Automatic generated HTML of A Grammar of Modern Indo-European at Indo-European Language Association

2. Letters and Sounds

2.1 The Alphabets of Modern Indo-European

2.1.1. Indo-European does not have an old writing system to be revived with. In the regions where PIE speakers dwelled four thousand years ago, caves and stones probably still keep some ancient pictographic writings, composed of logograms (graphemes) that represent a morpheme or a whole word, as did Egyptian hieroglyphic logographs.

2.1.2. The Indo-European dialects have adopted different alphabets during the last millennia, and all of them should be usable today – although the main alphabet for today’s European Union is clearly the Latin one. This is a summary table of Proto-Indo-European phonemes and their regular corresponding letters in MIE alphabets: Greek, Latin, Cyrillic, Perso-Arabic and (alphasyllabary) Devanāgarī.

A. Vowels and Vocalic Allophones










Α α

A a


Ա ա

А а


Ε ε

E e


Ե ե

E e


Ο ο

O o


Ո ո

О о


Ā ā

Ա ա

Ā ā


Ē ε̄

Ē ē


Է է

Ē ē


Ω ω

Ō ō


Ո ո

Ō ō









Ι ι

I i


Ի ի

И и


Ī ī


Ի ի

Ӣ ӣ


Υ υ

U u


Ւ ւ

У у


Ū ū


Ւ ւ

Ӯ ӯ











Ρ ρ

R r

Ռ ռ

Р р



Λ λ

L l


Լ լ

Л л



Μ μ

M m


Մ մ

М м


Ν ν

N n


Ն ն

Н н

NOTE. The underdot diacritic might be used to mark the sonorants, as , , , ; usually, however, sonorants appear between consonants, so it is not necessary to mark them, v.i.

B. Consonants and Consonantal Sounds










Π π

P p


Պ պ

П п


Β β

B b

Բ բ

Б б


Βη βη

Bh bh


Բհ բհ

Бх бх


Τ τ

T t

Տ տ

Т т


Θ θ

Th th

Թ թ

Тх тх


Δ δ

D d

Դ դ

Д д


Δη δη

Dh dh


Դհ դհ

Дх дх


Κ κ

K k


Կ կ

К к


Χ χ

Kh kh


Ք ք

Кх кх


Γ γ

G g


Գ գ

Г г


Γη γη

Gh gh


Գհ գհ

Гх гх


Ϙ ϙ

Q q


Ք ք

Къ къ


Γγ γγ  Omicron

C c

Ղ ղ

Гъ гъ


Γγη γγη














Ch ch

Ղհ ղհ

Гъх гъх


Η η













H h

Հ հ

Х х









Ι ι (J j)

J j


Յ յ, Ի ի

Й й / Ј ј


Ϝ ϝ

W w


Ւ ւ

У у (W w)


Ρ ρ

R r

Ռ ռ

Р р


Λ λ

L l


Լ լ

Л л


Μ μ

M m


Մ մ

М м


Ν ν

N n


Ն ն

Н н


Σ σ ς

S s

Ս ս

С с

2.1.2. The Latin Alphabet used for Modern Indo-European is similar to the English, which is in turn borrowed from the Late Latin abecedarium. We also consider some digraphs part of the alphabet, as they represent original Proto-Indo-European sounds, in contrast to those digraphs used mainly for transcriptions of loan words.

NOTE. The Latin alphabet was borrowed in very early times from the Greek alphabet and did not at first contain the letter G. The letters Y and Z were introduced still later, about 50 BC.

The names of the consonants in Indo-European are as follows - B, be (pronounced bay); Bh, bhe (bhay);  C, ce (gway); Ch, che (gwhay); D, de (day); Dh, dhe (dhay); F, ef; G, ge (gay); Gh, ghe (ghay); H, ha; K, ka; L, el; M, em; N, en; P, pe; Q, qu; R, er; S, es; T, te; V, ve; W, wa; X, eks; Z, zet.

2.1.3. The Latin character C originally meant [g], a value always retained in the abbreviations C. (for Gaius) and Cn. (for Gnaeus). That was probably due to Etruscan influence, which copied it from Greek Γ, Gamma, just as later Cyrillic Г, Ge.

NOTE 1. In early Latin C came also to be used for [k], and K disappeared except before in a few words, as Kal. (Kalendae), Karthago. Thus there was no distinction in writing between the sounds [g] and [k]. This defect was later remedied by forming (from C, the original [g]-letter) a new character G. Y and Z were introduced from the Greek about 50 B.C., and occur mainly in loan words in Modern Indo-European.

NOTE 2. In Modern Indo-European, C is used (taking its oldest value) to represent the Indo-European labiovelar [gw] in PIE words, while keeping its different European values –  [k], [ts], [s], [θ], [ʃ], etc. – when writing proper names in the different modern IE languages.

2.1.4. The Latin [w] semivowel developed into Romance [v]; therefore V no longer adequately represented [w] and the Latin alphabet had to develop an alternative letter. Modern Indo-European uses V mainly for loan words, representing [v], while W is left for the consonantal sound [w].

NOTE. V originally denoted the vowel sound [u] (Eng. oo), and F stood for the sound of consonant [w] (from Gk. ϝ, called digamma). When F acquired the value of our [f], V came to be used for consonant [w] as well as for the vowel [u].

2.1.5. The letter I stood for the vowel [i], and was also used in Latin (as in Modern Greek) for its consonant sound [j]. J was originally developed as a swash character to end some Roman numerals in place of I; both I and J represented [i], [iː], and [j]. In MIE, J represents the semivowel [j]. In the Latin script, Y is used to represent the vowel [y] in foreign words.

