Schwab2023: experimental results (generated)

Auto-generated from Linglib/Data/Experiments/Schwab2023.json by scripts/gen_experiments.py. Do not edit by hand: edit the JSON and re-run the generator.

Two speeded acceptability experiments on the German NPIs jemals 'ever' and so recht 'really', each crossing the NPI with the position of a negative quantifier: in the matrix clause, inside a relative clause, or absent. For each experiment the paper reports Bayes factors, from bridge sampling, comparing a model with the NPI illusion of each NPI, or with the interaction of the two illusions, to a model without it, and reads each as evidence of a stated strength. The posterior estimates are not recorded; Experiment 1 prints the estimate of the so recht illusion with the numbers of the jemals illusion, although its posterior in Figure 2 is centred near zero.

Raw data

• https://osf.io/s9rt8/: stimulus materials, code and data of both experiments, with supplementary materials

References

• [Sch23]

inductive Schwab2023.Experiment : Type

The two experiments.

• exp1 : Experiment

  1: subject-extracted relative clauses (section 2)

• exp2 : Experiment

  2: object-extracted relative clauses, faster presentation (section 3)

@[instance_reducible]

instance Schwab2023.instDecidableEqExperiment : DecidableEq Experiment

Equations

• Schwab2023.instDecidableEqExperiment x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯

def Schwab2023.instReprExperiment.repr : Experiment → ℕ → Std.Format

Equations

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

@[instance_reducible]

instance Schwab2023.instReprExperiment : Repr Experiment

Equations

• Schwab2023.instReprExperiment = { reprPrec := Schwab2023.instReprExperiment.repr }

@[instance_reducible]

instance Schwab2023.instFintypeExperiment : Fintype Experiment

Equations

• Schwab2023.instFintypeExperiment = { elems := { val := ↑Schwab2023.Experiment.enumList, nodup := Schwab2023.Experiment.enumList_nodup }, complete := Schwab2023.instFintypeExperiment._proof_1 }

inductive Schwab2023.Effect : Type

The effects the Bayes factors test, labelled as in Figures 2 and 4.

• illusionJemals : Effect

  NPI illusion: jemals: the illusory licensing of jemals by the relative-clause quantifier

• illusionSoRecht : Effect

  NPI illusion: so recht: the illusory licensing of so recht by the relative-clause quantifier

• interaction : Effect

  Interaction effect: illusory licensing: the difference between the two illusions

@[instance_reducible]

instance Schwab2023.instDecidableEqEffect : DecidableEq Effect

Equations

• Schwab2023.instDecidableEqEffect x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯

def Schwab2023.instReprEffect.repr : Effect → ℕ → Std.Format

Equations

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

@[instance_reducible]

instance Schwab2023.instReprEffect : Repr Effect

Equations

• Schwab2023.instReprEffect = { reprPrec := Schwab2023.instReprEffect.repr }

@[instance_reducible]

instance Schwab2023.instFintypeEffect : Fintype Effect

Equations

• Schwab2023.instFintypeEffect = { elems := { val := ↑Schwab2023.Effect.enumList, nodup := Schwab2023.Effect.enumList_nodup }, complete := Schwab2023.instFintypeEffect._proof_1 }

inductive Schwab2023.Verdict : Type

The paper's reading of a Bayes factor.

• veryStrongForEffect : Verdict

  very strong evidence for the effect: very strong evidence for the effect

• moderateForEffect : Verdict

  moderate evidence for the effect: moderate evidence in favour of the effect

• inconclusive : Verdict

  inconclusive: inconclusive

• slightlyForNull : Verdict

  slightly favours the null: slightly favoured the null hypothesis

• moderateForNull : Verdict

  moderate evidence for the null: moderate evidence for the null hypothesis

@[instance_reducible]

instance Schwab2023.instDecidableEqVerdict : DecidableEq Verdict

Equations

• Schwab2023.instDecidableEqVerdict x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯

@[instance_reducible]

instance Schwab2023.instReprVerdict : Repr Verdict

Equations

• Schwab2023.instReprVerdict = { reprPrec := Schwab2023.instReprVerdict.repr }

def Schwab2023.instReprVerdict.repr : Verdict → ℕ → Std.Format

Equations

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

@[instance_reducible]

instance Schwab2023.instFintypeVerdict : Fintype Verdict

Equations

• Schwab2023.instFintypeVerdict = { elems := { val := ↑Schwab2023.Verdict.enumList, nodup := Schwab2023.Verdict.enumList_nodup }, complete := Schwab2023.instFintypeVerdict._proof_1 }

structure Schwab2023.BayesFactor : Type

A row of sections 2.1.5 and 3.1.5, pp. 12-13 and 18: the Bayes factor for an effect and the paper's reading of it.

• bf10 : Data.Experiments.Decimal

  The Bayes factor for the effect over the null, BF10.

• verdict : Verdict

  The paper's reading of it.

@[instance_reducible]

instance Schwab2023.instDecidableEqBayesFactor : DecidableEq BayesFactor

Equations

• Schwab2023.instDecidableEqBayesFactor = Schwab2023.instDecidableEqBayesFactor.decEq

def Schwab2023.instDecidableEqBayesFactor.decEq (x✝ x✝¹ : BayesFactor) : Decidable (x✝ = x✝¹)

Equations

• Schwab2023.instDecidableEqBayesFactor.decEq { bf10 := a, verdict := a_1 } { bf10 := b, verdict := b_1 } = if h : a = b then h ▸ if h : a_1 = b_1 then h ▸ isTrue ⋯ else isFalse ⋯ else isFalse ⋯

def Schwab2023.instReprBayesFactor.repr : BayesFactor → ℕ → Std.Format

Equations

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

@[instance_reducible]

instance Schwab2023.instReprBayesFactor : Repr BayesFactor

Equations

• Schwab2023.instReprBayesFactor = { reprPrec := Schwab2023.instReprBayesFactor.repr }

def Schwab2023.bayesFactors : Experiment → Effect → BayesFactor

The cells of sections 2.1.5 and 3.1.5, pp. 12-13 and 18, by experiment and effect; checked against the page images.

Equations

• Schwab2023.bayesFactors Schwab2023.Experiment.exp1 Schwab2023.Effect.illusionJemals = { bf10 := { mantissa := 965, exponent := 2 }, verdict := Schwab2023.Verdict.moderateForEffect }
• Schwab2023.bayesFactors Schwab2023.Experiment.exp1 Schwab2023.Effect.illusionSoRecht = { bf10 := { mantissa := 43, exponent := 2 }, verdict := Schwab2023.Verdict.slightlyForNull }
• Schwab2023.bayesFactors Schwab2023.Experiment.exp1 Schwab2023.Effect.interaction = { bf10 := { mantissa := 116, exponent := 2 }, verdict := Schwab2023.Verdict.inconclusive }
• Schwab2023.bayesFactors Schwab2023.Experiment.exp2 Schwab2023.Effect.illusionJemals = { bf10 := { mantissa := 9874, exponent := 1 }, verdict := Schwab2023.Verdict.veryStrongForEffect }
• Schwab2023.bayesFactors Schwab2023.Experiment.exp2 Schwab2023.Effect.illusionSoRecht = { bf10 := { mantissa := 33, exponent := 2 }, verdict := Schwab2023.Verdict.moderateForNull }
• Schwab2023.bayesFactors Schwab2023.Experiment.exp2 Schwab2023.Effect.interaction = { bf10 := { mantissa := 654, exponent := 2 }, verdict := Schwab2023.Verdict.moderateForEffect }