Proto-Indo-European demonstrates a rich array of nominal morphology. PIE inflects nouns for gender, case, and number. Three numbers arise in PIE: singular, dual, and plural. We find evidence for eight cases in PIE -- nominative, accusative, instrumental, dative, ablative, genitive, locative, vocative -- whose significance has been discussed in Lesson 2. We discuss below further points of gender and case, as well as other important features of nouns in Proto-Indo-European.
In general ancient IE languages display three genders: masculine, feminine and neuter. As mentioned in the first lesson, the term gender is really only a shorthand for a system of agreement between nouns and adjectives, ensuring that the listener understands which adjective modifies which noun. As such, the general term 'gender', as well as the individual groupings 'masculine', 'feminine' and 'neuter', are misleading in that they have nothing in principle to do with biological gender. As it happens, however, when a noun refers to something that does possess a biological gender, those that have male gender are generally represented by nouns with masculine grammatical gender, and females with the feminine gender. But this division is not at all strict: Latin nauta 'sailor' generally denotes a male, even though it declines according to the pattern of feminine nouns; German Sonne and Old English sunne 'sun' decline as feminine nouns, even though the sun has no biological gender of any sort.
Evidence from Anatolian, in particular Hittite, suggests that the three-way split between masculine, feminine and neuter may not represent the earliest state of affairs in PIE. In particular, Hittite only shows a two-fold grammatical gender system: masculine and neuter in the terminology above. The particular distribution of these grammatical genders among Hittite nouns suggests that 'masculine' really denotes animate nouns, while 'neuter' denotes inanimate. The origin of the later 'feminine' gender in PIE thus remains somewhat of a mystery, but some evidence points to its development out of the neuter plural, reinterpreted as a singular collective.
One remarkable feature of PIE morphology in general is that, at the most basic level, noun structure parallels that of verb structure. By that I mean that any given noun may be decomposed into
just like any verb. Take for example Gk. géneos 'of a kind', from earlier *géne(s)os:
The nominal suffixes and endings generally do not overlap with those of the verbal morphology (we do need to distinguish verbs and nouns, after all! Or do we...?), but amazingly enough the nominal roots very often do overlap with verbal roots. That is, we find generally that IE languages have ways of employing the same root as either a noun or a verb. In English we experience this daily with such alternations as sing (verb) versus song (noun), clearly derived from the same root by a morphophonetic process (ablaut) that allows a formal change between categories. Equally, in English we may cross categories simply by verbing a noun, or nouning a verb... that is, we need not apply any special process to the root to convert from one category to another.
We also saw that the verbal suffix can be thematic or athematic, i.e. it can contain or not a thematic vowel *-e- or *-o-. Remarkably, the same holds true for PIE nouns: suffixes can be either thematic or athematic.
Any of the root, suffix, or ending could contain a vowel in *Ø-, full-, or long-grade. The grade often shows a curious interplay with the accent pattern of the noun throughout its declensional paradigm.To understand this interplay, we should first distinguish between two types of cases: strong cases and weak cases. Here strong and weak bear no relation to similar uses in describing the declension of adjectives or conjugation of verbs; it simply seems that, ironically, linguists are loathe to develop new terminology when simple misapplication of arcane terminology will suffice. At any rate, we have the following pattern of strong and weak cases:
All unmarked cases in the above chart are weak. We see that, properly, the strong vs. weak distinction is one of case and number, since scholars generally agree that the accusative was not strong in all numbers. The importance with regard to the nominal accent in PIE derives from the following occurrence:
Here 'to the right' should be understood as shorthand for 'toward the end of the word'. Of course, where the full grade of the vowel and where the accent wind up in the weak cases each depend on two factors:
Though this should lead in principle to four independent parameters (location of full vowel, location of accent, jump of full vowel, jump of accent), interestingly we find that the two tend to pattern together. That is, the accent tends to remain over the full vowel, so that the location of one dictates the location of the other, and similarly for the jump. Given these parameters and the correlation between them, we may therefore classify nouns into a few basic types: acrostatic, mesostatic, proterokinetic, hysterokinetic, amphikinetic (or holokinetic). We describe their accent and vowel patterns in the chart below.
In the above, R, S, and E naturally stand for root, suffix, and ending, respectively. The boldface letter denotes which of these three elements carries the full vowel and accent in that particular type of case. Below we find examples of each pattern, employing the nominative as the exemplar of the strong cases, the genitive for the weak. Though other full vowel and accent patterns could occur in principle, linguistic reconstruction has generally isolated the above as the dominant patterns in PIE nominal declension. Moreover, PIE speakers could form new nouns merely by shifting the accent of a given noun through a process known as internal derivation
Proto-Indo-European maintains a dizzying array of possible nominal suffixes. We list below some of the more productive suffixes:
To the suffix Proto-Indo-European adds endings to denote case. These case endings determine the grammatical function of a given noun in an utterance. Reconstruction has uncovered some differences between the endings employed for thematic and athematic nouns. In principle, we should also find different endings for the masculine, feminine, and neuter genders. But as mentioned above, we find that the feminine gender seems to be a later innovation, so that we are left in the earliest stages of PIE with a simple distinction between animate (masculine) and inanimate (neuter). Moreover, masculine and neuter agree in all cases other than nominative, accusative, and vocative. Thus, the actual inventory of desinences is greatly reduced. The following chart displays side-by-side the thematic and athematic endings for both animate and inanimate genders in all numbers (cf. Fortson 2004).
|Athem. Sg.||Them. Sg.||Athem. Du.||Them. Du.||Athem. Pl.||Them. Pl.|
|Nom. anim.||*-s||*-o-s||*-H₁||*-ō < *-o-H₁||*-es||*-ōs < *-o-es|
|Voc.||*-Ø||*-e||*-H₁||*-ō < *-o-H₁||*-es||*-ōs < *-o-es|
|Acc. anim.||*-m||*-o-m||*-H₁||*-ō < *-o-H₁||*-ns||*-o-ns|
|Nom./Acc. neut.||*-Ø||*-o-m||*-iH₁||*-o-iH₁||*-H₂||*-ā < *-e-H₂|
|Abl.||*-s||*-ōt < *-o-H₂at||*-bʰ-||*-o(i)-bʰ-|
|Dat.||*-ei||*-ōi < *-o-ei||*-bʰ-||*-o(i)-bʰ-|
|Instr.||*-H₁||*-ō < *-o-H₁||*-bʰ-||*-o(i)-bʰ-|
We see in the above chart that scholars have difficulty reconstructing the weak cases of the dual. In the other forms, however, we notice an incredible similarity between thematic and athematic endings: the thematic ending for a given case and number echoes its athematic counterpart, but with the insertion of the thematic vowel. Some scholars also take the nominative/accusative plural form of the neuter, *-ā < *-e-H₂, as the point of departure for the feminine gender: the PIE speakers evidently viewed this form as a collective, and hence singular, in regard to some nouns. The above chart does not include all forms reconstructed for PIE nominals, but rather only a selection of the most pervasive.
In the end, the form encountered for any given noun derives from some root, one of the suffixes of the previous section, and one of the endings listed in the chart above. We of course do not see the above chart reflected exactly in any of the ancient attested daughter languages. As in Tocharian, cases tend to fall away or be added, and as this happens certain endings may be reinterpreted and recast as endings for another case.
Even a cursory glance across the ancient Indo-European languages shows an obvious affinity among the pronouns. The details of this affinity, however, are often difficult to piece together. One thing is certain: the pronouns followed a declensional pattern that differed somewhat from that employed with other nouns and adjectives. Some of the endings are clear for certain types of pronouns, but other remain rather elusive.
