Qamus particle family P-02
Generated from repo ledgers at main@3396d79 (2026-08-05). This issue is a public mirror of the committed ledgers; on any conflict the ledgers win. Evidence in comments must cite repo paths, PR numbers, or blocker ids — never private-workspace packet contents or server topology.
What this family is
Particle work-ordering family P-02: 22 entries — إِنْ, إِذْ, إِذَا, إِذًا, قَدْ, ثُمَّ, لَمَّا, لَكِنْ, كَمَا, مِنْ, أَمْ, هَلْ, مَاذَا, لِمَ …
(pages: p014, p016, p017, p018, p019, p021, p023, p024, p025, p034, p040, p041, p045, p048, p051, p052, p054, p060, p077, p086, p099, p100; source: qamus/data/particle-tranche-membership.json). Function families: homograph: in (condition / negation / mukhaffafa); homograph: idh (temporal / taʿlīl); homograph: idhā (condition-temporal / fujāʾiyya); homograph: idhan (result/inferential vs idhā); homograph: qad (taḥqīq/taqlīl vs qudda); homograph: thumma (conj) vs thamma; homograph: lammā (ḥīniyya / jazm not-yet); homograph: lākin (istidrāk vs la+kunnā); homograph: ka+mā function split; homograph: min (jarr) vs man; homograph: am (ʿaṭf-istifhām vs umm); homograph: hal (istifhām, function-load); homograph: mādhā (mā+dhā split); homograph: lima (li+mā vs lam); homograph: bal (iḍrāb/istidrāk); homograph: kallā (radʿ vs kull); homograph: naʿam (response vs niʿma); homograph: law (condition / tamannī / maṣdarī); homograph: thamma (locative dem) vs thumma; homograph: anā (pronoun) vs annā; homograph: mā (relative/negation/istifhām/maṣdarī); homograph: man (relative/istifhām/condition) vs min.
Flags: 22 scholar/two-vote-required; 22 homograph; 21 dual-family-tension (rival family assignment recorded, never collapsed).
Contains p048 (لِمَ, homograph li+mā vs lam — scholar/two-vote flagged), the particle axis's sole live span defect (2:91:3, C1 stem_swallow debt row): closing this family closes it. CORRECTION 2026-08-05: an earlier revision of the P-03 issue mis-attributed p048 to P-03.
The denominator that makes this the top-yield lane: the particle universe is ~50,263 candidate
occurrences with ~109,364 appearances, and 99.1% of those appearances sit on V/N pages as
example-āyah context words — currently certified: none for every row. The particle pages' own
spans are 983/984 rich live, but that old-era richness does not transclude; only certification does.
Impact on transclusion & the sarf/naḥw lattice projection
This is the purest transclusion lane in the programme: certify one particle occurrence and the
projection-hash parity invariant renders it rich on every card it appears on — across all 24 VN
tranches — with zero per-page authoring (measured on the VN-00 readback pilot: ~2,690 rendered-span
edges corpus-wide from one window's accounting). Surface match never authorizes reuse (og-1/og-2):
membership goes through the candidate matrix → disposition → certification rungs, never lookalike copying.
When this family closes, post the measured unlock counts on the affected VN issues.
Impact on the flywheel & repo
Family certification exercises the sarf/naḥw skills on real corpus at scale; every abstention,
disagreement, or defect (e.g. clitic-host segmentation, allomorph families like لَهُمْ/لَكُمْ) returns
to /fusha/ as fixtures, misconception rows, or registry keys. Dual-function and homograph rows
exercise rival preservation — outcomes feed the naḥw skill's attributed-alternatives machinery.
Closure conditions (ladder)
Update & closure discipline
- The executor updates this issue as work lands: check boxes with a one-line evidence citation
(PR number + ledger path), and post a comment when another issue's closure changes this window's
measured state (e.g. a particle-family closure lighting up context words here — cite the count).
- This issue CLOSES only when BOTH are true: the repo ledgers record the closed state AND the live
website renders it. The deploy rung is owner-gated (docs/blockers.yaml → owner-deploy-packet);
everything above it is executor work. Candidate-plane completion with no deploy = issue stays open.
