Proto-Indo-European Phonology

6. Laryngeals in the Neighborhood of Gmc. /r l m n/

6.1. Gmc. reflexes of PIE /r l m n/ and laryngeals

The Gmc. dialects, in contrast with all other IE dialects but Armenian, do not show a distinction between the reflexes of PIE [r̥] and [r̥X], [l̥] and [l̥X], [m̥] and [m̥X], [n̥] and [n̥X], or according to Brugmann's notation, r̥ and ṝ, l̥ and ḹ, m̥ and ṃ̄, n̥ and ṇ̄. In Gmc. the reflexes of the vocalic allophones of PIE /r l m n/ followed by laryngeals is the same as that of these allophones when they were not followed by laryngeals; compare Skt. jātá ‘born’, Goth. aírþa-kunds ‘born on earth, earthly’ PIE [gn̥Xt-], with Skt. matá ‘thought’, Goth. gamunds ‘memory’, PIE [mn̥t-]. In terms of the laryngeal theory we should say that the laryngeal in /gnXtó-/ was lost in Gmc. without effect on the [n̥], while in most of the other dialects it was lost with lengthening of the preceding resonant.

The Gmc. development of the vocalic allophones of PIE /r l m n/ before consonant, vowel, or laryngeal is: [r̥] Gmc. ur (ru), [l̥] Gmc. ul (lu), [m̥] Gmc. um, [n̥] Gmc. un; e.g. Goth. þaúrsus, OIcel. þorn, OS thorn ‘thorn’, Skt. tŕ̥ṇa ‘blade of grass’. The consonantal allophones of these resonants show up regularly in Gmc. as r l m n; e.g. Goth. ráuþs, OIcel. rauðr, OE rēad ‘red’, Lat. ruber.

But we find in some words in the Gmc. dialects resonant followed by a vowel where we should expect only a consonantal resonant. In most such forms the vowel is u; beside u occur also i and a. Examples are: OE hærfest < *harubist ‘harvest’, cf. Skt. kr̥ṇāti ‘harms’; OHG birihha, cf. Lith. béržas ‘birch’. These forms have been noted, but no explanation has been given.¹ The unexpected vowel cannot be an excrescent (svarabhakti) vowel, because it causes umlaut, as in OE hærfest.² Nor can it be a reflex of PIE /_eX/ (ə); for we should then expect Gmc. a, which in these words is rarer than u. A reexamination of the material is needed in an endeavor to explain why such a vowel is found in a particular group of words and to seek an explanation for its origin.

6.2. Words in which vowels are found after Gmc. /r l m n/

Five examples may be cited for such a vowel after r: OHG birihha, OE hærfest, OE hærðan, OHG hiruz, OHG muruwi.

OHG birihha, OIcel. bjǫrk ‘birch’ is cognate with Skt. bhūrja ‘a kind of birch’, Lith. béržas ‘birch’. The Skt. and Lith. forms give evidence for a laryngeal after /r/. WP derive these words from PIE /bherX-g-/ (bherēg-).

OE hærfest (< *harubist) ‘autumn’ is from the PIE root /(s)ker-/ ‘cut’, from which a great number of forms have developed. Another with p determinative is Gk. κρώπιον ‘sickle’. Although WP refuse to admit that the ρω of κρώπιον developed from /rX/, there is ample evidence for laryngeal in many forms, e.g. Skt. kr̥ṇāti ‘harms’. I therefore assume that OE hærfest developed from a form of this root with laryngeal suffix.

OE hærðan (< *haruþjan) ‘scrotum’ is from the same extended form of this root but with t determinative.

OHG hiruz, OE heorot, OIcel. hjortr ‘deer’ is derived from PIE /ker-/ which has many forms with laryngeal suffix, e.g. Gk. κέρας ‘horn’; among these are forms with w determinative, such as Gk. κεραός ‘horned’, Welsh carw ‘deer’. We may therefore assume PIE /ker-_eX-w-/.

OHG muruwi, marawi, OE mearu ‘tender’ are cognate with Skt. mūrná ‘crushed’; this form and the Skt. verb mr̥ṇā́ti ‘crushes’ give us evidence to assume a base /mer-eX-/.

Three examples may be cited for such a vowel after l: OHG halam, OE hælfter, OHG skiluf.

OHG halam ‘blade of grass’ is cognate with Gk. κάλαμος ‘reed’ and Serbo-Cr. slȁma ‘straw’ which gives evidence for laryngeal; WP suggest PIE /kol_eXmo-s/ (k̑oləmo-s). In OHG halm, OIcel. halmr, OE healm the unaccented vowel was lost.

