diff --git a/src/Pure/Tools/prismjs.scala b/src/Pure/Tools/prismjs.scala
--- a/src/Pure/Tools/prismjs.scala
+++ b/src/Pure/Tools/prismjs.scala
@@ -1,105 +1,110 @@
 /*  Title:      Pure/Tools/prismjs.scala
     Author:     Makarius
 
 Support for the Prism.js syntax highlighter -- via external Node.js process.
 */
 
 package isabelle
 
 import scala.collection.mutable
 
 
 object Prismjs {
   /* component resources */
 
   val HOME: Path = Path.explode("$ISABELLE_PRISMJS_HOME")
 
   lazy val available_languages: List[String] = {
     val components_path = HOME + Path.explode("components.json")
     val components_json = JSON.parse(File.read(components_path))
     JSON.value(components_json, "languages") match {
       case Some(JSON.Object(langs)) =>
         (for ((lang, JSON.Object(info)) <- langs.iterator if lang != "meta") yield {
           val alias =
             JSON.Value.List.unapply(info.getOrElse("alias", Nil), JSON.Value.String.unapply)
               .getOrElse(Nil)
           lang :: alias
         }).toList.flatten.sorted
       case _ => error("Failed to determine languages from " + components_path)
     }
   }
 
 
   /* JavaScript prelude */
 
   def prelude(lang: JS.Source): JS.Source =
     cat_lines(List(
       Nodejs.require_fs,
       Nodejs.require_path("const prismjs", HOME),
       Nodejs.require_path("prismjs.load", HOME + Path.explode("components"), dir = true),
       JS.function("prismjs.load", lang),
       """
 function prismjs_content(t) {
-  if (Array.isArray(t)) { return t.map(prismjs_content).join() }
-  else if (t instanceof prismjs.Token) { return prismjs_content(t.content) }
+  if (t instanceof prismjs.Token) { return prismjs_content(t.content) }
+  else if (Array.isArray(t)) { return t.map(prismjs_content).join() }
   else { return t }
 }
 
 function prismjs_tokenize(text, lang) {
   return prismjs.tokenize(text, prismjs.languages[lang]).map(function (t) {
+    var types = []
+    var content = ""
     if (t instanceof prismjs.Token) {
-      return {"kind": t.type, "content": prismjs_content(t)}
+      types.push(t.type)
+      if (Array.isArray(t.alias)) { for (a of t.alias) { types.push(a) } }
+      else if (t.alias !== undefined) { types.push(t.alias) }
+      content = prismjs_content(t)
     }
-    else { return {"kind": "", "content": t} }
+    else { content = t }
+    return {"types": types, "content": content}
   })
 }
 """))
 
 
   /* tokenize */
 
   sealed case class Token(
-    kind: String,
+    types: List[String],
     content: String,
-    range: Text.Range = Text.Range.zero)
+    range: Text.Range)
 
   def tokenize(text: String, language: String): List[Token] = {
     if (!available_languages.contains(language)) {
       error("Unknown language " + quote(language))
     }
 
     val json =
       Isabelle_System.with_tmp_file("input", ext = "json") { input =>
         Isabelle_System.with_tmp_file("output", ext = "json") { output =>
           File.write(input, text)
           val lang = JS.value(language)
           Nodejs.execute(
             prelude(lang) + "\n" +
             Nodejs.write_file(output,
               JS.json_print(JS.function("prismjs_tokenize", Nodejs.read_file(input), lang))))
           JSON.parse(File.read(output))
         }
       }
 
     def parse_token(t: JSON.T): Token =
       (for {
-        kind <- JSON.string(t, "kind")
+        types <- JSON.strings(t, "types")
         content <- JSON.string(t, "content")
-      } yield Token(kind, content)).getOrElse(error("Malformed JSON token: " + t))
+      } yield Token(types, content, Text.Range.zero))
+        .getOrElse(error("Malformed JSON token: " + t))
 
     val tokens0 =
       JSON.Value.List.unapply(json, t => Some(parse_token(t)))
         .getOrElse(error("Malformed JSON: array expected"))
 
     var stop = 0
     val tokens = new mutable.ListBuffer[Token]
     for (tok <- tokens0) {
       val start = stop
       stop += tok.content.length
-      if (tok.kind.nonEmpty) {
-        tokens += tok.copy(range = Text.Range(start, stop))
-      }
+      tokens += tok.copy(range = Text.Range(start, stop))
     }
     tokens.toList
   }
 }