Garassino & Jacob (2018) @cite{garassino-jacob-2018}

Polarity focus and non-canonical syntax in Italian, French and Spanish: Clitic left dislocation and sì che / sí que-constructions. In @cite{dimroth-sudhoff-2018}, pp. 227–254. DOI 10.1075/la.249.08gar.

This study file anchors three Fragment entries — Fragments.Italian.PolarityMarking.siChe, Fragments.Spanish.PolarityMarking.siQue, and Fragments.French.PolarityMarking.si — to the chapter that compares them as Romance polarity-focus (PF) realization strategies.

Three substantive contributions of the chapter

1. A typology of PF-marking strategies in Italian, French and Spanish that carves the inventory differently from the Features.InformationStructure.Strategy enum: G&J split lexical means (adverbs, affirmative particles, embedded-clause structures) from syntactic means (non-focal fronting, cleft family, clitic dislocation, sì che / sí que clefts) — see §1 below for the typed constructors.

2. A corpus result (Table 1, p. 239) showing complementary distribution of polar left-dislocation and sì che / sí que across the three languages in Direct Europarl: Italian uses LDs (6 occurrences), Spanish uses sí que (61 occurrences); the two strategies are essentially non-overlapping — see §2 below.

3. An explicit theoretical commitment to @cite{matic-nikolaeva-2018}'s "salient polarity" framework (footnote 13, p. 236), against the form- class encoding that the substrate's Strategy records. See §4 below; the formal statement of the non-equivalence lives in Phenomena/Polarity/Studies/MaticNikolaeva2018.lean.

Cross-references

• Phenomena/Polarity/Studies/MaticNikolaeva2018.lean — same volume, the framework G&J endorse against form-class encoding.
• Phenomena/Polarity/Studies/TurcoBraunDimroth2014.lean — earlier Germanic-vs-Romance production study G&J cite as their typological starting point (their footnote 5, p. 230); the italian_spanish_cognates theorem squatting at TBD2014:446 was factually a G&J claim and has been relocated here as siChe_siQue_cognates_at_encoding_level.
• Fragments/Italian/PolarityMarking.lean::siChe — the entry whose docstring previously hosted G&J's corpus claim as prose; the claim now lives below as Lean data.

§1 G&J's PF-marking strategy taxonomy

G&J carve the Romance inventory into 8 strategies, organized as lexical (§2.1: adverbs, particles, embedded clauses, elliptic embedding) vs syntactic (§2.2: fronting, clefts, dislocation) plus the sì che / sí que construction analyzed separately in §2.3.

This is not the same carving as Strategy. The substrate enum collapses several of these into .polarityReversal, following the Blühdorn/TBD2014 form-class tradition. G&J's taxonomy is finer-grained and cuts on syntactic structure rather than discourse function. Both encodings are kept; the divergence is the point.

inductive Phenomena.Polarity.Studies.GarassinoJacob2018.GJStrategy : Type

• lexicalAdverb : GJStrategy

  Lexical adverbs of truth/certainty/fact: It. davvero, Fr. vraiment, Sp. de veras, de hecho — @cite{garassino-jacob-2018} §2.1 ¶1, p. 230.

• affirmativePolarityParticle : GJStrategy

  Bare affirmative polarity particle: It. sì, Sp. sí, Fr. bien, Sp. ya — §2.1 ¶2, p. 230.

• embeddedClauseStructure : GJStrategy

  Embedded-clause structure with full matrix speech-act verb: It. ti assicuro che, Sp. te digo que, Fr. je t'assure que — §2.1 ¶3, p. 231 (examples 5–7).

• ellipticEmbedding : GJStrategy

  Elliptic embedding with adjective-only matrix: It. certo che, Sp. claro que, Fr. bien sûr que — §2.1 ¶4, p. 231 (examples 8–10).

• nonFocalFronting : GJStrategy

  Non-focal fronting (typical of Spanish, marginal in Italian): Algo debe saber — §2.2, p. 232 (examples 11–13).

• faireCleft : GJStrategy

  Cleft-family construction with stressed semantically-empty matrix verb: Fr. c'est ce qu'elle fait — §2.2, p. 233 (example 14).

• cliticDislocation : GJStrategy

  Clitic dislocation (LD / RD) into a PF-supporting structure — §2.2 ¶ on dislocation, examples 15–16; §2.4, examples 21–22.

• siQueClass : GJStrategy

  Sì che / sí que cleft-or-cleft-like construction — §2.3, examples 17–19.

