The vowel/vocoid phoneme inventory of Proto-Indo-European, as usually reconstructed nowadays, is given in the following table.
| Front or central | Back | Diphthongs | Long vowels | ||||||
| Syllabic |
Non-syllabic | Syllabic | Non-syllabic | ||||||
| High |
*i [i] | *i [ĭ]~[j] | *u [u] | *u [ŭ]~[w] | *ih₁ [i:], *uh₁ [u:] | ||||
| Mid |
*e [e], *h₁ [ɘ] | *h₁ [ɘ̆]~[ɘ̥]~[h] | *o [o] |
*ei [eɪ], *oi [oɪ], *eu [eʊ]~[əʊ], *ou [oʊ] | *ē [e:], *ō [o:] | ||||
| Low |
*h₂ [a]~[ɐ] | *h₂ [ɐ̆]~[ɐ̥]~[ħ] | *h₃ [ɔ] | *h₃ [ʕ̰]~[ɔ̰̆] | *h₂i [ɐi], *h₂u [ɐu] | *eh₂ [ɐ:], *eh₃ [ɔ:] | |||
IPA transcriptions are based upon their manifestations in the surving daughter languages, as exemplified in the Audio Etymological Lexicon and in the table of SoundCorrespondences.html, discussed in detail below.
In the Audio Etymological Lexicon, *i is evidenced in part by *h₂eu-is [ħɐʊɪs] > Latin avis “bird” vs *h₂eus- [hɛus] > English east.
*e is evidenced by *te-ge > English thee vs *teg‑ > English thatch.
The contrast between *i and *e is seen in *piḱ‑ > Lithuanian piktas “angry” vs *peḱu > fee, and *wed‑ > wet vs. *wid- [wid] > English wit.
*u is evidenced by *seh₂l-s > Latvian sāls “salt” vs. *séh₂ul‑ [sɐʊɫ] > Lithuanian saulė “sun”.
The contrast between *i and *u is seen in the offglides of the diphthongs *ei and *eu in e.g. *bʰeid‑ > English bite vs. *bʰeudʰ‑ > Sanskrit बुद्ध buddha; though they are seen to be phonologically similar (possibly non-contrastive) in *gʰlei‑ > English glee ~ *gʰleu‑ > Bosnian glu(miti) “act, pretend”.
Ablaut is the technical term for the morphological variation between related words in which the vowels are varied while the consonants of the root remain fixed. For example:
| Root |
e grade | o grade | zero grade |
| √*s—d | *sed- > sit | *sod- > sat | *-sd-, in *ni-sd-o [nizdo] > nest |
| Lengthened e grade | Lengthened o grade | ||
| *se:d-i‑ > seat | *so:d-o- > soot |
*o is evidenced by its role in ablaut alternations such as *dʰouh₂ [dʱoʊɐ] > dew vs *dʰuh₂ [dʱuɐ] > Siraiki dhuul “dust”, and *h₂e-h₂oik- [ɐħɑik̟ʲ] > English owe, own vs *h₂iḱ- [ħik̟ʲ] > Sanskrit ईष्टे ishte “own”. Further examples of ablaut alternations are tabulated here.
All instances of an open vowel like *a are now analyzed as vocalised versions of *h₂, which is thought to have been a fricative such as perhaps [ħ] at an earlier stage of the language. This topic is discussed in much more detail in the next section.
Although the phonological system of vowels is, at bottom, a fairly common 5-vowel system (i, e, a, o, u), the vowels have rather diverse morphophonological patterns. Specifically:
*i and *u alternate with their non-syllabic variants, *j and *w; see examples below.
*e and *o alternate with one another (and their absence, zero) in the “ablaut” system.
Low vowels [a] and [ɔ] (this should possibly be [o]) are syllabic forms of the earlier “laryngeals” *h₂ and *h₃.
Non-syllabic *w and *j (often transcribed *y) are found in many
words (see entries from wain
to young in the
Audio Etymological Lexicon). *w and *j are in complementary distribution
with the vowels *u and *i, *w and *j occurring in non-syllabic/non-nuclear
position; it is therefore possible to analyse *w and *j (*y) as variants
of *u and *i. They might equally well be transcribed as [ŭ] and [ĭ],
therefore, but the use of separate consonant letters enables the use of
diacritics to be avoided.
In the zero grade, the offglides of *ei, *eu, and *oi alternate with a full vowel *i, and the offglide of *eu alternates with *u (more such examples will be given in the next section):
*ei vs *iLikewise in *h₁weidʰh₁- [hweidʱɘ̥], the ancestor of wide, vs *h₁widʰh₁-uh₂- [hwidʱəwaħ] > widow, Hindi विधवा vidhva.
