Phenomena.ArgumentStructure.Studies.Polinsky2013

@cite{polinsky-2013-antipassive} @cite{polinsky-2013-applicative} @cite{maslova-nedjalkov-2013} @cite{siewierska-2013} @cite{haspelmath-2013-ditransitive} @cite{song-2013-periphrastic} @cite{song-2013-nonperiphrastic} @cite{silverstein-1976}

Cross-linguistic analyses anchored on Polinsky's two WALS chapters -- Ch 108 (antipassive) and Ch 109 (applicative) -- and the cross-chapter correlations they support. The 19-language ValenceProfile sample is the testbed for these correlations, drawing on data from all of Ch 105--111.

Polinsky's central empirical claim (Ch 108) is the antipassive-ergativity correlation @cite{silverstein-1976}: antipassives are concentrated in ergative languages, but accusative-language antipassives exist (17 of them in WALS Table 1, against 30 ergative). This file ports Table 1 and tests the correlation against the ValenceProfile sample.

For Ch 109 (applicative), Polinsky observes that applicatives "are commonly found in those languages that have little or no case marking of noun phrases and that have sufficiently rich verbal morphology" -- correlating with morphological causatives (Song 2013, Ch 111). The applicative_implies_morph_causative theorem tests this in the sample.

Contents

• §1. The 19-language ValenceProfile sample.
• §2. Sample-grounded cross-chapter generalisations: passive prevalence, antipassive-ergativity, applicative-causative correlation, areal distribution, no_passive_ergative_has_antipassive.
• §3. The 47-language antipassive alignment data (@cite{polinsky-2013-antipassive} Table 1).
• §4. Causative morphology examples (Song 2013).
• §5. Fragment-bridge theorems (English reciprocals, Finnish passive).

Out of scope

The substrate types (ReciprocalType, ApplicativeType, ValenceProfile, WALS converters, corpus-only theorems) live in Linglib/Typology/ArgumentStructure.lean. Pylkkänen's structural Appl classification + WALS divergence is in Studies/Pylkkanen2008.lean. Nordlinger's extended reciprocal apparatus is in Studies/Nordlinger2023.lean. Siloni's lexical/syntactic distinction is in Studies/Siloni2012.lean.

def Phenomena.ArgumentStructure.Studies.Polinsky2013.english : Typology.ArgumentStructure.ValenceProfile

English: reciprocal distinct from reflexive ("each other" vs "themselves"), periphrastic passive ("was kicked"), no antipassive (accusative alignment), no applicative, no productive morphological causative (lexical causatives like "kill" only).

Equations

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

def Phenomena.ArgumentStructure.Studies.Polinsky2013.japanese : Typology.ArgumentStructure.ValenceProfile

Japanese: reciprocal distinct from reflexive ("otagai" vs "jibun"), passive ("-(r)are-"), no antipassive, no applicative, morphological causative suffix "-(s)ase".

Equations

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

def Phenomena.ArgumentStructure.Studies.Polinsky2013.turkish : Typology.ArgumentStructure.ValenceProfile

Turkish: reciprocal distinct from reflexive -- "birbirine" (reciprocal) is formally distinct from "kendi" (reflexive). WALS Ch 106 codes Turkish as Value 2 (distinct from reflexive). Passive ("-Il"), no antipassive (accusative), no applicative, morphological causative "-dUr".

Equations

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

def Phenomena.ArgumentStructure.Studies.Polinsky2013.swahili : Typology.ArgumentStructure.ValenceProfile

Swahili: reciprocal distinct ("-ana"), passive ("-w-"), no antipassive, applicative ("-i-" / "-e-" benefactive + locative from both bases), morphological causative ("-ish-" / "-esh-").

iso = "swh" (Standard Swahili individual code) matches WALS; "swa" is the macrolanguage code, which WALS does not use.

