On the accentual status of vocatives in proto-Indo-European

On the accentual status of vocatives in proto-Indo-European Dieter Gunkel (University of Richmond) [email protected] 29th Annual UCLA Indo-European Conference, November 3rd, 2017 1. How were vocatives accented in proto-Indo-European (PIE)? 2. A relatively widespread view — one that that I have held and will challenge here: vocatives were accented on the initial syllable (barytone) regardless of the accentuation of the other case forms; barytonesis arose via an accent retraction process. For example, Fortson 2010 on athematic nouns: “In athematic animate nouns, the [singular –DG] vocative ending was zero and there was retraction of the accent. What this meant for all practical intents and purposes is that the vocative was the same as the nominative without the *-s or without lengthened grade: Gk. nomin. pólis ‘city’ but voc. póli ‘(o) city’; Ved. nomin. hastī́ ‘having a hand’ (< *hastīń ), voc. hástin. Vedic Sanskrit best preserves the accent retraction.” (§6.7) “The vocative plural was the same as the nominative, but probably with accent retraction.” (§6.14) “To judge by the Ved. vocative dévās ‘gods!’ [vs. nomin. devā́ s, –DG], the vocative plural had accent retraction like the singular.” (§6.52) on thematic nouns: “There is complete agreement on the reconstruction of the animate nominative, vocative, accusative, and neuter nominative-accusative singular. Retraction of the accent in the vocative is evidenced by such forms as Ved. vīŕ -a (vs. nomin. vīrás) and Gk. ádelph-e (vs. nomin. adelphós)...” (§6.47) and on ā-stems: “The vocative had retracted accentuation (preserved in Sanskrit and in a few words in Greek) and ended in *-a (whence the -a of Greek and Umbrian and the -o of OCS; elsewhere it was remade using the nominative ending). This *-a probably resulted from *-ah2 with loss of the laryngeal before a pause.” (§6.70) 3. This retraction process is sometimes connected with the accent retraction that accompanies substantivization (Paradebeispiele: λευκή ‘white’ : λεύκη ‘white skin condition’, λευκός ‘bright, white’ : λεῦκος name of a white(?) fish), as in the following considerations from Höfler 2017: 123: “Über die Genese des Wortbildungsmechanismus der Akzentrückziehung kann freilich nur spekuliert werden. Es könnte durchaus eine Verbindung mit der Barytonese des grundsprachlichen Vokativs zusammenhängen, die durch die vedische und griechische Evidenz ja quasi erwiesen wird.” In some cases, at least, the accent retraction was probably not part of an active, synchronic conversion of Adj to N, but a side-effect of the passive, diachronic demorphologization likewise reflected by the non-compositional semantics (cf. Probert 2006: 233, Garrett 2011, Lundquist 2015: 64, Yates 2015: 167). 4. In essence, this view goes back to Benfey (1845: 907). The basic comparison is between synchronically irregular barytone vocatives in Greek and synchronically regular barytone accentuation of phrase-initial Vedic vocatives (vs. other case forms, represented here by the nominative) with. 1 BARYTONE VOC Greek: pítar Vedic: πάτερ VS OXYTONE NOM, ACC, ETC VS pitā́ , pitáram ... pitṛ ́ ṣu VS πατήρ, πατέρα ... πατράσι 5. Barytone (=recessively accented) vocatives to non-barytone stems appear to be lexicalized archaisms in Greek (cf. Probert 2006: 146 n. 29). They are unproductive, synchronically unmotivated, and had high token frequencies in everday speech. For a recent discussion of frequency effects, see Sandell 2015: 73ff. et passim. On forms of address in Classical Greek, see Dickey 2010. Semantically, they fall into three groups. 