NOTE. That [j] value is retained in English J only in foreign words, as Hallelujah or Jehovah. Because Romance languages developed new sounds (from former [j] and [ɡ]) that came to be represented as I and J, English J (from French J), as well as Spanish, Portuguese or Italian J have sound values quite different from [j]. Romanisation of the sound [j] from different writing systems (like Devanagari) as Y –  which originally represented in Latin script the Greek vowel [y] –, due to its modern value in English, French or Spanish, has spread a common representation of [j] as Y in Indo-European studies, while J is used to represent other sounds.

2.1.6. The consonant cluster [ks] was in Ancient Greece written as X (Chi) in Western Greek, Ξ (Xi) in Eastern Greek dialects. In the end, X was standardized as [kh] ([x] in modern Greek), while Ξ represented [ks]. In the Latin script, the X stands for [ks], as in English or Latin, whereas in the Cyrillic alphabet it stands for [h] (and aspiration), as well as for [x] in foreign words.

NOTE. The Etruscans took over X from Old Western Greek, therefore it stood for [ks] in Etruscan and then in Latin, and also in most languages which today use an alphabet derived from the Roman, including English. Cyrillic X was taken with its standard Greek value [x], but is also used as [h] in those languages that need it; as, Macedonian, and Bulgarian and Serbian dialects.

2.1.7. As in Ancient and Classic Greek, in the Greek alphabet X stands for [kh], Φ for [ph], and Θ for [th].

NOTE. Because of its use in Modern Greek, they also represent (mainly foreign) [x], [f] and [θ].

2.1.8. Ē represents [ɛː] in the Greek alphabet, because Η was originally used in most Greek dialects to represent the sound [h], and it is therefore used with this value in IE writings, as well as to mark aspirated phonemes.

NOTE. For more on the problem of historical Eta and its representation in the modern Greek alphabet, see <>.

2.2. Classification of Sounds

2.2.1. The Vowels are short [a], [e], [i], [o], [u], written a, e, i, o, u, and long [], [], [], [], [], written ā, ē, ī, ō, ū, respectively. The other sounds are Consonants.

The Modern Indo-European Diphthongs proper are [ei̯], [oi̯], [ai̯], written ei, oi, ai, and [eu̯], [ou̯], [au̯], written eu, ou, au. In these diphthongs both vowel sounds are heard, one following the other in the same syllable.

NOTE. For the so-called long diphthongs [i̯], [i̯], [i̯], written ēi, ōi, āi, and [u̯], [u̯], [u̯], written ēu, ōu, āu, which remained only in Indo-Iranian, Greek and partly in Baltic languages, Schulze (1885) interpreted a regular correspondence of the type āi/ā/ī, which came respectively from the full grade of the long diphthong, the full grade before consonant (where the second element was lost), and the zero-grade (a contraction of schwa with the semivowel). Martinet (1953) proposed that laryngeals were behind those long diphthongs.

In any case, in the languages in which they are retained, long diphthongs have not a longer duration than normal diphthongs; phonologically they are equivalent, as Vedic and Greek metric shows. After Adrados (1995), “The difference, therefore, is not on the duration of the group, but on the relative duration of their components; in other words, for example ei and ēi have the same phonological duration (they are long, as opposed to a brief vowel), but in ei both elements have approximately the same duration, whereas in ēi the duration of i is perceptibly shorter than e. Because of that, the name ‘long first element diphthongs’ is more appropriate to refer to these phonemes”. Cf. Allen (1976) for an analysis of these diphthongs.

Strictly speaking, phoneticians do not consider the so-called rising diphthongs, [je], [jo], [ja], [j], [j], [j], nor [we], [wo], [wa], [w], [w], [w], as diphthongs proper, but rather sequences of glide and vowel.

NOTE. Whilst most Indo-Europeanists differentiate between sequences of approximant and vowel (rising diphthongs) from true falling diphthongs in their transcriptions, i.e. writing [je] (from [i]+[e]) but [ei] or [ei̯] (from [e]+[i]), some use a different approach, considering all of them combinations of vowel plus glide or glide plus vowel, i.e. writing [je] and [ej], or [i̯e] and [ei̯].

Therefore, there are no real triphthongs. The formations usually called triphthongs are [jei̯], [joi̯], [jai̯], [jeu̯], [jou̯], [jau̯], as well as [wei̯], [woi̯], [wai̯], [weu̯], [wou̯], [wau̯]; and none can be named strictly triphthong, as there is a consonantal sound [j] or [w] followed by a diphthong. The rest of possible formations are made up of a diphthong and a vowel.

2.2.2. Consonants are either voiced (sonant) or voiceless (surd). Voiced consonants are pronounced with vocal cords vibration, as opposed to voiceless consonants, where the vocal cords are relaxed.

a. The voiced consonants are [b], [d], [g], [gw], [l], [r] and [ɾ], [m], [n], [z], [j], [w].

b. The voiceless consonants are [p], [t], [k], [kw], [s].

c. The digraphs bh, dh, gh and ch represent the Indo-European voiced aspirates proper, i.e. [bh], [dh], [gh], [gwh], whereas ph, th, and kh represent voiceless aspirates [ph], [th], [kh], mostly confined to words of Greek origin, as well as foreign [ɸ], [θ] and [x], respectively.

d. The consonants [r], [l], [m], [n], and the semivowels [j] and [w], can function both as consonants and vowels, i.e. they can serve as syllabic border or center.

