diff --git a/src/substrait/derivation_expression.py b/src/substrait/derivation_expression.py
index 0d8b153f..be9d9690 100644
--- a/src/substrait/derivation_expression.py
+++ b/src/substrait/derivation_expression.py
@@ -2,10 +2,36 @@
 from typing import Optional
 
 from antlr4 import CommonTokenStream, InputStream
+from antlr4.error.ErrorListener import ErrorListener
 from substrait.type_pb2 import NamedStruct, Type
 from substrait_antlr.substrait_type.SubstraitTypeLexer import SubstraitTypeLexer
 from substrait_antlr.substrait_type.SubstraitTypeParser import SubstraitTypeParser
 
+
+class DerivationExpressionParseError(Exception):
+    """Raised when a derivation expression cannot be parsed."""
+
+
+class _RaisingErrorListener(ErrorListener):
+    """ANTLR error listener that raises instead of recovering.
+
+    ANTLR's default behaviour on a syntax error is to print a warning to
+    stderr and continue with a partial (possibly ``None``-containing) parse
+    tree.  We want the parse to fail loudly with a message naming the offending
+    input instead.
+    """
+
+    def __init__(self, expression: str):
+        super().__init__()
+        self._expression = expression
+
+    def syntaxError(self, recognizer, offendingSymbol, line, column, msg, e):
+        raise DerivationExpressionParseError(
+            f"Could not parse derivation expression {self._expression!r} "
+            f"at line {line}:{column}: {msg}"
+        ) from e
+
+
 # Binary operators keyed by the token text the grammar attaches to `op`.
 # (Integer division: type parameters like precision/scale are integers.)
 _BINARY_OPS = {
@@ -245,9 +271,18 @@ def _evaluate(x, values: dict):
 
 
 def _parse(x: str):
+    error_listener = _RaisingErrorListener(x)
+
     lexer = SubstraitTypeLexer(InputStream(x))
+    lexer.removeErrorListeners()
+    lexer.addErrorListener(error_listener)
+
     stream = CommonTokenStream(lexer)
+
     parser = SubstraitTypeParser(stream)
+    parser.removeErrorListeners()
+    parser.addErrorListener(error_listener)
+
     return parser.expr()
 
 
diff --git a/tests/test_derivation_expression.py b/tests/test_derivation_expression.py
index 680c7c2c..2caa62ea 100644
--- a/tests/test_derivation_expression.py
+++ b/tests/test_derivation_expression.py
@@ -1,6 +1,9 @@
+import re
+
+import pytest
 from substrait.type_pb2 import NamedStruct, Type
 
-from substrait.derivation_expression import evaluate
+from substrait.derivation_expression import DerivationExpressionParseError, evaluate
 
 
 def test_simple_arithmetic():
@@ -146,7 +149,7 @@ def test_struct_simple():
 
 def test_nstruct_simple():
     """Test named struct with field names and types."""
-    result = evaluate("nStruct<a i32, b i32>", {})
+    result = evaluate("nStruct<a: i32, b: i32>", {})
     expected = NamedStruct(
         names=["a", "b"],
         struct=Type.Struct(
@@ -160,9 +163,33 @@ def test_nstruct_simple():
     assert result == expected
 
 
+@pytest.mark.parametrize(
+    "expression",
+    [
+        "decimal",
+        "interval_day",
+        "interval_compound",
+        "precision_time",
+        "precision_timestamp",
+        "precision_timestamp_tz",
+    ],
+)
+def test_parameterized_type_without_arguments_raises(expression):
+    """Precision-parameterized types used without their required arguments are
+    syntax errors and should raise a clear parse error, not fall through to the
+    confusing "Unknown token type <class 'NoneType'>" exception (see #216)."""
+    with pytest.raises(DerivationExpressionParseError, match=re.escape(expression)):
+        evaluate(expression)
+
+
+def test_invalid_syntax_raises():
+    with pytest.raises(DerivationExpressionParseError):
+        evaluate("i8 +")
+
+
 def test_nstruct_nested():
     """Test named struct with nested struct field."""
-    result = evaluate("nStruct<a i32, b i32, c struct<i32, fp32>>", {})
+    result = evaluate("nStruct<a: i32, b: i32, c: struct<i32, fp32>>", {})
     expected = NamedStruct(
         names=["a", "b", "c"],
         struct=Type.Struct(