Join Date: Jun 2010
Location: in your brain=بمُخك
The Indo-Hittite and Indo-European "urheimats"!?
From Martin Bernal quoting Marija Gimbutas, Colin Renfrew (Cambridge university), Tamaz Gamkrelidze (president of Georgain academy of sciences), Vyacheslav Ivanonv (Yale university) , James Mallory, Joseph Greenberg, David Anthony and other scholars...
In the first half of the nineteenth century, romantic scholars who believed in the creative powers of cold and altitude maintained that Indo- European originated in the Himalayas or some other Asian mountain range. As the century wore on, this Urheimat shifted west, and it was generally agreed that PIE was first spoken by nomads somewhere to the north of the Black Sea. In the last fifty years, this Urheimat has been generally identified with the so-called Kurgan culture (named after distinctive burial mounds) attested in this region in the Fourth and Third Millennia BCE. Possessors of this material culture appear to have spread west into Europe, southeast to Iran and India and south to the Balkans and Greece. The general scheme of expansion from Central Asia or the steppes developed before the decipherment of Hittite. The ability to read Hittite led to the discovery that it was a “primitive” Indo-European language and the further recognition of a whole Anatolian linguistic family. It is now generally agreed that Proto-Anatolian split from PIE before the latter disintegrated into its separate branches.1 It is impossible, however, to tell the length of time between the two events, which could be anywhere from five hundred years to ten thousand. In any event, the difference is sufficient to cause most general linguists to make the distinction between Indo-European and the larger grouping Indo-Hittite.2 If, as most historical linguists suppose, not merely Indo-European but also Indo-Hittite began north of the Black Sea, how and when did speakers of the Anatolian languages enter Anatolia? The terminus ante quem is provided by early Hittite names in merchants’ reports from the Assyrian commercial colony at Karum Kanesh in central Anatolia around 2000 BCE.3 Some authorities argue that the migration of Anatolian speakers into Anatolia took place early in the Third Millennium and was associated with destructions of the period known as Early Bronze Age II.4 Others prefer a later part of the Third Millennium when, Mesopotamian sources indicate, barbarians invaded Anatolia.5 These invaders would seem much more likely to have been Phrygian and Proto-Armenian speakers, that is to say Indo-Europeans in the narrow sense. The distinguished archaeologist James Mellaart has even suggested that the belt of destructions across northern Anatolia at the end of the twentieth century BCE recorded the arrival of the Hittites in central Anatolia.6 Early Hittite names attested to from before the destructions falsify this suggestion. Difficulties arise with other relatively recent scenarios from the Third Millennium. For example, linking an arrival in Anatolia with the primary split in Indo-Hittite would force the later dispersal of Indo-European languages to the late Third Millenium or even the Second. This dating would be difficult to reconcile with the association of the spread of Indo- European languages with that of the so-called Kurgan material culture that is attested archaeologically in the Fourth Millennium.7 If “Anatolian” speech only arrived in Anatolia at that time, it would also be difficult to explain the great and deep divisions among these languages, some already attested in the late Third and early Second Millennia BCE. They include not only the “central” Anatolian languages—Hittite, Luvian and Palaic—but also more remote ones—such as Lydian, Lycian and possibly even Carian and the Cretan language written in Linear A.8 It is even more difficult, if not impossible, to explain the extreme internal diversity of the Anatolian subfamily if it only disintegrated in the Third or even the late Fourth Millennium.9 An Anatolian origin for Indo-Hittite A more attractive possibility for the family’s origin is the scheme mentioned in Chapter 2; it was proposed by Colin Renfrew and, in much more linguistic detail, by the Georgian and Russian linguists Thomas Gamkrelidze and Vjac*eslav Ivanov. The scholars’ versions of the dispersal of Indo-European (I would prefer Indo-Hittite) are very different in two crucial respects: cause and date. Renfrew associates the extension of Indo-Hittite with the spread of agriculture and, therefore, dates it to the Seventh Millennium BCE. He maintains that the language was already spoken in central Anatolia by the makers of the great Neolithic cultures there. When Renfrew proposed this the Neolithic was supposed to have begun in central Anatolia in the Seventh or Eighth Millennia BCE.10 It is now known to go back to the late Ninth.11 The region was at the western end of the southwest Asian zone of agricultural domestication. Linguistically, the culture would seem, therefore, to be a descendent of both Euroasiatic and Nostratic. PIH was only one of a number of languages spoken in central and eastern Anatolia during this long period. Likely the Kartvelian (Georgian) family, a “sister” of Euroasiatic and Hurrian as well as the apparent isolate Hattic, greatly influenced Hittite and even provides it with its name. (The Hittites called themselves Nes and their language Nesili.) By contrast, Gamkrelidze and Ivanov attribute the successful expansion of Indo-Hittite to the development of wheeled transport sometime before the beginning of the Third Millennium BCE.12 They illustrate their contention with the argument that the reduplicated PIE word *khoekkholo “wheel, circle” has parallels in the Sumerian gigir, the Semitic gilgal/galgal, and the Georgian gorgal, all with the same meaning. They maintain that the single root *khoel from which the reduplication was made indicates that the original form was Indo-European.13 They then argue, on the basis of archaeology, that the greatest concentrations of carts and chariots have not been found on the steppe where conventional Indo- Europeanists site the Indo-European Urheimat but in southwest Asia. (They do not mention that the concentration is in Mesopotamia not Anatolia.14)
These loans will be discussed in some detail in the second half of this chapter. Another difference between Renfrew and Gamkrelidze and Ivanov is that while Renfrew sets the Urheimat of Indo-Hittite in the major Neolithic cluster in central Anatolia around Çatal Hüyük, Gamkrelidze and Ivanov see it as having been in eastern region of the peninsula.15 They then propose that the Anatolian (Hittite) family moved west to the center of the region.16 Armenians stayed in the homeland, while the Indo-Aryans and eastern Iranians moved east and south. The main body of Indo- Europeans, according to Gamkrelidze and Ivanov, moved east and then north, swinging east of the Caspian Sea to what the authors describe as a “secondary homeland” west of the Volga and north of the Black Sea (see Map 4). They correlate this secondary homeland with the Kurgan material culture of the steppe in the Fourth and Third Millennia BCE. From this region arose what they call the “Ancient European Dialect speakers” whose dispersal led to establishment of the branches of the Celtic, Italic, Illyrian, Germanic, Baltic and Slavic families.17 The two authors do not place Greek in this cluster. They see Greek as having been linked to Armenian and Indo-Aryan, in eastern Anatolia. Armenian remained in the homeland and Indo-Aryan moved east to Iran and eventually India. Meanwhile, Greek moved through the Anatolian speakers to the west coast and from there into the whole of the Aegean Basin. They back the hypothesis that Greek originated in eastern Anatolia by providing a number of Kartvelian etymologies.18 From the Aegean, Greek speakers moved northwards to meet those of the “Ancient European Dialects” who arrived from the north. Albanian