Fillmore, Kay & O'Connor (1988): Regularity and idiomaticity in grammatical constructions #
This file formalizes Fillmore, Kay and O'Connor's semantics of the let alone
construction. A let alone sentence F ⟨X A Y let alone B⟩ (20) asserts the two
propositions F′(X A Y) and F′(X B Y) (24) and presupposes a scalar model in which they are
distinct points, the full clause the stronger (§2.3). The Appendix defines the model in five
steps: an argument space, the product of at least two linearly ordered dimensions (A1), with its
product order as "lower than" (A2, lower_iff); a propositional function P such that, for
distinct points, P b entails P a exactly when a is lower than b (A3, IsScalarModel);
the same for the negated propositions (A4, IsScalarModel.not_le_not_iff); and informativeness
as one-way entailment (A5), the strict order on propositions, which A3 turns into the order on
points (IsScalarModel.lt_iff, IsScalarModel.not_lt_not_iff). The states of affairs of Tables
1–4, where truth propagates from the one-corner and falsity from the zero-corner, are the lower
sets of the argument space, and every lower-set state space is a scalar model
(isScalarModel_mem). The semantic conditions of §2.3.2 — one model, one polarity, the full
clause stronger (Felicitous) — thus reduce to an order on the foci: under negation A is lower
than B, under positive polarity higher (felicitous_negative_iff, felicitous_positive_iff).
Hence a fragment naming the lowest point of its scale is anomalous, (107) against (106)
(not_felicitous_of_isBot), and exchanging the foci of one dimension between points that
differ on two leaves them incomparable, (122) against (121) (swap_incomparable). The ranks of
(21), (106)–(107) and (130)–(132), the linguists and languages of Tables 1–4 and (109)–(112),
and the four dimensions of (121)–(122) instantiate the theory; the paper's judgments are the
rows of Data/Examples/FillmoreKayOConnor1988.json, and rankRows_predicted and
focusRows_predicted check them against Felicitous and the order in which the conjunction
presents the stronger clause (p. 533).
The construction itself is a formal idiom, a lexically open one in the sense of §1.1.3
(letAloneConstruction_isFormalIdiom). Its syntax (§2.2) — a
coordination that neither topicalizes nor clefts as a unit (31)–(34), admits no VP ellipsis
because the INFL-complex belongs to the trigger F (39)–(41), and is licensed in the affective
environments of Klima (62)–(70) yet attested under positive polarity when the fragment
denies the context proposition (71)–(73) — and its pragmatics (§2.4), the fragment answering
Relevance and the full clause Quantity, are the rows' comments and the docstrings below. Scales
are pragmatic, in the tradition of Fauconnier rather than the semantic scales of Horn and
Gazdar, and a scalar model needs a second dimension to bear its degrees (fn. 16, after
Cresswell).
Implementation notes #
IsScalarModelis stated over any partial order; the product order of (A1)–(A2) enters throughlower_iff, and the stipulation of at least two dimensions (fn. 16) is not part of the predicate. The one-dimensional rank chain is therefore treated as the slice of the persons × ranks model that fn. 16 envisages, and a slice of a scalar model is a scalar model (IsScalarModel.slice).- A state of affairs is a
LowerSetof the argument space and the propositional function is membership: "he made colonel" holds in a career exactly when colonel is among the ranks reached. The paper's Tables 2 and 4 draw four such states;table2cis the one that separates Brilliant can read English from Brilliant can read Greek. forcerenders F′ as negation or identity. The paper leaves open why only the negative part of barely reaches the fragment (§2.3.3, (118)–(119)); the barely rows are judgment data without a model, as are the constituency and licensing rows of §2.2.- The ranks are the three the paper names; a commissioned officer in (106)–(107) is "made the lowest commissioned rank", since in a lower-set state holding some rank is holding the lowest.
- The (104)/(105) contrast rests on the two-dimension stipulation and on a lottery model the paper describes but does not build; (104) is a row without a model.
TODO #
- The §1 typology of idioms (decoding and encoding, grammatical and extragrammatical, and §1.2's familiarity of pieces and of arrangement), once rows classify idioms by it. Its fourth cell, unfamiliar pieces familiarly arranged, is excluded by the argument of §1.2.1 that unique pieces have no standard principles of arrangement, which the typology should derive rather than omit.
References #
Scalar models (Appendix) #
In a product of linear orders, a is lower than b iff no coordinate of a is higher than
that of b and at least one is lower, which is the strict product order (A2).
By (A3), ⟨S, T, Dˣ, P⟩ is a scalar model iff, for distinct points a and b, P b entails
P a just in case a is lower than b. Propositions are ordered by entailment, the pointwise
order on S → Prop.
Equations
- FillmoreKayOConnor1988.IsScalarModel P = ∀ ⦃a b : D⦄, a ≠ b → (P b ≤ P a ↔ a < b)
Instances For
By (A4), ¬ P a entails ¬ P b just in case a is lower than b.
