Lalaith's Guide to Adûnaic Grammar

by Lalaith <>



Part I: The Three Adûnaics

Beyond a few preliminary scraps that are represented only by a couple of personal names in the various versions of „The Drowning of Anadûne“, the evolution of Adûnaic is marked by three important steps of development. It seems noteworthy that each of them yields a different word for „earth“: kamât, daira, and *aban, respectively (which does not necessarily mean that one replaced the other). For convenience, we may use these words as sort of a "key fossile" and reference the evolutionary steps by their initials as k-Adûnaic, d-Adûnaic, and a-Adûnaic. (Lowdham frequently writes Adunaic but here we agree with the translator of the Red Book and write Adûnaic throughout.)

k-Adûnaic is represented in early drafts for NC: It is a strongly inflecting, "indo-european"-style language. (Inflecting means that much of the vocabulary changes its appearance according to number, gender, case, tense, or other symptoms. Latin, Russian, or German are good examples of inflecting languages).

d-Adûnaic is the kind that Arundel Lowdham described in RA. Other than its predecessor, it is an agglutinating language that Lowdham attributes with „a faintly semitic flavour“ (NC), though similarities to, say, Hebrew grammar seem to be at least superficial. (Agglutinating means that the stem of a word may accumulate lots and lots of affixes in front, behind or even inside, and so express relationships that inflecting languages use prepositions for. Turkish and Finnish are examples of agglutinating languages. This kind has the profound disadvantage that you cannot take a dictionary and simply look up an unknown word: You first have to decide what may be the proper stem.)

a-Adûnaic is the „official version“ attested in the LR Appendices and in LE. It is known only through royal names and a few other specimen such as adûn „west“, pharaz „gold“. We do not know anything about its grammar but can fathom that some of the known vocabulary is not constructed with the grammar of d-Adûnaic in mind.

Mr. Lowdham in particular discusses the grammar of a kind that he calls „Classical Adûnaic“, seemingly that kind that was spoken in Númenor at the time of its Downfall. In this essay, we will first examine what Arundel Lowdham tells us about the grammars of k- and d-Adûnaic, apply the results to the "Lament of Atalante" that was in fragments found among the Notion Club Papers, and finally have a look on how a-Adûnaic might have changed the context that we know.

The Noun

Lowdham gives in his report many tables of nouns declined through the various cases, numbers, sexes, etc. There is nothing to add to them, and so there is no need to repeat them here, and we may just summarise the general rules and trends.

(A) The good news first: A couple of features that many European languages love to bother us with are fortunately absent from Adûnaic of any kind. For example, as Mr. Lowdham nicely put it: Adûnaic has no gender, but it has sex ("there is not strictly speaking any 'gender' in Adunaic") - by that it quite corresponds to the English language. This means that inanimate objects generally are devoid of grammatic genders and useless questions like "Why is a table female in France (la table) and male in Germany (der Tisch)?" do not need to be asked by pupils of Adûnaic. Another comforting feature of the grammatic sex is that it may determine the shape of nouns and pronouns, but not of other word forms such as adjectives - only that Lowdham tells us a little later that the system of suffixes which are used as "m[asculine] and f[eminine] signs runs through all Adunaic grammar". Well, it was him who said he understood Adûnaic.

The bad news is that other than in English, there are as many as four sexes. But they are not so difficult to grasp, as you will see:

  1. Masculine. This sex denominates everyone and everything male, including eventual professions and titles: Men, Elves, Balrogs, and animals (but not plants. How would a Númenórean address Treebeard?). Masculine nouns can be identified either by their last vowel being -u(-), -û(-), or -ô(-) which developed out of archaic -au(-) (a sound-shift that can also be observed in the development of vulgar Latin), or by certain suffixes that are considered „male“. Note that these suffixes do not apply to personal names which usually end in consonants regardless of sex.

  1. Feminine. Analogous for girls, Elf-maidens, Ungoliant, etc. Feminine nouns prefer -i(-), -î(-), and -ê(-) (from archaic -ai(-)) as their last vowel, but also certain other suffixes.

  1. Common. The sex that ultra-feminists regret to miss in modern languages. It strictly incorporates both sexes and thus is used whenever you want to leave the actual sex of the person(s) in question unspecified. Númenóreans use the Common when talking about mixed couples and groups, races and species, probably also about professions of unspecified gender - though we may seriously doubt that p.e. tamar "smith" ever developed a Common derivative. Many Common nouns, Lowdham says, feature -a(-) or -â(-) as their last vowel. But actually, these two vowels are the notorious troublemakers in the Adûnaic language, for they can as well occur in any other sex. So, word by word, we have to explicitly memorise the sex of a noun that ends on -a(-), -â(-).

  2. Neuter. Everything that is neither explicitly male or female or both: objects, abstract terms, philosophical ideas - and plants, p.e. kulub "root". It could also be called the „inanimate“ sex as was the case in Adûnaic's late descendant, Westron. Grammatically, the Neuter is also the least clearly specified sex, for any word on -u(-), -a(-), -i(-) or their long variants can be Neuter. The usage of this sex is more or less identical to that of the preposition „it“ in the English language, but unfortunately, it is rather inappropriate to describe mythological concepts. The Númenóreans therefore enjoyed to develop divine or angelic personifications by attributing a Neuter noun with one of the other sexes. We may thus for example construct the genius loci of a land or a city: the adjective anadûni "western" assumes a Feminine suffix and becomes Anadûnê, lit.: "She, Westernesse"; the Neuter noun agan "death" assumes a long â and becomes the personification Agân "Death" (would that be Mandos???), nîlu "moon" is personified as Nîlû "Man in the Moon" (the one whom fishermen sometimes find in "the windy bay of Bel"). Most interestingly, the Adûnaic word for God, Êru, can thus be perceived as either neuter or male.

(B) In addition to sex, Adûnaic nouns are altered by number. And worse than in most modern European languages, there are three of them, for the ancient Edain counted „one, two, many“:

  1. Singular, one item at a time. This should not pose any problem, but alas, sometimes a word that looks like a singular is in fact not. Lowdham insists that p. e. gimil is not one star but defines the totality of stars, vaguely similar to the collective plural of Elvish, while „one star“ actually was expressed by gimli. A corresponding example seems to be found in the two attested forms azar vs. azra, "sea".

  2. Dual, a pair of them. This one is the trickster among the Adûnaic numbers. The Dual is only used if two items actually belong together (p.e. two parents) or contrast (sun and moon, yin and yang, and so on). Other than in those European languages that still feature a dual, you must not use the Adûnaic dual if you only talk about two objects that only accidentally come together. For example, one of your parents, either father or mother, is nuphar. Both your parents are named by using the dual nuphrat. But if you happen to discuss one of your parents vs. one of your friend's parents, then you have to use the singular and add the number 2 behind: nuphar satta. Lowdham in a struggle for perfection gives many interesting examples of the Adûnaic dual, such as huznat "two ears", nuphat "two fools" (what an Adûnaic youngster would say for nuphrat?), and last not least banâthat "[my] two wives".
    If this should cause trouble, remember that English and some other languages also retain odd relics of an ancient dual. As soon as you put on your „sunglasses“, you are wearing a dual right on your nose. Slovene, one of those living language that still maintain a full-fledged dual, notoriously sees two pub doors (vrata) where there ought to be one. Don't ask me why.

  3. Plural, many items. Plurals enjoy their last vowel to be -î(-), quite regardless of sex. This should as well not be very problematic if you consider one caveat: Where two objects do not belong together, as in nuphar satta, you have to give the noun in Singular number and not in plural. But if you talk about three or more objects, your choice is the Plural number, thus we write gimlî hazid "seven stars" but gimli satta "two stars", balîk hazad "seven ships", and so on (Lowdham contradicts himself in whether the correct form is hazid or hazad. It seems conceivable that hazid was an irregular Subjective formation, but such an explanation attempt is probably too forced.)

The curious word order of noun + numeral results from the fact that other than in English, numbers higher than 1 are not adjectives but other nouns (Lowdham forgot to tell us what kind of word the 1 would be). The items that you are counting you do arrange in front of the numerals because you are actually constructing the equivalent of a genitive according to the Adûnaic rules of grammar. The proper translation of such constructions is "Two of ships", "Seven of nine", etc. Now, as your homework you will translate "Two out of seven ships".

