The Tungusic Research Group at Dartmouth College


Section 1

Section 2

Section 3

About the languages

Basic Typological Features of Tungusic Languages
Tungusic languages can be said to have a shared typological profile. That is, despite the many differences among them, there are certain general linguistic properties they have in common. These are reviewed in this section.

Tungusic languages are exhibit VOWEL HARMONY to some degree. The vowel harmony operates along two parameters: vowel rounding and vowel tenseness. The latter is more commonly called ATR [Advance Tongue Root] or RTR [Retracted Tongue Root] harmony.

In general the rules of harmony of harmony are: 1) All the vowels in a word must have the same ATR value; that is, all the vowels must be lax or all must be tense. 2) When a round vowel occurs in a root, all the vowels which follow it (both in the root and suffixes) will also be round. Consider the following examples from the Central dialect of Oroqen:

1) a. OlO 'fish' b. OlO-kO:n 'little/pretty fish'
2) a. Nênaxên 'dog' b. Nênaxê-ka:n 'pretty/dear/little dog'
3) a. b´y´n 'person' b. b´y´-k´:n 'pretty/dear person'
4) a. OlO-dø 'to the fish' b. b´yun-du 'to the person'

In 1a, both the vowels are lax (as predicted by the first rule of harmony) and there are no unround vowels to the right of a round vowel (as predicted by the second rule). When the diminutive suffix is added (1b), the suffix vowel is also lax and round. Compare this with 2b. The diminutive suffix once again contains a lax vowel, but in this case it is not round because the preceding stem vowel is not round. In 3, which exemplifies how vowel harmony operates with tense vowel stems , the diminutive suffix contains a tense vowel (NB: in Tungusic [a]/[´] constitute a tense/lax pair).

The diminutive represents a suffix with an underlying unround vowel. Some suffixes, however, have an underlying round vowel, such as the dative case suffix -du/-dø. A comparison of 4a and 4b reveals that the first rule of harmony (which requires ATR values line up for vowels) operates in the expected way. The second rule is now irrelevant since it does not require that all vowels carry the same value for rounding, but only that suffix vowels round if the final stem vowel is round. Therefore, in 4b, the suffix vowel does not become unround.

There are many complications to vowel harmony in Tungusic. First, some suffix vowels simply fail to undergo harmony processes even though they seem like they "should". For example, the first vowel of the suffix -tSir´/tSira, which is used to signal a lesser degree of the property expressed by a head (e.g. Central Oroqen gOrO 'far' gOrO-tSira 'somewhat far'), is invariant even though [i]/[ê] exists as a harmony pair in the language. Second, many Tungusic languages have lost one member of certain harmony pairs (this is particularly common for the front vowels). Third, as Tungusic languages become obsolescent, younger speakers tend not to employ harmonic processes in the same way as older fluent speakers. This may be due to the influence of contact languages such as Mandarin Chinese and Russian, or it simply may be an imperfect acquisition of the language.

A second notable phonological characteristic of Tungusic languages is the existence of PHONEMIC VOWEL LENGTH. This means simply that the length of a vowel is not predictable and that variations in vowel length can be as significant in lexical identification as vowel height, voicing , etc. Hence, in Northeastern Oroqen a:kin 'liver' and akin 'older sibling' are not homophones; they are perceived by Oroqen speakers as being pronounced differently due the variation in vowel length. Compare this with English, which has non-phonemic vowel length, i.e. vowel length which is predictable and does not give rise to distinct words. The presence of a syllable final voiced consonant lengthens the preceding vowel as in [pœ:d] 'pad' (compare [pœt] 'pat'). Not only can one identify specific environments where vowel length will be found, artificially manipulating the usual vowel length has little effect on whether English speakers understand the word. If we pronounce 'pat' as [pœ:t], a listener may pick up on the fact that the word is being said 'with an accent' or 'in an unusual way', but they will still correctly identify the word as 'pat'.

Most Tungusic languages also have, as part of their phonological profile, a PROHIBITION AGAINST COMPLEX CODAS in syllable structure. This means that one does not find syllables which end with two consonants. There is also a strong tendency to avoid consonant clusters at the beginning of words. One practical effect of these restrictions is that consonant sequences almost always only occur in the middle of the words: e.g. Evenki dJapkun 'eight'. In such cases the first consonant is syllabified in the coda of the first syllable. The second consonant is syllabified in the onset of the second syllable.

