The point of it
The idea is to seek out compound word forms in an existing word list. For example, a word list might contain daredevil and dare and devil, or eyeglass and both eye and glass. This function processes the original list and identifies such cases where there's a match. It lists them in a text file or to a spreadsheet. Meaning is not considered: the only requirement is that each type consists of two other types.
... or more than two different pairings:
These are some found using the BNC. As you see from item 3580, this is a purely mechanical process.
How to do it
With a word list loaded, choose Compute | Compounds.
Choose a minimum number of characters for each segment. In the list above each potential part of a compound had to have a minimum of 5 characters and frequency of 5. Separate them with a comma.
When choosing your file to save to, choose Excel or .TXT format.
WordSmith processes each word form which is long enough and frequent enough, seeking pairings with all other words in the same word list. It also ignores any prefix-forms or suffix-forms you have defined, and punctuation. If the list is lemmatised it will consider only head-words.
How the search works
It requires both parts of the potential compound to be at least two characters long and that the two parts together comprise the compound.
Three-part compounds like inasmuch will not get picked up but counterclockwise will if the list contains counter and clockwise, just as clockwise will be found if the list has both clock and wise.
See also: prefixes and suffixes.