@lokascript/semantic 1.0.0

package/src/tokenizers/morphology/turkish-normalizer.ts

@@ -0,0 +1,412 @@
+/**
+ * Turkish Morphological Normalizer
+ *
+ * Turkish is a highly agglutinative language with strict vowel harmony.
+ * Suffixes attach in sequence and their vowels change based on the last
+ * vowel of the stem (front/back, rounded/unrounded).
+ *
+ * Vowel Harmony Rules:
+ * - Back vowels (a, ı, o, u) take back vowel suffixes
+ * - Front vowels (e, i, ö, ü) take front vowel suffixes
+ *
+ * Common verb suffixes:
+ * - Infinitive: -mak/-mek (değiştirmek = to change)
+ * - Present continuous: -iyor/-ıyor/-üyor/-uyor (değiştiriyor = is changing)
+ * - Past: -di/-dı/-dü/-du (değiştirdi = changed)
+ * - Reported past: -miş/-mış/-müş/-muş (değiştirmiş = apparently changed)
+ * - Future: -ecek/-acak (değiştirecek = will change)
+ * - Negation: -me/-ma before tense (değiştirmiyor = is not changing)
+ * - Passive: -il/-ıl/-ül/-ul (değiştirildi = was changed)
+ * - Causative: -tir/-tır/-tür/-tur (değiştirtmek = to make change)
+ *
+ * Person suffixes (after tense):
+ * - 1sg: -im/-ım/-üm/-um or -m (yapıyorum = I am doing)
+ * - 2sg: -sin/-sın/-sün/-sun (yapıyorsun = you are doing)
+ * - 3sg: (no suffix) (yapıyor = he/she is doing)
+ * - 1pl: -iz/-ız/-üz/-uz (yapıyoruz = we are doing)
+ * - 2pl: -siniz/-sınız/-sünüz/-sunuz (yapıyorsunuz = you all are doing)
+ * - 3pl: -ler/-lar (yapıyorlar = they are doing)
+ *
+ * Examples:
+ *   değiştiriyorum → değiştir (I am changing)
+ *   değiştirmek → değiştir (to change)
+ *   gösterdi → göster (showed)
+ *   gizleniyor → gizle (is being hidden)
+ */
+
+import type {
+  MorphologicalNormalizer,
+  NormalizationResult,
+  SuffixRule,
+  ConjugationType,
+} from './types';
+import { noChange, normalized } from './types';
+
+/**
+ * Check if a character is a Turkish letter.
+ * Turkish uses Latin alphabet with special characters: ç, ğ, ı, ö, ş, ü
+ */
+function isTurkishLetter(char: string): boolean {
+  const code = char.charCodeAt(0);
+  // Basic Latin letters
+  if ((code >= 0x41 && code <= 0x5a) || (code >= 0x61 && code <= 0x7a)) {
+    return true;
+  }
+  // Turkish special characters
+  const turkishChars = 'çÇğĞıİöÖşŞüÜ';
+  return turkishChars.includes(char);
+}
+
+/**
+ * Check if a word contains Turkish characters (including special chars).
+ */
+function containsTurkish(word: string): boolean {
+  for (const char of word) {
+    if (isTurkishLetter(char)) return true;
+  }
+  return false;
+}
+
+/**
+ * Check if a vowel is a back vowel.
+ */
+function isBackVowel(char: string): boolean {
+  return 'aıouAIOU'.includes(char);
+}
+
+/**
+ * Check if a vowel is a front vowel.
+ */
+function isFrontVowel(char: string): boolean {
+  return 'eiöüEİÖÜ'.includes(char);
+}
+
+/**
+ * Check if a character is a vowel.
+ */
+function isVowel(char: string): boolean {
+  return isBackVowel(char) || isFrontVowel(char);
+}
+
+/**
+ * Get the last vowel in a word.
+ */
+function getLastVowel(word: string): string | null {
+  for (let i = word.length - 1; i >= 0; i--) {
+    if (isVowel(word[i])) {
+      return word[i];
+    }
+  }
+  return null;
+}
+
+/**
+ * Check if a suffix matches vowel harmony with the stem.
+ * This helps validate that a potential suffix actually belongs.
