The diachrony of "voice reversal" in Hittite

The diachrony of “voice reversal” in Hittite Anthony D. Yates University of California, Los Angeles adyates@ucla.edu 12th International Congress of Hittitology Istanbul 5 September 2023 Slides available at: www.adyates.com/research/ 1 / 34 Two voice alternations in Old Hittite (1) “Voice reversal” with ACTIVUM TANTUM pai– ‘go’: a. kinun=a natta kuwāpikki pāun now= TOP NEG anywhere go:1 SG . PST. ACT ‘But recently I haven’t gone anywhere.’ Ï (KBo 17.1 iv 13) (in sense of Grestenberger 2018), i.e., syntactically unaccusative verbs that exhibit active inflectional endings in basic stem forms like (1a). . . ACTIVA TANTUM 2 / 34 Two voice alternations in Old Hittite (1) “Voice reversal” with ACTIVUM TANTUM pai– ‘go’: 1 a. kinun=a natta kuwāpikki pāun now= TOP NEG anywhere go:1 SG . PST. ACT ‘But recently I haven’t gone anywhere.’ (KBo 17.1 iv 13) b. karū=ma [ŠÀ? ] É DUMU.MEŠ-an paišgahat ˘ :1 SG . PST. MID formerly= TOP inside house children:GEN . PL go:IPFV ‘I used to go to the children’s quarters.’ (KBo 17.1 iv 12–13) Ï . . . consistently undergo a “voice reversal” in marked imperfective forms (characterized by –ške/a–), instead taking middle inflectional endings as in (1b) (Melchert 2017b; Yates and Gluckman 2020). 2 1 Unaccusative already in OH (see Yates 2022). 2 Building on Watkins 1969:72, Neu 1968:86–9. 3 / 34 Two voice alternations in Old Hittite (2) “Voice reversal” with DEPONENT huett(i)– ‘pull, drag’: ˘ a. kinun=a 1 UDU LU-naš kāššaš=(š)aš huittiyanta now= TOP 1 sheep man:GEN . SG in.place.of=his ˘draw:3 PL . NPST. MID ‘But now in place of the man they shall drag in one sheep.’ (KBo 6.26 i 41) Ï DEPONENTS (in sense of Grestenberger 2014, 2016, 2018), i.e., syntactically transitive verbs that exhibit middle inflectional endings in basic stem forms like (2) . . . 4 / 34 Two voice alternations in Old Hittite (2) “Voice reversal” with DEPONENT huett(i)– ‘pull, drag’: ˘ a. kinun=a 1 UDU LU-naš kāššaš=(š)aš huittiyanta now= TOP 1 sheep man:GEN . SG in.place.of=his ˘draw:3 PL . NPST. MID ‘But now in place of the man they shall drag in one sheep.’ (KBo 6.26 i 41) b. kētt=a kētt=a GI-an huttiyannai this.side=CONJ this.side=CONJ arrow:ACC . SG . C ˘draw:IPFV:3 SG . NPST. ACT tarnai=m=an natta release:3 SG . NPST. ACT = TOP =3 SG . ACC . C NEG ‘He keeps drawing his arrow toward this side and that, but he does not let it go.’ (KBo 17.43 i 10–11) Ï . . . consistently undergo a “voice reversal” in marked imperfective forms (characterized by –anna/i–), instead taking active inflectional endings as in (2b) (Yates and Gluckman 2020). 5 / 34 When and why? (3) Voice reversal in Old Hittite: BASIC STEM IMPERFECTIVE ACTIVA TANTUM ACT → MID DEPONENT MID → ACT 6 / 34 When and why? (3) Voice reversal in Old Hittite: BASIC STEM IMPERFECTIVE ACTIVA TANTUM ACT → MID DEPONENT MID → ACT Ï Two principal questions addressed here: 6 / 34 When and why? (3) Voice reversal in Old Hittite: BASIC STEM IMPERFECTIVE ACTIVA TANTUM ACT → MID DEPONENT MID → ACT Ï Two principal questions addressed here: ◦ When did voice reversal develop historically? 6 / 34 When and why? (3) Voice reversal in Old Hittite: BASIC STEM IMPERFECTIVE ACTIVA TANTUM ACT → MID DEPONENT MID → ACT Ï Two principal questions addressed here: ◦ When did voice reversal develop historically? ◦ Why did voice reversal develop? 6 / 34 When and why? (3) Voice reversal in Old Hittite: BASIC STEM IMPERFECTIVE ACTIVA TANTUM ACT → MID DEPONENT MID → ACT Ï Two principal questions addressed here: ◦ When did voice reversal develop historically? ◦ Why did voice reversal develop? Ï On the demise of voice reversal after Old Hittite see Appendix I. 6 / 34 Roadmap I §1 Introduction §2 Old Hittite voice reversal as inheritance? Ï Voice reversal in Tocharian? Ï Absence of voice reversal in Hittite lexicalized imperfectives §3 Emergence of Old Hittite voice reversal §4 Voice reversal as an Anatolian phenomenon 7 / 34 Voice reversal as inheritance? (3) Voice reversal in Old Hittite: BASIC STEM IMPERFECTIVE ACTIVA TANTUM ACT → MID DEPONENT MID → ACT Ï Melchert (2017b:482–4) tentatively proposes that ACTIVA TANTUM suffixed with *–sk“e/o– underwent voice reversal (ACT → MID) already in Proto-Indo-European (PIE). ⇒ Voice reversal in Old Hittite ACTIVA TANTUM is inherited. 8 / 34 Voice reversal as inheritance? (3) Voice reversal in Old Hittite: BASIC STEM IMPERFECTIVE ACTIVA TANTUM ACT → MID DEPONENT MID → ACT < PIE Ï Melchert (2017b:482–4) tentatively proposes that ACTIVA TANTUM suffixed with *–sk“e/o– underwent voice reversal (ACT → MID) already in Proto-Indo-European (PIE). ⇒ Voice reversal in Old Hittite ACTIVA TANTUM is inherited. Ï Melchert adduces comparative support in Tocharian. 8 / 34 Voice reversal as inheritance? (4) a. b. c. d. e. f. TB TA/B TA/B TA/B TA TA/B skente mäsk– musk– wāsk– yutk– sätk– ‘they are’ ‘be(come)’ (Prs III) ‘disappear’ (Prs III) ‘move’ (Prs XII) ‘become agitated’ (Prs III) ‘spread (intr.)’ (Prs III) < *h1 s-(s)k“e⁄o – < *mn-sk“e⁄o – ˚ < *m(y)uhx -sk“e⁄o – < *ugh -sk“e⁄o – < *hx yudh -sk“e⁄o – < *(h2 )sut-sk“e⁄o – g. h. TA TB kātk– rätk– ‘arise’ (Prs VII) ‘(a)rise’ (Prs VII) < *gh ad-sk“e⁄o – < *h3 rihx -T-sk“e⁄o – Ï In (4) are Tocharian verbs that (i) have unaccusative semantics and (ii) are derived with *–sk“e⁄o – from PIE roots with primarily ACT forms. 