diff --git a/google/cloud/language/v1/BUILD.bazel b/google/cloud/language/v1/BUILD.bazel
index 2a2e305b581b7..b65c96a5a0d31 100644
--- a/google/cloud/language/v1/BUILD.bazel
+++ b/google/cloud/language/v1/BUILD.bazel
@@ -220,10 +220,10 @@ php_gapic_library(
     name = "language_php_gapic",
     srcs = [":language_proto_with_info"],
     grpc_service_config = "language_grpc_service_config.json",
+    migration_mode = "MIGRATING",
     rest_numeric_enums = True,
     service_yaml = "language_v1.yaml",
     transport = "grpc+rest",
-    migration_mode = "MIGRATING",
     deps = [
         ":language_php_grpc",
         ":language_php_proto",
diff --git a/google/cloud/language/v1/language_service.proto b/google/cloud/language/v1/language_service.proto
index 361a024ca07c5..d34240827c37b 100644
--- a/google/cloud/language/v1/language_service.proto
+++ b/google/cloud/language/v1/language_service.proto
@@ -92,6 +92,15 @@ service LanguageService {
     option (google.api.method_signature) = "document";
   }
 
+  // Moderates a document for harmful and sensitive categories.
+  rpc ModerateText(ModerateTextRequest) returns (ModerateTextResponse) {
+    option (google.api.http) = {
+      post: "/v1/documents:moderateText"
+      body: "*"
+    };
+    option (google.api.method_signature) = "document";
+  }
+
   // A convenience method that provides all the features that analyzeSentiment,
   // analyzeEntities, and analyzeSyntax provide in one call.
   rpc AnnotateText(AnnotateTextRequest) returns (AnnotateTextResponse) {
@@ -966,8 +975,7 @@ message TextSpan {
 
 // Represents a category returned from the text classifier.
 message ClassificationCategory {
-  // The name of the category representing the document, from the [predefined
-  // taxonomy](https://cloud.google.com/natural-language/docs/categories).
+  // The name of the category representing the document.
   string name = 1;
 
   // The classifier's confidence of the category. Number represents how certain
@@ -1118,6 +1126,18 @@ message ClassifyTextResponse {
   repeated ClassificationCategory categories = 1;
 }
 
+// The document moderation request message.
+message ModerateTextRequest {
+  // Required. Input document.
+  Document document = 1 [(google.api.field_behavior) = REQUIRED];
+}
+
+// The document moderation response message.
+message ModerateTextResponse {
+  // Harmful and sensitive categories representing the input document.
+  repeated ClassificationCategory moderation_categories = 1;
+}
+
 // The request message for the text annotation API, which can perform multiple
 // analysis types (sentiment, entities, and syntax) in one call.
 message AnnotateTextRequest {
@@ -1139,6 +1159,9 @@ message AnnotateTextRequest {
     // Classify the full document into categories.
     bool classify_text = 6;
 
+    // Moderate the document for harmful and sensitive categories.
+    bool moderate_text = 11;
+
     // The model options to use for classification. Defaults to v1 options
     // if not specified. Only used if `classify_text` is set to true.
     ClassificationModelOptions classification_model_options = 10;
@@ -1182,4 +1205,7 @@ message AnnotateTextResponse {
 
   // Categories identified in the input document.
   repeated ClassificationCategory categories = 6;
+
+  // Harmful and sensitive categories identified in the input document.
+  repeated ClassificationCategory moderation_categories = 7;
 }