Greek and Latin from an Indo-European Perspective

By Coulter George, Matthew McCullagh, Benedicte Nielsen and Antonia Ruppel

This volume presents new work exploring how the study of historical linguistics can advance our understanding of Greek and Latin and, conversely, how the classical languages can help us to reconstruct Proto-Indo-European and the culture of its speakers. Classical and Indo-European linguistics have been particularly exciting areas of research in recent years, and this book is intended to provide insight into some of the main areas of current debate. It stems from an international conference held in Cambridge in 2005 and includes contributions from keynote speakers Andreas Willi and Joshua Katz. The book covers a wide range of topics: phonology (the accentuation of Greek monosyllables, the development of laryngeals in Greek, and typological discussion of the glottalic theory); morphology (the prehistory of the past-tense augment, the iteratives and causatives of the Latin second conjugation, the origin of the Latin prefix co(m)- , Indo-European root nouns and s-stem neuters, Greek and Latin reflexive pronouns, the Greek comparative suffix); the etymologies of etymos, Achilles, adulare, and a Macedonian gloss; the significance of the Greek particle tar; and comparisons of Sanskrit matrimonial names and poetic terminology with their Greek counterparts. Greek and Latin from an Indo-European Perspective demonstrates the continuing relevance of linguistics for the study of ancient languages and literature, and will be of interest to classicists, Indo-European linguists, and historical linguists generally

• Language: English
• Publisher: Cambridge Philological Society
• Release date: Aug 30, 2020
• ISBN: 9781913701383

Book preview

Preface

In July 2005, nearly thirty classicists and Indo-Europeanists from eleven countries gave papers at a conference in Cambridge dedicated to the light which the classical languages shed on Indo-European linguistics and, conversely, the way in which historical linguistics can improve our understanding of Greek and Latin. We had organised the meeting primarily as an opportunity to bring together researchers in the early stages of their academic career: considering the number of new ideas circulating in the field, many of them spurred on by the work of a younger generation of linguists, we thought it would prove invaluable to the discipline to offer a platform where the latest theories could be discussed. Additionally, to provide a touch of seniorum experientia, Prof. Joshua Katz of Princeton and Prof. Andreas Willi of Oxford were kind enough to deliver two keynote talks. In the end, the high standard of the papers presented at the conference, combined with the usefulness of the discussion sessions that followed, convinced us that a published account of the proceedings would be an important contribution to the literature on the linguistics of Greek, Latin and Indo-European. Accordingly, we have collected here seventeen of the papers, all thoroughly revised, spanning the wide range of linguistic topics discussed at the conference.

The volume opens with three papers that focus on phonology: Thomas Olander offers a new explanation for why some Greek monosyllables have an acute accent, others a circumflex; Adam Hyllested and Paul S. Cohen propose a new sound law to account, inter alia, for the initial ὑ- in Greek ὑφαίνω; and Brett Miller presents a detailed survey of the most recent typological evidence affecting the plausibility of the Glottalic Theory. It then moves to two papers dealing with verbal morphology: Andreas Willi argues that reduplication marked perfectivity in Indo-European, and on this basis presents a new explanation for the verbal augment as a reanalysed reduplication syllable in *h1e-; Daniel Kölligan explains the functional dichotomy of iteratives and causatives in the Latin second conjugation with reference to the valency of the base verbs in Indo-European. Next, three papers on particles, preverbs and pronouns: Joshua T. Katz offers new arguments for positing a Greek particle ταp by examining the syntactic and narrative contexts in which it and its composite αὐτάρ occur; Dag Haug argues on the basis of evidence from Plautus that Latin co(m)- has an allative force, marking goal, rather than the comitative sense generally ascribed to it; and Nicoletta Puddu makes the case that we can only understand the Greek reflexive pronouns properly if we move beyond a purely syntactic description to a more pragmatic analysis of their use. Questions of nominal morphology are addressed in the next four papers: Jenny Helena Larsson argues that the reconstruction of an ablauting Indo-European root noun *ṵo k-/*ṵe k- cannot be supported on the basis of Greek and Baltic evidence; Roland Litscher proposes that the equation between κρέας and kravth is illusory, thus changing our view of the Indo-European s-stems; Ulla Remmer accounts for Greek female names in -ώ with reference to Sanskrit matrimonial names in -āyī-; and Carlo Vessella explains the anomalous long α in comparatives such as ϑᾶσσον, ἔλασσον, and ἆσσον. The next four papers deal with etymologies of individual words: Michiel de Vaan explains Latin adūlāre by connecting it with auidus; George Hinge equates ὅσιος and ἔτυμος with Sanskrit satyá- and tūtumá- by a careful look at the development of laryngeal + consonant clusters; Alexander Nikolaev presents a new etymology for the name of Achilles, namely *h2ṇĝhi-(h2/3)ṷl(h1/3)-o- ‘pain/death-slaying’ and suggests a revision to Rixs Law before nasals in Greek; and Wojciech Sowas account of the Macedonian gloss ἄλιζα reveals a new Balkan isogloss. The volume concludes with a paper on Indo-European poetics, in which Anne Mahoney draws our attention to similarities in the technical vocabulary used by Greek and Sanskrit poets to describe their craft.