1 ✓ (4a–f) exhibit only MID inflection. 1 TA/B verbal classes in (4) after Malzahn 2010. 9 / 34 Voice reversal as inheritance? (4) a. b. c. d. e. f. TB TA/B TA/B TA/B TA TA/B skente mäsk– musk– wāsk– yutk– sätk– ‘they are’ ‘be(come)’ (Prs III) ‘disappear’ (Prs III) ‘move’ (Prs XII) ‘become agitated’ (Prs III) ‘spread (intr.)’ (Prs III) < *h1 s-(s)k“e⁄o – < *mn-sk“e⁄o – ˚ < *m(y)uhx -sk“e⁄o – < *ugh -sk“e⁄o – < *hx yudh -sk“e⁄o – < *(h2 )sut-sk“e⁄o – g. h. TA TB kātk– rätk– ‘arise’ (Prs VII) ‘(a)rise’ (Prs VII) < *gh ad-sk“e⁄o – < *h3 rihx -T-sk“e⁄o – Ï In (4) are Tocharian verbs that (i) have unaccusative semantics and (ii) are derived with *–sk“e⁄o – from PIE roots with primarily ACT forms. 1 ✓ (4a–f) exhibit only MID inflection. ✘ (4g–h) exhibit only ACT inflection. 1 TA/B verbal classes in (4) after Malzahn 2010. 9 / 34 Voice reversal as inheritance? Ï Thus two weaknesses of Melchert’s (2017b) hypothesis: i. Tocharian support for reconstructing voice reversal in ACTIVA TANTUM is limited and mixed (“less than compelling” per Melchert 2017b:484). ii. Does not account for similar pattern of voice reversal in Hittite deponents. 10 / 34 Voice reversal as inheritance? Ï Thus two weaknesses of Melchert’s (2017b) hypothesis: i. Tocharian support for reconstructing voice reversal in ACTIVA TANTUM is limited and mixed (“less than compelling” per Melchert 2017b:484). ii. Does not account for similar pattern of voice reversal in Hittite deponents. Ï A third issue (↓): iii. Does not account for active inflection in lexicalized imperfectives of Hittite ACTIVA TANTUM. 10 / 34 Probing the diachrony of voice reversal in Hittite (5) a. iyanna/i– ‘set out (for); go’ b. ēške/a– ‘be/exist’ (< PIE *h1 ei– ‘go’ + –anna/i–) (< PIE *h1 es– ‘be’ + –ške–) Ï Two exceptional verbs in (5) may offer insight into the diachronic development of voice reversal in Hittite. 11 / 34 Probing the diachrony of voice reversal in Hittite (5) a. iyanna/i– ‘set out (for); go’ b. ēške/a– ‘be/exist’ (< PIE *h1 ei– ‘go’ + –anna/i–) (< PIE *h1 es– ‘be’ + –ške–) Ï Two exceptional verbs in (5) may offer insight into the diachronic development of voice reversal in Hittite. Ï Crucial shared properties of these verbs: Ï Are lexicalized derivatives of ACTIVA TANTUM roots, historically formed with imperfective suffixes. Ï Show only active inflection (viz., no voice reversal). 11 / 34 Hitt. iyanna/i– as active lexicalized imperfective (6) LUGAL-š=a IŠME š=aš iyanniš king:NOM . SG=TOP heard CONN =3 SG . NOM . C go:3 SG . NPST. ACT ‘The king heard (about it), then he set out.’ (KBo 22.2 rev. 7) Ï iyanna/i– ‘set out (for); go’ is attested already in Old Script with consistent active inflection — e.g., (6). 12 / 34 Hitt. iyanna/i– as active lexicalized imperfective (7) n=aš=za EGIR-pa parna=šša iyanniš CONN =3 SG . NOM = REFL back house:ALL=his: ALL go:3 SG . NPST. ACT ‘He then went back to his house.’ (KUB 24.8 ii 10–11; OH/NS) Ï Semantic bleaching — frequently ‘go’ as in (7) beside inceptive ‘set out’ in (6) — is indicative of lexicalization. 1 Ï Consistent with standard analysis as: ✓ Historical imperfective of *h1 ei– ‘go’ (cf. IMP Hitt. it/itten ‘go/y’all go!’) 2 ✘ Synchronic imperfective of unaccusative media tantum iya– ‘walk’ 1 2 “Lexicalized as a separate verb ‘to go’” per Hoffner and Melchert (2008:322). cf. HED 2: 326–8, HW 2 I: 1–4; Kloekhorst 2008:375–6 13 / 34 Hitt. ēške/a– as active lexicalized imperfective (8) MUNUS (dupl. ē [šd]u) NAPTIR〈TI=KA〉=ma=tta ēškanzi āra ˙ secondary.wife=your= TOP =2 SG . DAT rightfully be-3 PL . NPST. ACT ‘Rightfully they shall be yours as secondary wife.’ (KBo 5.3 iii 63–65 rev. 48–49; MH/NS) Ï ēške/a– is hapax in (8), first identified by Melchert (1984:31 n. 12) as: Ï –ške/a-suffixed form of copula Hitt. eš/aš– ‘be’ (< PIE *h1 es–). Ï Substantive (viz., existential) verb in a possessive construction. Ï Identification confirmed by duplicate (KBo 22.40+19.44 rev. 48–49). 1 Ï Exhibits active inflection. 1 Thus Beckman 1999:32, Cohen 2002:19, Klinger 2005:111; see further Melchert 2017c. 14 / 34 Hitt. ēške/a– as active lexicalized imperfective (8) MUNUS (dupl. ē [šd]u) NAPTIR〈TI=KA〉=ma=tta ēškanzi āra ˙ secondary.wife=your= TOP =2 SG . DAT rightfully be-3 PL . NPST. ACT ‘Rightfully they shall be yours as secondary wife.’ (KBo 5.3 iii 63–65 rev. 48–49; MH/NS) Ï Despite late(r) attestation ēškanzi has hallmarks of an archaism: Ï Combination of copula and –ške/a– may be synchronically restricted, given that high frequency verb and massively productive suffix do not otherwise cooccur. Ï Exhibits irregular root stress/strong allomorphy: ēške/a– ([é:sk:e/a-]) for expected x aškē/a– ([as-k:é:/á-]) (see Appendix II). 15 / 34 Hitt. ēške/a– as active lexicalized imperfective (8) MUNUS (dupl. ē [šd]u) NAPTIR〈TI=KA〉=ma=tta ēškanzi āra ˙ secondary.wife=your= TOP =2 SG . DAT rightfully be-3 PL . NPST. ACT ‘Rightfully they shall be yours as secondary wife.’ (KBo 5.3 iii 63–65 rev. 48–49; MH/NS) Ï Despite late(r) attestation ēškanzi has hallmarks of an archaism: Ï Combination of copula and –ške/a– may be synchronically restricted, given that high frequency verb and massively productive suffix do not otherwise cooccur. Ï Exhibits irregular root stress/strong allomorphy: ēške/a– ([é:sk:e/a-]) for expected x aškē/a– ([as-k:é:/á-]) (see Appendix II). Ï Formally and functionally matches cognate formations in early stages of other ancient IE languages. 15 / 34 Hitt. ēške/a– in comparative perspective (9) tis enth áde mántis a. éske be:3 SG . IPFC . ACT INDF there seer:NOM . SG . M ‘There was a certain seer there. . . ’ (Hom. Od. 9.508) Kaph eùs wanássōn AUG :be:3 SG . IPFC . ACT INDF K: NOM . SG . M rule: PTCP. PRS . ACT: NOM . SG . M ‘There was a certain Cepheus ruling. . . ’ (Alcm. 74 P.) b. é¯ske tis Ï Cognate of Hitt. ēške/a– is attested in early Greek – e.g., (9): 1 Ï Similarly functions as a substantive verb. Ï Reflects same irregular strong root allomorphy (< PIE *h1 és-(s)k“e⁄o –). Ï Exhibits same active inflection. 1 See Keller 1985, Watkins 1993:477, Zerdin 1999:309–16. 16 / 34 Hitt. ēške/a– in comparative perspective (10) cui suus heres nec escit 3 SG . DAT own:NOM . SG . M heir:NOM . SG . M NEG be:3 SG . PRS . ACT ‘He who has no heir of his own’ (Twelve Tables, V.4; Crawford 1996:580) Ï Cognate of Hitt. ēške/a– is attested in (Very) Old Latin — e.g., (10). 1 Ï Nearly identical usage, substantive verb in a possessive construction. Ï Reflects same irregular strong root allomorphy (< PIE *h1 és-(s)k“e⁄o –). Ï Exhibits same active inflection. 1 See Fraenkel 1925, Keller 1992:79–84, Haverling 2000:143–4, 395, Pezzini 2015:242–3, Weiss 2020:432. 17 / 34 Reconstructing Hitt. ēške/a– (11) a. PIE *h1 és-(s)k“e-t(i) > OLat. escit, Gk. éske > b. PIE *h1 és-(s)k“o-nti > Hitt. ēškanzi ([é:skantsi]), Lat. escunt Ï These archaic IE word equations suggest that Hitt. ēškanzi was inherited from PIE as in (11): 1 Ï In substantive usage. Ï With irregular root stress/strong allomorphy. Ï With active inflection. 1 For root full-grade cf. Weiss 2020:432 (but Hittite supports root stress, others non-probative), pace Hackstein 1995:299, LIV 2 : 241–2. 18 / 34 Reconstructing Hitt. ēške/a– (11) a. PIE *h1 és-(s)k“e-t(i) > OLat. escit, Gk. éske > b. PIE *h1 és-(s)k“o-nti > Hitt. ēškanzi ([é:skantsi]), Lat. escunt Ï These archaic IE word equations suggest that Hitt. ēškanzi was inherited from PIE as in (11): 1 Ï In substantive usage. Ï With irregular root stress/strong allomorphy. Ï With active inflection. Ï Reconstruction of (11) would be direct counter-evidence to Melchert’s (2017b) hypothesis of inherited voice reversal in *–sk“e⁄o -suffixed ACTIVA. Ï More on TB skente — middle with root zero-grade — in §3 below. Ï On Pal. iška see Appendix III. 1 For root full-grade cf. Weiss 2020:432 (but Hittite supports root stress, others non-probative), pace Hackstein 1995:299, LIV 2 : 241–2. 18 / 34 Active inflection in lexicalized imperfectives as archaism (5) a. iyanna/i– ‘set out (for); go’ b. ēške/a– ‘be/exist’ (< PIE *h1 ei– ‘go’ + –anna/i–) (< PIE *h1 es– ‘be’ + –ške–) Ï Active inflection of (11) — vs. middle in imperfectives of all other Old Hittite ACTIVA TANTUM — in any case awaits a satisfactory explanation. 19 / 34 Active inflection in lexicalized imperfectives as archaism (5) a. iyanna/i– ‘set out (for); go’ b. ēške/a– ‘be/exist’ (< PIE *h1 ei– ‘go’ + –anna/i–) (< PIE *h1 es– ‘be’ + –ške–) Ï Active inflection of (11) — vs. middle in imperfectives of all other Old Hittite ACTIVA TANTUM — in any case awaits a satisfactory explanation. ⋆ Proposal: Old Hittite voice reversal is an innovation vis-à-vis PIE. Ï (11) were lexicalized prior to development of voice reversal, thus retain active inflection as an archaism. Ï But applies systematically in newly created or renewed forms, thus to productively derived imperfectives of ACTIVA TANTUM (and deponents). 19 / 34 Active inflection in lexicalized imperfectives as archaism (5) a. iyanna/i– ‘set out (for); go’ b. ēške/a– ‘be/exist’ (< PIE *h1 ei– ‘go’ + –anna/i–) (< PIE *h1 es– ‘be’ + –ške–) Ï Active inflection of (11) — vs. middle in imperfectives of all other Old Hittite ACTIVA TANTUM — in any case awaits a satisfactory explanation. ⋆ Proposal: Old Hittite voice reversal is an innovation vis-à-vis PIE. Ï (11) were lexicalized prior to development of voice reversal, thus retain active inflection as an archaism. Ï But applies systematically in newly created or renewed forms, thus to productively derived imperfectives of ACTIVA TANTUM (and deponents). Ï More on the precise chronology of Hittite voice reversal in §4 below. 19 / 34 Roadmap II §1 Introduction §2 Old Hittite voice reversal as inheritance? §3 Emergence of Old Hittite voice reversal Ï Morphosyntax of voice in Hittite and IE Ï Motivating voice reversal Ï Emergence of (limited) voice reversal in IE §4 Voice reversal as an Anatolian phenomenon 20 / 34 Morphosyntax of voice in Hittite and IE (12) SYNTAX a. TR c. UN CLASS BASIC STEM S NON - DEPONENT appanzi ‘take’ C MEDIA TANTUM eša ‘sit’ AN AC Ï In Old Hittite — like in older IE languages and cross-linguistically — voice morphology and syntactic function generally align (“match”): 1 Ï Transitive, agentive verbs regularly exhibit active inflection in active syntactic contexts. Ï Unaccusative verbs regularly exhibit middle inflection (media tantum). 