The Language of the Modhupur Mandi (Garo), Volume 1

Verb Bases, Verb Stems, and Verbs

(A)

pp. 105

The Garo language allows two kinds of sentences. The great majority have a verb as their major constituent, but a few consist of nothing more than two noun phrases. These are "equational sentences" which say that the two noun phrases refer to the same thing. To express this meaning in English we have to link two noun phrases with some form of the verb be , but no such linking verb is needed in Garo: U-a me'-chik Bang-la-des-ni 'That woman (is) a Bangladeshi'. Equational sentences will be considered in the final chapter of this book. With the single exception of equational sentences, a verb is an obligatory constituent, and the only obligatory constituent, of every sentence.

Verbs are constructed from constituents of four types, two of them obligatory and two optional. First, is what I will call a "verb base". This conveys the central meaning of the verb, and it is the piece of the verb to which the other constituents, most of them suffixes, are attached. The other obligatory constinuent is the principal verb suffix. This may be a tense-aspect suffix that completes not only the verb but also the sentence, or it can be a subordinating or nominalizing suffix that shows how the verb is related to other parts of the sentence. Nothing else than a verb base and a principal verb suffix are needed to form a complete verb. The other two constituents are optional. The adverbial affixes fit between the verb stem and the principal verb suffix. A single verb can have anything from no adverbial affixes up to a half dozen or so. They modulate the meaning of the verb in various ways, often in ways that resemble English adverbs. Finally, a verb may have, but does not need to have, one or two terminalPage  106verb suffixes that come after the principal verb suffix. Terminal verb suffixes are used for such purposes as asking questions, expressing uncertainty, and citing another person's speech. Kat-no-a 'will run' has two meaningful parts: the verb base is kat 'run', and the principal verb suffix is -no-a 'future'. Kat-ang-no-a 'will run away' has a third constituent added to the first two, the adverbial affix -ang- 'in the direction away from the speaker'. This fits between the two obligatory parts of the verb. Kat-ang-no-a-ma? 'Will (he) run away?' adds the terminal verb suffix -ma that turns the sentence into a question. This chapter describes the two obligatory parts of the verb, while the optional affixes will be described in the chapter that follows.

The verb base is generally the first meaningful piece, or "morpheme" of a verb, as it is in kat-no-a 'will run' where kat- is the verb base. Mandi has a few bisyllabic verb bases, such as a-gan- 'say', at-chong- 'sit', and chan-chi- 'think, believe'. These consist of a single mropheme and they cannot be further divided into smaller meaningful parts constituents. Most Garo verb bases, however, are monosyllabic, consisting of just one syllable. The language has many hundreds of monosyllabic verb bases. These convey meanings of many sorts: states, movements, qualities, acts, manipulations and much else. But even hundreds of verb bases are not enough to convey all the meanings that people want to express, so large numbers of verbs include one or more morphemes that are added to the verb base, and that extend and refine the meaning of the verb bases. These are "derivational affixes". They are included among the larger group of "adverbial affixes and they will be treated in much more detail in the next chapter, but it needs to be stressed now that the affix that immediately follows the verb base often forms such a tight unit with the base that the pair needs to be considered as a lexical item. This combination of verb base and derivational affix is what I will call a "verb stem". The derivational affix typically has a fairly clear meaning of its own which it contributes to the meaning of the verb stem, but the meaning of the pairs cannot always be fully predicted from the meaning of their parts. An analogy from English would be a verb base such as design , which can attract a derivational prefix re- , to yield the verb stem redesign . The meaning of redesign is largely predictable from the meaning of its parts, but the meaning of some other verbs, such as recycle is less so. In English the derivational affix often precedes the verb base. In Mandi, the derivational affix characteristically follows.

For example, -chak- , is a derivational affix that often has a meaning that suggests some sort of action toward someone else. It often implies either help of some sort, or a reciprocal action, but the exact meanings of the verb stems in which it is used need to be separately described and separately learned: dak-chak-a 'help' from dak-a 'do'; hai'-chak-a 'be understanding,Page  107perceptive' from hai'-a 'know'; a-gan-chak-a 'answer, reply' from a-gan-a 'speak'; ron'-chak-a 'give help' from ron'-a 'give'; ka'-sa-chak-a 'sympathize' from ka'-sa-a 'love'; dok-chak-a 'hit back, return a hit' from dok-a 'hit, strike'. Thousands of verb stems of this sort are found in the language. In most respects these compound verb stems act just like simple verb bases, and they can take the full range of later affixes.

It is important to be clear about the difference between a "verb base", a "verb stem", and a "verb". The verb consists of a verb stem together with a principal verb suffix. Only a complete verb can be used as a word. There are several kinds of verb stems. If there is no derivational affix the verb base by itself acts as the verb stem, but there very large numbers of compound verb stems that have a derivational affix in addition to the verb base. Both simple verb bases and compound verb stems need a principal verb suffix to complete the verb.

Many, though by no means all, of the derivational affixes have an independent existence as a verb base. For example, chak- , as a verb base means 'support, side with, protect, defend, contribute to'. This is clearly related to the derivational affix used in the examples given just above, but most verb bases cannot be freely (productively) used as adverbial affixes and even those that can be used this way cannot generally be affixed to all verb bases, but only to a selection of them. The constructions of a verb base with a derivational affix, in other words, are conventional lexical items that must be learned by each speaker. Many examples will be given in the next chapter.

Mandi has two other kinds of compound verb stems in addition to those formed from a verb base and a following derivational affix. One of these has a first syllable that is derived from a noun. The other begins with a classifier. Neither of these is nearly as common as the verb stem with a following derivational affix, but they complicate the picture.

It was pointed out in the previous chapter that when an indefinite direct object immediately precedes the verb it does not take a case marker. Bol-ko den'-a means 'chop the wood'. The accusative marker -ko shows that some particular wood is being referred to. Bol den'-a , without the case marker, means 'chop wood, do wood chopping' where no particular wood is pointed to. The sentences are distinguished in their form by whether or not there is an accusative -ko , and they are distinguished in their meaning by whether the wood is definite or indefinite. Such phrases can be freely constructed, and nothing forces us to count the constituents, in this case bol and den'-a as a single word rather than two. Sometimes, however, the use of a particular noun with a particular verb has become so conventionalized that it seems more natural to consider the pair to belong to a single word.Page  108

The verb jak-su-a 'wash hands', which was described in the previous chapter, includes jak which means 'hand' and which is used easily as an independent noun. It is easy to identify -su- as meaning 'wash', even though Mandis never use it by itself as an independent verb. Thus jak-su-a needs to be considered as a single word. Jik-gal-a , from jik 'wife' and gal-a 'throw away', and se-gal-a , from se 'husband' and the same gal-a , have been conventionalized to mean 'divorce'. Song-dong-a from song 'village' and dong-a 'be at' means 'settle a village'. Gi-sik-pil'-a , literally 'mind-reverse' means 'reform'. In all of these, a noun has been incorporated into the verb stem as its first constituent. In some cases, the part of the construction that looks like a noun does not even have the same form as it does when it is used as a separate word. Mik-chip-a is formed from the root of the noun meaning 'eye' and chip-a 'close', but the word for 'eye' is mik-gon or mik-ron and mik is never used without being attached to something. The ku- of ku'-dal-it-a 'enlarge a mouth, such as the mouth of a basket' (lit. 'cause the mouth to be big') is clearly the root of the noun ku'-sik 'mouth'. The same ku'- when used with rak-a 'strong' in ku'-rak-a means 'speak loudly, shout'. Since ku'- is never used except in compounds, it cannot be interpreted as a separate word.

In some cases, nouns that do not appear to be the logical object are incorporated into the verb. The following examples are typical: ja'-neng-a 'tired, as from walking', (lit. 'leg-tired'); jak-neng-a 'have tired hands'; han'-sel-a 'healthy', (lit. 'body-healthy'); Jak 'hand, arm' is a special favorite as the first member of such compounds. In addition to jak-su-a 'wash hands', we find jak-rak-a 'quick, strong', (lit. 'having strong or quick hands'). jak-ro-a 'long, of sleeves', (lit. 'arm-long'); jak-wat-a 'let go of', (lit. 'hand-release'); jak-si-ot-a 'point to, point out'.

