@cite{schwarzer-2026} — Selection-Violating Coordination in German

Schwarzer, Luise. 2026. The law and order of selection-violating coordination: German DP-CP-coordination is not sensitive to linear or temporal order. Glossa 11(1). 1–19.

Main Question

When a CP appears coordinated with a DP in a DP-selecting position (selection-violating coordination), which conjunct must be the DP? Three families of analyses make different predictions:

1. Bottom-up (@cite{sag-etal-1985}, @cite{munn-1993}): asymmetric &P structure; the structurally prominent (first) conjunct must be the selected DP. Predicts DP-first universally.

2. Linear closeness (@cite{bruening-alkhalaf-2020}, @cite{bruening-2025}): left-to-right derivation; the linearly closest conjunct to the verb must satisfy selection. Predicts the preferred order depends on whether complements precede or follow the verb.

3. Temporal closeness (@cite{kim-lu-2024}): processing illusion; the temporally closest conjunct to the verb has its features checked. Makes the same predictions as linear closeness.

German as Test Case

German is OV in embedded clauses: complements precede the selecting verb. The linear/temporal accounts predict that in OV order, the rightmost (verb-adjacent) conjunct should be the DP, i.e., CP-first order should be preferred. The bottom-up account predicts DP-first regardless.

Results

Two experiments show German speakers uniformly prefer DP-first order (~77% in 2AFC) regardless of pre- or post-verbal position. This rules out linear/temporal closeness accounts and supports bottom-up analyses.

The paper's two experimental conditions correspond to different German clause types with different verb positions:

• Root declarative (V2): the verb moves to C°, landing in second position. Complements follow the verb → postverbal. Paper examples (17a/b).

• Embedded finite clause (verb-final): only V-to-I movement, not full V-to-C. Combined with SOV base order, the verb stays clause-final. Complements precede the verb → preverbal. Paper examples (16a/b).

The competing analyses make different predictions for the preverbal (= OV) condition. All agree on the postverbal (= VO) condition.

theorem Schwarzer2026.german_root_v2 : Minimalist.ForceHead.Decl ∈ Fragments.German.german

German has V2 in root declaratives: the finite verb moves to C°, placing it in second position. Complements follow the verb.

theorem Schwarzer2026.german_embedded_v_to_i : Minimalist.ForceHead.Fin ∈ Fragments.German.german

German has V-to-I movement in embedded finite clauses (not full V-to-C). Combined with the independently motivated SOV base order, this yields verb-final surface order in embedded clauses.

def Schwarzer2026.germanRootComplementPosition : Typology.WordOrder.VerbPosition

In German root declaratives, V2 places the verb before its complements. Coordination of complements is therefore postverbal.

This is the VO condition in @cite{schwarzer-2026}'s Experiment 2, corresponding to examples (17a/b) in the paper.

Equations

• Schwarzer2026.germanRootComplementPosition = Typology.WordOrder.VerbPosition.postverbal

def Schwarzer2026.germanEmbeddedComplementPosition : Typology.WordOrder.VerbPosition

In German embedded finite clauses, the verb remains clause-final (V-to-I in SOV base → verb at end of IP). Coordination of complements is therefore preverbal.

This is the OV condition in @cite{schwarzer-2026}'s Experiment 2, corresponding to examples (16a/b) in the paper.

Equations

• Schwarzer2026.germanEmbeddedComplementPosition = Typology.WordOrder.VerbPosition.preverbal

theorem Schwarzer2026.german_v2_grounds_postverbal : Minimalist.ForceHead.Decl ∈ Fragments.German.german ∧ germanRootComplementPosition = Typology.WordOrder.VerbPosition.postverbal

V2 root → postverbal complement position, grounded in V2Profile.

theorem Schwarzer2026.german_vfinal_grounds_preverbal : Minimalist.ForceHead.Fin ∈ Fragments.German.german ∧ germanEmbeddedComplementPosition = Typology.WordOrder.VerbPosition.preverbal

Embedded V-to-I + SOV base → preverbal complement position, grounded in V2Profile.

