sC clusters are the exception to the generalization against homorganic clusters (ex *pw, *bw), and against obstruent-obstruent clusters (ex *pt, *fk), and are the only obstruent-nasal ones…

—

All of these languages have both [t] and [s] in their inventories. So why do some choose one segment and some the other when encountered with the forbidden [ ]?
If claims about universals are to be made, we hope that they would be the same universals that are supported by the first language acquisition evidence. The L1 acquisition facts are clear when it comes to substitution of stops and fricatives:

By using OT, both both universal and developmental effects are essentially wielded as transfer

OT can account for L2 data without making,

“Data can lead to hypotheses about UG that can be tested on L1 data.”

: offending segments are mapped to the closest available alternative in the L2 inventory.