A third type of verb stem has a morpheme that is otherwise used as a classifier as its first constituent. The most common words of this type are formed from a classifier plus either -dal-a 'big' or -chon-a 'small', as in king-dal-a 'large, of flat things' where king- is the classifier for flat things; rong-chon-a 'small of round things' and many others. This is not a fully productive construction. *Sak-dal-a which one might suppose would mean 'big of a person' is never used. A few other words with adjective like meanings can form verbs with classifiers, but these are much less common than those with -dal-a or -chon-a and they are quite idiosyncratic: king-kil-ak-a 'pull up, of pant legs or a lunggi', 'open, of a book' ( king- 'classifier for flat things'); ding-rak-a 'strong, of long thin things' ( ding- 'classifier for long thin things', rak-a 'strong'); rong-git-ing 'not fully cooked, as rice' ( rong- 'classifier for round things', git-ing 'raw'); mik-to-a 'measure by forearm lengths' ( mik the length of a forearm, to-a 'to measure').Page  109

All three kinds of compound verb stems can take the same kinds of suffixes as simple verb bases: adverbial affixes, principal verb suffixes and terminal suffixes. They differ in their internal composition, not in the larger constructions into which they enter.

Parts of Speech: Verbs and Adjectives

(B)

pp. 109

The class of verbs can be defined as consisting of all those words that include a suffix such as -a 'neutral tense', -no-a 'future', -bo 'imperative', or any one of a dozen or so other suffixes. Many of these words have meanings that correspond to the words we call "verbs" in other languages, such as 'run, walk, cry, see, think', and 'worry'. As in most, perhaps all, languages, verbs form one of the two largest classes of words, and Garo has thousands of them. Nouns, of course, form the other large class of words, but Mandi nouns are formally distinguished from verbs, by taking a different set of suffixes and by entering into different kinds of constructions.

With verbs defined in this way, Garo is a very "verby" language. To put it less frivolously, Garo accomplishes many things with verbs that some other languages accomplish with nouns or other parts of speech. Among other things, most meanings that are conveyed by adjectives in English are conveyed in Garo by a type of verb. As a result, Garo hardly has a separate class of adjectives. Verbs take a much more complex set of affixes than nouns. Nothing but a verb is essential to make a complete sentence. Verbs can also be put into a form that can modify another verb, and most adverbs are derived from verbs.

Most English nouns are translated by Garo nouns, and most English verbs are translated by Garo verbs, but if we classify words as either nouns or as verbs according to the suffixes that they can take and by the larger constructions into which they enter, rather than by their meaning, it turns out that the ranges covered by Garo word classes and English word classes are by no means identical. The most striking difference between the languages is in the words that translate English adjectives. These take the same suffixes as verbs and so, by definition, they are verbs. The word that most naturally translates 'big', for example, is dal'-a . A literal translation of dal'-a into English should be 'to be big' rather than simply 'big', and such a translation would make its verbal nature explicit. Dal'-a can take the full range of verb suffixes and it is used much like any other verb: dal'-no-a 'will be big', dal'-no-a-ma 'will (it) be big?', dal'-tok-jok 'all have become big', and so on. This parallels the constructions of other intransitive verbs such as kat-a 'run': kat-no-a 'will run', kat-no-a-ma? 'will (he) run?', kat-tok-jok 'all have run'.Page  110

Dal'-a can be used to modify nouns in much the same way that adjectives can be used in English, but so can words that translate English verbs. This means that the ability to act as modifiers does not distinguish Garo adjectives from Garo verbs any more than does the ability to take verbal suffixes: dal'-gip-a man-de 'big person', kat-gip-a man-de 'the man who runs, the running man'. Whether a meaning is conveyed by an adjective or a verb in English is irrelevant for the assignment of a Garo word to a part of speech, and the crucial criterion is the type of constructions that it enters. What we find is that meanings such as 'big, narrow, hot, run', and 'sleep' are all conveyed in Garo by words that can be used both to modify nouns and to act as the main verb of a sentence. Since all these words are used in much the same way, they all have to be considered to belong to the same part of speech. Most of the Garo words that translate English adjectives are intransitive verbs. This does not mean that all intransitive verbs have exactly the same characteristics, any more than all nouns have exactly the same characteristics, but it takes a close look to find the differences. One difference will be described below in the section on "Nominalizing Suffixes".

Garo does have a small set of words that convey meanings that translate English adjectives and that are grammatically idiosyncratic. This set includes gip-bok 'white', gi-sim 'black', git-dal 'new', and a handful of others. As the examples suggest, they can be used without a principal verb suffix, something that is impossible for most verbs, even most verbs that translate English adjectives. These " gi- verbs" will be described as a separate word class in Chapter 11, "Minor Parts of Speech".

Parts of Speech: Nouns and Verbs

(B)

pp. 110

Verbs and adjectives are less distinct in Garo than in English, but verbs and nouns are more distinct. A very large number of English words can be used as either a verb or a noun. Think of hammer, rain, bite, walk, view, fish, and string We can sail (v.) a ship with a sail (n.). We can run (v.) or go for a run (n.). We use a can (n.) to can (v.) beans, and so forth. In any particular example, a word can be easily identified as either as a noun or a verb, but we need to know how it is used in order to know its part of speech. Garo has far fewer words than English that can be used in these two different ways. With a few exceptions, a Garo word is used only as a noun or only as a verb. This leads to a number of differences in the way things are expressed in Garo and in English. In Garo, for example, mik-ka is a noun meaning 'rain' while wa-a is a verb meaning 'to rain'. It is quite natural to say mik-ka wa-ing-a literally 'the rain is raining'. This strikes English speakers as odd but it is simply the consequence of Garo's morePage  111consistent distinction between nouns and verbs, and it is really no more odd than saying the wind is blowing , an example in which English happens to distinguish the noun from the verb.

In English we can often use the name of a tool for the action it performs, but this is not possible in Garo where tools are named by nouns and their actions are described by verbs. Thus in Garo, one cannot say, the equivalent of 'I hammer with a hammer'. The tool is called hat-dur but dok-a 'hit, beat' is what one does with it: Ang-a hat-dur-cha dok-a 'I hit with a hammer'. Perhaps it is the impossibility of using tool names as verbs that requires Garo to have large numbers of quite specialized verbs for the ways things are manipulated. Since these precise meanings cannot be conveyed by using a tool name as a verb, specialized verbs are needed instead. In English, we can 'shovel dirt, hoe dirt, or rake dirt', each with a different tool. Garos can bak-a 'dig out dirt when the dirt is to be used for something', bil-a 'dig out or extract something, such as an edible root', cho'-a 'dig a hole, as for a post', kit-a 'dig out dirt for a large hole such as a grave or a well', and kong-a 'dig by scooping and scraping'. None of these verbs tell us what tool used.

The clearer separation of nouns and verbs in Garo does not mean that meanings are always assigned to nouns and verbs in the same way as in English. Wal 'night', for example can be used as a verb in Garo: wal-jok 'it has become night, night has fallen'. Most Garo words for thunder and lightning are verbs. English thunder can be used as either a noun or a verb, but Mandi hil'-hil-a can only be a verb. English lightning , can only be used as a noun, but the Mandi words for 'lightning' are verbs: el-ep-a, jil'-au-a .

