Documentation

Linglib.Phenomena.Negation.Studies.Stakov2026Typology

Czech Three-Way Negation: Cross-Linguistic Typology #

@cite{repp-2013} @cite{romero-2024} @cite{stankova-2025} @cite{stankova-2025} @cite{gartner-gyuris-2017} @cite{simik-2024}

Bridges between the core three-way negation distinction (CzechThreeWayNeg.lean) and cross-linguistic frameworks: @cite{romero-2024} PQ typology, @cite{simik-2024} Czech PQ forms, @cite{stankova-2025} verb position / context sensitivity.

Also contains example data (CzechNegDatum), bias profiles, and corpus data.

Map Czech negation positions to Romero's PQ form typology.

Czech VSO (interrogative) word order = high negation = HiNQ. Czech SVO (declarative) word order = low negation = LoNQ. Inner and medial are both syntactically low (SVO) but differ in LF scope.

Staňková (2026 §2): "When a PQ has the SVO word order, I call it declarative; when a PQ has the VSO word order, I call it interrogative."

Equations
Instances For

    Map Czech negation positions to Romero's evidential bias strength.

    Inner negation: strong contextual evidence bias (□_ev > ¬, must be ¬p). Medial negation: weak contextual evidence bias (¬ > □_ev, needn't be p). Outer negation: no contextual evidence bias (FALSUM, not □_ev-based).

    Equations
    Instances For

      Whether the negation position requires obligatory focus.

      Only outer negation (FALSUM) is obligatorily focused — it targets discourse polarity and generates alternatives on whether p is or isn't in the CG (Staňková §3.2, §4).

      Equations
      Instances For

        Inner and medial both map to LoNQ (low negation = declarative word order). Czech low negation PQs are ambiguous between inner and medial readings, distinguished by polarity items and particles (Table 1).

        Verb position in Czech polar questions (@cite{stankova-2025} §2).

        Czech PQs use two word orders, determined by whether the finite verb moves to clause-initial position:

        • V1 (verb-initial): interrogative word order, verb+ne- in PolP
        • nonV1 (non-verb-initial): declarative word order, verb+ne- in TP

        Since Czech negation prefix ne- is inseparable from the verb, verb position directly determines the syntactic position of negation. This creates the surface syntax–negation interpretation mapping (@cite{zeijlstra-2004} Agree analysis, S&Š eqs. 11–12).

        • v1 : VerbPosition

          Verb-initial (interrogative) word order. The verb+ne- moves to PolP, within the scope of FALSUM[iNeg]. Only outer negation (FALSUM) is available (S&Š eq. 11).

        • nonV1 : VerbPosition

          Non-verb-initial (declarative) word order. The verb+ne- stays low in TP. Can be licensed by either FALSUM[iNeg] or Op¬[iNeg], but not both simultaneously (S&Š eq. 12).

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

            A Czech PQ negation example with its reading and Romero classification.

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

                Ex. (6a): Inner negation with NCI — grammatical.

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

                  Ex. (6b): Outer negation with NCI — ungrammatical.

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

                    Ex. (7a): Medial negation with PPI — grammatical.

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

                      Ex. (7b): Outer negation with PPI — grammatical.

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

                        Ex. (11): Outer negation with náhodou — grammatical.

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

                          Ex. (15a): Inner negation with fakt — grammatical.

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

                            Ex. (15d): Outer negation with fakt — ungrammatical.

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

                                Each Czech example carries its Romero PQ form automatically via NegPosition.toPQForm. These bridge theorems verify that the classification is consistent with the word order and bias conditions.

                                Romero's Tables 1–2 make predictions about which Czech examples should be felicitous. These bridge theorems verify that Staňková's Czech data is consistent with Romero's cross-linguistic generalizations.

                                Ex. 6b is a HiNQ. Romero Table 1 says HiNQ mandatorily conveys original bias for p. The example IS a HiNQ but is ungrammatical because the NCI žádný is incompatible with outer negation — a Czech-specific constraint layered on top of the Romero framework.

