@cite{ogihara-steinert-threlkeld-2024} — Data #
@cite{ogihara-steinert-threlkeld-2024}
Theory-neutral empirical data on the veridicality asymmetry between temporal connectives before and after.
Key Empirical Generalizations #
- After is veridical: "He left after she arrived" entails "she arrived".
- Before is non-veridical: "He left before she arrived" is compatible with her not arriving (the "barely prevented" reading).
- Before is non-veridical even with perfective complements: "The bomb exploded before anyone defused it" does not entail anyone defused it.
Data Sources #
- Ogihara, T. & Steinert-Threlkeld, S. (2024), §2–3.
- Anscombe, E. (1964), §3 (original observation of the asymmetry).
- Beaver, D. & Condoravdi, C. (2003), §2 (three readings of before).
An empirical judgment about whether a temporal connective entails the truth of its complement clause.
- sentence : String
The example sentence
- connective : String
The temporal connective
- complementEntailed : Bool
Does the sentence entail that the complement event occurred?
- gloss : String
Brief gloss of the entailment pattern
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
"He left after she arrived" entails "she arrived".
Equations
- One or more equations did not get rendered due to their size.
Instances For
"He left before she arrived" does NOT entail "she arrived". Compatible with: she did arrive (veridical reading), she didn't arrive (counterfactual reading, e.g. "before she could arrive"), or indeterminate (non-committal reading).
Equations
- One or more equations did not get rendered due to their size.
Instances For
"The bomb exploded before anyone defused it" — the complement event (defusing) did NOT occur. This is the counterfactual reading of before (@cite{beaver-condoravdi-2003}, "barely prevented").
Equations
- One or more equations did not get rendered due to their size.
Instances For
"She finished her coffee after he left" entails "he left".
Equations
- One or more equations did not get rendered due to their size.
Instances For
"The Supreme Court decided the election before the votes were counted" — non-committal: compatible with votes eventually being counted or never counted (@cite{beaver-condoravdi-2003}, ex. 22).
Equations
- One or more equations did not get rendered due to their size.
Instances For
"Mozart died before he finished the Requiem" — counterfactual: Mozart never finished the Requiem (@cite{beaver-condoravdi-2003}, ex. 24).
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
Before is antisymmetric: "Cleo was in America before David was" and "David was in America before Cleo was" cannot both be true (with non-overlapping intervals). (@cite{beaver-condoravdi-2003}, exx. 3-4)
Equations
- One or more equations did not get rendered due to their size.
Instances For
After is NOT antisymmetric: overlapping intervals allow both directions. (@cite{beaver-condoravdi-2003}, exx. 5-7, diagram 7)
Equations
- One or more equations did not get rendered due to their size.
Instances For
Before is transitive: if A before B and B before C, then A before C. (@cite{beaver-condoravdi-2003}, exx. 12-14)
Equations
- One or more equations did not get rendered due to their size.
Instances For
After is NOT transitive: overlapping intervals allow after(A,B) ∧ after(B,C) ∧ ¬after(A,C). (@cite{beaver-condoravdi-2003}, exx. 8-11)
Equations
- One or more equations did not get rendered due to their size.
Instances For
Before licenses NPIs; after does not. (@cite{beaver-condoravdi-2003}, exx. 15-18)
Equations
- One or more equations did not get rendered due to their size.
Instances For
"David won the race before he entered it" — pragmatically odd because winning temporally presupposes entering: there is no historical alternative where one wins before entering. (@cite{beaver-condoravdi-2003}, ex. 32)
Equations
- One or more equations did not get rendered due to their size.
Instances For
"David entered the race after he won it" — same temporal impossibility viewed through after: entering after winning reverses the natural temporal order. (@cite{beaver-condoravdi-2003}, ex. 33)
Equations
- One or more equations did not get rendered due to their size.
Instances For
A datum recording an empirical problem for B&C's branching-time analysis.
These are cases where before is used with complement events that are
in the past, which B&C's forward-branching alt(w,t) cannot handle.
- sentence : String
The example sentence
- reading : String
Which reading of before is involved?
- problem : String
Why B&C's analysis fails
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
"Shohei Ohtani was named the 2023 AL MVP before the 2023 MLB season ended."
The complement event (season ending) is in the PAST relative to the
naming event but also temporally precedes it. B&C's alt(w,t) at the
naming time cannot branch to alternatives where the season doesn't end,
because the season ending is already in the past. (O&@cite{ogihara-steinert-threlkeld-2024}, §5.1)
Equations
- One or more equations did not get rendered due to their size.
Instances For
"It snowed a lot in 2020 before the pandemic hit."
Both events are in the past. B&C's analysis requires alt(w,t) at the
snow time to include alternatives where the pandemic doesn't hit, but
the pandemic is also in the past. (O&@cite{ogihara-steinert-threlkeld-2024}, §5.1)
Equations
- One or more equations did not get rendered due to their size.
Instances For
"Nostradamus predicted many things before they happened." The complement events (predictions coming true) are in the past relative to utterance time. B&C would need alternatives where the predicted events never happen, but these events are already settled. (O&@cite{ogihara-steinert-threlkeld-2024}, §5.1)
Equations
- One or more equations did not get rendered due to their size.
Instances For
A datum recording asymmetries in the availability of non-committal readings, which B&C's Event Continuation Condition should but cannot fully predict.
- sentence : String
The example sentence
- nonCommittalAvailable : Bool
Is the non-committal reading available?
- explanation : String
Why the availability is as it is
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
"Mary will leave the party before Bill gets drunk." Non-committal reading is available: maybe Bill gets drunk, maybe not. B&C's Event Continuation Condition is satisfied (Bill getting drunk is a normal continuation). (O&@cite{ogihara-steinert-threlkeld-2024}, §5.2)
Equations
- One or more equations did not get rendered due to their size.
