Documentation

Linglib.Core.Person

Person #

@cite{cysouw-2009} @cite{siewierska-2004}

Two components of the person API:

§ 1–4: Person Features (@cite{cysouw-2009}, @cite{siewierska-2004}). Framework-neutral decomposition of person into binary features:

These features form a containment hierarchy: [+author] → [+participant]. An author (speaker) is necessarily a participant.

This decomposition is shared across theoretical frameworks:

The Minimalist-specific extension [±proximate] (@cite{pancheva-zubizarreta-2018}) is added in Theories/Syntax/Minimalism/Core/PersonGeometry.lean.

§ 5–9: Person Categories (@cite{cysouw-2009}). The 8 referential person categories from Cysouw's paradigmatic framework. Three singular categories (individual speech act roles) and five group categories (attested combinations of participants).

The full paradigmatic structure machinery (morpheme classes, homophony types, language data) remains in Phenomena/Agreement/PersonMarkingTypology.lean.

Binary person features: [±participant, ±author].

These two features suffice for the three-way person distinction:

  • 1st person: [+participant, +author]
  • 2nd person: [+participant, −author]
  • 3rd person: [−participant, −author]

The fourth combination [−participant, +author] is ill-formed: an author (speaker) is necessarily a speech-act participant.

  • hasParticipant : Bool

    [+participant]: referent includes a speech-act participant (1P or 2P).

  • hasAuthor : Bool

    [+author]: referent includes the speaker (1P only for singulars).

Instances For
    Equations
    • One or more equations did not get rendered due to their size.
    Instances For
      Equations
      Instances For
        Equations
        • One or more equations did not get rendered due to their size.
        Instances For

          Well-formedness: [+author] → [+participant]. An author (speaker) is necessarily a participant.

          Equations
          Instances For

            1st person features: [+participant, +author].

            Equations
            Instances For

              2nd person features: [+participant, −author].

              Equations
              Instances For

                3rd person features: [−participant, −author].

                Equations
                Instances For
                  theorem Core.Person.illFormed_only :
                  { hasParticipant := false, hasAuthor := true }.wellFormed = false

                  The ill-formed combination [−participant, +author] is the only combination that violates well-formedness.

                  theorem Core.Person.exactly_three_wellFormed :
                  (List.filter Features.wellFormed [{ hasParticipant := true, hasAuthor := true }, { hasParticipant := true, hasAuthor := false }, { hasParticipant := false, hasAuthor := true }, { hasParticipant := false, hasAuthor := false }]).length = 3

                  There are exactly 3 well-formed feature combinations (= 3 persons).

                  All person levels yield well-formed features.

                  PersonLevel.isSAP = Features.hasParticipant.

                  The 8 referential person categories (@cite{cysouw-2009}, Fig 10.1).

                  Three singular categories (individual speech act roles) and five group categories (attested combinations of participants).

                  Instances For
                    Equations
                    Instances For

                      All 8 categories in canonical order (singular, then group).

                      Equations
                      • One or more equations did not get rendered due to their size.
                      Instances For

                        Is this an inclusive category (contains both speaker and addressee)?

                        Equations
                        Instances For

                          UD conflates inclusive and exclusive under first person plural.

                          Decompose any Category into binary person features.

                          Features underdetermine group categories: excl, minIncl, and augIncl all map to ⟨true, true⟩. The full Category type is needed for number and inclusivity distinctions.

                          Equations
                          Instances For

                            All 8 categories yield well-formed features.

                            Map singular Category to PersonLevel (the canonical three-way person distinction used by PersonGeometry, DifferentialIndexing, etc.). Group categories map to none — they encode number distinctions that PersonLevel does not capture.

                            Equations
                            Instances For

                              Round-trip: PersonLevel → Category → PersonLevel is identity.

                              includesSpeaker on Category = hasParticipant ∧ hasAuthor on PersonLevel for singular categories: speaker (s1) = [+participant, +author], addressee (s2) = [+participant, −author], other (s3) = [−participant, −author]. This unifies the Category decomposition in Spanish/PersonFeatures.lean with PersonGeometry.decomposePerson.

                              SAP (speech-act participant) = includesSpeaker ∨ includesAddressee for singular categories. This matches PersonLevel.isSAP.

                              Singular categories: Category.toFeatures agrees with PersonLevel.toFeatures via the PersonLevel bridge.