There are exceptions to the clear separation of nouns and verbs. One exception is wal 'night', which can not only take verb suffixes, but also noun suffixes: wal-o-na 'until night'. Another is wal'-ku-a 'smoke' which, like English 'smoke', can be either a noun or a verb. The words for elimination are also exceptions: Mandi git-tam-a 'urine, urinate', ke'-em-a 'stools, defecate'. -a is usually a verb suffix, but these two words end in -a even when used as nouns. A bit more common than such identical forms are noun-verb pairs in which the verb is formed by adding a principal verb suffix to a verb base that is identical to the noun: bu-ra 'old man', bu-ra-a 'to become old, of men'; kram-chi 'sweat, perspiration', kram-chi-a 'to sweat'; jak-tom 'fist', jak-tom-a 'make a fist'; mi-dap 'a wrapped up portion of rice', mi-dap-a 'wrap up rice portions'. A'chik equivalents of the elimination words also have identical noun bases and verb bases: su-bu 'urine', su'-bu-a 'urinate', ki 'stools', ki-a 'defecate'. Even this kind of noun-verb pair is far from common in Garo, however. More often, nouns and verbs are entirely distinct.Page  112

Transitive and Intransitive Verbs

(B)

pp. 112

Like verbs in all languages, those in Garo fall into a number of subtypes that differ in the number of "arguments" that they can take. Intransitives can have a subject, but they cannot have an object. Kat-a 'run' and chon-a 'small' can have subjects, as in ang-a kat-a 'I run' and bi-a chon-a 'he is small' but they cannot have objects. A transitive verb such as nik-a 'see', however, can have an object, in addition to its subject: Ang-a bi-ko nik-a 'I see her', where bi-ko 'her' is shown to be the direct object by the accusative case marker -ko .

Just as Garo makes a shaper distinction between nouns and verbs than English does, so does it make a sharper distinction between intransitive and transitive verbs. A very large number of English verbs can be used either transitively or intransitively. A string can break (intransitive) or I can break (transitive) a string. A door can close (intransitive) or I can close (transitive) a door. Garo often has separate verbs for transitive and intransitive meanings, and when it does not, the transitive meaning must generally be distinguished from the intransitive by a causative affix. Some contrasting transitive and intransitive verbs are shown by the examples in Table 6—1. The examples suggest that some transitive-intransitive pairs are related. The first three pairs differ only in the b- of the intransitives and the p- of the transitives. Some Tibeto-Burmese languages have large numbers of verb pairs like these, with transitives and intransitives distinguished by the voicing of the initial consonant, but these are the only three pairs that I have come across in Garo. Perhaps they are the final remnants of a pattern that was more widespread at an earlier stage of the language. Git-chit-a , an intransitive, seems to be formed by adding the prefix git- to the transitive verb. The same relationship can probably be seen, even if a bit obscured, in pik-a 'pull out by the roots' (transitive), and gip-ik-a 'come out by the roots' (intransitive). Such prefixation is not a productive process in modern Garo, however, and this may be another example of a remnant of an earlier, more widespread pattern. Other intransitive/transitive pairs, such as chot-a/tet-a 'break (of long things such as string)' or kam-a/so-a 'burn', show no similarity at all.

	Transitive	Intransitive
burst	bret-a	pret-a
crush	bin-ek-a	pin-ek-a
break (solid things)	be'-a	pe'-a
break (string etc.)	chot-a	tet-a
boil	git-u-a	rit-a
come/pull out by roots	gip-ik-a	pik-a
tear	git-chit-a	chit-a
be covered/cover	git-dap-a	dap-a
come out/take out	ong'-kat-a	rong'-kat-a
dry	ran'-a	ram-a
tickle	ka'-kit-chu-a	rem-bu-a
burn	kam-a	so-a

When Garo does not have separate intransitive and transitive verbs it usually requires the difference to be shown by a causative affix. This is -et- or -it- in Mandi and -at- in A'chik: sat-ti krip-jok 'the umbrella is closed' (intransitive); ang-a sat-ti-ko krip-et-jok 'I have closed the umbrella' (transitive); a-bu-a 'to take a bath', a-bu-it-a 'to give a bath' as to a child.

Transitive verbs are almost always distinguished from intransitive verbs in Garo, but there are a few exceptions. I have heard speakers use am'-pil-a Page  113'turn over' both intransitively for what one does oneself, or transitively for what one does to something else. I have also heard wal-pil-a 'return' used as both a transitive and an intransitive. Nevertheless, one speaker insisted to me that these words could not be used transitively without adding a causative affix. In any case, examples of words that can be used without change both transitively and intransitively are quite rare. The much more common Garo practice is for transitive and intransitive verbs to be carefully distinguished.

Most transitives can have just one object, but Mandi also has verbs that can take both an indirect object and direct object: Bi-song ang-na boi-ko ron'-ing-a 'They are giving me a book'. In this example ang-na 'to me' is the dative form of the pronoun 'I' and it is the indirect object of the verb; boi-ko is the accusative of the word the Mandis use for 'book' and it is the direct object. Various other arguments, such as instrumentals and various kinds of locatives are also possible. These will be considered in the section on Case Markers in Chapter 8, "Nouns".

The subjects, direct and indirect objects, instrumentals, locatives and the rest are the "arguments" of Garo clauses and sentences. The arguments of Garo are much like those of English or any other language except that none of them is ever obligatory. English requires a subject, and English transitive verbs require both a subject and a direct object. In Mandi, by contrast, as long as one is willing to rely on the context to make the meaning clear, it is always possible to leave out arguments, even to use the verb without any arguments at all. This means that we cannot define a transitive verb as one that must take a direct object,Page  114but only as one that can take a direct object.

Incipient and Continuing Action

(C)

pp. 114

Many Mandi verbs have close English equivalents: kat-a 'run', mal-a 'crawl', nik-a 'see', and ni-a 'look' are all very close in meaning to their English translations. In some cases, however, one of the languages makes finer distinctions than the other. English, for example, more often distinguishes incipient actions from continuing states than does Garo. English distinguishes wake up , an incipient action, from be awake , a continuing state, but Garo uses the same verb for both: mik-rak-a . Nor does Garo distinguish get dressed (incipient) from wear (continuing), both of which can be named by the same Garo verbs.

Other distinctions are drawn more finely in Garo. No single Garo verb corresponds to English 'get dressed or wear', but the speaker must choose between several verbs according to the article of clothing that is being put on or worn, or according to the part of the body where it belongs. Sit-ik-a means 'put on or wear a hat'; chin-a is 'put on or wear a shirt or other garment for the upper body', especially when the garments are wrapped rather than put on by inserting a limb; gan-a is 'put on or wear a garment for the lower part of the body'; skrok-a 'put on or wear of bracelets, rings, shoes, socks', all of which require a body part to be inserted into the garment; ga'-a is 'wear or put on shoes', and so on. 'Lift', 'hold' and 'carry' are distinct in English, but the same Garo verbs can be used for all of them. However, the Garo verbs differ in the manner in which something is 'lifted, held' or 'carried'. Thus ol-a means 'hold from a tump line' (a strap across the forehead that supports a basket on the back) or 'carry by means of a tump line', or even 'lift it up to the position for carrying by a tump line'; rip-e-a means 'lift, hold or carry on the shoulder, as one carries a log'; ba'a means 'lift, hold or carry in a cloth, especially a child'; ke-a means 'lift, hold or carry something from a strap'.

Just as English has ways of distinguishing 'carry on the shoulder' from 'carry by a strap' when this is necessary (as I have just done in this very sentence), so Garo has ways of distinguishing incipient and continuing actions. Generally this is done by means of verbal affixes rather than by entirely separate verbs, however. (See -ba- and -ang- under "Adverbial Affixes" in Chapter 7). When the meaning is clear, as it often is, it is not necessary to make the distinction explicit.

Core Verbs

(A)

pp. 114

Garo has thousands of verb bases and verb stems. Many, of course, have quite specialized meanings, but like English, Garo has a few veryPage  115common verb bases that have very general meanings. The meanings of some of these overlap the meanings of such core English verbs as have and be , but none of them are simple translations of the English verbs. Several core verbs are common enough to deserve special notice. They are described here in roughly the order of their frequency, the most common first.

