The Silk Road refers to a complex of ancient trade routes extending roughly from eastern Europe to the heart of China. The term itself dates only to the 19th century and refers to what was merely one of the many goods transported over land via this web of mercantile passages. Though many of our ancient sources concerning these trade routes contain information either difficult to correlate with specific modern-day geographic and ethnographic knowledge or simply difficult to believe precisely because of modern-day knowledge, we nevertheless are certain of the important place in history the Silk Road holds, as a medium not only for the exchange of goods but also for the exchange of information and technology.
The route cuts through the heart of Central Asia, through the so-called "Stans" --- Kazakhstan, Kyrgyzstan, Tajikistan, Turkmenistan, Uzbekistan, as well as the northernmost reaches of Afghanistan and Pakistan --- a region known from time immemorial to be home to a wide variety of peoples from varied linguistic, ethnic and religious backgrounds. It is precisely where this trade route hits the westernmost extent of present-day China that we find the archaeological sites that provide us with Tocharian documents. To properly put the Tocharian documents and the people who wrote them in their cultural context, an admittedly cursory overview of some of the major ancient settlements and peoples of the region merits our attention.
The geographic setting is a striking mix of extremes. In the land surrounding the Aral Sea we find the southernmost extension of the great Siberian steppes. Into the western shore of the Aral Sea flows the Syr Darya, while into the southern shore flows the Amu Darya, feeding into the sea via a fertile delta suitable for cultivation. The plains continue to sweep south and east into the foothills of the so-called "Roof of the World." Here rise the Pamirs, sweeping southeast into the Hindu Kush. As one pushes farther east there loom the Qurum Mountains, to their east the Qaraqurum, and to the northeast of these the Altun. This provides the northern extent of Tibet, and if we return to the Pamirs and instead follow the southerly line to the east, we encounter the familiar and majestic Himalayas. If we backtrack once again to the Pamirs, we may this time push northeast and follow the line of the Tian Shan mountains, which provide the natural border between China's Xinjiang province to the south and Kyrgyzstan to the north.
The Tian Shan to the north and the Qurum, Qaraqurum and Altun to the south bracket with their claw the basin into which the Tarim river carries mountain waters until they finally dissipate into the sand. In the middle of all this is a forbidding, impassable desert called the Taklamakan (Turkic Täklimakan), all contained within the present borders of China. Such settlements as we find all line the rim of the desert, hovering close to the foothills of the looming mountain ranges.
Archaeological excavations show signs of habitation in Central Asia dating back at least to paleolithic times. This is evidenced by stone tools as well as cave dwellings that seem to have been seasonally inhabited along migration routes. We find a shift, some 10,000 years ago, from a typical hunter-gatherer lifestyle to one centered around agriculture and livestock. With this came a change to more permanent settlements, with roughly 30 houses built from clay grouped together in a community.
Agriculture was greatly facilitated in the sixth millenium BC by the advent of irrigation, at this early date mostly in the form of heaped up mounds of earth designed to divert flood water from the rivers. Naturally the fertile deltas of rivers such as the Amu Darya (historically known also as the Oxus), where it empties into the Aral Sea, were among the earliest sites of agriculture. Through the use of irrigation, agriculture was able to move to the upper banks of the river and then farther into the countryside in a region occasionally referred to as Transoxonia. Constructing such irrigation systems was of course a labor-intensive activity, and therefore required larger permanent populations (Frye, 1996; Roudik, 2007).
Not all inhabitants of the region chose to remain tied to a permanent settlement. In the third millenium BC we encounter evidence of northern tribes from parts of modern Kazakhstan that maintained a nomadic lifestyle while shifting from hunting to more pastoral pursuits of grazing livestock. At roughly the same time we find the appearance of copper and bronze tools, and only in the first millenium do we finally see the emergence of iron tools. Herdsmen seem early to have domesticated sheep and goats, but in the third and second millenia we also find domestication of camels and horses for pulling carts (Roudik, 2007; Kuzmina, 2008).
In connection with this southward expansion, certain specific archaeological sites are worth mentioning. In particular, archaeological finds identify three groups known as Afanasievo, Andronovo and Karasuk, which range over southern Siberia and Kazakhstan and appear to display a continuity supporting the notion of a spreading population characterized by warriors on chariots. The material finds of the Karasuk culture ultimately find their way to the Ferghana valley by roughly 1500 BC, and further evidence points to possible expansion beyond into Xinjiang (Frye, 1996; Roudik, 2007; Kuzmina, 2008).
As prehistory gives way to history at the beginning of the first millenium BC, we find the area in question encompassed by Parthia. By this time stable settlements had arisen in the river deltas near the Aral Sea, for example Khwarazm (or Khorezm, Gk. Chorasmia) at the mouth of the Amu Darya, and a Sogdian settlement along the Syr Darya. The Sogdians also appear to have established an oasis in Ferghana. We still however find a coexisting nomadic population, and by the middle of the first millenium BC we encounter the nomadic Massegetae inhabiting the region between the Amu Darya and the Syr Darya. Persian and Greek accounts typically find little by which to distinguish the various populations and seem to consider them as a group, in particular with a generally similar mode of dress including a short tunic, wide belt, and trousers, with the same customs among nomadic and settled peoples alike (Roudik, 2007).
State formation in the region appears to have proceeded by way of tribal confederation. In this way the city of Khwarazm rose to prominence and extended its sway from the Aral Sea to the mountainous region to the south. Khwarazm maintained close ties with Bactria, a neighboring state centered upon the modern city of Balkh. Their political ties mirrored the many cultural and linguistic similarities the two states already shared. We also find the state of Margiana, centered at Merv, and the long-lived Sogdian state taking foothold in the region. These four major states occupied the areas of what is now Turkmenistan, northern Afghanistan and Uzbekistan.
