After liquids r, l many spirants that arose from either aspirates or voiced stops once again became stops. Tolkien described these phonetic developments in several places in the Outline of Phonology (OP2) from the 1950s. For aspirates > voiceless spirants > voiceless stops after liquids:

[Aspirates] after r and l. Spirantal forms (if developed) were soon restopped so that the normal phonetic products were rp, lp / rt, lt / rk, lk, falling in with voiceless stops. r and l were (as the nasals) generally unfavourable to the development of following spirants. Compare the development of the aspirates after nasals above, and of the voiced stops below.

The cases other than rth, lth (which could be produced by suffixion) were infrequent, since they can only normally be produced for kalta-stems of KALAT-bases, in which final aspirates (other than th) were rare and ph not evidenc[ed]; or from occasional ancient derivatives of KAT-bases with prefix and suppression of medial sundoma as RATH > arthā. This latter is [the] only source of rph, lph, as in base RAPH “seize, grab”, arp(h)o “seizer, thief” > arpo (OP2: PE19/89).

The word Q. arpo given above is the only example of this change involving ph, but lkh, rkh sound changes can be seen in examples like: ✶tarkhildī > Q. tarkildi (PE17/101); Val. Tulukhastāz [> *Tulkhast-] > Q. Tulkas (WJ/399). As noted above, the aspirates may have become voiceless stops directly without passing through a spirantal phase; the evidence is ambiguous. Also note that combinations like lf, rf, lh, rh do occasionally appear in Quenya compounds, but these may be late formations or reformations.

There is strong evidence, however, that rth > rt was not the normal phonetic development. Most of the examples show rth > rs instead, with the usual change of [θ] to [s]. See the entry on how [rs] became [ss] for a list of examples. This may be a remnant of ideas from the 1930s, when voiceless spirants survived after liquids (see below). The closest we come to rth > rt is Q. certa vs. S. certh, but this Quenya word was a loan word from Sindarin, and thus represents phonetic adaptation rather than phonetic evolution. Lack of examples makes it hard to tell how lth really developed in the 1950s and 60s.

As for voiced spirants > voiced stops after liquids:

After r, l. See above under ii (g), for a discussion of the processes of change here involved. It is probable that in AQ the voiced stops were at first opened to spirants after r, l; but that a reaction soon supervened, while ƀ, ð, ʒ were still at that stage, and identical in place of articulation with the original stops (as surviving in mb, nd, ñg(w)), causing some at least of these spirants to be re-stopped (OP2: PE19/93).

There were some dialectical complications with these developments, however. In particular, Tolkien said that “In V. rƀ, lƀ > rb, lb. In N. TQ rƀ, lƀ > rv, lv” and furthermore, he said that “The products of rg, lg; rgw, lgw are in (N.) TQ rr, ll (but rgē̆, lgē̆ > rye, lye); rw, lw; in Vanyarin rg, lg, rw, lw” (PE19/93). Thus it seems the full restopping only occurred in Vanyarin. The spirants ƀ, ʒ seems to remain in the Ñoldorin Quenya dialect, and afterwards [β] became [v] and [lɣ], [rɣ] became [ll], [rr]:

The TQ (Noldorin) forms, and especially the fact that even in PQ the product of rgw, lgw is rw, lw, point to the opening of g > ʒ, where it had not become d [gy > dy]. That the PQ spellings are by the Noldor represented as rg, lg, independently of the Vanyar, may be due to an artificial reconstruction, since the Sindarin-speaking Noldor had no spoken sound [ʒ], but were familiar with the consonant [g] (though not in the groups rg, lg). In view of PQ rw, lw, this is more likely than the re-opening of lg, rg > lʒ, rʒ in Noldorin dialect (whence later ll, rr) (OP2: PE19/93-4).

As indicated by this note, the likeliest scenario in the Ñoldorin dialect is lg > lʒ > ll rather than lg > lʒ > lg > lʒ > ll (and likewise for rg). Thus in the Ñoldorin Quenya dialect, restopping of voiced spirants only occurred for lð, rð > ld, rd. The clearest example of this is:

• ✶galadā > ʒalaðā > ʒalðā > Q. alda “tree” (PE17/135).

Even in Vanyarin, lgw, rgw > lʒw, rʒw > lw, rw before restopping, making it likely this was a more ancient sound change; see the entry on how [lɣw], [rɣw] became [lw], [rw]. The notes above seem to indicate the spirantal combinations lʒ, rʒ otherwise survived into PQ. See the entry on how [lɣ], [rɣ] became [ll], [rr] for a discussion on the possible timing of these later sound changes.

Conceptual Development: There is similar stopping of spirants after liquids in Tolkien’s earliest conception of Qenya from the 1910s and 1920s, but these early phonetic rules are complicated by the fact that voiceless and voiced spirants were part of the phonetic inventory of Primitive Elvish in the 1910s and 1920s. Thus, the spirants + liquid combinations did not exclusively arise from the development of aspirates or voiced stops in these combinations (as they were from the 1930s forward), but were instead often part of words in the Primitive Elvish language of the 1910s and 1920s.

Tolkien described these sound changes in the Qenya Phonology written in the 1910s. In this conceptual period, it seems that voiceless spirants were first voiced after liquids, and that these together with original voiced spirants all became stops, so that (for example) lþ, rþ > lð, rð > ld, rd (PE12/19, 24). The relative order of these changes can be confirmed by the fact that combinations like lt, rt remained unchanged. There is no evidence in the 1910s and 1920 that ld > lð > ld; it seems likelier that in the earliest conceptual stages voiced stops did not become spirants after liquids.