NOTE. There is a clear difference between the vocalic allophones of the semivowels and those of the sonants, though: the first, [i] and [u], are very stable as syllabic center, while [r̥], [l̥], [m̥], [n̥], aren’t, as they cannot be pronounced more opened. Hence the big differences in their evolution, depending on the individual dialects.

2.2.3. The Mutes are classified as follows:





















Labialized velars or Labiovelars [kw] (written q), [gw] (written c), [gwh] (written ch), are pronounced like [k], [g], [gh] respectively, but with rounded lips.

NOTE 1. German Neogrammarians reconstructed a fourth series of phonemes, the voiceless aspirates *ph, *th, *kh, to explain some irregularities in the outputs of the voiceless row. Most Indo-Europeanists reject this fourth independent row of phonemes, and findings of Indo-Iranian, Armenian and Greek have been explained as 1) expressive in origin, 2) contact of a voiceless with a laryngeal phoneme, and 3) effect of a prior s. For support of the fourth row, cf. Szemerényi (1985).

NOTE 2. The modern mainstream Proto-Indo-European reconstruction, that accepts only these two rows of velars as the most logical PIE phonetic system, has been confronted with the question of the actual existence of the groups [kw], [gw], and [ghw], different from (and similar or identical in their dialectal outputs to) labialized [kw], [gw], and [gwh]. A distinction between both is often found, though, whether an independent row of palatalized velars is accepted or not; as, kwōn, dog, ekwos, horse, ghwer-, wild,  kweidos, white, kwet-, cook (cf. O.Ind. kwathati), tekw-, run, etc. which might be found reconstructed as *qōn, *eqos, *cher-, etc. For a defence of such unified forms, see e.g. Jussi Halla-aho at <>.

2.2.4. The so-called Liquids are l, which represents the alveolar lateral approximant [l], an r, pronounced in PIE and in most modern IE languages (at least occasionally) as alveolar trill [r], today often allophonic with an alveolar tap [ɾ], particularly in unstressed positions. These sounds are voiced.

NOTE. About Indo-European r and l, cf. Ban’czerowski (1968).

For foreign words, the group rh represents an [h] sound coming just after the [ɾ], mainly in words of Greek origin. Other groups include rr, the alveolar trill [r], and its aspirated counterpart rrh. The palatal lateral approximant [ʎ] shall be represented as lj.

2.2.5. The Nasals are labial [m], written m, and dental [n], written n. These are voiced. The velar nasal [ŋ] – as ng in English sing – could have existed in IE as allophone of [n] before velars.

NOTE. Erhart (1970) reconstructs three nasals, N, M1 and M2, this one a fricative seminasal with which he explains the results of alternating m and w in some suffixes and roots; as, -ment-/-went-, men-/wen-, etc. It was left unexplained, though, under which conditions did it change.

 The palatal nasal [ɲ] of foreign words (similar to the [n] sound in English onion or canyon) is represented by the pair nj.

2.2.6. The Fricatives are voiceless [s] and voiced [z], z being usually the output of s before voiced consonants.

NOTE. [z] was already heard in Late Proto-Indo-European, as a different pronunciation (allophone) of [s] before voiced consonants, as can be clearly seen in PIE nizdos (for nisdos), nest, which comes from PIE roots ni-, down, and zero-grade -sd- of sed, sit. Because of that it is preferred to write s for [z] in MIE.

It is also possible to write voiceless and voiced pairs from foreign words: labiodentals, f and v; dorsal voiceless h and [x], written kh; and postalveolar [ʃ] as sh and [ʒ] as zh. Possible groups include ks, ts, dz, tsh (for [tʃ]), dzh (for [dʒ]), etc.

2.2.7. The Semivowels are usually written j, and w. These are voiced.

NOTE. Some authors make a distinction between consonantal [j], [w], and vocalic [i], [u]. Actually, however, both appear as TIT and EYE (where T = consontant, E = vowel, I = i,u, Y = j,w), and never as TYT or EIE. Against it, see Schmitt-Brandt (1967) and Szemerényi (1985) and Mayrhofer (1986).

2.2.8. Gemination appears in phonemes whose duration is long enough to be perceived – their implosion and explosion, both audible – as distributed in two syllables. They existed in PIE: in stops, as appās, attās (and tātā), dad, pappājō, eat, or kakkājō, shit; in nasals, as anna-, ammā (and mammā), mother, mum; in liquids, as bōullā, buble; and in the sibilant, as kussō, kiss.

NOTE. They appear mostly in words of expressive origin, children vocabulary, onomatopoeia, etc., which makes it more likely that PIE inherited gemination as an expressive resource, different from its central phonological system; a resource that was retained for a long time by most IE languages as a recurrent possibility.

2.2.10. A synoptic table of the Proto-Indo-European phonetic system:













































s , z




h, *H



r , l












NOTE 1. The existence of a distinctive row of PIE ‘satemizable’ velars, the so-called palatovelars, has been the subject of much debate over the last century of IE studies. Today a majority of modern scholars support only two types of velars in Late PIE – generally Velars and Labiovelars, although other solutions have been proposed, see Appendix II.2.

Palatovelars could be found in PII, though, and are to be represented with Ķ ķ, Ģ ģ, Ģh ģh.

