The Proto-Indo-European Sound System

The purpose of this page is to help you work out my transcription. Of course, the Indo-Europeans had no writing system whatsoever, let alone an alphabetic script. The conventional way of writing reconstructed PIE is based on the practice of some nineteenth-century German scholars, most notably Karl Brugmann. With certain modifications made necessary by more recent discoveries and theoretical developments, the conventions codified by Brugmann are still followed by most Indo-Europeanists. They are ingenious and very useful in their own way. However, the people who originated them wrote by hand and had no qualms about using special symbols, indices, diacritic marks and so on. Such things are rather cumbersome if you work with HTML texts; modern alternatives to Brugmannian spelling are hardly better in this respect. I have therefore devised a simplified, HTML-friendly transcription system for writing PIE without any special characters. Here are the rules:

PIE had a large inventory of stops (plosive consonants). Their pronunciation was labial (involving the lips), coronal (involving the tip of the tongue) or dorsal (involving the back part of the tongue). Dorsal stops can be further classified into palatal (‘soft’, a bit like English /k/ in cube), plain (or simple velar, like /k/ in cut), and labiovelar (with lip rounding, like /kw/ in quote). The letters y and w in my system have no other function apart from marking the palatal or labialised character of the preceding consonant; they are not used to stand for independent speech segments. I mention here the IE palatals (and put them in the table below) because most reconstructions found in the standard handbooks require them in the protolanguage. However, it is not quite clear if they could really contrast with ‘plain’ velars; I am inclined to think they couldn’t, so in my reconstructions elsewhere on this site you will only find k etc. where many other people reconstruct ky.

PIE stops could have any of the following three manners of articulation: simple voiceless (like t), simple voiced (like d), or aspirated voiced (like dh, pronounced with a strong puff of breath). The exact pronunciation of these three types of sound is to some extent a matter of speculation, but it is at least certain that all three are needed to account for the observed contrasts in IE languages.

Here is a table of the stop system:
Plain Velar

There was one sibilant fricative, s, and probably a few more velar or glottal (h-like) sounds, known as the ‘laryngeals’. They are very poorly attested in the historical languages (with the notable exception of the Anatolian languages, e.g. Hittite), but the assumption of their existence in the protolanguage (the ‘laryngeal theory)’ is very important for understanding PIE morphology. I shall use the symbols x, xw, and h to refer to the three ‘laryngeals’ required by most versions of the theory. I consider it likely that x was a velar fricative like Scots ‘ch’ in loch, xw was its labialised counterpart, and h was a glottal ‘aspirate’, just like English /h/.

There were two nasal phonemes, m and n, the latter tending to assimilate to the point of articulation of the following consonant, so that e.g. nk was pronounced as in English think. There were also two liquids, l and r, and two phonemes which could function as vowels or glides depending on their position in the string of speech segments (see below): i and u. Finally, there were three other vowels, e, o, and a. Vowels had long variants. They were restricted to certain environments and their distribution was in principle predictable; but length will be explicitly marked here with a colon (:) for the reader’s convenience.

Like many other languages, PIE had syllables without any vowels; some consonants – the laryngeals, nasals and liquids could therefore be syllabic, like /n/ in English button. I shall use the symbol @ placed before a consonant to indicate that it functions like a vowel in being the central element of a syllable. The reader may regard @ as a ‘reduced’ vowel or a brief vocalic articulation accompanying the consonant. @ did not occur independently and was not a separate phoneme. Indeed, consonant syllabicity was entirely predictable from the phonetic environment and I use a separate symbol for it just in order to make the exposition easier.

The vowels i and u, on the other hand, had non-syllabic variants. When followed by another vowel or by @, they were pronounced like the English glides /y/ and /w/ (in yes and well). When followed by a consonant or word-final, they could form a diphthong with a preceding e, o, a, or their long counterparts.

PIE stress was not fixed in relation to the beginning or end of words; it could fall on any syllable, as in English or Russian. If need be, I shall indicate its location by typing the stressed syllable in boldface.

All PIE words are reconstructed, that is, not directly attested. Such indirectly recovered forms are traditionally marked with an asterisk (*).

A few examples of transcription:
(1) *hekuos or *hekyuos ‘horse’; note that in the second transcription ky is a single (palatal) consonant (see above on the palatal problem), and that in either case u is non-syllabic.
(2) *[email protected] ‘wolf’; note that the initial u is non-syllabic, that the following @l is a syllabic consonant as in English middle, and that kw stands for a single (labiovelar) consonant.
(3) *treies ‘three’; note that i is non-syllabic.
(4) *[email protected] ‘you(pl.) have struck/killed’; note that both ghw’s represent single consonants, and that @n stands for a syllabic nasal.
(5) *[email protected]te:r ‘father’; note the syllabic laryngeal and the long vowel. Note also that the second syllable is stressed.

Any questions? Ask Piotr

Back to Index

Hosted by