@cite{wood-2015} — Icelandic Morphosyntax and Argument Structure #
@cite{wood-2015} @cite{kratzer-1996} @cite{pylkkanen-2008} @cite{schaefer-2008} @cite{cuervo-2003} @cite{alexiadou-schaefer-2015} @cite{wood-marantz-2017}
@cite{wood-2015} establishes that Icelandic -st (from historical reflexive sik) is the morphological reflex of non-agentive Voice across multiple syntactic configurations, not a single "reflexive" or "anticausative" morpheme.
Layer position #
This file is the canonical Wood-2015 study site. The Icelandic
Fragment (Fragments.Icelandic.Predicates) carries only consensus
lexical data — verb forms, glosses, presence/absence of an active
variant. Wood's analytical apparatus lives here:
- the six-way -st classification (
StType), - the anticausative-marking enum -st / -na / -ka / unmarked
(
AnticausativeMarking), - the mapping from -st types to Voice flavors,
- per-verb projection bundles (
StVerbInfo) carrying stType, marking, and root structure for each verb.
The ±D notation is sourced from @cite{alexiadou-schaefer-2015} (External Arguments in Transitivity Alternations, OUP) and refined in @cite{wood-marantz-2017}. Earlier versions of this fragment attributed ±θ/±D to @cite{schaefer-2008} alone, which was imprecise.
Key Claims Formalized #
Voice–CAUSE independence (Ch. 3): The causal relation is shared across causative and anticausative alternants; Voice contributes only whether an external argument is introduced. Modeled here using @cite{cuervo-2003}'s VerbHead decomposition (@cite{wood-2015} himself uses a single v head; the multi-head decomposition is a linglib modeling convenience that captures the same independence structurally — see
Theories/Syntax/Minimalist/Voice.lean§5).-st as elsewhere exponent of Voice (Ch. 3, §3.3–3.5): -st spells out non-agentive Voice across anticausative, middle, reflexive, experiencer, inherent, and reciprocal configurations. The morphological uniformity masks syntactic diversity.
Voice projection locus (Ch. 3): @cite{wood-2015}'s Voice_{D}/Voice_{∅} distinction is captured structurally by
VoiceProjectionLocus(inVoice.lean), withAnticausativeMarking.voiceLocusprojecting from each morphological exponent (-st, -na, -ka) to its locus.-st and -na complementarity (Ch. 3, §3.3): -st and -na spell out different Voice projections (Voice_{D} vs Voice_{∅}) and never co-occur. -ka is an exponent of v, not Voice.
Applicative interaction (Ch. 5): @cite{wood-2015} shows -st cannot merge in SpecApplP because Appl assigns dative case and -st lacks case features. The high/low Appl interaction theorems below follow @cite{pylkkanen-2008} and @cite{schaefer-2008}, not @cite{wood-2015}'s Icelandic-specific analysis (which argues Icelandic lacks true high applicatives).
Equations
- Wood2015.instDecidableEqStType x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Equations
- Wood2015.instReprStType = { reprPrec := Wood2015.instReprStType.repr }
Equations
- One or more equations did not get rendered due to their size.
- Wood2015.instReprStType.repr Wood2015.StType.middle prec✝ = Repr.addAppParen (Std.Format.nest (if prec✝ ≥ 1024 then 1 else 2) (Std.Format.text "Wood2015.StType.middle")).group prec✝
- Wood2015.instReprStType.repr Wood2015.StType.reflexive prec✝ = Repr.addAppParen (Std.Format.nest (if prec✝ ≥ 1024 then 1 else 2) (Std.Format.text "Wood2015.StType.reflexive")).group prec✝
- Wood2015.instReprStType.repr Wood2015.StType.inherent prec✝ = Repr.addAppParen (Std.Format.nest (if prec✝ ≥ 1024 then 1 else 2) (Std.Format.text "Wood2015.StType.inherent")).group prec✝
- Wood2015.instReprStType.repr Wood2015.StType.subjectExp prec✝ = Repr.addAppParen (Std.Format.nest (if prec✝ ≥ 1024 then 1 else 2) (Std.Format.text "Wood2015.StType.subjectExp")).group prec✝
- Wood2015.instReprStType.repr Wood2015.StType.reciprocal prec✝ = Repr.addAppParen (Std.Format.nest (if prec✝ ≥ 1024 then 1 else 2) (Std.Format.text "Wood2015.StType.reciprocal")).group prec✝
Instances For
Map each -st type to its Voice flavor.