and the dead language of Thracian were formed by this merger.19 Gamkrelidze and Ivanov maintain that the great Neolithic civilizations of the Balkans in the Sixth and Fifth Millennia BCE were non–Indo- European speakers later “submerged by migratory waves of Indo- European speakers.”20 Linguistically, the clusters Gamkrelidze and Ivanov set out are plausible to most Indo-Europeanists. Specifically, most agree on the proposed bundle of Greek, Armenian and Indo-Iranian in which, for instance, unlike other Indo-European languages, some past tenses were marked by a prefixed e-. In general, more isoglosses or similarities unite the three than occur with Greek and Italic, let alone between Greek and Slavic or Germanic.21 Even so, the historical and geographical scheme set out by Gamkrelidze and Ivanov is not necessarily the best way to explain the linguistic divisions. The authors are vague in their chronology: they merely claim that the Proto-Greeks moved through Anatolia and across the Aegean before 3000 BCE.22 This date makes it difficult to see how their speech could have retained its special relationship with Armenian and Indo-Aryan, while passing through regions in which Anatolian languages were spoken without being affected by them. Colin Renfrew has linked the diffusion of Indo-Hittite with the spread of agriculture. As mentioned in Chapter 2, Renfrew has made a number of creative modifications to the views he set out on this subject, in his 1987 book.23 On two issues, however, he has remained constant: (1) The Urheimat of Indo-Hittite was the agricultural “cradle” in central—not eastern—Anatolia. (2) Indo-Hittite, accompanied by agriculture, spread west from this Urheimat to the Aegean around 7000 BCE. According to him, Indo-Hittite speakers (now Indo-European speakers, after the split with the Anatolian branch) moved on to the Balkans to western Europe and east to the north of the Black Sea. Gamkrelidze and Ivanov and Renfrew all modify the traditional view that the Ukrainian Steppe was the Indo-European Urheimat. Nevertheless, Gamkrelidze and Ivanov call the region “the secondary homeland” and correlate it with the Kurgan culture.24 Renfrew sees the steppes merely as the base from which the Indo-Aryan speakers moved southeast to Iran and India.25 Renfrew has always emphasized what he saw as the continuity of culture in Greece since the arrival of agriculture in the Seventh Millennium. In this he was opposing his onetime fellow excavator, the Lithuanian archaeologist and polymath Marija Gimbutas. Gimbutas linked Indo- European expansion to that of the Kurgan culture, which, according to her, had covered the Balkans including northern Greece as well as much of central Europe.26 Renfrew has argued that Indo-European spread into western Europe with agriculture, displacing the earlier languages of the hunters and gatherers there. Other archaeologists agree that there was no agricultural revolution in Europe and that agricultural techniques and pottery came into central and western Europe in the Sixth Millennium BCE from the east. They are divided, however, as to whether this was the result of a migration or adaptations to the new technology made by the native peoples who had previously been Mesolithic gatherers. Furthermore, non–Indo-European speakers have survived in western Europe into historic times. Basque, for instance, is still spoken today. Thus, the majority of scholars see the introduction of Indo-European languages to western Europe as coming after the spread of agriculture in a piecemeal process starting before 3000 BCE and continuing until the present. An eclectic hypothesis I see no reason why the hypotheses of Gamkrelidze and Ivanov, Renfrew and Gimbutas cannot be reconciled or fruitfully combined. We all accept the origin of Indo-Hittite in Anatolia, against the traditional vision of an Urheimat in the steppe. Where Gamkrelidze and Ivanov see “Greek” as having moved across the Aegean, however, I agree with Renfrew that the initial move was made much earlier with the spread of agriculture. I differ from Renfrew in seeing the migrants’ language not as Proto-Greek but as a branch of Indo-Hittite. Peoples speaking forms of this language spread north to create the Neolithic civilizations of the Sixth and Fifth Millennia BCE in the Balkans. Here I differ from Gamkrelidze, Ivanov and Gimbutas. I then follow the scheme set out by W. H. Goodenough in 1970. He argued that people from these agricultural civilizations on the edge of the steppe developed techniques of nomadism. From this mixed agricultural and nomadic population that spoke Indo-Hittite the Kurgan culture formed and Indo-European, in the narrow sense, developed in the Fourth Millennium.27 At this point, I accept the conventional view that the Kurgan culture and Indo-European languages spread out from the steppe. What Gamkrelidze and Ivanov call the Ancient European Dialects (Celtic, Italic, Illyrian, Germanic, Baltic Slavic and, probably, the Tocharian families) derived from northern dialects and migrated earlier, while the Indo-Aryan (Armenian and Greek) came from the southern. It seems that Indo-Iranian speakers had penetrated Iran from the north by the end of the Third Millennium BCE. During the Second Millennium, they entered the Near East and conquered much of northern India. Already they appear to have been calling themselves Arya or Aryans.
Given the racist and anti-Semitic uses to which this name has been put, it is wonderfully ironic that the word “Aryan” has an Afroasiatic origin. It is a loan from Semitic into Indo-Iranian. In Ugaritic, the name >ary was used as a gentilic (name of a people), but the word >ary “companion” is clearly related to the Egyptian ˆrˆ with the same meaning.28 This relationship is only one of a number of linguistic indications that the Indo-Iranians were in close contact with Semitic-speaking peoples of Mesopotamia and Syria. The broad-minded Indo-Europeanist Oswald Szemerényi has argued plausibly that the reduction from the PIE fivevowel system (a,e,i,o,u) to a three-vowel system (a,i,u) in Indo-Iranian was the result of contact with speakers of Semitic with its three-vowel system.29 As Szemerényi emphasizes, such a fundamental borrowing indicates very close contacts. Proto-Greeks and Phrygians migrated through the Balkans with the Kurgan culture in the late Third or early Second Millennia BCE. The Greeks stopped short of Crete and the eastern Aegean, where Indo- Hittite languages survived for some centuries. The Phrygians moved on into northwestern Anatolia. The Proto-Armenians appear to have been moved by Uratian rulers from the region of Phrygia to their later homeland only in the seventh century BCE.30This model of PIE speakers living for a time in a secondary homeland in the steppes away from Anatolian speakers and in relatively close proximity to each other explains the complex ways in which the isoglosses within Indo-European intersect.31 Specifically, it resolves such “problems” seen by Gamkrelidze and Ivanov as the Balto-Slavic-Indo isoglossess, when, according to them these languages belonged to fundamentally different branches.32 This model also ties linguistic divergence of Indo-European with archaeological evidence from the Kurgan culture. The spread of Indo-Hittite in eastern Europe was probably from Anatolia and appears to have followed the common pattern of linguistic expansion with the arrival of agriculture. On the other hand, the diffusion of Indo-European, in the narrow sense, was from the steppe north of the Black Sea and seems to have been the consequence of later conquests, migrations and cultural influences. These were possibly linked to the domestication of horses and the development of carts and, later, chariots.33 This case makes it clear that one cannot find single explanations for widespread developments. One must always be alert to the possibility that similar changes may be the results of very different processes.