@[implicit_reducible]

instance Phenomena.Polarity.Studies.GarassinoJacob2018.instDecidableEqGJStrategy : DecidableEq GJStrategy

Equations

• Phenomena.Polarity.Studies.GarassinoJacob2018.instDecidableEqGJStrategy x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯

@[implicit_reducible]

instance Phenomena.Polarity.Studies.GarassinoJacob2018.instReprGJStrategy : Repr GJStrategy

Equations

• Phenomena.Polarity.Studies.GarassinoJacob2018.instReprGJStrategy = { reprPrec := Phenomena.Polarity.Studies.GarassinoJacob2018.instReprGJStrategy.repr }

def Phenomena.Polarity.Studies.GarassinoJacob2018.instReprGJStrategy.repr : GJStrategy → ℕ → Std.Format

Equations

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

structure Phenomena.Polarity.Studies.GarassinoJacob2018.GJStrategyDatum : Type

A G&J taxonomy entry: paradigm form + which substrate strategy the Fragment file encodes for this entry (when one exists).

• gjStrategy : GJStrategy

  The G&J strategy class.

• exampleForm : String

  Representative Italian / French / Spanish example surface form.

• availability : Bool × Bool × Bool

  Whether this G&J strategy is available in Italian, French, Spanish per @cite{garassino-jacob-2018} §2 prose — ordered (it, fr, sp).

def Phenomena.Polarity.Studies.GarassinoJacob2018.davveroDatum : GJStrategyDatum

Equations

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

def Phenomena.Polarity.Studies.GarassinoJacob2018.siParticleDatum : GJStrategyDatum

Equations

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

def Phenomena.Polarity.Studies.GarassinoJacob2018.tiAssicuroDatum : GJStrategyDatum

Equations

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

def Phenomena.Polarity.Studies.GarassinoJacob2018.certoCheDatum : GJStrategyDatum

Equations

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

def Phenomena.Polarity.Studies.GarassinoJacob2018.algoDatum : GJStrategyDatum

Equations

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

def Phenomena.Polarity.Studies.GarassinoJacob2018.faireDatum : GJStrategyDatum

Equations

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

def Phenomena.Polarity.Studies.GarassinoJacob2018.dislocationDatum : GJStrategyDatum

Equations

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

def Phenomena.Polarity.Studies.GarassinoJacob2018.siCheDatum : GJStrategyDatum

Equations

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

def Phenomena.Polarity.Studies.GarassinoJacob2018.gjAllStrategies : List GJStrategyDatum

Equations

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

theorem Phenomena.Polarity.Studies.GarassinoJacob2018.lexical_means_pan_romance : davveroDatum.availability = (true, true, true) ∧ siParticleDatum.availability = (true, true, true) ∧ tiAssicuroDatum.availability = (true, true, true) ∧ certoCheDatum.availability = (true, true, true)

The four lexical strategies (§2.1) are available in all three languages; the chapter's lexical-means inventory is genuinely pan-Romance.

theorem Phenomena.Polarity.Studies.GarassinoJacob2018.syntactic_strategies_split_asymmetrically : algoDatum.availability = (true, false, true) ∧ faireDatum.availability = (false, true, false) ∧ siCheDatum.availability = (true, false, true)

The syntactic strategies (§2.2–2.3) split asymmetrically across the three languages, with each language licensing a different subset. This is the structural correlate of G&J's observation that "Spanish appears to be, so to speak, more Germanic than Romance" in PF marking (§4, p. 250). Non-focal fronting is marginally attested in Italian per G&J p. 232 — encoded as available (Bool collapses primary vs marginal).

theorem Phenomena.Polarity.Studies.GarassinoJacob2018.french_lacks_siQueClass : siCheDatum.availability.2.1 = false

French does not license the sì che / sí que class — G&J fn 11 (p. 234): "French si, unlike the corresponding forms in Spanish and Italian, is limited to dialogical contexts."

§2 Corpus result (Table 1, p. 239)

Distribution of polar CDs (clitic left-dislocations with PF reading) and sì che / sí que constructions in Direct Europarl. Italian sample 2.3M words, French 2.5M, Spanish 2.8M. Verified from the PDF.

structure Phenomena.Polarity.Studies.GarassinoJacob2018.GJCorpusDatum : Type

One row of @cite{garassino-jacob-2018} Table 1. siQueCount = none encodes the chapter's "NA" entry for French.

• language : String
• corpusSizeMillionWords : String
• polarLDCount : ℕ
• siQueCount : Option ℕ

@[implicit_reducible]

instance Phenomena.Polarity.Studies.GarassinoJacob2018.instDecidableEqGJCorpusDatum : DecidableEq GJCorpusDatum

Equations

• Phenomena.Polarity.Studies.GarassinoJacob2018.instDecidableEqGJCorpusDatum = Phenomena.Polarity.Studies.GarassinoJacob2018.instDecidableEqGJCorpusDatum.decEq

def Phenomena.Polarity.Studies.GarassinoJacob2018.instDecidableEqGJCorpusDatum.decEq (x✝ x✝¹ : GJCorpusDatum) : Decidable (x✝ = x✝¹)

Equations

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

def Phenomena.Polarity.Studies.GarassinoJacob2018.instReprGJCorpusDatum.repr : GJCorpusDatum → ℕ → Std.Format

