diff --git a/glossy/src/utils.typ b/glossy/src/utils.typ
index 0acab16..ea57668 100644
--- a/glossy/src/utils.typ
+++ b/glossy/src/utils.typ
@@ -36,3 +36,128 @@
     word + "s"
   }
 }
+
+// Determines the appropriate indefinite article ("a" or "an") for a given English word.
+//
+// Heuristic rules:
+// 1. If word looks like an acronym (all uppercase or letters separated by dots):
+//    - Use "an" if the first letter's name starts with a vowel sound, e.g. A (ay), E (ee), F (eff), H (aitch), I (eye), L (el), M (em), N (en), O (oh), R (ar), S (ess), X (ex).
+//    - Otherwise "a".
+//
+// 2. Otherwise, lowercase the word and apply standard rules:
+//    - Starts with a vowel (a, e, i, o): "an".
+//    - Starts with "u":
+//      * If it begins with "uni", "use", "user", or "eu" (e.g. "university", "use", "Europe"), we treat it as a "yoo" sound -> "a".
+//      * Otherwise -> "an".
+//    - Special silent 'h' words: if it starts with "hour", "honest", "honor", "honour", or other, use "an".
+//      Otherwise if starts with 'h', use "a".
+//    - Default: "a".
+//
+// Notes:
+// - This is a best-effort heuristic and won't be perfect.
+// - These are for singular forms. If plural form is requested alongside
+//   articles, the caller should panic before calling this function.
+//
+// Returns:
+//   "a" or "an"
+//
+// Example:
+//   __determine_article("RF") -> "an" (Ar-Ef)
+//   __determine_article("radio") -> "a"
+//   __determine_article("hour") -> "an"
+//   __determine_article("honest") -> "an"
+//   __determine_article("university") -> "a"
+//   __determine_article("umbrella") -> "an"
+//   __determine_article("Europe") -> "a" (because we assume "yoo-ruhp")
+#let __determine_article(word) = {
+  // Check input
+  if word == none or word.trim() == "" {
+    // If invalid, default to none
+    return none
+  }
+
+  // Normalize the word
+  let upper_word = word.trim()
+  let lower_word = lower(upper_word)
+
+  // When these letters are spelled out, they get an "an" (i.e. "an RFC")
+  let acronym_vowels = (
+    "A", // "ay"
+    "E", // "ee"
+    "F", // "eff"
+    "H", // "aitch"
+    "I", // "eye"
+    "L", // "el"
+    "M", // "em"
+    "N", // "en"
+    "O", // "oh"
+    "R", // "ar"
+    "S", // "ess"
+    "X", // "ex"
+  )
+
+  // Helper to detect if word looks like an acronym
+  let is_acronym = {
+    // All uppercase or letters separated by non-alphabetical chars
+    let cleaned = upper_word.replace(".", "").replace("-", "")
+    cleaned.codepoints().all(c => c == upper(c)) and cleaned.len() > 0
+  }
+
+  if is_acronym {
+    // Use acronym rules
+    let first_char = upper_word.slice(0,1)
+    if acronym_vowels.contains(first_char) {
+      "an"
+    } else {
+      "a"
+    }
+  } else {
+    // Non-acronym rules
+    let first_char = lower_word.slice(0,1)
+
+    // Special cases for words matching a set of known 'silent h' prefixes
+    let silent_h_words = (
+      "heir",   // "air"
+      "herb",   // "erb" (in American English)
+      "homage", // "oh-mij" or "ah-mij"
+      "honest", // "on-est"
+      "honor",  // "on-or" (American English)
+      "honour", // "on-our" (British English)
+      "hour",   // "our"
+    )
+    if silent_h_words.any(w => lower_word.starts-with(w)) {
+      return "an"
+    }
+
+    // Words starting with 'eu', which has the 'yoo' sound
+    if lower_word.starts-with("eu") {
+      return "a"
+    }
+
+    // Words starting with 'u', that get the 'yoo' sound
+    if first_char == "u" {
+      let yoo_sounds = (
+        "ubi",  // ubiquitous
+        "uni",  // unicycle, uniform, universe, university, unilateral, unique, union, etc.
