AI for linguistics Lean checks

← Anti-Dynamics · raw TODO.md

Anti-Dynamics (Schlenker 2007, JoLLI 16:325-356) — extraction checklist

Source: source/Schlenker-Anti-Dynamics-P-copy.{pdf,txt} (journal pages 325-356; the PDF was read visually wherever the text extraction lost logical symbols: pp. 336-342, 348, 351-353). Status legend: P proven in Lean as stated; PC proven after correction; CE counterexample found (checked in Lean); C confirmed (paper's claim checked in Lean); N not formalized (reason). All Lean names refer to lean/AntiDynamics/*.lean; line numbers are in RESULTS.md.

A. Definitions

#WhereContentStatus
A1§1.2 p.326Stalnaker's (i)-(iii): C[pp']=# unless presupposition holds on C; update with assertion; C[F and G]=C[F][G]N (historical background; (iii) is Heim's and, formalized in A6)
A2§1.3 (2) p.328deviant and*: C[F and* G]=C[G][F]C: andStarDef/Upd, andStar_eq_swap, andStar_agrees_trigfree, andStar_not_transparency, andStar_too_strong
A3§2.1 (5) p.330Definition of Transparency (α(d and γ)β ⇔ αγβ)formalized twice: contexts (Sys.Transp) and literal strings (Sys.StrTransp); equivalence PC/P strTransp_iff_transp
A4§2.1 (6) p.331Principle of Transparency (acceptability iff competitor ruled out)same as A3 (the "iff" is the definition of Transp)
A5§3.1 (18) p.335Syntax: predicates, propositions, formulas, Qi P . P, meta-languageFm, PLeaf, QLeaf, Pred (object language only; meta-language ∀d, ⇒, ⇔ are Lean's)
A6§3.2 (20)-(21) p.336Interpretation of lexical items; Heim/Beaver CCPs with definednessSys.Upd, Sys.Def, QModel.qupd/qdef
A7§3.2 (22) p.337Truth/falsity/presupposition failure at w ∈ CN (only a definition; Def and Upd are used directly)
A8§3.2 (25) p.338Static bivalent semantics (p̲p' = conjunction; if = material implication; quantifier via f(a,b))evalF, Sys.eval, QModel.qsem
A9§3.2 (26) p.338Principle of Transparency, concise form; Transp(C,F)= A3
A10§5.1 (28) p.344Non-TrivialitySys.NT
A11§5.3 p.348Definition of "accessed"/"parent" pairsAccessed
A12§5, intro p.343Heim's claims: (Q P̲P'. R) presupposes ∀d P(d); (Q P. R̲R') presupposes ∀d[P(d)⇒R(d)]C: heim_claim_i/ii, qdef_both

B. Lemmas

#WhereContentStatus
B1§2.2 (8), §3.2 (23) p.332,337Dynamic Transparency C[F]≠# ⇒ C[F]=C[F*]P: dynamic_transparency (propositional), dynamic_transparency_quant
B2§3.2 (24) p.337-8or* violates Dynamic Transparency (H=((not p) or* p̲p'))C: or_star_breaks_dynamic_transparency
B3§3.1 (19a) p.335Syntactic Lemma (a)P: syntactic_lemma_a (string level; proof in paper only sketched)
B4§3.1 (19b) p.335Syntactic Lemma (b)P: ser_append_inj (= unique readability; the paper's hypothesis "starts with (s, s not a parenthesis" is unnecessary)
B5§4 (27a,b) p.339Transparency LemmaP: transparency_lemma_a/b (via transp_conj/cond)
B6§5.1 (29) p.344-5Non-Triviality Corollary (i),(ii)P: nt_corollary_i/ii
B7§5.2 Lemma 1(i) p.345-7Transp(C,(Qi P̲P'.R)) iff C ⊨ ∀d P(d)PC: lemma1_i (proof of Case 2 has a gap: paper_zoneB_step_fails, repaired by find_step)
B8§5.2 Lemma 1(ii) p.347-8Transp(C,(Qi P.R̲R')) iff C ⊨ ∀d[P(d)⇒R(d)]P: lemma1_ii
B9§5.2 Remark p.345(c) can be replaced by Non-TrivialityP: quant_leaf (uses B6)
B10§5.3 Lemma 2 p.348-50NT is inherited by accessed pairsP: lemma2 (paper tacitly uses C[G]={w∈C:w⊨G}; here upd_eq_TS)
B11§4 proof, cases (a)-(f) p.339-42Inductive steps of Theorem 1P: transp_neg/conj/disj/cond, Sys.lift; typo found in (e)
B12§5.3 Thm 2 proof (a)-(g) p.351-2Inductive steps of Theorem 2 (incl. case (g))P: Sys.lift, quant_leaf, theorem2; typo found in (f)