Equations

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

@[implicit_reducible]

instance Phenomena.Polarity.Studies.GarassinoJacob2018.instReprGJCorpusDatum : Repr GJCorpusDatum

Equations

• Phenomena.Polarity.Studies.GarassinoJacob2018.instReprGJCorpusDatum = { reprPrec := Phenomena.Polarity.Studies.GarassinoJacob2018.instReprGJCorpusDatum.repr }

def Phenomena.Polarity.Studies.GarassinoJacob2018.italianRow : GJCorpusDatum

Equations

• Phenomena.Polarity.Studies.GarassinoJacob2018.italianRow = { language := "Italian", corpusSizeMillionWords := "2.3", polarLDCount := 6, siQueCount := some 0 }

def Phenomena.Polarity.Studies.GarassinoJacob2018.frenchRow : GJCorpusDatum

Equations

• Phenomena.Polarity.Studies.GarassinoJacob2018.frenchRow = { language := "French", corpusSizeMillionWords := "2.5", polarLDCount := 4, siQueCount := none }

def Phenomena.Polarity.Studies.GarassinoJacob2018.spanishRow : GJCorpusDatum

Equations

• Phenomena.Polarity.Studies.GarassinoJacob2018.spanishRow = { language := "Spanish", corpusSizeMillionWords := "2.8", polarLDCount := 0, siQueCount := some 61 }

def Phenomena.Polarity.Studies.GarassinoJacob2018.gjTable1 : List GJCorpusDatum

Equations

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

theorem Phenomena.Polarity.Studies.GarassinoJacob2018.italian_spanish_complementary_distribution : italianRow.polarLDCount > 0 ∧ italianRow.siQueCount = some 0 ∧ spanishRow.polarLDCount = 0 ∧ spanishRow.siQueCount = some 61

The central corpus claim of @cite{garassino-jacob-2018} (§3 Conclusions, p. 250): Italian and Spanish exhibit complementary distribution for the two PF-marking strategies — Italian uses LDs (6 / 0), Spanish uses sí que (0 / 61).

§3 Cognacy at the substrate-encoding level

@cite{garassino-jacob-2018} §2.3 (p. 234) and the broader Romance literature (@cite{bernini-1995}, @cite{batllori-hernanz-2013}, @cite{poletto-zanuttini-2013}) treat sì che and sí que as cognate constructions: an affirmative-polarity particle followed by a complementizer introducing an embedded clause carrying the asserted proposition.

The substrate-level cognacy theorem below records that the two Fragment entries agree on every substrate field. Crucially, this is cognacy at the encoding level only. §2's corpus theorem above shows the two constructions diverge in actual usage frequency: Spanish sí que appears 61× in Direct Europarl, Italian sì che zero times. Both "the constructions are formally cognate" and "they are pragmatically divergent" can hold, and G&J's chapter is precisely the documentation that they do.

(Relocated from TurcoBraunDimroth2014.lean:446; the cognacy claim chronologically anchors on Bernini 1995 / Batllori-Hernanz 2013 / G&J 2018, not TBD 2014.)

theorem Phenomena.Polarity.Studies.GarassinoJacob2018.siChe_siQue_cognates_at_encoding_level : Fragments.Italian.PolarityMarking.siChe.strategy = Fragments.Spanish.PolarityMarking.siQue.strategy ∧ Fragments.Italian.PolarityMarking.siChe.environments = Fragments.Spanish.PolarityMarking.siQue.environments

§4 G&J's framework position (footnote 13, p. 236)

G&J explicitly endorse @cite{matic-nikolaeva-2018}'s rejection of the form-class encoding of polarity focus. Footnote 13 reads, verbatim:

This view is similar to the one presented by Matić & Nikolaeva (this volume), according to whom PF (or salient polarity as they prefer to name this specific type of emphasis) is not directly encoded by certain linguistic forms in a given language but can be pragmatically conveyed by different structures under appropriate (contextual) conditions.

This is in tension with the substrate's Strategy enum, which assigns each entry a single .polarityReversal / .particle / .verumFocus strategy class as if it were a fixed form-meaning property. The contradiction is recorded at the substrate's def-site (Features/InformationStructure.lean::Strategy docstring) and formalized in Phenomena/Polarity/Studies/MaticNikolaeva2018.lean.

§5 Surface-class lumping caveat (footnote 11, p. 234)

G&J fn 11 distinguishes French si from the sì che / sí que class:

Si exists in French as an affirmative particle expressing positive polarity in a contrastive way; … However, French si, unlike the corresponding forms in Spanish and Italian, is limited to dialogical contexts, where it is used to answer a preceding opposite turn.

So the cross-linguistic lumping under Strategy.polarityReversal of French si, Italian sì, and Spanish sí records a shared functional role only — the surface category differs (French = response particle, Italian + Spanish = clause-initial cleft-like construction). The substrate flag in Strategy docstring records this caveat. The french_lacks_siQueClass theorem in §1 above proves the corresponding data-level claim.