Dong-a 'to be at, to exist, to have, there is, there are'. The meaning of dong-a overlaps several different expressions in English. Its most central meaning is 'to be in existence' or 'to be at a particular place' but depending upon the context it is often most naturally translated by 'there is', 'there are' or 'have'. Its meaning will be most easily understood from examples:

A sentence that has dong-a as its verb often includes a locative noun phrase (a noun phrase with the locative suffix -o ) as one of is arguments. That phrase is often best translated by a prepositional phrase ('at the village', 'on the table'), but sometimes its most natural translation is as the subject of the verb 'have': Ang-o dong-a 'I have it', 'It is at me'. The answer to questions such as Mi dong-a-ma? 'Do you have any rice?', 'Is there any rice' is likely to be either Dong-a 'There is, It exists, Yes', or Dong-ja 'There is none, It does not exist, No'. The important uses to which dong-a is put make it an exceedingly common verb, probably the most common in the language.

Ong'-a 'to be, to be the same as, to be equivalent to, to be true'. Among verbs ong'-a is probably second in frequency only to dong-a , andPage  116it translates a different sense of English be . It shows equivalence, that two things are the same or, when negative, not the same.

When ong'-a has the neutral tense-aspect marker, it is does not really need to be used at all, because Garo allows equational sentences that have no verb. Ang-a Man-di means 'I am a Mandi' and it usually needs no help from the verb ong'-a . Ang-a Mandi ong'-a would most likely be used only when it served to correct some contrary assumption. If someone had suggested that you might be a Bengali, the ong'-a would make a strong assertion, and correct the mistaken assumption, but a strong assertion could also be made by a forceful intonation without the ong'-a .

Lacking a verb, equational sentences have no way to convey the meanings that are carried by the verb suffixes. If the meaning of a verb suffix is required, ong'-a can be added to hold the suffixes, and it can be understood, in part, as a device that allows the suffixes to be used:

Bi-a Man-di ong'-ja. 'He is not a Mandi'.

Ang-ni su-e-tar ong'-ja. '(It) is not my sweater'.

As with any verb, it is quite possible to use ong'-a without any noun phrase arguments at all: Ong'-a-ma ong'-ja? 'Is it (the case) or isn't it? Is it true or not?'

Dak-a 'do', 'perform', 'act', 'make', 'construct'.

Dak-a is also used as an empty verb, a sort of dummy place holder that the language requires when there is no specialized a verb for a meaning thatPage  117is conveyed only by a noun (see the next section). Gu-ri is a noun meaning 'fog'. The verbal sense of 'to be foggy' can be conveyed by gu-ri dak-a literally 'fog does'. Bo-dol dak-a 'exchange', 'make an exchange'.

In addition, dak-a is often used with adverbs in a construction that can be seen as allowing the adverb to be used as a verb. Dak-a makes the subordination of the adverb unambiguous:

Expressions with dak-e become a part of longer sentences:

Man'-a 'to be able to, accomplish, achieve, manage, finish'.

U'-ko man'-no-a-ma? 'Will (you) be able to do that?'

Man'-jok . '(I) have done it, (I) have managed'.

Man'-a often follows an infinitive verb where it acts quite like an auxiliary:

Like dak-a , man'-a is sometimes used as an empty verb to allow a noun to assume a verb-like role: bal-wa man'-a 'wind does', 'wind blows'; ok man'-a 'be sick to the stomach' from ok 'stomach'; sol-di man'-a 'have a cold', from sol-di 'head cold'. In these examples the nouns are the subjects of man'-a rather than the objects, so 'a cold does' and 'the stomach does' would be more literal translations than 'have a cold' and 'be sick to the stomach'.

The rare Mandi passive is sometimes marked with man'-e a subordinating form of man'-a :Page  118

Pi'-sa a-ma-cha gam rim'-na hit-a man'-e cha'-a. child mother-Inst work (n) work-Inf (v) order-Neut able-Sub eat-Neut 'The child is ordered to work by (his) mother'. (lit. 'The child eats an order by (his) mother to do work'.)

Rim'-a, ka'-a 'work, act'. In its literal sense rim'-a is the usual Mandi word for 'work'. Ka'-a is the usual A'chik equivalent. Ka'-a is used in a number of fixed phrases by Modhupur Mandis ( bi-a ka'-a 'to get married'; ja'-chol ka'-a 'reach an agreement about marriage arrangements') but rim'-a is more usual for 'work'. The meanings of ka'-a and rim'-a overlap with that of dak-a . All can mean 'do' but dak-a has a somewhat more concrete sense and extends to 'make'. The meanings of ka'-a and rim'-a extend instead to 'work'. Both rim'-a and ka'-a exemplify the distinction between nouns and verbs. Kam and its variant gam are nouns meaning 'work' while ka'-a and rim'-a are verbs. It is quite natural to say either gam rim'-a in Mandi, or kam ka'-a in A'chik, both meaning, literally, 'work work'.

In A'chik, ka'-a also has the special job of helping to borrow verbs from Bengali and English. While Garos generally borrow words easily, A'chik speakers can be a bit reluctant about attaching Garo verb suffixes to a borrowed verb. When they want to incorporate a foreign verb into their speech, however, they can place ka'-a after the borrowed verb and attach the Garo verbal affixes to the ka'-a : o-pe-ret ka'-a 'operate, do surgery'. Ka'-a then becomes a device for incorporating borrowed words. Even in Mandi a few such phrases can be heard, such as ep-lai ka'-a 'apply', but the construction is used less freely than in A'chik. Bangladeshi Mandis are more often comfortably bilingual than A'chik speakers are, and they have less hesitation about using Mandi verbal suffixes with borrowed words, and less need for an empty verb to hold the affixes.

Empty Verbs

(B)

pp. 118

The sharp separation between Garo nouns and verbs means that it is generally impossible, in any simple way, to use a noun as a verb. Unlike English where hundreds of nouns such as shovel, water, ship, and sail are easily used as verbs with closely related meanings, Garo nouns generally remain nouns. In compensation, Garo has a very large number of verbs with quite specialized meanings that do some of the work that is done by English verbs such as shovel and ship . The language cannot have a verb that corresponds closely to every noun, however. As a result nouns are often coupled with a verb that carries little or no meaning of its own, but that holds the verb suffixes and, in a sense, allows the noun to be used in aPage  119verb-like way. Since subjects are not obligatory in Garo, the language has no need for an empty pronoun to act like English it in it is raining . It does need empty verbs.

For example, ga-dil-a is a noun meaning 'cloud' or 'clouds'. Ga-dil-a dak-a means literally 'clouds do' but it is used where English has 'it is cloudy. Where English has an empty subject, Garo has an empty verb. Similarly chak-ri is a noun borrowed from Bengali meaning 'one who works for wages' and chak-ri dak-a is 'do wage work', 'work for wages'. The language has several alternate empty verbs, and more than one can sometimes be used with the same noun. Gam is a noun meaning 'work'. 'Have (you) done the work?' can be expressed in Mandi as Gam-ko wa'-jok-ma?, Gam-ko dak-jok-ma? or Gam-ko rim'-jok-ma? . A'chik speakers are more likely to say Kam-ko ka'-jok-ma? . None of these verb bases adds much meaning of its own, but they can take any of the verb suffixes. The suffixes do add meaning, and they let the sentence be completed.

Empty verbs, or verbs that are almost empty, provide a way of incorporating borrowings, not only verbs but also nouns, from other languages into Garo. There is some resistance to attaching Garo verb suffixes to borrowed verbs, and it is even more difficult to turn a foreign noun into a verb by attaching Garo verb suffixes to it. The solution is to use an empty verb to hold the suffixes. Mik-ron ga-na ong'-jok (lit. 'eye(s) blind became') uses the Garo equatative verb ong'-a as a way of incorporating the Bengali ga-na 'blind' into a Garo sentence. Mas-ter Mo-hen-ko sa-sti on'-a , literally 'The teacher gives Mohen punishment' or, more naturally, 'The teacher punishes Mohen', uses the verb for 'give' to incorporate the borrowed noun sa-sti 'punishment'.

