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

Earlier chapters of this book have dealt with the most important parts of speech, including the two largest, nouns and verbs. Demonstratives, pronouns, and postpositions, all of which resemble nouns in some ways, have been described as constituents of noun phrases. Numerals, consisting of a classifier and a number, needed a chapter of their own. Several other parts of speech now need to be described. Of these, adverbs form by far the largest group and they, along with some adverb-like locatives, will be considered first. Next, a number of idiosyncratic adjective-like modifiers will be described, and then echoes and anticipations, interjections, courtesy expressions, and finally, conjunctions. The number of parts of speech recognized for a language depends upon how finely one wants to draw distinctions among words. Whether pronouns and demonstratives are considered to be nouns or distinct parts of speech, for example, is a matter of convenience, not fact. Pronouns resemble nouns in many ways, but differ in others. I find the categories that I describe here convenient, but they are not the only way to categorize the words of Mandi.

Adverbs

(A)

Adverbs form a very large word class, but the majority of them can quite naturally be regarded as having been derived from verbs. They are closely related to verbs, both in their form and in their meaning. Adverbs, however, take no suffixes at all, and in this respect, they differ as sharply from verbs as from nouns or even numerals. Adverbs often occur immediately before the verb of the clause in which they occur, but they can also occur earlier, with one or more noun phrases separating them fromPage  262the verb. They are distinguished from the noun phrases among which they occur in their lack of a case marker. Semantically they are distinguished by modifying the meaning of the verb rather than acting as one of its arguments. As the examples will show, adverbs can convey a very wide range of meanings.

It may not be obvious why I have labeled some words as "adverbs" since they may be most naturally translated into English by either adjectives or verbs. Jol'-kep-kep , for example, seems to be most naturally translated as 'narrow, especially of agricultural fields' but, unlike the English adjective, it is used to modify a verb rather than a noun. Bi-ni ha'-ba jol'-kep-kep dak-a , can be freely translated as 'his field is narrow' but a more literal translation would be 'his field is done narrowly'. Chau'-kok tim-e-tim-e a-song-ing-a might be translated as 'The thief is sitting and hiding' or even 'The hiding thief is sitting', but the literal construction is more like 'The thief is sitting hidingly, in a hiding fashion'. Tim-e-tim-e 'hiding' is an adverb which modifies a-song-ing-a 'sitting', and it describes the manner in which the sitting is being accomplished. To-ra bing'-chrip-chrip wat-bo , is more literally translated as 'weave the basket tightly' than as 'weave a tight basket' ( to-ra 'basket', bing'-chrip-chrip 'tight, of woven things', wat-a 'weave, of baskets').

Several kinds of adverbs can be distinguished by their form and by the manner in which they are derived from verbs. In particular, the majority of adverbs show some sort of reduplication, and many of these are transparently derived from verbs. The language has comparatively few adverbs that are not reduplicated and that consist of a single morpheme, and even the few words that can be used as simple adverbs can generally be used in some other way as well. The following words that lack any form of reduplication can be used as adverbs. They are a heterogeneous lot.

Skang is occasionally used in an adverb-like way to mean 'in the past'. Skang am'-pang-cha rik-a-ming '(We) formerly built with thatch'. More often, skang is used as a postposition, however: an-dal-na skang 'before dark'. Skang can also be used with a locative suffix, making it a "locative word" (see next section): Ang-a-de skang-o Tu-ra-o dong-a 'I used to be at Tura'.

Teng-re 'early, soon, fast, quickly' can be used as a simple adverb, as in teng-re cha'-gal-gal-bo 'eat quickly, soon'. It is frequently reduplicated to teng-re-teng-re , however, thereby assuming a more conventional adverbial form.

Da'-si 'a little while ago' and da'-an 'in a little while, a little while from now' are clearly related to each other and also clearly related to da'-o 'now', so they are not really monomorphemic. Nevertheless the second syllablesPage  263of da'-si and da'-an have no obvious etymologies and the two words are sufficiently specialized and lexicalized to be candidates for relatively simple adverbs. Like skang , they sometimes take a locative suffix.

Dam-san 'same, together, in one place' seems to be an adverb in na'-ching sak-gin-ing dam-san a-song-eng-a 'The two of us are sitting together', but dam-san is sometimes used with case markers and might be better considered a defective noun:

Teng-gil and a-gre can mean either 'very' or 'excessively': teng-gil ding'-ing-a 'very hot, too hot'. Not all speakers use both teng-gil and a-gre . My impression is that older speakers in Modhupur are more likely to use a-gre while younger speakers more often use teng-gil . A-gre is more common in A'chik. A-gre is also used as a postposition meaning 'in addition to, other than':

In addition to this mixed batch of adverb-like words, Mandi has borrowed a number of others from Bengali that are used more consistently as adverbs: kup 'very', be-si 'much', so-man 'evenly' etc. Much more characteristic of Mandi are the vast number of adverbs that have some overt indication in their form that they are adverbs. Their adverbial nature is sometimes shown by reduplication, sometimes by the suffix -e , and sometimes by both simultaneously.

Adverbs in -e . By using -e as the principal verb suffix, any verb can be put into a form that can act like an adverb and modify another verb. Strictly speaking, this is a construction that subordinates not only a verb, but an entire clause, to another verb. However, single verbs without any arguments are so often subordinated in this way that they need to be noted here as playing the role of adverbs: mik-tal-e chu-a 'sleep on one's back' ( mik-tal-a 'lie on one's back'); cheng-e don-ing-a 'put aside (a basket) before it is finished, when only the base has been woven', from cheng-a 'to weave the base of a basket'; dra-e i'-ba-a 'come on one's own, come against the will of someone else' from dra-a 'to force, act with force'.

The very common verb bang'-a 'to be much, many, a lot' and the equally common derived adverb bang'-e 'in large amounts' illustrate the adverbializing function of -e : man-de bang'-a 'there are a lot of people' (verb); man-de bang'-a i'-ba-jok 'a lot of people came' (nominalized verb); man-de bang'-e i'-ba-jok 'people came in large numbers' (adverb). As a verb base, bang'- can take any of the tense-aspect markers: in'-no bol-pang bang'-no-a 'there will many trees here'. As an adverb, bang'-e can modify a verb such as dong-no-a 'will be at': in'-no bol-pang bang'-e dong-no-a 'here the trees will be many' (adverb).

In the next example, bang'-a is a nominalized verb that modifies the noun man-de . In the example that follows, bang'-e is an adverb and itPage  264modifies the verb ron'-a-ming 'gave'.

Adverbs in -e are sometimes reduplicated, and thereby generally gain the meaning of 'long-lasting' or 'intermittent': dong-e-dong-e 'now and then'; gal-e-gal-e 'in part, omitting parts' ( gal-a 'throw away'); heng'-e-heng'-e 'widely spaced' (of the pieces from which a basket is woven, of rice plants spaced out after transplanting); tim-e-tim-e 'secretly, in hiding', where the implication is that the hiding lasts for some time.

Simple Reduplication . Many adverbs are formed by simple reduplication, sometimes of a single syllable, often of two. Examples of single syllable reduplication include chrap-chrap 'crowded together, as growing fruit'; bang-bang 'empty, of a hole, a building, of land cleared of trees'; dam-dam 'always at the same place', grang-grang 'with open spaces, as an open weave basket'; jring-jring 'always, repeatedly'; bom-bom 'lukewarm'; jem'-jem 'regularly, repeatedly, often'. Some two syllable adverbs have similar but not identical syllables: hang-jang 'without reason, with no purpose, randomly, hurriedly'; jrip-jrang 'soundlessly, silently'.

