@cite{rouillard-2026}: Temporal in-Adverbials — Empirical Data #
@cite{kennedy-2007} @cite{rouillard-2026} @cite{vendler-1957} @cite{ladusaw-1979}
Empirical distributional data and verification theorems for temporal in-adverbials (TIAs), following @cite{rouillard-2026} "Maximal informativity accounts for the distribution of temporal in-adverbials" (L&P 49:1–56).
Data Points #
E-TIAs (Event TIAs): measure event durations #
- (1a) "Mary wrote up a paper in three days." ✓ — telic VP (accomplishment)
- (1b) "*Mary was sick in three days." ✗ — atelic VP (state)
- (88) "The climber reached the summit in three days." ✓ — telic (achievement)
G-TIAs (Gap TIAs): measure gap durations, NPIs #
- (2a) "Mary hasn't been sick in three days." ✓ — negative perfect
- (2b) "*Mary has been sick in three days." ✗ — positive perfect
- (48) "*Mary wasn't sick in three days." ✗ — simple past, no perfect
E‑ and U‑Perfect Ambiguity #
- (53) "Mary hasn't written up a paper in three days." — ambiguous E-TIA / G-TIA
- (54) "Mary has been sick for three days." — ambiguous E-perfect / U-perfect
Architectural Role #
This file is a Phenomena file: it imports Theories and verifies that theoretical predictions match empirical observations. It does NOT define new theoretical machinery.
E-TIA acceptability datum: VP class → acceptable with E-TIA?
- vp : String
Description of the VP
- vendlerClass : Semantics.Tense.Aspect.LexicalAspect.VendlerClass
Vendler class of the VP
- acceptable : Bool
Whether "VP in three days" is acceptable
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
(1a) "Mary wrote up a paper in three days." — accomplishment, ✓
Equations
- One or more equations did not get rendered due to their size.
Instances For
(1b) "*Mary was sick in three days." — state, ✗
Equations
- Phenomena.TenseAspect.TemporalAdverbials.Rouillard2026.datum_1b = { vp := "be sick", vendlerClass := Semantics.Tense.Aspect.LexicalAspect.VendlerClass.state, acceptable := false }
Instances For
(88) "The climber reached the summit in three days." — achievement, ✓
Equations
- Phenomena.TenseAspect.TemporalAdverbials.Rouillard2026.datum_88 = { vp := "reach the summit", vendlerClass := Semantics.Tense.Aspect.LexicalAspect.VendlerClass.achievement, acceptable := true }
Instances For
(84) "*The dancers waltzed in one hour." — activity, ✗
Equations
- Phenomena.TenseAspect.TemporalAdverbials.Rouillard2026.datum_84 = { vp := "waltz", vendlerClass := Semantics.Tense.Aspect.LexicalAspect.VendlerClass.activity, acceptable := false }
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
E-TIA acceptability is predicted by telicity: telic VPs accept E-TIAs, atelic VPs reject them.
Equations
Instances For
All E-TIA data points are predicted by telicity.
(1a) Accomplishment → telic → E-TIA acceptable.
(1b) State → atelic → E-TIA unacceptable.
(88) Achievement → telic → E-TIA acceptable.
(84) Activity → atelic → E-TIA unacceptable.
E-TIA licensing follows from the Kennedy–Rouillard isomorphism: telic VPs map to closed/bounded scales, predicting licensing.
Atelic VPs map to open/unbounded scales, predicting blocking.
The prediction matches the data for every datum.
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
(2a) "Mary hasn't been sick in three days." — negative perfect, ✓
Equations
Instances For
(2b) "*Mary has been sick in three days." — positive perfect, ✗
Equations
Instances For
(48) "*Mary wasn't sick in three days." — negative, no perfect, ✗
Equations
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
G-TIA acceptability predicted by: requires BOTH negative polarity AND perfect. @cite{rouillard-2026} Table 1: only NEG + G-TIA + PFV reading is acceptable.
Equations
Instances For
All G-TIA data points match the polarity + perfect prediction.
G-TIA licensing is derived from Fragment entry fields, not stipulated.
The prediction isNegative && hasPerfect matches the Fragment's
requiresNegative && requiresPerfect on inGTIA.
E-TIA licensing is derived from Fragment: requires telicity, not polarity.
G-TIA NPI status from Fragment: G-TIA is an NPI, E-TIA is not.
Stacking constraint from Fragment: E-TIA is event-level (inner), G-TIA is perfect-level (outer). The syntactic positions determine the only acceptable stacking order.
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
- One or more equations did not get rendered due to their size.
- Phenomena.TenseAspect.TemporalAdverbials.Rouillard2026.instBEqTable1Entry.beq x✝¹ x✝ = false
Instances For
All 8 cells of Table 1 (sentence (112)) "*Mary has been sick in 3 days").
Equations
- One or more equations did not get rendered due to their size.
Instances For
Exactly one reading survives: NEG + G-TIA + PFV (E-perfect).
