Superset spellout over containment hierarchies #
The static core of the nanosyntax selection rule (no cyclic override
or spellout-driven movement), stated over the same ExponenceRule
vocabularies as the Elsewhere engine of
Morphology/Containment/Vocabulary.lean. A nanosyntax lexical entry
stores a constituent and carries no contextual restriction
(ContextFree); it can spell out any structure it contains — the
Superset Principle ([Sta09a]) — and competition selects the
smallest matching entry (Minimize Junk, which [Sta09a] himself
identifies with the Elsewhere Condition: the directional asymmetry
lives in the matching relation, not in the competition). The two
engines are dual: Elsewhere insertion maximizes the threshold over
rules the structure contains; Superset spellout minimizes the span
over entries that contain the structure.
The headline is the cross-framework theorem
spelloutGenerable_iff_generable: over a linear containment hierarchy,
Superset spellout from context-free antihomophonous entries and
Elsewhere insertion from terminal antihomophonous rules generate
exactly the same fully-realized patterns — the contiguous ones. The
frameworks are extensionally equivalent on this fragment while
differing intensionally: an overspecified entry realizes smaller
structures under Superset but yields a gap under Elsewhere insertion
(see the divergence examples at the end), which is how nanosyntax
derives ABC with no contextual apparatus — suppletion as pure phrasal
spellout, portmanteau by constituent size with no environment
restriction ([DCVW17] for exactly this Latin
degree case) — where DM needs the context-restricted portmanteau of
[Bob12] ch. 5.
Main declarations #
ExponenceRule.Matches— the Superset Principle: entry contains the target structureContextFree— the nanosyntax restriction on vocabulariesminSpan,spelloutWinner,spellout— Minimize-Junk competition and the realized patternisContiguous_spellout— *ABA for Superset spelloutisContiguous_iff_spelloutGenerable— spellable = contiguousspelloutGenerable_iff_generable— DM/nanosyntax equigenerativity
The older tree-based nanosyntax fragment
(Morphology/Nanosyntax/Basic.lean) predates this engine and is not
yet grounded in it.
The Superset Principle #
The Superset Principle ([Sta09a]): an entry can spell out
grade g when the constituent it stores contains grade g's
structure. Anti-monotone in g, dually to AppliesAt.
Instances For
Minimize Junk derived, dually to
ExponenceRule.moreSpecific_iff_threshold_le: an entry matches in a
subset of the structures another matches in iff it stores the smaller
constituent — smallest-match selection is Pāṇinian specificity under
superset matching.
The nanosyntax restriction, in the pointer-free idealization of [Cah09]: entries store bare constituents, with no DM-style contextual environment.
Equations
- Morphology.Containment.ContextFree v = ∀ it ∈ v, it.context = none
Instances For
Equations
- Morphology.Containment.instDecidableContextFree v = id inferInstance
Minimize-Junk competition #
The entries matching at grade g, in vocabulary order.
Equations
- Morphology.Containment.matching v g = List.filter (fun (it : Morphology.Containment.ExponenceRule n F) => decide (it.Matches g)) v
Instances For
The least matching span at grade g — Minimize Junk: the winning
entry stores as little unrealized structure as possible. ⊤ when no
entry matches.
Equations
- Morphology.Containment.minSpan v g = (List.map Morphology.Containment.ExponenceRule.spans (Morphology.Containment.matching v g)).minimum
Instances For
The key transfer lemma, dual to maxThreshold_eq_coe_of_le: a
winning span persists upward as long as the grade stays inside it.
The spellout winner at grade g: the first entry attaining the
least matching span.
Equations
- One or more equations did not get rendered due to their size.
Instances For
The spelled-out pattern: at each grade, the Minimize-Junk winner's
exponent (none when no entry contains the structure — a spellout
gap; full nanosyntax would attempt rescue operations before declaring
ineffability).
Equations
Instances For
Spellout gaps propagate upward: an entry matching a higher grade
would match the lower one too, so a gap at g forces gaps at every
g' ≥ g — [Dek21]'s paradigm-gap monotonicity for indefinites.
*ABA for Superset spellout #
The mirror image of isContiguous_realize: minSpan is monotone in
the grade (matching sets shrink upward, so the least span weakly
grows), the winner is a function of it, and antihomophony makes
the winner's exponent injective — so spellout fibers are convex. This
is the nanosyntax derivation of *ABA ([Cah09]), run on the same
vocabulary type as the DM derivation.
Completeness: spellable = contiguous #
The latest grade sharing g's form.
Equations
- Morphology.Containment.lastOcc p g = {j : Fin n | p j = p g}.max' ⋯
Instances For
The canonical nanosyntax lexicon of a pattern: one context-free entry per form, storing the form's largest constituent.
Equations
- One or more equations did not get rendered due to their size.
Instances For
A pattern is Superset-spellable: some context-free antihomophonous lexicon spells it out in full.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Spellable = contiguous: a fully realized pattern arises from Superset spellout over a context-free antihomophonous lexicon iff it is contiguous — [Cah09]'s Universal Contiguity as a theorem of the engine, mirroring Caha's own derivation of UC from the Superset Principle; the converse direction sharpens it to exact generative capacity.
DM and nanosyntax are equigenerative over linear containment hierarchies: Superset spellout from context-free antihomophonous lexicons and Elsewhere insertion from terminal antihomophonous vocabularies generate exactly the same fully realized patterns — the contiguous ones. The frameworks differ intensionally (rule format and selection direction; see the divergence examples below) but not in generative capacity on this fragment.
Where the frameworks diverge #
The equivalence is extensional, over total realizations. On partial lexicons the selection directions come apart: an overspecified entry realizes smaller structures under the Superset Principle but leaves a gap under Elsewhere insertion. This is nanosyntax's characteristic prediction — overspecified entries double as defaults for smaller structures — against DM's characteristic gaps.
Antihomophony is necessary for *ABA #
Two distinct entries sharing an exponent — accidental homophony, an
Antihomophonous violation — generate ABA: with "A" stored at grades 0
and 2 and "B" at grade 1, Minimize Junk yields A–B–A. [Cah09]
distinguishes accidental homophony (phonological) from systematic
syncretism (one item over a contiguous span); the Antihomophonous
hypothesis of isContiguous_spellout is exactly that distinction.