Perhaps an even larger number of adverbs are formed by reduplicating two syllables: tim-bong-tim-bong 'wide, of fields, built widely'; bi-dil-bi-dil 'growing densely'; dil'-mik-dil'-mik 'smokily'; se'-em-se'-em 'tiny'; re'-em-re'-em 'having a small pattern, such as the creases in one's skin'; rim-a-rim-a 'see or hear partially, unclear, blurred'; dil-ing-dil-ing 'flying about, as dust'; ting-ku-ting-ku 'squeal, of a pig, loud, of a rice pounder, of angry talking'; dim-prong-dim-prong 'swirling with a lot of smoke, dusty'; dip-u dip-u 'very soft'; pu-ta-pu-ta 'in drops, spotted'; tek-sram-tek-sram 'crumbly, brittle'; rik-rak-rik-rak 'reflecting, shining, of a mirror, tin roof, metal, water'. Ka'-sin-e-ka'-sin-e 'slowly' is a three-syllable reduplication of a word that is already shown to be an adverb by its suffixed -e . Three-syllable reduplications are less common that two-syllable reduplications, but clearly they are not impossible.

Sometimes each syllable of a two syllable word is reduplicated separately: ding'-ding-dal'-dal 'just cooked, ready to eat'; rau'-rau-tau'-tau 'speaking loudly and rapidly'; rik-rik-rak-rak 'restlessly'; so'-so-jeng-jeng 'going from this to that without finishing anything'. Notice that when aPage  265single syllable with a raka (glottal stop) is reduplicated the raka is not repeated in the second syllable, but if another syllable intervenes, the raka is reduplicated along with the rest of the syllable. Thus se'-em-se'-em has rakas in both the first and third syllable, while so'-so-jeng-jeng has the raka only on the first syllable. This follows the regular pattern of raka deletion in second, but not third, syllables.

Final Syllable Reduplication . Many adverbs have a reduplicated final syllable while earlier syllables remain without reduplication: jol'-kep-kep 'in a narrow manner, especially of fields, very thin, of a person'; bing'-chrip-chrip 'woven tightly, without gaps, of baskets'; sim-dim-dim 'having a dark color, as mountains when seen from away'; gong'-rak-sak-sak 'quickly, while running quickly'. Most adverbs of this type are derived from verbs which lack the reduplication: chong'-pong-pong 'growing out from a clump at the base' ( chong'-pong-a 'to grow out from a clump at the base, like bananas or bamboo'); mik-chok-chok 'having a long pointed face' ( mik-chok-a 'to have a long pointed face').

Incomplete Syllable Reduplication . Some adverbs have only partial reduplication of one or more syllables: rin-ek-sin-ek 'messy, knocked down, of growing plants'; bu'-rin-eng-bu'-sin-eng 'messily, of hair', pil-eng-pil-ung 'rocking back and forth, like a boat'; gu-rung-ga-rang 'aimlessly, of roaming about'.

Second and Fourth Syllable pairs . A considerable number of adverbs have identical second and fourth syllables, while their first and third syllables differ. In these adverbs, the third syllable is only a place holder, contributing no meaning of its own, but filling a phonological gap. The favorite third syllable is -ga- : ru-ra-ga-ra 'inconsistent'; ding-dang-ga-dang 'doing things in different ways'; dil-dek-ga-dek 'swing, wobble, move back and forth, of a bamboo bridge or a loose post'. Other third syllables occur, but less frequently: rit-dang-ban-dang 'separately, doing things in different ways'; geng'-gang-bak-gang 'disorderly, as of branches piled around'; re-wa-se-wa 'spoiled, filthy'. In some cases, alternate form of reduplication can be performed on a single underlying verb. Ring-reng-ring-reng and ring-reng-ga-reng 'swinging back and forth' are both derived from ring-reng-a 'to swing back and forth.

Finally, I have had heard two words that can only be described as having undergone double reduplication: dang'-dang-dang'-dang 'spaced out, as growing fruit' and sim-sim-sim-sim 'starting to get dark, starting to turn rotten'. Even though the Garo language came to this writer as a late second language, it feels natural to him that a syllable might be repeated four times. To imagine a syllable that had been repeated exactly three times is much more difficult.

It is not clear to me just how productively adverbs can be formed by reduplication from verbs. Speakers do not accept all of the reduplications that I have proposed, even when I follow common patterns, so there is no full productivity. On the other hand, I have heard many hundreds of reduplicated adverbs that conform to one of the patterns that I have described. The patterns seem quite varied, but of course an unlimited number of other patterns of reduplication might have been used instead. The patterns that are found in the language must come to seem very natural to its speakers, and it must not take much time or repetition for a new adverb to become established. My impression is that a considerable amount of both dialect and individual variation is found in the range of adverbs used. Speakers have denied the possibility of an adverb that I have heard someone else using. From this, I infer that new adverbs are coined fairly regularly, and that they become easily accepted as individual favorites, or established in one or another dialect. At the same time, large numbers of reduplicated adverbs are used in the same way by speakers from widely separated dialects, so they are by no means all idiosyncratic or local.

Locative Words

(C)

Garo has a large number of words that are so often suffixed with a locative -o that the suffix seems almost to have become a part of the word. Fully frozen -o 's are rare, however, so it would not be correct to consider these locative words to be a completely distinct part of speech. Rather, they are nouns of a special sort. They differ from other nouns in generally forming complete noun phrases all by themselves. Demonstratives, numerals, and modifiers can occur as constituents of most noun phrases, but they are rarely found in the same noun phrase as a locative word. An ordinary noun such as nok 'house' can occur with a locative case marker and still be associated with other words in a complex noun phrase such as i-a dal'-gip-a nok-o 'at this big house'. A locative word such as da'-o 'now', on the other hand, cannot be modified by a demonstrative, numeral or any other word. As this may suggest, these locative words are used in much the same way as some English adverbs, and they have a sufficiently distinct role to set them apart from other nouns, and even from other locatives. Like da'-o , many of these words have a temporal meaning.

The locative case marker seems rarely, if ever, to be totally frozen to the root. Even the da'- of da'-o 'now, at this time', is found in a number of related compounds such as da'-sal 'today' and da'-si 'a short time ago', although neither * da'-a nor * da is ever used alone with any related meaning. Other locative words that refer to points or periods of time include: me-ja-o Page  267'yesterday', min-i-o 'day before yesterday', am-bin-o 'tomorrow', a-tam-o 'afternoon, evening', wal-jat-chi-o 'in the middle of the night, midnight'. The extent of lexicalization that has taken place in these words can be seen in an expression such as am-bin-o-na skang 'before tomorrow' where am-bin-o 'tomorrow' has a dative case marker in anticipation of the postposition skang 'before'. Dative case markers cannot be freely suffixed to all words with locative -o , but in am-bin-o the -o has become so frozen to the base as to have almost become a part of it.

-cheng- is an adverbial affix meaning 'before'. The related cheng-o , with a nearly frozen locative, means 'some time ago, in the past' and it is used as an adverb. -ka-ri forms compounds that name the seasons of the year, as ding'-ka-ri 'hot season' and sin'-ka-ri 'cold season', and da'-sik-ka-ri means 'last year'. Ka-ri-o , almost always with a well attached locative, means 'next year'. Names for the days of the week are borrowed from Bengali, but when used by Mandis, they often carry the locative suffix: Suk-ro-bal-o 'on Friday'. These words are often compounded with sal 'day' and the locative suffix then moves to the end: Suk-ro-bal-sal-o . Months are most often followed by -ja 'month' and the locative -o : Mak-ja-o 'in the (Bengali) month of Mak (January-February)', Ep-ril-ja-o 'in April'. In one way or another, names for months and days of the week are usually used with an -o .

As discussed in Chapter 9, "Nominals", a considerable number of postpositions with both temporal and spatial meaning also occur regularly with the locative -o : sam-ba-o 'near, beside', gi-sep-o 'among', ja'-man-o 'after', sak-a-o 'above', and many others. These, too, have semi-frozen locative o 's. From a purely formal point of view, in'-o 'here' and un'-o 'there' are merely the locatives of the demonstratives i-a 'this' and u-a 'that'. These particular locatives occur in a remarkably wide range of phonological variants, however, and this suggests that they have been lexicalized with the special meanings of 'here' and 'there'.

