Morphosyntax of Dravidian Languages
Summary and Keywords
The Dravidian languages, spoken mainly in southern India and south Asia, were identified as a separate language family between 1816 and 1856. Four of the 26 Dravidian languages, namely Tamil, Telugu, Kannada, and Malayalam, have long literary traditions, the earliest dating back to the 1st century ce. Currently these four languages have among them over 200 million speakers in south Asia. The languages exhibit prototypical OV (object–verb) properties but relatively free word order, and are rich in nominal and verbal inflection; only Malayalam lacks verb agreement.
A typical characteristic of Dravidian, which is also an areal characteristic of south Asian languages, is that experiencers and inalienable possessors are case-marked dative. Another is the serialization of verbs by the use of participles, and the use of light verbs to indicate aspectual meaning such as completion, self- or nonself-benefaction, and reflexivization. Subjects, and arguments in general (e.g., direct and indirect objects), may be nonovert. So is the copula, except in Malayalam.
A number of properties of Dravidian are of interest from a universalist perspective, beginning with the observation that not all syntactic categories N, V, A, and P are primitive. Dravidian postpositions are nominal or verbal in origin. A mere 30 Proto-Dravidian roots have been identified as adjectival; the adjectival function is performed by inflected verbs (participles) and nouns. The nominal encoding of experiences (e.g., as fear rather than afraid/afeared) and the absence of the verb have arguably correlate with the appearance of dative case on experiencers. “Possessed” or genitive-marked N may fulfill the adjectival function, as noticed for languages like Ulwa (a less exotic parallel is the English of-possessive construction: circles of light, cloth of gold). More uniquely perhaps, Kannada instantiates dative-marked N as predicative adjectives. A recent argument that Malayalam verbs originate as dative-marked N suggests both that N is the only primitive syntactic category, and the seminal role of the dative case.
Other important aspects of Dravidian morphosyntax to receive attention are anaphors and pronouns (not discussed here; see separate article, anaphora in Dravidian), in particular the long-distance anaphor taan and the verbal reflexive morpheme; question (wh-) words and the question/disjunction morphemes, which combine in a semantically transparent way to form quantifier words like someone; the use of reduplication for distributive quantification; and the occurrence of ‘monstrous agreement’ (first-person agreement in clauses embedded under a speech predicate, triggered by matrix third-person antecedents).
Traditionally, agreement has been considered the finiteness marker in Dravidian. Modals, and a finite form of negation, also serve to mark finiteness. The nonfinite verbal complement to the finite negative may give the negative clause a tense interpretation. Dravidian thus attests matrix nonfinite verbs in finite clauses, challenging the equation of finiteness with tense.
The Dravidian languages are considered wh-in situ languages. However, wh-words in Malayalam appear in a pre-verbal position in the unmarked word order. The apparently rightward movement of some wh-arguments could be explained by assuming a universal VO order, and wh-movement to a preverbal focus phrase. An alternative analysis is that the verb undergoes V-to-C movement.
A Note on the Transcription
IPA symbols are used, with the following modifications that conform to the transliteration system prevalent in the Dravidian literature:
1. Long vowels are indicated by doubling a letter: e.g., /aa, ii, ee, uu/.
2. Dravidian has no /v/- /w/ contrast. Conventionally, /w/ has been used for Malayalam and Telugu, and /v/ for the other languages, for the bilabial semivowel.
3. Dental consonants have been left unmarked: thus /t, d, n/ represent dentals, as in Krishnamurti (2003). Where there is an alveolar–dental contrast (as in Malayalam), the alveolar is marked by a diacritic (subscript bar): e.g., ṉ.
4. The palatalized /k/ in the Malayalam causative suffix is conventionally transliterated /k’/.
5. The symbol R is used for the Malayalam alveolar tap.
1. The Dravidian Languages
The name ‘Dravidian’ was proposed by Bishop Robert Caldwell in his comparative grammar (1856) for a family of languages that were until then referred to as the ‘South Indian dialects’ or ‘South Indian vernaculars.’ A claim that Tamil is a ‘southern language’ of India as ancient as the ‘northern language’ Sanskrit is found in the earliest textual sources for this language (the first known work in Tamil is a treatise on grammar and poetics from the early pre-Christian era; Krishnamurti, 2003, p. 22). But this claim had to contend with a counterclaim that the languages of the south were mere corruptions of Sanskrit and its derivative Prakrits (Zvelebil, 1990, p. xix). In 1816, Francis Whyte Ellis, a British administrator in Madras, asserted that the south Indian dialects were in fact an independent family of genetically related languages, and he enumerated seven of them: Tamil, Telugu, Kannada, Malayalam, Tulu, Kodagu, and Malto. The concept of the language family had been mooted by Sir William Jones a mere 30 years earlier, in 1786.
Today there are at least 26 languages recognized as Dravidian. Brahui, the farthest removed from southern India, occurs in Baluchistan (Pakistan). Brahui, along with Kurukh and Malto (spoken in Nepal and in the north Indian states of Bihar, Madhya Pradesh, and Orissa), falls into the North Dravidian subgroup. A Central Dravidian subgroup comprises six languages. The four Dravidian languages that have long literary traditions, namely Tamil, Kannada, Malayalam, and Telugu (pronounced T̪amɨʐ, Kannaɖa, Malayaaɭam, and T̪elugu) are included in the South Dravidian subgroups 1 and 2. These four languages account for over 182 million of the 237 million speakers of Dravidian languages enumerated in the 1991 Census of India. The morphosyntax of Dravidian is discussed here with respect to these four languages.1
The Dravidian languages all exhibit prototypical SOV properties, such as overt case on the noun, possibly followed by a postposition. Auxiliaries follow the verb, while adverbs precede it. Complementizers and subordinators are clause-final. The languages are rich in nominal and verbal inflection; Malayalam alone lacks verb agreement. Dravidianists have described its morphology as agglutinative and suffixing (“suffixation is the only type of affixation that occurs in Dravidian,” Zvelebil, 1990, p. 16); however, Aronoff and Sridhar (1988) argue for a small set of productive prefixes in Kannada (see also Tirumalesh, 1997, for a reiteration of the traditional position). The languages are characterized by argument drop, copula drop (except for Malayalam again), and flexible word order.