Linguistic borrowings indicating an Anatolian origin for Indo-European Gamkrelidze and Ivanov reinforce their claims for an Anatolian Urheimat by listing what they see as striking parallels between PIH and languages known to have been in or near Anatolia. Their arguments on phonetics will be considered in Chapter 5. Here we shall look at some of the lexical items they consider to be loans into conventional PIE from other languages or language families. Gamkrelidze and Ivanov find one loan from Hattic -prass- into Hittite pars+ana “leopard.” They point out the importance given to leopard cults at Çatal Hüyük, as well as elsewhere in central Anatolia.34 This etymology is far from straightforward as pars+ana has been plausibly linked to roots ÷prs, ÷prd and ÷prq in Indo-European and ÷prq in Afroasiatic, all meaning “to tear, scratch.”35 Gamkrelidze and Ivanov provide clearer examples of PIH loans into Hattic: the PIH roots ÷wer “water” and ÷ai “to give, take” appear in Hattic.36 Gamkrelidze and Ivanov also claim PIH loans into Elamite; one of these ta “to put, place, stand” fits the PIE *d[h]eH very well. They also derive the Elamite luk “fire” from PIE *l(e)ukh.37 This, however, could equally come from a (pre-)Proto-Nostratic root (if Elamite is a Dravidian language, it belongs to the larger family). Bomhard and Kerns subsume the PIH *lew-k[h] under a Proto-Nostratic root *law-/lew “shine.” They refer to Afroasiatic forms with final -h. They do not, however, mention the Egyptian rqh≥ “light, fire” (generally rendered ro|kh or rokh in Coptic). This form indicates two possibilities: (1) that forms with a final /kh/ existed in Proto-Nostratic outside PIH and (2) that the Elamite word could be a loan from Afroasiatic. Similarly, Gamkrelidze and Ivanov provide an etymology for the Elamite pari “go on a campaign, march” from the PIH *phorH.38 This word seems ultimately related to the Proto- Nostratic root *÷phir, “to bear, bring forth”; the Egyptian pri, Coptic peire “to go, come out” is even closer. 39 Other Egyptian counterparts include pri in the sense of “to mount,” prt “ritual procession” (these are also used for the rising of Sothis/Sirius) and prw “procession” or “land emerging from the inundation.” Pari/e is only one of the PIH loans into Hurrian and its later form Urartian, proposed by Gamkrelidze and Ivanov. As these languages are always seen as northeast Caucasian and Nadene, there can be no shared Nostratic roots. On the other hand, if there are parallels with PIH we cannot be sure from which branch of Nostratic they came. For instance, where Gamkrelidze and Ivanov derive the Hurrian ass- from the PIH *es “to sit,” Bomhard and Kerns see a Proto-Nostratic root *>asy / *>esy “to put, place, be seated.” There are the Egyptian ist and the Sumerian as-te and es-de “seat, throne.”40 The Hurrian form could well be a borrowing from the latter. Similarly, where Gamkrelidze and Ivanov claim that the Hurrian-Urartian ag- comes from the PIH *÷ak& “to lead,” the root is seen by Bomhard and Kerns as Nostratic attested in both PIH and Proto- Afroasiatic.41 Therefore, the Hurrian form could also come from Semitic. The same is true of Gamkrelidze and Ivanov’s major claim that the Hattic kait “grain” and the Hurrian Kad/te “barley grain” derive from the PIH * Hat> “grain.” Thus, they argue, “The presence of a common word for grain in Proto-Indo-European, Hattic and Hurrian would be consistent with the claim that agriculture and the cultivation of particular grains developed in the range of Proto-Indo-European, Hattic and Hurrian.”42 The archaeological arguments in favor of multiple domestications of grains, including barley, were discussed in the last chapter.43 Even the lexical root itself presents problems for their argument. Dolgopolsky has proposed a Nostratic root *÷cänt “kernel, grain” found in Afroasiatic and Dravidian.44 Bomhard rejects this proposal forcefully.45 Nevertheless, a Semitic, or rather a Proto-Afroasiatic root *h≥int≥ (which Gamkrelidze and Ivanov see elsewhere as the origin of the PIE *Hand[h] “edible plant”) doubtless exists.46 However, *h≥int≥became h≥t≤t≥ in Ugaritic and Hebrew, a change that indicates assimilation of the /n/ (a similar process also took place in many Highland East Cushitic languages).47 Such forms could have been loaned into PIH or could have developed independently within it. Given the possible Nostratic root and the even more likely possibility that the Anatolian forms were borrowed from neighboring Semitic forms, there is no reason to believe that the cultivation of grains in general, and barley in particular, began in southwest Asia rather than farther south.
LOANS FROM OTHER LANGUAGES INTO PIH
While there are problems with some of the loans Gamkrelidze and Ivanov propose to and from PIH, Hattic, Hurrian and Elamite, their argument for an Anatolian origin of PIH has other supports. They provide a number of what they believe to be Kartvelian loans into PIH. The Kartvelian, or Georgian, language family was spoken in the southwest Caucasus and eastern Anatolia. Greenberg excluded Kartvelian from Euroasiatic, although he was willing to see it as in a larger Nostratic. He also followed the Czech linguist Václav Blaz*ek in seeing a large number of cognates between Kartvelian and Afroasiatic.48 Gamkrelidze and Ivanov provide twenty examples of Kartvelian cognates with PIH which they see as being borrowed from PIH into Kartvelian. Dolgopolsky and Bomhard, each, see one word (but different words) as deriving from Nostratic.49 The Indo-Europeanist J. P. Mallory believes that the parallels may have such a genetic origin.50 Nevertheless, some could well be loans, thus strengthening the case of Gamkrelidze and Ivanov for an Anatolian origin for PIE or, rather, PIH.51 Gamkrelidze and Ivanov also support their argument for an Anatolian Urheimat for PIH with what they see as the Egyptian origin of the PIH * b[h]ei “bee” from bˆt. This may be so, but the greater likelihood of its being a Nostratic or even Proto-World root will be considered below. Another possible loan from Egyptian into PIE, not mentioned by Gamkrelidze and Ivanov, is the root *÷k[h]alp for the related concepts “to hide, steal.” Bomhard sees a Nostratic root *÷k[h]aly / *k[h]Ely for these meanings attested in Indo-European and Dravidian.52This root as * kir also exists in Afroasiatic.53 All the Indo-European forms, however,end with a -p, forming a root *klep. This would seem cognate to the Egyptian kÅp “to cover, hide.”54 Gamkrelidze and Ivanov also propose some Sumerian loans into PIH. Two of these are connected with agriculture. The first they see is the Sumerian agar “irrigated territory, grainfield” as the etymon for the PIH * ak^ro “acre, field.” This word could have come through borrowing from Sumerian of the North Semitic >ikkår “laborer, peasant, cultivate.” The Sumerian word is, however, closer to the PIH root semantically.55 The second borrowing is still more complicated. Gamkrelidze and Ivanov see PIH *k>oou “bull, cow” as coming from the Sumerian *Nu[d] (=gud, gu) and possibly the Egyptian ngÅw “longhorned bull” sometimes shortened to ng or gw. Bomhard and Kerns identify the same root but claim it as Nostratic on the basis of the Sumerian, and a Dravidian, parallel. They