Agental suffixes and gender distinction in the Etymologies

Abbreviations used:

LR: The Lost Road and other writings
WJ: The War of the Jewels
VT: Vinyar Tengwar
PE: Parma Eldalamberon

Q: Quenya
N: Noldorin
S: Sindarin
EN: Exilic Noldorin
ON: Old Noldorin
CE: Common Eldarin
PQ: Primitive Quendian
Dor.: Doriathrin
Ilk.: Ilkorin


Agental suffixes (i.e. endings used to describe a person performing the action specified by a verb) are an important part of noun formation in languages. There are numerous instances where such endings appear in Tolkien's Elvish languages. The most prominent source are the Etymologies, since in this work one cannot only study the historical development of such endings but also the manifestation in the various branches of the Elvish languages.

A closer look reveals a rich structure of different primitive endings which in the various final languages are often blended and mixed. The Elvish languages do not have a grammatical gender, i.e. there is for example no gender agreement between noun and adjective or definite article and noun. However, Elvish agental suffixes do usually make a gender distinction (unlike in English but like in e.g. German). Thus, it is fruitful to expand the scope of an investigation to include not only endings characterizing persons doing some action but in general incarnates associated with a root meaning, i.e. to also endings describing how someone is and in general all gender-specific forms.

The majority of these endings can be traced back to Common Eldarin forms. While this implies in principle that these endings should be found in any of the Elvish languages, the peculiarities of language development, e.g. the loss of final vowels in Noldorin, often lead to certain forms being preferred over others. However, there is also evidence for 'modern' development of a class of endings evolving from former compounds which may not be realized in all languages. Thus, a closer look at even the seemingly simple matter of describing a person performing an action reveals again the scope of Tolkien's invention and his attention to details.

Primitive -ō, -ē, -ū and -ā

One of the most commonly seen schemes involes and appended directly to the primitive stem to mark masculine and feminine gender (or a male or female agent). The clearest indication for the assignment of endings comes from the root SEL-D- where we find derived forms Q: seldo 'child (m)' selde 'child (f)' selda 'child (n)' (VT46:13). Thus, we may also speculate that can be used to avoid making an explicit gender distinction. Primitive forms conforming to this scheme are attested for the root TÁWAR- with CE: tawarō Q: tavaro 'spirit of woods (m) and CE: tawarē, however the Quenya reflex of the latter is Q: tavaril 'dryad (f)' rather than ?taware (LR:391), involving a different ending to be discussed later. Support for the gender-neutral ending may come from the root BAL- cf. CE: bálā ON: Bala '*having power' where the forms with open gender distinction involve other endings, cf. Balano (m), Balane (f) (LR:350).

It is possible that in the same scheme marks a pair (e.g. a male and a female), this coule be argued from CE: besū Q: veru 'husband and wife' (LR:352, VT45:7,9), however the actual Quenya words for 'husband' and 'wife' are venno, vesse (ibid) and involve more complicated endings -nō, -sē. This (among other evidence) would again indicate that the attempt to identify different schemes of marking gender is rather futile when applied to Quenya or Noldorin as forms originating from different schemes would regularly merge at this later stage.

In Noldorin and Ilkorin an original gender distinction with these suffixes is lost along with the final vowel. Hence, if a distinction is desired, the female form must derive from a different ending which is not lost. Examples for the loss of distinction include CE: ndākō 'warrior, soldier' ON: ndōko N: daug (LR:375) or CE: berō Ilk: ber 'valiant man'.

There are several cases in which the above scheme does not apply, see e.g. CE: bánā Q: Vana T: Bana ON: Bana '*beautiful one' (LR:351) or CE:barádā Q: Varda T: Barada '*lofty one' where is used for the female, or Q: fire 'mortal man' (LR:381) where is used either general or male. Finally, Q: #tánie 'maker (f)' (LR:390, VT46:17), the counterpart to CE: tanō Q: tano 'craftsman, smith' (ibid) comes out somewhat unexpected, ?tane would be the expected form. However, an alternative gloss of #tánie is 'making', hence it could also be seen as a gerund/infinitive form involving the known ending -ie also seen in enyalien 'for recalling' (UT:305, 317).

Finally, in one case forms appear which cannot be explained by suffixes appended to the primitive root. From the root ANA- we find CE: anta- 'to present, to give' Q: anta- 'to give' and anto (m) ante (f) 'giver' (LR:348). It would hence appear from the context and the lack of a primitive form that these nouns are analoguous formation based on the verb rather than the result of primitive ?antaō, ?antaē or suffixes ?-tō, ?-tē.

Relevant forms in the Etymologies include:

and possibly also

Primitive -ū and -ī

There is a second system of Common Eldarin endings in which *-ū marks a male and a female, though this seems to be used rarely. In no instance are these affixes used with an agental meaning, they rather seem to denote exclusively gender distinction. The primitive *-ū is not attested directly but can be with reasonable certainty inferred from reflexes in Quenya and Doriathrin. For the female version, we have CE: Barathī from BARATH- '*lofty one' (LR:351). The clearest evidence for the ending *-ū as male counterpart to is seen from Q: Ainu (m) Aini (f) 'holy one' (LR:350) and Q: heru 'master' heri 'lady'. This cannot just be an ending peculiar to Quenya as we find multiple reflexes in other languages for the root 3AN-, cf. Q: hanu ON: anu N: anw Dor.: ganu 'a male, man' (LR:360, VT45:16).