The personal pronouns evidently showed differing forms depending on whether they retained their own accent in an utterance or whether they borrowed it, so to speak, from a neighboring word (the so-called enclitic usage). These pronouns only distinguish person, case, and number; they do not distinguish grammatical gender in PIE. In addition to the first ('I', 'we') and second ('thou', 'you all') person pronouns, we also find reason to reconstruct a reflexive pronoun (roughly equivalent to '-self' in English usage of 'himself', etc.). The following chart outlines the PIE forms of these pronouns (cf. Adams 1988).
|1st Pers.||2nd Pers.||Refl.|
|Nominative||*weys, weyom||*yuHs, yuHom|
Many of the above forms continue to be debated, and other forms are too uncertain to even be listed.
Proto-Indo-European also possesses a number of deictic (pointing) pronouns, akin to English 'this' or 'that'. One in particular finds employment as a third person pronoun ('he', 'she', 'it') built to the stems *so- and *to-. Though some of the details of the paradigm remain obscure, several PIE forms may be reconstructed with relative confidence.
Following a similar declension pattern is the interrogative ('who', 'what') pronoun with stem *kʷo-. This pronoun also served as an indefinite pronoun ('someone', 'anyone').
The following selection continues the Tocharian A text A255 (THT 888). The narrative continues with a list of the deeds of past Buddhas, their lifespans, and when they attained Nirvana.
29 - ṣäk-tmāṃ puklā wrasaśśi śolaṃ Viśvabʰū ñomā ptāñkät ṣeṣ.
säm penu -- -- -- -- -- tmāṃ-tri-wälts-puklyi puttiśparäṃ kälpāt.
tri-tmāṃ päñ-wälts puklā puttiśparṣināṃ wles wleṣāt.
tmāṃ we-wälts puklā śol lyalyipuräṣ ksaluneyaṃ kälk.
30 - śtwar-tmāṃ puklā wrasaśśi śolaṃ Krakasundi ñomā ptāñkät ṣeṣ.
säm penu kāsu kälko tmāṃ-puklyi puttiśparäṃ kälpāt.
we-tmāṃ we-wälts puklā puttiśparäṃ wleṣāt.
okät-wälts puklā śol lyalyipuräṣ ksaluneyaṃ kälk.
29 ṣäk-tmāṃ puklā wrasaśśi śolaṃ Viśvabʰū ñomā ptāñkät ṣeṣ. säm penu -- -- -- -- -- tmāṃ-tri-wälts-puklyi puttiśparäṃ kälpāt. tri-tmāṃ päñ-wälts puklā puttiśparṣināṃ wles wleṣāt. tmāṃ we-wälts puklā śol lyalyipuräṣ ksaluneyaṃ kälk. 30 śtwar-tmāṃ puklā wrasaśśi śolaṃ Krakasundi ñomā ptāñkät ṣeṣ. säm penu kāsu kälko tmāṃ-puklyi puttiśparäṃ kälpāt. we-tmāṃ we-wälts puklā puttiśparäṃ wleṣāt. okät-wälts puklā śol lyalyipuräṣ ksaluneyaṃ kälk.
29 For sixty thousand years in the life of beings there was a Buddhalord Visvabhu by name. This ... (man) as well, possessing thirteen thousand years, attained Buddhahood. For thirty-five thousand years he performed the service of Buddhahood. For twelve thousand years, having given his life, he attained Nirvana. 30 For forty thousand years in the life of beings there was a Buddhalord Krakasundi by name. This well gone (man) as well, possessing ten thousand years, attained Buddhahood. For twenty-two thousand years he performed Buddhahood. After eight thousand years, having given his life, he attained Nirvana.
The development of the PIE consonants in Tocharian provides a story no less interesting than that of the PIE vowels. The development of the consonants, in fact, revealed one of the most stunning early discoveries in the study of Tocharian: that the Tocharian languages lie within the Centum group, not within the Satem group of their geographic neighbors, and therefore undermine earlier notions of a simple geographic correlation of the Centum-Satem distinction.
Aficionados of smaller phonetic inventories will happily note that the Tocharian group largely streamlines the consonantal inventory of Proto-Indo-European. In particular, voicing and aspiration each lose their phonemic distinction, and so PIE consonants such as *b and *bʰ both become leveled with *p. One must keep in mind, however, that this denotes phonemic rather than phonetic leveling. That is, we cannot say whether the Proto-Tocharian *p into which all of *p, *b and *bʰ collapsed was pronounced with or without, say, aspiration; rather, we can only say that the distinction no longer remained important, as in English pot, where it makes no difference for understanding whether the initial p is deaspirated or not -- though it sounds a little unusual to a native speaker if it is.
Concomitant with this pruning of the PIE consonantal inventory, we find in Tocharian the rise of a process of palatalization which subsequently introduces some new phonemes into the inventory. This palatalization is distinct from that observed in the Satem languages and should be understood as a solely Tocharian process. Interestingly, we see Tocharian regularize the subsequent phonemic alternations that arise in certain common morphological paradigms and extend them beyond their original boundaries, in effect creating in certain paradigms a morphological palatalization.
We take as our starting point the Proto-Indo-European consonant inventory. The general consensus among scholars settles on the following system.
The association of the laryngeals *H with particular points of articulation leaves something to conjecture, and the above merely represents one possibility which has found some support among scholars (see, e.g., Ringe 1996). We include the laryngeals in the above chart for completeness, though we will treat their development separately in the following lesson.
Reconstruction of the Proto-Tocharian consonant inventory leaves us with the system outlined in the following chart.
The above chart depicts quite clearly the reduced number of stop consonants. Similarly we find the introduction of some palatal consonants, as well as palatalized variants of other, non-palatal consonants. For a given consonant C we denote by C' its palatalized variant.
In broad outline, the development from the PIE consonantal system to that of Proto-Tocharian follows from a few major principles:
We discuss each one of these processes in turn.
16.1.1 Merger of PIE Palatals and Velars
The first major principle, mentioned above, concerns the merger of the PIE palatal and velar stops, lining Tocharian up with the other Centum languages:
|PIE *Ḱ > *K,|
where *Ḱ represents any of *ḱ, *ǵ, *ǵʰ, and similarly *K represents any of *k, *g, *gʰ. A large swath of the ancient Indo-European languages share this merger, and this would lead us to suspect that this denotes a shared dialectal innovation within a subgrouping of the speakers of the parent language Proto-Indo-European. Though this likely contains some germ of truth, the actual state of affairs nevertheless remains somewhat murky. For example, we will see with Grassmann's Law below that Tocharian shares innovations with both Greek and Sanskrit, the latter belonging to the Satem group.
Tocharian naturally exhibits numerous examples of this change. Consider the following.
|*Ḱ||*K||*ḱṇtóm||*käntæ||A känt||Lat. centum|
|'hundred'||B kánte||Gk. hekatón|
|*ṇ-ǵneH₃-tiH₂||*ā̀knā́tsā̀||A āknāts||Lat. ignōtus|
|'not knowing, foolish'||B aknā́tsa|
|*ǵómbʰo-||*kæmβæ||A kam||Skt. jambʰa-|
|'tooth'||B keme||Gk. gómpʰos|
16.1.2 Grassmann's Law
Tocharian joins Greek and Sanskrit in a very select club among the ancient Indo-European languages which shows the operation of Grassmann's Law. According to this law, when successive syllables begin with aspirated consonants Cʰ, the consonant beginning the first syllable loses its aspiration:
|PIE *CʰVCʰ- > *CVCʰ-.|
In the case of Greek and Sanskrit, this seems to be an innovation produced independently in each language. Consider the following examples.
Tocharian too shows examples of this principle at work. Consider the examples listed below.
|B tsaik-||Lat. fingō|
|*dʰegʷʰ-||*dgʷʰ-||*dzk-||A tsäk-||Skt. dah-|
|(Ø-grade)||> *dzäk-||B tsäk-||Lat. foveō|
Evidently in Tocharian, as in Greek and Sanskrit, this is an independent development.