Equations
- Wood2015.StType.anticausative.voiceFlavor = Minimalist.VoiceFlavor.nonThematic
- Wood2015.StType.middle.voiceFlavor = Minimalist.VoiceFlavor.expletive
- Wood2015.StType.reflexive.voiceFlavor = Minimalist.VoiceFlavor.reflexive
- Wood2015.StType.inherent.voiceFlavor = Minimalist.VoiceFlavor.nonThematic
- Wood2015.StType.subjectExp.voiceFlavor = Minimalist.VoiceFlavor.experiencer
- Wood2015.StType.reciprocal.voiceFlavor = Minimalist.VoiceFlavor.reflexive
Instances For
How an anticausative is morphologically marked (@cite{wood-2015} Ch. 3, §3.3).
- -st spells out Voice_{D};
- -na spells out specifierless Voice_{∅};
- -ka spells out v (not Voice);
- unmarked: zero alternation, compatible with either Voice_{D} or Voice_{∅}, distinguished by independent diagnostics.
Per @cite{wood-2015} Ch. 3, §3.3, -st and -na never co-occur because they spell out different Voice types; -na and -ka never co-occur because -na requires v to be null/pruned.
- st : AnticausativeMarking
- na : AnticausativeMarking
- unmarked : AnticausativeMarking
- ka : AnticausativeMarking
Instances For
Equations
- Wood2015.instDecidableEqAnticausativeMarking x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Equations
Equations
- One or more equations did not get rendered due to their size.
Instances For
Each anticausative-marking morpheme picks out a position in
Voice/v space (@cite{wood-2015} Ch. 3, §3.3). The structural
distinction is visible to pattern matching, not hidden behind
string equality on labels like "Voice_{D}".
Equations
- Wood2015.AnticausativeMarking.st.voiceLocus = Minimalist.VoiceProjectionLocus.voiceD
- Wood2015.AnticausativeMarking.na.voiceLocus = Minimalist.VoiceProjectionLocus.voiceBare
- Wood2015.AnticausativeMarking.unmarked.voiceLocus = Minimalist.VoiceProjectionLocus.voiceDOrBare
- Wood2015.AnticausativeMarking.ka.voiceLocus = Minimalist.VoiceProjectionLocus.vHead
Instances For
PF realization of Voice in Icelandic.
-st is the elsewhere exponent for Voice — it appears whenever Voice is non-agentive (modulo antipassive, which is unproductive in Icelandic).
Equations
Instances For
Wood-2015 analytical projection of a single Icelandic -st verb.
Bundles the apparatus the slim Fragment doesn't carry: -st
classification, anticausative marking, root-structure decomposition
(@cite{cuervo-2003}-style notation per Voice.lean §5–§6), and the
possessive-dative diagnostic.
The lexical fragment entry.
- stType : StType
Wood's -st classification.
- marking : AnticausativeMarking
Anticausative-marking morpheme. Default -st.
- rootStructure : List Minimalist.VerbHead
@cite{cuervo-2003}-style event decomposition.
- licensesPossessiveDative : Bool
Does this verb license possessive datives (Wood Ch. 5 diagnostic)?
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- Wood2015.instReprStVerbInfo = { reprPrec := Wood2015.instReprStVerbInfo.repr }
Equations
- One or more equations did not get rendered due to their size.
- Wood2015.instBEqStVerbInfo.beq x✝¹ x✝ = false
Instances For
Equations
- Wood2015.instBEqStVerbInfo = { beq := Wood2015.instBEqStVerbInfo.beq }
Equations
- One or more equations did not get rendered due to their size.
Instances For
opna / opnast 'open' — anticausative -st (@cite{wood-2015} Ch. 3). Active: Jón opnaði dyrnar 'John opened the door'; -st: Dyrnar opnuðust 'The door opened'.
Equations
- One or more equations did not get rendered due to their size.
Instances For
splundra / splundrast 'shatter' — anticausative -st (@cite{wood-2015} Ch. 3, §3.5). Central example for the Voice–CAUSE independence argument.
Equations
- One or more equations did not get rendered due to their size.
Instances For
brjóta / brotna 'break' — anticausative with -na, NOT -st (@cite{wood-2015} Ch. 3, §3.3.2). Comparison case: -na spells out specifierless Voice_{∅} whereas -st picks Voice_{D}.
Equations
- One or more equations did not get rendered due to their size.
Instances For
selja / seljast 'sell' — middle -st (@cite{wood-2015}).
Equations
- One or more equations did not get rendered due to their size.
Instances For
lesa / lesast 'read' — middle -st (@cite{wood-2015}).
Equations
- One or more equations did not get rendered due to their size.
