-------------------------------------------
Simplex annotations and complex annotations

DG 2004-04-27, internal lexicography workshop
-------------------------------------------

DESCRIPTION OF A SPEECH OR LANGUAGE CORPUS

General term: annotation
Term used in speech technology: labelling
Term used in text technology: markup

A simplex annotation (e.g. label) is an event, e.g. an occurrence of a word,
phoneme, syllable, feature at a specific interval in a corpus.

A complex annotation consists of one or more tiers of events.

A tier is a sequence of events of the same type.

--------------------

Allen relations:
- calculus of intervals
- 13 relations

--------------------
Event logic: axiomatic relational logic
	=def	pair of a property and an interval
- Johan van Benthem, Amsterdam
- Applied to phonology in order to explicate autosegmental phonology
	by Steven Bird and Ewan Klein (1989) - "Event Phonology"
		event
		<property, interval>
		<[attribute: value], interval>
		<[attribute: value], <t_start, t_end>

Examples of annotation:

Xwaves (esps-waves+):	<property,t_end>
				 e.g.: `table	1.030'
	problem: the beginning is only implicit and has to be
		inferred by the user or added in ad hoc fashion
		- implicit, partial interval definition

SAM:		<property,<t_start,t_end>>
			e.g.: `table	1.030	1.659'
	corresponding to <[orth: table], <1.030, 1.659>>

Praat:	same as SAM, but with its own notation
TASX:	same as SAM, but with XML notation

-------------------------------------------------------------------------

HOW DOES THIS RELATE TO THE LEXICON?

- Lexical acquisition
	- list of lexical items, e.g. a wordlist
		- problem: what is a word? 
		- make list by...
			- converting the text to a list of words
			- sorting the list of words
			- removing duplicates
		- extract corpus properties of the list items
			(i.e. microstructure elements which can be
			inferred from corpus relations)
			by...
			- frequency count (absolute or relative / percent)
			- rank ordering 

- Lexical representation

	- macrostructure: overall structure of dictionary
	- mesostructure: generalisations over microstructures
		- definitions of grammar, pronunciation, ...
		- cross references (e.g. semantic relations)
		- references to corpus (e.g. concordance, examples)
	- microstructure: types of lexical information, `DATCATS',
		(data categories), e.g.
		- STRUCTURAL PROPERTIES (can be extracted from corpus)
			- external context, e.g. collocations
			- internal structure, e.g. derived, compound words,
				idioms
		- INTERPRETATIVE PROPERTIES
			- MEANING: semantic, pragmatic
			- FORM: phonetic, orthographic
		- METADATA PROPERTIES (local housekeeping properties)
			- lexicographer
			- source
			- dates of creation, modification
			Note: there are global metadata properties which
				which apply to the whole lexicon, e.g.
				- language
				- corpus used
				- publication details
	Note:
	- macrostructure contains mesostructure contains microstructure

- Lexical access