The editors of the volume are grateful for the assistance of many individuals over the course of its production, whom it is our pleasure now to thank. First, this work would never have been possible without the conference that gave rise to it. We would therefore like to thank all those who helped to make it so successful: our co-organizer, Oliver Simkin; the University of Cambridge’s Faculty of Classics, for providing not just funding, but also the ideal venue; James Clackson, for arranging an idyllic garden party in the warm hospitality of Jesus College; Newnham, Pembroke and Trinity Colleges for housing our speakers; John Donaldson, Aidan Foster and Rupert Thompson, for providing logistical assistance; and John Penney, for overseeing our application to the British Academy, whose generous financial support was crucial to the project. Tim Whitmarsh, who oversaw the publication of this collection in the supplement series of the Cambridge Classical Journal, deserves considerable thanks for his patience and flexibility in dealing with a volume that posed so many challenges in several different areas of book-production; Andrew Goodson, our copy-editor at the CCJ, and our anonymous reviewer both provided much careful and detailed advice that improved many aspects of all the papers assembled here. We would like to express our particular gratitude to Thomas Olander, whose typographical expertise has proved invaluable. His extraordinary skill at setting diacritical marks and formatting text to suit the complicated conventions of the field has ensured that the volume has a far more attractive and uniform appearance than would otherwise have been possible.

The accentuation of Greek monosyllabic words

THOMAS OLANDER

University of Sofia

1 The Greek tones are distinctive in accented long final syllables,¹ e.g. φυγών aor. pple. masc. ‘flee’ with acute tone (rising tone, i.e. high pitch on the second mora of the vowel) vs. φυγῶν fem. gen. pl. ‘flight’ with circumflex tone (falling tone, i.e. high pitch on the first mora of the vowel). The tonal distinction is also found in the only syllable of monosyllabic words, which may be considered a special case of final position, e.g. φὡς masc. man vs. φῶς neut. ‘light’; or πoὑς masc. ‘foot’ vs. βοῦς masc./fem. ‘ox’.

In polysyllabic words the acute tone basically appears on reflexes of Proto-Indo-European long vowels or short vowels followed by a laryngeal, while the circumflex tone is found in long syllables which have arisen as the result of post-Proto-Indo-European contractions. We might assume that the prosodic development of monosyllabic words would correspond to that of the last syllable of polysyllabic words. However, the basic accentuation principle of polysyllabic words is not valid for monosyllables, as is shown by e.g. μῦς masc. ‘mouse’, with circumflex tone on an uncontracted vowel. Thus I find it reasonable to analyze the accentuation of Greek monosyllabic words independently of that of polysyllabic words. In this paper I shall examine how and when the tonal opposition arose in monosyllabic words.

First, word-forms with an immediately transparent accentuation are filtered out, viz. finite verbal forms (§3), contracted word-forms (§4) and forms ending in PG *-ns (§5).² Then I will proceed to the remaining cases and try to ascertain if it is possible to reconstruct a language stage where the acute and circumflex tones were in complementary distribution, i.e. a stage where the choice of tone was automatically determined by the phonological shape of the word-form (§6).

