When deriving a primitive word from a root, one of its isolated consonants could be modified by nasal-infixion: inserting a homorganic nasal (from the same series) [m], [n] or [ŋ] before the consonant (SD/417). This nasal could be infixed before an isolated final or medial basic consonant, but never before an initial consonant (SD/434). Furthermore, nasal infixion was less frequent before the medial consonant (SD/417).

The only basic consonant to which the nasal could be infixed were stops or [z] (SD/434). In the case of [z], this was likely a remnant of earlier infixion to the palatal stop [ɟ] before it developed into [z] (SD/417). By the time of Classical Adûnaic, after nasals became voiceless stops before voiceless stops, aspirates and [s], the only surviving nasal-infixions would have been those before [z] and voiced stops, that is: [mb], [nd], [nz] and [ŋg] (“ng”).

When the nasal infixions was applied to a final consonant, the result was an illegal consonant cluster (see final-consonants). To resolve this problem, an additional vowel was suffixed to the word, such as: √KALAB > ✶kalambu, ✶aklambu (SD/425). Where the preceding vowel was a characteristic-vowel (as was usually the case), the new final vowel was required to be a different vowel.

Although nasal infixion could occur after any vowel, it was usually found after a characteristic-vowel of the root, so that forms like ✶Ad. √KALAB > ✶kalambu where more common than ✶kalimbu, ✶kalumbu (SD/425).