1 See especially Grestenberger 2014:19–62, 102–5, 2018:489–91 with references. 2 On DEPONENTS see Grestenberger 2014, 2016, 2018, 2019; on ACTIVA TANTUM Yates and Gluckman 2020 (cf. Weisser 2014; Rothstein-Dowden 2022 on “active deponency” in PIE). 21 / 34 Morphosyntax of voice in Hittite and IE (12) SYNTAX a. b. c. d. TR U AN NA S CC CLASS BASIC STEM NON - DEPONENT appanzi ‘take’ huittiyanta ‘drag’ DEPONENT MEDIA TANTUM ACTIVA TANTUM ˘ eša pāun ‘sit’ ‘go’ Ï In Old Hittite — like in older IE languages and cross-linguistically — voice morphology and syntactic function generally align (“match”): 1 Ï Transitive, agentive verbs regularly exhibit active inflection in active syntactic contexts. Ï Unaccusative verbs regularly exhibit middle inflection (media tantum). Ï DEPONENTS and ACTIVA TANTUM are mismatch verbs. 2 Ï DEPONENTS are transitive but exhibit middle inflection. Ï ACTIVA TANTUM are unaccusative but exhibit active inflection. 1 See especially Grestenberger 2014:19–62, 102–5, 2018:489–91 with references. 2 On DEPONENTS see Grestenberger 2014, 2016, 2018, 2019; on ACTIVA TANTUM Yates and Gluckman 2020 (cf. Weisser 2014; Rothstein-Dowden 2022 on “active deponency” in PIE). 21 / 34 Motivating voice reversal (12) SYNTAX a. b. c. d. TR U A NS C NA CLASS BASIC STEM NON - DEPONENT appanzi = appiskanzi huittiyanta 6= huttiyannai DEPONENT C MEDIA TANTUM ACTIVA TANTUM ˘ eša pāun IMPERFECTIVE ˘ = ēškedumat 6 = paišgahat ˘ ‘take’ ‘drag’ ‘sit’ ‘go’ Ï Two key observations (Yates and Gluckman 2020): Ï Only mismatch verbs undergo voice reversal in imperfective forms. Ï In such forms voice morphology and syntax are realigned (i.e., no mismatch). 22 / 34 Motivating voice reversal (12) SYNTAX a. b. c. d. TR U A NS C NA CLASS BASIC STEM NON - DEPONENT appanzi = appiskanzi huittiyanta 6= huttiyannai DEPONENT C MEDIA TANTUM ACTIVA TANTUM ˘ eša pāun IMPERFECTIVE ˘ = ēškedumat 6 = paišgahat ˘ ‘take’ ‘drag’ ‘sit’ ‘go’ Ï Two key observations (Yates and Gluckman 2020): Ï Only mismatch verbs undergo voice reversal in imperfective forms. Ï In such forms voice morphology and syntax are realigned (i.e., no mismatch). ⇒ Voice reversal is the emergence of syntactically expected voice morphology. 22 / 34 Motivating voice reversal (12) SYNTAX a. b. c. d. TR U A NS C NA CLASS BASIC STEM NON - DEPONENT appanzi = appiskanzi huittiyanta 6= huttiyannai DEPONENT C MEDIA TANTUM ACTIVA TANTUM ˘ eša pāun IMPERFECTIVE ˘ = ēškedumat 6 = paišgahat ˘ ‘take’ ‘drag’ ‘sit’ ‘go’ Ï Two key observations (Yates and Gluckman 2020): Ï Only mismatch verbs undergo voice reversal in imperfective forms. Ï In such forms voice morphology and syntax are realigned (i.e., no mismatch). ⇒ Voice reversal is the emergence of syntactically expected voice morphology. Ï No mismatches in imperfectives! 22 / 34 Motivating voice reversal — two proposals ◦ Why did voice reversal develop (in Old Hittite)? ⋆ General: Voice reversal is motivated by a (soft) cognitive bias against mismatches between voice morphology and syntax(/semantics) — and thus emerges diachronically. 1 Ï Mismatch (voice) morphology can be acquired (and thus be stable diachronically), especially in highly frequent words. Ï But learners are disposed to produce forms in which voice morphology and syntax are aligned, especially in words that are productively derived (and thus novel). 2 1 See, e.g., Sheehan et al. 2018 for a bias against case/agreement mismatches. 2 Learned as diacritic marking on (Hittite) verbal roots/stems per Yates and Gluckman (2020). 23 / 34 Motivating voice reversal — two proposals ◦ Why did voice reversal develop (in Old Hittite)? ⋆ General: Voice reversal is motivated by a (soft) cognitive bias against mismatches between voice morphology and syntax(/semantics) — and thus emerges diachronically. 1 Ï Mismatch (voice) morphology can be acquired (and thus be stable diachronically), especially in highly frequent words. Ï But learners are disposed to produce forms in which voice morphology and syntax are aligned, especially in words that are productively derived (and thus novel). 2 ⋆ Specific: Development of voice reversal in productively derived OH imperfectives is the grammaticalization of this emergent tendency. 1 See, e.g., Sheehan et al. 2018 for a bias against case/agreement mismatches. 2 Learned as diacritic marking on (Hittite) verbal roots/stems per Yates and Gluckman (2020). 23 / 34 Emergent voice reversal in IE — Tocharian (4) a. b. c. d. e. f. TB TA/B TA/B TA/B TA TA/B skente mäsk– musk– wāsk– yutk– sätk– ‘they are’ ‘be(come)’ (Prs III) ‘disappear’ (Prs III) ‘move’ (Prs XII) ‘become agitated’ (Prs III) ‘spread (intr.)’ (Prs III) < *h1 s-(s)k“e⁄o – < *mn-sk“e⁄o – ˚ < *m(y)uhx -sk“e⁄o – < *ugh -sk“e⁄o – < *hx yudh -sk“e⁄o – < *(h2 )sut-sk“e⁄o – Ï Expected on this analysis is that voice reversal will recur, emerging independently in IE languages that preserve inherited voice system. Ï Limited historical voice reversal would explain Tocharian verbs in (4): Ï (4b–f) lack IE cognates, thus likely inner-Tocharian formations. Ï Middle inflection emerges in novel derivatives of inherited ACTIVA “e⁄o –, aligning voice and syntax. TANTUM with productive –*sk 1 2 1 Thus LIV 2 , s.vv. 2 “ e⁄o – in prehistory of Tocharian see Malzahn 2010:460–1 with references. On productive –*sk 24 / 34 Emergent voice reversal in IE — Tocharian (4) a. b. c. d. e. f. TB TA/B TA/B TA/B TA TA/B skente mäsk– musk– wāsk– yutk– sätk– ‘they are’ ‘be(come)’ (Prs III) ‘disappear’ (Prs III) ‘move’ (Prs XII) ‘become agitated’ (Prs III) ‘spread (intr.)’ (Prs III) < *h1 s-(s)k“e⁄o – < *mn-sk“e⁄o – ˚ < *m(y)uhx -sk“e⁄o – < *ugh -sk“e⁄o – < *hx yudh -sk“e⁄o – < *(h2 )sut-sk“e⁄o – Ï Expected on this analysis is that voice reversal will recur, emerging independently in IE languages that preserve inherited voice system. Ï Limited historical voice reversal would explain Tocharian verbs in (4): Ï (4a) is inner-Tocharian renewal of inherited ACT *h1 és-(s)k“e-(ti) in (11), continued in early Latin, Greek, and Hitt. ēškanzi. Ï Both irregular root stress/full-grade and voice mismatch are eliminated in innovative formation. 25 / 34 Emergent voice reversal in IE — Ancient Greek (13) a. b. c. d. e. f. AGk. AGk. AGk. AGk. AGk. AGk. PRS . ACT baínō th éō ph eúgō eimì bióō páskh ō FUT. MID bé¯somai th eúsomai ph eúksomai és(s)omai bió¯somai peísomai ‘walk; go’ ‘run’ ‘flee’ ‘be’ ‘live’ ‘suffer’ Ï Greek verbs with prototypical unaccusative semantics often show “semi-deponent” pattern in (13): 1 Ï Active inflection in present stem. Ï Middle inflection in future with suffix –s e⁄o –. 1 See Rijksbaron 2007:156–7 for a list and alternative historical account (cf. Grestenberger 2019), but Fortson 2016 for a critique thereof (cf. Weiss 2020:446 n. 12). 26 / 34 Emergent voice reversal in IE — Ancient Greek (13) a. b. c. d. e. f. AGk. AGk. AGk. AGk. AGk. AGk. PRS . ACT baínō th éō ph eúgō eimì bióō páskh ō FUT. MID bé¯somai th eúsomai ph eúksomai és(s)omai bió¯somai peísomai ‘walk; go’ ‘run’ ‘flee’ ‘be’ ‘live’ ‘suffer’ Ï Greek verbs with prototypical unaccusative semantics often show “semi-deponent” pattern in (13): 1 Ï Active inflection in present stem. Ï Middle inflection in future with suffix –s e⁄o –. Ï Pattern in (13) attributable to limited historical voice reversal: Ï Middle inflection emerges in innovative forms derived with productive future suffix –s e⁄o – (< desiderative PIE *–h1 s e⁄o –). 1 See Rijksbaron 2007:156–7 for a list and alternative historical account (cf. Grestenberger 2019), but Fortson 2016 for a critique thereof (cf. Weiss 2020:446 n. 12). 26 / 34 Emergent voice reversal in IE deponents (14) “Activization” of DEPONENTS in IE: a. b. c. d. e. OH/OS OH/OS OH/OS OLat. OIr. huettianta ˘tuhša ˘ parašh a ˘ interpretari ·cuirethar >> >> >> >> >> OH/MS OH/MS MH/NS LLat. MIr. huettianzi ˘tuhhušzi ˘˘ paršiyami interpretāre cuirid ‘they drag’ ‘cuts off’ ‘I break’ ‘to explain’ ‘places’ Ï Well-established tendency across IE for middle inflection of DEPONENTS to be renewed by active inflection diachronically. 1 Ï Within Hittite — e.g., (14a–c). Ï Within Latin — e.g., (14d). Ï Within Irish — e.g., (14e). 1 On the tendency see Inglese 2020:216–8 with references; it is not a rule (see Grestenberger 2014:119–20 with references). 27 / 34 Emergent voice reversal in IE deponents (14) “Activization” of DEPONENTS in IE: a. b. c. d. e. OH/OS OH/OS OH/OS OLat. OIr. huettianta ˘tuhša ˘ parašh a ˘ interpretari ·cuirethar >> >> >> >> >> OH/MS OH/MS MH/NS LLat. MIr. huettianzi ˘tuhhušzi ˘˘ paršiyami interpretāre cuirid ‘they drag’ ‘cuts off’ ‘I break’ ‘to explain’ ‘places’ Ï Well-established tendency across IE for middle inflection of DEPONENTS to be renewed by active inflection diachronically. 1 Ï Within Hittite — e.g., (14a–c). Ï Within Latin — e.g., (14d). Ï Within Irish — e.g., (14e). ⇒ Historical voice reversal, eliminating mismatch between voice morphology and syntax. 1 On the tendency see Inglese 2020:216–8 with references; it is not a rule (see Grestenberger 2014:119–20 with references). 27 / 34 Roadmap III §1 Introduction §2 Old Hittite voice reversal as inheritance? §3 Emergence of Old Hittite voice reversal §4 Voice reversal as an Anatolian phenomenon Ï Voice reversal in Luwian and its chronological implications Ï Diachrony of voice reversal in Anatolian 28 / 34 Chronology of voice reversal ◦ When exactly did Old Hittite voice reversal develop? ⋆ Voice reversal is an innovation vis-à-vis PIE (§3). 29 / 34 Chronology of voice reversal ◦ When exactly did Old Hittite voice reversal develop? ⋆ Voice reversal is an innovation vis-à-vis PIE (§3). Ï Luwian evidence could allow for refining this chronology (↓). 29 / 34 Voice reversal in Luwian (15) “Voice reversal” with ACTIVUM TANTUM HLuw. hwiya– ‘move’: ˘ a. *a=wa/i=m=a-tà PRAE-na (PES2 )HWI/A-ya-ta CONN = QUOT =1 SG . ACC =3 PL . NOM . C before move:3 PL . PST. ACT ‘They (= Tarhunza, Karhuha, and Kubaba) proceeded before me.’ (KARKAMIŠ A11b+c §11) Ï HLuw. hwiya– ‘move’ (= Hitt. huw(a)i– ‘id.’) is an ACTIVUM TANTUM — ˘ ˘ inflection in basic stem forms like i.e., unaccusative with only active (15a). . . 