Aikhenvald (2004): Evidentiality typology #
[Aik04] [dH13b] [Bar84] [Osw86]
This file holds only the analytical disagreements: where
[Aik04]'s own paradigm-type letter for a language differs
from AikhenvaldSystem.fromInventory applied to the
Fragment-declared inventory, or where the resulting classification differs
from [dH13b]'s WALS Ch 77 coding.
Per-language Aikhenvald classifications are derived in
Typology/Evidentiality.lean; per-language inventories live in
Fragments/{Lang}/Evidentiality.lean. There are no per-language
verification examples here — those would be unit tests of the derivation
algorithm, not theorems about the typology.
Disagreements recorded #
- §1: 4 paper-vs-derived disagreements (Turkish, Bulgarian, Abkhaz, Kashaya). Each is an analytical choice about which morphemes count as marked evidentials.
- §2: Aikhenvald-vs-WALS disagreements for languages where the two classifications diverge. Three representative cases.
Evidentiality typology (consolidated from the former Typology/Modality + Typology/Evidentiality; single-paper [Aik04] apparatus) #
WALS Ch 77: how many evidential distinctions a language grammaticalises.
Extends the WALS 3-way classification with a threeOrMore value to
distinguish Quechua/Tuyuca-style rich systems from canonical 2-way
systems (Turkish/Bulgarian).
- noGrammatical : EvidentialSystem
No grammatical evidentials (e.g. English, French, Mandarin).
- indirectOnly : EvidentialSystem
Indirect evidential only (e.g. Georgian, West Greenlandic).
- directAndIndirect : EvidentialSystem
Two-choice direct vs indirect (e.g. Turkish, Bulgarian, Tibetan, Abkhaz).
- threeOrMore : EvidentialSystem
Three or more evidential choices (e.g. Quechua, Tuyuca, Aymara, Tariana).
Instances For
Equations
- Aikhenvald2004.instDecidableEqEvidentialSystem x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Equations
- Aikhenvald2004.instBEqEvidentialSystem.beq x✝ y✝ = (x✝.ctorIdx == y✝.ctorIdx)
Instances For
Equations
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
Whether a language has any grammatical evidential marking.
Equations
Instances For
Equations
- One or more equations did not get rendered due to their size.
Whether a language grammaticalizes a direct evidence category.
Equations
Instances For
Equations
- One or more equations did not get rendered due to their size.
Number of evidential choices in the system (0, 1, 2, or 3+).
Equations
Instances For
WALS Ch 78: how evidentiality is morphologically expressed.
- verbalAffix : EvidentialCoding
Evidential is a verbal affix or clitic (bound morpheme). Dominant strategy worldwide (131/418 in WALS Ch 78).
- clitic : EvidentialCoding
Evidential is a clitic (phrasal-level, not strictly verbal).
- particle : EvidentialCoding
Evidential is a free separate particle (65/418).
- partOfTAM : EvidentialCoding
Evidential distinctions fused into the TAM paradigm.
- notApplicable : EvidentialCoding
Not applicable: language has no grammatical evidentials.
Instances For
Equations
- Aikhenvald2004.instDecidableEqEvidentialCoding x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Equations
Equations
- Aikhenvald2004.instBEqEvidentialCoding.beq x✝ y✝ = (x✝.ctorIdx == y✝.ctorIdx)
Instances For
Equations
Equations
- One or more equations did not get rendered due to their size.
Instances For
Whether the coding strategy involves a bound morpheme (affix or clitic).
Equations
Instances For
Equations
- One or more equations did not get rendered due to their size.
WALS Ch 77 → EvidentialSystem. WALS Ch 77's 3-way classification maps
onto our 4-way; threeOrMore cannot be reached from WALS alone (WALS
lumps three-or-more systems with directAndIndirect).
Equations
- Aikhenvald2004.fromWALS77A Data.WALS.F77A.EvidentialityDistinctions.noGrammaticalEvidentials = Aikhenvald2004.EvidentialSystem.noGrammatical
- Aikhenvald2004.fromWALS77A Data.WALS.F77A.EvidentialityDistinctions.indirectOnly = Aikhenvald2004.EvidentialSystem.indirectOnly
- Aikhenvald2004.fromWALS77A Data.WALS.F77A.EvidentialityDistinctions.directAndIndirect = Aikhenvald2004.EvidentialSystem.directAndIndirect
Instances For
WALS Ch 78 → EvidentialCoding. WALS lumps "verbal affix or clitic";
the mixed and modalMorpheme categories are mapped to closest local
cells (best-effort).
