Removing Duplicate Words from a Cell

My most recent comment below discussed the following expression to remove punctuation characters from text in cell A1:
    TEXTJOIN("", , TEXTSPLIT(A1, {"?",".","!"}))
Here's a better version of that expression with the same result:
    CONCAT(TEXTSPLIT(A1, {"?",".","!"}))
I'm not sure Alice's column of text includes punctuation, and I doubt she wants to use a VBA UDF formula, but here's an alternative version of the Tip's UDF:

Function UniqueWords2(InputString As String) As String
    Dim sTemp As String, item As Variant, UniqueItems As New Collection
    Const PUNCTUATION = ". , ! ? : ; ( ) "" " 'two ("") become one (")
    sTemp = InputString
    For Each item In Split(WorksheetFunction.Trim(PUNCTUATION))
        sTemp = Replace(sTemp, item, vbNullString) 'remove punctuation
    Next item
    On Error Resume Next
        For Each item In Split(WorksheetFunction.Trim(sTemp))
            UniqueItems.Add Null, item 'case insensitive
            If Err = 0 Then 'item is unique
                UniqueWords2 = UniqueWords2 & " " & item
            Else
                Err.Clear
            End If
        Next item
    On Error GoTo 0
    UniqueWords2 = LTrim(UniqueWords2) 'remove leading space
End Function

Notice UniqueItems only collects keys, where each key is a unique word from InputString. Collection keys are case insensitive, so this formula
=UniqueWords2(A1)
might return a mixed-case result. Here’s an alternative that converts all the text to lower-case:
=UniqueWords2(LOWER(A1))
If A1 contains "red, RED (BLUE) blue!" the first formula returns "red BLUE" and the second one returns "red blue".
Alice might want to amend the PUNCTUATION constant, which was derived from the Tip's version; if so, each punctuation character must be followed by a space character.

Re. TEXTSPLIT the Tip says, "... if there are punctuation marks in the text, they are considered part of the word just before the mark." For example, the Tip's first formula and the formulas in my previous comments below would reduce text like "red red? red blue! blue! blue" to "red red? blue! blue".
If we consider only question (?), period (.), and exclamation (!), the following expression will remove those punctuation characters from cell A1:
    TEXTJOIN("", , TEXTSPLIT(A1, {"?",".","!"}))
So "red red? red blue! blue! blue" becomes "red red red blue blue blue". Therefore, using that expression instead of A1 in those formulas solves the punctuation issue noted in the Tip. The punctuation array {"?",".","!"} can be expanded if necessary.

Here’s a formula using a RegEx pattern suggested by Google Gemini to return text from cell A1 with duplicate words removed:
=TRIM(REGEXREPLACE(A1, "\b(\w+)\b(?=.*?\b\1\b)", "", 0, TRUE))
TRIM is necessary because the RegEx pattern results in extraneous space characters. The last argument (TRUE) is necessary to make REGEXREPLACE case insensitive; therefore, the result might be mixed-case. Here’s an alternative that converts all the text to lower-case:
=TRIM(REGEXREPLACE(LOWER(A1), "\b(\w+)\b(?=.*?\b\1\b)", "", 0))
The RegEx pattern in these formulas keeps the last instance of a word and removes earlier duplicates, so "red RED BLUE blue" becomes "blue RED" with the first formula and "blue red" with the second. I don’t believe Excel’s implementation of RegEx supports a pattern that keeps the first instance of a word and removes later duplicates.
REGEXREPLACE currently requires Excel 365, which also includes TRIMRANGE; therefore, the following formula returns an array with all duplicate words in column A reduced to single words (last instance first):
=TRIM(REGEXREPLACE(TRIMRANGE(A:A), "\b(\w+)\b(?=.*?\b\1\b)", "", 0, TRUE))
Here’s the alternative that returns lower-case to avoid a mixed-case result:
=TRIM(REGEXREPLACE(LOWER(TRIMRANGE(A:A)), "\b(\w+)\b(?=.*?\b\1\b)", "", 0))
And here's the equivalent using functions available in My Excel Toolbox:
=TRIM(ForEachCell(RangeTrim(A:A), "RegExSubstitute(LOWER(@), ""\b(\w+)\b(?=.*?\b\1\b)"", """", 0)"))
ForEachCell is discussed in my previous comment below. RangeTrim and RegExSubstitute work like TRIMRANGE and REGEXREPLACE.
See https://sites.google.com/view/MyExcelToolbox/

The Tip's first formula is
=TEXTJOIN(" ", , UNIQUE(TEXTSPLIT(TRIM(A1), " "), TRUE))
TRIM is used to remove extraneous space characters. But the default value of TEXTJOIN's second argument is TRUE, which tells it to ignore the single blank text value ("") received from UNIQUE that results when TEXTSPLIT is given text with extraneous space characters. Therefore, TRIM is unnecessary and the following formula is sufficient :
=TEXTJOIN(" ", , UNIQUE(TEXTSPLIT(A1, " "), TRUE))
These formulas keep the first instance of a word and remove later duplicates, so "red red blue blue red" becomes "red blue".
UNIQUE is case insensitive, so "red RED BLUE blue red" becomes "red BLUE". To remove mixed-case duplicates and return a lower-case result, use this formula:
=TEXTJOIN(" ", , UNIQUE(TEXTSPLIT(LOWER(A1), " "), TRUE))
TEXTSPLIT requires Excel 2024. Here's an alternative using My Excel Toolbox's SplitText function with Excel 2019's TEXTJOIN and Excel 2021's UNIQUE:
=TEXTJOIN(" ", , UNIQUE(SplitText(LOWER(A1), " "), TRUE))
Excel 2024 includes BYROW; this formula returns an array removing duplicate words from all cells in the range A1:A999
=BYROW(A1:A999, LAMBDA(cell, TEXTJOIN(" ", , UNIQUE(TEXTSPLIT(cell, " "), TRUE))))
If any cells are empty, BYROW returns #VALUE!, but this formula will fix that
=IFERROR(BYROW(A1:A999, LAMBDA(cell, TEXTJOIN(" ", , UNIQUE(TEXTSPLIT(cell, " "), TRUE)))), "")
Here's an alternative using My Excel Toolbox's ForEachCell and SplitText functions:
=IFERROR(ForEachCell(A1:A999,"TEXTJOIN("" "",,UNIQUE(SplitText(@,"" ""), TRUE))"), "")
ForEachCell is described in my comment dated 2025-01-20 here: https://excelribbon.tips.net/T011725
See https://sites.google.com/view/MyExcelToolbox/