# Every source the open page sets draw on: what it is, its licence, and what # it is used for. build.py writes this catalogue into each package's manifest # (ext_aurasense_pageset.sources), and `attribution` into every board's licence. # # Rule: only material whose licence is compatible with CC BY-SA 4.0. No # vendor vocabulary (Core First, Crescendo, WordPower, Gateway, Super Core, # Essence, LAMP Words for Life, Unity, Speak for Yourself) is a source of any # word, position, wording or name, and no PCS, SymbolStix, Widgit, METACOM or # ARASAAC symbol is used. attribution: >- Page sets of the AAC open standard (Transaction Science), CC BY-SA 4.0. Word choice draws on the Universal Core vocabulary (Center for Literacy and Disability Studies, UNC Chapel Hill; CC BY 4.0 per UNC, CC BY-SA 4.0 per Project Core; words only), on Banajee, DiCarlo & Buras Stricklin (2003), and on word frequencies from FrequencyWords 2018 by Hermit Dave (CC BY-SA 4.0; counts from OpenSubtitles 2018, http://www.opensubtitles.org). Word classes are coloured after the Fitzgerald Key (Fitzgerald 1954). Symbols are named from Mulberry Symbols, OpenMoji and Blissymbolics, each CC BY-SA 4.0, with their own attributions on each image. sources: universal_core: what: "Universal Core vocabulary: 36 words for beginning communicators" by: Center for Literacy and Disability Studies, University of North Carolina at Chapel Hill url: https://www.med.unc.edu/healthsciences/clds/universal-core-vocabulary/ also: https://project-core.com/communication-systems/ licence: CC-BY-4.0 (UNC) / CC-BY-SA-4.0 (Project Core); treated as CC-BY-SA-4.0 file: sources/universal_core.txt used_for: >- Membership only: every Universal Core word is on the Growing Core by 6x8 ("finished" as "all done", the app's word with a gloss and a symbol); Universal Core words weigh more in Fixed Paths and make up its starter filter. Not used: its symbols (they belong to their vendors) and its 6x6 layout. project_core: what: Project Core professional development modules and implementation model by: Center for Literacy and Disability Studies, UNC Chapel Hill url: https://project-core.com/professional-development-modules/ licence: CC-BY-SA-4.0 used_for: >- The principle behind the smallest grids: start with core words usable in every activity, no nouns, and grow without moving them (DESIGN-core.md). No text of the modules is reproduced. banajee_2003: what: >- Banajee, M., DiCarlo, C., & Buras Stricklin, S. (2003). Core vocabulary determination for toddlers. Augmentative and Alternative Communication, 19(2), 67-73. url: https://doi.org/10.1080/0743461031000112034 licence: a published fact (the list of words the study reports); cited file: sources/banajee_2003.txt used_for: >- Membership only: its 23 entries are on the Growing Core by 7x9 (all but "mine", on People); they weigh more in Fixed Paths and join its starter filter. subtitles_2018: what: FrequencyWords 2018, English (word counts from OPUS OpenSubtitles 2018) by: Hermit Dave url: https://github.com/hermitdave/FrequencyWords/tree/525f9b560de45753a5ea01069454e72e9aa541c6 licence: CC-BY-SA-4.0 (content); counts originate from OpenSubtitles (http://www.opensubtitles.org) file: sources/en_subtitles_2018_top5000.tsv used_for: >- Spoken frequency, the best open proxy for conversation we have: film and TV dialogue. It weighs Fixed Paths' words and orders its pages, orders the Growing Core's Actions page, and is the reference list of the "top N spoken words" metrics. Already a source of the app's own Korean, Malayalam, Telugu and Thai language packs (tools/lang/build_packs.py), at the same commit. wordfreq_en_pack: what: The app's English language pack (wordfreq 3.2.0 and Common Voice), data/lang/en.*.bin licence: CC-BY-SA-4.0 used_for: >- Not used for word choice (it mixes Wikipedia, news and books with speech). Cited because it is the frequency the app's own prediction uses. vocabulary_rs: what: "The app's English core vocabulary (aac-core src/vocabulary.rs)" cites: Banajee et al. 2003; Cross et al. 2006; Yorkston et al. 2010 (as cited there), and CORE_COVERAGE_EVIDENCE licence: Apache-2.0 (code); a word list used_for: >- Word classes: a word's ext_aurasense_word_class agrees with the app's own class wherever the app has one (the build lists any difference), and the app's level of each word is shown beside its stage. fitzgerald_key: what: >- Fitzgerald, E. (1954). Straight Language for the Deaf. Colour coding by word class as modified by Goossens', Crain & Elder (1992), as cited in aac-web defaults.rs. licence: a public method used_for: The colour of each word class (our own palette; DESIGN-core.md). symbols: what: Mulberry Symbols 3.6.1, OpenMoji 17.0.0, Blissymbolics (BCI-AV) licence: CC-BY-SA-4.0 each file: crates/aac-core/src/symbols/index.json (concept -> symbol id per set) used_for: >- Each button names the symbol the app draws for its words (or, for a page key, a chosen exemplar). Words with no symbol yet are listed, never given an invented id. glosses: what: "The app's glosses of its English words in 44 languages (aac-core translations_data.json)" audits: gloss_audit_2026-09-28.json, gloss_audit_2026-10-01.json (model readers; no native speaker yet) licence: Apache-2.0 (with the code) used_for: >- Every button's `translations`. These are glosses, never another language's core vocabulary (SPEC ยง4.1): every language but English is GlossOnly (data/pagesets/languages.json). nav_labels: what: "The words for the Back and Home controls in 44 languages (aac-core src/nav_labels.json, shared with the app)" licence: Apache-2.0 (with the code) used_for: The translations of Back and Home cells, where both readers agreed. unimorph_eng: what: UniMorph English (Wiktionary-derived inflection tables) url: https://github.com/unimorph/eng licence: CC-BY-SA-3.0 (adaptable under 4.0) used_for: Not yet. The source planned for irregular word forms (DESIGN-core.md, Word forms).