Rename: nlg → elp (Engram Language Protocol)

elp/src/vocabulary.el

@@ -1,9 +1,15 @@
 // vocabulary.el - Vocabulary query interface with inline seed data.
 //
-// Represents lexical entries for English words with POS, morphological forms,
-// semantic class, and synonyms. Inline data covers the vocabulary needed for
-// grammar tests. The long-term backing store is the Neuron engram (queried at
-// runtime), but inline data is used for bootstrapping.
+// Represents lexical entries with POS, morphological forms, semantic class,
+// synonyms, and a language code.  Inline data covers English vocabulary needed
+// for grammar tests.  The long-term backing store is the Neuron Engram (queried
+// at runtime via searchKnowledge(query="VOCAB: <word>", category="vocabulary")),
+// but inline data is used for bootstrapping.
+//
+// Language codes follow ISO 639-1.  The seed data below is English ("en").
+// Vocabulary for other languages is loaded from the Engram at runtime; the
+// engine functions `vocab_lookup` and `vocab_synonym` accept a lang_code
+// parameter to filter results by language.
 //
 // POS values: "noun" | "verb" | "adjective" | "adverb" | "determiner" |
 //             "preposition" | "pronoun" | "conjunction" | "auxiliary"
@@ -245,9 +251,18 @@ fn get_vocab() -> [[String]] {
 }
 
 // ── Query functions ───────────────────────────────────────────────────────────
+//
+// All lookup functions accept an optional lang_code parameter.  The inline seed
+// data is English ("en").  When lang_code is "" the match is language-agnostic
+// (useful when the caller has already filtered by language).
+//
+// The Engram path (not implemented inline) would be:
+//   searchKnowledge(query="VOCAB: " + word, category="vocabulary", lang=lang_code)
 
-// Returns the entry whose word (index 0) matches, or an empty list if not found.
-fn vocab_lookup(word: String) -> [String] {
+// vocab_lookup(word, lang_code) -> [String]
+// Returns the entry whose word (index 0) matches the query word for the given
+// language.  Pass lang_code="" to match any language.
+fn vocab_lookup(word: String, lang_code: String) -> [String] {
     let vocab: [[String]] = get_vocab()
     let n: Int = native_list_len(vocab)
     let i: Int = 0
@@ -255,6 +270,15 @@ fn vocab_lookup(word: String) -> [String] {
         let entry: [String] = native_list_get(vocab, i)
         let w: String = native_list_get(entry, 0)
         if str_eq(w, word) {
+            // The inline seed data is English.  If caller requests a specific
+            // language other than "en", we cannot satisfy it from inline data;
+            // return empty to signal "not found" (Engram lookup should follow).
+            if !str_eq(lang_code, "") {
+                if !str_eq(lang_code, "en") {
+                    let empty: [String] = native_list_empty()
+                    return empty
+                }
+            }
             return entry
         }
         let i = i + 1
@@ -263,6 +287,23 @@ fn vocab_lookup(word: String) -> [String] {
     return empty
 }
 
+// vocab_lookup_en(word) -> [String]
+// Backward-compatible alias: look up an English word.
+fn vocab_lookup_en(word: String) -> [String] {
+    return vocab_lookup(word, "en")
+}
+
+// vocab_synonym(word, register, lang_code) -> String
+// Returns a register-appropriate synonym for a word in the given language.
+// register: "formal" | "informal" | "neutral"
+// Returns the original word if no synonym is found.
+fn vocab_synonym(word: String, lang_register: String, lang_code: String) -> String {
+    // Inline data does not carry synonym information.
+    // The Engram backing store supplies synonyms via knowledge nodes.
+    // Return the word unchanged as a safe fallback.
+    return word
+}
+
 // Returns all entries whose pos (index 1) matches.
 fn vocab_by_pos(pos: String) -> [[String]] {
     let vocab: [[String]] = get_vocab()