The support of German Neogrammarians to the ‘palatals’ in Proto-Indo-European, as well as its acceptance in Brugmann’s Grundriß and Pokorny’s Wörterbuch, extended the distinction to many (mainly etymological) works, which didn’t deal with the phonological reconstruction problem directly. As Adrados (2005) puts it, about the standard [=Brugmannian] theories nowadays, “Indo-Europeanists keep working on a unitary and flat PIE, that of Brugmann’s reconstruction. A reconstruction prior to the decypherment of Hittite and the study of Anatolian! This is but other proof of the terrible conservadurism that has seized the scientific discipline that is or must be Indo-European linguistics: it moves forward in the study of individual languages, but the general theory is paralised. It is sad when our students go to Germany and come back brainwashed”.

NOTE 2. The cover symbol *H, traditionally *ə, stands for the uncertain Late PIE ouput of the (for Middle PIE) reconstructed laryngeal phonemes h1, h2, h3, which had evolved differently already by the time when Late PIE and Proto-Anatolian were independent languages. There is no consensus as to what these phonemes were like, or how many of them (if any) survived into Late PIE, but it is widely accepted that PIH *h2 was probably uvular or pharyngeal, and that *h3 was labialized. Commonly cited possibilities are ʔ, ʕ, ʕw and x, χ~ħ, xw. See Appendix II.3.

2.3. Sounds of the Letters

2.3.1 The following pronunciation scheme is substantially that used by the common Europe’s Indo-European speakers in ca. 2500 BC, when the laryngeal phonemes had already disappeared, having coloured following vowels, and lengthened preceding ones.

NOTE. MIE cannot permit dialectal phonetic differences, whether vocalic or consonantal – like Grimm’s Law effects in PGmc. consonants, already seen –, because a homogeneous pronunciation system is especially needed when targeting a comprehensible common language. Some differences exist in sister dialects Hellenic, Aryan and Anatolian, though.

2.3.2. Vowels:

ā  as in father

a  as in idea

ē  as in they

e  as in met

ī  as in meet

i  as in chip

ō  as in note

o  as in pot

ū  as in rude

u as in put

NOTE 1. Following the mainstream laryngeals’ theory, Proto-Indo-Hittite knew only two vowels, *e and *o, while the other commonly reconstructed vowels were earlier combinations with laryngeals. Thus, short vowels PIE a < *h2e; e < *(h1)e; o < *h3e, *(h1)o; long vowels ā < *eh2; ē < *eh1; ō < *eh3, *oh. The output of *h2o in Late PIE was either a or o, after the different schools. Short and long vowels and were just variants of the semivowels PIH *j and *w.

NOTE 2. The sonants may have been lengthened too (usually from older sequences of sonant + laryngeal, or because of compensatory lengthenings), especially in the conjugation of verbs, giving thus [r̥], [l̥], [m̥], [n̥], written as r̅, l̅, m̅, n̅. The semivowels can also have a prolonged pronunciation, giving allophones ij and uw. For more details on this see § 2.7.2.

2.3.3. Falling Diphthongs and equivalents in English:

i  as in vein

u   e (met) + u (put)

i  as in oil

u  as ow in know

i  as in Cairo

u  as ou in out

2.3.4. Consonants:

1. b, d, h, l, m, n, are pronounced as in English. n might also be pronounced as guttural [ŋ] when it is followed by another guttural, as in Eng. sing or bank.


There are several ways to generate breathy-voiced sounds:

1.  To hold the vocal cords apart, so that they are lax as they are for [h], but to increase the volume of airflow so that they vibrate loosely.

2. To bring the vocal cords closer together along their entire length than in voiceless [h], but not as close as in modally voiced sounds such as vowels. This results in an airflow intermediate between [h] and vowels, and is the case with English intervocalic [h].

3. To constrict the glottis, but separate the arytenoid cartilages that control one end. This results in the vocal cords being drawn together for voicing in the back, but separated to allow the passage of large volumes of air in the front. This is the situation with Hindustani.

3. p, k, t are plain as in Romance, Slavic or Greek languages, not aspirated as in English; t is never pronounced as sh, as in English oration or creation.

4. g always as in get. It had two dialectal pronunciations, the common simple velar and the ‘eastern’ (later generalized in PII) palatovelar. Compare the initial consonants in garlic and gear, whispering the two words, and it will be observed that before e and i the g is sounded farther forward in the mouth (more ‘palatal’) than before a or o. That is what we represent as ģ, similar to ķ, pronounced as k in key, compared to c in cold.

5. c is pronounced similar to [g] but with rounded lips. Compare the initial consonant in good with get to feel the different articulation. The voiceless q is similar to [k] but pronounced with rounded lips; as c in cool, compared to c in car.

6. bh, dh, gh, ch are uncertain in sound, but the recommended pronunciation is that of the Hindustānī’s “voiced aspirated stops” bh, dh, gh, as they are examples of living voiced aspirates in an Indo-European language (see note to the left).

7. The voiceless aspirated ph, kh, th, frequently of Hellenic origin, are pronounced very nearly like English word-initial p, k, t, as in pen, ten, Ken. Their sound is also described as equivalent to p+h, t+h, k+h, i.e. to the corresponding mutes with a following breath, as in loop-hole, hot-house, block-house.

8. j as the sound of y in yes, never the common English [dʒ], as j in join; w as w in will.

9. Indo-European r was probably slightly trilled with the tip of the tongue (still common today in many IE languages), as in Scottish English curd. Another pronunciation is common today among modern IE languages, and was possibly heard in PIE, the alveolar tap [ɾ], pronounced like the intervocalic t or d in American or Australian English, as in better.

10. s is voiceless as in sin, but there are situations in which it is voiced, depending on the surrounding phonemes. Like the aforementioned [r], modern speakers will probably pronounce [s] in slightly different ways, but this should not usually lead to misunderstandings, as there are no proper IE roots with original [z] or [ʃ], even though the former appeared in some phonetic environments, v.s.