+ */
+function matchesVowelHarmony(stem: string, suffix: string): boolean {
+  const stemLastVowel = getLastVowel(stem);
+  if (!stemLastVowel) return true; // No vowel in stem, can't validate
+
+  const suffixFirstVowel = suffix.split('').find(c => isVowel(c));
+  if (!suffixFirstVowel) return true; // No vowel in suffix, can't validate
+
+  // Back vowel stems take back vowel suffixes
+  if (isBackVowel(stemLastVowel)) {
+    return isBackVowel(suffixFirstVowel);
+  }
+  // Front vowel stems take front vowel suffixes
+  return isFrontVowel(suffixFirstVowel);
+}
+
+/**
+ * Suffix rules for Turkish verb conjugation.
+ * Each pattern includes all vowel harmony variants.
+ * Ordered by length (longest first) to ensure greedy matching.
+ */
+const TURKISH_SUFFIX_RULES: readonly SuffixRule[] = [
+  // Compound tense + person (longest patterns first)
+  // Present continuous + person
+  { pattern: 'iyorsunuz', confidence: 0.82, conjugationType: 'progressive', minStemLength: 2 },
+  { pattern: 'ıyorsunuz', confidence: 0.82, conjugationType: 'progressive', minStemLength: 2 },
+  { pattern: 'üyorsunuz', confidence: 0.82, conjugationType: 'progressive', minStemLength: 2 },
+  { pattern: 'uyorsunuz', confidence: 0.82, conjugationType: 'progressive', minStemLength: 2 },
+  { pattern: 'iyorsun', confidence: 0.82, conjugationType: 'progressive', minStemLength: 2 },
+  { pattern: 'ıyorsun', confidence: 0.82, conjugationType: 'progressive', minStemLength: 2 },
+  { pattern: 'üyorsun', confidence: 0.82, conjugationType: 'progressive', minStemLength: 2 },
+  { pattern: 'uyorsun', confidence: 0.82, conjugationType: 'progressive', minStemLength: 2 },
+  { pattern: 'iyoruz', confidence: 0.82, conjugationType: 'progressive', minStemLength: 2 },
+  { pattern: 'ıyoruz', confidence: 0.82, conjugationType: 'progressive', minStemLength: 2 },
+  { pattern: 'üyoruz', confidence: 0.82, conjugationType: 'progressive', minStemLength: 2 },
+  { pattern: 'uyoruz', confidence: 0.82, conjugationType: 'progressive', minStemLength: 2 },
+  { pattern: 'iyorum', confidence: 0.85, conjugationType: 'progressive', minStemLength: 2 },
+  { pattern: 'ıyorum', confidence: 0.85, conjugationType: 'progressive', minStemLength: 2 },
+  { pattern: 'üyorum', confidence: 0.85, conjugationType: 'progressive', minStemLength: 2 },
+  { pattern: 'uyorum', confidence: 0.85, conjugationType: 'progressive', minStemLength: 2 },
+  { pattern: 'iyorlar', confidence: 0.82, conjugationType: 'progressive', minStemLength: 2 },
+  { pattern: 'ıyorlar', confidence: 0.82, conjugationType: 'progressive', minStemLength: 2 },
+  { pattern: 'üyorlar', confidence: 0.82, conjugationType: 'progressive', minStemLength: 2 },
+  { pattern: 'uyorlar', confidence: 0.82, conjugationType: 'progressive', minStemLength: 2 },
+
+  // Future tense + person
+  { pattern: 'eceksiniz', confidence: 0.82, conjugationType: 'future', minStemLength: 2 },
+  { pattern: 'acaksınız', confidence: 0.82, conjugationType: 'future', minStemLength: 2 },
+  { pattern: 'eceksin', confidence: 0.82, conjugationType: 'future', minStemLength: 2 },
+  { pattern: 'acaksın', confidence: 0.82, conjugationType: 'future', minStemLength: 2 },
+  { pattern: 'eceğiz', confidence: 0.82, conjugationType: 'future', minStemLength: 2 },
+  { pattern: 'acağız', confidence: 0.82, conjugationType: 'future', minStemLength: 2 },
+  { pattern: 'eceğim', confidence: 0.85, conjugationType: 'future', minStemLength: 2 },
+  { pattern: 'acağım', confidence: 0.85, conjugationType: 'future', minStemLength: 2 },
+  { pattern: 'ecekler', confidence: 0.82, conjugationType: 'future', minStemLength: 2 },
+  { pattern: 'acaklar', confidence: 0.82, conjugationType: 'future', minStemLength: 2 },
+
+  // Reported past + person
