Documentation

Linglib.Studies.Burnett2019

[Bur19] — Signalling Games, Sociolinguistic Variation, and #

the Construction of Style

Linguistics and Philosophy 42: 419–450.

Overview #

Social Meaning Games (SMGs) model how sociolinguistic variant choice conveys social information. A speaker's use of -ing vs -in' induces listener inferences about persona traits (competent, friendly, etc.). The framework combines [Lew69]'s signalling games with RSA-style Bayesian reasoning to derive both style shifting (intra-speaker variation across contexts) and social stratification (inter-speaker variation across classes) from the same principles.

Architecture #

The meaning function is grounded in the Eckert–Montague lift from EckertMontague.emMeaningMI: each variant's Eckert field (a set of indexed properties) is lifted to persona compatibility via intersection semantics. The grounding theorem ingMeaning_eq_emMeaningMI verifies that the study's meaning function matches the theory-layer derivation.

Each context is a belief-based RSA over personae (worlds) and variants (utterances): the speaker Sk is the softmax of L₀⁶ (α = 6) and the listener L1k the Bayesian posterior (PMF.posterior) against the context-specific persona prior, which also weights L₀ (Burnett eq. 11). Predictions are exact PMF-value comparisons.

Key predictions #

  1. Per-persona variant preference: cool-guy prefers -in' ~69%
  2. Style shifting: casual→careful flips the cool-guy's preference
  3. Stern-leader exclusion: -in' is incompatible with stern leader
  4. Listener interpretation: Rice/Pelosi/Bush /t/ release predictions
  5. Bulletproofing: strong prior overwhelms variant effects (Bush)
  6. Cross-reference: model predictions close to [Lab12] data

Social properties (Burnett example (5)). Two bipolar dimensions: competence (competent/incompetent) and warmth (friendly/aloof).

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

      The four personae: maximally consistent subsets (Burnett example (6)). Each selects one pole per dimension.

      Instances For
        @[instance_reducible]
        Equations
        @[instance_reducible]
        Equations
        def Burnett2019.instReprPersona.repr :
        PersonaStd.Format
        Equations
        Instances For
          @[instance_reducible]
          Equations
          • One or more equations did not get rendered due to their size.

          Eckert fields (Burnett example (10)):

          The meaning function is derived via the Montagovian Individual / intersection semantics (Burnett footnote 14, Table 1): persona p is compatible with variant v iff p shares at least one property with v's Eckert field.

          The property space for Burnett's simplified example.

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

            The ING grounded field: both Eckert fields are consistent.

            Equations
            Instances For

              Grounding theorem: the inline meaning function equals the theory-layer emMeaningMI applied to the ING Eckert fields.

              -ing is compatible with 3 personae (Table 1: excludes doofus).

              -in' is compatible with 3 personae (Table 1: excludes stern leader).

              Each social context is a belief-based RSA over personae (worlds) and variants (utterances), α = 6 ([Bur19] p. 435):

              L₀(p | v) ∝ ⟦v⟧(p) · π(p)      (Bayesian literal listener, `L0k`)
              S₁(v | p) ∝ L₀(p | v)⁶         (`Sk`)
              L₁(p | v) ∝ π(p) · S₁(v | p)   (`L1k`, `PMF.posterior`)
              

              The persona prior enters L₀ directly (Burnett eq. 11): the "naive listener" weights compatible personae by π, not uniformly — this context-dependence drives style shifting. meaningE carries that prior-weighted meaning; the speaker is the softmax of L₀⁶ and the listener the Bayesian posterior against the same prior π (which sums to 1 for every context).

              def Burnett2019.meaningE (π : Persona) (v : Eckert2008.INGVariant) (p : Persona) :
              ENNReal

              Prior-weighted literal meaning ⟦v⟧(p) · π(p), lifted to ℝ≥0∞.

              Equations
              Instances For
                theorem Burnett2019.D_val {π : Persona} {v : Eckert2008.INGVariant} {mc ms md ma s : } (hc : meaningE π v Persona.coolGuy = ENNReal.ofReal mc) (hs : meaningE π v Persona.sternLeader = ENNReal.ofReal ms) (hd : meaningE π v Persona.doofus = ENNReal.ofReal md) (ha : meaningE π v Persona.asshole = ENNReal.ofReal ma) (hcnn : 0 mc) (hsnn : 0 ms) (hdnn : 0 md) (hann : 0 ma) (hsum : mc + ms + md + ma = s) :
                ∑' (p : Persona), meaningE π v p = ENNReal.ofReal s

                Denominator value D_v = Σ_p ⟦v⟧(p)·π(p), assembled from the four per-persona meaning values (ENNReal.ofReal of π p or 0).

                noncomputable def Burnett2019.L0k (π : Persona) ( : ∀ (p : Persona), 0 < π p) (v : Eckert2008.INGVariant) :

                Prior-weighted literal listener L₀(· | v) : PMF Persona.

