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

Introduction

(A)

pp. 213

Demonstratives, pronouns, question words, proper names, modifying verbs and postpositions share many but not all of the characteristics of nouns. All can take case markers, and all are found regularly as constituents of noun phrases. These similarities are enough to make it reasonable to call them all "nominals". Each type of nominal has its own special features, however, and each will be described in this chapter. Genitives and numerals are also nominals (because they can take case markers and appear as constituents of noun phrases) but genitives were described in Chapter 8, "Nouns", and numerals are important enough to rate a chapter of their own: Chapter 10. For learners, I have marked most of the sections of this chapter with "B" since they contain a good many details that early learners will probably not want to be burdened with, but each section also includes some information that should be useful quite soon. Learners may find it helpful to read the first parts of each section quite early, but leave remaining parts for later.

Demonstratives

(A)

pp. 213

The most important demonstratives are i-a 'this' and u-a 'that.' In a language without obligatory articles, one of these, most often u-a , is used when a definite meaning is essential: u-a man-de 'that man, the man'. As in this example, demonstrativess that occurs as modifiers within a larger noun phrase always have their -a form, but they lose their -a when anotherPage  214case marker is used. I-a and u-a are also used as the nearest equivalents to the English pronoun 'it':

Demonstratives are an area of the Mandi language where exuberant variation is at its most extreme. U-a and i-a are only the simplest of the possibilities. The words may, first, either have an initial h- or lack one. Second, they may have a glottal stop or lack one. This gives four possibilities: i-a, hi-a, i'-a, and hi'-a , with a parallel set for u-a . The locatives of these two demonstratives yield the ordinary words for 'here' and 'there'. Not only are these important and frequently used words, but they exhibit two additional dimensions of variation. An -a may be inserted as an extra syllable, and -no can be used as an alternative form of the locative suffix which, with most words, is uniformly -o . There are, then, sixteen possible ways to pronounce 'here': i-o, i'-o, i-no, i'-no, hi'-o, hi-no, hi'-no, hi-o, i-a-o, i'-a-o, i-a-no, i'-a-no, hi-a-o, hi'-a-o, hi-a-no, hi'-a-no . The longer the form, the more forceful and emphatic is the sense conveyed, but the basic meaning does not change. I have tried, on a number of occasions, to persuade speakers that one or more of these alternatives is impossible, but I have never met with any success. I have heard many of them in use, though I cannot swear to having heard every single one of the sixteen.

In addition, a third demonstrative, with several variants of its own, is used for things at a greater distance than u-a . Ai-a is the simplest pronunciation of this "remote" demonstrative. Ai-a is rarely used as a modifier within a larger noun phrase, and it is most often used with a locative, giving it the sense of 'way over there, at that distant place'. It is not really an equal third member of the demonstrative set because it is always emphatic and quite typically accompanied by a very distinctive intonation that emphasizes the great distance that is being described. I-a and u-a are opposites, like English 'this' and 'that. Ai-a lies outside their range. Ai-a , moreover, has a number of variants, including ai-wa and a-wa , even before the case marker is added. All these variants share the same rich variability of the less expressive u-a and i-a : optional initial -h , optional glottal stop, and -no as an optional locative form. I will not even try to list the possible ways to say 'way over there'. This variability seems extreme when language is written, but it is less surprising in the spoken language. In the absence of a codified orthography, colloquial speech is all we can observe, and the variation exceeds the resources of conventional phonology. We have reached the limits of rule-governed language and havePage  215entered realm of creative artistry, for it is possible to make up new ways to say 'way over there' almost without limit. Gesture and intonation can banish any risk of ambiguity from a great range segmental phonology.

Pronouns

(B)

pp. 215

Pronouns are the most noun-like of any of the classes of nominals, except for nouns themselves. Pronouns might even be considered to be a sub-class of nouns, but they do differ from other nouns in a few respects. For one thing, they rarely take modifiers of any sort. As genitives, they may, themselves, modify a noun, numeral or modifying verb, but they do not take modifiers of their own. Thus when they are not modifying something else, they almost always act alone as an argument of a verb. In this way they act like demonstrative pronouns, and they very frequently take case markers. The free forms of personal pronouns of Mandi, are shown in Table 9—1.

ang-a	'I'	ching-a	'we, exclusive'
		na'-ching	'we, inclusive'
na'-a	'you' sg.	'na'-song'	'you' pl.
bi-a	'he, she'	'bi-song'	'they' human

The -a that forms the final syllable of four of the pronouns can be considered to be a nominative case marker that is used only with pronouns that have monosyllabic bases. When any other case marker is used, it replaces the -a . Alternatively the -a form of the pronouns could be considered simply as the free and citation form. Bisyllabic pronouns, like nouns, lack any overt mark for the nominative. Na'-a 'you sg.' has an irregular combining form, nang'- . All case markers except -a are added to nang'- . The alternation between na'-a and nang'- counts as a major morphological irregularity in a language that is notably lacking in such irregularities. All case markers can be suffixed to pronouns just as they can be suffixed to nouns: ang-o 'with me, at me' (locative), nang'-ni 'your' (genitive singular), ching-ko 'us' (accusative), na'-song-na 'to you' (dative plural).

Garo pronouns do not have phonologically reduced forms. If a pronoun is pronounced at all, it is fully stressed. Where English might use a reduced form, Garo simply omits the pronoun altogether and relies on the contextPage  216to provide the sense. Because of the -a that is used for the nominative form of the monosyllabic pronouns, no personal pronoun or demonstrative is ever represented by an isolated monosyllable. Either a pronoun has two syllables of its own, or it is used with -a or with some other case marker. This gives pronouns more phonological substance than the many monosyllabic nouns of the language, which are easily cited in isolation or used in the nominative with no additional syllable.

A surprising fact, to anyone who is accustomed to languages with stable pronouns, is the dialectal variability of Garo pronouns. Several of the pronouns used in written Garo, and in the spoken dialect of many A'chik speakers, differ from those used in Mandi. (See Table 9—2.)

ang-a	'I'	ching-a	'we, exclusive'
		an'-ching	'we, inclusive'
na'-a	'you' sg.	'na'-sim-ang'	'you' pl.
u-a	'he, she'	'u-a-mang'	'they' human

The situation is actually more complex than implied by a simple two way division between Mandi and some sort of "standard" A'chik. An alternative second person plural, no'-ong is occasionally heard in Mandi. I have also heard both bi-si-mang and i-song as alternate forms for the third person plural, but I do not know in which dialects these are most often used. U-a-mang is a plural form of u-a 'that', 'it'. The use of u-a , and u-a-mang as personal pronouns in writing is probably based upon genuine oral usage in the dialect of the northeastern part of the Garo Hills, although in my experience, even educated and literate speakers more often use bi-a 'he, she' and either bi-si-mang or bi-song 'they' in their colloquial speech. Ang-a, na'-a, and ching-a are, so far as I know, used without variation by all Garo speakers.