+  { pattern: 'mişsiniz', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'mışsınız', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'müşsünüz', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'muşsunuz', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'mişsin', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'mışsın', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'müşsün', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'muşsun', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'mişiz', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'mışız', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'müşüz', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'muşuz', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'mişim', confidence: 0.85, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'mışım', confidence: 0.85, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'müşüm', confidence: 0.85, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'muşum', confidence: 0.85, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'mişler', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'mışlar', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'müşler', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'muşlar', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+
+  // Past tense + person
+  { pattern: 'diniz', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'dınız', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'dünüz', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'dunuz', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'tiniz', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'tınız', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'tünüz', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'tunuz', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'diler', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'dılar', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'düler', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'dular', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'tiler', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'tılar', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'tüler', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'tular', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'din', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'dın', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'dün', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'dun', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'tin', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'tın', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'tün', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'tun', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'dik', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'dık', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'dük', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'duk', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'tik', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'tık', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'tük', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'tuk', confidence: 0.82, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'dim', confidence: 0.85, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'dım', confidence: 0.85, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'düm', confidence: 0.85, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'dum', confidence: 0.85, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'tim', confidence: 0.85, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'tım', confidence: 0.85, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'tüm', confidence: 0.85, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'tum', confidence: 0.85, conjugationType: 'past', minStemLength: 2 },
+
+  // Present continuous (no person - 3rd person singular)
+  { pattern: 'iyor', confidence: 0.85, conjugationType: 'progressive', minStemLength: 2 },
+  { pattern: 'ıyor', confidence: 0.85, conjugationType: 'progressive', minStemLength: 2 },
+  { pattern: 'üyor', confidence: 0.85, conjugationType: 'progressive', minStemLength: 2 },
+  { pattern: 'uyor', confidence: 0.85, conjugationType: 'progressive', minStemLength: 2 },
+
+  // Future (no person - 3rd person singular)
+  { pattern: 'ecek', confidence: 0.85, conjugationType: 'future', minStemLength: 2 },
+  { pattern: 'acak', confidence: 0.85, conjugationType: 'future', minStemLength: 2 },
+
+  // Reported past (no person - 3rd person singular)
+  { pattern: 'miş', confidence: 0.85, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'mış', confidence: 0.85, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'müş', confidence: 0.85, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'muş', confidence: 0.85, conjugationType: 'past', minStemLength: 2 },
