[Ogi96]: Tense, Attitudes, and Scope #

[Ogi96] [Ogi89]

[Ogi96]'s theory: embedded past tense is ambiguous between a genuine past reading and a zero-tense reading. Zero tense is a bound variable that receives the matrix event time, producing the simultaneous reading.

The substrate enum PastReading lives in Semantics/Tense/SOT/Ambiguity.lean. This Studies file attributes the two-reading commitment to [Ogi96], derives the two predictions, and records the contrast with [Kra98a] (deletion, not ambiguity).

Core Mechanisms #

Ambiguous past: past morphology has two semantic values — genuine past (temporal precedence) vs zero tense (bound variable).
Zero tense = bound variable: receives binder time via lambda abstraction. The substrate primitive Tense.zeroTense_receives_binder_time proves the bound variable resolves to the binder.
SOT = zero tense: the simultaneous reading is the zero-tense reading.

Key Distinction from Kratzer #

Ogihara: past IS semantically ambiguous (two readings).
Kratzer: past is NEVER ambiguous; the simultaneous reading arises from morphosyntactic deletion of past at LF.

Both make identical predictions for the standard past-under-past data, but they disagree about the source of the simultaneous reading: ambiguity (Ogihara) vs deletion (Kratzer). The Phase F bridge program will land a typed contradiction witness on the embedded-present puzzle (where they actually diverge).

source

theorem Ogihara1996.ogihara_ambiguity_vs_deletion :

Tense.SOT.Ambiguity.PastReading.genuinePast ≠ Tense.SOT.Ambiguity.PastReading.zeroTense

[Ogi96]'s key claim: past IS ambiguous between genuine past and zero tense. This is a categorical structural difference from [Kra98a]'s deletion analysis. In Ogihara, the simultaneous reading = the zero-tense READING of past (semantic ambiguity); in Kratzer, it = deletion of past (morphological operation, no ambiguity).

source

theorem Ogihara1996.ogihara_derives_simultaneous {Time : Type u_1} (matrixFrame : _root_.Time.ReichenbachFrame Time) (g : Tense.TemporalAssignment Time) (n : ℕ) :

have embeddedR := Tense.interpTense n (Tense.updateTemporal g n matrixFrame.eventTime); (Tense.embeddedFrame matrixFrame embeddedR embeddedR).isPresent

[Ogi96] derives the simultaneous reading via the zero tense reading of past: the bound variable receives E_matrix. The derivation chain is zeroTense_receives_binder_time (substrate) → embeddedR = matrixFrame.eventTime → embeddedFrame.isPresent.

source

theorem Ogihara1996.ogihara_derives_shifted {Time : Type u_1} [LinearOrder Time] (matrixFrame : _root_.Time.ReichenbachFrame Time) (embeddedR embeddedE : Time) (hPast : embeddedR < matrixFrame.eventTime) :

(Tense.embeddedFrame matrixFrame embeddedR embeddedE).isPast

[Ogi96] derives the shifted reading via the genuine-past reading: the past tense contributes temporal precedence.

Reichenbach data frames for the Ogihara diagnostics that the schema cleanly admits. Verified against the book text: - (2b) Taroo-wa Hanako-ga byookidat-ta to it-ta (PAST matrix + PAST embedded) → ONLY shifted (embeddedByookiDatta). - (19d) "He said that Mary had been reading books yesterday" → past perfect with definite-past adverbial (pluperfectShifted).

**Schema-gap caveat: ex (2a) is intentionally not Lean-encoded.**
Ogihara ex (2a) `Taroo-wa Hanako-ga byooki-da to it-ta` (PAST matrix
+ PRESENT embedded) has a simultaneous reading: Hanako is sick at
Taro's saying time. The `ReichenbachFrame` schema cannot model this
faithfully — the embedded morphology says PRESENT (R = S), but the
simultaneous interpretation places the event at matrix E (≠ S).
A single (R, E) pair cannot simultaneously encode the morphological
tense AND the divergent event location. This is a general gap that
also bites counterfactual past, fake past, historical present, and
FID; it is not specific to Ogihara. The JSON record
`Examples.ex2a` captures the empirical pair (sentence + readings)
without committing to a Reichenbach analysis.

The pluperfect frame is built via `embeddedFrame` against the
local `matrixItta` (per CLAUDE.md "Theory-hub denotation as
study-file constraint" — `matrixItta`'s S/P/R/E = (0, 0, -2, -2)
matches the generic "X said Y" matrix shape, so the (19d)
"He said …" construction reuses it). The Japanese matrix is
root-clause; per Ogihara's absolute-tense analysis the embedded
(2b) frame keeps P = S = 0.

source

def Ogihara1996.matrixItta :

Time.ReichenbachFrame ℤ

Japanese matrix frame Taroo-wa ... to it-ta (Taro said ...). Past tense, perfective. S = P = 0, R = E = -2.

Equations

Ogihara1996.matrixItta = { speechTime := 0, perspectiveTime := 0, referenceTime := -2, eventTime := -2 }

Instances For

source

def Ogihara1996.embeddedByookiDatta :

Time.ReichenbachFrame ℤ

Ogihara ex (2b) embedded Hanako-ga byookidat-ta (Hanako had been sick) — PAST under PAST, ONLY the shifted reading. In non-SOT Japanese, embedded past is absolute (R < S, not R < matrix E). The unavailability of the simultaneous reading is Ogihara's cornerstone diagnostic.

Encoded as a standalone root-style frame (P = S = 0): per Ogihara Ch. 1 (p. 12), Japanese is an "absolute tense" language; the embedded clause does NOT shift its perspective time — that's precisely the empirical content of "absolute, not relative".

Equations

Ogihara1996.embeddedByookiDatta = { speechTime := 0, perspectiveTime := 0, referenceTime := -5, eventTime := -5 }

Instances For

source

def Ogihara1996.pluperfectShifted :

Time.ReichenbachFrame ℤ

Ogihara ex (19d) "He said that Mary had been reading books yesterday." English past perfect under past matrix. R < P (past) + E < R (perfect): the reading event precedes the reference time (within "yesterday"), which precedes the saying. Built via embeddedFrame against matrixItta (S/P/R/E = (0,0,-2,-2) — the generic "X said Y" matrix shape, reusable for the (19d) English example).

Equations