Locative words with a temporal meaning always have -o as their case marker, and those with spatial meaning often do as well, but some spatial postpositions can be constructed with the spatial locative -cha ( -chi in A'chik), as well as with -o : ki'-sang-cha 'behind, in back of, toward the back'; mik-kang-cha 'in front of, at the face of, toward the front'; pak-sa-cha 'beyond, on the other side of'. All of these occur with -o as well: ki'-sang-o, mik-kang-o and pak-sa-o . -cha often indicates motion, either 'toward' or 'from', while -o is more likely to indicate a fixed position, 'at', 'in', or 'on'.

Defective Nouns

(B)

While the great majority of English adjectives must be translated into Mandi as intransitive verbs, the language does have a number of words that are used to modify nouns but that, in most respects, behave more like nouns than like verbs. These are far fewer in number than the intransitive verbs, but individually they are both important and very common. They are also highly idiosyncratic. Many of them of them can be regarded as defective nouns, since they share a number of characteristics of nouns, but they cannot be used freely in all situations where most nouns are used. Many of these words name a quantity such as 'all, whole, most, half, some, part, a little, other', and so on, but other more diverse meanings are found as well: 'true, important, same, different, left, right, length, breadth'.

If these words can be characterized at all, it is by their diversity, indeed their eccentricity. They differ from each other in so many ways that it is difficult to make generalizations about them. All can modify nouns, and most can take case markers and stand alone as a noun phrase. In these ways, they are like nouns. Since most of them do not take verb suffixes, at least not at all easily, they are very different from the much larger number of intransitive verbs that can also modify nouns. As modifiers, some of these defective nouns can be placed only before the noun while others always follow and most can occur either before or after, just like ordinary modifying verbs. Most can take a -gip-a suffix and most can be reduplicated. Unlike nouns, however, many of these take a plural suffix only with difficulty or not at all. These are important and much used words. Perhaps only words that are urgently needed could maintain such rich eccentricities; they are used so frequently that they resist regimentation.

Consider, for examplegim-ik 'all, entire, whole', a very common word that can be used in several ways. Unusually among the defective nouns, gim-ik can be used with classifiers, and then it fits into the position where a number usually belongs. This is semantically understandable, of course, since 'all' is a quantity, just as numbers are. Ma'-su mang-gim-ik 'all the cattle', cher kan-gim-ik-ko ra'-ba-bo 'bring all the chairs' where kan- is the numeral classifier that is often used for constructed objects such as chairs. Gim-ik admits some possibility for ambiguity. Nok gim-ik could mean either 'the whole house' or 'all the houses', though the context rarely leaves much doubt. People insist to me that ma'-su gim-ik can only mean 'all the cattle'. To say 'the whole cow' would require ma'-su mang-gim-ik , with mang- the classifier for 'animals'.

Apart from its use with classifiers, gim-ik can be used as a noun and as a modifier of nouns. Used alone in a noun phrase, it can take case markers: gim-ik-na ron'-bo 'give to all, give to everyone'. Often it is usedPage  269to modify a noun: man-de gim-ik or gim-ik man-de 'all the people'; gim-ik mi-ko cha'-bo or mi gim-ik-ko cha'-bo 'eat all the rice'. In this last example gim-ik is the last member of the noun phrase, so it takes the case marker just as a nominalized verb would. In man-de-rang gim-ik 'all the men', the plural marker -rang remains as a suffix on man-de , again just as it would if used with an ordinary modifying verb. Speakers deny that *gim-ik-rang would be possible even if it were not modifying a noun, but this may be because of the semantic anomaly of pluralizing 'all'. In man-de-de, gim-ik-in bi-ka so-no-a 'all the people will be angry' the suffix -de 'but' which is generally suffixed to the final word of the noun phrase stays with the noun. This separates the gim-ik from the noun somewhat, as if the meaning is something like 'the people—-all will be angry'.

In most respects then, gim-ik acts like a noun, or at least a nominal. It is generally a constituent of a noun phrase or it acts by itself as a noun phrase. Most nouns cannot be used productively to modify another noun, in the way gim-ik does, however. That is usually the prerogative of nominalized verbs, but gi-mik is certainly not a verb since it does not take verb suffixes and does not need to be nominalized in order to modify a noun. Nor do speakers allow gip-a to be suffixed to gim-ik in the way it can be suffixed to verbs, though this may, again, be due to the semantic anomaly of attaching something that means 'the one that is' to a word meaning 'all':

Mit-am 'some' is as common and essential as gim-ik . Unlike gim-ik , mit-am cannot be used with a classifier, but it resembles gim-ik in being more of a noun than a verb. It does not take verb suffixes, but can be used either as a noun phrase itself or as a modifier of another noun. Unlike gim-ik , mit-am goes more easily before the noun that it modifies than after. Speakers even say that mit-am som-oi-o 'sometimes' is fine, but * so-moi mit-am-o is impossible. With some reluctance, some speakers have accepted man-de mit-am 'some people'. Nevertheless, when used alone, rather thanPage  270as a modifier, mit-am can take case markers: mit-am-ko-de nik-na man'-ja-ing-a 'cannot see some (of them)'. Unlike gim-ik , mit-am can take the -gip-a suffix, though only when used alone, not when preceding a noun that it modifies: mit-am-gip-a-de ga'-ak-jok 'some of them have fallen'.

A third example isrik-kit 'each'. Like gim-ik , rik-kit can be used with a classifier, as if it were a number. sak-rik-kit 'each person'. Unlike either gim-ik or mit-am , rik-kit always follows the noun it modifies and any case marker that the noun phrase requires is suffixed to it: man-de rik-kit-na a-gan-bo 'speak to each person'.

The eccentricities of words like these make it difficult to make coherent generalizations. The difficulties are compounded by the near impossibility, for a linguist who is not a native speaker, to sort out what is disallowed by syntactic rule from what is disallowed by semantic anomaly. I have worked with many highly intelligent native speakers, but I have never been able to train one of them to make a consistent distinction between syntactic and semantic anomalies. Repeatedly I have been told that some combination of words is impossible, whereupon I have devised some more or less bizarre context in which the sentence might apply. My consultants have then said "oh yes, in that case it is fine". How many examples have been rejected, and stayed rejected, because my imagination fell short of devising a sufficiently ingenious context, I will never know. If I had ever successfully translated 'colorless green ideas sleep furiously' into Mandi, I have no doubt that it would have been soundly and unanimously dismissed as impossible. In the end, I have begun to doubt the possibility of drawing such a sharp line between syntactic and semantic anomaly as linguists generally hope for. I have even wondered if it makes sense to expect to look for a syntax that is autonomous.

Rather than pretending to find order where I see so much chaos, I will limit my exposition of defective nouns to listing a selection and giving some examples.

Dam-san 'same, together'. Dam-san chu-bo 'sleep together!' (Imperative). Dam-san nok-o dong-bo 'Stay in the same house!':

This last sentence is constructed from two equational sentences connected by in-di-ba 'or'. The parts cannot be mistaken for verbal sentences because neither dam-san or ding-tang has a tense marker or any other verb suffix:

Gip-in 'other, another'. Gip-in ma'-su or ma'-su gip-in 'a different cow':

Gip-in can be used with classifiers:

Ding-tang 'different' can be used as a noun and take case markers: Ding-tang-ko cha'-bo 'eat something else'. More characteristically, however, it is used to modify another noun: Da'-sal-de ding-tang chol-a-ko ra'-bo 'bring a different shirt today'. With the help of dak-e 'in the manner of' it can be used adverbially to mean 'differently, in a different way': ding-tang dak-e nik-no-a 'will see (it) differently'.

