At every conceptual stage of Sindarin (including earlier Gnomish and Noldorin), all initial nasalized stops become stops: [mb-] > [b-], [nd-] > [d-], [ŋg-] > [g-], the only exception being when those stops became syllabic and developed a preceding vowel. These sound changes appeared in the phonetic Comparitive Tables from the 1930s (PE19/20), and Tolkien mentioned them on numerous occasions:

The initial PQ combinations of nd, ȵg, mb were anciently preserved in ON. In late ON for 5 b t appear 2 x w, i.e. d, g, b indicating a change by loss of nasal in absolute initial position or after a consonant. But the nasal forms are still preserved after vowels (as still occasionally in archaic Gondolic) (Feanorian Alphabet, late 1930s, PE22/29).

Since Telerin and Quenya both simplified mb, nd, ñg > nasals (not stops as in Noldorin [emphasis added] and Beleriandic), ñ- also = ñg - ñ = ñgy [in Telerin] (second version of Tengwesta Qenderinwa [TQ2], circa 1950, PE18/103-104).

The nasalized initial groups ... These groups [mb, nd, ñg] in Quenya normally simplified to nasals initially (in Telerin they became b, d, g) (Outline of Phonology [OP2], early 1950s, PE19/76).

These initial nasal groups [mb, nd, ñg] were normally simplified in the development of Quenya and (later) of Sindarin. The Quenya development was > mm´ > m´; in Sindarin in absolute initial position the nasal was lost: mb´ > b´, but if a proclitic word such as the article i “the” preceded, mb remained and developed as medially, > mm > m (from notes on the etymology of umbar, late 1960s, PE17/104).

The timing of this sound change is hard to nail down. The first note above indicates the sound change began in Late Old Noldorin. The last note indicates it was a “later” Sindarin change. As hinted at in the final note, the Sindarin sound change was late enough that it still played a role in the soft mutation system; it did so for both Noldorin and Sindarin (see below). The Eldamo data model places this phonetic rule after Old Sindarin but very early in the Sindarin period, which is more or less where David Salo put this sound change as well (GS/§4.78).

There are numerous examples of this phonetic rule. A representative sample:

• ✶mbandō/mbanda > S. and N. band “prison” (MR/350, Ety/MBAD).
• ✶ndorē > S. and N. dôr “land” (WJ/413, Ety/NDOR).
• ✶ñgolodō > S. and N. golodh “Noldo, one of the wise folk” (PM/360, Ety/ÑGOLOD).

As mentioned above, remnants of these ancient nasalized stops appear in the Sindarin and Noldorin systems of mutations. Though the initial sound of these words became b-, d-, g-, their mutated forms were distinct from words that originally began with simple stops, since they typically preserved the nasal in some forms. For example nguruthos was the lenited form of guruthos “shadow of death” (PE17/95), versus aear the lenited form of gaear “sea”, which shows the usual loss of initial g- in its mutated form (PE17/27). See the entry on soft mutation for further details.

There is one example that hints that this change might sometimes occur in the interior of words at morpheme boundaries: Aragorn < OS. Ára-ngorn (PE17/113), but since there are several other etymologies for the name Aragorn, it’s hard to say whether or not this was a general rule or a transient idea (the latter seems more likely).

Conceptual Development: As noted above, this sound change appeared in the earliest versions of Gnomish from the 1910s, and was retained throughout the conceptual evolution of Sindarin.

GS/§4.78 @@@ plus morpheme boundary: talagand