do not mention the Egyptian forms.56 Gamkrelidze and Ivanov claim that the sequence of the velar nasal /n/ and a pharyngeal in Egyptian is comparable to the glottalized labiovelar in Indo-European. They associate these forms with the “Old Chinese” *<kuo and *ngi÷e÷u.57 They then link these to Altaic forms and go on to claim once again that “The agreement among various forms of linguistic evidence for terms for ‘wild bull’ and ‘domesticated bull’ pinpoints the Near East as the area of the first acquaintance with the wild and the domestic bull.”58 As shown in the last chapter, there were at least three distinct domestications of cattle, two in Asia and another in Africa.59 The Egyptian ngÅw may be the result of the merging of different words for cattle. Thus it is puzzling and interesting that these components have possible African, as well as Asian, connections. Orel and Stolbova propose an Afroasiatic root *gar “calf, bull” that would provide an equally good cognate for ngÅw as *Nu[d] (=gud, gu).60 The possibly related form gw can be placed in another Afroasiatic cluster *gaw also attested in Berber, East Chadic, and Omotic. Finally there is the initial ng. The chief Proto-Afroasiatic root for bovine is *Òa. The basic root in Niger-Congo is *na which may or may not be related to Òa.61 The possibility of an extension with a final -g is suggested by the Wolof- nag “cow.”62 Wolof belongs to the Atlantic branch of Niger-Congo. In the archaic Bantu language Basa, spoken in Cameroon, the word is nyàga.63 Orel and Stolbova place the Egyptian ngÅ under a root *nag[i]H in which they also place the Arabic najat “sheep” and the West Chadic *nungi “cow.”64 They are uncertain about the West Chadic because the word could have been borrowed from nage, the equivalent word in Fulbe, the language of the Fulani and another Atlantic language.65 Takács does not list ng or ngaw among possible loans into Egyptian from non-Afroasiatic languages. Nevertheless, here too an African connection may appear. Gamkrelidze and Ivanov also propose two etymologies that suggest some Sumerian metallurgy was introduced to PIH speakers. The first is PIH *r(e)ud[h] “red, copper, ore” from the Sumerian urudu. They claim convincingly that metal names often came from color terms. In this connection, it is interesting to note that in a bilingual vocabulary the Eblaite gloss on the Sumerian urudu was kàpálu/ kàpáru.66 This is clearly derived from the West Semitic ÷kpr, a root with many meanings. One is found in the Hebrew ko\p(p)er “henna” used for red dye.67 If ko\p(p)er “red” also meant “copper,” Kypros/Cyprus seems to be a West Semitic name for the island famous for its copper, rather than the toponym originating the metal name found in the Latin cuprum etc.68 The second etymology proposed by Gamkrelidze and Ivanov is the metal name PIE *(a)wes(kh) “gold” from the Sumerian guskin “gold.” The PIE form may be shaky but their argument that names for gold, the first metal humans worked, are often used for copper and other metals is plausible.69 These forms indicate both that the speakers of PIH did know about the soft metals and that they learned about them from Sumerians.70 Loans between PIH/PIE and Semitic The most impressive group of loans to PIH, according to Gamkrelidze and Ivanov, are those from Semitic. It is not surprising, given the origins of their discipline that most Indo-Europeanists vehemently oppose the idea of these loans.71 Gamkrelidze and Ivanov’s proposals include the following: PIE *t[h]auro. “bull” from the Semitic ÷tawr. This etymology has been proposed at least since the work of August Friedrich Pott in 1836.72 Its absence from Hittite does not tie the loan to Anatolia, though contact with Semitic speakers would have been more likely there than with a PIH north of the Black Sea. The root’s absence from other branches of Afroasiatic makes the direction of the loan uncertain. Gamkrelidze and Ivanov follow conventional wisdom and argue for Semitic into Indo- European on the basis of root structure and the expectation that an Indo- European *th would have been rendered *th rather an interdental *t.73 This argument is not altogether convincing as what is reconstructed as a Proto-Semitic interdental *t by some scholars is seen by others as a voiceless aspirate *th.74 PIH *G[h]ait’. “kid, goat.” Gamkrelidze and Ivanov point out that this form occurs in only two branches of Indo-European, Italic and Germanic. As these subfamilies are extremely ancient and located far from southwest Asia, the word points to a loan from Semitic into PIE. Dolgopolsky claims it for Nostratic and gives parallels not only from other Afroasiatic terms but also from Dravidian.75 Bomhard, however, does not list it. Orel and Stolbova do not view this as an Afroasiatic root. David Cohen, however, lists *gady “kid” as a common Semitic root and Saul Levin sees the connection between *G[h]ait’ and the Semitic forms.76 Levin also cites the Chinese scholar Tsung-tung Chang who claims that the Archaic Chinese for goat was *kåt (the modern jie).77 Chang’s belief that it is a loan from Indo-European seems more plausible than that the parallel is a random coincidence or Proto-World. It has not, so far, been identified in any Tocharian language but if *kat, rather than *ziang (yang), was the principle name for goat, the archaeological evidence of this early presence of goats circa 6000 BP would suggest an early expansion of Indo-Europeans.78 All in all, it would seem that PIH *Ghait’ is probably a loan into PIH from Semitic, although it could be a shared Nostratic root for wild, not domestic, goat. PIE *ag[ho]no. “lamb, small sheep,” Semitic *oeigl “young animal.” The Geoeez form oe#gwl is even closer.79 Neither Dolgopolsky nor Bomhard and Kerns see it as Nostratic, but *oeigal “cow, calf ” is a clear Afroasiatic root and there seems little doubt that it was loaned from Semitic to PIE not vice versa.80 Gamkrelidze and Ivanov explain the Indo-European /n/ rather than /l/ as the result of the existence of “a large Indo- European class of domesticated animals with *-n-.”81 I find it simpler to believe that Indo-European preserved the original sound, as in the case of the prepositions referred to in Chapter 2.82 PIE *qhe/oph. “monkey, ape.” This form is found throughout Semitic as *qop and in Egyptian as gi/wf. Gamkrelidze and Ivanov have plausibly postulated that a voiceless aspirated stop in PIE could give rise to initial/k-/ or /ø/ from the Sanskrit kápi and the Greek ke\bos to the Icelandic api and the English “ape.” The two authors explain the variations as postvelar glottalized stops *q. The proposal that they derive from postvelar emphatic stops (not necessarily glottalic) is convincing, even though the alternations q/ø are not regular in the different daughter languages.83 PIH *b[h]ar. “grain, groats.” Gamkrelidze and Ivanov claim that this must be a borrowing because of the rare /a/ root vocalism. There is no doubt that ÷bar/÷bur is widespread not only in Semitic, as they claim, but throughout Afroasiatic.84 Bomhard and Kerns, however, see bar/ bEr as Nostratic found also in Dravidian and Sumerian.85 This weakens, although it does not destroy, the view of Gamkrelidze and Ivanov that * b[h]ar is a loan from Semitic into PIH.