Palatalization provides one of the dominant phonological processes in the historical evolution of the Tocharian system of consonants. Through this process a consonant C acquires a palatal off-glide, denoted C'. Many and varied languages undergo a similar process. For example, we find this process at work in the history of English itself with the modern pair Eng. kin vs. chin: we find OE cynn > Eng. kin, but OE cinn > Eng. chin (incidentally, the development from PIE is *genH₁- > *gnH₁-ieH₁ > PGmc *kun-jo > OE cynn > Eng. kin, but *genu- > kinnuz > OE cinn > Eng. chin). In the latter, the front vowel i palatalizes the c, ultimately resulting in the modern sound represented by ch; in the former, this process is prevented by the back quality of OE -y- ([ü], equivalent to Modern German ü). We find the same phenomenon acting on a synchronic level when speakers of English pronounce Got you! quickly as Gotcha!
We may summarize Tocharian palatalization symbolically as follows:
|*C > *C' before *e, *ē, *y, and sometimes *i, *ī.|
That is, a consonant palatalizes before an immediately following front vowel or glide, except that *i, *ī only affect certain types of consonants. In particular, *i, *ī fail to palatalize bilabial, dorsal (formed with the blade of the tongue), and labiovelar consonants, as well as the sibilant *s. See the discussion of the historical evolution of PIE *i for additional discussion. Further subtlety lies in which consonants form palatalized/non-palatalized pairs. In particular, we have the following correspondences.
|PIE||PToch Non-Palatalized||PToch Palatalized|
In the above we see that, for the most part, the particular palatalized reflex only depends on the Proto-Tocharian consonant. But notably we observe that palatalization does not treat the Indo-European dentals equally. Specifically, *d follows a treatment different from the other dentals, and moreover the palatalized reflex of dentals depends on whether or not the glide *y triggers the palatalization. We will discuss the evolution of the dentals further below.
Note also that *s has palatalized partner *ṣ, not *ś. The latter pairs with the velars, labiovelars, and affricate *ts; moreover it provides the reflex of PIE *t, *dʰ when palatalized by a following *y.
The following examples illustrate some of the above correspondences.
|*t||*c||*dʰugH₂tēr||*täkācær||A ckācar||Ved. duhitár-|
|'daughter'||B tkā́cer||Gk. tʰugátēr|
|'light'||B lyuke||Ved. rócas|
|'queen'||> *wlāntsā||B lāntsa||OIr. flaitʰ|
Deaspiration formally denotes the process by which an aspirated consonant loses the accompanying puff of air, leaving an unaspirated consonant. In Grassmann's Law, treated above, we see that at times in the course of historical evolution this process acts by changing an aspirated phoneme for its unaspirated counterpart. In Tocharian, however, the application of deaspiration was far more dramatic, ultimately removing all aspirated consonants from the Proto-Tocharian inventory. The result, therefore, is a system wherein no unaspirated consonant has an aspirated counterpart, and vice versa. Thus, in the context of Tocharian, documentary evidence only allows us to conclude that, after a certain point, aspiration no longer had any phonetic value. We cannot to date say whether or not there may have been allophonic variation in aspiration of the sort found in English: e.g. (aspirated) pot, but (unaspirated) spot.
On the phonemic level, however, way may state the rule simply:
|all PIE aspirated consonants lost their aspiration in Proto-Tocharian.|
The Tocharian languages display no phonemic distinction based on aspiration.
Just as drastically as deaspiration in Tocharian is the general devoicing that occurred in Proto-Tocharian:
|all PIE voiced stops lost their voicing in Proto-Tocharian.|
This of course carries the same caveat as that mentioned in the context of deaspiration: the loss of voicing applies on the phonemic level, though it may not apply strictly on the phonetic level. This rule applies only to the stop consonants; resonants, for example, retained their voicing. Combined with the above rule concerning deaspiration, this leads to a dramatic reduction in the stop consonant inventory of Proto-Tocharian. Consider the following examples.
|*G||*K||*dʰugH₂tēr||*täkācær||A ckācar||Ved. duhitár-|
|'daughter'||B tkā́cer||Gk. tʰugátēr|
|*dʰgʷʰ-||*dzk-||A tsäk-||Skt. dah-|
|> *dgʷʰ-||> *tsäk-||B tsäk-||Lat. foveō|
In the above, *G represents a general voiced stop in PIE, while *K represents the corresponding unvoiced consonant in PToch.
One of the great surprises in the historical phonology of Tocharian pertains to the treatment of the dental series PIE *t, *d, *dʰ. In particular, scholars have discovered a somewhat non-uniform development, wherein PIE *d is singled out from the group and undergoes changes different from PIE *t, *dʰ. At some point early in the Proto-Tocharian period, PIE *d developed into a voiced affricate PToch *dz. This change occurs, surprisingly enough, in both palatal and non-palatal contexts. The resulting affricate subsequently obeys the laws of devoicing and palatalization, though the reflex of the latter differs from that of its brethren *t, dʰ. In particular, devoicing causes a change PToch *dz > *ts, and subsequent palatalization yields PToch *dz > *ts > *ś; by contrast, palatalization regularly derives PToch *ts from both PIE *t and PIE *dʰ.
To further complicate the situation, we find that PToch *dz is regularly lost before non-palatal vowels and PToch *æ. In addition, PIE*d itself is lost before a nasal. The following examples help illustrate the behavior of PIE *d in Tocharian.
|*d||*dz||*dʰgʷʰ-||*dzk-||A tsäk-||Skt. dah-|
|> *dgʷʰ-||> *tsäk-||B tsäk-||Lat. foveō|
|*duk-||*dzäk-||A tsäk-||Lat. ducere|
|*d||*dz||*doru||*dzæru||A or||Gk. dóru|
|> *Ø||'wood' (sg.)||> *æru||B or|
|'thought'||B ime||Gk. ídmōn|
In the above chart, *N stands for a general nasal consonant. To further highlight the difference in treatment among the dental consonants, consider the following examples.
|'queen'||> *wlāntsā||B lāntsa||OIr. flaitʰ|
|*t||*c||*dʰugH₂tēr||*täkācær||A ckācar||Ved. duhitár-|
|'daughter'||B tkā́cer||Gk. tʰugátēr|
|(root aorist)||B lac|
|*d||*ś||*déḱṃ||*dzäkä(n)||A śäk||Lat. decem|
|'ten'||> *śäkä||B śak||Gk. déka|
The details of the evolution of the labiovelars continue to provide a forum for scholarly debate. Though the broad rules which govern how these consonants evolved seem to be clear, the data nevertheless provides some contradictory evidence. In general, a labiovelar PIE *Kʷ (representing any of *kʷ, *gʷ, *gʷʰ) obeys the following rules in the transition to Proto-Tocharian:
showing loss of the labial element; but
|PIE *Kʷ(i, u, Ṛ) > PToch *kʷä|
showing retention of the labial element; and
|PIE *Kʷ(e, ē, y) > PToch *ś(ä, æ, (y)),|
demonstrating palatalization. In the above C represents any consonant and *Ṛ represents any given vocalic resonant. Consider the following examples.
|*KʷC||*kC||*nokʷtew-yo||*nækcäw'yæ||A nakcu||Lat. noctū|
|'at night'||B nekciye|
|*Kʷ(o, a)||*k(o, a)||*gʷów-||*kæw||A ko||Skt. gáuḥ|
|'cow'||B keu||Gk. boũs|
|*Kʷ(i, u, Ṛ)||*kʷä||*gʷṃ-||*kʷäm||A kum-||Skt. gaccʰati|
|'come' (*Ø-grade)||B käm-||OE cuman|
|*Kʷ(e, ē, y)||*ś(ä, æ, (y))||*gʷēm-e-t||*śæm'ä||A||Skt. ágan|
|'come' (root aorist)||B śem|
In the last two examples we see the contrast within different forms of a single root between palatalized and non-palatalized reflexes of the labiovelar. In particular, we see the labial element of PToch *kʷ reflected in the u of A kum-.
