Pronunciation and transliteration

IPA is used only in the section on phonemes. Outside of the tables and figures, phonemes are enclosed by / as expected. In the remainder, which also covers the phonology beyond the phoneme, the transliteration is used. The transliterations are all set in bold.



Consonants are divided into light (non-aspirated, non-breathy) and heavy (aspirated, breathy). The heavy consonants all have in common that more air escapes as they are pronounced. Note the heavy fricatives.

For lack of a better way of showing it, the unvoiced heavy fricatives have been marked up with the extended IPA symbol for strong, unicode U0348 DOUBLE VERTICAL BAR BELOW ( ͈). They are more noisy than ordinary fricatives, pronounced while simultaneously forcing out extra air. The result may sound wheezy or breathless. It could be thought of as a co-articulation of the fricative and /h/.

/t/, /d/, /n/, /r/ and /l/, whether light or heavy, cover the entire range from dental to postalveolar, but are never retroflex.

There are many allowable consonant+consonant clusters in the onset of a syllable, but it is disputed whteher these are phonemic -as clusters-. They do not have their own symbols in the orthography.


  bilabial labiodental dental alveolar postalveolar palatal velar glottal
plosive p b   t d   k g  
nasal   m   n   ŋ  
trill   r
fricative   f v þ ð s z ʃ ʒ (c) h
approximant     j x
lateral approximant l

/c/ is in parentheses as not all dialects have it. It only occurs in onsets and is transliterated {tj}. It is pronounced [c], [cç], [ç], [ʃ], [tʃ]. Where it is not phonemic, {tj} is pronounced as to be expected: [tj].

/þ/ and /ð/ can be bidental.

/h/, /c/, and /j/ are the only light consonants that cannot be geminated.


  bilabial labiodental dental alveolar postalveolar palatal velar glottal
fricative   ʃ ʒ

There is one heavy phoneme that doesn't fit into the chart. It is transliterated {ř} and can be considered a sort of affricate. It is pronounced by sliding from an uvular trill or an uvular or velar fricative or approximant to a standard alveolar trill and it is always long. It is a gargly, growly sound with some chance of excess spit. The IPA transcription would be something along the lines of [ʀrː], [ʁrː], [xrː], [ɣrː] or [χrː].

The light /x/ is not a fricative but an approximant. The heavy /x/, being a fricative, is auditively stronger even without forcing out extra air, so the strong-mark might be considered redundant.

Acoustic differences between light and heavy consonants

The heavy /s/ is far noisier than the light /s/. A spectrogram of hsyn music reveals more noise above 5000 Hz than for an /s/ and especially a band resembling a formant at approx. 18000 Hz.


(vowel-chart) There are three close rounded vowels, /y/, /ʉ/ and /u/. They are all endolabially rounded, or protruded, /y/ being the most protruded of the three. The long vowels are in approximately the same positions in most dialects.

Foreign words are agressively adapted to these vowels. Allowing a too open /a/ or centering any of them is considered barbaric1.


Short consonants

Sound (IPA) UTF8 ASCII Name
p p puỳ
t t taỳ
k k keì
b b buò buv bō
d d daò
g g geò gev
ph pʰūr
th tʰār
kh kʰēr
bh bʰōl
dh dʰāl
gh gʰēl
m ɱ m moh om
n n nes en
ŋ ɴ ŋ N ŋaþ aŋ
mʱ ɱʱ hm hm hmās
hn hn hnef
f ɸ f fu
f͈ ɸ͈ hf hf hfē
θ þ T þam
s s sin
hs hs hsȳr
ʃ ʂ š c šoŋ
x x xa
hx hx hxō
v ʋ w v vy
vʱ ʋʱ wʱ hv hv hvī
ð ð D ðym
z z zen
ʒ ʐ ž j žaŋ
ɦ h ʰ ʱ ̤ ͈ h ʰ h harra
j ʲ ç j ʲ , jil
l l loỳ
r ɾ r raì
rʱ ɾʱ hr hr hreà

In the native script, h the phoneme and h the diacritic has different symbols. The latter is a diacritic placed above a consonant symbol. There is a third h, marking that /b/, /d/ and /g/ in the coda are pronounced heavy, that is: breathy. The transliteration cannot show the difference and is thus not quite satisfactory for this reason.