(C) The Adûnaic cases probably require the most effort. Not only is their structure thoroughly different from those which are found among most European languages, but they are also very distinct in k- and d-Adûnaic.

  1. Normal. The Normal case, present in both k- and d-Adûnaic, is of course the standard form of a noun, that one that you would find listed in an Adûnaic dictionary if there was any. You also find it in most places of a standard Adûnaic text. Sure, good grammarian that he is, Lowdham classifies many a situation in which a Normal is required. But why bother? Say that except for a very few cases the Normal is your choice, and leave it at that.

  2. Subjective. The Subjective case puts an emphasis on a noun. In cases where English speakers help themselves with constructions like: "This very one", "Him it was who...", Adûnaic replaces the Normal by a Subjective noun. US-marines prefer to talk in such Subjectives: "This recruit here has the trousers full of dung." Lowdham also insists that a Subjective is the subject of a verb, but this is somewhat misleading, for the Subjective can also replace the very verb where it ought to be a form of „to be“. Adûnaic behaves somewhat like Russian here. Better it was to say that the Subjective case marked the subject of a phrase.
    Most often, Subjectives either end on some vowel + n or they alter the last occuring vowel. We met already one possible case of a Subjective above in the word hazad „seven“ that belongs with the also attested Normal hazid. Note that Dual nouns form a Subjective only if they are of Neuter sex.
    (Hebrew has an equivalent of the Adûnaic Subjective: That is probably one reason why Lowdham attributed to Adûnaic "a faintly semitic flavour". For the same reason you may grant it a Scandinavian flavour, however.)

  3. Genitive. This case and the subsequent ones, present only in k-Adûnaic, are identical to those found in European languages. Like in English the Genitive answers questions that begin "Whose...?". k-Adûnaic masculine Genitives append to the Normal form but after vowels -vo. Example: bârun-adûnô, „Lord of the West“, from the Normal adûn.

  4. Dative. This case, like in English, answers questions that begin "To whom...?". k-Adûnaic Masculine Datives end on -s or, after consonants, -us. Example: dalad Ugrus, „under [the] Shadow“, from the Normal ugru.

  5. Instrumental. This case does not occur in English but it is familiar to Slavs. It answers questions that begin "With what...?" k-Adûnaic Masculine Instrumentals end on -mâ or, after consonants, -umâ. Lowdham insists that this suffix once was simply a word meaning "with". Example: sôbêthumâ, „with assent“, from a Normal *sôbêth.

Note the absence of one familiar English case from k-Adûnaic: the Accusative, answering questions that begin with "Whom...?" Lowdham's fragments give no hint to how it was expressed in Adûnaic.

d-Adûnaic does not possess cases (3) to (5). It rather imitates them by applying various prepositions to the noun, either before or after, in the way languages like Finnish or Hungarian do. A Dative is now expressed by adding -ô, -vô, a Genitive by supplying an-, 'n- before the word, and also simply by placing the noun in question right before the one it is related to (see our above case of counting items: gimli hazid. Sometimes, English may work the same way: kingfisher, a word that Tolkien discussed in L240, was actually a king's fisher), and the Instrumental still by adding -mâ which is now just no longer considered a case declension. Instead, d-Adûnaic has a new and rather unusual third case:

  1. Objective. We do not meet it so often, fortunately, for it appears only in compounds or in arrangements that serve as compound, p.e. noun + verb that has become a noun (p.e. „ship-building“) or noun + participle. The Objective is most of the time recognisable by an -u- as the last vowel of a noun. Not any compound features an Objective, though, for its presence depends on whether both nouns are simply aligned or actually have a link in meaning to each other. Lowdham gives as an example Minul-târik „Pillar supporting heaven“ vs. Minal-târik „Pillar [of clouds] in heaven“. If that sounds weird to you, you will see that we can do the same in English. Consider again the kingfisher of L240: d-Adûnaic uses the Objective arûn- here if the intended meaning is "someone who is fishing for the king" but the Normal arun- if the meaning simply was „a king who fishes“.
    One caveat, though: An Objective is always Singular even if its implied meaning is a Plural. Of course, d-Adûnaic just behaves as illogical as English here. Nimruzîr thus is certainly a friend of many Elves and not of a particular one, and yet his name is notoriously translated "Elf-friend" rather than „Elves' Friend“, Ar-Balkumagân is known as the "Shipwright" or "Ship-Builder" even though he fas reknown for building a lot of ships. If you really need a proper Plural within a compound then you only can resort to the Normal case and have to use some kind of infix such as an- to indicate what case you think is proper here.

(D) And then, Adûnaic distinguishes strong and weak nouns. Now this one is a real pet among grammarians: Strong nouns are too weak to resist internal deformation by case, number, or other grammatical features; weak nouns are too strong to allow for such intrusions. Any further questions? Well, for that reason grammarians lately prefer to talk about irregular and regular declensions instead. If you want an example, Proudfeet is a Strong or Irregular declension, Proudfoots is a Weak or Regular one.

Many modern languages seem slow by slow to abandon Irregular declensions because people consider them useless complications that only burden theirr memory. The vast majority of English or German nouns have by now acquired Regular declensions; new coinages always seem to be Regular. Adûnaic displays the same tendency. Regular inflexions have long become the rule in Classical Adûnaic and among the Masculine, Feminine, and Common sexes they all but completely replaced the Irregulars. Also, Neuter nouns that are either monosyllabic or have "a long vowel or diphthong in the final syllable" (RA) are all regular.

The proper inflexions of the Adûnaic noun depend on the last vowel of the stem, here marked in red for clarity: huzun "ear", azra "sea", zîr "love(r)", and so on. The Regular declension is quite easy to memorise.

  1. Singular declensions. The Normal case displays the standard, unchanged form of a word. Subjectives add a suffix: Masculine -un, Feminine -in, Common -an, and Neuter simply -a. Most Objectives ending on consonants add an -u to Masculines, Commons, and Neuters but -i to Feminines (not for any argument of logic but because the -u was routinely mistaken for signifying Masculine sex) while Objective ending on vowels replace this one by -u. But if any noun that is not Neuter ends in a (long) vowel, then the Objective case is identical to the Normal case.

  2. Dual declensions. Dual nouns add -ât throughout. There was once a group of words which used -at in the Normal case, but this had long since been dropped. Lowdham is unfortunately totally obscure with regard to existing distinctions between Normal and Subjective Duals. He does mention that Neuter nouns which display a long first vowel originally retained the last vowel in the Normal case but dropped it in Subjective. To illustrate this he gives an ancient Normal târikat. But how would Classical Adûnaic treat it since there is no suffix -at any longer? Does that mean there is a Normal Dual *târikât and a Subjective Dual *târkât? An as obscure statement seems to suggest that non-Neuter nouns use to shorten last long vowels in the Normal Dual: nardû "soldier", if this is interpreted correctly, thus produced the Normal Dual narduwat > narduwât and the Subjective Dual nardûwât. Strangely, Common nouns ending on seem to produce only a short -a- in the Dual. But this latter may be a typo or an error in the interpretation of the source texts.

  3. Plural declensions. The Normal Plural usually just adds an -î. But non-Neuter nouns ending in vowels behave differently and thus apparently serve to Lowdham as markers to classify his grammatical arrangements. Unless this is not another error, Normal Masculines ending in seem to be undecided about whether or not they should shorten the vowel when a glide consonant is inserted, and so we meet both narduwî and nardûwî (now what is a glide consonant? Adûnaic like Croatian abhors certain kinds of diphthongs and inserts a similar, hardly spoken consonant between the two vowels, resulting in narduwî rather than *narduî). On the other hand, Normal Masculines ending in always retain the long vowel, insert no glide consonant and add a short -i, thus manô develops manôi and not *manôwî. Further, Normal Feminines on do not add a Plural marker at all and thus look exactly like their Normal Singular form, and Feminines on develop an unusual glide consonant: izrê > izrênî – or should this be a typo for *izrêyî? Unfortunately, Lowdham fails to describe the behaviour of Normal Commons on -â. But from what he otherwise says we may deduce that they behave similar to Normal Commons on , and so, the Normal Singular of Adûnâi „Númenóreans“ was apparently *Adûnâ. Are you still with me?
    Despite the terrifying fuzz Lowdham makes about the Subjective Plural it is in fact very easy to handle: add -ya to the Normal Neuters and -m to all the others.