11. Doubled letters, like ll, mm, tt, etc., should be so pronounced that both members of the combination are distinctly articulated.

12. Regarding foreign sounds:

o    kh might represent [x], whether strong, with ‘ach-laut’, such as kh in Russian Khrushenko, or ch Chanukah, or soft, with ‘ich-laut’, such as ch in German Kirche or Lichtenstein. Also, th might be pronounced as English th in thing, and dh as th in this.

o    z, v, f, sh, are pronounced as in English.

o    zh is pronounced as s in English leisure.

o    tsh corresponds to English ch in chain, and tzh to j in jump.

2.4. Syllables

2.4.1. In many modern languages, there are as many syllables in a word as there are separate vowels and diphthongs. This is not exactly so in Modern Indo-European. It follows, indeed, this rule too:

swe-sōr, sister, skrei-bhō, write, ne-wā, new, ju-góm, yoke.

NOTE. The semivowels are always written j and w. So in trejes, three, newos, new, dghwās [‘dn̥gh-ws], languages, etc.

2.4.2. Indo-European has also consonant-only syllables. It is possible to hear similar sound sequences in English cattle or bottom, in German Haben, in Czech hlt, Serbian srpski, etc. In this kind of syllables, it is the vocalic sonant [r̥], [l̥], [m̥], or [n̥] –constrained allophones of [r], [l], [m], [n] –, the one which functions as syllabic centre, instead of a vowel proper:

k-di, heart, w-qos, wolf, de-k, ten, nō-m, name.

NOTE 1. Words derived from these groups, represented TRT (where T = consonant, R = sonant), are unstable and tend to add auxiliary vowels before or after the sonants, i.e. T°RT or TR°T. Because of that, their evolutions differ greatly in modern IE languages. For example, dghwā, language, evolved as [‘dən-ghwa:] into PGmc. tung(w)ō, and later English tongue or German Zunge, while in archaic Latin it was pronounced dingwa, and then the initial d became l in Classic Latin, written lingua, which is in turn the origin of Modern English words “linguistic” and “language”. For wqos (cf. Ved. vkas < PII wkas), it evolved either as [‘wəl-kwos], later into PGmc. *wulxwaz (cf. O.H.G. wolf) or BSl. *wilkas (cf. O.C.S. vьlkъ) or as [‘wlə-kwos], which gave Common Greek *wlukwos (cf. Gk. lykos), Ita. *wlupos (cf. Lat. lupus).

NOTE 2. Apart from the common scheme TRT, another, less stable scheme has been proposed for a common PIE, a certain TRE (where E = vowel); as, PIE *gw°nā, for MIE cenā, woman, or *k°rwos, for kerwos, deer, etc. – conventionally, the symbol ° under the sonant is placed before it in these schemes. Nevertheless, it is commonly accepted that Late PIE dialects did in fact add an auxiliary vowel to this sequence at early times, probably before the first dialectal split: as early Indo-Iranian and Balto-Slavic dialects show, vocalization of TRE had already happened when TRT hadn’t still been vocalized, i.e. T°RE > TERE. Also, many dialects show a common vocalization in [a] for the sonant in some TERE groups, while showing different outputs (even non-vocalization) for TRT. Therefore, even if this theory might make some irregularities fit into a common Late (or Middle) PIE sound, it is not applicable to those early PIE dialectal words, whose vocalization might be inferred using the comparative grammar. Some TRE groups persisted in early IE dialects, though, often from older sequences that included laryngeals, and they are kept in MIE.

2.4.3. In the division of words into syllables, these rules apply:

1. A single consonant is joined to the following vowel or diphthong; as ne-wos, me-dhjos, etc.

2. Combinations of two or more consonants (other than the vocalic ones) are regularly separated, and the first consonant of the combination is joined to the preceding vowel; as ok-tōu, eight, pen-qe, five, etc. but a-gros, field, sqa-los, squalus.

3. In compounds, the parts are usually separated; as a--sta-tis, distance, from apo + statis; or am-bhí-qo-los, servant, from ambhí + qolos.

2.4.4. The semivowels [j], [w] are more stable than sonants when they are syllable centres, i.e. [i] or [u]. However, when they are pronounced lento, they give the allophones (or allosyllables) ij, uw. Examples of alternating forms in PIE include médhijos (cf. Lat. medius), and medhjos (cf. O.Ind. mádhjas or Gk. μέσσος); dwōu, two (cf. Goth. twai, Gk. δω-,), and duwōu (cf. O.Ind. duva, Gk. δύω < *δύϝω, Lat. duo).

2.5. Quantity

2.5.1. Syllables are distinguished according to the length of time required for their pronunciation. Two degrees of Quantity are recognized, long and short.

NOTE. In syllables, quantity is measured from the beginning of the vowel or diphthong to the end of the syllable. Such distinctions of long and short are not arbitrary and artificial, but are purely natural, a long syllable requiring more time for its pronunciation than a short one.

2.5.3.  A syllable is long usually,

a. if it contains a long vowel; as,  mā-tr, mother, --, hide,

b. if it contains a diphthong; as, lai-wós, left, oi-nos, one,

c. if it contains any two non-syllabic consonants (except a mute followed by l or r); as, pneu-, breathe strongly, tmā-mi, cut.

2.5.4. A syllable is short usually,

a. if it contains a short vowel followed by a vowel or by a single consonant; as, pel-nis, skin, or e-í-mi, go,

b. if it contains a vocalic sonant; as, q-mis, worm, c-tis, march.

