Documentation

Linglib.Phenomena.WordOrder.Studies.Westergaard2009

Westergaard (2009): Micro-Cues, Information Structure, and Economy #

@cite{westergaard-2009}

Marit Westergaard. The Acquisition of Word Order: Micro-Cues, Information Structure, and Economy. Linguistik Aktuell/Linguistics Today 145. John Benjamins, 2009.

Core Claim #

V2 is not a single parameter. It decomposes into micro-parameters: one per clause-type head in a split-CP (ForceP) domain. Each micro-parameter is independently settable to + (verb movement to that head) or − (no verb movement). Different Germanic languages and dialects are characterized by different profiles of + and − across these heads.

The book distinguishes two levels:

Formalization #

  1. ForceHead: the seven clause-type heads (theory layer)
  2. V2Profile: Profile ForceHead (theory layer; set of active heads)
  3. Language profiles: per-language Fragment files write set literals
  4. MicroCue: syntactic templates from Ch. 3 §4
  5. Bridge theorems to SAI data, V2 data, and GermanicV2
  6. Information Structure: [±FOC] conditioning of "optional" V2

Adjacent literature #

The Split-ForceP Model #

@cite{westergaard-2009} splits @cite{rizzi-1997}'s ForceP into clause-type-specific projections. All seven heads are in the CP domain (above FinP). Crucially, the distinctions among Decl°, Int°, Pol°, Excl°, Imp° are finer than @cite{rizzi-1997}'s inventory — they are all "flavors of Force" that the existing Cat enum does not distinguish.

Fin° and Wh° do correspond to existing Cat heads (.Fin and .C respectively), but the five Force-level heads (Decl°, Int°, Pol°, Excl°, Imp°) are all at the Force level. Note: @cite{westergaard-2009}'s Pol° is a CP-domain head for yes/no-questions (the verb-fronting target; y/n questions surface as V1, with Spec-PolP either empty or hosting a covert Q-operator depending on analysis — @cite{roberts-1993}, Rizzi 1996 posit a Q-operator satisfying a wh-criterion; the V1 surface order is what .Pol records, theory-neutrally). NOT @cite{laka-1990}'s ΣP (which is Cat.Pol in linglib at F-value 2).

Shared types for describing V2 word order variation.

Clause types relevant to V2 variation.

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

      V2 status of a clause type in a given language/dialect.

      • obligatory : V2Status

        V2 is obligatory

      • impossible : V2Status

        V2 is impossible (verb stays low or appears finally)

      • optional : V2Status

        V2 alternates with non-V2, conditioned by other factors

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

          A single V2 observation: what happens in a given clause type.

          • sentence : String
          • language : String
          • clauseType : V2ClauseType
          • v2Status : V2Status
          • description : String
          • citation : String
          Instances For
            def Westergaard2009.instReprV2Datum.repr :
            V2DatumStd.Format
            Equations
            • One or more equations did not get rendered due to their size.
            Instances For

              @cite{westergaard-2009}'s Table 3.1 enumerates V2 micro-parameter settings for six Germanic varieties (Standard Norwegian, Standard English, Nordmøre Norwegian, Belfast English, German, Danish). Each row is a V2Profile set literal in the corresponding Fragment file; the theorem below pins down all six rows simultaneously, so flipping a single field in a Fragment breaks one conjunct. -- UNVERIFIED: page reference for Table 3.1 (cited as p. 41 in earlier -- drafts of this file) has not been independently checked against -- the published Benjamins edition.

              Ch. 3 §4 introduces the cues — the syntactic templates in the input that trigger each micro-parameter. A micro-cue is a piece of I-language structure that children produce on exposure to the relevant input. Ch. 10 §3 (34)–(37) gives the final formulations.

              The distinction from Table 3.1: micro-parameters are the *grammar's*
              settings; micro-cues are the *observable evidence* in the input
              that leads children to set each parameter.
              
              Final micro-cue formulations (Ch. 10 (34)–(37)):
              - (34) DeclP[XP Decl°[+V] ...] — V2 in declaratives
              - (35) IntP[wh Int°[+V] ...] — V2 in wh-questions (wh-phrase in SpecIntP)
              - (36) IntP[wh[Int°] ...] — non-V2 in wh-questions (wh-head *in* Int°)
              - (37) TopP[DP[−FOC] Top° IntP[wh[Int°] ...]] — given subject → non-V2
              
              NOTE: (36) and (37) are the two key innovations. (36) captures the
              wh-head/phrase distinction: monosyllabic wh-words are heads that
              occupy Int° directly, blocking verb movement. (37) captures the
              TopP/[±FOC] mechanism: given subjects ([−FOC]) move to SpecTopP,
              which is the structural basis for the information-structure
              conditioning of V2 in § 10 below. 
              

              A micro-cue: a syntactic template that serves as evidence for a particular micro-parameter setting in acquisition.

              • Which head this cue is evidence for

              • template : String

                The syntactic template (schematic notation)

              • description : String

                Description of the cue

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

                  Cue for V2 in wh-questions.

