1. Introduction

The Tamil Case system is analyzed in native and missionary grammars (henceforth NMG) as consisting of a finite number of cases1 (realized morphologically as nominal or pronominal suffixes), to some of which postpositional suffixes may be added. In these traditional analyses there is always a clear distinction made between postpositional morphemes and case endings. Thus the usual treatment of Tamil case (Arden 1942) is one where there are seven cases--the nominative (first case), accusative (second case), instrumental (third), dative (fourth), ablative (fifth), genitive (sixth), and locative (seventh). The vocative is sometimes given a place in the case system as an eighth case, although vocative forms do not participate in usual morphophonemic alternations, nor do they govern the use of any postpositions.

1What a typical NMG grammar of Tamil gives as a description of the case system of modern Literary Tamil (Arden 1942: 75) is given in Table 1.

The problem with such a rigid classification is that it fails in a number of important ways adequately to account for both the inventory of case morphemes, or for syntactic constraints of various sorts on the system. That is, it is neither an accurate description of the number and shape of the morphemes involved in the system, nor of the syntactic behavior of those morphemes (and other morphemes, especially verbs, that control the occurrence of particular case markers). It is based on an assumption that there is a clear and unerring way to distinguish between case and postpositional morphemes in the language, when in fact there is no clear distinction. It fails to deal with variation in the system, whether in the syntax or the morphology. In fact, none of these problems with the NMG analyses is news to anyone who has studied the case system in detail, but this study may be the first to catalogue these problems in a systematic way. Let us therefore begin by examining these problems in the order already presented.2 (I shall violate continually the rule that diachronic and synchronic descriptions should not be mixed, because to separate out descriptions of various stages of the history of Tamil for separate treatment would then require repeating what are essentially the same complaints about the analyses of the system--the problems tend to be the same, no matter what stage of the language we are dealing with.) To summarize the problems:

1. What are the case morphemes and their phonological shapes?
2. What is their syntactic behavior?
3. How do we distinguish between case morphemes and postpositional morphemes?
4. How do we deal with variation in the system, especially variation that is controlled by pragmatic considerations, rather than purely syntactic ones?
5. What special problems do we encounter when dealing with modern Spoken Tamil?
6. Would the best analysis of this system in fact be one that treats it as whole system rather than case versus postpositions?

1.1 Inventory and Distribution of Case Morphemes

The first problem is that of the failure of NMG analyses to describe the actual distribution of case morphemes, since in almost any stage of the language that one might want to examine there are a number of situations where case morphemes are in fact replaced by postpositions, or there is variation between the occurrence of one or another case ending, and/or one or another of the morphemes usually called postpositions. For example, NMG analyses fail to assign an appropriate separate place in the system for instrumental and sociative uses3 of the so-called third case (the third case in fact has separate suffixes for instrumental and sociative uses, but is still regarded as one case). NMG analyses also include an ablative case that is clearly formed from a locative case-marker (-il) plus a postposition (-iruntu). (In modern spoken Tamil, the system breaks down even further, with postpositional morphs completely replacing case suffixes in some instances, or combining with case suffixes to form what seem to be as genuine a kind of "case" suffix as is the ablative, which was long ago admitted to membership, despite its clear construction using a locative marker plus a postposition.) NMG’s also typically fail to provide an adequate explanation for the genitive, which often precedes other case markers (i.e. has other case markers suffixed to it) so that it is then relegated to the status of an "oblique" form, or is classified as an "adjectival" form, or a stem alternate; in any event it is demoted to something less than a "real" case marker, ostensibly because of some notion that a "true" case marker in Tamil could not have another genuine case marker affixed to it. This ambiguity of the status of the genitive is not so much of a problem when it comes to nouns, but with pronouns, where the oblique stem may function as a genitive, e.g. en pustakam "my book" one might wonder why this oblique stem can be genitive when case markers can be added to it that also function as genitive, e.g. eṉṉuṭaiya pustakam (spoken ennōḍe pustakam). In the modern spoken language various changes have also led to some homonymy in the system, with the Literary Tamil (henceforth LT) genitive form uṭaiya being pronounced in Spoken Tamil (ST) sometimes as ōḍe, in other dialects as ōḍu, which is homophonous with the "sociative" uḍan/ōḍu/ōḍe in some dialects; in others no such confusion may result, or some other morpheme may be used for "association", such as a postposition, e.g. kūḍa, toṇeyle or some others. The instrumental case marker itself (LT -āl, ST -āle) may also vary in ST, with some dialects employing postpositions instead of the official instrumental ending (LT kaiyāl).4

2Lest it appear that I am trying to build up suspense about the origins of this confusing system, only to show my great erudition when I reveal the true system, I should say that it has always been obvious5 that much of the case system has been modeled on that of another language, and that the natural system of Tamil has been forced into this other mold, with the result that what are clearly two different cases are made to fit into one because of some notion that the system had to have seven and only seven cases. To Indo-Aryanists it will be obvious that much of the above NMG system is modeled on the case system of Sanskrit, which has seven or eight cases (ablative and genitive are often subsumed under one, vocative and nominative under another, etc., depending on the paradigm of the declension in question). Even the order of Tamil cases is approximately the same as those given for Sanskrit. Since this system does not, as we have just seen, work very well, and is obviously a model imposed from another language, (just as Latin was once used as a grammatical model for modern European languages), it is obviously high time to abandon this foreign system. Since Tamil grammarians usually abjure any influence from or debt to "northern" grammatical models, there should be no difficulty in forsaking this inappropriate grammatical model in favor of one designed to fit the facts of the language. In fact when we look at the history of grammatical treatments of Tamil from the oldest records6 onward for an idea of what the case system was originally like, and how it has changed, we see that there has often been disagreement about how to analyze the system, and that there has also been a constant history of substitution of new morphemes for case endings, replacing older morphemes previously in use. What has not happened is a reinterpretation of the system to include new categories, or new cases, as well as the new morphemes. And of course when we get to the modern language, there is very little attempt by any grammarians to deal with the system at all, since the Spoken language does not officially exist.

1.2. Syntax and Case

The syntax of the Tamil case system is usually dealt with in approximately the same manner as the inventory of morphemes: we are told that the case in question is "governed" by various constraints, such as that the accusative is "governed" by the presence of certain transitive verbs, and that nouns that are the objects of verbs are marked accusative. Dative is controlled by verbs of motion, marking motion toward something, or by verbs of giving, etc. This can be illustrated by considering the text of a translation by Zvelebil (1982: 10) of the grammatical portion of Tolkāppiyam dealing with the case system.

75. The fourth is the case which is named -ku. Whatever substance it may be, it [denotes] receiving.
76. They say that the case denotes the object for which an action is done, that to which one subjects himself, that to which something is apportioned, that which is transformed, that which is suitable to something, that for which something is done, and to express friendship, enmity, love and greatness and so on."

My own suggestions for how the syntax of case relations should be handled is dealt with below in sections 1.3 and 1.4.

1.3 Case and Postpositions

The distinction between case and postpositions is also dealt with as a given requiring no justification or explanation by most NMGs. Thus Arden (1942: 74-77) gives the following explication about the Tamil case system:

102. Tamil Nouns have two Numbers (1) The Singular and (2) The Plural. In each Number there are eight Cases….
103. The Cases are known in Tamil grammar as the First Case, Second Case, Third Case and so on. […]
105. (A) Singular Number.
(i) The First Case or the Nominative Case Singular is the Noun itself; as nari a jackal, maram a tree. […]
Except the Vocative the remaining Cases are all formed by adding certain casal suffixes to the Inflexional Base…. It is sometimes called the "Oblique" Case. [His footnote, hs.]
The Inflexional Base is the form of the Noun that takes the casal suffices except the suffix of the Vocative. It is often the same as the Nominative; but it sometimes has a peculiar form of its own. Thus the Inflexional Base of vaṇṇāṉ is vaṇṇāṉ; of nari is nari; but the Inflexional Base of maram is marattu.

Postpositions are presented as equivalent to English prepositions, except that they are

[…] suffixes added to the words which they govern, often called Particles and sometimes called Postpositions." (Arden 1942: 125)
[…] These Postpositions are often Nouns or Verbs in origin, e.g. kīḻ is a Noun, a place below, pōtu is a Noun, time, oḻiya is a Verb, to cease or to set aside." (ibid.)

Arden then catalogues the various postpositions and the cases they are suffixed to; he notes of course that there is variability in the system:

Instead of the Accusative, some of these words occasionally take the Nominative case." (Arden 1942: 126)
[…] But they may also be added to the Inflexional Base." (ibid. p. 127)
[…] The following are added to the Inflexional Base. Neuter Nouns usually insert in between the Noun and the Postposition." (ibid.)

In yet another section of his grammar, entitled "Uses of the Cases of Nouns and Pronouns," one finds some interesting observations about syntactic problems involving cases and postpositions:

362. (i) The Genitive expresses possession. […]
(iii) But in many such cases the case ending is omitted and the Nominative form used. […]
(iv) Very often the Inflexional Base is used. […]
(vi) There is no rule as to which ending iṉ, uṭaiya, iṉuṭaiya, atu should be used. Westerners are apt to use the ending iṉ too frequently.
(vii) The Nominative Case should not be used for the Genitive if any ambiguity is likely. (Arden 1942: 190-1)

From Arden’s description we can see that although he accepts a priori the notion that case and postpositions are separate morphological categories, his own description of exceptions to the rules he gives us, especially in (362) belies the categorical differences he claims exist.

1.3.1 Postpositions

Having made the claim that there is no clear cut distinction between case and postpositions in Tamil except for the criterion of bound vs. unbound morphology, we are forced to examine all the postpositions as possible candidates for membership in the system. Actually this is probably going too far in the other direction (and probably beyond the scope of this paper), since then almost any verb in the language can be advanced to candidacy as a postposition. There are the clear-cut bound morphemes (-ai, -ku, -il, -āl, etc .) and then there is the set of almost-bound morphemes that occur quite regularly with one or more of the pakka case morphemes. These are usually de-nominal forms with some semantically-locative meaning (mēle, kīḻe, pakkattule, uḷḷe, the points of the compass, hand directions) but there are also a number of de-verbal forms that, by virtue of their frequency of use, seem to have acquired independent status as postpositions. These are either infinitival or pastparticipial forms: -āka, patti "about", tavira "besides, instead of", sēndu/sēttu "together, pāttu "at, toward" (psychologically), pōla "like", and a few others. In some cases, the verb from which these are derived is archaic or obsolete as a main verb, leaving the postposition "orphaned", at least in ST, with no verbal origin to act as its derivational source. Other de-verbal forms, such as the so-called ablative irundu (which is affixed to the locative -le, kiṭṭe, or semantically-locative postpositions such as mēle, kīḻe, or adverbs such as aṅge "there", have been considered from earliest times to fill the bill as case forms despite their clear formation on the locative,7 but because of the frequency of occurrence and unchallengeable independent status of the copula (iru) of which irundu is the past participle, a truly ambiguous situation results for which the grammarians have no comfortable explanation. Another good example of the questionable status of the ablative as a "true" case is shown by what happens when clitics ō and āvatu are affixed to ablative-marked adverbs such as eṅge. One would expect that clitics, which are always found word-finally in Tamil, would be suffixed to the "case" marker irundu but in fact parallel to examples where this occurs, i.e. eṅgerund-ō "from somewhere or other" we also get eṅgeyō-irundu "ibid." If the ablative were a true case it should not be possible to split it with something as word-final as a clitic. We could also, of course, include other verbs not commonly used as postpositions, such as the past participle of pōḍu, "put" and the past participle of the Literary verb koḷ "hold" (Asher 1982: 112) which can substitute for the instrumental, as in

a. katti pōṭṭu veṭṭuṅga "Cut it with a knife" [knife hvg-put cut]

or

b. katti koṇḍu veṭṭuṅga "Cut it with a knife" [knife hvg-held cut]

instead of

c. katti- āle veṭṭuṅga "Cut it with a knife" [knife instr. cut]

It would be convenient if we possessed a linguistically-universal metric for evaluating case systems with which we could compare particular linguistic systems to determine what is or is not a valid system. I know of no model that can serve as a touchstone, although there are various attempts, ranging from the Sanskrit kāraka system to Fillmore’s The Case for Case. Languages with elaborate case systems such as Finnish and Hungarian cry out for comparison, but I suspect that underlying the case morphemes in these luxuriant case systems there are postpositions derived from some other formclass.

1.4. Variation in the system

In ST we also get some variation in the system that is determined by pragmatic considerations, as well as by meaning, function, and syntax, so there are additional problems here (Schiffman 1985).

3For example, the postposition kiṭṭe (LT iṭam) is usually described as a "locative" marker, that is, a marker of location used only with human subjects. This is because the regular locative case marker -le (LT -il) cannot occur with human or animate noun phrases. However, when kitte does occur with human NP’s, its meaning is not so much location in (or at or on) as it is an expression of "in the presence of, in the possession of, near".

1. avan- kiṭṭe paṇam irukku "He has money" [he loc. money is]
(i.e. on his person)

This contrasts with

2. avan- ukku paṇam irukku "He has money" [he dat. money is]
(i.e. he is wealthy)

In 1, the notion conveyed is temporary location, while 2, which is marked for dative, conveys the notion of permanent possession, a permanent or habitual state, and inalienable possession. kiṭṭe also substitutes for the locative in ablative expressions ("from" something) since the ablative in Tamil is made up of a locative case marker plus a postposition irundu. Certain semantically locative expressions that lack (or cannot take) locative case markers add irundu directly, as in

3. aṅge-y-rundu "from there"
4. mēle-y-rundu "from above, from on top of"

and to express motion away from a person, (i) rundu is added to kiṭṭe instead of locative le:

5. vīṭṭu-le-rundu vandadu "it came from the house"
6. rām-kiṭṭe-rundu vandadu "it came from Ram"

What the grammars do not describe is that kitte is also used instead of dative case with certain verbs that ordinarily require the dative, or even, in some cases, the accusative. Such verbs as sollu "say", terivi "inform", kēḷu "ask", pō "go", and some others, such as sibārisu paṇṇu "recommend" and maṇṇu pōḍu "apply" in actual usage are now more often found with kiṭṭe instead of dative or accusative because of the use of the latter seems to imply a directness and bluntness that is not deferential enough.

7. avane kēḷu "ask him (outright)"
8. avan-kiṭṭe kēḷu "ask him (nicely)"
9. enakku sonnāru "he told me outright"
10. en-kiṭṭe sonnāru "he broke the news to me gently, he told me in a nice way"
11. ḍairekṭar-kiṭṭe teriviṅga "Please inform the Director"
12. ḍairekṭar-ukku teriviṅga "Please tell the Director (and don’t mince words).

Despite the difficulty of rendering the differences in these pairs with consistent English translations (having to resort to lexical variants of English verbs, etc.) it should be evident that what is expressed with kitte is a manner of operation that differs from the direct manner conveyed by the dative or accusative. This manner has little to do with politeness to the addressee or third person, since it makes no difference who the actor or agent or dramatis personae are--what kitte implies is that no matter who did what, it was done in a deferential manner. It can occur with or without normal politeness markers because it is not dealing with politeness as this is usually defined; thus there is a four-way variation possible here: with or without politeness, with or without deference.

1.5. Differences between Literary and Spoken

There are other differences between LT and ST case systems, and as it may already be obvious, the seven-case format adopted from Sanskrit is even less appropriate for ST than it is for LT. To place it temporarily in a format similar to the LT one, however, for the purposes of comparison, let us consider the schema in Table 2:

This systematization is rather oversimplified since it does not take into account social or regional dialect variation, nor does it deal with a number of other issues we have mentioned, in particular the problem of what is in effect a continuum that begins with the case markers (by which is traditionally meant bound morphemes not occurring elsewhere) and merges gradually with the set of morphemes known as postpositions, which have independent syntactic status (being derived from nouns or verbs). These postpositions now occasionally or regularly replace certain of the case markers, which is what all the fuss is about. Some of these forms are found only in certain phonological or syntactic environments.

4This schema also does not deal with variation in the system above and beyond the variation already mentioned involving deference. This variation is probably to be characterized as a kind of syntactic problem, because it does not involve any new morphemes, but rather involves use of a different case than the one specified by the grammars in collocation with certain verbs. This, of course, is what is involved in the variation already mentioned with the dative and the animate locative (kiṭṭe).

5Let us now examine a number of syntactic and semantic problems involved in case variation of this and other sorts. According to the general rules of uses of the dative, grammars usually state that the modal verb muḍiyum "an, be able" requires the dative case, as in enakku muḍiyum "o-me is-able" or" can, am able (to do s.t.)" In fact, however, the instrumental suffix -āle also occurs often with this modal verb, but the meaning is different: ennāle muḍiyum" am able" implies that the subject is not only able but willing to undertake some action. In other words, a sentence like idu uṅgaḷāle muḍiyumā? means not "re you able to do this?" but actually "re you willing and able to undertake to do this; are you willing to make an effort (on my behalf)?" ("Can you give this a go? Can you take on this task/effort? Can you put yourself out for me in this matter?" are some other possible ways of interpreting this kind of construction). As can be seen, unless such syntactic factors are taken into account in describing the Tamil case system, a mere listing of case morphemes and their approximate meanings is not an adequate kind of description.

1.5.1. Variation with the Dative

Another kind of variation that is not usually described adequately is the difference between uses of the dative ukku suffix and the extended dative involving ukku plus āka, which is usually glossed as "or the sake of". In fact it is not always clear in actual usage why one of these is used and not the other. For example, using the previous example and extending it somewhat, we might get the sentence

a. enakk-āka idu uṅgaḷāle muḍiyumā?

which varies with

b. enakku idu uṅgaḷāle muḍiyumā?

The difference in meaning is approximately as follows:

a. Can you make an special effort to do this for me?
b. Can you make an effort to do this for me?

Another example of this might be illustrated by the pairs c. and d.:

c. idukkāka nān vandirukkēn" came especially to do this"
d. idukku nān vandirukkēn" came to do this"

1.5.2. Dative with uḷḷe

Dative is also "xtended" by the addition of uḷḷe "ithin" to indicate "interiority", (perhaps the Finnish "nessive" as contrasted with "dessive" is a good term here) i.e. a kind of "deeper" locative than the kind of location provided by the plain seventh case. That is,

a. vīṭṭu-kk-uḷḷe irukku "It is within the house"

contrasts with

b. vīṭṭu-le irukku "It is in the house"

in the sense that (a) emphasizes the interiority, inclusion, surroundedness of an object--that something is deep in the heart of something, while (b) merely states that X is in something. When used with verbs of motion, dative + ulle emphasizes that something penetrated into the core or interior of something, (cf. Finnish illative) while plain dative may only indicate "allative" position:

c. vīṭṭu-kk-uḷḷe pōyirukkāru "He went right into the house"
d. vīṭṭu-kku pōyirukkāru "He went to the house, he went home"

Dative + uḷḷe may also be used also to express the equivalent of English "among", i.e.

e. avarkaḷukkuḷḷē nālu peyarai aḻaittāṉ. "He called four among them" (Arden 1942: 189)

In modern spoken, however, the locative would be more likely be used:

f. avaṅgaḷ-le nālu pēre aḻeccān. "idem."

1.5.3. Proximity vs. contact

Another variation with the dative and genitive is with certain postpositions indicating relative position, such as mēle, kīḻe, pakkattule and the points of the compass. When used with the genitive, they indicate that there is contact between the two objects, whereas with the dative they indicate that there is proximity but no contact:

1. marattu-mēle irundadu "It was on top the tree"
2. marattu-kku mēle irundadu "It was above the tree"
3. vīṭṭu mēle/kīḻe/pakkattule irundadu "It was on top of/on the underside of/on the side of the house"
4. vīṭṭ-ukku mēle/kīḻe/pakkattule irundadu "It was above, below, next to the house (but not touching it)"

1.6. Conclusions

Having presented a rather scathing criticism of native and missionary grammarians’ attempts to deal with the Tamil Case System, and having reviewed various problems that need to be dealt with, I am convinced that a taxonomic approach that attempts to categorize case morphemes on the one hand and postpositional morphemes on the other is missing the point. Obviously the whole system must be treated in its entirety, since when all the morphemes in question are viewed in toto, what emerges is an overall system consisting of (1) an inner core of indisputably bound case morphemes, (2) an intermediate layer of case/postpositional morphemes, (3) and an outer layer of morphemes that occur elsewhere in the language. The Tamil Case System is a kind of continuum or polarity, with the "true" case-like morphemes found at one end of the continuum, with less case-like but still bound morphemes next, followed by the commonly recognized postpositions, then finally nominal and verbal expressions that are synonymous with postpositions but not usually recognized as such at the other extreme. This results in a kind of "dendritic" system, with most, but not all,8 of the basic case nodes capable of being extended in various directions, sometimes overlapping with others, to produce a thicket of branches. The overlap, of course, results from the fact that some postpositions can occur after more than one case, usually with a slight difference in meaning, so that an either-or taxonomy simply does not capture the whole picture.

6I am also tempted to compare Dravidian case systems with those of the Altaic languages, which on typological and perhaps even historical grounds, show strong similarities. It is interesting to note that the genitive case, for which no satisfactory analysis emerges in Dravidian, is also a problem in Altaic languages, and is in fact not usually treated as a case in Proto-Altaic. Altaic languages are thus shown to have four cases at the "core", as it were, with other accretions similar to those found also in Dravidian. I do believe it is high time to abandon the rigid seven or eight-case system since doubts about its validity have been voiced since the earliest analyses of Tamil began, and the problems with it can be seen to be growing worse rather than better. Such a system also works poorly for the other Dravidian languages, as is obvious from the most cursory glance at the present-day grammars of Malayalam, Kannada and Telugu. It is also high time that Tamil scholars collaborated on a modern description of both Literary and Spoken Tamil that corresponds to the facts of the language, rather than to some outdated notion borrowed from another time and grammatical tradition.