Descendants and ancestry of a Proto-Indo-European phytonym *meh₂l-

Descendants and ancestry of a Proto-Indo-European phytonym *meh2l-1 Rhona S. H. Fenwick The University of Queensland Email address: [email protected] Human migrations often lead phytonyms to be repurposed over time to suit differences in local plant communities. This is particularly true of plants that have great cultural importance. In this article evidence from several IndoEuropean branches is marshalled to reconstruct a basic Early PIE phytonym *meh 2l-, from which a range of derived stems diverged in the descendant languages to refer to plants as diverse as the apple, mallow, pomegranate, mango, ivy, and grapevine. Internal reconstruction of *meh2l- is undertaken to hypothesise its origin in a cryptonym *móh2-l-s (genitive *méh2 -l-s) ‘growing or fruitful one’, one of a very small class of Early PIE l-stems. Introduction Names for a variety of plants are reconstructible for the period of PIE unity, including a wide array of both wild and cultivated species. Much, however, remains to be illuminated, and it was in the process of attempting to reconcile two tantalisingly similar but apparently irreducible protoforms for the apple – perhaps the most culturally important of PIE-period fruit trees – that patches of light began to fall upon a much broader set of phytonyms, not restricted to fruit trees or even to woody plants, all based upon a single PIE root. In this article, I seek not only to unify the two PIE terms for the apple, but to demonstrate how the apple terminology of PIE fits into a larger phytonymic constellation, derived from a single common origin through ordinary PIE morphological means. The beginning: bobbing for apples Initial efforts sought to identify a common ancestor for two PIE terms for the apple (Malus domestica L.). The establishment of a single parent term remains elusive, and IndoEuropeanists are faced with the problem of reconciling a pair of tantalisingly similar but apparently irreconcilable roots: *h 2éb(ō)l-2 and *méh 2l-. A basic stem referring to apples was already reconstructed for PIE by Pokorny (1959) – as *ā̆bel-, *ā̆bōl-, *abəl- – and is reflected in robust stem series from Baltic, Slavic, Celtic, and Germanic, as well as a single potential cognate in an Italic toponym (see also Hamp 1979; Adams 1985):3 1 Abbreviations: Alb = Albanian, Anat = Anatolian, Arm = Armenian, Av = Avestan, BSl = Balto-Slavic, Celt = Celtic, coll. = collective, CS = Church Slavonic, dial. = dialectic, EBalt = East Baltic, Eng = English, f. = feminine gender, Gaul = Gaulish, gen. = genitive, Gk = Greek, Gmc = Germanic, Goth = Gothic, Hellen = Hellenic, Hitt = Hittite, id. = idem, IE = Indo-European, IndIr = Indo-Iranian, Ital = Italic, Lat = Latin, Latg = Latgalian, Latv = Latvian, Lith = Lithuanian, m. = masculine gender, n. = neuter gender, nom. = nominative, NWC = North-West Caucasian, obl. = oblique, OCS = Old Church Slavonic, OE = Old English, OHG = Old High German, OIr = Old Irish, ON = Old Norse, Osc = Oscan, PIE = Proto-Indo-European, Pruss = Prussian, Russ = Russian, Skr = Sanskrit, Toch = Tocharian, WBalt = West Baltic. 2 For the amphikinetic collective, cf. Adams (1985:80) and Beekes (2011:195). 3 To maintain regularity in discussing morphophonological changes, stems are reconstructed using PIE phonology here regardless of the number of branches in which a single stem is reliably represented. PIE *h2ébl- ‘apple, apple tree’ → thematic *h2ébl-o- → Eng apple, OHG apful, Crimean Goth apel, ON epli,4 Old Pruss woble, Old Czech jáblo5 , OIr ubhall, ubhull,6 Welsh afal, Breton aval ‘apple’ → amphikinetic coll. *h2ébōl(-s) → Proto-EBalt *ā̂bōl (gen. *ābelés)7 ‘apple’ *ā̂bōl(s) → Latv ā̂bols, Latg uobals, Lith obuolỹs, óbuolas ‘apple’ *ābelés → Latv ā̂bele, Latg uobeļs, Lith obelìs, obelė̃ ‘apple tree’ PIE *h2ébl̥ -dʰro-m ‘apple-instrument’?8 → ON apaldr, OE apuldor, apuldre, OHG affoltra, affaltar ‘apple tree’ PIE *h2ébl̥ -ko- ‘having apples’ → Middle Ir ablach, abhlach ‘having apple trees’ PIE *h2ébl̥ -ni- ‘having apples’ → Old Pruss wobalni, Czech jabloň, OCS аблань ‘apple tree’ PIE *h2ébl̥ -no-s (m.), *h2ébl̥ -n-eh2 (f.) ‘of apples’ → Celt *áβallos (m.) → Gaul avallo ‘apples’ → Celt *áβallā (f.) → OIr abhall, Old Breton aballen ‘apple tree’ *h2ébl̥ -n-eh2 perhaps also → Osc Abella ‘a town famed for its apples’9 Pokorny had already seen similarities with the series of basic and derived Hellenic forms apparently going back to a PIE root *méh 2l-, particularly given that PIE *b is rare to the point that its very existence as a phoneme in the protolanguage is doubted (see e.g. Hopper 1973; Fortson 2004:54): PIE *méh2l-o- ‘apple, apple tree’ → Attic ῆλον, Doric ᾶλον ‘apple, tree-fruit’ PIE *meh2l-éy-o- ‘made of apples’ → Attic ηλέα, Ionic ηλέη ‘apple tree’ PIE *méh2l-eiw-o- ‘to do with apples’ → Attic ήλειος ‘apple-bearing’ PIE *méh2l-ih2 -no- ‘of apples’ → Attic ήλινος, Aeolic, Boeotian άλινος ‘of apples’ But Pokorny could only adduce an uncertain relationship, and Friedrich (1970:57-64) also concluded that the two could not be easily resolved. A single protoform remained long elusive, and many have simply avoided treating the two as monophyletic (see Adams 1985; Blažek 2004; Beekes 2011:195; Cheung and Aydemir 2015). Even a recent attempt to unify the Hellenic terminology with Latin mālum, Albanian mollë, and Hittite ša-ma-lu ~ ša-am-lu ‘apple; (?)apricot’ by way of a putative PIE *sm̥ h2l- (Kroonen 2016) dismisses the form *h2éb(ō)l- as fundamentally unrelated. One effort to adduce a single *h Xm̥ b- ‘swell, protrude, be rounded’ as a common origin of *h2ébl- and *méh2l- (Zavaroni 2007) cannot be taken seriously for an excessive degree of phonological tolerance; on phonologically safer grounds, Gamkrelidze and Ivanov (1995:50) use the Hittite forms to reconstruct an Indo-Hittite 4 Via spreading of umlaut from the Proto-Gmc dative *áplai. Diminutives have obliterated the basic forms in all modern Slavic languages (Derksen 2008:25-26). 