Phonological typology — WALS substrate (Chapters 1–19) #

@cite{dryer-haspelmath-2013} @cite{maddieson-2013} @cite{goedemans-van-der-hulst-2013}

Theory-neutral phonological-typology substrate distilled from WALS Chapters 1–19, in the same Linglib/Typology/{Domain}.lean mould as Case.lean, WordOrder.lean, etc. Per-chapter enums + WALS converters + WALS aggregate distributional theorems.

Authorship attribution (NOT one paper) #

The 19 chapters are NOT all by Maddieson — the file's previous incarnation (Phenomena/Phonology/Typology.lean) cited only @cite{maddieson-2013}, which is an attribution error. Correct split:

Chs 1–13, 18–19: Maddieson 2013 (segmental inventory + tone)
Chs 14–17: Goedemans & van der Hulst 2013 (stress location, weight- sensitivity, weight factors, rhythm). Different framework: StressTyp database, presupposes a metrical-grid view where every word has a primary head — exactly what Hyman 2006 (the file's main consumer) challenges.

Theory-laden chapters (acknowledged) #

This file is substrate in the sense that it just hosts WALS data; but several WALS chapters embed analytical commitments worth flagging:

Ch 13 (tone) is NOT framework-neutral. The noTones / simpleToneSystem / complexToneSystem partition operationalises tone as level inventory size, exactly the position Hyman 2006 challenges (he defines tone functionally: "pitch enters lexical realisation"). Hyman2006.lean collapses the simple/complex split back to a single +T bit and flags WALS as "coarser than Lionnet 2025".
Chs 14–17 (stress) embed StressTyp commitments. fixedStress presupposes obligatory metrical heads; languages without word-level heads (French in Hyman's analysis) are mis-coded. Hyman2006.lean records this as french_wals_mismatch.
Ch 12 (syllable structure) commits to syllable-as-unit. Languages analysed without syllables (Bella Coola per Bagemihl 1991) cannot be classified.

These commitments are NOT removed at the substrate level — the substrate records what WALS encodes, with the commitments. Theory-side critique lives in study files (Hyman2006.lean, future work).

Featural double-coding (deferred) #

WALS Chs 7 (glottalized), 8 (lateral), 9 (velar nasal), 11 (front rounded) are inventory-presence claims that could derive from the featural substrate in Linglib/Theories/Phonology/Featural/Features.lean ([+constricted glottis], [+lateral], [+nasal, +dorsal], [+front, +round]) applied to a PHOIBLE inventory. Bridge theorems are deferred to Phenomena/Phonology/Studies/Maddieson2013.lean.

PHOIBLE 2.0 as the post-2019 successor for inventory facts #

Maddieson's UPSID-based methodology (~451 inventories) is the 1980s baseline WALS Ch 1–11/18–19 inherits. The field-canonical post-2019 successor is PHOIBLE 2.0 (@cite{moran-mccloy-2019}; ~3000 inventories with full IPA transcription + 37-feature decomposition), housed in Linglib/Data/PHOIBLE/. WALS partitions remain useful for cross-paper classification references; PHOIBLE provides the underlying inventories. Bridge theorems WALS↔PHOIBLE live in Phenomena/Phonology/Studies/Maddieson2013.lean.