Tolkien generally referred to nouns ending in a consonants as “consonantal” nouns, for example on PE14/42 (1920s) and PE21/76 (1950s). As with all Quenya words, a noun can only end in one of the five valid final consonants: l, n, r, s, t. Consonantal nouns were often a reduction of longer ancient forms, however, and their inflections could reflect these more ancient endings. For example, the singular (uninflected) word for “nymph” is wingil, but its plural form is wingildi, reflecting a longer ancient form. These variant bases for inflected forms are referred to as the noun stem, and is given in parenthesis after the word in dictionary entries when it differs from the uninflected form: wingil (wingild-).
Many inflectional suffixes begin with a consonant, and therefore cannot (usually) be added directly to a consonantal noun. There are various ways that Quenya deals with this.
First, many inflections exist with vocalic and consonantal variants, the vocalic variant beginning with (or being) a consonant and the consonantal variant beginning with (or being) a vowel. For example, the Quenya plural suffix is -r after (most) vocalic nouns but -i after consonantal nouns. The dual suffix is -t after (most) vocalic nouns but -u (probably) after consonantal nouns.
Second, a suffix beginning with a consonant may be separated from a noun’s final consonant using a “joining vowel”. This strategy is most obvious with the adverbial cases (allative, ablative, locative) where the joining vowel is typically e in Tolkien’s later writing. Thus for the noun Ambar “World”, the vowel e is inserted between the ablative suffix -llo and the noun to give Ambarello = Ambar + e + -llo (Merin Sentence, MS).
Finally, a suffix beginning with a consonant may assimilate to the final consonant of a noun. This works best when (a) the noun ends in a single consonant and (b) that consonant matches the first consonant of the suffix. For example, with menel “heaven(s)”, using the joining vowel e with ablative -llo would produce the rather awkward *menelello, and this would naturally reduce to menello which is the form actually appearing in Tolkien’s writing (VT43/13).
Tolkien employed all three strategies in various contexts. For example, the partitive plural uses assimilation in many examples, such as:
- Kasalli “some Dwarves” = Kasar + -li (WJ/388) with rl > ll.
- elelli “some stars” = elen + -li (PE17/127) with nl > ll.
Unfortunately, Tolkien did not write about consonantal nouns often, and many of his declension charts are limited to vocalic nouns. This includes the famous Plotz letter, our best examples of Late Quenya declensions, which only gives declensions for nouns ending in vowels. Thus there are a fair number of cases where we don’t know how Tolkien imagined a suffix would be added to a consonantal noun. These are discussed in more detail in the entries on individual noun cases. A more general discussion of joining vowels also appears in the section on Conceptual Development below.
Origins of consonantal nouns: Some nouns were already consonantal in ancient Common Eldarin (CE), and these nouns frequently remained consonantal in Quenya, excluding rare cases such as final ʒ, ñ [IPA ɣ, ŋ], both of which vanished very early in the phonetic history of Elvish. Tolkien discussed these ancient consonantal nouns in Common Eldarin: Noun Structure written in the 1950s (PE21/70-71). Examples include ancient ✶ndē̆r “man” (Q. nér) and ✶menel “heaven(s)” (Q. menel). Of these, the ancient monosyllables had some specialized developments, as discussed on the entry on monosyllabic nouns.
However, there was a general trend in Elvish whereby short final [e], [a], [o] vanished. Thus many Quenya consonantal originally ended with short ĕ, ă or ŏ, later lost:
It seems nonetheless clear that this original class received additions early in the history of the separate Eldarin dialects by the coalescence with it of certain old paroxytone nouns ending ĕ, ă, ŏ which were in some branches soon lost phonetically, via an obscured [ǝ], especially after sonorous sonants such as n, m, l, r (PE21/71).
As noted in the entry on vocalic nouns, not every CE noun lost its short final vowel, but where they were lost it seems the loss was very early, leaving no trace even in inflected forms. Thus it is impossible in many cases to determine exactly what the lost final vowel was, through presumably in most cases it matched the base vowel (sundóma). However, the reduced vowel [ǝ] may have had some influence on the development of joining vowels before it ultimately vanished; see below for discussion.
Final vowels were more likely to be lost in longer words, especially compounds. This is how nouns whose stems ended in a consonant cluster most likely to originated:
[ǝ] in final syllables was lost – see Final Syllables – and at the end of long words, especially compounds, ă, ĕ, ŏ, reductions of ā, ē, ō (as described above) were also lost. Thus kwènedḗ > kwendḗ but móri-kwènedḗ > móri-kwènǝdĕ > móri-kwèndĕ > móri-kwèn(d), Q. Moriquen (PE19/59).
In this example, the stem of singular Moriquen is Moriquend-, as seen in its plural form Moriquendi (S/51, WJ/361).
Conceptual Development: Tolkien had a distinct class of consonantal nouns all the way back in the Early Qenya Grammar (EQG) of the 1920s (PE14/42, 71). In his lengthy Declension of Nouns from the early 1930s, he gave numerous examples of declensions for nouns ending in a variety of consonants. Frequently the case suffixes he used in this document are not compatible with Late Quenya, but these examples provided clues to how Tolkien might have declined consonantal nouns in a variety of situations. He employed all three of the strategies mentioned above for combining suffixes with consonantal nouns:
1) There were several noun cases with distinct forms for vocalic and consonant nouns, though not necessary the same variations seen in Tolkien’s later writing. For example, the vocalic/consonantal plural suffixes were -li/-i (as opposed to later -r/-i), the vocalic/consonantal duals were -t/-u, and the vocalic/consonantal dative suffixes were -r/-en (as opposed to later -n/-en). For vocalic nouns, the nominative singular was marked with a suffixal -n but the accusative was unmarked, whereas for consonantal nouns the nominative singular was (mostly) unmarked and the accusative had the suffix -a.