The Persian king Cyrus, with a victory against the Medes to the east of Parthia in the mid-first millenium BC, finally managed to bring these areas under the dominion of what became the Achaemenid Empire. The Iranian-speaking influence that had entered Central Asia with the rise of Parthia now strengthened its hold, along with a more palpable presence of the Zoroastrian religion (Frye, 1996). Thus Khwarazm, Bactria, and Sogdiana all found new identities as satrapies within the new empire, under the purview of a central governor seated at Samarkand.
The loss of independence was not without its benefits. To be sure, taxes were now levied for the simple right to open irrigation canals feeding from the rivers which were now the emperor's property. But the imperial infrastructure also provided for construction of new irrigation canals. Moreover, the bureaucratic tendencies of the empire demanded skilled bookkeepers, and this in turn resulted in a generally higher level of education. And, finally, inclusion within the empire now brought a vast military might to bear when border provinces like Sogdiana felt the pressures of skirmishes along the trade routes leading out of the empire (Roudik, 2007).
In 334 BC, Alexander the Great turned his gaze toward Central Asia and launched a campaign with such force that it caused Darius III, the Persian emperor at the time, to flee to Bactria. There he was killed in an internal plot, and shortly thereafter Alexander took Samarkand, the principal city in Central Asia. Perhaps in any other region this would have ended Alexander's need for further campaigning; but as luck would have it, this only spurred the Sogdians to revolt, thereby extending Alexander's foray and fanning the flames of military ferocity. Another year of military operations finally brought an end to the resistance, but at the price of hundreds of thousands of Sogdian casualties and diminishing the prosperity of Sogdiana and Bactria. Nor could Alexander accomplish this victory by sheer military might, but rather through a strategy of winning over local leaders and accepting the Zoroastrian religion so pervasive in the region.
The consolidation of Alexander's rule in the region added to the religious diversity. Not only did the Iranian peoples maintain their Zoroastrian beliefs, but large numbers of Greek settlers imported their own religious practices to Bactria. At roughly this time we also see the first Buddhist monks entering the region from India.
Within a relatively short time after Alexander's death, his empire began to disintegrate. The governance of the Central Asian provinces fell to the Seleucids. In the middle of the third century BC, Bactria and Parthia again began a rise to prominence. These states maintained their hold through confederation until roughly the middle of the second century BC. At this time nomads from the region surrounding Syr Darya began to mint coins and construct fortifications (Roudik, 2007). China was of course ever watchful of the nomadic tribes along its borders, and at the beginning of the second century we find it extending its commercial contacts into the region of Sogdiana. From this time onward China maintained a presence in the region, a presence either greater or lesser depending on the degree to which conflicts with the nomadic tribes allowed them access to the major commercial centers. This opened the doors to the Far East and the regions that produced silk, which the Chinese often used as a means of payment in commerce. This era marks the rise of what we know as the Silk Road.
As trade along the Silk Road increased, the ensuing prosperity fueled the growth of several settlements in the area of the Tarim Basin. These stations provided a staging ground for the interaction of various empires with interests in the region. Moreover the harsh terrain meant that, as these centers fell into the hands of one power or another, contact between Central Asia and China could effectively be cut off. Their strategic and commercial value therefore converted them into focal points for imperial aspirations in the surrounding regions.
Many of the most detailed reports about the size and makeup of these settlements come from Chinese sources. Mallory and Mair (2000) have provided an insightful and engaging study of the sources, and our discussion here will mainly review some of the highlights of their work.
Kašghar (Qäshqär, Chinese Shule) forms the gate between Bactria and Ferghana to the west and the Tarim Basin to the east. It is at this outpost that a traveller from the west must make a decision to follow a northerly or southerly route as the Silk Road forks and passes to either side of the Taklamakan desert. Chinese accounts recorded during the Han dynasty list Kašghar as containing 1,510 households, comprising 18,647 individuals, of which 2,000 could bear arms. From roughly the beginning of the second century BC to the beginning of the first century AD Kašghar formed an important garrison town in the Western Han dynasty, but subsequently fell under the control of Khotan to the south. General Ban Chao later retook the city and it remained intermittently thereafter part of the Chinese empire. Around the 7th century AD Kašghar pertained to the Tibetan empire, and in the 9th the encroaching Uyghurs took control.
But during the early part of the first century Kašghar formed part of the Kushan empire, rising to prominence for a time in the area. Evidently it was during this period that Buddhism came to Kašghar, and accounts mention that the city retained the Buddha's stone spittoon (though it was not the only city to make this claim). References to sacrifices to the sky god suggest that Buddhism by no means took hold in all pockets within the region, but rather that Zoroastrianism and perhaps other religions managed to coexist in Kašghar. By the 10th century Islam had mostly driven out opposing religions. There was little land suitable for cultivation, but the city was a primary source of wool, or felt, and carpets.
Kučā (Kucha, Chinese Qiuci) lies along the northern rim of the Tarim Basin, perched along the northern route of the Silk Road that skirts the Tängri Tagh (Chinese Tian Shan) mountain range. As mentioned above the northerly and southerly routes of the Silk Road divide at Kašghar to pass on either side of the Taklamakan desert. They rejoin again at Lopnur, to the east of the desert. The oasis of Kučā lies roughly at the midpoint of the northerly route. The northerly route itself divided at Kučā, one direction taking the traveler on to Lopnur, the other passing farther to the north over the Tängri Tagh themselves. In the middle of the first century BC Kučā was already under Chinese control, and in fact it was the seat of the governor of the Han dynasty's protectorate, a region that included all the outposts of the Tarim Basin. That status and the strategic location provided Kučā with a pivotal role in Chinese efforts to stave off the Xiongnu bands that constantly threatened attack from the north. At the same time, it appears that Kučā felt no particular allegiance to China, and its constant efforts to free itself from the empire's grip proved a thorn in China's side. In the 9th century, Kučā like Kašghar fell under the sway of the Uyghurs.