                                Ex. 7a (medial, nonV1, LoNQ) is felicitous in contexts with some evidence against p. Romero Table 2: LoNQ requires evidence against p. Staňková's medial negation carries weak evidential bias — weaker than inner, but still requires contextual evidence.

                                Ex. 6a (inner, nonV1, LoNQ) is felicitous only with strong contextual evidence against p. Romero Table 2: LoNQ requires evidence against p. Staňková adds the refinement that inner neg requires strong evidence (the inner/medial split within LoNQ).

                                Czech reveals a finer-grained distinction within Romero's LoNQ category.

                                Romero treats LoNQ as a single PQ form. Staňková shows that Czech SVO (= LoNQ) polar questions are actually ambiguous between inner and medial readings, distinguished by their diagnostic signatures (Table 1) and by the strength of their evidential bias.

                                This is a cross-linguistic prediction: languages with overt NCI/PPI contrasts and diagnostic particles may reveal the inner/medial split within LoNQ.

                                Czech polar question forms: the 2×2 grid of [Interrogative vs Declarative] × [Positive vs Negative] (@cite{simik-2024} §3.2, exx. 11–17).

                                Czech uses two independent formal strategies to express bias:

                                1. Negation: positive vs negative (ne- prefix)
                                2. Word order: interrogative (VSO, verb-initial) vs declarative (SVO)

                                This is finer-grained than Romero's PosQ/LoNQ/HiNQ, because Czech declarative PQs (DeclPQ) are a separate grammatical category not present in English (@cite{simik-2024}: "declarative PQs represent yet another type of utterance with a canonical SVO word order").

                                • interPPQ : CzechPQForm

                                  InterPPQ: Interrogative (VSO), Positive. Default unbiased PQ.

                                • interNPQ : CzechPQForm

                                  InterNPQ: Interrogative (VSO), Negative. Conveys positive epistemic bias. Broader distribution than English HiNQ (@cite{simik-2024} §5).

                                • declPPQ : CzechPQForm

                                  DeclPPQ: Declarative (SVO), Positive. Conveys positive evidential bias.

                                • declNPQ : CzechPQForm

                                  DeclNPQ: Declarative (SVO), Negative. Requires negative evidential bias.

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

                                    Czech bias profile (@cite{simik-2024} Table 2, based on @cite{stankova-2023}).

                                    Each cell records which Czech PQ forms are felicitous under a given combination of contextual evidence × original speaker bias. Empty list = no form is natural.

                                    Uses ContextualEvidence and OriginalBias from BiasedPQ rather than Czech-specific copies — these are the same bias dimensions.

                                    Table 2 (glossing over details): | Evid \ Epist | forP | neutral | againstP | |--------------|--------------------|--------------------|------------| | forP | | DeclPPQ, InterNPQ | DeclPPQ | | neutral | InterPPQ, InterNPQ | InterPPQ | | | againstP | DeclNPQ, InterNPQ | DeclNPQ | |

                                    *DeclPPQ with neutral/neutral is conditioned by information structure. InterNPQ with neutral epistemic requires explanation-seeking context.

                                    Equations
                                    Instances For

                                      InterPPQ is the default (unbiased) Czech PQ — the only form felicitous in quiz scenarios where no bias is intended (@cite{simik-2024} §4.1, ex. 25).

                                      @cite{simik-2024}'s key generalization: Declarative PQs are specialized for conveying evidential bias. The polarity of the DeclPQ matches the polarity of the evidential bias — DeclPPQ conveys positive evidential bias, DeclNPQ conveys negative evidential bias.

                                      Interrogative PQs, by contrast, convey epistemic bias (speaker's prior belief), and InterNPQ can additionally convey evidential bias.

                                      This is captured by: for any evidential bias, the DeclPQ that appears has matching polarity.