Instances For
baða / baðast 'bathe' — reflexive -st (@cite{wood-2015}).
Equations
- One or more equations did not get rendered due to their size.
Instances For
klæða / klæðast 'dress' — reflexive -st (@cite{wood-2015}).
Equations
- One or more equations did not get rendered due to their size.
Instances For
nálgast 'approach' — inherent -st (@cite{wood-2015}). No active variant; -st is lexicalized.
Equations
- Wood2015.nalgast_info = { verb := Fragments.Icelandic.Predicates.nalgast, stType := Wood2015.StType.inherent, rootStructure := [Minimalist.VerbHead.vGO] }
Instances For
minnast 'remember' — inherent -st (@cite{wood-2015}). No active variant.
Equations
- Wood2015.minnast_info = { verb := Fragments.Icelandic.Predicates.minnast, stType := Wood2015.StType.inherent, rootStructure := [Minimalist.VerbHead.vBE] }
Instances For
leiðast 'be bored' — subject-experiencer -st (@cite{wood-2015}). Mér leiðist í skólanum 'I am bored in school'.
Equations
- Wood2015.leidast_info = { verb := Fragments.Icelandic.Predicates.leidast, stType := Wood2015.StType.subjectExp, rootStructure := [Minimalist.VerbHead.vBE] }
Instances For
kyssa / kyssast 'kiss' — reciprocal -st (@cite{wood-2015}). Þau kyssast 'They kissed (each other)'.
Equations
- One or more equations did not get rendered due to their size.
Instances For
All -st verb projections (excludes brotna which is -na-marked).
Equations
- One or more equations did not get rendered due to their size.
Instances For
The causative alternation for opna / opnast: same root, different Voice.
Voice alone determines causativity for change-of-state roots.
-st spells out Voice across all six construction types. Despite morphological uniformity, the underlying Voice configurations differ in ±θ/±D parameters.
Despite shared morphology, the Voice configurations differ.
Anticausative and middle Voice do NOT assign θ; reflexive does.
Stated using the canonical Voice constants from Voice.lean.
Lookup table for StType.voiceFlavor on the four
typologically-distinct cells. Bundled into a single theorem so
the table appears as one proof obligation.
-st spells out all non-agentive Voice flavors.
Active Voice gets no suffix.
The six -st types occupy distinct cells in the ±θ/±D space.
@cite{wood-2015}'s Voice_{D}/Voice_{∅} distinction maps to the ±D
axis (selectsSpecifier); Voice semantics (Ø vs AGENT) maps to
the ±λx axis.
High applicatives are blocked when Voice has no event semantics (@cite{pylkkanen-2008}, @cite{schaefer-2008}). Note: @cite{wood-2015} Ch. 5 argues Icelandic lacks true high applicatives; the asymmetry formalized here follows the cross-linguistic typology.
Low applicatives survive when Voice has no event semantics because they relate to the theme, not the event (@cite{pylkkanen-2008}).
In anticausatives, possessive datives also survive.
High applicatives ARE licensed with agentive Voice.
The full asymmetry: high applicatives require Voice with event semantics; low applicatives are independent of Voice (@cite{pylkkanen-2008}, @cite{schaefer-2008}).
Reflexive Voice predicts agent θ-role (agent acts on self).
Experiencer Voice predicts experiencer θ-role.
Anticausative Voice predicts no external θ-role.
Middle Voice predicts no external θ-role.
@cite{wood-2015}'s key applicative claim: -st cannot merge in SpecApplP because Appl assigns dative case and -st lacks case features.
In ditransitive -st alternations, Appl datives are retained because Appl assigns case independently of Voice (@cite{wood-2015} Ch. 5, §5.3.1).
-st and -na spell out different Voice projection loci
(@cite{wood-2015} Ch. 3). -st picks out Voice_{D}; -na picks
out Voice_{∅}. They never co-occur. Stated structurally over
VoiceProjectionLocus, so the inequality is genuine pattern
distinction (not a string-equality coincidence on labels).
-ka spells out v, distinct from both Voice exponents.
brotna is -na marked, not -st marked.
Different anticausative roots select different Voice types: opnast takes -st (Voice_{D}), brotna takes -na (Voice_{∅}). The marking choice is lexically determined per root.
All anticausative -st verbs have inchoative root structure (no vDO, with vGO + vBE per @cite{cuervo-2003}).
All anticausative -st verbs have change-of-state root structure
(contains vGO; finer projection than isInchoative).
Inherent -st verbs lack active variants.
Subject-experiencer -st verbs lack active variants.
Ten -st verb info entries (excludes brotna, which is -na-marked).