+
+  // Simple past (no person - 3rd person singular)
+  { pattern: 'di', confidence: 0.85, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'dı', confidence: 0.85, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'dü', confidence: 0.85, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'du', confidence: 0.85, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'ti', confidence: 0.85, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'tı', confidence: 0.85, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'tü', confidence: 0.85, conjugationType: 'past', minStemLength: 2 },
+  { pattern: 'tu', confidence: 0.85, conjugationType: 'past', minStemLength: 2 },
+
+  // Infinitive
+  { pattern: 'mek', confidence: 0.88, conjugationType: 'dictionary', minStemLength: 2 },
+  { pattern: 'mak', confidence: 0.88, conjugationType: 'dictionary', minStemLength: 2 },
+
+  // Optative mood (let me/us...) - -eyim/-ayım/-elim/-alım
+  { pattern: 'eyelim', confidence: 0.82, conjugationType: 'optative', minStemLength: 2 },
+  { pattern: 'ayalım', confidence: 0.82, conjugationType: 'optative', minStemLength: 2 },
+  { pattern: 'eyim', confidence: 0.82, conjugationType: 'optative', minStemLength: 2 },
+  { pattern: 'ayım', confidence: 0.82, conjugationType: 'optative', minStemLength: 2 },
+  { pattern: 'elim', confidence: 0.82, conjugationType: 'optative', minStemLength: 2 },
+  { pattern: 'alım', confidence: 0.82, conjugationType: 'optative', minStemLength: 2 },
+
+  // Necessitative (must/should) - -meli/-malı
+  { pattern: 'melisiniz', confidence: 0.82, conjugationType: 'necessitative', minStemLength: 2 },
+  { pattern: 'malısınız', confidence: 0.82, conjugationType: 'necessitative', minStemLength: 2 },
+  { pattern: 'melisin', confidence: 0.82, conjugationType: 'necessitative', minStemLength: 2 },
+  { pattern: 'malısın', confidence: 0.82, conjugationType: 'necessitative', minStemLength: 2 },
+  { pattern: 'meliyiz', confidence: 0.82, conjugationType: 'necessitative', minStemLength: 2 },
+  { pattern: 'malıyız', confidence: 0.82, conjugationType: 'necessitative', minStemLength: 2 },
+  { pattern: 'meliyim', confidence: 0.85, conjugationType: 'necessitative', minStemLength: 2 },
+  { pattern: 'malıyım', confidence: 0.85, conjugationType: 'necessitative', minStemLength: 2 },
+  { pattern: 'meliler', confidence: 0.82, conjugationType: 'necessitative', minStemLength: 2 },
+  { pattern: 'malılar', confidence: 0.82, conjugationType: 'necessitative', minStemLength: 2 },
+  { pattern: 'meli', confidence: 0.85, conjugationType: 'necessitative', minStemLength: 2 },
+  { pattern: 'malı', confidence: 0.85, conjugationType: 'necessitative', minStemLength: 2 },
+
+  // Ability (can) - -ebil/-abil + tense suffixes
+  { pattern: 'ebiliyor', confidence: 0.82, conjugationType: 'potential', minStemLength: 2 },
+  { pattern: 'abiliyor', confidence: 0.82, conjugationType: 'potential', minStemLength: 2 },
+  { pattern: 'ebilir', confidence: 0.85, conjugationType: 'potential', minStemLength: 2 },
+  { pattern: 'abilir', confidence: 0.85, conjugationType: 'potential', minStemLength: 2 },
+  { pattern: 'ebildi', confidence: 0.82, conjugationType: 'potential', minStemLength: 2 },
+  { pattern: 'abildi', confidence: 0.82, conjugationType: 'potential', minStemLength: 2 },
+  { pattern: 'ebilmek', confidence: 0.85, conjugationType: 'potential', minStemLength: 2 },
+  { pattern: 'abilmek', confidence: 0.85, conjugationType: 'potential', minStemLength: 2 },
+
+  // Imperative (2nd person singular is just stem, 2nd person plural has suffix)
+  { pattern: 'iniz', confidence: 0.82, conjugationType: 'imperative', minStemLength: 2 },
+  { pattern: 'ınız', confidence: 0.82, conjugationType: 'imperative', minStemLength: 2 },
+  { pattern: 'ünüz', confidence: 0.82, conjugationType: 'imperative', minStemLength: 2 },
+  { pattern: 'unuz', confidence: 0.82, conjugationType: 'imperative', minStemLength: 2 },
+  { pattern: 'in', confidence: 0.8, conjugationType: 'imperative', minStemLength: 2 },
+  { pattern: 'ın', confidence: 0.8, conjugationType: 'imperative', minStemLength: 2 },
+  { pattern: 'ün', confidence: 0.8, conjugationType: 'imperative', minStemLength: 2 },
+  { pattern: 'un', confidence: 0.8, conjugationType: 'imperative', minStemLength: 2 },