Equations

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

def Phenomena.ArgumentStructure.Studies.Polinsky2013.dyirbal : Typology.ArgumentStructure.ValenceProfile

Dyirbal (Pama-Nyungan, Australia): reciprocal distinct from reflexive. Reflexive: stem + -riy (@cite{dixon-1972} §4.8.1). Reciprocal: reduplicated stem + -(n)bariy (@cite{dixon-1972} §4.8.2). The reciprocal "functions like an intransitive stem" (monovalent). No passive, dedicated antipassive with oblique patient, ergative alignment, no applicative, morphological causative. NOTE: Dyirbal is NOT in WALS Ch 106 sample (175 languages).

Equations

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

def Phenomena.ArgumentStructure.Studies.Polinsky2013.chukchi : Typology.ArgumentStructure.ValenceProfile

Chukchi (Chukotko-Kamchatkan, Russia): reciprocal distinct, no passive (inverse system instead), dedicated antipassive "ine-" with oblique patient (instrumental), ergative alignment, no applicative, morphological causative.

Equations

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

def Phenomena.ArgumentStructure.Studies.Polinsky2013.indonesian : Typology.ArgumentStructure.ValenceProfile

Indonesian: reciprocal distinct from reflexive -- "saling" (reciprocal) is formally different from "diri sendiri" (reflexive). WALS Ch 106 codes Indonesian as Value 2 (distinct from reflexive). Passive ("di-"), no antipassive, no applicative, morphological causative.

Equations

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

def Phenomena.ArgumentStructure.Studies.Polinsky2013.french : Typology.ArgumentStructure.ValenceProfile

French: mixed reciprocal type -- both reflexive-identical "se" and distinct "l'un l'autre" ('each other'). WALS Ch 106 codes French as Value 3 (mixed). Periphrastic passive ("être + past participle"), no antipassive, no applicative.

Nonperiphrastic causative (Ch 111): WALS codes French as .both. The periphrastic faire + INF construction is Ch 110 (periphrastic causatives), excluded from Ch 111 by definition.

Equations

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

def Phenomena.ArgumentStructure.Studies.Polinsky2013.russian : Typology.ArgumentStructure.ValenceProfile

Russian: mixed reciprocal type -- both reflexive-identical "-sja"/"-s'" and distinct "drug druga" ('each other'). WALS Ch 106 codes Russian as Value 3 (mixed), parallel to German ("sich" + "einander"). Passive (synthetic "-sja" + periphrastic), no antipassive, no applicative, morphological causative (zero-derivation, e.g. "lomat'-sja" anticausative).

Equations

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

def Phenomena.ArgumentStructure.Studies.Polinsky2013.arabic : Typology.ArgumentStructure.ValenceProfile

Arabic (Modern Standard): reciprocal via Form VI (tafaaʿal-a), formally distinct from Form V reflexive (tafaʿʿal-a): Form VI inserts long -aa- after the first root consonant, Form V doubles the medial consonant (@cite{ryding-2005} Ch 27 §1). Example: taʿaanaq-a 'to embrace one another' (Form VI, reciprocal) vs. takallam-a 'to speak' (Form V, reflexive/mediopassive). Passive via internal vowel change (kutiba), no antipassive, no applicative, morphological causative (Form IV ʾafʿala). NOTE: MSA is NOT in WALS Ch 106 sample; Arabic (Egyptian) is listed as Value 2 (distinct from reflexive).

Equations

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

def Phenomena.ArgumentStructure.Studies.Polinsky2013.hindi : Typology.ArgumentStructure.ValenceProfile

Hindi: reciprocal distinct ("ek duusre"), passive ("-aa" / "jaanaa" periphrastic), no antipassive, no applicative, morphological causative ("-aa" / "-vaa").

Equations

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

def Phenomena.ArgumentStructure.Studies.Polinsky2013.westGreenlandic : Typology.ArgumentStructure.ValenceProfile

West Greenlandic (Eskimo-Aleut): reciprocal-reflexive polysemy ("immi-" for both), passive, antipassive with oblique patient, ergative alignment, no applicative, morphological causative.

