Landau (2024): Control #
Formalizes [Lan24] (Cambridge Elements in Generative Syntax),
the survey synthesizing the state of control theory: the revised
Visser's generalization ((19), [vU13]), the direct-discourse
correlation behind embedded-speech-act theories ((33),
[Pos70]) and its Korean jussive realization ((35)), the
attitude-only distribution of partial control ((36)–(37)), the
OC/NOC/NC trichotomy ((41)), NOC's dual topic/logophoric licensing
((52)), and the strict-vs-alternating adjunct split with the
propositional-variant criterion ((84)/(89)). The dual theory of §5
rests on the neutral substrate (Syntax/Control/Defs.lean) and the tier
system of Studies/Landau2015.lean; this file holds the Element's own
empirical generalizations and the complement typing ((56)/(58)/(72)). Its (95)
challenges — backward control, PC's residue, agreement under
property theories, the OC-NC generalization's crosslinguistic
variation ([Gan19]), overt-PRO licensing, adjunct loci — are
the open frontier.
References #
The OC/NOC/NC trichotomy #
NOC is the non-OC behavior of clauses that can display OC; clauses that never can (finite complements, in most languages) are no control (NC). The Spanish paradigm (41): infinitival complements are OC, infinitival subjects NOC, finite clauses NC in either position.
Equations
- Landau2024.instDecidableEqStatus x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Equations
- Landau2024.instReprStatus = { reprPrec := Landau2024.instReprStatus.repr }
Equations
- Landau2024.instReprStatus.repr Landau2024.Status.oc prec✝ = Repr.addAppParen (Std.Format.nest (if prec✝ ≥ 1024 then 1 else 2) (Std.Format.text "Landau2024.Status.oc")).group prec✝
- Landau2024.instReprStatus.repr Landau2024.Status.noc prec✝ = Repr.addAppParen (Std.Format.nest (if prec✝ ≥ 1024 then 1 else 2) (Std.Format.text "Landau2024.Status.noc")).group prec✝
- Landau2024.instReprStatus.repr Landau2024.Status.nc prec✝ = Repr.addAppParen (Std.Format.nest (if prec✝ ≥ 1024 then 1 else 2) (Std.Format.text "Landau2024.Status.nc")).group prec✝
Instances For
A row of the Spanish paradigm (41): clause finiteness, whether the clause sits in complement position, and its control status.
- finite : Bool
- complement : Bool
- status : Status
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- Landau2024.instReprEx41Row = { reprPrec := Landau2024.instReprEx41Row.repr }
The four configurations of (41).
Equations
- One or more equations did not get rendered due to their size.
Instances For
NOC occurs only in clauses that can display OC: never in finite clauses, whose uncontrolled behavior is NC.
The revised Visser's generalization #
(19), [vU13]: implicit subjects cannot control if T agrees with a referential DP. The Norwegian minimal pair (20): the personal passive (T agrees with the promisee) blocks implicit control; the impersonal passive (expletive, no referential agreement) allows it. The interrogative complements of (22) are a genuine standing exception.
A passive control configuration: does T agree with a referential DP, and is implicit control licensed?
- tAgreesReferential : Bool
- implicitControlOK : Bool
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- Landau2024.instReprPassiveConfig = { reprPrec := Landau2024.instReprPassiveConfig.repr }
The Norwegian pair (20): personal vs impersonal passive of 'promise'.
Equations
- Landau2024.ex20 = [{ tAgreesReferential := true, implicitControlOK := false }, { tAgreesReferential := false, implicitControlOK := true }]
Instances For
The revised Visser's generalization on (20): implicit control is licensed exactly where T does not agree with a referential DP.
The direct-discourse correlation and Korean jussives #
[Pos70]'s generalization ((33)): if the direct-discourse counterpart's subject is second person, the controller is the object; if first person, the subject. The Korean jussive markers realize the correlation grammatically ((35), with mal 'say' held fixed): the volitional marker is speaker-oriented and induces subject control, the imperative addressee-oriented inducing object control, the exhortative speaker+addressee-oriented inducing split control; the indicative induces no control.
Equations
- Landau2024.instDecidableEqDDSubject x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Equations
- Landau2024.instReprDDSubject = { reprPrec := Landau2024.instReprDDSubject.repr }
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- Landau2024.instDecidableEqChoice x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Equations
- Landau2024.instReprChoice.repr Landau2024.Choice.subject prec✝ = Repr.addAppParen (Std.Format.nest (if prec✝ ≥ 1024 then 1 else 2) (Std.Format.text "Landau2024.Choice.subject")).group prec✝
- Landau2024.instReprChoice.repr Landau2024.Choice.object prec✝ = Repr.addAppParen (Std.Format.nest (if prec✝ ≥ 1024 then 1 else 2) (Std.Format.text "Landau2024.Choice.object")).group prec✝
- Landau2024.instReprChoice.repr Landau2024.Choice.split prec✝ = Repr.addAppParen (Std.Format.nest (if prec✝ ≥ 1024 then 1 else 2) (Std.Format.text "Landau2024.Choice.split")).group prec✝
Instances For
Equations
- Landau2024.instReprChoice = { reprPrec := Landau2024.instReprChoice.repr }
[Pos70]'s correlation (33), stated over discourse roles: speaker-subject direct discourse → subject control, addressee-subject → object control, joint → split.