2.6. Accent

2.6.1. There are stressed as well as unstressed words. The last could indicate words that are always enclitic, i.e., they are always bound to the accent of the preceding word, as -qe, and, -, for; while another can be proclitics, like prepositions.

2.6.2. The oldest PIE was a stress language in which syllable strength was chiefly a matter of pitch differences and, presumably, of intensity (loudness).

NOTE. Following Gąsiorowski, “[i]n this respect it was similar to Spanish or Polish, but not to English with its emphatic ‘expiratory’ stress (…) It thus stood close to the borderline between stress systems and pitch accent systems. Indeed, some linguists have attributed pitch accent contrasts to PIE on the strength of accentual correspondences between Balto-Slavic and Greek. However, scholars such as Jerzy Kuryłowicz and – more recently – Paul Kiparsky have convincingly argued that such contrasts arose independently in the branches in question. The best evidence for the original location of stress in PIE comes from Vedic (Classical Sanskrit developed its own stress system, similar to that of Latin). The location of pitch accent in Classical Greek (especially in Greek noun paradigms) also reflects the PIE stress pattern. There are, to be sure, some specifically Greek constraints on the distribution of pitch accents, but in the environments where such restrictions do not apply, Greek usually agrees with Vedic. In the Germanic languages the original location of stress is sometimes reconstructible thanks to the phonetic ‘fingerprints’ of Verner’s Law. Germanic spectacularly bears out the testimony of Vedic and Classical Greek. Finally, the evolution of pitch-accent systems in Balto-Slavic makes most sense if we adopt the stress system reconstructed on the basis of Vedic, Greek and Germanic as its starting-point”.

2.6.4. The Stress is free, but that does not mean anarchy. On the contrary, it means that each non-clitic word has an accent and only one accent, and one has to know – usually by way of practice – where it goes. Its location depended on the inflectional type to which a given word belonged.

NOTE. Indo-European stress is (at least partly) unpredictable. Rather, it is lexical: it comes as part of the word and must be memorized, although orthography can make stress unambiguous for a reader, and some stress patterns are ruled out. Otherwise homophonous words may differ only by the position of the stress, and it is thus possible to use stress as a grammatical device.

2.6.5. Adjectives are often stressed on the ending, especially if they are derivatives; as, ghtnós, golden, from ghtom, gold, gnōtós, unknown, from gnōskō, know. Nevertheless, nouns and adjective might be stressed on any syllable.

NOTE. There are some accent rules to be followed in the declension of nouns and in the conjugation of verbs, which will be later studied.

2.7. Vowel Change

2.7.1. Vowel Change was common in Proto-Indo-European. In many words the vowel varies because of old alternating forms that gave different derivatives.

NOTE. With the creation of zero-grade stems, vocalization appears, as the original radical vowels disappear and new ones are added. That happens, for example, in root bh- [bhr̥], carry, (cognate with English bear), which can be reconstructed from IE languages as bher-, bhor- or bh-. The same can be said of the semivowels [j] and [w] when they are syllable edges, being syllable centres [i] and [u] in zero-grades.

So for example in o-grade domos, house, which gives dómūnos, lord, as Lat. dominus, Skr. dámūnas; but full grade root dem-, which gives demspóts, master, lord, later despot, as Gk. δεσπότης (despótēs), Skr. dampati, Av. dəg patōiš, (with fem. demspotnjā).

NOTE. The forms attested in Indo-Iranian (and maybe Greek) come from i-stem potis, probably derived from the original Late PIE form dems-póts, cf. ghósti-pots, guest, as Lat. hospēs, hospitis, O.Russ. gospodь<*-ostьpot-; compare, for an original PIE ending -t in compounds, Lat. sacerdōs < MIE sákrodhots, O.Ind. devastút-, “who praises the gods”, etc. The compound is formed with pot-, lord, husband, and pot-njā, mistress, lady.

2.7.2. Different vocalizations appeared in IE dialects in some phonetic environments, especially between two occlusives in zero-grade, impossible to pronounce without adding a vowel; as e.g. skp-, which evolved as Lat. scabo or Got. skaban.

NOTE. Although the dialectal solutions to such consonantal groups aren’t unitary, we can find some general PIE timbres. As a, i with a following dental (especially in Gk. and BSl.) or u, also considered general, but probably influenced by the context, possibly when in contact with a labial, guttural or labiovelar.

2.7.3. Sometimes different reconstructions might account for some vowel differences; a for o, as *lawō for lowō, wash; a vocalic sonant for a or e plus sonant, as *Sos for Samos, summer, or *kwos for kerwos, deer, etc.

NOTE. Different reconstructions might be equally valid, depending on the criteria employed. Sometimes different PIE language stages have to be taken into account; as, for root neqt-, night, a common PIH full-grade *neqts is reconstructible, according to Hitt. nekut; however, Late PIE dialects show that an o-grade noun was later generalized; cf. O.Gk. nuks, nuktós, O.Lat. nox, noctis, for an old PIE consonant stem *noqts. The newer i-stem noqtis was the general Late PIE (and later also PII, EIE) form, cf. O.Ind. nakti, Gmc. naxti, Sla. notjь, Bal. nakti.

The phonological reconstruction of Late PIE includes generally the Schwa Indogermanicum, uncertain in sound, which usually stands for an older laryngeal *h2. In North-West IE, PIE reconstructed *ə usually appears as a; as, statis, standing post, from zero-grade *sth2- of root stā- (<steh2-) stay; or patr, from older *ph2tér-.

