Documentation

Linglib.Theories.Syntax.Minimalist.Copula

Copula Theory: HAVE, BE, and Delayed Gratification #

@cite{myler-2016} @cite{freeze-1992} @cite{kayne-1993} @cite{wood-2015}

@cite{myler-2016} proposes that the copula verb (v) is a semantically vacuous light verb: ⟦v⟧ = λx.x. The PF realization of v is determined by Vocabulary Insertion sensitive to the syntactic environment:

This captures HAVE = BE + transitivity: have is the spell-out of v in the environment of transitive, external-argument-introducing Voice.

Delayed Gratification #

A DP can satisfy a θ-role introduced by a head X without being merged in Spec,XP. Instead, the DP merges higher in the structure and the θ-role percolates up via λ-abstraction until it finds its argument. This is delayed gratification — distinct from both raising (which involves a syntactic copy/trace in the lower position) and control (which involves PRO). In delayed gratification, there is no syntactic representation of the argument in the lower position at all.

This mechanism is the key to the "too-many-structures" puzzle: since possession relations originate inside DP and the possessor role can be gratified at any position in the clausal spine, the cross-linguistic variation in possession constructions reduces to where in the structure the possessor is first-merged.

FreeP (@cite{myler-2016} §4.1.1.3) #

The Free head introduces an experiencer θ-role in eventive have constructions. Like Voice, Free varies cross-linguistically in whether it requires a specifier:

The surface realization of the copula, determined by Vocabulary Insertion.

Instances For
    @[implicit_reducible]
    Equations
    Equations
    Instances For

      Vocabulary Insertion rule for the copula.

      @cite{myler-2016} (89):

      • v ⇔ HAVE / __Voice{D},φ
      • v ⇔ BE / elsewhere

      The conditioning environment is transitive Voice: Voice that introduces an external argument (has a DP specifier with φ-features). This is exactly the hasD property of VoiceHead.

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

        The copula VI rule as a proper VocabItem from the DM VI framework.

        Two items compete via the Elsewhere Condition:

        • HAVE: specificity 2 (checks hasD = true AND flavor ∉ {nonThematic, passive})
        • BE: specificity 0 (elsewhere — matches any context)

        vocabularyInsertSimple copulaVIRules voice agrees with copulaVI voice.

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

          The VocabItem formulation agrees with the direct copulaVI function: "have" is inserted iff copulaVI returns .have.

          HAVE in the environment of agentive Voice.

          HAVE in the environment of causer Voice.

          HAVE in the environment of experiencer Voice.

          BE in the environment of middle/expletive Voice.

          BE in the environment of non-thematic Voice (anticausative).

          BE in the environment of passive Voice.

          The VI rule is equivalent to: HAVE ↔ Voice is transitive (has external argument that is not a PF-only marker and not a passive).

          How a θ-role introduced by a head X is satisfied.

          @cite{myler-2016} §1.3 (69–73) distinguishes three mechanisms:

          • Instant: the DP merges in Spec,XP — the standard case. The DP is both the syntactic and semantic argument of X.
          • Delayed: the DP merges higher in the structure (e.g., Spec,YP where Y ≠ X). The θ-role of X percolates up via λ-abstraction and is eventually saturated. There is NO syntactic representation of the argument in the lower Spec,XP position.
          • Raising: the DP is base-generated in Spec,XP (or a copy/trace is left there) and moves to a higher position. Unlike delayed gratification, there IS a syntactic reflex in the lower position.

          Delayed gratification is the key to Myler's account: possession relations originate DP-internally (in Spec,nP for inalienable, in Spec,PossP for alienable), but the possessor can be gratified at any point in the clausal spine — Spec,PredP, Spec,ApplP, Spec,VoiceP. The variation in WHERE gratification occurs generates the cross-linguistic typology of possession constructions.

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

              Does this gratification type leave a syntactic reflex in the lower (base) position? This is the key difference between delayed gratification and raising/control.

              Equations
              Instances For

                Delayed gratification is unique: the argument moves up but leaves no syntactic reflex. This has consequences for agreement, scope, and intervention — none of the effects triggered by a DP in the lower position arise.

                Raising has BOTH syntactic reflex and upward movement. This is what distinguishes it from delayed gratification.

                The Free head introduces an experiencer θ-role above an embedded VoiceP.

