GMEOW Languages Module

• IRI: https://blackcatinformatics.ca/gmeow/slices/languages
• Tier: extension
• Group: extensions

What This Slice Covers

This slice owns 165 terms and contributes 16 mapping or projection rows. Use it when its terms match the native fact you want to preserve; use the linkage tables to see how those facts leave GMEOW for consumer vocabularies.

Dependencies

• gmeow:slices/expertise
• gmeow:slices/kernel
• gmeow:slices/language
• gmeow:slices/observations
• gmeow:slices/provenance
• gmeow:slices/temporal

Consumers

• Rich sociolinguistics (proficiency, scripts, varieties, diachrony) extending the core language slim slice.

Local Map

Examples

Proficiency

• Source: slices/extensions/languages/examples/proficiency.ttl
• GMEOW terms: gmeow:LanguageProficiency, gmeow:Person, gmeow:ProficiencyLevel, gmeow:ProficiencyScale, gmeow:knowsLanguage, gmeow:langEnglish, gmeow:langFrench, gmeow:name, gmeow:nativeLanguage, gmeow:profModalityReading

# SPDX-FileCopyrightText: 2026 Blackcat Informatics® Inc. <paudley@blackcatinformatics.ca>
# SPDX-License-Identifier: CC-BY-4.0
#
# Worked example: language proficiency is PER-MODALITY . A flat
# gmeow:nativeLanguage / gmeow:knowsLanguage says WHICH languages an agent has;
# the reified gmeow:LanguageProficiency says HOW WELL — and crucially per
# gmeow:proficiencyModality (reading / speaking / listening / signing …), because
# the same person is often strong reading a language and weak speaking it. Each
# proficiency is read against an explicit named gmeow:proficiencyScale (here
# CEFR), so a level is never a bare adjective. Languages reuse the core seeds
# (gmeow:langEnglish / langFrench).
@prefix gmeow: <https://blackcatinformatics.ca/gmeow/> .
@prefix ex:    <https://blackcatinformatics.ca/gmeow/examples/languages/> .
@prefix rdfs:  <http://www.w3.org/2000/01/rdf-schema#> .

ex:dana a gmeow:Person ;
    gmeow:name           "Dana Reyes"@en ;
    gmeow:nativeLanguage gmeow:langEnglish ;
    gmeow:knowsLanguage  gmeow:langFrench .

ex:cefr a gmeow:ProficiencyScale ; rdfs:label "CEFR"@en .
ex:b2   a gmeow:ProficiencyLevel ; rdfs:label "B2 (upper intermediate)"@en .
ex:a2   a gmeow:ProficiencyLevel ; rdfs:label "A2 (elementary)"@en .

# --- Strong at READING French (B2) ...
ex:frenchReading a gmeow:LanguageProficiency ;
    gmeow:proficiencyAgent    ex:dana ;
    gmeow:proficiencyLanguage gmeow:langFrench ;
    gmeow:proficiencyModality gmeow:profModalityReading ;
    gmeow:proficiencyLevel    ex:b2 ;
    gmeow:proficiencyScale    ex:cefr .

# --- ... but only A2 SPEAKING it — same language, different modality.
ex:frenchSpeaking a gmeow:LanguageProficiency ;
    gmeow:proficiencyAgent    ex:dana ;
    gmeow:proficiencyLanguage gmeow:langFrench ;
    gmeow:proficiencyModality gmeow:profModalitySpeaking ;
    gmeow:proficiencyLevel    ex:a2 ;
    gmeow:proficiencyScale    ex:cefr .

Terms

Classes

Properties

Individuals

Linkages

• Rows: 16
• Projection profiles: schema-org
• External vocabularies: doap, schema, wd, wdt

Guide

Languages — modelling & interoperability guide

Most vocabularies treat a language as an opaque tag: inLanguage "ja". That single assumption — a language is its ISO/BCP-47 code — structurally excludes the languages that matter most to a forward-looking model:

GMEOW inverts the assumption, exactly as the names block did for personal names.

Governing tenet: registry-independent identity

A gmeow:Language has a self-minted IRI. Registry codes — BCP-47, ISO 639, Glottolog, Wikidata — are optional alignments (gmeow:languageCode, gmeow:authorityLink, skos:exactMatch), never the primary key. A code-less conlang or AI-language is therefore a fully first-class, co-equal language. tests/test_languages.py enforces that nothing requires a code.

To isolate the GMEOW graph from registry changes and support code-less conlangs, all internal literals must use private-use BCP-47 tags (e.g. @en) for any GMEOW-namespaced property. The language entity's gmeow:languageTag functional property links the entity to its internal tag.