Chinese sources for the first few centuries BC comment on how the population was essentially tied to a settled way of life and not given to the nomadic wanderings of the Xiongnu. The settlement contained 6,970 households comprising 81,317 people, and apparently another 21,076 able to bear arms. Kučā was therefore largest among the cities of eastern Central Asia and simply dwarfed the nearest arm of Chinese authority in Wulei. Later descriptions from the Jin dynasty (AD 265-419) describe a city with a walled inner citadel and a thousand Buddhist shrines and cloisters. The city's dimensions evidently rivaled those of the Chinese capital at Chang'an. Chinese sources record that the agricultural resources of Kučā comprised millet, wheat, rice, legumes, hemp, grapevines and pomegranates, as well as horses, cattle, sheep and camels. Kučā's proximity to the mountains also implied a wealth of mineral goods: copper, iron, lead, gold, and tin among others. Kučā produced felt and rugs, and it remained a central point for trading silk. Commerce was facilitated by the exchange of cotton goods or copper, silver, or gold coins.
It should be mentioned that accounts record a custom in both Kašghar and Kučā whereby the native population used boards to flatten the backs of children's skulls, much like what is found among native populations in North and Central America. In Kučā men and women wore hair down to the nape of their neck, and they employed wool garments and caps. From what can be ascertained from the historical accounts, Buddhism was the primary religion in Kučā in the earliest centuries AD and likely even earlier. In line with Buddhist practice, inhabitants of Kučā typically cremated the dead.
Khotan (Chinese Yutian) was a settlement along the southerly route of the Silk Road around the Tarim Basin. Contrary to some of the other large commercial centers in the region, Khotan enjoyed a wealth of natural resources. Chief among these fortuitous natural circumstances was its location between the Yurung-kāsh and the Qara-qāsh rivers, which stem from the Qurum (Chinese Kunlun) mountains and actually extend all the way to the Tarim river itself. With these two sources to draw from, Khotan possessed ample resources for agriculture through irrigation. Moreover, the mulberry tree was native to the area, so that with silkworms and Chinese know-how Khotan was able to turn itself into the center of silk production for the region. An added benefit was that the Chinese had also invented a process for turning the mulberry into pulp for paper, and so from Khotan the first paper found its way into the region. Khotan, in addition to cotton and wool, sent from its vast supplies large quantities of jade to China. Accounts from the Han dynasty report that the city contained 3,300 households, comprising 19,300 people, of which 2,400 could bear arms.
Several legends surround Khotan. One concerns the dual personalities that factor into its legendary founding. In particular a few documents suggest that Khotan was originally founded as two separate colonies. One came from northwest India and was supposedly led by the son of king Aśoka himself; the other came from the east, led by an exiled Chinese king. After a decisive battle between the two settlements, they eventually coalesced under the rule of one of the founders (stories disagree as to which one). If nothing else the story perhaps derives from a desire to reflect the coexistence of Prakrit-speaking Buddhists alongside Chinese speakers: we in fact find Chinese coins in the area that have Prakrit writing on the reverse side. Chinese accounts testify to a hundred Buddhist monasteries in Khotan, with some 5,000 monks; but here too we find references to a 'celestial god' that suggests the persistence of Zoroastrianism.
Further accounts show an expansion of Khotan's power in the early first century AD. Evidently with the assistance of the Xiongnu nomads, Khotan was able to extend its influence as far as Kašghar to its northwest. Thus control of the southerly route of the Silk Road fell under control of Khotan and Krorän. This situation did not appeal to the Chinese, and so the general Ban Chao likewise came and annexed Khotan to the empire. Not without efforts to the contrary, Khotan nevertheless remained under Chinese control for the succeeding several centuries. At times the Turks threatened to wrest Khotan from the grip of the Chinese, but in the 8th century it was actually Tibet that managed to annex Khotan and much of the Tarim region.
Krorän (Chinese Loulan) provides an antipode to Kašghar across the Tarim Basin: if Kašghar is where the Silk Road divides to bypass the Talkamakan desert, Krorän is where the routes again combine, lending Krorän great strategic importance from the point of view of the Chinese. Krorän lies near the salt marshes of Lopnur, in earlier times an area fed by the Tarim river but bracketed by harsh deserts. Sources from the Han dynasty put the population at 1,570 households, comprising 14,100 people, of which 2,912 could bear arms. The mixture of salt and sand prevented any hope of local agriculture, and the city survived on goods brought in by trade. This furnished a somewhat symbiotic relationship with nomadic peoples from the region, and we find there signs of numerous asses, horses and camels. Krorän for its part formed a center for trade in jade, rushes, tamarisk and balsam poplar.
Krorän appears to have been caught between warring states since its earliest history. In the early second century BC the nomadic Xiongnu of the north annexed the city and from there managed to harass the Han empire. The Chinese naturally found this situation intolerable, and at the end of the first century BC it fell to invading Chinese forces. Krorän nevertheless continued to vacillate between the two powers for the next two centuries. Finally Ban Chao's son, Ban Yong, sent a group of 500 colonists to the region in the first century AD to solidify China's hold. This militarized presence had the added benefit of managing against great odds to produce functional irrigation canals in a difficult river system. Documents uncovered on the site of the colony show that in the time leading up to its abandonment in the fourth century AD, Chinese was apparently spoken alongside Prakrit and another language native to its inhabitants.
The following excerpt continues the Tocharian B text B107 (THT 107) from the previous lesson.
We find in line 13 the collocation ostameṃ ltu, literally 'left the house'. This however is something of a set phrase in Buddhist literature (cf. Pinault, 2008), meaning that one has set out on religious wanderings. Thus the phrase in context means 'become a monk' or 'enter the saṃgʰa'. As Pinault points out, the usage has an exact parallel in A waṣtäṣ lantu, and its use in such context allows us to pin down the meaning of A waṣt B ost 'house'. Namely, the term represents more than merely a family dwelling structure: it denotes organized village society in general. This parallels the distinction in Sanskrit between the grāma- 'village' and araṇya- 'forest'.