Equations
- Aikhenvald2004.fromWALS78A Data.WALS.F78A.EvidentialityCoding.noGrammaticalEvidentials = Aikhenvald2004.EvidentialCoding.notApplicable
- Aikhenvald2004.fromWALS78A Data.WALS.F78A.EvidentialityCoding.verbalAffixOrClitic = Aikhenvald2004.EvidentialCoding.verbalAffix
- Aikhenvald2004.fromWALS78A Data.WALS.F78A.EvidentialityCoding.partOfTheTenseSystem = Aikhenvald2004.EvidentialCoding.partOfTAM
- Aikhenvald2004.fromWALS78A Data.WALS.F78A.EvidentialityCoding.separateParticle = Aikhenvald2004.EvidentialCoding.particle
- Aikhenvald2004.fromWALS78A Data.WALS.F78A.EvidentialityCoding.modalMorpheme = Aikhenvald2004.EvidentialCoding.particle
- Aikhenvald2004.fromWALS78A Data.WALS.F78A.EvidentialityCoding.mixed = Aikhenvald2004.EvidentialCoding.partOfTAM
Instances For
WALS Ch 77 lookup by ISO 639-3 code. none for languages WALS doesn't
cover. Used by Aikhenvald-vs-WALS divergence theorems.
Equations
- One or more equations did not get rendered due to their size.
Instances For
WALS Ch 78 lookup by ISO 639-3 code.
Equations
- Aikhenvald2004.walsEvidentialCoding iso = Option.map (fun (x : Data.WALS.Datapoint Data.WALS.F78A.EvidentialityCoding) => Aikhenvald2004.fromWALS78A x.value) (Data.WALS.F78A.lookupISO iso)
Instances For
[Aik04] Ch 2 paradigm system-type classification. The letter encodes term count (A=2, B=3, C=4, D=5); the digit encodes paradigm shape.
- A1 : AikhenvaldSystem
A1: Firsthand vs Non-firsthand. The most common 2-term type (Cherokee, Jarawara, Yukaghir, Tibetan). Aikhenvald 2004 Ch 2 §2.1.1.
- A2 : AikhenvaldSystem
A2: Non-firsthand vs 'everything else'. The single marked term covers inference and report. Common in Turkic and Caucasian languages (Turkish, Abkhaz, Hunzib). Aikhenvald 2004 Ch 2 §2.1.1.
- A3 : AikhenvaldSystem
A3: Reported (or 'hearsay') vs 'everything else'. The single marked term is the reportative. Widespread (Lezgian, Estonian, Enga, Kham, many North American Indian languages). Aikhenvald 2004 Ch 2 §2.1.1.
- A4 : AikhenvaldSystem
A4: Sensory evidence and Reported. Both terms marked (Ngiyambaa, Diyari, Lakondê/Latundê, reduced Wintu). Aikhenvald 2004 Ch 2 §2.1.1.
- A5 : AikhenvaldSystem
A5: Auditory vs 'everything else'. Attested only in Euchee. Aikhenvald 2004 Ch 2 §2.1.1.
- B1 : AikhenvaldSystem
B1: Direct (or Visual) + Inferred + Reported. The most common 3-term system. The Quechua family (
-mi,-chi,-shi), Wanka Quechua, Shasta, Amdo Tibetan, Qiang, Mosetén. Aikhenvald 2004 Ch 2 §2.2. - B2 : AikhenvaldSystem
B2: Visual + Non-visual sensory + Inferred. No reportative. Rare (Washo, Siona). Aikhenvald 2004 Ch 2 §2.2.
- B3 : AikhenvaldSystem
B3: Visual + Non-visual sensory + Reported. No inferential. Oksapmin, Maricopa, Dulong. Aikhenvald 2004 Ch 2 §2.2.