In addition to the personal pronouns, Mandi has a number of other pronouns or pronoun-like words, three of which are characterized by the monosyllabic -a . One of these three is mo-a 'what?' (sometimes pronounced mu-a ), a question word that has -a only in the nominative. Like other pronouns the -a is lost when another case marker is used. The other two are the demonstratives, i-a and u-a , which, in addition to their use as modifiers in larger noun phrases, can also be used alone as demonstrative pronouns. Since their final -a is usually lost when another case marker is added, they act like the monosyllabic personal pronouns.

Page  217

Two other monosyllabic words that are similar to pronouns show no signs of nominative -a . One of these is the relative pronoun je 'whatever, whoever'. This has no -a in the nominative or free form and it is this form to which case markers are suffixed. The other is the question word sa 'who?', which also does without a nominative -a . Mandi occasionally loses a syllable when two a 's might be expected to be adjacent. This happens most notably when a neutral verb suffix -a might be expected to follow the negative -ja , but we find only -ja , as if the non-occurring * ja-a has been reduced to a single syllable. Conceivably we could explain away the absence of a "nominative -a " from sa as a result of a similar syllable loss.

A'chik, like Mandi, uses sa- as the combining form of the question word meaning 'who' and it is that form that takes case markers, but A'chik has an irregular form, sa-wa for the nominative. -wa could be considered the nominative case marker in A'chik.

The only other form in which -wa appears is ja-wa 'someone else'. This refers to someone different from those who have been mentioned before and it can almost always be translated as 'someone else'. It looks as though it is parallel to A'chik sa-wa 'who', but in Mandi ja-wa can take the full range of case markers without losing a syllable so it does not count as a monosyllabic pronoun. It is used only for people, never for animals or physical objects: ja-wa i'-ba-ing-a 'someone else is coming', ja-wa-ko nik-a 'saw someone else', ja-wa-na ron'-bo 'give it to someone else', ja-wa-o da'-dong-a-bo 'don't stay with anyone else'. The regular genitive is ja-wa-ni but it is also possible to say ja-ni , a faint Mandi echo of the more regular loss of -wa in A'chik: ja-wa-ni nok, ja-ni nok 'someone else's house'. Mandis do not drop the -wa with case markers other than the genitive. Ja-wa can made plural by suffixing -rang : ja-wa-rang-git-a 'like some others'.

Mandi also has several indefinite words that can be used in a pronoun-like way:

Da-rang 'everyone' is usually used as da-rang-ba and coupled with a negative verb to mean 'not anyone':

Ke-o 'someone, anyone' is borrowed from Bengali but widely used by Mandis. Like da-rang it is often used in negative sentences:

Page  218

A-mik-ka 'someone, an unknown or hypothetical person' is used more easily in positive sentences. This may be a borrowing of Bengali o-muk but it is thoroughly assimilated. A-mik-ka i'-ba-ing-a 'Someone is coming'.

Reflexive Pronouns: an'-tang, an'-tang-tang 'self'. The Mandi reflexive pronouns an'-tang and an'-tang-tang are nearly synonymous. Some speakers have insisted to me that an'-tang is always singular, while an'-tang-tang may be either singular or plural. Others have said that both can be either singular or plural, and my own experience corresponds to this second view. Ordinarily they refer back to an earlier noun or pronoun, often to the immediately preceding word, but when the meaning is clear they can occur without an explicit antecedent. Depending upon the context, the reflexives can refer equally to first, second or third persons, to 'myself, yourself, herself, ourselves' and so on.

The two examples that follow illustrate two somewhat different uses of reflexives. In the first, the personal pronoun is nominative while the reflexive is accusative, and the action of 'seeing' is directed back toward 'myself'. In the second sentence, the two pronouns are both nominative and the reflexive simply emphasizes the fact that it is, indeed, 'I myself' who does the seeing, and not someone else:

Ang-a an'-tang-tang'-ko nik-a. 'I see myself'

Ang-a an'-tang-tang nang'-ko nik-a. 'I, myself, see you'.

The suffix -in , which foregrounds the word and focuses attention, and whose meaning resembles that of a reflexive, is so often suffixed to an'-tang and an'-tang-tang that it is easy to be fooled into supposing that it is simply a part of the word. In fact it is always possible to use the reflexive pronoun without -in , but using it gives the word some extra force:

Page  219

Reflexives may take any of the case markers, though only the genitive -ni is very common, probably because the occasions when it is semantically appropriate to use a reflexive in any case except nominative or genitive are limited. The genitive in the first example makes it clear that the possessor of the book is the same person as is referred to by the subject of the sentence:.

The next two examples illustrate the possibility of using two reflexive pronouns together. In the first, one reflexive emphasizes the nominative subject while the other is the direct object. In the second example, which is an imperative, the nominative an'-tang-in is understood to refer to the addressee, while an'-tang-ni modifies poi-sa-ko 'money':

Question Words

(B)

pp. 219

Like other languages, Garo has a large number of question words, but all of them are constructed from one of three basic roots: sa 'who?', mo- 'what?' and ba- whose central meaning is probably 'which?' (of more than one thing). The more specific meanings that are needed when asking questions emerge when case markers are added to one of these roots. The use of case markers with question words means that they are nominals, and they are used in sentences in the same positions as noun phrases. Usually they form noun phrases of just one word, but occasionally they modify another noun and in this case they always occur first, in the same positionPage  220as a demonstrative: ba-gip-a song-o grong-no-a? 'in which village will (we) meet?' A genitive question word, like any other genitive, can occur with a possessed noun: sa-ni pi'-sa grap-ing-a? 'whose child is crying?' The question particle -ma that is put at the very end of a sentence to form a yes-no question is never used in a sentence that has a question word.

Unlike ordinary nouns, a considerable number of irregularities beset the question words. In particular, not all combinations of roots with case markers are possible, and even among those that are possible some are much more common than others. The result is that certain combinations of root plus case marker have become lexicalized, with meanings that often correspond fairly closely to one of the more specialized questions words of English. For example, ba-no 'where?' is formed from the root ba- 'which?', together with one form of the locative suffix. The locative suffix is more often -o , but with demonstratives and ba- , -no is an optional alternative. Ba-o is possible for 'where', but ba-no is very common. Thus, on the one hand, if one simply wants a word that means 'where', it seems straightforward to translate it as ba-no or ba-o . On the other hand, from a more analytical point of view, these words are part of a larger paradigm and perhaps their etymological meaning should be considered as 'at which place'.