NOTE. Other examples are a-stems in *-ī/-jə, from older *-ih2, and neuter plural in *-ə<*-h2.

2.8. Consonant Change

2.8.1. Regarding Consonant Change, different reconstructions might appear, too; as, for ghortos, garden, enclosure, later town (cf. Gmc. gardan, Lat. hortus, Gk. khortos, Phry. -gordum, O.Ir. gort, Lith. gardas, O.C.S. gradu, Alb. garth, etc.), some would reconstruct an alternative *ghordhos, so that both forms (in -t- and -dh-) fit perfectly into the schemes of dialectal phonological laws.

2.8.2. The so called s-Mobile (mobile pronounced as in Latin, it is a neuter adjective) refers to the phenomenon of alternating word pairs, with and without s before initial consonants, in stems with similar or identical meaning. This “moveable” prefix s- is always followed by another consonant. Typical combinations are with voiceless stops (s)p-, (s)t-, (s)k-, with liquids and nasals, (s)l-, (s)m-, (s)n-; and rarely (s)w-.

NOTE. Examples include (s)ten-, compare O.Ind. stánati, Gk. sténō, O.Eng. stenan, Lith. stenù, O.Sla. stenjo, and without s- in O.Ind. tányati, Gk. Eol. ténnei, Lat. tonare, O.H.G. donar, Cel. Tanaros (name of a river). For (s)pek-, cf. O.Ind. spáśati, Av. spašta, Gk. skopós (<spokós), Lat. spektus, O.H.G. spehon, without s- in O.Ind. páśyati, Alb. pashë. For (s)ker-, cf. O.Ind. ava-, apa-skara-, Gk. skéraphos, O.Ir. scar(a)im, O.N. skera, Lith. skiriù, Illyr. Scardus, Alb. hurdhë (<*skd-), without s- in O.Ind. knáti, Av. kərəntaiti, Gk. keíro, Arm. kcorem, Alb. kjëth, Lat. caro, O.Ir. cert, O.N. horund, Lith. kkarnà, O.Sla. korŭcŭ, Hitt. kartai-, and so on.

Such pairs with and without s are found even within the same dialect, as Gk. (s)tégos, “roof”, (s)mikrós, “little”, O.Ind. (s)tṛ, “star”, and so on.

NOTE. Some scholars believe it was a prefix in PIE (which would have had a causative value), while others maintain that it is probably caused by assimilations of similar stems – some of them beginning with an s-, and some of them without it. It is possible, however, that the original stem actually had an initial s, and that it was lost by analogy in some situations, because of phonetic changes, probably due to some word compounds where the last -s of the first word assimilated to the first s- of the second one. That helps to explain why both stems (with and without s) are recorded in some languages, and why no regular evolution pattern may be ascertained: so for example in wqons spekjont, they saw wolves, becoming wqons ‘pekjont. See Adrados (1995).

2.8.3. Before a voiced or aspirated voiced consonant, s was articulated as voiced, by way of assimilation; as, nisdos [‘niz-dos], nest, misdhom [‘miz-dhom], meed, salary, or osdos [‘oz-dos], branch. When s forms a group with sonants there is usually assimilation, but such a trend was sometimes reversed by adding a consonant; as Lat. cerebrum (<Ita. kereθrom), from kersrom [‘kerz-rom], brain.

NOTE. Related to the later assimilation of [s] into [z] between vowels, they became very unstable in some IE dialects, showing sometimes rhotacism; as, snusós, daughter-in-law, cf. Lat. nurus, O.H.G. snur; or genos, race, stock, kind, cf. Lat. genus, generis (<*geneses).

2.8.4. Similarly, the manner of articulation of an occlusive usually depends on its environment. Thus, voiced stops turn voiceless in final position; as, pods, foot, gives voiceless O.Ind. pāt, qid gives O.Ind. cit, agtós gives voiceless Gk. ακτος (aktos) or Lat. actus. The same happens with voiced aspirates, as in legh-, lie (cognate to Eng. log), giving Gk. λεκτρον (lektron), Lat. lectus, O.H.G. Lehter. Voiceless occlusives become voiced before voiced consonants; as, zero-grade ped- in Gk. επιβδα (epi-bd-a).

2.8.5. A sequence of two dentals, such as -tt-, -dt-, -tdh-, -ddh-, etc. was eliminated in all Indo-European dialects, but the process of this suppression differed among branches; Vedic Sanskritshowing little change, some others an intermediate -sT-, and others -ss- or -s-. Compounds were not affected by this trend; as, kréd-dhēmi, believe.

NOTE. This trend began probably in Late PIE, and thus all IE speakers knew such evolutions, which we sum up into a common intermediate stage -st-, -sdh-, etc., which was followed in some early IE dialects, and probably known to the rest of them. See the section Conventions Used in this Book for more on this question. For phonetic changes in Aryan dialects, see Appendix II.

Examples in MIE are e.g. forms derived from PIE root weid-, know, see, which gave verb widējō, cf. Lat. vidēre, Goth. witan, O.C.S. viděti, Lith. pavydéti; p.p. wistós, seen, from wid--, (cf. O.Ind. vitta-, but Av. vista-, O.Pruss. waist, O.Sla. věstъ, or Gmc. wīssaz, Lat. vīsus, O.Gk. ϝιστος, O.Ir. rofess, etc.); noun wistis, sight, vision, from wid-ti-, cf. Goth wizzi, Lat. vīsiō; Greek wistōr, wise, learned man, from wid-tor, cf. Gk. στωρ<*ϝίστωρ (wístōr), PGk wistorjā, history, from Gk. στορία (historía); Imp. wisdhi! know!, from wid-dhí, cf. O.Ind. viddhí, O.Gk. ϝίσθι, O.Lith. veizdi, and so on.

