Documentation

Linglib.Theories.Semantics.Reference.Reciprocals

Reciprocal Semantics: Anaphoric Relations and Scope #

@cite{dalrymple-haug-2024} @cite{dalrymple-et-al-1998} @cite{haug-dalrymple-2020}

Two competing analyses of reciprocal expressions like each other:

  1. Quantificational (@cite{heim-lasnik-may-1991}): the reciprocal is (or contains) a quantifier that can raise to the matrix clause, yielding a wide-scope (I-)reading. The local antecedent is bound by the raised quantifier part.

  2. Relational (@cite{dalrymple-haug-2024}, @cite{sternefeld-1998}, @cite{beck-2001}, @cite{dotlacil-2013}, @cite{haug-dalrymple-2020}): the reciprocal is a pronoun bearing an anaphoric relation to its antecedent. The narrow/wide scope ambiguity reduces to the choice of anaphoric relation: group identity (∪) for narrow scope vs. binding (=) for wide scope.

Three Anaphoric Relations #

Following @cite{higginbotham-1985} and @cite{williams-1991}, anaphoric dependencies between a pronoun and its antecedent come in three types:

The formal semantics of these relations is defined over plural information states in Theories/Semantics/Dynamic/PPCDRT/Anaphora.lean. This file holds the enum-level classification — the abstract relation labels, the scope readings, the antecedent property bundle, and the prediction functions used by the Dalrymple–Haug 2024 cross-construction survey.

Two-parameter scope classification #

@cite{haug-dalrymple-2020} §3.3 (p. 24) makes the reciprocal-scope classification two-dimensional: the locus of the reciprocal in the matrix DRS (high or low) crossed with the type of anaphoric relation between the matrix subject and the embedded local antecedent (binding or group identity). Three of the four cells are attested; the (low, bound) cell is empty (paper p. 24: "the bound reading of the reciprocal's antecedent cannot cooccur with a low locus for the reciprocal, because it does not make available the plurality that the reciprocal needs").

The RecipReading structure exposes locus alongside the two anaphoric relations so that consumers can witness the three cells distinctly.

Key Prediction #

The two analyses diverge on whether properties of the local antecedent (the embedded-clause pronoun coreferent with the matrix subject) can constrain reciprocal scope. The relational analysis predicts they can; the quantificational analysis predicts they cannot for cases involving distributive operators (§5) and logophoric antecedents (§6) — see @cite{dalrymple-haug-2024} for the empirical contrast.

The three types of anaphoric relation between a pronoun and its antecedent. Properties of the resolution, not the expression: the same pronoun (e.g., they) can participate in binding or group identity depending on context.

  • binding : AnaphoricRelation

    Bound variable: pronoun gets its value from a c-commanding binder. The antecedent denotes an individual.

  • groupIdentity : AnaphoricRelation

    Group identity: pronoun denotes the same plurality as its antecedent. Cumulative identity across all contexts.

  • reciprocity : AnaphoricRelation

    Reciprocity: cumulative identity across situations (same group) but per-situation distinctness (different individuals in each pair).

Instances For
    @[implicit_reducible]
    Equations
    Equations
    • One or more equations did not get rendered due to their size.
    Instances For

      Locus of the reciprocal in the matrix DRS. @cite{haug-dalrymple-2020} §3.3 (p. 24): the reciprocal is either interpreted in-situ inside the embedded clause (low locus) or lifted to the matrix DRS (high locus). The locus is one of the two parameters in the §3.3 classification of reciprocal readings.

      • high : RecipLocus

        High locus: reciprocal lifted to matrix DRS. Required for wide scope and for the crossed reading.

      • low : RecipLocus

        Low locus: reciprocal interpreted in-situ inside the embedded clause. Required for narrow scope.

      Instances For
        @[implicit_reducible]
        Equations
        Equations
        • One or more equations did not get rendered due to their size.
        Instances For

          Scope reading of a reciprocal in a complex sentence.

          • narrow (we-reading): "Tracy and Chris each thought 'We saw each other.'" The reciprocal is interpreted inside the embedded clause.
          • wide (I-reading): "Tracy thought 'I saw Chris' and Chris thought 'I saw Tracy.'" The reciprocal's semantic contribution is in the matrix clause.
          Instances For
            @[implicit_reducible]
            Equations
            Equations
            • One or more equations did not get rendered due to their size.
            Instances For

              The two families of reciprocal analysis.

              • quantificational : RecipAnalysis

                Reciprocal is/contains a quantifier that can QR to the matrix clause. @cite{heim-lasnik-may-1991}.

              • relational : RecipAnalysis

                Reciprocal is a pronoun bearing an anaphoric relation on its antecedent. @cite{sternefeld-1998}, @cite{beck-2001}, @cite{dotlacil-2013}, @cite{haug-dalrymple-2020}.

