Adûnaic duals are much like Quenya duals in that they are used for “natural pairs” (SD/428). The dual is formed with the dual suffix -at for all classes of nouns. Unlike many other declensions, there is no general distinction between the duals of strong-nouns and weak-nouns (SD/429), at least not by the time of Classical Adûnaic. It does make a difference whether the noun ends in a consonant (Class I) or a vowel (Class II), however.

Dual Formation

Nouns ending in a consonant: For nouns ending in a consonant (Class I):

• The dual suffix -at is added to the noun: nûph “fool” → nûphat “pair of fools” (SD/438), batân “road” → batânat “pair of roads” (SD/431).
• For nouns ending in a consonant preceded by a short vowel (Strong I nouns), the dual generally undergoes the Adûnaic syncope, so that the last short vowel vanishes: huzun “ear” → huznat “pair of ears” (SD/430), nithil “girl” → nithlat “pair of girls” (SD/436).
• The exception to this rule is when the second-to-last syllable is long (has a long vowel, diphthong or ends in two consonants), which prevents the syncope from occurring: târik “pillar” → târikat “pair of pillars” (SD/430), igmil “star-shaped figure” → *igmilat “pair of star-shaped figures”. It is probably also true if the two short vowels are different from each other: minal “heaven” → *minalat.

Nouns ending in a vowel: For nouns ending in a vowel (Class II), the dual suffix -at combines with this vowel according to the standard Adûnaic rules for vowel-combinations:

• If the noun ends in -i or -u (long or short), the usual glide-consonant y or w is inserted between the vowel and the dual suffix: gimli “star” → gimliyat “pair of stars” (SD/431), nardû “soldier” → nardûwat “pair of soldiers” (SD/438).
• If the noun ends in a short -a, it combines with the dual suffix into -ât: azra “sea” → azrât “pair of seas” (SD/431).
• If the noun ends in a long -â, -ê or -ô, the final vowel absorbs the a of the dual suffix, leaving only -t: anâ “human” → anât “pair of humans” (SD/437), izrê “sweetheart” → izrêt “pair of sweethearts” (SD/438), manô “spirit” → manôt “pair of spirits” (SD/438).

Subjective duals: The dual may also be inflected for the subjective case. Since the dual generally has a short vowel in its final syllable, the strong subjective inflection is used. This means the subjective inflection -a- lengthens the last a via a-fortification: zadan “house” → zadnat “pair of houses” → zadnât “pair of houses (subjective)” (SD/430), nîlu “moon” → nîluwat “pair of moons” → nîluwât “pair of moons (subjective)” (SD/431).

Where the last vowel of the dual is already long, there is no change for the subjective: izrê “sweetheart” → izrêt “pair of sweethearts” → izrêt “pair of sweethearts (subjective)” (SD/438).

Archaic duals: Archaic Adûnaic had different dual formations in a few cases:

• For Strong II nouns (those ending in a short vowel), the dual suffix -at originally replaced the final vowel instead of combining with it: nîlu “moon” → †nîlat “pair of moons” instead of the later form nîluwat (SD/431).
• For nouns ending in -ê and -ô, the final vowel decomposed into -ay and -aw before adding the dual suffix, as discussed in the section on the archaic glide-consonant: manô “spirit” → †manawat “pair of spirits” instead of the later form manôt (SD/438).

By the time of Classical Adûnaic, these formations had been replaced by the rules above.

Later duals: Duals of nouns ending in -a (azra → azrât) were so common that the dual suffix eventually was perceived to be -ât instead of -at. The duals of all nouns were formed with this new suffix, incidentally making the subjective dual indistinguishable from the normal dual (SD/431).

Tolkien stated that “in all nouns the N[ormal] and S[ubjective] of Duals was only distinguished in earlier texts” and that this change occurred “before the Exilic periods”, but it is unclear how soon the change occurred before the fall of Númenor and whether it applied to Classical Adûnaic. Most of Tolkien’s examples in Lowdham’s Report do not use this rule, so the discussion here assume this change occurred soon before Exile, in either Late Classical and Exilic Adûnaic.

Irregular duals: There are two curious dual forms that appear on SD/438: nardū̆wā̆t as the dual of nardû “soldier” and zōrī̆yā̆t as the dual of zôrî “nurse”. The short-or-long ā̆ appears in several examples, and seems to be shorthand for the two variations of the dual: normal (-at) and subjective (-ât). The short-or-long ū̆ and ī̆ are more difficult to explain. Given the base nouns, these vowels should always be long. Perhaps under some conditions, the dual suffix can cause a long final vowel to shorten, though what those conditions might be is unclear.

Another example of an irregular dual form is nuphrât as the dual form of nuphâr “parent” (SD/434). This is a weak-noun (Weak I), but its dual seems to undergo the Adûnaic syncope despite have a long vowel in its lasts syllable. This could be some kind of archaic or irregular strong declension of the dual. Alternately, it could be a slip or a misreading for the regular dual *nuphârat.

Dual Usage

Like Quenya, Adûnaic uses the dual for natural pairs, such as two hands (pât) or two pillars supporting a single object (târikat). Where the two objects are unrelated, Adûnaic instead uses the number “two”: satta (SD/428). Two soldiers that are part of special unit might be nardûwat “soldier-pair”, but two unrelated soldiers would be nardû satta “two soldier[s]” (Adûnaic numbers follow the noun).

Unlike Quenya, the Adûnaic dual can also be used for two unlike nouns that were “contrasted or opposed” (SD/428): sun and moon (ûri and nîlu), heaven and earth (minal and dâira), Manwë and Varda (Amân and Avradî). Such duals can be expressed in two ways (SD/428):

• The nouns could be compounded and then declined into the dual: minaldâirât “heaven-earth-pair”.
• The first noun could be declined into the dual, with second noun omitted but understood: minalat “heaven-(earth)-pair”.

In the second cases, an otherwise unique item in the dual would be understood to be pair with its natural counterpart. Since “pair of heavens” makes no senses as the meaning of minalat, it must be paired with the Earth. Similarly, ûriyat makes no sense as “two suns”. Where unclear, the second noun could be added in the singular: minalat dâira (SD/428).

[The discussion above assumes that minal does not undergo the Adûnaic syncope so that its dual form is minalat instead of minlat. Neither dual is attested, however, so either is possible.]