Because of the special prosodic conditions that may apply to function words as opposed to content words with a full lexical meaning, these words will not be treated here. I also leave out of consideration word-forms that have been remodelled to such an extent that their accentuation cannot be regarded as original, e.g. χήν masc./fem. ‘wild goose’;³ and words for which a safe Proto-Greek reconstruction cannot be given, e.g. σής masc. ‘moth’, the enigmatic χρή ‘it is necessary’ and the type represented by Homeric δμώς masc. ‘slave’.

Certain words that are disyllabic in Greek were possibly monosyllabic at an earlier stage. This applies (a) to words like ἀνήρ masc. ‘man’, ἀστήρ masc. ‘star’, ὀδών masc. ‘tooth’, and (b) to ἰχϑῦς masc. ‘fish’ and ὀφρῦς fem. ‘eyebrow’. Notice the different accentuation of the words of the former group, where the prothetic vowel reflects a laryngeal, and those of the latter group, with a prothetic vowel probably of non-laryngeal origin.⁴ Interestingly, words of group (a) are apparently accented according to the rules of polysyllabic words, while the accentuation of words of group (b) seems to be in accordance with the accentuation rules of monosyllabic words that will be established in this paper. I leave these words out of further consideration.

2 The attempts to establish the original distribution of the acute and circumflex tones in Greek monosyllabic word-forms differ significantly from each other. The traditional explanation of the Greek tonal distinctions in terms of similar tonal distinctions in Proto-Indo-European is hardly sustainable within a laryngealistic framework. Eduard Schwyzer maintained that the only phonetically regular reflex on a long syllable is the circumflex tone.⁵ According to this hypothesis, we find the expected tone in word-forms like δῶ, κῆρ, κρῖ; βῆ, στῆ; Ζῆν, βοῦς, βοῦν, γραῦς, ναῦς, φϑοῖς; εἷς, πᾶς; πῦρ, δρῦς, μῦς. The accentuation of all monosyllabic word-forms that have an acute tone is, consequently, regarded as secondary. The acute tone of ϑείς and δούς is taken from the oblique case forms (ϑέντος, δόντος); that of μήν, μείς, κτείς, χϑών, ϑήρ, κήρ, χεἱρ, Κρής, σής, πούς is introduced by analogy with the polysyllabic n-, r- and dental stems. Similar explanations are given for ρ ς, ς, ϑώς, Τρώς; and in φὡς ‘man’ the acute is introduced to differentiate the word from φῶς Tight’. As the exceptions to Schwyzer’s theory constitute a large and heterogeneous group, it is not surprising that other proposals have been made.

The opposite position to that of Schwyzer is taken by Hermann Berger, according to whom the regular tone of Greek monosyllabic word-forms is the acute.⁶ Instances of circumflex tone are regarded as secondary, e.g. in βοῦς cow’ and ὗς/σῦς ‘pig’, where Berger proposes an onomatopoeic explanation of the unexpected tone. Although Bergers theory seems to account for a greater part of the material than that of Schwyzer, it still leaves a number of cases unexplained. The same applies to other hypotheses on the origin of the Greek tones, such as Jerzy Kuryłowiczs morphologically conditioned accentual developments,⁷ or Paul Garde’s idea of a connection between tone and vowel quality.⁸

3 Monosyllabic finite verbal forms with circumflex tone, e.g. Horn, βῆ ‘walk (aor. 3. sg.)’, Horn, βῆς ‘walk (aor. 2. sg.)’, σχῶ ‘have (subj. 1. sg.)’, constitute an immediately transparent case. The accentuation of these word-forms conforms to the principle that all finite verbal forms are accented on the leftmost mora possible within the limitations of the Dreisilbengesetz, a result of the circumstance that the forms were unaccented at an earlier stage. Since the accentuation of finite verbs is automatic in Greek, it does not help us clarify the rise of distinctive tones. A similar principle accounts for the accentuation of vocative forms such as Zεῦ and βοῦ.

