Documentation

Linglib.Theories.Syntax.Minimalism.Formal.ExtendedProjection.Basic

@cite{chomsky-1970}'s [±V, ±N] categorial features, adopted by @cite{grimshaw-2005} for Extended Projections. Cross-classifies the four lexical categories:

  • V = [+V, -N], N = [-V, +N], A = [+V, +N], P = [-V, -N]

Functional categories inherit these from their lexical anchor.

For an alternative where [N] and [V] carry semantic content (referentiality and temporal predication), see CategorialFeatures.

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

          Compute @cite{chomsky-1970}'s [±V, ±N] features from Cat. Functional categories inherit features from their lexical anchor:

          • v, T, C inherit [+V, -N] from V
          • n, Num, Q, D inherit [-V, +N] from N
          Equations
          Instances For

            Grimshaw's F-value: the functional level within an extended projection.

            F-values are globally aligned across category families to capture the verbal–nominal parallelism.

            The nominal spine follows @cite{borer-2005}'s ordering: Q (classifier / individuation, CL#) is at F2, below Num (number / counting, #) at F3. This reflects the semantic composition order: individuation must precede counting (you can't count what hasn't been individuated). See Borer2005.lean for the formal argument.

            F-levelRoleVerbalNominal
            F0Lexical (content)VN
            F1Categorizerv, Voice, Appln (gender/class)
            F2SpecificationT, Neg, Asp, ModQ (classifier/CL#)
            F3Inner edgeFinNum (number/#)
            F4Discourse/refFocD (definiteness)
            F5TopicTop, Rel
            F6Clause/forceC, Force
            F7Speech actSA

            Verbal–nominal parallelism: The parallelism is robust at F0 (lexical anchors) and F1 (categorizers: v ↔ n). At F2–F3, the verbal and nominal spines are analogous but not isomorphic: T/Asp specify temporal properties while Q specifies individuation; Fin types the clause while Num types the nominal. The semantic functions differ, but both occupy the same structural zone.

            The verbal C-domain is internally ordered per @cite{rizzi-1997}: Fin(F3) < Foc(F4) < Top(F5) < C(F6).

            Equations
            Instances For

              Category consistency: two categories share [±V, ±N] features. This is the core constraint on Extended Projections — V and T are consistent (both [+V, -N]), but V and D are not.

              Equations
              Instances For
                def Minimalism.fMonotone (lower upper : Cat) :

                F-value monotonicity: F-values must not decrease going up the tree. In an EP, each head's F-value is ≥ the one below it.

                Equations
                Instances For

                  Perfect projection: same [±V, ±N] AND same F-value. E.g., two V heads (F0, [+V, -N]) are perfect projections of each other. Distinct from EP extension, where F-value increases.

                  Equations
                  Instances For

                    Is this category a lexical head (F0)? L-heads are content categories: V, N, A, P.

                    Equations
                    Instances For

                      Is this category a functional head (F1+)? F-heads are functional categories: v, D, T, C.

                      Equations
                      Instances For

                        The four lexical category families, each defining an EP domain. All categories in an EP must belong to the same family.

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

                            Map a category to its family. This determines which EP it can participate.

                            Equations
                            Instances For

                              @cite{panagiotidis-2015} categorial features: [N] and [V] as substantive, LF-interpretable features with semantic content.

                              • [N] = sortal perspective / referentiality (capacity to introduce a discourse referent, following @cite{longobardi-1994}, @cite{longobardi-2005}; §4.3 p84)
                              • [V] = temporal perspective / eventivity (capacity to anchor to time/events; §4.3 p85)

                              This contrasts with @cite{chomsky-1970}'s [±V, ±N] diacritics (see CatFeatures): Chomsky's features are arbitrary binary cross-classifiers, while Panagiotidis's are grounded in semantic substance. The key empirical difference is the status of P: Chomsky treats P as actively bearing [-V, -N]; Panagiotidis treats P as the default categorizer lacking both [N] and [V] (§4.3).

                              Interpretability distinction (§5.8): On categorizers (v, n, a), these features are interpretable — they provide the LF-legible interpretive perspective (sortal or temporal). On higher functional heads (T, C, D, etc.), these features are uninterpretable copies that serve only for Agree/selection. This formalization does not encode the interpretable/uninterpretable distinction but records which features are present, which suffices for EP consistency.

                              Despite the conceptual difference, the two systems produce the same four equivalence classes over categories (see chomsky_panagiotidis_agree).

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

                                    Map a category to @cite{panagiotidis-2015}'s categorial features.

                                    Categorizers (n, v, a) bear the substantive features; functional heads in the same EP inherit them (just as in Grimshaw's consistency requirement). P and its extended projection bear neither feature — P is the default case.

                                    Equations
                                    Instances For

                                      Consistency under Panagiotidis's system: two categories share [N]/[V] features.

                                      Equations
                                      Instances For

                                        Chomsky's [±V, ±N] and Panagiotidis's [N]/[V] produce the same equivalence classes over all categories. They agree on which pairs are EP-consistent.