6 Via spreading of u-infection from the Proto-Goidelic dative *áuβaulu (Jaskuła 2006:199-202). 7 Backformation of tree-names from the Proto-East Baltic genitive *ābelés involved much declensional and tonic remodelling. For the ablaut, compare also PIE *swésōr (gen. *swesrés) → Lith sesuõ (gen. *seserés → sesers̃ ) ‘sister’ (Adams 1985:80; Fraenkel 1965:777). 8 While the PIE instrumental *-dʰro- usually accompanies verb roots, North and West Gmc forms regularly go back to Proto-Gmc *ápuldrą. Ascribing OE –dor, –dre to OE trēow ‘tree’ is phonologically implausible. 9 The Oscan place-name Abella is said to mean ‘apple-bearing’ by Virgil (Aeneid VII:740, Lat mālifera). Its cognacy with Celt *áβallā ← PIE *h2ébl̥ -n-eh2 is uncertain (Adams 1985:81), but the congruence is consistent with the proposed Italo-Celtic clade in IE (see e.g. Watkins 1966; Jasanoff 1997; Ringe et al. 2002). 5 *ŝamlu- [sic], and adduce a suite of Indo-Iranian terms for mangoes and pomegranates10 – notably Sanskrit āmráḥ ‘mango tree’ and āmrám ‘mango fruit’ – in support of a later development *h 2éml- → *h2ébl-. But despite these insights and suggestion of a likely IndoIranian cognate cluster, it is unnecessary to assert an Indo-Hittite ‘compact fricative’ *ŝ, which Gamkrelidze and Ivanov (1995) claim was distinct from *s and lost initially in all branches but Anatolian. It would be more parsimonious to treat the Hittite initial *š- as simply reflecting PIE s-mobile: PIE *(s)h3ngʰ-w- ‘nail, claw’ → *sh3n̥ ́ gʰ-w-oi- → Hitt ša-an-ku-wa-(a-)i- [*sankuwāi-] ‘nail, unit of linear measure’ → *h̥ ́ 3ngʰ-w-i- → Lat unguis ‘nail, claw, hoof’ PIE *(s)h2éḱr-u ‘bitter; tear’ → *sh2éḱr-u → *sh2éh2r-u (Kloekhorst 2008:391) → Hitt iš-ḫa-aḫ-ru [*isḫaḫru] ‘tear’ → *h2éḱr-u → Skr áśru, Toch A ākär ‘tear’ PIE *(s)h3ékʷ- ‘eye’ → *sh3ékʷ-o- → Hitt ša-ku-wa(-a)-, ša-a-ku-wa- [*sākuwa-] ‘eye’ → *h2ékʷ- → Gk ὤψ ‘eye, face’, Toch A ak, B ek ‘eye’ More problematically, the PIE neuter u-stems also comprise a highly stable class in the daughter languages, and especially so in Sanskrit: PIE *dór-u ‘tree’ → Skr dā́ ru ‘wood, timber’ PIE *ǵón-u → Skr jā́ nu ‘knee’ PIE *médʰ-u ‘honey, mead’ → Skr mádhu ‘honey’ PIE *h2éḱr-u ‘bitter’ → Skr áśru ‘tear’ PIE *péḱ-u → Skr páśu ‘livestock’ But āmráḥ and āmrám, the proposed Sanskrit cognates of Hittite ša-ma-lu ~ ša-am-lu, are thematic o-stems (as are the forms in the rest of IE), speaking against the possibility of treating the Hittite u-stem neuter as a simple descendant of the same PIE stem; Hamp’s (1979) and Adams’s (1985:80) reconstructions of a PIE u-stem neuter *āblu- (← *h2ébl-u) must similarly be discounted. Nonetheless, the Indo-Iranian cognates that Gamkrelidze and Ivanov propose show just the expected *h2eml- that might speak to a potential intermediary between Hellenic *meh2l- on one hand, and North-West IE *h 2ebl- – that is, a dialect group comprising Balto-Slavic, Germanic, Italic, and Celtic (see e.g. Pooth 2008:234; Salmons 1992:84; Watkins 1966:33) – on the other. The major remaining challenge, then, is to explain the link between *meh2l-, *h2eml-, and *h 2ebl-, and though the metathesis required to transform *meh 2l- into *h2eml- or vice versa (C1VC2C3- → C2 VC1C3-) is irregular, a rather straightforward explanation nevertheless exists for why it may have taken place, and in which direction. Upon thematisation, the intermediate *h 2éml-o- forms a near-minimal pair with PIE 10 The semantic shift from one tree-fruit to another is natural, as apples do not thrive in the subtropics of the Indian subcontinent. *h2em-ró- ‘sour, bitter, raw’,11 and it is conceivable that analogy under the latter’s influence could have transformed an original *méh 2l-o- into *h2éml-o-. The minimal pair continues into Sanskrit; indeed, the descendants became so closely associated here that they were soon interchanged, and the final stress of *h 2em-ró- levelled across both: PIE *h2em-ró- → Skr amláḥ ‘sour’, āmláḥ ‘tamarind’ (with -l-!) PIE *h2éml-o- → Skr āmráḥ ‘mango tree’, āmrám ‘mango fruit’ (with -r-!) A semantic connection between ‘sour’ and ‘apple, tree-fruit’ has a strong typological parallel in modern Faroese, where a shift in sense of the basic term epli – now usually ‘potato’12 – has necessitated the formation of a new compound súrepli ‘apple’ (← súr- ‘sour’). The semantic linkage is known outside of IE as well, most notably in Abkhaz (North-West Caucasian), where a-ʨʷ’á ‘apple’ corresponds almost exactly to a-ʨʷ’-rá ‘be sour’13 and thence to ProtoNorth-West Caucasian *ʨʷ’a- ‘be sour’ (see Christol 1986:17; Chirikba 1998:235). The semantic link between PIE *méh2l-o- and *h2ém-ro-, combined with their similar phonetic content, could thus have been a strong force directing a metathesis of the former in the direction of the latter. For the later shift from *h 2éml-o- to *h2ébl-o-, it is again necessary to rely on typological concerns. As Kroonen (2016) notes, there is no known regularly-occurring change *-ml- → *-bl- at any point during the period of PIE unity, and such developments are known to occur regularly only in Proto-Greek, and later, middle Celtic: PIE *mélit- ‘honey’ → *mélit-s → Gk έλι (gen. έλιτος) ‘honey’ → *mlít-yo- ‘gather honey’ → Gk βλίττω ‘I gather honey’ PIE *mélh2- ‘mill (v.)’ → *mólh2- → Lat molō ‘I mill’, Goth, OHG malan ‘mill (v.)’ → *ml̥ h2-tí- → OIr mláith → Middle Irish bláith ‘smooth’; Middle Welsh blawt ‘meal, flour’ However, the very rarity of PIE *b (Hopper 1973; Fortson 2004:54) speaks against the possibility that it could have spread to the general lexicon through regular sound change, for if it had, it would be more frequent in lexical roots. Indeed, *h2éb(ō)l- is the only strong candidate where *b can be attributed to neither sound-symbolism nor borrowing. Any PIE term containing *b must therefore be carefully critiqued, to rule out any possibility that *b actually reflects a later development of an earlier phoneme. In this case, it is a fact of phonetic mechanics that when articulating a cluster of [nasal + sibilant] or [nasal + resonant], unless the velum is lowered at exactly the same instant that the nasal is released, the phonetic result will include a transitional stop homorganic in place of articulation with the nasal (Whiting 2004:425-426). If velar lowering occurs sufficiently early in the articulation, 11 Cf. OEng ampre ‘sorrel’, Swedish amper ‘pungent’, Lat amārus ‘bitter’, and perhaps Alb ëmblë ‘sweet’. The sense ‘apple’ is still basic elsewhere in North Gmc: cf. Icelandic epli, Danish æble, Swedish äpple. 13 Abkhaz a- is a definite-generic article; -ra is a verbal noun formant. Other North-West Caucasian terms for apples come from an unrelated noun root: Ubykh mˁe, Circassian mə. 12 consonantal nasality is lost entirely: -nr- → [dr], -ms- → [ps], -ml- → [bl]. Moreover, this type of development is a process that normally lacks historical justification and is independent of the language (Whiting 2004:425-426), so although it may also occur systematically in a way that rises to the level of a regular sound law – as appears in ProtoGreek and middle Celtic – it is also a development that may with equal ease appear sporadically. In Slavic, for instance, мл- clusters are normally preserved: PIE *ml̥ w(hX)- ‘speak’ → OCS млъва ‘tumult, hubbub’, Czech mluva ‘speech’ PIE *ml̥ d-u- ‘tender, young’ → OCS младъ, Russ младший, Czech mladý ‘young’ PIE *ml̥ dʰ-n- ‘lightning’ → OCS млъни, Old Czech mlna ‘id.’ But in descendants from Proto-Slavic *mlinъ ‘pancake’, irregular forms with initial b- arise in Upper Sorbian blinc ~ mlinc and Russian блин (Derksen 2008:319), violating the regular preservation of PIE *-ml-. Both b- and m-forms survive together in Upper Sorbian, as well as Old Russian блинъ and млинъ, allowing a secure Proto-Slavic reconstruction *ml-inъ and a connection to PIE *m(e)lh2- ‘grind, mill’, despite the clear irregularity. For PIE *h 2éml-o- → *h2ébl-o-, unfortunately no single descendant preserves both *-mand *-b-forms, but what we do see is that the *-b-form is restricted to North-West IE, and thus is probably later and phonetically innovative. That it shares its semantic field so closely with *h 2éml-o-, and differs phonologically by only a single distinctive feature of a single rare phoneme in a cluster known to be the result of both regular and sporadic mechanical denasalisation, is beyond coincidence – particularly since even the Indo-Iranian *h2eml-oitself later develops a similar epenthetic -b- at least once in the Indic subbranch (and not in Iranian or Nuristani): later PIE *h2 éml-o- → IndIr *āmrá- ‘tree-fruit’ (Indic: ‘mango’, Nuristani: ‘pomegranate’) → Proto-Indic *āmrá- → Skr āmrá→ Proto-Dardic *āmbrá → Degano Pashayi āmbrék, Kashmiri amb, ambəri→ Middle Indic *āmb(r)a → Aśoka Prakrit aṃba, Pāli amba → Ardhamāgadhī Prakrit āmba → Assamese ām, Oriya āmba, Maithili amuo → Śauraseni Prakrit *āmba → Old Awadhi āṃba, Sindhi āmo, Lahnda amb, Hindi ā́ m → Mahārāṣṭri Prakrit *āṃba → Konkani āṃbo, Marathi āṃbā, Sinhala amba → Proto-Nuristani *āmrár → Kātávari ām(r)ā́ r, Waigali, Aṣkuňu āmā́ r later PIE *h2 éml̥ -no- ‘of tree-fruit’ → IndIr *āmṛna→ Proto-Iranian *āmṛna 14 → Pashto maṇa, Shughni mūn, Yidgha åmuno ‘mango’ The development from *méh2l-o- to *h2ébl-o- is thus understood, but reconstructing its original gender is complicated by bewildering gender variety in the daughters (masculine in West Germanic and Baltic, feminine in Insular Celtic, neuter in Hellenic and Slavic). Sanskrit likely preserves the original IE situation in a dichotomy between masculine āmráḥ ‘mango (tree)’ and neuter āmrám ‘mango (fruit)’. Lexical pairs are common in Greek, Latin, and Sanskrit where a single stem supplies an animate-gender tree-name – usually feminine in 14 Also seen in early loans into Finnic (cf. Finnish omena, Estonian õun ‘apple’, Erzya umarʲ, Moksha marʲ). Greek and Latin, usually masculine in Sanskrit – and a neuter fruit-name, and because of its productivity in these three ancient branches, this pattern was likely also found in PIE: Gk ἀ ύγδαλον (n.), –ος (f.) ‘almond, id. tree’; βάτον (n.), –ος (f.) ‘blackberry, bramble’; ἐρινόν (n.), –ος (m.) ‘wild fig, id. tree’; προῦ νον (n.), πρού νη (f.) ‘plum, id. tree’ Lat amygdalum (n.), –us (f.) ‘almond, id. tree’; mōrum (n.), –us (f.) ‘mulberry, id. tree’; prūnum (n.) , –us (f.) ‘plum, id. tree’ Skr badaram (n.), –aḥ (m.) ‘jujube, id. tree’; bilvam (n.), –aḥ (m.) ‘Bengal quince, id. tree’; jambum (n.), –uḥ (m.) ‘rose-apple, id. tree’ As such, there may be grounds for reconstructing PIE *méh2l- (→ *h 2éml- → *h 2ébl-) as also having two stem-forms – *méh 2l-o-s ‘apple-tree’ and *méh 2l-o-m ‘apple’ – though if Hellenic ever had *méh2l-o-s it was soon replaced, as were many other tree-names, 15 by *meh2l-éy-eh2 (f.) ‘bearing apples’ → Ionic Greek ηλέη ‘apple-tree’, Attic ηλέα ‘id.’. Reaching more widely through Anatolian and Greek An Early PIE *méh2l-, thematicised to *méh2l-o- and metathesising to *h2éml-o- and later *h2ébl-o- in North-West IE, seems at first to match Hittite ša-ma-lu ~ ša-am-lu poorly. But Gamkrelidze and Ivanov (1995) were likely right in adducing this Hittite term as ultimately cognate to PIE *méh 2l-o- ~ *h2éml-o-, though more indirectly than they propose. Though the neuter u-stems form a stable class probably going back to very early PIE, they are also quite few, and preserved only in the most archaic PIE branches; elsewhere, they were mostly thematised at an early date. But one neuter u-stem – this time in Greek, and long defying other attempts at etymology – may suggest a genuine link between PIE *méh 2l-o- and the Hittite u-stem neuter forms. The origin of Homeric Greek ῶλυ ‘moly’ largely remains as much of a mystery as the identity of the plant it names (Chantraine 