30 / 34 Voice reversal in Luwian (15) “Voice reversal” with ACTIVUM TANTUM HLuw. hwiya– ‘move’: ˘ a. *a=wa/i=m=a-tà PRAE-na (PES2 )HWI/A-ya-ta CONN = QUOT =1 SG . ACC =3 PL . NOM . C before move:3 PL . PST. ACT ‘They (= Tarhunza, Karhuha, and Kubaba) proceeded before me.’ (KARKAMIŠ A11b+c §11) b. | PRAE-pa=wa/i||=mu | za-a-sa | EXERCITUS-la/i/u-na-si-is before= TOP = QUOT =1 SG . DAT this:NOM . SG . C of.army:NOM . SG . C (DEUS)TONITRUS-sa | hu-ha-sà-ta-si RED :move: IPFV:3 SG . PST. MID StG:NOM . SG . C ‘This Tarhunza of the Army proceeded before me.’ (TELL AHMAR 6 §7) Ï . . . but in (reduplicated) marked imperfective forms (characterized by –sa–) it consistently switches to middle inflection, e.g., (15b). 1 1 Likewise in TELL AHMAR 6 §19, KARKAMIŠ 11b+c §8, and KARKAMIŠ A12 §2. 31 / 34 Refining the chronology of voice reversal ◦ When exactly did Old Hittite voice reversal develop? Ï Luwian evidence for voice reversal is thus slender, but no (?) apparent counter-evidence. 32 / 34 Refining the chronology of voice reversal ◦ When exactly did Old Hittite voice reversal develop? Ï Luwian evidence for voice reversal is thus slender, but no (?) apparent counter-evidence. Ï Two possibilities: Ï Voice reversal developed (as an optional process) in Proto-Anatolian (PA), then grammaticalized as obligatory in prehistory of Hittite. Ï (Limited) voice reversal is an independent innovation in Luwian. 32 / 34 Diachrony of voice reversal Ï Voice reversal was not a PIE feature, but developed independently in Anatolian and to a more limited extent in other IE branches. Ï Recurrence is motivated by learners’ dispreference for mismatches between voice and syntax. 1 cf. Melchert 2017a on the allative case. 33 / 34 Diachrony of voice reversal Ï Voice reversal was not a PIE feature, but developed independently in Anatolian and to a more limited extent in other IE branches. Ï Recurrence is motivated by learners’ dispreference for mismatches between voice and syntax. Ï Voice reversal developed in PA or in prehistories of Hittite and Luwian. Ï Grammaticalized as an obligatory process in Old Hittite. 1 cf. Melchert 2017a on the allative case. 33 / 34 Diachrony of voice reversal Ï Voice reversal was not a PIE feature, but developed independently in Anatolian and to a more limited extent in other IE branches. Ï Recurrence is motivated by learners’ dispreference for mismatches between voice and syntax. Ï Voice reversal developed in PA or in prehistories of Hittite and Luwian. Ï Grammaticalized as an obligatory process in Old Hittite. Ï Voice reversal in Hittite was ultimately a failed innovation. 1 Ï Breaks down in post-Old Hittite, likely due to broader changes in the function of the middle morphology (see Appendix I). 1 cf. Melchert 2017a on the allative case. 33 / 34 Thank you! • Special thanks to the members of the: · Indo-European & Modern Linguistic Theory research group · UCLA PIES Graduate Seminar • And to audiences at the: · 3rd Munich-UCLA Historical Linguistics Colloquium · 93rd Annual Meeting of the Linguistic Society of America • As well as to: · Craig Melchert, Brent Vine, Stephanie Jamison, John Gluckman, Dieter Gunkel, Ron Kim, Sergio Neri, and Olav Hackstein. 34 / 34 References I Beckman, Gary M. 1999. Hittite Diplomatic Texts, 2 edn. Atlanta, GA: Scholars Press. Cohen, Yoram. 2002. Taboos and Prohibitions in Hittite Society. Heidelberg: Winter. Crawford, Michael H. (ed.). 1996. Roman Statues. London: Institute of Classical Studies, University of London. Fortson, Benjamin W. 2016. The Greek future deponents. Paper presented at the 35th Annual East Coast Indo-European Conference, Athens, GA, 6–8 June 2016. Fraenkel, Ernst. 1925. Zum Texte Römischer Juristen. Hermes 60(415–443). Friedrich, Johannes, Annelies Kammenhuber, and Inge Hoffmann. 1975–. Hethitisches Wörterbuch, 2 edn. Heidelberg: Winter. Grestenberger, Laura. 2014. Feature Mismatch: Deponency in Indo-European Languages. Ph.D. diss., Harvard University. References II ———. 2016. Reconstructing Proto-Indo-European Deponents. Indo-European Linguistics 4.98–149. ———. 2018. Deponency in finite and non-finite contexts. Language 94(3).487–526. ———. 2019. Deponency in morphology. In Oxford Research Encyclopedia in Linguistics. Oxford / New York: Oxford University Press. Hackstein, Olav. 1995. Untersuchungen zu den sigmatischen Präsensstammbildungen des Tocharischen. Göttingen: Vandenhoeck & Ruprecht. Hackstein, Olav, Jared Miller, Elisabeth Rieken, and Ilya Yakubovich (eds.). 2014–. The Digital Philological-Etymological Dictionary of the Minor Anatolian Corpus Languages. https://www.ediana.gwi.uni-muenchen.de/. References III Haverling, Gerd. 2000. On –sco Verbs, Prefixes and Semantic Functions: A Study in the Development of Prefixed and Unprefixed Verbs from Early to Late Latin. Göteborg: Acta Universitatis Gothoburgensis. Hoffner, Harry A., and H. Craig Melchert. 2008. A Grammar of the Hittite Language. Vol. I: Reference Grammar. Winona Lake, IN: Eisenbrauns. Inglese, Guglielmo. 2020. The Hittite Middle Voice: Synchrony, Diachrony, Typology. Leiden / Boston: Brill. Keller, Madeleine. 1985. Latin escit, escunt at-il des correspondants? Revue de Philologie 59.27–44. ———. 