Francescotti (1995): Even: The Conventional Implicature Approach Reconsidered #
This file formalizes [Fra95]'s felicity condition for even: the sentence without
even must be more surprising than most of its contextually determined true neighbors, against
[Ben82]'s one neighbor and [KP79]'s all. The three conditions are one
threshold on the number of neighbors the prejacent surpasses (Meets); they are ordered
(meets_most_of_universal, meets_existential_of_most), and the paper's two counterexamples
are the two gaps between them: surpassing exactly one of several neighbors meets Bennett's
condition but not the majority (not_meets_most_of_one), and surpassing all but one meets the
majority but not Karttunen and Peters' (meets_most_of_all_but_one). The universal threshold
is the scalar presupposition Focus.Particles.evenPresup.
Implementation notes #
Neighbors are a list of alternatives with a surprise comparison, and the thresholds read the count of neighbors the prejacent surpasses, so the vagueness of most is the threshold itself; the paper's second gradient, the margin of surprise, is not formalized. The rows record the paper's scenarios by the counts its descriptions fix: how many neighbors the prejacent surpasses and how many there are.
References #
Equations
- Francescotti1995.instDecidableEqThreshold 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
- Francescotti1995.instReprThreshold = { reprPrec := Francescotti1995.instReprThreshold.repr }
Surpassing k of n neighbors meets the threshold.
Equations
- Francescotti1995.Meets Francescotti1995.Threshold.existential x✝¹ x✝ = (0 < x✝¹)
- Francescotti1995.Meets Francescotti1995.Threshold.universal x✝¹ x✝ = (x✝¹ = x✝)
- Francescotti1995.Meets Francescotti1995.Threshold.most x✝¹ x✝ = (x✝ < 2 * x✝¹)
Instances For
Equations
- One or more equations did not get rendered due to their size.
All is most, when there are neighbors.
Most is at least one.
Bennett's counterexample: surpassing exactly one of several neighbors, as Albert's passing surpasses Marie's, meets the existential threshold but not the majority.
The failing scenario: surpassing all but one neighbor, everyone but Marie, meets the majority but not the universal threshold.
Exactly half is not most: Andre barely in the taller half.
The number of neighbors the prejacent surpasses in surprise.
Equations
- Francescotti1995.surpassed prejacent neighbors moreSurprising = List.countP (fun (a : α) => decide (moreSurprising prejacent a)) neighbors
Instances For
The felicity condition at a threshold.
Equations
- Francescotti1995.Felicitous prejacent neighbors moreSurprising t = Francescotti1995.Meets t (Francescotti1995.surpassed prejacent neighbors moreSurprising) neighbors.length
Instances For
The existential threshold is Bennett's condition.
The universal threshold is Karttunen and Peters' condition.
The universal threshold is the scalar presupposition of even.
The paper's scenarios #
A scenario of the data: how many true neighbors the prejacent surpasses in surprise, how many there are, and whether the even-sentence is felicitous.
- surpassed : ℕ
- neighbors : ℕ
- felicitous : 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
Instances For
The majority threshold fits every scenario.
Bennett's threshold does not: it licenses Even Albert passed the exam.
Karttunen and Peters' threshold does not: it blocks Even Albert failed the exam.