Al-ak-si 'selfish'. This has a very strange meaning to be conveyed by a defective noun. One would expect that any Mandi word meaning 'selfish' would be an intransitive verb, but al-ak-si cannot take verb suffixes. My attempts to use it as a verb resulted in speakers adding a dummy verb to hold the suffixes: sak-sa al-ak-si ong'-a 'one person is selfish'. It readily modifies nouns: man-de al-ak-si, al-ak-si man-de 'selfish people'.

Pak-sa 'one half, one side, the other side'. This seems to have the form of a numeral (it includes sa 'one') but it is sufficiently lexicalized that it can itself be used as a classifier: pak-sa-gin-ing 'two halves, two sides'. Unlike most classifiers, however, it is enough of a nominal to take a locative case marker: Sal pak-sa-o re'-ba-bo 'come at noon', i.e. 'come when the day/sun is at the halfway point'; U-a bait-ni pak-sa-o nok dong-a 'there is a house on the other side of that low ground'.

Pak-ma 'most, majority, largest part'. This is related to pak-sa 'half', and it is unusual among these defective nouns in having a clear etymology. Pak means 'part', and -ma is used as the second element of many compounds in the sense of 'big', so it means the 'big part'. Pak-ma man-de-in chu-ing-a or man-de pak-ma-in chu-ing-a 'most of the people are sleeping'. Pak-ma cannot take a plural suffix but that limitation can be overcome by reduplication.

Ok-ki-sa 'a little, a small amount'. Ok-ki-sa mi dong-a, mi ok-ki-sa dong-a 'there is a little rice'.

Kak-ket 'true, correct'. Kak-ket gol-po, gol-po kak-ket 'true story'. One would expect the idea of 'correct' to be conveyed by an intransitive verb in Mandi, but kak-ket cannot take the verb suffixes, and in the following example it needs the empty verb ong'-a to hold the verb suffixes. Nang'-ni a-gan-a-in kak-ket ong'-a 'Your talk is correct'. It can act as a noun and take noun suffixes: Kak-ket-rang-ko kin-a-a 'hear the correct things'.

Dak-ri 'most, majority'. Mi dak-ri-ko cha'-bo 'eat most of the rice'. Dak-ri mi rat-na man'-jok 'have finished cutting most of the rice', Dak-ri neng-tak-jok 'done most of the resting', Ang-ni wak dak-ri dal'-jok 'my pig has mostly grown'.

Many, though not all, defective nouns name a quantity, but not all words for quantity are defective nouns. Bang'-a 'many, much, a lot', for example, names a quantity but rather than being a defective noun, it is a perfectly good intransitive verb. It takes all verb suffixes: bang'-no-a-ming-kon 'would probably have been a lot'. Like any other verb, bang'-a can be nominalized and then used to modify a noun: Bol-pang bang'-a-ko den'-jok '(he) cut many trees'. The -a- in the middle of bang'-a-ko is the nominalizing suffix which is needed to put this intransitive verb into a form that can modify a noun. The need to be nominalized sets bang'-a apart from the defective nouns discussed in this section which have no need for a nominalizing suffix. They are modifiers to start with.

Gi-type Adjectives

(B)

Mandi has a set of seven intransitive verbs with distinctive properties that make them more like English adjectives than are the other intransitive verbs of Mandi. All have meanings that are prototypically adjectival, and their first syllables are all gi-, gip- or git- . Comparison with Boro, a closely related language, suggests that these verbs once had the uniform first syllable gɨ- , a form that is still used in many adjective-like words in Boro. Over the course of time, in Mandi, this prefix has assimilated to the following syllable, so that it now appears in several somewhat different forms.

The seven words are: gip-bok 'white'; gi-sim 'black'; git-chak 'red'; git-ting 'raw, unripe'; git-tang 'living, fresh'; git-dal 'new'; git-cham 'old'. These can all be used as intransitive verbs and can all take the full range of principal verb suffixes. As their forms show, however, they have the unusual ability to be used without a principal verb suffix. A principal verb suffix is possible on all of them, but, unlike other intransitive verbs, they can be used to modify a noun without a nominalizing suffix: nok gi-sim 'black house', man-de git-dal 'new man' (often implying 'new son-in-law'), bi-te git-ting 'unripe fruit', etc.

With four of these seven words, it is possible to use the second syllable without the first. The remaining syllable then acts like an ordinary intransitive verb that must always be used with a principal verb suffix: sim-a 'black', bok-a , 'white', tang-a 'live, be alive, fresh', and cham'-a 'be worn out, old'. In their single syllable form, these take the full range of verb suffixes. Kin-i bok-jok 'the hair has turned white', pul-ko-bi tang-ing-a-ma? 'is the cauliflower fresh?'. Both the one-syllable forms and the two-syllable forms can be used as intransitive verbs. Bok-a and gip-bok-a , for example, can both mean 'to be white': ba'-ra gip-bok-jok-ma?, ba'-ra bok-jok-ma? 'did the cloth get white?'. Both can also be used as modifiers of nouns, but only the one syllable form requires the nominalizing suffix -a . In other words, bok-a, gip-ok-a, and gip-bok are allowed as modifiers, but *bok is not. Most intransitive verbs risk an ambiguity when the verb follows a noun and has the principal verb suffix -a . The verb could be nominalized to let it modify the noun, or it could be the main verb of the sentence: Chi ding'-a can mean either 'the hot water' or 'the water is hot'. Because the gi- words have two forms, this ambiguity can be partially avoided. Le-ka gi-sim can mean only 'the black paper'. Le-ka gi-sim-a remains ambiguous, however. It can mean either 'the paper is black' or 'the black paper'. As the following example shows, the one-syllable forms that lack the prefix can be used as modifiers as well as verbs: Sim-a do'-ko mu-a ming-a? 'what is the black bird called?' More often, however, it is the prefixed forms of the words that are used as modifiers.

Indefinite -ba words

(C)

Mandi and other dialects of Garo have a number of words that are suffixed with -ba and used in an indefinite sense in both negative sentences and in sentences with real or implied interrogatives. Their meanings overlap with the meanings of English words such as 'anyone', 'anything', 'someone' and 'something'. In negative sentences they are used where English uses 'nothing, nobody' and so on. Most of these words are constructed by adding -ba either to a question word, or to a numeral formed with a classifier and -sa 'one'. Perhaps the most common -ba word is ma-ming-ba 'anything, nothing', but it is an atypical example in one respect, because ma-ming is rarely used without the suffixed -ba . Most -ba words are formed from a word that is also used easily without the -ba . Ma-ming-ba dong-ja 'it is nothing, it doesn't matter, never mind' is a common fixed expression, but it is a typical example in which a -ba word is used with a negative verb. The two suffixes, -ba and -ja , reinforce one another. The next example is similar, but it has the accusative case marker -ko . Indefinite -ba always follows the case marker:

Several other indefinite -ba words, in addition to ma-ming-ba , are common:

Ba-no-ba 'somewhere, anywhere' is derived from ba-no 'where?'

Sal-sa-ba 'any time, never' is derived from sal-sa 'one day'.

Rang'-sa-ba 'sometime, at any time, never' is derived from rang'-sa 'one time, once':

Sak-sa-ba 'someone, anyone, no one' is derived from sak-sa 'one person':

In addition to the well lexicalized words that have been listed, any numeral that is formed from a classifier with -sa 'one' can by made into an indefinite word by having -ba suffixed to it. Such words are often used with negative verbs where they mean 'none, no': king-sa-ba 'some, of flat things', 'none, of flat things' when used with a negative verb ( king- 'classifier for flat things'); rong-sa-ba 'some, none, of round things' etc.

Kan-sa-ba dong-ja. 'There is none'.

A-chak mang-sa-ba chik-ja. 'No dog bites'.

