(p.251) Appendix II The Indo-European phonological system
(p.251) Appendix II The Indo-European phonological system
The main catalogue of information about the traditionally reconstructed Proto-Indo-European (PIE) phonological system is Mayrhofer (1986). Recent synopses in English are available in Meier-Brügger (2003), Ringe (2006), and Fortson (2010).
The inventory of traditional contrasting segments is laid out in (1).
(1) PIE phonological segments
The voiced aspirates were kept only in Indic, where they remain to this day. They are typically described as breathy voiced but they are in fact voiced and aspirated, as demonstrated instrumentally by Dixit (1975).
The palatal series merged with the velars in all the geographically Western IE languages (including Ancient Greek). For instance, PIE *ḱonk-, as in Vedic śaṅk-ate ‘worries, hesitates’, yields pre-Gmc. *kank-, whence Goth. hāhan* ‘to hang’, etc. Though easternmost of the Indo-European languages, Tocharian patterns in part with Western Indo-European in having a velar stop (e.g. TochA känt, B kante ‘hundred’ = L centum [kéntum], G he-katón, etc.) where the (other) eastern dialects have a palato-alveolar continuant, e.g. Skt. śatám, Lith. šim̃tas ‘hundred’ < PIE *(d)ḱm̥t-ó-m ‘id.’.
In Germanic, the dialectal IE stop system shifted by Grimm’s Law.
(2) Grimm’s Law
*gw (> b/g/w)
Grimm’s Law is responsible for such correspondences as G pod- / L ped- and E foot (< *pōd-) or L quod ‘what’ < *kwod > Gmc. *hwat (OE hwæt, E what). For a recent discussion, see Miller (2010: 1.83–93).
The non-vocalic syllabic sonorants (also called resonants) subdivide into three categories. The glides */y, w/ alternate with vowels */i, u/. The syllabic nasals remained intact nowhere. In Greek and Sanskrit, they became a, in Latin en/in, em/im, and in Germanic un, um. So, for instance the PIE negating particle *n̥ yields G a(n)-, as in a-theist or an-archist, L in-, as in in-secure, and Gmc. un-, as in un-likely.
The syllabic liquids remained intact only in Indo-Iranian. In Germanic, they developed like the syllabic nasals, i.e. -uR- (R = any resonant), e.g. PIE *wĺ̥kw-o-s > Skt. vŕ̥kaḥ, PGmc. *wulf-az ~ *wolf-az > wolf.
One of the archaic features of Anatolian is preservation of at least some of the Proto-Indo-European so-called laryngeals, usually transcribed ḫ in Hittite. The number and phonetic nature of these sounds is disputed (most scholars assume three), and they are variously transcribed. Watkins (2000), for instance, writes ə1, ə2, ə3; others h1, h2, h3. There is also a convention that writes h1 etc. when consonantal, ə1 etc. when syllabic (e.g. *pə2tḗr ‘father’ > Skt. pitā́ / pitár-, G patḗr, L pater),1 and H (or X) when the precise laryngeal is indeterminate or irrelevant. Some core reflexes follow.
(3) Core laryngeal reflexes
*h1e > /e/
*eh1 > /ē/
*h2e > /a/
*eh2 > /ā/
*h3e > /o/
*eh3 > /ō/
*Ho > /o/
*oH > /ō/
At least some long vowels result from contraction of a vowel plus laryngeal, and */o/ was not colored by any laryngeal. Note that *h1 is not written in any of the Anatolian scripts and had no vowel-coloring effects. With the possible exception of *h1, the laryngeals were all fricatives (Byrd 2010: 4).
