AQ. [rl] and [lr] became [ll]; [rl|lr] > [ll|ll]
Tolkien said that rl and lr became ll in both the Outline of Phonetic Development [OP1] from the 1930s and the Outline of Phonology [OP2] from the 1950s:
ll, lr, rl: These all gave phonetically ll; but lar, r-l occurred by grammatical analogy (OP1: PE19/49).
ll remained, and was a greatly favoured combination; rl and lr also became ll. In the case of rl (not lr) later reformations produced r-l with intruded vowel determined by the quality of the vowel preceding the r (OP2: PE19/100).
It is possible that the rule whereby [lr] became [ll] was actually a phonetic development in Primitive Elvish, since there is evidence of this phonetic change in all Eldarin language branches in The Etymologies of the 1930s:
- ᴹ✶kalrō > ᴹQ. kallo “hero”; ᴹ✶kalrondō > N. callon (Ety/KAL).
- ᴹ✶stalrā > Ilk. thall “steep” (Ety/STAL).
- ᴹ✶talrunya > ᴹQ. tallune, N. tellen “sole of foot” (Ety/RUN; Ety/TAL).
There is likewise evidence that rl > ll in (Old) Sindarin:
However, it is possible these were parallel sound changes rather than occurring during the Primitive Elvish period, as suggested by David Salo (GS/§4.160, §4.161). The combinations rl and lr are fairly common in Sindarin/Noldorin compounds, and rl appears in at least one Quenya compound: ᴹQ. Erlaire “June”, probably er + laire “*first summer” and thus it is likely an Exhilic compound.
Tolkien indicated that the assimilation rl > ll reoccurred later in Quenya’s phonetic history, after both [d] > [ð] > [r] and [s] > [z] > [r]: see the entry on how [rl] became [ll] in Parmaquesta for further details.
Conceptual Development: I have found no direct evidence of either lr or rl > ll as an ancient sound change in the Early Quenya of the 1910s and 1920s, but Tolkien did indicate rl > ll (or zl > ll) as an assimilation occurring after other sound changes:
[Medial] s gave z and then r before l, r, m, n, w̑, y̑, ƀ, ʒ, giving ll, ´r, rn, rm, rw, ry, rb [or] rw, rg [or] ´r (PE12/19).
See the entry on how [rl] became [ll] for more details.