The verbs used in constructions of this sort are not always entirely empty, though their meaning is almost always less specific than the nouns with which they are used. Ku'-chi dap-a and ku'-chi gal-a can both be used where an English speaker might use 'spit' as a verb. Ku'-chi is a noun meaning 'saliva' and it is used here with either dap-a 'cover' or gal-a 'throw away', effectively allowing ku'-chi to be used in a verbal expression meaning 'to spit'. In the examples, dak-a, man'-a, ka'-a, rim'-a and even ong'-a add little meaning that is not already given by the noun. They are required by the syntax, because except for equational sentences, every sentence needs a verb. When verb suffixes are needed, some verb is essential, and these verbs of rather general meaning are pressed into service.

Other verbs can be used in similar ways but contribute a bit more meaning. Cha-a is a verb that can mean 'grow' but in nok-ma cha-a it means 'act as a nok-ma (big man, headman)'. Ga-a often means 'climb'. Ma-rang ga-a means 'taboo exists, appears', where ma-rang means 'taboo'Page  120or 'pollution'. Ga-a may imply that the ma-rang rises or climbs to envelop people, but its literal meaning has been diluted and the more general meaning is simply that some sort of taboo is in force. 'Foggy' can be either gu-ri dak-a 'fog does', or gu-ri on-a . On-a means literally 'go down, descend' which, perhaps, is how fog arrives. Ra'-a 'take, bring' looses most of its literal meaning in the phrase heng-gok ra'-a 'to snore' formed from the noun heng-gok 'a snore'. So a very considerable range of verbs can pull nouns into a verb-like role. Some add almost no meaning of their own, others add a little, and of course in many sentences the verbs contribute at least as much meaning as their arguments. The verbs may even be so explicit as to let the arguments to be omitted entirely. Verbs are not so easily omitted as nouns and noun phrases. Mandi does not need an empty pronoun like the English it . It needs a great many empty verbs.

Principal Verb Suffixes

(A)

pp. 120

In addition to its verb stem, every verb requires one principal verb suffix, and together, these form a complete verb. Principle verb suffixes come in three varieties, each of which forms a different type of verb with a different role in the sentence:

1. Sentence completing suffixes. Except for equational sentences, every sentence requires one verb that has one of these sentence completing suffixes. Most often this is the last word in the sentence. This means that (again with the exception of equational sentences) it impossible to form a Garo sentence without a sentence completing suffix.

2. Subordinating suffixes. These turn a verb, together with any subject, object or other arguments that may come with it, into a form that can be subordinated to another verb. They provide one important way by which complex sentences can be built up from simpler parts.

3. Nominalizing suffixes. These turn a verb into a noun, or put the verb into a form that can modify a noun. Like subordinating suffixes, the nominalizing suffixes provide a way to build up complex sentences from simpler parts.

Since (almost) every sentence requires a sentence completing verb suffix, they are the most essential of the three types of principal verb suffixes, but it would be impossible to speak for very long, or to use anything beyond very simple sentences, without the help of subordinating and nominalizing suffixes. This section focuses on the principal verb suffixes that are used in the Mandi dialect, but many of these are used in A'chik as well. I also include several additional A'chik suffixes that are rarely heard in Mandi. Speakers generally understand the forms used in dialects other than theirPage  121own, so they are all part of the speaker's, or at least the listener's, passive linguistic knowledge, even when they are not actively used.

Sentence Completing Suffixes

(A)

pp. 121

Suffixes that can complete a sentence fall into two major subtypes and a few leftovers. First are tense-aspect markers that express the time or the manner of the event. Second are several imperatives. The third and minor subtype consists of some miscellaneous forms that fit into the same position in a verb as the others, but that less clearly belong to a systematic set of suffixes.

Tense-Aspect Suffixes . The tense-aspect suffixes give a temporal dimension to the verb and to the sentence. Mandi has a closed set of four of these tense-aspect suffixes. A'chik has five. Two of the A'chik suffixes are the same as in Mandi. Two have the same meaning but different form. Mandi speakers recognize these two A'chik suffixes as equivalent to their own, but rarely use them. The fifth A'chik suffix has no equivalent in Mandi.

-a 'neutral'. All dialects of Garo use -a where English would use a present tense, but -a can also be used in such a generalized sense that it amounts to a neutral suffix, one that indicates little more than the absence of any of the more specialized meanings that are carried by the other principal verb suffixes. Linguists might call it the "unmarked" tense. When citing verbs in isolation, even nonliterate speakers add the -a suffix, and as Garo has become a written language, it has seemed natural to use it as the form to list in dictionaries. When indicating actions taking place at the present time, -a is generally used along with -ing- the marker of the progressive. In the absence of -ing-, -a often indicates habitual action or a general situation, not unlike the simple present of English. In contexts where the meaning is clear, it may even be used where English would call for a past tense, but the language also has other, less ambiguous ways to indicate the past.

Sal-a-rik-it i'-ba-a. '(He) comes every day'.

Nok-o te-bil dong-a 'There is a table in the house'.

Man-de cha'-tok-ing-a. 'All the people are eating'.

Gai-ra skul-o po-ri-ing-a. '(She) is studying at Gaira school'.

-jok 'change of state'. -jok can often be translated by the English perfect tense (the tense formed with have ): ang-a cha'-jok 'I have eaten',Page  122 bi-song kat-tok-jok 'they have all run away'. For convenience, I will refer to -jok as the "perfect" tense marker, but the English perfect tense covers a wider range of meanings than -jok . Garo -jok means, specifically, a change of state. Thus ang-a cha'-jok really means 'my state has changed from one of not having eaten to one of having eaten.' A clumsy translation like this is forced on us most clearly when -jok is used with -ja- 'negative'. Ang-a cha'-ja-jok does not mean 'I have not eaten' but, instead, means 'I eat no more'. More precisely, the negative sentence means 'My state has changed from one of eating to one of not eating', so whether the verb is positive or negative -jok indicates a switch from one state to another. I'-ang-jok '(he) has gone', i'-ang-ja-jok '(he) does not go any more'; wal-jok 'it has become night', wal-ja-jok 'it is night no longer', tang-ja-jok 'not live any longer, changed from the state of living to the state of not living'.

-jok indicates not only that the state has changed but that it remains in that changed state at the time of speaking. Thus, to say i'-ang-jok means not merely that '(he) has gone' but also that '(he) has not yet come back again', not merely 'has gone', but 'is gone'. Ang-a cha'-jok means not merely 'I have eaten' but 'I have eaten sufficiently recently that I do not yet have to eat again'. I am in the state of not needing to eat. Though overlapping in meaning, the English perfect tense provides a much less than perfect translation of -jok .

-no-a, -no 'future'. -no is simply a phonologically reduced alternative of -no-a . It is especially likely in fast speech, but it is also perfectly pronounceable in slow and careful speech as well. Speakers recognize clearly that -no-a and -no are different in form but equivalent in meaning and use. This is probably the least problematic of the four tense-aspect suffixes since its meaning is a straightforward future:

-no-a sometimes implies desire or intent:

Ring-no-a-ma? 'Do (you) want to drink?'Page  123

Ring-no-a '(Yes, I) want to drink'.

Ring-ja-no-a '(No, I) do not want to drink'.

-gen 'future' (A'chik). The future is probably the most often cited example offered by Mandis who want to illustrate the way in which their dialect differs from A'chik. A'chik -gen is equivalent to Mandi -no-a , and every Mandi in Bangladesh seems to know that "they say -gen but we say -no-a ". -gen is used in both writing and speaking by most educated Garos in India, and in India it is spreading at the expense of other forms of the future. Even in the 1950's some educated Garos found it quite amusing to hear me using such a folksy suffix as -na-wa , still another alternative for the future.

By the usual standards of Garo, -gen is a highly irregular form. The A'chik negative future is -ja-wa instead of *ja-gen which would be regular but is impossible. -wa can be considered to be the form of the A'chik future suffix that is used only with the negative, while -gen is used only with non-negative verbs. Villagers in the area north of Tura, who speak a dialect that they call a "A'beng" or "Matabeng", use -na-wa as the positive future and, like A'chik speakers, use -ja-wa as the negative. Their dialect has a less irregular future than A'chik. Mandi speakers in Bangladesh, who use -no-a or -no for the positive future, and -ja-no-a or -ja-no for the negative future, have achieved complete regularity.

