In Ancient Quenya, any initial [s] appearing before a voiced continuant vanished, unvoicing the following consonant in the process. This occurred before nasals m, n, liquids l, r, and the semi-vowels y, w. Tolkien hinted at these phonetic developments in The Lord of the Rings Appendix E (LotR/1114-5):

LH represents this sound when voiceless (usually derived from initial sl-). In (archaic) Quenya this is written hl ...

RH represents a voiceless r (usually derived from older initial sr-). It was written hr in Quenya.

HY was usually derived from sy- and khy-

Tolkien described this process in some detail in both the first and second versions of his outlines of Quenya phonetic development (OP1, OP2) from the 1930s and 1950s respectively, and they can also be seen in the Comparative Tables of phonetic developments from the 1930s (PE19/21).

Initial sm-, sn-: In the 1930s (PE19/37), Tolkien described this process by saying that the s assimilated to the following nasal to become a voiceless nasal (which he denoted by m̌m, ňn) and then the following nasals unvoiced (m̌m̌, ňň). In the 1950s (PE19/79), Tolkien said that the following nasal unvoiced first (sm̌, sň) and then the s assimilated (m̌m̌, ňň). In both OP1 and OP2, the resulting long voiceless nasal reduced to a simple voiceless nasal (m̌, ň or [m̥], [n̥]), transcribed hm, hn. These sounds still appeared in texts from the Ancient Quenya period, but by Parmaquesta they had voiced again to ordinary nasals m, n. The net effect was that initial s vanished before nasals.

In the 1950s, Tolkien went on to say that the existence of hm, hn in Ancient Quenya texts was the primary evidence for primitive initial sm-, sn-, since initial s vanished before nasals in other Eldarin languages with the exception of Nandorin, itself not well recorded. Tolkien did not mention this in the 1930s, presumably because at this earlier conceptual stage Tolkien allowed voiceless initial nasals to survive in Noldorin [Pre-Sindarin], so there was ample evidence of primitive initial sm-, sn-.

Initial sl-, sr-: In both the 1930s (PE19/37) and 1950s (PE19/79), Tolkien said that these initial combinations became voiceless liquids transcribed hr, hl ([r̥], [l̥]). In the 1950s and 60s, these sounds survived in PQ, and initial hr-, hl- were both mentioned in The Lord of the Rings Appendix E as sounds appearing in Quenya (LotR/1114). In OP2, Tolkien stated that both hr-, hl- were revoiced in the Ñoldorin dialect of Quenya, but the voiceless pronunciation was generally restored after the Noldor encountered Sindarin (PE19/79). In Appendix E, however, he said only that hl- was pronounced as voiced [l] by the Third Age, and hr- remained voiceless (LotR/1114). In any case, the Ancient Quenya forms were preserved in writing, and both hl- and hr- appear frequently in Quenya words from the 1950s and 60s.

Presumably in the 1950s and 60s the voiceless sounds were preserved in the Vanyarin dialect of Quenya, but in the 1930s Tolkien stated that initial hr, hl were revoiced generally in the Quenya language (PE19/37). At this conceptual stage, only the Noldor retained these voiceless sounds in pronunciation because they were common in their native Noldorin [Pre-Sindarin] language. In the 1930s and 40s, initial hr-, hl- rarely appear in Quenya words. However, in the Comparative Tables of phonetic developments from the 1930s, Tolkien revised the developments of sl-, sr- from r-, l- to hr-, hl- (PE19/21, note #36), so it seems he at least considered retaining the unvoiced forms in the 1930s.

Initial sj-, sw-: In both the 1930s (PE19/37) and 1950s (PE19/79), Tolkien said that these initial combinations became voiceless sounds which he denoted ɧ, ƕ and transcribed hy, hw. This lexicon uses [j̊], [w̥] in its phonetic notations for consistency with other voiceless sounds. Unlike the other voiceless initial sounds discussed here, these were always preserved in the normal pronunciation of Tarquesta, and were common initial Quenya sounds from the 1930s through the 1960s. Tolkien mentioned both these as Quenya sounds in The Lord of the Rings Appendix E, where he described the pronunciation of hy as resembling the sound in English “hew” or “huge” (presumably [hj]), and the pronunciation of hw in as English “white” [ʍ] (LotR/1114).