Since final vowels are regularly dropped in Noldorin, reflexes of the forms are much obscured. In one case the ON form acquired an -l, presumably in analogy with the common ending -ril (see below), hence we find CE: Barathī > ON: Barathi(l) > EN: Berethil '*lofty one' (LR:351). In a different instance, the female form was actually coined after the male (and hence also ends with -w, cf. N: inw 'female' (after anw 'male') from the root INI- (LR:361). In a third case, the likely primitive ending is changed to -o in ON and lost in the transition to Noldorin, cf. CE: *kherū > ON: khēro > N: hîr 'master'. While the Quenya female form heri 'lady' shows the primitive ending, the ON form khíril and the Noldorin form hiril 'lady' again show a different ending which preserves the distinction in spite of the loss of final vowels.

The list of relevant forms found in the Etymologies includes:

Primitive -rō, -rē and -rū

The next step in complexity following simple vowel suffixes are primitive endings involving an additional consonant. The most commonly seen variant is male -rō, female -rē and possible dual -rū. The fist two are directly attested in their primitive form, cf. CE: weirē Q: Vaire 'weaver, Vaire' (LR:398) and CE: kalrō Q: kallo 'noble man, hero' (LR:362). The last is deduced from Q: ontaru 'parents' (LR:379, VT46:7). Neither ending must necessarily produce Quenya reflexes with the final vowel, cf. CE: ábārō 'refuser' leading to both Quenya variants Q: Avar and Avaro, hence a shortening of the ending is possible. One may assume that such forms are then, like Noldorin reflexes with a loss of final vowel, in general gender-neutral.

This latter form indicates an important difference to the previously discussed affixes: The longer forms -rō, -rē and -rū can apparently not only be appended to stems but also to derived verbs formed on the CE stage. Unfortunately there is little in terms of direct evidence for this, however the appearance of a -t- typical for the derived verb ending -tā inside a word in Quenya and Noldorin reflexes is quite characteristic. Consider for example the root MAK- from which we find Q: mahta- 'wield a weapon, fight' hence mahtar 'warrior' N: maethor (LR:371). Clearly, directly appending the ending to the stem would yield *mak(a)rō and lead to Q: makar N: magor - these forms are attested in the compounds Q: Menelmakar and N: Menelvagor (LOTR, WJ:324). Instead, the presence of the -t- indicates a derivation from *maktārō which would result in the given Noldorin and Quenya forms.

However, derivation does not seem to be confined to verbs, we find e.g. from the root PHAL- Q: falma 'wave' and falmar 'sea-spirit' (LR:381), which seems to be based on adding the ending to a noun.

When a form is derived from the root, the loss of the final vowel leads in Noldorin usually to a consonant cluster which is then vocalized. This is best apparent from the root TAM- where CE: tamrō 'knocker' leads to N: tafr, tavor 'woodpecker' (LR:390, VT46:17), indicating that the unvocalized version tafr is subsequently vocalized as tavor. In fact, Noldorin words ending in -or can usually be explained either as a vocalized ending -rō appended to a consonant or as a remnant of the same ending appended to a verb -Cārō where the ā is subsequently opened to au which ends up as o when unstressed. One may theorize from this that the plural form could then be different for these two groups - if no analogical levelling takes place, the expected plural of maethor would then be *maethuir whereas the plural of tafr, tavor would perhaps be *teifr, *teivor. There is one example which may support the first scenario - if N: Afor is derived from a verb ábā- (rather than from the longer root ABAR-) then the plural Efuir would be just in line with what has been indicated above.

Perhaps not entirely unexpected, there are also several forms which do not conform well to the scheme. CE: bálāre N: Balar (LR:350) shows an unexpected ending , the form Q: onóro 'brother' is paired with onóne 'sister' (LR:378) instead of the expected ?onóre and likewise CE: tāro Q: tár 'king' is paired with CE: tārī Q: tári 'queen' instead of ?tārē, the Ilkorin reflex again shows the use of a distinct ending Ilk. tóril 'queen' (LR:389).

A list of forms found in the Etymologies where the endings are apparently appended to CE verbs or nouns includes:

Forms where the endings are appended to the bare stem include:

Primitive -dō, -dē and -mō, -mē

While seen less frequent than the previous class, the endings -dō (male) and *-dē (female) are still an important part of word formation. They are most productive in Quenya, although Noldorin reflexes exist, especially for roots ending in -n, leading to a characteristic Noldorin ending -Vnd, -Vnn, -Vn. Only -dō is directly attested in CE: tuilelindō 'spring-singer' (LR:395), although the female counterpart can be inferred from examples like Q: nildo (m) nilde (f) 'friend' (LR:378).