4 As is shown by internal and external reconstruction, a number of synchronically monosyllabic word-forms in Attic are reflexes of contractions of originally disyllabic words. The accent remains where it was before the contraction: if the accent was on the second vowel, the contracted syllable receives acute tone; if it was on the first vowel, the contracted syllable receives circumflex tone. This explains the acute tone of the following words:

δᾁς fern, ‘fire-brand’ (acc. δᾷδα⁹); cf. Horn, δαΐς;

κλεἱς fern, ‘key’ (acc. κλεῖδα); cf. Ion. κληίς;

πρών masc. ‘foreland’ (acc. πρῶνα); cf. epic πρη ν;

and the circumflex tone of the following words:

νοῦς masc. ‘mind’, acc. νοῦν; cf. uncontracted Horn, νόος, νόον;

οἶς masc./fem. ‘sheep’, acc. olv; cf. uncontracted Horn, ὄϊς, ὄϊν, Arg. οƑις;

παῖς masc. ‘boy’ (acc. παῖδα); cf. the frequent disyllabic scansion, πάϊς, in Homer;¹⁰

τρεῖς masc./fem. nom. pi. ‘three’; cf. uncontracted Dor. τρεες;

φῶς neut. ‘light’ (gen. φωτός); cf. uncontracted Horn, φάος, Pamph. φαβος.

If the final accentuation of Horn, γρηΰς is authentic, we expect acute tone in γραῦς fem. ‘old woman’, acc. γραῦν (gen. γρᾱός);¹¹ however, the circumflex may be easily understood as an adaptation to the principle which will be presented below in §7 (cf. monosyllabic acc. γραῦν). In Π ν masc. ‘Pan’ (acc. Πᾶνα; cf. uncontracted Παονι (Melpea, sixth century BC)),¹² it is unclear how the uncontracted form was accented.

The circumflex tone of vαυς fem. ‘ship’ (Hom. acc. νῆα) is the expected outcome of PIE *náh2us with an intervocalic laryngeal.¹³ Notice, however, that even a hypothetical monosyllabic Indo-European proto-form would receive circumflex tone in accordance with the law of tonal distribution presented below in §6.

5 Another group of word-forms with transparent accentuation is constituted by words ending in PG *-Vn(t)s. In these forms, the acute tone is the regular outcome:¹⁴

ϑ ς masc./fem. ‘heap; beach’ (acc. ϑῖνα);

ς fem. ‘sinew, tendon (acc. ἶνα); the -n- may be a secondary addition to the stem;¹⁵

μείς masc. ‘month’ (acc. μῆνα); cf. Dor. nom. sg. μης; the alternative Att. nom. sg. μήν is analogical after the oblique cases;

ῥ ς fem. ‘nose’ (acc. ῥῖνα).

Monosyllabic aorist participles also follow this rule, e.g.:

β ς aor. ptc. masc. ‘walk’ (acc. βάντα);

δούς aor. ptc. masc. ‘give’ (acc. δόντα);

στ ς aor. ptc. masc. ‘stand’ (acc. στάντα).

A parallel case is that of χείρ ‘hand’ with PG *-ers, the accentuation of which may also be regarded as regular:

χεἱρ fem. ‘hand’ (acc. χεῖρα); cf. Dor. χήρ; Aeol. χέρρας.

There are two counter-examples to the rule that final PG *- n(t)s yields acute tone:

εἶς num. masc. ‘one’ (acc. ἕνα); but cf. the acute of μηδ-είς, οὐδ-είς;

πᾶς pron. masc. ‘all’ (acc. πάντα).

The right explanation of the circumflex tone of these two words is perhaps the one proposed by Berger, according to whom the unexpected tone is connected with the semantics of the words, ‘one’ and ‘all’ often being used with emphasis.¹⁶

6 The rules governing the accentuation of the monosyllabic word-forms mentioned so far are rather unproblematic. The remaining cases may, in my view, be divided into two groups: those ending in two consonants before the simplification of *-ts to -ς (groups (a) and (b)), and those ending in fewer than two consonants (group (c)):

(a)PG *|bēks masc./fem. cough’ > βήξ (acc. βήχα)

PG *|grūps masc. ‘griffin’ > γρ ψ (acc. γρῦπα)

PG *|hrāks/*|hrōks fem. ‘grape’ > ῥ ξ (acc. ῥᾶγα)/ῥώξ (acc. ῥῶγα)

PG *|klōps masc. ‘thief’ > κλὡψ (acc. κλῶπα)

PG *|psīks masc. ‘crumb’ > ψίξ (acc. ψῖχα)