- The closing comment must state the realized impact: occurrences accounted, facts certified, cards
now rich live, cross-page appearances lit elsewhere (name the issues), and what the flywheel
returned to /fusha/ (fixtures, misconception rows, drill keys — cite paths).
Qamus particle family P-02
What this family is
Particle work-ordering family P-02: 22 entries — إِنْ, إِذْ, إِذَا, إِذًا, قَدْ, ثُمَّ, لَمَّا, لَكِنْ, كَمَا, مِنْ, أَمْ, هَلْ, مَاذَا, لِمَ …
(pages: p014, p016, p017, p018, p019, p021, p023, p024, p025, p034, p040, p041, p045, p048, p051, p052, p054, p060, p077, p086, p099, p100; source:
qamus/data/particle-tranche-membership.json). Function families: homograph: in (condition / negation / mukhaffafa); homograph: idh (temporal / taʿlīl); homograph: idhā (condition-temporal / fujāʾiyya); homograph: idhan (result/inferential vs idhā); homograph: qad (taḥqīq/taqlīl vs qudda); homograph: thumma (conj) vs thamma; homograph: lammā (ḥīniyya / jazm not-yet); homograph: lākin (istidrāk vs la+kunnā); homograph: ka+mā function split; homograph: min (jarr) vs man; homograph: am (ʿaṭf-istifhām vs umm); homograph: hal (istifhām, function-load); homograph: mādhā (mā+dhā split); homograph: lima (li+mā vs lam); homograph: bal (iḍrāb/istidrāk); homograph: kallā (radʿ vs kull); homograph: naʿam (response vs niʿma); homograph: law (condition / tamannī / maṣdarī); homograph: thamma (locative dem) vs thumma; homograph: anā (pronoun) vs annā; homograph: mā (relative/negation/istifhām/maṣdarī); homograph: man (relative/istifhām/condition) vs min.Flags: 22 scholar/two-vote-required; 22 homograph; 21 dual-family-tension (rival family assignment recorded, never collapsed).
Contains p048 (لِمَ, homograph li+mā vs lam — scholar/two-vote flagged), the particle axis's sole live span defect (2:91:3, C1 stem_swallow debt row): closing this family closes it. CORRECTION 2026-08-05: an earlier revision of the P-03 issue mis-attributed p048 to P-03.
The denominator that makes this the top-yield lane: the particle universe is ~50,263 candidate
occurrences with ~109,364 appearances, and 99.1% of those appearances sit on V/N pages as
example-āyah context words — currently
certified: nonefor every row. The particle pages' ownspans are 983/984 rich live, but that old-era richness does not transclude; only certification does.
Impact on transclusion & the sarf/naḥw lattice projection
This is the purest transclusion lane in the programme: certify one particle occurrence and the
projection-hash parity invariant renders it rich on every card it appears on — across all 24 VN
tranches — with zero per-page authoring (measured on the VN-00 readback pilot: ~2,690 rendered-span
edges corpus-wide from one window's accounting). Surface match never authorizes reuse (og-1/og-2):
membership goes through the candidate matrix → disposition → certification rungs, never lookalike copying.
When this family closes, post the measured unlock counts on the affected VN issues.
Impact on the flywheel & repo
Family certification exercises the sarf/naḥw skills on real corpus at scale; every abstention,
disagreement, or defect (e.g. clitic-host segmentation, allomorph families like لَهُمْ/لَكُمْ) returns
to /fusha/ as fixtures, misconception rows, or registry keys. Dual-function and homograph rows
exercise rival preservation — outcomes feed the naḥw skill's attributed-alternatives machinery.
Closure conditions (ladder)
tools/certify_typed_fact.py), guarded projection, or explicit pending/scholar state with causeUpdate & closure discipline
(PR number + ledger path), and post a comment when another issue's closure changes this window's
measured state (e.g. a particle-family closure lighting up context words here — cite the count).
website renders it. The deploy rung is owner-gated (
docs/blockers.yaml→owner-deploy-packet);everything above it is executor work. Candidate-plane completion with no deploy = issue stays open.
now rich live, cross-page appearances lit elsewhere (name the issues), and what the flywheel
returned to /fusha/ (fixtures, misconception rows, drill keys — cite paths).