The surviving reading is the negative G-TIA with perfective aspect.
The subinterval property (homogeneity) corresponds to open-scale
boundedness in the Kennedy–Rouillard isomorphism.
homogeneous_iff_atelic (from Aspect.lean) + atelic_open (from
MaximalInformativity.lean) form the bridge chain:
homogeneous → atelic → open scale → E-TIA blocked.
Non-homogeneous (telic) → closed scale → E-TIA licensed.
@cite{rouillard-2026} §6.1 — argues that G-TIAs are NPIs licensed by maximal informativity, NOT by downward entailment. The key evidence:
1. DE-based accounts (@cite{hoeksema-2006}, @cite{gajewski-2005}/2007) incorrectly predict
E-TIAs should also be polarity-sensitive (they aren't — E-TIAs are aspect-
sensitive, not polarity-sensitive).
2. The MIP accounts for BOTH E-TIA (aspect) and G-TIA (polarity) restrictions
from a single principle.
3. G-TIAs pattern with strong NPIs (anti-additive, not merely DE), but their
licensing condition is orthogonal to the DE hierarchy.
NPI licensing mechanism: DE vs MIP. Rouillard argues MIP subsumes DE for G-TIAs.
- downwardEntailment : LicensingMechanism
- maximalInformativity : LicensingMechanism
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
NPI prediction: does a licensing mechanism correctly predict both E-TIA and G-TIA distributions?
- mechanism : LicensingMechanism
- predictsETIA : Bool
Correctly predicts E-TIA distribution (aspect-sensitive, not polarity)?
- predictsGTIA : Bool
Correctly predicts G-TIA polarity sensitivity?
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
DE incorrectly restricts E-TIAs by polarity (Rouillard §6.1, p. 49–50): a condition like (144) would block E-TIAs in ALL non-DE environments, but E-TIAs are fine in positive telic sentences.
Equations
- One or more equations did not get rendered due to their size.
Instances For
MIP correctly handles both E-TIAs (via telicity/informativity) and G-TIAs (via open PTS/information collapse).
Equations
- One or more equations did not get rendered due to their size.
Instances For
MIP is strictly more explanatory: it handles both distributions. DE handles only G-TIAs and makes wrong predictions for E-TIAs.
Equations
- One or more equations did not get rendered due to their size.
Instances For
(131) "Since when has Mary been sick?" -- U-perfect only. @cite{von-fintel-iatridou-2019}: since-when Qs lack E- vs U-perfect ambiguity.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Since-when questions always lack the E-perfect reading. This is predicted by the MIP applied to Hamblin answerhood (eq. 135): ANS requires a maximally informative true answer in the Hamblin set, but the E-perfect Hamblin set can never have one (by density).
Fragment bridge: since requires the perfect and specifies the LB of PTS, matching the since-when question's presupposition structure.
@cite{rouillard-2026} §3.2, ex. (60) — when two TIAs are stacked, the inner (VP-adjacent) one must be an E-TIA and the outer one a G-TIA. The reverse order is ungrammatical.
(60a) "Mary hasn't written up a paper in three days in two weeks." ✓
inner = "in three days" (E-TIA, modifies VP)
outer = "in two weeks" (G-TIA, modifies AspP)
(60b) "#Mary hasn't written up a paper in two weeks in three days." ✗
inner = "in two weeks" (G-TIA?), outer = "in three days" (E-TIA?)
— violates the syntactic position constraint
Equations
- One or more equations did not get rendered due to their size.
Instances For
(60a) Inner E-TIA + outer G-TIA: acceptable.
Equations
- One or more equations did not get rendered due to their size.
Instances For
(60b) Reversed order: unacceptable.
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
Stacking acceptability = inner is E-TIA ∧ outer is G-TIA. Derives from syntactic positions: E-TIAs are event-level (VP-adjacent), G-TIAs are perfect-level (AspP-adjacent). Proximity to VP determines reading (Rouillard §3.2, schemata (57), (61)).
Equations
Instances For
Connecting E-TIA data to the universal LicensingPipeline #
The E-TIA data (§1–3) is verified against scaleBoundedness (§3) and
telicity (§2). Here we extend the bridge to LicensingPipeline, the
universal interface that connects adjective licensing, path licensing,
mereological licensing, and temporal licensing through a single mechanism.
Every E-TIA datum is predicted by LicensingPipeline.isLicensed.
Equations
Instances For
Pipeline agrees with direct scaleBoundedness on all four Vendler classes.
With compositional chains, the pipeline instance and scaleBoundedness
both route through VendlerClass →.telicity Telicity →.toMereoTag MereoTag →.toBoundedness Boundedness, so agreement is definitional.
Atelic (state/activity) collapses like open adjective (tall): SituationBoundedness.unbounded ↔ Boundedness.open_.
Bounded situations license like closed adjectives (full): SituationBoundedness.bounded ↔ Boundedness.closed.