Section 2, “Normal Inflection: Gender, Number, and Person,” begins with the familiar topic of (pro)nominal inflection. The discussion is broadened to consider the morphological makeup of the Dravidian pronoun, and (given the idea of a universal merge order of elements in the DP, Cinque, 2005) notes a few facts relevant to a putative N-movement in the Dravidian DP. Section 3, “Nominal Inflection: Case,” touches on aspects of Dravidian case that are of general interest, such as case-layering (Caha, 2009) and the correlation of overt accusative case with the object’s animacy or specificity/definiteness.
A long-standing observation in areal or typological studies is the occurrence of dative case on experiencer “subjects” in south Asian languages. Section 4 (“Are There Adjectives in Dravidian?”) draws on research that situates this phenomenon in the paucity of the adjectival category in Dravidian, and the absence of have as a verb of possession.
Section 5, “Verbal Inflection,” turns to the verb: the causative, the light verb, and the reflexive constructions. A construction traditionally called the “conjunctive participle” has been analyzed as a serial verb, with the properties of participial adjuncts. The causative morpheme functions as a verbalizer, but there is evidence that the Kannada causative morpheme lexicalizes a notion of indirect causation.
Section 6 introduces recent research on the morphosemantics of quantifiers and questions in Dravidian. As in many other languages, the quantifiers and question words of Dravidian incorporate conjunctive and disjunctive particles. Their morphology appears to transparently reflect the underlying syntax and semantics of these words. This section also prefigures the discussion of finiteness in Section 7, noting that finite clauses cannot be conjoined in Dravidian.
Section 7, “Agreement and Finiteness,” outlines an argument that sentence-anchoring and tense-interpretation, which are standardly assumed to combine in the construct Tense, are separated in Dravidian. The anchoring elements identified are agreement, finite negation, and modals. The so-called tense morpheme, which (as in some other languages) is homophonous with the corresponding participial aspect morpheme, is argued to be in fact the latter, and thus a nonfinite element.
Finally, Section 8 concerns an argument that appears to challenge two fundamental notions about Dravidian—that it is a wh- in situ language, and that it is an SOV language—together with an alternative analysis of the facts that motivate this argument. The question word in Malayalam occurs in a position to the immediate left of the verb. This has been analyzed as a focus movement to a position within the IP, taking the underlying word order of Malayalam to conform to the (Kaynean) antisymmetric assumption of a universal VO order. The alternative analysis denies that wh-movement in Malayalam is a subcase of Focus movement, and proposes that the verb undergoes V-to-C movement.
2. Nominal Inflection: Gender, Number, and Person
2.1. Third-Person Gender-Number Inflection on the Pronoun and the Noun
A singular third-person pronoun in Dravidian consists of a deictic element suffixed with gender-number inflection. The latter is in boldface in Table 1.
Table 1. Singular third-person pronouns in Dravidian
Distal (a-) series
Proximal (i-) series
avan avaɭ adɨ
‘he’ ‘she’ ‘it’
ivan ivaɭ idɨ
‘he’ ‘she’ ‘it’
awan awaɭ adɨ
iwan iwaɭ idɨ
avanu avaɭu adu
ivanu ivaɭu idu
waaɖu aame adi
wiiɖu iime idi
Proto-Dravidian had a medial deictic element *u- in addition to distal a- and proximal i- (Krishnamurti, 2003, p. 253). Currently, the proximal pronouns take on deictic rather than pronominal functions (Jayaseelan & Hariprasad, 2001).
The plural paradigm neutralizes the gender distinction, the plural suffix signifying only [+/– human]. Proto-Dravidian distinguished exclusive and inclusive first-person plurals *ya(a)m and *ña(a)m; these forms “are preserved intact” in Central and North Dravidian (Krishnamurti, 2003, p. 248). Tamil has an exclusive ‘we’ naangaL and an inclusive ‘we’ naam (literary)/naambaL, traceable to proto South Dravidian. Telugu is said to have “restored” an inclusive plural manamu ‘we’ “by morphological innovation” (the exclusive ‘we’ is meemu).
Like the singular pronouns, the third-person plural pronouns consist of the plural suffix on a deictic element.
Wh- pronouns also inflect for gender and number. But the singular wh- forms may not always be in use (being seen as derogatory), or indeed may not exist.
On the noun as well, gender-number inflection can be identified, although it occurs less systematically. The nominal suffixes in (3) are traceable to proto-Dravidian (Zvelebil, 1990, pp. 20–21). Note the feminine singular suffix -i/-tti. The corresponding pronominal suffix -aɭ (introduced in Table 1) is an innovation in South Dravidian I, as are the Telugu feminine singular pronouns aa-me/ii-me innovations in South Dravidian 2.
2.2.Gender-Number Inflection on Postposed Numerals and Quantifiers
Pre-nominally, numerals and quantifiers are uninflected. Post-nominally, they inflect for a gender-number suffix (4). The noun may be said to move across the number word, triggering agreement on it. The numeral ‘one’ functions as the indefinite article in (4).
The numeral ‘two’ (and in Telugu, ‘three’) also has an inflected form: Tam./Mal. iruvar ‘two persons,’ Kan. ibbaru, Tel. iddaru. The inflected numerals all serve as pronouns (as do inflected quantifiers, cf. the discussion of (6–8) below). Larger numerals occur post-nominally with a [+human] noun or classifier word (Tam./Mal. peerɨ, Kan. jana, Tel. mandi).
Malayalam, which lacks verb agreement, displays the post-nominal agreement seen in (4). In the corresponding Kannada examples (5), only the numeral obb-, lexically specified [+human], can inflect and so occur post-nominally. The nonhuman numeral ondu ‘one,’ which cannot inflect, must occur only pre-nominally.
Gender-number inflection appears on the post-nominal numeral even if the noun is uninflected, for example, in loanwords: ondu sʈudenʈu ‘a student’ ~ sʈudenʈ obbanu/obbaɭu ‘a student (male/ female)’.
In (6) we see postposed quantifiers inflect for the plural.
The inflected quantifier can take the appropriate case inflection in a sentence (7), and occur as an independent argument (8). The inflected forms thus serve as indefinite pronouns.
2.3. Gender-Number Inflection on Postposed Adjectives and Predicate Adjectives
Only a small number of underived adjectives can be identified in Dravidian (see section (4)). Attributively, they do not agree with the noun (9a). Post-nominally, they display agreeing gender-number inflection. But adjectives do not readily postpose in the DP; an adjective seems to postpose within the DP only along with a numeral or quantifier (cf. (9b)).
