diff --git a/src/tokens.rs b/src/tokens.rs
index 0da2b07..d9c0f72 100644
--- a/src/tokens.rs
+++ b/src/tokens.rs
@@ -1,3 +1,11 @@
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub enum Case {
+    Title,
+    Lower,
+    Scream,
+    None,
+}
+
 #[derive(Debug, Clone, Copy, PartialEq, Eq)]
 pub struct Symbol<'t> {
     token: &'t str,
@@ -46,6 +54,10 @@ impl<'t> Symbol<'t> {
         self.token
     }
 
+    pub fn case(&self) -> Case {
+        Case::None
+    }
+
     pub fn offset(&self) -> usize {
         self.offset
     }
@@ -58,6 +70,7 @@ impl<'t> Symbol<'t> {
 #[derive(Debug, Clone, Copy, PartialEq, Eq)]
 pub struct Word<'t> {
     token: &'t str,
+    case: Case,
     offset: usize,
 }
 
@@ -84,14 +97,22 @@ impl<'t> Word<'t> {
         Ok(item)
     }
 
-    pub(crate) fn new_unchecked(token: &'t str, offset: usize) -> Self {
-        Self { token, offset }
+    pub(crate) fn new_unchecked(token: &'t str, case: Case, offset: usize) -> Self {
+        Self {
+            token,
+            case,
+            offset,
+        }
     }
 
     pub fn token(&self) -> &str {
         self.token
     }
 
+    pub fn case(&self) -> Case {
+        self.case
+    }
+
     pub fn offset(&self) -> usize {
         self.offset
     }
@@ -127,6 +148,22 @@ impl WordMode {
             WordMode::Boundary
         }
     }
+
+    fn case(self, last: WordMode) -> Case {
+        match (self, last) {
+            (WordMode::Uppercase, WordMode::Uppercase) => Case::Scream,
+            (WordMode::Uppercase, WordMode::Lowercase) => Case::Title,
+            (WordMode::Lowercase, WordMode::Lowercase) => Case::Lower,
+            (WordMode::Number, WordMode::Number) => Case::None,
+            (WordMode::Number, _)
+            | (_, WordMode::Number)
+            | (WordMode::Boundary, _)
+            | (_, WordMode::Boundary)
+            | (WordMode::Lowercase, WordMode::Uppercase) => {
+                unreachable!("Invalid case combination: ({:?}, {:?})", self, last)
+            }
+        }
+    }
 }
 
 fn split_symbol(symbol: &str, offset: usize) -> impl Iterator<Item = Word<'_>> {
@@ -135,6 +172,7 @@ fn split_symbol(symbol: &str, offset: usize) -> impl Iterator<Item = Word<'_>> {
     let mut char_indices = symbol.char_indices().peekable();
     let mut start = 0;
     let mut start_mode = WordMode::Boundary;
+    let mut last_mode = WordMode::Boundary;
     while let Some((i, c)) = char_indices.next() {
         let cur_mode = WordMode::classify(c);
         if cur_mode == WordMode::Boundary {
@@ -143,13 +181,16 @@ fn split_symbol(symbol: &str, offset: usize) -> impl Iterator<Item = Word<'_>> {
             }
             continue;
         }
+        if start_mode == WordMode::Boundary {
+            start_mode = cur_mode;
+        }
 
         if let Some(&(next_i, next)) = char_indices.peek() {
             // The mode including the current character, assuming the current character does
             // not result in a word boundary.
             let next_mode = WordMode::classify(next);
 