Unfortunately the data is not uniform, and we find evidence that goes against some of the above rules. For example, we also find PIE *wḷkʷo- 'wolf' > PToch *wälkwæ > B walkwe, with an unexpected sequence -kw-. Such examples might show influence from roots with like semantics and similar phonetic shape: PIE *(H₁)éḱwo- 'horse' > PToch *yäkwæ > A yuk B yakwe, where -kw- in this instance derives naturally from the palatal *ḱ followed by the labial glide *w.
For the most part, the sound (phoneme) inventory of Proto-Tocharian discussed above passes to the daughter languages rather mechanically. In some instances, however, Tocharian A and B treat Proto-Tocharian consonantal phonemes differently. We note the most salient of these differences in the following sections.
16.4.1 Proto-Tocharian to Tocharian A
A major feature of the evolution of simple consonants into Tocharian A is depalatalization. Specifically, we find PToch *w' > A w, so that Tocharian A loses the distinction between palatalized and non-palatalized variants of the bilabial glide *w.
We also find in Tocharian A a general elimination of labiovelars. That is, the labial element of the labiovelar *kʷ generally disappears, often with coloring of the adjacent vowel. This likewise applies to the sequence *kw, so that the *w falls away.
The following examples illustrate depalatalization and the loss of labiovelars.
|*w'||w||*H₂wéH₁-nt-o-||*w'æntæ||A want||Ved. vāta-|
|'wind'||B yente||Lat. ventus|
|*kʷ||k||*gʷṃ-||*kʷäm||A kum-||Skt. gaccʰati|
|'come' (*Ø-grade)||B käm-||OE cuman|
|*kuwonṃ||*kwænä(n)||A koṃ||Gk. kúōn|
|'dog' (acc.)||B kweṃ||Goth. hunds|
|*kuH-ti-||*kwācä||A kāc||Lat. cutis|
Several changes also affected consonant clusters in Tocharian A. The following points outline the major changes which occurred among consonant clusters:
The first two items denote simple changes affecting only the consonant cluster itself. The change *-kṣ- > A -pṣ-, however, occurs less frequently and its plausibility depends more heavily on the proposed reconstructions of the etyma involved. We also see that in Tocharian A the combination of a palatal consonant followed by *y resulted in loss of *y, with gemination (doubling) of the preceding palatal consonant. In addition, clusters involving nasals often tend to affect the preceding vowel. Specifically, not only is the nasal element of intervocalic *-ns- lost, but the preceding vowel acquires a palatal articulation. Moreover, the particular cluster *-ñc found in many nouns and the third person plural present active ending of verbs has the effect of palatalizing a preceding vowel. The palatalization applies in particular to PToch *a, ā, and *ä with the following results:
where *y denotes palatalization of the preceding vowel. Consider the examples listed below of the above mentioned sound changes.
|*st||ṣt||*-stH₂e||*-stā||A -ṣt||Skt. -tʰa|
|(2 sg. act. pret.)||B -sta|
|*-ks-||-ps-||*kʷleik-s-||*klaiks-||A kleps-||Skt. kliśyáte|
|(gen. sg.)||B (aśiyantse)|
|*-ns-||-ys-||*H₁ōmso-||*ānsæ||A *ays > es||Lat. umerus|
|'shoulder'||B āntse||Goth. ams|
|*-ñc||-yñc||*H₂eǵ-o-nti||*āk-æ-ncä||A ākeñc||Lat. agunt|
|'they lead' (3 pl. act. pres.)||B (ākeṃ)|
|*H₁i-Ø-nti||*yä-ncä||A yiñc||Skt. yánti|
|'they go' (3 pl. act. pres.)||B (yaneṃ)||Gk. íāsi|
The Tocharian B reflexes in parentheses are analogous in function but derive from different proto-forms.
16.4.2 Proto-Tocharian to Tocharian B
In contrast to Tocharian A, which depalatalizes PToch *w', Tocharian B in fact takes the process of palatalization one step further:
|PToch *w' > B y.|
This minor rebellion was not enough to satisfy Tocharian B, and so the language further distinguishes itself from its sibling by conservation of labiovelars. Tocharian B shows a tendency to preserve both PToch *kʷ and PToch *kw as B kw. The rule does not apply to all situations, however. Consider the following examples.
|*w'||y||*H₂wéH₁-nt-o-||*w'æntæ||A want||Ved. vāta-|
|'wind'||B yente||Lat. ventus|
|*kʷ||kw||*kuwonṃ||*kwænä(n)||A koṃ||Gk. kúōn|
|'dog' (acc.)||B kweṃ||Goth. hunds|
|'sharp'||B akwátse||OE āwel|
|*kʷis-so||*kwäsæ||A kus||Hitt. kuis|
|'who'||B kuse||Gk. hós-tis|
|*gʷṃ-||*kʷäm||A kum-||Skt. gaccʰati|
|'come' (*Ø-grade)||B käm-||OE cuman|
As the last example shows, Tocharian B does not retain the labiovelar in all situations.
Tocharian B also exhibits a number of changes to consonant clusters, differing from those displayed by Tocharian A (well, by definition of course -- if they were the same changes, then we would call them Proto-Tocharian changes!). In particular we find the following major changes:
Thus, PToch *w falls away immediately following a dental consonant -- be it stop, nasal, or sibilant (all represented here by *T) -- with gemination of the preceding consonant. This rule does not apply when the sequence lies at the beginning of a word. In addition, we find metathesis of the nasals in PToch *-mn- when between vowels. We also find epenthetic insertion of -t- in Proto-Tocharian consonant sequences consisting of any nasal (*N) or liquid (*L) followed by a sibilant (*S). When word-final, we find the simplification of the sequence PToch *-ns to solely B -n. Consider the following examples.
|*-Tw-||-TT-||*r(e)itw-||*ritw-||A ritw-||Av. raēṭwa-|
|'be attached'||B ritt-|
|*(s)p(e)n-w-||*pänw-||A pänw-||Gk. pénomai|
|'bring near' (Ø-grade)||B pänn-||OE spinnan|
|*-mn-||-nm-||*gʷṃ-||*kʷäm-näsk-||A kumnäs-||Skt. gaccʰati|
|'come' (*Ø-grade)||(CLASS X present)||B känmäsk-||OE cuman|
|*NS||NtS||*H₁ōmso-||*ānsæ||A *ays > es||Lat. umerus|
|'shoulder'||B āntse||Goth. ams|
|'birth' (perlative)||B cameltsa|
|*-ns#||-n#||*-(i,u)-ns||*-äns||A||Gk. *-ons > -ous|
|(acc. pl.)||B -äṃ|
We now turn to a closer inspection of the various primary case endings one encounters in the Tocharian texts, along with their historical development. In the present section we focus on the endings of the singular.
Simply put, the nominative case has no ending to distinguish it from other forms. We might expect this by considering, for example, thematic nouns in Proto-Indo-European with masculine nominative *-o-s and accusative *-o-m. Since PIE final consonants are lost in Tocharian, both forms result in PToch *-æ, and therefore lose any distinction. The evolution of PIE neuter nouns only serves to further reinforce this tendency, since these nouns have identical nominative and accusative even in PIE itself. In those situations in Tocharian where the oblique has acquired, or retains, a marker, we may distinguish the nominative case by the absence of this form.
The vocative only retains in Tocharian B a form distinct from the nominative; otherwise the two cases have identical forms. In Tocharian B we distinguish three basic vocative endings, whose use is determined by the noun's nominative form:
In the above C denotes an arbitrary consonant and C' an arbitrary palatalized consonant. Where palatalization is involved, the oblique may dictate the form of the vocative. Take for example the adjective B orotstse 'great', with masculine singular oblique B orocce; this has masculine singular vocative oroccu. The origin of these vocative endings remains obscure.
