Although the adverbial suffixes often used a joining vowel when combined with consonantal nouns, in cases where the final consonant of the noun matches the first consonant of the suffix, the suffix was usually assimilated to the noun: Amanna “to Aman” (VT49/26), menello “from heaven (menel)” (VT43/13). In a set of Quenya prayers from the 1950s (VT43, VT44), however, Tolkien experimented with some more complex assimilations.

One of these assimilations was an allative: menelda “to heaven” (VT43/13). Most were locative, however, and Tolkien considered a variety of possible forms:

• cemesse, cemenze, cemende “on Earth” (VT43/17).
• menelle, menelze, menelde “in Heaven” (VT43/16).
• Erumande, tarmenelde “in Heaven” (VT44/34).

The forms cemesse and menelle can be explained by Quenya’s phonetic history, where cemen + sē > cemesse because [ns] became [ss] and menel + sē > menelle because [ls] became [ll] as in ᴹ✶télesā > tella (Ety/TELES), although elsewhere Tolkien said that primitive [ls] survived unchanged in Quenya (PE19/47, 99). The suffix -ze might be explained by generalization of an intervocalic -Vze < -Vse, which was then used as a suffix for consonantal nouns as well.

The locative suffix -de is trickier to explain. It is possible that [ls] > [ld] indicates conceptual variation in the development of ancient [ls]. Alternately, in the Early Qenya Grammar (EQG) written in the 1920s, Tolkien said that there were two short locative suffixes: -se, -de (PE14/78). Thus the 1950s -de suffix might be a restoration of the second EQG form. Some of the above 1950s locatives had “adjectival” variants menellea, menelzea, meneldea (VT43/13, VT44/16), and such adjectival forms with added -a appeared in EQG as well, another indication that Tolkien may have been experimenting with restoring the 1920s paradigm.

As for the allative menelda “to heaven”, it can be explained by either (a) the normal phonetic development whereby [ln] became [ld] (menel + nā > menelda) or (b) by preservation of the primitive allative suffix ✶-da, normally lost in Quenya and replaced by -nna.

Outside these Quenya prayers, there is little evidence of these assimilated locative/allative/ablative forms in Tolkien’s later writings, aside from the simple cases where the consonants match. Wynne, Smith, and Hostetter mention “an unpublished declension of tāl, c. 1967, which gives the locative forms as talasse and talse” (VT43/16) but we don’t know if these are ancient or modern Quenya forms. Given the use of the ómataima (base vowel) a in talasse rather than the more common joining vowel e, I suspect these were ancient forms.

Conceptual Development: As indicated above, EQG from the 1920s had short allative, ablative and locative forms -ta, -lo and -se/-de and Tolkien said “true consonantal words employed shortened forms” (PE14/78), but he gave no real examples of when or how they assimilated to final consonants. In the more detailed Declension of Nouns written in the early 1930s, Tolkien gave numerous examples of consonantal nouns with assimilated suffixes (PE21/20-37), but the short locative -de seems to have been abandoned. Patterns include:

• Allative -ta:
  • tāl, hun, nēr, kas, sūt → talta, hunta, nerta, kasta, sutta
  • hōn (hom-) → honta
  • yāt (yak-) → yahta
  • sat (sap-) → sapsa
  • ambor (ambos-) → ambosta
• Ablative -lo:
  • tāl, hun, nēr, kas → tallo, hullo, nello, kallo
  • hōn (hom-) → holmo
  • sūt, yāt (yak-) → sutyo, yatyo
• Locative -se:
  • nēr, kas, sūt, yāt (yak-) → nerse, kasse, sutse, yakse
  • hun → hunte
  • tāl → talte or talse
  • hōn (hom-) → humpe or honse

Nouns whose stems ended in consonant clusters generally used the long suffixes: qen (qend-): qendanta, qendullo, qendesse; in the 1920s and early 1930s the long allative suffix was -nta. Similar assimilations appear in declension charts from the mid-1930s where the long allative was -nde (PE21/52), but after that there were no clear examples of assimilated allative, ablative, locative forms until the Quenya prayer examples given above from the 1950s.

Neo-Quenya: For purposes of Neo-Quenya, I would avoid these assimilated locative forms. We don’t know which (if any) of the variants Tolkien preferred or whether or not they were transient ideas. The declension paradigms of the early 1930s are simply too different to adopt into Quenya as Tolkien imagined it in the 1950s and 60s. Until we get more examples, I would stick to the longer suffixes -nna, -llo, -sse except in cases where the noun ends in a single matching consonant.