PIE *d[h]oHna|. “grain, bread.” This root is only found in Indo- Aryan Baltic and Tocharian. The root *duh≥n “millet” appears to be limited to Semitic.86 Gamkrelidze and Ivanov find the shift Proto-Semitic /h°/ to PIH laryngeal H easier than the other way around.
*Hand[h]. “edible plant.” Gamkrelidze and Ivanov see this in the Sanskrit ándha “plant from which the sacred soma was made,” the Greek ánthos “flower” and the Armenian and “field.” They derive this form from the Semitic *h≥int≥ “wheat grains.”88 The early twentieth century lexicographer of Indo-European Julius Pokorny sees a root *andh with the same word collection together with the Tocharian ant(e) “plain.” While the Armenian and Tocharian forms may well be related, I shall argue in Chapter 10 that it is much more plausible to derive the Greek ánthos “vegetable, life force” from the Egyptian ntr “vegetable, life force” than from a PIE etymon.89 PIH *k’oern. “millstone.” The parallels of this word with the Semitic root ÷gurn- have been observed at least since Hermann Möller wrote in 1911.90 It has not so far been found in any other branch of Afroasiatic. Bomhard and Kerns, who propose a Nostratic root, link *Gar-*GEr?- to Kartvelian and Dravidian as well as to the Indo-Hittite *gher, *ghor, from which come “grind” etc.91 Bomhard and Kerns disagree with Gamkrelidze and Ivanov and link “quern” not to *gurn but to another Nostratic root *k’wur-/*k’wor- “to crush, grind,” which has no Afroasiatic forms.92 Given the uncertainty on the voicing of emphatic consonants, I think that Bomhard and Kerns were too rigid here. Moreover, given the semantic identity, there is certainly a relationship between *k’wur-/*k’wor and * Gar-*GEr-.
The difficulty Bomhard and Kerns have in relating the Semitic to the PIE initial consonant genetically would be eased by proposing, as Gamkrelidze and Ivanov have done, a loan from Semitic into PIH.93 PIE *Med[h]u. “honey, beverage made from honey, mead.” Gamkrelidze and Ivanov propose this as a loan from the Semitic root ÷mtq. Their reasoning is that reduction of complexity is generally more common than the reverse and the root is attested only as a noun in Indo- European but as both a noun and a verb in Semitic. They, thus, see the loan as originating from Semitic.94 The root is, in fact, much more widely attested than merely in Indo- European and Semitic. Some of the clearest common roots in Nostratic are those dealing with honey. The stems *mit and *bit “bees” and their product *mel “honey” are found throughout Afroasiatic.95 The roots *mit and *bit are also found beyond Nostratic. The reconstructed Chinese word for bee and honey is *mi÷e*t.96 This could be a loan from Indo- European, although that is unlikely. Théophil Obenga, the Congolese specialist in African languages, has also pointed to apparent Niger-Congo parallels.97 Lionel Bender sees a root *bim, *bi or *mbe “bee, honey” in Proto-Nilo-Saharan.98 Similar forms even exist in Polynesian.99 Thus, there is a possibility that the roots of these forms may be Proto-World. Certainly, hunters and gatherers enjoy honey. PIH *P[h]elek[h]u-. “ax, poleax.” Since the decipherment of cuneiform in the mid-nineteenth century, scholars have been struck by the resemblance between the Greek pélekus found in Homer and the Akkadian pilaqqu both meaning “ax.”100 They also knew that pélekus, although with an unusual form for Indo-European, had a regularly derived cognate in the Sanskrit parasú. In addition, they realized that the Semitic roots ÷plq and ÷plg, with meanings of “cleave” and “cleaver” were widespread.101 Therefore, they found it impossible to link pilaqqu with pélekus . Gamkrildze and Ivanov’s solution to the riddle is to propose, convincingly, that the loan was made from Semitic when the Greek and Indo-Aryan branches of Indo-European were still united. The process must have taken place early before Indo-Aryan went through the satem shift from palatal k[h]y > s.102 PIE *Sek[h]u|r-. “ax, poleax.” Gamkrelidze and Ivanov see a loan from a Semitic root attested in the Akkadian sukurru “javelin” and the Hebrew segor “ax” to the Latin secu@ris and the Old Church Slavonic sekyra “ax.” The fact that these two languages are in what the authors describe as Ancient European Dialects convinces them that the form is an early loan.103 There are, however, some difficulties with this scheme. First, the word ságaris (5) “ax” is found in Greek and like almost all words in Greek with initial s-, is clearly a loan word. Since it is associated with Scythes and Persians, the word is supposed to come from one of these languages, but no trace of it has been found in Indo-Aryan. Ságaris is also widely thought to be the origin of the Hebrew sEgor, “battle ax.”104 The Semitic root ÷sgr is “to shut, close, imprison.” The word sEgôr means “to enclose, encase.” The Latin se\cu\rus is generally thought to derive from the Indo- European root *sek “cut” and could be a development of the Indo- European root *sek “cut” and cu\ra “cut off from care.” This charming and ancient explanation is clearly a folk etymology. A loan from the Canaanite (probably Punic) sEgôr or the passive participle of ÷sgr, sågur/ såkur is much more likely. The alternation sgr/skr is found in both Hebrew and Phoenician. PIE K[h]laHw. “lock, close: key.” Gamkrelidze and Ivanov derive this from a Semitic root *k-l “to hold back, restrain, lock.” Bomhard and Kerns, however, postulate *khal-*khEl “to guard, hold back, watch” as a Nostratic root, even though the only form they can find outside Indo- European and Semitic is the Sumerian kal “hold, keep, retain.”105 This could easily be derived from the Akkadian kalû. PIE *naHw-. “ship, vessel.” In this case Gamkrelidze and Ivanov have been ingenious or far-fetched. They derive *naHw- from the Semitic root *>unw(at) “vessel.” The derivation requires a metathesis of the initial >u- and the laryngeal lengthens the root vowel. Bomhard and Kern give a Nostratic etymology for what they describe as the PIE *ne?H- (glottal fricative) *no?H “sail, ship.” They emphasize the process and link it to the Afroasiatic *ne?-/*nE? “to come, go, arrive, travel, sail” in particular the Egyptian nOEˆ Coptic na “to travel by boat.”106 There is also the similar form nyw “pot, vessel,” which may be related.107 Either a common Nostratic root or a borrowing into PIE from Egyptian would seem more plausible than Gamkrelidze and Ivanov’s Semitic derivation. PIE *k[ho]r(e)i. “buy, trade, barter.” Early in the twentieth century, Herman Möller proposed deriving this root from the Semitic *kri.108 It fits well in both its phonetics and its semantics. PIE *t’ap[h]. “sacrifice.” This root is found widely both in Indo- European, from Latin to Tocharian, and in Semitic. Gamkrelidze and Ivanov’s proposal here is plausible, although it may at times have been confused with a Nostratic root *t[h]ap[h] “fire, burn.”109 PIE *Hast[h]er-. “star” and the Semitic *oettar “deified star, planet Venus.” While these forms are clearly related, the nature of that relationship is not at all obvious. Gamkrelidze and Ivanov contradict themselves here. They first argue that the loan was from Semitic to Indo-European as the initial oeayin in Semitic corresponds to the Indo- Hittite, laryngeal H. They further add that, very much as in Semitic languages, in Indo-European an interdental spirant /t/ produced an /s/ or ø in different branches.110 In a footnote, however, the authors argue for a loan from Indo-Hittite into Semitic on the grounds that the elements in *Hast[h]er can be explained within Indo-Hittite while those in *oettar cannot.111 On the other hand, John Pairman Brown argues that
