Agental suffixes and gender distinction in the Qenya Lexicon

Abbreviations used:

LR: The Lost Road and other writings
QL: The Qenya Lexicon
VT: Vinyar Tengwar
PE: Parma Eldalamberon

Q: Qenya
N: Noldorin


Agental and gender-distinguishing suffixes in Tolkien's invented Elvish languages have been studied in some detail for the 'Etymologies' [1-3] in the article Agental suffixes and gender distinction in the Etymologies [4]. While the Etymologies, providing in many cases the primitive form and the derived forms in different Elvish languages are uniquely suited for such a study, the concept of such derivational suffixes can be traced back much further. The Qenya Lexicon [5], providing one of the earliest accounts of the Finnish-inspired branch of the Elvish languages, contains a number of forms which allow insights into the structure of the scheme how such suffixes were employed in Qenya.

For all we know, Qenya did not have at any time a distinction of grammatical gender of nouns (implying e.g. a gender-specific form of the definite article as in German or of adjectives as in Latin), but there are a number of words for persons which appear in pairs, one ending denoting a male person, the other a female (in some cases, a gender-neutral form can also be found). A good example would be etta 'cousin' ettanu 'male cousin' ettani 'female cousin' from the root ETE 'cling together' (PE12:36).

In addition to such gender-pariring endings, there are also a number of words for persons for which the ending seems to have just a gender-neutral agental function, most clearly seen if the verb is attested as well, cf. lala- 'to chatter, to babble' lalar (lalard-) 'babbler' (PE12:50). In some cases, there appear to be blendings between endings in the agental and the gender-distincting function, for example -mo is an ending frequently seen with verbs in an agental function without an open gender distinction being apparent, cf. pili- 'to steal' pilmo 'thief, robber' (PE12:73,74), however in one case this ending is paired with another to show gender distinction in the pair Valmo 'male Vala' Valde 'female Vala' (PE12:99). Thus, it is worthwhile keeping in mind that there may not always be a clear-cut distinction between the function of different endings, and it seems conceiveable that -mo does not in all cases imply a male agent.

Also, when looking into the details, the concept of paired endings is somewhat misleading, as there are examples such as veru 'husband' †veruni, †veri, †vesse, vestin 'wife' from VERE (PE12:101) where a single male ending is paired with a number of female ones. Thus, the underlying linguistic scenario is probably characterized by lost forms and analogical replacements, attesting to the complexity of Tolkien's linguistic invention even in the early material of the Qenya Lexicon.

(In the following, due to technical reasons, the diacritics of the roots cited are often suppressed - the interested reader is asked to consult the QL for the original writing).

-u and -i

The most frequent endings seen distinguishing gender are -u and -i, in some cases appended to the stem directly, in other cases appended to some longer derived form. Clear-cut examples are Two other pairs show endings which can be interpreted as variants of the simple endings seem above, in the latter case the forms -u and -i become only visible in the stem: Presumably is of a similar origin, but involves in addition the root MAKA 'slay' (PE12:57) which explains the longer ending. Maybe also is of the same type, but maybe this pair is an example of a genuine male ending -o which does not disappear in the inflected stem. Note here that both first and second element change when going from male to female variant. In other examples, only either -u or -i is found, not always (but often) with a clear gender. Male examples are whereas female forms include

-nu and -ni

Perhaps in some way analogically formed from the previous examples, cf.anu 'a male' anī 'woman' (PE12:31) there is one pair in which the longer forms -nu and -ni are observed: In addition there is also the unpaired form (which, before any reader gets overly excited, clearly refers to a female dog).

-u and -uni

In what is presumably a further analogical development, in few instances the female form paired with the ending -u is not formed by the ending -i but by appending -ni to the male ending. We find and still less clear

-mo and -re/-de

There are two instances in which the male ending -mo is found paired with a female ending, in one case this is -re, in another case this appears to be -de. This ending appears also in hilde 'child' (PE12:40) which one is tempted to interpret as 'daughter', however the translation clearly does not show this meaning. However, in the majority of cases, -mo appears to be unpaired and shows a clear agental meaning (in some cases, -pm- > -mp- is assumed silently):

-ro/-ndo and -sse

Another pair observed is -ro/-ndo and -sse. Interestingly enough, the two male forms seem to be variants, they are always listed as alternatives. Furthermore, -ro leads in all examples to a lengthening of the preceding vowel, -ndo does so sometimes as well. As an example of the female form paired with a different male ending, one finds

-illo and -ille

In one instance, the pair -illo and -ille can be found in the lexicon, -ille also occurs in a separate, unpaired form:

-on and -is

One example of a pair showing the endings -on and -is can be found However, the ending -is occurs much more frequently unpaired (or paired with a different ending). As melitse 'darling' (PE12:60) could indicate, -is (-its-) seems to be a shortening of an original ending -itse.