Equations

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

def Phenomena.ArgumentStructure.Studies.Polinsky2013.kinyarwanda : Typology.ArgumentStructure.ValenceProfile

Kinyarwanda (Bantu, Rwanda): reciprocal suffix -an- (e.g., -saban- 'ask each other'), formally distinct from reflexive prefix -ii-/-iiy- (e.g., á-r-íi-reeb-a 'she watches herself') (@cite{kimenyi-1980} §4.2.1, p. 5). Passive -w-, no antipassive, applicative -ir-/-er- (benefactive + other from both bases), morphological causative. NOTE: Kinyarwanda is NOT in WALS Ch 106 sample.

Equations

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

def Phenomena.ArgumentStructure.Studies.Polinsky2013.lango : Typology.ArgumentStructure.ValenceProfile

Lango (Nilotic, Uganda): reciprocal identical to reflexive. WALS Ch 106 codes Lango as Value 4 (identical to reflexive). Passive absent, antipassive with implicit patient (per WALS Ch 108 coding -- accusative alignment, one of the accusative-language antipassives), no applicative, morphological causative.

Equations

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

def Phenomena.ArgumentStructure.Studies.Polinsky2013.chamorro : Typology.ArgumentStructure.ValenceProfile

Chamorro (Austronesian, Guam): reciprocal distinct, passive present, antipassive with oblique patient (accusative alignment), applicative (benefactive + other, both bases), morphological causative.

Equations

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

def Phenomena.ArgumentStructure.Studies.Polinsky2013.halkomelem : Typology.ArgumentStructure.ValenceProfile

Halkomelem (Salishan, Canada): reciprocal suffix -təl (e.g., ʔiyá·təl 'fight', yéyətəl 'make friends'), formally distinct from reflexive suffixes -lá·mət 'oneself' and -(ə)θət 'oneself, itself' (@cite{galloway-1993} §6.1.3, §11.2.1.14--15). Passive present, antipassive with oblique patient, ergative alignment, applicative (benefactive + other, both bases), morphological causative. NOTE: Halkomelem is NOT in WALS Ch 106 sample.

Equations

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

def Phenomena.ArgumentStructure.Studies.Polinsky2013.modernGreek : Typology.ArgumentStructure.ValenceProfile

Modern Greek: mixed reciprocal type -- both nonactive voice morphology (identical to reflexive) and distinct reciprocal constructions. WALS Ch 106 codes Modern Greek as Value 3 (mixed). Passive present ("periphrastic with nonactive morphology"), no antipassive, no applicative, NEITHER morphological nor compound causative (relies on periphrastic causative).

Equations

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

def Phenomena.ArgumentStructure.Studies.Polinsky2013.german : Typology.ArgumentStructure.ValenceProfile

German: mixed reciprocal type ("sich" reflexive/reciprocal + "einander" distinct reciprocal), passive (werden + participle), no antipassive, no applicative, morphological causative (zero-derivation).

Equations

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

def Phenomena.ArgumentStructure.Studies.Polinsky2013.finnish : Typology.ArgumentStructure.ValenceProfile

Finnish: reciprocal distinct from reflexive ("toisiaan" ≠ "itsensä"), impersonal "passive" (present -- the Fragment's finnishPassive has semantic content but does not project a syntactic agent), no antipassive, accusative alignment, no applicative, morphological causative -tta- / -ttä-.

Equations

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

def Phenomena.ArgumentStructure.Studies.Polinsky2013.allProfiles : List Typology.ArgumentStructure.ValenceProfile

19-language sample for cross-chapter generalisations.