Especially for roots ending in -R- or -L, a variant form -mo is seen in Quenya reflexes, such as Q: heldo, helmo (m) helde (f) 'friend' (VT46:3). Curiously enough, no female form ?helme is indicated, presumably because such forms would regularly clash with a distinct abstract suffix -me. Nevertheless, -me as a gender-marking suffix exists as indicated by Q: sermo (m) serme (f) 'friend' (LR:385).

All the suffixes of this class are only seen appended to the root, never to derived forms like verbs or nouns. The list of relevant forms includes:

Primitive -ndō and -llē

The endings -ndō (male) and -llē (female) can be considered the next step in complexity, as they are usually not applied alone but rather appended to other endings discussed previously, in particular and -rō (male) and and -rī 'female'. Thus, these suffixes are rather productive in Noldorin since they screen other suffixes and hence preserve them to some degree. This is especially apparent from contrasting Quenya and Noldorin in CE: kalrō Q: kallo 'noble man, hero' but the extended CE: kalrondō N: kallon 'hero' (LR:362). Moreover, the rather obvious distinction between -ndō and -llē is the regular way to mark gender in Noldorin, cf. CE: *kherū, *kherī N: hîr 'master, *lady' but CE: *kherondō N: heron 'lord, master' and CE: kherillē N: hiril 'lady' (LR:364, VT45:22). We must assume that hîr does not express gender as Elbereth is addressed as A Hîr Annûn gilthoniel in Lúthien's song (The Lays of Beleriand p. 354).

While not as frequent as in Noldorin, at least the ending -ndō is productive in Quenya as well, cf. CE: ulgundō Q: ulundo 'monster' (LR:396), although more commonly the shortened variant -on is found, e.g. CE: sayrondō Q: sairon 'wizard' (LR:385).

Primitive -indō and -issē

Possibly a longer variant form of the endings discussed above, *-indō (m) and *-issē (f) seem to be applicable directly to stems without another suffix coming before. This is apparent from Q: melindo (m) melisse (f) 'lover'. The evidence for these forms in the Etymologies is rather thin, but they appear to productive in at least Quenya and Noldorin, cf. CE: *rigissē N: rhîs 'queen' (LR:383). An abbreviated form seems to be responsible for Q: falmarin pl. falmarindi 'sea-spirit' (LR:381)

The relevant forms are:

Compound endings

There are several examples of words which are transparently not composed from CE root and suffix but can rather be seen as compounds of independent words (which have become so frequently used that the second part may be seen as an affix). For example Q: kentano 'potter' clearly involves tano 'maker' (LR:363, VT45:19) and also occurs as shortened suffix -tan, cf. Q: Martan 'Aule' (LR:358). Other (especially in Noldorin) very productive suffix elements based on compounds are -dir (male) and -dis 'female', cf. N: meldir (m) meldis (f) 'friend' (LR:372, VT45:34). A sample of forms occuring in the Etymologies is:

Agental and gender marking suffixes in post-Etymologies sources

The general structure of gender marking seems to be similar when taking later sources into account. Some forms, e.g. Q: #colindo 'bearer' (LOTR) conform perfectly with the ending -indo discussed above. In other places, the above interpretation is directly confirmed. Tolkien notes in WJ:400 that the ending -mo often appeared in names or titles, sometimes with an agental significance: Ulmo was interpreted as 'the Pourer' < *UL 'pour out'.

In Sindarin, a new compound ending appears in e.g. S: orodben 'mountaineer, *mountain-person' and rochben 'rider, *horse-person' (WJ:376). However, the possibly most intriguing development is the appearance of a new female suffix -eth.

This is seen in several places, cf. ellon 'elf-man' elleth 'elf-maid' (f) (WJ:363); firion 'mortal man' firieth 'mortal woman' (WJ:387, these forms show a common repetition of -i- as the stem vowel), adaneth 'mortal woman' (MR:349) and besoneth 'bread-giver' (PM:404). In spite of two similar forms in the Etymologies (bereth 'spouse' (LR:351) and dineth 'bride' (LR:377)) such an ending does not occur in the Etymologies - the former is from the root BARATH and the latter involves the adjective neth 'young'.

Appendix: Gender pairing

The following pairs of endings can be made out to distinguish male and female in the Etymologies:
Common Eldarin Quenya Noldorin Telerin Doriathrin Ilkorin
*-ō/-ē -o/-e        
*-ū/-ī -u/-i -w/-   -u/*-i  
-rō/-rē -ro/-re -r/-r -ro/*-re    
*-dō/-dē -do/-de        
*-mō/*-mē -mo/-me        
*-Vndō/*-Vllē -Vndo, -Vn/*-Vlle, *-Vl -Vnd, -Vnn, -Vn/-Vl -Vndo/?   -Vnd, -Vn/-Vl
*-indō/*-issē -indo, -in/*-isse *-ind/-is      


Helge Fauskanger's Quenya Affixes has been useful for cross-checking some of the information.


[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
Thorsten Renk

Back to the index