Each experimental verb is classified as CP-selecting or non-CP-selecting by deriving the classification from its complementType and altComplementType fields in the German fragment lexicon.

theorem Schwarzer2026.beenden_nonselecting : Fragments.German.Predicates.beenden.canTakeClausalComplement = false

beenden "end" does not take a dass-clause complement.

theorem Schwarzer2026.streichen_nonselecting : Fragments.German.Predicates.streichen.canTakeClausalComplement = false

streichen "cancel" does not take a dass-clause complement.

theorem Schwarzer2026.uebereilen_nonselecting : Fragments.German.Predicates.uebereilen.canTakeClausalComplement = false

übereilen "rush" does not take a dass-clause complement.

theorem Schwarzer2026.entwickeln_nonselecting : Fragments.German.Predicates.entwickeln.canTakeClausalComplement = false

entwickeln "develop" does not take a dass-clause complement.

theorem Schwarzer2026.veranlassen_selecting : Fragments.German.Predicates.veranlassen.canTakeClausalComplement = true

veranlassen "induce" takes both DP and dass-clause.

theorem Schwarzer2026.vergessen_selecting : Fragments.German.Predicates.vergessen.canTakeClausalComplement = true

vergessen "forget" takes both DP and dass-clause.

theorem Schwarzer2026.erwarten_selecting : Fragments.German.Predicates.erwarten.canTakeClausalComplement = true

erwarten "expect" takes both DP and dass-clause.

theorem Schwarzer2026.beschliessen_selecting : Fragments.German.Predicates.beschliessen.canTakeClausalComplement = true

beschließen "decide" takes both DP and dass-clause.

ConjunctOrder, bottomUpPrediction, and linearClosenessPrediction are defined in @cite{bruening-alkhalaf-2020} and imported via open. VerbPosition and OVOrder.verbPosition live in Typology.WordOrder substrate.

def Schwarzer2026.temporalClosenessPrediction : Typology.WordOrder.VerbPosition → BrueningAlKhalaf2020.ConjunctOrder

Temporal closeness prediction: the conjunct parsed closest in time to the verb has its features checked. Makes the same predictions as linear closeness.

Analysis: @cite{kim-lu-2024}.

Equations

• Schwarzer2026.temporalClosenessPrediction = BrueningAlKhalaf2020.linearClosenessPrediction

theorem Schwarzer2026.asymmetric_implies_position_invariant (pos : Typology.WordOrder.VerbPosition) : BrueningAlKhalaf2020.bottomUpPrediction pos = BrueningAlKhalaf2020.ConjunctOrder.dpFirst

Under asymmetric coordination, the bottom-up prediction is position-invariant: the structurally prominent conjunct is always the first, so DP-first is predicted regardless of verb position.

theorem Schwarzer2026.linear_predictions_differ : BrueningAlKhalaf2020.linearClosenessPrediction Typology.WordOrder.VerbPosition.preverbal ≠ BrueningAlKhalaf2020.linearClosenessPrediction Typology.WordOrder.VerbPosition.postverbal

Linear closeness predictions differ by position — this is the key empirical distinguisher.

theorem Schwarzer2026.temporal_equals_linear (pos : Typology.WordOrder.VerbPosition) : temporalClosenessPrediction pos = BrueningAlKhalaf2020.linearClosenessPrediction pos

Temporal closeness inherits linear closeness predictions exactly.

inductive Schwarzer2026.Exp1ComplementType : Type

Complement type in Experiment 1: DP-CP coordination vs bare dass-clause.

• coord : Exp1ComplementType
• dass : Exp1ComplementType

@[implicit_reducible]

instance Schwarzer2026.instDecidableEqExp1ComplementType : DecidableEq Exp1ComplementType

Equations

• Schwarzer2026.instDecidableEqExp1ComplementType x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯

def Schwarzer2026.instReprExp1ComplementType.repr : Exp1ComplementType → ℕ → Std.Format

Equations

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

@[implicit_reducible]

instance Schwarzer2026.instReprExp1ComplementType : Repr Exp1ComplementType

Equations

• Schwarzer2026.instReprExp1ComplementType = { reprPrec := Schwarzer2026.instReprExp1ComplementType.repr }

@[implicit_reducible]

instance Schwarzer2026.instBEqExp1ComplementType : BEq Exp1ComplementType

Equations

• Schwarzer2026.instBEqExp1ComplementType = { beq := Schwarzer2026.instBEqExp1ComplementType.beq }

def Schwarzer2026.instBEqExp1ComplementType.beq : Exp1ComplementType → Exp1ComplementType → Bool

Equations

• Schwarzer2026.instBEqExp1ComplementType.beq x✝ y✝ = (x✝.ctorIdx == y✝.ctorIdx)

structure Schwarzer2026.Exp1Cell : Type

Descriptive statistics from Experiment 1. Latin square design: SELECTION (yes/no) × COMPLEMENT (dass/coord). 50 participants on Prolific, 6 excluded; 44 per cell.