Two other -ba words that are rarely used in Modhupur are used in A'chik and sometimes in the dialects of Bangladesh that border the Garo Hills: Pang-na-ba is constructed from pang-nan 'always', and when used with a negative verb means 'never'. Da-rang-ba is like ma-ming-ba in being rarely used without the -ba . When used with a negative verb it means 'nobody'. In Modhupur sal-sa-ba and sak-sa-ba generally take the place of pang-na-ba and da-rang-ba .

When another word follows one of these -ba words within the same noun phrase, the -ba is sometimes moved to the later word, in the samePage  276way that case markers are moved to the end of their noun phrase. In the next example, the -ba has been moved to dal'-a 'big' from kan-sa where it is more often found:

-ba is not as regularly moved to the end of the noun phrase as are case markers, however. In the next example it has been moved as far as the noun, but not all the way to the modifier dal'-a 'big':

A wider variety of -ba words are used in positive sentences than in negative ones, and here they imply some degree of uncertainty. They may be used in embedded questions or they may, themselves, imply a question. With sufficient ingenuity it is possible to find a way to use a -ba with just about any question word in combination with any case marker. The meaning of the -ba words in positive sentences is, again, indefinite, as can be seen by comparing sentences that have the same question words, once with, and once without, the -ba , as in the following examples with sa 'who?'. Appropriate case markers can be used with both sa and sa-ba :

Sa-ko san-di-ing-a? 'Who are (you) looking for?'

Sa-ko-ba san-di-ing-a? 'Are (you) looking for someone?'

Sa-ko-ba san-di-ja-ing-a. '(I) am not looking for anyone'.

Sa-ni ma'-su? 'Whose cow?'

Sa-ni-ba ma'-su, ang-a-de hai'-ja? 'Whose cow?, I do not know'.

Sa-ming re'-ang-no-a? 'Who will (he) go with?'

Sa-ming-ba re'-ang-no-a? 'Will he go with anyone? I wonder whom he will go with'.

Courtesy Expressions

(B)

Like everyone else, Mandis have polite and impolite ways of speaking and they have expressions that convey courtesy, but the literal meanings of the expressions they use most are very different from those used in English or other western languages. Under the impact of English, and perhaps especially under the influence of western missionaries, a number of equivalentsPage  277for English politeness phrases have been developed. These are relatively recent additions to the language and they are quite artificial. People know them, but rarely use them except with a foreigner, or when trying to emulate foreign manners. They remain marginal.

Ka'-sa-pa'-e 'please' is a rather artificial translation of English 'please'. It is known but rarely used. Ke-ma ka'-bo or ke-ma ka'-pa-bo is supposed to be equivalent to 'excuse me, please forgive me' but it is no more common than ka'-sa-pa'-e . Mit-tel-a 'thank you' is used marginally more often than ka'-sa-pa'-e or ke-ma ka'-bo , but Mandis are apt to use it in situations where, by western standards, they should not do the thanking, but be thanked. When it is used at all, it is a sort of general purpose expression of courtesy, most often used when two people are leaving each other.

Sal-nam means, literally, 'day-good' and it is occasionally used as an equivalent for 'good day' or 'good morning', but since everyone recognizes it is an over-literal, even quite silly, translation of the English expression, it is as likely to said as a joke as to be serious. In the same way, wal-nam is occasionally used where English speakers would say 'good night'. Mandis seem to know the conventional English meanings of 'good morning' and 'good night' better than they know the meanings of 'please', 'excuse me', and 'thank you', or at least when they do use wal-nam or sal-nam they manage more often to use it at the right time of day and on the same kinds of occasions when the corresponding English expressions would be used.

Nam-e dong-a-ma? 'are things good?, how are you?' is probably as recent and artificial a construction as the others, but it has become much better established. The A'chik equivalent is nam-eng-a-ma? which is a no more than a syntactic variant. People do occasionally ask this question and receive the answer nam-eng-a 'good', 'fine' or, sometimes, al-a-mal-a 'so-so'.

Garos, like many Indians and Bangladeshis, simply do not often use courtesy phrases like these in the situations where English speakers find them so essential. This does not mean, of course, that Garos lack ways of showing courtesy, or that they lack the sense of gratitude that makes Westerners feel obliged to say 'thank you', but they do express their feelings in different ways. First, they signal more with the posture of the body than do westerners. They lower their heads a bit to show respect or gratitude. They avoid making themselves look tall or dominant, which might be taken as immodesty, and of course, they have polite and impolite ways to use language. Garos have less fear of silence and less need to fill all available space with talk than do Westerners, but this does not mean that they do not talk. Indeed, they chatter as incessantly as anyone else, but if they happen to have nothing to say they find no embarrassment in sitting inPage  278silence. Still, when arriving and departing, and when meeting and leaving one another, there is usually an exchange of language.

When people meet on the road one often asks the other bat-cha i'-ang-ing-a? 'Where are (you) going?' When a Garo who is innocent of English wants to know how to say something in that exotic language, the first thing he or she asks is likely to be "How do you say bat-cha i'-ang-ing-a? ". If the one addressed is heading in the direction of his home (and people usually know where everyone else lives) he may be asked, instead, bat-cha-ni i'-ba-ing-a? 'Where are you coming from?' These questions are as conventional as is 'How are you?' among English speakers, and they are no greater an invasion of privacy. Nevertheless, they should be answered with some information — 'to my uncle's', 'to Joloi', or 'from my rice field' — but it is possible simply to say 'from over there.' Usually only one person asks the question. The other answers, but does not echo the original question back. On arriving at someone's house it is polite to call out Man-de dong-a-ma? 'Is there anybody (home)' and one might get the answer dong-a 'there is'. When leaving a group it is courteous to say i'-ang-na-jok 'I intend to leave' and sometimes people add na'-song a-gan-ku-bo 'You all go on talking'.

There are also a few grammatical markers of courtesy. The -ne that can be added to an imperative softens it and acts almost like 'please': a-song-bo-ne 'please sit down'. When added to an infinitive, the same -ne , used with appropriate intonation and gestures, is a courteous way to make a request: Ang-a nap-na-ne? 'May I come in?' Garo does offer ways to speak politely, just as it offers ways of speaking impolitely, but it has fewer fixed courtesy phrases than Europeans are accustomed to.

Interjections

(B)

Like people everywhere, Garos have words that are generally used alone, rather than as constituents of sentences. Some of these words fit the conventional patterns of pronunciation of the language (its "phonology") even if they do not fit its syntax, but like speakers of all languages, Garos also use a number of conventional vocalizations that do not even conform to the usual phonology, so they hardly even deserve to be called "words". Most of these non-phonological forms indicate something about the emotions or attitudes of the speaker, and they can be understood without support from accompanying words. All of these forms, both those that do and those that do not conform to the usual phonology, can be called "interjections", and even the non-phonological ones, deserve to be considered because they are meaningful vocal noises, and because they are used along with language even if they are not integrated into its syntax or phonology.

Since the interjections are less well integrated into the language than other words, it is relatively easy to borrow them from one language to another. Indeed, the least language-like of these exclamations can hardly said to belong to a particular language at all. In view of the high level of Mandi bilingualism, it not surprising that they share a number of their interjections with their Bengali neighbors. I give a few examples here of the interjections that Mandis use. Some of these are used widely by speakers of other Garo dialects, while others are more narrowly restricted to Mandi. Additional examples will be found in the lexicon in Volume II.

Ai-au 'oh my, dear me, wow' is a general expression of either surprise or dismay. The second syllable typically receives a heavy stress, and the intonation will generally show clearly whether the dominant emotion is surprise, wonder, or dismay. Ai-au is used by A'chik speakers as well as by Mandis, but it is not used by Bengalis.

