With this first method I take some random words from directed-association results:
Example: officer, instructor, blame
and use the first three letters to trigger adjectives. Thus:
officer = official, off-side
instructor = instrumental, inside, insulting
blame = blameworthy, blatant, black
I can also use the letters to trigger verbs/actions and use them adjectivally:
instructor = insisting, insulting, inspecting
I can also make up new adjectives. So with a random word 'hair' I could create an adjective and a possible meaning:
Hairish = like hair
With a random word 'Constable' I could create:
Constable-ish = in the style of John Constable
With method two my aim is to go beyond the predictable: I name a specific thing or object before I consider possible adjectives (or phrases) that fit. I do this by using a template:
Name an X thing or object
and choosing a random word for the 'X'.
Name a crying thing = tears
From this I can name possible adjectives and adjectival phrases:
Wet, emotional, moving, from the eyes, onion-consequence.
I can also use the template:
Name an X X thing or object
to set more of a challenge and go beyond the obvious when naming a thing or object.
Setting a quota
With method two I can also set a quota for my adjectival phrase to go beyond the obvious choices.
So if I set a directive: "Name a parliament thing" and I choose "Debate" as an answer I can try to name adjectival phrases of several words:
Debate is: likely to be heated, among opposing sides, likely to make the news if it's deemed newsworthy.
See also: Listing adverbs and adverbial phrases