2) Longer suffixes were frequently assimilated to shorter nouns, especially monosyllables ending in a single consonant. The number of possible combinations and the variety of assimilations was staggering, too much to cover here. The more interesting combinations are discussed in the entries on individual noun cases.
3) Otherwise, longer suffixes employed some kind of joining vowel, such as -anen, -unta, -ullo, -esse for instrumental, allative, ablative and locative (the allative suffix was -nta at this conceptual stage). The variants with a joining vowel often coexisted with the variants showing assimilation, though typically one or the other was marked as archaic.
In these and other early declension charts, the joining vowel depended on the character of the suffix, not the character of the noun to which the suffix was added. In various charts they were:
Here EQG is Early Qenya Grammar (PE14/47, 78) and v3, v4 and v5c use the declension labeling scheme from PE16 and PE21 (PE16/113; PE21/16-37; PE21/52). The “Adv.” column is the “Adverbial” case Tolkien abandoned in the 1950s; in Eldamo entries I use the term similative for this noun case to avoid confusion with other grammatical functions. The EQG “instrumental” marked with a * is actually a partitive, but its form matches later instrumental inflections.
Given that the joining vowel varied by noun case, it is likely the they were produced by some phonetic effect. In EQG, the allative, ablative, locative suffixes are -anta, -ullo, -isse, it is not a coincidence that in this conceptual period syllabic ṇ, ḷ, ṣ often produced an, ul, is (PE12/3, 10). In the Qenya Phonology from the 1910s Tolkien said: “Connected with the evaporation of ə (ultimately < ŏ, ă, ĕ), ṇ, ḷ, ṛ can arise which were originally purely semi vocalic or consonantal” (PE12/11). In the typescript version of EQG, Tolkien made it clear that this phenomenon produced the joining vowels for many noun cases:
In the “consonant” declension the old lost ə (from -ĕ, -ă, -ŏ) should reappear with varying quality. This normally produces -ISSE, -ULLO, -ANTA (or -ATTA) … (PE14/78).
Phonetic effects do not explain the EQG suffixes -inen and -indon, but Tolkien indicated that these arose by generalization from plural forms:
… but the -I- is often extended (aided by the -I- that appears where these forms are added to the plural stems), especially to -INDON (PE14/78).
It seems likely that joining vowels in later declension charts arose from one of these two mechanisms: a phonetic effect produced by consonant syllabification from lost ə, or by generalization from some common form such as plurals. In Tolkien’s later writing from the 40s, 50s and 60s, the most typical joining vowel was e. As this is unlikely to be the result of syllabification, it probably was generalized from some common form. There are hints of this mechanism for the dative/genitive suffix from Primitive Quendian Structure: Final Consonants written in 1937:
In origin this was probably e-n being the stem (old base-extension) of such nouns as dēr, dere-n and of reduced ĕ-stems as khende-n generalized (PE21/59 note #44).
Thus dative/genitive -en was generalized from noun stems originally ending in e, a common ending for nouns. Tolkien deleted this sentence in this particular document, but it seems to be the most likely explanation of the widespread use of the joining vowel e for all the adverbial cases (allative, ablative, locative) in the 40s, 50s and 60s.
A third possible mechanism for deriving the joining vowel was that it was based on the base vowel of the noun stem rather than the suffix itself. There is no evidence of this in consonantal declensions from the 20s and 30s, but there are hints of it in notes on Common Eldarin: Noun Structure from the 1950s:
Thus the fortified forms added to stems yielding monosyllabic nouns always require a dissyllabic stem with ómataima: nenesse “in (the) water” (PE21/79, from stem nēn-).
A similar “joining base vowel” appeared in another monosyllabic declension mentioned but not fully described by Wynne, Smith, and Hostetter: “an unpublished declension of tāl, c. 1967, which gives the locative forms as talasse and talse” (VT43/16). A third possible example is ambartanen (S/223), the instrumental of ambar (ambart-). However, this last example could also represent a form derived from syllabification of ən > ṇ > an, since -anen was the common consonantal instrumental suffix in the Declension of Nouns from the early 1930s used for nouns with any base vowel (PE21/25-28) and ṇ > an remained the usual medial phonetic development in the 50s and 60s (PE22/150).
To summarize, Tolkien used a variety of techniques for deriving the joining vowel for consonantal nouns:
- Generalization from some widespread form, such as plural vowel i or frequent noun ending e.
- Syllabification from lost ə before the suffix, producing results similar to the phonetic development of ṇ, ḷ, ṣ.
- Use of the base vowel from the noun stem, so that the vowel varied by stem rather than by suffix.
Of these, I think the third is most weakly attested, and may only have been limited to monosyllabic consonantal nouns, and possibly only in archaic forms.
Neo-Quenya: Most Neo-Quenya authors throw up their hands at all this complexity, and just use the joining vowel e most of the time, since it was widely used in consonantal inflections in the 50s and 60s. Usually exceptions are made for (a) simple assimilations such as menel → ablative menello (VT43/13) and (b) cases where another joining vowel is well attested, such as i for the 1st person possessive suffix -nya: atar → atarinya (LR/61, UT/186), emil → emilinya (VT47/26). I generally recommend this approach as well, though not every Neo-Quenya authors agrees what the exact set of exceptions should be.