• complement : Exp1ComplementType
• selecting : Bool
• nObs : ℕ
• meanZ : ℤ
• medianZ : ℤ

@[implicit_reducible]

instance Schwarzer2026.instReprExp1Cell : Repr Exp1Cell

Equations

• Schwarzer2026.instReprExp1Cell = { reprPrec := Schwarzer2026.instReprExp1Cell.repr }

def Schwarzer2026.instReprExp1Cell.repr : Exp1Cell → ℕ → Std.Format

Equations

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

def Schwarzer2026.exp1_coord_nonsel : Exp1Cell

Equations

• Schwarzer2026.exp1_coord_nonsel = { complement := Schwarzer2026.Exp1ComplementType.coord, selecting := false, nObs := 44, meanZ := -253, medianZ := -244 }

def Schwarzer2026.exp1_coord_sel : Exp1Cell

Equations

• Schwarzer2026.exp1_coord_sel = { complement := Schwarzer2026.Exp1ComplementType.coord, selecting := true, nObs := 44, meanZ := 369, medianZ := 495 }

def Schwarzer2026.exp1_dass_nonsel : Exp1Cell

Equations

• Schwarzer2026.exp1_dass_nonsel = { complement := Schwarzer2026.Exp1ComplementType.dass, selecting := false, nObs := 44, meanZ := -526, medianZ := -684 }

def Schwarzer2026.exp1_dass_sel : Exp1Cell

Equations

• Schwarzer2026.exp1_dass_sel = { complement := Schwarzer2026.Exp1ComplementType.dass, selecting := true, nObs := 44, meanZ := 891, medianZ := 1080 }

theorem Schwarzer2026.coord_rescues_nonselected : exp1_coord_nonsel.meanZ > exp1_dass_nonsel.meanZ

Coordination is more acceptable than bare dass-clause in non-selecting contexts (the coordination "rescue" effect).

theorem Schwarzer2026.dass_better_when_selected : exp1_dass_sel.meanZ > exp1_coord_sel.meanZ

In selecting contexts, bare dass-clause is more acceptable than coordination (no rescue needed).

theorem Schwarzer2026.interaction_larger_unselected : exp1_coord_nonsel.meanZ - exp1_dass_nonsel.meanZ > exp1_coord_sel.meanZ - exp1_dass_sel.meanZ

The interaction: the coordination benefit is larger in non-selecting contexts than in selecting ones. Difference for coord vs dass: non-selecting: -253 - (-526) = +273; selecting: 369 - 891 = -522.

structure Schwarzer2026.Exp2Data : Type

2AFC experiment: participants chose between DP-first and CP-first order. 30 participants (48 recruited, 1 non-German excluded, 17 excluded for selecting ungrammatical option ≥2× in control).

• position : Typology.WordOrder.VerbPosition
• dpFirstCount : ℕ
• cpFirstCount : ℕ

@[implicit_reducible]

instance Schwarzer2026.instReprExp2Data : Repr Exp2Data

Equations

• Schwarzer2026.instReprExp2Data = { reprPrec := Schwarzer2026.instReprExp2Data.repr }

def Schwarzer2026.instReprExp2Data.repr : Exp2Data → ℕ → Std.Format

Equations

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

def Schwarzer2026.exp2_preverbal : Exp2Data

Equations

• Schwarzer2026.exp2_preverbal = { position := Schwarzer2026.germanEmbeddedComplementPosition, dpFirstCount := 23, cpFirstCount := 7 }

def Schwarzer2026.exp2_postverbal : Exp2Data

Equations

• Schwarzer2026.exp2_postverbal = { position := Schwarzer2026.germanRootComplementPosition, dpFirstCount := 23, cpFirstCount := 7 }

def Schwarzer2026.Exp2Data.total (d : Exp2Data) : ℕ

Total responses per condition.

Equations

• d.total = d.dpFirstCount + d.cpFirstCount

def Schwarzer2026.Exp2Data.majorityOrder (d : Exp2Data) : BrueningAlKhalaf2020.ConjunctOrder

The observed majority order in a 2AFC cell.

