Any etymology beyond this reconstructed form is controversial. Sagart (2004) considers it to have arisen after Proto-Austronesian proper, rather than from a compound from*ɣatep(“five”) +*təlu(“three”), and relates it toPazehxasebatulu(literally“five and three”), but Winter (2010) rejects this on phonological and historical grounds.
Consistency of sound changesp → w alsoə andt sound dropped,
"*ɣatepatəlu →*wa(tə)lu-"
"*ɣatepiSəpat →wiS(ə)wa(t) → *wiSwa/*Siwa"
except: ɣatepiduSa →pidu, becausep meetsd, so there is a sound reinforcement.