Tungusic languages have AGGLUTINATIVE MORPHOLOGY. Therefore, the sequences of morphemes which occur within a word are easily segmentable and tend to carry one piece of grammatical information:

5) aSi:-n-dula:-Bi:
to his wife

In 5, the four morphemes in the Evenki word are readily divided. When attached to other roots, they appear in the same form (though, of course, subject to regular allomorphic processes). They all mark a single grammatical category.

This sort of morphology stands against 'fusional' morphology, where it is hard to find sharp boundaries between morphemes and several bits of grammatical information are commonly included in a single morpheme.

Another typological property of Tungusic languages is that there is a WEAK DISTINCTION BETWEEN THE WORD CLASS OF ROOTS. Adjectives and nouns, in particular, are hard to differentiate outside of specific instances of use. Consider the root sagdi 'old'. In many Tungusic languages/dialects, there is no reason to classify the root as either a noun or an adjective other than the fact that it has the same meaning as adjectives from other languages that have a strong word class distinction. From a Tungusic internal perspective, though, there are no morphological reasons to label it as one or the other: it can take nominal morphology such as case, number, and possessives. Moreover, there are no syntactic properties inherent to the root which lead us to classify it as one way or the other. It can be used as a nominal modifier (in which case it functions as an adjective) or as a noun phrase head (in which case it functions as a noun). The distinction between nouns and verbs is also not nearly as clear as what one finds in familiar Indo-European languages. Many roots simply seem to be unmarked for word class. However, there are also a large number of lexical roots which must be assigned to either the verb category (or the noun category), so at least in this instance, one finds justification for claiming that noun vs. verb is a distinction made in Tungusic.

One fascinating and understudied aspect of Tungusic comes from the realm of derivational morphology. Not only are there an immense number of derivational morphemes (there are upwards of seventy derivation morphemes relevant to nouns alone), derivational morphology is also found on root which in other languages rarely host affixation, such as numerals. Consider the following derivational processes involving numerals (taken from Bulatova and Grenoble):

digi 'four'
digini 'four people'
digiNn´ 'four animals'
digill´ 'four days'
digir´:gd´ 'four things'
diginm´:n 'four layers'

In contrast, a language such as English allows limited derivation with numerals. One perhaps finds it in expressions such as two-way, three-way, etc., and frozen derivational morphology can be observed in once, twice, thrice, as well as in fourth, fifth, etc., but that it about the extent of it.

Unfortunately, Tungusologists have not always been careful to identify how productive particular derivational morphemes are or the range of meanings they can have. Therefore, derivational morphology remains one of the largest unexplored issues of the Tungusic lexicon.

Tungusic languages have COMPLEX CASE SYSTEMS. There is variability in the number of cases that individual languages have, with the low end represented by the five of Manchu and the high end represented by Evenki with thirteen or so. Although there are a fairly large number of case inflections, the basic structure of the system is of a fairly usual sort when taken from a cross-linguistic perspective. There are five core case categories: nominative (which is unmarked in all Tungusic languages), accusative, dative, instrumental, and locative. One somewhat unusual (though not unique) property of this system is that the accusative category is split into two sub-classes, which are usually referred to as definite-accusative and indefinite-accusative. As the names suggest, the definite-accusative case suffix is employed when a noun is identifiable from context and is specific, whereas indefinite-accusative is used elsewhere.

The locative category is where one finds most of the case markers in Tungusic languages with robust inventories of affixes. In Evenki, for example, one finds a general locative, an allative, an ablative, a prolative, an elative, an allative-locative, and an allative-prolative. (Don't worry if most of these terms don't mean much to you--they are merely rough descriptions of the semantics of the case affix using a Latin term).

Where one finds variation in the number of case affixes in Tungusic, it is almost always in the number of affixes dedicated to expressing locative notions. One other point of variation is the existence of a comitative affix. Historically, there does not seem to have been a comitative case in Tungusic; rather there was a comitative suffix which could appear together with case suffixes on nouns. However, there is reason to believe that in some Evenki dialects, at least, the suffix is becoming or has become a part of the case paradigm (and therefore is used in complementary distribution with other case inflections).