In the 1950s, Tolkien said that in the Vanyarin dialect of Quenya these sound became spirantal, with hw further developing into f and hy into a “sh” sound [ʃ] which he denoted š (PE19/75, 79); he posited a similar pronunciation change among the Lindar [pre-Vanyar] in the 1930s (PE19/34-35). In The Lord of the Rings Appendix E, he mentioned a similar substitution of “sh” for hy among Westron speakers (LotR/1114). At one point Tolkien considered having sj, sw unvoice medially as well, but ultimately abandoned this idea when he remembered that -rya < -sya appeared in Galadriel’s lament (PE19/102, note #164).

Early Qenya: In the 1910s and 1920s it seems that initial s ultimately vanished before nasals and liquids, much like the Quenya developments of the 1930s:

• ᴱ✶smaika- > ᴱQ. maika “cruel” (GL/33).
• ᴱ✶smeigé > ᴱQ. mie² “bit, small piece” (PE13/150).
• ᴱ✶sleiwa > ᴱQ. laiwa “pale” (PE13/149).

Whether these nasals and liquids passed through an unvoiced phase at this early conceptual stage is unclear, but it seems likely.

The development of sy, sw in the 1910s and 20s is confused by the existence of voiceless palatal and labio-velar spirants [ç] and [ʍ] (or [xʷ]?) in the phonetic inventory of Primitive Elvish at the earliest conceptual stage. Tolkien denoted these two sounds as χ̑, x͡w (PE12/15) or sometimes ẏ, ẇ, especially medially (PE12/16). These primitive voiceless spirants also combined with s initially and had similar results as sy, sw.

Initial hy- is quite common in Early Qenya, but it seems to be the result of both initial [ç-] and [sj-] as described by Tolkien in the early Qenya Phonology (PE12/19, 26), similar to how hy- was the result of both [kʰj-] and [sj-] later on. In the Qenya Lexicon, primitive initial [ç] seems to be denoted HY in roots (QL/41). There are several examples of these developments:

• ᴱQ. hyapa “shoe” < ᴱ✶saẏap- (PE12/26), ᴱ√SAYAPA (QL/82) or ᴱ✶χı̯ap (<< deleted sı̯ap, GL/47).
• ᴱQ. hyanda- “mow, cleave” < ᴱ✶χ̑ṇđ (GL/48).
• ᴱQ. hyanda “blade, share” < ᴱ√HYAŘA [ᴱ√HYAÐA] (QL/41).

Compare these early examples with ᴹ✶skyapat- > ᴹQ. hyapat “shoe” (Ety/SKYAP; VTE/46) and ᴹ✶syandō > ᴹQ. hyando “cleaver” from The Etymologies of the 1930s (Ety/SYAD). ᴱQ. hyanda and ᴹQ. hyando were the basis for the sword name Sangahyando “Throng-cleaver” in the 1910s and 1930s, respectively.

For both initial sw and sẇ, the normal development in Early Quenya seems to be to f (PE12/19): ᴱ✶swandǝ > ᴱQ. fan “dog” (PE12/26; QL/82). Initial hw almost never occurs in Early Quenya, the one exception being the masculine 3rd-singular possessive prefix ᴱQ. hwa- with variant form fa- (PE14/54). Later on, the development of hw- > f- was limited to the Lindarin/Vanyarin dialect, as noted above (PE19/34, PE19/79).

Summary of Conceptual Developments: The following table summarizes the conceptual developments:

Primitive Elvish	sm-	sn-	sr-	sl-	sy-	sw-
Early Qenya (1910s-20s)	m-	n-	r-	l-	hy-	f-
Middle Quenya (1930s-40s)	hm- > m-	hn- > n-	hr- > r-	hl- > l-	hy-	hw-
Late Quenya (1950s-60s)	hm- > m-	hn- > n-	hr-	hl-	hy-	hw-

In Early Qenya, initial sẏ-, sẇ- (s combined with voiceless palatal and labio-velar spirants) had the same developments as sy-, sw- > hy-, f-. In Middle Quenya, voiceless hr- and hl- were preserved in the Noldorin use of Quenya (which was not their native language in the 1930s). In Late Quenya, hl- was pronounced (voiced) l- by the Third Age but was preserved in spelling.