As I told before because of Ales' request, I have translated the
section on the variants of _r_, of the article about the _r_ in Quenya
that I am editing for Lambenor. Here it goes.
A foreword on this matter is that it is noticeable the resemblance
between the variants of _r_ (_rd_, _hr_, _ry_ and _rr_) and the
variants of _l_ in Quenya (_ld_, _hl_, _ly_ and _ll_). Both consonants
are what in traditional linguistics are called "liquid", i.e.
consonants which may form syllabic groups after another consonant. In
the earliest texts on Quenya Tolkien actually stated that the _r_ was
in the Eldarin tongues "a mere variant of _l_" (PE12:15). And although
there is no evidence of this idea to bee kept in later development of
Quenya, we do see that both consonants always had common features, as
observed in their parallel variations.
We may actually see a close relationship between the symbols of
_rómen_ (which represented some type of _r_) and _lambe_ (which
represented _l_). Not only they are in the same row of additional
_tengwar_ (between 25 and 28), but they both share a wavy line (in
_rómen_ something like a prolongation of the bow of _óre_); and of
them both there are variants for a "following _d_" (_arda_ and
_alda_). So it is likely that _rómen_ was originally designed for
denoting the liquid quality of _r_, not only a sound distinct of
So, although in _tengwar_ only the variants _rd_ and _ry_ are attested
(both as modifications of _rómen_), we may think with little risk that
the other variations (_hr_ and _rr_) should be also ortographically
related to _rómen_, so keeping the parallelism with _lambe_.
Phonologically speaking, these variations should be as well related to
the original phoneme of _rómen_, the trilled /r/.
Now, let us examine each one of them:
_Arda_ occurs in the contact of _r_ with _d_, what usually occurred by
loss of an intermediate vowel (as in _varda_ < _barádâ_, see BARÁD- in
Etym.), in compounds (as _mardil_ < _mar_ + -_(n)dil_), and so on. In
any case, they were still pronounced as two separate sounds ([rd]).
Moreover, each one belonged to one distinct syllable; but Quenya
writing favoured the rendering of two adjacent consonants in only one
symbol when the combinantion occurred often, as this does.
_Hr_ (supposedly rendered as _halla_ + _rómen_) was yet a single
sound, a "voiceless _r_" (LR:1088) - and breathed, according to a note
in LR:1097-, although it was usually derived from the contact of two
distinct sounds, initial _sr_-. It is remarkable that, in opposition
to the previous comment on _arda_, _hr_ is written with two different
_tengwar_ (even being a single phoneme!). But _halla_ may be regarded
a diacritic symbol rather than a _tengwa_, and moreover _hr_ does not
seem to be a so usual consonant in Quenya.
Regarding this matter and the previously commented about the proper
_tengwa_ to use in these variations of _r_, we may also note that the
only attested usage of _halla_ is actually with _rómen_, or rather a
variation of _rómen_ in which its left straight line is extended
vertically. It occurs in the tengwar MSS of Edwin Lowdham (IX:319-21);
but it is of relatively small interest for studying Quenya writing, since
those texts do not represent Quenya but Old English, and moreover
this symbol does not represent a variation of _r_, but of _w_.
Regarding the pronunciation of this _hr_, it is often compared to
Welsh voiceless _r_, i.e. a voiceless trilled alveolar, but note that
Tolkien stated in LR:1097 that it is also breathed (rendered in IPA's
code as a superscript _h_).
On _ry_, it is written with _rómen_ with two dots below, execpt in one
case: the version of _Namárië_ published in _Eorclanstanas_, where
_ómaryo_ is written with _óre_, as stated by Mans Björkman in his
article on "Classic Quenya Mode"
The author of the article states that this is "probably by mistake".
This group is controversial. In LR:1094 Tolkien states that "palatals
were represented by a Fëanorian diacritic denoting 'following y'
(usually two underposed dots)". This is obviously the case, so _ry_
could be regarded as a palatal _r_ (there is no IPA symbol for it, but
it may be represented as a retracted retroflex trill). But some
scholars think that it would be rather a palatalized trill (represent
it as /rj/, with _j_ in superscript), or maybe a group of /r/ + /j/,
where each sound would belong to a distinct syllable. There is a hint
which points to this hypothesis, in some notes on pronunciation
published in PE13:63, where Tolkien stated that _ty_ had originally an
unitary sound, the unvoiced palatal stop [c], but was later spelt as
[cj]. This extension could be common to the whole _tyelpetéma_, and
hence _ry_ would have originally the sound of palatal _r_, but later
the one of palatal _r_ + [j].
Finally we have _rr_ (as found in words like _earráme_, _mirroanwi_,
and so on), that according to the model of other double letters
(specially _ll_ as seen in the _tengwar_ version of _Namárië_, in
words such as _pella_, _tellumar_, Tintallë_..., but also other as
_ss_ in _lassi_, _lissë_...) should be most likely written as _rómen_
with a bar o tilde below. There is discussion on whether
phonologically it is a long _r_ (/r:/) or a geminate (/r.r/); in
LR:1089 Tolkien says that these double consonants are "long" (although
_rr_ is not cited) and so says in LR:1096 about the consonants written
with the underbar, hence we think that the first interpretation is the
Some have pointed that in words as _earráme_, compounded by _ear_ +
_ráme, where as separated words the former is written with _óre_ and
the second with _rómen_, this _rr_ could be written as _óre_ +
_rómen_. But this idea is not very likely, since the _tengwar_ writing
is usually based on the sequence of sounds in the words, not in their
isolated components; for instance, _elen_ ('star') ends with _númen_
and _tári_ ('queen'), starts in _tinco_, but the compound _Elentári_
(seen in the _tengwar_ version of _Namárië) is written with _anto_,
even when the phonemes of the adjacent consonants keep separated
(opposite to /r:/).