13 - Upage ājivike eṣerñāna etsuwai masa weñāmeś ṣerśkana ñiś ostameṃ ltu nesau wärpauca dakṣiṇāke. ñi ka ṣ tāṃ oṅkorñai pintwāt petes.
14 - toy weñāreneś sā oṅkorño tañ śwālya mā ste. pilycalñene lalālu laukito rṣāke tākaṃ rṣākeṃne śpālmeṃ cwi wes tā oṅkorñai pintwāt aiskem.
15 - Upagentse mañu kärstātene. ṣañ ytāri masa.
16 - Nānda cāla oṅkorñai Nandābala tāy ṣerśka postäṃ msāne. maitar yopar warttone dakṣiṇākeṃ ritasi wrocce rṣākeṃ.
13 Upage ājivike eṣerñāna etsuwai masa weñāmeś ṣerśkana ñiś ostameṃ ltu nesau wärpauca dakṣiṇāke. ñi ka ṣ tāṃ oṅkorñai pintwāt petes. 14 toy weñāreneś sā oṅkorño tañ śwālya mā ste. pilycalñene lalālu laukito rṣāke tākaṃ rṣākeṃne śpālmeṃ cwi wes tā oṅkorñai pintwāt aiskem. 15 Upagentse mañu kärstātene. ṣañ ytāri masa. 16 Nānda cāla oṅkorñai Nandābala tāy ṣerśka postäṃ msāne. maitar yopar warttone dakṣiṇākeṃ ritasi wrocce rṣākeṃ.
13 The mendicant Upaga went up to the sisters (and) said to them: "Sisters, I have left the house a worthy recipient. Therefore give me this porridge in charity." 14 They said to him, "This porridge is not to be eaten by you. Should there be a stranger, a sage having exerted himself in austerity, (who is) best among sages, to him we will give this porridge in charity." 15 Upaga's hopes were dashed. He went on his way. 16 Nanda lifted the porridge (and) her sister Nandabala went after her. They went (and) entered the forest to seek a great sage worthy (of reward).
In a rare burst of simplicity, the pared-down paradigm of the Tocharian relative and interrogative pronouns allows our weary minds a brief respite from the labors of memorizing seemingly innumerable classes of all kind of words, along with their various exceptions. We forthwith present the paradigm for A kus-ne 'he who, that which, etc.'; A kus 'who? which?'; and B kuse 'who? which?; he who, that which'.
|A Relative||A Interrogative||B Interrogative|
We pause for a moment so that you may admire....
Tocharian B employs kuse and its forms as both a relative pronoun and as an interrogative pronoun. Specifically, as a relative it occurs in situations similar to the English 'who' and 'whom' in 'I saw the man who was walking towards me...,' 'The man whom you met...,' and the like; as an interrogative it functions similar to 'who' and 'whom' in 'Who was walking towards me?', 'Whom did you meet yesterday?', etc. Tocharian A does not employ exactly the same forms for both relative and interrogative, but mercifully forms the former from the latter by the regular appending of the suffix -ne. In both languages a single form serves to denote any grammatical (or biological) gender and any number.
Tocharian A kus B kuse bear a striking resemblance to interrogative and relative pronouns in other archaic Indo-European languages, leaving little doubt as to their PIE ancestor. The likely evolution follows the pattern of the chart below.
|Nominative||*kʷi-s + so||*kʷiso||*kwäsæ||kus||kusé||Hitt. kuis, Lat. quis|
|Oblique||*kʷi-to-m||*kwätæ||*kwäcæ||kuc||kucé||Hitt. kuin, Lat. quem|
|Genitive||*kʷoyyo-||*kwæiyæ||*kai||ke||Gk. poĩos, Lat. cuius|
In the nominative form *kʷi-s + so we actually have a collocation of the proper nominative to the stem *kʷi- followed by the nominative masculine of the common deictic pronoun *so (cf. Section 31, Lesson 7). The geminate (doubled) consonant *-ss- that ensued was evidently already simplified to *-s- within Proto-Indo-European itself. The oblique (PIE accusative) form *kʷi-to-m also displays the deictic stem *to-; but we notice that in the Proto-Tocharian period the dental *-t- undergoes palatalization to *-c-, evidently in accordance with a general tendency among pronominal elements to exhibit palatalization outside of the nominative case. The genitive forms appear to have different origins between the two languages; in either situation however it appears that the most likely origin in PIE is an adjectival stem, *kʷoyyo- or *kʷwo-.
The relative pronoun in the oblique, A kuc-ne B kuce, may play the role of a causal conjunction: 'on account of, since, because'. This parallels the Sanskrit usage of yad, the neuter accusative of the relative pronoun, which may also be used to introduce subordinate clauses with causal sense.
We also find in the Tocharian languages a rather rare interrogative pronoun, A äntsaṃ B intsu 'who? which?'. The following chart lists the extant forms.
|A Relative||A Interrogative||B Interrogative|
As with A kus B kuse, Tocharian A marks a distinction between the relative and interrogative pronouns by appending -ne to the former; Tocharian B however employs the same forms for both relative and interrogative. If we imagine the -t- to have arisen from secondary phonetic processes like those we have encountered in the historical phonology, then we see rather clearly that this pronoun derives from a nasal element A än- B in- followed by forms of the demonstrative pronoun A saṃ B su. The origin of the nasal element remains somewhat mysterious.
We find in Tocharian B exclusively one further pronoun mäksu. As with the others in Tocharian B, this is employed as both interrogative and relative. The following chart lists the extant forms.
|B Masculine||B Feminine||B Neuter|
This pronoun evidently derives from the composition of a number of elements. We see in the paradigm the forms of the demonstrative B su or seṃ. The origin of the leading element mäk- remains somewhat obscure. One likely possibility however (Pinault 2008) is that this too derives from composition, in particular of the same PIE *kʷi- encountered above and another element mä- perhaps related ultimately to PIE *mo- as found, e.g., in Hitt. mān, mahhan 'how?'