- B4 : AikhenvaldSystem
B4: Non-visual sensory + Inferred + Reported. Visual is unmarked default (Nganasan, Enets, Retuarã). Aikhenvald 2004 Ch 2 §2.2.
- B5 : AikhenvaldSystem
B5: Reported + Quotative + 'everything else'. Two distinct reported-type evidentials, similar to A3 with quotative split (Comanche, Dakota, Tonkawa). Aikhenvald 2004 Ch 2 §2.2.
- C1 : AikhenvaldSystem
C1: Visual + Non-visual sensory + Inferred + Reported. The most common 4-term system; East Tucanoan languages (Tucano, Barasano, Tatuyo, Siriano, Macuna), Eastern Pomo, traditional Tariana. Aikhenvald 2004 Ch 2 §2.3.
- C2 : AikhenvaldSystem
C2: Direct (or Visual) + Inferred + Assumed + Reported. The sensory may be just visual (Tsafiki, Shipibo-Konibo, Pawnee, Mamainde). Aikhenvald 2004 Ch 2 §2.3.
- C3 : AikhenvaldSystem
C3: Direct + Inferred + Reported + Quotative. Two distinct reportative types (Cora, Northern Embera, SE Tepehuan). Aikhenvald 2004 Ch 2 §2.3.
- D1 : AikhenvaldSystem
D1: Visual + Non-visual sensory + Inferred + Assumed + Reported. The classical 5-term system: current Tariana, Tuyuca, Desano, traditional Wintu. Kashaya is comparable but distinguishes auditory within non-visual. Aikhenvald 2004 Ch 2 §2.4.
Instances For
Equations
- Aikhenvald2004.instDecidableEqAikhenvaldSystem x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Equations
Equations
- Aikhenvald2004.instBEqAikhenvaldSystem.beq x✝ y✝ = (x✝.ctorIdx == y✝.ctorIdx)
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
Equations
Number of evidential terms in the paradigm. The letter prefix encodes the count directly: A=2, B=3, C=4, D=5. Note: A2/A3/A5 have only one marked term but Aikhenvald counts the unmarked 'everything else' form, giving 2 (see Ch 2 §2.1).
Equations
- Aikhenvald2004.AikhenvaldSystem.A1.termCount = 2
- Aikhenvald2004.AikhenvaldSystem.A2.termCount = 2
- Aikhenvald2004.AikhenvaldSystem.A3.termCount = 2
- Aikhenvald2004.AikhenvaldSystem.A4.termCount = 2
- Aikhenvald2004.AikhenvaldSystem.A5.termCount = 2
- Aikhenvald2004.AikhenvaldSystem.B1.termCount = 3
- Aikhenvald2004.AikhenvaldSystem.B2.termCount = 3
- Aikhenvald2004.AikhenvaldSystem.B3.termCount = 3
- Aikhenvald2004.AikhenvaldSystem.B4.termCount = 3
- Aikhenvald2004.AikhenvaldSystem.B5.termCount = 3
- Aikhenvald2004.AikhenvaldSystem.C1.termCount = 4
- Aikhenvald2004.AikhenvaldSystem.C2.termCount = 4
- Aikhenvald2004.AikhenvaldSystem.C3.termCount = 4
- Aikhenvald2004.AikhenvaldSystem.D1.termCount = 5
Instances For
Projection to the WALS Ch 77 ([dH13b]) classification,
extended with threeOrMore. A-systems map per their direct/indirect
coverage; B/C/D map uniformly to .threeOrMore. WALS itself
collapses B/C/D into .directAndIndirect; the .threeOrMore
extension is linglib's, preserving Aikhenvald's richer typology
(see Typology/Modality.lean and Aikhenvald 2004 Ch 2 §2.2).