So much for the easy part. Now the Irregular inflexions fall apart into several categories or Classes which other than with the Regular inflexions never merged into one. These Classes are distinguished by whether or not their stem ends in a vowel, and in the plural forms they behave quite distinctly. Fortunately, we may recall that among the non-Neuter sexes, Classical Adûnaic had mostly erased out all the irregularities, and so, the definitions below concern only quite few words.

  1. Class I behaves irregularly only among Neuter nouns that have three consonants (i.e. are bisyllabic, like kalab „fall“) and end in one of them. In the Normal, their last vowel also always is short, and so, they derive their Subjectives simply via "a-fortification" of this vowel. Where necessary, resulting diphthongs are merged: -ai- > -ê-, etc. The other sexes all inflect the Subjective regularly, that is, they add those standard suffixes that we already discussed. Short last vowels are getting lost during this process. The Objective case replaces the last vowel with -u-; but this irregularity has much dwindled and is in Classical Adûnaic mostly replaced by regular declensions, either maintaining or dropping the last vowel. The Dual always is regularly declined. In the Plural, Normals replace long last vowels by -î-; the Subjective adds the suffix -a to the Neuter sex and -im to any of the others. In fact, the Subjective behaves mostly like in Singular; only where the Normal Singular has a short last vowel there the Subjective Plural drops it rather than replacing it by -î-.

  2. Class II correspondingly concerns nouns that end with a vowel; their stems may hold either two or three consonants. There are in this class no Neuter nouns that end on a long vowel. In Singular, their Subjectives use the a-fortification as they do in Class I while the other sexes again apply the regular suffixes. However, where a diphthong occurs it does in this Class not merge into one long vowel but kills off the first one: thus raba "dog" produces the Subjective raban, not *rabân. Quoth Lowdham. One may of course also say that the non-neuter Subjectives simply add a suffix -n to the Normal and leave it at that. The Objective replaces the last vowel in the Singular by -u (an exception is the word anâ "human being" that has an Objective anû-. This is the last irregular survivor of another, vanished, subclass.) The Dual always replaces the same vowel by -ât. Similar to Class I, Class II Plurals replace the last vowel by (except again for anâ that also developed a regular Plural anâi), Subjectives add a suffix -ya to Neuters but -m to other sexes, so that in contrast to Class I Subjectives they retain the Normal's -î- at the end of the truncated stem.

Note: Lowdham presents the history of the Class II Dual in a very awkward manner, making over-extensive use of the multiple meanings of the word "later". What he wants to tell is in fact simply this: Ancient Adûnaic gained out of the Neuter Singulars azra, gimli, nîlu the Normal Duals azrat, gimlat, nîlat (one may wonder under which conditions a Númenórean saw nîlat - "double moons"!?!) and the corresponding Subjectives azrât, gimlât, nîlât. While centuries passed, the idea stuck fixed in the mind of Númenórean pupils that the final -a/â- belonged to the suffix and should not be declined; this was an evident confusion because the non-Neuter sexes of the same class indeed featured suffixes of such kind. So, Middle Adûnaic invented a new kind of Normal Duals: azrat, gimliyat, nîluwat as well as corresponding Subjectives. Later-on, when pupils became even worse, as with all other nouns this system eventually eroded away and in Classical Adûnaic left only the Normals azrât, gimlât, nîlât.

Well, that was the worst of it. If you need to know anything further, I refer you to Mr. Lowdham's report.

The Verb

Lowdham's explanation of Adûnaic grammar never advanced beyond the noun and a few glimpses of other word forms. All information that we have on the verb are a few notes that CT summarised like that (some paragraphs inserted to increase clarity):

"There were three classes of verbs:
I Biconsonantal, as kan 'hold';
II Triconsonantal, as kalab 'fall down';
III Derivatives, as azgarâ- 'wage war', ugrudâ- 'overshadow'.

There were four tenses:
(1) aorist ('corresponding to English "present", but used more often than that as historic present or past in narrative');
(2) continuative (present);
(3) continuative (past);
(4) the past tense ('often used as pluperfect when aorist is used = past, or as future perfect when aorist = future').

The future, subjunctive, and optative were represented by auxiliaries; and the passive was rendered by the impersonal verb forms 'with subject in accusative'."


This sounds like approaching nightmare. But only so because grammarians enjoy to describe languages in such a way that they all look like Latin. For what the above statement tells us indeed is that the Adûnaic verb is a very simple thing and lacks many of the intricacies that ridiculously complicated languages like French or Macedonian bother us with.

The three Classes are not difficult to grasp at all. Classes I and II use a base directly as the stem of a verb. The bases are those things that Lowdham somewhat inconsistently writes VERSALLY: They never were actual words but are an agreement among grammarians to arrange words of common origin in their dictionaries. For example, "verb", "verbally", and "verbose" all have a common base VERB. Now, Class I features bases that display two consonants (and one or more vowels) while Class II has three of them. Both classes were presumably quite small in comparison to Class III, for therein, everything is found that distorts a base rather than using it unchanged. Lowdham gives a good example: azgarâ- "wage war" is evidently derived from the base ZAGAR (or Z'GAR) "sword". We may expect that this class features various subdivisions, strong and weak verbs, irregularities etc., but of those we do not learn anything.

(We may then also assume that there are regular and irregular verbs like there are regular and irregular nouns. But about these we have no informations at all.)

The Tenses may look funny to the eye of a European, but they are not that difficult, either.

  1. The Adûnaic past tense will of course not mean any problem. Like in English, it is used to tell of something that happened in the past, wherever in time the narrator's viewpoint is. In comparison, note the behaviour of the word "come" in the following example: If your viewpoint is in the present, the past tense is in the past: "I tell you that he has come." If you told about it already in the past, the past tense is even further in the past: "I told you that he had come." (English grammarians call that a pluperfect, for they never learned to pronounce the correct expression plusquamperfect - Latin: "more than perfect"). And if you are going to tell that in the future, then the past tense is located in a future somewhat closer to your present: "I will tell you when he will have come." Quite convenient that one tense can do all these things.

  2. Now the aorist is a feature that occurs in languages like Ancient Greek, Turkish or Quenya, though each time it means something different. That's probably why the grammarians gave it that name, for it means "unlimited" and its content of definition is thus actually nil. In Adûnaic it may serve as the present tense („I come“) but other than in English is also the preferred tense of written stories and histories („historic present or past“). Well, of course Mr. Lowdham could simply have informed us that Dúnedain authors prefer to tell their tales in present tense, couldn't he? But of course, he would not be a good grammarian if he did. English, BTW, can do this as well. "Me goes straight back into the pub and hits him right-away on the nose". That's the straight use of an aorist.

  3. Continuative present (which is not called "continuative aorist"!) and past are virtually the same two things again. But they have a tiny difference. This time, they refer to on-going or repeating actions, that is, in those cases where English uses these dreadful things on -ing that we foreign pupils despise so much ("I learn" - "I am learning" - never?). English grammarians call that the progressive, their Russian colleagues dub it the imperfective aspect, describing continuous or repetitive actions in contrast to the perfective or one-time action. But we better leave that all alone...

I suppose most inexperienced students of Adûnaic will latest have swooned when they read that "the future, subjunctive, and optative were represented by auxiliaries". But no reason for that. As ominous as it may sound, it is simply a sophisticated demonstration of the talented grammarian's magic skill to bring forth things that in the discussed language are not there at all. What Mr. Lowdham actually tells us here is that Adûnaic does not have anything of these. For while many languages torment us with different inflections for all or some of these gadgets and demand from us to learn all of them by heart, there is no need to do that in Adûnaic: like in many civilised languages, in such cases we leave the poor verbs all alone and only add some indicators. That could perhaps be a second verb which tells us "read future tense here". And that second verb, most often a form of „to be“ or „to have“ or „to will“, is the auxiliary! If you compare: Ancient Greek has all the features Lowdham mentions and more, and that is why "it is all Greek to me" became proverbial among the humanistically educated. French and Italian have a future and a subjunctive but the Italian subjunctive also conveniently covers the optative. German has no future but two subjunctives (here called conjunctives) which are rapidly declining. English has none, and so does Adûnaic.