By (A5), P b is more informative than P a, entailing it and not conversely, just in case
a is lower than b.
By (A5) for the negated propositions, ¬ P a is more informative than ¬ P b just in case
a is lower than b.
Fixing a coordinate of a scalar model leaves a scalar model, so the bearers of a degree (fn. 16) may be held fixed.
The states of affairs conforming to a scalar model are the lower sets of its argument space — truth propagating from the one-corner, falsity from the zero-corner (Tables 1–4) — and with membership as the propositional function every lower-set state space is a scalar model.
The semantic conditions on let alone sentences (§2.3.2) #
F′ of (24), the semantic operator derived from the trigger F, is negation under negative polarity and identity under positive.
Equations
- FillmoreKayOConnor1988.force Polarity.negative x✝ = fun (x : S) => ¬x✝ x
- FillmoreKayOConnor1988.force Polarity.positive x✝ = x✝
Instances For
The conditions of §2.3.2 on a let alone sentence with foci a and b are that F′(X A Y) and
F′(X B Y) are propositions of one scalar model and one polarity, and that the full clause F′(X A Y)
is the more informative (A5).
Equations
- FillmoreKayOConnor1988.Felicitous P pol a b = (FillmoreKayOConnor1988.force pol (P a) < FillmoreKayOConnor1988.force pol (P b))
Instances For
Under negation the full clause is the stronger exactly when A is the lower point, as in he didn't make colonel, let alone general.
Under positive polarity the full clause is the stronger exactly when A is the higher point: you've got enough material for a whole semester, let alone a week (71).
A fragment naming the lowest point of the scale is anomalous (107). Nothing is lower than it, so the a-fortiori inference from the full clause has no lower point to start from.
Exchanging the foci of one dimension between two points that differ on two dimensions leaves the points incomparable, so neither clause is the stronger (122).
The construction (§2.1) #
In the let alone construction F ⟨X A Y let alone B⟩ (20a) the paired foci A and B flank let alone; the shared material X and Y and the trigger F are elided from the typed form.
Equations
- One or more equations did not get rendered due to their size.
Instances For
The form of let alone is lexically open, since such sentences "must therefore be given treatment as the kind of formal idiom or special construction we have been discussing" (§2.1).
The incredulity type Him be a doctor? (14h), §1.1.4's formal idiom that exists "in the service of specific pragmatic or rhetorical purposes": a non-nominative subject with a bare-stem predicate.
Equations
- One or more equations did not get rendered due to their size.
Instances For
The conjunction family (p. 533) #
The fragment-taking conjunctions that presuppose a scale relating their conjuncts: let alone, much less and not to mention present the stronger clause first, in fact and if not present it second (130)–(132).
- letAlone : Conjunction
- muchLess : Conjunction
- notToMention : Conjunction
- inFact : Conjunction
- ifNot : Conjunction
Instances For
Equations
- FillmoreKayOConnor1988.instDecidableEqConjunction 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
Whether the conjunction presents the stronger clause first.
Equations
- FillmoreKayOConnor1988.Conjunction.letAlone.StrongerFirst = True
- FillmoreKayOConnor1988.Conjunction.muchLess.StrongerFirst = True
- FillmoreKayOConnor1988.Conjunction.notToMention.StrongerFirst = True
- FillmoreKayOConnor1988.Conjunction.inFact.StrongerFirst = False
- FillmoreKayOConnor1988.Conjunction.ifNot.StrongerFirst = False
Instances For
Equations
- One or more equations did not get rendered due to their size.
The paper's judgments #
A sentence of the family with its conjunction, the polarity of its trigger, the points of its two foci and the paper's judgment.
- conj : Conjunction
- pol : Polarity
- a : D
- b : D
- judgment : Judgment
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
- FillmoreKayOConnor1988.instReprRow = { reprPrec := FillmoreKayOConnor1988.instReprRow.repr }
The point of the clause the conjunction presents as the stronger.
Equations
- r.stronger = if r.conj.StrongerFirst then r.a else r.b
Instances For
The point of the clause the conjunction presents as the weaker.
Equations
- r.weaker = if r.conj.StrongerFirst then r.b else r.a
Instances For
The conditions of §2.3.2 on the row, the conjunction fixing which clause must be the stronger.
Equations
Instances For
The conditions order the foci so that the stronger clause's point is the lower under negation and the higher under positive polarity.
Equations
- r.FociOrdered = match r.pol with | Polarity.negative => r.stronger < r.weaker | Polarity.positive => r.weaker < r.stronger
Instances For
Equations
- One or more equations did not get rendered due to their size.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- FillmoreKayOConnor1988.polarityTable = [("negative", Polarity.negative), ("positive", Polarity.positive)]
Instances For
Military rank: (21), (106)–(107), (130)–(132) #
Equations
- FillmoreKayOConnor1988.instDecidableEqRank x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Equations
- FillmoreKayOConnor1988.instReprRank = { reprPrec := FillmoreKayOConnor1988.instReprRank.repr }
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.
