Blevins proposes the following sound correspondences: There is neutralization and sometimes loss of final nasals in Proto-Ongan, with final **n merged into Proto-Ongan *ŋ, and final **m and **ny partially merged into *ŋ. Final stops are lost in multisyllabic words in Proto-Ongan. Initial **b drops from Proto-Ongan before **u and perhaps before **i. **qw and **kw become *kw in Proto-Ongan, and *q/k or *w in Proto-AN. Proto-Ongan and Proto-AN share a typologically odd restriction against root-initial *m-. Vowel-initial words in Proto-Ongan correspond to *q in Proto-AN; because the Austronesian forms often include doublets, Blevins believes this is due to epenthesis in Proto-AN.
Proto-Austronesian
*i
*u
*a
*ə
*ay#
Proto-Ongan
*i
*u, *o
*a, *e
*e
*e#
Final **ay has become *e in Proto-Ongan. *e also derives from **a before palatals, word-finally, and when unstressed. Proto-Ongan *o typically derives from **u in a checked syllable, or from assimilation as in **wa. Proto-Ongan *ə is thought to have been an allophone of *e, found before coda nasals except after palatals.
Grammatical correspondences
Most derivational morphology and grammatical words are so short that the several resemblances between Proto-Ongan and Proto-AN may be chance. However, Ongan morphology does appear to explain an odd situation in Austronesian. Proto-Austronesian has a limited set of reconstructed vowel-initial roots, all of which are kin terms, body parts, or other readily possessed nouns. Ongan languages have inalienable possession, and inalienably possessed nouns are all vowel initial. Elsewhere, vowel-initial roots in Proto-Ongan correspond to initial *q- in Proto-Austronesian. The complete list of vowel-initial Proto-AN roots reconstructed by Blust is as follows: ;Kin: *aki grandfather; *ama father, paternal uncle; *aNak child; *apu grandparent/grandchild; *aya paternal aunt; *ina mother, maternal aunt ;Body: *ujung/ijung nose; *ikuR tail; *iSeq urine; *uRat vein, sinew; *utaq vomit ;Other: *asu/wasu dog; *aCab cover; *ian home; *uNay splinter These are all the kinds of words expected in inalienable-possession systems. Blevins suggests that inalienable possession was lost from Proto-Austronesian, presumably after epenthetic *q- was added to vowel-initial words. There are many Proto-AN doublets like *wasu, *asu 'dog'; initial *w- has also been lost from *w-anaN 'right side' and *w-iRi 'left side' in Pazeh and other languages, from what Blust describes as "some now-obscured morphological process". Blevins suggests that in all three cases, the Proto-Austronesian *w- reflects the Proto-Ongan possessive prefix *gw- 'his, her', which remained as a fossil in some daughter languages. Thus proto-Austronesian–Ongan may explain some of the odd patterns found in proto-Austronesian.
Criticism
The connection between Austronesian and Ongan has not been supported by Austronesianists. Robert Blust finds that Blevins' conclusions are not supported by her data: Of her first 25 reconstructions, none are reproducible using the comparative method, and Blust concludes that the grammatical comparison does not hold up. Blust also cites non-linguistic evidence against Blevins' hypothesis.