Let us take a standard phrase: "The king falls", and apply those features to it one after another. Obviously, we have two ways to tell something that will happen in the future. Either we use an aorist: "The king falls tomorrow", and rely that we will be understood AND survive the uproar among conservative grammarians. Or we add the little word "will" (and cut a tiny suffix): "The king will fall." - The little word "will" serves as our auxiliary: our grain of salt that we add to the phrase so that we may express the whole tense without having to bother about declensions, conjugations, and other frightening troubles.

Now subjunctives and optatives are together known as moods. This does not mean happy or angry but is an ancient misspelling of "modes" that, like the "pluperfect", English grammarians will never learn to set straight again. Our standard phrase "The king falls" they call the indicative mood, for it indicates what is really true (or what you think it is). The subjunctive mood is convenient if you are not convinced that it is true what you are saying (evidently, Anglo-Saxons use subjunctives VERY sparingly): "The king might fall". The optative mood expresses a wish, a desire, or a hope. "Long live the king" is one classic optative, another is: "The king shall fall". Now you may regret the lack of one mood that seems familiar from your nativ language: the imperative. How do you give orders in Númenor? Simply by wishing someone to do it, and so, your obvious choice is the optative (though, if your name is Ar-Pharazôn, it does not make much of a difference). Adûnaic behaves like ancient Hebrew here; the optative is thus the mood the Ten Commandments are written in, and you have to admit that "Thou shalt not kill" sounds more impressive than "Don't kill!", doesn’t it?

And finally there is the Passive. This belongs, together with the Active, to the Voice or Genus. The passive you use if you tell something that you do not want to be held responsible for. So, "The king fell" is an active, "The king was felled" is a typical passive.

With regard to Adûnaic it is said above that the Passive is constructed by using "the impersonal verb forms". This leads us to an important alternative approach to Adûnaic sexes. For we may now distinguish two verbal genders:

  1. Personal. A verb connected to a Normal noun must be personal, that is, some prefix signifying sex and number has to precede it. Lowdham's example is Bar ukallaba „The Lord fell“ where u- is the marker of the Masculine Singular person.

  2. Impersonal. A verb connected to a Subjective noun may be impersonal, i. e. either with or without the above prefix. Using a personal verb is thus not really wrong here but it increases the emphasis so much that if you overdo it you will sound like a fanatic prophet. Thus, Barun kallaba is more emphatical than Bar ukallaba but Barun ukallaba is even more.

(And if you now ask "Where is the Objective?", remember that it appears only in compounds and cannot go with a verb.)

So according to the above statement, the Adûnaic Passive is generally made from verbs that do not have pronouns in front. English uses in this case a participle ("felled"), but this may not be so in Adûnaic.

But there is one unsolvable problem left to us. What does the statement mean that the passive is accompanied by its "subject in accusative"? As we have learned above, the one case that Adûnaic never developed is the Accusative! Does that mean perhaps that the description of the Adûnaic verb belongs to yet another phase of development besides k- and d-Adûnaic that we know nothing else about? Hello, Mr. Lowdham?!?

The Adjective

We cannot tell much about Adûnaic adjective. In general, words like "old, far, crazy", etc. seem to behave the same way they do in English: Unlike verbs, they are not sensitive to sexes. Adjectives are all considered to be of common sex, but they react sensibly on numbers and in contact to nouns provide us with Singular, Plural, and probably also Dual prefixes. They seem as well to respect Normal, Subjective, and Objective cases: There is clearly an Objective involved in izindu-beth "prophet" (izindi "straight, true"), while burôda may be a Subjective of a Normal adj. *burôdi (this example was kindly suggested to me by Aleš Bičan).

Like in any language, there exists a predefined set of adjectives. But further ones can be formed out of nouns or verbs (i.e. participles) or even other adjectives. Thus, adûn "west" may first turn into a genitive an-adûn "of the west" and finally anadûni "western". This new adjective may become once again a noun: Anadûnê "Westernesse", and from that of course may follow *an-Anadûnê „of Westernesse“, and so on ad nauseam.

The Companions

Companions are little words that accompany a noun or replace it. They are commonly known as pronouns and as articles. We do not have a list of Adûnaic pronouns, but there are a couple of statements which allow some conclusions.

(A) As in any other language, we have to distinguish first person ("I"), second ("you"), and third ("he, she, it"). Probably there were also made distinctions between familiar and formal addressing, for Westron had such, but their nature is unknown to us. Fith regard to sex, Adûnaic behaves very much like p.e. Latin, for it "distinguishes gender (or rather sex) in the pronouns of the third person" but not in the others. That means that if you talk about yourself or address somebody, you do not have to question your or your opposite's sex. In the third person, however, the vowels found in the companions serve to distinguish the sexes: "u and i are the bases of pronominal stems for 'he' and 'she'", which, as early Adûnaic featured neither e nor o, leaves a to be used both for Common and Neuter pronouns. There also is a "variation between pronominal u- and hu-", arising from different stems. But we do not have any information about where either of them has to be applied.

(B) The companions also distinguish our usual numbers, Singular, Dual, and Plural. Different from English, they produce sex-specific plurals, so if you talk about some men, you will use a different "they" than if your objects are some women.

(C) There is no information on whether companions are sensitive to Case.

(D) Most often, the companions are getting affixed to other words, notably verbs (thus producing the Personal verb forms) and nouns. In pronouns of personal verbs, we again have to distinguish two classes:

  1. subject pronouns. The subject tells us who or what does what the verb says. In a phrase like "She loves him", "she" is the subject pronoun.

  2. object pronouns. In the above phrase, "him" is the object, that is to whom or what happens what the verb says.

NC seems to preserve one example of a verb with both subject and object pronoun: kitabdahê, *„you touch me“, from a verbal stem *tabad „touch“.

(E) We also meet fossilised pronouns in the Subjective case that "was originally made with pronominal affixes". These affixes seem to correspond somewhat to the demonstrative pronouns of Latin or the emphatical pronoun of Swedish, thus zigûr, "a wizard" produces zigûrun "the Wizard; Sauron". The full set that probably as well exists or once existed as independent pronouns, is:

  1. Neuter: in Singular and Dual there either occurs an "a-fortification of the last vowel of the stem" or -a is attached, Plural -iya

  2. Masculine: Singular -un, Plural -im

  3. Feminine: Singular -in, Plural -im

  4. Common: Singular -(a)n, Plural -im

(F) Pronouns may also serve to specify time, number, space, and so on, and grammarians enjoy to invent little filers into which they may sort any kind of companion according to their liking. Thus for example "here" is a local pronoun, "all" is - obviously - an indefinite pronoun, though Anglo-Saxon grammarians seem to prefer ranking it as an adjective, and "now", "once" etc. German grammarians classify as temporal pronouns but many English ones as adverbs (see below).

Articles are a baffling feature of Adûnaic. What does a language that has Subjectives and gender-suffixes require articles for? One obvious usage seems to be when you want to set an emphasis on the object of a phrase. In such a case, the Subjective case is not allowed, and the gender-suffix alone may not seem sufficient. And so, the difference between Akallabêth „She who has fallen“ (kalab „to fall“, *akallaba „something that has fallen“?) and hi-Akallabêth may perhaps be that the latter should rather read "It was She who has fallen". But as Lowdham tells us nothing about this peculiar issue, my attempt above remains the merest speculation.

The Particles

The German dictionary "Duden", vol. 6, gives a lovely definition of a particle: Take a text and scratch any noun, any verb, any adjective, and all the companions. Everything that is still left on the page is a particle.

The grammarians distinguish three main categories of particles: prepositions, adverbs, and conjunctions. However, if you did what I suggested to a d-Adûnaic text, you also wiped out all the prepositions, for these are words like up, down, before, of, at, etc, and they do not here exist as individual words: "Adunaic prepositions are in fact usually 'postpositions' following their noun" which is set in the Normal case, with the preposition suffixed to it. Lowdham's statement seems to imply a few exceptions to the rule, but we find such in k-Adûnaic only.