(4) Examples of laryngeal developments
*h1es-ti ‘is’: Hitt. ēš-zi, G es-tí, L est, PGmc. *isti > Goth. ist, E is
*séh1-mn̥ ‘seed’ (L sēmen ‘seed; SEMEN’) / collective *séh1-mō > PGmc. *sēmȭ > OS, OHG sāmo ‘seed’ (Ringe 2006: 74)
(p.253) *h2ént-i ‘in front’ > *h2ánti > Hitt. ḫānza ‘in front’ / ḫanti ‘opposite; against; facing; apart’ (see Miller 2010: 1.255), G antí ‘against; ANTI-’, L ante ‘in front; ANTE-’, PGmc. *andi ‘in addition; and’ > OE and AND
*méǵh2 ‘great’ > Hitt. mēg by laryngeal deletion, but the rest of Indo-European had epenthesis: *méǵh2ə > Skt. máhi, G méga (Byrd 2010: 85)
*peh2- ‘protect; feed’ (Skt. pā́-ti ‘protects’): *peh2-trom / *peh2-dhlom > *pah2-trom / *pah2-dhlom > Gmc. *fōðra- FODDER / L pābulum ‘food; fodder; nourishment’ PAB(U)-LUM; enlarged *peh2-s- > *pah2-s- > Hitt. paḫš- ‘protect’, L pāstor ‘shepherd’ PASTOR
*-éh2 (factitive suffix) > *-ah2, e.g. *new-eh2- ‘make new’ > Hitt. nēw-aḫḫ- ‘renew; recopy’, L (re)nov-ā-re ‘to renew’ RENOVATE (LSDE 240ff.)
*peh2wr̥ ‘fire’ > *pah2wr̥ > Hitt. pāḫḫur; zero-grade *ph2ur- > *puh2r- (by metathesis) > G pũr (PYRE); Gmc. *fūr-i- > OE fȳr FIRE
*h2ów-i- ‘sheep’ > Hitt. ḫāwi-, Lycian χawa-, L ovis (cf. OVINE), PGmc. *awiz > OS ewi ‘lamb’ (cf. *awjō > OE ēowu EWE)
*h2ost- ‘bone’ > L os / oss- (OSS-ify); cf. *h2ost-eí-o- > G ostéon OSTE(O)-; collective *h2ést-ōi > Hitt. ḫaštāi ‘bone(s)’
*h3ep-en-e/ont- > *h3op-en-e/ont- ‘rich’ > Hitt. ḫappenant-, L opulent-us OPULENT (same root [*op-1] as Lat. opera ‘works’ etc. LSDE 173f.)
*peh3(i)- > *poh3(i)- > *pō(i)- ‘drink’ > L pō-tiō ‘a drink’ POTION, POISON; zero-grade *ph3-tí- > *pə3-ti- > G *poti- > posi- in sumpósion ‘a drinking together; drinking party’ SYMPOSIUM; cf. enlarged *peh3-s- > Hitt. pāš- ‘take a swallow’ (Hittite lost *h3 in many environments)
Some of these examples, e.g. the last, raise the issue of PIE apophony. The most frequent alternations are between */e/, */o/, and Ø (zero-grade). While many details of these alternations are disputed, the three hypotheses in (5) seem generally accepted.
(5) Hypotheses of Indo-European apophony
(a) e-grade (or full grade) is the basic vocalism of most primary verbal roots, e.g. *leikw- > G leíp-ō ‘I leave’; *leǵ- ‘collect’ (LIV 397) > G lég-ō ‘I gather, count, tell, say’
(b) o-grade is conditioned by certain morphological categories; cf. the Greek perfect lé-loip-a ‘I have left’ (< *le-lóikw-h2e), or deverbal nouns like lógos ‘account; reason(ing); speech; word’ (< *lóǵ-o-s).
(c) zero-grade originally occurred when the root was unaccented; cf. the Greek aorist é-lip-on ‘I left’ (< *(é-)likw-óm), Latin past passive participle (re)lic-tus ‘(having been) left’ (< *likw-tó-).
This brief synopsis of PIE phonology is intended solely as background to the historical discussions in this work. For more details, the reader is referred to the basic handbooks.
One final caution: IE (or Indo-European) is frequently used as a non-technical shorthand, especially in reconstructions, as the equivalent of PIE (or Proto-Indo-European).
(1) The reason for the reconstruction of */ə2/ in *pə2tér- is that (i) since the root vowel shows up as /a/ in Greek, Latin, and Germanic, but as /i/ in Sanskrit pitár-, it cannot be simply */a/, which would yield /a/ also in Sanskrit; cf. *ǵhans- ‘goose’ (OHG gans) > Vedic haṃsá- ‘id.’ (IEL 82), and (ii) the specific choice of */ə2/ (as opposed to, e.g., */ə3/) is that Greek has the /a/ reflex, not e.g. the /o/ reflex of */ə3/ (see below).