PG *|ptōks masc. ‘hare’ > πτώζ (acc. πτῶκα; but Att. πτῶξ¹⁷)

PG *|(ś)knīps masc. ‘kind of insect’ > (σ)κνίψ (acc. (σ)κνῖπα)

PG *|sphāks masc. ‘wasp’ > σφήξ (acc. σφῆκα; cf. Dor. σφαξ)

PG *|ihōps masc. ‘flatterer’ > ϑώψ (acc. ϑῶπα)

(b)PG *|krēts masc. ‘Cretan’ > Kρής (acc. Kρῆτα)

PG *|pōts masc. ‘foot’ > ποὑς (for *πώς;¹⁸ acc. πόδα)

PG *|phōts masc. ‘man’ > φώς (acc. φῶτα)

PG *|thēts masc. ‘serf’ > ϑής (acc. ϑῆτα)

(c)PG *|dō neut. ‘house’ > Hom., Hes. δῶ

PG *|drūs fem. ‘tree’, acc. *|drūn > δρῦς, δρῦν

PG *|gw us masc./fem. ‘ox’, acc. *|gwōn > βοῦς, βοῦν (for βῶν, actually found in Homer and in Doric);

PG *|krī neut. ‘barley’ > epic κρῖ

PG *|mūs masc. ‘mouse’, acc. *|mūn > μῦς, μῦν

PG *|skōr neut. ‘dung’ > σκῶρ (gen. σκατός)

PG *|(s)ūs masc./fem. ‘pig’, acc. *|(s)ūn > σῦς/ὗς, σῦν/ὗν

At a language stage where the simplification of *-ts to *-s had not yet taken place, the distribution of tones is obvious:

Monosyllabic words ending in two consonants in Proto-Greek have acute tone in Attic, while those ending in fewer than two consonants have circumflex tone.

This law, which was automatic in a pre-stage of Greek and only later gained phonological relevance, constitutes the basis for the distribution of tones in monosyllabic words.

We may alternatively assume that the simplification of *-ts to *-s had already taken place when the distribution of tones was established. In this case, the acute tone of the words belonging to group (b) above should be regarded as secondary like that of the words mentioned in § 8 below.

An interesting consequence of the distributional law is that the rule that monosyllabic neuters regularly have circumflex tone – which is often given a morphological explanation¹⁹ – has a simple phonological basis, as neuters never end in more than one consonant in Proto-Greek.

7 On the basis of the material hitherto reviewed, the prosodic development of which was accounted for as phonologically regular, we can deduce a very simple paradigmatic correlation that enables us – or, more relevantly, enabled a speaker of Greek – to predict the tone of the nominative singular on the basis of the structure of the accusative singular:

If the accusative is disyllabic, the nominative has acute tone (e.g. nom. ῥ ς, acc. ῥῖνα; or nom. κλώψ, acc. κλῶπα); if the accusative is monosyllabic, the nominative has circumflex tone (e.g. nom. μῦς, acc. μῦν; or nom. οἶς, acc. οἶν).

The historical reason behind this principle is the prosodic law established in § 6 above combined with the fact that the Proto-Indo-European accusative singular morpheme had a consonantal realization (PIE *-m > -ν) after a stem-final vowel and a vocalic realization (PIE *-m > -α) after a stem-final consonant.

Notice that the principle is valid for almost all nouns, including those with unclear etymologies and the remodelled ones mentioned in § 1 above. We find only rare exceptions like παῖς (acc. παῖδα), the accentuation of which is due to a late contraction, see § 4 above; and πᾶς (acc. πάντα), εῖς (acc. ἕνα), which were mentioned in § 5 above.

The fact that the principle was synchronically transparent until a relatively recent stage in the development of Greek provides the key to understanding the apparent counterexamples to the distribution of the acute and circumflex tones proposed here.

