Your inclusion of accent is brilliant! No one else has ever discussed it. I had assumed it always was on the ultima, but you have several instances of penult length. When I worked through Yamada'a Itbayaten dictionary, I also noted sometimes he doubled a vowel, but did not think it significant. I thank you for your care in this matter.

adze *piku < actually Spanish pico 'pickaxe' | Andrew Pick found evidence to support *kawaL

all | Batanic *tata:vuh is an innovation specific to this group

anger | Nice to see that Itbayaten sorih has other cognates *sorih

ankle | Until your study, I assumed each dialect had its own word

ant | The data I have supports *balaho

bad | Batanic *maraLet seems to be an innovation somehow related to but changed from PMP *daqet
Does Iraralay really have two forms? I note that Iratay is blank.

banana | not enough data for a reconstruction? I find *binelbel a possibility, which might be related to Formosan = Blust's PAN *beNbeN, which Dyen would have transcribed *beLbeL (to emphasize the lateral as opposed to the nasal reflexes among Formosan languages).

beautiful | I'm surpised Yamada missed this word in Itbayaten

belly | Batanic *bulek is an innovation, replacing *tiyan

big | I agree that *rakuL is more appropriate for Proto-Batanic than my *dakuL. Given the larger picture (PPH) the *d is justified.

bone | I have Itb to7xang with glottal stop, and reconstruct *tu7Lang

148 leaf is a duplicate of 145

175 nail is a duplicate of 97