In practice, there is little difference in usage between B kuse and B mäksu. There does seem however to be a slight preference for the use of mäksu where Sanskrit employs the interrogative adjective Skt. katara-, katama- 'which one (of many)?'
Noun classes V and VI represent nouns with different nominative and oblique plurals, that is, nouns generally derived from original PIE animates. We find the following overarching division: Class V collects nouns whose plural reflects an alternation PIE *-i vs. *-ns in the nominative and accusative plural; Class VI by contrast shows an alternation PIE *-es vs. *-ns in nominative and accusative plural. The separation is not quite so clean, since loss of final consonants and subsequent contraction of vowels also means that many PIE *i- and *u-stems, with nominative plural in *-es, have also fallen into Class V; Class VI generally reflects original PIE *n-stem nouns. Each class contains further subdivisions based on the particular stem shapes encountered and their ensuing reflexes. The following chart displays the endings characterizing the subdivision.
|Class||Subclass||A N. Pl.||B N. Pl.||Notes|
|V||1||-e||-i||Original thematic *-o-|
In each instance, the oblique plural derives from PIE *-ns, giving uniformly A -s B -ṃ. The vowel preceding the final nasal in Tocharian B generally replicates the vowel of the nominative plural ending; therefore practically speaking the accusative plural form is superfluous for the purposes of classification and we may simply list the nominative plural as in the chart above. For a discussion of the historical evolution of these endings, see Section 27.1 in Lesson 6.
The chart below illustrates the declension of Class V.1 with the forms of the nouns A oṅk B eṅkwe masc. 'man' and A yuk B yakwe masc. 'horse'. The former illustrates the declension of nouns denoting humans, the latter non-human nouns. We have already seen the majority of these forms in Section 12.2 of Lesson 3; the salient forms are repeated here for convenience.
|A V.1||B V.1||A V.1||B V.1|
Note that with human nouns we generally do not see use of the instrumental case in Tocharian A.
The nouns A mañ B meñe masc. 'moon', A lyäk B lyak masc. 'thief', and A koṃ B kauṃ masc. 'day' illustrate the declension of classes V.2 and V.3.
|A V.2||B V.2||A V.3||B V.3||A V.3||B V.3|
The traditional classification into the above subgroups on synchronic grounds generally depends on the nominative and oblique singular in Tocharian B. In particular, Class V.2 shows B -e in the nominative singular and -Ø in the oblique; meanwhile Class V.3 shows -Ø in both the nominative and oblique singular.
Below we illustrate the declensions VI.1 and VI.2 with the nouns A ri B riye fem. 'city', A pyāpi B pyāpyo fem. 'flower', and A ak B ek alt. 'eye'. In terms of the classification into subclasses 1-4, it is enough to distinguish based on the nominative plural forms. Each of these subclasses does, however, have further subdivisions based on the forms of the singular. We need not focus on such fine distinctions for the moment.
|A VI.1||B VI.1||A VI.2||B VI.2||A VI.2||B VI.2|
Note that A ak B ek exhibits paral forms; in particular, we only find paral forms --- no plural --- for this word outside of the singular in Tocharian A. The traditional distinction between the above paradigms of subclass VI.2 is based synchronically on the forms of the nominative and oblique singular within Tocharian B, either nominative and oblique in -o and -ai respectively, or -Ø in both forms.
Finally we illustrate the subclasses VI.3 and VI.4 with the nouns A käntu (masc. or fem.) B kantwo (masc.) 'tongue', A āre 'plough', and B saswe masc. 'man'.
|A VI.3||B VI.3||A VI.4||B VI.4|
Tocharian possesses a further special verbal adjective known as the privative. As the name suggests (Lat. prīvāre 'rob, deprive'), this verbal adjective has a connotation of negation or lack. In particular, both Tocharian languages employ this construction in any of three basic senses: specifically the privative may denote
The first sense merely relates factual information from the point of view of the speaker, similar in sense to English "(left) undone" built to the verb "do". The latter two constitute assertions of a sort: the former maintains a connotation somewhat like a negated Gerundive I, with a sense similar to English "un-doable"; the latter takes something of the sense of a negated Gerundive II, similar to English "not to be done".
The formation of the privative involves addition of the suffix A -t B -tte, as well as a negative prefix A a(n)- B e(n)-. The construction is exceedingly rare in Tocharian A, leaving only a handful of isolated forms (here listed with their Tocharian B equivalents): A atäṅkät B etaṅkätte 'unhindered', A apälkāt B empalkaitte (empālkatte) 'unconcerned', A asinät B ontsoytte 'insatiable'. In Tocharian B the privative remains a productive construction. We list below the privative forms for the root B yām- 'do, make'.
|Privative||B Masc.||B Fem.|
The privative B ayāmätte has the sense 'not to be done', similar to Latin nōn faciendus. For the declension, compare the Class I derived adjectives in B -tstse (Section 13.2, Lesson 3).
We find similar constructions in the other early Indo-European languages. For example, take Gk. ámbrotos and Skt. amṛto-, both 'immortal', from PIE *ṇ-mṛ-to-, an adjective in *-to- built to the verbal root (here *mṛ-) and accompanied by the negative prefix *ṇ-. It will therefore come as little surprise that the Tocharian privative likewise appears to derive from an adjective based on PIE *-to- and adjoined to the negative prefix *ṇ-. The evolution of the prefix provides little surprise: PIE *ṇ- > PToch *æ(n)- > A a(n)- B e(n)-. We do however find variation, typically as the result of umlaut triggered by the following (root) vowel, or by the following (root-initial) consonant cluster. In particular we find the evolution outlined in the following chart.
|*ṇ-CV-||*æ(n)-CV-||*ā(n)-CV-||a(n)-||when V is PToch *ā, *ai or *au|
|*ṇ-CV-||*æ(n)-CV-||*o(n)-CV-||o(n)-||when V is PToch *o or when PToch *nC > *mp|
Thus we find in Tocharian B the prefix a(n)- when the root vowel or diphthong causes *a-umlaut; and we find o(n)- when either the root vowel causes *o-umlaut or the combination of *æ(n)- with the following root-initial consonant results in the cluster -mp-.