Equations

• d.majorityOrder = if d.dpFirstCount > d.cpFirstCount then BrueningAlKhalaf2020.ConjunctOrder.dpFirst else BrueningAlKhalaf2020.ConjunctOrder.cpFirst

theorem Schwarzer2026.position_invariance_dp : exp2_preverbal.dpFirstCount = exp2_postverbal.dpFirstCount

Position invariance: the counts are identical in pre- and post-verbal position. This is the central empirical finding.

theorem Schwarzer2026.position_invariance_cp : exp2_preverbal.cpFirstCount = exp2_postverbal.cpFirstCount

theorem Schwarzer2026.dp_majority_preverbal : exp2_preverbal.majorityOrder = BrueningAlKhalaf2020.ConjunctOrder.dpFirst

DP-first is the observed majority in both positions.

theorem Schwarzer2026.dp_majority_postverbal : exp2_postverbal.majorityOrder = BrueningAlKhalaf2020.ConjunctOrder.dpFirst

theorem Schwarzer2026.dp_preference_above_chance : exp2_preverbal.dpFirstCount * 2 > exp2_preverbal.total

DP-first probability is 23/30 ≈ 0.767, well above chance (0.5). Logistic regression: logit diff to control = +0.894, SE = 0.327, z = 2.74, p < 0.0001.

theorem Schwarzer2026.dp_preference_supermajority : exp2_preverbal.dpFirstCount * 4 ≥ exp2_preverbal.total * 3

DP-first accounts for more than 3/4 of responses.

theorem Schwarzer2026.bottomUp_correct_preverbal : BrueningAlKhalaf2020.bottomUpPrediction Typology.WordOrder.VerbPosition.preverbal = exp2_preverbal.majorityOrder

Bottom-up predicts correctly in preverbal position.

theorem Schwarzer2026.bottomUp_correct_postverbal : BrueningAlKhalaf2020.bottomUpPrediction Typology.WordOrder.VerbPosition.postverbal = exp2_postverbal.majorityOrder

Bottom-up predicts correctly in postverbal position.

theorem Schwarzer2026.linear_wrong_preverbal : BrueningAlKhalaf2020.linearClosenessPrediction Typology.WordOrder.VerbPosition.preverbal ≠ exp2_preverbal.majorityOrder

Linear closeness predicts incorrectly in preverbal position: it predicts CP-first, but DP-first is observed.

theorem Schwarzer2026.temporal_wrong_preverbal : temporalClosenessPrediction Typology.WordOrder.VerbPosition.preverbal ≠ exp2_preverbal.majorityOrder

Temporal closeness also wrong in preverbal position.

theorem Schwarzer2026.all_agree_postverbal : BrueningAlKhalaf2020.bottomUpPrediction Typology.WordOrder.VerbPosition.postverbal = BrueningAlKhalaf2020.linearClosenessPrediction Typology.WordOrder.VerbPosition.postverbal ∧ BrueningAlKhalaf2020.linearClosenessPrediction Typology.WordOrder.VerbPosition.postverbal = exp2_postverbal.majorityOrder

All three accounts agree in postverbal position (all predict DP-first, which is correct). The distinguishing power is preverbal only.

theorem Schwarzer2026.percolation_adjudication : BrueningAlKhalaf2020.predictOrder BrueningAlKhalaf2020.FeaturePercolation.structural Typology.WordOrder.VerbPosition.preverbal = exp2_preverbal.majorityOrder ∧ BrueningAlKhalaf2020.predictOrder BrueningAlKhalaf2020.FeaturePercolation.structural Typology.WordOrder.VerbPosition.postverbal = exp2_postverbal.majorityOrder ∧ BrueningAlKhalaf2020.predictOrder BrueningAlKhalaf2020.FeaturePercolation.linear Typology.WordOrder.VerbPosition.preverbal ≠ exp2_preverbal.majorityOrder

The German 2AFC data falsifies linear percolation: structural percolation (bottom-up) matches the observed majority in both positions, while linear percolation fails in the preverbal position that distinguishes them.

As @cite{schwarzer-2026} notes (§4), the results are "not straightforwardly an argument for" the bottom-up approach — there could be an independent DP-first preference — but they are decisive against linear/temporal closeness accounts.

theorem Schwarzer2026.temporal_is_linear_percolation (pos : Typology.WordOrder.VerbPosition) : temporalClosenessPrediction pos = BrueningAlKhalaf2020.predictOrder BrueningAlKhalaf2020.FeaturePercolation.linear pos

Temporal closeness (@cite{kim-lu-2024}) uses the same linear percolation mechanism as B&AK, applied to parsing time rather than surface string position.

def Schwarzer2026.structurePrediction : Features.Coordination.CoordSymmetry → Option (Typology.WordOrder.VerbPosition → BrueningAlKhalaf2020.ConjunctOrder)

Under asymmetric coordination structure, the bottom-up prediction follows: the first conjunct is structurally prominent and must satisfy selection. Under symmetric structure, no such prediction is made (both conjuncts are equally prominent).