8 Since monosyllabic words ending in fewer than two consonants regularly have circumflex tone, the following word-forms must be regarded as phonologically irregular:

PG *|kār fem. ‘goddess of death’ → Kήρ (acc. Kῆρα)

PG *|khthōn fem. earth’ → χϑὡν (acc. χϑόνα)

PG *|kwhēr masc. ‘beast’ → ϑήρ (acc. ϑῆρα)

PG *|psēn masc. ‘gall-insect’ → ψήν h(acc. ψῆνα)

PG *|phōr masc. ‘thief’ → φὡρ (acc. φῶρα)

PG *|phrēn fem. ‘midriff’ → φρήν (acc. φρένα)

The substitution of the expected circumflex tone with an acute is easily understandable as an accommodation to the principle established in § 7 above. Perhaps the regular acute tone of the nom. sg. of polysyllabic n- and r-stems like ποιμήν, σωτήρ also contributed to this analogical change of a circumflex tone to an acute.

The acute tone of Zευς masc. ‘Zeus’, reflecting PG *|dj us from PIE *d ṷs, is unexpected. It was probably introduced analogically from polysyllabic words of the βασιλεὑς type.²⁰

9 To recapitulate the law of distribution of tones in terms of moras, in a pre-stage of Greek a monosyllabic word-form containing a long vowel or a diphthong had high pitch on the second mora of the syllabic nucleus if ending in two consonants, but on the first mora if ending in fewer than two consonants. Monosyllables containing a short vowel had high pitch on the only mora of the vowel. At this stage, the two tones were in complementary distribution in monosyllabic words:²¹

The phonologization of the acute and circumflex tones in monosyllabic words probably started when *-ts was simplified to *-s. The functional load of the tonal opposition was significantly increased when *-ns was simplified to -s with compensatory lengthening of the preceding vowel; high pitch fell on the second mora of the new vowel, resulting in an acute tone. Analogical developments of the type *ϑήρ → ϑήρ contributed to blur the original distribution. Contractions of the type κλη ς > κλείς constitute a relatively recent development.

10 As I have tried to make plausible in the preceding paragraphs, it is possible to establish a stage in the prehistory of Greek where there were no distinctive tones in monosyllabic word-forms. Words ending in two consonants appear in Attic Greek with acute tone, whereas words ending in fewer than two consonants have circumflex tone. At this stage the distribution of the two tones was automatic, determined by the number of final consonants; the acute and circumflex tones were in complementary distribution. The tones were phonologized when certain final consonant clusters were simplified.

From a morphological point of view the different development of the PIE accusative marker *-m to Greek -ν after vowels and -α after consonants, together with the proposed distribution of tones, led to a simple correlation between the number of syllables of the accusative and the tone of the nominative (disyllabic accusative = acute tone in nominative; monosyllabic accusative = circumflex tone in nominative). This correlation, which arose regularly in the bulk of the monosyllabic nouns, was analogically extended to the few existing exceptions.

1Unless otherwise specified, ‘Greek’ refers to Attic. – Syllables are considered long if they contain a long vowel or a diphthong; short syllables do not show tonal distinctions and are left out of consideration. – In words ending in -οι or -αι, tones are also distinctive in the penultimate, a problem which is without relevance to the one examined here. – I owe thanks to Jenny Helena Larsson, who kindly reviewed an early draft version of this paper, and to Benedicte Nielsen, who did a great job in commenting on a later version of the paper, as well as to Andrew Goodson and Coulter George for their useful comments.

2For practical reasons, the term ‘Proto-Greek’ (abbreviated PG) is not used in a strict sense but refers to a pre-stage of Greek where changes like the simplification of final *-ts to *-s had not taken place, e.g. in PG *pōts ‘foot’ although such changes may actually have taken place in a pre-dialectal stage.

3See Sihler (1995) 158.

4For the accentuation of ιχϑύς and όφρϋς, see LSJ s.w.: ‘ιχϑύς (so Hdn. Gr.2.936, -ῦς and -ύς freq. in codd.), ύος, ό […]*; ‘όφρῦς, ύος, ή […] ῦ in nom. and acc., which are accented όφρῦς, -ῦν by Hdn. Gr.2.937: the accentuation όφρύς, όφρύν may be admitted in late writers […]’.

5Schwyzer (1939) 378: ‘Es wurde also wohl einmal Länge im einsilbigen Worte rein phonetisch zirkumflektiert’; this view is apparently supported by Kümmel (2002) 67.

6See Berger (1953).

7See Kuryłowicz (1932), (1934), (1958) 126–30, (1968) 88–90.

8Garde (1976) 304–6, with notes on pp. 453–4.

