According to Linked Open Vocabularies, LexInfo defines 592 entities, including:

271 Individual values (e.g., verb, thirdPerson, vulgarRegister)

182 Classes (e.g., Verb, VerbPOS, VerbPhrase, Tense)

135 Object Properties (e.g., synonym, translation, tense, voice)

4 Datatype Properties (e.g., pronunciation, romanization)

2 Annotation Properties (e.g., languageSpecific, example)
GitHub

This breadth allows for rich annotation of lexical entries, covering both syntactic and semantic details.

A logical representation to capture the data as listed in the turtle code.

An optional link from the Object/Fact-types in CaseTalk can be simply added. This indicates that any IRI in the world of semantic web, can be mentioned as a simple custom attribute in the FCO-IM models.

Since this is already a supported possibility, the only additional file needed is the SKOS definition file, or an external SKOS file, but linking becomes very straight forward.

SKOS export is now implemented. CaseTalk can export conceptual models as SKOS Turtle, including concept schemes, preferred labels, broader/narrower/related relations, and definitions. OntoLex-Lemon and LexInfo support still require further investigation.

How the three vocabularies layer on top of each other:

SKOS - concept organization layer
Concepts, preferred/alternate labels, definitions, hierarchies (broader/narrower) and associative links (related), concept schemes (taxonomies, thesauri). SKOS knows nothing about words as words - just that a concept has a label string.

OntoLex-Lemon - lexical layer
Models a LexicalEntry (a word or phrase) separately from the concept it expresses. Connects lexical entries to SKOS concepts via LexicalSense. Supports inflected forms (canonical form, past tense, plural, etc.) and multilingual translation at the sense level. Enables: 'the German word laufen and the English word run both refer to the same concept, but are different lexical entries with different morphological forms'.

For CaseTalk this is directly relevant: OFType names and verbalizations are already lexical entries in disguise. OntoLex-Lemon would let you export verbalizations as proper linguistic data, not just label strings.

LexInfo - linguistic metadata layer
Companion to OntoLex-Lemon, adds a controlled vocabulary for linguistic properties: part of speech (noun, verb, adjective), morphological features (tense, mood, number, gender, case), syntactic frames and argument structure.

Summary:

• SKOS: concept hierarchy + relations -> OFType taxonomy, broader/narrower
• OntoLex-Lemon: words linked to concepts, multilingual forms -> verbalizations as lexical entries
• LexInfo: POS + morphology on those words -> role names (noun/verb), verb forms in expressions

OntoLex-Lemon is the most relevant next step: CaseTalk's verbalization engine already produces natural language from fact types, so exporting that as OntoLex-Lemon data would give linguists and NLP tools something they can actually use. LexInfo is mostly annotation detail on top of that.

Related tickets created for the remaining investigation work: 0005519 (OntoLex-Lemon) and 0005520 (LexInfo).

SKOS Export module is build.