+
+  // Passive voice
+  { pattern: 'ildi', confidence: 0.82, conjugationType: 'passive', minStemLength: 2 },
+  { pattern: 'ıldı', confidence: 0.82, conjugationType: 'passive', minStemLength: 2 },
+  { pattern: 'üldü', confidence: 0.82, conjugationType: 'passive', minStemLength: 2 },
+  { pattern: 'uldu', confidence: 0.82, conjugationType: 'passive', minStemLength: 2 },
+  { pattern: 'ilir', confidence: 0.82, conjugationType: 'passive', minStemLength: 2 },
+  { pattern: 'ılır', confidence: 0.82, conjugationType: 'passive', minStemLength: 2 },
+  { pattern: 'ülür', confidence: 0.82, conjugationType: 'passive', minStemLength: 2 },
+  { pattern: 'ulur', confidence: 0.82, conjugationType: 'passive', minStemLength: 2 },
+
+  // Causative
+  { pattern: 'tirmek', confidence: 0.82, conjugationType: 'causative', minStemLength: 2 },
+  { pattern: 'tırmak', confidence: 0.82, conjugationType: 'causative', minStemLength: 2 },
+  { pattern: 'türmek', confidence: 0.82, conjugationType: 'causative', minStemLength: 2 },
+  { pattern: 'turmak', confidence: 0.82, conjugationType: 'causative', minStemLength: 2 },
+  { pattern: 'dirmek', confidence: 0.82, conjugationType: 'causative', minStemLength: 2 },
+  { pattern: 'dırmak', confidence: 0.82, conjugationType: 'causative', minStemLength: 2 },
+  { pattern: 'dürmek', confidence: 0.82, conjugationType: 'causative', minStemLength: 2 },
+  { pattern: 'durmak', confidence: 0.82, conjugationType: 'causative', minStemLength: 2 },
+
+  // Negation + tense combinations (very common)
+  { pattern: 'miyorsunuz', confidence: 0.8, conjugationType: 'negative', minStemLength: 2 },
+  { pattern: 'mıyorsunuz', confidence: 0.8, conjugationType: 'negative', minStemLength: 2 },
+  { pattern: 'müyorsunuz', confidence: 0.8, conjugationType: 'negative', minStemLength: 2 },
+  { pattern: 'muyorsunuz', confidence: 0.8, conjugationType: 'negative', minStemLength: 2 },
+  { pattern: 'miyorsun', confidence: 0.8, conjugationType: 'negative', minStemLength: 2 },
+  { pattern: 'mıyorsun', confidence: 0.8, conjugationType: 'negative', minStemLength: 2 },
+  { pattern: 'müyorsun', confidence: 0.8, conjugationType: 'negative', minStemLength: 2 },
+  { pattern: 'muyorsun', confidence: 0.8, conjugationType: 'negative', minStemLength: 2 },
+  { pattern: 'miyoruz', confidence: 0.8, conjugationType: 'negative', minStemLength: 2 },
+  { pattern: 'mıyoruz', confidence: 0.8, conjugationType: 'negative', minStemLength: 2 },
+  { pattern: 'müyoruz', confidence: 0.8, conjugationType: 'negative', minStemLength: 2 },
+  { pattern: 'muyoruz', confidence: 0.8, conjugationType: 'negative', minStemLength: 2 },
+  { pattern: 'miyorum', confidence: 0.82, conjugationType: 'negative', minStemLength: 2 },
+  { pattern: 'mıyorum', confidence: 0.82, conjugationType: 'negative', minStemLength: 2 },
+  { pattern: 'müyorum', confidence: 0.82, conjugationType: 'negative', minStemLength: 2 },
+  { pattern: 'muyorum', confidence: 0.82, conjugationType: 'negative', minStemLength: 2 },
+  { pattern: 'miyor', confidence: 0.82, conjugationType: 'negative', minStemLength: 2 },
+  { pattern: 'mıyor', confidence: 0.82, conjugationType: 'negative', minStemLength: 2 },
+  { pattern: 'müyor', confidence: 0.82, conjugationType: 'negative', minStemLength: 2 },
+  { pattern: 'muyor', confidence: 0.82, conjugationType: 'negative', minStemLength: 2 },
+  { pattern: 'medi', confidence: 0.82, conjugationType: 'negative', minStemLength: 2 },
+  { pattern: 'madı', confidence: 0.82, conjugationType: 'negative', minStemLength: 2 },
+  { pattern: 'me', confidence: 0.75, conjugationType: 'negative', minStemLength: 3 },
+  { pattern: 'ma', confidence: 0.75, conjugationType: 'negative', minStemLength: 3 },
+];
+
+/**
+ * Turkish morphological normalizer.
+ */
+export class TurkishMorphologicalNormalizer implements MorphologicalNormalizer {
+  readonly language = 'tr';
+
+  /**
+   * Check if a word might be a Turkish verb that can be normalized.
+   */
+  isNormalizable(word: string): boolean {
+    // Must contain Turkish characters
+    if (!containsTurkish(word)) return false;
+
+    // Must be at least 3 characters (Turkish verb stems are usually 2+ chars)
+    if (word.length < 3) return false;
+
+    return true;
+  }
+
+  /**
+   * Normalize a Turkish word to its stem form.