The meaning of -gen , like that of -no-a , is a straightforward future.

Ang-a kin-al-o re'-ang-gen . 'I will go tomorrow'.

A-chak chik-ja-wa . 'The dog will not bite'.

-na-jok 'immediate or intentional future'. This indicates that something is about to happen, or that the speaker is planning to do something:

-na-jok provides a polite way to announce one's intention.

Ching-a i'-ang-na-jok . 'We are about to/intend to leave'.

The negative of -na-jok parallels the meaning of the negative of -jok , for it means 'will not (do something) any more'. It does not mean 'do not intend' or 'not about to': cha'-na-jok '(I) am about to eat, I intend to eat',Page  124 cha'-ja-na-jok '(I) intend to eat no more, (I) am about to stop eating, I do not care for any more'. Like -jok, -na-jok , indicates a change of state, but it is a change of state that is about to take place or is in the process of taking place, rather than one that has already happened: Bol ga'-ak-na-jok 'The tree is about to fall, about to change its state from not yet fallen to fallen'; bi-a kam rim'-ja-na-jok 'he is about to stop working', 'he doesn't want to work anymore', 'he is about to change from a state of working to a state of not working'.

In both form and meaning -na-jok can be seen as combining the perfect -jok and future -no-a . Like -no-a , na-jok expresses future time. Like -jok it expresses a change of state. In form, also, it has a similarity to both of the other tense-aspect markers. The -jok 's of course are identical in the two suffixes. The -na of -na-jok is hardly identical to the future -no-a but it is close enough to suggest an etymological connection. The four Mandi tense-aspect markers thus form a tightly integrated set. jok shows change of state, -no-a shows future, -na-jok shows both, and -a shows neither.

-gin-ok 'immediate or intentional future' (A'chik). -gin-ok is a close synonym of -na-jok but it is more characteristic of A'chik, while -na-jok is the usual form heard in Mandi. Si-gin-ok 'about to die', re'-ang-gin-ok '(I) will now be going'. Like -na-jok , -gin-ok can probably be seen as combining two other tense-aspect suffixes, the A'chik -gen 'future', and -jok . The vowel change from -e to -i , is really a reduction to a shortened vowel and is found sporadically in other words. -ok has been more seriously distorted from -jok but the similarity between -gen -jok and -gin-ok is too close to be dismissed as mere coincidence, especially when we have the parallel example of na-jok .

These four tense-aspect suffixes are at the heart of the Garo, and Mandi tense system, but there is more to the expression of time that just these. In particular, the Mandi terminal verb suffix -ming 'past' and several adverbial affixes, -ing progressive, -ang- 'away', and -ba- 'in this direction', interact closely with the tense-aspect markers to allow the expression of complex and subtly different meanings.

-a-ha 'simple past' (A'chik). A'chik also makes heavy use of a fifth tense-aspect suffix that is rarely heard in Mandi and for which Mandi has no close equivalent, though its meaning is often conveyed by the morphologically more complex form -a-ming . -a-ha differs in meaning from -jok in indicating nothing about the present state. Bi-a kat-ang-a-ha means 'he ran away' but says nothing about whether he has come back or not. Bi-a kat-ang-jok means 'he has run away and is still gone'. -jok is also more likely to indicate the recent past, while -a-ha can refer as easily to 'long ago' as to 'just now'. When A'chik speakers use -ja- 'negative' together withPage  125 a-ha they collapse them together as -ja-ha , which means 'did not': chik-ja-ha 'did not bite'. The meaning of chik-ja-jok is 'does not bite any more'. Mandi speakers generally recognize and understand -a-ha even though they do not use it.

Imperative Suffixes (B). Mandi has several types of imperatives. These are formed by suffixes that occupy the same position in a verb that might otherwise be occupied by a tense-aspect suffix.

-bo 'positive second person imperative'. The normal way to tell or invite someone to do something is to use -bo as the principal verb subject. This can be used for an abrupt command, but imperatives are used as often when making an offer as when ordering, and they can then be very polite: At-chong-bo '(please) sit down' and cha'-ku-bo 'eat some more' can be very courteous, especially if said in a courteous tone of voice and with appropriate gestures. On the other hand, I'-ang-bo! 'Go away!' can be very abrupt and rude if said harshly or shouted.

Imperatives can be made stronger by adding -da , or more encouraging and courteous by adding -ne : i'-ba-bo-da! 'come here!', i'-ba-bo-ne! 'please come!' but the attitude of the speaker is conveyed at least as much by tone of voice and posture as by the added syllables. It is also possible, though not very common, to convey a rather abrupt imperative without any suffix. u-ko cha 'eat that'. This creates one of the very few situations in which a naked verb base or verb stem appears with no suffix at all. In accordance with the general rule that glottal stops do not appear in the final syllable of a word, Mandis drop the raka from a verb such as cha'-a 'eat' when it is used as a naked imperative.

-kan, ka-na 'third person imperative', 'let him', 'let them'. These give permission or instruct someone, but they do not necessarily imply a firm command. -kan and -ka-na are used only with third person subjects, explicit or implied, never with first or second persons. The two forms appear to be simple alternates. If they have any difference in meaning, I have failed to discover it:

Dong-kan. 'Let (him) be'.

I'-ba-ja-kan. 'Don't let (him) come'.

In a related, usage, -kan or -ka-na can be used as a subordinating suffix that typically subordinates its verb to a verb of saying or ordering. This will be described more fully in Chapter 13, "Subordination":Page  126

Occasionally -kan and -bo are used together. It seems that -kan-bo differs little in meaning from simple -kan :

Ji-ba ong'-kan-bo. 'Whatever, let it happen'.

Cha'-a-ri-kan-bo. 'Just let it be eaten.'

-et 'imperative'. Mandis use -et as an equivalent for either the second person imperative -bo or the third person imperative -kan , but -et is considerably less common. As a second person imperative it is a bit more forceful than -bo , even a bit abrupt, implying that something should be done right away:

As a third person imperative, -et is also more forceful than -kan or -ka-na , coming closer to giving an order than simply giving permission.

Bi-ko a-song-et. 'Tell him to sit down'

Bi-ko a-song-kan 'Let him sit down'.

-et is homophonous with one form of the causative affix and it seems to carry something of the causative or transitive force of that affix, but as an imperative it comes at the end of the verb in the same spot as the other principal verb suffixes. The alternative pronunciation -it- , that Mandis sometimes use for the causative, is not used for the imperative.

da'-, da'-. . .-a, da'-. . .-bo, da'-. . .-a-bo 'negative imperative'. The negative imperative in Mandi is most often shown by da'- prefixed to the verb, generally with a suffix on the verb as well. The suffix may be -a or -bo , or both of these together: da'-cha-a-bo 'don't eat', da'-i-ang-bo 'don't go'. Occasionally da'- is prefixed to a verb with no suffix at all, so that da'-kat, da'-kat-bo, da'-kat-a and da'-kat-a-bo are all possible ways to say 'don'tPage  127run'. da'- is the only productive inflectional prefix in the language, or at least this is true for Mandi speakers. A few speakers, generally younger ones I believe, can be heard to pronounce this as da'-a and to use it as a separate word. They act as if prefixes are simply beyond reasonable expectations for a Mandi speaker: da'-a ring-a 'don't drink'. When nothing at all is suffixed to the verb, the imperative is particularly forceful: da'-cha 'don't eat!', da'-at-chong 'don't sit!'. It is probably a bit easier to have a verb that is lacking any suffix with the negative imperative than with a positive imperative but even with the negative imperative a verb without a suffix is not very common. A few speakers insist that some suffix, whether -a, -bo or -a-bo is required with every negative imperative. Others ignore this restriction.

mo'-na. . .-a-bo . A rhetorically distinctive negative imperative can be made with mo'-na , which is more often a question word meaning 'why?', along with the negative imperative suffix -a-bo . The usual prefix da'- of the negative imperative is not used with mo'-na :

This construction might be thought of as meaning, literally, "why are you doing that bad thing", but its rhetorical sense is a negative imperative, telling someone not to do something. The mo'-na can be quite far from the verb, further than is allowed for da'-a , even for speakers who use da'-a as a separate word rather than a prefix.