The exact evolution of the adjectival suffix is slightly more tricky. The privative suffix in the documented languages is A -t B -tte. The geminate -tt- in Tocharian B has led some scholars to suspect the origin lies in a PIE suffix *-two-. Two basic problems arise: on the one hand, Tocharian A shows no geminate, and on the other hand, the Tocharian A reflex should then be A *-tu. For this reason some scholars (cf. Pinault 2008) suggest the double -t- in Tocharian B likely represents scribal practice to ensure pronunciation of a true dental stop rather than a fricative between vowels. If this is indeed the correct interpretation, that would allow a reconstruction A -t B -tte < PToch *-tæ < PIE *-to-, leading back to the same suffix found in other IE languages.
Though the affixal structure *ṇ-...-to- appears to parallel that found in Gk. ámbrotos and Skt. amṛto-, the Tocharian privative does not add these affixes directly to the root as in Greek and Sanskrit. Rather it seems that a likely origin for the privative stem lies with the subjunctive. Note the following examples.
|B klautk-||klautkotär (IV)||klautkaṃ (V)||aṅklautkatte||turn|
|B käm-||känmaṣṣäṃ (IX)||śämt (II)||ekamätte||come|
The subjunctive stem shows clearly in the privative of the root B klautk- 'turn'. With B käm-, however, it seems that in particular the privative stem must derive from the weak stem of the subjunctive, here B *kämä- rather than B *śämä-: PToch *æ(n)-kämä-tæ > B ekamätte. But in this particular example we might just as well find a construction built directly to the root (Pinault 2008), parallel to what we find elsewhere in IE: PIE *ṇ-gʷṃ-to- > PToch *æ(n)-kʷäm-tæ- > B ekamätte, though this requires the insertion of an anaptyctic vowel.
The Tocharian imperative does not exhibit any distinctive stem, but rather employs either a preterite or subjunctive stem. The imperative does however show a new set of personal endings. These endings only include forms for the second person singular, dual, and plural. There is however one isolated form, päklyossū, which appears to denote a third person singular imperative. The following chart lists the imperative endings for Tocharian A and B.
In addition to the above endings, the Tocharian imperative also employs a prefix AB p- whose vocalism varies depending on the root. In particular, though the prefix always appears in Tocharian A, it regularly falls away in Tocharian B when the root begins with p-: e.g. AB pärk- 'request' shows second person plural mediopassive imperative PToch *pä-pärksā-c > A ppärksāc B parksat. We also commonly find simplification of the ensuing consonant cluster in Tocharian B: tāka for ptāka, second person singular active imperative built to the suppletive past stem tāk- of the verb nes- 'be'. In Tocharian A, the prefix appears as pu- before consonants which we historically labiovelars: PToch *pä-kʷäm-äs > A pukmäs 'come you (all)!', from A kum- 'come'. In Tocharian B, we find pe- in the singular imperative pete, built to the root B ai- 'give'. We also find po- in a handful of verbs: for example B pokse < B āks- 'show', B pokkāka < B kāk- 'call', B poñ < B weñ- 'speak'. The conditions which govern the vowel quality of the prefix remain somewhat unclear. Only a small number of verbs display the consonant doubling we find in B pokkāka, from the root B kāk- 'call'.
The prefix may derive from the same source as the prefix po- found in Old Church Slavonic and pē- in Hittite. The assumption of an original PIE *po-, however, would require an evolution PIE *po- > PToch *pæ- > *pä-. This somewhat uncommon evolution does however find some support in a development *mo- > *mæ- > *mä- perhaps found with some pronouns and adverbs.
Tocharian displays five basic classes of imperative formation, with an additional class comprising those verbs with so-called irregular formations. In essence the imperative classes recapitulate the preterite classes, employing the same stem formations. The following chart displays the basic imperative class distinctions alongside the corresponding preterite formations.
|Class||Stem Shape||A Pret.||B Pret.||A Impt.||B Impt.||Comparanda|
|i||PIE *-H-||śärs||śarsa||päkras||pkārsa||Hitt. kars-, Lat. scīre|
|ii||PIE *CV-CVC, CēC||kakäl||kālaṃ||pkäl||Lat. colere, Skt. cárati|
|iii||PIE *-s||arsāt||ersate||parsār||persat||Gk. õrsa, Skt. ṛṇóti|
|iv||PToch *-ṣṣā-||kākätkṣuräṣ||kakātkäṣṣu||pkātkäṣṣat||Gk. gētʰéō|
|v||PToch *-ñ(ñ)-||weñār||weñāre||poñ||Gk. eĩpon, Lat. vōx|
Generally speaking, a verb will take the same class in the imperative that it takes in the preterite, but there are several exceptions where a verb's preterite belongs to one class but its imperative to another. The following chart provides some examples.