                  Equations
                  Instances For

                    Cue for V2 in declaratives.

                    Equations
                    Instances For

                      Cue for V2 in exclamatives.

                      Equations
                      Instances For

                        Cue for V2 in embedded questions.

                        Equations
                        Instances For

                          Cue for non-V2 in exclamatives.

                          Equations
                          Instances For

                            Cue for non-V2 in embedded questions.

                            Equations
                            Instances For

                              Cue for V2 in yes/no-questions.

                              Equations
                              Instances For

                                Cue for V2 in imperatives.

                                Equations
                                Instances For

                                  Cue for wh-head-in-Int° (non-V2 in wh-questions). Ch. 10 (36): IntP[wh[Int°] ...] — the monosyllabic wh-word occupies Int° itself, blocking verb movement to that position.

                                  Equations
                                  Instances For

                                    Cue for given-subject-blocking-V2 in Tromsø monosyllabic wh-questions. Ch. 10 (37): TopP[DP[−FOC] Top° IntP[wh[Int°] ...]] — the [−FOC] subject moves to SpecTopP, leaving Int° empty (verb stays low → non-V2). The TopP/[±FOC] mechanism is what derives the information-structure conditioning of "optional" V2 in § 10.

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

                                      A cue is expressed in a language iff its target head is in the language's V2Profile. Children exposed to an expressed cue will set the corresponding parameter to +.

                                      Equations
                                      Instances For

                                        V2 observations from across the book, organized by language.

                                        Non-subject-initial declaratives: V2 obligatory.

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

                                          Yes/no-questions: V2 obligatory.

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

                                            Wh-questions with long (polysyllabic) wh-phrases: V2 obligatory.

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

                                              Wh-questions with short (monosyllabic) wh-words: V2 optional, conditioned by information structure.

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

                                                Exclamatives: non-V2 obligatory.

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

                                                  Embedded declaratives: non-V2 (mostly).

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

                                                    Standard English: no V2 in declaratives (SVO base order).

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

                                                      Standard English: V2 in wh-questions (via SAI).

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

                                                        Belfast English: V2 in embedded questions too.

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

                                                          Danish: V2 in exclamatives (unlike Norwegian and English).

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

                                                            German root declaratives: V2 obligatory.

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

                                                              German embedded clauses with complementizer: verb-final (no V2). @cite{westergaard-2009}'s analysis: the verb raises to Fin° (hence +Fin° in Table 3.1) but not to C, surfacing clause-finally because Westergaard tacitly assumes a head-final FinP for German embedded clauses (Vikner-style V-to-I where the I-position itself is final). Alternative analyses (Haider 2010; @cite{harizanov-gribanova-2019}) derive the same surface order without V-to-Fin, leaving the verb in its base position. The codebase records the Westergaard +Fin° side; see HarizanovGribanova2019.lean for the formal contrast.

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

                                                                Standard Norwegian and Standard English differ only on Decl°. Captures the classic observation that English lost V2 in declaratives but retained it in questions.

                                                                All six languages in @cite{westergaard-2009} Table 3.1 agree on +Pol° (verb-fronting / V1 in yes/no-questions). NOT a Germanic universal beyond this sample: Yiddish embedded y/n questions with tsi and certain colloquial registers complicate the picture, and on some analyses Pol° is epiphenomenal on Int° rather than an independent micro-parameter.

                                                                Ch. 7 argues that monosyllabic wh-words are syntactic heads (X°) while polysyllabic wh-constituents are phrases (XP). When a wh-head occupies Int°, it blocks verb movement, making non-V2 possible. When a wh-phrase is in SpecIntP, Int° is free for the verb → V2 obligatory.

                                                                Tromsø Norwegian wh-words:
                                                                - Monosyllabic (heads): *ka* 'what' (1σ), *kem* 'who' (1σ),
                                                                  *kor* 'where' (1σ)
                                                                - Polysyllabic (phrases): *korfor* 'why' (2σ), *korsen* 'how' (2σ),
                                                                  *katti* 'when' (2σ) 
                                                                
                                                                def Westergaard2009.tromsøWhWords :
                                                                List (String × String × )

                                                                Tromsø wh-word data: (form, gloss, syllable count).

                                                                Equations
                                                                • Westergaard2009.tromsøWhWords = [("ka", "what", 1), ("kem", "who", 1), ("kor", "where", 1), ("korfor", "why", 2), ("korsen", "how", 2), ("katti", "when", 2)]
                                                                Instances For

                                                                  All monosyllabic Tromsø wh-words classify as heads.

                                                                  All polysyllabic Tromsø wh-words classify as phrases.

                                                                  English SAI (from SubjectAuxInversion.lean) is the surface reflex of +Int° and +Pol° in the English V2 profile.

                                                                  English declaratives lack V2: Decl° is not in the profile.

                                                                  Norwegian yes/no-questions are obligatorily V2, consistent with +Pol°.