1968:729-730). It was linked by Kretschmer (1896:386) to Sanskrit mū́ lam ‘root’ by way of the sense exhibited in mūlakarman ‘the use of roots for magical purposes’, and many later authors concur with this connection (see e.g. Liddell and Scott 1996:1158), but this linkage is difficult to support. While Greek -ω- can be cognate with Sanskrit -ū- if the ‘breaking’ of the PIE sequence *-uh 3- to *-ϝω- → -ω- is genuine (see Francis 1970; Normier 1977; Rasmussen 1999), PIE *múh 3l- is otherwise unknown either in this form or in any other ablaut grades, and as noted earlier, Sanskrit normally preserves PIE neuter u-stems even more faithfully than Greek does. Connections to terms for the mallow (Greek αλάχη, ολόχη, Latin malva) have also been criticised on semantic grounds, and Chantraine (1968:730) avoids the matter entirely by treating ῶλυ as a likely loan of unknown Mediterranean origin. 15 Greek tree-names for many arboreal products with neuter names in -ον are formed with the ending -έα ~ -έη (← PIE *X-éy-eh2 ‘bearing X’): ῥόδον ‘rose’ → ῥοδέα (Ionic ῥοδῆ ~ ῥοδέη); σῦκον ‘fig’ → συκέα (Ionic συκῆ ~ συκέη); κίτρον ‘citron’ → κιτρέα. Most notable is όρον ‘mulberry’ → ορέα (Epic ορέη), as it comes from a root with strong PIE heritage (Arm mor ‘blackberry’, Welsh merwydden ‘mulberry’, Ir (s)mér ‘id.’), and a continuing parallel in Latin mōrum (n.) ‘mulberry’, mōrus (f.) ‘mulberry tree’ may indicate that some Greek tree-names ending in -έα ~ -έη perhaps obliterated older names in *-ος. Yet what if ῶλυ is not a loan, but belongs to the native neuter u-stems, one of the most ancient IE noun classes?16 Homeric ῶλυ could descend from a PIE *móh2l-u, and the root is identical (but for ablaut grade) with the already-reconstructed *méh 2l-. And it is already known that proterokinetic stress patterning spread among early PIE resonant-final nominal stems to many archaic acrostatic-inflecting nouns, generating a new class of proterokinetic nouns with root o-grade in strong cases and zero in the weak (Fortson 2004:108-109). Excellent examples survive in other u-stem neuters: PIE *ǵón-u ‘knee’ → Skr jā́ nu ‘knee’, Toch A kanweṃ ‘knees’ → old obl. *ǵén-u- → Hitt *gēnu, Lat genū ‘knee’ → remodelled obl. *ǵn-éw- → OE cnēo, Goth kniu ‘knee’ PIE *dór-u ‘tree’ → Gk δόρυ ‘tree-trunk, spear’, Skr dā́ ru ‘wood’ → old obl. *dér-u- → *dérw-om → Breton derv, Welsh derw ‘oak’ → remodelled obl. *dr-éw- → OE trēo ‘tree’, Gk δρῦς ‘tree, oak’ As PIE *móh 2l-u is structurally identical with this class, a regular weak stem *m̥ h2l-éw- may be hypothesised for it, providing an exact phonological correspondence to Anatolian apple terminology (assuming, as earlier, that initial š- of Hittite ša-ma-lu ~ ša-am-lu may be accounted for by s-mobile, thus *(s)m̥ h2l-éw-). The Hittite orthographic renditions have previously been assumed to reflect underlying *šamlu (see Hoffner 1974:113-114; Weeks 1985:83; Gamkrelidze and Ivanov 1995:550-551, and a priori, Kloekhorst 2008:712-713). But as Kloekhorst (2008:27-29) himself argues persuasively, many Hittite words spelled with CaRC- sequences (where R is a resonant *r, *l, *m, *n) probably disguise underlying syllabic resonants preserved in *CR̥ C- sequences inherited from PIE. Since Hittite regularly loses interconsonantal laryngeals, moreover (Kloekhorst 2008:81), generalising weak *sm̥ h2l-éwto the strong cases would yield precisely the phonological form *sm̥ lu, reflected in the Hittite syllabic spellings ša-ma-lu and ša-am-lu. Of course, the correspondence between Hittite *sm̥ lu- and Homeric Greek ῶλυ is still semantically challenging; regardless of the difficulties with identifying the plant to which ῶλυ originally referred, it is at least unlikely to refer either to the apple tree, or any plant closely relatable to it. However, another potential cognate pair does exist in Anatolian and Greek involving apple terminology and exhibiting a similar semantic challenge – this time with the apple-name on the Greek side of the equation. As noted earlier, Attic ῆλον and Doric ᾶλον go back regularly to PIE *méh 2l-o-m which, despite several candidates, has found no unequivocal IE cognates.17 Close phonological matches do appear in Anatolian, where Old Hittite ma-a-aḫ-la-aš [*mā́ ḫlas] ‘branch of a grapevine’ and Lydian ῶλαξ ‘a type of wine’ likely go back to PIE *móh 2l-o-s; while the laryngeal *h 2 disappears before most 16 Sanskrit mā́ lu- (f.) ‘camelfoot creeper, Phanera vahlii’ is a tempting comparandum, but I avoid it as evidence here due to the gender mismatch and lack of further identifiable Indo-Iranian cognates. 17 Alb mollë and Lat mālum ‘apple, tree-fruit’ are probably loans from Doric Gk ᾶλον – the Latin especially in view of Oscan Abella, which shows *b rather than *m – and so cannot be relied upon to secure Italic and Albanian to PIE *méh 2l-o- (see Friedrich 1970:60). Phonologically and semantically problematic, Toch A malañ and B meli ‘nose’ are now usually discounted as potential cognates (Adams 1985:82; see also Penney 1977). consonants in Hittite, other forms are suggestive that it is retained before resonants (Kloekhorst 2008:539-540, 1023). As such, Anatolian *móh2l-o- has often been adduced as cognate with the Hellenic cluster (see e.g. Kuryłowicz 1927; Sturtevant 1928, 1931; later Friedrich 1970:61-62), though discounted by others on semantic grounds (e.g. Ehelolf 1933; Puhvel 2004:4-5; Kloekhorst 2008:539). Another likely comparandum is Classical Armenian mol ‘offshoot, sucker, runner’, also consistent with PIE *móh 2l-o-s, though isolated as it is in this branch it offers little further information. In any case, the semantic disconnect between Anatolian *móh2l-o-s ‘(branch of) grapevine’ and Hellenic *méh 2l-o-m ‘apple’ makes it hard to treat these forms as related a priori, but when considered alongside the stem *móh2l-u ~ *(s)m̥ h2l-éw-, a startling connection appears. Meanings of the descendants of *móh2l-o- and *móh2l-u have seemingly been simply exchanged in Anatolian: ‘apple’ became the referent of *móh2l-u rather than *méh 2l-o-m as elsewhere in PIE, while the thematic stem *móh2l-o-s was co-opted to refer to a different culturally important plant. Such semantic exchange is uncommon, but a similar such exchange has already been noted in Sanskrit āmláḥ ‘tamarind’ and āmráḥ ‘mango’, from PIE *h 2em-ró- and *h 2éml-o- respectively. The bidirectional connection shown in Anatolian and Hellenic descendants also shows a fundamental semantic interrelationship between PIE *móh2l-u and *móh2l-o-, and suggests that both are derived from a single, more ancient phytonym. Sufficient information exists even in Anatolian and Hellenic to reconstruct the original source, as the descendant forms reflect PIE o-grade and e-grade ablaut, indicative of an acrostatic root noun. And finally, a proterokinetic inflection pattern for the putative *ustem neuter *móh2l-u is also consistent with its derivation from earlier acrostatic inflection, following broader patterns in the PIE resonant-final nominal stems, as noted earlier (Fortson 2004:108-109). Two lines of evidence thus support an early PIE root acrostatic *móh2l-s (nom.), *méh 2l-s (gen.) ‘cultivated (fruit) plant’, from which three main stems were derived: (1) thematic *móh2l-o- (Anatolian, perhaps Armenian); (2) thematic *méh 2l-o- (Hellenic), along with a. thematic *h2éml-o- (Indo-Iranian), and b. thematic *h2ébl-o- (North-West IE); (3) proterokinetic *(s)móh2l-u (gen. *(s)m̥ h2l-éw-) (Anatolian, Hellenic) Further derived stems in Italic and Balto-Slavic This is not, however, the end of the story. The u-stem neuter *móh 2l-u in particular probably served as the basis for further derivatives surviving in isolated IE branches, and these further derivations suggest the earlier root noun did not originally refer to fruit trees, but was a phytonym referring to a broader range of plants. Explicit evidence for this surfaces in the additional branches of Balto-Slavic and Italic. The etymology of Latin malva ‘mallow’ has also been the subject of debate for many years (see e.g. Lee 1968:1068), and nothing similar survives elsewhere in Italic (see Buck 1904; Untermann 2000), but the PIE neuter u-stem *móh2l-u offers a useful basis for a PIE etymology; the Latin form is consistent with a regular Italic reflex of PIE *m̥ h 2l-w-eh2, formed regularly from the zero-grade of *móh2l-u. In Balto-Slavic, by contrast, we see the same oblique stem of the u-stem neuter *móh2l-u that underlies Anatolian *sm̥ lu and Homeric ῶλυ. But in Balto-Slavic, this oblique stem has been built upon with *-sk-y-o-, producing a thematic *m̥ h2l-éw-sk-yo-. The precise semantic shift caused by this suffix complex is not quite clear, but the sole phonological irregularity in the form’s descent is that after the regular loss of the interconsonantal laryngeal *-h 2-, the subsequent cluster *ml- was denasalised to *bl-, just as it was in the apple-term *h 2éml-o-s → *h2ébl-o-s. The resulting *bléuskyos then descended regularly into Balto-Slavic *bleûskyas: PIE *m̥ h2l-éw-sk-yo- → BSl *bleûskyas ‘a plant’ → Slav *bljúščъ → OCS блющъ, Polish bluszcz, Russ плющ (with dissimilation) ‘ivy’ → WBalt *bleûskyas → Old Pruss bleusky18 ‘reed’ → EBalt *bleûskyas → Lith pliū̃škis, pl(i)ūšìs, pliū̃šė19 ‘reed’ Parallels with *wód-r̥ (obl. ud-én-) ‘water’ and *ḱérh2- ‘head, horn’ With the adduction of these further stem-forms, four potential derivatives have now been reconstructed from Early PIE *móh 2l-: (1) thematic *móh 2l-o- and *méh2l-o- (along with a later deformation to *h2éml-o- and then *h2ébl-o-); (2) proterokinetic u-stem neuter *móh2l-u with the weak stem *m̥ h2l-éw-; (3) thematic animate *m̥ h2l-éw-sk-yo- built on the weak stem of (2); and (4) thematic animate *m̥ h2l-w-eh2, built on a zero-grade of the strong stem of (2). Though there is some variety in the plant-names the descendants refer to, raising uncertainty about their semantic development, another striking fact in support of the interrelatedness of this quartet of stem-forms is that they correspond, unit for unit, with forms from another seemingly derived PIE proterokinetic neuter: *wód-r̥ ‘water’, from early PIE *wed- ‘wet’.20 Although the neuter stem-formant for *wód-r̥ is the PIE heteroclitic *-r~n rather than *-u, the root *wed- ‘wet’ exhibits the same four distinct stem-types reconstructible for *meh2l-,21 and also virtually identical inflection throughout. The sole minor exceptions are from Anatolian, where initial s-mobile is found in the proterokinetic weak stem, and an additional o-grade form in the thematic descendant: PIE *móh2l-s (nom.), *méh2l-s (gen.) ‘type of cultivated plant or herb’ → thematic *móh2l-o- → Old Hitt *mā́ ḫlas ‘grapevine’, *méh2 l-o- → Doric Gk ᾶλον ‘apple’ → proterokinetic *móh2l-u (nom.), *m̥ h2l-éw-s (gen.) o-grade strong *móh2l-u → Hellen *mṓlu → Homeric Gk ῶλυ ‘moly, a fantastic herb’ o-grade weak *(s)m̥ h2l-éw- → Anat *sm̥ lu → Hitt ša-ma-lu, ša-am-lu [*sm̥ lu] ‘apple’ → *m̥ h2l-éw-sk-yo- → BSl *bleûskyas → Old Pruss bleusky ‘reed’ zero-grade strong *m̥ h2l-w- → *m̥ h2l-w-eh2 → Ital *málwā → Lat malva ‘mallow’ 18 Few instances of the Old Prussian diphthong -eu- are attested, but peuse ‘pine-tree’ (compare Lith pušìs ‘id.’) also shows clear descent from PIE *-eu- in *péuḱ-eh2 (cf. also Greek πεύκη ‘pine-tree’). 19 The vocalism -iū̃ - rather than -iaũ- is unexpected, perhaps influenced by Slavic forms with -ю-. One wonders if the name of the Lithuanian river Pliaušė may reflect a more archaic (and regular) form of the same word. 20 Not only on the basis of Luvian wida- ‘wet’, Old Armenian get ‘river’, and other forms showing PIE thematic *wéd-o- and a later s-stem *wéd-os, but also from the external comparandum of Proto-Uralic *wete ‘water’ (cf. Hungarian víz, Erzya vedʲ, Finnish and Estonian vesi). 