Unaccented u was preserved in OE hælfter ( < *haluftri) long enough to cause umlaut; we have no trace of it in OHG halfter ‘bridle’. These nouns are derived from PIE /(s)kel-/ with p determinative. While most developments from this root give no evidence of laryngeal suffixes, the Lith. nouns kálpa ‘crosspiece on a sled’ and kìlpa ‘stirrup’ have the intonation of a laryngeal base. Since the evidence for laryngeal is so slight, we can only hesitantly suggest that the Gmc. nouns developed from PIE /kel-X-p-/.

Although there are no words in other IE dialects corresponding to OHG skiluf, skilaf ‘reed’, WP suggest a derivation from the same root with b determinative.

The unaccented vowel of Goth. miluks, OHG miluk ‘milk’ is said to be similar in origin. Since the original provenience of these words is disputed, we can draw no conclusions from them. (See WP 2.298-9.)

Two examples may be cited for such a vowel after m: OHG demar, emiz.

OHG demar ‘dusk’ is cognate with Skt. támisrā, támas ‘darkness’. On the basis of these Skt. words and Lith. témti ‘get dark’ PIE /tem-X-/ may be assumed.

OHG emiz, emazzi ‘constant, busy’ is cognate with Skt. amīti ‘oppresses’, Gk. ὀμοίιος ‘oppressing’; WP assume a PIE base /om-eX-/ (omō-).

Two examples may be cited for such a vowel after n: OHG anut, OS wanum.

OHG anut, OS anad, OIcel. ǫnd ‘duck’ are cognate with Skt. ātí ‘a water-bird’, Gk. νη̑σσα ‘duck’, Lith. ántis ‘duck’. These are developments from PIE /anXti-/.

OS wanum, wanam ‘splendid’ are cognate with Skt. vāná ‘dear’; this and Skt. vanóti ‘loves’ are from PIE /wen-/ with laryngeal suffix.

The unaccented vowels of OHG kranuh, kranih, OE cranoc ‘crane’ are said to be similar in origin. WP 1.591-3 derive these words, like Gk. γέρην, OS krano, Lith. gérvė and other names for the crane, from the root /ger-/ ‘cry hoarsely’. Since there are no words in other dialects corresponding exactly to kranuh, we can base on it no conclusions for the origin of the unaccented vowel.

6.3. The conditions under which these developments took place

Those examples with u after resonants for which we have similar cognates in other IE dialects give us evidence for development from laryngeal bases. I assume that in these Gmc. words a reduced vowel was preserved between the resonant and laryngeal; in some forms the reduced vowel regularly became u; in others it was lost. Then the laryngeals disappeared. We do not have much evidence for this development, for, like svarabhakti vowels, u (a i) from PIE /_e/ was lost in most words in the various Gmc. dialects; as a matter of fact our primary evidence for them in OE is modification of the preceding vowel.

There are some Gmc. words in which we might expect developments from /_e/ that show no trace of them. OHG kerno, OIcel. kjarni, like Lat. grānum ‘corn’, are no-formations based on a laryngeal base /ger-X-/. But there is no trace of such a reduced vowel in kerno. It may have been lost by analogy with the zero grade form OHG, OIcel. korn, in which /_e/ would have disappeared after PIE [r̥]. OHG stirna ‘forehead’ likewise shows no trace of /_e/ although it is a development from a laryngeal base /sterX-/, cf. Skt. str̥ṇā́ti ‘strews’. I assume that in these words the reduced vowel was lost before the time our forms were written, as in OHG halm beside halam.

Hirt finds difficulty in explaining why the ‘middle vowel’ was lost without a trace in Goth. -kunds ‘born’, OHG korn ‘grain’, etc.³ These forms, however, are quite different in formation from those in which a reflex of /_e/ is preserved. In -kunds, korn, etc. the resonant was vocalic. In the forms listed above the resonant was consonantal. The reduced vowel /_e/ and the laryngeals were preserved only after the consonantal allophones of the resonants.

I conclude that in laryngeal bases /_eX/ did not develop to a after r l m n as it did elsewhere in Gmc., e.g. in Goth. staþs, Skt. sthíti, Gk. στάσις ‘standing’; Goth. raþjō, Lat. ratiō ‘reason’, rērī ‘believe’. In laryngeal bases on the other hand /_eX/ was maintained after /_eX/ elsewhere became a, and /X/ was then lost; finally /_e/ became Gmc. /u/, occasionally /i/ or /a/.

Footnotes

¹ Schmidt, Vocal. II.373 (1875); Bezzenberger, BB 17.216 fn.2 (1891); Hirt, IF 7.194 (1897) and IG 2.118 and 194; Noreen, Abriss 87; Persson, Beitr. 685 fn.

² BB 17.216.

³ IF 7.194-5.