Instances For
The non-committal reading is sensitive to contextual plausibility: available when the complement is a normal continuation, unavailable when it is pragmatically impossible.
Cross-linguistic morphological evidence for the veridicality asymmetry.
- language : String
Language
- connective : String
The temporal connective (in the object language)
- gloss : String
English gloss
- observation : String
Key morphological observation
- supportsNonveridicality : Bool
Does this support the non-veridical analysis of before?
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Japanese mae ('before') requires non-past tense in its complement, even when describing past events. This independently supports the non-veridical analysis: the complement is presented as unrealized from the perspective of the main-clause event. (O&@cite{ogihara-steinert-threlkeld-2024}, §3)
Equations
- One or more equations did not get rendered due to their size.
Instances For
Japanese ato ('after') allows past tense in its complement, consistent with the veridical analysis: the complement event is presented as having occurred. (O&@cite{ogihara-steinert-threlkeld-2024}, §3)
Equations
- One or more equations did not get rendered due to their size.
Instances For
The Japanese tense asymmetry mirrors the veridicality asymmetry: mae (non-past complement) patterns with non-veridical before, ato (past complement) patterns with veridical after.
After is uniformly veridical across examples.
Before is uniformly non-veridical across all examples.
The asymmetry: after and before differ on complement entailment.
Before and after are opposites on all logical properties tested.
The Fragment entry for after matches the empirical datum: both record complement veridicality as true.
The Fragment entry for before matches the empirical datum: both record complement veridicality as false.
The veridicality asymmetry is reflected in the Fragment entries.
O&ST's theory derives after's veridicality from the double-existential quantificational structure: ∃e₁∃e₂[P(e₁) ∧ Q(e₂) ∧...] entails ∃e₂, Q(e₂).
This is not a stipulation in the Fragment — it follows from the semantics.
O&ST's theory derives before's non-veridicality from the universal quantification over the complement: ∃e₁[P(e₁) ∧ ∀e₂[Q(e₂) →...]] is vacuously true when Q has no witnesses.
Concretely: any P-event with an empty Q yields before(P, Q).
Scenario: "He left₁ after she arrived₀" with punctual events.
- leaving event at time 1
- arriving event at time 0 O&ST predicts: after(leave, arrive) holds (τ(arrive) ≺ τ(leave)).
Scenario: "He left₁ before she arrived₃" with punctual events.
- leaving event at time 1
- arriving event at time 3 O&ST predicts: before(leave, arrive) holds (τ(leave) ≺ τ(arrive)).
Scenario: "The bomb exploded₅ before anyone defused it" (nobody defused it). O&ST predicts: before(explode, defuse) holds vacuously (no defuse-events).
The punctual after-scenario projects correctly through eventDenotation: O&ST.after implies Anscombe.after on the projected interval sets.
The punctual before-scenario projects correctly through eventDenotation.
The logical properties of before and after noted by B&C follow directly from the quantificational structure. We verify each on concrete interval scenarios over ℤ, matching the B&C diagrams.
Before is antisymmetric on non-overlapping statives: if A before B, then ¬(B before A). (@cite{beaver-condoravdi-2003}, exx. 3-4)
Scenario: Cleo [1,5], David [8,12]. Cleo before David holds; David before Cleo does not.
The ∀-quantifier in Anscombe.before enforces this: if all of B follows some point in A, then no point in B precedes all of A.
After is NOT antisymmetric: overlapping intervals allow both after(A,B) and after(B,A). (@cite{beaver-condoravdi-2003}, exx. 5-7, diagram 7)
Scenario: Cleo [1,8], David [5,12]. Both Cleo-after-David and David-after-Cleo hold because ∃ requires only one witness.
Before is transitive: A before B ∧ B before C → A before C. (@cite{beaver-condoravdi-2003}, exx. 12-14)
Scenario: Delores [1,3], Ginger [6,8], Fred [11,13].
After is NOT transitive: overlapping intervals allow after(A,B) ∧ after(B,C) ∧ ¬after(A,C). (@cite{beaver-condoravdi-2003}, exx. 8-11)
Scenario: Fred [1,3], Ginger [2,5], Delores [4,7]. Fred after Ginger: t=3, t'=2. ✓ Ginger after Delores: t=5, t'=4. ✓ Fred after Delores: need t ∈ [1,3], t' ∈ [4,7], t' < t — impossible since max(Fred)=3 < 4=min(Delores). ✗
Before is antisymmetric in general: before(A,B) → ¬before(B,A).
From the ∀ in Anscombe's before: if ∃t ∈ A, ∀t' ∈ B, t < t', then for any s ∈ B we get t < s. But before(B,A) gives ∀ t ∈ A, s < t, so s < t and t < s — contradiction.
Note: no non-emptiness assumption needed.
Before is transitive in general: before(A,B) → before(B,C) → before(A,C).
From before(A,B): ∃t ∈ A, ∀t' ∈ B, t < t'.
From before(B,C): ∃s ∈ B, ∀s' ∈ C, s < s'.
Then t < s (from the first, since s ∈ B), and for any u ∈ C, s < u
(from the second). So t < u by transitivity of <. Witness: t ∈ A.
Note: no non-emptiness assumption needed — s ∈ timeTrace B is
provided by the second hypothesis.
The NPI datum matches the Fragment entry: before licenses NPIs.
The Japanese Fragment entry for mae agrees with the cross-linguistic datum: both record that mae supports the non-veridicality analysis.
The Japanese Fragment entry for ato agrees with the cross-linguistic datum: ato is veridical and does not support non-veridicality.