Equations

• Schwarzer2026.structurePrediction Features.Coordination.CoordSymmetry.asymmetric = some BrueningAlKhalaf2020.bottomUpPrediction
• Schwarzer2026.structurePrediction Features.Coordination.CoordSymmetry.symmetric = none

theorem Schwarzer2026.asymmetric_entails_dp_first (pos : Typology.WordOrder.VerbPosition) : structurePrediction Features.Coordination.CoordSymmetry.asymmetric = some BrueningAlKhalaf2020.bottomUpPrediction ∧ BrueningAlKhalaf2020.bottomUpPrediction pos = BrueningAlKhalaf2020.ConjunctOrder.dpFirst

Asymmetric structure entails position-invariant DP-first prediction.

theorem Schwarzer2026.bottomUp_universal (pos : Typology.WordOrder.VerbPosition) : BrueningAlKhalaf2020.bottomUpPrediction pos = BrueningAlKhalaf2020.ConjunctOrder.dpFirst

The bottom-up prediction is universal: for any language, regardless of its OV/VO parameter, the predicted order is always DP-first.

theorem Schwarzer2026.ov_distinguishes_accounts : BrueningAlKhalaf2020.linearClosenessPrediction Typology.WordOrder.VerbPosition.preverbal = BrueningAlKhalaf2020.ConjunctOrder.cpFirst ∧ BrueningAlKhalaf2020.bottomUpPrediction Typology.WordOrder.VerbPosition.preverbal = BrueningAlKhalaf2020.ConjunctOrder.dpFirst

For OV languages, linear closeness makes the wrong prediction (CP-first in preverbal position), while bottom-up is correct.

theorem Schwarzer2026.ov_dpfirst_rules_out_linear (obs : BrueningAlKhalaf2020.ConjunctOrder) (h : obs = BrueningAlKhalaf2020.ConjunctOrder.dpFirst) : BrueningAlKhalaf2020.linearClosenessPrediction Typology.WordOrder.VerbPosition.preverbal ≠ obs

The German result generalizes: if an OV language shows DP-first preference in preverbal position, linear/temporal closeness accounts are ruled out for that language.

theorem Schwarzer2026.german_contradicts_bak_subject_evidence : BrueningAlKhalaf2020.linearClosenessPrediction Typology.WordOrder.VerbPosition.preverbal = BrueningAlKhalaf2020.ConjunctOrder.cpFirst ∧ exp2_preverbal.majorityOrder = BrueningAlKhalaf2020.ConjunctOrder.dpFirst

B&AK's English subject-position evidence (§3.1, examples (41a/b)) and Schwarzer's German preverbal data both test the preverbal configuration. B&AK predict CP-first for both (correct for English subjects per B&AK's judgments, but wrong for German complements). German data directly contradicts the closeness prediction in the preverbal environment where B&AK claim their strongest evidence.

def Schwarzer2026.nonSelectingVerbs : List Fragments.German.Predicates.GermanVerbEntry

The non-selecting verbs are exactly those used in the "not selected" condition of Experiment 1: they take only DP complements, so a dass-clause is unselected.

Equations

• Schwarzer2026.nonSelectingVerbs = [Fragments.German.Predicates.beenden, Fragments.German.Predicates.streichen, Fragments.German.Predicates.uebereilen, Fragments.German.Predicates.entwickeln]

def Schwarzer2026.selectingVerbs : List Fragments.German.Predicates.GermanVerbEntry

The CP-selecting verbs are exactly those used in the "selected" condition of Experiment 1.