                @cite{myler-2016} §4.1.1.3: Free is a functional head that merges above the embedded VoiceP inside have's complement. It is related to but distinct from Appl:

                • Like high Appl, Free relates an individual to an event
                • Unlike Appl, Free introduces specifically an experiencer role
                • Free's specifier behavior varies cross-linguistically (±D)

                The ±D parameter on Free generates the English/Spanish asymmetry:

                • Free_{} (English): no specifier → delayed gratification → experiencer HAVE exists, free datives do not
                • Free_{D} (Spanish): specifier required → instant gratification → free datives exist, eventive experiencer HAVE does not

                @cite{myler-2016} table (35): | Free heads | Phenomena | Languages | |-------------|-------------------------------------|--------------| | Free_{} | Eventive Exp HAVE, No free datives | English | | Free_{D} | No Eventive Exp HAVE, Free datives | Spanish | | Both | Both | None known | | None | Neither | None known |

                • hasD : Bool

                  Does Free require a specifier (DP)?

                Instances For
                  def Minimalist.instDecidableEqFreeHead.decEq (x✝ x✝¹ : FreeHead) :
                  Decidable (x✝ = x✝¹)
                  Equations
                  Instances For
                    def Minimalist.instReprFreeHead.repr :
                    FreeHeadStd.Format
                    Equations
                    • Minimalist.instReprFreeHead.repr x✝ prec✝ = Std.Format.bracket "{ " (Std.Format.nil ++ Std.Format.text "hasD" ++ Std.Format.text " := " ++ (Std.Format.nest 8 (repr x✝.hasD)).group) " }"
                    Instances For

                      English-type Free: cannot take a specifier. The experiencer role introduced by Free must be gratified higher, via delayed gratification to Spec,VoiceP → yields experiencer HAVE.

                      Equations
                      Instances For

                        Spanish-type Free: must take a specifier. The experiencer role is gratified instantly in Spec,FreeP → yields free datives. But delayed gratification to Spec,VoiceP is blocked (the role is already consumed) → no eventive experiencer HAVE.

                        Equations
                        Instances For

                          Does this Free head allow delayed gratification of its experiencer role to a higher position?

                          If Free has a specifier ({D}), the role is gratified instantly and cannot percolate further. If Free lacks a specifier ({}), the role remains unsaturated and percolates up.

                          Equations
                          Instances For

                            Does this Free head yield free datives? Free datives arise when the experiencer merges in Spec,FreeP.

                            Equations
                            Instances For

                              Does this Free head yield eventive experiencer HAVE? Eventive experiencer HAVE arises when the experiencer role percolates to Spec,VoiceP via delayed gratification.

                              Derived from allowsDelayedGratification: experiencer HAVE exists iff Free's θ-role can percolate (not consumed locally).

                              Equations
                              Instances For

                                English: eventive experiencer HAVE, no free datives.

                                Spanish: free datives, no eventive experiencer HAVE.

                                The complementarity: for any Free head, exactly one of eventive experiencer HAVE and free datives is available.

                                The complement of have (= the complement of v_BE).

                                @cite{myler-2016} table (100): the interpretation of a HAVE sentence depends on the interaction between the complement type and the Voice alloseme selected. The complement types are:

                                • possessedDP: a DP containing a possession relation (relational noun or Poss head). Voice is expletive → relational HAVE.
                                • eventDP: a DP denoting an event (light-verb HAVE: "had a bath"). Voice assigns agent/holder.
                                • saturatedEventiveVoiceP: a full VoiceP with an agent and event (engineer HAVE: "had John bathe the dog"). Voice selects engineer alloseme.
                                • stativeSC: a stative small clause — PredP, AP, PP (causer HAVE: "had me angry"; locative HAVE: "has nests in it"). Voice selects holder/causer.
                                • freeP: a FreeP embedding a VoiceP (experiencer HAVE: "had Johnny run off on us"). Voice is expletive; Free introduces the experiencer role.
                                • modalBase: a DP/set of worlds (modal HAVE: "has to leave"). Semantics is world-containment, not individual-event.
                                Instances For
                                  @[implicit_reducible]
                                  Equations
                                  Equations
                                  • One or more equations did not get rendered due to their size.
                                  Instances For
                                    @[implicit_reducible]
                                    Equations
                                    • One or more equations did not get rendered due to their size.

                                    Is the complement a stative predication (small clause)? This is the condition for the holder/causer Voice alloseme: Voice assigns a holder role when it combines with a stative predicate (AP, PP, PredP).