* oettar was derived from the Sumero-Akkadian name of the goddess Is¨tar.He then suggests that this form itself may be “a very old loan from the Indo-European for “star.”112 Whichever the direction, the relationship between the two roots indicates close contacts between early Indo-Hittite and Semitic speakers. PIE *Sep[h]t[h]m¢. “seven.” Gamkrelidze and Ivanov, following a tradition going back to Möller and beyond, derive from the Semitic *saboe feminine *saboe -at.113 They argue that “borrowing of numerals especially those higher than five is a widespread phenomenon attested in many languages and can be explained by particularly close contact and cultural interaction.” They also see Proto-Kartvelian as having borrowed the Semitic feminine form to form *swid.114 Gamkrelidze and Ivanov do not take up Möller’s parallel etymology from Semitic: that of the Indo-European root he refers to as *s-g:- “six.”115 Pokorny describes the root uncertainly as *su÷ek|s, *sek^s, *ksek|s, *ksu÷ek|s, u÷ek|s or uk^s.116 The Sanskrit form is xát but the Avestan is xsvas. The initial cluster xsv is unparalleled in Indo-European. This suggests a loan but from what Semitic form? The situation is further complicated by the medial -d- found for “six” in the masculine in Ge’ez and other Ethiopic languages. Akkadian and Hebrew have forms based on s--s/t. The conventional explanation is that the Semitic root is *sds, but this was dropped in the southwest Asian languages.117 Saul Levin uses Egyptian forms sis or srsw to postulate that the Afroasiatic root was *SeCS (S standing for a sibilant or related fricative and CS for an unspecified consonant). He goes on to argue that the medial -d- was inserted to avoid the confusing sibilants. He also insists on the importance of the linked numbers six and seven in Mesopotamian and western Semitic culture. There were for instance, the seven visible planets, leading to the days of the week, the seven days of creation and resting after the sixth day, not to mention the sexagesimal system.
PIE K[h]r¢-n. and the Semitic *qarn “horn.” Finally, Gamkrelidze and Ivanov take up this example. The Indo-Europeanist Alan Nussbaum devotes his book Head and Horn in Indo-European to this term.119 It is a significant indicator of Nussbaum’s cultural blinkers, as well as those of his teachers, colleagues and referees, that in this study of the Indo- European root *kher “head” and *khr¢-n- “horn” he does not mention, let alone discuss, the fact that the Semitic root for “horn” is *qarn. Gamkrelidze and Ivanov maintain that the loan is from PIE to Semitic not the other way around.120 They argue for this direction because *o k[h]r≥-n derives from a root *k[[h] r “top, head,” lacking in Afroasiatic.
Saul Levin, too, argues that *qarn is a loan from Indo-European into Semitic on the grounds that it is unattested elsewhere in Afroasiatic.121 Egyptian, however, has the words qÅ “be high” qÅÅ “hill,” which have cognates in Berber, Semitic, and Lowland East Cushitic.122 Orel and Stolbova see parallels to the Semitic *qarn in the Late Egyptian qrty “two horns” and the Omotic qar “horn.”123 Whatever, the direction of borrowing, the striking similarity of the two roots indicates close contacts between Semitic and PIE speakers. Hittite Íall-i-. “royal” Akkadian ßarr-um “king.” That this crucial term should have been borrowed from Akkadian is important but not surprising, given the contacts known to have taken place between Akkadian-speaking Assyrian merchants and Hittite speakers around 2000 BCE. Probably these contacts occurred much earlier.124 To conclude this section, Gamkrelidze and Ivanov have produced or repeated a number of significant lexical exchanges between PIH-PIE and many southwest Asian languages. Some of the etymologies seem unlikely; others are as well or better explained by derivations from common roots in Nostratic rather than from loans. Nevertheless, a substantial core remains, especially of loans from Semitic. Unless a loan is attested in both Anatolian and PIE, however, it cannot be used as evidence that Anatolia was the Indo-Hittite Urheimat. Sufficient other evidence in favor of this Urheimat exists elsewhere. In fact, very few of Gamkrelidze and Ivanov’s proposals can jump this double hurdle. Nevertheless, the lexical exchanges and the evidence of morphological and structural loans from Semitic into PIE, which I hope to show below, indicate very close relationships between Semitic and PIE speakers, particularly just before the breakup of PIE in the Fourth Millennium BCE.Beyond words: Further borrowings between Afroasiatic and Indo-Hittite Any aspect of one language can be transferred to another. Generally, however, there is a hierarchy in borrowing: The easiest to borrow are content words, chiefly nouns. Then comes the transfer of functional words, conjunctions and adverbial particles. With more intense contact one can find prepositions and postpositions; this involves some structural changes. Beyond that with strong cultural pressure more or less significant structural modification can occur.125 When looking at relations between Afroasiatic and PIH, the observer may find it difficult to distinguish these more fundamental exchanges from shared derivations from Nostratic. For instance, while the negation /n/ is common throughout Nostratic, the prepositions *en “in,” *an “on,” and *ad “to” (referred to in Chapter 2) are only attested in PIE and Afroasiatic.126Thus, they could either be forms common to Nostratic that have dropped out of other branches or they could be loans, probably from Semitic to PIE. A morphological feature with a similar ambiguity is the ending /-i¤/ used in Indo-European with both active and inactive nouns with the general sense “of pertaining to” and used later for the genitive or locative. This appears in other Euroasiatic languages, such as Ainu, Aleut, Inuit and, possibly, Chukchi and Korean.127 The parallels with Indo- European Afroasiatic, however, are even more striking. The so-called nisba form /-i¤/ “belonging to” or “the one who” is found in both Egyptian and Semitic: see such contemporary forms as Iraqi, Baghdadi, Jordani, Israeli. In Proto-Semitic, too, the nisba is associated with the genitive ending found in Eblaite, Akkadian and Arabic and is the same as that in PIE, -ı\:.128 Thus, these correspondences, like those of the prepositions, indicate either a Nostratic root or an Afroasiatic loan into PIE or both.
