New interview with Christopher Tolkien

[William Cloud Hicklin]

The problem there is that in attempting to reconstruct ancient pronunciations, we rarely have much to go on besides what we can deduce from the sound-patterning of the language's surviving verse, whether rhyme, assonance, or alliteration; and that doesn't help much with medial consonants and consonant-pairs.

Occasionally orthography helps, as a sound-shift can be traced in a spelling change over time and/or colloquial spellings; but that's no help with -ng since there wouldn't have been an alternate orthography, "J" not existing in OE (except as an alternate written form of the vowel "I")

So -ng is really anybody's guess, depending on whether one wants to use "frog DNA" from mod. German, or from Dutch-Frisian which is a closer cousin. FWIW, I'm disposed to think that dzh- in English, spelled either J or G, is a French import. But I certainly have no evidence to back up my gut there.