                Equations
                Instances For
                  theorem Burnett2019.L0_val (π : Persona) ( : ∀ (p : Persona), 0 < π p) {v : Eckert2008.INGVariant} {p : Persona} {m d q : } (hm : meaningE π v p = ENNReal.ofReal m) (hd : ∑' (p' : Persona), meaningE π v p' = ENNReal.ofReal d) (hmnn : 0 m) (hdpos : 0 < d) (hq : m / d = q) :
                  (L0k π v) p = ENNReal.ofReal q

                  L0 value: L₀(p | v) = ⟦v⟧(p)·π(p) / D_v, an exact ENNReal.ofReal.

                  noncomputable def Burnett2019.Sk (π : Persona) ( : ∀ (p : Persona), 0 < π p) (p : Persona) :

                  Speaker S₁(· | p) : PMF INGVariant, softmax of L₀⁶ (α = 6).

                  Equations
                  Instances For
                    theorem Burnett2019.Sk_val (π : Persona) ( : ∀ (p : Persona), 0 < π p) {v : Eckert2008.INGVariant} {p : Persona} {a z q : } (ha : (L0k π v) p = ENNReal.ofReal a) (hz : ∑' (v' : Eckert2008.INGVariant), (L0k π v') p ^ 6 = ENNReal.ofReal z) (hann : 0 a) (hzpos : 0 < z) (hq : a ^ 6 / z = q) :
                    (Sk π p) v = ENNReal.ofReal q

                    Speaker value: S₁(v | p) = L₀(p|v)⁶ / Σ_v' L₀(p|v')⁶.

                    theorem Burnett2019.SkZ_val (π : Persona) ( : ∀ (p : Persona), 0 < π p) {p : Persona} {a b z : } (ha : (L0k π Eckert2008.INGVariant.velar) p = ENNReal.ofReal a) (hb : (L0k π Eckert2008.INGVariant.apical) p = ENNReal.ofReal b) (hann : 0 a) (hbnn : 0 b) (hz : a ^ 6 + b ^ 6 = z) :
                    ∑' (v : Eckert2008.INGVariant), (L0k π v) p ^ 6 = ENNReal.ofReal z

                    Speaker Z value: Σ_v L₀(p|v)⁶, assembled from the two L0_vals.

                    theorem Burnett2019.Sk_lt_iff (π : Persona) ( : ∀ (p : Persona), 0 < π p) {p : Persona} (v₁ v₂ : Eckert2008.INGVariant) :
                    (Sk π p) v₁ < (Sk π p) v₂ (L0k π v₁) p ^ 6 < (L0k π v₂) p ^ 6

                    Endorsement reduction (<): S₁ prefers v₂ iff L₀⁶ does.

                    theorem Burnett2019.prior_pmf_sum (π : Persona) ( : ∀ (p : Persona), 0 π p) (h : π Persona.coolGuy + π Persona.sternLeader + π Persona.doofus + π Persona.asshole = 1) :
                    p : Persona, ENNReal.ofReal (π p) = 1

                    The persona prior sums to 1 (every context is a distribution).

                    noncomputable def Burnett2019.priorK (π : Persona) (h1 : p : Persona, ENNReal.ofReal (π p) = 1) :

                    The listener's persona prior π : PMF Persona.

                    Equations
                    Instances For
                      @[simp]
                      theorem Burnett2019.priorK_apply (π : Persona) (h1 : p : Persona, ENNReal.ofReal (π p) = 1) (p : Persona) :
                      (priorK π h1) p = ENNReal.ofReal (π p)
                      theorem Burnett2019.marg_ne_zero (π : Persona) ( : ∀ (p : Persona), 0 < π p) (h1 : p : Persona, ENNReal.ofReal (π p) = 1) (v : Eckert2008.INGVariant) :
                      PMF.marginal (Sk π ) (priorK π h1) v 0
                      noncomputable def Burnett2019.L1k (π : Persona) ( : ∀ (p : Persona), 0 < π p) (h1 : p : Persona, ENNReal.ofReal (π p) = 1) (v : Eckert2008.INGVariant) :

                      Pragmatic listener L₁(· | v) : PMF Persona, the Bayesian posterior.

                      Equations
                      Instances For
                        theorem Burnett2019.L1_lt_iff (π : Persona) ( : ∀ (p : Persona), 0 < π p) (h1 : p : Persona, ENNReal.ofReal (π p) = 1) (v : Eckert2008.INGVariant) (p₁ p₂ : Persona) :
                        (L1k π h1 v) p₁ < (L1k π h1 v) p₂ ENNReal.ofReal (π p₁) * (Sk π p₁) v < ENNReal.ofReal (π p₂) * (Sk π p₂) v

                        Listener reduction (<): L₁ prefers p₂ iff the prior-weighted speaker scores do; the marginal cancels.