                                        The conceptual difference — P as [-V, -N] vs. P as default — is invisible to the consistency check: both systems group P only with itself.

                                        An Extended Projection: a sequence of categories forming a projection chain with category consistency and F-value monotonicity.

                                        The spine lists categories from lowest (lexical head) to highest (functional).

                                        • Root syntactic object of the EP

                                        • spine : List Cat

                                          Categories along the spine, from lexical head (F0) upward

                                        • catConsistent : Bool

                                          All spine categories share [±V, ±N] features

                                        • fMonotoneChain : Bool

                                          F-values are non-decreasing along the spine

                                        Instances For

                                          Check that a list of categories is category-consistent (all share the same [±V, ±N] features).

                                          Equations
                                          Instances For

                                            Check that a list of categories has monotonically non-decreasing F-values.

                                            Equations
                                            Instances For

                                              Compute the EP spine from a syntactic object by collecting categories along the head-projection chain. Returns pairs of (SO, Cat) from the deepest lexical head up to the root.

                                              Build an ExtendedProjection from a syntactic object.

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

                                                Is this EP well-formed? (category consistent AND F-monotone)

                                                Equations
                                                Instances For

                                                  The family of an EP (determined by any element of the spine).

                                                  Equations
                                                  Instances For

                                                    The lexical anchor of an EP (the F0 head at the bottom).

                                                    Equations
                                                    Instances For

                                                      The highest functional head in the EP.

                                                      Equations
                                                      Instances For

                                                        The verbal chain V → v → T → C is category-consistent: all share [+V, -N] features.

                                                        The nominal chain N → n → Q → Num → D is category-consistent: all have [-V, +N] features. Q (CL#, individuation) is below Num (#, counting) per @cite{borer-2005}.

                                                        F-values increase monotonically along the verbal chain: V(0) ≤ v(1) ≤ T(2) ≤ C(3).

                                                        F-values increase along the nominal chain: N(0) ≤ n(1) ≤ Q(2) ≤ Num(3) ≤ D(4). Q (individuation) is below Num (counting) per @cite{borer-2005}.

                                                        V and N are NOT category-consistent (different [±V, ±N]).

                                                        V and D are NOT category-consistent (verbal ≠ nominal).

                                                        F0 is exactly the lexical heads.

                                                        Categories in the same family are category-consistent.

                                                        The full verbal EP spine [V, v, T, C] is category-consistent.

                                                        The full verbal EP spine [V, v, T, C] is F-monotone.

                                                        The full nominal EP spine [N, n, Q, Num, D] is category-consistent.

                                                        The full nominal EP spine [N, n, Q, Num, D] is F-monotone.

                                                        F0 categories correspond to BarLevel.zero (head), F1+ categories correspond to intermediate/maximal projections. This connects Grimshaw's F-level to X-bar bar levels.

                                                        The verbal and nominal spines are parallel at F0–F1: V ↔ N (lexical), v ↔ n (categorizer).

                                                        At F2–F3 the spines diverge: T (temporal specification, F2) pairs with Q (individuation, F2), while Fin (clause-typing, F3) pairs with Num (number, F3). These are structural analogs occupying the same EP zone, but their semantic functions differ. See borer_ordering_unique in Borer2005.lean for the formal argument that Q must be below Num.

                                                        Is this category a categorizer? Categorizers bear substantive, interpretable [N]/[V] features and combine with acategorial roots to yield categorized items.

                                                        Important: Panagiotidis (§4.5) argues categorizers are NOT functional heads — they are the only true lexical heads (roots being acategorial). Our F-value system (from @cite{grimshaw-2005}) places them at F1, which makes isFHead return true for categorizers. This reflects Grimshaw's architectural classification, not Panagiotidis's ontological claim about their nature.

                                                        Equations
                                                        Instances For

                                                          All three categorizers are at F1 in @cite{grimshaw-2005}'s F-value system. @cite{panagiotidis-2015} predicts this parallelism; the F1 encoding is Grimshaw's.

                                                          The three categorizers are parallel: all at the same F-level.

                                                          The adjectival categorizer is in the adjectival family (parallel to v→verbal, n→nominal).

                                                          The adpositional chain P → Place → Path is category-consistent: all share [-V, -N] features. @cite{dendikken-2010}: PlaceP (locational) and PathP (directional) are functional projections above P.

                                                          F-values increase along the adpositional chain: P(0) ≤ Place(1) ≤ Path(2). Parallel to V(0) ≤ v(1) ≤ T(2) in the verbal domain.

                                                          The adpositional EP spine [P, Place] is well-formed (locational PP).

                                                          The adpositional EP spine [P, Place, Path] is well-formed (directional PP). @cite{dendikken-2010}: directional PPs project PathP above PlaceP.

                                                          The verbal EP spine with @cite{rizzi-1997}'s split-CP layer: V → v → T → Fin → Foc → Top → C. Fin is the boundary between IP and CP; Foc and Top are discourse-related projections between Fin and C (= Force).

                                                          Equations
                                                          Instances For

                                                            The split-CP spine is category-consistent: all heads share [+V, -N].

