbers, names)

… or leave well alone

Phrase level elements

are often by convention typographically distinct

“data-like” (names, numbers, dates, times, addresses)

editorial intervention (corrections, regularizations, additions, omissions ...)

cross references and links (see later)

for example...

Direct speech

Use the who attribute to show speakers

Speeches can be nested in other speeches

“Foreign” language phrases

The lang attribute may be attached to any element

Use <foreign> if nothing else is available

Define each language in <langUsage> in header

Referring strings

The <rs> (referring string) element is used for any kind of name or reference

Dates, times, numbers

attributes can be used to quantify <date> and <dateRange> expressions

similarly, times <time>, <timeRange> and numbers <num>

Correction and Regularization

<corr> and <sic> for correction (or non-correction)

<reg> and <orig> for regularization (or the reverse)

Omissions, Deletions, Additions

<gap> omission by transcriber

<del> cancellation in source or by editor

<add> or <supplied> insertion in source or by editor

<unclear> material uncertain because illegible

<damage> physical damage to text carrier

The multiple hierarchy problem

XML allows only one hierarchy at a time

Is a document

chapter-paragraph-phrase

gathering-page-leaf

or both?

discontinuous segments

links and milestones

Boundary markers

page, column, and line breaks (<pb>, <cb>, <lb>)

generic <mileStone>

Some chunks are also phrases

<list> lists of all kinds

<note> notes (authorial or editorial)

<figure> pictures or figures

<formula> formulae

<table> tables

<bibl> bibliographic descriptions

Lists

use <list> for lists of any kind (use type attribute to distinguish)

use <label> in two-column lists as alternative to n attribute

may be nested as necessary

for example...

Figures and graphics

The presence of a graphic is indicated by the <figure> element

The title of the graphic is tagged as a <head>

A description of the graphic may be supplied (as a <figDesc>) for use by software unable to render the graphic

The graphic itself is specified as an external entity

for example...

Tables

a <table> element contains <row>s of <cell>s

spanning is indicated by rows and cols attributes

role attribute indicates whether row or column holds data or a label

embedded tables are permitted

for example...

Bibliography

Use simple <bibl> with optional subcomponents:

<respStmt> (for any kind of responsibility) or <author>, <editor>, etc.

<title> with optional level attribute

<imprint> groups publication details

<biblScope> adds page references etc.

Use <listBibl> for list of references

for example...

Notes

Use <note> for notes of any kind (editorial or authorial)

if in-line, use place attribute to specify location

if out of line, either

use target attribute to specify attachment point

or mark attachment point as a <ref>

for example...

Out of line bibliographic notes

<p>As Blenkinsop <ref target=“N32”> 32</ref> remarks …



<p>As Blenkinsop <ptr target=“N32”/> remarks …

Links and pointers

cross-referencing

association of text and annotation

association of image and text or audio and transcript

alignment of text and translation...

Terminology

A pointer points from here (where it is) to there (somewhere else)

A ref does the same, but has some content

A link points to two or more places and asserts some (linking) relation between them. Its own location is not significant

An anchor exists only to be pointed at

Cross References

Use <ptr> (empty element) or <ref> (with content)

use target to specify an identifier (ID value)

HTML-style pointers in TEI

a URL identifies an external entity

it is pointed to by an xref or xptr

TEI X-pointers

Allow you to point outside the current document

“location ladder” technique

15 different location methods

TEI X- pointers

target specified by location ladder (within external entity named by doc attribute)

The most reliable location methods are tree based

... and links

freestanding links can associate anything that has an ID, including x-pointers

can also be grouped and typed

Slide 47

A three way alignment

Front and back matter

contain generic divisions and <titlepage>

which contains arbitrary mixture of <titlePart>s and

stated author:<docAuthor> and <byLine>

original title and edition: <docTitle>, <docEdition>,

original imprint and date: <docImprint> <docDate>

Example titlepage elements

Not covered here...

specialised front and back matter

analytic tagging

segmentation

interpretations

the header

tags for documentation

Summary

How TEI Lite handles…

Structural divisions

Rendition vs. interpretation

Phrases, chunks, and chunky phrases

Pointers and links

Any dtd dealing with ordinary text will need a similar range


	what do most people want, most of the time?
		realistic for existing texts, e.g. OTA, UVA, HTI...
		realistic for document production, e.g. TEI technical documentation
	see http://www.tei-c.org/Lite/
	Tutorial now available in French, Italian, Russian, Japanese, and Korean