DEVELOPMENT OF AN INDO-EUROPEAN GENDER SYSTEM BASED ON SEX
The “borrowing” of an organizational system as fundamental as gender requires not only close contact between speakers of the two languages but also existing exchanges of vocabulary and other grammatical features. In Chapter 3, I considered the possibility that the Afroasiatic binary sexual gender system was derived from Khoisan. In that case dealing with the very distant past and since we have little knowledge of East African Khoisan, it is only possible to find a few lexical borrowings. In the case of Afroasiatic, or specifically Semitic and PIE, as suggested above, we have more evidence. According to the conventional definition, “genders are classes of nouns reflected in the behavior of associated words.”129 By this definition, gender is extremely common among the world’s languages. Here, however, I am concerned with the limited subsets of binary systems reflecting the oppositions animate-inanimate and masculine-feminine, which are considerably less frequent. Although a sex-gender system is found in every branch of Indo- European (in the narrow sense), the system provides two interesting problems. 130 First, Indo-European is the only Euroasiatic language in which sex-based gender occurs and, second, linguists generally agree that the system is not primary to Indo-European. Like some other languages, PIH had a predisposition for binary structures. In this case, the major one was between “active” and “inactive” nouns. There were also verb doublets to match these categories.131 The inactive nouns were marked with “-om” and the active with nothing or, later, with a final -s or -os. Plurals of active nouns were formed with an additional -s. The inactive nouns were not considered to have plurals; therefore, they required verbs in the singular. Sometimes, however, they were considered to have mass, which was marked with *-a|.132 This much is generally agreed. How these endings became attached to the feminine gender is, however, hotly disputed. In 1906 the Neo- Grammarian Karl Brugmann argued that after some semantic shifts “had caused no more than two or three /a|/ abstracts or collectives and i|/yaor i|/i|i|stems or forms to refer to females, the analogical attraction of this handful of forms was sufficient to draw the bulk of words denoting females into what came to be the distinctive feminine classes.”133 Brugmann implied that the common Indo-European word *gu÷ena| (*k’wena|) “woman” was originally an abstract word for “bearing, parturition.” Thus, it had the abstract plural suffix -a|.134 On equally flimsy grounds, he suggested that the PIE *eku÷a (*ekhoa|) “can have meant originally a drove of horses” [my italics]. These two examples provided his grounds for the semantic shift that led to the establishment of a feminine gender with the final -a|.135 Brugmann’s hypothesis was greatly strengthened by the decipherment of Hittite. In the first place, no obvious gender system was found in the language and, second, vowel lengthenings were explained by actual laryngeals in Hittite that had disappeared in Indo-European. Therefore, conventional Indo-Europeanists have always celebrated Brugmann’s discovery. 136 Paul Brosman, the linguist whose work on this topic Gamkrelidze and Ivanov follow, accepts Brugmann’s general hypothesis but no longer maintains that the final -a| in *k’wena| derived from an abstract plural. Instead he sees the coding simply as a coincidence that helped the shift.137 In the years that followed Brugmann’s work, a number of linguists dismissed his proposal as implausible.138 The only alternative was provided by the French Indo-Europeanists Antoine Meillet and André Martinet who saw the distinction between male and female as having begun with the demonstratives *so and *sa|.139 Others came up with still lesslikely explanations. For instance, the German historical linguist Götz Weinold proposed that the three-way distinctions of masculine-feminineneuter was a natural development corresponding to what he saw as the triadic structure of Indo-European society as a whole, so beloved by Dumézil and other Aryanist mystics.140 In 1975 the American linguist Rocky Miranda revived Brugmann’s original hypothesis by providing what he saw as a parallel example of one word having created a whole new gender. According to Miranda, the Indo-Aryan language of Konkani, spoken around Goa on the west coast of India, went through a major structural change when the word c¨edu “child,” but primarily “little girl,” was reclassified from neuter to a new form of feminine. This change affected the whole system and “the neuter gender became a second feminine gender.”141 This gender bending is not as drastic a change as the Indo-European transformation from a two gender system of active-inactive to a triadic masculine-feminineneuter. Nevertheless, it does illustrate the possible attractive power of a single central word. Paul Brosman uses Miranda’s work to buttress his modified restatement of Brugmann’s views.142 Others remain skeptical. Szemerényi, for instance, argues convincingly that the long vowels in PIE should not be derived from lost laryngeals unless the latter are attested in Hittite, which they are not in the collective or neuter plural forms.143 He also questions the general assumption that Hittite lacked a feminine gender.144 Specifically, he argues that the PIE root on which Brugmann based his hypothesis—*k’wena|—was originally *gwen (*k’wen) and that the final -a was already a feminine suffix.A Semitic origin for Indo-Hittite gender In all the convoluted arguments regarding gender no Indo-Europeanist has, to my knowledge, looked beyond the Indo-European family. Given what we know about lexical exchange and the possibility of grammatical influences, it would seem worthwhile to consider the likelihood of an Afroasiatic explanation for the rise of the Indo-European feminine gender. As stated in the last chapter, almost every branch of Afroasiatic is organized on the basis of a strict distinction between male and female gender. Given the contacts between speakers of North Afroasiatic and those of Indo-Hittite—which are indicated by archaeology, lexical borrowings and mere geographical proximity—some Indo-Hittite speakers undoubtedly know about the principle of sex-based gender. In 1959, Istvan Fodor made the important point that an organizational principle of sex gender is not the same as the suffixes marking gender that appear in many genderless languages.146 In this case, however, the two coincide and the similarity of the phonetic markers in Afroasiatic and Indo-European reinforces the idea that the central structural principle was borrowed. The feminine marker *-t (not found in Khoisan) appears in Semitic, Egyptian, Cushitic, Chadic, Berber and, in some instances, Omotic.147 The vocalization of the suffix is less secure but in Semitic and Egyptian, the two language families for which there is ancient attestation and with which PIE speakers are likely to have contact, the predominant form is * -at in the singular and *-a|t in the plural.148 The Egyptologist and linguist Antonio Loprieno views the overall situation of feminine markers in Egyptian in the singular as -at after consonantal and A-stems, as -u|t after U-stems and -it| after I-stems.149 In Akkadian in the normal state, the suffix was -at-(um) or -t-(um) and in the plural it was always -a|t. In the absolute state, it was -at in the singular and -a| in the plural.150 As mentioned above, since the nineteenth century linguists have associated development of the feminine -a|or -ı|in Indo-European with the neuter plural or abstract -a. It is just possible that this neuter plural was also borrowed from Semitic. In Akkadian, abstracts were formed with the suffix -u|t.151 According to Loprieno, the similar Egyptian suffix -wt was “morphologically feminine but applied to masculine nouns is often used in the formation of collectives.”152 The Egyptologist Jean Vergote saw the Middle Egyptian -wt as having two different meanings, which he reconstructed from Coptic: the first *-u|wat included “true collective nouns”; the second *-a|wat “is generally considered a category of abstract nouns.”153 It is not clear whether these reconstructions of the Egyptian -wt can be applied to the Akkadian –u|t. Nevertheless, the inaccuracy of early cuneiform makes such an idea quite possible. In Sumerian, for which the script appears to have been designed, /w/ was rare and probably secondary. Therefore, in the Third Millennium it was used to signify wa, we, wi and wu.154 If this is the case, *-a|wat and *-u|wat would seem sufficient not only to explain the Hittite collective –a|but also to provide a reason for its connection with the PIE feminine -a|. There is no trace of a final *-t in PIH or PIE. The phonetic obstacle here to a borrowing from a Semitic suffix -at is not, however, so great as one might suppose. In both Semitic and Egyptian, the -t in the final position was clearly very unstable. When it was exposed, as when case or unstressed verbal endings were lost, the -t too was dropped, lengthening the previous vowel in compensation. This process took place for different forms in different languages at different times. In Egyptian -t was dropped during Late Egyptian 1600-1000 BCE.155 In the Canaanite dialects the development -at>-a|h was, if anything, rather later.156 Such changes would, of course, have been too late to have affected PIE, let alone PIH. This was not the case, however, with Eblaite, which is attested from the middle of the Third Millennium.