This was also true in the Outline of Phonetic Development (OP1) from the 1930s, where Tolkien said:

Owing to the opening in Q. of all voiced stops (except after nasals and l, r) ... (OP1: PE19/40).

After l, r. In PQ the stops remained: ld, rd, lg, rg, lb, rb. lgw, rgw became lw, rw already in early PQ (OP1: PE19/46).

Some of the changes seen in the Ñoldorin dialect of the 1950s, Tolkien instead ascribed to the Lindarin dialect [pre-Vanyarin] in the 1930s:

In Lindarin ld, rd, lb still remained; but rb > rv; while lg, rg > lʒ, rʒ > ll, rr; but lgē̆, rgē̆ > lye, rye (OP1: PE19/46).

Because voiced stops did not become spirants after liquids in the 1930s, this must have been limited to the Lindarin dialect. However, lgw, rgw > lw, rw in all dialects in the 1930s (as they did also in the 1950s), hinting that some spirantalization happened in at least that combination.

In the first draft of OP2 in the 1950s, Tolkien still attributed some of these sound changes to only the Vanyarin dialect:

In V. rƀ, lƀ > rv, lv. In TQ. rƀ > rv, but lƀ > lb, with restopping (PE19/93 note #18).

The corrections to the final version of the Vanyarin/Ñoldorin dialectical developments in OP2 were made in green ball point, making them fairly late revisions from around 1970 (PE19/17). Phonetic notes from the late 1950s or early 1960s indicate occasional lv > lb restopping in Exhilic Quenya (EQ):

In Q. {both lb and lw > lw: both in EQ > lb} lb and lw existed, then both became lv. Consider[ed] correct but lv often to lb, which is sufficiently common to be allowed as a correct variant. Thus distinct from Vanyar ancient lb from lv but lw from l + w (PE17/129).

There are no attested Quenya words with rv, but there are lv/lb variations throughout the 1950s and 60s, such as: Q. olba “branch” (PM/341) vs. Q. olva “growing things with roots in the earth” (S/45); Q. alve “elm” with variant albe (PE17/146, PE17/153). Tolkien even mentioned that lb was a common Elvish pronunciation of lv in The Lord of the Rings Appendix E:

For lv, not for lw, many speakers, especially Elves, used lb: this was written with 27+6 [jw], since lmb could not occur (LotR/1121).

In the case of aspirates after liquids, the normal development in OP1 of the 1930s was to voiceless spirants without restopping:

rph, rth, rkh > rf, rþ, rh, in the latter h being in PQ a strong spirant: the Lindarin development of rh, lh to long voiceless ř, ľ being seldom intruded into TQ pronunciation and never by Noldor. lph, lth, lkh > lf, lþ, lh. Note ls, rs developed in TQ from PQ lþ, rþ remain as such, with strong voiceless s (OP1: PE19/44).

This can be seen, for example, in ᴹ√NOROTH [> *norþa] > ᴹQ. norsa “giant” rather than **norta (Ety/NOROTH). The same lack of restopping can be partly seen in the original draft of OP2:

r/l + aspirates: rf, {rþ} rh, lf, lh (OP2: PE19/81) [with rþ deleted per note #64].

Later in the original draft Tolkien wrote:

In the case of the aspirates after r, l spirantal forms were only fully developed or maintained in PQ in the sequence rph > rf. The normal products of the other groups were rt, rk; lp, lt, lk, with loss of aspiration, or more strictly with its reduction so that the aspirates and voiceless stops (which were lightly aspirated in Quenya initially and after r, l, m, n) coalesced (OP2: PE19/89 note #101).

Thus it seems in the 1950s Tolkien first decided that þ was restopped after liquids, and then decided that restopping was the general pattern with only rf surviving. Around 1970 he made further revisions in green ball point, first adding lf to the list of combinations that survived, then deciding that rf, lf only surviving in the Vanyarin dialect, before rejecting this as well and settling on the rule of universal restopping of voiceless spirants after liquids as described above, rewriting the entire section in green pen (OP2: PE19/89 note #101).

To summarize the conceptual developments of spirants after r, l:

• 1910s-1920s: Voiceless spirants were voiced and the along with original voiced spirant became voiced stops.
• 1930s: Aspirates became and remained voiceless spirants. Voiced stops after r, l also remained unchanged except in a few combinations, with more spirantalizations in the Lindarin [pre-Vanyarin] dialect.
• Around 1970: Aspirates became voiceless stops (whether they were ever spirantal is unclear). Voiced stops became spirants, but were then restopped universally in Vanyarin (except for a few combinations) but only lð, rð were restopped in the Ñoldorin dialect (and sporadically lv > lb according to The Lord of the Rings Appendix E, but this seems to be a late TQ change).

There was a gradually conceptual shift between the OP1 scenario of the 1930s and the final OP2 scenario of 1970, with various combinations moving incrementally between the two scenarios. Actually examples show rth > rs up through the 1960s, so the stopping of the voiceless dental spirant after liquids may never have been formally adopted.

@@@ voiceless spirants remained unchanged in this period: norsa “giant”, ilfirin, but norta, norto

@@@ but Elben >> Elwen