@prefix gmeow: <https://blackcatinformatics.ca/gmeow/> .
@prefix ex:    <https://example.org/lang/> .

# First-class language individuals define their private-use tags:
ex:english a gmeow:Language ;
    gmeow:languageTag "en" ;
    gmeow:languageCode "en" .

ex:ithkuil a gmeow:Language ;                          # NO languageCode — and that's fine
    gmeow:languageTag "und" ;
    gmeow:languageOrigin gmeow:originConstructedEngineered ;
    gmeow:designGoal "Maximal cognitive precision with minimal ambiguity." ;
    gmeow:wasAttributedTo ex:quijada ;                 # a human creator…
    gmeow:hasAppellation [
        gmeow:fullName "Ithkuil"@en ;
        gmeow:nameLanguage ex:english ;
        gmeow:namePurpose gmeow:namePurposeGlossonym
    ] .

ex:aiLang a gmeow:Language ;
    gmeow:languageTag "und" ;
    gmeow:languageOrigin gmeow:originAiGenerated ;
    gmeow:wasAttributedTo ex:modelAgent .              # …or a gmeow:SoftwareAgent

★ The registry tag isn't discarded — it's reconstructed on demand by the projection layer (see below), so schema.org consumers still get ja-Hani etc.

Scope: one umbrella, many kinds

gmeow:Language covers natural, constructed (auxiliary / engineered / artistic / ritual), AI-generated, historical / reconstructed, sign / whistled / tactile, and formal languages. The single structural split is gmeow:FormalLanguage → gmeow:ProgrammingLanguage (grammar-defined, machine modality, no native speakers); a software project links to one via gmeow:writtenInLanguage. Everything else is distinguished by value vocabularies: languageOrigin, languageModality (spoken / signed / written / whistled / tactile / machine / multimodal), languageStatus.

Writing systems are first-class — and co-mingled

A language uses many co-equal scripts at once. Japanese interleaves four, each in a distinct role — so "a language has one script" is as wrong as "a person has one name", and the fix is the same reified-usage relator (WritingSystemUsage, mirroring names' NameUsage):

ex:japanese gmeow:usesWritingSystem ex:wsHan , ex:wsHiragana , ex:wsKatakana , ex:wsRomaji .
ex:wsuJaHan a gmeow:WritingSystemUsage ;
    gmeow:usageLanguage ex:japanese ; gmeow:usageWritingSystem ex:wsHan ;
    gmeow:scriptRole gmeow:scriptRoleLogographicContent .
# …three more usages, one per script + role.

The usage also carries a period (gmeow:scriptUsageInterval / validFrom/validUntil), so a script change over time is just a closed usage — Turkish in Arabic script until 1928, then Latin. Bespoke and non-linear scripts (Ithkuil) are first-class: a WritingSystem may have no scriptCode and writingSystemType gmeow:wsTypeNonLinear.

Versions are a first-class lineage

A language evolves; AI languages version fast. Each version is itself a usable gmeow:Language (a gmeow:LanguageVersion), linked up to its lineage by gmeow:versionOf and ordered by gmeow:supersedes / gmeow:wasDerivedFrom — older versions stay first-class.

ex:ithkuil2011 a gmeow:LanguageVersion ;
    gmeow:versionOf ex:ithkuil ; gmeow:versionLabel "2011" ;
    gmeow:supersedes ex:ithkuil1993 ; gmeow:wasDerivedFrom ex:ithkuil1993 .

Proficiency — reified, per-skill, leveled

A person knows a language to a degree, and differently per skill. GMEOW reifies this (gmeow:LanguageProficiency, again the NameUsage idiom) — mint one per (agent, language, modality), so "native overall" and "B2 writing" coexist:

ex:profDeWriting a gmeow:LanguageProficiency ;
    gmeow:proficiencyAgent ex:learner ; gmeow:proficiencyLanguage ex:german ;
    gmeow:proficiencyModality gmeow:profModalityWriting ;
    gmeow:proficiencyLevel gmeow:cefrB2 ; gmeow:proficiencyScale gmeow:scaleCEFR .

The base gmeow:knowsLanguage (≈ schema:knowsLanguage) and gmeow:nativeLanguage relations state that an agent knows a language; gmeow:LanguageProficiency adds the level and the skill.