                                                                  German embedded clauses are verb-final (no V2), even though German has +Fin° (V-to-I). V2 = verb-to-C requires +Decl°/+Int° etc.; verb-final is consistent with −Wh° (no V-to-C in embedded contexts).

                                                                  Theories/Syntax/HPSG/Inversion.lean derives English matrix/embedded question word-order asymmetries from an [INV ±] feature on clauses, with matrixRequiresInvPlus and embeddedRequiresInvMinus constraints. @cite{westergaard-2009}'s English profile commits to +Int° (matrix wh) and +Pol° (matrix y/n), which are the V-to-C steps that surface as inversion. The two frameworks agree on the same surface contrast via different machinery; the bridge theorem makes the agreement visible.

                                                                  Westergaard and @cite{sag-wasow-bender-2003}-style HPSG agree on English matrix question inversion: Westergaard's V-to-C (.Int/.Pol ∈ stdEnglish) projects the same surface order that HPSG derives from [INV +]. Both frameworks could have committed otherwise (Westergaard could have ascribed −Int°; HPSG could have left matrix-question INV unconstrained), so the agreement is non-trivial.

                                                                  The same agreement on the embedded side: Westergaard records Standard English as −Wh° (no V-to-C in embedded questions); HPSG records embeddedRequiresInvMinus. Both frameworks predict subject-first surface order in embedded questions for Standard English; Belfast English breaks this on both sides (.Wh ∈ belfastEnglish per Westergaard; @cite{henry-1995}'s "I wonder could he come" per HPSG-style accounts).

                                                                  GermanicV2.lean proves that German V2 involves head-to-head movement of V to C, skipping T (HMC violation). This is the structural realization of +Decl° in the V2 profile: verb movement targets the Decl° head in the CP domain.

                                                                  German +Decl° is consistent with the V-to-C movement formalized in GermanicV2.lean.

                                                                  WALS classifies German as having "no dominant order" (Typology.lean). Westergaard's micro-parameters explain why: German has +Decl° (V2 in root declaratives) but also +Fin° (V-to-I in embedded clauses, yielding verb-final surface order due to SOV base). This split makes the "basic" order indeterminate — SVO on the surface in root clauses, SOV underlyingly and in embedded clauses.

                                                                  German's "no dominant order" classification in WALS is consistent with a profile that has BOTH +Decl° (V2 in roots → surface SVO) AND +Fin° (V-to-I in embedded → surface SOV).

                                                                  English is classified as SVO in WALS. Consistent with −Decl° (no verb movement in declaratives → surface SVO with SVO base) and −Fin° (no V-to-I in embedded → embedded order also SVO).

                                                                  In Tromsø wh-questions with monosyllabic wh-words, V2 vs. non-V2 correlates with the discourse status of the subject:

                                                                  - **[−FOC] / given subject** (pronoun) → non-V2 preferred.
                                                                    Subject moves to SpecTopP; verb stays low.
                                                                  - **[+FOC] / new subject** (full DP) → V2 preferred.
                                                                    Subject stays in SpecIP; verb moves to Top° to check [−FOC].
                                                                  
                                                                  The book *derives* this from TopP structure (pp. 46–47): given
                                                                  subjects carry [−FOC], which triggers movement to SpecTopP, leaving
                                                                  Int° empty (verb stays low). New subjects lack [−FOC], so they stay
                                                                  in SpecIP and the verb moves to Top°/Int° → V2. The [±FOC] feature
                                                                  already exists in `Features.lean` (`foc : Bool → FeatureVal`) but
                                                                  is not yet connected to an Agree-based derivation.
                                                                  
                                                                  TODO: Replace this stipulative pattern match with a derivation from
                                                                  [±FOC] feature checking on subjects + TopP Agree/movement. The
                                                                  current version captures the correct *empirical mapping* but does
                                                                  not explain *why* the mapping holds — the TopP mechanism does. 
                                                                  

                                                                  Preferred V2 status given subject givenness in Tromsø monosyllabic wh-questions.

                                                                  STIPULATIVE: pattern-matches on givenness directly. The book derives this from [±FOC]/TopP (see § 10 docstring). Focus marking belongs to a separate axis (Features.InformationStructure.Focus); a focus-driven V2 extension would parameterize over Focus α separately.

                                                                  Equations
                                                                  Instances For

                                                                    @cite{westergaard-2009}'s structural economy (p. 4):

                                                                    (9a) Only build as much structure as there is evidence for in the input.
                                                                    (9b) Only move elements as far as there is evidence for in the input.
                                                                    
                                                                    These principles constrain *children's grammars*: children build
                                                                    minimal structure, adding projections only when input evidence forces
                                                                    them.
                                                                    
                                                                    The corollary below: languages with fewer active micro-parameters
                                                                    require less structure to be built. Our own derivation, not a claim
                                                                    directly stated in the book. `Profile.activeCount` instances the
                                                                    polymorphic counter from `Typology/Profile.lean`.