Equations

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

theorem Phenomena.ArgumentStructure.Studies.Polinsky2013.majority_of_sample_has_passive : (List.filter (fun (x : Typology.ArgumentStructure.ValenceProfile) => x.hasPassive) allProfiles).length ≥ 14

In our sample, most languages have a passive: at least 14 of 19.

theorem Phenomena.ArgumentStructure.Studies.Polinsky2013.ergative_implies_antipassive_in_sample : have ergatives := List.filter (fun (p : Typology.ArgumentStructure.ValenceProfile) => p.alignment == Core.AlignmentFamily.ergative) allProfiles; (ergatives.all fun (p : Typology.ArgumentStructure.ValenceProfile) => p.antipassive.hasAntipassive) = true

@cite{polinsky-2013-antipassive}'s key generalisation in the sample: every ergative-alignment language has an antipassive.

theorem Phenomena.ArgumentStructure.Studies.Polinsky2013.most_accusative_lack_antipassive : have accusatives := List.filter (fun (p : Typology.ArgumentStructure.ValenceProfile) => p.alignment == Core.AlignmentFamily.accusative) allProfiles; have withoutAP := List.filter (fun (p : Typology.ArgumentStructure.ValenceProfile) => !p.antipassive.hasAntipassive) accusatives; withoutAP.length > accusatives.length - withoutAP.length

@cite{polinsky-2013-antipassive}'s flip side: most accusative languages in the sample lack an antipassive (the converse fails: not all accusative languages lack one -- Lango is a counterexample).

theorem Phenomena.ArgumentStructure.Studies.Polinsky2013.almost_all_have_nonperiphr_causative : (List.filter (fun (p : Typology.ArgumentStructure.ValenceProfile) => p.causative != Typology.ArgumentStructure.NonperiphrCausativeType.neither) allProfiles).length ≥ 16

@cite{song-2013-nonperiphrastic} dominance reproduced in the sample: almost every language has at least a morphological or compound causative (Modern Greek is the lone exception -- only periphrastic).

theorem Phenomena.ArgumentStructure.Studies.Polinsky2013.applicative_implies_morph_causative : have withApplicative := List.filter (fun (p : Typology.ArgumentStructure.ValenceProfile) => p.applicative.hasApplicative) allProfiles; (withApplicative.all fun (p : Typology.ArgumentStructure.ValenceProfile) => p.causative.hasMorphological) = true

@cite{polinsky-2013-applicative}'s observation: applicatives "are commonly found in those languages that have... sufficiently rich verbal morphology." Since morphological causatives also require rich verbal morphology, we expect a correlation. In the sample: every applicative language also has a morphological causative.

def Phenomena.ArgumentStructure.Studies.Polinsky2013.isEurasian (p : Typology.ArgumentStructure.ValenceProfile) : Bool

WALS Ch 106 areal pattern: Value 1 (no non-iconic reciprocal constructions) is concentrated in Oceania, Southeast Asia, and parts of Africa. Every Eurasian language in the sample has at least one productive grammatical reciprocal construction.

Equations

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

theorem Phenomena.ArgumentStructure.Studies.Polinsky2013.eurasian_all_have_productive_reciprocal : ((List.filter isEurasian allProfiles).all fun (p : Typology.ArgumentStructure.ValenceProfile) => p.reciprocal != Typology.ArgumentStructure.ReciprocalType.noDedicated) = true

theorem Phenomena.ArgumentStructure.Studies.Polinsky2013.no_passive_ergative_has_antipassive : have noPassiveErgative := List.filter (fun (p : Typology.ArgumentStructure.ValenceProfile) => p.passive == Typology.ArgumentStructure.PassivePresence.absent && p.alignment == Core.AlignmentFamily.ergative) allProfiles; (noPassiveErgative.all fun (p : Typology.ArgumentStructure.ValenceProfile) => p.antipassive.hasAntipassive) = true

Cross-chapter: in the sample, every passive-absent ergative language has an antipassive (consistent with ergative languages using antipassive in roles where accusative languages use passive).

theorem Phenomena.ArgumentStructure.Studies.Polinsky2013.some_languages_have_both_app_and_antipass : (List.filter (fun (p : Typology.ArgumentStructure.ValenceProfile) => p.applicative.hasApplicative && p.antipassive.hasAntipassive) allProfiles).length = 2

Applicatives and antipassives are dual voice operations: applicatives increase valence, antipassives decrease it. In the sample, two languages combine them (Chamorro and Halkomelem).