C. Theorems

#WhereContentStatus
C1§4 Theorem 1(i) p.338Transp(C,F) iff C[F]≠# (propositional)P: theorem1_i, theorem1_i_strings
C2§4 Theorem 1(ii) p.339C[F]≠# ⇒ C[F]={w∈C: w⊨F}P: theorem1_ii
C3§5.3 Theorem 2 p.350quantified case under (i) constant domain size, (ii) constant restrictor size, (iii) NTCE as stated (theorem2_needs_expressiveness); PC: theorem2 with the extra hypothesis Expressive
C4§5.3 Thm 2 (ii)C'[F']={w∈C':w⊨F'}P: part of theorem2 (Sys.upd_eq_TS, unconditional)

D. Worked derivations / examples

#WhereContentStatus
D1§1.1 (1), §2.1 (3),(4),(7)Moldavia, Pavarotti dataN (empirical data)
D2§2.2 (10) p.332naive C ⊨ F ⇔ F* cannot yield asymmetryC: naive_conj_symmetric, transp_conj_asymmetric
D3§2.2 (11)-(12) p.332-3atomic case: naive = C ⊨ p'⇒p, too weakC: naive_atomic, naive_strictly_weaker
D4§2.2 (13) p.333Transparency of atom iff C ⊨ pC: transp_atomic
D5§2.3 (14) p.333(p̲p' and q) presupposes pC: ex14
D6§2.3 (15) p.333(p and q̲q') presupposes p⇒qC: ex15
D7§2.3 (16) p.334(if p̲p'. q) presupposes pC: ex16
D8§2.3 (17) p.334(if p. q̲q') presupposes p⇒qC: ex17

E. Quantifier discussion

#WhereContentStatus
E1§5.1 p.343scenario: 2 P-individuals, Q = less than three: Transp trivially holdsC: less_than_three_scenario
E2§5.1 p.343NT rules out (E1) when all worlds have < 3 PC: not_nt_of_lt3
E3§5.1 p.343-4three-world scenario (sizes 2,4,4): NT holds, Transp holds, Heim fails; Constancy neededC: constancy_needed
E4fn 12 p.344Q = infinitely many: if P^w−R^w finite, Transp holdsC: transpInf_of_finite; new converse not_transpInf_of_infinite, iff transpInf_iff
E5fn 16 p.355integers / lucky-not-to-be-13: P−Q a singleton so Transp holdsC: footnote16
E6§6.5 (34) p.354-5soccer example: Transp holds, Heim predicts failureC: soccer_example

F. §6 Conclusion claims

#WhereContentStatus
F1§6.1 (32)a-d p.352four candidate CCPs for unlessC, with typo in (32)d (unlessD_paper_is_wrong); (32)b,c leave the update undefined (unlessB_update_ill_defined)
F2§6.1 (33) p.352-3predictions of (32)a-d for "Unless John didn't come, Mary will know he is here"C: unless_33, unless_rules_differ
F3§6.1 p.353Transparency: unless ~ if notC: transp_unless_eq_if_not (Transparency verdict = (32)a: unless_33, last clause)
F4§6.1 p.353"same syntax + same bivalent content ⇒ same projection"true by construction of Transp (depends only on syntax and static semantics); instance F3
F5§6.2 p.353global/local accommodationN (informal proposal)
F6§6.3 p.354Transparency as meta-constraint on lexical CCPsN (research suggestion; not a claim)
F7§6.4, 6.6 p.354-5linear order vs processing order; DRT rivalsN (open questions)

G. Verification tasks (hunt for errors)

  • Statement of Theorem 1: hypotheses checked (needs ⊤,⊥ in the language; nothing else). No error.
  • Each case of the proof of Theorem 1: checked; typos Q4(a),(b),(c) in RESULTS.md.
  • Lemma 1 proofs step by step: gap in Case 2/Zone B (Q2), typo in Case 1 (Q3).
  • Statement of Theorem 2: missing hypothesis (Q1); notational ambiguity (Q5); circular-looking use of C[G] (Q6).
  • Quantifier equivalence "weaker than propositional": confirmed with explicit finite countermodels (E1, E3, E6, Q1).