6. Vocatives used to address family members and members of the household: ἄδελφε ‘brother’ : ἀδελφός, ἀδελφοῦ [Attic vs. regular ἀδελφέ in the koiné] ἄνερ ‘husband’ : ἀνήρ, ἀνδρός ‘man, husband’ γύναι ‘woman, wife’ : γυνή, γυναικός δᾶερ ‘brother-in-law’ : δᾱήρ, δᾱέρος πάτερ ‘father’ : πατήρ, πατρός θύγατερ ‘daughter’ : θυγάτηρ, θυγατρός δέσποτα ‘master’ : δεσπότης ... δεσπόται [contrast στρατιῶτα]. 7. Vocatives used to address/invoke gods and swear: Ζεῦ ‘Zeus’ : Ζεύς, Διός [also Ζεῦ πάτερ, Ζεῦ βασιλεῦ] σῶτερ ‘savior’ : σωτήρ, σωτῆρος [also Ζεῦ.. σῶτερ] The fact that the e-vowel of σῶτερ is also unpredictable/irregular provides additional evidence that this vocative, at least, is lexicalized (contra Steriade 1988: 281, 2014: 26). 8. Vocatives used “expressively” to convey insults, sympathy, pity, and the like (cf. Dickey 2010: 330f.), to judge by the accentuation of the following duo in the Aristophanic ms. tradition (cf. Probert 2006: 263): µόχθηρε ‘wretched’ : µοχθηρός πόνηρε ‘base’ : πονηρός. There are good reasons to reconstruct a stage of Attic with βδέλυρε : βδελυρός ‘disgusting’, µῶρε : µωρός ‘dull, stupid’, and πῆρε : πηρός ‘blind, disabled’ (cf. Probert 2006: 146 n. 29). 9. Note that other frequent, everyday Greek vocatives such as µῆτερ ‘mother’, ἄνθρωπε ‘man’, τέκνον ‘child’, oὗτος ‘you there’ are also barytone, but they are predictably so, since the other case forms are barytone as well: µήτηρ, ἄνθρωπος, τέκνον, oὗτος. 10. Like Greek, Slavic preserves irregular barytone vocatives, e.g. the Serbo-Croatian word for ‘woman’, where the accentual contrast between the sg. voc. žȅna (short falling initial) and nom. žèna, gen. žèni, acc. žènu (short rising) goes back to a contrast between barytone and oxytone accentuation (cf. Weiss 2011: 201 with refs), which is reminiscent of oxytone nom. γυνή vs. barytone voc. γύναι. 2 11. This has a couple of potentially weird consequences. 12. Byrd (2015: 31) points out a phonological peculiarity with the reconstruction based on Vedic phraseinitial pítar and Greek πάτερ: “aside from barytone vocatives of the type *pə́ h2ter ‘o father’ (Gk. πάτερ), there are no securely attested cases of accented vocalized laryngeals.” 13. Fortson (2010: §6.44) notes a morphophonological peculiarity: “There are no certain accent–ablaut distinctions within thematic paradigms except in the vocative, where the accent was retracted.” (§6.44) 14. What have we done? We have identified likely archaisms in Greek and Slavic, found synchronically regular — but phrase-initial — barytone comparanda in Vedic (πάτερ : pítar), and reconstructed barytonesis for PIE (*pə́ h2ter). But what about phrase-internal unaccented vocatives in Vedic? Let’s revisit the accentual grammar of Vedic. 15. In Vedic, all phrases begin with an accented word: with one exception (1.2.8b), all ca. 40,000 pādas of the Rigveda begin with an accented word. 16. Barytonesis of vocatives is phrasally conditioned: vocatives are unaccented unless they are phraseinitial. 17. The accentuation of finite verbs is partly phrasally conditioned as well. In main clauses, verbs are unaccented unless they are phrase-initial (or contrastive). 18. There is an important contrast between phrase-initial vocatives and finite verbs. 19. The verbs display their lexical accent — their accentuation is phonologically unpredictable. 20. Verb, non-initial in RC, displaying the lexical accent: gṛṇánti ‘they sing’ yé tvā gṛṇánti váhnayaḥ (1.48.11d) ‘who sing to you (as) conveyors (of the oblation)’ 21. Verb, non-initial in MC, unaccented: gṛṇanti víprā ukthébhiḥ kaváyo gṛṇanti (3.34.7d) ‘the inspired ones, the kavis, sing (them) with hymns’ 22. Verb, initial in MC, accented as in RC, displaying the lexical accent: gṛṇánti — not *gṛ ́ ṇanti gṛṇánti vipra te dhíyaḥ (1.14.2b) ‘they sing your thoughts, o inspired one’ 23. Vocatives are accented on the initial syllable, i.e. in a phonologically predictable way. 24. The ablative devébhyo displays the lexical accent of devá-, and the non-initial vocative deva