                                                            The split-CP spine is F-monotone: 0 ≤ 1 ≤ 2 ≤ 3 ≤ 4 ≤ 5 ≤ 6. This is the key payoff of the fValue fix — before the fix, Fin/Foc/Top/C all had fValue 3, so any permutation would trivially pass.

                                                            The reverse split-CP ordering [Top, Foc, Fin] is NOT monotone. This correctly rules out pathological orderings that the collapsed fValues (all = 3) would have accepted.

                                                            Verbal EP with NegP: V → v → Neg → T → Fin → C. Represents a clause with an IP-internal negation layer.

                                                            Equations
                                                            Instances For

                                                              The verbal EP with NegP is category-consistent.

                                                              The verbal EP with NegP is F-monotone: 0 ≤ 1 ≤ 2 ≤ 2 ≤ 3 ≤ 6.

                                                              Full Rizzi left periphery: V → v → T → Fin → Foc → Top → Force. Force is the clause-typing head when the CP is fully split.

                                                              Equations
                                                              Instances For

                                                                The full Rizzi left periphery is category-consistent.

                                                                The full Rizzi left periphery is F-monotone.

                                                                Force and C have the same fValue (they're the same position when unsplit).

                                                                Neg is in the specification domain (same F-level as T and Q).

                                                                The structural size of a clausal complement, determined by the highest functional head projected.

                                                                Complement size matters for tense Agree locality: a CP complement constitutes a phase boundary that blocks upward Agree for [uPAST], while a TP complement is transparent.

                                                                Also relevant for @cite{wurmbrand-2014}'s three-way infinitival classification (restructuring ≈ vP, propositional ≈ TP, full finite ≈ CP).

                                                                • highestHead : Cat

                                                                  The highest functional head in the complement

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

                                                                      The F-level of a complement (derived from fValue).

                                                                      Equations
                                                                      Instances For

                                                                        A complement is phase-sized (≥ CP) if its highest head is at or above the C level in the functional sequence.

                                                                        Equations
                                                                        Instances For

                                                                          A complement is transparent to tense Agree if it is smaller than a full CP — i.e., the highest head is below C in the fseq.

                                                                          @cite{egressy-2026}: TP complements (fValue 2) are transparent; CP complements (fValue 6) are opaque.

                                                                          Equations
                                                                          Instances For

                                                                            Standard complement sizes.

                                                                            Equations
                                                                            Instances For

                                                                              vP complements are transparent to tense Agree.

                                                                              TP complements are transparent to tense Agree.

                                                                              FinP complements are transparent to tense Agree.

                                                                              CP complements are opaque to tense Agree.

                                                                              ForceP complements are opaque to tense Agree.

                                                                              SAP complements are opaque to tense Agree.

                                                                              The seven clause-type heads in @cite{westergaard-2009}'s split-ForceP. Each represents a possible target for verb movement.

                                                                              @cite{westergaard-2009} splits @cite{rizzi-1997}'s ForceP into clause-type-specific projections: DeclP, IntP, PolP, ExclP, ImpP are all "flavors of Force" in the CP domain, while FinP and WhP handle embedded contexts. This decomposition allows V2 to be treated as multiple independent micro-parameters rather than a single macro-parameter.

                                                                              All seven heads are at or above FinP. The five root-clause heads (Decl, Int, Pol, Excl, Imp) are finer-grained than @cite{rizzi-1997}'s single Force head — they are all at the Force level (F6). Fin° corresponds to Cat.Fin (F3); Wh° is at the Force level (F6).

                                                                              Instances For

                                                                                All ForceHead values are in the verbal EP family.

                                                                                A V2 profile: for each clause-type head, whether verb movement to that head is required (+) or absent (−) in a given language/dialect.

                                                                                This is the formalization of @cite{westergaard-2009}'s micro-parameter model: V2 is not one parameter but seven independent ones.

                                                                                Instances For

                                                                                  Count how many heads trigger verb movement in a profile.

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

                                                                                    Whether two profiles differ on exactly one head.

                                                                                    Equations
                                                                                    Instances For

                                                                                      The syntactic status of a wh-element: head (X°) or phrase (XP).

                                                                                      @cite{westergaard-2009} argues that monosyllabic wh-words (ka 'what', kem 'who', kor 'where' in the Tromsø dialect) are syntactic heads, while polysyllabic wh-constituents (korfor 'why', korsen 'how', katti 'when') are phrases. This distinction is supported by similar patterns in Italian dialects (@cite{poletto-pollock-2004}).

                                                                                      The distinction matters for V2: when a wh-head occupies Int°, it blocks verb movement to that position, making non-V2 possible. When a wh-phrase is in SpecIntP, Int° is free for the verb → V2 obligatory.

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

                                                                                          Determine wh-element status from syllable count. Monosyllabic → head; polysyllabic → phrase.

                                                                                          Equations
                                                                                          Instances For

                                                                                            When a wh-element is a head in Int°, verb movement to Int° is blocked, making non-V2 possible. When it's a phrase in SpecIntP, Int° is free for the verb → V2 obligatory.

                                                                                            Equations
                                                                                            Instances For