As mentioned previously, for a large number of nouns the nominative and oblique are identical. This results in a situation where, for a number of nouns, no particular ending distinguishes the oblique from other cases. The most interesting and conspicuous departure from this state of affairs occurs with substantives denoting reasoning beings.
17.3.1 Oblique Ending AB -ṃ
In the Tocharian languages, despite the general lack of any morphological marker for the oblique based on grammatical classification, we find more interestingly the rise of a marker AB -ṃ /-n/ based on semantic classification. In particular, substantives which denote reasoning or rational beings employ the ending -ṃ in the oblique singular. Scholars often denote this semantic distinction by writing [+human] or [-human], but one must be careful to understand that this is shorthand: substantives characterized by [+human] may in fact be other animals being personified, or even immortal gods. Nevertheless, Tocharian seems to adhere strictly to the notion of reasoning, since e.g. animals which are not being personified do not take the ending -ṃ in the oblique, nor do e.g. other divine 'things' that do not reason. For example, A ñkät B ñakte 'divine' takes the oblique ending -ṃ when referring to the Buddha, but not when referring to the Sun or Moon.
Adjectives (when not used as a substantive) may also take the oblique ending -ṃ, without necessarily modifying a substantive that is [+human]. Tocharian A allows this ending both in the masculine and the feminine singular of adjectives of any class; Tocharian B restricts the usage of this ending to adjectives ending in B-re, -śke, -ṣke, -i and -o.
Now if we sit back and think a moment, from an Indo-European perspective the above description of the distribution of the oblique -ṃ is pretty amazing. Certainly the ending can not derive directly from the *-m which provides the accusative ending in Proto-Indo-European, since PIE final consonants are lost in Tocharian. What then might AB -ṃ derive from? We see above two major clues:
From the above characteristics, it would not be hard to imagine a relation to PIE *n-stems. In particular, *n-stems are well attested in the germanic family as a major adjective formation, the so-called weak adjective formation. In particular, these *n-stem adjectives are employed when the noun modified is determined, e.g. accompanied by a definite article, or when the adjective is used as a substantive: Gothic þái ana aírþái þizái gōdōn saianans 'they that are sown on the good ground' (Mark 4.20). The use of such formations pervades Germanic, but finds attestation also in other branches of IE: Latin catus 'sly' vs. Catō (genitive: Catōnis) 'the Sly One'; Greek platús 'broad' vs. Plátōn 'the Broad (Shouldered) One'.
Given the preceding observation on *n-stems outside Tocharian, we may suppose the following state of affairs. At some stage leading to Proto-Tocharian, thematic nouns and adjectives could alternately employ weak (*n-stem) or strong (non-*n-stem) inflection:
|Strong Inflection||Weak Inflection|
|Nominative||*o-s > *-æ > A -Ø B -e||*-o-ō(n) > *-u > AB Ø|
|Accusative||*o-m > *-æ > A -Ø B -e||*-o-on-ṃ > *-ænä(n) > A -aṃ B -eṃ|
Strong inflection led, through phonological change, to indistinguishable nominative and accusative forms; this agreed with the situation inherited from PIE for neuters: nominative and accusative *-o-m > *-æ > A -a B -e for thematic neuters, and *-os > *-æ > A -a B -e for *s-stem neuters. Thus Tocharian speakers naturally associated strong nominal inflection with the characteristic [-human] appropriate to neuters. The final -ṃ inherited from the weak declension, by contrast, became a marker denoting [+human].
The above explanation, however, must certainly be overly simplistic and only an approximation to the true situation. But it nevertheless provides plausible motivation both for the distribution of -ṃ among nouns and for its general pervasiveness throughout the adjective classes.
17.3.2 Other Consonant Endings
We also encounter other consonant endings in the oblique case. As above, loss of the final *-m of the accusative means these typically preserve some consonant proper to the stem of the original PIE form. The two major endings preserve the *-r- and *-nt-, respectively, of original PIE *r-stem nouns and substantives in *-ent- or *-ont-. The following table provides examples of the evolution of some of these forms.
17.3.3 Removal of the Vowel of the Nominative
In Tocharian B we find a number of nouns, both masculine and feminine, which have nominative singular ending in -C'e (-e following a palatalized consonant), but whose oblique derives from removal of the final -e. These nouns generally remain confined to classes VI.1 and V.2:
|VI.1 Sg.||VI.1 Pl.||V.2 Sg.|
This declension pattern generally follows from nouns in PIE with stems in *-éy- and *-én-, and with nominative singular forms *-ḗy and *-ḗn, respectively. Consider, for example, the forms of B maśce 'fist':
In the above we find confusion in the accusative between the éy-stem nouns and *i-stem nouns.
17.3.4 Palatalization of Suffixed Consonants
Several types of adjectives, primarily in Tocharian B, employ palatalization of a suffixed consonant to denote the masculine oblique singular. Note the following forms.
|A Nom. Sg.||A Obl. Sg.||B Nom. Sg.||B Obl. Sg.|
In some instances this palatalization also spreads to the masculine genitive singular, as well as to the masculine plural and feminine singular. The exact origin of this palatalization remains a source of scholarly debate.
17.3.5 Oblique in A -e B -ai
The ending A -e B -ai derives from Proto-Tocharian -ai and, at least in Tocharian B, provides one of the more frequent markers of the oblique case. This ending pertains to both masculine and feminine nouns, with varied nominative forms:
|Feminine Abstracts||N||-o, -iye||prosko, proskiye|
|Feminine Substs.||N||-o, -iye||pyāpyo|
|Feminine Substs.||N||-yo, -ya||wertsiya, wertsyo|
Because of the pervasiveness of the ending A -e in Tocharian A, it is difficult to discern exactly when this ending corresponds to the oblique singular B -ai in Tocharian B. We do find one example which appears more or less certain: A kli 'woman', with oblique singular kule; cf. B klyiye, with oblique singular klai. In general the ending -ai undoubtedly derives from many sources; however the details of the development remain a subject of scholarly debate.
17.3.6 Oblique in B -o, -a
This category comprises a smaller number of substantives within Tocharian B. In particular, only two feminine nouns -- B śana 'woman, wife' and lāntsa 'queen' -- show the oblique ending -o. A small group of nouns both masculine and feminine employ the oblique ending -a. In the former group, the particular combination of nominative and oblique endings derives from the interplay between the position of the PIE accent and the grade of a thematic suffix containing a laryngeal; in the latter group, secondary reformation may have occurred to distinguish what would otherwise have been identical nominative and oblique forms:
|Acc.||*ǵṇdʰweH₂-m||*käntwå||*kantwo > kantwa||Got. tuggo|
The reformation of the oblique form may have patterned after oblique forms in -ai. Note in the last example that we find metathesis of PIE *dṇǵʰweH₂ to *ǵṇdʰweH₂.
17.4.1 Peculiar Genitive Endings
The Tocharian languages display a small number of genitive singular endings which are particularly peculiar from a Proto-Indo-European standpoint. The first of these endings, A -Ø B -e, is remarkable because, despite deriving from by far the most common genitive ending in PIE, it survives only in a very limited number of Tocharian nouns. In particular, it remains in certain adjectives with stem *-ent-/*-ont-, and in the lone familial noun B tkācer 'daughter'.
The second of these endings, AB -i, bears note because its historical development remains a bit of a mystery. Scholars continue to debate its origin, but one proposal suggests a likely source lies in the Proto-Indo-European dative case, formed on top of a genitive ending (Pinault 2008). It occurs only with personal nouns, in particular the remaining familial nouns, personal pronouns and demonstratives, and proper nouns borrowed from Sanskrit with nominative singular in -e.
|Gen.||*gʷen-éH₂-s+ey||*śänå+'äy||*śänó+i > śnoy||Ved. jáni|
We further find genitive singular endings A -i B -ñ in certain nouns borrowed from Sanskrit. The origin of these endings is less well understood than those treated above.
