ngavyaw’ ghu = canine’s baby or baby of the canine = puppy

ghu ngavyaw’ = baby’s canine or canine of the baby = baby’s pet

You forget the more general genitive interpretation of the noun-noun construction.

I'm not going to address this question directly, because baby and canine are both nouns and adjectives, and this confuses the translations. Instead, I'm going to translate an idea that deals strictly in nouns. How about a student who is a child?

ghojwI' puq student child
puq ghojwI' child student

Which is it? Is it a child (head noun) of the type student (modifying noun)? Or is it a student (head noun) of the type child (modifying noun)?

It's both. Either is correct. It's a child who is a student and a student who is a child.

So, completely ignoring semantics, is it

ngavyaw' ghu canine baby
or
ghu ngavyaw' baby canine
?

It's both. Either is correct. It's a baby that is a canine, and it's a canine that is a baby.

*puppy* would be pronounced “poop-pee”, if it were pronounceable at all, given the lack of vowel in the second syllable, since a {y} is never a vowel in Klingon. There is no “uh” sound in Klingon, so it would be difficult to transliterate without making it easily confused with “poppy” *{papIy}*.

Qa'yIn was not proposing to transliterate the word puppy, but instead keep it as a foreign word.

But on transliterating puppy: remember that Klingon nouns tend to place stress on the final syllable of the root noun, so papIy would sound like "pa-PEE." You can add a glottal stop to draw the stress: pa'pIy "PA-pee."

-- 
SuStel
http://trimboli.name