Da'-nang 'oh dear' is a common expression of dismay. It may be used along with a sentence even if it is not really a part of it. Often it is placed after a verb, a position that is outside most of the organized syntax of the language. It is the position for afterthoughts that are moved from the more ordinary pre-verbal position. The intonation of da'-nang sets it apart as clearly as it sets apart any bit that is displaced from an earlier position. bi-a si-jok, da'-nang 'He died, alas'

At-cha 'okay, right, good' is used by speakers of many languages throughout both Bangladesh and much of India. It originally meant, literally, 'good', but as a borrowed word in Mandi it has largely lost that specific meaning. It is used in all its extended meanings, which are quite similar to the meanings of 'okay'.

Ki-ja-ni is a Bengali expression meaning, literally, 'who knows?' but it is often used by Mandis, even in otherwise Mandi contexts. It carries the general sense of 'I don't know, how can one know?'

Bap-a-ri-bap, ba-ba, o-ba-ba 'my gosh, wow!' are Bengali expressions of surprise that have been adopted by the Mandis.

Ha 'here, please take it'. This is used when giving something to someone.

[m] This consists of a sharply falling intonation, sometimes with a bilabial nasal m , or if the lips are open, with an ill-defined [ng] or [n]. It is an invitation for agreement, a sort of informal tag question, 'isn't that so? don't you agree?' This is widely used by both Mandi and Bengali speakers and by foreigners who have lived for a long period in Bangladesh, even when speaking English. It must have come originally from Bengali.

Iis, i-si with a falling intonation and sometimes with prolonged vowels, is an expression of disgust: 'yuck'.

Ai-a 'ow!' is an expression of sudden pain.

Finally, Mandi has a number of more or less conventional words that either fill in space when one has to think for just a moment before finding the right words, or that indicate that the speaker has made a mistake and is about make a correction. These come in the midst of a sentence, but they interrupt the syntax rather than forming a part of it.

a-kai has a reasonably consistent pronunciation that fits Mandi phonology, but it interrupts the syntax. It indicates that one has misspoken or made a mistake of some kind. It is frequently used when one has stumbled in speech or used an incorrect name by mistake. It can be used where English speakers might say 'I mean', but it has no meaning apart from its use in correcting errors. It is usually followed immediately by some sort of rephrasing or correction of what one has just said.

Ma'-bai, mai-ba, ma'-ki, ma-bi-ko, and na-bi-ko are all hesitation forms, said while collecting one's thoughts. They may be used in situations where an English speaker might say uh , but the availability of these conventional interjections does not stop Mandis from saying something very much like uh as well.

Answers to yes-no questions . Like interjections, answers to yes-no questions can be single words that are used outside of any syntactic context. Like interjections, also, they are sometimes phonologically anomalous.

All dialects of Garo, as far as I know, have straightforward words meaning 'yes' that are used when answering a yes-no question in the affirmative. Ha'-we is the fullest form in Mandi, but this is an area of many alternatives, and all of the following are possible: ha'-we, a-we, we, u-e, a'-we, ha'-wi, ho'-we, hu-e-ro, ho and even te . The accepted form for 'yes' in A'chik is we .

A'chik, as spoken in the Garo Hills, lacks a word that is equivalent to 'no', and the only linguistic way to give a negative answer is to repeat the verb of the question and to add a negative affix to it: Mandi dong-a-ma? 'is there anyone there?', dong-ja 'there is not'. Am-bin-o i'-ang-no-a-ma? 'Will (you) go tomorrow?' I'-ang-ja-no-a '(I) will not go'. Mandis have borrowed na , the Bengali word for 'no', and they use it freely to answer questions, often together with a single quick shake of the head.

Garos also have conventional ways of indicating 'yes' and 'no' with vocalizations that do not conform to the usual phonology of the language. 'No' can be indicated by a slightly rising intonation. This may be voiceless when it starts but it is then voiced, sometimes with the lips closed so that an [m] sound is produced but sometimes with the lips spread to produce some other, not very well defined, nasalized vowel or consonant. This isPage  281then abruptly terminated by a glottal stop. The [m] or other nasal is similar to [m] of the interjection that seeks confirmation (see above), but the rising intonation and the glottal stop make the meaning of 'no' unambiguous.

'Yes' can be indicated by a vocalization that can either rise slightly or fall slightly. Its distinguishing feature is a sharp interruption in voicing made by a glottal stop. Both the closure of the vocal cords and their subsequent release are distinctly audible, and there is voicing both before and after the stop. The lips often remain closed so that an [m] is produced, but an abrupt glottal closure interrupts it. Americans are likely to misinterpret this, at first, as meaning 'no' because it sounds rather like an American noise with that meaning, but the Garo 'yes' usually rises a bit in pitch and has more stress at the end, while the American 'no' is more likely to fall and to have more stress at the beginning.

A'chik has a word that means 'I don't know': hai-da . This word must be related to the Mandi word for 'know, understand', which is hai'-a , with hai'-ja as the negative, 'don't know'. A'chik has ui-a rather than hai'-a for 'know', and A'chik hai-da is not analyzable. Mandi manages without a separate lexical item for 'I don't know' and gets along fine with hai'-ja .

Conjunctions and Relatives

(B)

The Garo dialects have many ways by which words, phrases, clauses and sentences can be joined together into larger constructions, but the language makes much heavier use of suffixes for this purpose than does English. It has correspondingly less need for independent words such as conjunctions and relative pronouns. Many of the suffixes have been described in earlier chapters, especially in the section on "Subordinating Suffixes" in Chapter 6, "Verbs". The ways these can be used to build up complex constructions will be the main topic of the remaining chapters of this book. Nevertheless, the language does have a number of independent words that act as conjunctions, and it is these that form the topic of this section.

By English standards, older forms of Garo are impoverished in simple conjunctions. Most strikingly, there is no simple native Garo equivalent for and . Where two nouns would be connected by and in English, they are often simply placed side by side in Garo. Many such pairings are so conventional that they can be considered compounds: jik-se 'wife and husband', mi-bi-jak 'rice and curry'. The construction is productive, however, so that it is possible to join two nouns in a way that implies conjunction, without using an explicit marker of conjunction: hol-di chi-ni bre-jok '(I) bought turmeric (and) sugar'. To be more explicit, it is also possible to suffix two nouns with -ba , as in na'-a-ba ang-a-ba 'both you and I', chi-baPage  282chu-ba 'both water and rice beer'. Longer phrases can be joined in the same way, but if one of a pair is suffixed with -ba the other must have its -ba as well:

The case marker -ming 'with, together, along with' can also be suffixed to a pair of nouns or pronouns: ching-ming na'-song-ming 'both we and you all', mi-ming ja-ba-ming 'both rice and curry'. Nominalized verbs can also be joined with -ba 's: dal'-a-ba chon-a-ba 'both big ones and small ones'. Tensed verbs cannot take the suffix -ba , but they can be conjoined implicitly simply by being placed side by side. They always have the same series of suffixes, and these are enough to link them:

In addition to these ways of joining words and phrases, which are presumably old and native Garo, Mandi speakers have borrowed the word ar from Bengali and use it in a way that is close in both meaning and syntax to English 'and'. It is well assimilated into the language and is used freely to join two nouns, two verbs, or two longer phrases:

It is even possible to use ar along with suffixed -ba 's, although the ar is redundant and optional:

Some speakers in the Garo Hills, where Bengali influence is less strong, use a-ro instead of ar to mean 'and'. Occasionally even Modhupur Mandis use a-ro in this way, but in Modhupur a-ro is more often used to mean 'more' or 'moreover', which is closer to its Bengali sense:

Ar-ba is used only in negative sentences and, together with the negative, it means 'no more, not anymore': ar-ba dong-ja-jok 'there is no more', ar-ba re'-ba-ja-na-jok 'will not come any more'.

A-ro-ba is used only in positive sentences, where it means 'and also, in addition': a-ro-ba re'-ang-no-a 'and also (I) will go'.

Occasionally, Mandis even use the more elaborate and explicit Bengali e-bong which also means 'and' or 'in addition'.