+   */
+  normalize(word: string): NormalizationResult {
+    // Convert to lowercase for matching
+    const lowerWord = word.toLowerCase();
+
+    // Try suffix rules
+    for (const rule of TURKISH_SUFFIX_RULES) {
+      if (lowerWord.endsWith(rule.pattern)) {
+        const stem = lowerWord.slice(0, -rule.pattern.length);
+
+        // Validate stem length
+        const minLength = rule.minStemLength ?? 2;
+        if (stem.length < minLength) continue;
+
+        // Validate vowel harmony (optional, can help avoid false positives)
+        if (!matchesVowelHarmony(stem, rule.pattern)) {
+          // Lower confidence if vowel harmony doesn't match
+          // but still allow it since there are exceptions
+          const adjustedConfidence = rule.confidence * 0.9;
+
+          const metadata: { removedSuffixes: string[]; conjugationType?: ConjugationType } = {
+            removedSuffixes: [rule.pattern],
+          };
+          if (rule.conjugationType) {
+            metadata.conjugationType = rule.conjugationType;
+          }
+          return normalized(stem, adjustedConfidence, metadata);
+        }
+
+        const metadata: { removedSuffixes: string[]; conjugationType?: ConjugationType } = {
+          removedSuffixes: [rule.pattern],
+        };
+        if (rule.conjugationType) {
+          metadata.conjugationType = rule.conjugationType;
+        }
+        return normalized(stem, rule.confidence, metadata);
+      }
+    }
+
+    // No normalization needed
+    return noChange(word);
+  }
+}
+
+// Export singleton instance
+export const turkishMorphologicalNormalizer = new TurkishMorphologicalNormalizer();

package/src/tokenizers/morphology/types.ts

@@ -0,0 +1,211 @@
+/**
+ * Morphological Normalizer Types
+ *
+ * Defines interfaces for language-specific morphological analysis.
+ * Normalizers reduce conjugated/inflected forms to canonical stems
+ * that can be matched against keyword dictionaries.
+ */
+
+/**
+ * Result of morphological normalization.
+ */
+export interface NormalizationResult {
+  /** The extracted stem/root form */
+  readonly stem: string;
+
+  /** Confidence in the normalization (0.0-1.0) */
+  readonly confidence: number;
+
+  /** Optional metadata about the transformation */
+  readonly metadata?: NormalizationMetadata;
+}
+
+/**
+ * Metadata about morphological transformations applied.
+ */
+export interface NormalizationMetadata {
+  /** Prefixes that were removed */
+  readonly removedPrefixes?: readonly string[];
+
+  /** Suffixes that were removed */
+  readonly removedSuffixes?: readonly string[];
+
+  /** Type of conjugation detected */
+  readonly conjugationType?: ConjugationType;
+
+  /** Original form classification */
+  readonly originalForm?: string;
+
+  /** Applied transformation rules (for debugging) */
+  readonly appliedRules?: readonly string[];
+}
+
+/**
+ * Types of verb conjugation/inflection.
+ */
+export type ConjugationType =
+  // Tense
+  | 'present'
+  | 'past'
+  | 'future'
+  | 'progressive'
+  | 'perfect'
+  // Mood
+  | 'imperative'
+  | 'subjunctive'
+  | 'conditional'
+  // Voice
+  | 'passive'
+  | 'causative'
+  // Politeness (Japanese/Korean)
+  | 'polite'
+  | 'humble'
+  | 'honorific'
+  // Form
+  | 'negative'
+  | 'potential'
+  | 'volitional'
+  // Japanese conditional forms
+  | 'conditional-tara' // たら/したら - if/when (completed action)
+  | 'conditional-to' // と/すると - when (habitual/expected)
+  | 'conditional-ba' // ば/すれば - if (hypothetical)
+  // Korean-specific
+  | 'connective' // 하고, 해서 etc.