17.4.2 Productive Endings
The most productive genitive singular endings on a synchronic level are A -s B -ntse. As with the -ṃ of the accusative, the origin of the genitive A -s B -ntse most likely lies in PIE *n-stem nouns (Pinault 2008). In particular, it appears that *n-stem nouns with nominative-accusative in *-ṇ and genitive *-en-s, where the *-e- formed part of the suffix, incurred a secondary genitive formation, *-ens-os, on top of that already in place. This likewise occurred in the weak formation of thematic nouns (see the discussion of oblique -ṃ above): the typical genitive *-o-n-os was remodeled as *-ons-os. In the Proto-Tocharian period, the ensuing *-onsos > *-ænsæ was further reanalyzed as *-æ-nsæ, since the *-æ- agreed with the accusative of the strong declension of thematic nouns (cf. *yäkwæ). This of course falls in line with the general Tocharian tendency to build secondary cases using postpositions appended to the oblique (accusative). This last ending, *-nsæ, then follows regular sound changes to yield A -s (due to loss of final vowels and the simplification *-ns > A -s in Tocharian A) and -ntse (due to the insertion of -t- between *-ns- in Tocharian B). Consider the following examples.
In the above chart, PIE *stH₂-mṇ illustrates the *n-stem nouns, in this instance with a reformed genitive. We see the parallel with the weak declension of *H₁éḱw- in the genitive, which is also reformed and further reanalyzed as an ending appended to the accusative.
We also find the productive endings A -āp, -yāp B -epi for the genitive singular. These generally pertain to masculine adjectives and participles, and in Tocharian A also to substantives characterized as [+human]. These endings likely show the adoption of a genitive ending first occurring among the demonstrative pronouns. In particular, the labial element perhaps derives from an archaic thematic dative: PIE *te-smoy > *cäzβu > *cä-βä. The final element *-βä is appended in Tocharian A to *-yā-, perhaps a reflex of feminine *-iH₂; in Tocharian B the pronoun further inherits another dative marker, *cä-β-i > B cwi, cpi, and the element -pi is imported into adjectives and participles.
Class II comprises athematic adjectives deriving from original PIE *n-stems. As we have seen in the discussion of the oblique singular ending -ṃ, Tocharian has employed a number of *n-stem formations throughout its evolution. It should therefore come as no surprise that a number of adjectives fall into Class II. As exemplar we take A klyom B klyomo 'noble'. The forms are given below.
|A Masculine||A Feminine||B Masculine||B Feminine|
We see in the Tocharian A masculine genitive and oblique forms the vestiges of a formation in *-nt-, and likewise in the masculine oblique singular of Tocharian B. Apart from these forms, the paradigm clearly shows the influence of the *n-stems. The nominative reflects PIE *ḱléu-mō(n), with cognates Got. hliuma 'hearing', ON hljómr 'sound', Skt. śrutá- 'heard, famous', Gk. klutós 'renowned'. The development of the form involves umlaut: *ḱléu-mō(n) > *klyäumõ > *klyumo > *klyomo (umlaut) > A klyom B klyomo. The final -o of the Tocharian B masculine nominative singular has spread to other cases, replacing the expected vocalism of the *n-stem forms:
|PIE||PToch||B Expected||B Actual|
This replacement even changes the vowel of the typical -epi of the genitive ending.
As the name implies, the verbs of the present thematic classes insert the theme, or thematic vowel, PIE *-e / o- between root and ending. In particular, the first person forms (both singular and plural), as well as the third person plural (not singular), show *o-grade of the thematic vowel; the remaining forms show *e-grade. The following chart lists the PIE combinations of thematic vowel and ending that led to the Proto-Tocharian forms.
|3||*-o-nti||*-ænt 'ä||*-añc > -eñc|
The apostrophe denotes palatalization of the preceding consonant. One immediately notices the most salient feature of thematic conjugations: where PIE originally has the theme *-e-, the change PIE *e > PToch *'ä often results in vowel loss; but palatalization of the preceding consonant remains. That is, the lasting effect of the PIE theme *-e- is Tocharian palatalization of the consonant preceding the ending.
1st Person Singular Active: The theme vowel of the first person singular is PIE *-o-, but the endings are heterogeneous between the two Tocharian languages. Tocharian A derives the form from the ending *-mi, importing the athematic ending into the thematic conjugation. Such reformation of the first person singular is found outside of Tocharian as well, e.g. Skt. bʰarāmi 'I bear'. Tocharian B, by contrast, does not use the athematic ending, but rather employs the usual thematic ending. Not content to stop there, Tocharian B moreover adds the morpheme *-u. Its origin is uncertain, though perhaps it derives from *-wi, possibly harkening back to the -vī found in Latin perfects: e.g. clamāvī 'I cried'.
2nd Person Singular Active: Interestingly, only this active form agrees between the two languages. The derivation follows regular sound rules; the only difficulty being the loss of the final *-ā# > B -a#, which normall should remain in Tocharian B.
3rd Person Singular Active: This form shows the most marked difference between the two languages, and the most notable departure from PIE. The general loss of final consonants in Tocharian implies that such forms must have been analyzed at some point as a phrasal unit with a following adverbial or other part of speech, or with some other suffix which differed between the two languages. If in fact these endings derive from the addition of some enclitic elements, the most likely candidates are the sentence connectives PIE *se and *nu 'now'. In particular the meaning of the latter stands in accord with the fact that this ending only appears in the non-past formation.
1st Person Plural Active: The first person plural active ending is almost as problematic in PIE as in Tocharian. The only certain element is PIE *-m-, since Lat. -mus, Gk. -mes (Doric), and Vedic -mā (occasionally alternating with -ma) all leave the vowel, grade, and terminal element in doubt. It is no wonder then that Tocharian A and B differ on precisely these grounds. Given the Vedic form -masi and its resemblance to Gk. -mes, it is tempting to take the PIE form as *-mesi. This provides a plausible antecedent of A -mäs, but would require that PIE *i not palatalize the preceding PIE *-s-, otherwise atypical in Tocharian. Moreover, if the antecedent of the Tocharian B form is PIE *-mō, this requires that PIE *ō develop into PToch *-o, which, though possible, is generally restricted to the environment PIE *-ōn#.
2nd Person Plural Active: The two languages diverge on this form. Tocharian B shows a lengthened grade of the vowel, necessary to explain both the palatalization and that fact that the vowel itself is retained.
3rd Person Plural Active: The third person plural endings follow normal phonological evolution. The Tocharian A form results in -eñc < *-añc due to palatalization of the vowel in the environment of a cluster of palatalized consonants. The Tocharian A form derives from the PIE primary (non-past) ending *-nti, while that of Tocharian B derives from the original secondary ending *-nt.
1st Person Singular Mediopassive: This ending is a composite structure. The most salient element is the the PIE *-r-, a marker which in general characterizes the entire mediopassive paradigm in PIE, cf. Lat. orior 'I rise'. Tocharian shows also the element *-h₂, the PIE primary thematic active ending of the first person singular, and further preposes the secondary ending *-m. The final *-i is likely again the deictic particle encountered in the active endings. Greek shows a similar structure, e.g. Gk. pʰér-o-mai < PIE pʰér-o-mh₂i, differing from Tocharian only in the zero grade of the ending, and lack of the mediopassive marker *-r. The thematic vowel often drops in Tocharian A forms. The change B *-mār > -mar results from the accent falling on root or stem, leaving the vowel of the ending unaccented.
2nd Person Singular Mediopassive: This ending essentially recapitulates the active ending, merely adding the mediopassive marker *-r and the deictic *-i.
3rd Person Singular Mediopassive: Whereas Latin shows an *o-grade in the third person mediopassive ending (*-tor > Lat. -tur), Tocharian shows a zero grade followed by the mediopassive marker *- and the deictic *-i.
1st Person Plural Mediopassive: This reflects the personal ending found in, e.g., Gk. -metʰa and Vedic -mahi, augmented by the mediopassive *-r and deictic *-i.
