Merge pull request 'Sa'ueia workshop first cut: table, checker, sound proposal' (#5) from weksa/sa-ueia-workshop into main

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

settings/eusocial-interbeing/README.md

@@ -20,7 +20,7 @@ register, and history.
 - Weksa owns the language-project pipeline: ontology projection, grammar,
   morphology, phonology, diachrony, rendering, glossing, and critique.
 - A future language project owns one inhabitant language. Airawa imperial
-  register, pre-imperial Airawa registers, Sa'auei'a family-unit speech, ritual
+  register, pre-imperial Airawa registers, Sa'ueia family-unit speech, ritual
   breeding-ground registers, and ecological-partner contact systems should not
   collapse into one universal naming soup.
 
@@ -40,7 +40,7 @@ surfaces:
 - Airawa continent: mother-tree memory networks, archival dependency,
   disconnected networks, and the Airawa Empire's biological-memetic
   infrastructure.
-- Sa'auei'a continent: advanced nomadic civilization, family-unit political
+- Sa'ueia continent: advanced nomadic civilization, family-unit political
   movement, sacred breeding-ground commons, remembered routes, and reciprocal
   ecological relationships.
 

settings/eusocial-interbeing/language-design.md

@@ -49,7 +49,7 @@ rights, witnessed warrior promise, and honorable scar evidence. See
 ### Provisional Airawa Phonology Notes
 
 Airawa phonology is not finalized, but the current design pressure distinguishes
-it from Sa'auei'a softness. Airawa names should feel sharper, more dorsal and
+it from Sa'ueia softness. Airawa names should feel sharper, more dorsal and
 coronal, and more willing to close around hard edges.
 
 Provisional tendencies:
@@ -58,9 +58,9 @@ Provisional tendencies:
 - sibilant and postalveolar fricatives should carry social texture: `s`, `z`,
   `sh` / `zh`;
 - `r` may be trilled in some Airawa registers;
-- syllables can tolerate clipped or closed shapes where Sa'auei'a forms would
+- syllables can tolerate clipped or closed shapes where Sa'ueia forms would
   flow longer and wetter;
-- avoid making native Airawa names read like smooth Sa'auei'a route names unless
+- avoid making native Airawa names read like smooth Sa'ueia route names unless
   the form is explicitly a loan, exonym, or cross-continental register.
 
 The current room consensus pushes this further for introductory speech: Airawa
@@ -75,9 +75,9 @@ voiced postalveolar fricative /zh/ [ʒ]. The name is being considered for a
 Zyphos Face drawn from a non-imperial Airawa archival/fungal-road tradition: an
 archivist shaped by memory as kinship, not content.
 
-## Sa'auei'a Pressures
+## Sa'ueia Pressures
 
-Sa'auei'a language work must account for:
+Sa'ueia language work must account for:
 
 - advanced nomadic civilization without permanent urban dominance
 - squad-sized family units
@@ -88,10 +88,46 @@ Sa'auei'a language work must account for:
 - reciprocity routes and seasonal corridors
 - camps, commons, and partner sites as distinct name targets
 
-The Sa'auei'a continent should not read as empty wilderness. Its language should
+The Sa'ueia continent should not read as empty wilderness. Its language should
 have ways to name civilization that moves, recurs, bargains, and is remembered by
 ecology.
 
+## Sa'ueia Sound (Proposal For Deru To Cut)
+
+Status: proposal. Deru approved the outline in #zyphos on 2026-11-04; only the
+species name Sa'ueia is canon. Everything below is a working draft in
+`sa-ueia/table.yaml`, each entry marked `deru` or `weksa-placeholder`, and
+`sa-ueia/check.py` enforces what is mechanical.
+
+Taken from Deru's words:
+
+- five vowels: `a e i o u`;
+- no `h`, `q`, `y`; `k` where a `q` might have been;
+- short words, soft and open;
+- suffixing morphology; the one suffix so far is `-um`, a verb suffix;
+- identical letters merge at a morpheme join: `weru` + `um` gives `werum`,
+  `ila` + `werum` gives `ilawerum` (walk of knowledge);
+- a vowel only where one is needed.
+
+First roots, all Deru's: `ila` walk, `weru` know, `Olune` breeding ground
+(unanalyzed).
+
+Proposed, for Deru to accept or cut:
+
+- the apostrophe in Sa'ueia is orthographic, and is not counted as a letter;
+- "short" means at most 10 letters in a joined word, as a warning only;
+- no doubled letter inside a morpheme, so a double can only arise at a join
+  and the merge rule removes it.
+
+Open, not guessed:
+
+- the consonant inventory. Only `l m n r s w` appear in Deru's words so far,
+  and that is attestation, not an inventory;
+- whether the know root is `weru` or `wer-` (both give `werum`);
+- syllable shapes, stress, and what "vowel only where needed" means once the
+  inventory is known (the checker does not test it);
+- the phonological role of the apostrophe.
+
 ## Naming Brief Shape
 