9In δᾷϑα, as in all words with an accented long penultimate vowel and a short final vowel, the circumflex tone is automatic and thus of secondary linguistic importance.

10Chantraine (1948) 29; Hinge, on the other hand, mainly on the basis of the Doric material (nom. sg. παίς, acc. sg. παίδα) reconstructs ‘*/pa.(w)ís/’ with final accentuation (Hinge (2006) 125); according to this view, the tone of the Attic nom. sg. παῖς is analogical.

11Cf. Schwyzer (1939) 480, 574; LSJ s.v.; according to Chantraine (1948) 224, on the other hand, the common disyllabic scansion ‘s’explique sans doute comme un artifice métrique’; cf. Simkin (2004) 20 with n. 205.

12See LSJ s.v.

13Berger (1953) 8–9.

14The same outcome is found in the last syllable of polysyllabic words, e.g. PG o-stem acc. pl. *a|grons > ἀγρούς.

15See LSJ s.v.

16Berger (1953) 15–16; cf. Bally (1945) 22 (εἶς probably analogical with τρεῖς).

17See Schwyzer (1939) 377.

18The Attic form arose by an analogical replacement of the open vowel *[ɔ:] of *πως with the closed vowel [o:] (written <ου>) under influence of the closed quality of the short stem-vowel of ποδ-.

19E.g. Kuryłowicz (1968) 89: ‘Die einsilbigen Neutra haben den Zirkumflex auf langer Wurzelsilbe, da alle mehrsilbigen Neutra der 3. Deklination rezessiv betont werden: δῶ, Kῆρ, Kρῖ, πῦρ, σκῶρ. Auch hier handelt es sich natürlich um eine morphologische Erscheinung’; cf. Kuryłowicz (1949) 29; Kiparsky (1973) 798.

20Thus also Martinez García (1997); notice that a disyllabic Lindeman variant PIE *di ṷs would regularly yield acute tone in accordance with the rules proposed here (this possibility was kindly pointed out to me by Karl Praust).

21The acute accent mark (’) denotes high pitch.

Monophthong for expected υ-diphthong in Greek*

ADAM HYLLESTED

University of Copenhagen

PAUL S. COHEN

Yorktown Heights, NY

1 Rix (1970)¹ has shown that a PIE laryngeal followed by a syllabic liquid or nasal yields a Greek sequence of vowel + liquid/nasal, with the vowel quality determined by the identity of the laryngeal – a development that has come to be called Lex Rix. Peters (1980) in his detailed examination of all Greek lexical items having potential reflexes of PIE laryngeals, comes to the conclusion, inter alia, that while PIE initial *Hi- yields ἰ-, initial *Hu-, in a perfect parallel to the situation with liquids and nasals, yields the same result in Greek as its full-grade counterpart *Heṷ-, namely Vὐ-, with the quality of the vowel being determined by the identity of the initial laryngeal. We can refer to this posited development conveniently as Extended Lex Rix. Peters discusses a total of eight items (or item families) that in his analysis indicate, with reasonable reliability, PIE initial *Hu-. In a summary table (p. 72), he lists seven of them as following the rule² and just one – ὑφαίνω weave’ – as not doing so. In an attempt to explain the apparently anomalous behaviour of ὑφαίνω (p. 125), he invokes the possibilities of lexical diffusion and of some (yet unknown) conditioning factors that were no longer operative.

Our focus in the present paper will be on offering a rule-governed phonological explanation of why the vocalism of ὐφαίvω is different and also on the discussion of other relevant examples.

2 Extended Lex Rix for *Hu- is widely (though not universally) accepted. Thus, e.g., LIV (658) at the entry for *ṷebh- ‘to wind, weave’ explicitly rejects the Hittite verb ḫuppiye- ‘interlace, intertwine, mix together’ (and implicitly, therefore, Hitt. ḫuwapp-‘interlace, entangle, ensnare, commingle’³ and ḫuppala- ‘net’ as well) as a cognate of G ὑφαίνω (and, inter alia, Ved. ubhn ti ‘bind, chain’, OHG weban ‘weave, plait’, Toch. A wpantär ‘weave’) in part because of the Hitt, -pp- (i.e., the Doppelschreibung taken by most authorities as pointing to a voiceless stop), but chiefly because the initial ḫ- in Hittite shows the presence of an *h2- or *h3- that, in order to accord with Extended Lex Rix, would demand G †αὐφαίνω or †οὐφαίνω respectively.⁴ Thus, we find in LIV (loc. cit.) ‘Aorist ?*ṷebh-/ṷbh- heth. wepta webte’, which is footnoted as follows: ‘Vgl. Neu, HS 111 (1998) 59¹⁷; der Wurzelstamm ist wegen des Nasalpräsens vermutlich als Aorist zu bestimmen.’ The relevant portion of Neu reads