is unaccented. devébhyo deva dāśúṣe (1.74.9c) ‘(win ...) from the gods, o god, for the pious man’ 3 25. The initial vocative is accented on the first syllable déva. The lexical accent does not surface (in the same position it appears in non-vocative case forms such as dat. pl. devébhiyo). déva devébhiyo havíḥ (1.3.11b) ‘(release,) o god, the oblation for the gods’ 26. The process that locates the accent on phrase-initial vocatives has access to the vowels in the initial syllables of Lindeman variants. Recall that monosyllabic case forms of dyáu- ‘heaven’ have disyllabic variants diyáu-. Note the contrast between the initial lexical accent of the nominative singular and that of the vocative singular (cf. Whitney 1869). nominative: dyáuṣ, diyáuṣ vocative: dyáuṣ, díyauṣ 27. Nom. sg. Accent on diphthong: dyáuṣ. dyáur me pitā́ janitā́ nā́ bhir átra (1.164.33, 11σ) ‘My father, my progenitor, is Heaven; here is my navel.’ (J&B) 28. Nom. sg. Lindeman variant. Accent on diphthong: diyáuṣ. diyáuṣ pitā́ janitā́ satyám ukṣan (4.1.10c, 11σ) ‘Father Heaven was the begetter – they sprinkled it real’ (following J&B) 29. Voc. sg. Accent on diphthong: dyáuṣ. dyáur dehí lokáṃ vájrāya viṣkábhe (8.100.12b, 11σ) ‘Heaven, grant a place for the mace to prop apart.’ (J&B) 30. Voc. sg. Lindeman variant. Accent on initial vowel: díyauṣ. díyàuṣ pítaḥ pṛ ́ thivi mā́ tar ádhrug (6.51.5a, 11σ) O Heaven, Father, o Earth, Mother, lie-less one. 31. We are free to view accent assignment to vocatives as a phrasal phonological process. Under a compositional approach to Vedic accentuation (Yates, tomorrow), the “Basic Accentuation Principle” (BAP) assigns initial stress to the nominative dyáuṣ but it does not assign an initial accent to nominative *díyauṣ (rather diyáuṣ). We might conclude that the initial vowel of Lindeman variants arises in a post-lexical (or post-BAP) domain. Note that if Lindeman variants arose in Vedic via phrasal phonology, we would have to view vocative accent assignment as phrasal phonology. As things stand (Barber 2013: 55–64, Byrd 2015: 204–206), the conditioning of Lindeman variants is unclear. 32. I suggest interpreting this in the following way. Verbs are lexically accented; their accent is suppressed in main clauses. Vocatives are lexically unaccented. There is a requirement in Vedic that the phrase begin with an accented word. We can capture this by positing a phrase-initial high tone. (H Similar suggestions have been made, e.g. “... the accentuation of a verb in sentence-initial position is an inheritance from Proto-IndoEuropean, where all sentence-initial words were accented. Accentuation in this position seems to have possessed a demarcative value ...” (Klein 1992: 2; cf. also 86–87) 4 33. If the first word is lexically accented (as it usually is), the tone “docks” on the lexically accented syllable. H ( agním 34. This holds for finite verbs in main clauses, since it is more important for the high tone to dock on the first word than for the verbal accent to be suppressed. H ( gṛṇánti 35. If the first word is not lexically accented — and this applies to vocatives — the tone “docks” on the phrase-initial syllable, which may be the first vowel of a disyllabic variant of the word H ( diyauṣ or the first vowel of the monosyllabic variant. H ( diyauṣ 36. The notation of the phrasal high tone in Vedic is identical to the notation of the lexically accented syllable, whence phrase initial dyáuṣ, díyauṣ, déva, etc. The phonetic realization may have been similar or identical as well. The grammar will be learned correctly as long as speakers are paying attention to the contrast between phrase-initial vocatives vs. non-initial vocatives. 