Adverbs define the circumstances of an event, such as "The king fell violently". In English, if you meet a word on -ly, it is an adverb, though not all adverbs finish like that. In Adûnaic, this is much less clear.

Conjunctions are little words like and, or, but etc. that serve to make VERY BIG phrases out of many very small ones. Plato, Cicero, and Umberto Eco use them to create single phrases that extend across several pages. Biblical Hebrew that did not know about dots and commas used a conjunction and/then to separate one phrase from another, and that in the translation produces the endless repetitive sound of the KJV. In Adûnaic, we do not meet conjunctions often for there only are very tiny phrases preserved.

Word order

(A) Adûnaic phrases generally display the same sequence as English phrases: Subject-Predicate-Object, Predicate being a nifty way to say "verb". Well, English grammarians usually say "verb" in the first place, but others don't agree, especially those who do not reckon participles as verbs – forget it... Sometimes, the word order may be altered so that the Object locates before the Subject. That happens when the two closely relate, as in the case of Adûnaic's simulated genitive or the English kingfisher in which the fisher is the Subject and the king the Object. But as a rule, the Object is never found immediately before the Predicate. Adûnaic, though, accumulates its pronouns - "agglutinates", as the grammarians say - into a single word: "sheloveshim".

If there is any uncertainty about the identity of the Subject in your phrase, remember that you may set it in Subjective case to make it obvious, while if you want to set the stress on the object you tie it up with an article.

(B) Adjectives "normally precede nouns", says Lowdham, so an expression like "the old man" you may just literally translate. But sometimes it happens to be found behind the noun and this then should mean "the man is old", though there seem to be examples where this is not so obvious, for participles, where they are met, always seem to follow their noun without implying a similar is-relationship.

(C) Auxiliaries defining the mood seem most commonly to be found directly in front of the Predicate. This is again similar to the English way of using them. Thus, in the example "The king has fallen", the auxiliary "has" precedes the Predicate "fallen".

(D) Conjunctions apparently do not affect the word order as they do not in English (unlike, for example, in German).

All said above is only true for standard indicative phrases. We have no information on how you ask a question in Adûnaic, for example.

Part II: The Lament

The Lament of Atalante exists in three distinct versions L1, L2, L3. Each of them is more or less completely given in NC. L1 and L2 are manuscripts while L3 exists in both a manuscript and a typescript version that will be referred to as L3m and L3t. They are closely related to each other but not identical, and CT felt unable to decide which of them was later. However, L3m seems somewhat closer to L2 than L3t, also, the latter displays a somewhat more complex grammatical structure. It might thus be safe to conclude that L3t was the final version of the Lament.

In the following, each version is given in full and then its grammar will be analysed. The versions are officially divided into "fragments" I and II; to ease reference, I will number the individual phrases as well. For example, I/3 refers to the third phrase of fragment I.

Differing from the way they are written in NC, I give personal names and titles with capital initials and add in red the glosses that Lowdham omitted. The macrons indicating long vowels in NC are, in accordance with the spelling system of WS, replaced by circumflexes, for not all browsers support the display of macrons.

Version L1

L1 is clearly using k-Adûnaic. The initial form was much amended, both by ink and pencil. NC gives the emendations in ink (bracketed here) but, very regrettably, not those in pencil.


1: Kadô Zigûrun zabathân hunekkû [> unekkû] ...
And so / the Wizard (Sauron) / humbled / he-came

Phrase I/1 resolves into the following components:

The phrase is very emphatical with a Subjective noun and a personal verb. Its proper translation was apparently "And so it was him, the Wizard, who humbled came". Or, as the translation of an Old English version of the same passage suggests: "... who was humbled and came". If this latter is true, then zabathân may be a verb *zabatha- with a suffix -an „is, was“.

2: Eruhîn udûbanim dalad Ugrus ...
The-Children-of-God / fell / under / Horror? Shadow?

Noteworthy in this phrase is the unusual isolated position of the preposition, much as if Adûnaic shaped its sentences like English.

3: Arpharazôn azgaranâdu Avalôi-men [> Avalôi-si] ...
Ar-Pharazôn / was waging war? / Powers on

4: Bârun-Adûnô rakkhatû kamât sôbêthumâ Eruvô ...
The Lord of West / broke asunder / earth / assent-with / of God

5: azrê nai phurusam [> phurrusim] Akhâs-ada. Anadûni akallabi.
seas / might flow / Chasm-into / Westernesse / fell in ruin.


1: Adûnâim azûlada ...
The-Adunai (Men of W.) / eastward

The proper predicate („escaped“? „fled“?) should apparently have followed azûlada but is missing from this fragment.

2: agannûlô burudan nênum ...
death-shade / heavy-is / on-us

II/2 provides an interesting order of words, for it seems to conflict with Lowdham's claim that adjectives normally precede the noun. Yet Adûnaic behaves just like English here: There is a difference in sound between „Heavy death-shade is on us“ and „Death-shade is heavy on us“. However, we would expect the Subj. agânnûlô to contain the copula, hence „Death-shade-is“. That this is not so seems to indicate a variation between k- and d-Adûnaic.

3: adûn batân akhaini ezendi îdô kathî batânî rôkhî-nam ...
West / road / lay / straight / lo! now / all / ways / bent-are

As in II/2, the standard word-order adjective-noun is reversed and some auxiliaries are added in order to place the stress on the adjective. English behaves just the same way.

4: êphalek îdôn Akallabêth...
far away / lo! now is / She-that-is-fallen

Remark that we meet here a case in which the predicate („is“) precedes the subject noun, something which according to Lowdham should never happen.

5: êphal êphalek îdôn Athanâtê
far / far away / is now / Athanâtê (the Land of Gift)


Listing the nouns (affixes here in red), we are able to add several further forms to the table of masculine declensions that Lowdham provided for k-Adûnaic:






With regard to the recorded verbs, we may admit the following observations:

Version L2

In L2, we find the huge revolution that turned k- into d-Adûnaic fully in place. L2 has no English glosses and looks like a draft to test how d-Adûnaic grammar would affect the Lament. In SD it was not published in full but only the differences to L3 were listed. From this list the following text can be reconstructed, later changes bracketed [], glosses added in red. The numbering of fragments and phrases corresponds to that in L1.


1: Kadô Zigûrun zabathân unakkha [> yadda > unakkha]...
And so / the Wizard / humbled / he came [> went > he came]

I/1 features the only change in the shape of the predicate: With the declared absence of short -e- from d-Adûnaic, a verb unekku was no longer possible and so it became unakkha, temporarily replaced by yadda that we meet again in II/3 with the translation „went“.

2: Êruhînim dubbudam [>dubdam] Ugru-dalad ...
The Children of Êru / fell / Shadow-under

Eruhîn for similar reasons becomes Êruhînim and at the same time assumes a Subjective case (Êru- may in fact represent an internal Objective, for the Obj. of a Normal Êru- is again Êru-). Hence it may be followed by an impersonal Plural dub(bu)dam, the successor of udûbanim with a slightly changed base. The old Dative is of course lost, and thus the preposition becomes, in accordance with Lowdham's statements, a postposition: Ugru-dalad.

3: Ar-Pharazôn[un] azagrâra/azaggara Avalôiyada ...
Ar-Pharazôn / was waging war / Valar-against

The change to Subjective seems to represent the very decision that only this case may tolerate an impersonal noun, otherwise azagrâra/azaggara would have had to be personalised by a pronoun. We may wonder how it would have looked like: certainly *uazaggara would have looked rather uncouth to a Númenórean?