 A Weksa naming brief should capture at least:

settings/eusocial-interbeing/sa-ueia/check.py

@@ -0,0 +1,102 @@
+#!/usr/bin/env python3
+"""Sa'ueia workshop checker: join morphemes by table.yaml's rules, flag violations.
+
+Reads table.yaml next to this file (or a path argument). The consonant inventory
+is open, so no consonant is ever flagged; only rules in the table are enforced.
+Exit status is 1 when any error is found; warnings do not fail the run.
+"""
+
+from __future__ import annotations
+
+import sys
+from dataclasses import dataclass
+from pathlib import Path
+
+import yaml
+
+TABLE = Path(__file__).with_name("table.yaml")
+PROVENANCES = {"deru", "weksa-placeholder"}
+
+
+@dataclass(frozen=True)
+class Finding:
+    severity: str  # "error" | "warning"
+    where: str
+    message: str
+
+    def __str__(self) -> str:
+        return f"{self.severity}: {self.where}: {self.message}"
+
+
+def letters(form: str) -> str:
+    return form.replace("'", "").lower()
+
+
+def join(morphemes: list[str]) -> str:
+    """Concatenate, merging one identical letter at each join."""
+    out = ""
+    for m in morphemes:
+        if out and m and out[-1] == m[0]:
+            m = m[1:]
+        out += m
+    return out
+
+
+def check_form(form: str, rules: dict, where: str, *, inside_morpheme: bool = False) -> list[Finding]:
+    found = []
+    text = letters(form)
+    for ch in rules["forbidden_letters"]["value"]:
+        if ch in text:
+            found.append(Finding("error", where, f"{form!r} contains forbidden letter {ch!r}"))
+    for a, b in zip(text, text[1:]):
+        if a == b and (not inside_morpheme or rules["no_doubled_letters_inside_morpheme"]["value"]):
+            found.append(Finding("error", where, f"{form!r} has doubled {a!r}"))
+    limit = rules["max_word_letters"]
+    if not inside_morpheme and len(text) > limit["value"]:
+        found.append(Finding(limit["severity"], where, f"{form!r} is {len(text)} letters, over {limit['value']}"))
+    return found
+
+
+def check_table(table: dict) -> list[Finding]:
+    rules = table["rules"]
+    found = []
+    for r_name, r in rules.items():
+        if r.get("provenance") not in PROVENANCES:
+            found.append(Finding("error", f"rules.{r_name}", "missing or unknown provenance"))
+    if table["consonants"]["inventory"] != "open":
+        found.append(Finding("error", "consonants", "inventory is not Deru's to guess; keep it open until given"))
+    forms: dict[str, str] = {}
+    for kind in ("roots", "suffixes"):
+        for entry in table[kind]:
+            form = entry["form"]
+            where = f"{kind}.{form}"
+            if entry.get("provenance") not in PROVENANCES:
+                found.append(Finding("error", where, "missing or unknown provenance"))
+            forms[form] = kind
+            found += check_form(form, rules, where, inside_morpheme=True)
+    for comp in table["compositions"]:
+        where = "compositions." + "+".join(comp["morphemes"])
+        if comp.get("provenance") not in PROVENANCES:
+            found.append(Finding("error", where, "missing or unknown provenance"))
+        for m in comp["morphemes"]:
+            if m not in forms and not any(m == f2 for f2 in (c["expected"] for c in table["compositions"])):
+                found.append(Finding("error", where, f"morpheme {m!r} is in no table row or earlier composition"))
+        got = join(comp["morphemes"])
+        if got != comp["expected"]:
+            found.append(Finding("error", where, f"joins to {got!r}, table says {comp['expected']!r}"))
+        found += check_form(comp["expected"], rules, where)
+    return found
+
+
+def main(argv: list[str]) -> int:
+    path = Path(argv[1]) if len(argv) > 1 else TABLE
+    findings = check_table(yaml.safe_load(path.read_text(encoding="utf-8")))
+    for f in findings:
+        print(f)
+    print(f"{sum(f.severity == 'error' for f in findings)} error(s), "
+          f"{sum(f.severity == 'warning' for f in findings)} warning(s)")
+    return 1 if any(f.severity == "error" for f in findings) else 0
+
+
+if __name__ == "__main__":
+    sys.exit(main(sys.argv))