Questions words are also made complex by showing more dialectal variability than is found in most areas of the vocabulary. Mandi and A'chik share the three basic roots, and most forms used in one dialect would probably be understood by speakers of the other, but some forms are much more characteristic of one dialect and some forms more characteristic of the other. There is also considerable individual variability and, even for a single dialect, it is almost impossible to draw sharp lines between what is common and widely acceptable, what is allowed and grammatical but rare, and what is simply impossible. I have constructed combinations of a root with a particular case marker which I felt I had never heard in use, and tried it out on native speakers. Sometimes they tell me it is impossible, sometimes they have to think about it for a while but then find a context where they feel that it can be or might be used, and sometimes they immediately accept it. Which should a linguist count as grammatical? Perhaps it is linguistic methodology rather than nature of language that pushes us to insist on a sharp line between what is grammatical and what is not.

I will describe each root in turn, and survey the way case markers and some other suffixes are used with each of them. My description will refer primarily to the Mandi dialect, and while I will point out a few of the major ways in which it differs from A'chik, I will not try to give a complete account of the similarities and differences between the dialects. I start withPage  221 sa 'who', which is the least irregular of the three,

Sa 'who?'. Sa is the only one of the three interrogative roots that Mandis regularly use without modification in the nominative. This makes it a candidate, along with je 'whoever, whatever' for a monosyllabic pronoun that does not have a suffixed -a for the nominative and free form.

Used alone, sa asks the question 'who?' 'who's that?' Of course sa can also be used in longer sentences. The word order of a sentence with sa is no different from the word order of a sentence with any other noun, but since the order of noun phrases is so free this does not much restrict its position. At any rate, there is no need for any movement rule with questions words:

Sa takes case markers easily and without irregularities, and this makes it, by a wide margin, the most regular of the three question word roots: Sa-ko nik-jok? 'Who did (you) see' (accusative); Sa-na ron'-no-a? 'Who will (I) give (it) to?' (dative); Bi-a sa-ni pi'-sa? 'Whose child is that?' (genitive); Sa-ming re'-ba-a? 'Who did (you) come with?', Sa-git-a dak-a? 'Who is (he) like?'

Question words can, of course, be used with postpositions:

Sa leaves the number unspecified. The answer could be either singular or plural. Sa-sa 'who all?' expects a plural answer, and often implies some sort of distributive sense. This is one example of the use of reduplication to convey a distributive meaning. Mi-ko sa-sa-na ron'-a? 'to whom all should rice be given?' implies that portions are being given to each, not that a single batch is being given to a collection of several people. I have been told that sa-rang is also a possible plural for sa though it feels marginal to me and it is must be used less than sa-sa . It would probably be used in a more collective sense than sa-sa , as Sa-rang re'-ang-no-a? 'Who all will go? Which group of people will go?'. People have denied the possibility of using the other plural suffixes with sa but I suspect that if they were pushed, at least a few of them would allow sa-dang and sa-drang . Sa-gip-a 'which (of several people)?' is possible but not common. Ba-gip-a is more usual in the sense of 'which', not only for objects, but also for people.

Mo'-, ma'-, mu'-, mai 'what?' This root has several pronunciations. Some pronunciations are more common with some case markers, othersPage  222with others, but there is much individual and dialectal variability as well. Mai is the A'chik root and that is occasionally heard from Mandis as well, but Mandis consider mai to belong primarily to another dialect than their own. Mo'-, ma'- and mu'- are all more common than mai- in Mandi, with mo'- probably in the lead with the most cases markers. Another variable is the raka (glottal stop) that is sometimes omitted, and it seems always to be absent from the nominative. The usual nominative and free form in Modhupur Mandi is mo-a or mu-a , an example of a monosyllabic pronoun with -a in the nominative. Occasionally one also hears moi . Like other question words, mo-a fits into the same position of the sentence as a corresponding non-interrogative noun phrase would: te-bil-o gles dong-a '(a) glass is on the table', te-bil-o mo-a dong-a? 'What is on the table?'

The dative of mo- may be pronounced mo'-na, mu'-na or ma'-na . Generally, but not always, it has a raka. Mo'-na means, quite literally 'for what?', and it is the usual way to ask 'why?' Mai-na is the A'chik version of 'why?' Na'-a mo'-na u-ko cha'-jok? 'Why did you eat that?'

The most common genitive form is ma'-ni , though mo'-ni occurs as well. It is used in such expressions as ma'-ni sal-o 'on what day?'. The word often means 'what kind of', as in mo'-ni chol-a 'what kind of shirt?' This question could elicit such answers as 'cotton', 'sleeveless' etc. In ma'-ni gim-in 'because of what?', 'why?', the genitive case marker is followed by a postposition that usually means 'because'.

The accusative occurs easily in questions such as mo'-ko dak-ing-a 'what are (you) doing?':

Like the locative of the demonstratives, the locative of mu-a is one of the places in Mandi where variation seems almost out of control. I have been given the following forms as possible locatives and I doubt if this exhausts the inventory: mo'-o, mo'-no, mu-a-o, mu-a-no, moi-o, moi-no, ma'-o, ma'-no . All of these would mean 'at what?, 'near what?', but in fact, none of them are very common. Locative questions are much more often formed with the root ba- (see below) and none of those with mu- or with other ma- roots are common enough for me to have a sense of their naturalness or relative frequency. Still, speakers insist that all are possible. Other locative case markers, such as -cha, -cha-na, o-na, etc. are even less commonly suffixed to ma'- or mo'- than are -o or -no .

Ma-ming 'what with?' seems more natural than mo-ming or other conceivable forms. Ma-ming is a part of a conventional courtesy expressionPage  223in which one person asks if another has had his rice (i.e. eaten a meal). If the answer is 'yes' it is likely to be followed by a second question: ma-ming? 'what with?', that is 'what vegetables or curry did you have with it?'

Like sa , mu-a and its inflected forms can be reduplicated to give a plural and, especially, a distributive meaning: mu-a mu-a dong-a? 'what all is there?', mo'-ko mo'-ko nik-jok? 'what all did (you) see?'