4: Bârun an-Adûn urahhata [> urahta] dâira sâibêth-mâ Êruvô ...
The Lord / of West (= Manwe) / broke / Earth / assenth-with / Êru-from

The subject noun corrsponds to Lowdham's description concerning d-Adûnaic's idea of a genitive. Bârun an-Adûn represents the unabbreviated form with Bârun in the Subjective case. This time, it is followed by a personal verb, urah(ha)ta, instead of the seemingly impersonal rakkhatû of ^L1. The poet put a really strong emphasis here! A change in vocabulary hits the word kamât that now becomes dâira and thus provides us with the formal distinction between k- and d-Adûnaic. Quite remarkably, the old Instrumental sôbêthumâ hardly changes and turns simply into sâibêth-mâ, lacking the glide vowel – one could of course say that this still was an Instrumental case, and whether Lowdham accepted it or not seems merely a matter of convenience.

5: azrîya du-phursâ Akhâsada. ... Anadûnê zîrân hikallaba [> hikalba]...
The seas / might gush / Chasm-into ... Westernesse / the beloved / she fell

Again, the subject noun azra turns into a Subjective case. Most interesting is the combination with the subjunctive du-phursâ, evidently from the same base as L1 nai phurrusim, but differently made. This is clearly an impersonal verb – and it raises the question whether or not subjunctives can at all be created using personal verbs. There also appears a new participle zîrân „beloved“, made the same way as zabathân from stem + suffix -ân. The akallabi of L1 is now personalised as hikal(la)ba, a Feminine formation determined by the Feminine interpretation of Anadûnê.

6: dulgu bawîb ... balîk hazad Nimruzîr azûlada ...
black / winds ... ships-of / seven-of / Elendil / eastward

This is a sgnificant expansion of II/1 in Version L1. The corresponding Old English version and its translation by Mr. Rashbold shows that between its two fragments, only the verbs „arose and drove away“ are missing, so they are in fact parts of one single phrase. The section dulgu bawîb illustrates Lowdham's claim that adjectives normally precede the noun and are not affected by sex. The second part shows an arrangement of genitives: in balîk hazad, the Normal Pl. balîk is apposed in genitival position to the Normal number hazad, thus „seven of ships“, while the whole complex is again genitival to Nimruzîr (Elendil). Genitival affixes are in this construction unnecessary. Nimruzîr illustrates the usage of the Objective case, for Nimru- is the Objective (singular!) of Nimir „an Elf“.The direction azûlada reappears unchanged from L1.


1: Adûnâim azûlada ... (struck out)

2: agannâlô buruda nênu ...
The death-shade (is) / heavy / us-on

agannûlô of L1 becomes the Subj. agannâlô, and we may correspondingly assume that the also attested basic noun nûlu (NC p. 306) becomes d-Adûnaic *nâlu. The adj. now is no longer suffixed by an auxiliary of „to be“, instead, the copula is, confirming Lowdham, expressed by the very use of the Subjective case in the noun; fittingly, the adj. here follows the noun. The double pronoun has lost the final -m, why? Does it in L1 express a modification analogous to the now absent -n of burudan?

3: adûn batân êluk izindi yadda: îdô katha batîna lôkhî ...
west / (a) road / once? / straight / went: / now / all / roads-are / crooked

This fragment was much changed from L1. êluk is a new word of doubtful meaning, perhaps an expansion of a stem *êl-, analogous to êphal-ak in II/4 and II/5. The old ezendi is forced into izindi by the absence of short -e- from d-Adûnaic. The impersonal new verb yadda (going with the Subjective batân) replaces akhaini: We saw it already popping up in I/1 and vanishing again. The particle katha now goes without numerical suffix while batânî of L1 assumes the Subj. case, again signifying an inclusion of the copula „to be“ that also is lost from the final adjective/participle (now lôkhî instead of rôkhî-nam).

4: êphalak îdôn Yôzâyan...
far away / now-is / the Land of Gift

Again, the absence of short -e- from d-Adûnaic requires the change of L1's êphalek. The particle îdôn is retained but now apparently identifies the Subjective case of îdô. The two names of Númenor exchange their places, with „Land of Gift“ assuming a new, and final, translation Yôzâyan, a compound including zâyan „land“. A better translation was in fact „Gift-Land“, for the two components are not in genitival apposition. Yô- could in fact be an Objective, but nothing is certain.

5: êphal êphalak îdôn Akallabêth
far / far away / now-is / She who fell

Aside of the changes already noted in II/4, there is nothing else to add.


Despite the upheaval of the grammar, the changes in vocabulary hardly transgress matters of orthography: new words are kamât > dâira, Athanâtê > Yôzâyan, akhaini > yadda (this being in fact two words of different meaning), zîrân, êluk, the female pronoun hi- (in hikallaba) and the phrase I/6, replacing II/1. The standard word order subject-predicate-object is retained, but I/4 accumulates a sequence of objects that follows the same order as it might do in an English phrase.

The adjective ezendi > izindi now precedes the verb, so that instead of "[a road] lay straight" II/3 reads more emphatically "straight went". English tolerates the same shifts. Most interestingly, it now seems to be established as a general rule that other than adjectives, participles directly follow the related noun, for zîrân finds itself in the same position to its subject as zabathân (though attached to a Normal noun, not a Subjective). Prepositions are in d-Adûnaic now generally suffixed: Ugru-dalad.

The nouns reflect the three cases of d-Adûnaic.




Adjectives ignore sex but apparently respect numbers and case. The normal form of a standard adjective seems to feature -i as last vowel. II/3 has izindi "straight", and RA adds anadûni "western" (but adûn "west" in II/3!). If the adjective is connected to a Subjective noun, it seems to acquire -a: buruda. The Normal plural indicator is , II/3 lôkhî, and this is perhaps accompanied by a Subjective plural *lôkhiya. One unusual formation is dulgu: Perhaps it indicates a different class of adjectives, or it may even be an Objective of *dulgi so that the compound should properly be interpreted as "winds of blackness".

The verbs represent two internal stages of development. In a first phase, all active verbs feature a doubled second consonant and, in their singular form, the last vowel -a: unakkha, dubbuda(m), azaggara, urahhata, hikallaba, yadda. In a second phase, half of them assume a shortened structure, lacking the second vowel: dubdam, urahta, hikalba. Can we propose that the alterations reflects a change of tense? It would make sense to assume that the doubled forms represent the „narrative or historic past“, that is the aorist. But what about the shortened forms? The trouble is that they never seem to change the tense of the translation: In fact, Lowdham in RA provides us both with usaphda „he understood“ and ukallaba „he fell“. Now it could be argued that ukallaba indicates a one-time-action and thus an aorist, while usaphda describes a lasting condition for which one of the two continuative tenses might be appropriate. But certainly, hikalba is not a suitable continuative! Can the shortened forms be past tenses, thus dubdam „they had fallen“, urahta „he had broken“, hikalba „she had fallen“, and finally usaphda, „he had understood“? Probably not, either. The most likely interpretation seems to be this: The doubled verbs actually represent the past tense, originally used throughout the Lament, and it was at this very point that the decision was made to use the present tense/aorist as „narrative or historic past“, thus changing several verbs into their short aorist forms.

There are also several special cases to be noted: azaggara has as a secondary form suggested azagrara which is a rather spectacular declension of the stem azgarâ-. Given the translation in L1, this could indeed represent a proper continuative – but as we will see, it was discarded in L3! And then we have du-phursâ which is another shortened form, but one with a fortified final -a that is probably determined by the usage of the auxiliary du- which seems to mark the (or one) d-Adûnaic subjunctive. Most remarkably, this subjunctive has no Plural ending, though the relevant noun is „seas“. We meet similar forms in Lowdham's short description of the Adûnaic tenses: azgarâ-, ugrudâ-. Very likely, phursâ- is as well a bare, undeclined stem of a Class III verb. There are other examples of compounds formed with bare stems: bêth in izindu-bêth, saibêth; zîr in Nimruzîr, etc. This could as well explain now the participles: zabathân, zîrân.

The verb-noun Akallabêth is hard to explain in terms of d-Adûnaic. It suggests the combination of a past tense *akallaba "it fell" with the female attribute -ith that Lowdham described in RA (mith "girl"). But what then is the meaning of the initial a-? It cannot identify a Neuter; neither is it a part of the stem as in azaggara. How would a Númenórean distinguish a proper Neuter from whatever this a- may mean?

Versions L3m and L3t

Both versions feature (deliberately incomplete) English glosses that substantially vary against each other. The differences in their Adûnaic are minor, though. Since as mentioned above, L3m seems to be earlier, it is given here with the changes to L3t bracketed.