Thus, an inflected but isolated post-nominal adjective is interpreted as a predicate adjective: compare (10) with (9b). (The equative copula is null in Dravidian, except in Malayalam: see (12)). The person inflection of a post-nominal adjective is held invariant at third person.
Telugu predicative adjectives (however) show person agreement. To account for this apparent challenge to the generalization that adjectives do not agree for person (Baker, 2008), Balusu (2014) proposes that agreement is located on a Pred head.3
2.4.Gender, Number, and Person in Verb Agreement
The third-person gender-number suffixes identified above appear in verb agreement (cf. (13)). Malayalam has no verb agreement. Verb agreement exhibits other person (i.e., first and second person) and tense inflections; individual language grammars may be consulted for the complete paradigms.
3. Nominal Inflection: Case
The nominative case marker is Ø. The genitive, accusative, and dative case markers are reconstructible to proto-Dravidian (Krishnamurti, 2003, p. 227). Other cases (e.g., locative, ablative, instrumental, and/or associative) are variously attested in all the languages.
3.1. The Oblique Stem
The non-nominative case markers attach to an “oblique stem” that is identical to the genitive-marked noun. Traditionally, the residual genitive case marker on the noun to which the oblique cases attach (see (15)) has been described as an “augment.” However, the layering of oblique cases on a genitive stem is clearly discernible in the pronominal paradigm:
The genitive or oblique stem is also seen in some inflected nouns. Its presence may (however) be obscured by morphophonemic changes in some paradigms. For example, in the Kannada examples in (15), the genitive stem is not obviously present in the accusative and dative paradigms, but it is identifiable in the locative and ablative paradigms, as also in the complement to the postposition:
Not all postpositions take a genitive noun complement. Some take a dative or accusative complement:
Traditionally, adpositions are said to “govern” the case on the noun they attach to. Caha (2009) includes prepositions into a case hierarchy, and suggests that the case of the preposition’s complement indicates the limits of N-movement in the hierarchy. Assuming a case hierarchy for Dravidian, if postpositions in Dravidian are indeed a separate category located among the cases, they must occur at least above the genitive case projection. Some P must occur as high as above the dative. (The Dravidian P is nominal or verbal in origin. Some putative P currently inflect for case, arguing that they are nouns: Kan. meele-(in)-inda ‘on/top- gen -from’ ‘from on top of.’)
3.2. The Accusative Case Marker
The overt realization of the accusative case marker is contingent on the animacy of the object or its specificity/definiteness, as noted for languages like Hindi and Turkish. For a discussion of the interaction of specificity with overt case and the syntactic position of the object in Kannada, see Lidz (2006).
3.3.The Dative Case in Dravidian Morphosyntax
It is well known that experiencers are case marked dative in the Dravidian languages (and in south Asian languages more generally). Long considered an areal, typological feature of a “dative subject,” it has been argued from a universalist perspective that dative case on experiencers and inalienable possessors correlates with the paucity of the adjectival category in Dravidian (Amritavalli & Jayaseelan, 2003).
4. Are There Adjectives in Dravidian?
Syntactic theory acknowledges that not all the lexical categories may be primitive, many languages lacking A and P (Hale & Keyser, 1993). The question whether underived adjectives exist in Dravidian has been extensively debated.5 In defense of their existence, a handful of attributive adjectives are usually cited, including the number and quantifier words discussed in Section 2.2; there are about 30 such proto-Dravidian roots.
Given the paucity of adjectives, the Dravidian noun fulfils adjectival functions, for example, in the dative experiencer construction (see Sections 4.1 and 4.2). Another very productive morphological strategy “adjectivalizes” or “adverbializes” a noun by suffixing it with -aa (Tam.), aayi- (Mal.), -aagi (Kan.), or -gaa (Tel.). The first three suffixes clearly have their provenance in the past participial form of a verb ‘to become’ in these languages; Telugu gaa is perhaps related to an infinitival form kaa of that verb. Balusu (2016) explores in detail the properties of the Telugu -gaa adjective.
4.1. Genitive and Dative Possessors, Experiencers and Properties
The dative experiencer construction (17) below is seen to differ from its English translation not only in the experiencer’s case (as commonly noted), but in the syntactic category that encodes the experience. The experience is realized as a noun, ‘anger,’ and not as an adjective ‘angry.’
Property nouns with an adjectival function (e.g., koopa ‘anger’ in (17))6 take the genitive case inflection when they occur attributively (18):
The occurrence of “possessed” or genitive nouns in an adjectival function has been noticed elsewhere (see Koontz-Garboden & Francez’s (2010) discussion of Ulwa), and indeed a vestigial of-genitive adjectival construction is seen even in English: cf. a man of property/honor, circle of light, ring of gold, words of wisdom, and the translations in (18).
Ulwa apparently uses the same possessive morphosyntax for property nouns in attributive as well as predicative positions. Dravidian and English use a different possessive morphosyntax for predicative nouns. English uses the have construction (The man has property). Dravidian uses the (so-called) dative-of-possession construction, instantiated in (19):
Comparing (19) with (17), the dative experiencer construction is seen to be an instance of a predicative possessive construction in Dravidian (‘He has anger.’). The next section further explores its syntax.
4.2.A “Floating” Dative Case
In Kannada, a set of nouns denoting physical properties, which occur attributively in the genitive case (20), appear predicatively in the dative case (21).
The construction in (21) contrasts with the dative experiencer construction (17). In (17), the predicative property noun is bare, and the experiencer is dative-marked. In (21), dative case appears on the predicative property noun, and the subject is nominative. The alternating realization of dative case on either the predicative property noun, or the experiencer/possessor “subject,” is clearly seen in (22–23):
In Kannada, as in English, (23) is fully idiomatic in a context where a requisite height—its presence or absence—is the topic of discourse. Under similar discourse conditions, the sentences in (21) can also occur in the “dative experiencer” or “dative subject” construction: raste-ge agala ide/illa ‘The road does/does not have (the) width.’