21 Though since PIE *wódr̥ ‘water’ has such basic meaning, further derived stems are of course also reconstructible: collective *wédōr, zero-grade thematic *údr-o-, etc. PIE *wed- ‘wet, damp, moisture’ → thematic *wéd-o- → Luvian *wida- ‘wet’, Old Arm get ‘river’ → proterokinetic *wód-r̥ (nom.), *ud-én-s (gen.) o-grade strong *wód-r̥ → Anat *wódr̥ → Hitt *wā́ tr̥ ‘water’ o-grade weak *ud-én- → IndIr *udán- → Skr udán ‘water’ → *ud-én-sk-yo- → Celt *udenskyo- → Proto-Goidelic *udʲnʲsʲkʲə → OIr uisce ‘water’ zero-grade strong *úd-r- → *úd-r-eh2 → BSl *ū́ dra → Lith ūdra ‘otter’ As no other root is yet known to exhibit this full set of stem forms, it is unknown how widely applicable this pattern is. The PIE suffix combination *-sk-yo- is unusual (Matasović 2009:395), although also found in *tus-sk-yo- ‘empty’ (→ Sanskrit tucchyá-, Church Slavonic тъщь, Lithuanian tùščias ‘id.’) (Derksen 2008:502) and perhaps also the productive Tocharian B suffix -ṣṣe ‘pertaining to’ (Meillet 1914:17).22 Moreover, with the exception of forms derived with the *-u- of the Caland derivational system (Fortson 2004:123; Beekes 2011:181), polymorphemic origins of most u-stems are hard to identify. This is especially true of u-stem neuter nouns, even though their inflectional morphology is suggestive of systematic suffixal derivation from earlier root nouns; it is therefore challenging to argue for the common origin of a neuter u-stem and a thematic series from the same PIE root. But such a pattern nonetheless surfaces in PIE *ḱerh 2- ‘head; horn’ (see Nussbaum 1986), surviving in several thematic stems as well as a root noun:23 PIE *ḱérh2 -s ‘head; horn’ → Epic Gk κέρας (gen. κέραος) ‘horn’ → *ḱr̥ ́ h2 -o- ‘head’ → IndIr *ćṛʜas → Skr śíraḥ, Av sara, sāra ‘head’24 → *ḱr̥ ́ h2 -no- ‘having horns’ → Gmc *hurną ‘horn’, CS сръна ‘roe deer’ → *ḱr̥ ́ h2 -s-ro- ‘one with horns’ → Lat crābrō, Lith šìršė ‘hornet’ → *ḱórh2-u ‘horn, crest’ (not directly attested) → *ḱórh2-u-dʰ- ‘crest, headgear’ → Gk κόρυς (gen. κόρυθος) ‘helmet, scalp’ → *ḱerh2-w-ó-s ‘horned’ → Gk κεραός (← *κεραϝός) ‘horned’ → *ḱérh2-w-o-s ‘horned one’ → Lat cervus ‘deer, stag’ → *ḱr̥ ́ h2-w-o-s25 → Welsh carw ‘deer’, Old Pruss sirwis → *ḱr̥ h2-w-eh2 ‘horn’ → IndIr *ćṛwaʜ → Av sṛwā ‘horn’26 No less than three stem-forms derived from initial *ḱerh 2- show near-identical forms to those hypothesised for *meh 2l-: a plain thematic o-stem (*ḱr̥ ́ h2-o-, beside *móh2l-o- ~ *méh2l-o-), a u-stem neuter (*ḱórh 2-u, beside *móh 2l-u), and a stem-extension *-eh2 added to the zero22 Toch B -ṣṣ- descends from PIE *-sk- before a front vowel or sonorant, a shift still visible in verbal inflections: PIE *néh2-sk-onti ‘they swim’ → Toch B nāskeṃ, but PIE *néh2-sk-e ‘it swam’ → Toch B nāṣṣa. 23 Though PIE *ḱérh2-s may itself be a fossilised ‘collective’ in *-h2, derived in earlier PIE from *ḱer(Nussbaum 1986:1-18; see also Kloekhorst 2008:446-447). 24 Compare PIE *tr̥ h2ós ‘through’ → Skr tirás, Av tarō. 25 Also reflected in loans from Baltic into Finnic: cf. Finnish hirvas ‘stag’, hirvi ‘elk’. 26 Nussbaum (1986:16) writes this as sruuā-, using the usual convention whereby Avestan glides are rendered by doubled glyphs for the corresponding vowels. This seems to lead him to take the underlying Avestan stem to be *sruw- rather than *sṛw-, and thus erroneously discount *ḱr̥ -w-eh2 as a potential ancestral form. As Avestan preserves PIE syllabic *ṛ (see e.g. Beekes 1988:93), the Avestan form here is not *sruwā, but sṛwā, with syllabic ṛ and thus regularly descended from *ḱr̥ -w-eh2. grade of the u-stem neuter (*ḱr̥ h 2-w-eh 2, beside *m̥ h2l-w-eh2). Although the o-stem parallel differs in root vocalism, and the u-stem neuter *ḱórh2-u is attested only indirectly through further thematic and athematic derivations, the set of parallelisms shows that the reconstruction of similar derivatives is quite plausible for *móh 2l-s as well. Internal reconstruction of *méh2l-: a possible new PIE *l-stem? But despite the likely existence of a cognate cluster arising from PIE *móh2l-s (nom.), *méh 2l-s (gen.) ‘type of cultivated plant’, reconstruction of this original acrostatic is challenging, because known PIE *-l-stems are so few. *sh 2él-s (gen. *sh 2l-ó-s) ‘salt’ and *séh 2-wl̥ (gen. *sh2-wén-s) ‘sun’ are the only other robustly-attested PIE *l-stems, and heteroclisis in the latter suggests that it could be deformed from an earlier *séh2-wr̥ in any case. *wébʰ-(e)l- ‘beetle, weevil’ is another candidate, but cognates appear only in Germanic and Balto-Slavic and the non-epenthesised *wébʰ-l- may survive only in dialectic Russian: PIE *webʰ- ‘weave’? → *wébʰ-(e)l- ‘beetle, weevil’ → BSl *wébl- → Proto-Slavic *wébl-ica → dial. Russian ве́блица ‘intestinal worm’ → coll. *wébōl- → East Baltic *vábōl-a- → Lith vãbuolas, Latv vabole ‘beetle’ → Gmc *wíbilaz → OE wifel, Old Saxon wivil, OHG wibil ‘beetle, weevil’ The apparent derivational origin of *wébʰ-(e)l- nonetheless tempts one to attempt an internal PIE reconstruction for *meh2l- too. Assuming for the moment that *-l represents a rare survival of an Early PIE consonant-stem deverbal noun formant, as for *wébʰ-(e)l-, what then remains is Early PIE *meh 2-. And such a root is in fact confidently reconstructed for PIE (see e.g. Jasanoff 1991; Kloekhorst 2008:541; Yakubovich 2010), meaning ‘grow, increase, mature, be fruitful’ and reflected richly in the daughters: PIE *meh2- ‘grow, increase, mature, be fruitful’ PIE *mh2-oi- (ablaut *mh2-i-) ‘grow, thrive, be fruitful’ → Hitt *māi- (*mi-) ‘grow, thrive, be born’ (see Kloekhorst 2006) → participle *méh2-i-ent- → Hitt *mayant- ‘adult, powerful’ → negative participle *n̥ ́ -mh2-i-ent- → Hitt *amiyant- ‘small, miniature’ PIE *méh2-tero- ‘mature, full-grown’ → Lat mātūrus, Russ матёрый ‘full-grown’, Bulgarian матор, CS маторъ, матеръ ‘old’ (Derksen 2008:303-304; Yakubovich 2010:487-488) PIE *méh2-ter-s ‘fruitful one, one who gives birth’?27 → *méh2-tēr by Szemerényi’s law → Gk ήτηρ, Lat māter, Toch B mācer (etc.) ‘mother’ PIE *méh2-ro- ‘great, large’ *móh2-ro- → OIr mór ‘great, large’ *méh2-ro- → OIr már, Middle Welsh mawr, Gaul –maros ‘great, large’ → elative *méh2-yos, *méh2-is ‘greater, especially large’ 27 As Fortson (2004:112) notes, efforts to extend the pattern of nomina agentis more widely through the kinship terms in -tēr have been unconvincing. The semantics here, though, are suggestive. Compare also: PIE *dʰugh2-tḗr ‘daughter’ ← *dʰeugh2- ‘to suckle, to draw milk’, cf. Skr dṓgdhum ‘id.’ PIE *ph2 -tḗr ‘father’ ← *peh2- ‘to guard’, cf. Toch B pāsk-, Skr pā́ tum ‘id.’, Lat pāscere ‘to pasture’ o-grade *méh2-yos → OIr mó(o), Middle Welsh mwy, Lat māior (m.), māius (n.) ‘larger, greater’?28 zero-grade *méh2-is → Osc mais ‘more’, Goth mais, OE mā ‘more’ → oppositional *méh2-is-tero- → Umbrian mestru ‘more, greater’ → superlative *méh2-is-to- → ON mestr, OE māst, mǣst, Goth maist ‘most, greatest’ → superlative *méh2 -is-m̥ mo- → Osc maimas, Old Breton meham, Old Welsh muihiam ‘most, greatest’ (see Matasović 2009:258) *méh 2-l- would thus mean ‘growing thing, one which is fruitful’.29 If *méh 2l- did carry these semantics, it would also neatly explain the semantic divergence between Anatolian forms meaning ‘grapevine’ and western IE forms meaning ‘apple’, drawing them both under the semantic umbrella of culturally-important growing things. Finally, the existence of a phytonym *méh2-l- from a verb ‘grow, be fruitful’ is paralleled by yet another set of IndoEuropean phytonyms from *werdʰ-, with the similar meaning ‘grow, increase, mature’ (see Gamkrelidze and Ivanov 1995:556): PIE *werdʰ- ‘grow, increase, mature’ → IndIr *várdʰ- → Skr várdhatum (past part. vṛ́ddha-) ‘grow, increase, mature’ → e-grade *wérdʰ-o-s ‘growth’ → IndIr *wárdʰas → Skr várdhaḥ ‘growth, prospering’, but also ‘tubeflower, Clerodendrum sp.’ → zero-grade *wr̥ ́ dʰ-o-s ‘growth’ → Alb (h)urdhë ‘ivy’, Proto-Gmc *wórdaz → OE word ‘gooseberry bush’, dial. Norwegian orr, erre ‘bush’ with metathesis → *rúdʰ-o-s → Ital *rubus → Lat rubus ‘blackberry bramble’30 zero-grade → *wr̥ ́ dʰ-eh2 → IndIr *vṛ́dʰā → Av varəδā ‘flower’ The stem *méh2-l- would seem therefore to mean ‘growing thing, fruitful thing’, from which further stem-extensions produced a range of terms for specific plants, both cultivated and wild. While the semantic divergence between reflexes referring to cultivated (apple, grape) and wild species (ivy, reed, mallow) is substantial, a similar semantic divergence in reflexes of *wr̥ ́ dʰ-o-s ‘growth’ – yielding Proto-Germanic *wórdaz ‘(berry-)bush’ on the one hand, Albanian (h)urdhë ‘ivy’ on the other – shows that even terms for fruit trees and ivy may arise plausibly from a single common ancestor: in this case, an acrostatic deverbal noun *móh2-l-s (gen. *méh 2-l-s) ‘thing which grows or is fruitful’, from which the phonological connections 28 Lat māior may rather arise from *méǵh2-yos ← *meǵh2 - ‘large, big’ (Schrijver 1991:480; de Vaan 2008:358359; cp. Lat magis ‘moreso’), but no reflex of PIE *ǵ survives elsewhere in Ital, Celt or Gmc. Comparatives *méǵh2-yos ‘larger’ and *méh 2-yos ‘greater’ probably coexisted in North-West PIE (see OIr maige ‘grand, great’ ← *méǵh2-yos, but mó(o) ‘greater, larger’ ← *méh2-yos), and later were confounded in Ital. 29 Though Cheung and Aydemir (2015:82) also see a PIE deverbal from *méh2 - underlying the Anat and Hellen forms, by postulating *méh2-lo- directly they avoid adducing Anat u-stem apple terminology and so their model thus lacks the further connections to Greek ῶλυ, Latin malva, and BSl *bleûskyas. 30 Many take Latin rubus as cognate with ruber ‘red’ ← PIE *h1rudʰ-ró- (see e.g. Lee 1968:1664; Livingston 2004:28, 30), but no other IE reflexes of *h1r(o)udʰ-ó- mean ‘bramble’ or similar, only being colour terms elsewhere (notably, even in dialectic Latin): PIE *h1rudʰ-ó- → Latv ruds, Lith rudàs ‘red, reddish’ PIE *h1roudʰ-ó- → OIr rúad, Welsh rhudd, ON rauðr, Goth rauþs, dial. Lat rōbus, rūfus ‘red, reddish’ Alternately, both hypotheses may be partly correct, and a metathesis *wr̥ dʰ- → *rudʰ- may have occurred under influence of *h 1rudʰ-: indeed, just as proposed earlier for *méh2l- → *h2éml- by analogy with *h2ém-ro- ‘sour’. are clear enough to establish a rich set of plant terminology spreading across most IndoEuropean branches: PIE *méh2- ‘grow, mature, be fruitful’ → *móh2-l-s (gen. *méh2-l-s) ‘growing thing, plant’ → *(s)móh2l-u (gen. *(s)m̥ h2l-éw-s) ‘a plant’ → Homeric Gk ῶλυ ‘moly, a fantastic herb’, Hitt ša-ma-lu, ša-am-lu [*sṃlu] ‘apple’ → *m̥ h2l-w-eh2 ‘a plant’ → Lat malva ‘mallow’ → *m̥ h2l-éw-sk-yo- ‘a plant’ → Old Pruss bleusky, Lith pliū̃škis ‘reed’, OCS блющъ, Polish bluszcz, Russ плющ ‘ivy’ → *móh2l-o- ‘fruit vine, grapevine’ → Lydian ῶλαξ ‘type of wine’, Old Hitt ma-(a-)aḫ-la-aš [*mā́ ḫlas] ‘branch of a grapevine’, perhaps Classical Arm mol ‘sucker, runner, stolon’ → *méh2l-o- ‘fruit of a tree (n.), fruit tree (f.)?’ → Doric Gk ᾶλον ‘apple’ → h2éml-o- by analogy with *h2ém-ro- ‘sour’ → Skr āmráḥ ‘mango tree’, āmrám ‘mango’ → *h2éml̥ -no- → Pashto maṇa, Shughni mūn, Yidgha åmuno ‘mango’ → h2ébl-o- by assimilation → the rich North-West IE apple terminology Acknowledgments My sincere thanks go to Guus Kroonen and two anonymous reviewers for their comments and suggestions. I’d also like to thank Nick Sims-Williams, who kindly alerted me to transcriptional errors in some of the Indo-Iranian forms I had cited. References Adams, D. Q. 1985 The Indo-European word for ‘apple’ again. Indogermanisches Forschung 90: 79–82. Beekes, R. S. P. 2011 Comparative Indo-European Linguistics: an introduction, 2nd edition. Amsterdam and Philadelphia: John Benjamins. Blažek, V. 1979 Indo-European “apple(s)”. In: Poschenrieder, Thorwald (ed.) Die Indogermanistik und ihre Anrainer, 11–30. Innsbruck: IBS Verlag. Buck, C. D. 1904 A Grammar of Oscan and Umbrian, with a collection of inscriptions and a glossary. Boston: The Athenaeum Press. Chantraine, P. 1968 Dictionnaire étymologique de la langue grecque. Paris: Editions Klincksieck. Cheung, J., and H. Aydemir. 