Transformations are functions (FnO)

Transliteration, transcription and translation are functions — script→script or language→language — so GMEOW catalogues them declaratively as fno:Functions (projections/transforms.fno.ttl): Hepburn, Kunrei, Pinyin, Wade-Giles, Revised Romanization, ISO 233 / 15919, IPA transcription, and a generic translate. This closes a loop in the names block: gmeow:romanization now names the system that produced it via gmeow:transliterationScheme, and each scheme links to its FnO function.

ex:yamadaName gmeow:romanization "Yamada Tarō"@und ;
    gmeow:transliterationScheme gmeow:schemeHepburn .   # records HOW, not just WHAT

Language varieties — contested classifications without a winner

Dialect, sociolect, register, idiolect, localized variant, generational slang, standard, creole, pidgin, koine — these are all modeled as gmeow:LanguageVariety, a gufo:SubKind of gmeow:Language that inherits every language property (names, scripts, tags, status, provenance). The classification itself is a standpointed claim, not an OWL subclass decision.

The language-vs-dialect distinction is not an OWL class hierarchy decision. A single LanguageVariety entity can carry multiple varietyKind assertions from different standpoints, and none is privileged (Principle 9).

ex:scots a gmeow:LanguageVariety ;
    gmeow:languageTag "sco" ;
    gmeow:varietyKind gmeow:kindLanguage , gmeow:kindDialect ;
    gmeow:varietyOf ex:english .

# Standpoint reifiers — both coexist, neither wins.
ex:ax-scots-language a owl:Axiom ;
    owl:annotatedSource ex:scots ; owl:annotatedProperty gmeow:varietyKind ;
    owl:annotatedTarget gmeow:kindLanguage ;
    gmeow:accordingTo ex:standpoint-snl ; gmeow:confidence 0.85 .

ex:ax-scots-dialect a owl:Axiom ;
    owl:annotatedSource ex:scots ; owl:annotatedProperty gmeow:varietyKind ;
    owl:annotatedTarget gmeow:kindDialect ;
    gmeow:accordingTo ex:standpoint-academic ; gmeow:confidence 0.90 .

Querying all varietyKind assertions with their standpoints:

SELECT ?variety ?kind ?standpoint ?confidence WHERE {
    ?variety gmeow:varietyKind ?kind .
    OPTIONAL {
        ?ax a owl:Axiom ;
            owl:annotatedSource ?variety ;
            owl:annotatedProperty gmeow:varietyKind ;
            owl:annotatedTarget ?kind ;
            gmeow:accordingTo ?standpoint ;
            gmeow:confidence ?confidence .
    }
}

varietyKind and varietyOf are both non-functional, so a single variety can hold multiple classifications and multiple parentage claims simultaneously. A superseded classification is suppressed with gmeow:displayable false (Principle 10), never erased.

LanguageVersion vs LanguageState — distinct purposes

A version can have states, and a state can describe a version, but neither is defined as the other:

ex:ithkuil2011 a gmeow:LanguageVersion ;
    gmeow:versionOf ex:ithkuil ; gmeow:versionLabel "2011" .

ex:ithkuil2011State a gmeow:LanguageState ;
    gmeow:stateLanguage ex:ithkuil2011 ;
    gmeow:stateStatusValue gmeow:statusConstructedActive ;
    gmeow:stateAuthority ex:standpoint-academic ;
    gmeow:stateInterval ex:modernEnglishInterval .

LanguageState follows the VersionMembership relator pattern: it binds {language} × {status} × {authority} × {interval}, inherits confidence / displayable / temporal scope from Observation, and bridges to the universal claim stack via stateLanguage ⊑ observedFeature and stateAuthority ⊑ vantage.

LanguageChangeEvent — diachronic arcs

Historical linguistic changes are first-class events: sound shifts, borrowing, standardization, extinction, revival, and more.

ex:greatVowelShift a gmeow:LanguageChangeEvent ;
    gmeow:changeType gmeow:changeSoundShift ;
    gmeow:affectedLanguage ex:english ;
    gmeow:eventInterval ex:greatVowelShiftInterval .

Because LanguageState is an Observation, the existing bitemporal query (slices/core/temporal/queries/tql/bitemporal.rq) works out of the box: ask "what was the status of English as of 1200 CE?" and receive the Middle English state with its standpoint and confidence.

Interoperability — the four-layer stack

The schema.org projection (run gmeow project) emits schema:Language / schema:ComputerLanguage, both endonym and exonym as co-equal schema:names, the flattened schema:knowsLanguage, and — via fnComposeBcp47 — the BCP-47 tag as schema:alternateName (de+Latn → de-Latn; Japanese yields ja-Hani, ja-Hira, ja-Kana, ja-Latn). Lossy drops: WritingSystemUsage, origin / modality / status, version lineage, LanguageCreation, and the proficiency level. Language instances coreference Lexvo (http://lexvo.org/id/iso639-3/…), Glottolog and Wikidata in data via skos:exactMatch + gmeow:authorityLink.