	A TEI-conformant document comprises a header followed by a text
	the header is essential for:
		bibliographic control and identification
		resource documentation and processing (see later)


	A text may be unitary or composite
	a unitary text contains
		front matter
		back matter
		a body
	in a composite text, the body is a group of texts (or nested groups)


	generic, hierarchic subdivisions
	vanilla or numbered
	type attribute
	associated head and trailer elements from the divtop class


	Applicable to all elements
		id for unique identification
		n for (non-unique) name or number
		rend for rendition (appearance)
		lang for language and hence writing-system
	Extensible, like other classes


	non normative
	extend, using standard entity sets or transliteration
	document transliteration scheme with formal Writing System Declaration


		What are divisions made of?
			prose is mostly paragraphs (<p>)
			verse is mostly lines (<l>), sometimes in hierarchic groups (<lg>)
			drama is mostly speeches (<sp>) containing <p> or <l> and interspersed with stage directions (<stage>)
		These may be mixed, and may also appear directly within undivided texts.


	… but probably only people know that
	an encoding may claim to capture
		just visual salience,
		just its assumed causes
		both
	encoding makes explicit one (or more) sets of interpretations


	And this Indenture further witnesseth that the said Walter Shandy, merchant, in consideration of the said intended marriage...


	TEI scheme allows for close reading -- and the reverse
	can tag very detailed features of discourse function
	can normalise or simplify (e.g. dates numbers, names)
	… or leave well alone


	are often by convention typographically distinct
	“data-like” (names, numbers, dates, times, addresses)
	editorial intervention (corrections, regularizations, additions, omissions ...)
	cross references and links (see later)


	Use the who attribute to show speakers
	Speeches can be nested in other speeches


		The lang attribute may be attached to any element
		Use <foreign> if nothing else is available
		Define each language in <langUsage> in header


	The <rs> (referring string) element is used for any kind of name or reference


	attributes can be used to quantify <date> and <dateRange> expressions
	similarly, times <time>, <timeRange> and numbers <num>


	<corr> and <sic> for correction (or non-correction)
	<reg> and <orig> for regularization (or the reverse)


		<gap> omission by transcriber
		<del> cancellation in source or by editor
		<add> or <supplied> insertion in source or by editor
		<unclear> material uncertain because illegible
		<damage> physical damage to text carrier


	XML allows only one hierarchy at a time
	Is a document
		chapter-paragraph-phrase
		gathering-page-leaf
		or both?
	discontinuous segments
	links and milestones


	page, column, and line breaks (<pb>, <cb>, <lb>)
	generic <mileStone>


	<list> lists of all kinds
	<note> notes (authorial or editorial)
	<figure> pictures or figures
	<formula> formulae
	<table> tables
	<bibl> bibliographic descriptions


	use <list> for lists of any kind (use type attribute to distinguish)
	use <label> in two-column lists as alternative to n attribute
	may be nested as necessary


		The presence of a graphic is indicated by the <figure> element
		The title of the graphic is tagged as a <head>
		A description of the graphic may be supplied (as a <figDesc>) for use by software unable to render the graphic
		The graphic itself is specified as an external entity


	a <table> element contains <row>s of <cell>s
	spanning is indicated by rows and cols attributes
	role attribute indicates whether row or column holds data or a label
	embedded tables are permitted


	Use simple <bibl> with optional subcomponents:
		<respStmt> (for any kind of responsibility) or <author>, <editor>, etc.
		<title> with optional level attribute
		<imprint> groups publication details
		<biblScope> adds page references etc.
	Use <listBibl> for list of references


	Use <note> for notes of any kind (editorial or authorial)
	if in-line, use place attribute to specify location
	if out of line, either
		use target attribute to specify attachment point
		or mark attachment point as a <ref>


	<p>As Blenkinsop <ref target=“N32”> 32</ref> remarks …
	<!-- or equivalently -->
	<p>As Blenkinsop <ptr target=“N32”/> remarks …


	cross-referencing
	association of text and annotation
	association of image and text or audio and transcript
	alignment of text and translation...


		A pointer points from here (where it is) to there (somewhere else)
		A ref does the same, but has some content
		A link points to two or more places and asserts some (linking) relation between them. Its own location is not significant
		An anchor exists only to be pointed at


	Use <ptr> (empty element) or <ref> (with content)
	use target to specify an identifier (ID value)


	a URL identifies an external entity
	it is pointed to by an xref or xptr