                                      Which negation readings are available for each verb position.

                                      V1 → only outer (FALSUM) — S&Š eq. (11): [ForceP Q [StrengthP FALSUM[iNeg] [PolP NEG-V[uNeg] [CP … [TP SUBJECT tV …]]]]]

                                      nonV1 → inner, medial, or outer — S&Š eq. (12), plus Staňková 2026's medial: [ForceP Q [StrengthP {FALSUM[iNeg]} [CP SUBJECT [TP {Op¬[iNeg]} NEG-V[uNeg] …]]]]

                                      Note: outer negation in nonV1 requires contrastive topic on the subject and contrastive focus on the verb (S&Š ex. 18).

                                      Equations
                                      Instances For

                                        Whether a verb position's default negation reading requires contextual evidence (evidential bias) to be felicitous.

                                        V1 (FALSUM): no — insensitive to context (S&Š §5.2, §5.3) nonV1 (inner): yes — requires negative evidential bias (S&Š §5.2)

                                        This is the key experimental finding of S&Š: the syntactic position of negation determines sensitivity to contextual evidence.

                                        Equations
                                        Instances For

                                          Context sensitivity correlates with evidential bias strength: V1/outer has no evidential bias requirement (FALSUM is epistemic). nonV1/inner has strong evidential bias (requires contextual evidence for ¬p).

                                          Czech outer negation (FALSUM) is compatible with all types of evidential bias — positive, negative, and neutral — unlike English HiNQs which require negative or neutral evidence.

                                          This is confirmed by the positive-evidence subexperiment (S&Š ex. 14): V1 PQs with positive evidential bias were rated very natural (median 6/7).

                                          The broader distribution follows from Czech FALSUM being tied to epistemic bias (speaker's possibility assessment) rather than to evidential bias. FALSUM^CZ (@cite{simik-2024} eq. 44) only requires that the speaker considers p epistemically possible, regardless of evidence.

                                          Corpus data for náhodou in PQs (@cite{simik-2024} fn. 56).

                                          From 100 random occurrences of náhodou in PQs (SYN v11, Czech National Corpus):

                                          • All 100 involved negation
                                          • 6 had indefinite pronoun/determiners, all were PPIs (no NCIs)
                                          • All diagnosed outer negation

                                          This confirms the fragment-level claim that náhodou requires outer negation and is incompatible with NCIs.

                                          • sampleSize :

                                            Total random sample size

                                          • withNegation :

                                            Number involving negation

                                          • withIndefinites :

                                            Number with indefinite determiners

                                          • ppiIndefinites :

                                            Number of those indefinites that were PPIs

                                          Instances For
                                            Equations
                                            Instances For

                                              Categories of InterNPQ use with náhodou (@cite{simik-2024} §5.2, fn. 59).

                                              From 100 random occurrences, the four main categories are:

                                              1. Prior speaker belief (conflict-resolving PQ) — 14%
                                              2. Explanation-seeking (no prior bias, situational evidence) — 40%
                                              3. Relevance PQ (suggesting p is worth discussing) — 20%
                                              4. Hope (speaker hopes p is true) — 20%
                                              • belief : InterNPQUseCategory

                                                Conflict between prior epistemic bias for p and evidential bias for ¬p. The prototypical biased HiNQ use.

                                              • explanationSeeking : InterNPQUseCategory

                                                No prior epistemic bias; evidential bias from observed situation. Not available in English HiNQ.

                                              • relevance : InterNPQUseCategory

                                                Speaker suggests p is relevant for further discussion.

                                              • hope : InterNPQUseCategory

                                                Speaker hopes that p, and considers it epistemically possible.

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

                                                  Explanation-seeking is the most common InterNPQ+náhodou use. This motivates @cite{simik-2024}'s weaker FALSUM^CZ: the attitude holder merely considers p epistemically possible, not that they believe p.