Sagey (1986) @cite{sagey-1986} #
The Representation of Features and Relations in Non-Linear Phonology. PhD dissertation, Massachusetts Institute of Technology.
@cite{sagey-1986} proposes a hierarchical feature geometry organized by
vocal tract articulator, establishing the labial, coronal, dorsal, and
soft palate nodes that are now standard in phonological theory
(Theories/Phonology/FeatureGeometry.lean). The geometry predicts which
multiply-articulated (complex) segments are possible in human language
(Theories/Phonology/ComplexSegments.lean).
This study file formalizes Sagey-specific contributions that go beyond the consensus geometry:
Formalized contributions #
Major/minor articulator distinction (Ch. 3): in complex segments, one articulator is major (determines degree of closure) and one is minor.
Degree of closure as articulator-level property (Ch. 3): [continuant] and [consonantal] describe the root-to-articulator relationship, not the segment as a whole. This allows different degrees of closure at different articulators within one segment (e.g., clicks).
Soft palate independence (Ch. 2): nasal assimilation (spreading the soft palate node) is structurally simpler than place assimilation (spreading the place node), explaining its cross-linguistic frequency.
Empirical arguments: Nupe labiovelars [k͡p]/[g͡b] as labio-dorsal complex segments with dorsal as major articulator.
In a complex segment, one articulator is designated as major (determines the primary degree of closure seen by syllable structure) and the other as minor. This distinction is Sagey-specific — modern phonology does not uniformly adopt it.
- major_is_articulator : self.major.IsArticulator
- minor_is_articulator : self.minor.IsArticulator
Instances For
Nupe labiovelar /k͡p/: dorsal is major (stop closure), labial is minor. Sagey argues this based on nasalization patterns: labiovelars in Nupe nasalize to [ŋ͡m], with the velar nasal first, showing dorsal is the major articulator.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Sagey's degree of closure: a property of an articulator node, not a terminal feature. This is the Sagey-specific treatment; the modern geometry encodes closure as [±continuant] at the supralaryngeal node.
- stop : DegreeOfClosure
- fricative : DegreeOfClosure
- approximant : DegreeOfClosure
Instances For
Equations
- Sagey1986.instDecidableEqDegreeOfClosure x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- Sagey1986.instReprDegreeOfClosure = { reprPrec := Sagey1986.instReprDegreeOfClosure.repr }
An articulator paired with its degree of closure. In Sagey's framework, a click can be [−cont] at coronal and [−cont] at dorsal independently.
- closure : DegreeOfClosure
- node_is_articulator : self.node.IsArticulator = (true = true)
Instances For
A click's anterior closure (coronal, full stop).
Equations
- Sagey1986.click_anterior = { node := Phonology.FeatureGeometry.GeomNode.coronal, closure := Sagey1986.DegreeOfClosure.stop, node_is_articulator := Sagey1986.click_anterior._proof_1 }
Instances For
A click's posterior closure (dorsal, full stop).
Equations
- Sagey1986.click_posterior = { node := Phonology.FeatureGeometry.GeomNode.dorsal, closure := Sagey1986.DegreeOfClosure.stop, node_is_articulator := Sagey1986.click_posterior._proof_1 }
Instances For
Nasal assimilation spreads the soft palate node (1 feature), while place assimilation spreads the place node (14 features). The soft palate's independence from place explains why nasal assimilation is cross-linguistically simpler and more common than place assimilation: it involves spreading a smaller constituent.
Nasality is NOT under the place node — spreading place does not spread nasality. This is Sagey's core structural argument for the soft palate node as a separate constituent.
Nasality IS under supralaryngeal (via the soft palate node), so total assimilation (spreading supralaryngeal) does spread nasality.
A Nupe labiovelar stop /k͡p/: specified for both labial and dorsal, voiceless, non-continuant.
Equations
- One or more equations did not get rendered due to their size.
Instances For
The Nupe /k͡p/ is a complex segment (two active place articulators).
The Nupe /k͡p/ is well-formed: labial ≠ dorsal.
A simple /p/ (labial only) is not complex.
Equations
- One or more equations did not get rendered due to their size.
Instances For
A velar nasal /ŋ/ is NOT complex despite activating both the dorsal articulator and the soft palate (velum lowering). The soft palate is structurally independent of place, so nasal + place combinations are simple segments — this is Sagey's core argument for the soft palate node's independence.
Equations
- One or more equations did not get rendered due to their size.
Instances For
The velar nasal is well-formed (only one place articulator: dorsal).
A coronal-only segment (alveolar /t/) with multiple coronal features is NOT complex: [+cor, +ant, −dist] all fall under the single coronal articulator. This formalizes Sagey's key prediction: palatal–velar stops are impossible because palatals and velars are both dorsal — no combination of features under a single articulator can produce a complex segment.
Equations
- One or more equations did not get rendered due to their size.
Instances For
An alveopalatal (postalveolar) is [+cor, −ant, +dist] — still just one articulator (coronal), so not complex. An alveolar-alveopalatal doubly-articulated stop is therefore impossible: both articulations use the coronal articulator (@cite{sagey-1986} §2.2).
Equations
- One or more equations did not get rendered due to their size.
Instances For
Temporal derivation of the No-Crossing Constraint #
@cite{sagey-1986} Ch. 5 derives the ban on crossing association lines from temporal precedence. This section demonstrates the derived constraint with concrete integer-valued time instances.
A concrete geminate /t:/ occupying timing slots [0,1] and [2,3], with the melodic element spanning [0,3].
Equations
- One or more equations did not get rendered due to their size.
Instances For
Both associations in the geminate are valid (timing overlaps melody).
A concrete falling contour tone: timing [0,4], H tone [0,2], L tone [2,4].
Equations
- One or more equations did not get rendered due to their size.
Instances For
Both associations in the contour tone are valid.
Crossing forces invalidity (@cite{sagey-1986} §5.3): a crossing configuration — timing₁ at [0,1] → melody₁ at [4,5], timing₂ at [2,3] → melody₂ at [0,1] — has its timing positions correctly ordered and melody positions reversed, but the first association is invalid because timing [0,1] does not overlap melody [4,5]. This demonstrates the mechanism: crossing is impossible not because it's stipulated, but because validity (temporal overlap) cannot be satisfied for both associations simultaneously when they cross.