Mo'-rang, ma'-rang and mo'-drang are possible, or at least conceivable plurals, but people have denied the possibility of using the plural -dang with this root:

Ba-gip-a 'which' is very common, and sa-gip-a 'who?, which people?' is at least possible, but speakers have denied the possibility of suffixing -gip-a to any form of the mo'- interrogative. Whatever meanings might have been assigned to the nonexistent * mo'-gip-a have apparently been preempted by ba-gip-a .

Other suffixes that can be used with mo'- include -git-a 'like', -sik-ni 'how much': mo'-sik-ni ro'-a 'how long?, how much long?' mo'- with the appropriate case marker can also be followed by a postposition:

Yes-no questions are formed by placing a -ma at the end of a sentence. This is surely related to this question word root, but it occurs in an entirely different syntactic position. It is set decisively apart from the question words by its invariant and uninflected form. All by itself, Ma? can also be used to ask 'what did you say?'

Ba- 'what?' The third and final question word root is ba- or, as it is sometimes pronounced, bat- . Both semantically and morphologically it is the least regular of the three. It overlaps in meaning with mo- , and aPage  224few forms that are constructed from mo- and ba- are reasonably close synonyms. However mo- simply means 'what?', while ba- often means 'which of several?' While their meanings are by no means identical, the two roots seem, in some cases, to have achieved a sort of division of labor, with one case being used more often with one of the roots and another case more often with the other, although absolute prohibitions on the use of a case marker with a root are hard to find. For example locatives can be used with ma- or mo- but they are not at all common. Locatives with ba- , on the other hand, are very frequent.

Ba- has the odd property of never occurringin the nominative, so the question of whether or not it has the monosyllabic -a does not arise. For the nominative, either the more elaborate ba-gip-a or one of the variants of mo-a is needed.

The accusative ba-ko is possible, and the difference in meaning between ba- and mo- shows up with special clarity in the accusative. Ma-ko dak-ing-a means 'what are you doing?'. Ba-ko dak-ing-a? means 'which (of more than one thing) are you doing?' Ba-gip-a-ko often replaces ba-ko , however. Ba-ko seems abrupt, a bit too abbreviated.

Ba- can occur with the genitive case in such expressions as ba-ni sal-o 'on what day?, on which day?' and ba-ni song-ni man-de 'People from what village?, from which village?' It can also be reduplicated: Ba-ni-ba-ni song-ni man-de 'People from what all villages?' Ma'-ni , the genitive of ma'- means 'why?', and thus differs sharply in meaning from the genitive form of ba- .

Ba- is not impossible with accusative, dative, or instrumental case markers but none of these are common. Both the other interrogative roots, sa- and ma- , take these case markers more easily. When offered examples of ba- with one of these cases, speakers sometimes unreflectively substitute ba-gip-a , which can more easily have these cases than can ba- . Ba-git-a? 'In what way? how?' is natural: Ba-git-a re'-ang-no-a? 'How will (you) go?, By which route will (you) go?'

Ba- comes into its own with locatives, where mo- is restricted. The -o locative often has its -no alternative when used with ba- , and either ba-no or ba-o , literally 'at which place' is the usual way of asking 'where?' In many other circumstances, the locative -o has a temporal meaning, but ba-o and ba-no always mean 'where', never 'when?' as in bi-song ba-no dong-ing-a 'where are they?' Ba- can also take the augmented locative suffixes -o-na and -o-ni : Ba-no-ni ba-no-na mal-a-ming? 'from where to where did (he) crawl?' With the locative -cha , which is also spatial, but generally implies motion, the root assumes the form bat- , probably an assimilation to the following ch . Two of the most common expressions, used when peoplePage  225greet each other on the road, are Bat-cha i'-ang-ing-a? 'where are you going?' and bat-cha-ni i'-ba-ing-a? 'Where are you coming from?', 'Where have you been?'

In addition to being used with these case markers, ba- is used in a number of well lexicalized compounds that have more specialized interrogative meanings. Most of these compounds are, themselves, nouns, and some case markers can be more easily suffixed to these longer forms than to ba- alone.

bai-sik, ba-sik 'how many', 'how much'. Bai-sik dal'-a 'how big', 'how much big'; bai-sik ba-ji-jok 'what time is it?', 'how much (time) has struck?'; bai-sik a-rim-a 'how late?' In Mandi, bai-sik is often used in a construction with a numeral. The classifier is chosen according to the thing counted, and the number is always -sa 'one': bai-sik ak-sa 'how many people?'; bai-sik mang-sa 'how many animals'; a-chak bai-sik mang-sa 'how many dogs?'; bai-sik sal-sa? 'how many days?' As far as I know, this expression is not used in A'chik. Used with verbs for dimensions, bai-sik-ni? asks 'how much along that dimension?': bai-sik-ni ro-a? 'how long?'; ba-sik-ni chrim-a? 'how heavy; ba-sik-ni ke-ji? 'how many kilos?'. Bai-sik-na 'for how much' asks for a price: bai-sik-na bre-a 'how much did (you) buy (it) for?' Ba-sik-git-a? can also ask a price:

Ba-di-sik is an expansion of ba-sik without much change in meaning by those who use it, but not everyone uses it easily. Ba-di-sik bre-a-ming 'for how much did (you) buy (it)?' Ba-di-sik-na asks for the price; ba-di-sik-ni asks for the quantity.

ba-sok-o, ba-sik-o, bai-sik-o 'when?' This word looks as if it might be a locative of ba-sik 'how much', but whatever its etymology, this word is now thoroughly lexicalized in the sense of 'when?'. The A'chik form is ba-sak-o , a pronunciation rarely heard from Mandis. I have had Mandis deny the existence of ba-sik-o but I have also heard it used. Perhaps it should be regarded as a fast speech form. Occasionally ba-sik-o-na 'until when' and ba-sik-o-ni 'since when' are used as well:

ba-gip-a 'which'. The root ba- has 'which' as its central meaning, but the root must be a bit short to carry the meaning all alone, for it is not used in the nominative. Even with other case markers, ba-gip-a- , which can bePage  226translated as 'which one?', is generally preferred over simple ba- when the meaning is 'which'. This is especially true when asking about people, where -gip-a is particularly appropriate, but it can be used for non-human and inanimate objects as well. The word means, specifically 'which of several?' Ba-gip-a is a noun and it can take all case markers: ba-gip-a-ko cha'-ing-a? 'which ones are (you) eating?', ba-gip-a-ni chol-a-rang? 'whose shirts?' Both ba-gip-a-ni pi'-sa and sa-ni pi'-sa mean 'whose child (is this)?' Ba-ni pi'-sa would be more likely to be understood as 'a child of which village, of which place?': Ba-gip-a-ko ra'-no-a 'which shall (I) take?'; ba-gip-a song-o 'in which village?'

The locative form ba-gip-a-o 'where?' hardly differs in meaning from ba-no , but in this case it is the shorter form that is preferred. When modifying a noun, ba-gip-a usually means 'what sort of?':

Ba'-gip-a-rang-nipi'-sa? what kind-Plu-Gen child 'What kind of people's child?'

Ba-di 'how', 'in what way' is an interrogative adverb, rather than a nominal, and it is used to modify a verb. It does not take case markers and it often immediately precedes its verb: Ba-di chu-no-a 'how will (you) sleep (e.g. on which side of the bed)'; ba-di re'-ang-no-a 'how will (you) go (e.g. by which road)?' An expanded form, ba-dik-e , is more common than the simpler ba-di . An even further expansion yields ba-dik-e dak-e , which may be the most common of all. Ba-dik-e dak-e is a compound expression that acts like an adverb meaning 'how?', which is what ba-di means all by itself:

Page  227

Ba-dak-a? 'how?, what way?, what kind of?': ba-dak-aman-de? 'what kind of person?'. Ba-dak-a often means 'how does (it) look?'

Most question words can be reduplicated to give a plural meaning: ba-gip-a-ba-gip-a? 'which ones?', sa-sa? 'who all?', mo-ko-mo-ko? 'what all? (accusative)'. As the examples show, nominalizations and case markers are reduplicated along with the roots. sa-ko-sa-ko dok-a? means 'who all should be hit?' Sa sa-ko dok-a? is not a reduplicated form but has two separate questions words and would be understood to mean 'who hits whom?'

Proper Names

(B)

pp. 227

Among Garos, as among speakers of every language, proper names form a substantial part of each individual's vocabulary. Every Mandi, of course, knows hundreds, perhaps thousands, of place names, and an additional hundreds or thousands of personal names. To these we might add the names of gods and spirits. Place names are taken so freely from one language into another that they seem hardly to count as belonging to a particular language, although the conventional pronunciations in each language make them quite variable. Place names in the Garo Hills are typically Garo words, or at least they conform closely to Garo phonology, but most place names do not have obvious meanings except as labels for the places. In the Mandi areas of Bangladesh the majority of place names appear to be Bengali in origin. Pir-ga-cha , a village name, for example, has a syllable final r which would not be expected in a native Mandi word. A village may have a Pos-chim-pa-ra 'western section', where pos-chim is Bengali for 'west', and pa-ra is Bengali for 'village neighborhood'. In addition to local names for villages, neighborhoods within villages, rivers and, where they exist, hills and mountains, Garos now have considerable knowledge of geography beyond their own areas, and they know the names of distant cities, countries, continents, and seas. All are subject to some phonological modification as they are adapted to Garo speech habits, but they are generally understandable, even to someone without knowledge of Garo.

Place names can productively form compounds with names of particular features of the places: Gai-ra skul 'Gaira school', Dhaka so-hor 'Dhaka city'. Of course, they take locative and augmented locagive case markers easily: Gai-ra-o 'in Gaira', A-me-ri-ka-cha-ni 'from America' etc. Genitives are also common with place names: Gaira-ni ra-ma 'road to Gaira,Page  228road in Gaira'; Jo-loi-ni man-de 'a person from Joloi' etc. Other case markers are less common, but not impossible. Place names do not, however, normally form the head of complex noun phrases. They are rarely used with demonstratives, numerals, or even modifying verbs. The conventions for personal names differ from the practices of other societies more than do those for place names. Each Mandi has two family names and at least one personal name. In accordance with the matrilineal kinship system of the Garos, their family names are taken automatically from the mother. One of the family names is almost always Marak, Sangma, Momin, Areng or Shira . These are, or were, exogamous groups, and until recently no one was supposed to marry within his or her own group. The smaller family groups are known as "ma'chongs" and more than one hundred ma'chong names are probably in use. For modern purposes, Garos living in the Garo hills usually use one of the five exogamous groups names as their "last" or "family" name, and they use their ma'chong name as a "middle" name, often abbreviated to its initial. In Bangladesh, the ma'chong name is usually used as the "last name" though people are well aware of their membership in the larger groups and they continue to avoid marriage with someone from their own group.

Mandis share their family names with large numbers of people, but they dislike sharing a personal name with someone else, and they go to considerable lengths to find unique names for their children. Older Mandi names often had two syllables and their final syllables quite often gave a hint about the person's gender. Even now, female names tend to end with a vowel and male names with a consonant, although there are exceptions. Except for this gender distinction, personal names were, and often still are, simply made up, invented sequences of sound, conforming, of course, to the patterns of Garo phonology and selected to have a pleasant sound. People deliberately avoid using the name of anyone with whom they are acquainted. They cannot know the names of all the hundreds of thousands of Garos, however, so it happens occasionally that two people with the same name do meet, a situation regarded as distinctly unpleasant (Hvenekilde, Marak, and Burling 2000).

As an increasing number of Garos have gained a modern education and come in contact with members of other ethnic groups, the urge to find a unique name must have encouraged them to look to other languages and other groups for names that would be safely unique. It seems not to have bothered people to bestow a name of some foreigner, though the choice seems to have depended almost entirely on the sounds of the name rather than on admiration for the earlier holder of the name. Garos have been named 'Hitler' as well as 'Roosevelt', 'Truman', 'Aristotle', 'Milton'Page  229and 'Knudsen Berg'. Hitler was probably not named in admiration of the original, but the others were probably not either. Some people have Western sounding, though in fact made up, names. The ending 'son' became popular for men's personal names, and 'Wilson, Jackson, Milikson, Nekson' and others have all been used. Many people in Bangladesh have Bengali names, or names with endings that make them sound Bengali: Mo-hen-dro, Bi-jon-sing .

Whatever the name, it acts linguistically much like any other noun. It can take the full complement of case markers, and in can be used as an argument of a verb. Personal names are less likely to be used with the other constituents of noun phrases, but it is not impossible to find some reason to use demonstratives, numerals, genitives, or other modifiers with a proper name. In the right circumstances one can distinguish two people unfortunate enough to have the same name as dal'-a Rong-sen 'big Rong-sen' and chon-a Rong-sen 'little Rong-sen', but such modification is not common, especially since Garos try so hard to avoid using the same name for more than one person. Personal names are regularly used along with ma'chong and exogamous group names, just as given and family names are used in the West.

Modifying Verbs

(B)

pp. 229

Modifying verbs are described in Chapter 6, "Verbs". There it is explained that verbs that translate English adjectives and those that translate English verbs, can both be "nominalized" in a way that allows them either to modify a noun or to act as nouns themselves. Along with demonstratives, pronouns, and numerals, nominalized verbs count among the "nominals" of the language because they can take case markers and because they can act as a noun phrase or be used within a noun phrase. Nominalized verbs appear in sentences both as independent arguments of verbs, and as modifiers of other nouns. As modifiers, they may be single words that English speakers find natural to think of as "adjectives", but a verb can pull several of its arguments along with it when it is nominalized, and it is then more like a relative clause than an adjective. Whether or not it brings it arguments along, a verb needs a nominalizing suffix to convert it into a form that can act as a modifier or stand alone as a noun. A nominalized verb can take any of the noun suffixes that are appropriate to its position in the clause. See the section "Nominalizing Suffixes" in Chapter 6, "Verbs", and the section "Nominalized Verbs as Clause Constituents" in Chapter 12, "Complex Noun Phrases" for further discussion.Page  230

Postpositions

(B)

pp. 230

From a comparative point of view, Mandi postpositions do the same job and cover approximately the same range of meanings as do the prepositions and postpositions of other languages. They carry such meanings as position in space, direction of movement, location in time, and causation. From the point of view of their place in the Garo language, postpositions refine and extend the meanings of the case markers. The dozen or so case markers indicate the general role that each noun phrase plays in its sentence or clause. When more precision is needed, it is time to call on a postposition.

Some postpositions are transparently derived from nouns. Indeed the line between a postposition and a noun is not entirely sharp. Consider the following Mandi sentence:

As a noun, ki'-sang 'buttocks' is an uncomplicated term for a part of the body. In its locative form, it is the ordinary Mandi way of saying 'behind, in back of', not only for people but for any object that can be considered to have a front and a rear. Thus ang-ni ki'-sang-o means, very literally 'at the rear of me' with the four meaningful parts given in the exact reverse order from English. The term, of course, exactly parallels English 'in back of' except that it exploits a word for a different part of the anatomy. In A'chik, the usual term for 'in back of' is jang-gil-o , derived from another body part word, jang-gil 'back'. The construction by which ki'-sang-o is formed is a fully productive one in Mandi. For example ang-ni jep-o means 'in my pocket'. Only two characteristics of ki'-sang-o make it, but not jep-o , a postposition: the frequency of its use, and its conventionalized meaning. As a postposition ki'-sang is sufficiently distinct from its meaning as a body part to refer to the 'back' of things which do not have body parts at all.

Only a few postpositions show their derivation from nouns as clearly as ki'-sang-o and jang-gil-o , but many others are constructed with the same possessive construction. U'-ni gim-in , for example, means 'because of that'. Once again the order of the meaningful parts exactly reverses that of English, and except for the absence of the locative case marker, its structure parallels that of ang-ni ki'-sang-o . In this case, however, the postposition gim-in 'because' has no existence except as a postposition. Indeed it so regularly follows a -ni in the previous word that it is almost tempting to consider the gim-in to be an additional noun suffix. Gim-in always follows a noun or nominalized verb in the genitive case. The more transparent derivation of postpositions such as ki'-sang-o , however,Page  231suggests that postpositions, in general, should be considered as nouns that have become specialized in this use. As postpositions, however, they take only a restricted set of case markers.

Postpositions are often used with simple nouns (i.e. nouns that are not formed by nominalizing verbs) and the construction is then quite straightforward. Usually the noun has a case marker and the postposition follows directly after the case marker. Just as ki'-sang-o always follows a genitive, most postpositions are associated with a single case, but there are complications, and a few postpositions are able to follow more than just one case marker. Sometimes the meaning varies with the case.

Nominalized verbs can take postpositions just as simple nouns can. The verb base needs a nominalizing suffix before its case marker, so the sequence of suffixes is a bit more complex than with simple nouns, but in most cases the case marker is the same whether it follows a simple noun or a nominalized verb. This means that postpositions are closely associated with the case marker that characteristically precedes them. Case marker and postposition form a closely unified pair. The most common nominalizing suffix used before a postposition is -a , and the most common case marker is the genitive -ni . This means that a large number of postpositions follow -a-ni , which in turn follows a verb base: re'-ang-a-ni gim-in 'because of going', cha'-a-ni ja'-man-o 'after eating'.

Like ki'-sang-o 'in back of', ja'-man-o 'after' ends with an -o , presumably the locative case marker. This suggests that -ja'-man should be regarded as a nominal, and indeed it should, but, unlike ki'-sang and jang-gil , discussed above, ja'-man has no separate existence as an ordinary noun. Ja'-man is occasionally used as a postposition even without the locative case marker, though that is relatively rare. It can also be used as an adverb meaning 'later', 'afterwards'. Still, it must have the locative -o in a considerable majority of its occurrences. This groups ja'-man-o with the words that are so characteristically locative that one can imagine the locative suffix eventually fusing with the noun into an indivisible word. Because it remains possible to use ja'-man without a case marker, the fusion, if that is what it is, has only begun. Nevertheless, the class of nouns that carry semipermanent locative o 's is a salient one in the language. (See Chapter 11, "Minor Parts of Speech".)

Ja'-man is occasionally used with some other case marker than -o , most often with one of the augmented locatives:

Page  232

These case markers are not common with ja'-man , and only ja'-man-o , and to a lesser extent ja'-man , are really frequent.

Gim-in 'because of', like ja'-man-o , is regularly used with verbs that have been nominalized with -a which is followed, in turn, by a genitive: re'-ang-a-ni gim-in 'because of going'. An entire clause may be nominalized rather than just a single verb, so a verb may pull one or more noun phrase arguments with it. Once it is nominalized, the original subject of the verb is put into the genitive case. Ang-a re'-ang-a 'I go' can be nominalized to ang-ni re'-ang-a 'my going', and this in turn used with a genitive and a following postposition:

In the next few pages, a number of postpositions will be described, grouped according to the case marker with which they are characteristically used.

Postpositions that follow genitive -ni. By far the largest number of postpositions follow the genitive case marker. This is especially characteristic of spatial postpositions with such meanings as 'inside, beside, behind' and so on, but many other postpositions follow the genitive as well. The spatial postpositions are most often used with nouns, though most can, sometimes with a bit of effort, also be made to follow verbs that have been nominalized with -a . Two common postpositions that follow genitive -ni have already been described: Gim-in 'because' and ja'-man-o 'after'. Several others are used in similar ways.