The sense of 'but' is often conveyed with the final noun suffix -de : Ang-a cha'-no-a, bi-a-de cha'-ja-no-a 'I will eat but he will not'. In addition, Garo has closer native equivalents for 'but' than it does for 'and'. In-di-ba , or in Mandi, hin-di-ba or even hin'-di-ba with a raka (glottal stop), can connect two clauses with a meaning close to 'but':

The closely related in-di-o-ba (or hin-di-o-ba in Mandi) means 'but anyway' or 'in spite of':

Mandi speakers also occasionally use the Bengali kin-tu , which is equivalent to in-di-ba , and to-bu , which is equivalent to in-di-o-ba , even when the rest of their sentence is in Mandi.

Ba can be used between two nouns to mean 'or'. As a suffix -ba means 'also', but when two nouns are joined with the -ba that means 'also', both must have the suffix. When ba means 'or' only one is used. Intonational differences between the suffix and the conjunction also help to avoid ambiguity. When 'or' is intended, the syllable is more clearly separated from the preceding word by juncture and stress than when the meaning is 'also':

A more explicit way of saying 'or' is with ong'-ja-o-de , literally 'if (it) is not': Ro-nen ong'-ja-o-de bi-ni jik-gip-a 'Ronen or if (it) is not (him) his wife':

Clausal links . Garos often link two sentences together by means of a word that means little more than 'and then' or 'and so'. It is difficult to find clear criteria by which to decide whether this should count as a conjunction that joins the two shorter sentences into a single longer sentence, or instead, serves as an introduction to a new sentence. A considerable variety of forms are used for this linking, but they seem hardly to differ in meaning: u-ni-ko, i-ni-ko, u'-ni-ko, i'-ni-ko, u-ni-ku-no, i-ni-ku-no, u'-ni-ku-no, i'-ni-ku-no and sometimes even in-di-tan . As if such variability is not enough, a variety of Bengali forms have been taken into Mandi that are sometimesPage  285used in much the same way: tar-por, tar-po-re, ta-ba-de, tar-ba-de, tar-ba-de-se, ta-har, ta-ho-le, tai-le, ta-le, to-le, ta, te . Perhaps this is the kind of variation that is allowed when there is no clear written standard that defines or restricts what is regarded as a real part of the language:

Reduplication

(C)

The final two sections of this chapter describe processes that are both phonological and grammatical. Reduplication is a widespread process that effects many parts of speech and its uses will be summarized in this section. The final section with then consider echoes and anticipations. Echoes repeat some of the sounds of the preceding word but add more bulk than meaning to the sentence. Phonological anticipations reduplicate some or all of an initial syllable.

Reduplication is a pervasive process in the language and it has been dealt with in these pages along with the word classes where it is found. Here I will only summarize and compare the various types. ReduplicationPage  286is used to convey three different meanings, plurality, a distributive sense, and continuing or repetitive actions.

Plurality . Several kinds of words can be reduplicated to give the sense of plurality. Important among these are questions words:

The meaning can change depending upon how much appears to be reduplicated, and the first of the following two examples is not really a reduplication. The two sa-ming s refer to different people, as do the 'who' and 'whom' of the English translation. Sa-sa-ming of the second example is a genuine reduplication:

Indefinite -ba words and nominalized verbs can also be reduplicated to show plurality:

A number of the defective nouns that are used as modifiers can be reduplicated to indicate plurality: gip-in-gip-in 'others', from gip-in 'another'; ro-kom-ro-kom 'kinds' (see the section "Defective Nouns" earlier in this chapter).

Distributive sense . Numerals and numbers can be reduplicated to give a distributive sense: rong-bri-bri cha'-a-ming '(they) ate four each', pak-sa-pak-sa-ko chok-bo 'write on only one side of each', kan-sa-kan-sa-ko ron'-bo 'give one to each'. See section on "Numeral Suffixes and Reduplication" in Chapter 10, "Numerals".

Continuing or Repetitive sense . A number of verbs with meanings that describe continuous or repetitive action have a reduplicated or partially reduplicated form: sim-sim-a 'grow dark, as at sunset' from sim-a 'to be black, dark', tim'-tim-a 'rumble, of thunder, a motor, or a train', hil'-hil'-a 'rumble, of thunder or an engine', rip-rip-a 'rain a few drops', dim-dim-a 'burn vigorously'.

Many adverbial affixes also have a reduplicated form, often, though by no means always, carrying a sense of repeated or continuous action: -jem-jem- 'continuously', -pak-pak- 'a lot, heavily, as of rain', -bek-bek- 'freely, continuously', -rik-rik- 'repeatedly, piece by piece'.

A high proportion of adverbs show partial or full reduplication. Here, reduplication indicates not so much a specific meaning as the class to which the word belongs. Reduplicated adverbs have all sorts of meanings, but a good many do suggest a continuing or repetitive sense: pil-eng-pil-ung 'rocking back and forth, like a boat', gu-rung-ga-rang 'aimlessly, of roaming about', da'-an-da'-an 'often, again and again' from da'-an 'in a little while, soon'. When adverbs with an -e suffix are reduplicated, they usually gain a sense of continuous or repetitive action: tim-e-tim-e 'secretly, in hiding', implies that the hiding lasts for some time. Adverbial reduplication is described in the first section of this chapter.

One verb suffix is never used except with a reduplicated verb and its explicit meaning is protracted action. When -te is the principal verb suffix the whole verb is always reduplicated. The pair then means that the action continues over an extended period: i'-ang-te i'-ang-te 'going and going, continue to go'.

Other partial reduplications . Garo allows some other words, parts of words, and parts of phrases to be repeated in ways that have less specific meanings than those just described. While not strictly reduplications, these add to the repetitive and echoing nature of the language.

More nouns have partial reduplication than would be expected by sheer chance, but these words do not convey any consistent meaning: beng-beng-ka-ri 'a bee that makes honey', gu-ang-dom-dom-pa 'a kind of spider that lives in the forest', ha'-rong-rong 'sloping land'.

Echoes and anticipations are partial reduplications, but they add more rhetorical color than concrete meaning to a speaker's flow or words (see the next section.)

Finally, certain balanced constructions should be noted. These are not true reduplications. Nevertheless, the two parts of the construction echo each other in ways that are reminiscent of reduplications. Balanced questions are particularly important: Neng'-jok-ma, neng'-ku-ja? 'Are you tired (or) not yet tired?' (See Chapter 14, "Restructuring".)

Echoes and Anticipations

(C)

Mandis often increase the bulk of a word by adding syllables that either anticipate the sounds to come, or that echo the sounds that have just past. These anticipations and echoes add color to a speaker's style, and people explain clearly that they are used to make the language sound better. They make speech more pleasing to hear, or make it seem more important. They include some productive reduplications that are little more than phonological manipulations, but the language also has a large number of conventionalized reduplications.

Anticipations: -de . Almost any word can be expanded by reduplicating its first syllable and putting -de between the two identical syllables. There appear to be no restrictions on the parts of speech to which this expansion can apply. Anticipations with -de are often used in negative sentences, but they are not restricted to them. When used with a verb, the construction conveys a special assurance that the act was carried out as stated or, if the sentence is negative, that it was not carried out.Page  289With words belonging to other parts of speech, -de calls special attention to that word. This expansion is not used easily with one syllable words, however. It seems to need the phonological balance of two syllables: re'-de-re'-ang-ja 'no indeed, he did not go', re'-de-re'-ang-a-in 'he did indeed go', cha'-de-cha'-ja-jok 'has indeed eaten' (verb), a-de-a-chak 'dog' (noun), sak-de-sak-sa 'one person' (numeral), ok-de-ok-ki-sa 'a little' (defective noun), ang-de-ang-a-in 'I' (pronoun).