                        Casual-context prior (Burnett Table 2): voters at the barbecue think Obama is aloof (personae with aloof get more weight).

                        Equations
                        Instances For
                          @[reducible, inline]
                          noncomputable abbrev Burnett2019.casualSk :

                          Speaker at the casual context.

                          Equations
                          Instances For

                            Careful-context prior (Burnett Table 5): journalists think Obama is incompetent (incompetent personae get more weight).

                            Equations
                            Instances For
                              @[reducible, inline]

                              Speaker at the careful context.

                              Equations
                              Instances For

                                Rice: uniform prior — unfamiliar politician (Burnett Table 10).

                                Equations
                                Instances For
                                  theorem Burnett2019.riceSum1 :
                                  p : Persona, ENNReal.ofReal (ricePrior p) = 1
                                  @[reducible, inline]
                                  noncomputable abbrev Burnett2019.riceL1 :

                                  Listener for the Rice item (uniform prior).

                                  Equations
                                  Instances For
                                    theorem Burnett2019.pelosiSum1 :
                                    p : Persona, ENNReal.ofReal (pelosiPrior p) = 1
                                    @[reducible, inline]
                                    noncomputable abbrev Burnett2019.pelosiL1 :

                                    Listener for the Pelosi item.

                                    Equations
                                    Instances For

                                      Bush: listeners almost certain he is {inarticulate, aloof} (Burnett Table 15).

                                      Equations
                                      Instances For
                                        theorem Burnett2019.bushSum1 :
                                        p : Persona, ENNReal.ofReal (bushPrior p) = 1
                                        @[reducible, inline]
                                        noncomputable abbrev Burnett2019.bushL1 :

                                        Listener for the Bush item.

                                        Equations
                                        Instances For

                                          Denominators, literal-listener, and speaker values #

                                          Bush bulletproofing: exact posterior values #

                                          Cool-guy at the barbecue prefers -in' over -ing (~69% vs ~31%). Burnett (p. 435): "we predict that Obama will use -in' around 69% of the time [...] which is close to what Labov found" (72%).

                                          Stern leader only uses -ing: -in' is incompatible (Table 1). This predicts ~0% -in' in formal contexts where Obama constructs the stern leader.

                                          In the careful context, the cool-guy now prefers -ing over -in'. The prior shift reverses the informativity ranking.

                                          The /t/ release variable has the same mathematical structure as (ING). Relabeling: articulate↔competent, inarticulate↔incompetent (same friendly/aloof). Variants: released [tʰ]↔-ing, flapped [ɾ]↔-in'. The Eckert fields are structurally identical (Burnett example (19)): [tʰ] = {articulate, aloof}, [ɾ] = {inarticulate, friendly}.

                                          We reuse the same types and meaning function, since the math is isomorphic. The personae reinterpret as: coolGuy ↔ {articulate, friendly}, sternLeader ↔ {articulate, aloof}, doofus ↔ {inarticulate, friendly}, asshole ↔ {inarticulate, aloof}.

                                          The asshole prefers -in' in the casual context (both variants are compatible, but -in' is more informative given the prior).

                                          Rice: released /t/ triggers {articulate, aloof} = stern leader (Burnett Table 11). With uniform prior, the exclusive variant (only -ing compatible) gets double the L1 weight.

                                          Rice: flapped /t/ triggers {inarticulate, friendly} = doofus (Burnett Table 11). Symmetric to the released case.

                                          Pelosi: released /t/ predominantly triggers {inarticulate, aloof} — the strong prior that she is inarticulate overwhelms the released /t/ association with articulateness (Burnett Table 14).

                                          Bush "bulletproofing" (Burnett p. 444, Table 16): the prior is so extreme that variant choice has no practical effect. Both released and flapped /t/ yield >90% {inarticulate, aloof}.

                                          Cross-reference: the SMG model's qualitative predictions match the directional pattern observed in [Lab12]'s data on Obama's (ING) rates. The model predicts the cool-guy persona prefers -in' in casual context and -ing in careful context; the data shows Obama's -in' rate decreasing monotonically from casual (72%) through careful (33%) to formal (3%).

                                          Burnett's Social Meaning Game (SMG): a signalling game in which a speaker's variant choice conveys social information about their persona. An SMG is an interpretation game in [Fra11]'s sense, extended with the speaker's value function μ over personae ([Bur19] Def. 4.3). toInterpGame converts any SMG onto the shared signaling-game carrier.

                                          Burnett's own agent dynamics — the informativity speaker (eqs. (12)–(13)), soft-max persona selection (eqs. (14)–(15)), the Bayesian naive listener (eq. (18)) and the value-inferring uncovering listener (eq. (20)) — are not formalized in this section; the quantitative eq.-(13)-style speaker lives in the Sk pipeline of §3. What this section verifies is the structural layer: the carrier's literal listener on the converted game captures the exclusion behavior of the meaning function.