-na-be 'negative imperative' (A'chik). This is the usual negative imperative in A'chik but it is rarely used by Mandi speakers. Dak-na-be! 'Don't do it!' U-a nok-o-na nap-na-be! 'Don't go into that house'.

Other Sentence Completing Verb Suffixes (C). The most frequently used and most important sentence completing suffixes of Mandi are the tense-aspect markers and imperatives, but the membership of Mandi word and affix classes is rarely sharply bounded. A number of items are always clearly and unambiguously members of a class, but there are generally others that are more marginal. The latter may be difficult to exclude onPage  128formal grounds, but they are so much less common and so much more specialized in meaning that they seem hardly to belong with the others. The two suffixes given here are less central than those described above. Among other things, they are simply longer than the suffixes already listed, and some sort of derivation from constituent morphemes seems plausible. As they are now used, however, the bits are fused together quite securely.

-a-ha-jok 'past'. This appears to combine -a-ha ,which is rarely used by itself in Mandi, with the very common -jok . It has a bit of the meaning of both of its parts. Like -jok , it describes a state that has changed and not yet reverted to its original condition. It can suggest a kind of finality. Like -a-ha , it refers to a point a bit further back in time than a simple -jok . It would be incorrect to say *da'-o-han re'-ang-a-ha-jok 'just now he went' because the immediate past referred to by da'-o-han 'just now' conflicts with the more remote time of -a-ha-jok :

Si-a-ko dap-a-ha-jok . '(They) buried the dead one'.

I have heard -a-ha-jok more often from Modhupur speakers than from others, and have had the impression that it is something of a Modhupur specialty, but I have been told that it is used in other areas as well. Even in Modhupur it is much less common than the tense-aspect markers listed earlier, but it is available to speakers and it offers a subtle difference in meaning from -jok . I find it odd that -a-ha-jok is used considerably more often in Modhupur than -a-ha is used by itself, even though -a-ha appears to be one of the constituents of the longer suffix..

-na-ba-gin-ang 'probably, probably will'. This looks like a complex morphological form that is constructed from several parts, but it is not clear where the parts come from or what contribution each makes to the whole, so it can only be treated as a unit. It is well known in Modhupur even if it is less common than the tense-aspect suffixes. -na-ba-gin-ang often, though not always, refers to the future. The terminal verb suffix -kon , which overlaps in meaning with -na-ba-gin-ang can refer to the past as easily as to the future. -na-ba-gin-ang also implies a somewhat greater likelihood than -kon : si-na-ba-gin-ang 'will probably die, seems to be about to die'. Ong'-na-ba-gin-ang 'probably will', can be used by itself in the sense of 'that is probably the case':

Bon'-ku-ja-na-ba-gin-ang. '(He) probably isn't finished yet'.Page  129

Subordinating Suffixes

(B)

pp. 129

The members of the second subgroup of principal verb suffixes fit into the same position of the verb as the tense-aspect suffixes, but instead of forming verbs that can end a sentence, they form verbs that can be subordinated to another verb. They provide one of the chief ways of building up complex sentences from simpler parts. They will be introduced briefly here, but the constructions into which they enter will be considered in more detail in Chapter 13, "Subordination".

-na 'infinitive', 'to, for, in order to'. -na provides a means for joining two verbs in much the same way as English infinitives do, though it has a somewhat wider use. It can mean 'in order to', and it can also be used to ask permission. The infinitive verb with the -na suffix always comes first, and the two verbs always have the same subject. As in English, verbs of wishing, needing, and ability are prominent among those which are used with infinitive verbs:

When -na has the meaning 'in order to' the verb that follows the infinitive is not limited to meanings such wishing and needing:

When asking permission, man'-a 'can, may' can follow the infinitive but the infinitive can also stand alone. When no verb follows, the infinitive often has a polite -ne suffixed to it:

Ang-a ni-na-ne. 'May I see?', 'I'd like to see'.

Page  130

-e, -e-ming, -e-min-a . Garo speakers often join what might otherwise have been independent sentences by means of various principal verb suffixes that are suffixed to the verb of the first clause, and thus come at the very end of that clause. The first clause is, in this way, made subordinate to the second, main, clause. The simplest of these subordinating suffixes is -e , typically pronounced -i in the Mandi dialect. -e-ming and -e-min-a ( -i-ming and -i-min-a in Mandi) are elaborate synonyms.

-e turns a verb into a form that can modify or be subordinated to another verb. Verbs with -e can act like adverbs, since they modify another verb, although they would rarely be translated into English as adverbs. It is often more natural to translate such verb sequences as if they were joined by an 'and':

Two or more verbs in -e can be used together.

-o-de 'if',-o-de 'although',-o-sa 'only if'. Like other subordinating suffixes, these three complete one clause, and are followed by another clause. Although they all begin with -o they do not have a locative meaning:

Na'-a re'-ang-o-ba, ang-a-de re'-ang-ja-no-a. 'Even if you go I will not go'.

Na'-a re'-ang-o-sa, ang-a re'-ang-no-a. 'Only if you go will I go'.

These three subordinating suffixes can also follow -jok or -na-jok as -jok-o-de or -na-jok-o-de , but they cannot follow -no-a 'future'.

Miscellaneous Subordinators . By comparison with the subordinating suffixes already given, those that remain are distinctly marginal. They are less common and have more specialized meanings. Nevertheless they fitPage  131into the same grammatical constructions as the others, so they need to be included with them.

-a-ri 'just, merely, right after'. The same -a-ri , or one that is homophonous and has the same meaning, is a common adverbial affix (see the next chapter). When it is placed last in a verb, however, it subordinates its clause to the next one, like any other subordinating suffix:

The following example has two -a-ri 's. The first subordinates its clause to the next verb. The second is an adverbial affix that is followed by the principal verb suffix -jok :

-mit-ing, -mit-ing-o 'while'. This affix is oddly specific in meaning for a principal verb suffix, but like any ordinary principal verb suffix, it is added directly to verb stems. Words such as skang-o 'before' and ja'-man-o 'after', which seem to fall into the same semantic area, are postpositions and have a very different syntax from mit-ing-o . When mit-ing is used without the locative -o it means 'right now, right in the midst of'. It might be said when calling out to someone to explain a delay: mi cha'-mit-ing '(I'm) in the midst of eating'. More often, mit-ing-o is used with the locative suffix. This nominalizes it and allows the phrase to become the argument of a verb. It is not so narrow in its time referent as -mit-ing :

-te. . .-te 'On and on'. This is used to indicate long lasting, continuous action, The same verb stem is used twice, each time suffixed with -te :Page  132 cha'-te cha'-te 'eating and eating', dak-te dak-te 'doing and doing'. It is most characteristic of a narrative story telling style where some sort of lengthy and time consuming action is described:

Nominalizing Suffixes

(B)

pp. 132

Nominalizing suffixes are the third type of principal verb suffix, and they put a verb into a form that can be used as a noun or that can modify a noun. Nominalized verbs are typically used in noun phrases, either as head nouns or as modifiers of head nouns, and like other noun phrase constituents, they can take suffixed case markers. Before nominalization, the underlying verb may have had one or more arguments (subject, object etc.), and these arguments can be carried into the nominalized construction. This is the way that relative clauses are constructed. Nominalization, along with subordination, helps to built up complex constructions from simpler parts. There are four important nominalizing suffixes: a-, gip-a, -a-ni , and -o .

-a . In nonnegative sentences, the neutral tense-aspect suffix and this nominalizing suffix are homophonous and both have the same unmarked or neutral character. Nevertheless, they differ in the constructions into which the enter, so they need to be distinguished. A verb nominalized with -a can be used as a noun. It can take the case markers and postpositions and act as the head of a noun phrase.

