shema: record grammatical voice trajectory on every ShemaContext

The Shema (Devarim 6:4-9, 11:13-21, Bamidbar 15:37-41) deliberately
oscillates between singular ("you, sg.") and plural ("you, pl.")
voice. Devarim 11:14 famously switches mid-verse from אַרְצְכֶם (2mp)
to וְאָסַפְתָּ דְגָנֶךָ (2ms). The text refuses to stabilize at
either pole; that refusal IS the anti-capture mechanism:

  • stuck at 2ms     → private capture (gnostic interior)
  • stuck at 2mp     → mob theology (collective without person)
  • stuck at 1cs/1cp → operator-as-deity (the calf)

Before this commit, vav_pipeline/lib/vav_pipeline/shema.ex — the
module literally named after the Shema — encoded none of this. No
markers for person/singular/plural anywhere in the file. The chat
review surfaced this as "load-bearing anti-capture, missing".

Changes:

L1Reader.extract_voice/1 walks every word of a verse, parses the
OSHB morphology codes already present in L1 (Vqq2ms, Sp2mp, Pp1cs,
Vqi2mp, …), and returns a profile with markers, summary by
voice-key, dominant pole, and oscillates_within_verse boolean.
Strips the H/A language marker that prefixes the first /-segment of
each morph string — without that, every verse-initial verb (HVqi…)
was being silently dropped.

ShemaContext now carries voice_profile on every reception (both
receive_context/1 and receive_context_at/2). The voice is computed
fresh from L1, never cached.

8 new tests in test/shema_voice_trajectory_test.exs verify the
extractor against four anchoring verses:
  Devarim 6:5    — uniformly second-singular (5 markers)
  Devarim 11:13  — dominantly second-plural (5 second-plural markers)
  Devarim 11:14  — oscillates_within_verse: true (the mid-verse switch)
  Bereishit 1:1  — only third-singular (no addressee)
Plus three context-level tests confirming ShemaContext exposes the
field, computes it fresh, and traces a non-stabilizing trajectory
across 11:13 → 11:14 → 11:19.

Full vav_pipeline suite: 288 tests, 3 failures (all pre-existing
Shabbat-day calendar failures unrelated to this change), 19 excluded
(L3 PostgreSQL).

Still open: the across-verse stabilization-detector. ShemaContext
now KNOWS its voice; nothing yet refuses a sequence of receptions
that all stabilize at one pole. That belongs downstream — probably
the kushya generator (kal-vachomer challenge when voice has not
moved in N receptions). Logged in Pass 2 of
~/shorashim-conversation-review.md (Finding A).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

vav_pipeline/lib/vav_pipeline/l1_reader.ex

@@ -143,6 +143,61 @@ defmodule VavPipeline.L1Reader do
     Map.get(verse, :smikhut, [])
   end
 
+  @doc """
+  Extract grammatical voice profile from a verse.
+
+  Walks every word and reads the OSHB morphology codes, returning the
+  person/gender/number of every verb (V…), pronominal suffix (Sp…),
+  and independent personal pronoun (Pp…) in the verse.
+
+  Why this exists (anti-capture):
+    The Shema (Devarim 6:4-9, 11:13-21, Bamidbar 15:37-41) deliberately
+    oscillates between singular ("you, sg.") and plural ("you, pl.")
+    voice — and Devarim 11:14 famously switches mid-verse from plural
+    (אַרְצְכֶם, "your land") to singular (וְאָסַפְתָּ דְגָנֶךָ, "and YOU
+    [sg.] gather YOUR [sg.] grain"). Voice that stabilizes at either
+    pole is anti-Shema:
+      • stuck at second-singular → private capture (gnostic individualism)
+      • stuck at second-plural   → collective capture (mob theology)
+      • stuck at first-anything  → operator-as-deity (the calf)
+    The trajectory IS the anti-capture mechanism. The system records
+    voice per-verse so downstream stages can detect stabilization.
+
+  Returns:
+    %{
+      markers: [%{position, source, person, gender, number, morph}],
+      summary: %{voice_key => count},
+      dominant: voice_key | :mixed | :none,
+      oscillates_within_verse: boolean    # true iff both 2-sg AND 2-pl appear
+    }
+
+  voice_key is one of:
+    :first_singular, :first_plural,
+    :second_singular, :second_plural,
+    :third_singular, :third_plural
+  """
+  @spec extract_voice(verse()) :: map()
+  def extract_voice(verse) do
+    markers =
+      verse.words
+      |> Enum.flat_map(&voice_markers_in_word/1)
+
+    summary =
+      markers
+      |> Enum.frequencies_by(fn m -> voice_key(m.person, m.number) end)
+
+    oscillates =
+      Enum.any?(markers, fn m -> m.person == 2 and m.number == :singular end) and
+        Enum.any?(markers, fn m -> m.person == 2 and m.number == :plural end)
+
+    %{
+      markers: markers,
+      summary: summary,
+      dominant: dominant_voice(summary),
+      oscillates_within_verse: oscillates
+    }
+  end
+
   @doc """
   Extract enlarged letters (otiot gedolot) from a verse.
 