Occasionally more than a single syllable is reduplicated with the -de inserted after two or three syllables: me-de-me-seng-a, me-seng-de-me-seng-a, me-seng-a-de-me-seng-a 'sour'.

-ba : The syllable -ba can form an anticipation in the same way as -de . With a negative verb, -ba implies that some instruction was not carried out or that some expectation was left unmet: re'-ba-re'-ang-ja '(you) didn't go as was expected of you!', re'-ba-re'-ang-a-in '(you) did go', nam-ba-nam-a 'good', me-seng-ba-me-seng-a 'sour'.

-in, -an . Either -in or -an can form anticipations similar to those formed by -de and -ba , except that -in and -an are restricted to negative verbs. They are used neither with positive verbs nor with other parts of speech. As with -ba and -de these are most often added to the first syllable of a word, but they can occasionally be placed after two or three syllables: re'-in-re'-ba-ja-ing-a 'not going', re'-ang-tok-in-re'-ang-tok-ja 'not all go', a-in-a-song-ja 'not sit', gu-in-gu-al-ja or gu-al-in-gu-al-ja 'not forget'.

A final type of anticipatory reduplication occurs when the first consonant and first vowel of a word are repeated and the syllable completed with an -n . That syllable is then placed before the word, and it contributes a sense of thoroughness. This is usually done with negative verbs: kan-kat-ja 'not run at all', kin-kin-a-ja-ing-a 'not hear at all', gan-ga-du-ja 'don't climb at all, don-dong-ja 'none at all', an-a-song-ja 'don't sit at all'.

It is a curious fact that three of the syllables with which anticipations can be formed, de, ba and -in are homophonous with final noun suffixes. Even -an , which is an alternative to -in , is also an alternative of -in as a final noun suffix, but it is used more often by A'chik speakers than by Mandis. The only final noun suffix that has no equivalent here is -sa/-ha 'only'. I have tried, but failed, to persuade speakers to use -sa or -ha to form anticipations. I can see no similarity in meaning between the final noun suffixes and these anticipatory syllables, and I find the coincidence in form quite strange.

Echoes . The anticipations just described come at the beginning of words. Echoes come at the end. Echoes are more like full words than are anticipations. They almost always have the same number of syllables as the words they echo, and they almost always repeat some of the consoPage  290nants and vowels as well, but they are not identical to the earlier word. Some of the echoes are little more than phonological fill, but echoes are considerably more varied and complex than the more mechanical phonological anticipations. Some echoes are used as independent words with a similar, but generally not identical, meaning as the word that they echo. The echo construction is salient enough for native speakers to recognize it clearly, and to have given it a name: ku'-pa me-li-a 'words that go well together, harmonious words'. People enjoy the echoes and feel that they make the language more graceful and interesting.

The most productive and least imaginative echoes repeat a word except for a change in the first consonant. Often, the echo is given an initial t- . Any word at all can be echoed in this way, though some are more likely candidates than others, and nouns are probably the favorite choice: pu-kur tu-kur 'tank, constructed pond for water', pul tul 'flowers', me'-chik te'-chik 'women', me'-a-sa te'-a-sa 'men', gam-u tam-u 'work'.

Instead of changing the initial to t- , other modest changes can be made that are less productive and more idiosyncratic and conventionalized: al-si-a mal-si-a 'lazy', me'-chik mo'-chik 'women', ja-brang-e ju-brang-e 'worried, frightened'.

Other echoes deviate more from the base word. These are well conventionalized, but they add no meaning to the base word except for an indication that it is important in the sentence, for it is important words that attract echoes. Some well conventionalized echoes are phonologically quite different from the word they echo: a-je-a an'-cha-a 'to chant in a special way', rang'-kan-ta chu'-kan-ta 'angry', dong-ru-rak dong-su-sak 'restless, wiggly', kat-ru-rak kat-su-sak 'running back and forth', git-ak-a gi-jem-a 'cut into small pieces', wa-chi wa-da 'rainy season'. A very few echoes even have a different number of syllables from their antecedent: bret-a man-chal-a 'burst'.

A few words can be used as echoes with more than one word. Mai-ja , in particular, can be used to echo almost any negative verb, and it can violate the usual rule that the number of syllables is preserved in the echo. Since the verb base can vary, it is only the -ja that actually provides the echo: cha'-ja mai-ja 'does not eat', re'-ang-ja mai-ja 'does not go', a-gan-ja mai-ja 'does not talk'. A particularly common use of mai-ja is in dong-ja mai-ja 'does not exist'.

Echoes amount to partial reduplications, and any part of speech may have an echo. Most adverbs have some degree of reduplication, and some reduplicated adverbs differ little from echoes. Ding-dang is a common word meaning 'different'. It is possible to build this word into an echoing pair by adding -ga-dang and forming a new word, ding-dang-ga-dang . The addedPage  291 ga-dang hardly changes the meaning, but unlike most echoes, it does change the grammatical category of the word. Ding-dang-ga-dang is an adverb that can modify a verb, while the original ding-dang is a nominal that can modify a noun. The adverbial form is like that of many other adverbs, with identical second and fourth syllables and a meaningless -ga- as the third syllable, but it is formed in very much the same way as an echoing pair. In ding-dang-ko chok'-bo 'write (something) different' ding-dang is a nominal with an accusative suffix, and it is the object of the verb chok-bo 'write (imperative)'. In ding-dang-ga-dang chok-bo 'write differently' ding-dang-ga-dang is an adverb that modifies chok-bo 'write'. Echoing pairs, then, grade into reduplicated adverbs.

Echoing pairs also grade into a different kind of word pair. The wife-husband words (described in Chapter 8, "Nouns") are pairs of words that have related but different meanings. When used together, the pair generally has a more abstract meaning than either word does when used alone. Unlike ordinary echoes, wife-husband words need not have any phonological similarity. Nevertheless, some conventional pairs that imply an abstract sense, and that therefore count as wife-husband words, also have some echoing phonology so that they count as echo words as well. Some wife-husband words have no common phonology and some echoing pairs have meaningless second members, so a pair can be either echoing or a wife-husband word without being both. Some pairs are both, however, so it is impossible to draw a sharp line between echoes and wife-husband words.

The pair sut-kut 'fine clothes, dress up clothes', for example, nicely illustrates the overlap between echoing pairs and wife-husband words, at the same time that it illustrates borrowing. Sut comes from English 'suit' and kut from 'coat', but since the meaning of sut-kut is more generalized and abstract than the meanings of its parts, it counts as a fine wife-husband word. The words do not rhyme in English, but they do in Mandi, so they also form a satisfying echoing pair.

Borrowings of both echoing pairs and wife-husband words from Bengali are much more common than borrowings from English, because Bengali and Garo have very similar echoes and they can be easily borrowed whole: o-suk be-suk 'illness', ni-ti ri-ti 'law, custom', nas-ta nos-ta 'breakfast', mon-tro ton-tro 'ritual chant, mantra', bi-sun ti-sun 'seeds'. The echoing portions of these pairs are as meaningless in Bengali as in Mandi, but the first parts have been borrowed as ordinary words, and the echoes have come along with them. Quite possibly the entire process of echoing was borrowed from Bengali, but if so it has become a very well established process in Mandi. My examples all come from Mandi, but echoes are not limited to the dialects that have been most strongly impacted by Bengali. They are also found inPage  292A'chik where borrowing has been much less intense than in Mandi.

Echoes begin as phonological expansions and with little meaning of their own. Nevertheless, well established echoes are used repeatedly with the same antecedent so they have plenty of chance to be infected by the earlier word's meaning, and speakers will sometimes explain the meaning of what seems to be an echo as if it is an independent word. It seems quite possible that, now and then, an echo manages to escape the companionship of its antecedent and starts to be used as an independent word with the same or closely related meaning. It is difficult to be confidant in particular cases, however, because two independent words are sometimes coupled together in what could be mistaken for an echo, as in some kata jik-se (wife-husband words).