2015 Turco-Afghanica: on East Iranian *amarnā and Turkic alma, alïmla, almïla ‘apple’. In: Pelevin, M. S. (ed.) Na Pastbishche mysli blagoj: sbornik statej k jubileju I. M. Steblin-Kamenskogo, 73-94. Kontrast: Sankt-Peterburg. Chirikba, V. A. 1996 Common West Caucasian: the reconstruction of its phonological system and parts of its lexicon and morphology. Leiden: Research School CNWS. Christol, A. 1986 Notes Abkhaz 3: pomme et étoile. Revue des Études Géorgiennes et Caucasiennes 2: 1–20. Derksen, R. 2008 Etymological Dictionary of the Slavic Inherited Lexicon. Leiden and Boston: Brill. Ehelolf, H. 1933 Heth. milit = “Honig”. Orientalistische Literaturzeitung 36: 1-7. Fortson, B. W. IV. 2004 Indo-European Language and Culture: an introduction. Malden: Blackwell. Fraenkel, E. 1965 Litauisches etymologisches Wörterbuch, band II. Heidelberg: Universitätsverlag C. Winter. Francis, E. D. 1970 Greek disyllabic roots: the aorist formations. Dissertation submitted to Yale University. Friedrich, P. 1970 Proto-Indo-European Trees. Chicago and London: The University of Chicago Press. Gamkrelidze, T., and V. I. Ivanov. 1995 Indo-European and the Indo-Europeans: a reconstruction and historical analysis of a proto-language and a proto-culture. Berlin and New York: Mouton de Gruyter. Hamp, E. P. 1979 The North European word for ‘apple’. Zeitschrift für celtische Philologie 37 (1): 158-166. Hoffner, H. A. 1974 Alimenta Hethaeorum: Food Production in Hittite Asia Minor. New Haven: American Oriental Society. Hopper, P. J. 1973 Glottalized and murmured occlusives in Indo-European. Glossa 7 (2): 141-166. Jasanoff, J. H. 1991 The origin of the Celtic comparative type OIr. tressa, MW trech ‘stronger’. Die Sprache 34: 171-189. Jasanoff, J. H. 1997 An Italo-Celtic isogloss: the 3pl. mediopassive in *-ntró. In: Adams, D. Q. (ed.) Festschrift for Eric Hamp, Volume I, 146-161. Washington DC: Institute for the Study of Man. Jaskuła, K. 2006 Ancient Sound Changes and Old Irish Phonology. Lublin: Wydawnictwo KUL. Kloekhorst, A. 2008 Etymological Dictionary of the Hittite Inherited Lexicon. Leiden and Boston: Brill. Kretschmer, P. 1892 Indogermanische accent- und lautstudien. In: Kuhn, Adalbert (ed.) Zeitschrift für vergleichende Sprachforschung auf dem Gebiete der indogermanischen Sprachen 31, 325–472. Gütersloh: C. Bertelsmann. Kroonen, G. 2016 On the Origin of Greek ῆλον, Latin mālum, Albanian mollë and Hittite šam(a)lu- ‘apple’. Journal of Indo-European Studies 44 (1-2): 1-7. Kuryłowicz, J. 1927 ə indo-européen et ḫ hittite. In: Rozwadowski, Jan M. (ed.) Symbolae grammaticae in honorem Ioannis Rozwadowski 1, 95-104. Kraków: Gebethner & Wolff. Liddell, H. G., and R. Scott. 1996 A Greek-English Lexicon, 9th edition, with supplement. Oxford: Clarendon Press. Livingston, I. 2004 A linguistic commentary on Titus Andronicus. New York: Routledge. Mallory, J. P. and D. Q. Adams (ed.). 1997 Encyclopedia of Indo-European culture. London and Chicago: Fitzroy Dearborn Publishers. Matasović, R. 2009 Etymological Dictionary of Proto-Celtic. Leiden and Boston: Brill. Meillet, A. 1914 Le tokharien. Indogermanisches Jahrbuch 1: 1-19. Normier, R. 1977 Idg. Konsonantismus, germ. “Lautverschiebung” und Verners Gesetz. Historische Sprachforschung 91: 171-218. Nussbaum, A. J. 1986 Head and Horn in Indo-European. Berlin and New York: Walter de Gruyter. Penney, J. H. W. 1977 The treatment of Indo-European vowels in Tocharian. Transactions of the Philological Society 75 (1): 66-91. Pokorny, J. 1959 Indogermanisches etymologisches Wörterbuch. Bern: Francke. Pooth, R. A. 2008 Proto-Indo-European ablaut and root inflection: an internal reconstruction and inner-PIE morphological analysis. In: Rasmussen, J. E., and T. Olander (eds.) Internal reconstruction in Indo-European: methods, results, and problems, 229-254. Copenhagen: Museum Tusculanum Press. Puhvel, J. 2004 Hittite Etymological Dictionary, Volume 6: Words beginning with M. Berlin and New York: Mouton de Gruyter. Rasmussen, J. E. 1999 IH, UH and ṚH in Indo-European: a phonetic interpretation. In: Rasmussen, J. E. (ed.) Selected Papers on Indo-European Linguistics, with a Section on Comparative Eskimo Linguistics, 442-458. Copenhagen: Museum Tusculanum Press. Ringe, D., T. Warnow, and A. Taylor. 2002 Indo-European and computational cladistics. Transactions of the Philological Society 100 (1): 59–129. Salmons, J. 1992 Accentual Change and Language Contact: comparative survey and a case study of Early Northern Europe. London: Routledge. Schrijver, P. 1991 The Reflexes of the Proto-Indo-European Laryngeals in Latin. Amsterdam and Atlanta: Editions Rodopi. Sturtevant, E. H. 1928 Original h in Hittite and the medio-passive in r. Language 4: 159-170. Sturtevant, E. H. 1931 Changes of quantity caused by Indo-Hittite h. Language 7: 115-124. Untermann, J. 2000 Wörterbuch des Oskisch-Umbrischen. Heidelberg: Universitätsverlag C. Winter. de Vaan, M. 2008 Etymological Dictionary of Latin and the other Italic Languages. Leiden and Boston: Brill. Watkins, C. 1966 Italo-Celtic revisited. In: Birnbaum, Henrik, and Jaan Puhvel (ed.) Ancient Indo-European Dialects, 29-50. Berkeley and Los Angeles: University of California Press. Weeks, D. M. 1985 Hittite Vocabulary: an Anatolian appendix to Buck’s Dictionary of Selected Synonyms in the Principal Indo-European Languages. Dissertation submitted to the University of California, Los Angeles. Whiting, R. 2004 Šamaš, šapaš, and Murphy’s Law. Studia Orientalia 99: 425-433. Yakubovich, I. 2010 Indo-European *mā- ‘to grow’. In: Indoevropejskoe Jazykoznanie i Klassicheskaja Filologija XIV, 478-492. Nauka: Sankt-Peterburg. Zavaroni, A. 2007 I-E. “apple”, Hamito-Semitic “genitals” and roots beginning with *HmB-. Historische Sprachforschung 120: 20-41.