What's deliberately non-standard (and why)

Terms

This section anchors the guide to the terms declared in this slice. The base gmeow:Language / WritingSystem / version-lineage spine and the proficiency-level individuals live in adjacent slices and are not redeclared here.

gmeow:LanguageVersion · gmeow:LanguageVariety · gmeow:versionLabel

A LanguageVersion is a dated, fully usable revision of a language (a SubKind of Language), bearing its own scripts, names, and status. A LanguageVariety is a speech variety that IS a language — dialect/sociolect/register classifications are standpointed claims, not OWL subclasses. versionLabel carries any versioned artifact's designation as a literal.

gmeow:languageOrigin · gmeow:LanguageOrigin · gmeow:languageModality · gmeow:LanguageModality · gmeow:languageStatus · gmeow:LanguageStatus · gmeow:designGoal

The non-functional typing axis: languageOrigin (natural, constructed, AI-generated, formal, pidgin, creole, reconstructed, …), languageModality (spoken, signed, written, whistled, tactile, machine, multimodal), and languageStatus (living, historical, extinct, dormant, revived, emerging, …) draw from open value vocabularies; designGoal is a literal stating an engineered language's purpose.

gmeow:varietyKind · gmeow:LanguageVarietyKind · gmeow:varietyOf

The variety classification axis: varietyKind (dialect, sociolect, register, slang, standard, creole, pidgin, koine, lingua franca, …) and varietyOf (the parent language) are both non-functional and standpoint-dependent, so a single variety can be a language to one standpoint and a dialect to another.

gmeow:WritingSystemUsage · gmeow:usageLanguage · gmeow:usageWritingSystem · gmeow:scriptRole · gmeow:ScriptRole · gmeow:scriptUsageInterval

The reified co-mingling relator binding {language} × {writing system} × {role} × {period}, mirroring names' NameUsage. usageLanguage and usageWritingSystem are its functional constituents; scriptRole is the functional ScriptRole value (primary, logographic content, syllabic grammar, loanword, transliteration, liturgical, historical, decorative); scriptUsageInterval carries its period (Turkish-in-Arabic until 1928).

gmeow:usesWritingSystem · gmeow:scriptCode · gmeow:writingSystemType · gmeow:WritingSystemType · gmeow:textDirection · gmeow:TextDirection

The direct writing-system properties: usesWritingSystem is the non-functional, co-equal language→script relation; scriptCode is the optional ISO 15924 code; writingSystemType and textDirection draw open value vocabularies for structural kind (alphabet, abjad, abugida, syllabary, logographic, …) and direction (ltr, rtl, vertical, boustrophedon, non-linear, contextual).

gmeow:LanguageProficiency · gmeow:proficiencyAgent · gmeow:proficiencyLanguage · gmeow:proficiencyModality · gmeow:proficiencyLevel · gmeow:proficiencyScale · gmeow:proficiencyInterval · gmeow:levelScale

The reified, per-skill, leveled proficiency relator (the NameUsage idiom): mint one per (agent, language, modality) so "native overall" and "B2 writing" coexist. The functional constituents fix the agent, language, skill modality, level, and measuring scale; proficiencyInterval carries its period; levelScale ties a level individual to its framework.

gmeow:knowsLanguage · gmeow:nativeLanguage

The flat base relations: knowsLanguage states that an agent knows a language; nativeLanguage is its mother-tongue specialisation. Both are non-functional; the leveled per-skill detail rides LanguageProficiency.

gmeow:LanguageState · gmeow:stateLanguage · gmeow:stateStatusValue · gmeow:stateAuthority · gmeow:stateInterval

The standpoint-scoped, analytic/historical status relator (the VersionMembership pattern, also an Observation): it binds {language} × {status} × {authority} × {interval}, so competing reconstructions of a language's vitality over a period coexist. stateLanguage ⊑ observedFeature, stateAuthority ⊑ vantage.

gmeow:LanguageChangeEvent · gmeow:changeType · gmeow:LanguageChangeType · gmeow:affectedLanguage

The diachronic-arc event (Activity subclass): changeType draws the open LanguageChangeType vocabulary (sound shift, semantic drift, borrowing, spelling reform, standardization, extinction, revival, split, merger, …) and affectedLanguage links the one or more languages a change touches.

gmeow:LanguageCreation

The activity of creating or generating a language — a conlanger's design work or an AI model's generation run — linked to its product via wasGeneratedBy and ascribed to its creator (a SoftwareAgent for an AI language) via wasAttributedTo.