-ar -and -ni

Two cases in which a male ending -Xar (where X is an additional consonant) is paired with a female -ni can be found: A few more examples show the male version of the ending in an agental use:

-ion and -wen

Chiefly in denoting family relations (son of..., daughter of...), the two endings -ion and -wen are used. They are as endings in a sense apart from others, as they are abbreviations of the independent words for 'son' and 'daughter', not merely derivative endings. A few cases where both endings are paired can be found: In addition, a few unpaired female forms are attested:

-ner and -qin

Quite similar to the forms listed above, there is also a pair of endings derived from the words for 'man' and 'woman' and a paired and an unpaired example of their usage:

Female -inqi and -in

Two endings are found which denote clearly female entities, but are not obviously paired with a distinct male ending:

Agental endings: -r, -rdo

A number of endings -r, often in direct connection with a verb, indicate the agent performing the action of the verb. Although there appears to be a tendency for some of these verbs to denote male agents, the evidence that this ending is actually meant to be male is inconclusive. From moar (moard-), moardo 'shepherd' (PE12:60) on can infer that the ending -r is presumably a shortened version of an original -rdo, agreeing well with the stem -rd- which is indicated for a number of forms. There is a very similar group of words for which an ending -r appears as well, however no expanded stem is indicated. One is left to wonder of this group of words actually uses the same ending (and Tolkien simply did not indicate the stem variation) or if this is a genuinely distinct group.

Special developments

Finally, there are a number of words in which clearly either a gender-distinction or an agental meaning is implied, but which cannot be readily be classified in any of the above groups.

Comparison with the Etymologies

While providing roots, the Qenya Lexicon nevertheless contains much less information about primitive forms in the Etymologies and likewise only for a few forms corresponding Gnomish words. This naturally limits the scope on which we can compare the system with agental and gender marking in the Etymologies.

Qualitatively, the system found in the QL does not seem to be substantially less in scope: There is a number of different endings and pairs of endings. We have evidence that these endings cannot only be appended to bare roots but also to longer words, cf. AYA: aimaktu (aimaktar-) 'martyr (m)' aimaksin (aimaksi-) 'martyr (f)' (PE12:34) where the root MAKA appears to be present as well. There is some reason to believe that the attested forms represent a development in invented history in which analogical replacements took place, cf. VERE: veru 'husband' †veruni, †veri, †vesse, vestin 'wife' (PE12:101) where a number of forms are marked as obsolete, among them the 'natural' partner of veru, i.e. †veri. As in the Etymologies, there is also a distinction between mere endings (presumably used in primitive Elvish and directly appended to the root) and endings which are shortened versions of full words (presumably developed somewhat later), cf. Noldorin -dir and -dis from the Etymologies with -ner and -qin from the QL. Even complicated developments like a double gender marking is found, as in heruvesto 'husband' herivesti 'wife' (PE12:40).

However, when looking at the actual distribution of forms, the two systems of endings are clearly not the same. The Qenya of the Etyomologies shows extremely productive primitive suffixes and which seem to be completely absent from the scenario of the Qenya Lexicon. Likewise, the very productive group -rō and -rē found in the Etymologies does not play any role in the QL. In contrast, while -u/-i, the most frequently encountered pair of endings does exist in the scenario of the Etymologies as well, merely four instances of the ending are found.

Of the less productive suffixes, -mo/-de can be found in the Etymologies in e.g. Q: heldo, helmo (m) helde (f) 'friend' (VT46:3), -ndo/-sse from the QL correspond to -indo/isse attested in Q: melindo (m), melisse (f) 'lover' (LR:372, VT45:34), the ending -lle can be glimpsed in N: hiril 'lady' (LR:364, VT45:22), although the corresponding -llo from the QL is not found in the Etymologies.

While many of the forms attested with a final -r in the Etymologies and in the QL look similar, cf. Q: ehtar, ehtyar 'spearman' (LR:355) and ektar (ektard-) 'swordsman' (PE12:35), it has to be remarked that the development leading to the forms appears to be rather different. While the form from the QL presumably originates from a longer ending -rdo, no such ending can be found in the Etymologies, and most forms ending in -r are rather derived from primitive -rō/-rē, and the stem still indicates the difference.

Thus, it has to be concluded that the system of the Etymologies in essence represents a new scheme. The overlap in forms between the QL and the Etymologies is hence in all likelihood driven by the words and names which were fixed in Tolkien's concept and needed to be carried over, but the vast majority of the material was substantially reworked.


[1] The 'Etymologies' in 'The Lost Roand and Other Writings', edited by Christopher Tolkien
[2] Addenda and Corrigenda to the 'Etymologies' part I , Vinyar Tengwar 45, edited by Patrick H. Wynne and Carl F. Hostetter
[3] Addenda and Corrigenda to the 'Etymologies' part II , Vinyar Tengwar 46, edited by Patrick H. Wynne and Carl F. Hostetter
[4] Agental suffixes and gender distinction in the Etymologies by Thorsten Renk
[5] The 'Qenya Lexicon', Parma Eldalamberon 12, edited by Christopher Gilson, Carl F. Hostetter, Patrick H. Wynne and Arden R. Smith

Thorsten Renk

Back to the index