Design principles

Binding terminology and architecture decisions.

LIBRORUM models the book along two axes: content versus form, and the book as read versus the book as produced. Everything else follows from this.

Six regions

Region Content
Λ The book as read, in the general linguistic sense (written or spoken): graphemes, writing systems, word forms, sentences, verse lines, divisions, references, judgements.
Φ The observable book: what a reader sees, hears or touches (pages, sounding book, tactile book).
C_cont Digital content: the encoded text, markup, encoded grammars.
C_form Digital form: typeset documents, speech synthesis, braille files; the concrete presentation grammar.
M_cont Material content: copy marked for the press, manuscripts of the author's copy.
M_form Material form: hot-metal composition, plates, ribbons, braille plates.

The physical book is a carrier, not a region. Region names are always capitalized (C_form, C_cont, M_form, M_cont).

Modelling criterion

A linguistic unit is modelled only when some form depends on it: a presentation grammar acts on it, an encoding labels it, or something observable realizes it. Internal linguistic structure with no consequence for form is left to linguistics.

Writing systems, scripts, graphemes

A writing system is the written modality of a language; it belongs to exactly one language, while a language may have several. A script is an inventory of shapes (visual or tactile) shared by many writing systems. Graphemes belong to writing systems, never to scripts.

Linear order is linguistic

Units of Λ stand in a linear order that no presentation grammar can change. The grammar places only the more loosely connected elements (notes, running heads, illustrations, tables).

Presentation grammars

Abstract presentation grammars are relational, value-free conventions of a tradition; concrete ones hold the effective values for a given rendition. They exist for visual, sounding and tactile modalities.

An ontology decides nothing

Structural constraints belong to validation; open competences (lexicality, orthographic correctness, grammaticality) are recorded as judgements, results of interpretive acts, never as facts. Meaning is only ever a claim scoped to a reading event.

Stores of states

The production rows store states of the book. Mechanical transitions (:typesetFrom, :printedFrom, :synthesizedFrom…) are direct; human transitions summarize a path through Φ and Λ.

Peritext, epitext, metatext

Following Genette, paratext = peritext + epitext. Whatever is encoded, typeset or recognized on the page is peritext. Metatext (:Didascalia) is a functional notion: what directs the realization.

Terminology

Say illustration, never figure. A table and an illustration are distinct: a table may break across pages, an illustration is kept whole.