Part 1
Proto-Indo-European
Chapter 1
Proto-Indo-European phonology
Mate Kapović
Introduction
Proto-Indo-European phonology is reconstructed on the basis of its reflexes in Indo-European languages – mostly older ones, generally preserving more archaic features, i.e., closer to PIE. The principal languages referenced in this chapter for the reconstruction of PIE are Old Indic/Indo-Aryan (Vedic and Sanskrit), Avestan (Old/Gatha- and Young/Late Avestan), (Ancient) Greek, Latin, Old Church Slavic, Lithuanian (Old and Modern), Gothic, Hittite, Old Irish, Tocharian (A and B), (Classical) Armenian, and Albanian. Of course, other languages (Sabellic, Latvian, Old Prussian, Old High German, Old English, Luwian, Gaulish, Old Persian, etc.) are also used when necessary or when they have a form or feature not present in a language from the first group. Ideally, PIE reconstruction should take into account all the relevant information present in any IE language/dialect. This overview tries to present the reconstructed PIE phonological system and its main reflexes in major ancient IE languages. A more detailed overview of reflexes for specific languages/branches can be found in separate chapters in this volume.
The following phonological system can be reconstructed for the last phase of PIE:
The phonemes *b and *a (together with *ay and *aw) can be considered marginal, or at least infrequent, phonemes (p. 18, 42), *a also being an allophone of *e when in contact with *h2 (p. 43). Reconstruction of *ā not stemming from original *eh2 is very questionable (p. 42). Short *i and *u most likely had long allophones in monosyllabic forms (p. 55).
PIE stops were voiceless, voiced, or (voiced) aspirated. The velar stops were unmarked (*k, *g, *gh), palatalized (*kˊ, *ǵ, *ǵh) (IPA [kj, gj, gjh]), or labialized (*kw, *gw, *gwh). Besides *s, the “laryngeals” *h2 and *h3 (marked numerically) were almost certainly fricatives as well (some kind of h-type sounds). The *h1 “laryngeal” (provisionally labeled a fricative in the above scheme) might also have been a fricative, but this is less certain (p. 30). The sign *H is used for all three laryngeals when it is not known which one is to be reconstructed or when we need a mark for all three laryngeals. Resonants and semivowels were either asyllabic (*m, *n, *l, *r, *w, *y (IPA [j])) or syllabic (*m̥, *n̥, *l̥, *r̥, *u, *i) (IPA [m̩, n̩, l̩, r̩]). The vowels *i and *u act phonologically in the same way as syllabic resonants – the relation of *i/u to *y/w is the same as that of *m̥/n̥/l̥/r̥ to *m/n/l/r. Some reconstructable segments appeared only as allophones – cf. the non-phonemic fricatives *z (p. 29), *þ (IPA [θ]), and *ð (and *ðh) (p. 53) and the vowel *ǝ (p. 46–47, 52).
What is immediately typologically noticeable in the PIE phonological system, among other things, is the paucity of “real” vowels (if one does not take into account syllabic glides, long vowels, and diphthongs), the very unusual marginality (or even complete lack of) the vowel a, and the complete lack of affricates (no phonemic *c [ʦ], *č [ʧ], *ǯ [ʤ], etc.). On the other hand, the stop system was rather rich (with 15 of them). In later IE languages, phonological systems usually developed in the following direction: the vowel system became typologically more common, with a becoming more frequent (additionally, i and u lost their phonological relation to j and w/v, while diphthongs often became monophthongs); syllabic resonants (*m̥, *n̥, *l̥, *r̥) disappeared everywhere (later they appeared again in some languages); a lot of languages developed new affricates; the “laryngeals” (*h1, *h2, *h3) disappeared in most languages in most positions, but new fricatives appeared; the stop system simplified everywhere; and languages preserved only one additional distinctive feature – aspiration or labialization. Velar palatalization as such was nowhere preserved – it either completely disappeared or was the origin for new affricates and fricatives.
Note on the PIE “orthography”: various authors/schools write reconstructed PIE phonemes differently. Palatalized velars can also be written as *k’, *k̂, *kj, *ky, etc.; *ku̯ can be written instead of *kw; aspirated stops can be written as *bh, *dh, *gh; laryngeals can be written as *H1, *H2, *H3 or *ə1, *ə2, *ə3 (the latter usually for “syllabic” laryngeals), and an unknown laryngeal as *hx (instead of *H); the glides *y, *w can be written as *i̯/j, *u̯; *eu/eu̯ instead of *ew, etc.
Table 1.1 PIE stops | | voiceless | | voiced | | aspirated |
labials | | *p | | *b | | *bh |
dentals | | *t | | *d | | *dh |
velars | | *k | | *g | | *gh |
palatalized velars | | *kˊ | | *ǵ | | *ǵ h |
labialized velars | | *kw | | *gw | | *gwh |
No modern IE language has preserved PIE palatalized velars as such. On the other hand, labiovelars are preserved even in some modern IE languages – cf. the phonetics of English queen [kwiːn] (p. 25). PIE aspirated stops are traditionally reconstructed as voiced because of their reflexes, which are mostly voiced (p. 28). However, voiced aspirated stops are typologically less common than voiceless aspirated stops – cf. the initial allophonic aspirated [ph], [th], [kh] in English pin, tin, kin. In PIE, velars could be both palatalized and aspirated (*ǵh) and labialized and aspirated (*gwh) at the same time.
PIE voiceless stops do not change in most of the languages (p, t, k), the main exceptions being Germanic (where they change to voiceless fricatives f, þ, x) and Armenian (where they change to voiceless aspirates *pc, tc, kc). The phoneme *p weakens and disappears in Celtic. In most languages, PIE voiced stops remain unchanged (b, d, g). However, in Germanic and Armenian (perhaps also in Anatolian) they become voiceless p, t, k (in Tocharian all stops merge into p, t/ts, k). The phonemes traditionally reconstructed as voiced aspirated stops change in most IE languages. They are preserved only in Indo-Aryan (bh, dh, gh); in Greek they remain aspirated but become voiceless (φ [ph], ϑ/θ [th], χ [kh]); in Italic they change to fricatives (voiceless initially: *f-, *þ-, *x-; voiced medially: *-β-, *-ð-, *-γ-), and in other branches they change to plain voiced stops (b, d, g). PIE palatovelars change to plain velars in some languages (k, g) and to affricates (c[ʦ]/ć[ʨ]/ č[ʧ], d˘z[ʣ]/d˘ź[ʥ]/d˘ž[ʤ] …) and, further on, to fricatives (s/ś[ɕ]/š[ ʃ ], z/ź[ʑ]/ž[ʒ] …) in others. PIE labiovelars remain labialized velars in some languages (kw, xw, gw), or change further to full l...