The early chronology of long vowels in Balto-Slavic (original) (raw)

The early chronology of long vowels in Balto-Slavic As Ferdinand de Saussure observed at the end of the 19th century, "à part deux ou trois cas spéciaux (allongement du nominatif, allongement de l'aoriste sigmatique, etc.), l'alternance e-ē n'est pas indo-européenne" (1894: 428), and in these few cases we do not find an acute tone in Balto-Slavic. The original distribution has been obscured by various types of metatony and analogy. A few years later, Jakob Wackernagel pointed out that Sanskrit vṛddhi is found in three categories with seven subdivisions (1896: 66-68): (1) Secondary nominal derivation. (2) Roots with a full grade vowel, (a) in monosyllabic nouns, e.g.-hārd-'heart', (b) before primary suffixes, e.g. hāŕdi 'heart', (c) in the singular of some root presents, e.g. mārṣṭi 'wipes', (d) in the active forms of the sigmatic aorist, e.g. ajaiṣam 'I won'. (3) Final syllables of nominal stems, e.g. (a) nom.sg. sákhā 'friend', (b) loc.sg. agnā́ 'fire'. He concludes that vṛddhi is a variety ("Spielart") of the full grade which arose phonetically in monosyllabic word forms, for which he refers to parallels in Germanic and Afghan. The third category is best explained as phonetic lengthening before word-final resonants (cf. Kortlandt 1975: 85). Most 20th century scholars "accept the existence of an archaic layer of PIE formations characterized by apophonic or invariant lengthened grade" (Villanueva 2011: 7) such as Narten presents (Vedic stáuti 'praises') and causatives (Latin sōpīre 'put to sleep'), heteroclitics (Hittite šēḫur 'urine' < *sēH2ur), s-stem nouns (Greek γῆρας 'old age'), vṛddhi derivatives (OHG swāgur 'brother-in-law'), thematic nouns (OHG āz 'food') and ā-stem nouns (Greek κώμη 'village'). In Balto-Slavic Villanueva finds evidence for Narten presents, causatives and desideratives, lengthened grade iteratives, root nouns, "Narten nouns" and vṛddhi derivatives with an acute tone (2011: 21-32). He acknowledges that we do not find an acute tone in word-final position (Lith. akmuõ, duktė) and in monosyllables (Latvian sāls, gùovs, Lith. duõs, SCr. dònijeh 'I brought'), for which he assumes a type of metatony. Thus, he effectively agrees with the present author on the data identified by Saussure and Wackernagel as representing original lengthened grade vowels, albeit at the cost of introducing an additional rule of metatony. In the following I shall not give a detailed account of the many differences between the two of us (for which I refer to a forthcoming article by Tijmen Pronk) but rather focus the attention on the methodological issues underlying these differences. It appears that there are two basic issues where I find myself in disagreement with the majority of my 20th century colleagues. Firstly, my approach is reductionist in the sense that the number of possible reconstructions must be kept to a minimum. Thus, I reconstruct two, not three PIE velar series, viz. palatovelars and labiovelars, as found in Circassian (e.g. Kuipers 1960: 18), Ubykh (Vogt 1963: 13), and in Salish and Wakashan languages. There are two reasons for this. First, the alleged plain velar series is largely in complementary