1: Kadô Zigûrun zabathân unakkha...
And so / Sauron / humbled / he-came

No changes from L2.

2: Êruhînim dubdam ugru-dalad ...
The-Children-of-Êru / fell / ?shadow beneath [> ?shadow under]

I/2 copies down the final version given in L2. There is some uncertainty about whether dalad means „beneath“ or „under“ - probably both.

3: Ar-Pharazônun azaggara Avalôiyada ...
Ar-Pharazôn / was warring / against Powers

azaggara is now accepted as the final version. The translation into progressive tense „was warring“ seems not to be supported by the form of the original verb, though.

4: Bârim an-Adûn yurahtam dâira sâibêth-mâ Êruvô ...
Lords / of-West / they rent [> broke] / Earth / with assent [> assent-with] / from Êru [> ?-from]

Bârun an-Adûn of L2 is now set into Subjective Plural, and the Personal verb urahta follows in sex and number, providing us with Adûnaic's Masculine Plural pronoun: yu-.

5: azrîya du-phursâ akhâsada. ... Anadûnê zîrân hikalba [> hikallaba]...
that seas [> seas] / should gush [> so-as-to-gush] / into Chasm [> into chasm]. ... Anadune [> Númenor] / the beloved [> beloved] / she fell [> she-fell down]

The only trouble against L2 here is a lasting hesitation between hikalba and hikallaba. Why this would be so problematic is hard to tell. Is a clue maybe hidden in the translations: hikalba „she fell“, hikallaba „she fell down“? Unfortunately, Lowdham repeatedly translates ukallaba as „he fell“, so there probably cannot be put much weight onto the proper translation. Rather noteworthy is the translation of azrîya du-phursâ: either „that seas should gush“ or „seas so-as-to-gush“. The translation seems to pay special heed to the Subjective azrîya which here, in the second part of a phrase, seems to imply a „so that“. Rashbold's translation of OE suggests that hardly anything is missing between I/4 and this fragment: maybe nothing more than a mere conjunction.

6: Bawîba dulgî... balîk hazad anNimruzîr [> an-Nimruzîr] azûlada ...
winds (were) [> winds] / black ... ships / seven / of-Elendil / eastwards [> eastward]

I/6 has some baffling changes against L2. The Normal bawîb is changed into a Subjectiv bawîba with the curious effect that the adjective dulgu removes behind the noun and assumes a Plural ending . According to Lowdham's rules, this should be interpreted as „the winds were black“, and that is indeed the translation of L3m. Yet Rashbold translates from OE „black winds“. But there is probably no contradiction here. Lowdham only says that adjectives normally precede nouns: one exception could be whenever the adjective (or a participle) meets a Subjective. With the simple association balîk hazad Nimruzîr becoming balîk hazad an(-)Nimruzîr, the slightly awkward sequence of genitives belonging to other genitives is finally resolved, giving us both ways of formation at once.


1: (cancelled)

2: Agannâlô burôda nênud ... zâira nênud ...
death-shadow / very heavy [> heavy] / on us ... longing [> longing (is)] / on us

L2's buruda is fortified into burôda: Is this a change of the base or an indication that adjectives also may assume Subjective cases? L3m's translation „very heavy“ indeed seems to suggest the latter. (Aleš Bičan pointed out to me that an a-infixion used as intensifier is a feature that d-Adûnaic may have adopted in prehistorical times from Elvish. Quenya allows for similar possibilities.) Instead of nênu we now meet nênud, twice repeated in a slight expansion of the fragment that now includes a Subjective noun zâira (CT is not entirely clear about whether or not this expansion was already present in L2). It seems possible that this indicates a rectification of the vocabulary, *-ud thus drawing close to -ada „on, towards, into“.

3: adûn izindi batân tâidô ayadda: îdô katha [> kâtha] batîna lôkhî.
west / straight (right?) [> straight] / road / then (once?) [> once] / it went [> went] / now / all / ways (are) [> roads] / bent [> crooked]

izindi changes position and alters the meaning of II/3 from „a road went straight“ to „straight a road went“. The somewhat uncouth êluk is replaced by tâidô that evidently is closely related to îdô; but that does not necessarily mean tâidô had the same meaning as êluk. The verb yadda turns into a Personal Neuter ayadda „it went“, adding even more emphasis to the phrase. Finally, katha assumes an accent, but this probably bears not much significance: katha/kathu- is the form Lowdham uses in RA.

4: Êphalak îdô [>îdôn] Yôzâyan.
far-away / now [> now (is)] / Gift-land

Rather noteworthy is the hesitation between Normal îdô and Subjective îdôn while all previous versions of this phrase had îdôn. Since that is repeated in II/5, it was evidently not a mere slip.

5: Êphal-êphalak [> êphal êphalak] îdô [> îdôn] hiAkallabêth [> hi-Akallabêth].
far far-away / now [> now (is)] / She-that-hath-fallen

The fleeting stroke may or may not indicate a REALLY HEAVY emphasis by prefixing the same adverb again and again. And then, we meet an article hi-, identical to the Feminine pronoun of hikallaba. Well, if Akallabêth already was already a trouble in d-Adûnaic, hi-Akallabêth is a nightmare. What does a noun that has a Feminine suffix need as well a Feminine article for??? Well, there is a possible explanation. Someone wanted to place another intensifier here. But unfortunately, Akallabêth is the Object noun of îdô(n), so the Subjective *Akallabêthin is ruled out: It may, as you perhaps recall, only be used for Subject nouns. The insertion of an article may just mean the required solution to this problem.

The smaller samples

Belonging with L3t, there are three little d-Adûnaic phrases found in NC. Before we interpret the grammar of L3, it will be useful also to consider these:

NC1: Bâ kitabdahê!
Don't touch me!

This one is really remarkable: It is translated by an imperative, but remember that Adûnaic does not have such a mood at all. Evidently, NC1 is our only example of an optative and should more literally be translated by "You shall not touch me." It can be reduced to a verbal stem tabda-, a Singular subject pronoun ki- "you" and apparently an object pronoun - "I“ or „me". According to TT17, Westron constructs personal verbs the same way: meputeke „one blows me“ (!!! And then people think Elf-friends were puritanical...) from a stem put- to which Westron suffixes an -e like Adûnaic suffixes -a. Connected to it is an auxiliary that probably indicates a negation in Optative mood: "shall not". Latin also had such things, really desirable for a language that is meant to suit rhetors and manipulators.

The structure of the phrase, alas, is very ambiguous. Compare with our other sample of how to use an auxiliary: du-phursâ. While both verbs feature shortened forms, one has a lengthened final , the other simply a short -a. It may be the object pronoun that makes the difference here. But why is unlike du- not connected to the verb by a stroke? A mere slip of the transscriber or an actual difference in grammar? And can the order of the pronouns simply be reversed without further alterations? In short: would *hêtabdaki indeed mean „I touch you“?

NC2: Narîka 'nBâri 'nAdûn yanâkhim.
The Eagles of the Lords of the West are at hand.

In contrast to II/6, we meet here the opposite way to endlessly chain genitives: Narîka 'nBâri 'nAdûn or, as it could also have been written, *Narîka an-Bâri an-Adûn. In contrast to I/4 we meet here the Normal Plural Bâri, properly attached to a Subjective Plural narîka that evidently derives from a Normal Singular *narak „eagle“. And there also is an unforeseen verb: yanâkhim. Its base is evidently NAKH that also gave I/1 unakkha, yet it seems that nâkh- is distinct from nakh- which would simply mean „to come“, not „to be at hand“. The prefix ya- clearly is the Common equivalent of masculine yu- „they“ in I/4 yurahtam – may we conclude from the distribution of articles that the feminine form was *yi-? But there is one stupefying feature, and that is the Plural suffix -im whereas the Lament provides only -am. Does this signify a different word-class or a different tense? It seems conceivable by the somewhat free translation that yanâkhim is a Continuative Present rather than an Aorist, for the verb seems to describe a condition, not a moment in a process. It may be allowed to conclude that analogous to yurahtam there also was a verb *yanakham „they come/came“, and that the fortified vowel -â- indicates a different derivation from the same base with a subtly different meaning.

