Noun & Adjective Inflections
The existing corpus of Khuzdul is almost entirely nouns and a few adjectives, with absolutely no verbal inflection shown. The purpose of this section is to look at what inflections exist in Khuzdul nouns and adjectives, and/or which ones are probably. In the pages following this one where I analyze each Khuzdul example individually, the information here should provide an understanding of why I interpret the words in the way that I do. This is highly important, because without a solid analysis of the attested lexicon, an extension of Khuzdul will have almost no hope of capturing the language's aesthetic.
Nouns and Adjectives are the Same Part of Speech
Before listing out word forms for nouns and adjectives, it should be understood that they are, in all probability, the same parts of speech. In Arabic and other Semitic languages, there are several different classes of words, as there are in English. However, they are organized differently. A noun or adjective in Arabic is called an "ism". Some words that would be considered adjectives in English are actually considered as verbs in Arabic (and other Semitic languages).
The main reason this matters here is that, as a result, nouns and adjectives share the same inflections, which are listed below.
Singular - As should be expected, Khuzdul has a singular number. Tolkien says that several Khuzdul words are singular, notably khuzd "Dwarf". The singular, as is familiar to most people, represents a single unit of something. It is also considered the "least marked" form; that is, the basic form a noun or adjective will appear in. This is the form that will appear in the dictionary. Hebrew, Arabic, other Semitic langauges, and Adunaic all have a singular number, so it's not surprising that Khuzdul does as well. As in Arabic, the singular consists of "templates" of vowel patterns for the consonantal roots that is paired with another template for the plural number. As previously noted, Tolkien said that Khuzdul resembles Arabic in its broken plurals, so I will follow Arabic's example of the singular here as well. Without it, the plural form wouldn't be "broken".
Plural - Tolkien states for a number of Khuzdul words that they are plurals. Again, the main examples would be Khazâd "Dwarves" and baruk "axes". The plural will be mostly "broken" in Arabic style, following the above statement. The plural is more than one unit of something: anywhere from two to infinity.
Collective - Tolkien wrote that the word dûm is "either a true plural or a collective singular". Arabic makes use of a collective number. A collective is identical in form to the singular, which is described above. Like the singular, words that make use of the collective number will show the collective form in the dictionary because it is the least marked. However, unlike the singular, the collective represents more than one unit of something, just like the plural. Think of the word "fish" in English as an example. Because Tolkien mentions that dûm is either plural or collective, Khuzdul seems to mimic Arabic's noun formations, and Arabic makes extensive use of a collective, it's easy to see the collective being used in Khuzdul. It becomes even more likely when you start to analyze words in light of the other numbers that go with the collective. Note that nouns that have a collective form do not have a singular or plural form. Instead, they have a "singulative" and "plurative" form described below.
Singulative - Languages that have a collective number, including Arabic, generally have a form which identifies a single unit of a given collective. This form, called the "singulative", is more marked than the collective, usually through some kind of affix. A good example is "djinn" (more than 1) vs. "djinni" (only 1), where the single djinni is more marked than the collective using the -i suffix. In Arabic, nouns in the singulative number are called "nouns of unity". Khuzdul's singulative marker will usually be -ûn, but other forms such as -ân or -în will occur. Evidence for this is in found in Tumunzahar, Buzundush, Tharkûn, Nargûn, Gabilân, and Nuluk-khizdîn. See those individual descriptions to see why I believe Khuzdul has collective and singulative numbers, along with the singulative affix -ûn.
Plurative - Languages that make use of collectives and singulatives often also have a plurative number. The plurative would take the same form as the plural above, but will essentially be a "plural of paucity". This means it is used for more than one unit of something, but it is limited to a small number. In Arabic, the plurative is used for entities numbering only up to 10 or so. For more units that (approximately) ten, the collective number should be used.
Dual - Hebrew, Arabic, and Adunaic all have a dual number. Adunaic apparently gets its dual from Quenya since they both have a dual number formed by a -t suffix. Also, both Adunaic and Quenya use the dual in much the same way as many Semitic languages, which is mostly for objects that occur naturally in pairs, like hands. Because Khuzdul was said to be most similar to Adunaic where Adunaic differed from Quenya, I'm going to leave out a dual form. It doesn't seem to add that much complexity to the system anyway, most of which comes from simply having broken plurals.
Definite - Khuzdul shows no evidence for a definite article, which in English is "the". Vowel changes could potentially indicate definiteness (khazâd vs. khazad), but Tolkien refers to this as "composition form". Almost all of the corpus is names, and the only phrases we have can be interpreted as not requiring a definite article. I have a hard time picturing Khuzdul as a Semitic style language without a definite article, so I'm including it in Quasi-Khuzdul. Adunaic apparently doesn't have one, or Tolkien doesn't write about it at all, so I may be bucking the trend a bit here of making Khuzdul similar to Adunaic where it differs from Quenya (which has a definite article).
Indefinite - Khuzdul definite shows no evidence of an indefinite article, either (English "a" or "an"). However, this is not at all surprising because Hebrew & Arabic only have a definite article. The result is that the same will be true for Quasi-Khuzdul. The indefinite will be considered the basic state since it is unmarked.
