Dryer & Haspelmath (eds., 2013): WALS aggregate generalizations #
@cite{dryer-haspelmath-2013} @cite{dryer-1992}
Aggregate cross-linguistic generalizations derived directly from the WALS
Online corpus (@cite{dryer-haspelmath-2013}). Each theorem here is a
distributional fact about WALS chapter values: a count or comparative-count
predicate over Data.WALS.Features.F{N}A.allData.
These are not Greenberg-1963 implicational universals (which are conditional,
per-language statements; see Studies/Greenberg1963.lean). They are
unconditional aggregate claims about how WALS values distribute, of the form
"value X is more common than value Y in chapter N".
The chapter authorship within WALS Online is per-chapter (typically Dryer for Ch 81–88, 95–96; Hammarström for Ch 97; etc.); the entire atlas is @cite{dryer-haspelmath-2013}. @cite{dryer-1992} is the canonical earlier synthesis of many of these correlations, predating WALS.
What this file proves #
§1. Basic constituent order (Ch 81–83): SOV-most-common, SOV+SVO majority, OV-dominant in Ch 83, subject-first vs verb-first dominance, object- initial rarity, SV vs VS dominance.
§2. Per-WALS-chapter generalizations (Ch 87, 88, 90, 94, 95, 96, 97, 81B, 90B, 90C, 60, 61): per-chapter dominant-value claims. Several link to Gibson 2025's single-word-exception argument (Ch 87/88/97 all involve single-word dependents where head direction is less predictive).
Ch 81: SOV is the most common basic order.
Ch 81: SOV + SVO together exceed 75% of all sampled languages.
Ch 83: OV slightly outnumbers VO.
Ch 81: subject-first orders (SOV + SVO) far outnumber verb-first orders (VSO + VOS).
Ch 81: object-initial orders (OVS + OSV) are extremely rare — less than 2% of the sample.
Ch 82: SV overwhelmingly dominates VS. SV languages outnumber VS languages by more than 5 to 1.
Ch 87: N-Adj order dominates cross-linguistically (one of Gibson's single-word exceptions: adjectives are typically leaves).
Ch 88: Dem-N vs N-Dem is roughly balanced (another single-word exception: demonstratives are single words, so head direction is less predictive).
Ch 90: N-RelCl strongly dominates (relative clauses are recursive phrases, not single words — head direction matters).
Ch 94: Initial subordinator words are the most common strategy for adverbial subordination.
Ch 95: Harmonic pairs (OV+Postp, VO+Prep) vastly outnumber disharmonic (OV+Prep, VO+Postp).
Ch 96: VO+NRel strongly dominates VO+RelN (relative clauses are recursive, so head direction matters). OV languages are more mixed due to the N-RelCl strategy.
Ch 97: OV languages split between AdjN and NAdj (weak correlation, single-word exception). This contrasts with Ch 95 where OV+Prep is nearly absent.
Ch 81B: SOV-or-SVO is the most common dual-order pattern, consistent with the general dominance of subject-first orders.
Ch 90B: Dominant-only prenominal (RelN) is the majority among languages with this strategy.
Ch 90C: Dominant-only postnominal (NRel) is overwhelmingly the majority among languages with this strategy.
Ch 60: "Highly differentiated" between genitives, adjectives, and relative clauses is the majority value.
Ch 61: "Without marking" is the majority strategy for headless adjective phrases.