theorem Phenomena.ArgumentStructure.Studies.Polinsky2013.applicatives_only_in_bantu_austronesian_salishan : ((List.filter (fun (p : Typology.ArgumentStructure.ValenceProfile) => p.applicative.hasApplicative) allProfiles).all fun (p : Typology.ArgumentStructure.ValenceProfile) => p.iso == "swh" || p.iso == "kin" || p.iso == "cha" || p.iso == "hur") = true

@cite{polinsky-2013-applicative} areal observation: applicatives cluster in three areas -- Bantu Africa, Western Pacific (Austronesian), and North/Meso-America (Salish, Mayan, Uto-Aztecan). The dearth in Eurasia correlates with rich case marking. In the sample, applicatives appear only in Bantu, Austronesian, and Salishan languages.

structure Phenomena.ArgumentStructure.Studies.Polinsky2013.AntipassiveAlignmentDatum : Type

Languages with antipassives classified by alignment, from @cite{polinsky-2013-antipassive} Table 1. Key empirical evidence for the antipassive-ergativity debate (@cite{silverstein-1976}).

• language : String
• alignment : Typology.ArgumentStructure.AlignmentType

def Phenomena.ArgumentStructure.Studies.Polinsky2013.instReprAntipassiveAlignmentDatum.repr : AntipassiveAlignmentDatum → ℕ → Std.Format

Equations

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

@[implicit_reducible]

instance Phenomena.ArgumentStructure.Studies.Polinsky2013.instReprAntipassiveAlignmentDatum : Repr AntipassiveAlignmentDatum

Equations

• Phenomena.ArgumentStructure.Studies.Polinsky2013.instReprAntipassiveAlignmentDatum = { reprPrec := Phenomena.ArgumentStructure.Studies.Polinsky2013.instReprAntipassiveAlignmentDatum.repr }

def Phenomena.ArgumentStructure.Studies.Polinsky2013.instDecidableEqAntipassiveAlignmentDatum.decEq (x✝ x✝¹ : AntipassiveAlignmentDatum) : Decidable (x✝ = x✝¹)

Equations

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

@[implicit_reducible]

instance Phenomena.ArgumentStructure.Studies.Polinsky2013.instDecidableEqAntipassiveAlignmentDatum : DecidableEq AntipassiveAlignmentDatum

Equations

• Phenomena.ArgumentStructure.Studies.Polinsky2013.instDecidableEqAntipassiveAlignmentDatum = Phenomena.ArgumentStructure.Studies.Polinsky2013.instDecidableEqAntipassiveAlignmentDatum.decEq

def Phenomena.ArgumentStructure.Studies.Polinsky2013.accusativeAntipassiveLangs : List AntipassiveAlignmentDatum

Accusative languages with antipassives (17 in WALS Table 1). These are the key counterexamples to the strong claim that antipassives are limited to ergative languages.

Equations

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

def Phenomena.ArgumentStructure.Studies.Polinsky2013.ergativeAntipassiveLangs : List AntipassiveAlignmentDatum

Ergative languages with antipassives (30 in WALS Table 1).

Equations

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

theorem Phenomena.ArgumentStructure.Studies.Polinsky2013.AP_accusative_count : accusativeAntipassiveLangs.length = 17

Sample-size and consistency for the Polinsky 2013 Table 1 data.

theorem Phenomena.ArgumentStructure.Studies.Polinsky2013.AP_ergative_count : ergativeAntipassiveLangs.length = 30

theorem Phenomena.ArgumentStructure.Studies.Polinsky2013.accusative_AP_all_accusative : (accusativeAntipassiveLangs.all fun (d : AntipassiveAlignmentDatum) => d.alignment == Core.AlignmentFamily.accusative) = true

theorem Phenomena.ArgumentStructure.Studies.Polinsky2013.ergative_AP_all_ergative : (ergativeAntipassiveLangs.all fun (d : AntipassiveAlignmentDatum) => d.alignment == Core.AlignmentFamily.ergative) = true

theorem Phenomena.ArgumentStructure.Studies.Polinsky2013.ergative_AP_more_common : ergativeAntipassiveLangs.length > accusativeAntipassiveLangs.length

