diff --git a/R/detect.R b/R/detect.R
index 5c62f337..bdfdd970 100644
--- a/R/detect.R
+++ b/R/detect.R
@@ -21,7 +21,7 @@
 #'   [boundary()]. An empty pattern, "", is equivalent to
 #'   `boundary("character")`.
 #'
-#' @param negate If `TRUE`, return non-matching elements.
+#' @param negate If `TRUE`, inverts the resulting boolean vector.
 #' @return A logical vector the same length as `string`/`pattern`.
 #' @seealso [stringi::stri_detect()] which this function wraps,
 #'   [str_subset()] for a convenient wrapper around
diff --git a/man/str_detect.Rd b/man/str_detect.Rd
index fa8b5d1b..965af1a1 100644
--- a/man/str_detect.Rd
+++ b/man/str_detect.Rd
@@ -25,7 +25,7 @@ Match character, word, line and sentence boundaries with
 \code{\link[=boundary]{boundary()}}. An empty pattern, "", is equivalent to
 \code{boundary("character")}.}
 
-\item{negate}{If \code{TRUE}, return non-matching elements.}
+\item{negate}{If \code{TRUE}, inverts the resulting boolean vector.}
 }
 \value{
 A logical vector the same length as \code{string}/\code{pattern}.
diff --git a/man/str_starts.Rd b/man/str_starts.Rd
index 2ea0a567..37b4b6c7 100644
--- a/man/str_starts.Rd
+++ b/man/str_starts.Rd
@@ -22,7 +22,7 @@ Match a fixed string (i.e. by comparing only bytes), using \code{\link[=fixed]{f
 is fast, but approximate. Generally, for matching human text, you'll want
 \code{\link[=coll]{coll()}} which respects character matching rules for the specified locale.}
 
-\item{negate}{If \code{TRUE}, return non-matching elements.}
+\item{negate}{If \code{TRUE}, inverts the resulting boolean vector.}
 }
 \value{
 A logical vector.
diff --git a/man/str_subset.Rd b/man/str_subset.Rd
index 5f72bd17..61a259b8 100644
--- a/man/str_subset.Rd
+++ b/man/str_subset.Rd
@@ -25,7 +25,7 @@ Match character, word, line and sentence boundaries with
 \code{\link[=boundary]{boundary()}}. An empty pattern, "", is equivalent to
 \code{boundary("character")}.}
 
-\item{negate}{If \code{TRUE}, return non-matching elements.}
+\item{negate}{If \code{TRUE}, inverts the resulting boolean vector.}
 }
 \value{
 A character vector, usually smaller than \code{string}.
diff --git a/man/str_which.Rd b/man/str_which.Rd
index 872c95c1..26539db6 100644
--- a/man/str_which.Rd
+++ b/man/str_which.Rd
@@ -25,7 +25,7 @@ Match character, word, line and sentence boundaries with
 \code{\link[=boundary]{boundary()}}. An empty pattern, "", is equivalent to
 \code{boundary("character")}.}
 
-\item{negate}{If \code{TRUE}, return non-matching elements.}
+\item{negate}{If \code{TRUE}, inverts the resulting boolean vector.}
 }
 \value{
 An integer vector, usually smaller than \code{string}.