37. There are typological parallels for this type of grammar. Like Vedic, Tokyo Japanese and Northern Bizkaian Basque have lexically accented and lexically unaccented words; if the accentual phrase begins with a lexically unaccented word, that word hosts a phrase-initial rise (LH) (cf Gussenhoven 2004: 170–208). Vedic would differ in having a phrase-initial high (H) as opposed to a rise (LH). 38. On this analysis, other lexically unaccented words should host the phrasal accent in the same way the vocatives do. However, other lexically unaccented words are (true) enclitics (that require a preceding phonological word to host them). Two things perhaps worth thinking more about. 39. Consider the pronominal forms asya, asyá, and ásya described in Jamison 2002: “these forms have three distinct accentual states, which correspond two three different syntactic values (1) unaccented (asya): non-initial anaphoric pronoun (‘of him/it’). [quite common] (2) accented on final (asyá): demonstrative adjective (i.e. modifying a substantive) ‘of this X’ can be initial or not. [quite common] (3) accented on initial (ásya): emphatic (hic et nunc) demonstrative, either adjectival or pronominal (‘of this X here, of this one here’); always initial. [quite rare]” 40. Banti (1984: 68) attempts to relate barytone initial ásya and unaccented non-initial asya to the vocatives, arguing that they “ ... are basically forms of a single anaphoric pronoun á-, which is restricted to the non-direct cases and bears no accent when it occurs within a verse or a sentence, but assumes an accent on its first syllable like the vocatives when it occurs at the beginning of a verse or a sentence.” 5 It is unclear whether this can be squared with the functional difference between ásya and asya. 41. Recall the relative independence of the two members of dvandva compounds such as mitrā́ -váruṇā ‘Mitra and Varuṇa’ dyā́ vā-pṛthivī́ ‘Heaven and Earth’. 42. They have two accents, two inflectional endings, and the two members can straddle the caesura in trimeter verse, e.g. RV 1.71.9c rā́ jānā mitrā́ (!)váruṇā supāṇī́ the two kings, Mitra and Varuṇa, good-handed, which is something the two members of standard compounds cannot do. 43. Since there are two accents, an analysis on which vocatives undergo retraction of the lexical accent /devá/ → déva might predict phrase-initial vocatives such as *mítrāváruṇā *dyā́ vāpṛ ́ thivī, where the retraction applies to each word-like member of the compound separately. Such are not attested, though. Vocative dvandvas only have one accent, e.g. mítrāvaruṇā. Note that this would remove half of the contradiction between the single and double wordhood discussed by Kiparsky 2010, the other being the retroflexion in agnīṣ́ ómā. 44. Can the Greek and Slavic fact derive from the grammar posited for Vedic? In short, yes. As far as I know, the Slavic word for ‘woman’ can have gone from an unaccented PIE form to an “enclinomenon.” We would assume that speakers did not attend to the distinction between phrase-initial and noninitial position, but assumed that the accent they heard in, e.g. mā́ ter was lexical, as — I suggest — we have been doing for Vedic phrase-initial vocatives. 45. Note what happens if we posit that the phrasal accent was (mis)learned as lexical after the distinction between acute and circumflex arose in Greek via the contraction of an accented and unaccented vowel after laryngeal loss (e.g. dat. sg. PIE *bhoráh2ai > phoráai > phorâːi, φορῆι). H ( zdeu with a relatively early pitch peak might be reinterpreted as Ζεῦ (vs. nom. Ζεύς), whence perhaps (Ζεῦ) βασιλεῦ, ἱππεῦ, etc. 46. Finally, if we posit that vocatives were unaccented in PIE, except in phrase-initial position, where they hosted a phrasal high tone, there is no (vocatival) reason to posit accentual mobility in thematic paradigms (see §13), nor is there a (vocatival) reason to posit an accented schwa primum, since we predict that the phrasal accent would have “docked” on the first accentable vowel in PIE H ( p(ə)h2ter 6 as still in Vedic, where the first vowel happens to have developed from a schwa primum H ( pitar and in pre-Greek, where the first vowel also happens to have developed from a schwa primum H ( pater References Banti, Giorgio. 