settings/eusocial-interbeing/sa-ueia/table.yaml

@@ -0,0 +1,81 @@
+# Sa'ueia workshop table, first cut.
+# Source: Deru's words in #zyphos, 2026-11-04. Setting owner: Deru (the Zyphos vault).
+# Every entry carries provenance:
+#   deru               - said by Deru, or part of the sound outline Deru approved
+#   weksa-placeholder  - Weksa's stand-in so the checker can run; Deru may cut it
+# Only the species name Sa'ueia is canon. Everything else here is a working
+# draft, even when its provenance is deru.
+
+roots:
+  - form: ila
+    gloss: walk
+    provenance: deru
+  - form: weru
+    gloss: know
+    provenance: deru
+    note: >
+      Deru gave "werum" (knowledge, maybe a verb) and was asked whether -um
+      suffixes wer- or weru. "weru" is the reading under which the join rule
+      does visible work (weru + um -> werum); wer- gives the same surface form.
+      Open for Deru.
+  - form: olune
+    gloss: breeding ground
+    provenance: deru
+    analysis: unanalyzed   # no morpheme boundaries known
+  - form: sa'ueia
+    gloss: the species name
+    provenance: deru
+    canon: true
+    analysis: unanalyzed
+    note: Apostrophe is part of the canon spelling; its phonological role is open.
+
+suffixes:
+  - form: um
+    gloss: verb suffix
+    provenance: deru
+
+rules:
+  vowels:
+    value: [a, e, i, o, u]
+    provenance: deru
+  forbidden_letters:
+    value: [h, q, y]
+    provenance: deru
+    note: k is used where q might have been.
+  join_merge_identical:
+    value: true
+    provenance: deru
+    note: >
+      When one morpheme ends in the letter the next begins with, they merge to
+      one letter at the join (weru + um -> werum, ila + werum -> ilawerum).
+  morphology_direction:
+    value: suffixing
+    provenance: deru
+  short_words:
+    provenance: deru
+    checked: false   # no number from Deru
+  max_word_letters:
+    value: 10
+    provenance: weksa-placeholder
+    severity: warning
+    note: Stand-in for "short words"; apostrophes not counted. Deru sets the real bound.
+  no_doubled_letters_inside_morpheme:
+    value: true
+    provenance: weksa-placeholder
+    note: Follows from the merge rule only if doubles never occur inside a morpheme either.
+  vowel_only_where_needed:
+    provenance: deru
+    checked: false   # not mechanically decidable without the consonant inventory
+
+consonants:
+  inventory: open   # Deru has not given it. The checker does not guess.
+  seen_in_deru_words: [l, m, n, r, s, w]   # attested above, not an inventory
+
+# Joined forms the checker must reproduce. Each: morphemes in order, expected surface.
+compositions:
+  - morphemes: [weru, um]
+    expected: werum
+    provenance: deru      # "werum" is Deru's form
+  - morphemes: [ila, werum]
+    expected: ilawerum
+    provenance: deru      # "Ilawerum", walk of knowledge