Equations

• Schwarzer2026.selectingVerbs = [Fragments.German.Predicates.veranlassen, Fragments.German.Predicates.vergessen, Fragments.German.Predicates.erwarten, Fragments.German.Predicates.beschliessen]

theorem Schwarzer2026.nonSelecting_all_false : (nonSelectingVerbs.all fun (v : Fragments.German.Predicates.GermanVerbEntry) => !v.canTakeClausalComplement) = true

All non-selecting verbs lack clausal complement capability.

theorem Schwarzer2026.selecting_all_true : (selectingVerbs.all fun (v : Fragments.German.Predicates.GermanVerbEntry) => v.canTakeClausalComplement) = true

All selecting verbs have clausal complement capability.

theorem Schwarzer2026.selecting_nonselecting_disjoint : (List.filter (fun (v : Fragments.German.Predicates.GermanVerbEntry) => selectingVerbs.any fun (x : Fragments.German.Predicates.GermanVerbEntry) => x == v) nonSelectingVerbs).length = 0

No verb is in both lists.

theorem Schwarzer2026.german_wals_vs_clausetype : Fragments.German.wordOrder.ovOrder = Typology.WordOrder.OVOrder.noDominant ∧ germanRootComplementPosition = Typology.WordOrder.VerbPosition.postverbal ∧ germanEmbeddedComplementPosition = Typology.WordOrder.VerbPosition.preverbal

German's WALS OV order is "no dominant" (V2 root vs SOV embedded), but the experimental conditions derive their verb positions from clause type: root V2 → postverbal, embedded verb-final → preverbal (§ 1).

theorem Schwarzer2026.ov_languages_preverbal : Typology.WordOrder.OVOrder.ov.verbPosition = some Typology.WordOrder.VerbPosition.preverbal

For any OV language in the typological sample, the OVOrder→VerbPosition bridge yields preverbal position.

theorem Schwarzer2026.vo_languages_postverbal : Typology.WordOrder.OVOrder.vo.verbPosition = some Typology.WordOrder.VerbPosition.postverbal

For any VO language in the typological sample, the OVOrder→VerbPosition bridge yields postverbal position.

theorem Schwarzer2026.linear_wrong_for_ov_languages : Option.map BrueningAlKhalaf2020.linearClosenessPrediction Typology.WordOrder.OVOrder.ov.verbPosition = some BrueningAlKhalaf2020.ConjunctOrder.cpFirst

Linear closeness makes the wrong prediction for all OV languages: deriving VerbPosition from OVOrder and then applying linear closeness yields CP-first, which is refuted by German data.

theorem Schwarzer2026.bottomUp_correct_for_ov_languages : Option.map BrueningAlKhalaf2020.bottomUpPrediction Typology.WordOrder.OVOrder.ov.verbPosition = some BrueningAlKhalaf2020.ConjunctOrder.dpFirst

Bottom-up makes the correct prediction for all OV languages.

theorem Schwarzer2026.experimental_verbs_in_allVerbs : (Fragments.German.Predicates.allVerbs.any fun (x : Fragments.German.Predicates.GermanVerbEntry) => x == Fragments.German.Predicates.beenden) = true ∧ (Fragments.German.Predicates.allVerbs.any fun (x : Fragments.German.Predicates.GermanVerbEntry) => x == Fragments.German.Predicates.streichen) = true ∧ (Fragments.German.Predicates.allVerbs.any fun (x : Fragments.German.Predicates.GermanVerbEntry) => x == Fragments.German.Predicates.uebereilen) = true ∧ (Fragments.German.Predicates.allVerbs.any fun (x : Fragments.German.Predicates.GermanVerbEntry) => x == Fragments.German.Predicates.entwickeln) = true ∧ (Fragments.German.Predicates.allVerbs.any fun (x : Fragments.German.Predicates.GermanVerbEntry) => x == Fragments.German.Predicates.veranlassen) = true ∧ (Fragments.German.Predicates.allVerbs.any fun (x : Fragments.German.Predicates.GermanVerbEntry) => x == Fragments.German.Predicates.vergessen) = true ∧ (Fragments.German.Predicates.allVerbs.any fun (x : Fragments.German.Predicates.GermanVerbEntry) => x == Fragments.German.Predicates.erwarten) = true ∧ (Fragments.German.Predicates.allVerbs.any fun (x : Fragments.German.Predicates.GermanVerbEntry) => x == Fragments.German.Predicates.beschliessen) = true

All 8 experimental verbs are in the German fragment's allVerbs list.

theorem Schwarzer2026.experimental_conjunction_is_j : Fragments.German.Coordination.und.role = Features.Coordination.CoordRole.j

The coordination particle used in the experiments is German und, which is the J particle in the Mitrović & Sauerland decomposition.

theorem Schwarzer2026.german_j_only : Fragments.German.Coordination.conjunctionStrategy = Features.Coordination.ConjunctionStrategy.jOnly

German uses J-only conjunction strategy (overt und, covert MU).