+  | 'conditional-myeon' // -(으)면 - if/when (general conditional)
+  | 'temporal-ttae' // -(으)ㄹ 때 - when (at the time of)
+  | 'causal-nikka' // -(으)니까 - because/since
+  // Korean honorific forms (-시- infix)
+  | 'honorific-conditional' // -하시면 - if (honorific)
+  | 'honorific-temporal' // -하실 때 - when (honorific)
+  | 'honorific-causal' // -하시니까 - because (honorific)
+  | 'honorific-past' // -하셨어요 - past (honorific)
+  | 'honorific-polite' // -하십니다 - polite (honorific)
+  // Korean sequential forms
+  | 'sequential-after' // -고 나서 - after doing
+  | 'sequential-before' // -기 전에 - before doing
+  | 'immediate' // -자마자 - as soon as
+  | 'obligation' // -아야/어야 해 - must do, should do
+  // Spanish-specific
+  | 'reflexive'
+  | 'reflexive-imperative'
+  | 'gerund'
+  | 'participle'
+  // Arabic-specific
+  | 'conditional-idha' // إذا - if/when (hypothetical)
+  | 'temporal-indama' // عندما - when (temporal conjunction)
+  | 'temporal-hina' // حين - at the time of
+  | 'temporal-lamma' // لمّا - when (past emphasis)
+  | 'past-verb' // فعل ماضي - past tense verb
+  // Turkish-specific
+  | 'conditional-se' // -se/-sa - if (hypothetical)
+  | 'temporal-ince' // -ince/-ınca/-unca/-ünce - when/as
+  | 'temporal-dikce' // -dikçe/-dıkça/-dukça/-dükçe - as/while
+  | 'aorist' // -ir/-ar - habitual/general
+  | 'optative' // -eyim/-ayım/-elim/-alım - let me/us
+  | 'necessitative' // -meli/-malı - must/should
+  // Japanese request/contracted forms
+  | 'request' // てください/でください - polite request
+  | 'casual-request' // てくれ/でくれ - casual request
+  | 'contracted' // ちゃう/じゃう - contracted completion (てしまう)
+  | 'contracted-past' // ちゃった/じゃった - contracted past completion
+  // Compound
+  | 'compound' // Multi-layer suffixes (ていなかった, 하고나서였어)
+  | 'te-form' // Japanese て-form
+  | 'dictionary'; // Base/infinitive form
+
+/**
+ * Interface for language-specific morphological normalizers.
+ *
+ * Normalizers attempt to reduce inflected word forms to their
+ * canonical stems. This enables matching conjugated verbs against
+ * keyword dictionaries that only contain base forms.
+ *
+ * Example (Japanese):
+ *   切り替えた (past) → { stem: '切り替え', confidence: 0.85 }
+ *   切り替えます (polite) → { stem: '切り替え', confidence: 0.85 }
+ *
+ * Example (Spanish):
+ *   mostrarse (reflexive infinitive) → { stem: 'mostrar', confidence: 0.85 }
+ *   alternando (gerund) → { stem: 'alternar', confidence: 0.85 }
+ */
+export interface MorphologicalNormalizer {
+  /** Language code this normalizer handles */
+  readonly language: string;
+
+  /**
+   * Normalize a word to its canonical stem form.
+   *
+   * @param word - The word to normalize
+   * @returns Normalization result with stem and confidence
+   */
+  normalize(word: string): NormalizationResult;
+
+  /**
+   * Check if a word appears to be a verb form that can be normalized.
+   * Optional optimization to skip normalization for non-verb tokens.
+   *
+   * @param word - The word to check
+   * @returns true if the word might be a normalizable verb form
+   */
+  isNormalizable?(word: string): boolean;
+}
+
+/**
+ * Configuration for suffix-based normalization rules.
+ * Used by agglutinative languages (Japanese, Korean, Turkish).
+ */
+export interface SuffixRule {
+  /** The suffix pattern to match */
+  readonly pattern: string;
+
+  /** Confidence when this suffix is stripped */
+  readonly confidence: number;
+
+  /** What to replace the suffix with (empty string for simple removal) */
+  readonly replacement?: string;
+
+  /** Conjugation type this suffix indicates */
+  readonly conjugationType?: ConjugationType;
+
+  /** Minimum stem length after stripping (to avoid over-stripping) */
+  readonly minStemLength?: number;
+}
+
+/**
+ * Configuration for prefix-based normalization rules.
+ * Used primarily by Arabic for article/conjunction prefixes.
+ */
+export interface PrefixRule {
+  /** The prefix pattern to match */
+  readonly pattern: string;
+
+  /** Confidence penalty when this prefix is stripped */
+  readonly confidencePenalty: number;
+
+  /** What the prefix indicates (for metadata) */
+  readonly prefixType?: 'article' | 'conjunction' | 'preposition' | 'verb-marker';
+
+  /** Minimum remaining characters after stripping (to avoid over-stripping) */
+  readonly minRemaining?: number;
+}
+
+/**
+ * Helper to create a "no change" normalization result.
+ */
+export function noChange(word: string): NormalizationResult {
+  return { stem: word, confidence: 1.0 };
+}
+
+/**
+ * Helper to create a normalization result with metadata.
+ */
+export function normalized(
+  stem: string,
+  confidence: number,
+  metadata?: NormalizationMetadata
+): NormalizationResult {
+  if (metadata) {
+    return { stem, confidence, metadata };
+  }
+  return { stem, confidence };
+}