2nd Person Plural Mediopassive: This ending proves difficult to explain. Both Tocharian A and B forms show the presence of the mediopassive *-r and deictic *-i which pervade the entire mediopassive paradigm. In other regards the two languages diverge. The Tocharian A form likely results from a PIE ending -dʰ(w)e, as found in Greek -s-tʰe. The Tocharian B form, however, lacking palatalization must result from a variant such as PIE *-dʰuwe, as found in e.g. Hitt. -duma < *-duwa. If so, then the form likely developed as PIE *-dʰuwe > PToch *-täw'ä > B *-täyä > -tä, with the last step following from contraction.
3rd Person Plural Mediopassive: This form parallels the singular, and whereas Latin has again the *o-grade *-ntor > Lat. -ntur, Tocharian once more shows a zero grade.
CLASS II presents follow a root thematic present formation. That is, CLASS II presents begin with the root, add the thematic vowel, then the endings. As such, they form the backbone of the majority of the thematic present conjugations. Of course this characterization must be viewed in Proto-Indo-European terms, and so strictly applies to those roots in the class which may be traced back to PIE heritage; the class then extended by analogy within Tocharian itself. In general the roots of this class show PIE full grade, *CéC-. Compare AB pär- 'bear' with its Indo-European cousins.
Neither PToch *p nor *r are subject to palatalization (or, they were palatalized and regularly depalatalized), and so the forms of this particular verb do not show the effects of the thematic vowel in Tocharian.
The structure of CLASS II also forms the basis for other thematic classes, in particular the present classes in *-se / o- and *-ske / o-.
The paradigm of AB āk- < PIE *h₂eǵ- 'lead' serves to illustrate the active forms of CLASS II verbs; A klyos- B klyaus- < PIE *kleus- 'hear' illustrate the mediopassive forms.
Note that the interplay of the accent and the weak vowel *ä often result in a lack of thematic vowel between root and ending in the final Tocharian form. The palatalization, or lack thereof, is the only remnant of the original vowel.
Though CLASS III and CLASS IV presents are thematic, several features distinguish them from other present classes. In particular, the verbs of both classes are deponent, that is, the presents exhibit only mediopassive forms. Moreover the thematic vowel shows differences from other thematic classes, in many respects suggesting a thoroughgoing *o-vocalism throughout the paradigm, rather than the *e / o alternation of the other thematic classes. The verb AB mäsk- 'be located, be' illustrates the forms of CLASS III.
The verb AB plānt- 'be happy' serves to illustrate the forms of CLASS IV.
Several plausible explanations exist for the formation of CLASS III and CLASS IV presents, and scholarly opinion to date lacks complete consensus.
One line of reasoning, set forth in Adams (1988), concerns the most archaic PIE forms of the middle voice. In particular, some evidence suggests that the most archaic form of the third person singular mediopassive ending was PIE *-o, in the plural *-ro. This, combined with the deictic *-, apparently forms the basis for the Sanskrit forms duhé < PIE *dʰugʰó + i and duhré < PIE *dʰugʰró + i. Though an archaic formation within PIE itself, the Tocharian B forms ste (3 sg.) < PIE stH₂ó, stare < PIE *stH₂ró support the possibility of this formation surviving into the documented Tocharian languages. The fact that the ending was archaic in PIE, however, may have led to its being augmented by the athematic mediopassive endings. This, combined with a generalization of the *-o- as thematic vowel throughout the mediopassive paradigm when the root showed no alternation with an active present, may have ultimately led to a present conjugation characterized by thematic *-o- through, and the more usual mediopassive endings. The change PIE *o > PToch *æ would then explain the thematic vowels of CLASS III in Tocharian A and B. CLASS IV might then be explained by a process of rounding: those roots with Proto-Tocharian root vowel *-ā- undergo a shift *(C)ā(C)æ- > *(C)å(C)å-.
Another possibility (Pinault, notes 2006) invokes the process of contraction. In particular, as many of the verbs of CLASS IV are synchronically denominatives, one might seek an origin in the denominative suffix PIE *-eh₂-, possibly followed by PIE *-ye / o-. The compound suffix then undergoes contraction across the *-y-, a process found elsewhere, e.g. PIE *tréyes > PToch *träyä > *træ+i > A tre B trai. This process does not appear to occur finally, but only in initial and intermediate syllables.
CLASS VIII presents comprise the so-called sigmatic presents. That is, these verbs append -s- to the verbal root, followed by the thematic vowel: PIE *-se / o-. The thematic vowel of course causes palatalization of the preceding *-s- in the expected positions. The forms of A ar- B er- 'evoke' < PIE *h₁or- (cf. Gk. órnūmi 'urge, incite', Lat. orior 'rise', Skt. ṛṇóti 'raises', Hitt. arāi 'rises') serve to illustrate the paradigm.
|Pres. Ppl.||aräsmāṃ (arsamāṃ)||ersemane|
The third person singular form in Tocharian A results from a regular simplification: A *arṣä-ṣ > *arṣṣ > *arṣ > aräṣ (a change also reflected in the third person singular present of 'to be': *naṣäṣ > A naṣ). The geminate consonant remains in the form with suffixed personal pronoun.
One generally finds that in verbs where stem-final *-se- becomes *-ṣ(ä)-, this palatalized *-ṣ- subsequently depalatalizes upon contact with a following -tV-, i.e. an ending containing t followed by a vowel. This is a regular, apparently morphologically conditioned phonetic change, and the striking point is that this does not occur in the Tocharian B mediopassive paradigm above: B -ṣ- is retained in the second and third person forms, in keeping with the remainder of the paradigm. Compare CLASS IX, where this change does occur.
The linguist of course wonders: what is the origin of the *-s- formation in the Tocharian present? The basic fact is that to date the origin remains elusive. Scholars typically discuss the sigmatic presents in conjunction with presents in *-sḱ-, which they view for the most part as causative formations within Tocharian. The *-s- also frequently denotes a causative, more so in Tocharian A, where the *-sḱ- formation has all but disappeared. The difficulty in associating the *-s- extension with an original causative lies in the fact that no such causative formation exists in Indo-European outside of Tocharian.
Of course sigmatic formations abound elsewhere in Indo-European. Some scholars link the Tocharian formation with original sigmatic aorists in PIE, in particular with aorist subjunctives. There are two primary difficulties with such a theory: (1) these aorists generally show full grade of the root, whereas the Tocharian presents show zero grade; (2) the PIE cognates of Tocharian *s-presents generally show root aorists, rather than *s-aorists. Other scholars propose linking the Tocharian formation with sigmatic presents found in Greek, as in e.g. Gk. aéksō 'increase, make grow' and dépsō 'knead, soften'. Again difficulties arise: (1) as above, the Greek forms show full grade, while the Tocharian forms derive from zero grade; (2) the verb A ok- B auks- 'sprout, grow up' is the only cognate of an IE sigmatic present (cf. Gk. aéksō; Skt. vakṣáyati 'makes grow', but perfect vavakṣa 'grew'; ON vexa 'make grow', but vaxa 'grow'; Goth. wahsjan 'grow'; OE weaxan 'wax, grow'), but has present A oksis- B auks-äsk-. Contary to expectation, the *-s- appended immediately to the root does not appear to have a causative function across the IE cognates; moreover, this holds equally true in Tocharian, as evidenced by the secondary suffix applied after the *-s-. The Tocharian form is in fact not causative, cf. B ostn=auksäṣṣäṃ 'he grows up at home' (121a1).