The most common use of a verb nominalized with -a , however, is to modify another noun. This is to say that these nominalized verbs fill the role of adjectives. In this role they can either precede or follow the nouns that they modify. The modifier-last position is probably the most neutral, least marked order, but there is little difference in meaning: nok dal'-a, dal'-a nok 'big house'.Page  133

The negative of the nominalizing -a is -gi-ja rather than -ja which is the negative of the tense-aspect marker. The fuller form surely helps the listener to recognize the word as a nominalized verb and to identify he construction into which it has entered:

skosim'-gi-jaha-bu-a. head wet-NNomz bathe-Neut 'bathe without a wet head, without getting the head wet'

gip-a . This nominalizing suffix overlaps in meaning with -a and the two are often interchangeable. However the homophony of the -a nominalizer with the neutral tense-aspect marker poses some danger of ambiguity. Probably for this reason modifiers formed with -a are most often found in relatively simple syntactic contexts. -gip-a has the considerable advantage over -a that its status as a nominalizer is unambiguous, and as constructions grow more complex it becomes more likely that -gip-a will be used in place of -a : dok-gip-a 'the one who hits'; dal'-gip-a a-chak, a-chak dal'-gip-a 'big dog'; Tu-ra-o-na kat-ang-gip-a me'-a-sa 'the man who ran away to Tura'.

Nor do -a and -gip-a have exactly the same meaning. Nominalizations with -a often convey a relatively abstract meaning, while those with -gip-a are more individualizing. -gip-a can often be translated as 'the one who' and it may pick out one example from many, as when pointing to a group of peopleand saying dal'-gip-a man-de 'the one who is big'. -gip-a often, though not always, implies an actor, often a human actor. The difference is shown clearly by two nominalizations of ra-sot-a , a verb meaning 'to butcher'. Ra-sot-a means 'butchering', 'the act of butchering'. A ra-sot-gip-a is a 'butcher', 'the person who does the butchering'.Page  134

Even when -gip-a does not refer to a human actor it does pick out particular individuals. In the next example books is indefinite. In the example that follows -gip-a indicates particular books:

The negative of -gip-a is -gi-ja-gip-a 'The one who is not', 'those who are not'. This combines -gi-ja , which is the negative of the simpler -a nominalization, with -gip-a :

Transitive verbs can modify nouns just as intransitivescan, but they are most naturally translated intoEnglish as relative clauses:

Page  135

Not only single verbs but even entire clauses can be nominalized. The arguments of the verb are then drawn into the resulting relative clause. The next sentence can be turned into a relative clause, as in the sentence that follows. In that second sentence the relative clause modifies man-de 'person':

As the examples suggest, the line between modification by an adjective and modification by a relative clause is less sharp in Garo than in English. However, as soon as arguments are pulled into a modifier along with the verb, the resulting clause is much more likely to precede the noun than to follow the noun that it modifies. As befits its greater complexity, it is also more likely to be marked by -gip-a than by -a . Single-word modifiers are somewhat more likely to follow the noun. Relative clauses will be considered in more detail in Chapter 12, "Complex Noun Phrases".

The details by which modification is accomplished reveal one difference among Garo intransitive verbs that parallels the difference between English verbs and adjectives. In the two following sentences dal'-a 'to be big' and kat-a 'to run' are used in what appears to be exactly the same way. Both of these sentences are fully grammatical. Sentences like this give us no grounds for setting adjectives apart as in any way different from other intransitive verbs:

However speakers allow only the first of the next two sentences. These two differ from the previous pair in having the modifier before the noun rather than after it. Page  136

Nevertheless, kat- 'run' can be placed in front of the noun by using the alternative nominalizer -gip-a . The following sentence is fully grammatical:

Dal'-a , dal'-gip-a and kat-gip-a can be placed on either side of the noun that is being modified. Only kat-a , is limited to the post-nominal position. The difference reflects the difference in the meaning of the two nominalizing suffixes. -gip-a is more individualizing. It has the meaning of 'the one that, the one who'. -a is more general and more abstract. Kat-gip-a ma'-su means very concretely 'the cow that runs'. *Kat-a ma'-su is impossible. In this detail, Garo makes a distinction among its intransitive verbs that parallels the English distinction between adjectives and intransitive verbs. In most respects, both kinds of Garo intransitive verbs are used in exactly the same way.

-a-ni . 'abstract nominalizer'. -a-ni may be related to, or derived from, the nominalizer -a with a genitive case marker which, like other genitives can modify nouns. When used as a noun, however, a verb nominalized with -a-ni has an abstract meaning. Such nouns are not so productively constructed as nominalizations in -a and -gip-a , and their meanings are not entirely predictable from the meanings of their parts: an'-seng-a-ni 'joy', from an'-seng-a 'to be happy'; dak-chak-a-ni 'help' (n.) from dak-chak-a 'help' (v.); i'-ang-a-ni 'going, trip' from i'-ang-a 'go'; cha'-a-ni 'food' from cha'-a 'eat'; chan-chi-a-ni 'idea, thoughts' from chan-chi-a 'think'; a-gan-grik-a-ni 'conversation' from a-gan-a 'talk, say' and -grik- 'each other'; in-grik-a-ni 'conversation, often angry, argument' from in-a 'say'; at-chong-chak-a-ni 'things with a place for sitting, chairs, stools, etc.' from at-chong-a 'sit' and -chak-a 'fit'; dok-a-sik-a-ni 'musical instruments', literally 'things beaten and blown'. Verbs with adverbial affixes can be nominalized with -a-ni : nik-ja-ing-a-ni 'things not seen'.

Verbs nominalized with -a-ni can modify nouns just as genitive nouns can. The noun sil-ni 'of iron' modifies sang-gong 'bracelet' in sil-ni sang-gong 'iron bracelet, bracelet of iron', and verbs with -a-ni can act in the same way: bi-a-ni sal 'wedding day', from bi-a 'marry'; mal-a-ni sam 'love potion', from mal-a 'crawl' and sam 'herb, medicine', literally 'medicine to make someone crawl':Page  137

Locative Nominalizers . -o is a common locative case marker with nouns. In addition -o and several other more complex suffixes that include -o are used to nominalize verbs. Unlike the homophonous case marker, the -o nominalizers always have a temporal meaning, never a spatial one.

The simplest of the locative subordinators is simply -o 'when, at the time of':

-o can also be coupled with two of the tense-aspect markers to form morphologically complex nominalizing suffixes. -jok-o 'at the time of finishing', 'when (past)' combines the completive meaning of -jok with the locative meaning of -o . -na-jok-o 'when (future)' combines the future meaning of -na-jok with the locative meaning of -o . -o cannot be used together with the future tense-aspect marker -no-a :

Locative nominalizers are more fully descried in Chapter 12, "Complex Noun Phrases".

Verbs and Postpositions

(C)

pp. 137

One important function of the nominalizing suffixes is to put verbs into a form that allows them to be used with postpositions. As might be expected, this is especially important for the temporal postpositions such as skang 'before' and ja'-man-o 'after', though other postpositions can also be used after nominalized verbs. While postpositions can follow nouns formed with any of the nominalizing suffixes, -a is the nominalizer most often used, perhaps because the postposition itself makes the construction sufficiently transparent to render the more elaborate -gip-a redundant. Postpositions generally follow a case marker so when they are used with nominalized verbsPage  138the sequence is: verb stem—-nominalizer—-case marker—-postposition. The use of postpositions with nouns and nominalized verbs can be compared in the following table. The examples with nouns follow a case marker. The examples with verbs follow the same case marker, but that, in turn, follows the nominalizing suffix -a , except for the last example, in which it follows nominalizing -o . Postpositions will be described in detail in Chapter 9, Nominals.

Postpositions with:	Nouns	Nominalized verbs
Follows genitive:	nas-ta-ni ja'-man-o	re'-ang-a-ni ja'-man-o
	'after breakfast'	'after going'
	ang-ni gim-in	neng'-a-ni gimin
	'because of me'	'because of being tired'
Follows dative:	ang-na gan-da	dok-a-na gan-da
	'because of me'	'because of the hitting'
Follows extended	me-ja-o-ni dip-at	neng'-o-ni dip-at
locative:	'since yesterday'	'since being tired'