@@ -308,4 +363,92 @@ defmodule VavPipeline.L1Reader do
       acc + Map.get(@gematria_values, char, 0)
     end)
   end
+
+  # --- Voice profile (OSHB morphology code parser) ---
+  #
+  # OSHB morph format uses '/' to separate language-segment groups, e.g.
+  #   "HC/Vqq2ms"      — H = Hebrew, C = conjunction prefix, then Vqq2ms verb
+  #   "HNcmsc/Sp2ms"   — Hebrew noun construct, then 2ms pronominal suffix
+  #   "HC/Vqq2mp"      — verb 2nd masc plural
+  # We extract person/gender/number from segments starting with V (verb),
+  # Sp (pronominal suffix), or Pp (independent personal pronoun).
+
+  defp voice_markers_in_word(word) do
+    case word.morph do
+      nil -> []
+      "" -> []
+      morph ->
+        morph
+        |> String.split("/")
+        |> Enum.flat_map(fn seg ->
+          case parse_morph_segment(seg) do
+            nil -> []
+            parsed -> [Map.merge(parsed, %{position: word.position, morph: morph})]
+          end
+        end)
+    end
+  end
+
+  defp parse_morph_segment(seg) do
+    # OSHB: the first segment of a morph string is prefixed with H (Hebrew)
+    # or A (Aramaic) language indicator. Strip it before POS detection;
+    # the same `/`-segment can be e.g. "HVqi2mp" (תִּשְׁמְעוּ) — without
+    # stripping, it would miss every verse-initial verb in the Torah.
+    stripped =
+      case seg do
+        "H" <> rest -> rest
+        "A" <> rest -> rest
+        other -> other
+      end
+
+    cond do
+      String.starts_with?(stripped, "Sp") -> attempt_pgn(stripped, :suffix)
+      String.starts_with?(stripped, "Pp") -> attempt_pgn(stripped, :pronoun)
+      String.starts_with?(stripped, "V")  -> attempt_pgn(stripped, :verb)
+      true -> nil
+    end
+  end
+
+  defp attempt_pgn(seg, source) when byte_size(seg) >= 3 do
+    pgn = String.slice(seg, -3, 3)
+    case parse_pgn(pgn) do
+      {:ok, p, g, n} -> %{source: source, person: p, gender: g, number: n}
+      :error -> nil
+    end
+  end
+  defp attempt_pgn(_, _), do: nil
+
+  defp parse_pgn(<<p, g, n>>) when p in ?1..?3 and g in [?m, ?f, ?c] and n in [?s, ?p] do
+    person = p - ?0
+    gender =
+      case g do
+        ?m -> :masculine
+        ?f -> :feminine
+        ?c -> :common
+      end
+    number =
+      case n do
+        ?s -> :singular
+        ?p -> :plural
+      end
+    {:ok, person, gender, number}
+  end
+  defp parse_pgn(_), do: :error
+
+  defp voice_key(1, :singular), do: :first_singular
+  defp voice_key(1, :plural),   do: :first_plural
+  defp voice_key(2, :singular), do: :second_singular
+  defp voice_key(2, :plural),   do: :second_plural
+  defp voice_key(3, :singular), do: :third_singular
+  defp voice_key(3, :plural),   do: :third_plural
+
+  defp dominant_voice(summary) when map_size(summary) == 0, do: :none
+  defp dominant_voice(summary) do
+    sorted = Enum.sort_by(summary, fn {_k, c} -> -c end)
+    case sorted do
+      [{key, _}] -> key
+      [{key1, c1}, {_key2, c2} | _] when c1 > c2 -> key1
+      _ -> :mixed
+    end
+  end
 end

vav_pipeline/lib/vav_pipeline/shema.ex

@@ -24,6 +24,12 @@ defmodule VavPipeline.Shema do
           ketiv_qere: [map()],
           root_gematria: map(),
           smikhut: [map()],
+          # Voice profile — anti-capture: the Shema deliberately oscillates
+          # between "you sg." and "you pl." (e.g. Devarim 11:14 switches
+          # mid-verse from אַרְצְכֶם to וְאָסַפְתָּ דְגָנֶךָ). Stabilization
+          # at either pole is anti-Shema; the trajectory IS the protection.
+          # See L1Reader.extract_voice/1 for full rationale.
+          voice_profile: map(),
           # From cloud (Torah calendar + L1 integrity)
           cloud_state: Cloud.cloud_state(),
           # From manna
@@ -56,6 +62,7 @@ defmodule VavPipeline.Shema do
         ketiv_qere: L1Reader.extract_ketiv_qere(verse),
         root_gematria: L1Reader.extract_roots(verse),
         smikhut: L1Reader.extract_smikhut(verse),
+        voice_profile: L1Reader.extract_voice(verse),
         # Cloud state — derived from Torah (L1 integrity + calendar)
         cloud_state: cloud_state,
         # Manna — daily provision
@@ -84,6 +91,7 @@ defmodule VavPipeline.Shema do
         ketiv_qere: L1Reader.extract_ketiv_qere(verse),
         root_gematria: L1Reader.extract_roots(verse),
         smikhut: L1Reader.extract_smikhut(verse),
+        voice_profile: L1Reader.extract_voice(verse),
         cloud_state: cloud_state,
         daily_provision: manna,
         scope: scope,