Glossary
Terms as Glycerine Aksara uses them. Several have a narrower or simply different sense here than in ordinary editorial usage; those are flagged.
Akṣara
One outlined region of a manuscript image together with an ordered sequence of graphemes.
Not a syllable
The system attaches no syllabic structure to an akṣara and imposes no rule about what may go inside one. śu is ś + u, two graphemes; it is not an entry in any catalogue. A word boundary may fall inside an akṣara. See Akṣaras and Transliteration.
Called a segment in the IIIF endpoints and in URLs.
Annotation
A record that points at something and says something about it. An edition is a stack of them: the akṣara annotates a region of the image, and every layer above annotates the layer below. See Annotations.
Attestation
An occurrence of a lemma in a particular place in a particular edition. The lemma is shared across the corpus; the attestations shown to you are those in the edition you have open.
Canonical lemma
A lemma that has been through review and approved, becoming shared property. Approval removes its owner, which is why its creator can no longer edit it.
Not a Sanskrit or Pali headword
"Canonical" is a review status, nothing to do with reference languages. The Sanskrit or Pali form recorded beside a Gāndhārī headword is a cognate, and private lemmas can carry cognates too. See The Shared Lexicon.
Canvas
The IIIF term for one image surface in a manifest — here, one manuscript photograph. Akṣara outlines are recorded against a canvas.
Chāyā
The Sanskrit or Middle Indic rendering of the Gāndhārī. Stored per sequence alongside the translation, and readable as continuous prose in the Viewer.
CKM
The manuscript's catalogue designation, recorded on the text and carried into the manifest metadata. Distinct from the reference, which is the short citation form.
Clarity
A controlled vocabulary describing how legible a grapheme is. A property of the writing's survival, recorded separately from any editorial intervention — it is not a text-critical mark.
Cognate
A (language, form) pair recorded on a lemma — the Sanskrit, Pali, or other comparandum. A structured value rather than a free-text note, so it can be queried: the Editor searches lemmas by their Skt. cognate, and treats lemmas sharing one as related.
Collection
A flat grouping of texts, by provenance, project, or repository. Many-to-many: a text can belong to several collections or none, and nothing else depends on the grouping.
Deprecated lemma
A canonical lemma withdrawn from circulation. Tokens already linked to it keep the link; it can no longer be chosen for new work. Deprecation exists because canonical lemmas cannot safely be deleted.
Diplomatic, hybrid, reconstructed
The three display modes for a transliteration. Diplomatic shows what survives; reconstructed shows the editor's full restored reading; hybrid falls between, marking restorations in place. The report shows all three stacked. See Display modes.
Edition
One scholar's reading of one text: its akṣaras, word division, lemmatisation, structure, translation, and chāyā. Owned by exactly one account, never co-edited, and never merged with another. A text may carry any number of editions and none ranks above the rest. See Editions.
Grapheme
A written sign from a fixed catalogue — the unit an akṣara is spelled with. Each carries a type (consonant, vowel, punctuation) and a sort code fixing its place in Indic alphabetical order. Editors choose from the catalogue; they do not add to it.
Homograph order
A number distinguishing lemmas that share a spelling. Part of the sort key, after the sort code.
Inflection
A bundle of grammatical values drawn from controlled vocabularies — part of speech, and whichever of case, number, gender, person, tense, and voice that part of speech admits. It holds no free text, and which fields appear is computed from the part of speech. Shared across editions like a lemma. See Inflections.
Ink mask
A per-image map, made by a machine-learning model, of how likely each pixel is to be ink. Used only by the Editor's Fit to ink, to trace akṣara outlines; it is not part of an edition. See Ink Masks.
Lemma
The dictionary headword a token is filed under, with its translation, homograph number, cognates, and optional dictionary link. Lemmas are shared by every user of the installation, not owned by an edition — the reason for the review workflow. See The Shared Lexicon.
Line
An ordered run of akṣaras — a physical line of writing. Sits directly on the akṣara layer, independently of tokens, so a word may cross a line break without difficulty.
Manifest
The IIIF document describing one text: its images as canvases, its metadata, and its editions' annotations. Generated live from the database, and the thing the Viewer's iiif-content parameter names. See IIIF and Deep-Zoom Images.
Obscuration
A controlled vocabulary recording what is covering or damaging a grapheme — as against clarity, which records how legible the result is.
Plain image
An image referenced by URL as a single ordinary file, with no tiling and no derivative service, as opposed to one served by the IIIF image server. Displays correctly but downloads in full. See IIIF and Deep-Zoom Images.
READ
The Research Environment for Ancient Documents, from which this system's shared lexicon was seeded and whose lemma-sorting algorithm it reproduces. Relevant when lemma ordering or homograph numbering looks familiar.
Reference
The short citation form of a text, such as 22-2 — the prefix appearing in attestation labels throughout the Viewer. Distinct from the CKM designation.
Segment
The name for an akṣara in the IIIF endpoints and in URLs. Same thing.
Sequence
An editorial unit — a section, a sentence, a verse — grouping tokens and carrying that unit's translation and chāyā. Sequences nest to any depth. The only layer that is purely editorial: everything else asserts something about the manuscript. See Sequences.
Sort code
A numeric code per grapheme fixing its position in Indic alphabetical order, and by extension the ordering of lemmas and glossaries. Variant forms sort immediately after their base, so ḵ follows k. Not the Latin alphabetical order of the transliteration. See Sorting follows the script.
Text
One manuscript: its title, CKM designation, reference, and photographs.
Images belong to the text
Every edition of a text shares one set of images. There is no per-edition image set. See The Data Model.
Text-critical mark
An editorial sign recording an intervention — restoration, deletion, addition, and the rest — with its conventional brackets. Applied per grapheme, not per akṣara, so a vowel can be restored while its consonant stands as read. Six are available. See Text-Critical Marks.
Token
A word: an ordered run of graphemes drawn from one or more akṣaras, carrying a lemma and an inflection. A token holds no coordinates of its own — its position is the akṣaras it names — and it records where within the first akṣara it starts, which is how word boundaries fall inside a sign. See Tokens and Lines.
Transliteration
The rendering of the recorded graphemes in Latin script. Generated from the data rather than typed: the vowel carrier is stripped, marks are bracketed, and the result depends on the display mode.
Vowel carrier
A synthetic grapheme, ʔ, stored before any vowel that opens an akṣara or follows another vowel, so that every vowel has a consonant before it for sorting purposes. Hidden from the picker and stripped on display — it is why a + i renders as aï. You never enter it. See The vowel carrier.