CLASS IX presents, extant only in Tocharian B, employ the suffix *-sḱ- followed by the thematic vowel: PIE *-sḱ-e / o- > -ṣṣ(ä)- / -skæ-. Where the thematic vowel was originally *-o-, the cluster *-sk- remains intact; where the thematic vowel was *-e-, the cluster *-sk- palatalizes as *-ṣṣ-. Though the class contains predominantly causative verbs, some non-causatives likewise pertain to this formation. The verb B kälp- 'attain' illustrates the paradigm.
|Present IX||B Base||B Causative|
The etymology of B kälp- remains uncertain, though it perhaps derives from the root PIE *kelp- found in Old English helma 'rudder, tiller' and hielfe 'handle'. The verb retains another formation based on the roots with final laryngeal, showing a CLASS VI present A kälp-n-ā-tär and Class I preterite A kälpāt B kälpāte. This construction is reflected in the non-causative stem *kälp-ā-sk-, whereas the causative stem derives from *kälp-äsk-.
The paradigm also illustrates the phonetic rule mentioned in the context of CLASS VIII (the latter notable because of the rule's absence): the palatalized suffix -ṣṣ(ä)- is depalatalized preceding -tV, i.e. an ending beginning with -t- followed by a vowel. Thus depalatalized kälpāstär and kalpästär, where by contrast CLASS VIII shows the palatalized erṣtär. Interestingly, this rule does not apply when the *-sk- is no longer felt as a suffix, but as part of the root itself. Consider for example ñäsk- 'demand, require; (med.) seek' < PIE *nes-sḱe / o- < *nes- 'move (back) toward a good position/state' (cf. Gk. néomai 'return home', Skt. násate 'unite with, approach', OE genesen 'save'): CLASS II present mediopassive 3rd sg. ñaṣtär, pl. ñaskentär. (In Tocharian A, the cluster *-st- does not appear; one only encounters -ṣt-.) Thus this phonetic rule is not only sensitive to the phonological environment, but to the morphological environment as well.
A point of critical importance surrounds the vocalization of the vowel *-ä- inserted through anaptyxis between the root and *-sk- suffix. This joining vowel appears variously as the reduced -ä- and the full -a-. When the former occurs, the ä of the root strengthens to a; when the latter, the root vowel remains in the reduced form. This alternation betokens a shift in the position of the accent: in Tocharian B the accent shifts to the first syllable in causative verb forms, while it remains further to the rigth in non-causative forms. This provides a very important distinction between causative and non-causative, since even non-causatives can employ the *-sk- suffix: B ai- 'give', with 1st sg. aiskau, 3rd aiṣṣäṃ; we- 'speak', with 1st sg. weskau, 3rd weṣṣäṃ. Thus the shift in accent is the only strict marker of causative forms, accompanied by initial palatalization.
CLASS X presents show an extension by *-sk- of verbs with nasal present stems: *-nā+sk-, or *-n+sk- with anaptyxis, yielding A -nās-/-näs- B -nāsk-/-näsk-. Both causative and non-causative verbs belong to this class. The verb täm- 'be born, (caus.) beget' illustrates the paradigm.
|Present X||A||B Base||B Causative|
|Pres. Ppl.||*tämnäsmāṃ (tmäṃsamāṃ)||tänmaskemane|
The verbal stems derive from *täm-näsk- > A *tämnäs- B *tänmäsk-, where the cluster -mn- undergoes metathesis in Tocharian B. Note also the simplification in the third person singular active of Tocharian A: *tämnäṣäṣ > A tämnäṣ (cf. A aräṣ in CLASS VIII). The Tocharian B forms, when causative, shift the accent forward to the initial syllable, as with the presents in simple *-sk-: causative third singular active tanmäṣṣäṃ, mediopassive tanmästär.
Three verbs of CLASS VIII -- näk- 'destroy'; päk- 'bring to maturity, cook'; tsäk- 'incinerate, burn (trans.)' -- have in Tocharian A special intransitive formation employing mediopassive forms of CLASS X. For example, tsäk- has active third person singular tskäṣ 'incinerates, makes burn' < *tsäkṣä-ṣ (cf. B tsakṣäṃ), with corresponding intransitive tsäknäṣtär 'is burning, is on fire'. Similarly, näknäṣtär 'dies' and päknäṣtär 'matures, cooks (intrans.)'.
CLASS XI presents also employ the *-sk- extension, but applied to sigmatic stems of the sort encountered in CLASS VIII. The root extension therefore becomes *-säsk- > A -sis- B -sask-/-säsk- (note the vowel in the Tocharian A suffix). The verb AB āks- < PIE *H₁ǵ-s- 'say' (cf. Lat. aiō 'I say, affirm' and adagium 'proverb'; Gk. ẽ 'he said' < *H₁ēǵt, with reformed present ẽmi) illustrates the paradigm.
Tocharian A exhibits few CLASS XI forms, and the -i- of the suffix is difficult to explain. It perhaps reflects a simple phonological change of -ä- to -i- between sibilants. Possibly arguing against this is the form A eṃtsäṣtär of the verb A ents- 'grasp', related to B eṅk- and perhaps derived from PIE *H₁neḱ-; cf. Skt. a'nóti 'reaches, gets' and Gk. enegkeĩn < *H₁neH1ṅke / o-, suppletive aorist to pʰérein 'carry'. Compare also A swāsäsmāṃ (mediopassive) B swāsäṣṣenca (active), from *swā-s-(ä)sk- 'make rain'.
CLASS XII presents show a suffix -(ä)ññ-, and comprise both primary and denominative verbs. The suffix has two likely origins: *-n-ye / o-, a denominative based on a root with final *-n; and *-ṇ(H)-ye / o-, a deverbative based on a nasal present formation. This latter formation parallels that of the *-nāsk- class. One finds a similar interplay between denominative and deverbative constructions in Sanskrit: gṛbʰāyáti vs. gṛbʰṇā́ti, both from the root grabʰ- 'grab'. The verbs A śew- 'yawn' and B mänt- 'injure' illustrate the paradigm.
Note the depalatalization of the *(ä)ññ-suffix before endings in -tV, as encountered in in CLASS IX, and notably absent in CLASS VIII.
CLASS XII marks the only surviving, productive class of denominatives in Tocharian. Consider, for example, the noun A tuṅk B taṅkw 'love': compare the present participles A tuṅkiññant B täṅwaññenca '(the one) loving' from *tänkw-äññ- 'to love'. Similarly the verb B lare-ññ- 'to cherish' from the adjective B lare 'dear'. CLASS XII is not restricted however to denominatives, showing forms derived from primary verbal roots: B arcantär from the root ārc-äññ-, based on ārc- 'have to, must'; similarly mänt- 'injure' above.
We distinguish three basic uses of the nominative case in Tocharian. The first and principal usage is to denote the grammatical subject of a finite verb form, in the same way that in English we distinguish I from me: only the former, the nominative form, can represent the grammatical subject of a verb.
The second principal use is as predicate nominative, generally in conjunction with an infinitive. That is, the infinitive and the noun in the nominative case are equated to one another, generally via a copula, whether explicit or implied. Consider the following examples, with bold typeface denoting the noun in the nominative and with italic typeface denoting the infinitive:
The third major use of the nominative is adverbial. In particular, the nominative masculine singular of an adjective may function as an adverb with cognate meaning. In other classical languages such as Greek, Latin or Sanskrit, one employs the neuter accusative; the Tocharian usage of the nominative derives from the coalescence of masculine and neuter in the singular, and from the identity of nominative and accusative in the neuter. The Tocharian form employs a neuter substantive, as discussed in the first lesson.
The oblique case serves many and varied roles within Tocharian. Chief among these, and the major point of departure from the accusative of other ancient IE languages, is as base of the secondary cases. That is, Tocharian forms the secondary cases of nouns by appending suffixes to the oblique form of the noun, either singular or plural according to context. This analysis of secondary cases particularly comes to the fore in group inflection, where one secondary case ending governs a string of nominals in the oblique. In particular, many prepositions and postpositions govern nouns in the oblique.
The remaining functions of the oblique case in Tocharian by and large parallel its uses in other ancient Indo-European languages. We identify the following major uses: