Describing the overt order of words within a sentence may appear to be a simple matter. To begin with, a combination of units may permit interruption by a break of some kind or by units from some other combination; alternatively the combination will demand contiguity and not tolerate interruption. Sentences are necessarily contiguous combination of words (rather than of formatives) within a sentence. Both words and sentences are, so to say, solid units.


Next, the combination may be thought of as a permutation with a specified sequence. A given language may permit any permutation of a given combination of words into a sentence. The word order may be free, fluid, with maximal randomness or entropy, as a mathematician would say, Thus, in Sanskrit with the combination naraya Nah (male given name) (nominative), rasam juice (accusative), pibati drinks, any of the six mathematically possible permutations are permissible.


(1)   (a) nārāya No rasam pibati.

(b) nārāya Nah$ pibati rasam.



All of these six mean Narayan drinks juice. Alternatively, the word order may be fixed, rigid, with minimal randomness or maximal predictability, thus, in English with a similar combination, John, juice, drinks only a single permutation is permissible.


(2)   John drinks juice.


In between, only some of the permutations may be permissible but not all. Thus, in Sindhi with combination hU he, AmbUmango khai eating, tho is only a subset of seven permutations are permissible out of the mathematically possible twenty-four. This is semi-fluid order.


(3)   (a) hU ambU khai tho

(b)   hU AmbU tho khai

(c)    hU khai tho ambU

(d)   hU tho AmbU khai

(e)    AmbU hU khai

(f)     AmbU khai tho hU

(g)    khai tho hU AmbU


All of these mean He is eating a mango.


This account seems to coincide with the common sense view. But at once doubts begin to assail us that things are not so simpleespecially to someone not knowing a language and asking for explicit instructions on proper word order.


The first question that may strike us is whether the word and the sentence are the only necessarily contiguous or solid combinations in an utterance. What about the phrase? (A phrase built around a finite verb is only a special case we shall call such phrases clauses.) Some phrases do not appear to be wholly solid: khai tho in Sindhi seems to be only partially solid in that it gets interrupted in 3rd. But what about tam rasam cold juice in Sanksrit? At least in highly literary Sanskrit the combination nārā yaNah, ītam, rasam, pibati appears to be as fluid as the shorter combination we considered earlier and permit separation of tam from rasam. But cold juice in English (possibly juice cold in poetic diction) and miho AmbU sweet mango in Sindhi are quite solid. even in Sanskrit clauses will be solid. (of course phrases within a clause need not be solid and the order within a clause may be fluid.) The wiser policy seems to be that we should inquire into the overt order of solid phrases within a sentence rather than into the order of words with in a sentence. A good empirical test for identifying solid phrases may be based on looking for potential points in a sentence for pauses or for parenthetical insertions.


The second question concerns the assumption that the combination has a specific grammatical structure. Think of the combination in English, a man, a dog, bites. Obviously the two following permutations are far from meaning the same thing. (Compare example-2 with2.)


(4) (a) A dog bites a man.

(b) A man bites a dog.


Surely this is not to be thought of as a departure from rigidity in English. The word order is fixed, but it is also distinctive; thats all. Actually we are dealing with two different grammatical combinations of the same grammatical combination.


(5)   (a) a man (subject),

a dog (object), bites

(b) a dog (subject),

a man (object), bites


The English words here do not show overt formatives (as in Sanskrit) to mark the grammatical functions, thats all. We are dealing here with the overt order of a combination of solid phrases that together constitute a grammatical structure (the covert order, so to say) within a sentence.


The third question concerns the observation that the various permutations in a non-rigid order (examples 1,3) mean the same thing (namely, Narayan drinks juice, He is eating a mango) Now do they really mean the very same or do they merely mean roughly the same thing? Obviously the differences between la, lb, etc. or between 3a, 3b, etc. are negligible meaning differences while the differences between 4a, 4b are far from negligible, they are major meaning differences. In other words, the different formations may weakly or strongly contrastive. As early as1957, Joos underlined the importance of this distinction between major and negligible meaning differences for the purposes of grammatical analysis. Items 5a, 5b are different grammatical combinations precisely because they lead to major meaning differences. The same grammatical combination is said to underlie permutations in fluid order examples la, lb etc.) or in semi-fluid order (examples 3a, 3b etc.) precisely because the resulting mean in differences are negligible. Consider the following ---


(6)   (a) John is ready.

(b) Is John ready?


In some sense, the difference between a statement and a question is a major meaning difference. But what about the underlying grammatical combination? Is it the same combination, namely, John, ready, is or are there two different combinations here? Again, there is some plausibility in saying that there is the same combination. This seems to suggest that it will be wiser to recognize that major differences may be carried by overt order (as in 6) no less than by grammatical structure (as in 4).


The fourth question concerns the carries of negligible meaning differences. We have identified one such carrier, namely, fluidity in overt order. Whether all mathematically possible permutations are admissible (example 1) or only a subset of them is admissible (example 3) is not all that important. But there are other carriers too besides overt order and grammatical structure. There are special formatives---Hindi has to, nA, hi, bhi and others; Marathi has tAr, na, aplA, ka, c, hi and others: German has doch, also auf and others. We shall call them sentence particles. And of course there are features of sentence accent, sentence tone, and sentence juncture. We shall together call such features sentence prosodies. In conclusion, it will note unreasonable to look for an at least loosely unified system of all these carriesovert order, sentence particles, sentence prosodies, and certain grammatical functions that we call sentential functions-conveying a family of negligible meaning difference, and at least some kind seen in example 6. The different carries often tend to support or even replace one another.


The fifth and last question concerns the correlation between the carriers and the meaning differences. It is not enough to say, for example, that in a given language with a given grammatical structure one meets with some grammatical structure one meets with some fluidity in overt order. Thus, in example 3 even A superficial knowledge of fifth language will reveal that a specific order is the normal, colorless, unmarked order, the rest being in some sense deviant, colourful, marked orders. Only a more intimate knowledge of the language will enable one to sort out which order is associated with which sentence particles and sentence prosodies and with which specific color or meaning difference. Only then the answer to the question, namely, which subset of permutations, holds any interest for us.


In an earlier presentation (Kelkar 1982), I offered grounds for recognizing two sets of grammatical functions at the rank of a sentence in Marathi: (i) a set of dietetic functions, namely, Agent, Tenant, and Object that are closely associated with selection of the verbal: (ii) a set of sentential function, namely, Subject, Substrate, and Theme that in some ways parallel the dietetic functions but that are distinct from them: (iii) a set of the remaining sentential functions, namely, Circumstantial, Manner, and Verbal. The relevance of these considerations to the present concerns stem two facts, namely


(i)                 The unmarked overt can be set out in terms of the sentential functions as follows: Circumstantial, Subject, Substrates, Theme, Manners, Verbal

(ii)                The marked overt orders can be seen sometimes as the assignment of given constituents variably to different sentential functions and sometimes As playing around with the sentential functions themselves.


Some illustrations are called for at this point.



cora ni mitra la pivi; bAkSis dili:

theif by friend to bag gift gave

Agent Tenant Object Complement verb

Subject Substrate verbal

the thief gave the bag to the friend as a present


(8) (a) Cora ni pivi: mitra la dIli:

thief by bag friend to gave

Subject Theme Verbal

(b) pivi: cora ni mitra la dili:

bag thief by friend to gave

Subject Substrate Substrate Verbal

(source) (goal)

(9) tikDe cor jhopla

there thief slept

Circum Subject Verbal

(10) cor tikDe jhopla

thief there slept

Subject Manner Verbal


Whether the thief (the Agent) is cast in the rle of the Subject (as in 7, 8a) or the source-Substrate (as in 8b), whether to the friend (the Tenant) is cast in the role of the Substrate (as in 7, 8b) or as a part of the Verbal Subtle and therefore negligible difference between The thief gave the bag to the friend and The thief gave the friend the bag), whether there is cast in the rle of the Circumstantial (as in 9) or the Manner (as in 10)all these depend not so much on the facts of the case but on the way they are construed or interpreted in the speakers intent.


We have thus illustrated how the deviant order (examples 8, 10) may differ from the normal order (respectively examples 7, 9) in the assignment of certain features of the event or the state of affairs being presented (the agency of the thief, the object hood of the bag, the tenancy of the friend, the location of the sleep) to this or that sentential function.


Now we shall illustrate the other alternative, namely, the playing around with the sentential functions themselves.


(11) (a) cora ni mitra la pivi: dili:║

the thief gave the bag to the friend.

(b)               cora ni | mitra la pivi dili:║

(c) cora ni mitra la| pivi: dili:║

(c)               cora ni mitra la pivi: | dili:║



Here the overt order (to gather with the sentential functions) is maintained intact, but the medial sentence juncture is placed differently in separating the presupposed background (absent in 1la)) from the offered foreground. (The placement of the sentence tone and accents also shifts concurrently. One may observe in passing that Marathi television news readers tend to make eye contact with the viewer in uttering the first constituent of the offered foreground that bears also a sentence accent and the beginning of the tone nucleus.) Any global question or negation affects the offered foreground.




(b) tyani mitra la | pivi: dili:║

(c) tyani tyala | pivi: dili:║

(d) tyani tyala ti| dili:║


One or more elements of the presupposed background is liable to situation-dependent pronominalization or even to situation-dependent ellipsis (that is, tyani, tyala, ti in 12 may be left understood).


In contrast to these modes of de-emphasis, there are modes of emphasis of different kinds. Any focal question or negation affects the emphasis-bearing constituent. Apart from contrastive sentence accents and sentence particles, departures from the normal covert order of sentential functions also act as carriers of emphasis. There are four such departures:


(i)                 Front-shift of the emphasized element which is then separated by a sentence juncture.

(ii)                End-shift of the emphasized element which is then separated by a sentence juncture.

(iii)              Shift near the Verbal with which the emphasized element forms a solid constituent.

(iv)              Front-shift of the solid constituent (made up of the emphasized element and the Verbal ) which is then separated by a sentence juncture.


Some examples follow. The emphasized element is always chosen from the offered foreground.


(13) (a) cora ni | mitra la pivi dili:║

(b) mitra la | cora ni pivi dili:║

(c) pivi | cora ni mitra la dili:║

(d)dili: | cora ni mitra la pivi:║


(14) (a) mitra la pivi dili: | cora ni║

(b) cora ni pivi: dili: | mitra la║

(c) cora ni mutra la dili: pivi:║

(d) cora ni mitra la pivi| dili: ║


Note that the front-shift of the normally first element and the end-shift of the normally last element are not wholly vacuous (examples 13a, 14d, ) in that sentence prosodies still mark the shift.


(15) (a) mitra la pivi |cora ni dili: ║

(b) cora ni pivi | mitra la dili: ║

(c) cora ni mitra la| pivi dili: ║


Note that the shift of the element normally adjacent to the Verbal to the Verbal-adjacent position is not wholly vacuous (example 15ac) in that sentence prosodies stil mark the shift.


(16) (a) cora ni dili: | mitra la pivi ║

(b) mitra la dili: | cora ni pivi ║

(c) pivi dili:|cora ni muitra la║

It will be noticed that the medial sentence juncture separating the emphasized element chosen from the offered foreground obliterates the medial juncture separating the offered for-ground from the presupposed background. Other details concerning the associated sentence prosodies need to be worked out.


As for the meaning differences associated with the four kinds of shifts of emphasized element, only some general observations can be made pending a closer investigation:


(i) This emphasis is contrastive emphasis and should not be confused with intensive emphasis conveyed in Marathi by drawl accent, clip accent, and other carriers.

(ii) Front-shift strengthens the foregrounding effect; end-shift may have the effect of an afterthought. Front-shift is the more frequent; end-shift is the more forceful. Front-shift is analogous to cleft sentences in English. End-shift is analogous to pseudo-cleft sentences in English. (It is the thief that... versus The one who. is the thief)


(iii)               Verbal adjacencies is weaker than simple end-shift. Simple front-shift is weaker than front-shift with Verbal adjacency.


While this is by no means a full-scale presentation of Marathi word order or rather Marathi phrase order, I hope I have presented enough material to substantiate the approach to this problem proposed in the opening discussion of some of the questions that need to be raised against the common sense view that is too often uncritically accepted by Indian students of language whether traditional grammarians or linguistically trained analysis.






Ioos, martin 1957. An axiomatic approach to English grammar. Talk, Linguistic Institute, Ann Arbor. 6 August.



Kelkar, Ashok R. 1982. Diathesis in Marathi. Talk, 3rd International Conference of South Asian Languages and Linguistics Mysore, Jan.

Suggested Marathi terminology

ellipsis(n.) ָ

adjacent (a) ׭֍֟

emphasis (n.) ָ֬-ק

agent (n)

facts of the case(n.) ֣, ßãן

background (n.) ״ ; š״

filler (n.) ã֭֯

circumstantial (n.) ֌׾ßָ

fluid(free) (a.) ׿ףֻ

clause (n.) ־֌

foreground (n.) ״

complement (n.)

function (n.) ֵ

constituent(n.) ֙

goal(n.) ֭֬

contiguity (n.) ִ߯, ע

insertion(n.) ׭־֭

contrastive(a.) 1 (conveying difference) ־֓ 2 (conveying contrast or exclusion) ֵ־֟ strongly c (major contrast) ï™-־֓ weakly c (negligible contrast) ߝ-ֵ־֓


covert(a.) ϓ

manner(n.) ׾֬׾ßָ

de-emphasis(n.) ָ֬-ֵ

marked(a.) ׸

diathesis (n.) ׍ֵ֭

medial(a.) ֬ß

overt (a.) ύ

object (n.)

parenthesis (n.) 1 ֵָ2 ֵָ֟

offered (a.) ß

phrase (n.) ϤӬ

order (n.)

presupposed (a.) ߟ

situs (n.) ׭֭֬

rank (n.)

slot (n.) ã֭

rigid (fixed) (n.)

solid (n.) Ӭ

sentence accent ֌ ֻ

source (n.) ֭֬

sentence juncture ֌-ߴ֍

speakers intent ׾׾ց

sentence particle ֌-׭֯֟

subject (n.)

sentence prosody ֌-ֻפ

substrate (n.) ָ֬, ֵ

sentence tone ֌-

tenant (n.) ָ̍

sentential (a.) ֌֐֟

theme (n.) ׾ֵ

sequence (n.) ԯֵ, ԯָ

unmarked (a.) ׸

situation (n.) Ӑ

verbal (n.) ׾֬


word (n.) 1(grammatical rank) ֤ 2 (phonological rank) ֲword order(n.) ֤-




This paper was presented in the Symposium Linguistic aspects of Marathi at the 14th All India Conference of Linguists held at Nagpur on 5-6 July 1985. It was published in Professor R.G. Bhandarkar Commemoration Volume, Bulletin of the Deccan College Postgraduate and Research Institute 47-48: 129-34, 1988-9, published January 1989. A Marathi version appeared Bāhā ān&i jīvan 3:4; Divati: 1985