|A ken-/kāk-||kāk (I)||pukāks-äṃ (i)||call|
|B kwā-/kāk-||kāka (I)||pokkāka (i)||call|
|A kärs-||krasaṣ (v)||śärs (I)||päkras (i)||know|
|B kärs-||kārsaṃ (v)||śarsa (I)||pkārsa (i)||know|
|A ṣäm-/läm-||lamaṣ (v)||lyäm (I)||pälmäs (i)||sit|
|B ṣäm-/läm-||lāmaṃ (v)||lyama (I)||plāma (i)||sit|
|A käl-||kakäl (II)||pkäl (ii)||endure|
|A wätk-||wotäk (II)||putäk (ii)||order|
|B wätk-||yātka (II)||pitka (ii)||order|
|B käl-||kelu (i)||keltsa (III)||pkel (iii)||endure|
|A ar-||aräñtär (vii)||arsāt (III)||parsār (iii)||evoke|
|B er-||ertar (i)||ersate (III)||persat (iii)||evoke|
|A yām-||yāmam (ii)||yāmäs (III)||pyāmtsār (iii)||make|
|B yām-||yāmäṃ (i)||yamaṣṣa (IV)||pyāmtsar (iii)||make|
|A ṣäm-/läm- (caus.)||lmāṣiṣ (ix, Opt.)||lyalyäm (II)||pälmāṣar (iv)||set|
|B kätk- (caus.)||katkäs(t)si (ix)||śatkātai (II)||katkäṣṣar (iv)||have proceed|
|A āks-||ākṣiññim (xii, Opt.)||ākṣiññā (V)||pākṣiññā-ñi (v)||proclaim|
|B taṅkw-||täṅkwaṃ (xii)||pätaṅkwaññe (v)||love|
|B āks-||ākṣäṃ (ii)||ākṣa (I)||pokse (vi)||proclaim|
We notice in the above chart that occasionally the imperative form bears a closer resemblance to the corresponding subjunctive than to the corresponding preterite. This is particularly true of verbs of imperative Class i, such as the stem AB kärs- 'know' shown above. Recall from the discussion of Section 34.1, Lesson 7, that Class I preterites exhibit the results of an original PIE alternation between *e-grade in the active singular forms, and *Ø-grade elsewhere. This in particular leads to the palatalization of the root-initial consonant *ke- > PToch *śä-. The imperative however lacks this palatalization altogether. If we recall on the other hand the discussion of Section 29.2.2, Lesson 6, we note that CLASS v subjunctives generally derive from verbs with root-final laryngeals (the same source as the PToch *ā-marker of Class I preterites); moreover this stem-final *-ā- causes umlaut of a preceding PToch *æ < *o, yielding PToch root vowel *ā. Therefore CLASS v subjunctives show an ablaut pattern of PToch *-ā- vs. *-ä-, with no attendant palatalization of the root-initial consonants. This is precisely the situation we find in the paradigm of certain Class i imperatives like AB kärs-. Another such verb is PIE *TerK- > AB tärk- 'let go'. We place the subjunctive (Section 29.2.2) and preterite (Section 34.1) paradigms of this verb alongside the imperative paradigm for comparison.
|A Subj. v||B Subj. v||A Pret. I||B Pret. I||A Impt. i||B Impt. i|
One interesting feature of the relation between imperative and subjunctive is that, even when the imperative does not derive from the preterite stem, it still derives from another stem (subjunctive) with perfective aspect. This provides a nice counterpoint to the prefixation with *pä-, if in fact derived from PIE *po-, since the latter figures prominently in perfective verbal aspect among the Slavic languages.
Thus we see that Class i imperatives are generally associated with verbs exhibiting Class I preterites or CLASS v subjunctives. The ablaut pattern of the root follows that of the CLASS v subjunctives. Class ii imperatives, in like fashion, are typically associated with verbs showing Class II preterites. In Tocharian B these typically display second person singular active ending -a after a root with short root vowel; the ending is lacking when the root vowel is long. Class iii imperatives show the ṣ-suffix characteristic of the Class III preterites, though this suffix only appears regularly in the mediopassive forms. Class iv imperatives are all causatives. They display the ṣṣ-suffix (-āṣ- in Tocharian A) appropriate to either the Class IV preterites or the CLASS ix subjunctives (or, equivalently, CLASS IX presents). The verbs A läm- 'sit', B kätk- 'proceed', B mäsk- 'exchange', and B tsälp- 'be free (of)' form Class iv imperatives, in natural association with their CLASS ix (causative) subjunctives, even though they have Class II preterites. Imperative Class v, containing few verbs, shows the suffix -ññ-, generally in line with associated preterite Class V or subjunctive CLASS xii.
We show below imperative forms of the verb AB käl- 'lead, bring' to illustrate the Class i conjugation. Alongside these forms we display those of A klyos- B klyaus- 'hear', which shows palatalization of the root-final consonant.
|Imperative i||A i.1||B i.1||A i.2||B i.2|
Most imperatives follow the same basic pattern, the main difference being the suffix added to the root, which depends on the particular imperative class. As a further example, we list below the imperative forms of AB yām- 'make'.
Class vi consists of a small number of verbs that show imperative forms differing from the patterns discussed above. We collect them in this section. Certain verbs pertain to the sixth class of imperatives in both Tocharian languages. The following chart lists their extant forms.
|A i-||B i-||A e-||B ai-||B täl-|
|2 Pl.||pic(äs)||pcīso (cisso)||pac||petso (petes)|
We do not find mediopassive forms for the above verbs. Other verbs, by contrast, show irregular imperatives in Tocharian B, but regular imperatives in Tocharian A. The following chart displays the Class vi forms in Tocharian B alongside the corresponding forms for the cognate verb in Tocharian A.
|A āks- (v)||B āks-||A lä-n-t- (i)||B lä-n-t-||A we- (i)||B we-|
|2 Pl.||pokses||pälcäs||platstso||penäs||poñes (pontso)|
|2 Sg.||plyatstar (caus.)|
We have scattered throughout these lessons various comments pertaining to verbal aspect as exhibited in Tocharian. We collect here some of these observations to put them in clearer perspective.
Recall that linguists distinguish two basic features of a verb: tense and aspect. Tense refers to when an action occurs relative to the speaker: before (past), simultaneous with (present), or after (future) the moment of utterance. Aspect, by contrast, refers to how the speaker conceives of an action: completed (perfective), ongoing (imperfective or durative), denoting a state of existence (stative). Where tense and aspect exist in a linguistic system, tense is typically marked morphologically; aspect may be marked by the semantics of the verb itself, or by morphological means, or by a combination of the two.