Wie das eingangs behandelte Textfragment 731/t der Boğazköy-Grabungskampagne 1961 wiederum anschaulich zeigt, lassen sich auch aus kleinen Tafelbruchstücken immer wieder wichtige Informationen gewinnen. So könnte mit Blick auf das hethitische Lexikon in der Figura etymologica ú-e-pu-uš (Akkus. Plur.) ú-e-ep-ta aus dem noch unveröffentlichten Textfragment Bo 96/98 Vs. 9’, das eine Abschrift aus der zweiten Hälfte des 13. Jahrhunderts v. Chr. bietet, die idg. Wurzel *ṷebh- weben, flechten, knüpfen (IEW I 1114) vorliegen, was die wörtliche Übersetzung Webstücke webte er/sie ergäbe. Hethitisch ṷeb- wäre dann ein altes Wurzelverb vergleichbar ṷaš kaufen, erwerben. Die hier gegebene Interpretation von hethitisch ṷeb- möge vorerst noch unter Vorbehalt stehen.

We can benefit from a look at the context, which has been published in the meantime:

KBo 42.6 Vs.?

(8) […-]zi-mi-is ÍD-as ar-ru-ma-ar e-ep-t[a …]

(9) […-]x-ni ú-e-pu-us ú-e-ep-ta nu=mu TUG-an=mi-i[t…]

‘[someone] took the washing of the river [or in the rivers],

[someone] ṷep-ed ṷepa-s and […-ed] my cloth for me’.

The context is quite fragmentary, and on the basis of this passage alone, we cannot be sure of the meaning. The word ‘cloth’ is indeed mentioned in the sentence after the ṷep-sentence, but would that be enough to determine that ṷep- means ‘to weave’? To Alwin Kloekhorst (p.c.), it seems like a typically unconvincing case of an interpretation based solely on a supposed etymological connection.

3 In general, the PIE root for ‘weave’ has been, in consequence of Extended Lex Rix, reconstructed as *ṷebh-⁵ without initial laryngeal, notwithstanding the derivation in Peters (1980) and the existence of the seemingly related Myc. fut. midd. pple. e-we-pe-se-so-me-na ‘(intended) to be woven/to be used for textiles’ (a hapax), the initial vowel of which points to a PIE laryngeal.

There is, however, long-standing and widespread acceptance of an additional PIE root for ‘weave’: it occurs both in a basic form *Heṷ- (vel sim.) > Ved. váyati ‘weave’, ótu- ‘woof’, Skt. uta- ‘woven’ (and perhaps Alb. ven weave*), and with a *-dh-extension, as found in Lith. áudžiu, áusti ‘weave’ and OE wǣd garment, cloth, rope’, ON váð ‘fabric, weaving’. Watkins (2000a) and LIV each have both roots for ‘weave’, but do not choose either to connect or to contrast them. Mallory & Adams (1997) 572, however, give both *h2/3eu-⁶ ‘weave’ and *h2/3ṷebh- ‘weave’, and explicitly connect them; thus we see under *h2/3eu-, ‘This is the most basic term for weave that is reconstructable for PIE. It has largely been supplanted, within PIE itself, by the enlarged *h2/3ṷebh- of the next entry’; and under *h2/3ṷebh-, ‘This enlargement of *h2/3eu- would appear to have been the usual word for "weave’ in later PIE …’. Moreover, under *h2/3ṷebh-, along with Albanian, Germanic, Indic, Iranian and Tocharian reflexes, we find: ‘Grk ὑφαίνω, Hit huppai- entangle, ensnare; interlace, ḫuppala net (though -pp-, reflecting PIE *-p- rather than *-bh- is not well explained⁷),