+        "usa",  // usage, usability
+        "use",  // use, useful, useless
+        "usu",  // usual, usually
+        "usur", // usurp
+        "ute",  // utensil, uterus
+        "uti",  // utility, utilize, utilization
+        "uto",  // utopia, utopian
+      )
+      if yoo_sounds.any(w => lower_word.starts-with(w)) {
+        return "a"
+      }
+    }
+
+    // Vowel start
+    if "aeiou".contains(first_char) {
+      return "an"
+    }
+
+    // the default case, plus 'h' words not in the silent list
+    "a"
+  }
+}
diff --git a/glossy/tests/determine-article/ref/1.png b/glossy/tests/determine-article/ref/1.png
new file mode 100644
index 0000000..d16c201
Binary files /dev/null and b/glossy/tests/determine-article/ref/1.png differ
diff --git a/glossy/tests/determine-article/test.typ b/glossy/tests/determine-article/test.typ
new file mode 100644
index 0000000..6440e4c
--- /dev/null
+++ b/glossy/tests/determine-article/test.typ
@@ -0,0 +1,289 @@
+#import "/src/utils.typ": *
+
+#let tests = (
+  // Main tests (covers most of English language)
+  ("Europe","a"),
+  ("European","a"),
+  ("Hannah","a"),
+  ("T-shirt","a"),
+  ("Xerox","a"),
+  ("Zebra","a"),
+  ("aardvark","an"),
+  ("apple","an"),
+  ("boat","a"),
+  ("car","a"),
+  ("cat","a"),
+  ("chair","a"),
+  ("dog","a"),
+  ("egg","an"),
+  ("elephant","an"),
+  ("emu","an"),
+  ("eucalyptus","a"),
+  ("eulogy","a"),
+  ("euphemism","a"),
+  ("exhausted","an"),
+  ("ghost","a"),
+  ("hat","a"),
+  ("heir","an"),
+  ("heiress","an"),
+  ("herb","an"),
+  ("herbal","an"),
+  ("historic","a"),
+  ("history","a"),
+  ("homage","an"),
+  ("honest","an"),
+  ("honestly","an"),
+  ("honor","an"),
+  ("honorary","an"),
+  ("honour","an"),
+  ("horse","a"),
+  ("hour","an"),
+  ("hourglass","an"),
+  ("house","a"),
+  ("iris","an"),
+  ("orange","an"),
+  ("owl","an"),
+  ("table","a"),
+  ("ugly","an"),
+  ("umbrella","an"),
+  ("umpire","an"),
+  ("unicorn","a"),
+  ("unique","a"),
+  ("university","a"),
+  ("urn","an"),
+  ("usage","a"),
+  ("user","a"),
+  ("utensil","a"),
+
+  // Acronym tests
+  ("ASCII","an"),
+  ("CFO","a"),
+  ("F.A.Q.","an"),
+  ("FBI","an"),
+  ("GIF","a"),
+  ("HIV","an"),
+  ("JPEG","a"),
+  ("LED","an"),
+  ("MRI","an"),
+  ("NATO","an"),
+  ("PNG","a"),
+  ("Q&A","a"),
+  ("RF","an"),
+  ("RNG","an"),
+  ("SAS","an"),
+  ("SMS","an"),
+  ("SOS","an"),
+  ("SQL","an"),
+  ("UFO","a"),
+  ("URL","a"),
+  ("VCR","a"),
+  ("WWW","a"),
+  ("X.Y.Z.","an"),
+  ("YAML","a"),
+
+  // Extra tests (courtesy of LLM)
+
+  // Common English words starting with vowels
+  ("idea","an"),
+  ("island","an"),
+  ("offer","an"),
+  ("option","an"),
+  ("operation","an"),
+  ("appletree","an"),
+  ("iguana","an"),
+  ("octopus","an"),
+  ("atom","an"),
+  ("item","an"),
+
+  // Words starting with vowel but "yoo" sound (thus 'a')