Mik-kang-o 'in front of', is used much as is ki'-sang-o . As a noun, mik-kang means 'face'. Gi-sep-o means 'between', 'among':

Page  233

Among many other spatial postpositions are sak-a-o 'above, on top of', sam-ba-o 'beside', ning'-a-o 'inside, under':

-git-a is used as a case marker, but it can also be used as a postposition following genitive -ni , where it means 'so that', 'according to', 'for the sake of', 'in conformity with', and it can follow either a noun or a nominalized verb:

Mandis also use several synonyms for gim-in 'because', most of which also follow -ni . Just why so many are needed is a bit of a mystery: ang-ni gim-in, ang-ni pal, ang-ni a'-sel, ang-ni mik-ta 'because of me', re'-ang-a-niPage  234a'-sel 'because of going', cha'-a-ni pal 'because of eating'. Gan-da is used with the dative -na and, less often, with genitive -ni in the same meaning:

Postpositions that most often follow dative -na . While the largest number of postpositions follow nouns with a genitive case marker, several are most characteristically used after the dative -na .

Gan-da has two quite different uses, although some similarity can be seen in their meanings. As a postposition that follows a dative, or less commonly, the genitive, it means 'because, therefore, on account of, due to'. Speakers claim that it is synonymous with gim-in, -a-sel, pal , and mik-ta , all of which follow the genitive -ni . Gan-da then stands apart from the other postpositions with related meanings, since the others uniformly follow the genitive rather than the dative:

In addition, gan-da can follow a variety of case markers to convey a meaning something like 'as for, concerning, by contrast'. In this usage, gan-da calls attention to something, generally though not always, with an explicit contrast to something else. In the next example, the pronouns have the nominative case because they are the subjects of the verb:

In the next example the pronouns are in the dative because they mark the recipients:

Page  235

The next example has the pronoun in the accusative because it is the object of the verb:

Bat-e means 'more than' and it is the usual Garo way of making a comparison. It can follow a noun in the dative case, or a verb nominalized with -a and then made dative:

Skang 'before' and git-a 'so that, suitable for' follow nouns in the dative case. Oddly, skang and git-a can also follow a verb suffixed with -na but without any overt nominalizing suffix. Ordinarily postpositions are used with a verb only when it has been nominalized, usually with -a . With a verb that has not been nominalized -na might be expected to be an infinitive, but the homophonous dative would be more likely before a postposition. Whether dative or infinitive, both noun stems and verb stems can be followed directly by -na skang or -na git-a . Could verbs followed by one of these postpositions have been nominalized with a zero nominalizer? That would be very strange, since no hint of a zero nominalizer is found elsewhere in the language, and there is not much hint of any sort of zero anywhere. To interpret -na as an infinitive suffix would mean that in this situation, alone, postpositions follow different affixes when following verbs than when following nouns. To interpret it as a dative allows a casePage  236marker to be suffixed to a verb, an equally odd state of affairs. To speakers, of course, it makes no difference whether -na is a dative or an infinitive marker. Skang follows -na after both nouns and verbs. It is a problem only to linguists: an-dal-na skang 'before dark', am-bin-o-na skang 'before tomorrow', gu-al-na skang 'before forgetting'.

Like skang, git-a 'so that, suitable for' is able to follow a noun with dative -na (as well as one with genitive -ni ), and it can also follow verbs with -na :

A-ge 'before' is a synonym of skang that has been borrowed from Bengali. Like its native Mandi synonym, it follows -na without an explicit preceding nominalizer: cha'-na a-ge 'before eating', ring-na a-ge 'before drinking'.

Postpositions that follow locatives . A few postpositions can follow one of the augmented locatives -o-ni 'from, after' and -o-na 'to, until'.

Dip-at is unusual in being able to follow both -o-ni where it means 'since, after', and -o-na where it means 'until, up to, as long as'. While dip-at is probably the most common pronunciation for this postposition, several other variants are also used: dip-et, stip-at, stip-et . Some individuals seem to prefer one or the other, though they also recognize the others and agree that all are possible. I detect no difference in meaning and Mandis tell me they are synonyms. This is one of those places where, for no apparent reason, Mandi has a large number of alternative forms: am-bin-o-na dip-at 'until tomorrow' am-bin-o-ni dip-at 'after tomorrow, from tomorrow on'; da'-o-ni-dip-at 'from now on'.

Page  237

Dip-at can be used with nominalized verbs as well as with nouns. Since nominalization can be accomplished with -o plus either -ni or -na , yielding -o-ni 'since, after' and -o-na 'until, as long as', both nouns and verbs have identical suffixes before dip-at :

King-king 'until, as long as' follows the augmented locative -o-na when it is suffixed to a noun. Som-bal-o-na king-king 'until Friday':

King-king can also follow verbs with what must be regarded as an -o nominalization, plus a -na augment, yielding the same form that is suffixed to nouns: re'-ang-o-na king-king 'until going', neng'-ja-o-na king-king 'as long as (you) are not tired'. sal-ja-chi-o-na king-king 'until noon, until midday'.

Postpositions that follow -ming . The case marker -ming 'along with' can be followed by the postpositions grim or lo-ge both of which mean 'together'. Lo-ge is borrowed from Bengali. I believe that I have even heard both used together as grim-lo-ge which amounts to a double redundancy and makes it impossible to miss the togetherness:

Page  238

Postposition that follows locative -cha. -cha can take the postposition bak 'side, in the direction of': mik-kang-cha bak 'at the front side', bring-cha bak 'in the direction of the forest'. Bak is often used after another postposition. Postpositions are themselves nominals and many regularly have the locative case marker -o . Less often they have the alternative locative -cha , but one of the uses of -cha is then to take the postposition bak , and this allows a sequence of two postpositions. In the next example, ki'-sang-cha 'at my behind' is a postposition following the genitive case, and bak is a postposition following the -cha locative:

Postpositions following the nominative . The handful of postpositions that follow the nominative are distinguished from case markers only by the fact that they follow the -a or "free" form of the monosyllabic pronouns. If this -a is regarded as a case marker then they follow the nominative case marker.