                                          structure Burnett2019.SocialMeaningGame (P V : Type) [Fintype P] [Fintype V] [DecidableEq P] [DecidableEq V] :

                                          A Social Meaning Game ([Bur19] Def. 4.1, extended per Def. 4.3 with the value function): a signalling game where variant choice conveys social information.

                                          • P: persona types (social categories the listener is trying to infer)
                                          • V: variant types (linguistic forms the speaker chooses)
                                          • prior: probability distribution over personae
                                          • meaning: whether a variant is compatible with a persona (derived from the EM field: v means t iff the EM lift of v includes persona t)
                                          • socialEval: Def. 4.3's μ — the value the speaker assigns to constructing each persona in the context (persona-only, as in the paper)
                                          • prior : P

                                            Prior probability over personae.

                                          • prior_nonneg (t : P) : 0 self.prior t

                                            Prior is non-negative.

                                          • meaning : VPBool

                                            Semantic meaning: is variant v compatible with persona t?

                                          • socialEval : P

                                            Def. 4.3's value function μ: how much the speaker values constructing persona t in the context.

                                          Instances For
                                            def Burnett2019.SocialMeaningGame.toInterpGame {P V : Type} [Fintype P] [Fintype V] [DecidableEq P] [DecidableEq V] (smg : SocialMeaningGame P V) :

                                            Convert a Social Meaning Game to an interpretation game ([Fra11]) on the shared signaling-game carrier.

                                            The mapping:
                                            - Types = Personae (what the listener tries to infer)
                                            - Messages = Variants (what the speaker chooses)
                                            - meaning = SMG meaning (EM field compatibility)
                                            - prior = SMG prior over personae 
                                            
                                            Equations
                                            • smg.toInterpGame = { meaning := fun (v : V) (p : P) => smg.meaning v p = true, prior := smg.prior, meaningDecidable := fun (m : V) (a : P) => instDecidableEqBool (smg.meaning m a) true }
                                            Instances For
                                              def Burnett2019.fromGroundedField {V : Type} [Fintype V] [DecidableEq V] (ps : SocialMeaning.PropertySpace) (gf : SocialMeaning.EckertMontague.GroundedField V ps) (personaeSets : Finset (Finset ps.Property)) [Fintype personaeSets] [DecidableEq personaeSets] (prior : personaeSets) (prior_nonneg : ∀ (t : personaeSets), 0 prior t) (socialEval : personaeSets) :
                                              SocialMeaningGame (↥personaeSets) V

                                              Construct a Social Meaning Game from a grounded field, prior, and social evaluation function.

                                              The meaning function is derived from the EM field: variant v is compatible with a persona set p iff v's indexed properties are a subset of p's properties.

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

                                                The carrier's literal listener on the converted game is uniform over compatible personae. This captures the exclusion structure (which personae are ruled out by each variant) but not the prior-weighted refinement — [Bur19]'s naive listener proper (eq. (18)) is Bayesian over the speaker's production rule. The quantitative predictions (≈69% -in' for cool guy at α = α′ = 6, eq. (15); style shifting) use the belief-based speaker Sk of §3 (Burnett eq. (13): P_S(m|π) ∝ L₀(π|m)^α), which incorporates the context-specific prior into the meaning function to recover Bayesian conditioning. We construct an SMG from the study's types and prove structural properties.

                                                Obama's social value function μ at the barbecue ([Bur19], Table 6, p. 438).

                                                Cool guy ({competent, friendly}) is most valued (μ = 2); asshole ({incompetent, aloof}) is least (μ = 0). The μ function reflects what the speaker (Obama) most wants the listener to infer about him in this context.

                                                Equations
                                                Instances For

                                                  The (ING) Social Meaning Game for the casual context ([Bur19], Defs. 4.1/4.3 + Table 2 + Table 6).

                                                  This connects the study's types to the §8 game structure, exercising SocialMeaningGame and toInterpGame.

                                                  Equations
                                                  Instances For

                                                    The SMG meaning is grounded in the Eckert–Montague intersection lift — connecting the game structure to the compositional semantics layer via ingMeaning_eq_emMeaningMI.

                                                    The literal listener excludes stern leader after hearing -in' (incompatible: stern leader = {competent, aloof} shares no property with [-in'] = {incompetent, friendly}).

                                                    The literal listener excludes doofus after hearing -ing (incompatible: doofus = {incompetent, friendly} shares no property with [-ing] = {competent, aloof}).

                                                    The literal listener assigns equal probability (1/3) to all compatible personae — uniform over ⟦v⟧: since 3 personae are compatible with each variant, each gets 1/3.

                                                    This is the structural content of the meaning function: each variant partitions personae into compatible (1/3) and incompatible (0).