                                    Crucially, possessedDP and modalBase are NOT stative predicates from Voice's perspective — the possession relation originates DP-internally and Voice is vacuous (expletive).

                                    Equations
                                    Instances For

                                      Is the complement a saturated eventive VoiceP? This is the most specific environment: a full clause with its own agent and event, triggering the engineer alloseme.

                                      Equations
                                      Instances For

                                        Is the complement an eventive DP (not a full VoiceP)? Event-denoting DPs (light-verb HAVE: "had a bath") trigger agentive Voice, unlike FreeP (which triggers expletive Voice).

                                        Equations
                                        Instances For

                                          The reading (interpretation) of a HAVE sentence.

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

                                              @cite{myler-2016} table (100): Which Voice alloseme is selected given a particular complement type.

                                              Derived from complement properties via VoiceAlloseme.fromComplement (Allosemy.lean), with one extension: event-denoting DPs (light-verb HAVE) trigger the agent alloseme, which fromComplement does not cover (it only distinguishes saturated VoiceP from stative from elsewhere).

                                              The cascade:

                                              1. Saturated eventive VoiceP → engineer (most specific)
                                              2. Event-denoting DP → agent (eventive but not a VoiceP)
                                              3. Stative SC → holder (stative predicate)
                                              4. Elsewhere (possessedDP, FreeP, modalBase) → expletive (Voice is vacuous)
                                              Equations
                                              • One or more equations did not get rendered due to their size.
                                              Instances For

                                                The non-eventDP cases agree with VoiceAlloseme.fromComplement: when the complement is not an event-denoting DP, the alloseme can be derived purely from the IsSaturatedEventiveVoiceP and IsStativePredicate properties — which is exactly what fromComplement does.

                                                Agent + stative SC = * : agentive Voice requires a dynamic event, but a stative SC does not provide one.

                                                Engineer is ONLY available with saturated eventive VoiceP. All other complement types yield a different alloseme.

                                                When Voice is expletive (relational, experiencer, modal HAVE), the meaning comes entirely from the complement — Voice contributes nothing. This is the "meaning = complement's meaning" generalization.

                                                The complete theta-role prediction chain for HAVE sentences:

                                                HaveComplement → VoiceAlloseme → VoiceFlavor → Option ThetaRole
                                                

                                                This composes three independently motivated mappings:

                                                1. Complement type determines Voice alloseme (§5, table (100))
                                                2. Alloseme maps to syntactic VoiceFlavor (Allosemy.lean)
                                                3. VoiceFlavor.thetaRole determines theta role (Voice.lean)

                                                The result: each HAVE reading predicts a specific external θ-role (or none, for expletive Voice).

                                                Equations
                                                Instances For

                                                  Engineer HAVE assigns agent (to the "engineer" orchestrator).

                                                  Causer HAVE assigns experiencer (holder alloseme → experiencer flavor).

                                                  Relational HAVE assigns no theta role (expletive Voice).

                                                  Experiencer HAVE (FreeP) assigns no theta role from Voice (the experiencer role comes from Free, not Voice).

                                                  Modal HAVE assigns no theta role.

                                                  Whether a nominal (DP complement of HAVE) can undergo delayed gratification — i.e., can its possessor role percolate up to be gratified at Spec,VoiceP?

                                                  @cite{myler-2016} §4.1.2.1 observes that complex event nominals (CENs) resist delayed gratification and thus cannot appear in relational have constructions:

                                                    • John had a/the destruction of the city.
                                                    • The city had a/the destruction.

                                                  The reason: CENs contain a v head (verbal substructure), and v forces instant gratification — the possessor must be realized DP-internally. In contrast, simplex event nominals (SENs) and relational nouns lack v, so their possessor role can percolate.

                                                  Derived from v allosemy (Allosemy.lean): VAlloseme.introducesEvent is the independently motivated property that determines whether v contributes an event variable. When v is eventive, the DP has verbal substructure that blocks delayed gratification. When v is zero (identity), delayed gratification is available.

                                                  Equations
                                                  Instances For

                                                    Relational nouns (v = zero, no event variable) allow delayed gratification.

                                                    CENs (v = eventive, introduces event variable) block delayed gratification.

                                                    The bridge: delayed gratification is blocked iff v introduces an event. This derives the CEN restriction from the allosemy framework rather than stipulating it as a separate parameter.