              Instances For
                @[implicit_reducible]
                Equations
                Equations
                • One or more equations did not get rendered due to their size.
                Instances For

                  Properties of the local antecedent of the reciprocal (the embedded-clause pronoun coreferent with the matrix subject) that affect scopal possibilities. @cite{dalrymple-haug-2024}.

                  • isBound : Bool

                    Whether the local antecedent is syntactically bound (=) by the matrix subject.

                  • hasCollectiveConjunct : Bool

                    Whether the embedded predicate is conjoined with a necessarily collective predicate.

                  • isExhaustiveControl : Bool

                    Whether the construction involves exhaustive control vs. partial control.

                  • controllerIsCollective : Bool

                    Whether the controller is interpreted collectively.

                  • forcesGroupIdentity : Bool

                    Whether the pronoun type forces group identity (∪), excluding the binding (=) option. Japanese zibun-tati (plural reflexive), @cite{nishigauchi-1992}.

                  • isLogophoric : Bool

                    Whether the antecedent is a logophoric pronoun.

                  • hasDistributiveOperator : Bool

                    Whether a distributive operator (each, each of them) is present in the matrix clause.

                  Instances For
                    Equations
                    • One or more equations did not get rendered due to their size.
                    Instances For

                      Scope readings predicted by the relational analysis.

                      Distributive operators are orthogonal on the relational analysis: each other is a pronoun, not a quantified NP, so distribution is orthogonal — both readings remain available. Logophoric antecedents restrict to narrow only.

                      Equations
                      • One or more equations did not get rendered due to their size.
                      Instances For

                        Scope readings predicted by the quantificational analysis.

                        Diverges from relationalPrediction on:

                        • Distributive operators: predicts narrow only (incorrect; both attested).
                        • Logophoric antecedents: predicts both (incorrect; only narrow attested).
                        Equations
                        • One or more equations did not get rendered due to their size.
                        Instances For

                          A synthetic Strongest Meaning Hypothesis at the scope-ambiguity layer (@cite{dalrymple-et-al-1998}'s SMH idea, applied to the choice between narrow and wide scope): when both readings are available, pick the logically stronger one. Narrow scope is stronger than wide in the sense that the narrow-scope reciprocity reading is more restrictive on doxastic alternatives.

                          Caveat. This is NOT what @cite{haug-dalrymple-2020} §6.1 actually argues against. The paper's §6.1 contrast (eq 132–133) is about SMH applied to reciprocal-reading STRENGTH (Strong vs Weak Reciprocity) under downward-entailing contexts — a different dimension from scope. Properly formalising the §6.1 argument would require a Strong/Weak Reciprocity gradation in the substrate, which PPCDRT does not currently expose. The SMH_diverges_from_relational theorem below is synthetic divergence at the scope layer — a Lean fact, not a paper-faithful refutation.

                          Equations
                          • One or more equations did not get rendered due to their size.
                          Instances For

                            Synthetic SMH-vs-relational divergence at the scope layer. On the default property bundle, the relational analysis leaves both readings available; the synthetic SMH commits to narrow only. See the caveat on strongestMeaningPrediction — this is not the paper's §6.1 argument, which is about Strong/Weak Reciprocity not formalised here.

                            A reciprocal reading per the @cite{haug-dalrymple-2020} two-parameter classification: locus of the reciprocal × type of antecedent relation × type of reciprocal-to-antecedent relation. The three valid cells are exhibited by narrowScopeReading, wideScopeReading, crossedReading; the fourth cell (low locus + bound antecedent) is empirically empty per paper p. 24.

                            • locus : RecipLocus

                              Locus of the reciprocal in the matrix DRS.

                            • antecedentRel : AnaphoricRelation

                              Anaphoric relation between the matrix subject and the embedded local antecedent.

                            • reciprocalRel : AnaphoricRelation

                              Anaphoric relation between the embedded local antecedent and the reciprocal pronoun itself.

                            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

                                  Narrow-scope reading (we-reading): low locus, group-identity antecedent, in-situ reciprocity. @cite{haug-dalrymple-2020} eq 52.

                                  Equations
                                  • One or more equations did not get rendered due to their size.
                                  Instances For

                                    Wide-scope reading (I-reading): high locus, bound antecedent, matrix-clause reciprocity. @cite{haug-dalrymple-2020} eq 54.

                                    Equations
                                    • One or more equations did not get rendered due to their size.
                                    Instances For

                                      Crossed reading (paper §3.3, eq 56): high locus, group-identity antecedent, group-identity reciprocal slot — reciprocity is contributed by the DRS distinctness presupposition ∂(u₃ ≠ u₂), not by an anaphoric reciprocity relation.

                                      Equations
                                      • One or more equations did not get rendered due to their size.
                                      Instances For

                                        The three attested cells. The empirically-empty fourth cell ({low, bound, _}) is not listed.

                                        Equations
                                        • One or more equations did not get rendered due to their size.
                                        Instances For

                                          The three readings are pairwise distinct as RecipReading records. Sanity check that the four-cell classification produces three different cells (not three name-aliases for the same cell).