NC3: Urîd yakalubim!
The mountains lean over!

Here we encounter a verb of the same kind as in NC2: yakalubim, clearly related to the base KALAB that also gives hikalba, hikallaba and finally Akallabêth. But a stem kalub- could probably not produce a tense kalba: Here we have to conclude on a different verb-class with different flections. Like yanâkhim, yakalubim may tentatively be interpreted as a Continuative Present: The mountains (*urud > urîd, cf. Sindarin orod from which urud would be a typical Adûnaic loan according to RA) are in the state of tumbling or at the verge of falling, but the process is not accomplished yet, thus not justifying a simple aorist.

Some more specimen found in RA discuss the proper use of the Subjective but do not need to be discussed here in detail:

RA1a: Ar-Pharazônun Bâr "King Pharazôn [is] Lord"

RA1b: Ar-Pharazôn kathuphazgânun "King Pharazôn the Conqueror"

Iincluding katha „all“, here as a noun – would a more literal translation be „*taker of everything“?

RA1c: Ar-Pharazônun kathuphazgân "King Pharazôn [is/was] a conqueror"

RA2a: Bâr ukallaba "the Lord fell"

RA2b: Bârun (u)kallaba "it was the Lord who fell"

But cf. Zigûrun unakkha "Sauron came" in I/1, more properly "it was Sauron who came".


The known personal pronouns now add up to:







-hê? (object)














„We (us)“















Summing up the known verbs in impersonal Singular results in the following:


We would naturally expect that the most common form of the verb also represents the most common tense, the aorist used as „historic or narrative past“. This seems to be the simple short form with either one consonant (Class I) or two consonants joining (Class II and III). With regard to the doubled forms it seems decisive that tâidô ayadda means „once went“, so ayadda would describe a past, no longer current condition and thus be set in the actual Adûnaic past tense „used as pluperfect“. This interpretation also seems legitimate for unakkha, (u)kallaba, hikallaba, seeing the author hesitating between aorist hikalba and the proper pluperfect. And yet there is again a problem: the troublesome azaggara that of all represented verbs should be the one to represent a past continuative! Now the two verbs yanâkhim and yakalubim are clearly present tenses, but they both represent somewhat expanded derivatives of the base: It would seem unlikely that kalub- was identical to the short form kalba- in a different tense. Alas, we have too little material to decide whether the -i- in the -im suffix, contrasting with -a- in yurahtam, signifies the use of a continuative tense (but if this was the case, what then about azaggara?) or simply a different formation of the verbal stem.

The known verb forms, reduced to impersonal singular types, thus can be listed like that:


Class I

Class II

Class III


bêth-, zîr-, kan-, rûkh-, bêl-


azgarâ-, ugrudâ-; *zabathâ-, *abrazâ-?

(1) aorist

*nakha?, *nâkha?

dubda, rahta, kalba, *kaluba?, saphda, tabda,


(2) cont. present


*dubdi etc.?


(3) cont. past




(4) past

nakkha, yadda

dubbuda, rahhata, kallaba




*phursân, *dubdân etc.?

zabathân, abrazân

And then there was a-Adûnaic

At the end of RA, Christopher Tolkien expressed his belief that his father „abandoned the further development of Adunaic and never returned to it“. This may be true as far as written accounts of phonology and grammar are concerned; and yet it is evident that Adûnaic was further developed, introducing profound changes in grammar, vocabulary, and even phonology. This step brought d-Adûnaic to a-Adûnaic, and it can be dated with some certainty to the moment when the Line of Elros expanded from 12 or 13 to more than 20 kings. For the isolated page that CT describes in HA § 24,25 gives Adûnaic royal names that can still be explained by d-Adûnaic: Ar-Balkumagân (Tar-Ciryatan, „King Shipwright“) that displays the Objective case of balak, Arpharazôn. But this is no longer true for the Adûnaic names assumed by the later kings of Númenor.

Possibly a main cause for the profound change was the idea that one of the Númenórean kings would assume the throne-name „Lord of the West“ in blasphemous imitation of Manwë. For the name that we meet in the Line of Elros is neither k-Adûnaic Bârun-adûnô nor d-Adûnaic Bâr an-adûn but a new compound: Ar-Adûnakhôr. It may seem that *Ar-Bâr an-adûn looked too clumsy for a royal name, yet the idea of a „Lord of the West“ seemed too appealing to be given up. How then is the new Adûnaic compound to be explained? Is it Adûna-khôr or Adûn-akhôr? The former looks interesting for it suggests a word *khôr that is very similar to Elvish heru. But how then was the unusual Adûna- to be explained that clearly is not an Objective nor a typical Genitive, and would heru really be imported by the Númenóreans as khôr since Lowdham suggests that this should result in *hiru?

Another one of the royal names may serve to even further narrow down the date of the change: Ar-Abattârik aka. Tar-Ardamin, the „lost king“ whose name was unintentionally dropped from KR and never restored. Both names are intended to mean „Pillar of Earth“ and seem to be a purposeful pun on the „Pillar of Heaven“, the central mountain of Númenor. For Abattârik clearly imitates d-Adûnaic Minul-târik, given by Lowdham as the native name of said mountain, composed of the Objective of minal „heaven“ and of târik „pillar“ (but how? For since Minul-târik contains an Objective, so should Abattârik. Yet there is none. We only perceive a Normal *aban „Earth“ vs. d-Adûnaic daira, that is assimilated by the following -t- into abat-, and this would according to RA suggest that Tar-Ardamin's name meant simply „Pillar on Earth“, not supporting it, actually having an effect on it, as in the case of Minul-târik. Had a-Adûnaic given up the Objective altogether since we do not meet it in any compound in this final variation of the Númenórean tongue?). Correspondingly, Ardamin mimics Menelmin that once was the Quenya name of the same mountain. This must mean that the names of the Númenórean kings were already fixated before Menelmin assumed its final name Meneltarma, for Ardamin's name did not accordingly mutate into *Ardatarma. Unfortunately, this poses a problem in the chronology for Menelmin occurs only in an older version of the Akallabêth: In the earliest draft of what would become the final Akallabêth it is already Menelmindon followed by Meneltarma, but this already in a time when there were still only 13 kings conceived of and d-Adûnaic was still fully valid. Was Ar-Abattârik's name perhaps already conceived of long before it appeared in any written text?

A third royal name seems to suggest that this may indeed be the case: Ar-Gimilzôr that in d-Adûnaic was a translation of Elros but in a-Adûnaic refers to Tar-Telemnar and seems to translate his name, „Silver flame“. Yet Lowdham informs us that d-Adûnaic gimil means „star“ and thus closely corresponds to Elros that was then translated „Star-foam“. Has gimil assumed a different meaning in a-Adûnaic? Not always, for another name recorded in the Akallabêth is Gimilzagar that d-Adûnaic would simply translate as „star-sword“ and thus make it correspond exactly to the Elvish name Elemmacil. Unless Gimilzôr is not intended at all to translate Telemnar, this may well be a case of an Adûnaic homonym: gimil „silver“, maybe in origin a distorted loan of Khuzdul khibil.

Further changes in the vocabulary include the usage of azar/azra as "star" rather than „sea“ and the change of the patronymic suffix -(ô)hîn (< Azrabêlôhîn = Eärendilion, „Son of Eärendil“) into -thôr (Ar-Sakalthôr = Tar-Falassion). Also, it is unlikely that a-Adûnaic retained the name Êru which was by the Notes for Revision dated 1951 transferred to Elvish (see SD and subsequent volumes). It might have become directly adopted by the Hadorians, as were the names of most of the Valar, but if the import rules of d-Adûnaic still applied, Elvish Éru would have emerged as *Îru.

And as a final note it may be observed that RA was definitely not Tolkien's final word on Adûnaic. Aside of the discussion of the abandoned root ROTH in PR, the Silmarillion Index states that Adûnaic adûn was directly borrowed from Sindarin adún. Since RA and related texts contain nothing at all to this extent, this statement proves that there must be somewhere some unpublished further material on the Adûnaic language.

List of Abbreviations