1984. On the Morphology of Vedic Gender-Distinguishing Pronominals. Rome: Dipartimento di studi glottoantropologici universita “La Sapienza.” Barber, Peter J. 2013. Sievers’ Law and the History of Semivowel Syllabicity in Indo-European and Ancient Greek. Oxford: Oxford University Press. Benfey, Theodor. 1845. Review of Böhtlingk, Otto, “Ein erster Versuch über den Accent im Sanskrit” (Mémoires de l’Académie impériale des Sciences de St. Petersbourg, vi. Série, 1845). Hallesche Allgemeine Literatur-Zeitung 114: 897ff. Byrd, Andrew Miles. 2015. The Indo-European Syllable. Leiden: Brill. Dickey, Eleanor. 2010. Forms of Address and Markers of Status. In A Companion to the Ancient Greek Language, ed. Egbert J. Bakker, 327–337. Leiden: Brill. Fortson, Benjamin W. IV. 2010. Indo-European Language and Culture: An Introduction (Blackwell Textbooks in Linguistics 19). 2nd ed. Malden, MA: Wiley-Blackwell. Garrett, Andrew. 2011. Verner’s Law Nominal Doublets: Bidirectional Leveling or Accent Shift? Paper presented at the 30th East Coast Indo-European Conference, Harvard University. Gussenhoven, Carlos. 2004. The Phonology of Tone and Intonation. Cambridge: Cambridge University Press. Höfler, Stefan. 2017. Der Stier, der Stärke hat. Possessive Adjektive und ihre Substantivierung im Indogermanischen. Ph.D. diss., University of Vienna. Jamison, Stephanie W. 2002. Rigvedic sīm and īm. In Indian Linguistic Studies: Festschrift in Honor of George Cardona, ed. Madhav Deshpande and Peter E. Hook, 290ff. Delhi: Motilal Banarsidass. Kim, Ronald I. 2002. The Continuation of Proto-Indo-European Lexical Accent in Ancient Greek: Preservation and Reanalysis. University of Pennsylvania Working Papers in Linguistics 7.2: 59–93. Kiparsky, Paul. 2010. Dvandvas, Blocking, and the Associative: The Bumpy Ride from Phrase to Word. Language 86.2:302–331. Klein, Jared S. 1992. On Verbal Accentuation in the Rigveda (American Oriental Society Essay 11). New Haven: American Oriental Society. Lundquist, Jesse. 2015. On the Accentuation of Vedic -ti-Abstracts. Indo-European Linguistics 3:42–72. 7 Probert, Philomen. 2003. A New Short Guide to the Accentuation of Ancient Greek. London: Bristol Classical Press. Probert, Philomen. 2006. Ancient Greek Accentuation: Synchronic Patterns, Frequency Effects, and Prehistory. Oxford: Oxford University Press. Sandell, Ryan. 2015. Productivity in Historical Linguistics: Computational Perspectives on WordFormation in Ancient Greek and Sanskrit. Ph.D. dissertation, University of California, Los Angeles. Steriade, Donca. 1988. Greek Accent: A Case for Preserving Structure. Linguistic Inquiry 19.2:271–314. Steriade, Donca. 2014. A Synchronic Analysis of A[ncient] Greek Accent. Paper presented at the Harvard GSAS Colloquium. Wackernagel, Jakob. 1877. Der griechische Verbalaccent. Zeitrschrift für vergleichende Sprachforschung 23:457–470. Weiss, Michael. 2011. Outline of the Historical and Comparative Grammar of Latin. 2nd ed. Ann Arbor: Beech Stave. Whitney, William Dwight. 1869. On the Nature and Designation of the Accent in Sanskrit. Transactions of the American Philological Association 1:20–45. Whitney, William Dwight. 1889. A Sanskrit Grammar, Including Both the Classical Language, and the Older Dialects, of Veda and Brahmana. 2nd ed. Leipzig. Yates, Anthony. 2015. Anatolian Default Accentuation and its Diachronic Consequences. IndoEuropean Linguistics 3:145–187. Yates, Anthony. 2017. Lexical Accent in Cupeño, Hittite, and Indo-European. Ph.D. diss., University of California, Los Angeles. 8