Taken together, these data suggest that dative case plays a seminal role in the possessor-experiencer construction in languages. First, there is a (Kaynean) hypothesis that the verb have derives from be, by the incorporation of dative case. Second, in have-less languages where dative case does not incorporate into be, it appears on the possessor argument of be (as in the Dravidian possessive construction). The complement of be is here a property noun that discharges the adjectival function. Third, in Kannada, perhaps uniquely, dative case manifests on the property noun. The subject of be now occurs in the nominative case, as in languages with adjectives, suggesting that dative case serves to “adjectivalize” the noun.7
The dative experiencer/possessor construction has been argued to generalize with the double object construction in English, where a covert possessive verb relates the indirect and direct objects (Amritavalli, 2014a; Harley, 2002).
4.3. Other Theoretical Accounts of Dative Case
In Malayalam, the verbalizing and causativising suffix-ik’k’ is homophonous with dative case. Pursuing the leading idea of reducing the number of lexically primitive categories, Jayaseelan (2013) suggests that the Malayalam verb originates as a nominal that incorporates into dative case. Amritavalli (2007b) explores the role of dative case in Kannada in building up expressions denoting a region or place, designated ‘axial parts’ by Svenonius (2006) (e.g., in front of), from part-denoting nouns or stems (e.g., the front of).
5. Verbal Inflection
5.1.1. Suffixal Causatives
The suffixes -ik’k’ -ipp, -is, and -inc/-imp in Malayalam, Kannada, and Telugu (respectively) are commonly deemed “causative,” as they occur on transitive verbs to derive causative verbs:
It has been noted that these suffixes may be mere verbalizers. They derive verbs from category-neutral stems (25), and from nouns, especially of Sanskrit origin (26):
5.1.2. Light Verb Causatives
Tamil uses the light verbs vey- ‘put’ and cey-, paɳɳ- ‘do’ for causativization. The light verb takes an infinitival complement. (An older causative suffix -vi/-(p)pi is seen in some verbs; Lehmann, 1993, p. 52.)
Lehmann (1993, pp. 219ff.) suggests that the Tamil causative is a “simplex clause” with “verb raising,” noting that in sentences like (27), the “inner verb” (‘study’) does not assign nominative case to its subject ‘he.’ The surface case-frame displayed by the causatives of transitives for the Tamil light verb causative, no less than the Kannada suffixal causative, falls in with the so-called Type I causatives (Baker, 1988). In Type I causatives, the causee argument of the inner, transitive verb is optional. If present, it gets an oblique (e.g., instrumental) case. The object of the inner verb surfaces with accusative case. Thus in (28), the beatee is accusative, and the beater-causee optionally surfaces as an instrumental:
Suffixal causative languages also have available to them the light verb strategy of causativization or verbalization. Indeed, it is this strategy all four languages discussed here employ to license English verbs (borrowed as stems): apply ceyy-/maaɖ-/ceey- ‘apply’ (lit. ‘do apply’). The suffixal causative cannot occur on these borrowed stems: *apply-ik’k’/-is/-inc.
5.1.3. Transitive Morphology; Direct and Indirect Causation
In Tamil and Telugu, intransitive and transitive verbs display stem alternants with morphology resembling causative morphology, traceable to proto-Dravidian tense and voice morphology (Krishnamurti, 2003, pp. 278ff.):
Kannada (which does not have such stem alternants) relates some intransitive-transitive pairs with the causative suffix -is, leading to a suggestion that this morpheme is a lexicalization of the “little v” node in syntax (Lidz, 2004). A problem for this account is that unergative intransitive verbs, standardly considered to project a v node, never manifest -is.
Second, not all transitives of inchoatives (change-of-state verbs) manifest -is, forcing Lidz to hypothesize that -is must be lexically selected. However, what licenses -is in transitive verbs may be the indirect causation of self-propelled action by the causee (Kodandaram, 2015), as opposed to direct causation. Thus, the transitives of unergative verbs are always marked by –is:
Among transitives of inchoatives, the boiling of water is marked with -is (31a), whereas the spilling of water is not (31b). That is, the unaccusative verb ‘boil’ appears to differ from ‘spill’ in denoting an action that inheres in the causee, and so ‘boil’ patterns with unergative verbs, requiring -is in its transitive occurrence. (Interestingly, the English glosses suggest that the causative verb make similarly distinguishes between these two inchoative verbs.)
5.2. Verb Serialization and the Light Verb
This section considers a set of participle-plus-verb constructions, grouping them into a “serial verb” or “participial adjunct” construction and a “light verb” construction. (The Malayalam and Kannada examples are from Jayaseelan, 2004, and Amritavalli, 2007a, respectively.)
5.2.1. The Serial Verb
A pervasive feature of Dravidian is a (potentially infinite) string of participles that precedes a final, finite verb, to indicate a series of actions or events.
Only the final verb inflects for tense: past tense (as in (32)), or the present (33a), or the future (33b).
The citation form of this construction instantiates perfect participles (as in the examples above), and this is admittedly the most commonly attested type of serial verb. But the construction also permits the imperfect and the negative participles. The latter have participial adjunct counterparts in English.
Each participle in this construction can have its own object and/or adverbial modifiers, arguing that the “serial verb” must consist of at least VPs (verb phrases):
In the majority of instances, the subject is shared by all the verbs, suggesting that the construction instantiates clauses with controlled PRO subjects. The clausal analysis receives support from the few instances where each verb has a lexical subject. Sridhar (1990, p. 237) observes that nonidentical subjects typically occur where the clauses stand in a cause–effect relation:
Jayaseelan points out that it is the perfect aspect morpheme iʈʈɨ in (39) that licenses a lexical subject for the participle. He draws a parallel with the English gerundial “absolutive” construction.
5.2.2. The Light Verb
The structurally similar light verb construction differs from the serial verb construction (SVC) in that only one verb—the first verb—in the light verb construction has its lexical meaning. There may be up to two light verbs (40b, 40c) that add aspectual meaning. Only the final verb is finite.
Sentences may be ambiguous between an SVC and a light verb construction.
5.2.3. “Self-Benefactive” koɭɭ- and “Non-Self-Benefactive” koɖ-
In Tamil, Kannada, and Telugu, a verb koɭɭ- (literally ‘take’) and its variants appear in a light verb position. This element indicates that the action of a predicate is for the benefit of the subject. It also licenses strictly local anaphora in reflexive sentences.
Unlike the aspectual light verbs in (41), koɭɭ- takes only a past participle complement. It can be followed by another auxiliary or light verb. Tense-agreement morphology may appear on koɭɭ-, or the following element. Some dialects of Tamil use kiʈʈ- (perhaps a form of an older verb ‘get’) instead of koɭɭ-.
Let us first differentiate the forms under discussion from a progressive auxiliary koɭɭ-/kiʈʈ-, which occurs also in Malayalam (example from Jayaseelan, 2004):
Malayalam however lacks the self-benefactive koɭɭ- illustrated in (45). (All koɭɭ- examples below are from Kannada.)
Self-benefactive koɭɭ- has a morphosyntactic reflex. In (46), the subject (who does the work for her own benefit) is understood as the “possessor” of the work. An overt possessor can occur, provided it is the first person; that is, a possessive pronoun disjoint with the subject cannot occur.
Koɭɭ- ‘take’ has a converse counterpart koɖ- ‘give.’ In the koɖ- sentence (47a), the “possessor” of the work is understood as disjoint in reference with the subject, and any overt possessors must be disjoint with the subject. In (47b), a dative argument is licensed as the beneficiary. This argument, the possible “possessor” of the work, is again disjoint in reference with the subject.
Both koɭɭ- and koɖ- occur as parts of words where their meanings are lexicalized: for example, karedu-koɭɭ- ‘call-koɭɭ’ ‘bring along (animate object),’ heeLi-koɖ- ‘tell-give’ ‘teach.’
Koɭɭ- may occur in unaccusative sentences, such as (48). Its semantic contribution in such sentences is not well understood (for some discussion, see Lidz, 2004).
Koɭɭ- must occur in reflexive sentences to license a local antecedent for the long-distance anaphor taan- ‘self.’ (See the separate article on anaphora in Dravidian for more on taan.)
Sundaresan (2012) analyzes ‘reflexive’ koɭɭ- as a restructuring verb.
6. Questions, Quantifiers, and the Conjunctive and Disjunctive Particles
The particles -um and -oo, and their variants, appear in Dravidian as conjunction and disjunction markers and in quantifiers formed from question words. -Um is also an emphatic particle. -Oo occurs in questions and also in correlatives.
The morphosyntax of question words and quantifiers in Dravidian has been argued to be a transparent reflection of the semantics of these elements.
6.1. Conjunction and Disjunction
The conjunction and disjunction markers appear on each conjunct or disjunct. They follow the case marker or postposition, if any.
Dravidian does not allow the conjunction of finite clauses (see Jayaseelan, 2014, and references).
When finite clauses are coordinated by the disjunction marker, the coordinated clauses are interpreted as questions (Amritavalli, 2003).
6.2. Quantifiers from Question Words
In Dravidian, as in many other languages (Bangla, Chinese, Japanese), quantifiers are found to incorporate question words (the Malayalam examples below are from Jayaseelan, 2011). The disjunction morpheme -oo suffixed to a question word yields an existential quantifier; the conjunction morpheme -uu suffixed to a question word yields a universal quantifier. The latter is polarity sensitive.
There is a second form of the universal quantifier, which incorporates a conditional morpheme.
As the glosses indicate, the two universal quantifiers occur in different domains. In Kannada, wh-uu forms are negative polarity items; wh-if-uu forms occur in modal or generic contexts.
Malayalam (however) allows some wh-uu forms to occur in modal/generic contexts [compare (56) with (55a)], and prohibits some wh-uu forms in negative contexts (57). [In (56), the modal of permission -aam induces dative case on the subject.]
In polarity questions, only wh-if-um forms are licensed, in Malayalam and in Kannada.
6.2.1. Distributive Quantification by Reduplication
In the Dravidian languages, reduplication of numerals and pronouns gives rise to distributive meanings. The Telugu example (59) has the three kinds of possible distributive readings (60) (Balusu & Jayaseelan, 2013):
A similar sentence without the reduplicated numeral (61) has only the interpretations in (62):
Dravidian also reduplicates the plural third-person pronoun to form a distributive quantifier that is a strict (local) anaphor, illustrated for Malayalam in (63) (from Balusu & Jayaseelan, 2013):
The antecedent of awar-awar need not be the subject, as in:
6.3. Um as an Emphatic Particle
Across languages, emphasis is marked by conjunction markers (e.g., Japanese mo). Dravidian um- also conveys emphasis, or an “additive” meaning, illustrated in (65) (Jayaseelan, 2016):
The conjunction -um also forms what have been called “minimizer” negative polarity items:
6.4. Questions and Disjunctions
In Malayalam, the disjunction marker -oo is also the yes–no question particle. This homophony, which holds in many other languages (e.g., Sinhala, Japanese, Dutch), arguably reflects the syntax and semantics of questions (Jayaseelan, 2012). In Kannada, the disjunction marker -oo is restricted to embedded questions and needs licensing (Amritavalli, 2003, 2013a).
7. Agreement and Finiteness
Traditionally, agreement has been considered the marker of finiteness in Dravidian. The question of what constitutes finiteness in Kannada was first raised afresh in Amritavalli (2000), who noted and discussed the problem of the finite negative clause in Kannada. This clause attests a nonfinite matrix verb complement—an infinitive, or a gerund—to a finite negative element. The negative clause has a tense interpretation corresponding to each of the two nonfinite verb forms. The facts argue for separating the two elements that constitute Tense as standardly understood: the temporal interpretation of the verb, and the finiteness marker in the sentence.
Subsequently, Amritavalli and Jayaseelan (2005) and Amritavalli (2014b) have enlarged upon the argument for separating finiteness, or sentence-anchoring, from tense. In Tamil, Telugu, and Kannada, finite negative clauses assume quite a different form than their affirmative counterparts; finite negative clauses lack agreement, the verb (as already mentioned for Kannada) is nonfinite, and it may or may not have a tense interpretation—in Tamil, thus, the negative clause is entirely free with respect to tense interpretation. Negation has different forms in finite and nonfinite sentences; the finite negative is the anchoring element in finite negative clauses. Dravidian, it is argued, anchors the sentence with agreement, or with a modal, or with a finite negative element. What looks like a tense morpheme is therefore an aspect morpheme that appears in a participial verb.
Malayalam is argued to conform to the general Dravidian pattern. Although its affirmative and negative sentences do not differ in the manner described, Malayalam, too, provides for finite and nonfinite forms of negation. The difference is that it has no overt agreement. In the three languages that have agreement morphemes, the participial forms of the verb are selected only by the agreement element. They appear to be positive polarity items, and the finite negative must thus select other nonfinite verb forms (gerunds and infinitives). In Malayalam, however, the loss of agreement has led to a loss of polarity specification for the participial forms, which are therefore free to occur in positive and negative sentence contexts.
Jayaseelan (2014), adopting the view of finiteness outlined above, explains the impossibility of conjoining finite clauses (noted in Section 6.1) in terms of a “narrow C” in Dravidian.
7.1. ‘Monstrous Agreement’
An interesting fact about agreement in Dravidian is that the verb in a clause embedded under a speech predicate may show first-person agreement, notwithstanding the remote antecedent being a matrix third-person subject.
It can be shown that the agreement in the embedded clause is not due to a shift to direct speech in it. Sundaresan (2013) examines this phenomenon in Tamil, which she labels "monstrous agreement,” as an instance of indexical shift; she proposes the enrichment of the left periphery with contextual features, to explain its range of properties.
For agreement in languages of the Central Dravidian group (Munda, Kuvi, Kui, Pengo), see Ramakrishna Reddy (2003).
8.Wh-movement and Word Order in Dravidian
A well-known characteristic of SOV languages is that the question word does not move to the position where the complementizer occurs, that is, to the clause-final position. Dravidian is typical in this respect, and has therefore standardly been considered to be a wh-in-situ language. However, it has been observed for questions in Malayalam (Jayaseelan, 2003) that a wh-phrase does not in fact remain in the canonical position of the corresponding argument. The Malayalam question word moves to a position to the immediate left of the verb. This is seen from the licit and illicit word orders in the declarative and interrogative sentences below:
The question word in Malayalam appears to obligatorily “scramble” rightward, next to the verb. Similar facts hold for the other Dravidian languages, which show—if not an obligatory movement, as in Malayalam—at the very least a strong tendency to move the question word next to the verb.
These facts also have a bearing on word order in Dravidian. If the surface order, SOV, is taken to be the underlying order, the subject appears to “lower” into the clause, moving rightward into a Comp-like position between the direct object and the verb (an impossible movement under current theoretical assumptions).
The solution Jayaseelan (2003) proposes makes two crucial assumptions. The first is that question words move into a focus position, and that there is a focus position immediately to the left of the verb (an “IP internal” position). The second is that, in accordance with antisymmetry (Kayne, 1994), only the VO order is a possible underlying order, and the OV order that obtains on the surface must be due to movement of the verb’s arguments out of the VP, to the left.
Mathew (2015) argues against analyzing wh-movement in Malayalam as a subcase of focus movement. She proposes that the verb undergoes V-to-C movement, and the wh-word does not need to move. Mathew differs also in her account of what Jayaseelan (2003) calls a cleft construction (a default strategy for question formation in Malayalam). She analyzes the verb aaɳɨ not as a cleft copula, but as an auxiliary.
Bayer, J., Bhattacharya, T., & Hany Babu, M. T. (Eds.) (2007). Linguistic theory and South Asian languages: Essays in honour of K. A. Jayaseelan. Amsterdam: John Benjamins.Find this resource:
Bhaskararao, P., & Subbarao, K. V. (Eds.) (2004). Non-nominative Subjects (vols. 1 and 2). Amsterdam: John Benjamins.Find this resource:
Bhat, D. N. S. (1994). The adjectival category. Amsterdam: John Benjamins.Find this resource:
Bhat, D. N. S. (1999). The prominence of tense, aspect and mood. Amsterdam: John Benjamins.Find this resource:
Bhat, D. N. S. (2002). Grammatical relations. London and New York: Routledge.Find this resource:
Dayal, V., & Mahajan, A. (Eds.). (2004). Clause structure in South Asian languages. Boston: Kluwer.Find this resource:
Hariprasad, M., Nagarajan, H., Madhavan, P., & Vijayakrishnan, K. G. (Eds.). (1997). Phases and interfaces of morphology (vols. 1 and 2). Hyderabad: CIEFL.Find this resource:
Jayaseelan, K. A., & Amritavalli, R. (2017). Dravidian syntax and universal grammar. New York: Oxford University Press.Find this resource:
Lust, B., Wali, K., Gair, J., & Subbarao, K. V. (Eds.). (2000). Lexical anaphors and pronouns in selected South Asian languages: A principled typology. The Hague: Mouton.Find this resource:
Schiffman, H. F. (1999). A reference grammar of spoken Tamil. Cambridge, U.K.: Cambridge University Press.Find this resource:
Schiffman, H. F. (1979). A grammar of spoken Tamil. Original from Christian Literature Society, the University of California, Digitized 4 Mar 2008. Retrieved from ccat.sas.upenn.edu/plc/tamilweb/book.html.
