Quenya words for “tell” and “tale” began with nyar- for much of Tolkien’s life, but their derivation shifted over time. The earliest root for such words was ᴱ√NYAŘA [NYAÐA] “relate, tell” from the Qenya Lexicon of the 1910s with derivatives like ᴱQ. nyara- “relate, tell” and ᴱQ. nyara “tale” (QL/68), but it had no obvious derivatives in the contemporaneous Gnomish Lexicon. In The Etymologies of the 1930s Tolkien first gave this root as ᴹ√NYAR “tell, relate”, then changed it to ᴹ√NAR² with a Quenya-only variant nyar- (Ety/NAR²; EtyAC/NAR²). This change may have been in keeping with Tolkien notion from the 1930s that palatalized dentals might have been a Quenya-only innovation (PE18/44). The root ᴹ√NYAR/NAR had derivatives like ᴹQ. nyáre/N. narn “tale” and ᴹQ. nyar- “tell”.
However, in both the Outline of Phonetic Development (OP1) from the 1930s and Outline of Phonology (OP2) from the early 1950s Tolkien indicated the primitive form of the Quenya verb was ñgyar- (OP1: PE19/36, OP2: PE19/76). This may have been in keeping with Tolkien’s later decision that initial palatalized dentals were not a feature of Primitive Elvish at all; see the entry on how [[at|[j] was lost after initial dentals]] in the Welsh-like branch of the Elvish languages for discussion of this transition. Despite this decision, Tolkien continued to use S. narn “tale” regularly in his later writings, despite there being no way it could be derived from ñ(g)yar-.
Neo-Eldarin: For purposes of Neo-Eldarin, I think it is best to ignore Tolkien’s decision to remove initial palatalized dentals from Primitive Elvish, and I recommend retaining √NYAR as the root for “tell” and “tale” in Elvish, this being the only reasonable option for retaining both Q. nyar- and S. narn.
A rejected root in The Etymologies of the 1930s with a single derivative: ᴹQ. narwe “sign, token”, along with an unglossed (and probably incomplete) form nar... (EtyAC/NDAR).