*eu vs *u
*h₂eus- [hɛus] vs *h₂us‑ “east” > Sanskrit उषस् ushas
*meus‑a- vs *mus-o‑ “moss”
*kreuh₂- [kɾeʊħ] vs *krouh₂- [kɾoʊħḁ] vs *kruh₂- [kɾʊħḁ] “raw, blood” > Persian خون khun
*oi vs *i
In *(h₂e-)h₂oik- [ɐħɑik̟ʲ], the ancestor of owe, *i is an offglide in the diphthong *oi, but in the zero grade of this word, *h₂e-h₂iḱ- [ɐħik̟ʲ], the ancestor of Sanskrit ईष्टे ishte “own”, *i is a full vowel.
Likewise in *h₂oisk- [ħoisk] > ask vs *h₂isk-o [ħisko] > Sanskrit इच्छा ichaa “desire”
*póiḱ‑o- “foe” vs *piḱ‑ > Sanskrit पिशुन pishuna
*ou vs *u
*dʰouh₂ [dʱoʊɐ] “dew” vs. *dʰuh₂ [dʱuɐ] > Siraiki dhuul “dust”, Latin fumos “smoke”
*h₂i, as in *dh₂i-tí- [dɐití] > tide; *dh₂i-mon [dɐimon] > time
*h₂u, as in *ph₂u- [pɐu] > few, Latin paucus
The Audio Etymological Lexicon does not contain any clear examples of *h₃i or *h₃u, perhaps because they would probably come to be homophonous with *oi or *ou. In *gʷeh₃-u-s [gwous] > cow, *h₃ was an off-glide to the preceding vowel, colouring it to [o]. If there was a glide between this [o] and the following [u], it might have been just a phonetic transition rather than a true diphthongal offglide.
Alternations between on-glides and full vowels are also sometimes found. In *wed‑, the ancestor of wet, the initial vocoid is non-syllabic *w, which we might well transcribe instead as an on-glide *ŭ, because in the zero grade form of this word *ud-r-o‑ “otter”, *u was a full vowel.
Long mid vowels *ē [e:] and *ō [o:] are found in the ablaut system. For example:
*sed‑ > English sit vs *sēd-i‑ > English seat
*h₂ous [hɔus] > English ear vs *h₂ṓus-ih₁ “ears” > Persian هوش hoosh “intellect”
The contrast between long and short vowels survived through Proto-Germanic and Old English into Middle English, and subsequently led to a series of extensive alterations to the long vowels known as the Great Vowel Shift.
Long mid-vowels [e:] and [o:] may also arise by combination of short e or o with a following “laryngeal” *h₁, which has a lengthening effect on preceding vowels, an effect that is also found with close vowels [i:] (*ih₁) and [u:] (*uh₁).
*ih₁
*kʷih₁- [kʷi:] > while
*wih₁ro- [ʋi:ɾo] > were(wolf) i.e. “man”
*uh₁
*h₃bʰruh₁-s [ŏ̥bʱɾu:s] > brow
*muh₁s‑ [mu:s] > mouse
*syuh₁- [sju:] > sew
*suh₁-nus [su:nus] > son
*suh₁-īno- [su:hi:no] > swine
*h₂yuh₁n-ḱos [ɐju:nkǝs] > young
*eh₁
*bʰleh₁- [bʱle:] > bleat
*dʰéh₁‑ “do” > Sanskrit धा dhaa “something put down”
*ǵʰeh₁- [g̟ʱe:] > go
*leh₁d- [le:d] > led
*seh₁-to- [se:to] > seed
*speh₁-ro- [spe:ro] > Latin spēs “hope”
*h₂weh₁- [ɐwe:] > Ancient Greek ἄημι ae:mi “wind”
*yeh₁-ro- [je:ɾɵ] > year
*oh₁
*dʰoh₁- [dʱo:] > do
*oh₃ is also expected to lead to a long [o:], but there are no examples in the selection of words given in the Audio Etymological Lexicon.*eh₃ also gives rise to long [o:], as in *gʷeh₃-u-s [gwous] > cow
*eh₂
Long [a:] is attributed to *eh₂ according to “laryngeal” theory, as in *prih₁-eh₂- [pʰri:ɐ:] > friend
[NB: In some of the simulations, e.g. *tri-tih₁o- [triti:o], *kʷieh₁-to- [kʷie:to], and *h₂weh₁-nt- [ɐwent], the supposed long vowels are not actually very long, but this is a deficiency of the simulations to be corrected in future improvements, not a defect of the theoretical analysis.]
For further discussion of the “laryngeals” *h₁, *h₂ and *h₃ and their roles in the vocoid system, see the next section.
| Previous: Fricatives and approximants | Next: Laryngeals |