Long consonants

Sound (IPA) UTF8 ASCII Name
ʀrː ʀʁː ř H řadh

Other long consonants are written twice gg, mm. Their names are derived from the names of the short consonants with the prefix že-: žegev long g.

Short vowels

Sound (IPA) UTF8 ASCII Name
ɐ ɑ a a aga
e ɛ e ele
i ɪ i iji
u o o oro
ʉ ʊ u užu
y ʏ y yny

Long vowels

Sound (IPA) UTF8 ASCII Name
ɑː ā A akān
ē E elēdh
ī I ijīŋ
ō O orōm
ʉː ū U ulūv
ȳ Y ynȳs
ai̯ - aìma
au̯̯ - aòra
a̯u - aóro
ay̯ aỳ - aỳla
ea̯ - eàke
e̯a - eáka
ei̯̯ - eìne
eu̯̯ - eòme
ey eỳ - eỳve
ia̯ - iàgi
i̯a - iága
u̯̯a - oáma
ue̯ - oéne
ui̯̯ - oìŋo
uy̯ oỳ - oỳvo
ʉ̯a - uála
ʉi̯̯ - uìtʰu
ʉ̯i - uíji
y̯a - yáþa
ye̯ - yèny
y̯e - yére
y̯o - yólo
y̯u - yúšu

The transliteration used also has umlauted vowels. These mark mark roots and affixes that have a symbol of their own in the script. A common example is the the "i" in the generalizer-marker or the negation-marker ë. Umlauted affixes never take stress and are untouched by sandhi.


Long, geminate sounds are 1.5 to 2 times longer than the short ones. All vowels can be long, as can all consonants except h, j, and c. ř is always long. Length is phonemic in most dialects and spelled out in the various orthographies, here exemplified by run promise, oath vs. rūn finger.

Timing, stress, tone

This is dialect-dependent. xaldea street for instance, is mora-timed and has a pitch accent. dišuì is syllable-timed and kaìrtin is stress-timed.

Stress is not lexical. It falls on the first long syllable of the root, or the penultimate syllable if there are no long syllables. Secondary stress may fall on other long syllables throughout the entire word, for a maximum of two long syllables in a row. Likewise, the pitch change in those dialects that have it falls on the syllable that receives primary stress in a stress-timed dialect.

Double consonants

Two short consonants of the same type that wind up next to eachother are converted to a single geminate one:

air + ra > airra

Such geminates are never syllabic.

Multiple vowels

Two neighboring vowels can be pronounced as:

  1. Two separate sounds, in two separate syllables, like Anaïs (/ana.i/). This is the default case and therefore unmarked, thus Anaïs would be written anai.

  2. A double vowel. This is two short vowels of the same type, as in summl. The "aa" is either pronounced as two "a"s following eachother, stress on the second "a" and written , or as two "a"s following eachother, stress on the first "a", . The first case, stress on the second "a", is the most usual, and can safely be used for all words where the opposite isn't clearly stated. Double vowels are considered to be a single sound, it could be argued that they are phonemic. However, they are also always in two different syllables.

  3. A diphthong. Taruven has both rising and falling diphthongs, that is the first vowel, then a slide towards the second vowel, then the second vowel. This isn't quite like diphthongs, but close. See Glides below and Double consonants & double vowels above.

Three or more vowels are a combination of one of the above and a short vowel, for instance first a double vowel, then a separate sound, as in xaáil or oòo, falling diphthong followed by a short vowel as in sïaòy, short vowel folloed by a rising diphthong as in sïyál or three or more vowels separated by hiatus.

Taruven is not afraid of multiple vowels: sïaòyen /si.ʹau̯̯.y.en/ winds, uleìneaþ /ulʹei̯̯ne.aθ/ song (accusative), uísge /ʹʉ̯isge/ unicorn, oòo /ʹu.u.u/ owl, aòō /ʹau̯̯.uː/ howl...

