A good argument for determiners being silent can be this: names of different kinds of trees and names of wooden things were preceded by the determiner G̃IŠ (tree, wood, tool), for example:
- G̃IŠ.nàd, G̃IŠ.ná: bed, couch (
nú-a, 'to lie down' + nominative
- G̃IŠ.kun4: ladder, stairs, threshold
- G̃IŠ.kiri6: orchard, garden, palm grove (
ki, 'place', +
ru5, 'to send
forth shoots, buds, or blossoms')
- G̃IŠ.apin: n., seeder plow (
a, 'seed', +
bun, 'to blow')
- G̃IŠ.zar: wagon shaft (Akkadian loanword, from serru(m) II)
In Sumerian nominative phrases, the head noun always came first, e.g. LUGAL 'king': LU 'man' + GAL 'big'. If we assume that the determiners were pronounced, then it follows that such words as 'ladder' or 'bed' were constructed as 'such-and-such tree/tool' where 'such-and-such' stands for attributes that help us distinguish, in my example, "lying-down tree/tool" from "ladder-like tree/tool", which sounds weird, just as weird as the analysis of the noun G̃IŠ.kiri6 'orchard' as "tree of blossom place" since orchard is actually a place, not a tree.
Also note the last word, G̃IŠ.zar 'wagon shaft' which was borrowed from Akkadian serru.
And now imagine an early Bronze Age language which is so logical that it groups together the bed, the ladder, the garden, the wagon shaft by adding the same morpheme to all such words. And even when borrowing a word for a new thing from a neighboring tribe, the speakers of the language (primitive illiterate peasants and fishermen) intuitively add the "tool" morpheme to it. As for me, it is more likely that it was the scribes who added that G̃IŠ sign for easier understanding of the written text than the Sumerian language originally was so logically classifying. In fact, the only known classification of nouns in Sumerian is as for animacy, there are two genders, animate and inanimate. This gender distinction goes through the whole system of the language, many noun cases are different for each gender, the verb agrees with both subject and object in that animacy. Now imagine that those determiners had been a thing in the Sumerian language. Wouldn't we have seen that logical system of categories reflected in the noun case and verb conjugation morphology? We can see nothing like that, though.