settings/eusocial-interbeing/sa-ueia/test_check.py

@@ -0,0 +1,94 @@
+import copy
+import unittest
+
+import yaml
+
+import check
+
+
+def table():
+    return yaml.safe_load(check.TABLE.read_text(encoding="utf-8"))
+
+
+def errors(t):
+    return [f for f in check.check_table(t) if f.severity == "error"]
+
+
+class JoinTests(unittest.TestCase):
+    def test_identical_letters_merge_at_join(self):
+        self.assertEqual(check.join(["weru", "um"]), "werum")
+        self.assertEqual(check.join(["ila", "werum"]), "ilawerum")
+
+    def test_merge_is_one_letter_not_all(self):
+        self.assertEqual(check.join(["ila", "aba"]), "ilaba")
+        self.assertEqual(check.join(["ila", "ala"]), "ilala")
+
+    def test_different_letters_concatenate(self):
+        self.assertEqual(check.join(["ila", "um"]), "ilaum")
+
+    def test_merge_applies_only_at_joins(self):
+        # a doubled letter inside a morpheme is not collapsed by join
+        self.assertEqual(check.join(["aa", "b"]), "aab")
+
+
+class TableTests(unittest.TestCase):
+    def test_shipped_table_is_clean(self):
+        self.assertEqual(errors(table()), [])
+
+    def test_forbidden_letter_in_root_flagged(self):
+        t = table()
+        t["roots"].append({"form": "haru", "gloss": "x", "provenance": "weksa-placeholder"})
+        self.assertTrue(any("forbidden letter 'h'" in str(f) for f in errors(t)))
+
+    def test_q_and_y_flagged(self):
+        for ch in "qy":
+            t = table()
+            t["roots"].append({"form": f"a{ch}a", "gloss": "x", "provenance": "deru"})
+            self.assertTrue(any(f"forbidden letter '{ch}'" in str(f) for f in errors(t)), ch)
+
+    def test_k_allowed(self):
+        t = table()
+        t["roots"].append({"form": "kela", "gloss": "x", "provenance": "weksa-placeholder"})
+        self.assertEqual(errors(t), [])
+
+    def test_missing_or_unknown_provenance_flagged(self):
+        for bad in (None, "guess"):
+            t = table()
+            t["suffixes"][0]["provenance"] = bad
+            self.assertTrue(any("provenance" in str(f) for f in errors(t)), bad)
+
+    def test_unmerged_composition_flagged(self):
+        t = table()
+        t["compositions"][0]["expected"] = "weruum"
+        msgs = [str(f) for f in errors(t)]
+        self.assertTrue(any("joins to 'werum'" in m for m in msgs))
+        self.assertTrue(any("doubled 'u'" in m for m in msgs))
+
+    def test_doubled_letter_inside_morpheme_flagged(self):
+        t = table()
+        t["roots"].append({"form": "ilaa", "gloss": "x", "provenance": "weksa-placeholder"})
+        self.assertTrue(any("doubled 'a'" in str(f) for f in errors(t)))
+
+    def test_unknown_morpheme_in_composition_flagged(self):
+        t = table()
+        t["compositions"].append({"morphemes": ["ila", "zo"], "expected": "ilazo", "provenance": "weksa-placeholder"})
+        self.assertTrue(any("'zo'" in str(f) for f in errors(t)))
+
+    def test_long_word_is_warning_not_error(self):
+        t = table()
+        t["compositions"].append({"morphemes": ["ila", "werum", "ila", "werum"], "expected": "ilawerumilawerum",
+                                  "provenance": "weksa-placeholder"})
+        self.assertEqual(errors(t), [])
+        self.assertTrue(any(f.severity == "warning" for f in check.check_table(t)))
+
+    def test_consonant_inventory_must_stay_open(self):
+        t = table()
+        t["consonants"]["inventory"] = ["l", "m"]
+        self.assertTrue(any("inventory" in str(f) for f in errors(t)))
+
+    def test_apostrophe_is_not_a_letter(self):
+        self.assertEqual(check.letters("Sa'ueia"), "saueia")
+
+
+if __name__ == "__main__":
+    unittest.main()

state/map.yaml

@@ -221,10 +221,10 @@ open_questions:
   - What is the minimal Rust to Dust language-project slice that proves the
     engine without hard-coding the setting into it?
   - What minimal Eusocial Interbeing language-project slice can prove naming for
-    Airawa, Sa'auei'a, ecological partner, and imperial registers without
+    Airawa, Sa'ueia, ecological partner, and imperial registers without
     collapsing them into one universal language?
   - Which Zyphos language project should be first: Airawa, imperial Airawa,
-    Sa'auei'a, or a narrow naming-brief interlingua?
+    Sa'ueia, or a narrow naming-brief interlingua?
   - What is the minimal Airawa starter lexicon and permission model needed to
     translate the first chieftain/visitor/mother-tree scene honestly?
   - How should Airawa encode honorable scars: bodily evidence, honor claim, debt