Documentation

Linglib.Phenomena.Modality.ComparePosition

@cite{hacquard-2010} @cite{cinque-1999} @cite{hacquard-2006}Two approaches to the correlation between modal position (high vs low in the clause) and modal flavor (epistemic vs root/circumstantial):

@cite{cinque-1999}: Cartographic Stipulation #

Cinque proposes a universal hierarchy of functional projections with dedicated heads for each modal flavor:

Mod_epistemic > Mod_irrealis >... > Mod_root > Mod_ability

Each modal type occupies a fixed position in the clause. The position–flavor correlation is STIPULATED: epistemic modals are high because there is an Epistemic Modality Phrase above TP; root modals are low because there is an Ability/Root Modality Phrase below AspP.

@cite{hacquard-2006}: Content Licensing Derivation #

Hacquard derives the same correlation from a single principle: epistemic modal bases require a contentful event (one with propositional content). VP events lack content; speech act and attitude events have content. Since aspect binds modals to VP events, a modal below AspP cannot be epistemic. No dedicated functional heads are needed.

Comparison Axes #

Property	Cinque	Hacquard
Mechanism	Dedicated functional heads	Content licensing
Position correlation	Stipulated	Derived
Cross-linguistic prediction	Universal hierarchy	Universal content predicate
Flexibility	None (fixed heads)	Varies with event structure
Embedded contexts	Must re-stipulate	Falls out from attitude events

inductive Phenomena.Modality.ComparePosition.CinqueModHead :

A Cinque-style functional projection for modality. Each modal flavor occupies a dedicated syntactic position.

modEpistemic : CinqueModHead
Mod_epistemic: above TP (high)
modIrrealis : CinqueModHead
Mod_irrealis: above TP (high)
modRoot : CinqueModHead
Mod_root: below AspP (low)
modAbility : CinqueModHead
Mod_ability: below AspP (low)

Instances For

instance Phenomena.Modality.ComparePosition.instDecidableEqCinqueModHead :

DecidableEq CinqueModHead

Equations

Phenomena.Modality.ComparePosition.instDecidableEqCinqueModHead x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯

def Phenomena.Modality.ComparePosition.instBEqCinqueModHead.beq :

CinqueModHead → CinqueModHead → Bool

Equations

Phenomena.Modality.ComparePosition.instBEqCinqueModHead.beq x✝ y✝ = (x✝.ctorIdx == y✝.ctorIdx)

Instances For

instance Phenomena.Modality.ComparePosition.instBEqCinqueModHead :

BEq CinqueModHead

Equations

Phenomena.Modality.ComparePosition.instBEqCinqueModHead = { beq := Phenomena.Modality.ComparePosition.instBEqCinqueModHead.beq }

def Phenomena.Modality.ComparePosition.instReprCinqueModHead.repr :

CinqueModHead → ℕ → Std.Format

Equations

One or more equations did not get rendered due to their size.

Instances For

instance Phenomena.Modality.ComparePosition.instReprCinqueModHead :

Repr CinqueModHead

Equations

Phenomena.Modality.ComparePosition.instReprCinqueModHead = { reprPrec := Phenomena.Modality.ComparePosition.instReprCinqueModHead.repr }

def Phenomena.Modality.ComparePosition.CinqueModHead.isHigh :

CinqueModHead → Bool

Cinque's stipulated height for each head.

Equations

Instances For

theorem Phenomena.Modality.ComparePosition.cinque_high_epistemic (h : CinqueModHead) :

h.isHigh = (h.flavor == Core.Modality.ModalFlavor.epistemic)

In Cinque's system, high = epistemic and low = circumstantial. This correlation is true BY STIPULATION — it's built into the functional head inventory.

Hacquard derives the SAME correlation from content licensing, without stipulating dedicated functional heads.

High modals (above AspP) are bound to contentful events → epistemic available. Low modals (below AspP) are bound to VP events → no epistemic (content licensing blocks it).

The derivation:

VP events lack content (EventBinder.vpEvent.hasContent = false)
Low modals are bound to VP events (ModalPosition.belowAsp.defaultBinder =.vpEvent)
Epistemic requires content (EventBinder.canProjectEpistemic = hasContent)
Therefore: low modals cannot be epistemic

Both accounts predict the SAME position–flavor correlation for matrix clauses: high modals get epistemic, low modals get circumstantial. The accounts are extensionally equivalent here.

The difference is explanatory depth: Cinque stipulates the correlation; Hacquard derives it from content licensing.

The accounts make different predictions for EMBEDDED modals (under attitude verbs like "believe", "want"). Hacquard predicts that a high modal in an embedded clause binds to the ATTITUDE event (not the speech event), yielding the attitude holder's epistemic state. This falls out from the same content licensing principle: attitude events are contentful → epistemic R available.

Cinque must re-stipulate: the embedded clause has its own Mod_epistemic head, and the modal "knows" to occupy it. The question of WHY the modal gets the attitude holder's knowledge (not the speaker's) has no structural answer in the cartographic account. In Hacquard's account, it follows from the event binding chain:

[VP_att believe [ModP might [AspP PFV [VP be pregnant]]]]
            e₁ ←───────┘
CON(e₁) = Jane's beliefs → epistemic R for *might*

Hacquard's system handles embedded epistemic modals via ModalPosition.withAttitude: a high modal under an attitude verb binds to the attitude event (not the speech event). The attitude event is contentful, so epistemic R is available.

Cinque requires n functional heads (one per modal flavor at each height level). Hacquard requires 1 principle (content licensing) + 1 binary distinction (contentful vs contentless events).

We count: Cinque uses 4 heads (§1); Hacquard uses 3 event binders with 1 content predicate. The number of combinatoric PRIMITIVES is smaller in Hacquard's system.

Content licensing unifies the contentful/contentless distinction across ALL three binders. Cinque would need separate stipulations for each functional head.

@cite{narrog-2012} provides a third perspective on the position–flavor correlation, complementing Cinque (stipulation) and Hacquard (derivation). Narrog's empirical scope hierarchy from Japanese (@cite{narrog-2009a}) shows that epistemic categories empirically outscope deontic categories, which in turn outscope dynamic (ability) categories, which outscope voice and aspect.

This agrees with both Cinque and Hacquard on the basic prediction (epistemic is high, root is low) but adds a finer-grained picture with ~10 scope levels vs. Cinque's 2 (high/low) or Hacquard's 2 (above/below AspP). The detailed hierarchy is formalized in Phenomena.Modality.Studies.Narrog2012.

The key insight: Narrog's event-oriented / speaker-oriented cut aligns with Hacquard's AspP boundary. Below aspect: event-oriented (no propositional content, root modality only). Above aspect: speaker-oriented (propositional content, epistemic available). The diachronic claim (meanings climb the hierarchy) and the synchronic claim (content licensing) are two views of the same structural fact.

def Phenomena.Modality.ComparePosition.narrogOrientationToPosition :

Semantics.Modality.Narrog.SpeakerOrientationLevel → Semantics.Modality.EventRelativity.ModalPosition

Narrog's speaker-orientation maps to Hacquard's position: event-oriented = belowAsp, speaker-oriented/mood = aboveAsp.

Equations

Instances For

The Narrog → Hacquard bridge preserves the epistemic availability prediction: event-oriented categories (belowAsp) lack content and cannot project epistemic; speaker-oriented categories (aboveAsp) have content and can project epistemic.