Diphthongs are either falling, where the first vowel dominates the pair, or rising, where the second likewise dominates. The <uí> in uísge rises towards /i/, while the <eì> in uleìne falls from /e/. But in both, and for that matter all, cases, both the first and the second sound is clearly heard, the /i/ in <eì>, for instance, is just very short.

In the transliteration the type is marked on the second in the pair, with a grave accent marking a falling diphthong and an acute accent marking a rising diphthong.


A syllable consists of at least one, and only one, vowel or long sound, plus zero or more short sounds.

There can be none, one short, one long or two short consonants in the onset. The nuclues may consist of one short consonant, one short vowel, one long vowel, one diphhtong or one long consonant. The coda can contain none, one short, or one long consonant. The maximum syllable structure is (C)(C)V(V)(C). Thus, saì consists of one syllable (/sai̯/), nnta consists of two (/nn/+/ta/), and sïelle consists of three (/si/+/el/+/le/ or /si/+/ell/+/e/ or /si/+/e/+/lle/) or four (/si/+/e/+/ll/+/e/).

The possible combinations of consonants in the onsets are limited.

Word initial onsets

All single consonants may be used in the onset, regardless of length. Long consonants and nasals never take part in a consonant-cluster.

No voiced fricative is used as the first part of a cluster.

All complex onsets, most frequent first: tš dž sk ks ts tr br tj vr kr gv xv tl šv št xd tv sv pt ŋg nd kv xs fr fl dz sr

ddz dž
tts tštl trtv tj
ffl fr

a Onomatopoeia and interjections only
b Only words derived from the roots *sk "uncertainty, irrealis" and possibly skar2 "thorn"

It could be argued that dz dž ts tš are affricates.

Word final codas

There may be no light plosives in the coda. /b/, /d/, /g/, /p/, /t/ and /k/ are thus either made heavy through the extra-h, or are followed by en epenthetic /e/ known as the extra-e. This ensures that the plosives are released.

All codas, most frequent first: n r l nn v m dh š s gh h ŋ þ ll x th kh ff rr bh ss šš ph ð z vv ř ŋŋ mm

Short codas: n r l v m dh š s gh h ŋ þ x th kh bh ph ð z

Long codas: nn ll ff rr ss šš vv ř ŋŋ mm

Light short codas: r l v m š s h ŋ þ x ð z

Heavy short codas: dh gh th kh bh ph

The Extra "e" and "h"

Browse through any dictionary of Taruven and you will sooner or later remark on the amount of words ending with e or h, or to be more precise: <ge>, <bh> and <dh>. One of them is uísge3, another is dubh.

The extra e is epenthetic. It is not needed when followed by a vowel or long sound in the same word, and is removed.

uísgan two uísge
uísgin a few uísge
uísgen many uísge

Otherwise, the extra e is kept:

uísgegal large uísge

The extra h is only present at the end of a word, never in a word:

duban two knives
dubin a few knives
duben many knives

dubgal large knife


.. barbaric1
Some historians claim this is due to the many wars against the mařaz empire, whose language mařam had the vowels /a/ /e/ /i/ /o/ /u/ and /ə/. On the other hand, the same historians claim /ʀrː/ was borrowed from mařam, it being necessary to be able to refer to the language and culture in question. Other historians like to point out that 1) Taruven is much older than mařam and 2) all the standard AI-templates produce AI that only understand correctly pronounced Taruven even though they are more than capable of learning other languages. The techs of xaldea claim the AIs learn and understand other languages perfectly well but pretend not to.
.. skar2
It is unclear whether skar noun, thorn, is back-formation from skaran or if skaran comes from the dual of skar. skaran (singular) is a short sword, the ends of the crossguard carrying sharp prongs pointing forward. The prongs are also called skaran (dual).
.. uìsge3
uìsge noun, uplifted four-legged, three-hooved omnivore with a mane and a single, straight horn on the forehead. Resembles a cross between a horse and an antelope.