Equations
- Aikhenvald2004.AikhenvaldSystem.A1.toWALS = Aikhenvald2004.EvidentialSystem.directAndIndirect
- Aikhenvald2004.AikhenvaldSystem.A2.toWALS = Aikhenvald2004.EvidentialSystem.indirectOnly
- Aikhenvald2004.AikhenvaldSystem.A3.toWALS = Aikhenvald2004.EvidentialSystem.indirectOnly
- Aikhenvald2004.AikhenvaldSystem.A4.toWALS = Aikhenvald2004.EvidentialSystem.directAndIndirect
- Aikhenvald2004.AikhenvaldSystem.A5.toWALS = Aikhenvald2004.EvidentialSystem.directAndIndirect
- Aikhenvald2004.AikhenvaldSystem.B1.toWALS = Aikhenvald2004.EvidentialSystem.threeOrMore
- Aikhenvald2004.AikhenvaldSystem.B2.toWALS = Aikhenvald2004.EvidentialSystem.threeOrMore
- Aikhenvald2004.AikhenvaldSystem.B3.toWALS = Aikhenvald2004.EvidentialSystem.threeOrMore
- Aikhenvald2004.AikhenvaldSystem.B4.toWALS = Aikhenvald2004.EvidentialSystem.threeOrMore
- Aikhenvald2004.AikhenvaldSystem.B5.toWALS = Aikhenvald2004.EvidentialSystem.threeOrMore
- Aikhenvald2004.AikhenvaldSystem.C1.toWALS = Aikhenvald2004.EvidentialSystem.threeOrMore
- Aikhenvald2004.AikhenvaldSystem.C2.toWALS = Aikhenvald2004.EvidentialSystem.threeOrMore
- Aikhenvald2004.AikhenvaldSystem.C3.toWALS = Aikhenvald2004.EvidentialSystem.threeOrMore
- Aikhenvald2004.AikhenvaldSystem.D1.toWALS = Aikhenvald2004.EvidentialSystem.threeOrMore
Instances For
The Aikhenvald typology is always richer-or-equal to the WALS one:
.threeOrMore for all 3+-term systems (B/C/D), the appropriate
2-way value for 2-term systems (A).
Derivation from a typed inventory #
Derive [Aik04]'s paradigm system-type letter from a
declared inventory by inspecting which cells of her paradigm space are
covered. Mirrors Determiner.markingStrategy's derivation pattern.
Returns none for inventories whose cell pattern doesn't fit any
standard A–D letter; per-paper editorial overrides for such cases
live in the relevant Studies/AuthorYear.lean.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Worked examples documenting the API #
§1. Paper-vs-derived disagreements #
[Aik04]'s own letter assignment for the four sample languages
where it differs from AikhenvaldSystem.fromInventory:
- Turkish, Bulgarian, Abkhaz: Aikhenvald treats one term as unmarked default (A2 = single marked non-firsthand); Fragment counts both terms (A1).
- Kashaya: Aikhenvald assigns D1 as closest fit; derivation returns
nonebecause the auditory cell plus missing nonvisualSensory and assumed cells break D1's structural pattern.
Aikhenvald 2004's own letter for the four divergent languages. none
for any language whose derivation matches the paper.
Equations
- Aikhenvald2004.paperType "tur" = some Aikhenvald2004.AikhenvaldSystem.A2
- Aikhenvald2004.paperType "bul" = some Aikhenvald2004.AikhenvaldSystem.A2
- Aikhenvald2004.paperType "abk" = some Aikhenvald2004.AikhenvaldSystem.A2
- Aikhenvald2004.paperType "kju" = some Aikhenvald2004.AikhenvaldSystem.D1
- Aikhenvald2004.paperType x✝ = none
Instances For
§2. Aikhenvald-vs-WALS disagreements #
Representative cases — same structural pattern repeats across the other indirectOnly-via-modal cases (German, Korean, Finnish) and the other WALS-lumps-rich-systems cases (Tariana, Kashaya).
Empty inventory per Aikhenvald-strict vs WALS indirectOnly. French
is the canonical case: the journalistic conditional has reportative
use that WALS counts but Aikhenvald excludes. Same pattern holds for
German, Japanese, Korean, Finnish.
The signature divergence: Tuyuca's D1 5-term system per Aikhenvald
(projecting to threeOrMore) vs WALS's directAndIndirect lump. Same
paradigm, two classifications, both kernel-verified.
Andean B1: Aikhenvald has classification; WALS has no entry.
Turkish: derivation's A1 happens to project to the same WALS coding
(directAndIndirect) WALS assigns directly; the Aikhenvald paper
A2 projects to indirectOnly. Triple-comparison divergence.