pi normally acts like it introduces a noun followed by modifiers, a nominalizer if one could temporarily agree on a word for it.
First the evidence, then the question.
The phrase "kama jo" almost never appears anywhere except after mi, sina, li.
The phrase "pi kama jo" looks like it's only been used by two people in my incomplete current corpus files , e.g.
"...mi tawa ma tomo Tokijo li tawa tomo suli pi kama jo ijo pi nimi "Tokyu Hands" li kama jo e poki ni...." (jan Yoh) (definitely being used like a verb phrase with the "e" phrase dropping the "e")
".... jan li pali kin e ilo pi kama jo. ona li kama..." (Gilgamesh)... intransitive verbalizer (looks just like as nominalizer so one could quibble)
The following words (ostrich and department store) I got from jan Kipo's spreadhsheet.
It looks like they were created by transforming "li" and/or "e" into "pi" or sometimes just dropping one or both.
(?) Verbalizer for intransitives
Ostrich: waso ma pi tawa mute. (jan Mako 2007)
Read as a noun: bird of many trips
They meant: land bird of going a lot--- the idea here is that the bird runs around a lot
I think they got here via this transform:
waso ma li tawa mute => waso ma pi tawa mute.
(?) Verbalizer for transitives
Department store, Castle: tomo suli pi kama jo ijo mute (jan Luka 2007)
read as a noun following pi: big house owning, inanimate, numerous arrival
They meant big house, coming to own (aux + v) many things (acc)
I think these transforms were what was in mind:
tomo suli li kama jo e ijo mute => tomo suli pi kama jo ijo mute
tomo suli li kama jo e ijo mute => tomo suli pi kama jo pi ijo mute (pi replaces "e" and "li")
I think these are mistakes. Putting verb-y things after pi (including the accusative complement!) is a language innovation, I suspect will happen, but so far isn't official.
Any other opinions?