Gin-ang 'with, along with'. A few speakers have assured me that gin-ang can follow the combining form of the pronouns. That would turn gin-ang into a case marker. The wider consensus, however, is that gin-ang must follow the -a form of the pronoun, and I therefore count it as a postposition, but this is one of the cases where the line between case markers and postpositions is blurred. In A'chik gin-ang and gri (see just below) can apparently be used in a rather eccentric but verb-like way where they mean 'to be with' and 'to be without', but examples are only rarely heard in Mandi, and I do not fully understand the nature of this verb-like construction. Even as a postposition, gin-ang is somewhat unusual in Modhupur, and it seems to be dying out among younger speakers. A few examples will suggest the range of its uses:

Page  239

Gri 'without'. Just about everything that has just been said about gin-ang could be equally well said about gri . Indeed, they form a pair of opposites, meaning 'with' and 'without', respectively. Like gin-ang , gri flirts with being a case marker, since it is occasionally, though not generally, used after the combining form of one of the monosyllabic pronouns. Like gin-ang also, it is sometimes used in a verb-like way meaning 'to be without', 'to not have', though neither gin-ang nor gri ever take verb suffixes. Gri may be marginally better established in Modhupur than gin-ang , but it also shows signs of dying out among younger speakers: chol-a gri 'without a shirt', mik-gil gri 'without eyelids, shameless':

Jang-chi-o 'at the middle of' can follow the nominative, as in song jang-chi-o 'at the middle of the village'. The nearly synonymous jat-chi-o Page  240more often appears in the fixed phrase sal jat-chi-o 'at the middle of the day, noon'. Postpositions, unlike adverbs, are not usually reduplicated, but jol-jol 'along, via, by way of' is used as a postposition in such phrases as a'-sam jol-jol 'along the bank (of a river, etc.)', or bring jol-jol 'by way of the forest'.

Sa-ra 'without' is a Bengali borrowing that is often used where gri might once have been used. It follows the combining form of the pronouns and so counts as a postposition rather than a case marker:

Borrowed Bengali Case Markers and Postpositions

(C)

pp. 240

The depth of influence that the Bengali language has had on Mandi is nowhere more dramatically illustrated than with the number of suffixes and postpositions that have been borrowed. Given the political and cultural pressures that have influenced the Mandis, it is hardly surprising to find large numbers of borrowed nouns, and even verbs, but Bengali words are borrowed with such ease that they drag their suffixes along with them. My impression is that Bengali noun suffixes are taken into Garo more easily than verb suffixes, but this may be because when verb suffixes are used I tend to interpret the resulting sentence as being fundamentally Bengali rather than Mandi. At that point I begin to feel that the remaining Mandi words have been borrowed into a sentence that is basically Bengali rather than the reverse. Whatever the case, borrowing can be sufficiently intense to make it difficult to be sure, in every instance, whether an utterance should be regarded as Mandi that has been strongly influenced by Bengali or Bengali that has been strongly influenced by Mandi. Still, I do think that there has been more resistance to borrowing verbal morphology than to borrowing nominal morphology. At least people rarely mix verb suffixes from the two languages into the same verb. Even the fact that, in the course of leaning Mandi, I have absorbed a considerable amount of Bengali noun morphology but very little verb morphology suggests that the noun morphology has been better integrated into Mandi.

Modern borrowing is made relatively easy by the typological similarity of the two languages, and the resemblances must be the result of some thousands of years of contact between the language families to which Garo and Bengali belong. Although Bengali is Indo-Aryian and Garo is Tibeto-Burman, both are verb final languages characterized by case markers, postpositions and rather free order of noun phrases before the verb. In bothPage  241languages, case markers are suffixed to the noun and postpositions follow the case markers. Bilingual Mandi speakers surely have some tendency to use Bengali case markers with Bengali words and Mandi case markers with Mandi words but that is only a tendency, not a rule. Bengali case markers can be heard on Mandi words and vice versa. I once heard a Bengali case marker attached to an English word which was embedded in a sentence that was otherwise Mandi.

The extent of borrowing varies with the amount of contact that speakers have had with native Bengalis, and it also varies with the speaker's expectations about the listener's knowledge of Bengali. Mandi residents of Dhaka use more Bengali words and more Bengali suffixes than villagers whose daily conversation is more often with other Mandis, and where standards of Bengali are not so high. Even in rural villages, however, Mandis in Bangladesh use a good many Bengali postpositions very easily, even when they seem to be synonymous with Mandi postpositions. Sometimes they even use the appropriate Bengali case marker on the word before the postposition, but more often they retain the Mandi case marker. It is striking that Bengali postpositions tend to follow the same case markers as Mandi postpositions of the same or similar meaning. Thus Bengali a-ge 'before, ago' follows the dative case markers, as does the synonymous Mandi skang . Mik-ta 'because', borrowed from Bengali, follows the genitive -ni as does the synonymous Mandi gim-in .

The use of a-ge 'before, ago' and po-re 'after' in place of their Mandi synonyms is probably encouraged by their use with time expressions that include numbers. People who have only Bengali words for the numbers higher than five, find it awkward to say something like 'six months ago' or 'six months from now' without putting the whole thing into Bengali. It is possible to say choi-mas-na skang 'six months ago' where choi-mas 'six months' is Bengali, but the dative suffix -na and the postposition skang 'ago' are in Mandi, but since there is no way of avoiding the choi-mas , the a-ge follows very easily to give choi-mas age . When the Bengali lo-ge 'together with' is used as a postposition following the Mandi case marker -ming 'along with', as it very often is, the effect is simply to strengthen the statement by bringing together two near-synonyms. When Bengali por-jun-to 'as long as, until' is used in place of the synonymous Mandi king-king , it is not easy to see what is gained, but it may come along with borrowed nouns or even suggest a hint of sophistication on the part of the speaker. Bengali sa-ra 'without' is used so often that some younger speakers feel uncertain about the use of the older Mandi gri . Still, large numbers of Mandi postpositions continue to be used even as Bengali synonyms enter the language.

Page  242

Case markers are borrowed less often than postpositions, and usually only when they are pulled along by borrowed words, but Mandis in Bangladesh speak Bengali well enough to have no difficulty at all in understanding the odd borrowed case marker when it crops up in a Mandi context. A woman who was speaking to me in Mandi once instructed me to get off a bus pa-ni tank-er ka-che 'beside the water tank'. Tank , of course, comes ultimately from English but pa-ni is Bengali for 'water' so pa-ni tank has a Bengali feel about it, and the phrase has nearly become the place name for this particular landmark. This makes it easy to use when speaking some other language than Bengali, just as place names like Dhaka or Bangladesh can be used in any language. The speaker might have said pa-ni tank-ni sam-ba-o , with a Mandi case marker ( -ni 'genitive') and postposition ( sam-ba-o 'beside'), but having used what felt like a Bengali expression it was easy to complete the phrase with Bengali forms instead: -er the Bengali genitive marker, and ka-che 'beside, next to', a Bengali postposition.