Sindarin has a number of consonant mutations with various effects. Tolkien said:

The lenitions or “mutations” of S. were deliberately devised to resemble those of W[elsh] in phonetic origin and grammatical use; but are not the same in either p[honetic] o[rigin] or g[rammatical] u[se] ... [in a footnote] though of phonetic origin, they are used grammatically, and so may occur or be absent in cases where this is not phonetically justified by descent (from a 1972 letter to Richard Jeffrey, Let/426).

Indeed, the two most frequent Sindarin consonant mutations are soft mutation and nasal mutation, and both bear a strong resemblance to the corresponding mutations in Welsh but differ in a few particulars. The principle effect of mutations is the modification of a following consonant, which typically undergoes sound changes similar to its historical medial development under the influence of a closely preceding word or clitic.

Of the Sindarin mutations, the best understood are:

• The soft mutation which is the result of a (often vanished) preceding vowel.
• The nasal mutation which is the result of a (often vanished) preceding n.
• The mixed mutation which is a mixture of soft and nasal mutation.

Soft mutation is the one that is most “grammaticalized” and is introduced into a number of contexts where it could not be the result of historical sound changes. One example given by Tolkien:

palan-tîriel [which appears in LotR as soft mutated palan-díriel] should phonetically > -thíriel, past participle “having gazed afar”; but grammatically before actual forms of verbs, the soft mutation only was normally used in later S., to avoid the confusion with other verb stems (ibid., Let/427).

In this context, the soft mutation is probably the result of an adverb preceding a verbal form: “far (palan) having-gazed (tíriel)”. It is used as a grammatical mutation rather than the nasal mutation that would result from purely phonological changes. In her book A Fan’s Guide to Neo-Sindarin, Fiona Jallings reserves the term “lenition” (strictly speaking just another term for “soft mutation”) for soft mutations that are grammatical rather than phonological in nature, and I have adopted this convention as well.

There are three other Sindarin mutations that are less understood and are to some degree speculative:

• The stop mutation which is the result of a preceding stop d or t.
• The liquid mutation which is the result of a preceding liquid l, r.
• The sibilant mutation which is the result of a (lost) preceding sibilant s.

Compared to other mutations, they are obscure and poorly attested and may or may not have been a regular feature of Sindarin as Tolkien conceived of it. Of these, stop mutation has the best support and is mentioned in several places. Sibilant mutation is described in only a single source, and has no clearly attested examples or counter examples. Liquid mutation is not described in any published source, examples of it appear in only a single source, and a several counter examples appear elsewhere. In terms of how likely they are to be part of Sindarin, I would rank them (from most to least likely) as stop, sibilant, then liquid, but all three can probably be ignored in Neo-Sindarin writing as they arise relatively rarely.

As a final note, some mutations also modify the preceding consonant as well as the following consonant. This is most notable in nasal, mixed and stop mutations, where the preceding n or d may also undergo changes, vanishing or being modified along with (or instead of) the following consonant.