Gradient At-Issueness and Projectivity #
[Rob12] [TBD18]
Tonhauser-Beaver-Degen 2018's gradient at-issueness model: degrees
and thresholds in Rat01, with at-issueness anti-correlated with
projectivity. Mirrors the gradable-adjective pattern (degree > θ →
positive meaning).
Degree Types #
A degree of at-issueness ∈ [0, 1]. 0 = fully backgrounded (not at-issue), 1 = fully at-issue.
Instances For
A degree of projectivity ∈ [0, 1]. 0 = no projection, 1 = obligatory projection.
Instances For
Contextual threshold for at-issueness classification. Content with degree above this threshold counts as at-issue.
Instances For
Contextual threshold for projectivity classification.
Instances For
Threshold Semantics #
Content is at-issue when its degree exceeds the threshold.
Equations
Instances For
Content is projective when its projectivity degree exceeds the threshold.
Equations
Instances For
Classical Recovery #
Binary at-issueness, recoverable from gradient degree + threshold.
- atIssue : AtIssuenessClassical
- notAtIssue : AtIssuenessClassical
Instances For
Equations
- Discourse.AtIssueness.instDecidableEqAtIssuenessClassical 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
Recover binary classification from gradient degree and threshold.
Equations
Instances For
Default threshold at 0.5, matching the midpoint of the [0, 1] scale.
Instances For
Default projectivity threshold at 0.5.
Instances For
Anti-Correlation #
A pair of at-issueness and projectivity ratings for a single expression.
- atIssueness : ℚ
- projectivity : ℚ
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
A monotone-anti-correlation bundle: every pair ordering on at-issueness forces the reverse weak ordering on projectivity. This is a structural strengthening of the empirical Gradient Projection Principle of [TBD18], who establish a gradient negative correlation across content classes (with substantial within-class and item-level variance), not a deterministic monotone pairing across every stimulus pair. Suitable for analytical scenarios where the strict-monotone fragment is the object of study; not a faithful summary of the paper's empirical claim.
- pairs : List GradientPair
Instances For
QUD Connection #
Qualitative QUD-based at-issueness: content varying within QUD cells counts as at-issue ([Rob12]).
Equations
- One or more equations did not get rendered due to their size.
Instances For
Boundedness #
Both at-issueness and projectivity are closed-bounded scales on [0, 1].
Instances For
Smart Constructors #
Construct a degree from [0, 100], normalizing to [0, 1].
Equations
- Discourse.AtIssueness.ofPercent n h0 h1 = ⟨n / 100, ⋯⟩