Composition - Tolkien mentions this in reference to the difference between Khazâd "dwarves" and Khazad-dûm "Dwarf-mansion". Note the difference in a long vs. short A. The composition form seems to be pretty much the same thing as the Semitic construct state or "bound form". There, the bound form is used in the formation of genitive expressions (as in "X of Y"). Khuzdul's composition form also seems to obey similar formation to the Hebrew bound form in that it comes from vowel reduction. Biblical Hebrew especially uses the bound form to indicate genitive relations such as ownership, and therefore does not have a "genitive case". Khuzdul also shows no evidence of a genitive case, per se. Look at the phrase Baruk Khazâd "Axes of the Dwarves". The second word, Khazâd, doesn't have an extra suffix compared to other places we see it. Tolkien does describes a suffix -ul, found in the word Fundinul from Balin's tomb inscription, as a "genitive ending of patronymics". That suffix can be viewed more as an "adjectival" suffix, which is similar. They have slightly different uses, though. Khuzdul then seems to use the composition (bound) form and apposition to indicate genitives. It appears to be more similar to Hebrew than Arabic in this respect.
Nominative - The nominative case is used for the subject noun of a sentence, be it transitive or intransitive. In English think of "I" vs. "me", where "I" is nominative. Hebrew & Arabic both have a nominative case, and it is the unmarked form. Adunaic is different in that is marks the nominative (through suffixes and/or vowel changes) and calls it the "subjective". Tolkien stated that mênu is "plural, accusative 'you' ". The vast majority of languages that have an "accusative" case have a "nominative" case as well, so we can safely say that Khuzdul has this.
Accusative - As just mentioned, Tolkien refers to mênu as "plural, accusative 'you' ", so Khuzdul obviously has an accusative case. Hebrew & Arabic have it, so Khuzdul follows suit. The accusative is used as the object of a verb. From the previous examples "I" and "me", it would be "me" that is the accusative. For Khuzdul's accusative case, the form will be a -u suffix or infix. This comes from what is found in Adunaic's "objective genitive". Khuzdul seems to have the same construction, seen in "Felak-gundu", where gundu "cave" receives the action of the felak "hewer". Another example is Uzbad Khazad-dûmu "Lord of Moria", where the radicals Z-B-D apparently signify a verb meaning "to rule, reign over", and uzbad would then be "ruler, lord". Khazad-dûm then receives the action of the lord, which is "ruling". Note that, here, Khazad-dûmu has the extra -u suffix compared to other places we see the name. That -u suffix appears in gundu and mênu, and also is the same in Adunaic. That can't be coincidence! If the form were simply Uzbad Khazad-dûm, without the -u, then the meaning would probably change to something more like "a lord from Moria". That is, one who is simply associated with Moria in someway, but doesn't act upon Moria in any significant manner, at least that is indicated in its root meaning. Also noteworthy is that apparently the -u accusative ending doesn't cause a shortening of vowels in the root word, unlike what most suffixes would. Perhaps that is because this suffix isn't changing the basic meaning of the word to which it is attached.
There is no direct evidence that Khuzdul has gender, but that doesn't necessarily mean it isn't there. Quenya and Sindarin have no gender, while Adunaic has some gender that is limited to natural, "semantic" gender. Hebrew and Arabic both have gender systems, with "grammatical" gender being especially prevalent in Hebrew. Including gender that had grammatical, morphological, and semantic aspects to it would make nouns and adjectives much more complex, and Khuzdul was said to be very difficult.
Khuzdul shows a hint of possible gender in Nargûn, Tharkûn, Gabilân, and Nuluk-khizdîn. The difference is found in the vowels that make up the "suffix of specificity", which is -ûn / -ân / -în. To some extent, affixes in Semitic languages can be found with different vowels for the same affix. It could be viewed that the consonant is the "true" affix, and the vowel is added after as a necessary vocalization. On the other hand, because the suffix seems to make the root indicate a single entity, it could be seen as evidence of a gender system. Viewed this way, it's interesting that there are three different vowels. Is this three genders then? Hebrew and Arabic only have masculine and feminine, but Adunaic has the "neuter" gender as well, even though Adunaic's system is limited. Yiddish, though, does have three genders - masculine, feminine, and neuter - which is grammatical, morphological, and semantic in nature. Such a system could appear in Khuzdul simply due to the small body of Khuzdul examples. If -ûn / -ân / -în are viewed as indications of gender, then a Yiddish-style system would fit there and also explain the three genders that show up in Adunaic.
Looking closer, Adunaic uses -i(-), -î(-), and -ê(-) as feminine affixes. In Arabic, the names of towns, cities, and countries are usually feminine. Nuluk-khizdîn fits into that category, and it has the -în suffix. Adunaic apparently uses feminine forms for countries as well, as seen in Anadûnê "Westernesse", as opposed to the adjective anadûni "western". Similarly, the affixes -u(-), -û(-), and -ô(-) are used for masculine nouns in Adunaic. The dwarves referred to Gandalf as Tharkûn, which should therefore be masculine. The name for Mordor, Nargûn, also shows the same vowel, and seems to contradict the use of a feminine form for countries. However, it may be that Nargûn was the dwarves' name for Sauron, and Mordor was so closely associated with him that the two were used as synonyms. That leaves Gabilân, which would have a neuter gender since it has a different vowel in the suffix -ân, and the other vowels seem to be used for masculine and feminine. Again, this lines up with Adunaic, which uses the affixes -a(-) or -â(-) for its "common" gender, which is a combination of both masculine and feminine. Any vowel may represent the neuter in Adunaic, so again the -ân in Gabilân makes sense there.
The other words that might have the singulative suffix (or "suffix of specificity") are Tumunzahar and Buzundush. It's very easy to suppose that the stem words tûm and buz (or bûz) are grammatically masculine, and therefore take the masculine -ûn suffix. Words that are grammatically neuter or feminine would then take the -ân and -în suffixes to form the singulative number respectively.
Finally, the word aglâb may have a suffix -âb. This might be a neuter form since it uses <â> just like -ân, and its easy to view "language" as a neuter. There could very well be masculine -ûb and feminine -îb forms of the suffix. Alternatively, -âb could be invariable and used to form collectives as well as derive feminine nouns.