@cite{polinsky-2013-antipassive}'s headline ratio: ergative languages with antipassives outnumber accusative ones with antipassives.

structure Phenomena.ArgumentStructure.Studies.Polinsky2013.CausativeMorphologyExample : Type

Examples of morphological causatives by morpheme position. @cite{song-2013-nonperiphrastic} reports that suffixation is by far the most common pattern, paralleling Greenberg's Universal 27.

• language : String
• morpheme : String
• position : String

@[implicit_reducible]

instance Phenomena.ArgumentStructure.Studies.Polinsky2013.instReprCausativeMorphologyExample : Repr CausativeMorphologyExample

Equations

• Phenomena.ArgumentStructure.Studies.Polinsky2013.instReprCausativeMorphologyExample = { reprPrec := Phenomena.ArgumentStructure.Studies.Polinsky2013.instReprCausativeMorphologyExample.repr }

def Phenomena.ArgumentStructure.Studies.Polinsky2013.instReprCausativeMorphologyExample.repr : CausativeMorphologyExample → ℕ → Std.Format

Equations

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

def Phenomena.ArgumentStructure.Studies.Polinsky2013.instDecidableEqCausativeMorphologyExample.decEq (x✝ x✝¹ : CausativeMorphologyExample) : Decidable (x✝ = x✝¹)

Equations

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

@[implicit_reducible]

instance Phenomena.ArgumentStructure.Studies.Polinsky2013.instDecidableEqCausativeMorphologyExample : DecidableEq CausativeMorphologyExample

Equations

• Phenomena.ArgumentStructure.Studies.Polinsky2013.instDecidableEqCausativeMorphologyExample = Phenomena.ArgumentStructure.Studies.Polinsky2013.instDecidableEqCausativeMorphologyExample.decEq

def Phenomena.ArgumentStructure.Studies.Polinsky2013.causativeMorphExamples : List CausativeMorphologyExample

Equations

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

theorem Phenomena.ArgumentStructure.Studies.Polinsky2013.suffixing_dominates_causatives : have suffixes := List.filter (fun (e : CausativeMorphologyExample) => e.position == "suffix") causativeMorphExamples; have nonSuffixes := List.filter (fun (e : CausativeMorphologyExample) => e.position != "suffix") causativeMorphExamples; suffixes.length > nonSuffixes.length * 2

Suffixing dominates: most morphological causative examples are suffixes (@cite{song-2013-nonperiphrastic}).

theorem Phenomena.ArgumentStructure.Studies.Polinsky2013.english_reciprocal_distinct_from_reflexive : Fragments.English.Pronouns.eachOther.pronounType ≠ Fragments.English.Pronouns.PronounType.reflexive ∧ Fragments.English.Pronouns.oneAnother.pronounType ≠ Fragments.English.Pronouns.PronounType.reflexive

English reciprocal forms ("each other", "one another") are formally distinct from reflexive forms ("themselves", etc.) -- derived from the Fragment's pronoun entries rather than stipulated in the profile.

theorem Phenomena.ArgumentStructure.Studies.Polinsky2013.finnish_passive_has_semantics : Fragments.Finnish.Predicates.finnishPassive.HasSemantics

Finnish impersonal "passive" has semantic content (existential closure over agent) -- derived from the Fragment's voice head.

theorem Phenomena.ArgumentStructure.Studies.Polinsky2013.finnish_passive_no_agent : ¬Fragments.Finnish.Predicates.finnishPassive.AssignsTheta

Finnish impersonal "passive" does NOT project a syntactic agent -- derived from the Fragment's voice head. The Fragment-grounded discrepancy with WALS Ch 107 (which classifies Finnish as .present) is real: the Finnish "passive" has agent semantics without projecting syntax.