157 In this language, written in north Syria and, therefore, close to any Urheimat of Indo-Hittite or Indo- European, -t was frequently dropped, not only when exposed as a final, but also while case endings remained.
158 If this construction was expressed in writing, the process had probably been going on for a long time previously in speech. Then -at and -a|t- were replaced by “-a” or “-a-”—the length of the vowel could not be expressed in Eblaite cuneiform. However, /a|/ is very likely the sound, not only through analogy with the normal compensation but also because in Amorite, spoken in north Syria around 2000 BCE, “-at in the absolute state was replaced by “-a|.”
159 Given the structural similarities of the gender system in Semitic and PIE, as well as those between the markers for the feminine and collective, the probability of a borrowing at some level is very high. Its time, place and nature, however, is much less certain. Before investigating these factors, we should consider the question of what kind of gender existed in the Anatolian languages. As mentioned above, the scholarly consensus is that there is no trace of gender in Hittite. Moreover, the active and inactive gender system common in Euroasiatic was still strong in that language. Some have argued that the existence of suppletive indicators of sex (using different stems for pairs of the type: ram-ewe, pig-sow or boy-girl) demonstrates that a sex-gender system did not exist. Fodor has denied this line of thinking, pointing out that such lexically distinct forms have existed throughout Indo-European with its strong gender system.
160 As mentioned above, most agree that the -a|for collectives did exist in Hittite.
161 Furthermore, a minority of scholars insist that there was a three-way gender system in PIH before the separation of PIE from Anatolian, but as Szemerényi stated in 1985, “the question is still not settled.”
162 Nevertheless, the probability is that a three-way system did not exist in the Anatolian languages.
The minimal hypothesis of borrowing from Semitic is the purely structural one that speakers of PIE were aware of a language organized by sex-gender and, therefore, they used their own collective -a| for the new gender. Not far removed from this reasoning, is holding that PIE speakers in the late Fourth Millennium were reinforced in their choice of a marker for the new gender by the knowledge that at least one of the closest Semitic languages indicated the feminine with an -a|. The hypothesis that the PIH collective -a| itself derived from an Afroasiatic *-a|wat, although attractive, is obviously much more speculative.
163 In the first place, if one places the breakup of Indo-Hittite in the Fifth Millennium, the introduction of such a collective before that date would require a relatively nearby Afroasiatic language in which the final -t had been lost. One certainly could not project Eblaite that far into the past, although as shown above later evidence indicates the vulnerability of final -t throughout Afroasiatic. It is also strongly probable that at that time Semitic speakers were already ensconced in north Syria.
164 Another serious problem is the improbability that a collective would have been introduced into Anatolian before or without the feminine singular. There are in fact some slight indications that the Hittites were using -a as a female marker. If Szemerényi is right, the PIH root was * ÷gwen (*k’wen) and the final -a|, found in the Hittite koinna, was probably already a feminine suffix in Anatolian.
165 The use of this marker, however, does not mean that any of the Anatolian languages had set up a sexgender system. It is also possible that *k’wen itself was borrowed from Semitic. Orel and Stolbova postulate an Afroasiatic root *kün “woman, co-wife.” The Akkadian kinı\:tu and Arabic kann-at are found elsewhere, in Berber and West Chadic. Orel and Stolbova see the development in the Agaw *kwin- “woman” as irregular. Even so, it is possible that the Afroasiatic root, like the Indo-European one, was originally a labiovelar.
166 No doubt fewer traces of sex-gender can be found in the Anatolian languages than in other branches of Indo-Hittite. This may be because the Semitic influence developed further within Indo-European. Alternatively, gender in the Anatolian languages could have been limited or counteracted by the surrounding and underlying languages that lacked a sex-gender system: Hattic, Kartvelian, Hurrian, Sumerian, Elamite etc. It is, in fact, easy to see how Semitic languages could have influenced the Anatolian in the Fourth Millennium. Urban life began at Ebla in the middle of the Fourth Millennium as part of the trading system made up of so-called Uruk, Sumero-Semitic speakers, and connecting Mesopotamia to Iran, Syria and central Anatolia. This is 1,500 years before Assyrian merchants were recorded in Hittite-speaking Karum Kanesh.
167 As far as I am aware, little archaeological evidence of contact between southwest Asia and the steppe has been found from before the disintegration of PIE in the Fourth Millennium. Nevertheless, using the grounds of vocabulary, some have argued that there was trading around the Black Sea at this time.
168 If the archaeological evidence is thinly stretched, the linguistic evidence of exchange between Semitic and PIE is strong. The lexical borrowings are supplemented by the grammatical borrowings mentioned above. In this context, there would seem no reason to deny that the concept of a feminine gender came to PIE from Semitic and little reason to deny that the gender’s marker -a| had the same origin. As stated above, the derivation of the Indo-Hittite collective plural -a|from an Afroasiatic * -u|wat or *-a|wat is much less clear-cut.
That Indo-Hittite or Indo-European speakers borrowed the feminine gender does not mean the rigid Afroasiatic binary sex-gender system was reproduced. The earlier “active” gender of Indo-Hittite was split between masculine and feminine but the “inactive” one remained as neuter, with the original marker *-om preserved in the singular. The borrowed collective *-a| being used for the plural, but, as a collective, it took verbs in the singular. In this way Indo-European developed its unique three-way gender system.
The failure of Indo-Europeanists and other historical linguists even to consider the possibility of some relationship between the strikingly obvious similarities shared by the Indo-European and Afroasiatic gender systems is an example of the general academic tendency to avoid the obvious. In this particular case it is an indication that the men and women formed in the linguistic tradition of the Neo-Grammarians are still reluctant, or unable, to “think outside the box.” The limiting effects of this tradition must also be taken into account when considering Indo-Europeanists’ approaches to the possibilities of exchanges between individual languages belonging to the Afroasiatic and Indo-Hittite. This chapter has been concerned the origins of Indo-Hittite and Indo- European and their contacts with Afroasiatic languages before the Third Millennium BCE. In the rest of the book I shall look at the linguistic relations between one Indo-European language, Greek, and two Afroasiatic, West Semitic and Ancient Egyptian, in the following three millennia.
Last edited by Colin Wilson; 09-15-2010 at 01:42 PM.