+  ("universe","a"),
+  ("unicycle","a"),
+  ("uniform","a"),
+  ("union","a"),
+  ("united","a"),
+  ("unique","a"), // already tested, reaffirmed
+  ("usual","a"),
+  ("useful","a"),
+  ("utility","a"),
+  ("utopia","a"),
+
+  // More common vowel words with a/an
+  ("umbrella","an"), // repeated but reaffirming
+  ("eggplant","an"),
+  ("insect","an"),
+  ("eleven","an"),
+  ("olive","an"),
+  ("impulse","an"),
+  ("ox","an"),
+  ("aroma","an"),
+  ("ear","an"),
+  ("ocean","an"),
+
+  // Words starting with 'h' but pronounced, so 'a'
+  ("house","a"), // repeated for emphasis
+  ("hammer","a"),
+  ("helmet","a"),
+  ("history","a"), // repeated
+  ("hero","a"),
+  ("hill","a"),
+  ("hobby","a"),
+  ("holiday","a"),
+  ("harmony","a"),
+  ("hover","a"),
+
+  // More silent 'h' words (already in set, but variations)
+  ("honorific","an"),
+  ("honorarium","an"),
+  ("honesties","an"),
+  ("hourly","an"),
+  ("heirloom","an"),
+
+  // Words starting with 'e' + 'u' combo with "yoo" sound
+  ("Europe","a"),    // repeated
+  ("European","a"),  // repeated
+  ("euphoric","a"),
+  ("eureka","a"),
+  ("euphemistic","a"),
+  ("euphonious","a"),
+  ("eucalyptus","a"), // repeated
+  ("euthanasia","a"),
+  ("eugenics","a"),
+  ("Eurasia","a"),
+
+  // Common consonant-start words
+  ("book","a"),
+  ("pen","a"),
+  ("fork","a"),
+  ("spoon","a"),
+  ("knife","a"),
+  ("ball","a"),
+  ("tree","a"),
+  ("river","a"),
+  ("mountain","a"),
+  ("cloud","a"),
+
+  // More vowel-start words
+  ("ostrich","an"),
+  ("amphibian","an"),
+  ("eleven","an"), // repeated
+  ("issue","an"),
+  ("incense","an"),
+  ("antenna","an"),
+  ("echo","an"),
+  ("aisle","an"),
+  ("oracle","an"),
+  ("overcoat","an"),
+
+  // More acronym/initialism tests:
+  ("ATM","an"),  // A = 'ay'
+  ("EPA","an"),  // E = 'ee'
+  ("FAQ","an"),  // F = 'eff', repeated as F.A.Q.
+  ("HTML","an"), // H = 'aitch'
+  ("IBM","an"),  // I = 'eye'
+  ("LAN","an"),  // L = 'el'
+  ("NFL","an"),  // N = 'en'
+  ("OSHA","an"), // O = 'oh'
+  ("RAM","an"),  // R = 'ar'
+  ("SMS","an"),  // repeated, S = 'ess'
+
+  // Acronyms starting with consonant sound letters
+  ("GIF","a"),  // G = 'jee', repeated
+  ("CFO","a"),  // C = 'see', repeated (consonant sound 's')
+  ("CEO","a"),  // C = 'see'
+  ("DNA","a"),  // D = 'dee' starts with 'd' consonant sound
+  ("PGA","a"),  // P = 'pee' starts with 'p' consonant sound
+  ("QVC","a"),  // Q = 'kyoo' starts with 'k' consonant sound
+  ("TNT","a"),  // T = 'tee' starts with 't' consonant sound
+  ("UPS","a"),  // U = 'yoo' sound, hence 'a'
+  ("VIP","a"),  // V = 'vee' starts with 'v' consonant sound
+  ("WIPO","a"), // W = 'double-u' starts with 'd' consonant sound
+
+  // More tricky 'u' words with 'yoo' sound for 'a'
+  ("unify","a"),
+  ("unilateral","a"),
+  ("ubiquitous","a"), // 'yoo-bik...'