2.9. Peculiarities of Orthography

2.9.1. Indo-European words may show a variable orthography, although a unified one should be strongly encouraged.

2.9.1. Vowel Changes that influence the way MIE is written include the alternating PIE forms that gave different frozen derivatives. 

A vowel change that should not affect MIE orthography is what many reconstruct as PIE [ə] or schwa, generally evolved as North-West IE a; as, PIH *ph2tér- PIE *pətér- EIE patér-, father; PIH *bhh2tis → PIE *bhətis → EIE bhatis, appearance; PIH *anh2mos → PIE *anəmos → EIE ánamos, breath, and so on.

NOTE. This Late PIE reconstructed schwa (see §2.2.1) is important for the different vocalism of EIE, PII and PGk; cf. MIE patér- with Aryan pitár-, or MIE ánamos with Hellenic ánemos. 

2.9.2. Consonant Changes that should not affect MIE orthography, already seen, include voiced sibilants, as nisdos [‘niz-dos], kersrom [‘kerz-rom]; and voiceless occlusives, as pods [pots], agtós [ak-‘tos], leghtrom [‘lek-trom], -pd- [bd].

NOTE. Although the accuracy of some allophones in PIE is certain, for practical reasons the phonetically correct notation is therefore avoided in favour of the phonemically correct notation.

Changes that usually affect how MIE is written include commonly reconstructed variants, as egh-, ek-, outside, out, from; and doubious cognates, as necr-, dark, and neqt-, night, maybe from a common PIH suffixed *negw-, to dawn.

2.9.3. About semivowels, as a general exception, they are not written when the semivowel is the last sound of the first word in a compound; e.g., for triathlom (from tri-, three, and Gk. athlon, “contest”), triathlon, we won’t write *trjathlom;  sindhueurōpājóm, and not *sindhweurōpājóm; etc.

NOTE. In Modern Indo-European, compounds may be written with and without hyphen, as in the different modern Indo-European languages. Nevertheless, the older, not hyphenated version is preferred for formal writings; as, sindhu-eurōpājóm for sindhueurōpājóm, compare Eng. Indo-European, Ger. Indoeuropäisch, Fr. Indo-européen, It., Sp. indoeuropeo, Gal.-Pt. Indo-européu, Cat. indoeuropeu, Du. Indo-Europees, Pol. indoeuropejski, Lit. indoeuropiečių, Ir. Ind-Eorpach, Russ. индоевропейский, Gk. ινδοευρωπαϊκή,  Ira. هندواروپایی, Hin. हिन्द-यूरोपीय, etc.

2.9.4. The vocallic allophones [r̥], [l̥], [m̥], [n̥] may be written, as in Latin transliterations of Sanskrit texts, as , , , and , to help the reader clearly identify the sonants; therefore, alternative writings ṇmṛtós, inmortal, któm, hundred, wodṛ, water, etc. are also possible.

2.9.5. An Apostrophe is used to mark the ommited letter of a contraction in word-final position, usually in elisions at the end of imperative verbs, especially in spoken language; as cemj’ for cemje, come here; or takej’ for takēje, shut up.

2.9.6. An Acute Accent is written over the vowel or semivowel in the stressed syllable, except when stress is on the second to last syllable (or paenultima) and in monosyllabic words. Accented long vowels and sonants are represented with special characters. The weak vowel of a possible diphthong is also accented; so in eími [e-‘i-mi], I go, and not *eimi, pronounced [‘ei-mi] if left unaccented.

2.9.7. The forms with the copulative -qe, and, and disjunctive -w, or, are usually written by adding it to the preceding word, as in Latin -que, but with a hyphen.

2.9.8. The capital letters are used at the beginning of the following kinds of words:

a.  the names of days, months, seasons and public holidays; as, Jānwārjos, January, Samos, summer, Newos Atnos, New Year, etc.

b. the names of people and places, including stars and planets; as, Sāwel, Sun, Aleksanr, Alexander, Deiwos, God, Sindhu, Indus (river), Teutiskolondhom, Germany (cf. O.H.G. Diutisk-lant<*þeudiska-landam), etc.

NOTE. In old IE languages demonyms were not written in capital letters; as, Eur, eurōpājós; Angljā or Angljolondhom, England (cf. O.E. Engla-land, “land of the Angles”), but angljós, English; Hispānja, Spain, but hispānós, Spanish; teutiskós, German; and so on.

c.  people’s titles, as Prōbhastr, Professor, Kelomnelis, Colonel, Rēgtr, rector,

d. Skeuros, North, Déksinā, South, Áusteros, East, Éperom, West, and their derivatives.

NOTE. Germanic Nertros, lower, later North, from ner-, lower, bottom, and Suntos, “of the Sun”, later South, possibly from alternative root Sun- of Sāwel, sun; Gmc. West comes probably from the same root as wespros, evening.

e. in official or well-established place names; as Plátejā, the Square, etc.

2.10. Kindred Forms

Compare the following Europe’s Indo-European words and their evolution in Germanic and Latin, with their common derivatives in Modern English.





English (Lat.)

patr,  father




father (paternal)

sept, seven




seven (September)

trebhō, dwell




thorp (trabecula)

globjō, hold, clench




clip (globe)

bhrātēr, brother




brother (fraternal)

bherō, carry




bear (infer)

wertō, turn




worth (versus)

trejes, three




three (trinity)

dek, ten




ten (