Equations
Instances For
Equations
- Landau2024.instDecidableEqJussive x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Equations
- Landau2024.instReprJussive = { reprPrec := Landau2024.instReprJussive.repr }
Equations
- One or more equations did not get rendered due to their size.
Instances For
The discourse orientation of each jussive marker.
Equations
Instances For
The controller each marker induces under mal 'say' ((35b–d)).
Equations
Instances For
The Korean jussive table realizes [Pos70]'s correlation: controller choice is the direct-discourse role of the marker's orientation — determined by the embedded mood, not the (fixed) matrix verb.
Partial control is attitude-only #
(36)–(37): PC is attested in attitude complements and unavailable
under implicatives — *John managed to gather at 6 (37a). On the
dual theory this is Control.IsSaturating.not_isPartial: implicative
complements are predicative, and predication saturates, sharing the
referent exhaustively. The (37a) configuration refutes saturating
status for its hypothetical PC reading, never the other way around.
The semantic layer a tier's complement inhabits ((56)/(58)): predicative complements are properties, logophoric ones propositions.
Equations
Instances For
Generalization (72): a lexical subject saturates a property, so exactly the propositional (logophoric, attitude) complements license one (the generalization originates with [Gra15]).
The (37a) configuration: matrix controller position 0, embedded
subject position 1.
Equations
- Landau2024.ex37Dependency = {(0, 1)}
Instances For
The PC reading's referent sizes: the controller (John) is properly contained in the gathering group.
Equations
- Landau2024.ex37Val p = if p = 0 then 1 else 2
Instances For
A partial-control reading is incompatible with a saturating dependency: (37a)'s star, from exhaustive sharing.
NOC: dual licensing by topic and logophoric center #
(52): topicality and logophoricity are each
sufficient for NOC antecedence and neither is necessary — (52a) is
NOC by a [−top, +log] antecedent (an implicit passive agent), (52b)
by [+top, −log] (an established topic with no mental perspective on
the event). The [+human] character of logophoric antecedents
follows from mental perspective; topical antecedents are only
preferentially human — the (55b) default [+topic] → [+human], with
(49) the resisting data and (50) the attested overrides.
A candidate NOC antecedent's discourse status.
- topic : Bool
- logophoricCenter : Bool
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- Landau2024.instReprAntecedent = { reprPrec := Landau2024.instReprAntecedent.repr }
(55a): a DP may serve as NOC controller iff it is a topic or a logophoric center.
Equations
- a.MayControl = (a.topic = true ∨ a.logophoricCenter = true)
Instances For
(52a): the implicit passive agent — a logophoric center that is no topic — controls.
(52b): the established topic with no perspective on the event controls.
Neither licensor is necessary — each (52) case lacks the other's.
A DP with neither status cannot serve as NOC controller.
Adjunct control: strict OC vs alternating OC/NOC #
(84): English controlled adjuncts split into strict OC adjuncts and adjuncts alternating OC/NOC; strict NOC adjuncts are unattested. The propositional-variant criterion ((89)): an adjunct type alternates iff its head also builds a variant hosting a lexical subject — the P head s-selects a property only (strict) or a property/proposition ambiguously (alternating).
The English controlled-adjunct types of (84).
- goal : AdjunctType
- result : AdjunctType
- stimulus : AdjunctType
- subjectPurpose : AdjunctType
- objectPurpose : AdjunctType
- rationale : AdjunctType
- temporal : AdjunctType
- absolutive : AdjunctType
- justification : AdjunctType
- telic : AdjunctType
Instances For
Equations
- Landau2024.instDecidableEqAdjunctType x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- Landau2024.instReprAdjunctType = { reprPrec := Landau2024.instReprAdjunctType.repr }
(84)'s observed split: does the adjunct type ever alternate into NOC?
Equations
- Landau2024.observedAlternates Landau2024.AdjunctType.goal = false
- Landau2024.observedAlternates Landau2024.AdjunctType.result = false
- Landau2024.observedAlternates Landau2024.AdjunctType.stimulus = false
- Landau2024.observedAlternates Landau2024.AdjunctType.subjectPurpose = false
- Landau2024.observedAlternates x✝ = true
Instances For
(89)'s structural column: does the adjunct head build a propositional variant (one hosting a lexical subject)?
Equations
- Landau2024.hasPropositionalVariant Landau2024.AdjunctType.goal = false
- Landau2024.hasPropositionalVariant Landau2024.AdjunctType.result = false
- Landau2024.hasPropositionalVariant Landau2024.AdjunctType.stimulus = false
- Landau2024.hasPropositionalVariant Landau2024.AdjunctType.subjectPurpose = false
- Landau2024.hasPropositionalVariant x✝ = true
Instances For
The propositional-variant criterion, validated on the English inventory: an adjunct type alternates into NOC exactly when its head has a propositional variant.