+  ("usurp","a"),
+  ("usuary","a"),    // 'yoo-zhoo-ary' (rare, but treat like usual)
+  ("uterus","a"),    // 'yoo-terus'
+  ("usher","an"),    // 'uh-sher' vowel sound
+  ("ugly","an"),     // repeated, 'uh-glee'
+  ("upper","an"),     // 'uh-pur' (vowel sound but actually 'uh' sound = 'an'? Actually "upper" starts with a simple vowel sound 'uh', so should it be 'an'? It's commonly "an upper hand".
+  ("update","an"),    // 'up-date' starts with 'uh' vowel sound → 'an update' normally. Actually we say "an update" commonly. Correcting to 'an':
+
+  // Words starting with 'u' that actually take "an"
+  // (because they start with a vowel sound 'uh'):
+  ("under","an"),
+  ("understand","an"),
+  ("unusual","an"),
+  ("unreal","an"),
+  ("unemployed","an"),
+  ("unequal","an"),
+  ("unopened","an"),
+  ("unarmed","an"),
+  ("uneasy","an"),
+  ("united","a"), // 'yoo-nited' is 'a united front'
+  ("universal","a"),
+  ("umbrella","an"), // repeated
+  ("upset","an"),    // 'uh-pset' vowel start
+
+  // More vowel words to ensure coverage:
+  ("animal","an"),
+  ("image","an"),
+  ("order","an"),
+  ("arrow","an"),
+  ("ace","an"),
+  ("acre","an"),
+  ("agent","an"),
+  ("artist","an"),
+  ("idea","an"), // repeated
+  ("aim","an"),
+  ("item","an"), // repeated
+  ("isotope","an"),
+  ("opera","an"),
+
+  // Words starting with vowels but pronounced with consonant sound 'y' → 'a'
+  ("euphemistic","a"), // repeated
+  ("euphoric","a"),
+  ("eulogy","a"), // repeated
+  ("Europe","a"), // repeated
+  ("European","a"), // repeated
+  ("eugenics","a"), // repeated
+  ("euphemism","a"), // repeated
+  ("eureka","a"), // repeated
+  ("euphonious","a"),
+  ("Eurasia","a"),
+
+  // Common consonant-start words:
+  ("milk","a"),
+  ("road","a"),
+  ("stone","a"),
+  ("book","a"), // repeated
+  ("pen","a"),  // repeated
+  ("fork","a"), // repeated
+  ("spoon","a"), // repeated
+  ("knife","a"), // 'knife' actually starts with /n/ sound because 'k' is silent. "a knife" or "an knife"? It's "a knife" (starts with 'n' consonant sound?), actually 'kn' in 'knife' is silent, so 'knife' is pronounced 'nife' with a consonant sound 'n', so "a knife" is correct.
+  ("glass","a"),
+  ("window","a"),
+
+  // Final checks with tricky silent letters:
+  ("gnome","a"), // 'g' silent, pronounced 'nome' with 'n' consonant sound → 'a gnome'
+  ("gnat","a"),  // 'nat' consonant sound → 'a gnat'
+  ("xylophone","a"), // 'zylophone' → 'z' consonant sound → 'a xylophone'
+  ("xenon","a"), // 'zee-non' consonant sound 'z' → 'a xenon'
+  ("honorific","an"), // repeated but reaffirming
+  ("hourly","an"), // repeated but reaffirming
+  ("honorably","an"),
+  ("honorifics","an"),
+  ("honors","an"),
+)
+
+#for (word, article) in tests {
+  let chosen = __determine_article(word)
+  if chosen != article {
+    panic("bad choice! word: " + word + ", article: " + article + ", chosen: " + chosen)
+  }
+}
+
+#set page(height: auto, width: auto, margin: 1em)
+
+Success! All articles determined properly!