Tense and aspect are independent characterizations of a verb, that is, any aspect may occur in any tense, at least in principle: past perfective, present perfective, past imperfective, present stative, etc. In practice, however, certain aspects maintain certain connotations that may lead to certain co-occurrences of tense-aspect pairs: for example, completed (perfective) aspect generally prohibits an action from co-occurring with the entire duration of an utterance; the verbal notion marked as perfective therefore likely has already occurred or is as yet about to occur. In the event that it has already occurred, this can predispose the speaker to choosing past tense along with perfective aspect, or to allow perfective aspect alone to connote prior completion, in effect producing a past tense without the need to mark it so. Similarly, if the event has not yet occurred, the speaker may choose to associate the perfective aspect with future tense, or allow the perfective aspect alone to connote futurity. In this way we see how many languages make do solely with a morphological system of aspectual distinctions, with no overt tense marking: the connotations of the verbal aspect go a long way to placing an action in time relative to the speaker.
Within the Tocharian verbal system we may describe certain harbingers of a system of verbal aspect. The present tense generally expresses imperfective aspect. When the semantics of a verbal root inherently denote perfectivity (think of, e.g., arrive in English -- you either arrive or you don't, so it's inherently perfective; there is no ongoing, unfinished arrival), the present conjugation of the verb typically belongs to a class employing a suffix with durative connotations. In addition the Tocharian imperfect typically imparts imperfect aspect on a verb form: this largely follows as a result of Tocharian building the imperfect on the present verbal stem.
Perfective aspect likewise finds expression in two principal manners, but these remain more distinct due to the fact that they generally involve independent verbal stems. In particular, the Tocharian preterite denotes a past action with perfective aspect, while the Tocharian subjunctive denotes a present action with perfective aspect. This latter helps account for its frequent role as a simple future tense. Finally the Tocharian imperative evidently expresses perfective aspect; as with the imperfect tense, this seems to arise secondarily. Specifically, the Tocharian imperative takes its stem from the preterite, or occasionally the subjunctive, which themselves mark perfective aspect. The prefix typical of the imperative, with the Slavic languages in mind, likely bolsters this perfective aspect.
Consider a specific example of how such patterns of aspect and tense play out in Tocharian. In particular we may focus on CLASS vi subjunctives. As we noted in Lesson 6, Section 29.3, verbs with CLASS vi subjunctives (suffix *-nā-) generally show corresponding presents in CLASS X (suffix *-nā-sk-). Across the early Indo-European languages, the suffix *-sḱ- often retains an iterative sense. This suggests that, as *-nā- came to be associated with the subjunctive and perfective aspect, there arose among the verbs employing *-nā- in the subjunctive the need for a renewed imperfective formation. For this Tocharian co-opted the iterative *-sḱ- suffix and produced a durative, hence present, marker *-nāsk-.
We find in Tocharian that the aspectual distinction imparted by a particular verb class suffix depends substantially on the particular history of that verb stemming from Proto-Indo-European times. For instance the nasal infix, for its part, generally denotes durative or imperfective aspect in Indo-European. This likely underlies its appearance in CLASS VI presents (Lesson 5, Section 24.3). The fact that the verbs of this class historically employed in PIE an additional marker, *-n-, to elicit an imperfective or durative sense suggests that these verbs were in fact root-perfective or aoristic --- that is, the root without the infix denoted a perfective action in PIE terms. Recall from the discussion of tense and aspect in PIE from the Lesson 3 introduction that these roots therefore typically formed unmarked, or root, preterites in PIE itself. But root preterites evidently formed the backbone of the subsequent Class I preterites in Tocharian. We should therefore expect an association in Tocharian between nasal-infixed presents and root preterites, and this is what we find: verbs with CLASS VI (nasal infix) presents typically show Class I preterites (root-aorists). We similarly find that the root thematic subjunctives (CLASS ii) tend to employ suffixes *-s- or *-sk- in their corresponding present formations (cf. Lesson 6, Section 29.3.1).
Thus we find a rather intricate historical development that leads for the most part logically to the particular verb classes and class associations we find within Tocharian. We must however allow for occasional variation in what stem (present, subjunctive, or preterite) the Tocharian speakers feel to be 'dominant' or 'primary' for a given verb, or we should allow different stems of the same verb to 'dissociate' over time. For example, consider English ought: formerly the only past tense form of owe, its modal use so came to dominate its usage that it was no longer felt to be part of the same verbal complex as owe; the latter thereby developed a new past tense form owed. Returning to Tocharian, suppose for example that we take a PIE verb with root-final laryngeal, PIE *CXH-, where X represents any unknown element preceding the laryngeal. If the lexical semantics of the root are perfective, as with English arrive, then we might expect this root to form a root-preterite, and perhaps employ a nasal infix *CXnH- to create a durative present. Now, if the semantics stay the same within Tocharian, we subsequently expect a root-preterite formation (Class I) and a nasal infix present (CLASS VI). Thus the reflex PToch *-nā- of the final *-nH- of the root logically becomes associated with present durativity in Proto-Tocharian. On the other hand, the new durative stem PIE *CXnH- could be taken as primary, or dissociated, within Proto-Tocharian. If the verb nevertheless maintains the same semantics (lexical perfectivity), then *-nH- > PToch *-nā- will become associated with perfective aspect, and Tocharian will seek a new, additional durative marker. Thus we might expect a nasal infix subjunctive (CLASS vi) and a present with a nasal compound suffix (CLASS X).
In this way we may start to make sense of the particular associations we find between different classes of Tocharian present, subjunctive and preterite formation. Though English he owes denotes the present and formerly he ought simply denoted the past tense of the same utterance (equivalent to he owed), ought subsequently shifted categories and currently only stands as a modal denoting a (tenseless) obligation (he ought = he should). In Tocharian we find similar category shifts between present, subjunctive and preterite. And as verbs of similar phonological and morphological shape undergo the same shifts, we find that certain PIE markers like *-n- which had a unique valence (e.g. durative) in PIE now have a plurality of valences (e.g. present and durative, or subjunctive and perfective) within Tocharian.