-            match (start_mode, cur_mode, next_mode) {
+            match (last_mode, cur_mode, next_mode) {
                 // cur_mode is last of current word
                 (_, _, WordMode::Boundary)
                 | (_, WordMode::Lowercase, WordMode::Number)
@@ -157,24 +198,36 @@ fn split_symbol(symbol: &str, offset: usize) -> impl Iterator<Item = Word<'_>> {
                 | (_, WordMode::Number, WordMode::Lowercase)
                 | (_, WordMode::Number, WordMode::Uppercase)
                 | (_, WordMode::Lowercase, WordMode::Uppercase) => {
-                    result.push(Word::new_unchecked(&symbol[start..next_i], start + offset));
+                    let case = start_mode.case(cur_mode);
+                    result.push(Word::new_unchecked(
+                        &symbol[start..next_i],
+                        case,
+                        start + offset,
+                    ));
                     start = next_i;
                     start_mode = WordMode::Boundary;
+                    last_mode = WordMode::Boundary;
                 }
                 // cur_mode is start of next word
                 (WordMode::Uppercase, WordMode::Uppercase, WordMode::Lowercase) => {
-                    result.push(Word::new_unchecked(&symbol[start..i], start + offset));
+                    result.push(Word::new_unchecked(
+                        &symbol[start..i],
+                        Case::Scream,
+                        start + offset,
+                    ));
                     start = i;
-                    start_mode = WordMode::Boundary;
+                    start_mode = cur_mode;
+                    last_mode = WordMode::Boundary;
                 }
                 // No word boundary
                 (_, _, _) => {
-                    start_mode = cur_mode;
+                    last_mode = cur_mode;
                 }
             }
         } else {
             // Collect trailing characters as a word
-            result.push(Word::new_unchecked(&symbol[start..], start + offset));
+            let case = start_mode.case(cur_mode);
+            result.push(Word::new_unchecked(&symbol[start..], case, start + offset));
             break;
         }
     }
@@ -240,23 +293,65 @@ mod test {
     #[test]
     fn split_symbol() {
         let cases = [
-            ("lowercase", &["lowercase"] as &[&str]),
-            ("Class", &["Class"]),
-            ("MyClass", &["My", "Class"]),
-            ("MyC", &["My", "C"]),
-            ("HTML", &["HTML"]),
-            ("PDFLoader", &["PDF", "Loader"]),
-            ("AString", &["A", "String"]),
-            ("SimpleXMLParser", &["Simple", "XML", "Parser"]),
-            ("vimRPCPlugin", &["vim", "RPC", "Plugin"]),
-            ("GL11Version", &["GL", "11", "Version"]),
-            ("99Bottles", &["99", "Bottles"]),
-            ("May5", &["May", "5"]),
-            ("BFG9000", &["BFG", "9000"]),
+            (
+                "lowercase",
+                &[("lowercase", Case::Lower, 0usize)] as &[(&str, Case, usize)],
+            ),
+            ("Class", &[("Class", Case::Title, 0)]),
+            (
+                "MyClass",
+                &[("My", Case::Title, 0), ("Class", Case::Title, 2)],
+            ),
+            ("MyC", &[("My", Case::Title, 0), ("C", Case::Scream, 2)]),
+            ("HTML", &[("HTML", Case::Scream, 0)]),
+            (
+                "PDFLoader",
+                &[("PDF", Case::Scream, 0), ("Loader", Case::Title, 3)],
+            ),
+            (
+                "AString",
+                &[("A", Case::Scream, 0), ("String", Case::Title, 1)],
+            ),
+            (
+                "SimpleXMLParser",
+                &[
+                    ("Simple", Case::Title, 0),
+                    ("XML", Case::Scream, 6),
+                    ("Parser", Case::Title, 9),
+                ],
+            ),
+            (
+                "vimRPCPlugin",
+                &[
+                    ("vim", Case::Lower, 0),
+                    ("RPC", Case::Scream, 3),
+                    ("Plugin", Case::Title, 6),
+                ],
+            ),
+            (
+                "GL11Version",
+                &[
+                    ("GL", Case::Scream, 0),
+                    ("11", Case::None, 2),
+                    ("Version", Case::Title, 4),
+                ],
+            ),
+            (
+                "99Bottles",
+                &[("99", Case::None, 0), ("Bottles", Case::Title, 2)],
+            ),
+            ("May5", &[("May", Case::Title, 0), ("5", Case::None, 3)]),
+            (
+                "BFG9000",
+                &[("BFG", Case::Scream, 0), ("9000", Case::None, 3)],
+            ),
         ];
         for (input, expected) in cases.iter() {
             let symbol = Symbol::new(input, 0).unwrap();
-            let result: Vec<_> = symbol.split().map(|w| w.token).collect();
+            let result: Vec<_> = symbol
+                .split()
+                .map(|w| (w.token, w.case, w.offset))
+                .collect();
             assert_eq!(&result, expected);
         }
     }