Steever, S. B. (1988). The serial verb formation in the Dravidian languages. Delhi: Motilal Banarsidass.Find this resource:
Steever, S. B. (Ed.). (1988). The Dravidian languages. London and New York: Routledge.Find this resource:
Steever, S. B. (1993). Analysis to synthesis: The development of complex verb morphology in the Dravidian languages. Oxford: Oxford University Press.Find this resource:
Subrahmanyam, P. S. (2013). The morphosyntax of Dravidian languages. Thiruvananthapuram: Dravidian Linguistics Association.Find this resource:
Amritavalli, R. (2001). Morphological causatives, UTAH and monoclausality. In A. Abbi, R. S. Gupta, & A. Kidwai (Eds.), Linguistic structure and language dynamics in South Asia: Proceedings of the SALA XVIII Roundtable (pp. 388–400). Delhi: Motilal Banarsidass.Find this resource:
Amritavalli, R. (2006). Review of Krishnamurti, The Dravidian Languages. Lingua, 116, 2254–2258.Find this resource:
Balusu, R. (2014). Lexical semantics of transitivizer light verbs in Telugu. In P. Chandra & R. Srishti (Eds.), The lexicon-syntax interface: Perspectives from South Asian languages (pp. 101–126). Amsterdam: John Benjamins.Find this resource:
Hany Babu, M. T., & Madhavan, P. (2007). Argument structure and aspect: The case of two imperfectives in Malayalam. In E. Reuland, T. Bhattacharya, & G. Spathas (Eds.), Argument structure (pp. 105–120). Amsterdam: John Benjamins.Find this resource:
Jayaseelan, K. A. (1991). Review of Steever (1988). Lingua, 29, 543–548.Find this resource:
Lidz, J. (2004). Causation and reflexivity in Kannada. In V. Dayal & A. Mahajan (Eds.), Clause structure in South Asian languages (pp. 93–130). Boston: Kluwer.Find this resource:
Madhavan, P. (2001). Review of Asher and Kumari (1997). Lingua, 111, 629–638.Find this resource:
Mohanan, K. P. (1991). Malayalam. In B. Comrie, W. Bright, & D. Crystal (Eds.), Oxford Encyclopaedia of Linguistics. Oxford: Oxford University Press.Find this resource:
Mohanan, K. P., & Mohanan, T. (2009). Multiple tenses in the Malayalam word. In S. Inkelas & K. Hansen (Eds.), The nature of the word: Essays in honor of Paul Kiparsky. Cambridge, MA: MIT Press.Find this resource:
Subrahmanyam, P. S. (2015). The Dravidian family: Characteristic features. In G. K. Panikkar, B. Ramakrishna Reddy, K. Rangan, & B. B. Rajapurohit (Eds.), V. I. Subramoniam Commemoration Volume I, Studies on Dravidian (pp. 19–42). Thiruvananthapuram: International School of Dravidian Linguistics.Find this resource:
Zvelebil, K. V. (2003). Dravidian Languages. Encyclopaedia Britannica. Retrieved from Encyclopaedia Britannica Online. http://www.search.eb.com/eb/article?eu=118132
Amritavalli, R. (2000). Kannada clause structure. In R. Singh (Ed.), Yearbook of South Asian languages (pp. 11–30). New Delhi: SAGE.Find this resource:
Amritavalli, R. (2003). Question and negative polarity in the disjunction phrase. Syntax, 6(1), 1–18.Find this resource:
Amritavalli, R. (2007a). Syntactic categories and lexical argument structure. In E. Reuland, T. Bhattacharya, & G. Spathas (Eds.), Argument structure (pp. 49–61). Amsterdam: John Benjamins.Find this resource:
Amritavalli, R. (2007b). Parts, axial parts and next parts in Kannada. In M. Baˇsi´c, M. Pantcheva, M. Son, & P. Svenonius (Eds.), Tromsø Working Papers on Language & Linguistics: Nordlyd 34.2, special issue on Space, Motion, and Result, (pp. 86–101). CASTL, Tromsø. Retrieved from http://www.ub.uit.no/baser/nordlyd/http://septentrio.uit.no/index.php/nordlyd/issue/view/12Find this resource:
Amritavalli, R. (2013a). Nominal and interrogative complements in Kannada. In Y. Miyamoto, D. Takahashi, H. Maki, M. Ochi, K. Sugisaki, & A. Uchibori (Eds.), Deep insights, broad perspectives: Essays in honor of Mamoru Saito (pp. 1–21). Tokyo: Kaitakusha.Find this resource:
Amritavalli, R. (2013b). Result phrases and dative experiencers in Kannada. Nineteenth ICL (International Congress of Linguists), Geneva. Geneva: Département de Linguistique de l’Université de Genève. Retrieved from www.cil19.org/Find this resource:
Amritavalli, R. (2014a). Rich results. In P. Chandra & R. Srishti (Eds.), The lexicon-syntax interface: Perspectives from South Asian languages (pp. 71–100). Amsterdam: John Benjamins.Find this resource:
Amritavalli, R. (2014b). Separating tense and finiteness: Anchoring in Dravidian. Natural Language and Linguistic Theory, 32, 283–306.Find this resource:
Amritavalli, R., & Jayaseelan, K. A. (2003). The genesis of syntactic categories and parametric variation. In H.-J. Yoon (Ed.), Generative grammar in a broader perspective: Proceedings of the 4th GLOW in Asia 2003 (pp. 19–41). Seoul: Hankook: KGGC and Seoul National University.Find this resource:
Amritavalli, R., & Jayaseelan, K. A. (2005). Finiteness and negation in Dravidian. In G. Cinque and R. S. Kayne (Eds.) The Oxford handbook of comparative syntax (pp. 178–220). New York: Oxford University Press.Find this resource:
Aronoff, M., & Sridhar, S. N. (1988). Prefixation in Kannada. In M. Hammond & M. Noonan (Eds.), Theoretical morphology (pp. 179–191). San Diego, CA: Academic Press.Find this resource:
Asher, R. E., & Kumari, T. C. (1997). Malayalam. London and New York: Routledge.Find this resource:
Baker, M. C. (1988). Incorporation: A theory of grammatical function changing. Chicago: University of Chicago Press.Find this resource:
Baker, M. C. (2008). The syntax of agreement and concord. Cambridge, U.K.: Cambridge University Press.Find this resource:
Balusu, R. (2014). Person agreement in adjectival predication in Telugu. Paper presented at the LISSIM 8 Workshop, Solang, June 2, 2014.Find this resource:
Balusu, R. (2015). Comparison, predication, and lexical semantics of PC nouns in Telugu. Indian Linguistics, 76(1–2), 1–18.Find this resource:
Balusu, R. (2016). The eventive predicator -gaa in Telugu. Linguistic Analysis, 40, 3–4.Find this resource:
Balusu, R., & Jayaseelan, K. A. (2013). Distributive quantification by reduplication in Dravidian. In K.-H. Gil, S. Harlow, & G. Tsoulas (Eds.), Strategies of quantification (pp. 60–86). Oxford: Oxford University Press.Find this resource:
Caha, P. (2009). The nanosyntax of case (PhD diss., University of Tromsø).Find this resource:
Caldwell, R. (1856). A comparative grammar of the Dravidian or South-Indian family of languages. London: Harrison.Find this resource:
Cinque, G. (2005). Deriving Greenberg’s universal 20 and its exceptions. Linguistic Inquiry, 36, 315–332. Retrieved from http://www.mitpressjournals.org/doi/abs/10.1162/0024389054396917#.WMJeQdKGOUkFind this resource:
Francez, I., & Koontz-Garboden, A. (2013). Semantic variation and the grammar of property concepts. Ms.Find this resource:
Harley, H. (2002). Possession and the double object construction. Linguistic Variation Yearbook, 2, 29–68.Find this resource:
Kayne, R. S. (1994). The antisymmetry of syntax. Cambridge, MA: MIT Press.Find this resource:
Kodandaram, S. (2015). The light verb in Kannada in a minimalist framework. Paper presented at the 37th International Conference of the Linguistic Society of India (ICOLSI), JNU, New Delhi, October 15–17.Find this resource:
Koontz-Garboden, A., & Francez, I. (2010). Possessed properties in Ulwa. Natural Language Semantics, 18, 197–240.Find this resource:
Hale, K., & Keyser, S. J. (1993). On argument structure and the lexical expression of syntactic relations. In K. Hale & S. J. Keyser (Eds.), The View from Building 20: Essays in honor of Sylvain Bromberger (pp. 53–109). Cambridge, MA: MIT Press.Find this resource:
Jayaseelan, K. A. (2001). Questions and question-word incorporating quantifiers in Malayalam. Syntax, 4(2), 63–93.Find this resource:
Jayaseelan, K. A. (2003). Question words in focus positions. Linguistic Variation Yearbook, 3, 69–99.Find this resource:
Jayaseelan, K. A. (2004). The serial verb construction in Malayalam. In V. Dayal & A. Mahajan (Eds.), Clause structure in South Asian languages (pp. 67–91). Boston: Kluwer.Find this resource:
Jayaseelan, K. A. (2011). Comparative morphology of quantifiers. Lingua, 121(2), 269–286.Find this resource:
Jayaseelan, K. A. (2012). Question particles and disjunction. Linguistic Analysis, 38(1–2), 35–51.Find this resource:
Jayaseelan, K. A. (2013). The dative case in the Malayalam verb. In Y. Miyamoto, D. Takahashi, H. Maki, M. Ochi, K. Sugisaki, & A. Uchibori (Eds.), Deep insights, broad perspectives: Essays in honor of Mamoru Saito (pp. 139–166). Tokyo: Kaitakusha.Find this resource:
Jayaseelan, K. A. (2014). Coordination, relativization and finiteness in Dravidian. Natural Language and Linguistic Theory, 32, 191–211.Find this resource:
Jayaseelan, K. A. (2016). Decomposing coordination: The two operators of coordination. Linguistic Analysis, 40, 3–4.Find this resource:
Jayaseelan, K. A., & Hariprasad, M. (2001). Deixis in pronouns and noun phrases. Linguistic Analysis, 31(1–2), 132–149.Find this resource:
Krishnamurti, B. (2003). The Dravidian languages. Cambridge, U.K.: Cambridge University Press.Find this resource:
Lehmann, T. (1993). A grammar of modern Tamil. Pondicherry: Pondicherry Institute of Linguistics and Culture.Find this resource:
Lidz, J. (2004). Causation and reflexivity in Kannada. In V. Dayal & A. Mahajan (Eds.), Clause structure in South Asian languages (pp. 93–130). Boston: Kluwer.Find this resource:
Lidz, J. (2006). The grammar of accusative case in Kannada. Language, 82(1), 10–32.Find this resource:
Mathew, R. (2015). Head movement in syntax. Amsterdam: John Benjamins.Find this resource:
Menon, M. (2014). Property concepts and the apparent lack of adjectives in Dravidian. In P. Chandra & R. Srishti (Eds.), The lexicon-syntax interface: Perspectives from South Asian languages (pp. 25–52). Amsterdam: John Benjamins.Find this resource:
Ramakrishna Reddy, B. (Ed.). (2003). Agreement in Dravidian languages. Chennai: International Institute of Tamil Studies.Find this resource:
Sridhar, S. N. (1990). Kannada. London and New York: Routledge. Reprinted (2007) as Modern Kannada grammar. Delhi: Manohar.Find this resource:
Sundaresan, S. (2012). Context and (co)reference in the syntax and its interfaces (PhD diss., Universities of Stuttgart and Tromso).Find this resource:
Sundaresan, S. (2013). Monstrous agreement, anaphora, and context: Evidence from Tamil. In Proceedings of the 41st Annual Meeting of the North East Linguistic Society, University of Pennsylvania, PA.Find this resource:
Svenonius, P. (2006). The emergence of axial parts. In P. Svenonius (Ed.), Tromsø working papers on language and linguistics: Nordlyd 33.1, special issue on Adpositions (pp. 1–22). Tromsø: University of Tromsø.Find this resource:
Tirumalesh, K. V. (1997). Kannada prefixation. PILC Journal of Dravidian Studies, 7(2), 201–208.Find this resource:
Zvelebil, K. V. (1990). Dravidian linguistics: An introduction. Pondicherry: Pondicherry Institute of Linguistics and Culture.Find this resource:
(3.) Predicative adjectives like oɭɭeyavanu (lit. ‘good-he’) are sometimes said to “incorporate” a pronoun avanu ‘he’ (e.g., Sridhar, 1990, p. 275). But (in Kannada as in English) pronouns do not co-occur with attributive adjectives: *oɭɭeya naanu/niinu ‘good me/you.’ The putative third-person “pronoun” in the predicate adjective is not referential. Inflected adjectives serve as indefinite noun phrases (oɭɭeyavanu beeku ‘A good man is called for’), and pronominal inflection occurs also in headless relative clauses: baruv-a avaru barali ‘come-relative ppl.-they come-let’ ‘Let whoever comes, come; let those who come, come.’
(5.) “Among those who tend to deny or do deny the existence of adjectives as a separate ‘part-of-speech,’ the most prominent are Jules Bloch and M.S. Andronov. Master, Burrow, and Zvelebil, on the other hand, accept adjectives as a separate word-class” (Zvelebil, 1990, p. 27, n. 75).
(7.) This morphosyntactic account (due to Amritavalli & Jayaseelan, 2003) of the syntactic frames in which adjectives and property nouns occur may be said to anticipate (and perhaps go beyond) the “Lexical Semantic Variation Hypothesis” (Francez & Koontz-Garboden, 2013), which assumes the lexical categories of nouns and adjectives, and attributes the occurrence of possessive morphosyntax with only nouns to a semantic requirement: “Substance denoting P(roperty)C(oncept) lexemes require possessive semantics to achieve the truth conditions of PC constructions. Adjectivally denoting PC lexemes do not. Possessive morphosyntax contributes possessive semantics, and hence surfaces only with PC constructions built on substance denoting PC lexemes.”