COMPLEX TENSE-ASPECT MARKING is another typological feature of Tungusic languages. In general terms, we might say that verbs contain two 'slots' relevant to tense/aspect. On slot indicates tense, the other indicates aspect (rather than indicating position in time with respect to the speech event like 'tense', aspect describes the internal temporal make-up of the action being described by a verb, e.g. whether it was extended in time or punctuated, whether it was repeated, whether it has been completed, etc.). Consider the following example from Oroqen:

6) yabø-ìin-ìa-l
They just left.

Immediately following the verb root is an 'inchoative' aspect suffix. The suffix is employed to indicate close proximity between two events (in this example, the speech event and the event of leaving indicated by the verb). The tense suffix ìa is placed next.

Because all Tungusic languages have no less than two (and as many as seven) tense affixes and multiple aspect suffixes (perhaps as many as fifteen in some languages), the potential combinations of tense and aspect affixes can be very large. Of course, not every possible combination necessarily appears: some may be ruled out for semantic reasons. Even so, mastering the tense/aspect system in all its subtleties is a large task even for native speakers, and is one area of Tungusic grammar which is lost quickly as the languages obsolesce.

A final notable property of Tungusic morphology is the existence of opposition between NOMINATIVE AND NON-NOMINATIVE PRONOUN ROOTS. In nominative case, one set of pronouns is employed (bi 1S, si 2S, etc.). For all other cases, a different root is employed (min- 1S, sin- 2S, etc.). This unusual feature of personal pronouns is also found in Turkic and Mongolic languages, which is perhaps the single best piece of evidence that these three language families form a genetic unity, i.e. Altaic.

Tungusic languages have SUBJECT-OBJECT-VERB CONSTITUENT ORDER. The order shows up with great regularity in narrative texts and is almost inevitably the order given in elicited sentences. However, other orders are possible when a speaker is topicalizing or focusing non-subject constituents.

As is expected with OV languages, Tungusic tends to place MODIFIERS BEFORE HEADS. Thus, adjectives, numerals, and relative clauses precede the nouns they modify; adverbs precede verbs; degree words precede adjectives; and so on. Some violations to this patterns do occur. For example, non-finite verbs follow tensed auxiliary verbs contrary to expectations.

Despite these regular patterns in the linear order of elements, constituent order does not seem to be rigid in Tungusic as it is in languages such as English. In many Tungusic languages, speakers tend to allow nearly any permutation of words when asked. This fact suggests that Tungusic languages are marked by WEAK CONFIGURATIONALITY, which is typical of languages with relative free constituent orders. One interesting research issue in Tungusic is whether one can find evidence for major phrase types (verb phrase, noun phrase, adjective phrase, and adpositional phrase), which are assumed to be universal in some syntactic theories.

CONVERBS are another feature of Tungusic languages. These are verb forms which are subordinate to a main verb and express their subordinate status through verbal affixation. In many ways, a converb is similar to a participle. One major difference, however, is that converbs indicate the semantic relationship that they have to main verbs, whereas the semantic relationship between adverbial participles and main verbs is not explicit and must be determined from context. An example of an Oroqen converb is given in 7.

7) dJanda-ksan dJ´bt´-r´-n
sing-CONVERB sing-PROG-3S
After singing, she ate.

The suffix -ksan, which appears exclusively on converbs, signals that the action of the converb occurs prior to the action of the main verb (NB: in languages with participles, the temporal relationship to a main verb can also sometimes be expressed, but this is done with tense morphology which is not restricted to participles).

Another defining property of Tungusic syntax is the existence of a TENSED NEGATIVE AUXILIARY. Rather than employing a particle to negate verbs (as in most Indo-European languages), Tungusic languages use an auxiliary (e-)that is inflected for tense and subject agreement. The main verb, in this construction, is also marked for tense, but it does not take subject agreement (8).

8) ´-tS´-n ´m´-r´
She didn't come.

© Trustees of Dartmouth College
Last updated 20 Nov 1998