1992. Les verbes latins à infectum in –sc–, étude morphologique (à partir des formations attestées dès l’époque préclassique). Brussels: Latomus. References IV Klinger, Jörg. 2005. Der Vertrag Šuppiluliumas I mit Hukkana von Hajaša. ˘ ˘ In Bernd Jankowsky and Gernot Wilhelm (eds.), Staatsvertrage, Herrscherinschriften, und andere Dokumente zur politschen Geschichte, 107–112. Gütersloh: Gütersloher Verlagshaus. Kloekhorst, Alwin. 2008. Etymological Dictionary of the Hittite Inherited Lexicon. Leiden / Boston: Brill. Malzahn, Melanie. 2010. The Tocharian Verbal System. Leiden ; Boston: Brill. Melchert, H. Craig. 1984. Notes on Palaic. Zeitschrift für vergleichende Sprachforschung 97(1).22–43. ———. 2017a. An Allative Case in Proto-Indo-European? In Bjarne S.S. Hansen, Adam Hyllested, Anders R. Jörgensen, Guus Kroonen, Jenny H. Larsson, Benedicte N. Whitehead, Thomas Olander and Tobias M. Söborg (eds.), Usque ad Radices: Indo-European studies in honour of Birgit Anne Olsen, 527–539. Copenhagen: Museum Tusculanum Press. References V ———. 2017b. Mediopassives in *-sk“e/o- to Active Intransitives. In Ivo Hajnal, Daniel Kölligan and Katharina Zipser (eds.), Miscellanea Indogermanica. Festschrift für José Luis García Ramón zum 65. Geburtstag, 477–486. Innsbruck: Institut für Sprachen und Literaturen der Universität Innsbruck. ———. 2017c. A PIE *h1 (e)sk“e/o– as Substantive Verb Revisited. Paper presented at the 36th Annual East Coast Indo-European Conference, Ithaca, 4 June 2017. Neu, Erich. 1968. Das hethitische Mediopassiv und seine indogermanischen Grundlagen. Wiesbaden: Harrassowitz. Pezzini, Giuseppe. 2015. Terence and the Verb ‘To Be’ in Latin. Oxford / New York: Oxford University Press. Puhvel, Jaan. 1984–. Hittite Etymological Dictionary. Berlin / New York: Mouton. References VI Rijksbaron, Albert. 2007. The Syntax and Semantics of the Verb in Classical Greek. Leiden / Boston: Brill. Rix, Helmut, and Martin J. Kümmel (eds.). 2001. Lexikon der indogermanischen Verben: Die Wurzeln und ihre Primärstammbildungen, 2 edn. Wiesbaden: Reichert. Rothstein-Dowden, Zachary. 2022. Dental-aspirate presents in Greek and Indo-European. Ph.D. diss., Harvard University. Sheehan, Michelle, Albertyna Paciorek, and John N. Williams. 2018. Case/agreement matching: Evidence for a cognitive bias. Glossa 3(1).92/1–23. Watkins, Calvert. 1969. Geschichte der indogermanischen Verbalflexion (Indogermanische Grammatik: III/1, Formenlehre). Heidelberg: Winter. References VII ———. 1993. Some Anatolian Words and Forms. In Gerhard Meiser (ed.), Indogermanica et Italica: Festschrift für Helmut Rix zum 65. Geburtstag, 469–478. Innsbruck: Innsbrucker Beiträge zur Sprachwissenschaft. Weiss, Michael. 2020. Outline of the Historical and Comparative Grammar of Latin, 2 edn. Ann Arbor / New York: Beech Stave Press. Weisser, Phillip. 2014. Mismatch verbs: A unified account of unaccusatives and deponents. In Franz Rainer, Francesco Gardani, Hans C. Luschützky and Wolfgang U. Dressler (eds.), Morphology and Meaning: Selected Papers from the 15th International Morphology Meeting, Vienna, February 2012. Amsterdam / Philadelphia: John Benjamins. Yakubovich, Ilya. 2006. Were Hittite Kings Divinely Anointed? A Palaic Invocation for Hittite Religion. Journal of Ancient Near Eastern Religions 5.107–137. References VIII Yates, Anthony D. 2022. Motion verbs in Anatolian and Indo-European. Invited lecture, University of California, Los Angeles (Slides available at: http://www.adyates.com/research/). Yates, Anthony D., and John Gluckman. 2020. Voice Reversals and Syntactic Structure: Evidence from Hittite. Glossa 5(120).120/1–39. Zerdin, Jason. 1999. Studies in the Ancient Greek Verbs in -SKŌ. Ph.D. diss., University of Oxford. The demise of voice reversal (A1) Voice reversal in post-Old Hittite: BASIC STEM ACTIVA TANTUM DEPONENT IMPERFECTIVE ACT → ✘) ACT (✘ MID → ACT Ï Voice reversal in DEPONENTS disappears as a side effect of their diachronic renewal with ACT morphology in basic stem forms (cf. §3). 1/8 The demise of voice reversal (A1) Voice reversal in post-Old Hittite: BASIC STEM ACTIVA TANTUM DEPONENT IMPERFECTIVE ACT → ✘) ACT (✘ MID ✘) ACT (✘ MID → ACT Ï Voice reversal in DEPONENTS disappears as a side effect of their diachronic renewal with ACT morphology in basic stem forms (cf. §3). Ï Demise of voice reversal in ACTIVA TANTUM — whence new imperfectives with ACT inflection — requires an explanation. 1/8 The demise of voice reversal (A1) Voice reversal in post-Old Hittite: BASIC STEM ACTIVA TANTUM DEPONENT IMPERFECTIVE ACT → ✘) ACT (✘ MID ✘) ACT (✘ MID → ACT Ï Voice reversal in DEPONENTS disappears as a side effect of their diachronic renewal with ACT morphology in basic stem forms (cf. §3). Ï Demise of voice reversal in ACTIVA TANTUM — whence new imperfectives with ACT inflection — requires an explanation. Ï Core of the solution (Melchert 2017b:482): “This peculiar pattern broke down in later Hittite as part of the general elimination of mediopassive inflection for any function other than the passive.” 1/8 The demise of voice reversal (A1) Voice reversal in post-Old Hittite: BASIC STEM ACTIVA TANTUM DEPONENT IMPERFECTIVE ACT → ✘) ACT (✘ MID ✘) ACT (✘ MID → ACT ⋆ Proposal: Post-OH systemic change in voice morphosyntax, such that: Ï Unaccusative syntax is no longer realized by middle but active. Ï Only passive is realized by middle inflection. Ï Consequences of the change: Ï New imperfectives to historical ACTIVA TANTUM receive regular ACTIVE inflection. Ï Unaccusative historical media tantum retain middle inflection but as an exceptional lexical property. 2/8 Prosody of Hittite imperfectives (A2) a. b. c. PIE *prk“-sk“é-ti ˚ PIE *gw m-sk“é ˚ PIE *kw r-sk“é-ti ˚ > Ved. prcháti ‘asks’, Lat. poscit ‘demands’ ˚ > Ved. gácha ‘go!’, Gk. báske ‘come!’ > > Hitt. kuwarškezzi ([kw ar-sk:é:-tsi]) ‘cuts’ d. Hitt. /ekw -sk:é-si/ → akkuškēši ([akw :-usk:é:-si]) ‘you drink’ > e. Hitt. /et-sk:é-∅/ → azzikkı̄ ([at-sik:ı́:) ‘eat!’ > > f. Hitt. /ep:-sk:é-ántsi/ → appiškanzi ([ap:-isk:á-ntsi]) ‘they take’ Ï PIE *–sk“e/o-stems regularly had zero-grade of the root and suffixal stress (cf. LIV 2 : 209–10, 490–1) — e.g., (A2a–c). Ï This pattern stably continued in Hittite, where productive imperfectives (with –ške/a-) of e/a-ablauting verbs regularly exhibit weak root allomorphy and suffixal stress — e.g., (A2d–f ). 3/8 Prosody of Hittite imperfectives (A2) a. b. c. PIE *prk“-sk“é-ti ˚ PIE *gw m-sk“é ˚ PIE *kw r-sk“é-ti ˚ > Ved. prcháti ‘asks’, Lat. poscit ‘demands’ ˚ > Ved. gácha ‘go!’, Gk. báske ‘come!’ > > Hitt. kuwarškezzi ([kw ar-sk:é:-tsi]) ‘cuts’ d. Hitt. /ekw -sk:é-si/ → akkuškēši ([akw :-usk:é:-si]) ‘you drink’ > e. Hitt. /et-sk:é-∅/ → azzikkı̄ ([at-sik:ı́:) ‘eat!’ > > f. Hitt. /ep:-sk:é-ántsi/ → appiškanzi ([ap:-isk:á-ntsi]) ‘they take’ Ï Hitt. ēške/a– with root stress/strong allomorphy is thus exceptional from synchronic and diachronic perspectives. Ï Inner-Hittite Neubildung would be x aškē/a– ([as-k:é:/á-]), like (A2d–f). Ï Historically irregular full-grade is the lectio difficilor — and matched by early Latin and Greek cognates. 1 ⇒ Clear case for reconstructing PIE *h1 és-(s)k“e⁄o –. 1 Contra LIV 2 : 242 n. 10. 4/8 On Palaic iška (A3) KUB 35.165 obv. 21–22 + KUB 32.17 7´–8´: 21 22 [(nu-ku)] pa-aš-hu-ul-la-ša-aš ti-[ya-]az ta-ba-ar-ni LUGAL-i ˘ ti-i pa-a-pa-az-ku-ar [(a-an-na-]az-ku-ar ti-i iš-ka[ n]u-uš-ši-am-pí ti-i a-ri nu-uš-ši-ya-am-pí ti-i a. ‘And now, o Sun-god p, may the tabarna have you as his very father, you as his very mother.’ b. ‘Anoint him, and exalt him now!’ Ï Interpretation of Pal. iška in (A3) is disputed. Ï Melchert (1984:30–1) takes as a form of verb ‘be’ in substantive usage, reads with preceding text as in (A3a) (followed by Watkins 1993:477–8). Ï Yakubovich (2006:121) takes as form of verb ‘anoint’, reads with following text as in (A3b). Ï Yakubovich (2018, p.c.) supports meaning in (A3b), but proposes a different morphological analysis such that “iška” does not exist. 5/8 On Palaic iška (A3) KUB 35.165 obv. 21–22 + KUB 32.17 7´–8´: 21 22 [(nu-ku)] pa-aš-hu-ul-la-ša-aš ti-[ya-]az ta-ba-ar-ni LUGAL-i ˘ ti-i pa-a-pa-az-ku-ar [(a-an-na-]az-ku-ar ti-i iš-ka[ n]u-uš-ši-am-pí ti-i a-ri nu-uš-ši-ya-am-pí ti-i a. ‘And now, o Sun-god p, may the tabarna have you as his very father, you as his very mother.’ b. ‘Anoint him, and exalt him now!’ Ï If Pal. iška in (A3) (i) exists and (ii) is a form of the verb ‘be’, its pre-form is controversial: Ï Per Melchert (1984:30) from *h1 és-(s)k“e with active inflection (IMP *–∅). Ï Per Watkins (1993:30) from *h1 s-(s)k“-ó with middle inflection (IMP *–o). 1 1 Possibly salvageable (pace Watkins 1993:477), if 〈iš〉 and 〈eš〉 are really interchangeable in Palaic texts (cf. eDiAna #634). 6/8 On Palaic iška (A3) KUB 35.165 obv. 21–22 + KUB 32.17 7´–8´: 21 22 [(nu-ku)] pa-aš-hu-ul-la-ša-aš ti-[ya-]az ta-ba-ar-ni LUGAL-i ˘ ti-i pa-a-pa-az-ku-ar [(a-an-na-]az-ku-ar ti-i iš-ka[ n]u-uš-ši-am-pí ti-i a-ri nu-uš-ši-ya-am-pí ti-i a. ‘And now, o Sun-god p, may the tabarna have you as his very father, you as his very mother.’ b. ‘Anoint him, and exalt him now!’ Ï Overall assessment Pal. iška in (A3): Ï Highly uncertain that it (i) exists; (ii) is a form of the verb ‘be’; and (iii) exhibits middle inflection. Ï But if (i–iii) are correct, the emergence of middle inflection could be a post-PA development as in Luwian. 7/8 Voice (non-)reversal in HLuw. hwiya– (A4) (a=)w=a/i-sá | za-ti LOCUS-la/i-ti-i this:LOC . SG place:LOC . SG | (“PES2 ”)HWI/A-HWI/A-ta RED :move:3 SG . PST. ACT ‘He used to go to this place.’ CONN = QUOT =3 SG . NOM (KARKAMIŠ A6 §9) Ï HLuw. hwihwiya– — the reduplicated stem of hwiya– ‘move’ — exhibits active inflection. ⇒ Reduplication on its own (viz., in absence of imperfective –sa–) does not trigger voice reversal. 8/8 Voice (non-)reversal in HLuw. hwiya– (A4) (a=)w=a/i-sá | za-ti LOCUS-la/i-ti-i this:LOC . SG place:LOC . SG | (“PES2 ”)HWI/A-HWI/A-ta RED :move:3 SG . PST. ACT ‘He used to go to this place.’ CONN = QUOT =3 SG . NOM (KARKAMIŠ A6 §9) Ï HLuw. hwihwiya– — the reduplicated stem of hwiya– ‘move’ — exhibits active inflection. ⇒ Reduplication on its own (viz., in absence of imperfective –sa–) does not trigger voice reversal. Ï Absence of voice reversal consistent with Yates and Gluckman’s (2020) analysis of ACTIVA TANTUM: lexically marked for [+ACT] on root, revert to MID inflection when root and voice are non-local. 8/8