Equations
Instances For
Equations
He made rank r holds in a career, the lower set of the ranks reached, when r is in it.
Equations
- FillmoreKayOConnor1988.MadeRank x1✝ x2✝ = (x1✝ ∈ x2✝)
Instances For
In (107), He wasn't even a commissioned officer, let alone a second lieutenant, the fragment names the lowest point.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- FillmoreKayOConnor1988.rankFoci? ex = do let a ← ex.parse? "a" FillmoreKayOConnor1988.rankTable let b ← ex.parse? "b" FillmoreKayOConnor1988.rankTable pure (a, b)
Instances For
Equations
Instances For
Among (21), (106)–(107) and (130)–(132), the acceptable sentences are exactly those meeting the conditions of §2.3.2 with the conjunction's ordering of the stronger clause.
Linguists and languages: Tables 1–4, (109)–(112) #
The professors of Indo-European linguistics in order of erudition: "Apotheosis knows every language that Brilliant knows, Brilliant knows every language that Competent knows, and Competent knows every language that Dimm knows".
Instances For
Equations
- FillmoreKayOConnor1988.instDecidableEqLinguist 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
Equations
- One or more equations did not get rendered due to their size.
Equations
- FillmoreKayOConnor1988.instDecidableEqLang x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Equations
- FillmoreKayOConnor1988.instReprLang = { reprPrec := FillmoreKayOConnor1988.instReprLang.repr }
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.
Equations
Instances For
Equations
Equations
- FillmoreKayOConnor1988.instDecidableLTProdLinguistLang x✝¹ x✝ = { decide := decide (x✝¹ ≤ x✝) && decide ¬x✝ ≤ x✝¹, reflects_decide := ⋯ }
Professor p.1 can read language p.2 in a state of affairs.
Equations
- FillmoreKayOConnor1988.CanRead x1✝ x2✝ = (x1✝ ∈ x2✝)
Instances For
In Table 2c, Apotheosis reads English and French, and Brilliant reads English.
Equations
- One or more equations did not get rendered due to their size.
Instances For
In the Appendix's illustration, Brilliant can read English holds in Table 2c and Brilliant can read Greek does not, so the first does not entail the second, while the second entails the first because (Brilliant, English) is the lower point.
At the corners of Table 1, Dimm can read Hittite entails that every linguist reads every language, and Apotheosis can't read English that none reads any.
In (109)–(112) each (a) sentence is more informative than its (b) sentence, Brilliant can read Hittite than Brilliant can read French, Brilliant can't read French than Brilliant can't read Hittite, Competent can read Hittite than Brilliant can read French, and Brilliant can't read French than Competent can't read French.
Complex scales: (121)–(122) #
Equations
- FillmoreKayOConnor1988.instDecidableEqDim 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
- FillmoreKayOConnor1988.instReprDim = { reprPrec := FillmoreKayOConnor1988.instReprDim.repr }
Equations
- One or more equations did not get rendered due to their size.
A point of the four-dimensional argument space.
Equations
Instances For
Equations
- FillmoreKayOConnor1988.instDecidableLEPoint x y = { decide := decide (∀ a ∈ Finset.univ, (fun (d : FillmoreKayOConnor1988.Dim) => x d ≤ y d) a), reflects_decide := ⋯ }
Equations
- FillmoreKayOConnor1988.instDecidableLTPoint x✝¹ x✝ = { decide := decide (x✝¹ ≤ x✝) && decide ¬x✝ ≤ x✝¹, reflects_decide := ⋯ }
You could get a wealth man to task your vehicle for fee in a state of affairs.
Equations
- FillmoreKayOConnor1988.CanGet x1✝ x2✝ = (x1✝ ∈ x2✝)
Instances For
Equations
- FillmoreKayOConnor1988.Point.mk w t v f FillmoreKayOConnor1988.Dim.wealth = w
- FillmoreKayOConnor1988.Point.mk w t v f FillmoreKayOConnor1988.Dim.task = t
- FillmoreKayOConnor1988.Point.mk w t v f FillmoreKayOConnor1988.Dim.vehicle = v
- FillmoreKayOConnor1988.Point.mk w t v f FillmoreKayOConnor1988.Dim.fee = f
Instances For
Equations
- FillmoreKayOConnor1988.wealthTable = [("poor rich", 0, 1), ("rich poor", 1, 0)]
Instances For
Equations
- FillmoreKayOConnor1988.taskTable = [("wash wax", 0, 1), ("wax wash", 1, 0)]
Instances For
Equations
- FillmoreKayOConnor1988.vehicleTable = [("car truck", 0, 1), ("truck car", 1, 0)]
Instances For
Equations
- FillmoreKayOConnor1988.feeTable = [("$2 $1", 0, 1), ("$1 $2", 1, 0)]
Instances For
The foci of a row, one pair per dimension in the order A B.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
Instances For
In (121)–(122) the sentence with the likelier hiring in the full clause is acceptable and each exchange of one dimension's foci is not.
Every example that names a conjunction is read into one of the two models.